[llvm-branch-commits] [llvm] [RISCV] Update the getCSRCost from 5 to 32768 (PR #239534)

Piyou Chen via llvm-branch-commits llvm-branch-commits at lists.llvm.org
Sun Nov 8 20:34:42 PST 2026


https://github.com/BeMg updated https://github.com/llvm/llvm-project/pull/239534

>From 233f02a8f5bb2e34fc917e68f52492a046e85588 Mon Sep 17 00:00:00 2001
From: Piyou Chen <piyou.chen at sifive.com>
Date: Thu, 5 Nov 2026 23:18:30 -0800
Subject: [PATCH] [RISCV] Update the getCSRCost value

---
 llvm/lib/Target/RISCV/RISCVRegisterInfo.h     |    5 +-
 .../RISCV/GlobalISel/double-convert.ll        |  135 +-
 .../CodeGen/RISCV/GlobalISel/fastcc-float.ll  |   76 +-
 .../CodeGen/RISCV/GlobalISel/float-convert.ll |  323 +-
 llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll   |  956 ++--
 .../CodeGen/RISCV/GlobalISel/rotl-rotr.ll     |   51 +-
 .../RISCV/GlobalISel/rv64-double-convert.ll   |   56 +-
 .../RISCV/GlobalISel/rv64-float-convert.ll    |   56 +-
 llvm/test/CodeGen/RISCV/GlobalISel/rv64zbb.ll |   26 +-
 llvm/test/CodeGen/RISCV/GlobalISel/shifts.ll  |  179 +-
 ...lar-shift-by-byte-multiple-legalization.ll |  600 +-
 .../CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll |    8 +-
 .../CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll |    8 +-
 llvm/test/CodeGen/RISCV/bfloat-convert.ll     |   57 +-
 llvm/test/CodeGen/RISCV/bitint-fp-conv-200.ll | 2200 ++++----
 llvm/test/CodeGen/RISCV/calling-conv-half.ll  |   28 +-
 llvm/test/CodeGen/RISCV/clmulh.ll             |  102 +-
 llvm/test/CodeGen/RISCV/clmulr.ll             |  446 +-
 llvm/test/CodeGen/RISCV/condops.ll            |   85 +-
 llvm/test/CodeGen/RISCV/ctlz-cttz-ctpop.ll    |  113 +-
 .../CodeGen/RISCV/ctz_zero_return_test.ll     |   86 +-
 llvm/test/CodeGen/RISCV/double-convert.ll     |  267 +-
 .../CodeGen/RISCV/double-round-conv-sat.ll    |  139 +-
 .../RISCV/double-stack-spill-restore.ll       |   29 +-
 .../RISCV/exception-pointer-register.ll       |   20 +-
 llvm/test/CodeGen/RISCV/float-convert.ll      |  309 +-
 .../CodeGen/RISCV/float-round-conv-sat.ll     |  318 +-
 llvm/test/CodeGen/RISCV/fp128.ll              |  343 +-
 llvm/test/CodeGen/RISCV/fpclamptosat.ll       | 4906 +++++++++--------
 llvm/test/CodeGen/RISCV/fpenv.ll              |   42 +-
 llvm/test/CodeGen/RISCV/half-convert.ll       |  563 +-
 .../test/CodeGen/RISCV/half-round-conv-sat.ll |  456 +-
 llvm/test/CodeGen/RISCV/idiv_large.ll         | 1607 +++---
 .../CodeGen/RISCV/interrupt-attr-nocall.ll    |  100 +-
 ...e-outliner-and-machine-copy-propagation.ll |   16 +-
 .../RISCV/machine-sink-load-immediate.ll      |  128 +-
 llvm/test/CodeGen/RISCV/miss-sp-restore-eh.ll |    9 +-
 .../test/CodeGen/RISCV/overflow-intrinsics.ll |  137 +-
 .../CodeGen/RISCV/qci-interrupt-attr-fpr.ll   |  280 +-
 llvm/test/CodeGen/RISCV/qci-interrupt-attr.ll |   40 +-
 .../RISCV/regalloc-csr-cost-patterns.ll       |  695 ++-
 llvm/test/CodeGen/RISCV/rv32xtheadbb.ll       |   70 +-
 llvm/test/CodeGen/RISCV/rv32zbb.ll            |   70 +-
 .../test/CodeGen/RISCV/rv64-double-convert.ll |   93 +-
 llvm/test/CodeGen/RISCV/rv64-float-convert.ll |   93 +-
 llvm/test/CodeGen/RISCV/rv64-half-convert.ll  |   93 +-
 .../RISCV/rv64-statepoint-call-lowering.ll    |    4 +-
 llvm/test/CodeGen/RISCV/rvv/clmulh-sdnode.ll  |  168 +-
 .../RISCV/rvv/fixed-vectors-fp-buildvec.ll    |  326 +-
 .../RISCV/rvv/fixed-vectors-masked-scatter.ll | 1166 ++--
 .../RISCV/rvv/fixed-vectors-shuffle-rotate.ll |   80 +-
 .../CodeGen/RISCV/rvv/fpclamptosat_vec.ll     |  175 +-
 .../RISCV/rvv/interrupt-attr-nocall.ll        |    6 +-
 llvm/test/CodeGen/RISCV/rvv/localvar.ll       |   56 +-
 .../RISCV/rvv/nontemporal-vp-scalable.ll      | 1340 ++---
 llvm/test/CodeGen/RISCV/rvv/pr95865.ll        |  106 +-
 llvm/test/CodeGen/RISCV/rvv/shrinkwrap.ll     |   60 +-
 .../RISCV/rvv/vsetvli-insert-crossbb.ll       |   65 +-
 .../RISCV/rvv/vxrm-insert-out-of-loop.ll      |   60 +-
 llvm/test/CodeGen/RISCV/select-cc.ll          |   66 +-
 llvm/test/CodeGen/RISCV/sext-zext-trunc.ll    |   32 +-
 ...-branch-opt-load-atomic-acquire-seq_cst.ll |   84 +-
 .../RISCV/short-forward-branch-opt-load.ll    |   42 +-
 ...h-opt-with-branch-with-immediates_32_eq.ll |   12 +-
 ...h-opt-with-branch-with-immediates_32_ne.ll |   12 +-
 ...-opt-with-branch-with-immediates_32_sge.ll |   12 +-
 ...-opt-with-branch-with-immediates_32_slt.ll |   12 +-
 ...-opt-with-branch-with-immediates_32_uge.ll |   12 +-
 ...-opt-with-branch-with-immediates_32_ult.ll |   12 +-
 ...h-opt-with-branch-with-immediates_48_eq.ll |   12 +-
 ...h-opt-with-branch-with-immediates_48_ne.ll |   12 +-
 ...-opt-with-branch-with-immediates_48_sge.ll |   12 +-
 ...-opt-with-branch-with-immediates_48_slt.ll |   12 +-
 ...-opt-with-branch-with-immediates_48_uge.ll |   12 +-
 ...-opt-with-branch-with-immediates_48_ult.ll |   12 +-
 .../RISCV/shrinkwrap-save-restore-debug.ll    |   28 +-
 llvm/test/CodeGen/RISCV/shrinkwrap.ll         |  150 +-
 llvm/test/CodeGen/RISCV/simplify-condbr.ll    |   10 +-
 llvm/test/CodeGen/RISCV/smulh.ll              |  424 +-
 llvm/test/CodeGen/RISCV/umulh.ll              |  182 +-
 .../test/CodeGen/RISCV/xqcibm-cto-clo-brev.ll |  116 +-
 .../RISCV/xqcilo-xqcilia-frame-index.ll       |   96 +-
 .../CodeGen/RISCV/zicond-fp-select-zfinx.ll   |   12 +-
 .../RISCV/dw_op_entry_value_32bit.ll          |    2 +-
 .../RISCV/dw_op_entry_value_64bit.ll          |    2 +-
 85 files changed, 10763 insertions(+), 10658 deletions(-)

diff --git a/llvm/lib/Target/RISCV/RISCVRegisterInfo.h b/llvm/lib/Target/RISCV/RISCVRegisterInfo.h
index e022a2ec8328fc..56cc3871399011 100644
--- a/llvm/lib/Target/RISCV/RISCVRegisterInfo.h
+++ b/llvm/lib/Target/RISCV/RISCVRegisterInfo.h
@@ -73,9 +73,8 @@ struct RISCVRegisterInfo : public RISCVGenRegisterInfo {
 
   unsigned getCSRCost() const override {
     // The cost will be compared against BlockFrequency where entry has the
-    // value of 1 << 14. A value of 5 will choose to spill or split cold
-    // path instead of using a callee-saved register.
-    return 5;
+    // value of 1 << 14. 32768 mean (1 << 14) * 2 instruction.
+    return 32768;
   }
 
   bool enableCSRFirstTimeMinCost(const MachineFunction &MF) const override;
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/double-convert.ll b/llvm/test/CodeGen/RISCV/GlobalISel/double-convert.ll
index f781a97f8dab54..96fda144dabfa2 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/double-convert.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/double-convert.ll
@@ -1159,71 +1159,62 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ;
 ; RV32I-LABEL: fcvt_l_d_sat:
 ; RV32I:       # %bb.0: # %start
-; RV32I-NEXT:    addi sp, sp, -48
-; RV32I-NEXT:    sw ra, 44(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 40(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 36(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 32(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s6, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s7, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    addi sp, sp, -32
+; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s1, a0
 ; RV32I-NEXT:    mv s2, a1
 ; RV32I-NEXT:    call __fixdfdi
-; RV32I-NEXT:    mv s3, a0
-; RV32I-NEXT:    mv s4, a1
+; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    li s0, 0
 ; RV32I-NEXT:    lui a3, 802304
 ; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    mv a1, s2
 ; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    call __gedf2
-; RV32I-NEXT:    lui s6, 524288
-; RV32I-NEXT:    li s7, 0
-; RV32I-NEXT:    lui s5, 524288
+; RV32I-NEXT:    li s4, 0
+; RV32I-NEXT:    lui s3, 524288
 ; RV32I-NEXT:    bltz a0, .LBB27_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s7, s3
-; RV32I-NEXT:    mv s5, s4
+; RV32I-NEXT:    lw s4, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s3, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB27_2: # %start
 ; RV32I-NEXT:    lui a3, 278016
-; RV32I-NEXT:    li s3, -1
 ; RV32I-NEXT:    addi a3, a3, -1
 ; RV32I-NEXT:    li a2, -1
 ; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    mv a1, s2
 ; RV32I-NEXT:    call __gtdf2
-; RV32I-NEXT:    blt s0, a0, .LBB27_4
-; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    mv s3, s7
-; RV32I-NEXT:    j .LBB27_5
-; RV32I-NEXT:  .LBB27_4:
-; RV32I-NEXT:    addi s5, s6, -1
-; RV32I-NEXT:  .LBB27_5: # %start
+; RV32I-NEXT:    bge s0, a0, .LBB27_4
+; RV32I-NEXT:  # %bb.3:
+; RV32I-NEXT:    lui s3, 524288
+; RV32I-NEXT:    addi s3, s3, -1
+; RV32I-NEXT:    li s4, -1
+; RV32I-NEXT:  .LBB27_4: # %start
 ; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    mv a1, s2
 ; RV32I-NEXT:    mv a2, s1
 ; RV32I-NEXT:    mv a3, s2
 ; RV32I-NEXT:    call __unorddf2
 ; RV32I-NEXT:    mv a1, s0
-; RV32I-NEXT:    bne a0, s0, .LBB27_7
-; RV32I-NEXT:  # %bb.6: # %start
-; RV32I-NEXT:    mv s0, s3
-; RV32I-NEXT:    mv a1, s5
-; RV32I-NEXT:  .LBB27_7: # %start
+; RV32I-NEXT:    bne a0, s0, .LBB27_6
+; RV32I-NEXT:  # %bb.5: # %start
+; RV32I-NEXT:    mv s0, s4
+; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:  .LBB27_6: # %start
 ; RV32I-NEXT:    mv a0, s0
-; RV32I-NEXT:    lw ra, 44(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 40(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 36(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 32(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s6, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s7, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 48
+; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_l_d_sat:
@@ -1232,7 +1223,6 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s1, a0
 ; RV64I-NEXT:    call __fixdfdi
 ; RV64I-NEXT:    mv s0, a0
@@ -1241,10 +1231,10 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    mv a0, s1
 ; RV64I-NEXT:    call __gedf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s2, -1
 ; RV64I-NEXT:    bgez a0, .LBB27_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    slli s0, s2, 63
+; RV64I-NEXT:    li s0, -1
+; RV64I-NEXT:    slli s0, s0, 63
 ; RV64I-NEXT:  .LBB27_2: # %start
 ; RV64I-NEXT:    li a0, 543
 ; RV64I-NEXT:    slli a0, a0, 53
@@ -1254,7 +1244,8 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    sext.w a0, a0
 ; RV64I-NEXT:    blez a0, .LBB27_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    srli s0, s2, 1
+; RV64I-NEXT:    li s0, -1
+; RV64I-NEXT:    srli s0, s0, 1
 ; RV64I-NEXT:  .LBB27_4: # %start
 ; RV64I-NEXT:    mv a0, s1
 ; RV64I-NEXT:    mv a1, s1
@@ -1268,7 +1259,6 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -1335,26 +1325,25 @@ define i64 @fcvt_lu_d_sat(double %a) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    mv s1, a1
 ; RV32I-NEXT:    call __fixunsdfdi
 ; RV32I-NEXT:    mv s2, a0
-; RV32I-NEXT:    mv s3, a1
+; RV32I-NEXT:    sw a1, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s1
 ; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __gedf2
-; RV32I-NEXT:    li s4, 0
-; RV32I-NEXT:    li s5, 0
+; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    bltz a0, .LBB28_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s4, s2
-; RV32I-NEXT:    mv s5, s3
+; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    lw a2, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB28_2: # %start
+; RV32I-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a3, 278272
 ; RV32I-NEXT:    li s2, -1
 ; RV32I-NEXT:    addi a3, a3, -1
@@ -1365,17 +1354,14 @@ define i64 @fcvt_lu_d_sat(double %a) nounwind {
 ; RV32I-NEXT:    li a1, -1
 ; RV32I-NEXT:    bgtz a0, .LBB28_4
 ; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    mv s2, s4
-; RV32I-NEXT:    mv a1, s5
+; RV32I-NEXT:    lw s2, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB28_4: # %start
 ; RV32I-NEXT:    mv a0, s2
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -1384,35 +1370,34 @@ define i64 @fcvt_lu_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixunsdfdi
-; RV64I-NEXT:    mv s2, a0
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gedf2
-; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s0, 0
-; RV64I-NEXT:    bltz a0, .LBB28_2
+; RV64I-NEXT:    sext.w a1, a0
+; RV64I-NEXT:    li a0, 0
+; RV64I-NEXT:    bltz a1, .LBB28_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    mv s0, s2
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB28_2: # %start
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    li a0, 1087
 ; RV64I-NEXT:    slli a0, a0, 52
 ; RV64I-NEXT:    addi a1, a0, -1
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    blez a0, .LBB28_4
-; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    li s0, -1
-; RV64I-NEXT:  .LBB28_4: # %start
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    bgtz a0, .LBB28_4
+; RV64I-NEXT:  # %bb.3: # %start
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    j .LBB28_5
+; RV64I-NEXT:  .LBB28_4:
+; RV64I-NEXT:    li a0, -1
+; RV64I-NEXT:  .LBB28_5: # %start
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/fastcc-float.ll b/llvm/test/CodeGen/RISCV/GlobalISel/fastcc-float.ll
index 6c928bba46c551..d368b0a7c27278 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/fastcc-float.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/fastcc-float.ll
@@ -16,37 +16,27 @@ define fastcc float @callee([32 x float] %A) nounwind {
 define float @caller([32 x float] %A) nounwind {
 ; CHECK-LABEL: caller:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    addi sp, sp, -80
-; CHECK-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    flw ft0, 80(sp)
-; CHECK-NEXT:    flw ft1, 84(sp)
-; CHECK-NEXT:    flw ft2, 88(sp)
-; CHECK-NEXT:    flw ft3, 92(sp)
-; CHECK-NEXT:    flw ft4, 96(sp)
-; CHECK-NEXT:    flw ft5, 100(sp)
-; CHECK-NEXT:    flw ft6, 104(sp)
-; CHECK-NEXT:    flw ft7, 108(sp)
-; CHECK-NEXT:    flw ft8, 112(sp)
-; CHECK-NEXT:    flw ft9, 116(sp)
-; CHECK-NEXT:    flw ft10, 120(sp)
-; CHECK-NEXT:    flw ft11, 124(sp)
-; CHECK-NEXT:    lw t0, 128(sp)
-; CHECK-NEXT:    lw t1, 132(sp)
-; CHECK-NEXT:    lw t2, 136(sp)
-; CHECK-NEXT:    lw t3, 140(sp)
-; CHECK-NEXT:    lw t4, 144(sp)
-; CHECK-NEXT:    lw t5, 148(sp)
-; CHECK-NEXT:    lw t6, 152(sp)
-; CHECK-NEXT:    lw s0, 156(sp)
-; CHECK-NEXT:    lw s1, 160(sp)
-; CHECK-NEXT:    lw s2, 164(sp)
-; CHECK-NEXT:    lw s3, 168(sp)
-; CHECK-NEXT:    lw s4, 172(sp)
+; CHECK-NEXT:    addi sp, sp, -64
+; CHECK-NEXT:    sw ra, 60(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    flw ft0, 64(sp)
+; CHECK-NEXT:    flw ft1, 68(sp)
+; CHECK-NEXT:    flw ft2, 72(sp)
+; CHECK-NEXT:    flw ft3, 76(sp)
+; CHECK-NEXT:    flw ft4, 80(sp)
+; CHECK-NEXT:    flw ft5, 84(sp)
+; CHECK-NEXT:    flw ft6, 88(sp)
+; CHECK-NEXT:    flw ft7, 92(sp)
+; CHECK-NEXT:    flw ft8, 96(sp)
+; CHECK-NEXT:    flw ft9, 100(sp)
+; CHECK-NEXT:    flw ft10, 104(sp)
+; CHECK-NEXT:    flw ft11, 108(sp)
+; CHECK-NEXT:    lw t0, 112(sp)
+; CHECK-NEXT:    lw t1, 116(sp)
+; CHECK-NEXT:    lw t2, 120(sp)
+; CHECK-NEXT:    lw t3, 124(sp)
+; CHECK-NEXT:    lw t4, 128(sp)
+; CHECK-NEXT:    lw t5, 132(sp)
+; CHECK-NEXT:    lw t6, 136(sp)
 ; CHECK-NEXT:    sw t0, 0(sp)
 ; CHECK-NEXT:    sw t1, 4(sp)
 ; CHECK-NEXT:    sw t2, 8(sp)
@@ -54,9 +44,12 @@ define float @caller([32 x float] %A) nounwind {
 ; CHECK-NEXT:    sw t4, 16(sp)
 ; CHECK-NEXT:    sw t5, 20(sp)
 ; CHECK-NEXT:    sw t6, 24(sp)
-; CHECK-NEXT:    sw s0, 28(sp)
-; CHECK-NEXT:    sw s1, 32(sp)
-; CHECK-NEXT:    sw s2, 36(sp)
+; CHECK-NEXT:    lw t0, 140(sp)
+; CHECK-NEXT:    sw t0, 28(sp)
+; CHECK-NEXT:    lw t0, 144(sp)
+; CHECK-NEXT:    sw t0, 32(sp)
+; CHECK-NEXT:    lw t0, 148(sp)
+; CHECK-NEXT:    sw t0, 36(sp)
 ; CHECK-NEXT:    fmv.w.x fa0, a0
 ; CHECK-NEXT:    fmv.w.x fa1, a1
 ; CHECK-NEXT:    fmv.w.x fa2, a2
@@ -65,16 +58,13 @@ define float @caller([32 x float] %A) nounwind {
 ; CHECK-NEXT:    fmv.w.x fa5, a5
 ; CHECK-NEXT:    fmv.w.x fa6, a6
 ; CHECK-NEXT:    fmv.w.x fa7, a7
-; CHECK-NEXT:    sw s3, 40(sp)
-; CHECK-NEXT:    sw s4, 44(sp)
+; CHECK-NEXT:    lw a0, 152(sp)
+; CHECK-NEXT:    sw a0, 40(sp)
+; CHECK-NEXT:    lw a0, 156(sp)
+; CHECK-NEXT:    sw a0, 44(sp)
 ; CHECK-NEXT:    call callee
-; CHECK-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 80
+; CHECK-NEXT:    lw ra, 60(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 64
 ; CHECK-NEXT:    ret
   %C = call fastcc float @callee([32 x float] %A)
   ret float %C
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/float-convert.ll b/llvm/test/CodeGen/RISCV/GlobalISel/float-convert.ll
index bd51cc583b5d48..1bcf3a841e7e20 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/float-convert.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/float-convert.ll
@@ -833,20 +833,17 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    call __fixsfsi
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 847872
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gesf2
 ; RV32I-NEXT:    li s1, 0
-; RV32I-NEXT:    lui s4, 524288
-; RV32I-NEXT:    lui s3, 524288
+; RV32I-NEXT:    lui s2, 524288
 ; RV32I-NEXT:    bltz a0, .LBB23_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s3, s2
+; RV32I-NEXT:    lw s2, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB23_2: # %start
 ; RV32I-NEXT:    lui a1, 323584
 ; RV32I-NEXT:    addi a1, a1, -1
@@ -854,45 +851,41 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    bge s1, a0, .LBB23_4
 ; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    addi s3, s4, -1
+; RV32I-NEXT:    lui s2, 524288
+; RV32I-NEXT:    addi s2, s2, -1
 ; RV32I-NEXT:  .LBB23_4: # %start
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    call __unordsf2
 ; RV32I-NEXT:    bne a0, s1, .LBB23_6
 ; RV32I-NEXT:  # %bb.5: # %start
-; RV32I-NEXT:    mv s1, s3
+; RV32I-NEXT:    mv s1, s2
 ; RV32I-NEXT:  .LBB23_6: # %start
 ; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_w_s_sat:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixsfsi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 847872
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gesf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    lui s3, 524288
-; RV64I-NEXT:    lui s2, 524288
+; RV64I-NEXT:    lui s1, 524288
 ; RV64I-NEXT:    bltz a0, .LBB23_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    mv s2, s1
+; RV64I-NEXT:    ld s1, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB23_2: # %start
 ; RV64I-NEXT:    lui a1, 323584
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -901,7 +894,8 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    sext.w a0, a0
 ; RV64I-NEXT:    blez a0, .LBB23_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    addiw s2, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addiw s1, s1, -1
 ; RV64I-NEXT:  .LBB23_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -910,14 +904,12 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    li a0, 0
 ; RV64I-NEXT:    bnez a1, .LBB23_6
 ; RV64I-NEXT:  # %bb.5: # %start
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s1
 ; RV64I-NEXT:  .LBB23_6: # %start
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i32 @llvm.fptosi.sat.i32.f32(float %a)
@@ -941,32 +933,31 @@ define i32 @fcvt_wu_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    call __fixunssfsi
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    li s1, 0
+; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    bltz a0, .LBB24_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s1, s2
+; RV32I-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB24_2: # %start
+; RV32I-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 325632
 ; RV32I-NEXT:    addi a1, a1, -1
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gtsf2
-; RV32I-NEXT:    blez a0, .LBB24_4
-; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    li s1, -1
-; RV32I-NEXT:  .LBB24_4: # %start
-; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    bgtz a0, .LBB24_4
+; RV32I-NEXT:  # %bb.3: # %start
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    j .LBB24_5
+; RV32I-NEXT:  .LBB24_4:
+; RV32I-NEXT:    li a0, -1
+; RV32I-NEXT:  .LBB24_5: # %start
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -975,34 +966,33 @@ define i32 @fcvt_wu_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixunssfsi
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s1, 0
-; RV64I-NEXT:    bltz a0, .LBB24_2
+; RV64I-NEXT:    sext.w a1, a0
+; RV64I-NEXT:    li a0, 0
+; RV64I-NEXT:    bltz a1, .LBB24_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    mv s1, s2
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB24_2: # %start
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 325632
 ; RV64I-NEXT:    addi a1, a1, -1
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    blez a0, .LBB24_4
-; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    li s1, -1
-; RV64I-NEXT:  .LBB24_4: # %start
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    bgtz a0, .LBB24_4
+; RV64I-NEXT:  # %bb.3: # %start
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    j .LBB24_5
+; RV64I-NEXT:  .LBB24_4:
+; RV64I-NEXT:    li a0, -1
+; RV64I-NEXT:  .LBB24_5: # %start
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -1073,24 +1063,20 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s6, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    call __fixsfdi
-; RV32I-NEXT:    mv s2, a0
-; RV32I-NEXT:    mv s3, a1
+; RV32I-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 913408
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gesf2
 ; RV32I-NEXT:    li s1, 0
-; RV32I-NEXT:    lui s6, 524288
-; RV32I-NEXT:    li s4, 0
-; RV32I-NEXT:    lui s5, 524288
+; RV32I-NEXT:    li s2, 0
+; RV32I-NEXT:    lui s3, 524288
 ; RV32I-NEXT:    bltz a0, .LBB25_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s4, s2
-; RV32I-NEXT:    mv s5, s3
+; RV32I-NEXT:    lw s2, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s3, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB25_2: # %start
 ; RV32I-NEXT:    lui a1, 389120
 ; RV32I-NEXT:    addi a1, a1, -1
@@ -1098,8 +1084,9 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    bge s1, a0, .LBB25_4
 ; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    li s4, -1
-; RV32I-NEXT:    addi s5, s6, -1
+; RV32I-NEXT:    li s2, -1
+; RV32I-NEXT:    lui s3, 524288
+; RV32I-NEXT:    addi s3, s3, -1
 ; RV32I-NEXT:  .LBB25_4: # %start
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s0
@@ -1107,8 +1094,8 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    mv a1, s1
 ; RV32I-NEXT:    bne a0, s1, .LBB25_6
 ; RV32I-NEXT:  # %bb.5: # %start
-; RV32I-NEXT:    mv s1, s4
-; RV32I-NEXT:    mv a1, s5
+; RV32I-NEXT:    mv s1, s2
+; RV32I-NEXT:    mv a1, s3
 ; RV32I-NEXT:  .LBB25_6: # %start
 ; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
@@ -1116,9 +1103,6 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s6, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -1128,7 +1112,6 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s1, a0
 ; RV64I-NEXT:    call __fixsfdi
 ; RV64I-NEXT:    mv s0, a0
@@ -1136,10 +1119,10 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    mv a0, s1
 ; RV64I-NEXT:    call __gesf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s2, -1
 ; RV64I-NEXT:    bgez a0, .LBB25_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    slli s0, s2, 63
+; RV64I-NEXT:    li s0, -1
+; RV64I-NEXT:    slli s0, s0, 63
 ; RV64I-NEXT:  .LBB25_2: # %start
 ; RV64I-NEXT:    lui a1, 389120
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -1148,7 +1131,8 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    sext.w a0, a0
 ; RV64I-NEXT:    blez a0, .LBB25_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    srli s0, s2, 1
+; RV64I-NEXT:    li s0, -1
+; RV64I-NEXT:    srli s0, s0, 1
 ; RV64I-NEXT:  .LBB25_4: # %start
 ; RV64I-NEXT:    mv a0, s1
 ; RV64I-NEXT:    mv a1, s1
@@ -1162,7 +1146,6 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -1217,45 +1200,41 @@ define i64 @fcvt_lu_s_sat(float %a) nounwind {
 ;
 ; RV32I-LABEL: fcvt_lu_s_sat:
 ; RV32I:       # %bb.0: # %start
-; RV32I-NEXT:    addi sp, sp, -32
-; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s1, a0
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    call __fixunssfdi
-; RV32I-NEXT:    mv s3, a0
-; RV32I-NEXT:    mv s4, a1
-; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    li s0, 0
-; RV32I-NEXT:    li s2, 0
+; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    bltz a0, .LBB26_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s0, s3
-; RV32I-NEXT:    mv s2, s4
+; RV32I-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB26_2: # %start
+; RV32I-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 391168
 ; RV32I-NEXT:    addi a1, a1, -1
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    call __gtsf2
-; RV32I-NEXT:    blez a0, .LBB26_4
-; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    li s0, -1
-; RV32I-NEXT:    li s2, -1
-; RV32I-NEXT:  .LBB26_4: # %start
 ; RV32I-NEXT:    mv a0, s0
-; RV32I-NEXT:    mv a1, s2
-; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    call __gtsf2
+; RV32I-NEXT:    bgtz a0, .LBB26_4
+; RV32I-NEXT:  # %bb.3: # %start
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    j .LBB26_5
+; RV32I-NEXT:  .LBB26_4:
+; RV32I-NEXT:    li a0, -1
+; RV32I-NEXT:    li a1, -1
+; RV32I-NEXT:  .LBB26_5: # %start
+; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_lu_s_sat:
@@ -1263,34 +1242,33 @@ define i64 @fcvt_lu_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s2, a0
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s0, 0
-; RV64I-NEXT:    bltz a0, .LBB26_2
+; RV64I-NEXT:    sext.w a1, a0
+; RV64I-NEXT:    li a0, 0
+; RV64I-NEXT:    bltz a1, .LBB26_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    mv s0, s2
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB26_2: # %start
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 391168
 ; RV64I-NEXT:    addi a1, a1, -1
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    blez a0, .LBB26_4
-; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    li s0, -1
-; RV64I-NEXT:  .LBB26_4: # %start
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    bgtz a0, .LBB26_4
+; RV64I-NEXT:  # %bb.3: # %start
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    j .LBB26_5
+; RV64I-NEXT:  .LBB26_4:
+; RV64I-NEXT:    li a0, -1
+; RV64I-NEXT:  .LBB26_5: # %start
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -1823,32 +1801,31 @@ define zeroext i32 @fcvt_wu_s_sat_zext(float %a) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    call __fixunssfsi
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    li s1, 0
+; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    bltz a0, .LBB31_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s1, s2
+; RV32I-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB31_2: # %start
+; RV32I-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 325632
 ; RV32I-NEXT:    addi a1, a1, -1
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gtsf2
-; RV32I-NEXT:    blez a0, .LBB31_4
-; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    li s1, -1
-; RV32I-NEXT:  .LBB31_4: # %start
-; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    bgtz a0, .LBB31_4
+; RV32I-NEXT:  # %bb.3: # %start
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    j .LBB31_5
+; RV32I-NEXT:  .LBB31_4:
+; RV32I-NEXT:    li a0, -1
+; RV32I-NEXT:  .LBB31_5: # %start
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -1857,35 +1834,35 @@ define zeroext i32 @fcvt_wu_s_sat_zext(float %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixunssfsi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s2, 0
-; RV64I-NEXT:    bltz a0, .LBB31_2
+; RV64I-NEXT:    sext.w a1, a0
+; RV64I-NEXT:    li a0, 0
+; RV64I-NEXT:    bltz a1, .LBB31_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    mv s2, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB31_2: # %start
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 325632
 ; RV64I-NEXT:    addi a1, a1, -1
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    blez a0, .LBB31_4
-; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    li s2, -1
-; RV64I-NEXT:  .LBB31_4: # %start
-; RV64I-NEXT:    slli a0, s2, 32
+; RV64I-NEXT:    bgtz a0, .LBB31_4
+; RV64I-NEXT:  # %bb.3: # %start
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    j .LBB31_5
+; RV64I-NEXT:  .LBB31_4:
+; RV64I-NEXT:    li a0, -1
+; RV64I-NEXT:  .LBB31_5: # %start
+; RV64I-NEXT:    slli a0, a0, 32
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -1912,20 +1889,17 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    call __fixsfsi
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 847872
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gesf2
 ; RV32I-NEXT:    li s1, 0
-; RV32I-NEXT:    lui s4, 524288
-; RV32I-NEXT:    lui s3, 524288
+; RV32I-NEXT:    lui s2, 524288
 ; RV32I-NEXT:    bltz a0, .LBB32_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv s3, s2
+; RV32I-NEXT:    lw s2, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB32_2: # %start
 ; RV32I-NEXT:    lui a1, 323584
 ; RV32I-NEXT:    addi a1, a1, -1
@@ -1933,45 +1907,41 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    bge s1, a0, .LBB32_4
 ; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    addi s3, s4, -1
+; RV32I-NEXT:    lui s2, 524288
+; RV32I-NEXT:    addi s2, s2, -1
 ; RV32I-NEXT:  .LBB32_4: # %start
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    call __unordsf2
 ; RV32I-NEXT:    bne a0, s1, .LBB32_6
 ; RV32I-NEXT:  # %bb.5: # %start
-; RV32I-NEXT:    mv s1, s3
+; RV32I-NEXT:    mv s1, s2
 ; RV32I-NEXT:  .LBB32_6: # %start
 ; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_w_s_sat_sext:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixsfsi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 847872
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gesf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    lui s3, 524288
-; RV64I-NEXT:    lui s2, 524288
+; RV64I-NEXT:    lui s1, 524288
 ; RV64I-NEXT:    bltz a0, .LBB32_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    mv s2, s1
+; RV64I-NEXT:    ld s1, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB32_2: # %start
 ; RV64I-NEXT:    lui a1, 323584
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -1980,7 +1950,8 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV64I-NEXT:    sext.w a0, a0
 ; RV64I-NEXT:    blez a0, .LBB32_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    addi s2, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addi s1, s1, -1
 ; RV64I-NEXT:  .LBB32_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -1989,15 +1960,13 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV64I-NEXT:    li a0, 0
 ; RV64I-NEXT:    bnez a1, .LBB32_6
 ; RV64I-NEXT:  # %bb.5: # %start
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s1
 ; RV64I-NEXT:  .LBB32_6: # %start
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i32 @llvm.fptosi.sat.i32.f32(float %a)
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll b/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll
index e0471c92edbbc7..00de7eea38be6e 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll
@@ -942,244 +942,276 @@ define i128 @fptosi_i128(fp128 %x) nounwind {
 ; RV32:       # %bb.0: # %fp-to-i-entry
 ; RV32-NEXT:    addi sp, sp, -128
 ; RV32-NEXT:    sw ra, 124(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 120(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 116(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 112(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 108(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s4, 104(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s5, 100(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 96(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 92(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s8, 88(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s9, 84(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s10, 80(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s11, 76(sp) # 4-byte Folded Spill
-; RV32-NEXT:    lui a5, 4
-; RV32-NEXT:    lw s1, 12(a1)
-; RV32-NEXT:    slli a4, s1, 1
-; RV32-NEXT:    srli a4, a4, 17
-; RV32-NEXT:    addi a2, a5, -1
-; RV32-NEXT:    mv s0, a0
-; RV32-NEXT:    bgeu a4, a2, .LBB19_2
+; RV32-NEXT:    lui a4, 4
+; RV32-NEXT:    lw a6, 12(a1)
+; RV32-NEXT:    slli a5, a6, 1
+; RV32-NEXT:    srli a5, a5, 17
+; RV32-NEXT:    addi a2, a4, -1
+; RV32-NEXT:    bgeu a5, a2, .LBB19_2
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    li s3, 0
-; RV32-NEXT:    li s10, 0
-; RV32-NEXT:    li a0, 0
+; RV32-NEXT:    li a2, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    j .LBB19_5
+; RV32-NEXT:    li a3, 0
+; RV32-NEXT:    li a4, 0
+; RV32-NEXT:    j .LBB19_6
 ; RV32-NEXT:  .LBB19_2: # %fp-to-i-if-check.exp.size
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    lw a3, 0(a1)
 ; RV32-NEXT:    lw a2, 4(a1)
 ; RV32-NEXT:    lw a0, 8(a1)
-; RV32-NEXT:    slli a1, s1, 16
-; RV32-NEXT:    srai s1, s1, 31
-; RV32-NEXT:    srli a1, a1, 16
-; RV32-NEXT:    lui a6, 16
-; RV32-NEXT:    addi a5, a5, 110
-; RV32-NEXT:    ori s9, s1, 1
-; RV32-NEXT:    or a1, a1, a6
-; RV32-NEXT:    bltu a5, a4, .LBB19_4
+; RV32-NEXT:    slli a7, a6, 16
+; RV32-NEXT:    srai a1, a6, 31
+; RV32-NEXT:    srli a6, a7, 16
+; RV32-NEXT:    lui a7, 16
+; RV32-NEXT:    addi t0, a4, 110
+; RV32-NEXT:    ori a4, a1, 1
+; RV32-NEXT:    or a6, a6, a7
+; RV32-NEXT:    sw a1, 52(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a4, 48(sp) # 4-byte Folded Spill
+; RV32-NEXT:    bltu t0, a5, .LBB19_4
 ; RV32-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32-NEXT:    lui a5, 4
-; RV32-NEXT:    addi a5, a5, 111
-; RV32-NEXT:    sw zero, 24(sp)
-; RV32-NEXT:    sw zero, 28(sp)
-; RV32-NEXT:    sw zero, 32(sp)
-; RV32-NEXT:    sw zero, 36(sp)
-; RV32-NEXT:    sub a5, a5, a4
-; RV32-NEXT:    srli a4, a5, 3
-; RV32-NEXT:    addi a6, sp, 8
-; RV32-NEXT:    andi a4, a4, 12
-; RV32-NEXT:    sw a3, 8(sp)
-; RV32-NEXT:    sw a2, 12(sp)
-; RV32-NEXT:    sw a0, 16(sp)
-; RV32-NEXT:    sw a1, 20(sp)
-; RV32-NEXT:    add a4, a6, a4
-; RV32-NEXT:    lw a0, 12(a4)
-; RV32-NEXT:    lw a1, 8(a4)
-; RV32-NEXT:    lw a2, 4(a4)
-; RV32-NEXT:    lw a3, 0(a4)
-; RV32-NEXT:    andi a4, a5, 31
-; RV32-NEXT:    xori a4, a4, 31
-; RV32-NEXT:    slli a6, a0, 1
-; RV32-NEXT:    srl a7, a1, a5
-; RV32-NEXT:    sll a6, a6, a4
-; RV32-NEXT:    slli a1, a1, 1
-; RV32-NEXT:    or a6, a7, a6
-; RV32-NEXT:    sw a6, 4(sp) # 4-byte Folded Spill
-; RV32-NEXT:    srl a6, a2, a5
-; RV32-NEXT:    sll a1, a1, a4
-; RV32-NEXT:    or s6, a6, a1
-; RV32-NEXT:    slli a2, a2, 1
-; RV32-NEXT:    srl a1, a3, a5
-; RV32-NEXT:    sll a2, a2, a4
-; RV32-NEXT:    or s7, a1, a2
-; RV32-NEXT:    srl a0, a0, a5
-; RV32-NEXT:    sw a0, 0(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv a0, s7
+; RV32-NEXT:    lui a7, 4
+; RV32-NEXT:    addi a7, a7, 111
+; RV32-NEXT:    sw zero, 72(sp)
+; RV32-NEXT:    sw zero, 76(sp)
+; RV32-NEXT:    sw zero, 80(sp)
+; RV32-NEXT:    sw zero, 84(sp)
+; RV32-NEXT:    sub a5, a7, a5
+; RV32-NEXT:    srli a7, a5, 3
+; RV32-NEXT:    addi t0, sp, 56
+; RV32-NEXT:    andi a7, a7, 12
+; RV32-NEXT:    sw a3, 56(sp)
+; RV32-NEXT:    sw a2, 60(sp)
+; RV32-NEXT:    sw a0, 64(sp)
+; RV32-NEXT:    sw a6, 68(sp)
+; RV32-NEXT:    add a7, t0, a7
+; RV32-NEXT:    lw a2, 12(a7)
+; RV32-NEXT:    lw a0, 8(a7)
+; RV32-NEXT:    lw a3, 4(a7)
+; RV32-NEXT:    lw a6, 0(a7)
+; RV32-NEXT:    andi a7, a5, 31
+; RV32-NEXT:    xori a7, a7, 31
+; RV32-NEXT:    slli t0, a2, 1
+; RV32-NEXT:    srl t1, a0, a5
+; RV32-NEXT:    sll t0, t0, a7
+; RV32-NEXT:    slli t2, a0, 1
+; RV32-NEXT:    or a0, t1, t0
+; RV32-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srl a0, a3, a5
+; RV32-NEXT:    sll t0, t2, a7
+; RV32-NEXT:    or a0, a0, t0
+; RV32-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32-NEXT:    slli a3, a3, 1
+; RV32-NEXT:    srl a0, a6, a5
+; RV32-NEXT:    sll a3, a3, a7
+; RV32-NEXT:    or a0, a0, a3
+; RV32-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srl a0, a2, a5
+; RV32-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s9
+; RV32-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s3, a0
-; RV32-NEXT:    mv s8, a1
-; RV32-NEXT:    mv a0, s6
+; RV32-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 44(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s9
+; RV32-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s8, a0, s8
-; RV32-NEXT:    sltu a0, s8, a0
-; RV32-NEXT:    add s2, a1, a0
-; RV32-NEXT:    mv a0, s7
+; RV32-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 44(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s10, a0, s8
-; RV32-NEXT:    sltu a0, s10, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s11, s2, a0
-; RV32-NEXT:    mv a0, s6
+; RV32-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 44(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s4, a0
-; RV32-NEXT:    mv s5, s9
-; RV32-NEXT:    mv s9, a1
-; RV32-NEXT:    add s8, a0, s11
-; RV32-NEXT:    mv a0, s1
-; RV32-NEXT:    mv a1, s1
-; RV32-NEXT:    mv a2, s7
-; RV32-NEXT:    mv a3, s6
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 52(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 52(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s6, a0
-; RV32-NEXT:    mv s7, a1
-; RV32-NEXT:    mv a0, s5
-; RV32-NEXT:    mv a1, s1
-; RV32-NEXT:    lw a2, 4(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a3, 0(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 48(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 52(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 4(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sltu a2, s8, s4
-; RV32-NEXT:    add a3, s6, a0
-; RV32-NEXT:    sltu a4, s11, s2
-; RV32-NEXT:    add a0, s8, a3
-; RV32-NEXT:    add a4, s9, a4
-; RV32-NEXT:    sltu a3, a3, s6
-; RV32-NEXT:    add a1, s7, a1
-; RV32-NEXT:    add a2, a4, a2
-; RV32-NEXT:    add a1, a1, a3
-; RV32-NEXT:    sltu a3, a0, s8
-; RV32-NEXT:    add a1, a2, a1
-; RV32-NEXT:    add a1, a1, a3
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, a4, a3
+; RV32-NEXT:    lw a3, 40(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a3, a0
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a6, a3, a0
+; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a3, a5
+; RV32-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a6, a0, a6
+; RV32-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a5, a5, a0
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a4, a6, a4
+; RV32-NEXT:    add a5, a0, a5
+; RV32-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a3, a0
+; RV32-NEXT:    add a4, a4, a5
+; RV32-NEXT:    add a4, a4, a0
 ; RV32-NEXT:    j .LBB19_5
 ; RV32-NEXT:  .LBB19_4: # %fp-to-i-if-exp.large
-; RV32-NEXT:    lui a5, 1048572
-; RV32-NEXT:    addi a5, a5, -111
-; RV32-NEXT:    sw zero, 40(sp)
-; RV32-NEXT:    sw zero, 44(sp)
-; RV32-NEXT:    sw zero, 48(sp)
-; RV32-NEXT:    sw zero, 52(sp)
-; RV32-NEXT:    add a4, a4, a5
-; RV32-NEXT:    srli a5, a4, 3
-; RV32-NEXT:    addi a6, sp, 56
-; RV32-NEXT:    andi a5, a5, 12
-; RV32-NEXT:    sw a3, 56(sp)
-; RV32-NEXT:    sw a2, 60(sp)
-; RV32-NEXT:    sw a0, 64(sp)
-; RV32-NEXT:    sw a1, 68(sp)
-; RV32-NEXT:    sub a0, a6, a5
-; RV32-NEXT:    lw a1, 0(a0)
+; RV32-NEXT:    lui a7, 1048572
+; RV32-NEXT:    addi a7, a7, -111
+; RV32-NEXT:    sw zero, 88(sp)
+; RV32-NEXT:    sw zero, 92(sp)
+; RV32-NEXT:    sw zero, 96(sp)
+; RV32-NEXT:    sw zero, 100(sp)
+; RV32-NEXT:    add a5, a5, a7
+; RV32-NEXT:    srli a7, a5, 3
+; RV32-NEXT:    addi t0, sp, 104
+; RV32-NEXT:    andi a7, a7, 12
+; RV32-NEXT:    sw a3, 104(sp)
+; RV32-NEXT:    sw a2, 108(sp)
+; RV32-NEXT:    sw a0, 112(sp)
+; RV32-NEXT:    sw a6, 116(sp)
+; RV32-NEXT:    sub a0, t0, a7
+; RV32-NEXT:    lw a6, 0(a0)
 ; RV32-NEXT:    lw a2, 4(a0)
-; RV32-NEXT:    andi a3, a4, 31
+; RV32-NEXT:    andi a3, a5, 31
 ; RV32-NEXT:    xori a3, a3, 31
-; RV32-NEXT:    lw a5, 8(a0)
-; RV32-NEXT:    lw a0, 12(a0)
-; RV32-NEXT:    srli a6, a1, 1
-; RV32-NEXT:    sll a7, a2, a4
-; RV32-NEXT:    srl a6, a6, a3
+; RV32-NEXT:    lw a7, 8(a0)
+; RV32-NEXT:    lw t0, 12(a0)
+; RV32-NEXT:    srli a0, a6, 1
+; RV32-NEXT:    sll t1, a2, a5
+; RV32-NEXT:    srl a0, a0, a3
 ; RV32-NEXT:    srli a2, a2, 1
-; RV32-NEXT:    or s4, a7, a6
-; RV32-NEXT:    sll a6, a5, a4
+; RV32-NEXT:    or a0, t1, a0
+; RV32-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sll a0, a7, a5
 ; RV32-NEXT:    srl a2, a2, a3
-; RV32-NEXT:    or a2, a6, a2
-; RV32-NEXT:    srli a5, a5, 1
-; RV32-NEXT:    sll a0, a0, a4
-; RV32-NEXT:    srl a3, a5, a3
-; RV32-NEXT:    or a3, a0, a3
-; RV32-NEXT:    sll s5, a1, a4
-; RV32-NEXT:    mv a0, s9
-; RV32-NEXT:    mv a1, s1
+; RV32-NEXT:    or a2, a0, a2
+; RV32-NEXT:    srli a0, a7, 1
+; RV32-NEXT:    sll a7, t0, a5
+; RV32-NEXT:    srl a0, a0, a3
+; RV32-NEXT:    or a3, a7, a0
+; RV32-NEXT:    sll a0, a6, a5
+; RV32-NEXT:    sw a0, 44(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, a4
+; RV32-NEXT:    lw a1, 52(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s3, a0
-; RV32-NEXT:    mv s6, a1
-; RV32-NEXT:    mv a0, s1
-; RV32-NEXT:    mv a1, s1
-; RV32-NEXT:    mv a2, s5
-; RV32-NEXT:    mv a3, s4
+; RV32-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 52(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 52(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s3
-; RV32-NEXT:    add a1, a1, s6
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add s8, a1, a0
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a1, a2
+; RV32-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    add a0, a2, a0
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s9
+; RV32-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s6, a0
-; RV32-NEXT:    mv s7, a1
-; RV32-NEXT:    mv a0, s5
+; RV32-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s9
+; RV32-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s3, a0
-; RV32-NEXT:    add s10, s6, a1
-; RV32-NEXT:    sltu a0, s10, s6
-; RV32-NEXT:    add s7, s7, a0
-; RV32-NEXT:    mv a0, s5
+; RV32-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 48(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 48(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a1, a0
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s10, a0, s10
-; RV32-NEXT:    sltu a0, s10, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s5, s7, a0
-; RV32-NEXT:    sltu s6, s5, s7
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 44(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 48(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 48(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s5, a0, s5
-; RV32-NEXT:    add a1, a1, s6
-; RV32-NEXT:    sltu a0, s5, a0
-; RV32-NEXT:    add a1, a1, a0
-; RV32-NEXT:    add a0, s5, s2
-; RV32-NEXT:    add a1, a1, s8
-; RV32-NEXT:    sltu a2, a0, s5
-; RV32-NEXT:    add a1, a1, a2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 48(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a4, a0, a3
+; RV32-NEXT:    lw a3, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a1, a3
+; RV32-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a4, a0
+; RV32-NEXT:    add a5, a3, a0
+; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a4, a3
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a5, a0
+; RV32-NEXT:    sltu a4, a3, a4
+; RV32-NEXT:    add a4, a0, a4
 ; RV32-NEXT:  .LBB19_5: # %fp-to-i-cleanup
-; RV32-NEXT:    sw s3, 0(s0)
-; RV32-NEXT:    sw s10, 4(s0)
-; RV32-NEXT:    sw a0, 8(s0)
-; RV32-NEXT:    sw a1, 12(s0)
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:  .LBB19_6: # %fp-to-i-cleanup
+; RV32-NEXT:    sw a2, 0(a0)
+; RV32-NEXT:    sw a1, 4(a0)
+; RV32-NEXT:    sw a3, 8(a0)
+; RV32-NEXT:    sw a4, 12(a0)
 ; RV32-NEXT:    lw ra, 124(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 120(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 116(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 112(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 108(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s4, 104(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s5, 100(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 96(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s7, 92(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s8, 88(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s9, 84(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s10, 80(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s11, 76(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    addi sp, sp, 128
 ; RV32-NEXT:    ret
 ;
@@ -1339,22 +1371,19 @@ define i128 @fptoui_i128(fp128 %x) nounwind {
 ; RV32I-NEXT:  # %bb.3: # %fp-to-i-entry
 ; RV32I-NEXT:    mv a6, a4
 ; RV32I-NEXT:  .LBB24_4: # %fp-to-i-entry
-; RV32I-NEXT:    j .LBB24_6
 ; RV32I-NEXT:  # %bb.5: # %fp-to-i-entry
-; RV32I-NEXT:    mv a6, a4
-; RV32I-NEXT:  .LBB24_6: # %fp-to-i-entry
 ; RV32I-NEXT:    addi sp, sp, -16
-; RV32I-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    beqz a6, .LBB24_9
-; RV32I-NEXT:  # %bb.7:
+; RV32I-NEXT:  .LBB24_6:
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    li a2, 0
-; RV32I-NEXT:  .LBB24_8: # %fp-to-i-if-exp.small
+; RV32I-NEXT:  .LBB24_7: # %fp-to-i-if-exp.small
 ; RV32I-NEXT:    li a5, 0
 ; RV32I-NEXT:    li a7, 0
-; RV32I-NEXT:    j .LBB24_67
+; RV32I-NEXT:    j .LBB24_66
+; RV32I-NEXT:  # %bb.8: # %fp-to-i-entry
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    bnez a4, .LBB24_6
 ; RV32I-NEXT:  .LBB24_9: # %fp-to-i-if-check.exp.size
 ; RV32I-NEXT:    addi a6, a3, 111
 ; RV32I-NEXT:    li a3, 1
@@ -1386,8 +1415,9 @@ define i128 @fptoui_i128(fp128 %x) nounwind {
 ; RV32I-NEXT:    sub a6, a6, a5
 ; RV32I-NEXT:    li t0, 32
 ; RV32I-NEXT:    neg t5, a6
-; RV32I-NEXT:    srl t2, a1, a6
+; RV32I-NEXT:    srl a5, a1, a6
 ; RV32I-NEXT:    sll t6, a4, t5
+; RV32I-NEXT:    sw a5, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    bltu a6, t0, .LBB24_20
 ; RV32I-NEXT:  # %bb.17: # %fp-to-i-if-exp.small
 ; RV32I-NEXT:    srl a7, a4, a6
@@ -1398,18 +1428,19 @@ define i128 @fptoui_i128(fp128 %x) nounwind {
 ; RV32I-NEXT:    lui a6, 1048572
 ; RV32I-NEXT:    addi a6, a6, -111
 ; RV32I-NEXT:    add a5, a5, a6
-; RV32I-NEXT:    li a7, 64
+; RV32I-NEXT:    li a6, 64
 ; RV32I-NEXT:    li t2, 32
 ; RV32I-NEXT:    neg t5, a5
 ; RV32I-NEXT:    srl t3, a3, t5
-; RV32I-NEXT:    sll t1, a2, a5
+; RV32I-NEXT:    sll a7, a2, a5
+; RV32I-NEXT:    sw a7, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    bltu a5, t2, .LBB24_24
 ; RV32I-NEXT:  # %bb.19: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    li t0, 0
-; RV32I-NEXT:    sll t6, a3, a5
+; RV32I-NEXT:    sll t4, a3, a5
 ; RV32I-NEXT:    j .LBB24_25
 ; RV32I-NEXT:  .LBB24_20:
-; RV32I-NEXT:    or a7, t2, t6
+; RV32I-NEXT:    or a7, a5, t6
 ; RV32I-NEXT:    mv a5, a1
 ; RV32I-NEXT:    beqz a6, .LBB24_22
 ; RV32I-NEXT:  .LBB24_21: # %fp-to-i-if-exp.small
@@ -1418,169 +1449,176 @@ define i128 @fptoui_i128(fp128 %x) nounwind {
 ; RV32I-NEXT:    bltu a6, t0, .LBB24_29
 ; RV32I-NEXT:  # %bb.23: # %fp-to-i-if-exp.small
 ; RV32I-NEXT:    li a7, 0
-; RV32I-NEXT:    srl t4, a2, a6
-; RV32I-NEXT:    j .LBB24_30
+; RV32I-NEXT:    srl t1, a2, a6
+; RV32I-NEXT:    mv t3, a3
+; RV32I-NEXT:    bnez a6, .LBB24_30
+; RV32I-NEXT:    j .LBB24_31
 ; RV32I-NEXT:  .LBB24_24:
 ; RV32I-NEXT:    sll t0, a3, a5
-; RV32I-NEXT:    or t6, t3, t1
+; RV32I-NEXT:    or t4, t3, a7
 ; RV32I-NEXT:  .LBB24_25: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    sub t4, a7, a5
+; RV32I-NEXT:    sub a7, a6, a5
 ; RV32I-NEXT:    mv a6, a2
 ; RV32I-NEXT:    beqz a5, .LBB24_27
 ; RV32I-NEXT:  # %bb.26: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    mv a6, t6
+; RV32I-NEXT:    mv a6, t4
 ; RV32I-NEXT:  .LBB24_27: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    bltu t4, t2, .LBB24_35
+; RV32I-NEXT:    bltu a7, t2, .LBB24_34
 ; RV32I-NEXT:  # %bb.28: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    srl t6, a2, t4
+; RV32I-NEXT:    srl t1, a2, a7
 ; RV32I-NEXT:    mv t3, a3
-; RV32I-NEXT:    bnez t4, .LBB24_36
-; RV32I-NEXT:    j .LBB24_37
+; RV32I-NEXT:    bnez a7, .LBB24_35
+; RV32I-NEXT:    j .LBB24_36
 ; RV32I-NEXT:  .LBB24_29:
 ; RV32I-NEXT:    srl t1, a3, a6
-; RV32I-NEXT:    sll t3, a2, t5
+; RV32I-NEXT:    sll t2, a2, t5
 ; RV32I-NEXT:    srl a7, a4, a6
-; RV32I-NEXT:    or t4, t1, t3
+; RV32I-NEXT:    or t1, t1, t2
+; RV32I-NEXT:    mv t3, a3
+; RV32I-NEXT:    beqz a6, .LBB24_31
 ; RV32I-NEXT:  .LBB24_30: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    mv t3, t1
+; RV32I-NEXT:  .LBB24_31: # %fp-to-i-if-exp.small
 ; RV32I-NEXT:    li t1, 64
-; RV32I-NEXT:    mv t3, a3
-; RV32I-NEXT:    beqz a6, .LBB24_32
-; RV32I-NEXT:  # %bb.31: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    mv t3, t4
-; RV32I-NEXT:  .LBB24_32: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    sub s0, t1, a6
-; RV32I-NEXT:    bltu a6, t0, .LBB24_40
-; RV32I-NEXT:  # %bb.33: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    li t4, 0
-; RV32I-NEXT:    bgeu s0, t0, .LBB24_41
-; RV32I-NEXT:  .LBB24_34:
-; RV32I-NEXT:    neg s1, s0
-; RV32I-NEXT:    srl s1, a1, s1
+; RV32I-NEXT:    sub t1, t1, a6
+; RV32I-NEXT:    bltu a6, t0, .LBB24_39
+; RV32I-NEXT:  # %bb.32: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    sw zero, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    bgeu t1, t0, .LBB24_40
+; RV32I-NEXT:  .LBB24_33:
+; RV32I-NEXT:    neg t2, t1
+; RV32I-NEXT:    srl t2, a1, t2
 ; RV32I-NEXT:    sll t5, a1, t5
-; RV32I-NEXT:    or s2, s1, t6
-; RV32I-NEXT:    j .LBB24_42
-; RV32I-NEXT:  .LBB24_35:
-; RV32I-NEXT:    neg t6, t4
-; RV32I-NEXT:    sll t6, a2, t6
-; RV32I-NEXT:    or t6, t3, t6
+; RV32I-NEXT:    or t6, t2, t6
+; RV32I-NEXT:    j .LBB24_41
+; RV32I-NEXT:  .LBB24_34:
+; RV32I-NEXT:    neg t1, a7
+; RV32I-NEXT:    sll t1, a2, t1
+; RV32I-NEXT:    or t1, t3, t1
 ; RV32I-NEXT:    mv t3, a3
-; RV32I-NEXT:    beqz t4, .LBB24_37
+; RV32I-NEXT:    beqz a7, .LBB24_36
+; RV32I-NEXT:  .LBB24_35: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    mv t3, t1
 ; RV32I-NEXT:  .LBB24_36: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    mv t3, t6
-; RV32I-NEXT:  .LBB24_37: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    bltu t4, t2, .LBB24_46
-; RV32I-NEXT:  # %bb.38: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    li t4, 0
-; RV32I-NEXT:    bgeu a5, t2, .LBB24_47
-; RV32I-NEXT:  .LBB24_39:
-; RV32I-NEXT:    srl t6, a1, t5
-; RV32I-NEXT:    sll s0, a4, a5
+; RV32I-NEXT:    bltu a7, t2, .LBB24_45
+; RV32I-NEXT:  # %bb.37: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    li t1, 0
+; RV32I-NEXT:    mv t4, t0
+; RV32I-NEXT:    bgeu a5, t2, .LBB24_46
+; RV32I-NEXT:  .LBB24_38:
+; RV32I-NEXT:    srl a7, a1, t5
+; RV32I-NEXT:    sll t0, a4, a5
 ; RV32I-NEXT:    sll t5, a1, a5
-; RV32I-NEXT:    or s1, t6, s0
-; RV32I-NEXT:    j .LBB24_48
-; RV32I-NEXT:  .LBB24_40:
-; RV32I-NEXT:    srl t4, a2, a6
-; RV32I-NEXT:    bltu s0, t0, .LBB24_34
+; RV32I-NEXT:    or t6, a7, t0
+; RV32I-NEXT:    j .LBB24_47
+; RV32I-NEXT:  .LBB24_39:
+; RV32I-NEXT:    srl t2, a2, a6
+; RV32I-NEXT:    sw t2, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    bltu t1, t0, .LBB24_33
+; RV32I-NEXT:  .LBB24_40: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    li t5, 0
+; RV32I-NEXT:    sll t6, a1, t1
 ; RV32I-NEXT:  .LBB24_41: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    addi t4, a6, -64
+; RV32I-NEXT:    mv t2, a4
+; RV32I-NEXT:    beqz t1, .LBB24_43
+; RV32I-NEXT:  # %bb.42: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    mv t2, t6
+; RV32I-NEXT:  .LBB24_43: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    bltu t4, t0, .LBB24_51
+; RV32I-NEXT:  # %bb.44: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    srl t6, a4, t4
+; RV32I-NEXT:    li t1, 64
+; RV32I-NEXT:    bnez t4, .LBB24_52
+; RV32I-NEXT:    j .LBB24_53
+; RV32I-NEXT:  .LBB24_45:
+; RV32I-NEXT:    srl t1, a2, t5
+; RV32I-NEXT:    mv t4, t0
+; RV32I-NEXT:    bltu a5, t2, .LBB24_38
+; RV32I-NEXT:  .LBB24_46: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    li t5, 0
-; RV32I-NEXT:    sll s2, a1, s0
-; RV32I-NEXT:  .LBB24_42: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    addi s1, a6, -64
-; RV32I-NEXT:    mv t6, a4
-; RV32I-NEXT:    beqz s0, .LBB24_44
-; RV32I-NEXT:  # %bb.43: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    mv t6, s2
-; RV32I-NEXT:  .LBB24_44: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    bltu s1, t0, .LBB24_52
-; RV32I-NEXT:  # %bb.45: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    srl t2, a4, s1
-; RV32I-NEXT:    bnez s1, .LBB24_53
-; RV32I-NEXT:    j .LBB24_54
-; RV32I-NEXT:  .LBB24_46:
-; RV32I-NEXT:    srl t4, a2, t5
-; RV32I-NEXT:    bltu a5, t2, .LBB24_39
+; RV32I-NEXT:    sll t6, a1, a5
 ; RV32I-NEXT:  .LBB24_47: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    li t5, 0
-; RV32I-NEXT:    sll s1, a1, a5
-; RV32I-NEXT:  .LBB24_48: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    addi s0, a5, -64
-; RV32I-NEXT:    mv t6, a4
-; RV32I-NEXT:    beqz a5, .LBB24_50
-; RV32I-NEXT:  # %bb.49: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    mv t6, s1
-; RV32I-NEXT:  .LBB24_50: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    bltu s0, t2, .LBB24_56
-; RV32I-NEXT:  # %bb.51: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    addi a7, a5, -64
+; RV32I-NEXT:    mv t0, a4
+; RV32I-NEXT:    beqz a5, .LBB24_49
+; RV32I-NEXT:  # %bb.48: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    mv a4, t6
+; RV32I-NEXT:  .LBB24_49: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    bltu a7, t2, .LBB24_55
+; RV32I-NEXT:  # %bb.50: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    li t2, 0
-; RV32I-NEXT:    sll a3, a3, s0
-; RV32I-NEXT:    bnez s0, .LBB24_57
-; RV32I-NEXT:    j .LBB24_58
-; RV32I-NEXT:  .LBB24_52:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    sll s0, a4, s0
-; RV32I-NEXT:    or t2, t2, s0
-; RV32I-NEXT:    beqz s1, .LBB24_54
+; RV32I-NEXT:    sll a3, a3, a7
+; RV32I-NEXT:    bnez a7, .LBB24_56
+; RV32I-NEXT:    j .LBB24_57
+; RV32I-NEXT:  .LBB24_51:
+; RV32I-NEXT:    neg t1, t4
+; RV32I-NEXT:    sll t6, a4, t1
+; RV32I-NEXT:    lw t1, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    or t6, t1, t6
+; RV32I-NEXT:    li t1, 64
+; RV32I-NEXT:    beqz t4, .LBB24_53
+; RV32I-NEXT:  .LBB24_52: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    mv a1, t6
 ; RV32I-NEXT:  .LBB24_53: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    mv a1, t2
-; RV32I-NEXT:  .LBB24_54: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    bltu s1, t0, .LBB24_60
-; RV32I-NEXT:  # %bb.55: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    bltu t4, t0, .LBB24_59
+; RV32I-NEXT:  # %bb.54: # %fp-to-i-if-exp.small
 ; RV32I-NEXT:    li a4, 0
-; RV32I-NEXT:    bltu a6, t1, .LBB24_61
-; RV32I-NEXT:    j .LBB24_62
-; RV32I-NEXT:  .LBB24_56:
-; RV32I-NEXT:    neg t2, s0
-; RV32I-NEXT:    srl s1, a3, t2
+; RV32I-NEXT:    bltu a6, t1, .LBB24_60
+; RV32I-NEXT:    j .LBB24_61
+; RV32I-NEXT:  .LBB24_55:
+; RV32I-NEXT:    neg t2, a7
+; RV32I-NEXT:    srl t6, a3, t2
 ; RV32I-NEXT:    sll t2, a3, a5
-; RV32I-NEXT:    or a3, s1, t1
-; RV32I-NEXT:    beqz s0, .LBB24_58
-; RV32I-NEXT:  .LBB24_57: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    or a3, t6, a3
+; RV32I-NEXT:    beqz a7, .LBB24_57
+; RV32I-NEXT:  .LBB24_56: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    mv a2, a3
-; RV32I-NEXT:  .LBB24_58: # %fp-to-i-if-exp.large
-; RV32I-NEXT:    bltu a5, a7, .LBB24_64
-; RV32I-NEXT:  # %bb.59: # %fp-to-i-if-exp.large
+; RV32I-NEXT:  .LBB24_57: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    li a3, 64
+; RV32I-NEXT:    bltu a5, a3, .LBB24_63
+; RV32I-NEXT:  # %bb.58: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    li a6, 0
-; RV32I-NEXT:    bnez a5, .LBB24_65
-; RV32I-NEXT:    j .LBB24_66
-; RV32I-NEXT:  .LBB24_60:
+; RV32I-NEXT:    bnez a5, .LBB24_64
+; RV32I-NEXT:    j .LBB24_65
+; RV32I-NEXT:  .LBB24_59:
 ; RV32I-NEXT:    srl a4, a4, a6
-; RV32I-NEXT:    bgeu a6, t1, .LBB24_62
-; RV32I-NEXT:  .LBB24_61:
+; RV32I-NEXT:    bgeu a6, t1, .LBB24_61
+; RV32I-NEXT:  .LBB24_60:
 ; RV32I-NEXT:    or a1, t3, t5
-; RV32I-NEXT:    or a4, t4, t6
-; RV32I-NEXT:  .LBB24_62: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    bnez a6, .LBB24_68
-; RV32I-NEXT:  # %bb.63: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    bgeu a6, t1, .LBB24_8
-; RV32I-NEXT:    j .LBB24_67
-; RV32I-NEXT:  .LBB24_64:
+; RV32I-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    or a4, a4, t2
+; RV32I-NEXT:  .LBB24_61: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    bnez a6, .LBB24_67
+; RV32I-NEXT:  # %bb.62: # %fp-to-i-if-exp.small
+; RV32I-NEXT:    bgeu a6, t1, .LBB24_7
+; RV32I-NEXT:    j .LBB24_66
+; RV32I-NEXT:  .LBB24_63:
 ; RV32I-NEXT:    or t2, t3, t5
-; RV32I-NEXT:    or a2, t4, t6
-; RV32I-NEXT:    mv a3, t0
-; RV32I-NEXT:    beqz a5, .LBB24_66
-; RV32I-NEXT:  .LBB24_65: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    or a2, t1, a4
+; RV32I-NEXT:    mv a3, t4
+; RV32I-NEXT:    beqz a5, .LBB24_65
+; RV32I-NEXT:  .LBB24_64: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    mv a1, t2
-; RV32I-NEXT:    mv a4, a2
-; RV32I-NEXT:  .LBB24_66: # %fp-to-i-if-exp.large
+; RV32I-NEXT:    mv t0, a2
+; RV32I-NEXT:  .LBB24_65: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    mv a2, a6
 ; RV32I-NEXT:    mv a5, a1
-; RV32I-NEXT:    mv a7, a4
-; RV32I-NEXT:  .LBB24_67: # %fp-to-i-cleanup
+; RV32I-NEXT:    mv a7, t0
+; RV32I-NEXT:  .LBB24_66: # %fp-to-i-cleanup
 ; RV32I-NEXT:    sw a3, 0(a0)
 ; RV32I-NEXT:    sw a2, 4(a0)
 ; RV32I-NEXT:    sw a5, 8(a0)
 ; RV32I-NEXT:    sw a7, 12(a0)
-; RV32I-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
-; RV32I-NEXT:  .LBB24_68: # %fp-to-i-if-exp.small
+; RV32I-NEXT:  .LBB24_67: # %fp-to-i-if-exp.small
 ; RV32I-NEXT:    mv a3, a1
 ; RV32I-NEXT:    mv a2, a4
-; RV32I-NEXT:    bgeu a6, t1, .LBB24_8
-; RV32I-NEXT:    j .LBB24_67
+; RV32I-NEXT:    bgeu a6, t1, .LBB24_7
+; RV32I-NEXT:    j .LBB24_66
 ;
 ; RV32D-LABEL: fptoui_i128:
 ; RV32D:       # %bb.0: # %fp-to-i-entry
@@ -2034,11 +2072,8 @@ define fp128 @sitofp_i64(i64 %x) nounwind {
 define fp128 @sitofp_i128(i128 %x) nounwind {
 ; RV32-LABEL: sitofp_i128:
 ; RV32:       # %bb.0: # %itofp-entry
-; RV32-NEXT:    addi sp, sp, -112
-; RV32-NEXT:    sw s0, 108(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 104(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 100(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 96(sp) # 4-byte Folded Spill
+; RV32-NEXT:    addi sp, sp, -128
+; RV32-NEXT:    sw s0, 124(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    lw a4, 0(a1)
 ; RV32-NEXT:    lw a2, 4(a1)
 ; RV32-NEXT:    lw a3, 8(a1)
@@ -2226,10 +2261,13 @@ define fp128 @sitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    sub a6, a6, t0
 ; RV32-NEXT:    bnez t3, .LBB29_25
 ; RV32-NEXT:  # %bb.17: # %itofp-if-then4
+; RV32-NEXT:    sw a6, 16(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    xori t3, a7, 115
 ; RV32-NEXT:    or t3, t3, t1
 ; RV32-NEXT:    or t2, t2, t1
 ; RV32-NEXT:    or t3, t3, t2
+; RV32-NEXT:    sw a5, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a7, 12(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    beqz t3, .LBB29_20
 ; RV32-NEXT:  # %bb.18: # %itofp-if-then4
 ; RV32-NEXT:    xori t3, a7, 114
@@ -2240,72 +2278,72 @@ define fp128 @sitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    li t1, 13
 ; RV32-NEXT:    sub t1, t1, t0
 ; RV32-NEXT:    srli t2, t1, 3
-; RV32-NEXT:    sw zero, 48(sp)
-; RV32-NEXT:    sw zero, 52(sp)
-; RV32-NEXT:    sw zero, 56(sp)
-; RV32-NEXT:    sw zero, 60(sp)
+; RV32-NEXT:    sw zero, 72(sp)
+; RV32-NEXT:    sw zero, 76(sp)
+; RV32-NEXT:    sw zero, 80(sp)
+; RV32-NEXT:    sw zero, 84(sp)
 ; RV32-NEXT:    andi t2, t2, 12
-; RV32-NEXT:    addi t3, sp, 32
-; RV32-NEXT:    sw a4, 32(sp)
-; RV32-NEXT:    sw a2, 36(sp)
-; RV32-NEXT:    sw a3, 40(sp)
-; RV32-NEXT:    sw a1, 44(sp)
+; RV32-NEXT:    addi t3, sp, 56
+; RV32-NEXT:    sw a4, 56(sp)
+; RV32-NEXT:    sw a2, 60(sp)
+; RV32-NEXT:    sw a3, 64(sp)
+; RV32-NEXT:    sw a1, 68(sp)
 ; RV32-NEXT:    add t3, t3, t2
 ; RV32-NEXT:    lw t4, 8(t3)
 ; RV32-NEXT:    lw t2, 12(t3)
 ; RV32-NEXT:    lw t5, 4(t3)
 ; RV32-NEXT:    lw t3, 0(t3)
 ; RV32-NEXT:    andi t6, t1, 31
-; RV32-NEXT:    srl s0, t4, t1
+; RV32-NEXT:    srl a5, t4, t1
 ; RV32-NEXT:    xori t6, t6, 31
-; RV32-NEXT:    slli s1, t2, 1
-; RV32-NEXT:    sll s1, s1, t6
+; RV32-NEXT:    slli s0, t2, 1
+; RV32-NEXT:    sll s0, s0, t6
 ; RV32-NEXT:    addi t0, t0, 115
-; RV32-NEXT:    srli s2, t0, 3
-; RV32-NEXT:    sw zero, 0(sp)
-; RV32-NEXT:    sw zero, 4(sp)
-; RV32-NEXT:    sw zero, 8(sp)
-; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    andi s2, s2, 12
-; RV32-NEXT:    addi s3, sp, 16
-; RV32-NEXT:    sw a4, 16(sp)
-; RV32-NEXT:    sw a2, 20(sp)
-; RV32-NEXT:    sw a3, 24(sp)
-; RV32-NEXT:    sw a1, 28(sp)
-; RV32-NEXT:    sub a1, s3, s2
+; RV32-NEXT:    srli a6, t0, 3
+; RV32-NEXT:    sw zero, 24(sp)
+; RV32-NEXT:    sw zero, 28(sp)
+; RV32-NEXT:    sw zero, 32(sp)
+; RV32-NEXT:    sw zero, 36(sp)
+; RV32-NEXT:    andi a6, a6, 12
+; RV32-NEXT:    addi a7, sp, 40
+; RV32-NEXT:    sw a4, 40(sp)
+; RV32-NEXT:    sw a2, 44(sp)
+; RV32-NEXT:    sw a3, 48(sp)
+; RV32-NEXT:    sw a1, 52(sp)
+; RV32-NEXT:    sub a1, a7, a6
 ; RV32-NEXT:    slli t4, t4, 1
 ; RV32-NEXT:    lw a2, 0(a1)
 ; RV32-NEXT:    lw a3, 4(a1)
 ; RV32-NEXT:    lw a4, 8(a1)
-; RV32-NEXT:    srl s2, t5, t1
-; RV32-NEXT:    sll t4, t4, t6
-; RV32-NEXT:    or s0, s0, s1
-; RV32-NEXT:    or t4, s2, t4
-; RV32-NEXT:    srl t3, t3, t1
+; RV32-NEXT:    srl a6, t5, t1
+; RV32-NEXT:    sll a7, t4, t6
+; RV32-NEXT:    or a5, a5, s0
+; RV32-NEXT:    or a6, a6, a7
+; RV32-NEXT:    srl a7, t3, t1
 ; RV32-NEXT:    lw a1, 12(a1)
 ; RV32-NEXT:    slli t5, t5, 1
 ; RV32-NEXT:    or a2, a4, a2
 ; RV32-NEXT:    sll a4, t5, t6
 ; RV32-NEXT:    or a2, a3, a2
 ; RV32-NEXT:    srli a3, a2, 1
-; RV32-NEXT:    andi t5, t0, 31
-; RV32-NEXT:    xori t5, t5, 31
+; RV32-NEXT:    andi t3, t0, 31
+; RV32-NEXT:    xori t3, t3, 31
 ; RV32-NEXT:    or a1, a2, a1
-; RV32-NEXT:    srl a2, a3, t5
+; RV32-NEXT:    srl a2, a3, t3
 ; RV32-NEXT:    sll a1, a1, t0
-; RV32-NEXT:    or a3, t3, a4
+; RV32-NEXT:    or a3, a7, a4
 ; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    srl t0, t2, t1
+; RV32-NEXT:    srl a7, t2, t1
 ; RV32-NEXT:    snez a4, a1
 ; RV32-NEXT:    or a4, a3, a4
-; RV32-NEXT:    mv a2, t4
-; RV32-NEXT:    mv a3, s0
-; RV32-NEXT:    mv a1, t0
+; RV32-NEXT:    mv a2, a6
+; RV32-NEXT:    mv a3, a5
+; RV32-NEXT:    mv a1, a7
 ; RV32-NEXT:  .LBB29_20: # %itofp-sw-epilog
-; RV32-NEXT:    slli t0, a4, 29
-; RV32-NEXT:    srli t0, t0, 31
-; RV32-NEXT:    or t2, a4, t0
-; RV32-NEXT:    addi t2, t2, 1
+; RV32-NEXT:    slli a5, a4, 29
+; RV32-NEXT:    srli a5, a5, 31
+; RV32-NEXT:    or a4, a4, a5
+; RV32-NEXT:    addi t2, a4, 1
 ; RV32-NEXT:    seqz t0, t2
 ; RV32-NEXT:    add t0, a2, t0
 ; RV32-NEXT:    or a2, t2, t0
@@ -2314,18 +2352,20 @@ define fp128 @sitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    sltu a4, t1, a3
 ; RV32-NEXT:    add a4, a1, a4
 ; RV32-NEXT:    slli a1, a4, 12
+; RV32-NEXT:    lw a5, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    bltz a1, .LBB29_26
 ; RV32-NEXT:  # %bb.21:
 ; RV32-NEXT:    srli a1, t1, 2
 ; RV32-NEXT:    slli a2, a4, 30
 ; RV32-NEXT:    srli a3, t2, 2
-; RV32-NEXT:    slli a7, t0, 30
+; RV32-NEXT:    slli a6, t0, 30
 ; RV32-NEXT:    or a2, a1, a2
-; RV32-NEXT:    or a1, a3, a7
+; RV32-NEXT:    or a1, a3, a6
 ; RV32-NEXT:    slli t1, t1, 30
 ; RV32-NEXT:    srli a3, t0, 2
 ; RV32-NEXT:    or a3, a3, t1
 ; RV32-NEXT:    srai a4, a4, 2
+; RV32-NEXT:    lw a6, 16(sp) # 4-byte Folded Reload
 ; RV32-NEXT:  .LBB29_22: # %itofp-if-end26
 ; RV32-NEXT:    lui a7, 524288
 ; RV32-NEXT:    slli a6, a6, 16
@@ -2341,11 +2381,8 @@ define fp128 @sitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    sw a3, 4(a0)
 ; RV32-NEXT:    sw a2, 8(a0)
 ; RV32-NEXT:    sw a4, 12(a0)
-; RV32-NEXT:    lw s0, 108(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 104(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 100(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 96(sp) # 4-byte Folded Reload
-; RV32-NEXT:    addi sp, sp, 112
+; RV32-NEXT:    lw s0, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    addi sp, sp, 128
 ; RV32-NEXT:    ret
 ; RV32-NEXT:  .LBB29_24:
 ; RV32-NEXT:    li a3, 0
@@ -2355,16 +2392,16 @@ define fp128 @sitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:  .LBB29_25: # %itofp-if-else
 ; RV32-NEXT:    addi t0, t0, -15
 ; RV32-NEXT:    srli a7, t0, 3
-; RV32-NEXT:    sw zero, 64(sp)
-; RV32-NEXT:    sw zero, 68(sp)
-; RV32-NEXT:    sw zero, 72(sp)
-; RV32-NEXT:    sw zero, 76(sp)
+; RV32-NEXT:    sw zero, 88(sp)
+; RV32-NEXT:    sw zero, 92(sp)
+; RV32-NEXT:    sw zero, 96(sp)
+; RV32-NEXT:    sw zero, 100(sp)
 ; RV32-NEXT:    andi a7, a7, 12
-; RV32-NEXT:    addi t1, sp, 80
-; RV32-NEXT:    sw a4, 80(sp)
-; RV32-NEXT:    sw a2, 84(sp)
-; RV32-NEXT:    sw a3, 88(sp)
-; RV32-NEXT:    sw a1, 92(sp)
+; RV32-NEXT:    addi t1, sp, 104
+; RV32-NEXT:    sw a4, 104(sp)
+; RV32-NEXT:    sw a2, 108(sp)
+; RV32-NEXT:    sw a3, 112(sp)
+; RV32-NEXT:    sw a1, 116(sp)
 ; RV32-NEXT:    sub a1, t1, a7
 ; RV32-NEXT:    lw a2, 8(a1)
 ; RV32-NEXT:    lw a3, 12(a1)
@@ -2393,11 +2430,11 @@ define fp128 @sitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    or a1, a1, a2
 ; RV32-NEXT:    slli a2, t1, 29
 ; RV32-NEXT:    srli a6, t1, 3
-; RV32-NEXT:    slli t0, a4, 29
+; RV32-NEXT:    slli a7, a4, 29
 ; RV32-NEXT:    or a3, a3, a2
-; RV32-NEXT:    or a2, a6, t0
+; RV32-NEXT:    or a2, a6, a7
 ; RV32-NEXT:    srai a4, a4, 3
-; RV32-NEXT:    mv a6, a7
+; RV32-NEXT:    lw a6, 12(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    j .LBB29_22
 ; RV32-NEXT:  .LBB29_27: # %itofp-sw-bb
 ; RV32-NEXT:    srli t0, a4, 31
@@ -2572,11 +2609,8 @@ define fp128 @uitofp_i64(i64 %x) nounwind {
 define fp128 @uitofp_i128(i128 %x) nounwind {
 ; RV32-LABEL: uitofp_i128:
 ; RV32:       # %bb.0: # %itofp-entry
-; RV32-NEXT:    addi sp, sp, -112
-; RV32-NEXT:    sw s0, 108(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 104(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 100(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 96(sp) # 4-byte Folded Spill
+; RV32-NEXT:    addi sp, sp, -128
+; RV32-NEXT:    sw s0, 124(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    lw a2, 0(a1)
 ; RV32-NEXT:    lw a4, 4(a1)
 ; RV32-NEXT:    lw a3, 8(a1)
@@ -2745,10 +2779,13 @@ define fp128 @uitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    sub a6, a6, t0
 ; RV32-NEXT:    bnez t3, .LBB34_22
 ; RV32-NEXT:  # %bb.14: # %itofp-if-then4
+; RV32-NEXT:    sw a6, 16(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    xori t3, a7, 115
 ; RV32-NEXT:    or t3, t3, t1
 ; RV32-NEXT:    or t2, t2, t1
 ; RV32-NEXT:    or t3, t3, t2
+; RV32-NEXT:    sw a5, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a7, 12(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    beqz t3, .LBB34_17
 ; RV32-NEXT:  # %bb.15: # %itofp-if-then4
 ; RV32-NEXT:    xori t3, a7, 114
@@ -2759,71 +2796,71 @@ define fp128 @uitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    li t1, 13
 ; RV32-NEXT:    sub t1, t1, t0
 ; RV32-NEXT:    srli t2, t1, 3
-; RV32-NEXT:    sw zero, 48(sp)
-; RV32-NEXT:    sw zero, 52(sp)
-; RV32-NEXT:    sw zero, 56(sp)
-; RV32-NEXT:    sw zero, 60(sp)
+; RV32-NEXT:    sw zero, 72(sp)
+; RV32-NEXT:    sw zero, 76(sp)
+; RV32-NEXT:    sw zero, 80(sp)
+; RV32-NEXT:    sw zero, 84(sp)
 ; RV32-NEXT:    andi t2, t2, 12
-; RV32-NEXT:    addi t3, sp, 32
-; RV32-NEXT:    sw a2, 32(sp)
-; RV32-NEXT:    sw a4, 36(sp)
-; RV32-NEXT:    sw a3, 40(sp)
-; RV32-NEXT:    sw a1, 44(sp)
+; RV32-NEXT:    addi t3, sp, 56
+; RV32-NEXT:    sw a2, 56(sp)
+; RV32-NEXT:    sw a4, 60(sp)
+; RV32-NEXT:    sw a3, 64(sp)
+; RV32-NEXT:    sw a1, 68(sp)
 ; RV32-NEXT:    add t3, t3, t2
 ; RV32-NEXT:    lw t4, 8(t3)
 ; RV32-NEXT:    lw t2, 12(t3)
 ; RV32-NEXT:    lw t5, 4(t3)
 ; RV32-NEXT:    lw t3, 0(t3)
 ; RV32-NEXT:    andi t6, t1, 31
-; RV32-NEXT:    srl s0, t4, t1
+; RV32-NEXT:    srl a5, t4, t1
 ; RV32-NEXT:    xori t6, t6, 31
-; RV32-NEXT:    slli s1, t2, 1
-; RV32-NEXT:    sll s1, s1, t6
+; RV32-NEXT:    slli s0, t2, 1
+; RV32-NEXT:    sll s0, s0, t6
 ; RV32-NEXT:    addi t0, t0, 115
-; RV32-NEXT:    srli s2, t0, 3
-; RV32-NEXT:    sw zero, 0(sp)
-; RV32-NEXT:    sw zero, 4(sp)
-; RV32-NEXT:    sw zero, 8(sp)
-; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    andi s2, s2, 12
-; RV32-NEXT:    addi s3, sp, 16
-; RV32-NEXT:    sw a2, 16(sp)
-; RV32-NEXT:    sw a4, 20(sp)
-; RV32-NEXT:    sw a3, 24(sp)
-; RV32-NEXT:    sw a1, 28(sp)
-; RV32-NEXT:    sub a1, s3, s2
+; RV32-NEXT:    srli a6, t0, 3
+; RV32-NEXT:    sw zero, 24(sp)
+; RV32-NEXT:    sw zero, 28(sp)
+; RV32-NEXT:    sw zero, 32(sp)
+; RV32-NEXT:    sw zero, 36(sp)
+; RV32-NEXT:    andi a6, a6, 12
+; RV32-NEXT:    addi a7, sp, 40
+; RV32-NEXT:    sw a2, 40(sp)
+; RV32-NEXT:    sw a4, 44(sp)
+; RV32-NEXT:    sw a3, 48(sp)
+; RV32-NEXT:    sw a1, 52(sp)
+; RV32-NEXT:    sub a1, a7, a6
 ; RV32-NEXT:    slli t4, t4, 1
 ; RV32-NEXT:    lw a2, 0(a1)
 ; RV32-NEXT:    lw a3, 4(a1)
 ; RV32-NEXT:    lw a4, 8(a1)
-; RV32-NEXT:    srl s2, t5, t1
-; RV32-NEXT:    sll t4, t4, t6
-; RV32-NEXT:    or s0, s0, s1
-; RV32-NEXT:    or t4, s2, t4
-; RV32-NEXT:    srl t3, t3, t1
+; RV32-NEXT:    srl a6, t5, t1
+; RV32-NEXT:    sll a7, t4, t6
+; RV32-NEXT:    or a5, a5, s0
+; RV32-NEXT:    or a6, a6, a7
+; RV32-NEXT:    srl a7, t3, t1
 ; RV32-NEXT:    lw a1, 12(a1)
 ; RV32-NEXT:    slli t5, t5, 1
 ; RV32-NEXT:    or a2, a4, a2
 ; RV32-NEXT:    sll a4, t5, t6
 ; RV32-NEXT:    or a2, a3, a2
 ; RV32-NEXT:    srli a3, a2, 1
-; RV32-NEXT:    andi t5, t0, 31
-; RV32-NEXT:    xori t5, t5, 31
+; RV32-NEXT:    andi t3, t0, 31
+; RV32-NEXT:    xori t3, t3, 31
 ; RV32-NEXT:    or a1, a2, a1
-; RV32-NEXT:    srl a2, a3, t5
+; RV32-NEXT:    srl a2, a3, t3
 ; RV32-NEXT:    sll a1, a1, t0
-; RV32-NEXT:    or a3, t3, a4
+; RV32-NEXT:    or a3, a7, a4
 ; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    srl t0, t2, t1
+; RV32-NEXT:    srl a7, t2, t1
 ; RV32-NEXT:    snez a2, a1
 ; RV32-NEXT:    or a2, a3, a2
-; RV32-NEXT:    mv a4, t4
-; RV32-NEXT:    mv a3, s0
-; RV32-NEXT:    mv a1, t0
+; RV32-NEXT:    mv a4, a6
+; RV32-NEXT:    mv a3, a5
+; RV32-NEXT:    mv a1, a7
 ; RV32-NEXT:  .LBB34_17: # %itofp-sw-epilog
-; RV32-NEXT:    slli t0, a2, 29
-; RV32-NEXT:    srli t0, t0, 31
-; RV32-NEXT:    or a2, a2, t0
+; RV32-NEXT:    slli a5, a2, 29
+; RV32-NEXT:    srli a5, a5, 31
+; RV32-NEXT:    or a2, a2, a5
 ; RV32-NEXT:    addi a2, a2, 1
 ; RV32-NEXT:    seqz t0, a2
 ; RV32-NEXT:    add t0, a4, t0
@@ -2833,18 +2870,20 @@ define fp128 @uitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    sltu a4, t1, a3
 ; RV32-NEXT:    add a4, a1, a4
 ; RV32-NEXT:    slli a1, a4, 12
+; RV32-NEXT:    lw a5, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    bltz a1, .LBB34_23
 ; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    srli a1, t1, 2
 ; RV32-NEXT:    slli a3, a4, 30
-; RV32-NEXT:    srli a7, a2, 2
-; RV32-NEXT:    slli t2, t0, 30
+; RV32-NEXT:    srli a6, a2, 2
+; RV32-NEXT:    slli a7, t0, 30
 ; RV32-NEXT:    or a2, a1, a3
-; RV32-NEXT:    or a1, a7, t2
+; RV32-NEXT:    or a1, a6, a7
 ; RV32-NEXT:    slli t1, t1, 30
 ; RV32-NEXT:    srli a3, t0, 2
 ; RV32-NEXT:    or a3, a3, t1
 ; RV32-NEXT:    srli a4, a4, 2
+; RV32-NEXT:    lw a6, 16(sp) # 4-byte Folded Reload
 ; RV32-NEXT:  .LBB34_19: # %itofp-if-end26
 ; RV32-NEXT:    lui a7, 524288
 ; RV32-NEXT:    slli a6, a6, 16
@@ -2860,11 +2899,8 @@ define fp128 @uitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    sw a3, 4(a0)
 ; RV32-NEXT:    sw a2, 8(a0)
 ; RV32-NEXT:    sw a4, 12(a0)
-; RV32-NEXT:    lw s0, 108(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 104(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 100(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 96(sp) # 4-byte Folded Reload
-; RV32-NEXT:    addi sp, sp, 112
+; RV32-NEXT:    lw s0, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    addi sp, sp, 128
 ; RV32-NEXT:    ret
 ; RV32-NEXT:  .LBB34_21:
 ; RV32-NEXT:    li a1, 0
@@ -2875,16 +2911,16 @@ define fp128 @uitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:  .LBB34_22: # %itofp-if-else
 ; RV32-NEXT:    addi t0, t0, -15
 ; RV32-NEXT:    srli a7, t0, 3
-; RV32-NEXT:    sw zero, 64(sp)
-; RV32-NEXT:    sw zero, 68(sp)
-; RV32-NEXT:    sw zero, 72(sp)
-; RV32-NEXT:    sw zero, 76(sp)
+; RV32-NEXT:    sw zero, 88(sp)
+; RV32-NEXT:    sw zero, 92(sp)
+; RV32-NEXT:    sw zero, 96(sp)
+; RV32-NEXT:    sw zero, 100(sp)
 ; RV32-NEXT:    andi a7, a7, 12
-; RV32-NEXT:    addi t1, sp, 80
-; RV32-NEXT:    sw a2, 80(sp)
-; RV32-NEXT:    sw a4, 84(sp)
-; RV32-NEXT:    sw a3, 88(sp)
-; RV32-NEXT:    sw a1, 92(sp)
+; RV32-NEXT:    addi t1, sp, 104
+; RV32-NEXT:    sw a2, 104(sp)
+; RV32-NEXT:    sw a4, 108(sp)
+; RV32-NEXT:    sw a3, 112(sp)
+; RV32-NEXT:    sw a1, 116(sp)
 ; RV32-NEXT:    sub a1, t1, a7
 ; RV32-NEXT:    lw a2, 8(a1)
 ; RV32-NEXT:    lw a3, 12(a1)
@@ -2913,11 +2949,11 @@ define fp128 @uitofp_i128(i128 %x) nounwind {
 ; RV32-NEXT:    or a1, a2, a1
 ; RV32-NEXT:    slli a2, t1, 29
 ; RV32-NEXT:    srli a6, t1, 3
-; RV32-NEXT:    slli t0, a4, 29
+; RV32-NEXT:    slli a7, a4, 29
 ; RV32-NEXT:    or a3, a3, a2
-; RV32-NEXT:    or a2, a6, t0
+; RV32-NEXT:    or a2, a6, a7
 ; RV32-NEXT:    srli a4, a4, 3
-; RV32-NEXT:    mv a6, a7
+; RV32-NEXT:    lw a6, 12(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    j .LBB34_19
 ; RV32-NEXT:  .LBB34_24: # %itofp-sw-bb
 ; RV32-NEXT:    srli t0, a2, 31
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/rotl-rotr.ll b/llvm/test/CodeGen/RISCV/GlobalISel/rotl-rotr.ll
index b9ba15074e150d..00121d8c7f5ae2 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/rotl-rotr.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/rotl-rotr.ll
@@ -2787,14 +2787,19 @@ define i64 @rotr_64_mask_multiple(i64 %a, i64 %b, i64 %amt) nounwind {
 ; RV32I-NEXT:    bnez t1, .LBB23_10
 ; RV32I-NEXT:    j .LBB23_11
 ; RV32I-NEXT:  .LBB23_9:
-; RV32I-NEXT:    addi sp, sp, -16
-; RV32I-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    srl t6, a0, t5
-; RV32I-NEXT:    sll s0, a1, t2
+; RV32I-NEXT:    mv a6, t5
+; RV32I-NEXT:    mv t5, t4
+; RV32I-NEXT:    mv t4, a5
+; RV32I-NEXT:    mv a5, a7
+; RV32I-NEXT:    sll a7, a1, t2
 ; RV32I-NEXT:    sll t3, a0, t2
-; RV32I-NEXT:    or a0, t6, s0
-; RV32I-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 16
+; RV32I-NEXT:    or a0, t6, a7
+; RV32I-NEXT:    mv a7, a5
+; RV32I-NEXT:    mv a5, t4
+; RV32I-NEXT:    mv t4, t5
+; RV32I-NEXT:    mv t5, a6
+; RV32I-NEXT:    li a6, 32
 ; RV32I-NEXT:    beqz t1, .LBB23_11
 ; RV32I-NEXT:  .LBB23_10:
 ; RV32I-NEXT:    mv a1, a0
@@ -2895,14 +2900,19 @@ define i64 @rotr_64_mask_multiple(i64 %a, i64 %b, i64 %amt) nounwind {
 ; RV32ZBB-NEXT:    bnez t1, .LBB23_10
 ; RV32ZBB-NEXT:    j .LBB23_11
 ; RV32ZBB-NEXT:  .LBB23_9:
-; RV32ZBB-NEXT:    addi sp, sp, -16
-; RV32ZBB-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
 ; RV32ZBB-NEXT:    srl t6, a0, t5
-; RV32ZBB-NEXT:    sll s0, a1, t2
+; RV32ZBB-NEXT:    mv a6, t5
+; RV32ZBB-NEXT:    mv t5, t4
+; RV32ZBB-NEXT:    mv t4, a5
+; RV32ZBB-NEXT:    mv a5, a7
+; RV32ZBB-NEXT:    sll a7, a1, t2
 ; RV32ZBB-NEXT:    sll t3, a0, t2
-; RV32ZBB-NEXT:    or a0, t6, s0
-; RV32ZBB-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZBB-NEXT:    addi sp, sp, 16
+; RV32ZBB-NEXT:    or a0, t6, a7
+; RV32ZBB-NEXT:    mv a7, a5
+; RV32ZBB-NEXT:    mv a5, t4
+; RV32ZBB-NEXT:    mv t4, t5
+; RV32ZBB-NEXT:    mv t5, a6
+; RV32ZBB-NEXT:    li a6, 32
 ; RV32ZBB-NEXT:    beqz t1, .LBB23_11
 ; RV32ZBB-NEXT:  .LBB23_10:
 ; RV32ZBB-NEXT:    mv a1, a0
@@ -2997,14 +3007,19 @@ define i64 @rotr_64_mask_multiple(i64 %a, i64 %b, i64 %amt) nounwind {
 ; RV32XTHEADBB-NEXT:    bnez t1, .LBB23_10
 ; RV32XTHEADBB-NEXT:    j .LBB23_11
 ; RV32XTHEADBB-NEXT:  .LBB23_9:
-; RV32XTHEADBB-NEXT:    addi sp, sp, -16
-; RV32XTHEADBB-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
 ; RV32XTHEADBB-NEXT:    srl t6, a0, t5
-; RV32XTHEADBB-NEXT:    sll s0, a1, t2
+; RV32XTHEADBB-NEXT:    mv a6, t5
+; RV32XTHEADBB-NEXT:    mv t5, t4
+; RV32XTHEADBB-NEXT:    mv t4, a5
+; RV32XTHEADBB-NEXT:    mv a5, a7
+; RV32XTHEADBB-NEXT:    sll a7, a1, t2
 ; RV32XTHEADBB-NEXT:    sll t3, a0, t2
-; RV32XTHEADBB-NEXT:    or a0, t6, s0
-; RV32XTHEADBB-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32XTHEADBB-NEXT:    addi sp, sp, 16
+; RV32XTHEADBB-NEXT:    or a0, t6, a7
+; RV32XTHEADBB-NEXT:    mv a7, a5
+; RV32XTHEADBB-NEXT:    mv a5, t4
+; RV32XTHEADBB-NEXT:    mv t4, t5
+; RV32XTHEADBB-NEXT:    mv t5, a6
+; RV32XTHEADBB-NEXT:    li a6, 32
 ; RV32XTHEADBB-NEXT:    beqz t1, .LBB23_11
 ; RV32XTHEADBB-NEXT:  .LBB23_10:
 ; RV32XTHEADBB-NEXT:    mv a1, a0
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/rv64-double-convert.ll b/llvm/test/CodeGen/RISCV/GlobalISel/rv64-double-convert.ll
index 3effdecbadeab5..8ff41b338c65f8 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/rv64-double-convert.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/rv64-double-convert.ll
@@ -169,48 +169,44 @@ define i128 @fptosi_sat_f64_to_i128(double %a) nounwind {
 define i128 @fptoui_sat_f64_to_i128(double %a) nounwind {
 ; RV64I-LABEL: fptoui_sat_f64_to_i128:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixunsdfti
-; RV64I-NEXT:    mv s3, a0
-; RV64I-NEXT:    mv s4, a1
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gedf2
-; RV64I-NEXT:    li s0, 0
-; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s2, 0
-; RV64I-NEXT:    bltz a0, .LBB5_2
+; RV64I-NEXT:    li a1, 0
+; RV64I-NEXT:    sext.w a2, a0
+; RV64I-NEXT:    li a0, 0
+; RV64I-NEXT:    bltz a2, .LBB5_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    mv s0, s3
-; RV64I-NEXT:    mv s2, s4
+; RV64I-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB5_2:
+; RV64I-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    li a0, 1151
 ; RV64I-NEXT:    slli a0, a0, 52
 ; RV64I-NEXT:    addi a1, a0, -1
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    blez a0, .LBB5_4
+; RV64I-NEXT:    bgtz a0, .LBB5_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    li s0, -1
-; RV64I-NEXT:    li s2, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    j .LBB5_5
 ; RV64I-NEXT:  .LBB5_4:
-; RV64I-NEXT:    mv a0, s0
-; RV64I-NEXT:    mv a1, s2
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    li a0, -1
+; RV64I-NEXT:    li a1, -1
+; RV64I-NEXT:  .LBB5_5:
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
 ; RV64ID-LABEL: fptoui_sat_f64_to_i128:
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/rv64-float-convert.ll b/llvm/test/CodeGen/RISCV/GlobalISel/rv64-float-convert.ll
index 2b733cef4ae35e..e8ffecd7c1d3f9 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/rv64-float-convert.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/rv64-float-convert.ll
@@ -165,47 +165,43 @@ define i128 @fptosi_sat_f32_to_i128(float %a) nounwind {
 define i128 @fptoui_sat_f32_to_i128(float %a) nounwind {
 ; RV64I-LABEL: fptoui_sat_f32_to_i128:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixunssfti
-; RV64I-NEXT:    mv s3, a0
-; RV64I-NEXT:    mv s4, a1
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    li s0, 0
-; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    li s2, 0
-; RV64I-NEXT:    bltz a0, .LBB5_2
+; RV64I-NEXT:    li a1, 0
+; RV64I-NEXT:    sext.w a2, a0
+; RV64I-NEXT:    li a0, 0
+; RV64I-NEXT:    bltz a2, .LBB5_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    mv s0, s3
-; RV64I-NEXT:    mv s2, s4
+; RV64I-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:  .LBB5_2:
+; RV64I-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 522240
 ; RV64I-NEXT:    addi a1, a1, -1
-; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    blez a0, .LBB5_4
+; RV64I-NEXT:    bgtz a0, .LBB5_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    li s0, -1
-; RV64I-NEXT:    li s2, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    j .LBB5_5
 ; RV64I-NEXT:  .LBB5_4:
-; RV64I-NEXT:    mv a0, s0
-; RV64I-NEXT:    mv a1, s2
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    li a0, -1
+; RV64I-NEXT:    li a1, -1
+; RV64I-NEXT:  .LBB5_5:
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
 ; RV64IF-LABEL: fptoui_sat_f32_to_i128:
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/rv64zbb.ll b/llvm/test/CodeGen/RISCV/GlobalISel/rv64zbb.ll
index 852b6369fc06dc..58c62923bee85b 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/rv64zbb.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/rv64zbb.ll
@@ -61,12 +61,10 @@ define signext i32 @ctlz_i32(i32 signext %a) nounwind {
 define signext i32 @log2_i32(i32 signext %a) nounwind {
 ; RV64I-LABEL: log2_i32:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -16
-; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    li s0, 31
 ; RV64I-NEXT:    beqz a0, .LBB1_2
 ; RV64I-NEXT:  # %bb.1: # %cond.false
+; RV64I-NEXT:    addi sp, sp, -16
+; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    srliw a1, a0, 1
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srliw a1, a0, 2
@@ -99,14 +97,14 @@ define signext i32 @log2_i32(i32 signext %a) nounwind {
 ; RV64I-NEXT:    li a1, 32
 ; RV64I-NEXT:    srliw a0, a0, 24
 ; RV64I-NEXT:    sub a0, a1, a0
+; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    j .LBB1_3
 ; RV64I-NEXT:  .LBB1_2:
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:  .LBB1_3: # %cond.end
-; RV64I-NEXT:    subw a0, s0, a0
-; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 16
+; RV64I-NEXT:    li a1, 31
+; RV64I-NEXT:    subw a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV64ZBB-LABEL: log2_i32:
@@ -123,14 +121,12 @@ define signext i32 @log2_i32(i32 signext %a) nounwind {
 define signext i32 @log2_ceil_i32(i32 signext %a) nounwind {
 ; RV64I-LABEL: log2_ceil_i32:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -16
-; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    addiw a1, a0, -1
-; RV64I-NEXT:    li s0, 32
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    beqz a1, .LBB2_2
 ; RV64I-NEXT:  # %bb.1: # %cond.false
+; RV64I-NEXT:    addi sp, sp, -16
+; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    srliw a0, a1, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    srliw a1, a0, 2
@@ -163,11 +159,11 @@ define signext i32 @log2_ceil_i32(i32 signext %a) nounwind {
 ; RV64I-NEXT:    li a1, 32
 ; RV64I-NEXT:    srliw a0, a0, 24
 ; RV64I-NEXT:    sub a0, a1, a0
-; RV64I-NEXT:  .LBB2_2: # %cond.end
-; RV64I-NEXT:    subw a0, s0, a0
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
+; RV64I-NEXT:  .LBB2_2: # %cond.end
+; RV64I-NEXT:    li a1, 32
+; RV64I-NEXT:    subw a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV64ZBB-LABEL: log2_ceil_i32:
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/shifts.ll b/llvm/test/CodeGen/RISCV/GlobalISel/shifts.ll
index 35f386cd3cdc84..c575a73d3c62db 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/shifts.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/shifts.ll
@@ -260,20 +260,20 @@ define i128 @lshr128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    neg s1, s0
 ; RV32I-NEXT:    srl s1, a4, s1
 ; RV32I-NEXT:    sll t5, a4, t5
-; RV32I-NEXT:    or s2, s1, t6
+; RV32I-NEXT:    or t6, s1, t6
 ; RV32I-NEXT:    j .LBB6_14
 ; RV32I-NEXT:  .LBB6_12:
 ; RV32I-NEXT:    srl t4, a1, a2
 ; RV32I-NEXT:    bltu s0, t0, .LBB6_11
 ; RV32I-NEXT:  .LBB6_13:
 ; RV32I-NEXT:    li t5, 0
-; RV32I-NEXT:    sll s2, a4, s0
+; RV32I-NEXT:    sll t6, a4, s0
 ; RV32I-NEXT:  .LBB6_14:
 ; RV32I-NEXT:    addi s1, a2, -64
-; RV32I-NEXT:    mv t6, a7
+; RV32I-NEXT:    mv s2, a7
 ; RV32I-NEXT:    beqz s0, .LBB6_16
 ; RV32I-NEXT:  # %bb.15:
-; RV32I-NEXT:    mv t6, s2
+; RV32I-NEXT:    mv s2, t6
 ; RV32I-NEXT:  .LBB6_16:
 ; RV32I-NEXT:    bltu s1, t0, .LBB6_18
 ; RV32I-NEXT:  # %bb.17:
@@ -281,9 +281,9 @@ define i128 @lshr128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    bnez s1, .LBB6_19
 ; RV32I-NEXT:    j .LBB6_20
 ; RV32I-NEXT:  .LBB6_18:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    sll s0, a7, s0
-; RV32I-NEXT:    or t2, t2, s0
+; RV32I-NEXT:    neg t6, s1
+; RV32I-NEXT:    sll t6, a7, t6
+; RV32I-NEXT:    or t2, t2, t6
 ; RV32I-NEXT:    beqz s1, .LBB6_20
 ; RV32I-NEXT:  .LBB6_19:
 ; RV32I-NEXT:    mv a4, t2
@@ -298,7 +298,7 @@ define i128 @lshr128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    bgeu a2, t1, .LBB6_24
 ; RV32I-NEXT:  .LBB6_23:
 ; RV32I-NEXT:    or a4, t3, t5
-; RV32I-NEXT:    or a7, t4, t6
+; RV32I-NEXT:    or a7, t4, s2
 ; RV32I-NEXT:  .LBB6_24:
 ; RV32I-NEXT:    bnez a2, .LBB6_28
 ; RV32I-NEXT:  # %bb.25:
@@ -406,20 +406,20 @@ define i128 @ashr128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    neg s1, s0
 ; RV32I-NEXT:    srl s1, a5, s1
 ; RV32I-NEXT:    sll t5, a5, t5
-; RV32I-NEXT:    or s2, s1, t6
+; RV32I-NEXT:    or t6, s1, t6
 ; RV32I-NEXT:    j .LBB7_14
 ; RV32I-NEXT:  .LBB7_12:
 ; RV32I-NEXT:    srl t4, a1, a2
 ; RV32I-NEXT:    bltu s0, t0, .LBB7_11
 ; RV32I-NEXT:  .LBB7_13:
 ; RV32I-NEXT:    li t5, 0
-; RV32I-NEXT:    sll s2, a5, s0
+; RV32I-NEXT:    sll t6, a5, s0
 ; RV32I-NEXT:  .LBB7_14:
 ; RV32I-NEXT:    addi s1, a2, -64
-; RV32I-NEXT:    mv t6, a3
+; RV32I-NEXT:    mv s2, a3
 ; RV32I-NEXT:    beqz s0, .LBB7_16
 ; RV32I-NEXT:  # %bb.15:
-; RV32I-NEXT:    mv t6, s2
+; RV32I-NEXT:    mv s2, t6
 ; RV32I-NEXT:  .LBB7_16:
 ; RV32I-NEXT:    bltu s1, t0, .LBB7_18
 ; RV32I-NEXT:  # %bb.17:
@@ -427,9 +427,9 @@ define i128 @ashr128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    bnez s1, .LBB7_19
 ; RV32I-NEXT:    j .LBB7_20
 ; RV32I-NEXT:  .LBB7_18:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    sll s0, a3, s0
-; RV32I-NEXT:    or t2, t2, s0
+; RV32I-NEXT:    neg t6, s1
+; RV32I-NEXT:    sll t6, a3, t6
+; RV32I-NEXT:    or t2, t2, t6
 ; RV32I-NEXT:    beqz s1, .LBB7_20
 ; RV32I-NEXT:  .LBB7_19:
 ; RV32I-NEXT:    mv a5, t2
@@ -444,7 +444,7 @@ define i128 @ashr128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    bgeu a2, t1, .LBB7_24
 ; RV32I-NEXT:  .LBB7_23:
 ; RV32I-NEXT:    or a5, t3, t5
-; RV32I-NEXT:    or t0, t4, t6
+; RV32I-NEXT:    or t0, t4, s2
 ; RV32I-NEXT:  .LBB7_24:
 ; RV32I-NEXT:    bnez a2, .LBB7_28
 ; RV32I-NEXT:  # %bb.25:
@@ -553,19 +553,19 @@ define i128 @shl128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    bltu a2, t2, .LBB8_14
 ; RV32I-NEXT:  # %bb.13:
 ; RV32I-NEXT:    li t5, 0
-; RV32I-NEXT:    sll s1, t1, a2
+; RV32I-NEXT:    sll t6, t1, a2
 ; RV32I-NEXT:    j .LBB8_15
 ; RV32I-NEXT:  .LBB8_14:
 ; RV32I-NEXT:    srl t6, t1, t5
 ; RV32I-NEXT:    sll s0, a1, a2
 ; RV32I-NEXT:    sll t5, t1, a2
-; RV32I-NEXT:    or s1, t6, s0
+; RV32I-NEXT:    or t6, t6, s0
 ; RV32I-NEXT:  .LBB8_15:
 ; RV32I-NEXT:    addi s0, a2, -64
-; RV32I-NEXT:    mv t6, a1
+; RV32I-NEXT:    mv s1, a1
 ; RV32I-NEXT:    beqz a2, .LBB8_17
 ; RV32I-NEXT:  # %bb.16:
-; RV32I-NEXT:    mv t6, s1
+; RV32I-NEXT:    mv s1, t6
 ; RV32I-NEXT:  .LBB8_17:
 ; RV32I-NEXT:    bltu s0, t2, .LBB8_19
 ; RV32I-NEXT:  # %bb.18:
@@ -575,9 +575,9 @@ define i128 @shl128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    j .LBB8_21
 ; RV32I-NEXT:  .LBB8_19:
 ; RV32I-NEXT:    neg t2, s0
-; RV32I-NEXT:    srl s1, a7, t2
+; RV32I-NEXT:    srl t6, a7, t2
 ; RV32I-NEXT:    sll t2, a7, a2
-; RV32I-NEXT:    or a7, s1, t0
+; RV32I-NEXT:    or a7, t6, t0
 ; RV32I-NEXT:    beqz s0, .LBB8_21
 ; RV32I-NEXT:  .LBB8_20:
 ; RV32I-NEXT:    mv a3, a7
@@ -590,7 +590,7 @@ define i128 @shl128(i128 %a, i128 %b) nounwind {
 ; RV32I-NEXT:    j .LBB8_25
 ; RV32I-NEXT:  .LBB8_23:
 ; RV32I-NEXT:    or t2, t3, t5
-; RV32I-NEXT:    or a3, t4, t6
+; RV32I-NEXT:    or a3, t4, s1
 ; RV32I-NEXT:    beqz a2, .LBB8_25
 ; RV32I-NEXT:  .LBB8_24:
 ; RV32I-NEXT:    mv t1, t2
@@ -714,99 +714,98 @@ define i128 @fshr128_minsize(i128 %a, i128 %b) minsize nounwind {
 ; RV32I-NEXT:  # %bb.4:
 ; RV32I-NEXT:    mv a4, a5
 ; RV32I-NEXT:  .LBB10_5:
+; RV32I-NEXT:    addi sp, sp, -32
+; RV32I-NEXT:    sw s0, 28(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s3, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s4, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lw a7, 0(a1)
 ; RV32I-NEXT:    lw a5, 4(a1)
 ; RV32I-NEXT:    bltu t4, a6, .LBB10_7
 ; RV32I-NEXT:  # %bb.6:
-; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    srl t2, a5, t4
+; RV32I-NEXT:    li s1, 0
+; RV32I-NEXT:    srl a1, a5, t4
 ; RV32I-NEXT:    j .LBB10_8
 ; RV32I-NEXT:  .LBB10_7:
-; RV32I-NEXT:    srl t0, a7, t3
-; RV32I-NEXT:    sll t1, a5, t6
-; RV32I-NEXT:    srl a1, a2, t3
-; RV32I-NEXT:    or t2, t0, t1
+; RV32I-NEXT:    srl a1, a7, t3
+; RV32I-NEXT:    sll t0, a5, t6
+; RV32I-NEXT:    srl s1, a2, t3
+; RV32I-NEXT:    or a1, a1, t0
 ; RV32I-NEXT:  .LBB10_8:
 ; RV32I-NEXT:    li t0, 64
 ; RV32I-NEXT:    mv t1, a7
 ; RV32I-NEXT:    beqz t4, .LBB10_10
 ; RV32I-NEXT:  # %bb.9:
-; RV32I-NEXT:    mv t1, t2
+; RV32I-NEXT:    mv t1, a1
 ; RV32I-NEXT:  .LBB10_10:
-; RV32I-NEXT:    addi sp, sp, -32
-; RV32I-NEXT:    sw s0, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sub s0, t0, t4
 ; RV32I-NEXT:    bltu t4, a6, .LBB10_13
 ; RV32I-NEXT:  # %bb.11:
 ; RV32I-NEXT:    li t2, 0
 ; RV32I-NEXT:    bgeu s0, a6, .LBB10_14
 ; RV32I-NEXT:  .LBB10_12:
-; RV32I-NEXT:    neg s1, s0
-; RV32I-NEXT:    srl s1, a3, s1
-; RV32I-NEXT:    sll t6, a3, t6
-; RV32I-NEXT:    or s2, s1, t5
+; RV32I-NEXT:    neg a1, s0
+; RV32I-NEXT:    srl s2, a3, a1
+; RV32I-NEXT:    sll a1, a3, t6
+; RV32I-NEXT:    or t6, s2, t5
 ; RV32I-NEXT:    j .LBB10_15
 ; RV32I-NEXT:  .LBB10_13:
 ; RV32I-NEXT:    srl t2, a5, t3
 ; RV32I-NEXT:    bltu s0, a6, .LBB10_12
 ; RV32I-NEXT:  .LBB10_14:
-; RV32I-NEXT:    li t6, 0
-; RV32I-NEXT:    sll s2, a3, s0
+; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    sll t6, a3, s0
 ; RV32I-NEXT:  .LBB10_15:
-; RV32I-NEXT:    addi s1, t4, -64
+; RV32I-NEXT:    addi s2, t4, -64
 ; RV32I-NEXT:    mv t5, a2
 ; RV32I-NEXT:    beqz s0, .LBB10_17
 ; RV32I-NEXT:  # %bb.16:
-; RV32I-NEXT:    mv t5, s2
+; RV32I-NEXT:    mv t5, t6
 ; RV32I-NEXT:  .LBB10_17:
-; RV32I-NEXT:    bltu s1, a6, .LBB10_19
+; RV32I-NEXT:    bltu s2, a6, .LBB10_19
 ; RV32I-NEXT:  # %bb.18:
-; RV32I-NEXT:    srl s2, a2, s1
+; RV32I-NEXT:    srl s0, a2, s2
 ; RV32I-NEXT:    j .LBB10_20
 ; RV32I-NEXT:  .LBB10_19:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    srl s2, a3, t4
-; RV32I-NEXT:    sll s0, a2, s0
-; RV32I-NEXT:    or s2, s2, s0
+; RV32I-NEXT:    neg t6, s2
+; RV32I-NEXT:    srl s0, a3, t4
+; RV32I-NEXT:    sll t6, a2, t6
+; RV32I-NEXT:    or s0, s0, t6
 ; RV32I-NEXT:  .LBB10_20:
-; RV32I-NEXT:    mv s0, a3
-; RV32I-NEXT:    beqz s1, .LBB10_22
+; RV32I-NEXT:    mv t6, a3
+; RV32I-NEXT:    beqz s2, .LBB10_22
 ; RV32I-NEXT:  # %bb.21:
-; RV32I-NEXT:    mv s0, s2
+; RV32I-NEXT:    mv t6, s0
 ; RV32I-NEXT:  .LBB10_22:
-; RV32I-NEXT:    bltu s1, a6, .LBB10_24
+; RV32I-NEXT:    bltu s2, a6, .LBB10_24
 ; RV32I-NEXT:  # %bb.23:
-; RV32I-NEXT:    li s1, 0
+; RV32I-NEXT:    li s0, 0
 ; RV32I-NEXT:    bltu t4, t0, .LBB10_25
 ; RV32I-NEXT:    j .LBB10_26
 ; RV32I-NEXT:  .LBB10_24:
-; RV32I-NEXT:    srl s1, a2, t4
+; RV32I-NEXT:    srl s0, a2, t4
 ; RV32I-NEXT:    bgeu t4, t0, .LBB10_26
 ; RV32I-NEXT:  .LBB10_25:
-; RV32I-NEXT:    or s0, t1, t6
-; RV32I-NEXT:    or s1, t2, t5
+; RV32I-NEXT:    or t6, t1, a1
+; RV32I-NEXT:    or s0, t2, t5
 ; RV32I-NEXT:  .LBB10_26:
 ; RV32I-NEXT:    mv t1, a7
 ; RV32I-NEXT:    mv t2, a5
 ; RV32I-NEXT:    beqz t4, .LBB10_28
 ; RV32I-NEXT:  # %bb.27:
-; RV32I-NEXT:    mv t1, s0
-; RV32I-NEXT:    mv t2, s1
+; RV32I-NEXT:    mv t1, t6
+; RV32I-NEXT:    mv t2, s0
 ; RV32I-NEXT:  .LBB10_28:
 ; RV32I-NEXT:    neg t6, t3
 ; RV32I-NEXT:    bltu t4, t0, .LBB10_30
 ; RV32I-NEXT:  # %bb.29:
 ; RV32I-NEXT:    li a4, 0
-; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    li s1, 0
 ; RV32I-NEXT:  .LBB10_30:
 ; RV32I-NEXT:    andi t3, t6, 127
-; RV32I-NEXT:    neg s2, t3
-; RV32I-NEXT:    srl s0, a7, s2
+; RV32I-NEXT:    neg a1, t3
+; RV32I-NEXT:    srl s0, a7, a1
 ; RV32I-NEXT:    bltu t3, a6, .LBB10_32
 ; RV32I-NEXT:  # %bb.31:
 ; RV32I-NEXT:    li t4, 0
@@ -817,48 +816,50 @@ define i128 @fshr128_minsize(i128 %a, i128 %b) minsize nounwind {
 ; RV32I-NEXT:    sll t4, a7, t6
 ; RV32I-NEXT:    or s3, s0, t5
 ; RV32I-NEXT:  .LBB10_33:
-; RV32I-NEXT:    sub s1, t0, t3
+; RV32I-NEXT:    sub s2, t0, t3
 ; RV32I-NEXT:    mv t5, a5
 ; RV32I-NEXT:    beqz t3, .LBB10_35
 ; RV32I-NEXT:  # %bb.34:
 ; RV32I-NEXT:    mv t5, s3
 ; RV32I-NEXT:  .LBB10_35:
-; RV32I-NEXT:    bltu s1, a6, .LBB10_37
+; RV32I-NEXT:    bltu s2, a6, .LBB10_37
 ; RV32I-NEXT:  # %bb.36:
-; RV32I-NEXT:    srl s3, a5, s1
+; RV32I-NEXT:    srl s3, a5, s2
 ; RV32I-NEXT:    j .LBB10_38
 ; RV32I-NEXT:  .LBB10_37:
-; RV32I-NEXT:    neg s3, s1
+; RV32I-NEXT:    neg s3, s2
 ; RV32I-NEXT:    sll s3, a5, s3
 ; RV32I-NEXT:    or s3, s0, s3
 ; RV32I-NEXT:  .LBB10_38:
 ; RV32I-NEXT:    mv s0, a7
-; RV32I-NEXT:    beqz s1, .LBB10_40
+; RV32I-NEXT:    beqz s2, .LBB10_40
 ; RV32I-NEXT:  # %bb.39:
 ; RV32I-NEXT:    mv s0, s3
 ; RV32I-NEXT:  .LBB10_40:
-; RV32I-NEXT:    bltu s1, a6, .LBB10_43
+; RV32I-NEXT:    bltu s2, a6, .LBB10_43
 ; RV32I-NEXT:  # %bb.41:
-; RV32I-NEXT:    li s1, 0
+; RV32I-NEXT:    mv s2, a1
+; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    bgeu t3, a6, .LBB10_44
 ; RV32I-NEXT:  .LBB10_42:
 ; RV32I-NEXT:    srl s2, a3, s2
 ; RV32I-NEXT:    sll s3, a2, t6
 ; RV32I-NEXT:    sll t6, a3, t6
-; RV32I-NEXT:    or s4, s2, s3
+; RV32I-NEXT:    or s2, s2, s3
 ; RV32I-NEXT:    j .LBB10_45
 ; RV32I-NEXT:  .LBB10_43:
-; RV32I-NEXT:    srl s1, a5, s2
+; RV32I-NEXT:    mv s2, a1
+; RV32I-NEXT:    srl a1, a5, a1
 ; RV32I-NEXT:    bltu t3, a6, .LBB10_42
 ; RV32I-NEXT:  .LBB10_44:
 ; RV32I-NEXT:    li t6, 0
-; RV32I-NEXT:    sll s4, a3, t3
+; RV32I-NEXT:    sll s2, a3, t3
 ; RV32I-NEXT:  .LBB10_45:
 ; RV32I-NEXT:    addi s3, t3, -64
-; RV32I-NEXT:    mv s2, a2
+; RV32I-NEXT:    mv s4, a2
 ; RV32I-NEXT:    beqz t3, .LBB10_47
 ; RV32I-NEXT:  # %bb.46:
-; RV32I-NEXT:    mv s2, s4
+; RV32I-NEXT:    mv s4, s2
 ; RV32I-NEXT:  .LBB10_47:
 ; RV32I-NEXT:    bltu s3, a6, .LBB10_49
 ; RV32I-NEXT:  # %bb.48:
@@ -868,10 +869,15 @@ define i128 @fshr128_minsize(i128 %a, i128 %b) minsize nounwind {
 ; RV32I-NEXT:    j .LBB10_51
 ; RV32I-NEXT:  .LBB10_49:
 ; RV32I-NEXT:    neg a6, s3
-; RV32I-NEXT:    srl s4, a7, a6
-; RV32I-NEXT:    sll s5, a5, t3
+; RV32I-NEXT:    srl s2, a7, a6
+; RV32I-NEXT:    mv t0, a4
+; RV32I-NEXT:    mv a4, s1
+; RV32I-NEXT:    sll s1, a5, t3
 ; RV32I-NEXT:    sll a6, a7, t3
-; RV32I-NEXT:    or a7, s4, s5
+; RV32I-NEXT:    or a7, s2, s1
+; RV32I-NEXT:    mv s1, a4
+; RV32I-NEXT:    mv a4, t0
+; RV32I-NEXT:    li t0, 64
 ; RV32I-NEXT:    beqz s3, .LBB10_51
 ; RV32I-NEXT:  .LBB10_50:
 ; RV32I-NEXT:    mv a5, a7
@@ -884,26 +890,25 @@ define i128 @fshr128_minsize(i128 %a, i128 %b) minsize nounwind {
 ; RV32I-NEXT:    j .LBB10_55
 ; RV32I-NEXT:  .LBB10_53:
 ; RV32I-NEXT:    or a6, s0, t6
-; RV32I-NEXT:    or a5, s1, s2
+; RV32I-NEXT:    or a5, a1, s4
 ; RV32I-NEXT:    beqz t3, .LBB10_55
 ; RV32I-NEXT:  .LBB10_54:
 ; RV32I-NEXT:    mv a3, a6
 ; RV32I-NEXT:    mv a2, a5
 ; RV32I-NEXT:  .LBB10_55:
-; RV32I-NEXT:    or a5, t1, t4
-; RV32I-NEXT:    or a6, t2, t5
+; RV32I-NEXT:    or a1, t1, t4
+; RV32I-NEXT:    or a5, t2, t5
 ; RV32I-NEXT:    or a3, a4, a3
-; RV32I-NEXT:    or a1, a1, a2
-; RV32I-NEXT:    sw a5, 0(a0)
-; RV32I-NEXT:    sw a6, 4(a0)
+; RV32I-NEXT:    or a2, s1, a2
+; RV32I-NEXT:    sw a1, 0(a0)
+; RV32I-NEXT:    sw a5, 4(a0)
 ; RV32I-NEXT:    sw a3, 8(a0)
-; RV32I-NEXT:    sw a1, 12(a0)
+; RV32I-NEXT:    sw a2, 12(a0)
 ; RV32I-NEXT:    lw s0, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s3, 16(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s4, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/wide-scalar-shift-by-byte-multiple-legalization.ll b/llvm/test/CodeGen/RISCV/GlobalISel/wide-scalar-shift-by-byte-multiple-legalization.ll
index 131d90d85d8101..c02d183ffa9ccd 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/wide-scalar-shift-by-byte-multiple-legalization.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/wide-scalar-shift-by-byte-multiple-legalization.ll
@@ -913,20 +913,20 @@ define void @lshr_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    neg s1, s0
 ; RV32I-NEXT:    srl s1, a3, s1
 ; RV32I-NEXT:    sll t4, a3, t4
-; RV32I-NEXT:    or s2, s1, t5
+; RV32I-NEXT:    or t5, s1, t5
 ; RV32I-NEXT:    j .LBB6_15
 ; RV32I-NEXT:  .LBB6_13:
 ; RV32I-NEXT:    srl t6, a5, a4
 ; RV32I-NEXT:    bltu s0, t0, .LBB6_12
 ; RV32I-NEXT:  .LBB6_14:
 ; RV32I-NEXT:    li t4, 0
-; RV32I-NEXT:    sll s2, a3, s0
+; RV32I-NEXT:    sll t5, a3, s0
 ; RV32I-NEXT:  .LBB6_15:
 ; RV32I-NEXT:    addi s1, a4, -64
-; RV32I-NEXT:    mv t5, a7
+; RV32I-NEXT:    mv s2, a7
 ; RV32I-NEXT:    beqz s0, .LBB6_17
 ; RV32I-NEXT:  # %bb.16:
-; RV32I-NEXT:    mv t5, s2
+; RV32I-NEXT:    mv s2, t5
 ; RV32I-NEXT:  .LBB6_17:
 ; RV32I-NEXT:    bltu s1, t0, .LBB6_19
 ; RV32I-NEXT:  # %bb.18:
@@ -934,9 +934,9 @@ define void @lshr_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    bnez s1, .LBB6_20
 ; RV32I-NEXT:    j .LBB6_21
 ; RV32I-NEXT:  .LBB6_19:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    sll s0, a7, s0
-; RV32I-NEXT:    or t2, t2, s0
+; RV32I-NEXT:    neg t5, s1
+; RV32I-NEXT:    sll t5, a7, t5
+; RV32I-NEXT:    or t2, t2, t5
 ; RV32I-NEXT:    beqz s1, .LBB6_21
 ; RV32I-NEXT:  .LBB6_20:
 ; RV32I-NEXT:    mv a3, t2
@@ -951,7 +951,7 @@ define void @lshr_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    bgeu a4, t1, .LBB6_25
 ; RV32I-NEXT:  .LBB6_24:
 ; RV32I-NEXT:    or a3, t3, t4
-; RV32I-NEXT:    or a7, t6, t5
+; RV32I-NEXT:    or a7, t6, s2
 ; RV32I-NEXT:  .LBB6_25:
 ; RV32I-NEXT:    bnez a4, .LBB6_29
 ; RV32I-NEXT:  # %bb.26:
@@ -1242,20 +1242,20 @@ define void @lshr_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV32I-NEXT:    neg s1, s0
 ; RV32I-NEXT:    srl s1, a3, s1
 ; RV32I-NEXT:    sll t4, a3, t4
-; RV32I-NEXT:    or s2, s1, t5
+; RV32I-NEXT:    or t5, s1, t5
 ; RV32I-NEXT:    j .LBB7_15
 ; RV32I-NEXT:  .LBB7_13:
 ; RV32I-NEXT:    srl t6, a5, a4
 ; RV32I-NEXT:    bltu s0, t0, .LBB7_12
 ; RV32I-NEXT:  .LBB7_14:
 ; RV32I-NEXT:    li t4, 0
-; RV32I-NEXT:    sll s2, a3, s0
+; RV32I-NEXT:    sll t5, a3, s0
 ; RV32I-NEXT:  .LBB7_15:
 ; RV32I-NEXT:    addi s1, a4, -64
-; RV32I-NEXT:    mv t5, a7
+; RV32I-NEXT:    mv s2, a7
 ; RV32I-NEXT:    beqz s0, .LBB7_17
 ; RV32I-NEXT:  # %bb.16:
-; RV32I-NEXT:    mv t5, s2
+; RV32I-NEXT:    mv s2, t5
 ; RV32I-NEXT:  .LBB7_17:
 ; RV32I-NEXT:    bltu s1, t0, .LBB7_19
 ; RV32I-NEXT:  # %bb.18:
@@ -1263,9 +1263,9 @@ define void @lshr_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV32I-NEXT:    bnez s1, .LBB7_20
 ; RV32I-NEXT:    j .LBB7_21
 ; RV32I-NEXT:  .LBB7_19:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    sll s0, a7, s0
-; RV32I-NEXT:    or t2, t2, s0
+; RV32I-NEXT:    neg t5, s1
+; RV32I-NEXT:    sll t5, a7, t5
+; RV32I-NEXT:    or t2, t2, t5
 ; RV32I-NEXT:    beqz s1, .LBB7_21
 ; RV32I-NEXT:  .LBB7_20:
 ; RV32I-NEXT:    mv a3, t2
@@ -1280,7 +1280,7 @@ define void @lshr_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV32I-NEXT:    bgeu a4, t1, .LBB7_25
 ; RV32I-NEXT:  .LBB7_24:
 ; RV32I-NEXT:    or a3, t3, t4
-; RV32I-NEXT:    or a7, t6, t5
+; RV32I-NEXT:    or a7, t6, s2
 ; RV32I-NEXT:  .LBB7_25:
 ; RV32I-NEXT:    bnez a4, .LBB7_29
 ; RV32I-NEXT:  # %bb.26:
@@ -1537,28 +1537,28 @@ define void @shl_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    slli s2, s2, 8
 ; RV32I-NEXT:    bltu t0, t2, .LBB8_7
 ; RV32I-NEXT:  # %bb.6:
-; RV32I-NEXT:    srl s4, a3, t0
+; RV32I-NEXT:    srl t4, a3, t0
 ; RV32I-NEXT:    j .LBB8_8
 ; RV32I-NEXT:  .LBB8_7:
 ; RV32I-NEXT:    neg s4, t0
 ; RV32I-NEXT:    sll s4, a3, s4
-; RV32I-NEXT:    or s4, t4, s4
+; RV32I-NEXT:    or t4, t4, s4
 ; RV32I-NEXT:  .LBB8_8:
-; RV32I-NEXT:    lbu s5, 8(a0)
+; RV32I-NEXT:    lbu s4, 8(a0)
 ; RV32I-NEXT:    lbu a0, 12(a0)
-; RV32I-NEXT:    slli t6, t6, 8
-; RV32I-NEXT:    or s3, s3, t5
+; RV32I-NEXT:    slli s5, t6, 8
+; RV32I-NEXT:    or t6, s3, t5
 ; RV32I-NEXT:    slli t5, s1, 8
-; RV32I-NEXT:    or s1, s2, s0
-; RV32I-NEXT:    mv t4, a7
+; RV32I-NEXT:    or s0, s2, s0
+; RV32I-NEXT:    mv s1, a7
 ; RV32I-NEXT:    beqz t0, .LBB8_10
 ; RV32I-NEXT:  # %bb.9:
-; RV32I-NEXT:    mv t4, s4
+; RV32I-NEXT:    mv s1, t4
 ; RV32I-NEXT:  .LBB8_10:
-; RV32I-NEXT:    or t6, t6, s5
-; RV32I-NEXT:    slli s0, s3, 16
+; RV32I-NEXT:    or t4, s5, s4
+; RV32I-NEXT:    slli t6, t6, 16
 ; RV32I-NEXT:    or a0, t5, a0
-; RV32I-NEXT:    slli s1, s1, 16
+; RV32I-NEXT:    slli s0, s0, 16
 ; RV32I-NEXT:    bltu t0, t2, .LBB8_12
 ; RV32I-NEXT:  # %bb.11:
 ; RV32I-NEXT:    li t5, 0
@@ -1566,24 +1566,24 @@ define void @shl_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:  .LBB8_12:
 ; RV32I-NEXT:    srl t5, a3, t3
 ; RV32I-NEXT:  .LBB8_13:
-; RV32I-NEXT:    or t0, s0, t6
-; RV32I-NEXT:    or a0, s1, a0
+; RV32I-NEXT:    or t0, t6, t4
+; RV32I-NEXT:    or a0, s0, a0
 ; RV32I-NEXT:    bltu a4, t2, .LBB8_15
 ; RV32I-NEXT:  # %bb.14:
 ; RV32I-NEXT:    li t3, 0
-; RV32I-NEXT:    sll s1, t0, a4
+; RV32I-NEXT:    sll t4, t0, a4
 ; RV32I-NEXT:    j .LBB8_16
 ; RV32I-NEXT:  .LBB8_15:
-; RV32I-NEXT:    srl t6, t0, t3
-; RV32I-NEXT:    sll s0, a0, a4
+; RV32I-NEXT:    srl t4, t0, t3
+; RV32I-NEXT:    sll t6, a0, a4
 ; RV32I-NEXT:    sll t3, t0, a4
-; RV32I-NEXT:    or s1, t6, s0
+; RV32I-NEXT:    or t4, t4, t6
 ; RV32I-NEXT:  .LBB8_16:
 ; RV32I-NEXT:    addi s0, a4, -64
 ; RV32I-NEXT:    mv t6, a0
 ; RV32I-NEXT:    beqz a4, .LBB8_18
 ; RV32I-NEXT:  # %bb.17:
-; RV32I-NEXT:    mv t6, s1
+; RV32I-NEXT:    mv t6, t4
 ; RV32I-NEXT:  .LBB8_18:
 ; RV32I-NEXT:    bltu s0, t2, .LBB8_20
 ; RV32I-NEXT:  # %bb.19:
@@ -1593,34 +1593,34 @@ define void @shl_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    j .LBB8_22
 ; RV32I-NEXT:  .LBB8_20:
 ; RV32I-NEXT:    neg t2, s0
-; RV32I-NEXT:    srl s1, a7, t2
+; RV32I-NEXT:    srl t4, a7, t2
 ; RV32I-NEXT:    sll t2, a7, a4
-; RV32I-NEXT:    or a7, s1, t1
+; RV32I-NEXT:    or a7, t4, t1
 ; RV32I-NEXT:    beqz s0, .LBB8_22
 ; RV32I-NEXT:  .LBB8_21:
 ; RV32I-NEXT:    mv a3, a7
 ; RV32I-NEXT:  .LBB8_22:
-; RV32I-NEXT:    lw s0, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    bltu a4, a6, .LBB8_24
 ; RV32I-NEXT:  # %bb.23:
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    li a5, 0
-; RV32I-NEXT:    bnez a4, .LBB8_25
-; RV32I-NEXT:    j .LBB8_26
+; RV32I-NEXT:    j .LBB8_25
 ; RV32I-NEXT:  .LBB8_24:
-; RV32I-NEXT:    or t2, t4, t3
+; RV32I-NEXT:    or t2, s1, t3
 ; RV32I-NEXT:    or a3, t5, t6
-; RV32I-NEXT:    beqz a4, .LBB8_26
 ; RV32I-NEXT:  .LBB8_25:
+; RV32I-NEXT:    lw s0, 28(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 24(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s3, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s4, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s5, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    beqz a4, .LBB8_27
+; RV32I-NEXT:  # %bb.26:
 ; RV32I-NEXT:    mv t0, t2
 ; RV32I-NEXT:    mv a0, a3
-; RV32I-NEXT:  .LBB8_26:
+; RV32I-NEXT:  .LBB8_27:
 ; RV32I-NEXT:    lui a3, 16
 ; RV32I-NEXT:    addi a3, a3, -1
 ; RV32I-NEXT:    srli a4, a1, 16
@@ -1860,28 +1860,28 @@ define void @shl_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) nounw
 ; RV32I-NEXT:    slli s2, s2, 8
 ; RV32I-NEXT:    bltu t0, t2, .LBB9_7
 ; RV32I-NEXT:  # %bb.6:
-; RV32I-NEXT:    srl s4, a3, t0
+; RV32I-NEXT:    srl t4, a3, t0
 ; RV32I-NEXT:    j .LBB9_8
 ; RV32I-NEXT:  .LBB9_7:
 ; RV32I-NEXT:    neg s4, t0
 ; RV32I-NEXT:    sll s4, a3, s4
-; RV32I-NEXT:    or s4, t4, s4
+; RV32I-NEXT:    or t4, t4, s4
 ; RV32I-NEXT:  .LBB9_8:
-; RV32I-NEXT:    lbu s5, 8(a0)
+; RV32I-NEXT:    lbu s4, 8(a0)
 ; RV32I-NEXT:    lbu a0, 12(a0)
-; RV32I-NEXT:    slli t6, t6, 8
-; RV32I-NEXT:    or s3, s3, t5
+; RV32I-NEXT:    slli s5, t6, 8
+; RV32I-NEXT:    or t6, s3, t5
 ; RV32I-NEXT:    slli t5, s1, 8
-; RV32I-NEXT:    or s1, s2, s0
-; RV32I-NEXT:    mv t4, a7
+; RV32I-NEXT:    or s0, s2, s0
+; RV32I-NEXT:    mv s1, a7
 ; RV32I-NEXT:    beqz t0, .LBB9_10
 ; RV32I-NEXT:  # %bb.9:
-; RV32I-NEXT:    mv t4, s4
+; RV32I-NEXT:    mv s1, t4
 ; RV32I-NEXT:  .LBB9_10:
-; RV32I-NEXT:    or t6, t6, s5
-; RV32I-NEXT:    slli s0, s3, 16
+; RV32I-NEXT:    or t4, s5, s4
+; RV32I-NEXT:    slli t6, t6, 16
 ; RV32I-NEXT:    or a0, t5, a0
-; RV32I-NEXT:    slli s1, s1, 16
+; RV32I-NEXT:    slli s0, s0, 16
 ; RV32I-NEXT:    bltu t0, t2, .LBB9_12
 ; RV32I-NEXT:  # %bb.11:
 ; RV32I-NEXT:    li t5, 0
@@ -1889,24 +1889,24 @@ define void @shl_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) nounw
 ; RV32I-NEXT:  .LBB9_12:
 ; RV32I-NEXT:    srl t5, a3, t3
 ; RV32I-NEXT:  .LBB9_13:
-; RV32I-NEXT:    or t0, s0, t6
-; RV32I-NEXT:    or a0, s1, a0
+; RV32I-NEXT:    or t0, t6, t4
+; RV32I-NEXT:    or a0, s0, a0
 ; RV32I-NEXT:    bltu a4, t2, .LBB9_15
 ; RV32I-NEXT:  # %bb.14:
 ; RV32I-NEXT:    li t3, 0
-; RV32I-NEXT:    sll s1, t0, a4
+; RV32I-NEXT:    sll t4, t0, a4
 ; RV32I-NEXT:    j .LBB9_16
 ; RV32I-NEXT:  .LBB9_15:
-; RV32I-NEXT:    srl t6, t0, t3
-; RV32I-NEXT:    sll s0, a0, a4
+; RV32I-NEXT:    srl t4, t0, t3
+; RV32I-NEXT:    sll t6, a0, a4
 ; RV32I-NEXT:    sll t3, t0, a4
-; RV32I-NEXT:    or s1, t6, s0
+; RV32I-NEXT:    or t4, t4, t6
 ; RV32I-NEXT:  .LBB9_16:
 ; RV32I-NEXT:    addi s0, a4, -64
 ; RV32I-NEXT:    mv t6, a0
 ; RV32I-NEXT:    beqz a4, .LBB9_18
 ; RV32I-NEXT:  # %bb.17:
-; RV32I-NEXT:    mv t6, s1
+; RV32I-NEXT:    mv t6, t4
 ; RV32I-NEXT:  .LBB9_18:
 ; RV32I-NEXT:    bltu s0, t2, .LBB9_20
 ; RV32I-NEXT:  # %bb.19:
@@ -1916,34 +1916,34 @@ define void @shl_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) nounw
 ; RV32I-NEXT:    j .LBB9_22
 ; RV32I-NEXT:  .LBB9_20:
 ; RV32I-NEXT:    neg t2, s0
-; RV32I-NEXT:    srl s1, a7, t2
+; RV32I-NEXT:    srl t4, a7, t2
 ; RV32I-NEXT:    sll t2, a7, a4
-; RV32I-NEXT:    or a7, s1, t1
+; RV32I-NEXT:    or a7, t4, t1
 ; RV32I-NEXT:    beqz s0, .LBB9_22
 ; RV32I-NEXT:  .LBB9_21:
 ; RV32I-NEXT:    mv a3, a7
 ; RV32I-NEXT:  .LBB9_22:
-; RV32I-NEXT:    lw s0, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    bltu a4, a6, .LBB9_24
 ; RV32I-NEXT:  # %bb.23:
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    li a5, 0
-; RV32I-NEXT:    bnez a4, .LBB9_25
-; RV32I-NEXT:    j .LBB9_26
+; RV32I-NEXT:    j .LBB9_25
 ; RV32I-NEXT:  .LBB9_24:
-; RV32I-NEXT:    or t2, t4, t3
+; RV32I-NEXT:    or t2, s1, t3
 ; RV32I-NEXT:    or a3, t5, t6
-; RV32I-NEXT:    beqz a4, .LBB9_26
 ; RV32I-NEXT:  .LBB9_25:
+; RV32I-NEXT:    lw s0, 28(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 24(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s3, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s4, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s5, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    beqz a4, .LBB9_27
+; RV32I-NEXT:  # %bb.26:
 ; RV32I-NEXT:    mv t0, t2
 ; RV32I-NEXT:    mv a0, a3
-; RV32I-NEXT:  .LBB9_26:
+; RV32I-NEXT:  .LBB9_27:
 ; RV32I-NEXT:    lui a3, 16
 ; RV32I-NEXT:    addi a3, a3, -1
 ; RV32I-NEXT:    srli a4, a1, 16
@@ -2218,20 +2218,20 @@ define void @ashr_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    neg s1, s0
 ; RV32I-NEXT:    srl s1, a3, s1
 ; RV32I-NEXT:    sll t4, a3, t4
-; RV32I-NEXT:    or s2, s1, t5
+; RV32I-NEXT:    or t5, s1, t5
 ; RV32I-NEXT:    j .LBB10_15
 ; RV32I-NEXT:  .LBB10_13:
 ; RV32I-NEXT:    srl t6, a6, a5
 ; RV32I-NEXT:    bltu s0, t0, .LBB10_12
 ; RV32I-NEXT:  .LBB10_14:
 ; RV32I-NEXT:    li t4, 0
-; RV32I-NEXT:    sll s2, a3, s0
+; RV32I-NEXT:    sll t5, a3, s0
 ; RV32I-NEXT:  .LBB10_15:
 ; RV32I-NEXT:    addi s1, a5, -64
-; RV32I-NEXT:    mv t5, a4
+; RV32I-NEXT:    mv s2, a4
 ; RV32I-NEXT:    beqz s0, .LBB10_17
 ; RV32I-NEXT:  # %bb.16:
-; RV32I-NEXT:    mv t5, s2
+; RV32I-NEXT:    mv s2, t5
 ; RV32I-NEXT:  .LBB10_17:
 ; RV32I-NEXT:    bltu s1, t0, .LBB10_19
 ; RV32I-NEXT:  # %bb.18:
@@ -2239,9 +2239,9 @@ define void @ashr_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    bnez s1, .LBB10_20
 ; RV32I-NEXT:    j .LBB10_21
 ; RV32I-NEXT:  .LBB10_19:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    sll s0, a4, s0
-; RV32I-NEXT:    or t2, t2, s0
+; RV32I-NEXT:    neg t5, s1
+; RV32I-NEXT:    sll t5, a4, t5
+; RV32I-NEXT:    or t2, t2, t5
 ; RV32I-NEXT:    beqz s1, .LBB10_21
 ; RV32I-NEXT:  .LBB10_20:
 ; RV32I-NEXT:    mv a3, t2
@@ -2256,7 +2256,7 @@ define void @ashr_16bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV32I-NEXT:    bgeu a5, t1, .LBB10_25
 ; RV32I-NEXT:  .LBB10_24:
 ; RV32I-NEXT:    or a3, t3, t4
-; RV32I-NEXT:    or t0, t6, t5
+; RV32I-NEXT:    or t0, t6, s2
 ; RV32I-NEXT:  .LBB10_25:
 ; RV32I-NEXT:    bnez a5, .LBB10_29
 ; RV32I-NEXT:  # %bb.26:
@@ -2547,20 +2547,20 @@ define void @ashr_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV32I-NEXT:    neg s1, s0
 ; RV32I-NEXT:    srl s1, a3, s1
 ; RV32I-NEXT:    sll t4, a3, t4
-; RV32I-NEXT:    or s2, s1, t5
+; RV32I-NEXT:    or t5, s1, t5
 ; RV32I-NEXT:    j .LBB11_15
 ; RV32I-NEXT:  .LBB11_13:
 ; RV32I-NEXT:    srl t6, a6, a5
 ; RV32I-NEXT:    bltu s0, t0, .LBB11_12
 ; RV32I-NEXT:  .LBB11_14:
 ; RV32I-NEXT:    li t4, 0
-; RV32I-NEXT:    sll s2, a3, s0
+; RV32I-NEXT:    sll t5, a3, s0
 ; RV32I-NEXT:  .LBB11_15:
 ; RV32I-NEXT:    addi s1, a5, -64
-; RV32I-NEXT:    mv t5, a4
+; RV32I-NEXT:    mv s2, a4
 ; RV32I-NEXT:    beqz s0, .LBB11_17
 ; RV32I-NEXT:  # %bb.16:
-; RV32I-NEXT:    mv t5, s2
+; RV32I-NEXT:    mv s2, t5
 ; RV32I-NEXT:  .LBB11_17:
 ; RV32I-NEXT:    bltu s1, t0, .LBB11_19
 ; RV32I-NEXT:  # %bb.18:
@@ -2568,9 +2568,9 @@ define void @ashr_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV32I-NEXT:    bnez s1, .LBB11_20
 ; RV32I-NEXT:    j .LBB11_21
 ; RV32I-NEXT:  .LBB11_19:
-; RV32I-NEXT:    neg s0, s1
-; RV32I-NEXT:    sll s0, a4, s0
-; RV32I-NEXT:    or t2, t2, s0
+; RV32I-NEXT:    neg t5, s1
+; RV32I-NEXT:    sll t5, a4, t5
+; RV32I-NEXT:    or t2, t2, t5
 ; RV32I-NEXT:    beqz s1, .LBB11_21
 ; RV32I-NEXT:  .LBB11_20:
 ; RV32I-NEXT:    mv a3, t2
@@ -2585,7 +2585,7 @@ define void @ashr_16bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV32I-NEXT:    bgeu a5, t1, .LBB11_25
 ; RV32I-NEXT:  .LBB11_24:
 ; RV32I-NEXT:    or a3, t3, t4
-; RV32I-NEXT:    or t0, t6, t5
+; RV32I-NEXT:    or t0, t6, s2
 ; RV32I-NEXT:  .LBB11_25:
 ; RV32I-NEXT:    bnez a5, .LBB11_29
 ; RV32I-NEXT:  # %bb.26:
@@ -2809,7 +2809,7 @@ define void @lshr_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:    neg s0, t6
 ; RV64I-NEXT:    srl s0, a3, s0
 ; RV64I-NEXT:    sll t3, a3, t3
-; RV64I-NEXT:    or s1, s0, t5
+; RV64I-NEXT:    or t5, s0, t5
 ; RV64I-NEXT:    j .LBB12_15
 ; RV64I-NEXT:  .LBB12_13:
 ; RV64I-NEXT:    srl t4, a4, a6
@@ -2817,28 +2817,28 @@ define void @lshr_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:  .LBB12_14:
 ; RV64I-NEXT:    li t3, 0
 ; RV64I-NEXT:    sub t5, t6, t1
-; RV64I-NEXT:    sll s1, a3, t5
+; RV64I-NEXT:    sll t5, a3, t5
 ; RV64I-NEXT:  .LBB12_15:
 ; RV64I-NEXT:    sub s0, a6, a7
-; RV64I-NEXT:    mv t5, t0
+; RV64I-NEXT:    mv s1, t0
 ; RV64I-NEXT:    beqz t6, .LBB12_17
 ; RV64I-NEXT:  # %bb.16:
-; RV64I-NEXT:    mv t5, s1
+; RV64I-NEXT:    mv s1, t5
 ; RV64I-NEXT:  .LBB12_17:
 ; RV64I-NEXT:    bltu s0, t1, .LBB12_19
 ; RV64I-NEXT:  # %bb.18:
-; RV64I-NEXT:    sub t6, s0, t1
-; RV64I-NEXT:    srl t6, t0, t6
+; RV64I-NEXT:    sub t5, s0, t1
+; RV64I-NEXT:    srl t5, t0, t5
 ; RV64I-NEXT:    bnez s0, .LBB12_20
 ; RV64I-NEXT:    j .LBB12_21
 ; RV64I-NEXT:  .LBB12_19:
-; RV64I-NEXT:    neg t6, s0
-; RV64I-NEXT:    srl s1, a3, s0
-; RV64I-NEXT:    sll t6, t0, t6
-; RV64I-NEXT:    or t6, s1, t6
+; RV64I-NEXT:    neg t5, s0
+; RV64I-NEXT:    srl t6, a3, s0
+; RV64I-NEXT:    sll t5, t0, t5
+; RV64I-NEXT:    or t5, t6, t5
 ; RV64I-NEXT:    beqz s0, .LBB12_21
 ; RV64I-NEXT:  .LBB12_20:
-; RV64I-NEXT:    mv a3, t6
+; RV64I-NEXT:    mv a3, t5
 ; RV64I-NEXT:  .LBB12_21:
 ; RV64I-NEXT:    bltu s0, t1, .LBB12_23
 ; RV64I-NEXT:  # %bb.22:
@@ -2850,7 +2850,7 @@ define void @lshr_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:    bgeu a6, a7, .LBB12_25
 ; RV64I-NEXT:  .LBB12_24:
 ; RV64I-NEXT:    or a3, t2, t3
-; RV64I-NEXT:    or t0, t4, t5
+; RV64I-NEXT:    or t0, t4, s1
 ; RV64I-NEXT:  .LBB12_25:
 ; RV64I-NEXT:    bnez a6, .LBB12_29
 ; RV64I-NEXT:  # %bb.26:
@@ -3942,7 +3942,7 @@ define void @lshr_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV64I-NEXT:    neg s0, t6
 ; RV64I-NEXT:    srl s0, a3, s0
 ; RV64I-NEXT:    sll t3, a3, t3
-; RV64I-NEXT:    or s1, s0, t5
+; RV64I-NEXT:    or t5, s0, t5
 ; RV64I-NEXT:    j .LBB13_15
 ; RV64I-NEXT:  .LBB13_13:
 ; RV64I-NEXT:    srl t4, a4, a6
@@ -3950,28 +3950,28 @@ define void @lshr_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV64I-NEXT:  .LBB13_14:
 ; RV64I-NEXT:    li t3, 0
 ; RV64I-NEXT:    sub t5, t6, t1
-; RV64I-NEXT:    sll s1, a3, t5
+; RV64I-NEXT:    sll t5, a3, t5
 ; RV64I-NEXT:  .LBB13_15:
 ; RV64I-NEXT:    sub s0, a6, a7
-; RV64I-NEXT:    mv t5, t0
+; RV64I-NEXT:    mv s1, t0
 ; RV64I-NEXT:    beqz t6, .LBB13_17
 ; RV64I-NEXT:  # %bb.16:
-; RV64I-NEXT:    mv t5, s1
+; RV64I-NEXT:    mv s1, t5
 ; RV64I-NEXT:  .LBB13_17:
 ; RV64I-NEXT:    bltu s0, t1, .LBB13_19
 ; RV64I-NEXT:  # %bb.18:
-; RV64I-NEXT:    sub t6, s0, t1
-; RV64I-NEXT:    srl t6, t0, t6
+; RV64I-NEXT:    sub t5, s0, t1
+; RV64I-NEXT:    srl t5, t0, t5
 ; RV64I-NEXT:    bnez s0, .LBB13_20
 ; RV64I-NEXT:    j .LBB13_21
 ; RV64I-NEXT:  .LBB13_19:
-; RV64I-NEXT:    neg t6, s0
-; RV64I-NEXT:    srl s1, a3, s0
-; RV64I-NEXT:    sll t6, t0, t6
-; RV64I-NEXT:    or t6, s1, t6
+; RV64I-NEXT:    neg t5, s0
+; RV64I-NEXT:    srl t6, a3, s0
+; RV64I-NEXT:    sll t5, t0, t5
+; RV64I-NEXT:    or t5, t6, t5
 ; RV64I-NEXT:    beqz s0, .LBB13_21
 ; RV64I-NEXT:  .LBB13_20:
-; RV64I-NEXT:    mv a3, t6
+; RV64I-NEXT:    mv a3, t5
 ; RV64I-NEXT:  .LBB13_21:
 ; RV64I-NEXT:    bltu s0, t1, .LBB13_23
 ; RV64I-NEXT:  # %bb.22:
@@ -3983,7 +3983,7 @@ define void @lshr_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV64I-NEXT:    bgeu a6, a7, .LBB13_25
 ; RV64I-NEXT:  .LBB13_24:
 ; RV64I-NEXT:    or a3, t2, t3
-; RV64I-NEXT:    or t0, t4, t5
+; RV64I-NEXT:    or t0, t4, s1
 ; RV64I-NEXT:  .LBB13_25:
 ; RV64I-NEXT:    bnez a6, .LBB13_29
 ; RV64I-NEXT:  # %bb.26:
@@ -5075,7 +5075,7 @@ define void @lshr_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) no
 ; RV64I-NEXT:    neg s0, t6
 ; RV64I-NEXT:    srl s0, a3, s0
 ; RV64I-NEXT:    sll t3, a3, t3
-; RV64I-NEXT:    or s1, s0, t5
+; RV64I-NEXT:    or t5, s0, t5
 ; RV64I-NEXT:    j .LBB14_15
 ; RV64I-NEXT:  .LBB14_13:
 ; RV64I-NEXT:    srl t4, a4, a6
@@ -5083,28 +5083,28 @@ define void @lshr_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) no
 ; RV64I-NEXT:  .LBB14_14:
 ; RV64I-NEXT:    li t3, 0
 ; RV64I-NEXT:    sub t5, t6, t1
-; RV64I-NEXT:    sll s1, a3, t5
+; RV64I-NEXT:    sll t5, a3, t5
 ; RV64I-NEXT:  .LBB14_15:
 ; RV64I-NEXT:    sub s0, a6, a7
-; RV64I-NEXT:    mv t5, t0
+; RV64I-NEXT:    mv s1, t0
 ; RV64I-NEXT:    beqz t6, .LBB14_17
 ; RV64I-NEXT:  # %bb.16:
-; RV64I-NEXT:    mv t5, s1
+; RV64I-NEXT:    mv s1, t5
 ; RV64I-NEXT:  .LBB14_17:
 ; RV64I-NEXT:    bltu s0, t1, .LBB14_19
 ; RV64I-NEXT:  # %bb.18:
-; RV64I-NEXT:    sub t6, s0, t1
-; RV64I-NEXT:    srl t6, t0, t6
+; RV64I-NEXT:    sub t5, s0, t1
+; RV64I-NEXT:    srl t5, t0, t5
 ; RV64I-NEXT:    bnez s0, .LBB14_20
 ; RV64I-NEXT:    j .LBB14_21
 ; RV64I-NEXT:  .LBB14_19:
-; RV64I-NEXT:    neg t6, s0
-; RV64I-NEXT:    srl s1, a3, s0
-; RV64I-NEXT:    sll t6, t0, t6
-; RV64I-NEXT:    or t6, s1, t6
+; RV64I-NEXT:    neg t5, s0
+; RV64I-NEXT:    srl t6, a3, s0
+; RV64I-NEXT:    sll t5, t0, t5
+; RV64I-NEXT:    or t5, t6, t5
 ; RV64I-NEXT:    beqz s0, .LBB14_21
 ; RV64I-NEXT:  .LBB14_20:
-; RV64I-NEXT:    mv a3, t6
+; RV64I-NEXT:    mv a3, t5
 ; RV64I-NEXT:  .LBB14_21:
 ; RV64I-NEXT:    bltu s0, t1, .LBB14_23
 ; RV64I-NEXT:  # %bb.22:
@@ -5116,7 +5116,7 @@ define void @lshr_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) no
 ; RV64I-NEXT:    bgeu a6, a7, .LBB14_25
 ; RV64I-NEXT:  .LBB14_24:
 ; RV64I-NEXT:    or a3, t2, t3
-; RV64I-NEXT:    or t0, t4, t5
+; RV64I-NEXT:    or t0, t4, s1
 ; RV64I-NEXT:  .LBB14_25:
 ; RV64I-NEXT:    bnez a6, .LBB14_29
 ; RV64I-NEXT:  # %bb.26:
@@ -6146,67 +6146,67 @@ define void @shl_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:    bltu a6, t1, .LBB15_2
 ; RV64I-NEXT:  # %bb.1:
 ; RV64I-NEXT:    li a1, 0
-; RV64I-NEXT:    sll s4, a4, t3
+; RV64I-NEXT:    sll a7, a4, t3
 ; RV64I-NEXT:    j .LBB15_3
 ; RV64I-NEXT:  .LBB15_2:
 ; RV64I-NEXT:    sll a7, a3, a6
 ; RV64I-NEXT:    sll a1, a4, a6
-; RV64I-NEXT:    or s4, t6, a7
+; RV64I-NEXT:    or a7, t6, a7
 ; RV64I-NEXT:  .LBB15_3:
 ; RV64I-NEXT:    slli s8, t4, 8
 ; RV64I-NEXT:    slli t4, t0, 8
 ; RV64I-NEXT:    lbu ra, 17(a0)
 ; RV64I-NEXT:    lbu s11, 18(a0)
-; RV64I-NEXT:    lbu a7, 20(a0)
+; RV64I-NEXT:    lbu s4, 20(a0)
 ; RV64I-NEXT:    lbu s6, 25(a0)
 ; RV64I-NEXT:    lbu s5, 26(a0)
 ; RV64I-NEXT:    lbu s9, 28(a0)
 ; RV64I-NEXT:    or s7, t5, a5
 ; RV64I-NEXT:    slli s10, s2, 8
-; RV64I-NEXT:    slli s2, s1, 8
-; RV64I-NEXT:    or s3, s3, s0
+; RV64I-NEXT:    slli s1, s1, 8
+; RV64I-NEXT:    or s2, s3, s0
 ; RV64I-NEXT:    li a5, 128
 ; RV64I-NEXT:    sub t0, a5, a6
 ; RV64I-NEXT:    mv a5, a3
 ; RV64I-NEXT:    beqz a6, .LBB15_5
 ; RV64I-NEXT:  # %bb.4:
-; RV64I-NEXT:    mv a5, s4
+; RV64I-NEXT:    mv a5, a7
 ; RV64I-NEXT:  .LBB15_5:
-; RV64I-NEXT:    slli s0, ra, 8
-; RV64I-NEXT:    or t5, s8, s11
-; RV64I-NEXT:    lbu s1, 16(a0)
+; RV64I-NEXT:    slli t5, ra, 8
+; RV64I-NEXT:    or a7, s8, s11
+; RV64I-NEXT:    lbu s0, 16(a0)
 ; RV64I-NEXT:    lbu a0, 24(a0)
-; RV64I-NEXT:    or t4, t4, a7
+; RV64I-NEXT:    or t4, t4, s4
 ; RV64I-NEXT:    slli s7, s7, 16
 ; RV64I-NEXT:    slli s6, s6, 8
-; RV64I-NEXT:    or s4, s10, s5
-; RV64I-NEXT:    or s2, s2, s9
-; RV64I-NEXT:    slli s3, s3, 16
+; RV64I-NEXT:    or s3, s10, s5
+; RV64I-NEXT:    or s1, s1, s9
+; RV64I-NEXT:    slli s2, s2, 16
 ; RV64I-NEXT:    bltu t0, t1, .LBB15_7
 ; RV64I-NEXT:  # %bb.6:
-; RV64I-NEXT:    sub a7, t0, t1
-; RV64I-NEXT:    srl a7, a3, a7
+; RV64I-NEXT:    sub t6, t0, t1
+; RV64I-NEXT:    srl t6, a3, t6
 ; RV64I-NEXT:    j .LBB15_8
 ; RV64I-NEXT:  .LBB15_7:
-; RV64I-NEXT:    neg a7, t0
-; RV64I-NEXT:    sll a7, a3, a7
-; RV64I-NEXT:    or a7, t6, a7
+; RV64I-NEXT:    neg s4, t0
+; RV64I-NEXT:    sll s4, a3, s4
+; RV64I-NEXT:    or t6, t6, s4
 ; RV64I-NEXT:  .LBB15_8:
-; RV64I-NEXT:    or s0, s0, s1
-; RV64I-NEXT:    slli s1, t5, 16
-; RV64I-NEXT:    or t6, s7, t4
-; RV64I-NEXT:    or t5, s6, a0
-; RV64I-NEXT:    slli s4, s4, 16
-; RV64I-NEXT:    or s2, s3, s2
+; RV64I-NEXT:    or t5, t5, s0
+; RV64I-NEXT:    slli s4, a7, 16
+; RV64I-NEXT:    or s0, s7, t4
+; RV64I-NEXT:    or a7, s6, a0
+; RV64I-NEXT:    slli s3, s3, 16
+; RV64I-NEXT:    or s1, s2, s1
 ; RV64I-NEXT:    mv t4, a4
 ; RV64I-NEXT:    beqz t0, .LBB15_10
 ; RV64I-NEXT:  # %bb.9:
-; RV64I-NEXT:    mv t4, a7
+; RV64I-NEXT:    mv t4, t6
 ; RV64I-NEXT:  .LBB15_10:
-; RV64I-NEXT:    or a0, s1, s0
-; RV64I-NEXT:    slli t6, t6, 32
-; RV64I-NEXT:    or a7, s4, t5
-; RV64I-NEXT:    slli s0, s2, 32
+; RV64I-NEXT:    or a0, s4, t5
+; RV64I-NEXT:    slli t6, s0, 32
+; RV64I-NEXT:    or a7, s3, a7
+; RV64I-NEXT:    slli s0, s1, 32
 ; RV64I-NEXT:    bltu t0, t1, .LBB15_12
 ; RV64I-NEXT:  # %bb.11:
 ; RV64I-NEXT:    li t5, 0
@@ -6219,35 +6219,35 @@ define void @shl_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:    bltu a6, t1, .LBB15_15
 ; RV64I-NEXT:  # %bb.14:
 ; RV64I-NEXT:    li t2, 0
-; RV64I-NEXT:    sll a7, t0, t3
+; RV64I-NEXT:    sll t3, t0, t3
 ; RV64I-NEXT:    j .LBB15_16
 ; RV64I-NEXT:  .LBB15_15:
 ; RV64I-NEXT:    srl a7, t0, t2
 ; RV64I-NEXT:    sll t3, a0, a6
 ; RV64I-NEXT:    sll t2, t0, a6
-; RV64I-NEXT:    or a7, a7, t3
+; RV64I-NEXT:    or t3, a7, t3
 ; RV64I-NEXT:  .LBB15_16:
-; RV64I-NEXT:    li t3, 128
-; RV64I-NEXT:    sub s0, a6, t3
+; RV64I-NEXT:    li a7, 128
+; RV64I-NEXT:    sub a7, a6, a7
 ; RV64I-NEXT:    mv t6, a0
 ; RV64I-NEXT:    beqz a6, .LBB15_18
 ; RV64I-NEXT:  # %bb.17:
-; RV64I-NEXT:    mv t6, a7
+; RV64I-NEXT:    mv t6, t3
 ; RV64I-NEXT:  .LBB15_18:
-; RV64I-NEXT:    bltu s0, t1, .LBB15_20
+; RV64I-NEXT:    bltu a7, t1, .LBB15_20
 ; RV64I-NEXT:  # %bb.19:
 ; RV64I-NEXT:    li t3, 0
-; RV64I-NEXT:    sub a7, s0, t1
-; RV64I-NEXT:    sll a4, a4, a7
-; RV64I-NEXT:    bnez s0, .LBB15_21
+; RV64I-NEXT:    sub t1, a7, t1
+; RV64I-NEXT:    sll a4, a4, t1
+; RV64I-NEXT:    bnez a7, .LBB15_21
 ; RV64I-NEXT:    j .LBB15_22
 ; RV64I-NEXT:  .LBB15_20:
-; RV64I-NEXT:    neg a7, s0
-; RV64I-NEXT:    srl a7, a4, a7
-; RV64I-NEXT:    sll t1, a3, s0
-; RV64I-NEXT:    sll t3, a4, s0
-; RV64I-NEXT:    or a4, a7, t1
-; RV64I-NEXT:    beqz s0, .LBB15_22
+; RV64I-NEXT:    neg t1, a7
+; RV64I-NEXT:    srl t1, a4, t1
+; RV64I-NEXT:    sll s0, a3, a7
+; RV64I-NEXT:    sll t3, a4, a7
+; RV64I-NEXT:    or a4, t1, s0
+; RV64I-NEXT:    beqz a7, .LBB15_22
 ; RV64I-NEXT:  .LBB15_21:
 ; RV64I-NEXT:    mv a3, a4
 ; RV64I-NEXT:  .LBB15_22:
@@ -7269,67 +7269,67 @@ define void @shl_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) nounw
 ; RV64I-NEXT:    bltu a6, t1, .LBB16_2
 ; RV64I-NEXT:  # %bb.1:
 ; RV64I-NEXT:    li a1, 0
-; RV64I-NEXT:    sll s4, a4, t3
+; RV64I-NEXT:    sll a7, a4, t3
 ; RV64I-NEXT:    j .LBB16_3
 ; RV64I-NEXT:  .LBB16_2:
 ; RV64I-NEXT:    sll a7, a3, a6
 ; RV64I-NEXT:    sll a1, a4, a6
-; RV64I-NEXT:    or s4, t6, a7
+; RV64I-NEXT:    or a7, t6, a7
 ; RV64I-NEXT:  .LBB16_3:
 ; RV64I-NEXT:    slli s8, t4, 8
 ; RV64I-NEXT:    slli t4, t0, 8
 ; RV64I-NEXT:    lbu ra, 17(a0)
 ; RV64I-NEXT:    lbu s11, 18(a0)
-; RV64I-NEXT:    lbu a7, 20(a0)
+; RV64I-NEXT:    lbu s4, 20(a0)
 ; RV64I-NEXT:    lbu s6, 25(a0)
 ; RV64I-NEXT:    lbu s5, 26(a0)
 ; RV64I-NEXT:    lbu s9, 28(a0)
 ; RV64I-NEXT:    or s7, t5, a5
 ; RV64I-NEXT:    slli s10, s2, 8
-; RV64I-NEXT:    slli s2, s1, 8
-; RV64I-NEXT:    or s3, s3, s0
+; RV64I-NEXT:    slli s1, s1, 8
+; RV64I-NEXT:    or s2, s3, s0
 ; RV64I-NEXT:    li a5, 128
 ; RV64I-NEXT:    sub t0, a5, a6
 ; RV64I-NEXT:    mv a5, a3
 ; RV64I-NEXT:    beqz a6, .LBB16_5
 ; RV64I-NEXT:  # %bb.4:
-; RV64I-NEXT:    mv a5, s4
+; RV64I-NEXT:    mv a5, a7
 ; RV64I-NEXT:  .LBB16_5:
-; RV64I-NEXT:    slli s0, ra, 8
-; RV64I-NEXT:    or t5, s8, s11
-; RV64I-NEXT:    lbu s1, 16(a0)
+; RV64I-NEXT:    slli t5, ra, 8
+; RV64I-NEXT:    or a7, s8, s11
+; RV64I-NEXT:    lbu s0, 16(a0)
 ; RV64I-NEXT:    lbu a0, 24(a0)
-; RV64I-NEXT:    or t4, t4, a7
+; RV64I-NEXT:    or t4, t4, s4
 ; RV64I-NEXT:    slli s7, s7, 16
 ; RV64I-NEXT:    slli s6, s6, 8
-; RV64I-NEXT:    or s4, s10, s5
-; RV64I-NEXT:    or s2, s2, s9
-; RV64I-NEXT:    slli s3, s3, 16
+; RV64I-NEXT:    or s3, s10, s5
+; RV64I-NEXT:    or s1, s1, s9
+; RV64I-NEXT:    slli s2, s2, 16
 ; RV64I-NEXT:    bltu t0, t1, .LBB16_7
 ; RV64I-NEXT:  # %bb.6:
-; RV64I-NEXT:    sub a7, t0, t1
-; RV64I-NEXT:    srl a7, a3, a7
+; RV64I-NEXT:    sub t6, t0, t1
+; RV64I-NEXT:    srl t6, a3, t6
 ; RV64I-NEXT:    j .LBB16_8
 ; RV64I-NEXT:  .LBB16_7:
-; RV64I-NEXT:    neg a7, t0
-; RV64I-NEXT:    sll a7, a3, a7
-; RV64I-NEXT:    or a7, t6, a7
+; RV64I-NEXT:    neg s4, t0
+; RV64I-NEXT:    sll s4, a3, s4
+; RV64I-NEXT:    or t6, t6, s4
 ; RV64I-NEXT:  .LBB16_8:
-; RV64I-NEXT:    or s0, s0, s1
-; RV64I-NEXT:    slli s1, t5, 16
-; RV64I-NEXT:    or t6, s7, t4
-; RV64I-NEXT:    or t5, s6, a0
-; RV64I-NEXT:    slli s4, s4, 16
-; RV64I-NEXT:    or s2, s3, s2
+; RV64I-NEXT:    or t5, t5, s0
+; RV64I-NEXT:    slli s4, a7, 16
+; RV64I-NEXT:    or s0, s7, t4
+; RV64I-NEXT:    or a7, s6, a0
+; RV64I-NEXT:    slli s3, s3, 16
+; RV64I-NEXT:    or s1, s2, s1
 ; RV64I-NEXT:    mv t4, a4
 ; RV64I-NEXT:    beqz t0, .LBB16_10
 ; RV64I-NEXT:  # %bb.9:
-; RV64I-NEXT:    mv t4, a7
+; RV64I-NEXT:    mv t4, t6
 ; RV64I-NEXT:  .LBB16_10:
-; RV64I-NEXT:    or a0, s1, s0
-; RV64I-NEXT:    slli t6, t6, 32
-; RV64I-NEXT:    or a7, s4, t5
-; RV64I-NEXT:    slli s0, s2, 32
+; RV64I-NEXT:    or a0, s4, t5
+; RV64I-NEXT:    slli t6, s0, 32
+; RV64I-NEXT:    or a7, s3, a7
+; RV64I-NEXT:    slli s0, s1, 32
 ; RV64I-NEXT:    bltu t0, t1, .LBB16_12
 ; RV64I-NEXT:  # %bb.11:
 ; RV64I-NEXT:    li t5, 0
@@ -7342,35 +7342,35 @@ define void @shl_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) nounw
 ; RV64I-NEXT:    bltu a6, t1, .LBB16_15
 ; RV64I-NEXT:  # %bb.14:
 ; RV64I-NEXT:    li t2, 0
-; RV64I-NEXT:    sll a7, t0, t3
+; RV64I-NEXT:    sll t3, t0, t3
 ; RV64I-NEXT:    j .LBB16_16
 ; RV64I-NEXT:  .LBB16_15:
 ; RV64I-NEXT:    srl a7, t0, t2
 ; RV64I-NEXT:    sll t3, a0, a6
 ; RV64I-NEXT:    sll t2, t0, a6
-; RV64I-NEXT:    or a7, a7, t3
+; RV64I-NEXT:    or t3, a7, t3
 ; RV64I-NEXT:  .LBB16_16:
-; RV64I-NEXT:    li t3, 128
-; RV64I-NEXT:    sub s0, a6, t3
+; RV64I-NEXT:    li a7, 128
+; RV64I-NEXT:    sub a7, a6, a7
 ; RV64I-NEXT:    mv t6, a0
 ; RV64I-NEXT:    beqz a6, .LBB16_18
 ; RV64I-NEXT:  # %bb.17:
-; RV64I-NEXT:    mv t6, a7
+; RV64I-NEXT:    mv t6, t3
 ; RV64I-NEXT:  .LBB16_18:
-; RV64I-NEXT:    bltu s0, t1, .LBB16_20
+; RV64I-NEXT:    bltu a7, t1, .LBB16_20
 ; RV64I-NEXT:  # %bb.19:
 ; RV64I-NEXT:    li t3, 0
-; RV64I-NEXT:    sub a7, s0, t1
-; RV64I-NEXT:    sll a4, a4, a7
-; RV64I-NEXT:    bnez s0, .LBB16_21
+; RV64I-NEXT:    sub t1, a7, t1
+; RV64I-NEXT:    sll a4, a4, t1
+; RV64I-NEXT:    bnez a7, .LBB16_21
 ; RV64I-NEXT:    j .LBB16_22
 ; RV64I-NEXT:  .LBB16_20:
-; RV64I-NEXT:    neg a7, s0
-; RV64I-NEXT:    srl a7, a4, a7
-; RV64I-NEXT:    sll t1, a3, s0
-; RV64I-NEXT:    sll t3, a4, s0
-; RV64I-NEXT:    or a4, a7, t1
-; RV64I-NEXT:    beqz s0, .LBB16_22
+; RV64I-NEXT:    neg t1, a7
+; RV64I-NEXT:    srl t1, a4, t1
+; RV64I-NEXT:    sll s0, a3, a7
+; RV64I-NEXT:    sll t3, a4, a7
+; RV64I-NEXT:    or a4, t1, s0
+; RV64I-NEXT:    beqz a7, .LBB16_22
 ; RV64I-NEXT:  .LBB16_21:
 ; RV64I-NEXT:    mv a3, a4
 ; RV64I-NEXT:  .LBB16_22:
@@ -8401,67 +8401,67 @@ define void @shl_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) nou
 ; RV64I-NEXT:    bltu a6, t1, .LBB17_2
 ; RV64I-NEXT:  # %bb.1:
 ; RV64I-NEXT:    li a1, 0
-; RV64I-NEXT:    sll s4, a4, t3
+; RV64I-NEXT:    sll a7, a4, t3
 ; RV64I-NEXT:    j .LBB17_3
 ; RV64I-NEXT:  .LBB17_2:
 ; RV64I-NEXT:    sll a7, a3, a6
 ; RV64I-NEXT:    sll a1, a4, a6
-; RV64I-NEXT:    or s4, t6, a7
+; RV64I-NEXT:    or a7, t6, a7
 ; RV64I-NEXT:  .LBB17_3:
 ; RV64I-NEXT:    slli s8, t4, 8
 ; RV64I-NEXT:    slli t4, t0, 8
 ; RV64I-NEXT:    lbu ra, 17(a0)
 ; RV64I-NEXT:    lbu s11, 18(a0)
-; RV64I-NEXT:    lbu a7, 20(a0)
+; RV64I-NEXT:    lbu s4, 20(a0)
 ; RV64I-NEXT:    lbu s6, 25(a0)
 ; RV64I-NEXT:    lbu s5, 26(a0)
 ; RV64I-NEXT:    lbu s9, 28(a0)
 ; RV64I-NEXT:    or s7, t5, a5
 ; RV64I-NEXT:    slli s10, s2, 8
-; RV64I-NEXT:    slli s2, s1, 8
-; RV64I-NEXT:    or s3, s3, s0
+; RV64I-NEXT:    slli s1, s1, 8
+; RV64I-NEXT:    or s2, s3, s0
 ; RV64I-NEXT:    li a5, 128
 ; RV64I-NEXT:    sub t0, a5, a6
 ; RV64I-NEXT:    mv a5, a3
 ; RV64I-NEXT:    beqz a6, .LBB17_5
 ; RV64I-NEXT:  # %bb.4:
-; RV64I-NEXT:    mv a5, s4
+; RV64I-NEXT:    mv a5, a7
 ; RV64I-NEXT:  .LBB17_5:
-; RV64I-NEXT:    slli s0, ra, 8
-; RV64I-NEXT:    or t5, s8, s11
-; RV64I-NEXT:    lbu s1, 16(a0)
+; RV64I-NEXT:    slli t5, ra, 8
+; RV64I-NEXT:    or a7, s8, s11
+; RV64I-NEXT:    lbu s0, 16(a0)
 ; RV64I-NEXT:    lbu a0, 24(a0)
-; RV64I-NEXT:    or t4, t4, a7
+; RV64I-NEXT:    or t4, t4, s4
 ; RV64I-NEXT:    slli s7, s7, 16
 ; RV64I-NEXT:    slli s6, s6, 8
-; RV64I-NEXT:    or s4, s10, s5
-; RV64I-NEXT:    or s2, s2, s9
-; RV64I-NEXT:    slli s3, s3, 16
+; RV64I-NEXT:    or s3, s10, s5
+; RV64I-NEXT:    or s1, s1, s9
+; RV64I-NEXT:    slli s2, s2, 16
 ; RV64I-NEXT:    bltu t0, t1, .LBB17_7
 ; RV64I-NEXT:  # %bb.6:
-; RV64I-NEXT:    sub a7, t0, t1
-; RV64I-NEXT:    srl a7, a3, a7
+; RV64I-NEXT:    sub t6, t0, t1
+; RV64I-NEXT:    srl t6, a3, t6
 ; RV64I-NEXT:    j .LBB17_8
 ; RV64I-NEXT:  .LBB17_7:
-; RV64I-NEXT:    neg a7, t0
-; RV64I-NEXT:    sll a7, a3, a7
-; RV64I-NEXT:    or a7, t6, a7
+; RV64I-NEXT:    neg s4, t0
+; RV64I-NEXT:    sll s4, a3, s4
+; RV64I-NEXT:    or t6, t6, s4
 ; RV64I-NEXT:  .LBB17_8:
-; RV64I-NEXT:    or s0, s0, s1
-; RV64I-NEXT:    slli s1, t5, 16
-; RV64I-NEXT:    or t6, s7, t4
-; RV64I-NEXT:    or t5, s6, a0
-; RV64I-NEXT:    slli s4, s4, 16
-; RV64I-NEXT:    or s2, s3, s2
+; RV64I-NEXT:    or t5, t5, s0
+; RV64I-NEXT:    slli s4, a7, 16
+; RV64I-NEXT:    or s0, s7, t4
+; RV64I-NEXT:    or a7, s6, a0
+; RV64I-NEXT:    slli s3, s3, 16
+; RV64I-NEXT:    or s1, s2, s1
 ; RV64I-NEXT:    mv t4, a4
 ; RV64I-NEXT:    beqz t0, .LBB17_10
 ; RV64I-NEXT:  # %bb.9:
-; RV64I-NEXT:    mv t4, a7
+; RV64I-NEXT:    mv t4, t6
 ; RV64I-NEXT:  .LBB17_10:
-; RV64I-NEXT:    or a0, s1, s0
-; RV64I-NEXT:    slli t6, t6, 32
-; RV64I-NEXT:    or a7, s4, t5
-; RV64I-NEXT:    slli s0, s2, 32
+; RV64I-NEXT:    or a0, s4, t5
+; RV64I-NEXT:    slli t6, s0, 32
+; RV64I-NEXT:    or a7, s3, a7
+; RV64I-NEXT:    slli s0, s1, 32
 ; RV64I-NEXT:    bltu t0, t1, .LBB17_12
 ; RV64I-NEXT:  # %bb.11:
 ; RV64I-NEXT:    li t5, 0
@@ -8474,35 +8474,35 @@ define void @shl_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) nou
 ; RV64I-NEXT:    bltu a6, t1, .LBB17_15
 ; RV64I-NEXT:  # %bb.14:
 ; RV64I-NEXT:    li t2, 0
-; RV64I-NEXT:    sll a7, t0, t3
+; RV64I-NEXT:    sll t3, t0, t3
 ; RV64I-NEXT:    j .LBB17_16
 ; RV64I-NEXT:  .LBB17_15:
 ; RV64I-NEXT:    srl a7, t0, t2
 ; RV64I-NEXT:    sll t3, a0, a6
 ; RV64I-NEXT:    sll t2, t0, a6
-; RV64I-NEXT:    or a7, a7, t3
+; RV64I-NEXT:    or t3, a7, t3
 ; RV64I-NEXT:  .LBB17_16:
-; RV64I-NEXT:    li t3, 128
-; RV64I-NEXT:    sub s0, a6, t3
+; RV64I-NEXT:    li a7, 128
+; RV64I-NEXT:    sub a7, a6, a7
 ; RV64I-NEXT:    mv t6, a0
 ; RV64I-NEXT:    beqz a6, .LBB17_18
 ; RV64I-NEXT:  # %bb.17:
-; RV64I-NEXT:    mv t6, a7
+; RV64I-NEXT:    mv t6, t3
 ; RV64I-NEXT:  .LBB17_18:
-; RV64I-NEXT:    bltu s0, t1, .LBB17_20
+; RV64I-NEXT:    bltu a7, t1, .LBB17_20
 ; RV64I-NEXT:  # %bb.19:
 ; RV64I-NEXT:    li t3, 0
-; RV64I-NEXT:    sub a7, s0, t1
-; RV64I-NEXT:    sll a4, a4, a7
-; RV64I-NEXT:    bnez s0, .LBB17_21
+; RV64I-NEXT:    sub t1, a7, t1
+; RV64I-NEXT:    sll a4, a4, t1
+; RV64I-NEXT:    bnez a7, .LBB17_21
 ; RV64I-NEXT:    j .LBB17_22
 ; RV64I-NEXT:  .LBB17_20:
-; RV64I-NEXT:    neg a7, s0
-; RV64I-NEXT:    srl a7, a4, a7
-; RV64I-NEXT:    sll t1, a3, s0
-; RV64I-NEXT:    sll t3, a4, s0
-; RV64I-NEXT:    or a4, a7, t1
-; RV64I-NEXT:    beqz s0, .LBB17_22
+; RV64I-NEXT:    neg t1, a7
+; RV64I-NEXT:    srl t1, a4, t1
+; RV64I-NEXT:    sll s0, a3, a7
+; RV64I-NEXT:    sll t3, a4, a7
+; RV64I-NEXT:    or a4, t1, s0
+; RV64I-NEXT:    beqz a7, .LBB17_22
 ; RV64I-NEXT:  .LBB17_21:
 ; RV64I-NEXT:    mv a3, a4
 ; RV64I-NEXT:  .LBB17_22:
@@ -9583,7 +9583,7 @@ define void @ashr_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:    neg s0, t6
 ; RV64I-NEXT:    srl s0, a3, s0
 ; RV64I-NEXT:    sll t4, a3, t4
-; RV64I-NEXT:    or s1, s0, t5
+; RV64I-NEXT:    or t5, s0, t5
 ; RV64I-NEXT:    j .LBB18_15
 ; RV64I-NEXT:  .LBB18_13:
 ; RV64I-NEXT:    srl t3, a4, a7
@@ -9591,28 +9591,28 @@ define void @ashr_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:  .LBB18_14:
 ; RV64I-NEXT:    li t4, 0
 ; RV64I-NEXT:    sub t5, t6, t1
-; RV64I-NEXT:    sll s1, a3, t5
+; RV64I-NEXT:    sll t5, a3, t5
 ; RV64I-NEXT:  .LBB18_15:
 ; RV64I-NEXT:    sub s0, a7, t0
-; RV64I-NEXT:    mv t5, a5
+; RV64I-NEXT:    mv s1, a5
 ; RV64I-NEXT:    beqz t6, .LBB18_17
 ; RV64I-NEXT:  # %bb.16:
-; RV64I-NEXT:    mv t5, s1
+; RV64I-NEXT:    mv s1, t5
 ; RV64I-NEXT:  .LBB18_17:
 ; RV64I-NEXT:    bltu s0, t1, .LBB18_19
 ; RV64I-NEXT:  # %bb.18:
-; RV64I-NEXT:    sub t6, s0, t1
-; RV64I-NEXT:    sra t6, a5, t6
+; RV64I-NEXT:    sub t5, s0, t1
+; RV64I-NEXT:    sra t5, a5, t5
 ; RV64I-NEXT:    bnez s0, .LBB18_20
 ; RV64I-NEXT:    j .LBB18_21
 ; RV64I-NEXT:  .LBB18_19:
-; RV64I-NEXT:    neg t6, s0
-; RV64I-NEXT:    srl s1, a3, s0
-; RV64I-NEXT:    sll t6, a5, t6
-; RV64I-NEXT:    or t6, s1, t6
+; RV64I-NEXT:    neg t5, s0
+; RV64I-NEXT:    srl t6, a3, s0
+; RV64I-NEXT:    sll t5, a5, t5
+; RV64I-NEXT:    or t5, t6, t5
 ; RV64I-NEXT:    beqz s0, .LBB18_21
 ; RV64I-NEXT:  .LBB18_20:
-; RV64I-NEXT:    mv a3, t6
+; RV64I-NEXT:    mv a3, t5
 ; RV64I-NEXT:  .LBB18_21:
 ; RV64I-NEXT:    bltu s0, t1, .LBB18_23
 ; RV64I-NEXT:  # %bb.22:
@@ -9624,7 +9624,7 @@ define void @ashr_32bytes(ptr %src.ptr, ptr %byteOff.ptr, ptr %dst) nounwind {
 ; RV64I-NEXT:    bgeu a7, t0, .LBB18_25
 ; RV64I-NEXT:  .LBB18_24:
 ; RV64I-NEXT:    or a3, t2, t4
-; RV64I-NEXT:    or t1, t3, t5
+; RV64I-NEXT:    or t1, t3, s1
 ; RV64I-NEXT:  .LBB18_25:
 ; RV64I-NEXT:    bnez a7, .LBB18_29
 ; RV64I-NEXT:  # %bb.26:
@@ -10771,7 +10771,7 @@ define void @ashr_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV64I-NEXT:    neg s0, t6
 ; RV64I-NEXT:    srl s0, a3, s0
 ; RV64I-NEXT:    sll t4, a3, t4
-; RV64I-NEXT:    or s1, s0, t5
+; RV64I-NEXT:    or t5, s0, t5
 ; RV64I-NEXT:    j .LBB19_15
 ; RV64I-NEXT:  .LBB19_13:
 ; RV64I-NEXT:    srl t3, a4, a7
@@ -10779,28 +10779,28 @@ define void @ashr_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV64I-NEXT:  .LBB19_14:
 ; RV64I-NEXT:    li t4, 0
 ; RV64I-NEXT:    sub t5, t6, t1
-; RV64I-NEXT:    sll s1, a3, t5
+; RV64I-NEXT:    sll t5, a3, t5
 ; RV64I-NEXT:  .LBB19_15:
 ; RV64I-NEXT:    sub s0, a7, t0
-; RV64I-NEXT:    mv t5, a5
+; RV64I-NEXT:    mv s1, a5
 ; RV64I-NEXT:    beqz t6, .LBB19_17
 ; RV64I-NEXT:  # %bb.16:
-; RV64I-NEXT:    mv t5, s1
+; RV64I-NEXT:    mv s1, t5
 ; RV64I-NEXT:  .LBB19_17:
 ; RV64I-NEXT:    bltu s0, t1, .LBB19_19
 ; RV64I-NEXT:  # %bb.18:
-; RV64I-NEXT:    sub t6, s0, t1
-; RV64I-NEXT:    sra t6, a5, t6
+; RV64I-NEXT:    sub t5, s0, t1
+; RV64I-NEXT:    sra t5, a5, t5
 ; RV64I-NEXT:    bnez s0, .LBB19_20
 ; RV64I-NEXT:    j .LBB19_21
 ; RV64I-NEXT:  .LBB19_19:
-; RV64I-NEXT:    neg t6, s0
-; RV64I-NEXT:    srl s1, a3, s0
-; RV64I-NEXT:    sll t6, a5, t6
-; RV64I-NEXT:    or t6, s1, t6
+; RV64I-NEXT:    neg t5, s0
+; RV64I-NEXT:    srl t6, a3, s0
+; RV64I-NEXT:    sll t5, a5, t5
+; RV64I-NEXT:    or t5, t6, t5
 ; RV64I-NEXT:    beqz s0, .LBB19_21
 ; RV64I-NEXT:  .LBB19_20:
-; RV64I-NEXT:    mv a3, t6
+; RV64I-NEXT:    mv a3, t5
 ; RV64I-NEXT:  .LBB19_21:
 ; RV64I-NEXT:    bltu s0, t1, .LBB19_23
 ; RV64I-NEXT:  # %bb.22:
@@ -10812,7 +10812,7 @@ define void @ashr_32bytes_wordOff(ptr %src.ptr, ptr %wordOff.ptr, ptr %dst) noun
 ; RV64I-NEXT:    bgeu a7, t0, .LBB19_25
 ; RV64I-NEXT:  .LBB19_24:
 ; RV64I-NEXT:    or a3, t2, t4
-; RV64I-NEXT:    or t1, t3, t5
+; RV64I-NEXT:    or t1, t3, s1
 ; RV64I-NEXT:  .LBB19_25:
 ; RV64I-NEXT:    bnez a7, .LBB19_29
 ; RV64I-NEXT:  # %bb.26:
@@ -11959,7 +11959,7 @@ define void @ashr_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) no
 ; RV64I-NEXT:    neg s0, t6
 ; RV64I-NEXT:    srl s0, a3, s0
 ; RV64I-NEXT:    sll t4, a3, t4
-; RV64I-NEXT:    or s1, s0, t5
+; RV64I-NEXT:    or t5, s0, t5
 ; RV64I-NEXT:    j .LBB20_15
 ; RV64I-NEXT:  .LBB20_13:
 ; RV64I-NEXT:    srl t3, a4, a7
@@ -11967,28 +11967,28 @@ define void @ashr_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) no
 ; RV64I-NEXT:  .LBB20_14:
 ; RV64I-NEXT:    li t4, 0
 ; RV64I-NEXT:    sub t5, t6, t1
-; RV64I-NEXT:    sll s1, a3, t5
+; RV64I-NEXT:    sll t5, a3, t5
 ; RV64I-NEXT:  .LBB20_15:
 ; RV64I-NEXT:    sub s0, a7, t0
-; RV64I-NEXT:    mv t5, a5
+; RV64I-NEXT:    mv s1, a5
 ; RV64I-NEXT:    beqz t6, .LBB20_17
 ; RV64I-NEXT:  # %bb.16:
-; RV64I-NEXT:    mv t5, s1
+; RV64I-NEXT:    mv s1, t5
 ; RV64I-NEXT:  .LBB20_17:
 ; RV64I-NEXT:    bltu s0, t1, .LBB20_19
 ; RV64I-NEXT:  # %bb.18:
-; RV64I-NEXT:    sub t6, s0, t1
-; RV64I-NEXT:    sra t6, a5, t6
+; RV64I-NEXT:    sub t5, s0, t1
+; RV64I-NEXT:    sra t5, a5, t5
 ; RV64I-NEXT:    bnez s0, .LBB20_20
 ; RV64I-NEXT:    j .LBB20_21
 ; RV64I-NEXT:  .LBB20_19:
-; RV64I-NEXT:    neg t6, s0
-; RV64I-NEXT:    srl s1, a3, s0
-; RV64I-NEXT:    sll t6, a5, t6
-; RV64I-NEXT:    or t6, s1, t6
+; RV64I-NEXT:    neg t5, s0
+; RV64I-NEXT:    srl t6, a3, s0
+; RV64I-NEXT:    sll t5, a5, t5
+; RV64I-NEXT:    or t5, t6, t5
 ; RV64I-NEXT:    beqz s0, .LBB20_21
 ; RV64I-NEXT:  .LBB20_20:
-; RV64I-NEXT:    mv a3, t6
+; RV64I-NEXT:    mv a3, t5
 ; RV64I-NEXT:  .LBB20_21:
 ; RV64I-NEXT:    bltu s0, t1, .LBB20_23
 ; RV64I-NEXT:  # %bb.22:
@@ -12000,7 +12000,7 @@ define void @ashr_32bytes_dwordOff(ptr %src.ptr, ptr %dwordOff.ptr, ptr %dst) no
 ; RV64I-NEXT:    bgeu a7, t0, .LBB20_25
 ; RV64I-NEXT:  .LBB20_24:
 ; RV64I-NEXT:    or a3, t2, t4
-; RV64I-NEXT:    or t1, t3, t5
+; RV64I-NEXT:    or t1, t3, s1
 ; RV64I-NEXT:  .LBB20_25:
 ; RV64I-NEXT:    bnez a7, .LBB20_29
 ; RV64I-NEXT:  # %bb.26:
diff --git a/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll b/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll
index b4e744808721fb..c9f12a405743c9 100644
--- a/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll
+++ b/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll
@@ -897,8 +897,8 @@ define i16 @atomicrmw_usub_sat_i16(ptr %ptr, i16 %val) {
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __atomic_load_2
 ; RV32I-NEXT:    mv a1, a0
-; RV32I-NEXT:    lui s2, 16
-; RV32I-NEXT:    addi s2, s2, -1
+; RV32I-NEXT:    lui a0, 16
+; RV32I-NEXT:    addi s2, a0, -1
 ; RV32I-NEXT:    and s1, s1, s2
 ; RV32I-NEXT:  .LBB5_1: # %atomicrmw.start
 ; RV32I-NEXT:    # =>This Inner Loop Header: Depth=1
@@ -984,8 +984,8 @@ define i16 @atomicrmw_usub_sat_i16(ptr %ptr, i16 %val) {
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __atomic_load_2
 ; RV64I-NEXT:    mv a1, a0
-; RV64I-NEXT:    lui s2, 16
-; RV64I-NEXT:    addi s2, s2, -1
+; RV64I-NEXT:    lui a0, 16
+; RV64I-NEXT:    addi s2, a0, -1
 ; RV64I-NEXT:    and s1, s1, s2
 ; RV64I-NEXT:  .LBB5_1: # %atomicrmw.start
 ; RV64I-NEXT:    # =>This Inner Loop Header: Depth=1
diff --git a/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll b/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll
index b2550986888319..36385b379f8ac9 100644
--- a/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll
+++ b/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll
@@ -199,8 +199,8 @@ define i16 @atomicrmw_uinc_wrap_i16(ptr %ptr, i16 %val) {
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __atomic_load_2
 ; RV32I-NEXT:    mv a1, a0
-; RV32I-NEXT:    lui s2, 16
-; RV32I-NEXT:    addi s2, s2, -1
+; RV32I-NEXT:    lui a0, 16
+; RV32I-NEXT:    addi s2, a0, -1
 ; RV32I-NEXT:    and s1, s1, s2
 ; RV32I-NEXT:  .LBB1_1: # %atomicrmw.start
 ; RV32I-NEXT:    # =>This Inner Loop Header: Depth=1
@@ -288,8 +288,8 @@ define i16 @atomicrmw_uinc_wrap_i16(ptr %ptr, i16 %val) {
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __atomic_load_2
 ; RV64I-NEXT:    mv a1, a0
-; RV64I-NEXT:    lui s2, 16
-; RV64I-NEXT:    addi s2, s2, -1
+; RV64I-NEXT:    lui a0, 16
+; RV64I-NEXT:    addi s2, a0, -1
 ; RV64I-NEXT:    and s1, s1, s2
 ; RV64I-NEXT:  .LBB1_1: # %atomicrmw.start
 ; RV64I-NEXT:    # =>This Inner Loop Header: Depth=1
diff --git a/llvm/test/CodeGen/RISCV/bfloat-convert.ll b/llvm/test/CodeGen/RISCV/bfloat-convert.ll
index b453334f349c02..b8911a6abc1cb4 100644
--- a/llvm/test/CodeGen/RISCV/bfloat-convert.ll
+++ b/llvm/test/CodeGen/RISCV/bfloat-convert.ll
@@ -470,11 +470,12 @@ define i64 @fcvt_l_bf16_sat(bfloat %a) nounwind {
 ; RV32IZFBFMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFBFMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IZFBFMIN-NEXT:    call __fixsfdi
+; RV32IZFBFMIN-NEXT:    mv a2, s0
+; RV32IZFBFMIN-NEXT:    lui a4, 524288
 ; RV32IZFBFMIN-NEXT:    lui a3, 524288
-; RV32IZFBFMIN-NEXT:    lui a2, 524288
 ; RV32IZFBFMIN-NEXT:    beqz s0, .LBB10_2
 ; RV32IZFBFMIN-NEXT:  # %bb.1: # %start
-; RV32IZFBFMIN-NEXT:    mv a2, a1
+; RV32IZFBFMIN-NEXT:    mv a3, a1
 ; RV32IZFBFMIN-NEXT:  .LBB10_2: # %start
 ; RV32IZFBFMIN-NEXT:    lui a1, 389120
 ; RV32IZFBFMIN-NEXT:    addi a1, a1, -1
@@ -482,16 +483,16 @@ define i64 @fcvt_l_bf16_sat(bfloat %a) nounwind {
 ; RV32IZFBFMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFBFMIN-NEXT:    beqz a1, .LBB10_4
 ; RV32IZFBFMIN-NEXT:  # %bb.3:
-; RV32IZFBFMIN-NEXT:    addi a2, a3, -1
+; RV32IZFBFMIN-NEXT:    addi a3, a4, -1
 ; RV32IZFBFMIN-NEXT:  .LBB10_4: # %start
-; RV32IZFBFMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFBFMIN-NEXT:    neg a4, s0
-; RV32IZFBFMIN-NEXT:    and a0, a4, a0
+; RV32IZFBFMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFBFMIN-NEXT:    neg a2, a2
+; RV32IZFBFMIN-NEXT:    and a0, a2, a0
 ; RV32IZFBFMIN-NEXT:    neg a1, a1
-; RV32IZFBFMIN-NEXT:    neg a3, a3
+; RV32IZFBFMIN-NEXT:    neg a2, a4
 ; RV32IZFBFMIN-NEXT:    or a0, a1, a0
-; RV32IZFBFMIN-NEXT:    and a1, a3, a2
-; RV32IZFBFMIN-NEXT:    and a0, a3, a0
+; RV32IZFBFMIN-NEXT:    and a1, a2, a3
+; RV32IZFBFMIN-NEXT:    and a0, a2, a0
 ; RV32IZFBFMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFBFMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFBFMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -510,11 +511,12 @@ define i64 @fcvt_l_bf16_sat(bfloat %a) nounwind {
 ; R32IDZFBFMIN-NEXT:    fle.s s0, fa5, fs0
 ; R32IDZFBFMIN-NEXT:    fmv.s fa0, fs0
 ; R32IDZFBFMIN-NEXT:    call __fixsfdi
+; R32IDZFBFMIN-NEXT:    mv a2, s0
+; R32IDZFBFMIN-NEXT:    lui a4, 524288
 ; R32IDZFBFMIN-NEXT:    lui a3, 524288
-; R32IDZFBFMIN-NEXT:    lui a2, 524288
 ; R32IDZFBFMIN-NEXT:    beqz s0, .LBB10_2
 ; R32IDZFBFMIN-NEXT:  # %bb.1: # %start
-; R32IDZFBFMIN-NEXT:    mv a2, a1
+; R32IDZFBFMIN-NEXT:    mv a3, a1
 ; R32IDZFBFMIN-NEXT:  .LBB10_2: # %start
 ; R32IDZFBFMIN-NEXT:    lui a1, 389120
 ; R32IDZFBFMIN-NEXT:    addi a1, a1, -1
@@ -522,16 +524,16 @@ define i64 @fcvt_l_bf16_sat(bfloat %a) nounwind {
 ; R32IDZFBFMIN-NEXT:    flt.s a1, fa5, fs0
 ; R32IDZFBFMIN-NEXT:    beqz a1, .LBB10_4
 ; R32IDZFBFMIN-NEXT:  # %bb.3:
-; R32IDZFBFMIN-NEXT:    addi a2, a3, -1
+; R32IDZFBFMIN-NEXT:    addi a3, a4, -1
 ; R32IDZFBFMIN-NEXT:  .LBB10_4: # %start
-; R32IDZFBFMIN-NEXT:    feq.s a3, fs0, fs0
-; R32IDZFBFMIN-NEXT:    neg a4, s0
-; R32IDZFBFMIN-NEXT:    and a0, a4, a0
+; R32IDZFBFMIN-NEXT:    feq.s a4, fs0, fs0
+; R32IDZFBFMIN-NEXT:    neg a2, a2
+; R32IDZFBFMIN-NEXT:    and a0, a2, a0
 ; R32IDZFBFMIN-NEXT:    neg a1, a1
-; R32IDZFBFMIN-NEXT:    neg a3, a3
+; R32IDZFBFMIN-NEXT:    neg a2, a4
 ; R32IDZFBFMIN-NEXT:    or a0, a1, a0
-; R32IDZFBFMIN-NEXT:    and a1, a3, a2
-; R32IDZFBFMIN-NEXT:    and a0, a3, a0
+; R32IDZFBFMIN-NEXT:    and a1, a2, a3
+; R32IDZFBFMIN-NEXT:    and a0, a2, a0
 ; R32IDZFBFMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; R32IDZFBFMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; R32IDZFBFMIN-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -552,11 +554,12 @@ define i64 @fcvt_l_bf16_sat(bfloat %a) nounwind {
 ; RV32ID-NEXT:    fle.s s0, fa5, fs0
 ; RV32ID-NEXT:    fmv.s fa0, fs0
 ; RV32ID-NEXT:    call __fixsfdi
+; RV32ID-NEXT:    mv a2, s0
+; RV32ID-NEXT:    lui a4, 524288
 ; RV32ID-NEXT:    lui a3, 524288
-; RV32ID-NEXT:    lui a2, 524288
 ; RV32ID-NEXT:    beqz s0, .LBB10_2
 ; RV32ID-NEXT:  # %bb.1: # %start
-; RV32ID-NEXT:    mv a2, a1
+; RV32ID-NEXT:    mv a3, a1
 ; RV32ID-NEXT:  .LBB10_2: # %start
 ; RV32ID-NEXT:    lui a1, 389120
 ; RV32ID-NEXT:    addi a1, a1, -1
@@ -564,16 +567,16 @@ define i64 @fcvt_l_bf16_sat(bfloat %a) nounwind {
 ; RV32ID-NEXT:    flt.s a1, fa5, fs0
 ; RV32ID-NEXT:    beqz a1, .LBB10_4
 ; RV32ID-NEXT:  # %bb.3:
-; RV32ID-NEXT:    addi a2, a3, -1
+; RV32ID-NEXT:    addi a3, a4, -1
 ; RV32ID-NEXT:  .LBB10_4: # %start
-; RV32ID-NEXT:    feq.s a3, fs0, fs0
-; RV32ID-NEXT:    neg a4, s0
+; RV32ID-NEXT:    feq.s a4, fs0, fs0
+; RV32ID-NEXT:    neg a2, a2
 ; RV32ID-NEXT:    neg a1, a1
-; RV32ID-NEXT:    and a0, a4, a0
-; RV32ID-NEXT:    neg a3, a3
+; RV32ID-NEXT:    and a0, a2, a0
+; RV32ID-NEXT:    neg a2, a4
 ; RV32ID-NEXT:    or a0, a1, a0
-; RV32ID-NEXT:    and a1, a3, a2
-; RV32ID-NEXT:    and a0, a3, a0
+; RV32ID-NEXT:    and a1, a2, a3
+; RV32ID-NEXT:    and a0, a2, a0
 ; RV32ID-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32ID-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32ID-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/bitint-fp-conv-200.ll b/llvm/test/CodeGen/RISCV/bitint-fp-conv-200.ll
index 780aebc1478958..a8e8b9479a31ff 100644
--- a/llvm/test/CodeGen/RISCV/bitint-fp-conv-200.ll
+++ b/llvm/test/CodeGen/RISCV/bitint-fp-conv-200.ll
@@ -305,9 +305,6 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV64-NEXT:    addi t1, t2, 128
 ; RV64-NEXT:  .LBB1_10: # %itofp-if-end
 ; RV64-NEXT:    addi sp, sp, -224
-; RV64-NEXT:    sd s0, 216(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s1, 208(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s2, 200(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    li a7, 200
 ; RV64-NEXT:    sub t0, a7, t1
 ; RV64-NEXT:    li a7, 199
@@ -316,6 +313,9 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV64-NEXT:    blt t0, t2, .LBB1_23
 ; RV64-NEXT:  # %bb.11: # %itofp-if-then4
 ; RV64-NEXT:    li t2, 26
+; RV64-NEXT:    sd a2, 24(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a7, 16(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    beq t0, t2, .LBB1_14
 ; RV64-NEXT:  # %bb.12: # %itofp-if-then4
 ; RV64-NEXT:    li t2, 25
@@ -323,79 +323,82 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV64-NEXT:  # %bb.13: # %itofp-sw-default
 ; RV64-NEXT:    li t2, 174
 ; RV64-NEXT:    sub t2, t2, t1
-; RV64-NEXT:    sd zero, 96(sp)
-; RV64-NEXT:    sd zero, 104(sp)
-; RV64-NEXT:    sd zero, 112(sp)
-; RV64-NEXT:    sd zero, 120(sp)
+; RV64-NEXT:    sd zero, 128(sp)
+; RV64-NEXT:    sd zero, 136(sp)
+; RV64-NEXT:    sd zero, 144(sp)
+; RV64-NEXT:    sd zero, 152(sp)
 ; RV64-NEXT:    srli t3, t2, 3
 ; RV64-NEXT:    zext.b t4, a5
 ; RV64-NEXT:    andi t3, t3, 24
-; RV64-NEXT:    addi t5, sp, 64
-; RV64-NEXT:    sd a6, 64(sp)
-; RV64-NEXT:    sd a3, 72(sp)
-; RV64-NEXT:    sd a4, 80(sp)
-; RV64-NEXT:    sd t4, 88(sp)
+; RV64-NEXT:    addi t5, sp, 96
+; RV64-NEXT:    sd a6, 96(sp)
+; RV64-NEXT:    sd a3, 104(sp)
+; RV64-NEXT:    sd a4, 112(sp)
+; RV64-NEXT:    sd t4, 120(sp)
 ; RV64-NEXT:    add t3, t5, t3
 ; RV64-NEXT:    ld t5, 16(t3)
 ; RV64-NEXT:    andi t6, t2, 63
 ; RV64-NEXT:    ld t4, 8(t3)
-; RV64-NEXT:    ld s0, 0(t3)
+; RV64-NEXT:    ld a2, 0(t3)
 ; RV64-NEXT:    xori t3, t6, 63
 ; RV64-NEXT:    slli t5, t5, 1
 ; RV64-NEXT:    addi t1, t1, 26
 ; RV64-NEXT:    srli t6, t1, 3
-; RV64-NEXT:    sd zero, 0(sp)
-; RV64-NEXT:    sd zero, 8(sp)
-; RV64-NEXT:    sd zero, 16(sp)
-; RV64-NEXT:    sd zero, 24(sp)
+; RV64-NEXT:    sd zero, 32(sp)
+; RV64-NEXT:    sd zero, 40(sp)
+; RV64-NEXT:    sd zero, 48(sp)
+; RV64-NEXT:    sd zero, 56(sp)
 ; RV64-NEXT:    andi t6, t6, 24
-; RV64-NEXT:    addi s1, sp, 32
-; RV64-NEXT:    sd a6, 32(sp)
-; RV64-NEXT:    sd a3, 40(sp)
-; RV64-NEXT:    sd a4, 48(sp)
-; RV64-NEXT:    sd a5, 56(sp)
-; RV64-NEXT:    sub a3, s1, t6
+; RV64-NEXT:    addi a7, sp, 64
+; RV64-NEXT:    sd a6, 64(sp)
+; RV64-NEXT:    sd a3, 72(sp)
+; RV64-NEXT:    sd a4, 80(sp)
+; RV64-NEXT:    sd a5, 88(sp)
+; RV64-NEXT:    sub a3, a7, t6
 ; RV64-NEXT:    ld a4, 0(a3)
 ; RV64-NEXT:    ld a5, 8(a3)
 ; RV64-NEXT:    srl a6, t4, t2
-; RV64-NEXT:    sll t5, t5, t3
-; RV64-NEXT:    or t5, a6, t5
+; RV64-NEXT:    sll a7, t5, t3
+; RV64-NEXT:    or a7, a6, a7
 ; RV64-NEXT:    ld a6, 16(a3)
-; RV64-NEXT:    srl t2, s0, t2
+; RV64-NEXT:    srl a2, a2, t2
 ; RV64-NEXT:    ld a3, 24(a3)
-; RV64-NEXT:    sll t6, a5, t1
-; RV64-NEXT:    andi s0, t1, 63
-; RV64-NEXT:    srli s1, a4, 1
-; RV64-NEXT:    xori s0, s0, 63
-; RV64-NEXT:    srl s1, s1, s0
-; RV64-NEXT:    srli s2, a6, 1
+; RV64-NEXT:    sll t2, a5, t1
+; RV64-NEXT:    andi t5, t1, 63
+; RV64-NEXT:    srli t6, a4, 1
+; RV64-NEXT:    xori t5, t5, 63
+; RV64-NEXT:    srl t6, t6, t5
+; RV64-NEXT:    srli a0, a6, 1
 ; RV64-NEXT:    sll a3, a3, t1
-; RV64-NEXT:    srl s2, s2, s0
-; RV64-NEXT:    or t6, t6, s1
-; RV64-NEXT:    or a3, a3, s2
+; RV64-NEXT:    srl a0, a0, t5
+; RV64-NEXT:    or t2, t2, t6
+; RV64-NEXT:    or a0, a3, a0
 ; RV64-NEXT:    slli t4, t4, 1
-; RV64-NEXT:    zext.b a3, a3
-; RV64-NEXT:    or a4, a6, a4
+; RV64-NEXT:    zext.b a0, a0
+; RV64-NEXT:    or a3, a6, a4
 ; RV64-NEXT:    srli a5, a5, 1
-; RV64-NEXT:    sll a4, a4, t1
-; RV64-NEXT:    srl a5, a5, s0
-; RV64-NEXT:    or a3, t6, a3
-; RV64-NEXT:    or a4, a4, a5
-; RV64-NEXT:    sll a5, t4, t3
-; RV64-NEXT:    or a3, a4, a3
-; RV64-NEXT:    or a4, t2, a5
-; RV64-NEXT:    snez a3, a3
-; RV64-NEXT:    or a6, a4, a3
-; RV64-NEXT:    mv a3, t5
+; RV64-NEXT:    sll a3, a3, t1
+; RV64-NEXT:    srl a4, a5, t5
+; RV64-NEXT:    or a0, t2, a0
+; RV64-NEXT:    or a3, a3, a4
+; RV64-NEXT:    sll a4, t4, t3
+; RV64-NEXT:    or a0, a3, a0
+; RV64-NEXT:    or a2, a2, a4
+; RV64-NEXT:    snez a0, a0
+; RV64-NEXT:    or a6, a2, a0
+; RV64-NEXT:    mv a3, a7
 ; RV64-NEXT:  .LBB1_14: # %itofp-sw-epilog
-; RV64-NEXT:    slli a4, a6, 61
-; RV64-NEXT:    srli a4, a4, 63
-; RV64-NEXT:    or a4, a6, a4
+; RV64-NEXT:    slli a0, a6, 61
+; RV64-NEXT:    srli a0, a0, 63
+; RV64-NEXT:    or a4, a6, a0
 ; RV64-NEXT:    addi a4, a4, 1
-; RV64-NEXT:    seqz a5, a4
-; RV64-NEXT:    slli a6, a4, 37
-; RV64-NEXT:    add a3, a3, a5
-; RV64-NEXT:    bltz a6, .LBB1_24
+; RV64-NEXT:    seqz a0, a4
+; RV64-NEXT:    slli a5, a4, 37
+; RV64-NEXT:    add a3, a3, a0
+; RV64-NEXT:    ld a2, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a7, 16(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    bltz a5, .LBB1_24
 ; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    srli a4, a4, 2
 ; RV64-NEXT:    slli a3, a3, 62
@@ -404,9 +407,6 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV64-NEXT:    mv t0, a7
 ; RV64-NEXT:  .LBB1_17: # %itofp-if-end26
 ; RV64-NEXT:    lui a4, 1046528
-; RV64-NEXT:    ld s0, 216(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s1, 208(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s2, 200(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    addi sp, sp, 224
 ; RV64-NEXT:    bgez a2, .LBB1_20
 ; RV64-NEXT:  # %bb.18: # %itofp-if-end26
@@ -436,14 +436,14 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV64-NEXT:    ret
 ; RV64-NEXT:  .LBB1_23: # %itofp-if-else
 ; RV64-NEXT:    addi a3, t1, -176
-; RV64-NEXT:    sd a6, 160(sp)
-; RV64-NEXT:    addi a4, sp, 160
+; RV64-NEXT:    sd a6, 192(sp)
+; RV64-NEXT:    addi a4, sp, 192
 ; RV64-NEXT:    srli a5, a3, 3
 ; RV64-NEXT:    andi a5, a5, 24
-; RV64-NEXT:    sd zero, 128(sp)
-; RV64-NEXT:    sd zero, 136(sp)
-; RV64-NEXT:    sd zero, 144(sp)
-; RV64-NEXT:    sd zero, 152(sp)
+; RV64-NEXT:    sd zero, 160(sp)
+; RV64-NEXT:    sd zero, 168(sp)
+; RV64-NEXT:    sd zero, 176(sp)
+; RV64-NEXT:    sd zero, 184(sp)
 ; RV64-NEXT:    sub a4, a4, a5
 ; RV64-NEXT:    ld a4, 0(a4)
 ; RV64-NEXT:    sll a3, a4, a3
@@ -464,17 +464,17 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV32:       # %bb.0: # %itofp-entry
 ; RV32-NEXT:    lw a4, 16(a0)
 ; RV32-NEXT:    lw a3, 20(a0)
-; RV32-NEXT:    lbu a2, 24(a0)
-; RV32-NEXT:    lw t0, 8(a0)
+; RV32-NEXT:    lbu a5, 24(a0)
+; RV32-NEXT:    lw a2, 8(a0)
 ; RV32-NEXT:    lw t1, 0(a0)
 ; RV32-NEXT:    lw t5, 4(a0)
 ; RV32-NEXT:    lw t2, 12(a0)
-; RV32-NEXT:    or a0, t0, a2
-; RV32-NEXT:    or a5, t1, a4
-; RV32-NEXT:    or a6, t5, a3
-; RV32-NEXT:    or a0, a5, a0
-; RV32-NEXT:    or a5, a6, t2
-; RV32-NEXT:    or a0, a0, a5
+; RV32-NEXT:    or a0, a2, a5
+; RV32-NEXT:    or a6, t1, a4
+; RV32-NEXT:    or a7, t5, a3
+; RV32-NEXT:    or a0, a6, a0
+; RV32-NEXT:    or a6, a7, t2
+; RV32-NEXT:    or a0, a0, a6
 ; RV32-NEXT:    beqz a0, .LBB1_37
 ; RV32-NEXT:  # %bb.1: # %itofp-if-end
 ; RV32-NEXT:    addi sp, sp, -224
@@ -484,63 +484,63 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV32-NEXT:    sw s3, 208(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s4, 204(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s5, 200(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 196(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 192(sp) # 4-byte Folded Spill
-; RV32-NEXT:    slli a2, a2, 24
-; RV32-NEXT:    srai a2, a2, 24
-; RV32-NEXT:    slli a0, a2, 24
+; RV32-NEXT:    slli a5, a5, 24
+; RV32-NEXT:    srai a5, a5, 24
+; RV32-NEXT:    slli a0, a5, 24
 ; RV32-NEXT:    srai s2, a0, 31
 ; RV32-NEXT:    srai a0, a0, 24
-; RV32-NEXT:    slli a6, s2, 25
+; RV32-NEXT:    slli a7, s2, 25
 ; RV32-NEXT:    srli a0, a0, 7
-; RV32-NEXT:    xor a5, t0, s2
-; RV32-NEXT:    or a0, a0, a6
-; RV32-NEXT:    sltu a6, a5, s2
+; RV32-NEXT:    xor a6, a2, s2
+; RV32-NEXT:    or a0, a0, a7
+; RV32-NEXT:    sltu t0, a6, s2
 ; RV32-NEXT:    xor a7, t2, s2
-; RV32-NEXT:    mv t6, a6
+; RV32-NEXT:    mv t6, t0
 ; RV32-NEXT:    beqz t2, .LBB1_3
 ; RV32-NEXT:  # %bb.2: # %itofp-if-end
 ; RV32-NEXT:    sltu t6, a7, s2
 ; RV32-NEXT:  .LBB1_3: # %itofp-if-end
 ; RV32-NEXT:    xor t4, t1, a0
 ; RV32-NEXT:    sltu t3, t4, a0
-; RV32-NEXT:    xor s3, t5, s2
+; RV32-NEXT:    xor s0, t5, s2
 ; RV32-NEXT:    mv t1, t3
 ; RV32-NEXT:    beqz t5, .LBB1_5
 ; RV32-NEXT:  # %bb.4: # %itofp-if-end
-; RV32-NEXT:    sltu t1, s3, s2
+; RV32-NEXT:    sltu t1, s0, s2
 ; RV32-NEXT:  .LBB1_5: # %itofp-if-end
-; RV32-NEXT:    or t2, t0, t2
-; RV32-NEXT:    mv t0, t1
-; RV32-NEXT:    beqz t2, .LBB1_7
+; RV32-NEXT:    or a2, a2, t2
+; RV32-NEXT:    mv t2, t1
+; RV32-NEXT:    beqz a2, .LBB1_7
 ; RV32-NEXT:  # %bb.6: # %itofp-if-end
-; RV32-NEXT:    mv t0, t6
+; RV32-NEXT:    mv t2, t6
 ; RV32-NEXT:  .LBB1_7: # %itofp-if-end
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    xor a4, a4, s2
-; RV32-NEXT:    xor s1, a3, s2
-; RV32-NEXT:    sltu s0, a4, s2
-; RV32-NEXT:    sub t2, s1, s2
+; RV32-NEXT:    xor s3, a3, s2
+; RV32-NEXT:    sltu s1, a4, s2
+; RV32-NEXT:    sub t5, s3, s2
 ; RV32-NEXT:    sub a4, a4, s2
-; RV32-NEXT:    sub t2, t2, s0
-; RV32-NEXT:    sltu t5, a4, t0
-; RV32-NEXT:    seqz t6, t2
-; RV32-NEXT:    and t6, t6, t5
+; RV32-NEXT:    sub t5, t5, s1
+; RV32-NEXT:    sltu t6, a4, t2
+; RV32-NEXT:    seqz s0, t5
+; RV32-NEXT:    and s0, s0, t6
 ; RV32-NEXT:    beqz a3, .LBB1_9
 ; RV32-NEXT:  # %bb.8: # %itofp-if-end
-; RV32-NEXT:    sltu s0, s1, s2
+; RV32-NEXT:    sltu s1, s3, s2
 ; RV32-NEXT:  .LBB1_9: # %itofp-if-end
-; RV32-NEXT:    xor a3, a2, s2
-; RV32-NEXT:    add s0, s2, s0
-; RV32-NEXT:    sub s1, a7, s2
-; RV32-NEXT:    sub a3, a3, s0
-; RV32-NEXT:    sub s0, a5, s2
-; RV32-NEXT:    sub a7, a3, t6
-; RV32-NEXT:    sub a3, s1, a6
-; RV32-NEXT:    sltu a5, s0, t1
+; RV32-NEXT:    sw a5, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    xor a3, a5, s2
+; RV32-NEXT:    add s1, s2, s1
+; RV32-NEXT:    sub a5, a7, s2
+; RV32-NEXT:    sub a3, a3, s1
+; RV32-NEXT:    sub a6, a6, s2
+; RV32-NEXT:    sub a7, a3, s0
+; RV32-NEXT:    sub a3, a5, t0
+; RV32-NEXT:    sltu a5, a6, t1
 ; RV32-NEXT:    sub a5, a3, a5
-; RV32-NEXT:    sub a3, t2, t5
-; RV32-NEXT:    sub a4, a4, t0
-; RV32-NEXT:    sub a6, s0, t1
+; RV32-NEXT:    sub a3, t5, t6
+; RV32-NEXT:    sub a4, a4, t2
+; RV32-NEXT:    sub a6, a6, t1
 ; RV32-NEXT:    srli t0, a4, 8
 ; RV32-NEXT:    slli t1, a3, 24
 ; RV32-NEXT:    or t1, t0, t1
@@ -548,109 +548,183 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV32-NEXT:    addi t5, t5, 1365
 ; RV32-NEXT:    lui t0, 209715
 ; RV32-NEXT:    srli t2, a3, 8
-; RV32-NEXT:    slli t6, a7, 24
-; RV32-NEXT:    lui s1, 61681
-; RV32-NEXT:    or t2, t6, t2
+; RV32-NEXT:    sw a7, 0(sp) # 4-byte Folded Spill
+; RV32-NEXT:    slli a7, a7, 24
+; RV32-NEXT:    lui t6, 61681
+; RV32-NEXT:    or t2, a7, t2
 ; RV32-NEXT:    addi s0, t0, 819
-; RV32-NEXT:    addi t6, s1, -241
+; RV32-NEXT:    addi t6, t6, -241
 ; RV32-NEXT:    bnez t2, .LBB1_11
 ; RV32-NEXT:  # %bb.10: # %itofp-if-end
-; RV32-NEXT:    srli t0, t1, 1
-; RV32-NEXT:    or t0, t1, t0
-; RV32-NEXT:    srli s1, t0, 2
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 4
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 8
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 16
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    not t0, t0
-; RV32-NEXT:    srli s1, t0, 1
-; RV32-NEXT:    and s1, s1, t5
-; RV32-NEXT:    sub t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 2
+; RV32-NEXT:    srli a7, t1, 1
+; RV32-NEXT:    or a7, t1, a7
+; RV32-NEXT:    srli t0, a7, 2
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 4
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 8
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 16
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    not a7, a7
+; RV32-NEXT:    srli t0, a7, 1
+; RV32-NEXT:    and t0, t0, t5
+; RV32-NEXT:    sub a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 2
+; RV32-NEXT:    and a7, a7, s0
 ; RV32-NEXT:    and t0, t0, s0
-; RV32-NEXT:    and s1, s1, s0
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 4
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    and t0, t0, t6
-; RV32-NEXT:    slli s1, t0, 8
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    slli s1, t0, 16
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    srli t0, t0, 24
-; RV32-NEXT:    addi t0, t0, 32
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 4
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    and a7, a7, t6
+; RV32-NEXT:    slli t0, a7, 8
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    slli t0, a7, 16
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    srli a7, a7, 24
+; RV32-NEXT:    addi t0, a7, 32
 ; RV32-NEXT:    j .LBB1_12
 ; RV32-NEXT:  .LBB1_11:
-; RV32-NEXT:    srli t0, t2, 1
-; RV32-NEXT:    or t0, t2, t0
-; RV32-NEXT:    srli s1, t0, 2
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 4
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 8
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 16
-; RV32-NEXT:    or t0, t0, s1
-; RV32-NEXT:    not t0, t0
-; RV32-NEXT:    srli s1, t0, 1
-; RV32-NEXT:    and s1, s1, t5
-; RV32-NEXT:    sub t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 2
+; RV32-NEXT:    srli a7, t2, 1
+; RV32-NEXT:    or a7, t2, a7
+; RV32-NEXT:    srli t0, a7, 2
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 4
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 8
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 16
+; RV32-NEXT:    or a7, a7, t0
+; RV32-NEXT:    not a7, a7
+; RV32-NEXT:    srli t0, a7, 1
+; RV32-NEXT:    and t0, t0, t5
+; RV32-NEXT:    sub a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 2
+; RV32-NEXT:    and a7, a7, s0
 ; RV32-NEXT:    and t0, t0, s0
-; RV32-NEXT:    and s1, s1, s0
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    srli s1, t0, 4
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    and t0, t0, t6
-; RV32-NEXT:    slli s1, t0, 8
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    slli s1, t0, 16
-; RV32-NEXT:    add t0, t0, s1
-; RV32-NEXT:    srli t0, t0, 24
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 4
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    and a7, a7, t6
+; RV32-NEXT:    slli t0, a7, 8
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    slli t0, a7, 16
+; RV32-NEXT:    add a7, a7, t0
+; RV32-NEXT:    srli t0, a7, 24
 ; RV32-NEXT:  .LBB1_12: # %itofp-if-end
-; RV32-NEXT:    srli s1, a5, 8
-; RV32-NEXT:    slli s4, a4, 24
-; RV32-NEXT:    or s1, s4, s1
-; RV32-NEXT:    slli s4, a5, 24
-; RV32-NEXT:    srli s5, a6, 8
-; RV32-NEXT:    sub s3, s3, s2
-; RV32-NEXT:    or s2, s5, s4
+; RV32-NEXT:    srli a7, a5, 8
+; RV32-NEXT:    slli s1, a4, 24
+; RV32-NEXT:    or s1, s1, a7
+; RV32-NEXT:    slli a7, a5, 24
+; RV32-NEXT:    srli s3, a6, 8
+; RV32-NEXT:    sub a2, a2, s2
+; RV32-NEXT:    or s2, s3, a7
 ; RV32-NEXT:    bnez s1, .LBB1_14
 ; RV32-NEXT:  # %bb.13: # %itofp-if-end
-; RV32-NEXT:    srli s4, s2, 1
-; RV32-NEXT:    or s4, s2, s4
-; RV32-NEXT:    srli s5, s4, 2
-; RV32-NEXT:    or s4, s4, s5
-; RV32-NEXT:    srli s5, s4, 4
-; RV32-NEXT:    or s4, s4, s5
-; RV32-NEXT:    srli s5, s4, 8
-; RV32-NEXT:    or s4, s4, s5
-; RV32-NEXT:    srli s5, s4, 16
-; RV32-NEXT:    or s4, s4, s5
-; RV32-NEXT:    not s4, s4
-; RV32-NEXT:    srli s5, s4, 1
-; RV32-NEXT:    and s5, s5, t5
-; RV32-NEXT:    sub s4, s4, s5
-; RV32-NEXT:    srli s5, s4, 2
-; RV32-NEXT:    and s4, s4, s0
-; RV32-NEXT:    and s5, s5, s0
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    srli s5, s4, 4
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    and s4, s4, t6
-; RV32-NEXT:    slli s5, s4, 8
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    slli s5, s4, 16
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    srli s4, s4, 24
-; RV32-NEXT:    addi s4, s4, 32
+; RV32-NEXT:    srli a7, s2, 1
+; RV32-NEXT:    or a7, s2, a7
+; RV32-NEXT:    srli s3, a7, 2
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 4
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 8
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 16
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    not a7, a7
+; RV32-NEXT:    srli s3, a7, 1
+; RV32-NEXT:    and s3, s3, t5
+; RV32-NEXT:    sub a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 2
+; RV32-NEXT:    and a7, a7, s0
+; RV32-NEXT:    and s3, s3, s0
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 4
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    and a7, a7, t6
+; RV32-NEXT:    slli s3, a7, 8
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    slli s3, a7, 16
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    srli a7, a7, 24
+; RV32-NEXT:    addi a7, a7, 32
 ; RV32-NEXT:    j .LBB1_15
 ; RV32-NEXT:  .LBB1_14:
-; RV32-NEXT:    srli s4, s1, 1
-; RV32-NEXT:    or s4, s1, s4
+; RV32-NEXT:    srli a7, s1, 1
+; RV32-NEXT:    or a7, s1, a7
+; RV32-NEXT:    srli s3, a7, 2
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 4
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 8
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 16
+; RV32-NEXT:    or a7, a7, s3
+; RV32-NEXT:    not a7, a7
+; RV32-NEXT:    srli s3, a7, 1
+; RV32-NEXT:    and s3, s3, t5
+; RV32-NEXT:    sub a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 2
+; RV32-NEXT:    and a7, a7, s0
+; RV32-NEXT:    and s3, s3, s0
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    srli s3, a7, 4
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    and a7, a7, t6
+; RV32-NEXT:    slli s3, a7, 8
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    slli s3, a7, 16
+; RV32-NEXT:    add a7, a7, s3
+; RV32-NEXT:    srli a7, a7, 24
+; RV32-NEXT:  .LBB1_15: # %itofp-if-end
+; RV32-NEXT:    or s3, t1, t2
+; RV32-NEXT:    sub t3, a2, t3
+; RV32-NEXT:    sub t4, t4, a0
+; RV32-NEXT:    bnez s3, .LBB1_17
+; RV32-NEXT:  # %bb.16: # %itofp-if-end
+; RV32-NEXT:    addi t0, a7, 64
+; RV32-NEXT:  .LBB1_17: # %itofp-if-end
+; RV32-NEXT:    mv a7, a0
+; RV32-NEXT:    srli a0, t4, 8
+; RV32-NEXT:    srli a2, t3, 8
+; RV32-NEXT:    slli s3, a6, 24
+; RV32-NEXT:    or a2, s3, a2
+; RV32-NEXT:    slli s3, t3, 24
+; RV32-NEXT:    or s3, a0, s3
+; RV32-NEXT:    bnez a2, .LBB1_20
+; RV32-NEXT:  # %bb.18: # %itofp-if-end
+; RV32-NEXT:    srli a0, s3, 1
+; RV32-NEXT:    or a0, s3, a0
+; RV32-NEXT:    srli s4, a0, 2
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 4
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 8
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 16
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    not a0, a0
+; RV32-NEXT:    srli s4, a0, 1
+; RV32-NEXT:    and s4, s4, t5
+; RV32-NEXT:    sub a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 2
+; RV32-NEXT:    and a0, a0, s0
+; RV32-NEXT:    and s4, s4, s0
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 4
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    and a0, a0, t6
+; RV32-NEXT:    slli s4, a0, 8
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    slli s4, a0, 16
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    srli a0, a0, 24
+; RV32-NEXT:    addi a0, a0, 32
+; RV32-NEXT:    slli s4, t4, 24
+; RV32-NEXT:    beqz s4, .LBB1_21
+; RV32-NEXT:  .LBB1_19:
+; RV32-NEXT:    srli s5, s4, 1
+; RV32-NEXT:    or s4, s4, s5
 ; RV32-NEXT:    srli s5, s4, 2
 ; RV32-NEXT:    or s4, s4, s5
 ; RV32-NEXT:    srli s5, s4, 4
@@ -661,83 +735,11 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV32-NEXT:    or s4, s4, s5
 ; RV32-NEXT:    not s4, s4
 ; RV32-NEXT:    srli s5, s4, 1
-; RV32-NEXT:    and s5, s5, t5
-; RV32-NEXT:    sub s4, s4, s5
-; RV32-NEXT:    srli s5, s4, 2
-; RV32-NEXT:    and s4, s4, s0
-; RV32-NEXT:    and s5, s5, s0
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    srli s5, s4, 4
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    and s4, s4, t6
-; RV32-NEXT:    slli s5, s4, 8
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    slli s5, s4, 16
-; RV32-NEXT:    add s4, s4, s5
-; RV32-NEXT:    srli s4, s4, 24
-; RV32-NEXT:  .LBB1_15: # %itofp-if-end
-; RV32-NEXT:    or s5, t1, t2
-; RV32-NEXT:    sub t3, s3, t3
-; RV32-NEXT:    sub t4, t4, a0
-; RV32-NEXT:    bnez s5, .LBB1_17
-; RV32-NEXT:  # %bb.16: # %itofp-if-end
-; RV32-NEXT:    addi t0, s4, 64
-; RV32-NEXT:  .LBB1_17: # %itofp-if-end
-; RV32-NEXT:    srli s4, t4, 8
-; RV32-NEXT:    srli s3, t3, 8
-; RV32-NEXT:    slli s5, a6, 24
-; RV32-NEXT:    or s3, s5, s3
-; RV32-NEXT:    slli s5, t3, 24
-; RV32-NEXT:    or s4, s4, s5
-; RV32-NEXT:    bnez s3, .LBB1_20
-; RV32-NEXT:  # %bb.18: # %itofp-if-end
-; RV32-NEXT:    srli s5, s4, 1
-; RV32-NEXT:    or s5, s4, s5
-; RV32-NEXT:    srli s6, s5, 2
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 4
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 8
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 16
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    not s5, s5
-; RV32-NEXT:    srli s6, s5, 1
-; RV32-NEXT:    and s6, s6, t5
-; RV32-NEXT:    sub s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 2
-; RV32-NEXT:    and s5, s5, s0
-; RV32-NEXT:    and s6, s6, s0
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 4
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    and s5, s5, t6
-; RV32-NEXT:    slli s6, s5, 8
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    slli s6, s5, 16
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    srli s5, s5, 24
-; RV32-NEXT:    addi s5, s5, 32
-; RV32-NEXT:    slli s6, t4, 24
-; RV32-NEXT:    beqz s6, .LBB1_21
-; RV32-NEXT:  .LBB1_19:
-; RV32-NEXT:    srli s7, s6, 1
-; RV32-NEXT:    or s6, s6, s7
-; RV32-NEXT:    srli s7, s6, 2
-; RV32-NEXT:    or s6, s6, s7
-; RV32-NEXT:    srli s7, s6, 4
-; RV32-NEXT:    or s6, s6, s7
-; RV32-NEXT:    srli s7, s6, 8
-; RV32-NEXT:    or s6, s6, s7
-; RV32-NEXT:    srli s7, s6, 16
-; RV32-NEXT:    or s6, s6, s7
-; RV32-NEXT:    not s6, s6
-; RV32-NEXT:    srli s7, s6, 1
-; RV32-NEXT:    and t5, s7, t5
-; RV32-NEXT:    sub t5, s6, t5
-; RV32-NEXT:    srli s6, t5, 2
+; RV32-NEXT:    and t5, s5, t5
+; RV32-NEXT:    sub t5, s4, t5
+; RV32-NEXT:    srli s4, t5, 2
 ; RV32-NEXT:    and t5, t5, s0
-; RV32-NEXT:    and s0, s6, s0
+; RV32-NEXT:    and s0, s4, s0
 ; RV32-NEXT:    add t5, t5, s0
 ; RV32-NEXT:    srli s0, t5, 4
 ; RV32-NEXT:    add t5, t5, s0
@@ -746,172 +748,174 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV32-NEXT:    add t5, t5, t6
 ; RV32-NEXT:    slli t6, t5, 16
 ; RV32-NEXT:    add t5, t5, t6
-; RV32-NEXT:    srli t5, t5, 24
-; RV32-NEXT:    or t6, s4, s3
-; RV32-NEXT:    beqz t6, .LBB1_22
-; RV32-NEXT:    j .LBB1_23
+; RV32-NEXT:    srli t6, t5, 24
+; RV32-NEXT:    j .LBB1_22
 ; RV32-NEXT:  .LBB1_20:
-; RV32-NEXT:    srli s5, s3, 1
-; RV32-NEXT:    or s5, s3, s5
-; RV32-NEXT:    srli s6, s5, 2
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 4
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 8
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 16
-; RV32-NEXT:    or s5, s5, s6
-; RV32-NEXT:    not s5, s5
-; RV32-NEXT:    srli s6, s5, 1
-; RV32-NEXT:    and s6, s6, t5
-; RV32-NEXT:    sub s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 2
-; RV32-NEXT:    and s5, s5, s0
-; RV32-NEXT:    and s6, s6, s0
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    srli s6, s5, 4
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    and s5, s5, t6
-; RV32-NEXT:    slli s6, s5, 8
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    slli s6, s5, 16
-; RV32-NEXT:    add s5, s5, s6
-; RV32-NEXT:    srli s5, s5, 24
-; RV32-NEXT:    slli s6, t4, 24
-; RV32-NEXT:    bnez s6, .LBB1_19
+; RV32-NEXT:    srli a0, a2, 1
+; RV32-NEXT:    or a0, a2, a0
+; RV32-NEXT:    srli s4, a0, 2
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 4
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 8
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 16
+; RV32-NEXT:    or a0, a0, s4
+; RV32-NEXT:    not a0, a0
+; RV32-NEXT:    srli s4, a0, 1
+; RV32-NEXT:    and s4, s4, t5
+; RV32-NEXT:    sub a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 2
+; RV32-NEXT:    and a0, a0, s0
+; RV32-NEXT:    and s4, s4, s0
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    srli s4, a0, 4
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    and a0, a0, t6
+; RV32-NEXT:    slli s4, a0, 8
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    slli s4, a0, 16
+; RV32-NEXT:    add a0, a0, s4
+; RV32-NEXT:    srli a0, a0, 24
+; RV32-NEXT:    slli s4, t4, 24
+; RV32-NEXT:    bnez s4, .LBB1_19
 ; RV32-NEXT:  .LBB1_21: # %itofp-if-end
-; RV32-NEXT:    li t5, 64
-; RV32-NEXT:    or t6, s4, s3
-; RV32-NEXT:    bnez t6, .LBB1_23
+; RV32-NEXT:    li t6, 64
 ; RV32-NEXT:  .LBB1_22: # %itofp-if-end
-; RV32-NEXT:    addi s5, t5, 64
-; RV32-NEXT:  .LBB1_23: # %itofp-if-end
-; RV32-NEXT:    or t2, s1, t2
-; RV32-NEXT:    or t1, s2, t1
-; RV32-NEXT:    or t1, t1, t2
-; RV32-NEXT:    bnez t1, .LBB1_25
-; RV32-NEXT:  # %bb.24: # %itofp-if-end
-; RV32-NEXT:    addi t0, s5, 128
-; RV32-NEXT:  .LBB1_25: # %itofp-if-end
-; RV32-NEXT:    li t1, 200
-; RV32-NEXT:    sub t2, t1, t0
-; RV32-NEXT:    li t1, 199
+; RV32-NEXT:    or a2, s3, a2
+; RV32-NEXT:    mv t5, a0
+; RV32-NEXT:    bnez a2, .LBB1_24
+; RV32-NEXT:  # %bb.23: # %itofp-if-end
+; RV32-NEXT:    addi t5, t6, 64
+; RV32-NEXT:  .LBB1_24: # %itofp-if-end
+; RV32-NEXT:    or a0, s1, t2
+; RV32-NEXT:    or a2, s2, t1
+; RV32-NEXT:    or t1, a2, a0
+; RV32-NEXT:    mv a0, a7
+; RV32-NEXT:    lw a2, 0(sp) # 4-byte Folded Reload
+; RV32-NEXT:    bnez t1, .LBB1_26
+; RV32-NEXT:  # %bb.25: # %itofp-if-end
+; RV32-NEXT:    addi t0, t5, 128
+; RV32-NEXT:  .LBB1_26: # %itofp-if-end
+; RV32-NEXT:    li a7, 200
+; RV32-NEXT:    sub t2, a7, t0
+; RV32-NEXT:    li a7, 199
 ; RV32-NEXT:    li t5, 25
-; RV32-NEXT:    sub t1, t1, t0
+; RV32-NEXT:    sub t1, a7, t0
 ; RV32-NEXT:    blt t2, t5, .LBB1_38
-; RV32-NEXT:  # %bb.26: # %itofp-if-then4
-; RV32-NEXT:    li t5, 26
-; RV32-NEXT:    beq t2, t5, .LBB1_29
 ; RV32-NEXT:  # %bb.27: # %itofp-if-then4
-; RV32-NEXT:    li t5, 25
-; RV32-NEXT:    beq t2, t5, .LBB1_40
-; RV32-NEXT:  # %bb.28: # %itofp-sw-default
-; RV32-NEXT:    li t5, 174
-; RV32-NEXT:    zext.b t6, a7
-; RV32-NEXT:    sub a7, t5, t0
-; RV32-NEXT:    sw zero, 112(sp)
-; RV32-NEXT:    sw zero, 116(sp)
+; RV32-NEXT:    li a7, 26
+; RV32-NEXT:    beq t2, a7, .LBB1_30
+; RV32-NEXT:  # %bb.28: # %itofp-if-then4
+; RV32-NEXT:    li a7, 25
+; RV32-NEXT:    beq t2, a7, .LBB1_40
+; RV32-NEXT:  # %bb.29: # %itofp-sw-default
+; RV32-NEXT:    li a7, 174
+; RV32-NEXT:    zext.b t5, a2
+; RV32-NEXT:    sub a2, a7, t0
 ; RV32-NEXT:    sw zero, 120(sp)
 ; RV32-NEXT:    sw zero, 124(sp)
-; RV32-NEXT:    sw zero, 96(sp)
-; RV32-NEXT:    sw zero, 100(sp)
+; RV32-NEXT:    sw zero, 128(sp)
+; RV32-NEXT:    sw zero, 132(sp)
 ; RV32-NEXT:    sw zero, 104(sp)
 ; RV32-NEXT:    sw zero, 108(sp)
-; RV32-NEXT:    srli t5, a7, 3
-; RV32-NEXT:    sw a4, 80(sp)
-; RV32-NEXT:    sw a3, 84(sp)
-; RV32-NEXT:    sw t6, 88(sp)
-; RV32-NEXT:    sw zero, 92(sp)
-; RV32-NEXT:    andi t5, t5, 28
-; RV32-NEXT:    addi s0, sp, 64
-; RV32-NEXT:    sw t4, 64(sp)
-; RV32-NEXT:    sw t3, 68(sp)
-; RV32-NEXT:    sw a6, 72(sp)
-; RV32-NEXT:    sw a5, 76(sp)
-; RV32-NEXT:    add t5, s0, t5
-; RV32-NEXT:    lw s0, 4(t5)
-; RV32-NEXT:    lw s1, 8(t5)
-; RV32-NEXT:    andi s2, a7, 31
-; RV32-NEXT:    lw s3, 0(t5)
-; RV32-NEXT:    xori t5, s2, 31
-; RV32-NEXT:    srl s2, s0, a7
-; RV32-NEXT:    slli s1, s1, 1
+; RV32-NEXT:    sw zero, 112(sp)
+; RV32-NEXT:    sw zero, 116(sp)
+; RV32-NEXT:    srli a7, a2, 3
+; RV32-NEXT:    sw a4, 88(sp)
+; RV32-NEXT:    sw a3, 92(sp)
+; RV32-NEXT:    sw t5, 96(sp)
+; RV32-NEXT:    sw zero, 100(sp)
+; RV32-NEXT:    andi a7, a7, 28
+; RV32-NEXT:    addi t6, sp, 72
+; RV32-NEXT:    sw t4, 72(sp)
+; RV32-NEXT:    sw t3, 76(sp)
+; RV32-NEXT:    sw a6, 80(sp)
+; RV32-NEXT:    sw a5, 84(sp)
+; RV32-NEXT:    add a7, t6, a7
+; RV32-NEXT:    lw t6, 4(a7)
+; RV32-NEXT:    lw s0, 8(a7)
+; RV32-NEXT:    andi s1, a2, 31
+; RV32-NEXT:    lw s2, 0(a7)
+; RV32-NEXT:    xori a7, s1, 31
+; RV32-NEXT:    srl s1, t6, a2
+; RV32-NEXT:    slli s0, s0, 1
 ; RV32-NEXT:    addi t0, t0, 26
-; RV32-NEXT:    sw zero, 16(sp)
-; RV32-NEXT:    sw zero, 20(sp)
 ; RV32-NEXT:    sw zero, 24(sp)
 ; RV32-NEXT:    sw zero, 28(sp)
-; RV32-NEXT:    sw zero, 0(sp)
-; RV32-NEXT:    sw zero, 4(sp)
+; RV32-NEXT:    sw zero, 32(sp)
+; RV32-NEXT:    sw zero, 36(sp)
 ; RV32-NEXT:    sw zero, 8(sp)
 ; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    sw t4, 32(sp)
-; RV32-NEXT:    sw t3, 36(sp)
-; RV32-NEXT:    sw a6, 40(sp)
-; RV32-NEXT:    sw a5, 44(sp)
+; RV32-NEXT:    sw zero, 16(sp)
+; RV32-NEXT:    sw zero, 20(sp)
+; RV32-NEXT:    sw t4, 40(sp)
+; RV32-NEXT:    sw t3, 44(sp)
+; RV32-NEXT:    sw a6, 48(sp)
+; RV32-NEXT:    sw a5, 52(sp)
 ; RV32-NEXT:    srli a5, t0, 3
 ; RV32-NEXT:    andi a5, a5, 28
-; RV32-NEXT:    addi a6, sp, 32
-; RV32-NEXT:    sw a4, 48(sp)
-; RV32-NEXT:    sw a3, 52(sp)
-; RV32-NEXT:    sw t6, 56(sp)
+; RV32-NEXT:    addi a6, sp, 40
+; RV32-NEXT:    sw a4, 56(sp)
+; RV32-NEXT:    sw a3, 60(sp)
+; RV32-NEXT:    sw t5, 64(sp)
 ; RV32-NEXT:    sub a3, a6, a5
-; RV32-NEXT:    sll a4, s1, t5
+; RV32-NEXT:    sll a4, s0, a7
 ; RV32-NEXT:    lw a5, 16(a3)
 ; RV32-NEXT:    lw a6, 20(a3)
 ; RV32-NEXT:    lw t3, 4(a3)
 ; RV32-NEXT:    lw t4, 0(a3)
-; RV32-NEXT:    lw t6, 12(a3)
-; RV32-NEXT:    lw s1, 8(a3)
-; RV32-NEXT:    or a4, s2, a4
-; RV32-NEXT:    srl a7, s3, a7
-; RV32-NEXT:    slli s0, s0, 1
+; RV32-NEXT:    lw t5, 12(a3)
+; RV32-NEXT:    lw s0, 8(a3)
+; RV32-NEXT:    or a4, s1, a4
+; RV32-NEXT:    srl a2, s2, a2
+; RV32-NEXT:    slli t6, t6, 1
 ; RV32-NEXT:    lw a3, 24(a3)
-; RV32-NEXT:    or s2, t3, a6
-; RV32-NEXT:    or s2, s2, t6
+; RV32-NEXT:    or s1, t3, a6
+; RV32-NEXT:    or s1, s1, t5
 ; RV32-NEXT:    or a5, a5, t4
-; RV32-NEXT:    or t4, a5, s1
-; RV32-NEXT:    andi s3, t0, 31
+; RV32-NEXT:    or t4, a5, s0
+; RV32-NEXT:    andi s2, t0, 31
 ; RV32-NEXT:    srli t4, t4, 1
-; RV32-NEXT:    xori s3, s3, 31
-; RV32-NEXT:    sll s2, s2, t0
-; RV32-NEXT:    srl t4, t4, s3
+; RV32-NEXT:    xori s2, s2, 31
 ; RV32-NEXT:    sll s1, s1, t0
+; RV32-NEXT:    srl t4, t4, s2
+; RV32-NEXT:    sll s0, s0, t0
 ; RV32-NEXT:    srli t3, t3, 1
-; RV32-NEXT:    srl t3, t3, s3
+; RV32-NEXT:    srl t3, t3, s2
 ; RV32-NEXT:    srli a6, a6, 1
 ; RV32-NEXT:    sll a3, a3, t0
-; RV32-NEXT:    srl a6, a6, s3
-; RV32-NEXT:    or t3, s1, t3
+; RV32-NEXT:    srl a6, a6, s2
+; RV32-NEXT:    or t3, s0, t3
 ; RV32-NEXT:    or a3, a3, a6
 ; RV32-NEXT:    zext.b a3, a3
-; RV32-NEXT:    srli a6, t6, 1
+; RV32-NEXT:    srli a6, t5, 1
 ; RV32-NEXT:    sll a5, a5, t0
-; RV32-NEXT:    srl a6, a6, s3
+; RV32-NEXT:    srl a6, a6, s2
 ; RV32-NEXT:    or a3, t3, a3
 ; RV32-NEXT:    or a5, a5, a6
-; RV32-NEXT:    or a6, t4, s2
+; RV32-NEXT:    or a6, t4, s1
 ; RV32-NEXT:    or a3, a5, a3
-; RV32-NEXT:    sll a5, s0, t5
+; RV32-NEXT:    sll a5, t6, a7
 ; RV32-NEXT:    or a3, a3, a6
-; RV32-NEXT:    or a5, a7, a5
+; RV32-NEXT:    or a2, a2, a5
 ; RV32-NEXT:    snez a3, a3
-; RV32-NEXT:    or t4, a5, a3
+; RV32-NEXT:    or t4, a2, a3
 ; RV32-NEXT:    mv t3, a4
-; RV32-NEXT:  .LBB1_29: # %itofp-sw-epilog
-; RV32-NEXT:    slli a3, t4, 29
-; RV32-NEXT:    srli a3, a3, 31
-; RV32-NEXT:    or a3, t4, a3
-; RV32-NEXT:    addi a3, a3, 1
-; RV32-NEXT:    seqz a4, a3
-; RV32-NEXT:    slli a5, a3, 5
-; RV32-NEXT:    add t3, t3, a4
-; RV32-NEXT:    bltz a5, .LBB1_39
-; RV32-NEXT:  # %bb.30:
-; RV32-NEXT:    srli a3, a3, 2
+; RV32-NEXT:  .LBB1_30: # %itofp-sw-epilog
+; RV32-NEXT:    slli a2, t4, 29
+; RV32-NEXT:    srli a2, a2, 31
+; RV32-NEXT:    or a2, t4, a2
+; RV32-NEXT:    addi a2, a2, 1
+; RV32-NEXT:    seqz a3, a2
+; RV32-NEXT:    slli a4, a2, 5
+; RV32-NEXT:    add t3, t3, a3
+; RV32-NEXT:    lw a3, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    bltz a4, .LBB1_39
+; RV32-NEXT:  # %bb.31:
+; RV32-NEXT:    srli a2, a2, 2
 ; RV32-NEXT:    slli t3, t3, 30
-; RV32-NEXT:    or a3, a3, t3
-; RV32-NEXT:  .LBB1_31: # %itofp-if-end26
+; RV32-NEXT:    or a2, a2, t3
 ; RV32-NEXT:    mv t2, t1
 ; RV32-NEXT:  .LBB1_32: # %itofp-if-end26
 ; RV32-NEXT:    lui a4, 1046528
@@ -921,64 +925,64 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 ; RV32-NEXT:    lw s3, 208(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s4, 204(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s5, 200(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 196(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s7, 192(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    addi sp, sp, 224
-; RV32-NEXT:    bgez a2, .LBB1_35
+; RV32-NEXT:    bgez a3, .LBB1_35
 ; RV32-NEXT:  # %bb.33: # %itofp-if-end26
-; RV32-NEXT:    li a2, 127
-; RV32-NEXT:    bgeu a2, t1, .LBB1_36
+; RV32-NEXT:    li a3, 127
+; RV32-NEXT:    bgeu a3, t1, .LBB1_36
 ; RV32-NEXT:  .LBB1_34: # %itofp-return
 ; RV32-NEXT:    sw a4, 0(a1)
 ; RV32-NEXT:    ret
 ; RV32-NEXT:  .LBB1_35: # %itofp-if-end26
 ; RV32-NEXT:    lui a4, 522240
-; RV32-NEXT:    li a2, 127
-; RV32-NEXT:    bltu a2, t1, .LBB1_34
+; RV32-NEXT:    li a3, 127
+; RV32-NEXT:    bltu a3, t1, .LBB1_34
 ; RV32-NEXT:  .LBB1_36: # %itofp-if-end26
-; RV32-NEXT:    lui a2, 524288
-; RV32-NEXT:    slli a3, a3, 9
+; RV32-NEXT:    lui a3, 524288
+; RV32-NEXT:    slli a2, a2, 9
 ; RV32-NEXT:    slli t2, t2, 23
-; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    lui a2, 260096
-; RV32-NEXT:    srli a3, a3, 9
-; RV32-NEXT:    add a2, t2, a2
-; RV32-NEXT:    or a0, a3, a0
-; RV32-NEXT:    or a4, a0, a2
+; RV32-NEXT:    and a0, a0, a3
+; RV32-NEXT:    lui a3, 260096
+; RV32-NEXT:    srli a2, a2, 9
+; RV32-NEXT:    add a3, t2, a3
+; RV32-NEXT:    or a0, a2, a0
+; RV32-NEXT:    or a4, a0, a3
 ; RV32-NEXT:    sw a4, 0(a1)
 ; RV32-NEXT:    ret
 ; RV32-NEXT:  .LBB1_37:
 ; RV32-NEXT:    sw zero, 0(a1)
 ; RV32-NEXT:    ret
 ; RV32-NEXT:  .LBB1_38: # %itofp-if-else
-; RV32-NEXT:    addi a3, t0, -176
-; RV32-NEXT:    sw t4, 160(sp)
-; RV32-NEXT:    srli a4, a3, 3
-; RV32-NEXT:    sw zero, 144(sp)
-; RV32-NEXT:    sw zero, 148(sp)
+; RV32-NEXT:    addi a2, t0, -176
+; RV32-NEXT:    sw t4, 168(sp)
+; RV32-NEXT:    srli a3, a2, 3
 ; RV32-NEXT:    sw zero, 152(sp)
 ; RV32-NEXT:    sw zero, 156(sp)
-; RV32-NEXT:    andi a4, a4, 28
-; RV32-NEXT:    addi a5, sp, 160
-; RV32-NEXT:    sw zero, 128(sp)
-; RV32-NEXT:    sw zero, 132(sp)
+; RV32-NEXT:    sw zero, 160(sp)
+; RV32-NEXT:    sw zero, 164(sp)
+; RV32-NEXT:    andi a3, a3, 28
+; RV32-NEXT:    addi a4, sp, 168
 ; RV32-NEXT:    sw zero, 136(sp)
 ; RV32-NEXT:    sw zero, 140(sp)
-; RV32-NEXT:    sub a5, a5, a4
-; RV32-NEXT:    lw a4, 0(a5)
-; RV32-NEXT:    sll a3, a4, a3
-; RV32-NEXT:    j .LBB1_31
+; RV32-NEXT:    sw zero, 144(sp)
+; RV32-NEXT:    sw zero, 148(sp)
+; RV32-NEXT:    sub a4, a4, a3
+; RV32-NEXT:    lw a3, 0(a4)
+; RV32-NEXT:    sll a2, a3, a2
+; RV32-NEXT:    mv t2, t1
+; RV32-NEXT:    lw a3, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    j .LBB1_32
 ; RV32-NEXT:  .LBB1_39: # %itofp-if-then20
-; RV32-NEXT:    srli a3, a3, 3
+; RV32-NEXT:    srli a2, a2, 3
 ; RV32-NEXT:    slli t3, t3, 29
-; RV32-NEXT:    or a3, a3, t3
+; RV32-NEXT:    or a2, a2, t3
 ; RV32-NEXT:    j .LBB1_32
 ; RV32-NEXT:  .LBB1_40: # %itofp-sw-bb
-; RV32-NEXT:    srli a3, t4, 31
+; RV32-NEXT:    srli a2, t4, 31
 ; RV32-NEXT:    slli t3, t3, 1
-; RV32-NEXT:    or t3, t3, a3
+; RV32-NEXT:    or t3, t3, a2
 ; RV32-NEXT:    slli t4, t4, 1
-; RV32-NEXT:    j .LBB1_29
+; RV32-NEXT:    j .LBB1_30
   %1 = load i200, ptr %in, align 8
   %2 = sitofp i200 %1 to float
   store float %2, ptr %out, align 4
@@ -988,836 +992,972 @@ define void @test_bitint_200_to_float(ptr %in, ptr %out) nounwind {
 define void @test_float_to_bitint_200(ptr %in, ptr %out) nounwind {
 ; RV64-LABEL: test_float_to_bitint_200:
 ; RV64:       # %bb.0: # %fp-to-i-entry
-; RV64-NEXT:    addi sp, sp, -160
-; RV64-NEXT:    sd ra, 152(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s0, 144(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s1, 136(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s2, 128(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s3, 120(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s4, 112(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s5, 104(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s6, 96(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s7, 88(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s8, 80(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s9, 72(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    li a3, 127
 ; RV64-NEXT:    lw a2, 0(a0)
 ; RV64-NEXT:    slli a0, a2, 33
 ; RV64-NEXT:    srli a0, a0, 56
-; RV64-NEXT:    mv s0, a1
 ; RV64-NEXT:    bgeu a0, a3, .LBB2_2
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    li s2, 0
-; RV64-NEXT:    li s6, 0
+; RV64-NEXT:    li a2, 0
 ; RV64-NEXT:    li a0, 0
-; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    j .LBB2_5
+; RV64-NEXT:    li a3, 0
+; RV64-NEXT:    li a4, 0
+; RV64-NEXT:    j .LBB2_6
 ; RV64-NEXT:  .LBB2_2: # %fp-to-i-if-check.exp.size
+; RV64-NEXT:    addi sp, sp, -160
+; RV64-NEXT:    sd ra, 152(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a1, 32(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    slli a1, a2, 41
-; RV64-NEXT:    sraiw s1, a2, 31
+; RV64-NEXT:    sraiw a4, a2, 31
 ; RV64-NEXT:    srli a1, a1, 41
 ; RV64-NEXT:    lui a2, 2048
 ; RV64-NEXT:    li a3, 149
 ; RV64-NEXT:    or a1, a1, a2
-; RV64-NEXT:    ori s3, s1, 1
+; RV64-NEXT:    ori a5, a4, 1
+; RV64-NEXT:    sd a4, 72(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a5, 64(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    bltu a3, a0, .LBB2_4
 ; RV64-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
 ; RV64-NEXT:    li a2, 150
 ; RV64-NEXT:    sub a2, a2, a0
-; RV64-NEXT:    srlw s4, a1, a2
-; RV64-NEXT:    mv a0, s4
+; RV64-NEXT:    srlw a0, a1, a2
+; RV64-NEXT:    sd a0, 56(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 56(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s3
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    mv s2, a0
-; RV64-NEXT:    mv s5, a1
+; RV64-NEXT:    sd a0, 48(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a1, 40(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    li a0, 0
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s3
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    add s5, a0, s5
-; RV64-NEXT:    sltu a0, s5, a0
-; RV64-NEXT:    add s7, a1, a0
-; RV64-NEXT:    mv a0, s4
+; RV64-NEXT:    ld a2, 40(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a2, a0, a2
+; RV64-NEXT:    sd a2, 64(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a0, a2, a0
+; RV64-NEXT:    add a0, a1, a0
+; RV64-NEXT:    sd a0, 16(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 56(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s1
+; RV64-NEXT:    ld a2, 72(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    add s6, a0, s5
-; RV64-NEXT:    sltu a0, s6, a0
-; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    add s8, s7, a0
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a2, a0, a2
+; RV64-NEXT:    sd a2, 64(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a0, a2, a0
+; RV64-NEXT:    add a1, a1, a0
+; RV64-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    li a0, 0
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s1
+; RV64-NEXT:    ld a2, 72(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    mv s3, a0
-; RV64-NEXT:    mv s5, a1
-; RV64-NEXT:    add s9, a0, s8
-; RV64-NEXT:    mv a0, s1
-; RV64-NEXT:    mv a1, s1
-; RV64-NEXT:    mv a2, s4
+; RV64-NEXT:    sd a0, 24(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    sd a0, 40(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 72(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a1, 72(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a2, 56(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    sltu a2, s8, s7
-; RV64-NEXT:    sltu a3, s9, s3
-; RV64-NEXT:    add a2, s5, a2
-; RV64-NEXT:    add a0, s9, a0
-; RV64-NEXT:    add a2, a2, a3
-; RV64-NEXT:    sltu a3, a0, s9
-; RV64-NEXT:    add a1, a2, a1
-; RV64-NEXT:    add a1, a1, a3
+; RV64-NEXT:    ld a2, 48(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a3, 16(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a4, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a5, a4, a3
+; RV64-NEXT:    ld a3, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a4, 40(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a4, a4, a3
+; RV64-NEXT:    ld a3, 0(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a5, a3, a5
+; RV64-NEXT:    ld a3, 40(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a3, a3, a0
+; RV64-NEXT:    ld a0, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a5, a5, a4
+; RV64-NEXT:    ld a4, 40(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a4, a3, a4
+; RV64-NEXT:    add a1, a5, a1
 ; RV64-NEXT:    j .LBB2_5
 ; RV64-NEXT:  .LBB2_4: # %fp-to-i-if-exp.large
-; RV64-NEXT:    addi a0, a0, -150
-; RV64-NEXT:    srli a2, a0, 3
-; RV64-NEXT:    sd a1, 32(sp)
-; RV64-NEXT:    sd zero, 40(sp)
-; RV64-NEXT:    sd zero, 48(sp)
-; RV64-NEXT:    sd zero, 56(sp)
-; RV64-NEXT:    andi a2, a2, 24
-; RV64-NEXT:    addi a1, sp, 32
-; RV64-NEXT:    sd zero, 0(sp)
-; RV64-NEXT:    sd zero, 8(sp)
-; RV64-NEXT:    sd zero, 16(sp)
-; RV64-NEXT:    sd zero, 24(sp)
-; RV64-NEXT:    sub a1, a1, a2
-; RV64-NEXT:    ld a4, 0(a1)
+; RV64-NEXT:    addi a6, a0, -150
+; RV64-NEXT:    srli a0, a6, 3
+; RV64-NEXT:    sd a1, 112(sp)
+; RV64-NEXT:    sd zero, 120(sp)
+; RV64-NEXT:    sd zero, 128(sp)
+; RV64-NEXT:    sd zero, 136(sp)
+; RV64-NEXT:    andi a0, a0, 24
+; RV64-NEXT:    addi a1, sp, 112
+; RV64-NEXT:    sd zero, 80(sp)
+; RV64-NEXT:    sd zero, 88(sp)
+; RV64-NEXT:    sd zero, 96(sp)
+; RV64-NEXT:    sd zero, 104(sp)
+; RV64-NEXT:    sub a1, a1, a0
+; RV64-NEXT:    ld a7, 0(a1)
 ; RV64-NEXT:    ld a2, 8(a1)
-; RV64-NEXT:    andi a3, a0, 63
-; RV64-NEXT:    ld a5, 16(a1)
+; RV64-NEXT:    andi a0, a6, 63
+; RV64-NEXT:    ld a3, 16(a1)
 ; RV64-NEXT:    ld a1, 24(a1)
-; RV64-NEXT:    srli a6, a4, 1
-; RV64-NEXT:    xori a3, a3, 63
-; RV64-NEXT:    sll a7, a2, a0
-; RV64-NEXT:    srl a6, a6, a3
-; RV64-NEXT:    or s4, a7, a6
+; RV64-NEXT:    srli t0, a7, 1
+; RV64-NEXT:    xori t1, a0, 63
+; RV64-NEXT:    sll a0, a2, a6
+; RV64-NEXT:    srl t0, t0, t1
+; RV64-NEXT:    or a0, a0, t0
+; RV64-NEXT:    sd a0, 56(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    srli a2, a2, 1
-; RV64-NEXT:    sll a6, a5, a0
-; RV64-NEXT:    srl a2, a2, a3
-; RV64-NEXT:    or a2, a6, a2
-; RV64-NEXT:    srli a5, a5, 1
-; RV64-NEXT:    sll a1, a1, a0
-; RV64-NEXT:    srl a3, a5, a3
-; RV64-NEXT:    or a3, a1, a3
-; RV64-NEXT:    sll s5, a4, a0
-; RV64-NEXT:    mv a0, s3
-; RV64-NEXT:    mv a1, s1
+; RV64-NEXT:    sll a0, a3, a6
+; RV64-NEXT:    srl a2, a2, t1
+; RV64-NEXT:    or a2, a0, a2
+; RV64-NEXT:    srli a3, a3, 1
+; RV64-NEXT:    sll a0, a1, a6
+; RV64-NEXT:    srl a3, a3, t1
+; RV64-NEXT:    or a3, a0, a3
+; RV64-NEXT:    sll a0, a7, a6
+; RV64-NEXT:    sd a0, 40(sp) # 8-byte Folded Spill
+; RV64-NEXT:    mv a0, a5
+; RV64-NEXT:    mv a1, a4
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    mv s2, a0
-; RV64-NEXT:    mv s6, a1
-; RV64-NEXT:    mv a0, s1
-; RV64-NEXT:    mv a1, s1
-; RV64-NEXT:    mv a2, s5
-; RV64-NEXT:    mv a3, s4
+; RV64-NEXT:    sd a0, 48(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a1, 24(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 72(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a1, 72(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a2, 40(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a3, 56(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    add s8, a0, s2
-; RV64-NEXT:    add a1, a1, s6
-; RV64-NEXT:    sltu a0, s8, a0
-; RV64-NEXT:    add s9, a1, a0
-; RV64-NEXT:    mv a0, s4
+; RV64-NEXT:    ld a2, 48(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a2, a0, a2
+; RV64-NEXT:    sd a2, 16(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a2, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a2, a1, a2
+; RV64-NEXT:    ld a1, 16(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a0, a1, a0
+; RV64-NEXT:    add a0, a2, a0
+; RV64-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 56(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s3
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    mv s6, a0
-; RV64-NEXT:    mv s7, a1
-; RV64-NEXT:    mv a0, s5
+; RV64-NEXT:    sd a0, 24(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a1, 8(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 40(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s3
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    mv s2, a0
-; RV64-NEXT:    add s3, s6, a1
-; RV64-NEXT:    sltu a0, s3, s6
-; RV64-NEXT:    add s7, s7, a0
-; RV64-NEXT:    mv a0, s5
+; RV64-NEXT:    sd a0, 48(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    sd a0, 64(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a1, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a1, a1, a0
+; RV64-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    sd a0, 24(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 40(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s1
+; RV64-NEXT:    ld a2, 72(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    add s6, a0, s3
-; RV64-NEXT:    sltu a0, s6, a0
-; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    add s3, s7, a0
-; RV64-NEXT:    sltu s5, s3, s7
-; RV64-NEXT:    mv a0, s4
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a2, a0, a2
+; RV64-NEXT:    sd a2, 64(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a0, a2, a0
+; RV64-NEXT:    add a1, a1, a0
+; RV64-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    sd a0, 40(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a1, 40(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sltu a0, a1, a0
+; RV64-NEXT:    sd a0, 24(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 56(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    mv a2, s1
+; RV64-NEXT:    ld a2, 72(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    li a3, 0
 ; RV64-NEXT:    call __multi3
-; RV64-NEXT:    add s3, a0, s3
-; RV64-NEXT:    add a1, a1, s5
-; RV64-NEXT:    sltu a0, s3, a0
-; RV64-NEXT:    add a1, a1, a0
-; RV64-NEXT:    add a0, s3, s8
-; RV64-NEXT:    add a1, a1, s9
-; RV64-NEXT:    sltu a2, a0, s3
-; RV64-NEXT:    add a1, a1, a2
-; RV64-NEXT:  .LBB2_5: # %fp-to-i-cleanup
-; RV64-NEXT:    sd s2, 0(s0)
-; RV64-NEXT:    sd s6, 8(s0)
-; RV64-NEXT:    sd a0, 16(s0)
-; RV64-NEXT:    sb a1, 24(s0)
+; RV64-NEXT:    ld a2, 48(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a3, 40(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a4, a0, a3
+; RV64-NEXT:    ld a3, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a1, a1, a3
+; RV64-NEXT:    sltu a3, a4, a0
+; RV64-NEXT:    ld a0, 64(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a5, a1, a3
+; RV64-NEXT:    ld a3, 16(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a3, a4, a3
+; RV64-NEXT:    ld a1, 0(sp) # 8-byte Folded Reload
+; RV64-NEXT:    add a1, a5, a1
+; RV64-NEXT:    sltu a4, a3, a4
+; RV64-NEXT:  .LBB2_5:
+; RV64-NEXT:    add a4, a1, a4
+; RV64-NEXT:    ld a1, 32(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    ld ra, 152(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s0, 144(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s1, 136(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s2, 128(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s3, 120(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s4, 112(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s5, 104(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s6, 96(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s7, 88(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s8, 80(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s9, 72(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    addi sp, sp, 160
+; RV64-NEXT:  .LBB2_6: # %fp-to-i-cleanup
+; RV64-NEXT:    sd a2, 0(a1)
+; RV64-NEXT:    sd a0, 8(a1)
+; RV64-NEXT:    sd a3, 16(a1)
+; RV64-NEXT:    sb a4, 24(a1)
 ; RV64-NEXT:    ret
 ;
 ; RV32-LABEL: test_float_to_bitint_200:
 ; RV32:       # %bb.0: # %fp-to-i-entry
-; RV32-NEXT:    addi sp, sp, -256
-; RV32-NEXT:    sw ra, 252(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 248(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 244(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 240(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 236(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s4, 232(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s5, 228(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 224(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 220(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s8, 216(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s9, 212(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s10, 208(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s11, 204(sp) # 4-byte Folded Spill
+; RV32-NEXT:    addi sp, sp, -272
+; RV32-NEXT:    sw ra, 268(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s0, 264(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s1, 260(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s2, 256(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a3, 127
 ; RV32-NEXT:    lw a2, 0(a0)
 ; RV32-NEXT:    slli a0, a2, 1
 ; RV32-NEXT:    srli a0, a0, 24
-; RV32-NEXT:    mv s0, a1
 ; RV32-NEXT:    bgeu a0, a3, .LBB2_2
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    li a2, 0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    li a4, 0
-; RV32-NEXT:    li s8, 0
-; RV32-NEXT:    li s10, 0
+; RV32-NEXT:    li a5, 0
+; RV32-NEXT:    li a6, 0
+; RV32-NEXT:    li s2, 0
+; RV32-NEXT:    li a2, 0
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    j .LBB2_21
 ; RV32-NEXT:  .LBB2_2: # %fp-to-i-if-check.exp.size
-; RV32-NEXT:    srai s1, a2, 31
+; RV32-NEXT:    sw a1, 148(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srai s0, a2, 31
 ; RV32-NEXT:    slli a2, a2, 9
-; RV32-NEXT:    zext.b s11, s1
+; RV32-NEXT:    zext.b a1, s0
+; RV32-NEXT:    sw a1, 164(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    srli a2, a2, 9
 ; RV32-NEXT:    lui a1, 2048
 ; RV32-NEXT:    li a3, 149
 ; RV32-NEXT:    or a1, a2, a1
-; RV32-NEXT:    ori s3, s1, 1
+; RV32-NEXT:    sw s0, 184(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    bltu a3, a0, .LBB2_8
 ; RV32-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
 ; RV32-NEXT:    li a2, 150
 ; RV32-NEXT:    sub a2, a2, a0
-; RV32-NEXT:    srl s5, a1, a2
-; RV32-NEXT:    mv a0, s5
+; RV32-NEXT:    srl a0, a1, a2
+; RV32-NEXT:    sw a0, 188(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s3
+; RV32-NEXT:    ori a2, s0, 1
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sw a0, 120(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a0, 160(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    mv s2, a1
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s3
+; RV32-NEXT:    ori a2, s0, 1
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
 ; RV32-NEXT:    add s2, a0, s2
 ; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add s3, a1, a0
-; RV32-NEXT:    mv a0, s5
+; RV32-NEXT:    add s1, a1, a0
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
 ; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    sw s2, 116(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s2, 144(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sltu a0, s2, a0
 ; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s2, s3, a0
+; RV32-NEXT:    add s2, s1, a0
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s7, a0, s2
-; RV32-NEXT:    sltu a2, s2, s3
-; RV32-NEXT:    sltu a0, s7, a0
-; RV32-NEXT:    add a1, a1, a2
-; RV32-NEXT:    add s8, a1, a0
-; RV32-NEXT:    mv a0, s5
+; RV32-NEXT:    add a2, a0, s2
+; RV32-NEXT:    sw a2, 180(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a3, s2, s1
+; RV32-NEXT:    lw a2, 180(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a3
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sw a0, 156(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s4, a0
-; RV32-NEXT:    mv s6, a1
+; RV32-NEXT:    mv s2, a0
+; RV32-NEXT:    sw a1, 168(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s3, a0
-; RV32-NEXT:    mv s2, a1
-; RV32-NEXT:    add s6, a0, s6
-; RV32-NEXT:    mv a0, s5
+; RV32-NEXT:    sw a0, 176(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 172(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s1, 176(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, s1, a0
+; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s7, s4, s7
-; RV32-NEXT:    add a2, a0, s6
-; RV32-NEXT:    add s8, a2, s8
-; RV32-NEXT:    sw s7, 112(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu s4, s7, s4
-; RV32-NEXT:    add s8, s8, s4
-; RV32-NEXT:    beq s8, a2, .LBB2_5
+; RV32-NEXT:    lw a3, 180(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, s2, a3
+; RV32-NEXT:    add a4, a0, s1
+; RV32-NEXT:    lw a3, 156(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a6, a4, a3
+; RV32-NEXT:    sw a5, 140(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a3, a5, s2
+; RV32-NEXT:    add a5, a6, a3
+; RV32-NEXT:    beq a5, a4, .LBB2_5
 ; RV32-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32-NEXT:    sltu s4, s8, a2
+; RV32-NEXT:    sltu a3, a5, a4
 ; RV32-NEXT:  .LBB2_5: # %fp-to-i-if-exp.small
-; RV32-NEXT:    sw s8, 108(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu a3, s6, s3
-; RV32-NEXT:    sltu a0, a2, a0
-; RV32-NEXT:    add s6, s2, a3
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s7, s6, a0
+; RV32-NEXT:    sw a3, 168(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a5, 136(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a5, 176(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a2, s1, a5
+; RV32-NEXT:    sltu a4, a4, a0
+; RV32-NEXT:    lw a0, 172(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    sw a0, 180(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add a1, a1, a4
+; RV32-NEXT:    lw s2, 180(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, s2, a1
+; RV32-NEXT:    sw a1, 176(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s2, a0
-; RV32-NEXT:    mv s3, a1
-; RV32-NEXT:    add s8, a0, s7
-; RV32-NEXT:    add s9, s8, s4
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    sw a0, 172(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 156(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s1, 172(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, s1, a0
+; RV32-NEXT:    lw a0, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s2, s1, a0
+; RV32-NEXT:    mv a0, s0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s5
+; RV32-NEXT:    lw a2, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s4, a1
-; RV32-NEXT:    sltu a1, s7, s6
-; RV32-NEXT:    add a2, s9, a0
-; RV32-NEXT:    sltu a0, s8, s2
-; RV32-NEXT:    add a1, s3, a1
-; RV32-NEXT:    sltu s7, s9, s8
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    sltu s6, a2, s9
-; RV32-NEXT:    mv s8, a2
-; RV32-NEXT:    add s7, a0, s7
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    sw a1, 168(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a1, 180(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 176(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a2, a1
+; RV32-NEXT:    add a2, s2, a0
+; RV32-NEXT:    lw a0, 172(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a3, s1, a0
+; RV32-NEXT:    lw a0, 156(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sltu a1, s2, s1
+; RV32-NEXT:    add a0, a0, a3
+; RV32-NEXT:    sltu a3, a2, s2
+; RV32-NEXT:    sw a3, 180(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv s2, a2
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 172(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 184(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s5
+; RV32-NEXT:    lw a2, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s3, a0
-; RV32-NEXT:    mv s2, a1
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    mv s0, a0
+; RV32-NEXT:    sw a1, 176(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, s0, a0
+; RV32-NEXT:    lw a0, 184(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    li a2, 0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add a2, a0, s4
-; RV32-NEXT:    add s10, s7, a2
-; RV32-NEXT:    add s10, s10, s6
-; RV32-NEXT:    beq s10, s7, .LBB2_7
+; RV32-NEXT:    lw a6, 172(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 180(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a0, s1
+; RV32-NEXT:    add a3, a6, a5
+; RV32-NEXT:    add a3, a3, a2
+; RV32-NEXT:    beq a3, a6, .LBB2_7
 ; RV32-NEXT:  # %bb.6: # %fp-to-i-if-exp.small
-; RV32-NEXT:    sltu s6, s10, s7
+; RV32-NEXT:    sltu a2, a3, a6
 ; RV32-NEXT:  .LBB2_7: # %fp-to-i-if-exp.small
-; RV32-NEXT:    sltu a3, s4, s3
-; RV32-NEXT:    sltu a0, a2, a0
-; RV32-NEXT:    add a3, s2, a3
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s2, a3, a0
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    sw a2, 180(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a4, s1, s0
+; RV32-NEXT:    sltu a5, a5, a0
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a4
+; RV32-NEXT:    add a1, a1, a5
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 176(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 184(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
+; RV32-NEXT:    mv s1, a3
 ; RV32-NEXT:    li a2, 0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv a0, s5
+; RV32-NEXT:    mv s0, a0
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s11
+; RV32-NEXT:    lw a2, 164(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add a0, s1, a0
-; RV32-NEXT:    add a0, a0, s6
-; RV32-NEXT:    add a0, a0, s2
+; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    add s0, s0, a0
+; RV32-NEXT:    lw a0, 180(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s0, s0, a0
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, s0, a0
 ; RV32-NEXT:    zext.b a0, a0
-; RV32-NEXT:    lw a1, 120(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a2, 116(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a3, 112(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a4, 108(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 148(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 160(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 144(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 140(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a6, 136(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    j .LBB2_21
 ; RV32-NEXT:  .LBB2_8: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sw s11, 88(sp) # 4-byte Folded Spill
-; RV32-NEXT:    addi a7, a0, -150
-; RV32-NEXT:    sw zero, 184(sp)
-; RV32-NEXT:    sw zero, 188(sp)
+; RV32-NEXT:    addi a0, a0, -150
+; RV32-NEXT:    sw zero, 240(sp)
+; RV32-NEXT:    sw zero, 244(sp)
+; RV32-NEXT:    sw zero, 248(sp)
+; RV32-NEXT:    sw zero, 252(sp)
+; RV32-NEXT:    sw a1, 224(sp)
+; RV32-NEXT:    sw zero, 228(sp)
+; RV32-NEXT:    sw zero, 232(sp)
+; RV32-NEXT:    sw zero, 236(sp)
+; RV32-NEXT:    srli a1, a0, 3
+; RV32-NEXT:    sw zero, 208(sp)
+; RV32-NEXT:    sw zero, 212(sp)
+; RV32-NEXT:    sw zero, 216(sp)
+; RV32-NEXT:    sw zero, 220(sp)
+; RV32-NEXT:    andi a1, a1, 28
+; RV32-NEXT:    addi a2, sp, 224
 ; RV32-NEXT:    sw zero, 192(sp)
 ; RV32-NEXT:    sw zero, 196(sp)
-; RV32-NEXT:    sw a1, 168(sp)
-; RV32-NEXT:    sw zero, 172(sp)
-; RV32-NEXT:    sw zero, 176(sp)
-; RV32-NEXT:    sw zero, 180(sp)
-; RV32-NEXT:    srli a0, a7, 3
-; RV32-NEXT:    sw zero, 152(sp)
-; RV32-NEXT:    sw zero, 156(sp)
-; RV32-NEXT:    sw zero, 160(sp)
-; RV32-NEXT:    sw zero, 164(sp)
-; RV32-NEXT:    andi a0, a0, 28
-; RV32-NEXT:    addi a1, sp, 168
-; RV32-NEXT:    sw zero, 136(sp)
-; RV32-NEXT:    sw zero, 140(sp)
-; RV32-NEXT:    sw zero, 144(sp)
-; RV32-NEXT:    sw zero, 148(sp)
-; RV32-NEXT:    sub a1, a1, a0
-; RV32-NEXT:    lw a0, 16(a1)
-; RV32-NEXT:    lw t0, 20(a1)
-; RV32-NEXT:    andi a2, a7, 31
-; RV32-NEXT:    lw a3, 0(a1)
-; RV32-NEXT:    lw a4, 4(a1)
-; RV32-NEXT:    lw a5, 8(a1)
-; RV32-NEXT:    lw a6, 12(a1)
-; RV32-NEXT:    xori a2, a2, 31
-; RV32-NEXT:    lw t1, 24(a1)
-; RV32-NEXT:    sw t1, 68(sp) # 4-byte Folded Spill
-; RV32-NEXT:    lw a1, 28(a1)
-; RV32-NEXT:    sw a1, 72(sp) # 4-byte Folded Spill
-; RV32-NEXT:    srli a1, a0, 1
-; RV32-NEXT:    sw t0, 80(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sll t0, t0, a7
-; RV32-NEXT:    sw t0, 104(sp) # 4-byte Folded Spill
-; RV32-NEXT:    srl s7, a1, a2
-; RV32-NEXT:    sll a0, a0, a7
-; RV32-NEXT:    sw a0, 112(sp) # 4-byte Folded Spill
-; RV32-NEXT:    srli a0, a6, 1
-; RV32-NEXT:    srl a0, a0, a2
-; RV32-NEXT:    sw a0, 108(sp) # 4-byte Folded Spill
-; RV32-NEXT:    srli a0, a5, 1
-; RV32-NEXT:    sll a1, a6, a7
-; RV32-NEXT:    srl a0, a0, a2
-; RV32-NEXT:    or s8, a1, a0
-; RV32-NEXT:    srli a0, a4, 1
-; RV32-NEXT:    sll a1, a5, a7
-; RV32-NEXT:    srl a0, a0, a2
-; RV32-NEXT:    or s9, a1, a0
-; RV32-NEXT:    srli a0, a3, 1
-; RV32-NEXT:    sll a1, a4, a7
-; RV32-NEXT:    sw a2, 76(sp) # 4-byte Folded Spill
-; RV32-NEXT:    srl a0, a0, a2
-; RV32-NEXT:    or s6, a1, a0
-; RV32-NEXT:    sw a7, 84(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sll s10, a3, a7
-; RV32-NEXT:    mv a0, s6
+; RV32-NEXT:    sw zero, 200(sp)
+; RV32-NEXT:    sw zero, 204(sp)
+; RV32-NEXT:    sub a2, a2, a1
+; RV32-NEXT:    lw a3, 16(a2)
+; RV32-NEXT:    lw a1, 20(a2)
+; RV32-NEXT:    sw a1, 104(sp) # 4-byte Folded Spill
+; RV32-NEXT:    andi a1, a0, 31
+; RV32-NEXT:    lw a4, 0(a2)
+; RV32-NEXT:    lw a5, 4(a2)
+; RV32-NEXT:    lw a6, 8(a2)
+; RV32-NEXT:    lw a7, 12(a2)
+; RV32-NEXT:    xori t0, a1, 31
+; RV32-NEXT:    lw a1, 24(a2)
+; RV32-NEXT:    sw a1, 52(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a1, 28(a2)
+; RV32-NEXT:    sw a1, 56(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srli a2, a3, 1
+; RV32-NEXT:    lw a1, 104(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sll a1, a1, a0
+; RV32-NEXT:    sw a1, 156(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srl a1, a2, t0
+; RV32-NEXT:    sw a1, 128(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sll a1, a3, a0
+; RV32-NEXT:    sw a1, 136(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srli a1, a7, 1
+; RV32-NEXT:    srl a1, a1, t0
+; RV32-NEXT:    sw a1, 132(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srli a1, a6, 1
+; RV32-NEXT:    sll a2, a7, a0
+; RV32-NEXT:    srl a1, a1, t0
+; RV32-NEXT:    or a1, a2, a1
+; RV32-NEXT:    sw a1, 180(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srli a1, a5, 1
+; RV32-NEXT:    sll a2, a6, a0
+; RV32-NEXT:    srl a1, a1, t0
+; RV32-NEXT:    or a1, a2, a1
+; RV32-NEXT:    sw a1, 176(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srli a1, a4, 1
+; RV32-NEXT:    sll a2, a5, a0
+; RV32-NEXT:    sw t0, 48(sp) # 4-byte Folded Spill
+; RV32-NEXT:    srl a1, a1, t0
+; RV32-NEXT:    or a1, a2, a1
+; RV32-NEXT:    sw a1, 188(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a0, 44(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sll s1, a4, a0
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s3
+; RV32-NEXT:    ori a2, s0, 1
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s2, a0
-; RV32-NEXT:    mv s5, s3
-; RV32-NEXT:    mv s3, a1
-; RV32-NEXT:    mv a0, s10
+; RV32-NEXT:    mv s0, a0
+; RV32-NEXT:    sw a1, 172(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, s1
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s5
+; RV32-NEXT:    lw a2, 184(sp) # 4-byte Folded Reload
+; RV32-NEXT:    ori a2, a2, 1
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sw a0, 120(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s4, s2, a1
-; RV32-NEXT:    sltu a0, s4, s2
-; RV32-NEXT:    add s3, s3, a0
-; RV32-NEXT:    sw s10, 124(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv a0, s10
+; RV32-NEXT:    sw a0, 160(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add s2, s0, a1
+; RV32-NEXT:    sltu a1, s2, s0
+; RV32-NEXT:    lw a0, 172(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s0, a0, a1
+; RV32-NEXT:    sw s1, 172(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, s1
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    lw a2, 184(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    sw s4, 116(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu a0, s4, a0
+; RV32-NEXT:    add s2, a0, s2
+; RV32-NEXT:    sw s2, 144(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a0, s2, a0
 ; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s2, s3, a0
-; RV32-NEXT:    sltu s3, s2, s3
-; RV32-NEXT:    sw s6, 132(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv a0, s6
+; RV32-NEXT:    add s1, s0, a0
+; RV32-NEXT:    sltu s2, s1, s0
+; RV32-NEXT:    lw s0, 184(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s11, a0, s2
-; RV32-NEXT:    add a1, a1, s3
-; RV32-NEXT:    sltu a0, s11, a0
-; RV32-NEXT:    add s10, a1, a0
-; RV32-NEXT:    mv a0, s9
+; RV32-NEXT:    add s1, a0, s1
+; RV32-NEXT:    sw s1, 168(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add s2, a1, s2
+; RV32-NEXT:    lw a1, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    add a0, s2, a0
+; RV32-NEXT:    sw a0, 120(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s5
+; RV32-NEXT:    ori a2, s0, 1
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s4, a0
-; RV32-NEXT:    mv s6, a1
-; RV32-NEXT:    sw s8, 128(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv a0, s8
+; RV32-NEXT:    mv s1, a0
+; RV32-NEXT:    mv s2, a1
+; RV32-NEXT:    lw a0, 180(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s5
+; RV32-NEXT:    ori a2, s0, 1
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s5, a0
-; RV32-NEXT:    mv s2, a1
-; RV32-NEXT:    add s3, a0, s6
-; RV32-NEXT:    sw s9, 100(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv a0, s9
+; RV32-NEXT:    sw a0, 124(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 152(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add s2, a0, s2
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s11, s4, s11
-; RV32-NEXT:    add a2, a0, s3
-; RV32-NEXT:    add s10, a2, s10
-; RV32-NEXT:    sltu s6, s11, s4
-; RV32-NEXT:    add s10, s10, s6
-; RV32-NEXT:    beq s10, a2, .LBB2_10
+; RV32-NEXT:    lw a2, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, s1, a2
+; RV32-NEXT:    sw a2, 140(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add a3, a0, s2
+; RV32-NEXT:    lw a2, 120(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a4, a3, a2
+; RV32-NEXT:    lw a2, 140(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu s1, a2, s1
+; RV32-NEXT:    add a2, a4, s1
+; RV32-NEXT:    beq a2, a3, .LBB2_10
 ; RV32-NEXT:  # %bb.9: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sltu s6, s10, a2
+; RV32-NEXT:    sltu s1, a2, a3
 ; RV32-NEXT:  .LBB2_10: # %fp-to-i-if-exp.large
-; RV32-NEXT:    lw a3, 104(sp) # 4-byte Folded Reload
-; RV32-NEXT:    or a3, a3, s7
-; RV32-NEXT:    sw a3, 96(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu a3, s3, s5
-; RV32-NEXT:    sltu a0, a2, a0
-; RV32-NEXT:    add a3, s2, a3
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    lw a1, 112(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a2, 108(sp) # 4-byte Folded Reload
-; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    sw a1, 104(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a3, 92(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s2, a3, a0
-; RV32-NEXT:    lw a0, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 156(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    or a4, a4, a5
+; RV32-NEXT:    sw a4, 156(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a4, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, s2, a4
+; RV32-NEXT:    sltu a3, a3, a0
+; RV32-NEXT:    lw a0, 152(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a4
+; RV32-NEXT:    sw a0, 108(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add a3, a1, a3
+; RV32-NEXT:    lw a0, 136(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    or a0, a0, a1
+; RV32-NEXT:    sw a0, 152(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 108(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a3
+; RV32-NEXT:    sw a0, 100(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 180(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    sw a2, 136(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sw a1, 60(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a0, 52(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    sw s2, 56(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s4, s2, s6
+; RV32-NEXT:    sw a0, 124(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 76(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 100(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 132(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    lw a0, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, a0, s1
+; RV32-NEXT:    sw s1, 168(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s6, a0
-; RV32-NEXT:    mv s2, a1
-; RV32-NEXT:    lw s3, 124(sp) # 4-byte Folded Reload
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    sw a0, 128(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 120(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw s1, 172(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv a0, s1
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s8, a0
-; RV32-NEXT:    add s7, s6, a1
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    mv s2, a0
+; RV32-NEXT:    lw a0, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 116(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, s1
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s11, s8, s11
-; RV32-NEXT:    add a2, a0, s7
-; RV32-NEXT:    add s10, a2, s10
-; RV32-NEXT:    sw s11, 112(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu s3, s11, s8
-; RV32-NEXT:    add s10, s10, s3
-; RV32-NEXT:    beq s10, a2, .LBB2_12
+; RV32-NEXT:    lw a2, 116(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 140(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, s2, a3
+; RV32-NEXT:    add a3, a0, a2
+; RV32-NEXT:    lw a4, 136(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a4, a3, a4
+; RV32-NEXT:    sw a5, 140(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu s1, a5, s2
+; RV32-NEXT:    add a4, a4, s1
+; RV32-NEXT:    beq a4, a3, .LBB2_12
 ; RV32-NEXT:  # %bb.11: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sltu s3, s10, a2
+; RV32-NEXT:    sltu s1, a4, a3
 ; RV32-NEXT:  .LBB2_12: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sw s10, 108(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu a3, s7, s6
-; RV32-NEXT:    sltu a0, a2, a0
-; RV32-NEXT:    add s10, s2, a3
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s6, s10, a0
-; RV32-NEXT:    lw a0, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sw a4, 136(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a4, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a2, a2, a4
+; RV32-NEXT:    sltu a3, a3, a0
+; RV32-NEXT:    lw a0, 120(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    sw a0, 92(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add a1, a1, a3
+; RV32-NEXT:    lw a0, 92(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 88(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s9, a0
-; RV32-NEXT:    sw a1, 36(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s8, a0, s6
-; RV32-NEXT:    add s3, s8, s3
-; RV32-NEXT:    add s5, s4, s3
-; RV32-NEXT:    lw s11, 100(sp) # 4-byte Folded Reload
-; RV32-NEXT:    mv a0, s11
+; RV32-NEXT:    sw a0, 112(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 64(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 112(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 88(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s2, a0, a1
+; RV32-NEXT:    add s1, s2, s1
+; RV32-NEXT:    sw s1, 84(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 84(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 80(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sw a0, 48(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a1, 28(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s7, a0, s5
-; RV32-NEXT:    ori a0, s1, 1
+; RV32-NEXT:    sw a0, 116(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 60(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw s1, 116(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a0, 80(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, s1, a0
+; RV32-NEXT:    sw s1, 128(sp) # 4-byte Folded Spill
+; RV32-NEXT:    ori a0, s0, 1
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    lw a2, 104(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 152(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s2, a0
-; RV32-NEXT:    sw a1, 40(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    mv s1, a0
+; RV32-NEXT:    sw a1, 72(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, s0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    lw a2, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 172(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv a2, a0
-; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a1, 44(sp) # 4-byte Folded Spill
-; RV32-NEXT:    lw a3, 56(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a0, 52(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a0, a3, a0
-; RV32-NEXT:    add s2, a2, s2
-; RV32-NEXT:    lw a1, 92(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a2, 64(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a1, a2, a1
-; RV32-NEXT:    add a5, s7, s2
-; RV32-NEXT:    lw a2, 60(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a1, a2, a1
-; RV32-NEXT:    sltu a2, s4, a3
-; RV32-NEXT:    add a0, a0, a2
-; RV32-NEXT:    sltu a2, s6, s10
-; RV32-NEXT:    lw a3, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sw a0, 120(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 68(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a2, a1, a0
+; RV32-NEXT:    lw a0, 120(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, s1
+; RV32-NEXT:    sw a0, 96(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 108(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 100(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a3, a1, a0
+; RV32-NEXT:    lw a0, 96(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a4, a1, a0
+; RV32-NEXT:    lw a0, 76(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a0, a3
+; RV32-NEXT:    lw a0, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a0, a1
+; RV32-NEXT:    add a2, a2, a0
+; RV32-NEXT:    lw a0, 92(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 88(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a1, a0
+; RV32-NEXT:    lw a0, 64(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a0, a1
+; RV32-NEXT:    lw a0, 112(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a5, s2, a0
+; RV32-NEXT:    lw a0, 84(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a0, s2
+; RV32-NEXT:    add a1, a1, a5
 ; RV32-NEXT:    add a2, a3, a2
-; RV32-NEXT:    sltu a3, s8, s9
-; RV32-NEXT:    sltu a4, s3, s8
+; RV32-NEXT:    add a3, a1, a0
+; RV32-NEXT:    lw a0, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 80(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    sw a0, 124(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    add s8, a1, a0
-; RV32-NEXT:    add a2, a2, a4
-; RV32-NEXT:    sltu s6, s5, s4
-; RV32-NEXT:    add a2, s8, a2
-; RV32-NEXT:    sw a5, 64(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu s10, a5, s7
-; RV32-NEXT:    sw s6, 36(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s6, a2, s6
+; RV32-NEXT:    sw a4, 64(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    lw a0, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a4, a0
+; RV32-NEXT:    sw a0, 132(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, a2, a0
+; RV32-NEXT:    lw a0, 180(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sw a1, 32(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
-; RV32-NEXT:    lw s3, 28(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add s3, a0, s3
-; RV32-NEXT:    mv a0, s11
+; RV32-NEXT:    sw a1, 80(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a1, 60(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s2, a0, a1
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s9, a0
-; RV32-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 16(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s3, a0, s3
-; RV32-NEXT:    lw a0, 48(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu s5, s7, a0
-; RV32-NEXT:    add s7, s3, s5
-; RV32-NEXT:    add s7, s7, s6
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    sw a1, 76(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s2, 60(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add s2, a0, s2
+; RV32-NEXT:    lw a0, 116(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    sw a0, 128(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sw s2, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add a0, s2, a0
+; RV32-NEXT:    sw s1, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add a0, a0, s1
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, s0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    lw a2, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s11, a0
-; RV32-NEXT:    sw a1, 60(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    sw a0, 116(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 92(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, s0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    lw a2, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 172(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sw a1, 52(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a0, 48(sp) # 4-byte Folded Spill
-; RV32-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a0, a0, a1
-; RV32-NEXT:    sw s11, 56(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a0, 44(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s11, s11, a0
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    sw a0, 112(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 88(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw s2, 112(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a0, 68(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s2, s2, a0
+; RV32-NEXT:    lw s1, 116(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, s1, s2
+; RV32-NEXT:    sw s1, 168(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, s0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    lw a2, 104(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 152(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sw a1, 28(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
-; RV32-NEXT:    lw s4, 40(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    ori a0, s1, 1
+; RV32-NEXT:    sw a0, 108(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 84(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 72(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 108(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s1, a1, a0
+; RV32-NEXT:    ori a0, s0, 1
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    lw a2, 96(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 156(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    lw a2, 12(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a2, s2, a2
-; RV32-NEXT:    sw s4, 12(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    add s2, s11, s4
-; RV32-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s2, s2, a2
-; RV32-NEXT:    sw s10, 92(sp) # 4-byte Folded Spill
-; RV32-NEXT:    add s10, s2, s10
-; RV32-NEXT:    add s10, s7, s10
-; RV32-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
-; RV32-NEXT:    beq s10, s7, .LBB2_14
+; RV32-NEXT:    sw a0, 100(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a0, 120(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 96(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    sw a0, 120(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a6, 100(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a6, a6, s1
+; RV32-NEXT:    lw a2, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a2, a6
+; RV32-NEXT:    lw a7, 120(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a7, a2, a7
+; RV32-NEXT:    add a2, a7, a0
+; RV32-NEXT:    add a2, a4, a2
+; RV32-NEXT:    beq a2, a4, .LBB2_14
 ; RV32-NEXT:  # %bb.13: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sltu a0, s10, s7
-; RV32-NEXT:    sw a0, 92(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a0, a2, a4
 ; RV32-NEXT:  .LBB2_14: # %fp-to-i-if-exp.large
-; RV32-NEXT:    lw a0, 80(sp) # 4-byte Folded Reload
-; RV32-NEXT:    srli a0, a0, 1
-; RV32-NEXT:    lw a2, 68(sp) # 4-byte Folded Reload
-; RV32-NEXT:    srli a1, a2, 1
-; RV32-NEXT:    beq s7, s3, .LBB2_16
+; RV32-NEXT:    lw a5, 104(sp) # 4-byte Folded Reload
+; RV32-NEXT:    srli t1, a5, 1
+; RV32-NEXT:    lw t4, 52(sp) # 4-byte Folded Reload
+; RV32-NEXT:    srli t0, t4, 1
+; RV32-NEXT:    lw t3, 44(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw t2, 48(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw t5, 56(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    beq a4, a5, .LBB2_16
 ; RV32-NEXT:  # %bb.15: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sltu s5, s7, s3
+; RV32-NEXT:    sltu a4, a4, a5
+; RV32-NEXT:    sw a4, 128(sp) # 4-byte Folded Spill
 ; RV32-NEXT:  .LBB2_16: # %fp-to-i-if-exp.large
-; RV32-NEXT:    lw a4, 84(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sll a2, a2, a4
-; RV32-NEXT:    lw a5, 76(sp) # 4-byte Folded Reload
-; RV32-NEXT:    srl a3, a0, a5
-; RV32-NEXT:    lw a0, 72(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sll a0, a0, a4
-; RV32-NEXT:    srl a1, a1, a5
-; RV32-NEXT:    lw s7, 36(sp) # 4-byte Folded Reload
-; RV32-NEXT:    beq s6, s8, .LBB2_18
+; RV32-NEXT:    sw a7, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a6, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s1, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s2, 72(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 68(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sll a7, t4, t3
+; RV32-NEXT:    srl t1, t1, t2
+; RV32-NEXT:    sll a1, t5, t3
+; RV32-NEXT:    srl a3, t0, t2
+; RV32-NEXT:    lw a6, 80(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw t2, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 76(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw t0, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw t4, 40(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw t5, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw t3, 60(sp) # 4-byte Folded Reload
+; RV32-NEXT:    beq t5, t4, .LBB2_18
 ; RV32-NEXT:  # %bb.17: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sltu s7, s6, s8
+; RV32-NEXT:    sltu t4, t5, t4
+; RV32-NEXT:    sw t4, 124(sp) # 4-byte Folded Spill
 ; RV32-NEXT:  .LBB2_18: # %fp-to-i-if-exp.large
-; RV32-NEXT:    or s8, a2, a3
-; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a2, a3, a2
-; RV32-NEXT:    sltu a3, s3, s9
-; RV32-NEXT:    lw a4, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sw a0, 132(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a2, 96(sp) # 4-byte Folded Spill
+; RV32-NEXT:    or a0, a7, t1
+; RV32-NEXT:    sw a0, 104(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a0, t3, t2
+; RV32-NEXT:    sltu a2, a5, t0
+; RV32-NEXT:    add a0, a6, a0
 ; RV32-NEXT:    add a2, a4, a2
-; RV32-NEXT:    lw a4, 8(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a3, a4, a3
-; RV32-NEXT:    or s3, a0, a1
-; RV32-NEXT:    add s6, a2, a3
-; RV32-NEXT:    lw a0, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    or a1, a1, a3
+; RV32-NEXT:    sw a1, 80(sp) # 4-byte Folded Spill
+; RV32-NEXT:    add s1, a0, a2
+; RV32-NEXT:    lw a0, 180(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s1
+; RV32-NEXT:    mv a2, s0
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add a0, a0, s7
-; RV32-NEXT:    add a0, a0, s5
-; RV32-NEXT:    add s5, a0, s6
-; RV32-NEXT:    lw s7, 40(sp) # 4-byte Folded Reload
-; RV32-NEXT:    beq s2, s11, .LBB2_20
+; RV32-NEXT:    lw a1, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a0, a1
+; RV32-NEXT:    lw a0, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    add a0, a0, s1
+; RV32-NEXT:    sw a0, 128(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    beq a0, a1, .LBB2_20
 ; RV32-NEXT:  # %bb.19: # %fp-to-i-if-exp.large
-; RV32-NEXT:    sltu s7, s2, s11
+; RV32-NEXT:    lw a1, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a0, a1
+; RV32-NEXT:    sw a0, 120(sp) # 4-byte Folded Spill
 ; RV32-NEXT:  .LBB2_20: # %fp-to-i-if-exp.large
-; RV32-NEXT:    lw a0, 104(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 96(sp) # 4-byte Folded Reload
-; RV32-NEXT:    mv a1, s6
-; RV32-NEXT:    mv a2, s1
-; RV32-NEXT:    mv a3, s1
+; RV32-NEXT:    lw a0, 152(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 156(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    mv a3, s0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s9, a0
-; RV32-NEXT:    mv a0, s8
-; RV32-NEXT:    mv a1, s3
-; RV32-NEXT:    ori a2, s1, 1
-; RV32-NEXT:    mv a3, s1
+; RV32-NEXT:    mv s2, a0
+; RV32-NEXT:    lw a0, 104(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 80(sp) # 4-byte Folded Reload
+; RV32-NEXT:    ori a2, s0, 1
+; RV32-NEXT:    mv a3, s0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a2, 12(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a1, a2, a1
-; RV32-NEXT:    lw a2, 4(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a2, s4, a2
-; RV32-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a1, a3, a1
-; RV32-NEXT:    lw a3, 0(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a2, a3, a2
-; RV32-NEXT:    add s9, a0, s9
-; RV32-NEXT:    add s2, a1, a2
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    lw a1, 108(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a3, a2, a1
+; RV32-NEXT:    lw a1, 100(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a2, a2, a1
+; RV32-NEXT:    lw a1, 84(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a1, a3
+; RV32-NEXT:    lw a1, 68(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a1, a2
+; RV32-NEXT:    add s2, a0, s2
+; RV32-NEXT:    add s1, a3, a1
+; RV32-NEXT:    mv a0, s0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s6
+; RV32-NEXT:    lw a2, 156(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add a0, a0, s9
-; RV32-NEXT:    add s3, a0, s2
-; RV32-NEXT:    lw a0, 100(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a1, 128(sp) # 4-byte Folded Reload
-; RV32-NEXT:    mv a2, s1
-; RV32-NEXT:    mv a3, s1
+; RV32-NEXT:    add a0, a0, s2
+; RV32-NEXT:    add a0, a0, s1
+; RV32-NEXT:    sw a0, 156(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 176(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 180(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    mv a3, s0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s2, a0
-; RV32-NEXT:    lw a0, 124(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 132(sp) # 4-byte Folded Reload
-; RV32-NEXT:    mv a1, s6
-; RV32-NEXT:    lw a2, 88(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv s1, a0
+; RV32-NEXT:    lw a0, 172(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 188(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 164(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    lw a1, 56(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a1, s11, a1
-; RV32-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a3, 44(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a2, a3, a2
-; RV32-NEXT:    lw a3, 60(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a1, a3, a1
-; RV32-NEXT:    lw a3, 52(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a2, a3, a2
-; RV32-NEXT:    add s2, s2, a0
-; RV32-NEXT:    add s4, a2, a1
-; RV32-NEXT:    mv a0, s1
+; RV32-NEXT:    lw a1, 116(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 168(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a3, a2, a1
+; RV32-NEXT:    lw a1, 112(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 72(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a2, a2, a1
+; RV32-NEXT:    lw a1, 92(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a1, a3
+; RV32-NEXT:    lw a1, 88(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a1, a2
+; RV32-NEXT:    add s1, s1, a0
+; RV32-NEXT:    add s2, a1, a3
+; RV32-NEXT:    mv a0, s0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s6
+; RV32-NEXT:    lw a2, 188(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    lw a1, 92(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add s5, s5, a1
-; RV32-NEXT:    add a0, a0, s2
-; RV32-NEXT:    add a0, a0, s4
-; RV32-NEXT:    add a0, a0, s3
-; RV32-NEXT:    add a0, a0, s7
-; RV32-NEXT:    add a0, s5, a0
+; RV32-NEXT:    lw a1, 132(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 128(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a2, a1
+; RV32-NEXT:    add a0, a0, s1
+; RV32-NEXT:    add s2, a0, s2
+; RV32-NEXT:    lw a0, 156(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add s2, s2, a0
+; RV32-NEXT:    lw a0, 120(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, s2, a0
+; RV32-NEXT:    add a0, a1, a0
 ; RV32-NEXT:    zext.b a0, a0
-; RV32-NEXT:    lw a1, 120(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a2, 116(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a3, 112(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a4, 108(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s8, 64(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 148(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 160(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 144(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 140(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a6, 136(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 96(sp) # 4-byte Folded Reload
 ; RV32-NEXT:  .LBB2_21: # %fp-to-i-cleanup
-; RV32-NEXT:    sw a1, 0(s0)
-; RV32-NEXT:    sw a2, 4(s0)
-; RV32-NEXT:    sw a3, 8(s0)
-; RV32-NEXT:    sw a4, 12(s0)
-; RV32-NEXT:    sw s8, 16(s0)
-; RV32-NEXT:    sw s10, 20(s0)
-; RV32-NEXT:    sb a0, 24(s0)
-; RV32-NEXT:    lw ra, 252(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 248(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 244(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 240(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 236(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s4, 232(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s5, 228(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 224(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s7, 220(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s8, 216(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s9, 212(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s10, 208(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s11, 204(sp) # 4-byte Folded Reload
-; RV32-NEXT:    addi sp, sp, 256
+; RV32-NEXT:    sw a3, 0(a1)
+; RV32-NEXT:    sw a4, 4(a1)
+; RV32-NEXT:    sw a5, 8(a1)
+; RV32-NEXT:    sw a6, 12(a1)
+; RV32-NEXT:    sw s2, 16(a1)
+; RV32-NEXT:    sw a2, 20(a1)
+; RV32-NEXT:    sb a0, 24(a1)
+; RV32-NEXT:    lw ra, 268(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s0, 264(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s1, 260(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s2, 256(sp) # 4-byte Folded Reload
+; RV32-NEXT:    addi sp, sp, 272
 ; RV32-NEXT:    ret
   %1 = load float, ptr %in, align 4
   %2 = fptosi float %1 to i200
diff --git a/llvm/test/CodeGen/RISCV/calling-conv-half.ll b/llvm/test/CodeGen/RISCV/calling-conv-half.ll
index 4c3fb428bc9cf5..63a23f5361bffb 100644
--- a/llvm/test/CodeGen/RISCV/calling-conv-half.ll
+++ b/llvm/test/CodeGen/RISCV/calling-conv-half.ll
@@ -685,14 +685,12 @@ define i32 @callee_half_on_stack_exhausted_gprs_fprs(i32 %a, float %fa, i32 %b,
 ; RV32I:       # %bb.0:
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    lw s0, 40(sp)
 ; RV32I-NEXT:    lhu a0, 48(sp)
 ; RV32I-NEXT:    call __extendhfsf2
 ; RV32I-NEXT:    call __fixsfsi
-; RV32I-NEXT:    add a0, s0, a0
+; RV32I-NEXT:    lw a1, 40(sp)
+; RV32I-NEXT:    add a0, a1, a0
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -700,14 +698,12 @@ define i32 @callee_half_on_stack_exhausted_gprs_fprs(i32 %a, float %fa, i32 %b,
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    addi sp, sp, -16
 ; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    lw s0, 64(sp)
 ; RV64I-NEXT:    lhu a0, 80(sp)
 ; RV64I-NEXT:    call __extendhfsf2
 ; RV64I-NEXT:    call __fixsfdi
-; RV64I-NEXT:    addw a0, s0, a0
+; RV64I-NEXT:    lw a1, 64(sp)
+; RV64I-NEXT:    addw a0, a1, a0
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    ret
 ;
@@ -715,15 +711,13 @@ define i32 @callee_half_on_stack_exhausted_gprs_fprs(i32 %a, float %fa, i32 %b,
 ; RV32IF:       # %bb.0:
 ; RV32IF-NEXT:    addi sp, sp, -16
 ; RV32IF-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    lw s0, 40(sp)
 ; RV32IF-NEXT:    lhu a0, 48(sp)
 ; RV32IF-NEXT:    call __extendhfsf2
 ; RV32IF-NEXT:    fmv.w.x fa5, a0
-; RV32IF-NEXT:    fcvt.w.s a0, fa5, rtz
-; RV32IF-NEXT:    add a0, s0, a0
+; RV32IF-NEXT:    fcvt.w.s a1, fa5, rtz
+; RV32IF-NEXT:    lw a0, 40(sp)
+; RV32IF-NEXT:    add a0, a0, a1
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    addi sp, sp, 16
 ; RV32IF-NEXT:    ret
 ;
@@ -731,15 +725,13 @@ define i32 @callee_half_on_stack_exhausted_gprs_fprs(i32 %a, float %fa, i32 %b,
 ; RV64IF:       # %bb.0:
 ; RV64IF-NEXT:    addi sp, sp, -16
 ; RV64IF-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64IF-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64IF-NEXT:    lw s0, 64(sp)
 ; RV64IF-NEXT:    lhu a0, 80(sp)
 ; RV64IF-NEXT:    call __extendhfsf2
 ; RV64IF-NEXT:    fmv.w.x fa5, a0
-; RV64IF-NEXT:    fcvt.l.s a0, fa5, rtz
-; RV64IF-NEXT:    addw a0, s0, a0
+; RV64IF-NEXT:    fcvt.l.s a1, fa5, rtz
+; RV64IF-NEXT:    lw a0, 64(sp)
+; RV64IF-NEXT:    addw a0, a0, a1
 ; RV64IF-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64IF-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64IF-NEXT:    addi sp, sp, 16
 ; RV64IF-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/clmulh.ll b/llvm/test/CodeGen/RISCV/clmulh.ll
index e59a414706c874..8d1e9a430b6b46 100644
--- a/llvm/test/CodeGen/RISCV/clmulh.ll
+++ b/llvm/test/CodeGen/RISCV/clmulh.ll
@@ -1641,16 +1641,16 @@ define i32 @clmulh_i32(i32 %a, i32 %b) nounwind {
 ; RV32IM-NEXT:    and t2, a7, t0
 ; RV32IM-NEXT:    and t3, a6, t0
 ; RV32IM-NEXT:    and t4, a7, a0
-; RV32IM-NEXT:    mul t5, t2, t1
+; RV32IM-NEXT:    mul s1, t2, t1
 ; RV32IM-NEXT:    mul t6, t4, t3
-; RV32IM-NEXT:    lui s0, 559241
-; RV32IM-NEXT:    lui s1, 279620
-; RV32IM-NEXT:    addi s0, s0, -1912
-; RV32IM-NEXT:    addi s1, s1, 1092
-; RV32IM-NEXT:    and s2, a6, s0
-; RV32IM-NEXT:    and s3, a7, s1
-; RV32IM-NEXT:    and a6, a6, s1
-; RV32IM-NEXT:    and a7, a7, s0
+; RV32IM-NEXT:    lui t5, 559241
+; RV32IM-NEXT:    lui s0, 279620
+; RV32IM-NEXT:    addi t5, t5, -1912
+; RV32IM-NEXT:    addi s0, s0, 1092
+; RV32IM-NEXT:    and s2, a6, t5
+; RV32IM-NEXT:    and s3, a7, s0
+; RV32IM-NEXT:    and a6, a6, s0
+; RV32IM-NEXT:    and a7, a7, t5
 ; RV32IM-NEXT:    mul s4, s3, s2
 ; RV32IM-NEXT:    mul s5, a7, a6
 ; RV32IM-NEXT:    mul s6, t2, s2
@@ -1665,22 +1665,22 @@ define i32 @clmulh_i32(i32 %a, i32 %b) nounwind {
 ; RV32IM-NEXT:    mul s2, a7, s2
 ; RV32IM-NEXT:    mul t3, s3, t3
 ; RV32IM-NEXT:    mul a7, a7, t1
-; RV32IM-NEXT:    xor t1, t6, t5
-; RV32IM-NEXT:    xor t5, s4, s5
-; RV32IM-NEXT:    xor t6, s7, s6
+; RV32IM-NEXT:    xor t1, t6, s1
+; RV32IM-NEXT:    xor t6, s4, s5
+; RV32IM-NEXT:    xor s1, s7, s6
 ; RV32IM-NEXT:    xor s3, s8, s9
-; RV32IM-NEXT:    xor t1, t1, t5
-; RV32IM-NEXT:    xor t5, t6, s3
+; RV32IM-NEXT:    xor t1, t1, t6
+; RV32IM-NEXT:    xor t6, s1, s3
 ; RV32IM-NEXT:    and t0, t1, t0
-; RV32IM-NEXT:    and a0, t5, a0
+; RV32IM-NEXT:    and a0, t6, a0
 ; RV32IM-NEXT:    xor t1, s11, s10
 ; RV32IM-NEXT:    xor t2, t2, s2
 ; RV32IM-NEXT:    xor a6, t4, a6
 ; RV32IM-NEXT:    xor a7, t3, a7
 ; RV32IM-NEXT:    xor t1, t1, t2
 ; RV32IM-NEXT:    xor a6, a6, a7
-; RV32IM-NEXT:    and a7, t1, s1
-; RV32IM-NEXT:    and a6, a6, s0
+; RV32IM-NEXT:    and a7, t1, s0
+; RV32IM-NEXT:    and a6, a6, t5
 ; RV32IM-NEXT:    or a0, a0, t0
 ; RV32IM-NEXT:    or a6, a7, a6
 ; RV32IM-NEXT:    or a0, a0, a6
@@ -1875,16 +1875,16 @@ define i32 @clmulh_i32(i32 %a, i32 %b) nounwind {
 ; RV32IMZBS-NEXT:    and t2, a7, t0
 ; RV32IMZBS-NEXT:    and t3, a6, t0
 ; RV32IMZBS-NEXT:    and t4, a7, a0
-; RV32IMZBS-NEXT:    mul t5, t2, t1
+; RV32IMZBS-NEXT:    mul s1, t2, t1
 ; RV32IMZBS-NEXT:    mul t6, t4, t3
-; RV32IMZBS-NEXT:    lui s0, 559241
-; RV32IMZBS-NEXT:    lui s1, 279620
-; RV32IMZBS-NEXT:    addi s0, s0, -1912
-; RV32IMZBS-NEXT:    addi s1, s1, 1092
-; RV32IMZBS-NEXT:    and s2, a6, s0
-; RV32IMZBS-NEXT:    and s3, a7, s1
-; RV32IMZBS-NEXT:    and a6, a6, s1
-; RV32IMZBS-NEXT:    and a7, a7, s0
+; RV32IMZBS-NEXT:    lui t5, 559241
+; RV32IMZBS-NEXT:    lui s0, 279620
+; RV32IMZBS-NEXT:    addi t5, t5, -1912
+; RV32IMZBS-NEXT:    addi s0, s0, 1092
+; RV32IMZBS-NEXT:    and s2, a6, t5
+; RV32IMZBS-NEXT:    and s3, a7, s0
+; RV32IMZBS-NEXT:    and a6, a6, s0
+; RV32IMZBS-NEXT:    and a7, a7, t5
 ; RV32IMZBS-NEXT:    mul s4, s3, s2
 ; RV32IMZBS-NEXT:    mul s5, a7, a6
 ; RV32IMZBS-NEXT:    mul s6, t2, s2
@@ -1899,22 +1899,22 @@ define i32 @clmulh_i32(i32 %a, i32 %b) nounwind {
 ; RV32IMZBS-NEXT:    mul s2, a7, s2
 ; RV32IMZBS-NEXT:    mul t3, s3, t3
 ; RV32IMZBS-NEXT:    mul a7, a7, t1
-; RV32IMZBS-NEXT:    xor t1, t6, t5
-; RV32IMZBS-NEXT:    xor t5, s4, s5
-; RV32IMZBS-NEXT:    xor t6, s7, s6
+; RV32IMZBS-NEXT:    xor t1, t6, s1
+; RV32IMZBS-NEXT:    xor t6, s4, s5
+; RV32IMZBS-NEXT:    xor s1, s7, s6
 ; RV32IMZBS-NEXT:    xor s3, s8, s9
-; RV32IMZBS-NEXT:    xor t1, t1, t5
-; RV32IMZBS-NEXT:    xor t5, t6, s3
+; RV32IMZBS-NEXT:    xor t1, t1, t6
+; RV32IMZBS-NEXT:    xor t6, s1, s3
 ; RV32IMZBS-NEXT:    and t0, t1, t0
-; RV32IMZBS-NEXT:    and a0, t5, a0
+; RV32IMZBS-NEXT:    and a0, t6, a0
 ; RV32IMZBS-NEXT:    xor t1, s11, s10
 ; RV32IMZBS-NEXT:    xor t2, t2, s2
 ; RV32IMZBS-NEXT:    xor a6, t4, a6
 ; RV32IMZBS-NEXT:    xor a7, t3, a7
 ; RV32IMZBS-NEXT:    xor t1, t1, t2
 ; RV32IMZBS-NEXT:    xor a6, a6, a7
-; RV32IMZBS-NEXT:    and a7, t1, s1
-; RV32IMZBS-NEXT:    and a6, a6, s0
+; RV32IMZBS-NEXT:    and a7, t1, s0
+; RV32IMZBS-NEXT:    and a6, a6, t5
 ; RV32IMZBS-NEXT:    or a0, a0, t0
 ; RV32IMZBS-NEXT:    or a6, a7, a6
 ; RV32IMZBS-NEXT:    or a0, a0, a6
@@ -2122,16 +2122,16 @@ define i32 @clmulh_i32(i32 %a, i32 %b) nounwind {
 ; RV32IMZVBC-NEXT:    and t2, a7, t0
 ; RV32IMZVBC-NEXT:    and t3, a6, t0
 ; RV32IMZVBC-NEXT:    and t4, a7, a0
-; RV32IMZVBC-NEXT:    mul t5, t2, t1
+; RV32IMZVBC-NEXT:    mul s1, t2, t1
 ; RV32IMZVBC-NEXT:    mul t6, t4, t3
-; RV32IMZVBC-NEXT:    lui s0, 559241
-; RV32IMZVBC-NEXT:    lui s1, 279620
-; RV32IMZVBC-NEXT:    addi s0, s0, -1912
-; RV32IMZVBC-NEXT:    addi s1, s1, 1092
-; RV32IMZVBC-NEXT:    and s2, a6, s0
-; RV32IMZVBC-NEXT:    and s3, a7, s1
-; RV32IMZVBC-NEXT:    and a6, a6, s1
-; RV32IMZVBC-NEXT:    and a7, a7, s0
+; RV32IMZVBC-NEXT:    lui t5, 559241
+; RV32IMZVBC-NEXT:    lui s0, 279620
+; RV32IMZVBC-NEXT:    addi t5, t5, -1912
+; RV32IMZVBC-NEXT:    addi s0, s0, 1092
+; RV32IMZVBC-NEXT:    and s2, a6, t5
+; RV32IMZVBC-NEXT:    and s3, a7, s0
+; RV32IMZVBC-NEXT:    and a6, a6, s0
+; RV32IMZVBC-NEXT:    and a7, a7, t5
 ; RV32IMZVBC-NEXT:    mul s4, s3, s2
 ; RV32IMZVBC-NEXT:    mul s5, a7, a6
 ; RV32IMZVBC-NEXT:    mul s6, t2, s2
@@ -2146,22 +2146,22 @@ define i32 @clmulh_i32(i32 %a, i32 %b) nounwind {
 ; RV32IMZVBC-NEXT:    mul s2, a7, s2
 ; RV32IMZVBC-NEXT:    mul t3, s3, t3
 ; RV32IMZVBC-NEXT:    mul a7, a7, t1
-; RV32IMZVBC-NEXT:    xor t1, t6, t5
-; RV32IMZVBC-NEXT:    xor t5, s4, s5
-; RV32IMZVBC-NEXT:    xor t6, s7, s6
+; RV32IMZVBC-NEXT:    xor t1, t6, s1
+; RV32IMZVBC-NEXT:    xor t6, s4, s5
+; RV32IMZVBC-NEXT:    xor s1, s7, s6
 ; RV32IMZVBC-NEXT:    xor s3, s8, s9
-; RV32IMZVBC-NEXT:    xor t1, t1, t5
-; RV32IMZVBC-NEXT:    xor t5, t6, s3
+; RV32IMZVBC-NEXT:    xor t1, t1, t6
+; RV32IMZVBC-NEXT:    xor t6, s1, s3
 ; RV32IMZVBC-NEXT:    and t0, t1, t0
-; RV32IMZVBC-NEXT:    and a0, t5, a0
+; RV32IMZVBC-NEXT:    and a0, t6, a0
 ; RV32IMZVBC-NEXT:    xor t1, s11, s10
 ; RV32IMZVBC-NEXT:    xor t2, t2, s2
 ; RV32IMZVBC-NEXT:    xor a6, t4, a6
 ; RV32IMZVBC-NEXT:    xor a7, t3, a7
 ; RV32IMZVBC-NEXT:    xor t1, t1, t2
 ; RV32IMZVBC-NEXT:    xor a6, a6, a7
-; RV32IMZVBC-NEXT:    and a7, t1, s1
-; RV32IMZVBC-NEXT:    and a6, a6, s0
+; RV32IMZVBC-NEXT:    and a7, t1, s0
+; RV32IMZVBC-NEXT:    and a6, a6, t5
 ; RV32IMZVBC-NEXT:    or a0, a0, t0
 ; RV32IMZVBC-NEXT:    or a6, a7, a6
 ; RV32IMZVBC-NEXT:    or a0, a0, a6
diff --git a/llvm/test/CodeGen/RISCV/clmulr.ll b/llvm/test/CodeGen/RISCV/clmulr.ll
index 2d8df3dff44a87..8bfc16e5b81e12 100644
--- a/llvm/test/CodeGen/RISCV/clmulr.ll
+++ b/llvm/test/CodeGen/RISCV/clmulr.ll
@@ -1456,16 +1456,16 @@ define i32 @clmulr_i32(i32 %a, i32 %b) nounwind {
 ; RV32IM-NEXT:    and t1, a6, a7
 ; RV32IM-NEXT:    and t2, a5, a7
 ; RV32IM-NEXT:    and t3, a6, a0
-; RV32IM-NEXT:    mul t4, t1, t0
-; RV32IM-NEXT:    mul t5, t3, t2
-; RV32IM-NEXT:    lui t6, 559241
-; RV32IM-NEXT:    lui s0, 279620
-; RV32IM-NEXT:    addi t6, t6, -1912
-; RV32IM-NEXT:    addi s0, s0, 1092
-; RV32IM-NEXT:    and s1, a5, t6
-; RV32IM-NEXT:    and s2, a6, s0
-; RV32IM-NEXT:    and a5, a5, s0
-; RV32IM-NEXT:    and a6, a6, t6
+; RV32IM-NEXT:    mul t6, t1, t0
+; RV32IM-NEXT:    mul s0, t3, t2
+; RV32IM-NEXT:    lui t5, 559241
+; RV32IM-NEXT:    lui t4, 279620
+; RV32IM-NEXT:    addi t5, t5, -1912
+; RV32IM-NEXT:    addi t4, t4, 1092
+; RV32IM-NEXT:    and s1, a5, t5
+; RV32IM-NEXT:    and s2, a6, t4
+; RV32IM-NEXT:    and a5, a5, t4
+; RV32IM-NEXT:    and a6, a6, t5
 ; RV32IM-NEXT:    mul s3, s2, s1
 ; RV32IM-NEXT:    mul s4, a6, a5
 ; RV32IM-NEXT:    mul s5, t1, s1
@@ -1480,22 +1480,22 @@ define i32 @clmulr_i32(i32 %a, i32 %b) nounwind {
 ; RV32IM-NEXT:    mul s1, a6, s1
 ; RV32IM-NEXT:    mul t2, s2, t2
 ; RV32IM-NEXT:    mul a6, a6, t0
-; RV32IM-NEXT:    xor t0, t5, t4
-; RV32IM-NEXT:    xor t4, s3, s4
-; RV32IM-NEXT:    xor t5, s6, s5
+; RV32IM-NEXT:    xor t0, s0, t6
+; RV32IM-NEXT:    xor t6, s3, s4
+; RV32IM-NEXT:    xor s0, s6, s5
 ; RV32IM-NEXT:    xor s2, s7, s8
-; RV32IM-NEXT:    xor t0, t0, t4
-; RV32IM-NEXT:    xor t4, t5, s2
+; RV32IM-NEXT:    xor t0, t0, t6
+; RV32IM-NEXT:    xor t6, s0, s2
 ; RV32IM-NEXT:    and a7, t0, a7
-; RV32IM-NEXT:    and a0, t4, a0
+; RV32IM-NEXT:    and a0, t6, a0
 ; RV32IM-NEXT:    xor t0, s10, s9
 ; RV32IM-NEXT:    xor t1, t1, s1
 ; RV32IM-NEXT:    xor a5, t3, a5
 ; RV32IM-NEXT:    xor a6, t2, a6
 ; RV32IM-NEXT:    xor t0, t0, t1
 ; RV32IM-NEXT:    xor a5, a5, a6
-; RV32IM-NEXT:    and a6, t0, s0
-; RV32IM-NEXT:    and a5, a5, t6
+; RV32IM-NEXT:    and a6, t0, t4
+; RV32IM-NEXT:    and a5, a5, t5
 ; RV32IM-NEXT:    or a0, a0, a7
 ; RV32IM-NEXT:    or a5, a6, a5
 ; RV32IM-NEXT:    or a0, a0, a5
@@ -1686,16 +1686,16 @@ define i32 @clmulr_i32(i32 %a, i32 %b) nounwind {
 ; RV32IMZBS-NEXT:    and t1, a6, a7
 ; RV32IMZBS-NEXT:    and t2, a5, a7
 ; RV32IMZBS-NEXT:    and t3, a6, a0
-; RV32IMZBS-NEXT:    mul t4, t1, t0
-; RV32IMZBS-NEXT:    mul t5, t3, t2
-; RV32IMZBS-NEXT:    lui t6, 559241
-; RV32IMZBS-NEXT:    lui s0, 279620
-; RV32IMZBS-NEXT:    addi t6, t6, -1912
-; RV32IMZBS-NEXT:    addi s0, s0, 1092
-; RV32IMZBS-NEXT:    and s1, a5, t6
-; RV32IMZBS-NEXT:    and s2, a6, s0
-; RV32IMZBS-NEXT:    and a5, a5, s0
-; RV32IMZBS-NEXT:    and a6, a6, t6
+; RV32IMZBS-NEXT:    mul t6, t1, t0
+; RV32IMZBS-NEXT:    mul s0, t3, t2
+; RV32IMZBS-NEXT:    lui t5, 559241
+; RV32IMZBS-NEXT:    lui t4, 279620
+; RV32IMZBS-NEXT:    addi t5, t5, -1912
+; RV32IMZBS-NEXT:    addi t4, t4, 1092
+; RV32IMZBS-NEXT:    and s1, a5, t5
+; RV32IMZBS-NEXT:    and s2, a6, t4
+; RV32IMZBS-NEXT:    and a5, a5, t4
+; RV32IMZBS-NEXT:    and a6, a6, t5
 ; RV32IMZBS-NEXT:    mul s3, s2, s1
 ; RV32IMZBS-NEXT:    mul s4, a6, a5
 ; RV32IMZBS-NEXT:    mul s5, t1, s1
@@ -1710,22 +1710,22 @@ define i32 @clmulr_i32(i32 %a, i32 %b) nounwind {
 ; RV32IMZBS-NEXT:    mul s1, a6, s1
 ; RV32IMZBS-NEXT:    mul t2, s2, t2
 ; RV32IMZBS-NEXT:    mul a6, a6, t0
-; RV32IMZBS-NEXT:    xor t0, t5, t4
-; RV32IMZBS-NEXT:    xor t4, s3, s4
-; RV32IMZBS-NEXT:    xor t5, s6, s5
+; RV32IMZBS-NEXT:    xor t0, s0, t6
+; RV32IMZBS-NEXT:    xor t6, s3, s4
+; RV32IMZBS-NEXT:    xor s0, s6, s5
 ; RV32IMZBS-NEXT:    xor s2, s7, s8
-; RV32IMZBS-NEXT:    xor t0, t0, t4
-; RV32IMZBS-NEXT:    xor t4, t5, s2
+; RV32IMZBS-NEXT:    xor t0, t0, t6
+; RV32IMZBS-NEXT:    xor t6, s0, s2
 ; RV32IMZBS-NEXT:    and a7, t0, a7
-; RV32IMZBS-NEXT:    and a0, t4, a0
+; RV32IMZBS-NEXT:    and a0, t6, a0
 ; RV32IMZBS-NEXT:    xor t0, s10, s9
 ; RV32IMZBS-NEXT:    xor t1, t1, s1
 ; RV32IMZBS-NEXT:    xor a5, t3, a5
 ; RV32IMZBS-NEXT:    xor a6, t2, a6
 ; RV32IMZBS-NEXT:    xor t0, t0, t1
 ; RV32IMZBS-NEXT:    xor a5, a5, a6
-; RV32IMZBS-NEXT:    and a6, t0, s0
-; RV32IMZBS-NEXT:    and a5, a5, t6
+; RV32IMZBS-NEXT:    and a6, t0, t4
+; RV32IMZBS-NEXT:    and a5, a5, t5
 ; RV32IMZBS-NEXT:    or a0, a0, a7
 ; RV32IMZBS-NEXT:    or a5, a6, a5
 ; RV32IMZBS-NEXT:    or a0, a0, a5
@@ -8216,18 +8216,19 @@ define void @commutative_clmulr_v2i64(<2 x i64> %x, <2 x i64> %y, ptr %p0, ptr %
 ; RV32IMZBC-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
 ; RV32IMZBC-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
 ; RV32IMZBC-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32IMZBC-NEXT:    lw a4, 4(a0)
+; RV32IMZBC-NEXT:    sw s3, 0(sp) # 4-byte Folded Spill
+; RV32IMZBC-NEXT:    lw s3, 4(a0)
 ; RV32IMZBC-NEXT:    lui a6, 16
 ; RV32IMZBC-NEXT:    lw t2, 8(a0)
 ; RV32IMZBC-NEXT:    lw a5, 12(a0)
 ; RV32IMZBC-NEXT:    lw t4, 0(a1)
 ; RV32IMZBC-NEXT:    addi a6, a6, -256
-; RV32IMZBC-NEXT:    srli a7, a4, 8
-; RV32IMZBC-NEXT:    srli t0, a4, 24
-; RV32IMZBC-NEXT:    and t1, a4, a6
+; RV32IMZBC-NEXT:    srli a7, s3, 8
+; RV32IMZBC-NEXT:    srli t0, s3, 24
+; RV32IMZBC-NEXT:    and t1, s3, a6
 ; RV32IMZBC-NEXT:    and a7, a7, a6
 ; RV32IMZBC-NEXT:    slli t1, t1, 8
-; RV32IMZBC-NEXT:    slli t3, a4, 24
+; RV32IMZBC-NEXT:    slli t3, s3, 24
 ; RV32IMZBC-NEXT:    or a7, a7, t0
 ; RV32IMZBC-NEXT:    or t0, t3, t1
 ; RV32IMZBC-NEXT:    or a7, t0, a7
@@ -8250,251 +8251,252 @@ define void @commutative_clmulr_v2i64(<2 x i64> %x, <2 x i64> %y, ptr %p0, ptr %
 ; RV32IMZBC-NEXT:    lw t3, 8(a1)
 ; RV32IMZBC-NEXT:    lw a0, 12(a1)
 ; RV32IMZBC-NEXT:    srli s0, t6, 1
-; RV32IMZBC-NEXT:    lui a1, 349525
-; RV32IMZBC-NEXT:    srli s1, t4, 8
-; RV32IMZBC-NEXT:    and s1, s1, a6
-; RV32IMZBC-NEXT:    srli s2, t4, 24
-; RV32IMZBC-NEXT:    or s1, s1, s2
-; RV32IMZBC-NEXT:    and s2, t4, a6
-; RV32IMZBC-NEXT:    slli s2, s2, 8
+; RV32IMZBC-NEXT:    lui s2, 349525
+; RV32IMZBC-NEXT:    srli a1, t4, 8
+; RV32IMZBC-NEXT:    and a1, a1, a6
+; RV32IMZBC-NEXT:    srli a4, t4, 24
+; RV32IMZBC-NEXT:    or a4, a1, a4
+; RV32IMZBC-NEXT:    and a1, t4, a6
+; RV32IMZBC-NEXT:    slli s1, a1, 8
 ; RV32IMZBC-NEXT:    slli t4, t4, 24
-; RV32IMZBC-NEXT:    addi a1, a1, 1365
-; RV32IMZBC-NEXT:    or t4, t4, s2
-; RV32IMZBC-NEXT:    and s0, s0, a1
+; RV32IMZBC-NEXT:    addi a1, s2, 1365
 ; RV32IMZBC-NEXT:    or t4, t4, s1
-; RV32IMZBC-NEXT:    srli s1, t4, 4
+; RV32IMZBC-NEXT:    and s0, s0, a1
+; RV32IMZBC-NEXT:    or a4, t4, a4
+; RV32IMZBC-NEXT:    srli t4, a4, 4
+; RV32IMZBC-NEXT:    and a4, a4, t0
 ; RV32IMZBC-NEXT:    and t4, t4, t0
-; RV32IMZBC-NEXT:    and s1, s1, t0
-; RV32IMZBC-NEXT:    slli t4, t4, 4
+; RV32IMZBC-NEXT:    slli a4, a4, 4
 ; RV32IMZBC-NEXT:    and t6, t6, a1
-; RV32IMZBC-NEXT:    or t4, s1, t4
-; RV32IMZBC-NEXT:    srli s1, t4, 2
+; RV32IMZBC-NEXT:    or a4, t4, a4
+; RV32IMZBC-NEXT:    srli t4, a4, 2
+; RV32IMZBC-NEXT:    and a4, a4, t1
 ; RV32IMZBC-NEXT:    and t4, t4, t1
-; RV32IMZBC-NEXT:    and s1, s1, t1
-; RV32IMZBC-NEXT:    slli t4, t4, 2
+; RV32IMZBC-NEXT:    slli a4, a4, 2
 ; RV32IMZBC-NEXT:    slli t6, t6, 1
-; RV32IMZBC-NEXT:    or s1, s1, t4
+; RV32IMZBC-NEXT:    or a4, t4, a4
 ; RV32IMZBC-NEXT:    or t4, s0, t6
-; RV32IMZBC-NEXT:    srli t6, s1, 1
+; RV32IMZBC-NEXT:    srli t6, a4, 1
 ; RV32IMZBC-NEXT:    and t6, t6, a1
 ; RV32IMZBC-NEXT:    srli s0, t5, 8
 ; RV32IMZBC-NEXT:    and s0, s0, a6
-; RV32IMZBC-NEXT:    srli s2, t5, 24
-; RV32IMZBC-NEXT:    or s0, s0, s2
-; RV32IMZBC-NEXT:    and s2, t5, a6
-; RV32IMZBC-NEXT:    slli s2, s2, 8
+; RV32IMZBC-NEXT:    srli s1, t5, 24
+; RV32IMZBC-NEXT:    or s0, s0, s1
+; RV32IMZBC-NEXT:    and s1, t5, a6
+; RV32IMZBC-NEXT:    slli s1, s1, 8
 ; RV32IMZBC-NEXT:    slli t5, t5, 24
-; RV32IMZBC-NEXT:    and s1, s1, a1
-; RV32IMZBC-NEXT:    or t5, t5, s2
-; RV32IMZBC-NEXT:    slli s1, s1, 1
+; RV32IMZBC-NEXT:    and a4, a4, a1
+; RV32IMZBC-NEXT:    or t5, t5, s1
+; RV32IMZBC-NEXT:    slli a4, a4, 1
 ; RV32IMZBC-NEXT:    or t5, t5, s0
 ; RV32IMZBC-NEXT:    srli s0, t5, 4
 ; RV32IMZBC-NEXT:    and t5, t5, t0
 ; RV32IMZBC-NEXT:    and s0, s0, t0
 ; RV32IMZBC-NEXT:    slli t5, t5, 4
-; RV32IMZBC-NEXT:    or t6, t6, s1
+; RV32IMZBC-NEXT:    or a4, t6, a4
 ; RV32IMZBC-NEXT:    or t5, s0, t5
-; RV32IMZBC-NEXT:    srli s0, t5, 2
+; RV32IMZBC-NEXT:    srli t6, t5, 2
 ; RV32IMZBC-NEXT:    and t5, t5, t1
-; RV32IMZBC-NEXT:    and s0, s0, t1
+; RV32IMZBC-NEXT:    and t6, t6, t1
 ; RV32IMZBC-NEXT:    slli t5, t5, 2
-; RV32IMZBC-NEXT:    or t5, s0, t5
-; RV32IMZBC-NEXT:    srli s0, a7, 8
-; RV32IMZBC-NEXT:    and s0, s0, a6
-; RV32IMZBC-NEXT:    srli s1, a7, 24
-; RV32IMZBC-NEXT:    or s0, s0, s1
-; RV32IMZBC-NEXT:    and s1, a7, a6
-; RV32IMZBC-NEXT:    slli s1, s1, 8
-; RV32IMZBC-NEXT:    slli s2, a7, 24
-; RV32IMZBC-NEXT:    or s1, s2, s1
-; RV32IMZBC-NEXT:    srli s2, t5, 1
-; RV32IMZBC-NEXT:    and s2, s2, a1
+; RV32IMZBC-NEXT:    or t5, t6, t5
+; RV32IMZBC-NEXT:    srli t6, a7, 8
+; RV32IMZBC-NEXT:    and t6, t6, a6
+; RV32IMZBC-NEXT:    srli s0, a7, 24
+; RV32IMZBC-NEXT:    or t6, t6, s0
+; RV32IMZBC-NEXT:    and s0, a7, a6
+; RV32IMZBC-NEXT:    slli s0, s0, 8
+; RV32IMZBC-NEXT:    slli s1, a7, 24
 ; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    srli s1, s0, 4
+; RV32IMZBC-NEXT:    srli s1, t5, 1
+; RV32IMZBC-NEXT:    and s1, s1, a1
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    srli s0, t6, 4
+; RV32IMZBC-NEXT:    and t6, t6, t0
 ; RV32IMZBC-NEXT:    and s0, s0, t0
-; RV32IMZBC-NEXT:    and s1, s1, t0
-; RV32IMZBC-NEXT:    slli s0, s0, 4
+; RV32IMZBC-NEXT:    slli t6, t6, 4
 ; RV32IMZBC-NEXT:    and t5, t5, a1
-; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    srli s1, s0, 2
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    srli s0, t6, 2
+; RV32IMZBC-NEXT:    and t6, t6, t1
 ; RV32IMZBC-NEXT:    and s0, s0, t1
-; RV32IMZBC-NEXT:    and s1, s1, t1
-; RV32IMZBC-NEXT:    slli s0, s0, 2
+; RV32IMZBC-NEXT:    slli t6, t6, 2
 ; RV32IMZBC-NEXT:    slli t5, t5, 1
-; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    srli s1, s0, 1
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    srli s0, t6, 1
+; RV32IMZBC-NEXT:    and t6, t6, a1
 ; RV32IMZBC-NEXT:    and s0, s0, a1
-; RV32IMZBC-NEXT:    and s1, s1, a1
-; RV32IMZBC-NEXT:    slli s0, s0, 1
-; RV32IMZBC-NEXT:    or t5, s2, t5
-; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    clmul t6, t6, t4
-; RV32IMZBC-NEXT:    clmul t5, s0, t5
-; RV32IMZBC-NEXT:    clmulh t4, s0, t4
-; RV32IMZBC-NEXT:    xor t5, t5, t6
-; RV32IMZBC-NEXT:    xor t4, t4, t5
-; RV32IMZBC-NEXT:    srli t5, t4, 8
+; RV32IMZBC-NEXT:    slli t6, t6, 1
+; RV32IMZBC-NEXT:    or t5, s1, t5
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    clmul a4, a4, t4
+; RV32IMZBC-NEXT:    clmul t5, t6, t5
+; RV32IMZBC-NEXT:    clmulh t4, t6, t4
+; RV32IMZBC-NEXT:    xor a4, t5, a4
+; RV32IMZBC-NEXT:    xor a4, t4, a4
+; RV32IMZBC-NEXT:    srli t4, a4, 8
+; RV32IMZBC-NEXT:    and t4, t4, a6
+; RV32IMZBC-NEXT:    srli t5, a4, 24
+; RV32IMZBC-NEXT:    and t6, a4, a6
+; RV32IMZBC-NEXT:    slli a4, a4, 24
+; RV32IMZBC-NEXT:    slli t6, t6, 8
+; RV32IMZBC-NEXT:    or t4, t4, t5
+; RV32IMZBC-NEXT:    or a4, a4, t6
+; RV32IMZBC-NEXT:    or a4, a4, t4
+; RV32IMZBC-NEXT:    srli t4, a4, 4
+; RV32IMZBC-NEXT:    and t4, t4, t0
+; RV32IMZBC-NEXT:    and a4, a4, t0
+; RV32IMZBC-NEXT:    srli t5, a5, 8
+; RV32IMZBC-NEXT:    slli a4, a4, 4
 ; RV32IMZBC-NEXT:    and t5, t5, a6
-; RV32IMZBC-NEXT:    srli t6, t4, 24
-; RV32IMZBC-NEXT:    and s0, t4, a6
-; RV32IMZBC-NEXT:    slli t4, t4, 24
+; RV32IMZBC-NEXT:    srli t6, a5, 24
+; RV32IMZBC-NEXT:    and s0, a5, a6
 ; RV32IMZBC-NEXT:    slli s0, s0, 8
+; RV32IMZBC-NEXT:    slli s1, a5, 24
 ; RV32IMZBC-NEXT:    or t5, t5, t6
-; RV32IMZBC-NEXT:    or t4, t4, s0
-; RV32IMZBC-NEXT:    or t4, t4, t5
-; RV32IMZBC-NEXT:    srli t5, t4, 4
-; RV32IMZBC-NEXT:    and t5, t5, t0
-; RV32IMZBC-NEXT:    and t4, t4, t0
-; RV32IMZBC-NEXT:    srli t6, a5, 8
-; RV32IMZBC-NEXT:    slli t4, t4, 4
-; RV32IMZBC-NEXT:    and t6, t6, a6
-; RV32IMZBC-NEXT:    srli s0, a5, 24
-; RV32IMZBC-NEXT:    and s1, a5, a6
-; RV32IMZBC-NEXT:    slli s1, s1, 8
-; RV32IMZBC-NEXT:    slli s2, a5, 24
-; RV32IMZBC-NEXT:    or t6, t6, s0
-; RV32IMZBC-NEXT:    or s0, s2, s1
-; RV32IMZBC-NEXT:    or t4, t5, t4
-; RV32IMZBC-NEXT:    or t5, s0, t6
-; RV32IMZBC-NEXT:    srli t6, t5, 4
-; RV32IMZBC-NEXT:    and t5, t5, t0
-; RV32IMZBC-NEXT:    and t6, t6, t0
-; RV32IMZBC-NEXT:    slli s0, t5, 4
+; RV32IMZBC-NEXT:    or s0, s1, s0
+; RV32IMZBC-NEXT:    or t4, t4, a4
+; RV32IMZBC-NEXT:    or a4, s0, t5
+; RV32IMZBC-NEXT:    srli t5, a4, 4
+; RV32IMZBC-NEXT:    and a4, a4, t0
+; RV32IMZBC-NEXT:    and t6, t5, t0
+; RV32IMZBC-NEXT:    slli a4, a4, 4
 ; RV32IMZBC-NEXT:    srli t5, t4, 2
-; RV32IMZBC-NEXT:    or t6, t6, s0
-; RV32IMZBC-NEXT:    srli s0, t6, 2
+; RV32IMZBC-NEXT:    or a4, t6, a4
+; RV32IMZBC-NEXT:    srli t6, a4, 2
+; RV32IMZBC-NEXT:    and a4, a4, t1
 ; RV32IMZBC-NEXT:    and t6, t6, t1
-; RV32IMZBC-NEXT:    and s0, s0, t1
-; RV32IMZBC-NEXT:    slli t6, t6, 2
-; RV32IMZBC-NEXT:    or t6, s0, t6
-; RV32IMZBC-NEXT:    srli s0, t3, 8
-; RV32IMZBC-NEXT:    and s0, s0, a6
-; RV32IMZBC-NEXT:    srli s1, t3, 24
-; RV32IMZBC-NEXT:    or s0, s0, s1
-; RV32IMZBC-NEXT:    and s1, t3, a6
-; RV32IMZBC-NEXT:    slli s1, s1, 8
+; RV32IMZBC-NEXT:    slli a4, a4, 2
+; RV32IMZBC-NEXT:    or a4, t6, a4
+; RV32IMZBC-NEXT:    srli t6, t3, 8
+; RV32IMZBC-NEXT:    and t6, t6, a6
+; RV32IMZBC-NEXT:    srli s0, t3, 24
+; RV32IMZBC-NEXT:    or t6, t6, s0
+; RV32IMZBC-NEXT:    and s0, t3, a6
+; RV32IMZBC-NEXT:    slli s0, s0, 8
 ; RV32IMZBC-NEXT:    slli t3, t3, 24
-; RV32IMZBC-NEXT:    srli s2, t6, 1
-; RV32IMZBC-NEXT:    or t3, t3, s1
-; RV32IMZBC-NEXT:    and s1, s2, a1
+; RV32IMZBC-NEXT:    srli s1, a4, 1
 ; RV32IMZBC-NEXT:    or t3, t3, s0
-; RV32IMZBC-NEXT:    srli s0, t3, 4
+; RV32IMZBC-NEXT:    and s1, s1, a1
+; RV32IMZBC-NEXT:    or t3, t3, t6
+; RV32IMZBC-NEXT:    srli t6, t3, 4
 ; RV32IMZBC-NEXT:    and t3, t3, t0
-; RV32IMZBC-NEXT:    and s0, s0, t0
+; RV32IMZBC-NEXT:    and t6, t6, t0
 ; RV32IMZBC-NEXT:    slli t3, t3, 4
-; RV32IMZBC-NEXT:    and t6, t6, a1
-; RV32IMZBC-NEXT:    or t3, s0, t3
-; RV32IMZBC-NEXT:    srli s0, t3, 2
+; RV32IMZBC-NEXT:    and a4, a4, a1
+; RV32IMZBC-NEXT:    or t3, t6, t3
+; RV32IMZBC-NEXT:    srli t6, t3, 2
 ; RV32IMZBC-NEXT:    and t3, t3, t1
-; RV32IMZBC-NEXT:    and s0, s0, t1
+; RV32IMZBC-NEXT:    and t6, t6, t1
 ; RV32IMZBC-NEXT:    slli t3, t3, 2
-; RV32IMZBC-NEXT:    slli t6, t6, 1
-; RV32IMZBC-NEXT:    or s0, s0, t3
-; RV32IMZBC-NEXT:    or t3, s1, t6
-; RV32IMZBC-NEXT:    srli t6, s0, 1
-; RV32IMZBC-NEXT:    and t6, t6, a1
-; RV32IMZBC-NEXT:    srli s1, t2, 8
-; RV32IMZBC-NEXT:    and s1, s1, a6
-; RV32IMZBC-NEXT:    srli s2, t2, 24
-; RV32IMZBC-NEXT:    or s1, s1, s2
-; RV32IMZBC-NEXT:    and s2, t2, a6
-; RV32IMZBC-NEXT:    slli s2, s2, 8
+; RV32IMZBC-NEXT:    slli a4, a4, 1
+; RV32IMZBC-NEXT:    or t6, t6, t3
+; RV32IMZBC-NEXT:    or t3, s1, a4
+; RV32IMZBC-NEXT:    srli a4, t6, 1
+; RV32IMZBC-NEXT:    and a4, a4, a1
+; RV32IMZBC-NEXT:    srli s0, t2, 8
+; RV32IMZBC-NEXT:    and s0, s0, a6
+; RV32IMZBC-NEXT:    srli s1, t2, 24
+; RV32IMZBC-NEXT:    or s0, s0, s1
+; RV32IMZBC-NEXT:    and s1, t2, a6
+; RV32IMZBC-NEXT:    slli s1, s1, 8
 ; RV32IMZBC-NEXT:    slli t2, t2, 24
-; RV32IMZBC-NEXT:    and s0, s0, a1
-; RV32IMZBC-NEXT:    or t2, t2, s2
-; RV32IMZBC-NEXT:    slli s0, s0, 1
+; RV32IMZBC-NEXT:    and t6, t6, a1
 ; RV32IMZBC-NEXT:    or t2, t2, s1
-; RV32IMZBC-NEXT:    srli s1, t2, 4
+; RV32IMZBC-NEXT:    slli t6, t6, 1
+; RV32IMZBC-NEXT:    or t2, t2, s0
+; RV32IMZBC-NEXT:    srli s0, t2, 4
 ; RV32IMZBC-NEXT:    and t2, t2, t0
-; RV32IMZBC-NEXT:    and s1, s1, t0
+; RV32IMZBC-NEXT:    and s0, s0, t0
 ; RV32IMZBC-NEXT:    slli t2, t2, 4
-; RV32IMZBC-NEXT:    or t6, t6, s0
-; RV32IMZBC-NEXT:    or t2, s1, t2
-; RV32IMZBC-NEXT:    srli s0, t2, 2
+; RV32IMZBC-NEXT:    or a4, a4, t6
+; RV32IMZBC-NEXT:    or t2, s0, t2
+; RV32IMZBC-NEXT:    srli t6, t2, 2
 ; RV32IMZBC-NEXT:    and t2, t2, t1
-; RV32IMZBC-NEXT:    and s0, s0, t1
+; RV32IMZBC-NEXT:    and t6, t6, t1
 ; RV32IMZBC-NEXT:    slli t2, t2, 2
-; RV32IMZBC-NEXT:    or t2, s0, t2
-; RV32IMZBC-NEXT:    srli s0, a0, 8
-; RV32IMZBC-NEXT:    and s0, s0, a6
-; RV32IMZBC-NEXT:    srli s1, a0, 24
-; RV32IMZBC-NEXT:    or s0, s0, s1
-; RV32IMZBC-NEXT:    and s1, a0, a6
-; RV32IMZBC-NEXT:    slli s1, s1, 8
-; RV32IMZBC-NEXT:    slli s2, a0, 24
-; RV32IMZBC-NEXT:    or s1, s2, s1
-; RV32IMZBC-NEXT:    srli s2, t2, 1
-; RV32IMZBC-NEXT:    and s2, s2, a1
+; RV32IMZBC-NEXT:    or t2, t6, t2
+; RV32IMZBC-NEXT:    srli t6, a0, 8
+; RV32IMZBC-NEXT:    and t6, t6, a6
+; RV32IMZBC-NEXT:    srli s0, a0, 24
+; RV32IMZBC-NEXT:    or t6, t6, s0
+; RV32IMZBC-NEXT:    and s0, a0, a6
+; RV32IMZBC-NEXT:    slli s0, s0, 8
+; RV32IMZBC-NEXT:    slli s1, a0, 24
 ; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    srli s1, s0, 4
+; RV32IMZBC-NEXT:    srli s1, t2, 1
+; RV32IMZBC-NEXT:    and s1, s1, a1
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    srli s0, t6, 4
+; RV32IMZBC-NEXT:    and t6, t6, t0
 ; RV32IMZBC-NEXT:    and s0, s0, t0
-; RV32IMZBC-NEXT:    and s1, s1, t0
-; RV32IMZBC-NEXT:    slli s0, s0, 4
+; RV32IMZBC-NEXT:    slli t6, t6, 4
 ; RV32IMZBC-NEXT:    and t2, t2, a1
-; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    srli s1, s0, 2
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    srli s0, t6, 2
+; RV32IMZBC-NEXT:    and t6, t6, t1
 ; RV32IMZBC-NEXT:    and s0, s0, t1
-; RV32IMZBC-NEXT:    and s1, s1, t1
-; RV32IMZBC-NEXT:    slli s0, s0, 2
+; RV32IMZBC-NEXT:    slli t6, t6, 2
 ; RV32IMZBC-NEXT:    slli t2, t2, 1
-; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    srli s1, s0, 1
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    srli s0, t6, 1
+; RV32IMZBC-NEXT:    and t6, t6, a1
 ; RV32IMZBC-NEXT:    and s0, s0, a1
-; RV32IMZBC-NEXT:    and s1, s1, a1
-; RV32IMZBC-NEXT:    slli s0, s0, 1
-; RV32IMZBC-NEXT:    or t2, s2, t2
-; RV32IMZBC-NEXT:    or s0, s1, s0
-; RV32IMZBC-NEXT:    clmul t6, t6, t3
-; RV32IMZBC-NEXT:    clmul t2, s0, t2
+; RV32IMZBC-NEXT:    slli t6, t6, 1
+; RV32IMZBC-NEXT:    or t2, s1, t2
+; RV32IMZBC-NEXT:    or t6, s0, t6
+; RV32IMZBC-NEXT:    clmul a4, a4, t3
+; RV32IMZBC-NEXT:    clmul t2, t6, t2
 ; RV32IMZBC-NEXT:    and t5, t5, t1
-; RV32IMZBC-NEXT:    clmulh t3, s0, t3
+; RV32IMZBC-NEXT:    clmulh t3, t6, t3
 ; RV32IMZBC-NEXT:    and t4, t4, t1
-; RV32IMZBC-NEXT:    xor t2, t2, t6
+; RV32IMZBC-NEXT:    xor a4, t2, a4
 ; RV32IMZBC-NEXT:    slli t4, t4, 2
-; RV32IMZBC-NEXT:    xor t2, t3, t2
-; RV32IMZBC-NEXT:    or t3, t5, t4
-; RV32IMZBC-NEXT:    srli t4, t2, 8
-; RV32IMZBC-NEXT:    srli t5, t3, 1
-; RV32IMZBC-NEXT:    and t4, t4, a6
-; RV32IMZBC-NEXT:    srli t6, t2, 24
-; RV32IMZBC-NEXT:    and a6, t2, a6
-; RV32IMZBC-NEXT:    slli t2, t2, 24
+; RV32IMZBC-NEXT:    xor a4, t3, a4
+; RV32IMZBC-NEXT:    or t2, t5, t4
+; RV32IMZBC-NEXT:    srli t3, a4, 8
+; RV32IMZBC-NEXT:    srli t4, t2, 1
+; RV32IMZBC-NEXT:    and t3, t3, a6
+; RV32IMZBC-NEXT:    srli t5, a4, 24
+; RV32IMZBC-NEXT:    and a6, a4, a6
+; RV32IMZBC-NEXT:    slli a4, a4, 24
 ; RV32IMZBC-NEXT:    slli a6, a6, 8
-; RV32IMZBC-NEXT:    or t4, t4, t6
-; RV32IMZBC-NEXT:    or a6, t2, a6
-; RV32IMZBC-NEXT:    and t2, t5, a1
-; RV32IMZBC-NEXT:    or a6, a6, t4
-; RV32IMZBC-NEXT:    srli t4, a6, 4
-; RV32IMZBC-NEXT:    and a6, a6, t0
-; RV32IMZBC-NEXT:    and t0, t4, t0
-; RV32IMZBC-NEXT:    slli a6, a6, 4
-; RV32IMZBC-NEXT:    and t3, t3, a1
-; RV32IMZBC-NEXT:    or a6, t0, a6
-; RV32IMZBC-NEXT:    srli t0, a6, 2
-; RV32IMZBC-NEXT:    and a6, a6, t1
+; RV32IMZBC-NEXT:    or t3, t3, t5
+; RV32IMZBC-NEXT:    or a4, a4, a6
+; RV32IMZBC-NEXT:    and a6, t4, a1
+; RV32IMZBC-NEXT:    or a4, a4, t3
+; RV32IMZBC-NEXT:    srli t3, a4, 4
+; RV32IMZBC-NEXT:    and a4, a4, t0
+; RV32IMZBC-NEXT:    and t0, t3, t0
+; RV32IMZBC-NEXT:    slli a4, a4, 4
+; RV32IMZBC-NEXT:    and t2, t2, a1
+; RV32IMZBC-NEXT:    or a4, t0, a4
+; RV32IMZBC-NEXT:    srli t0, a4, 2
+; RV32IMZBC-NEXT:    and a4, a4, t1
 ; RV32IMZBC-NEXT:    and t0, t0, t1
-; RV32IMZBC-NEXT:    slli a6, a6, 2
-; RV32IMZBC-NEXT:    slli t3, t3, 1
-; RV32IMZBC-NEXT:    or a6, t0, a6
-; RV32IMZBC-NEXT:    or t0, t2, t3
-; RV32IMZBC-NEXT:    srli t1, a6, 1
-; RV32IMZBC-NEXT:    and t1, t1, a1
-; RV32IMZBC-NEXT:    and a1, a6, a1
+; RV32IMZBC-NEXT:    slli a4, a4, 2
+; RV32IMZBC-NEXT:    slli t2, t2, 1
+; RV32IMZBC-NEXT:    or a4, t0, a4
+; RV32IMZBC-NEXT:    or a6, a6, t2
+; RV32IMZBC-NEXT:    srli t0, a4, 1
+; RV32IMZBC-NEXT:    and t0, t0, a1
+; RV32IMZBC-NEXT:    and a1, a4, a1
 ; RV32IMZBC-NEXT:    slli a1, a1, 1
-; RV32IMZBC-NEXT:    clmulr a4, a7, a4
-; RV32IMZBC-NEXT:    or a1, t1, a1
+; RV32IMZBC-NEXT:    clmulr a4, a7, s3
+; RV32IMZBC-NEXT:    or a1, t0, a1
 ; RV32IMZBC-NEXT:    clmulr a0, a0, a5
-; RV32IMZBC-NEXT:    sw t0, 0(a2)
+; RV32IMZBC-NEXT:    sw a6, 0(a2)
 ; RV32IMZBC-NEXT:    sw a4, 4(a2)
 ; RV32IMZBC-NEXT:    sw a1, 8(a2)
 ; RV32IMZBC-NEXT:    sw a0, 12(a2)
-; RV32IMZBC-NEXT:    sw t0, 0(a3)
+; RV32IMZBC-NEXT:    sw a6, 0(a3)
 ; RV32IMZBC-NEXT:    sw a4, 4(a3)
 ; RV32IMZBC-NEXT:    sw a1, 8(a3)
 ; RV32IMZBC-NEXT:    sw a0, 12(a3)
 ; RV32IMZBC-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
 ; RV32IMZBC-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
 ; RV32IMZBC-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
+; RV32IMZBC-NEXT:    lw s3, 0(sp) # 4-byte Folded Reload
 ; RV32IMZBC-NEXT:    addi sp, sp, 16
 ; RV32IMZBC-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/condops.ll b/llvm/test/CodeGen/RISCV/condops.ll
index a8de56b79644d0..59a1e29c0eb989 100644
--- a/llvm/test/CodeGen/RISCV/condops.ll
+++ b/llvm/test/CodeGen/RISCV/condops.ll
@@ -3073,100 +3073,95 @@ define i64 @setune_64(float %a, float %b, i64 %rs1, i64 %rs2) {
 define signext i16 @numsignbits(i16 signext %0, i16 signext %1, i16 signext %2, i16 signext %3) nounwind {
 ; RV32I-LABEL: numsignbits:
 ; RV32I:       # %bb.0:
-; RV32I-NEXT:    addi sp, sp, -16
-; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a3
 ; RV32I-NEXT:    beqz a0, .LBB60_2
 ; RV32I-NEXT:  # %bb.1:
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv a3, a2
 ; RV32I-NEXT:  .LBB60_2:
 ; RV32I-NEXT:    beqz a1, .LBB60_4
 ; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, a3
+; RV32I-NEXT:    sw a3, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call bat
-; RV32I-NEXT:  .LBB60_4:
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
+; RV32I-NEXT:  .LBB60_4:
+; RV32I-NEXT:    mv a0, a3
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: numsignbits:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -16
-; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s0, a3
 ; RV64I-NEXT:    beqz a0, .LBB60_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    mv s0, a2
+; RV64I-NEXT:    mv a3, a2
 ; RV64I-NEXT:  .LBB60_2:
 ; RV64I-NEXT:    beqz a1, .LBB60_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    addi sp, sp, -16
+; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, a3
+; RV64I-NEXT:    sd a3, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    call bat
-; RV64I-NEXT:  .LBB60_4:
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    ld a3, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
+; RV64I-NEXT:  .LBB60_4:
+; RV64I-NEXT:    mv a0, a3
 ; RV64I-NEXT:    ret
 ;
 ; RV64XTHEADCONDMOV-LABEL: numsignbits:
 ; RV64XTHEADCONDMOV:       # %bb.0:
-; RV64XTHEADCONDMOV-NEXT:    addi sp, sp, -16
-; RV64XTHEADCONDMOV-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64XTHEADCONDMOV-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64XTHEADCONDMOV-NEXT:    mv s0, a2
-; RV64XTHEADCONDMOV-NEXT:    th.mveqz s0, a3, a0
+; RV64XTHEADCONDMOV-NEXT:    th.mveqz a2, a3, a0
 ; RV64XTHEADCONDMOV-NEXT:    beqz a1, .LBB60_2
 ; RV64XTHEADCONDMOV-NEXT:  # %bb.1:
-; RV64XTHEADCONDMOV-NEXT:    mv a0, s0
+; RV64XTHEADCONDMOV-NEXT:    addi sp, sp, -16
+; RV64XTHEADCONDMOV-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
+; RV64XTHEADCONDMOV-NEXT:    mv a0, a2
+; RV64XTHEADCONDMOV-NEXT:    sd a2, 0(sp) # 8-byte Folded Spill
 ; RV64XTHEADCONDMOV-NEXT:    call bat
-; RV64XTHEADCONDMOV-NEXT:  .LBB60_2:
-; RV64XTHEADCONDMOV-NEXT:    mv a0, s0
+; RV64XTHEADCONDMOV-NEXT:    ld a2, 0(sp) # 8-byte Folded Reload
 ; RV64XTHEADCONDMOV-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64XTHEADCONDMOV-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64XTHEADCONDMOV-NEXT:    addi sp, sp, 16
+; RV64XTHEADCONDMOV-NEXT:  .LBB60_2:
+; RV64XTHEADCONDMOV-NEXT:    mv a0, a2
 ; RV64XTHEADCONDMOV-NEXT:    ret
 ;
 ; RV32ZICOND-LABEL: numsignbits:
 ; RV32ZICOND:       # %bb.0:
-; RV32ZICOND-NEXT:    addi sp, sp, -16
-; RV32ZICOND-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32ZICOND-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32ZICOND-NEXT:    czero.eqz a2, a2, a0
-; RV32ZICOND-NEXT:    czero.nez s0, a3, a0
-; RV32ZICOND-NEXT:    or s0, s0, a2
+; RV32ZICOND-NEXT:    czero.nez a0, a3, a0
+; RV32ZICOND-NEXT:    or a0, a0, a2
 ; RV32ZICOND-NEXT:    beqz a1, .LBB60_2
 ; RV32ZICOND-NEXT:  # %bb.1:
-; RV32ZICOND-NEXT:    mv a0, s0
+; RV32ZICOND-NEXT:    addi sp, sp, -16
+; RV32ZICOND-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32ZICOND-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32ZICOND-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
 ; RV32ZICOND-NEXT:    call bat
-; RV32ZICOND-NEXT:  .LBB60_2:
-; RV32ZICOND-NEXT:    mv a0, s0
+; RV32ZICOND-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
 ; RV32ZICOND-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32ZICOND-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32ZICOND-NEXT:    addi sp, sp, 16
+; RV32ZICOND-NEXT:  .LBB60_2:
 ; RV32ZICOND-NEXT:    ret
 ;
 ; RV64ZICOND-LABEL: numsignbits:
 ; RV64ZICOND:       # %bb.0:
-; RV64ZICOND-NEXT:    addi sp, sp, -16
-; RV64ZICOND-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64ZICOND-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
 ; RV64ZICOND-NEXT:    czero.eqz a2, a2, a0
-; RV64ZICOND-NEXT:    czero.nez s0, a3, a0
-; RV64ZICOND-NEXT:    or s0, s0, a2
+; RV64ZICOND-NEXT:    czero.nez a0, a3, a0
+; RV64ZICOND-NEXT:    or a0, a0, a2
 ; RV64ZICOND-NEXT:    beqz a1, .LBB60_2
 ; RV64ZICOND-NEXT:  # %bb.1:
-; RV64ZICOND-NEXT:    mv a0, s0
+; RV64ZICOND-NEXT:    addi sp, sp, -16
+; RV64ZICOND-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
+; RV64ZICOND-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
+; RV64ZICOND-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64ZICOND-NEXT:    call bat
-; RV64ZICOND-NEXT:  .LBB60_2:
-; RV64ZICOND-NEXT:    mv a0, s0
+; RV64ZICOND-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64ZICOND-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64ZICOND-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64ZICOND-NEXT:    addi sp, sp, 16
+; RV64ZICOND-NEXT:  .LBB60_2:
 ; RV64ZICOND-NEXT:    ret
   %5 = icmp eq i16 %0, 0
   %6 = select i1 %5, i16 %3, i16 %2
diff --git a/llvm/test/CodeGen/RISCV/ctlz-cttz-ctpop.ll b/llvm/test/CodeGen/RISCV/ctlz-cttz-ctpop.ll
index 45b29fd970ef3d..99b3f1191957ce 100644
--- a/llvm/test/CodeGen/RISCV/ctlz-cttz-ctpop.ll
+++ b/llvm/test/CodeGen/RISCV/ctlz-cttz-ctpop.ll
@@ -355,55 +355,53 @@ define i32 @test_cttz_i32(i32 %a) nounwind {
 define i64 @test_cttz_i64(i64 %a) nounwind {
 ; RV32I-LABEL: test_cttz_i64:
 ; RV32I:       # %bb.0:
+; RV32I-NEXT:    or a2, a0, a1
+; RV32I-NEXT:    beqz a2, .LBB3_3
+; RV32I-NEXT:  # %bb.1: # %cond.false
 ; RV32I-NEXT:    addi sp, sp, -32
 ; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a1
-; RV32I-NEXT:    or a1, a0, a1
-; RV32I-NEXT:    beqz a1, .LBB3_3
-; RV32I-NEXT:  # %bb.1: # %cond.false
-; RV32I-NEXT:    neg a1, a0
-; RV32I-NEXT:    and a1, a0, a1
-; RV32I-NEXT:    lui s2, 30667
-; RV32I-NEXT:    addi s2, s2, 1329
-; RV32I-NEXT:    mv s4, a0
-; RV32I-NEXT:    mv a0, a1
-; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    neg a2, a0
+; RV32I-NEXT:    and a2, a0, a2
+; RV32I-NEXT:    lui a3, 30667
+; RV32I-NEXT:    addi a3, a3, 1329
+; RV32I-NEXT:    sw a3, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, a2
+; RV32I-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    neg a0, s0
-; RV32I-NEXT:    lui s3, %hi(.LCPI3_0)
-; RV32I-NEXT:    addi s3, s3, %lo(.LCPI3_0)
-; RV32I-NEXT:    and a0, s0, a0
-; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    neg a1, a0
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a0, a1
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s4, .LBB3_4
+; RV32I-NEXT:    lw a1, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    bnez a1, .LBB3_4
 ; RV32I-NEXT:  # %bb.2: # %cond.false
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s3, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    j .LBB3_5
 ; RV32I-NEXT:  .LBB3_3:
 ; RV32I-NEXT:    li a0, 64
-; RV32I-NEXT:    j .LBB3_5
-; RV32I-NEXT:  .LBB3_4:
-; RV32I-NEXT:    srli s1, s1, 27
-; RV32I-NEXT:    add s1, s3, s1
-; RV32I-NEXT:    lbu a0, 0(s1)
-; RV32I-NEXT:  .LBB3_5: # %cond.end
 ; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    ret
+; RV32I-NEXT:  .LBB3_4:
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
+; RV32I-NEXT:  .LBB3_5: # %cond.false
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_cttz_i64:
@@ -799,42 +797,41 @@ define i64 @test_cttz_i64_zero_poison(i64 %a) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s1, a1
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv s0, a1
+; RV32I-NEXT:    mv s1, a0
 ; RV32I-NEXT:    neg a0, a0
-; RV32I-NEXT:    and a0, s2, a0
-; RV32I-NEXT:    lui s3, 30667
-; RV32I-NEXT:    addi s3, s3, 1329
-; RV32I-NEXT:    mv a1, s3
-; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    neg a0, s1
-; RV32I-NEXT:    lui s4, %hi(.LCPI7_0)
-; RV32I-NEXT:    addi s4, s4, %lo(.LCPI7_0)
 ; RV32I-NEXT:    and a0, s1, a0
-; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:    lui s2, 30667
+; RV32I-NEXT:    addi s2, s2, 1329
+; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    call __mulsi3
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    neg a0, s0
+; RV32I-NEXT:    and a0, s0, a0
+; RV32I-NEXT:    mv a1, s2
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s2, .LBB7_2
+; RV32I-NEXT:    bnez s1, .LBB7_2
 ; RV32I-NEXT:  # %bb.1:
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s4, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI7_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI7_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    j .LBB7_3
 ; RV32I-NEXT:  .LBB7_2:
-; RV32I-NEXT:    srli s0, s0, 27
-; RV32I-NEXT:    add s0, s4, s0
-; RV32I-NEXT:    lbu a0, 0(s0)
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI7_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI7_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:  .LBB7_3:
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/ctz_zero_return_test.ll b/llvm/test/CodeGen/RISCV/ctz_zero_return_test.ll
index 6e1b4eb7e9ede1..2f11428f30cd0d 100644
--- a/llvm/test/CodeGen/RISCV/ctz_zero_return_test.ll
+++ b/llvm/test/CodeGen/RISCV/ctz_zero_return_test.ll
@@ -37,37 +37,38 @@ define signext i32 @ctz_dereferencing_pointer(ptr %b) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    lw s2, 0(a0)
-; RV32I-NEXT:    lw s4, 4(a0)
+; RV32I-NEXT:    lw s1, 0(a0)
+; RV32I-NEXT:    lw s2, 4(a0)
+; RV32I-NEXT:    neg a0, s1
+; RV32I-NEXT:    and a0, s1, a0
+; RV32I-NEXT:    lui s0, 30667
+; RV32I-NEXT:    addi s0, s0, 1329
+; RV32I-NEXT:    mv a1, s0
+; RV32I-NEXT:    call __mulsi3
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    neg a0, s2
 ; RV32I-NEXT:    and a0, s2, a0
-; RV32I-NEXT:    lui s1, 30667
-; RV32I-NEXT:    addi s1, s1, 1329
-; RV32I-NEXT:    mv a1, s1
-; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    neg a0, s4
-; RV32I-NEXT:    lui s3, %hi(.LCPI0_0)
-; RV32I-NEXT:    addi s3, s3, %lo(.LCPI0_0)
-; RV32I-NEXT:    and a0, s4, a0
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s4, .LBB0_3
+; RV32I-NEXT:    bnez s2, .LBB0_3
 ; RV32I-NEXT:  # %bb.1: # %entry
 ; RV32I-NEXT:    li a0, 32
-; RV32I-NEXT:    beqz s2, .LBB0_4
+; RV32I-NEXT:    beqz s1, .LBB0_4
 ; RV32I-NEXT:  .LBB0_2:
-; RV32I-NEXT:    srli s0, s0, 27
-; RV32I-NEXT:    add s0, s3, s0
-; RV32I-NEXT:    lbu a0, 0(s0)
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI0_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI0_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    j .LBB0_5
 ; RV32I-NEXT:  .LBB0_3:
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s3, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI0_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI0_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
-; RV32I-NEXT:    bnez s2, .LBB0_2
+; RV32I-NEXT:    bnez s1, .LBB0_2
 ; RV32I-NEXT:  .LBB0_4: # %entry
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:  .LBB0_5: # %entry
@@ -76,8 +77,6 @@ define signext i32 @ctz_dereferencing_pointer(ptr %b) nounwind {
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -539,35 +538,36 @@ define signext i32 @ctz4(i64 %b) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s2, a1
+; RV32I-NEXT:    mv s1, a1
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    neg a0, a0
 ; RV32I-NEXT:    and a0, s0, a0
-; RV32I-NEXT:    lui s3, 30667
-; RV32I-NEXT:    addi s3, s3, 1329
-; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:    lui s2, 30667
+; RV32I-NEXT:    addi s2, s2, 1329
+; RV32I-NEXT:    mv a1, s2
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    neg a0, s2
-; RV32I-NEXT:    lui s4, %hi(.LCPI6_0)
-; RV32I-NEXT:    addi s4, s4, %lo(.LCPI6_0)
-; RV32I-NEXT:    and a0, s2, a0
-; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    neg a0, s1
+; RV32I-NEXT:    and a0, s1, a0
+; RV32I-NEXT:    mv a1, s2
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s2, .LBB6_3
+; RV32I-NEXT:    bnez s1, .LBB6_3
 ; RV32I-NEXT:  # %bb.1: # %entry
 ; RV32I-NEXT:    li a0, 32
 ; RV32I-NEXT:    beqz s0, .LBB6_4
 ; RV32I-NEXT:  .LBB6_2:
-; RV32I-NEXT:    srli s1, s1, 27
-; RV32I-NEXT:    add s1, s4, s1
-; RV32I-NEXT:    lbu a0, 0(s1)
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI6_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI6_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    j .LBB6_5
 ; RV32I-NEXT:  .LBB6_3:
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s4, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI6_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI6_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    bnez s0, .LBB6_2
 ; RV32I-NEXT:  .LBB6_4: # %entry
@@ -578,8 +578,6 @@ define signext i32 @ctz4(i64 %b) nounwind {
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/double-convert.ll b/llvm/test/CodeGen/RISCV/double-convert.ll
index 3c1cb9b554ac1d..df3b6c2eaef974 100644
--- a/llvm/test/CodeGen/RISCV/double-convert.ll
+++ b/llvm/test/CodeGen/RISCV/double-convert.ll
@@ -205,12 +205,11 @@ define i32 @fcvt_w_d_sat(double %a) nounwind {
 ;
 ; RV64I-LABEL: fcvt_w_d_sat:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, -497
 ; RV64I-NEXT:    slli a1, a1, 53
@@ -219,7 +218,6 @@ define i32 @fcvt_w_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixdfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    lui s3, 524288
 ; RV64I-NEXT:    bgez s2, .LBB3_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    lui s1, 524288
@@ -232,7 +230,8 @@ define i32 @fcvt_w_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    blez a0, .LBB3_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    addiw s1, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addiw s1, s1, -1
 ; RV64I-NEXT:  .LBB3_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -240,12 +239,11 @@ define i32 @fcvt_w_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    snez a0, a0
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i32 @llvm.fptosi.sat.i32.f64(double %a)
@@ -424,20 +422,18 @@ define i32 @fcvt_wu_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gedf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunsdfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    li a0, 1055
 ; RV64I-NEXT:    slli a0, a0, 31
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    slli a1, a0, 21
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    blez a0, .LBB6_2
 ; RV64I-NEXT:  # %bb.1: # %start
@@ -445,14 +441,14 @@ define i32 @fcvt_wu_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    j .LBB6_3
 ; RV64I-NEXT:  .LBB6_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB6_3: # %start
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -684,27 +680,28 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV32IFD-NEXT:    fmv.d fs0, fa0
 ; RV32IFD-NEXT:    fle.d s0, fa5, fa0
 ; RV32IFD-NEXT:    call __fixdfdi
+; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lui a4, 524288
 ; RV32IFD-NEXT:    lui a3, 524288
-; RV32IFD-NEXT:    lui a2, 524288
 ; RV32IFD-NEXT:    beqz s0, .LBB12_2
 ; RV32IFD-NEXT:  # %bb.1: # %start
-; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    mv a3, a1
 ; RV32IFD-NEXT:  .LBB12_2: # %start
 ; RV32IFD-NEXT:    lui a1, %hi(.LCPI12_1)
 ; RV32IFD-NEXT:    fld fa5, %lo(.LCPI12_1)(a1)
 ; RV32IFD-NEXT:    flt.d a1, fa5, fs0
 ; RV32IFD-NEXT:    beqz a1, .LBB12_4
 ; RV32IFD-NEXT:  # %bb.3:
-; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    addi a3, a4, -1
 ; RV32IFD-NEXT:  .LBB12_4: # %start
-; RV32IFD-NEXT:    feq.d a3, fs0, fs0
-; RV32IFD-NEXT:    neg a4, s0
+; RV32IFD-NEXT:    feq.d a4, fs0, fs0
+; RV32IFD-NEXT:    neg a2, a2
 ; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a4, a0
-; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a0, a2, a0
+; RV32IFD-NEXT:    neg a2, a4
 ; RV32IFD-NEXT:    or a0, a1, a0
-; RV32IFD-NEXT:    and a1, a3, a2
-; RV32IFD-NEXT:    and a0, a3, a0
+; RV32IFD-NEXT:    and a1, a2, a3
+; RV32IFD-NEXT:    and a0, a2, a0
 ; RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -734,29 +731,30 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV32IZFINXZDINX-NEXT:    fmv.d s0, a0
 ; RV32IZFINXZDINX-NEXT:    fle.d s2, a4, s0
 ; RV32IZFINXZDINX-NEXT:    call __fixdfdi
+; RV32IZFINXZDINX-NEXT:    mv a2, s2
+; RV32IZFINXZDINX-NEXT:    lui a4, 524288
 ; RV32IZFINXZDINX-NEXT:    lui a3, 524288
-; RV32IZFINXZDINX-NEXT:    lui a2, 524288
 ; RV32IZFINXZDINX-NEXT:    beqz s2, .LBB12_2
 ; RV32IZFINXZDINX-NEXT:  # %bb.1: # %start
-; RV32IZFINXZDINX-NEXT:    mv a2, a1
+; RV32IZFINXZDINX-NEXT:    mv a3, a1
 ; RV32IZFINXZDINX-NEXT:  .LBB12_2: # %start
 ; RV32IZFINXZDINX-NEXT:    lui a1, %hi(.LCPI12_1)
-; RV32IZFINXZDINX-NEXT:    lw a4, %lo(.LCPI12_1)(a1)
+; RV32IZFINXZDINX-NEXT:    lw a6, %lo(.LCPI12_1)(a1)
 ; RV32IZFINXZDINX-NEXT:    addi a1, a1, %lo(.LCPI12_1)
-; RV32IZFINXZDINX-NEXT:    lw a5, 4(a1)
-; RV32IZFINXZDINX-NEXT:    flt.d a1, a4, s0
+; RV32IZFINXZDINX-NEXT:    lw a7, 4(a1)
+; RV32IZFINXZDINX-NEXT:    flt.d a1, a6, s0
 ; RV32IZFINXZDINX-NEXT:    beqz a1, .LBB12_4
 ; RV32IZFINXZDINX-NEXT:  # %bb.3:
-; RV32IZFINXZDINX-NEXT:    addi a2, a3, -1
+; RV32IZFINXZDINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINXZDINX-NEXT:  .LBB12_4: # %start
-; RV32IZFINXZDINX-NEXT:    feq.d a3, s0, s0
-; RV32IZFINXZDINX-NEXT:    neg a4, s2
+; RV32IZFINXZDINX-NEXT:    feq.d a4, s0, s0
+; RV32IZFINXZDINX-NEXT:    neg a2, a2
 ; RV32IZFINXZDINX-NEXT:    neg a1, a1
-; RV32IZFINXZDINX-NEXT:    and a0, a4, a0
-; RV32IZFINXZDINX-NEXT:    neg a3, a3
+; RV32IZFINXZDINX-NEXT:    and a0, a2, a0
+; RV32IZFINXZDINX-NEXT:    neg a2, a4
 ; RV32IZFINXZDINX-NEXT:    or a0, a1, a0
-; RV32IZFINXZDINX-NEXT:    and a1, a3, a2
-; RV32IZFINXZDINX-NEXT:    and a0, a3, a0
+; RV32IZFINXZDINX-NEXT:    and a1, a2, a3
+; RV32IZFINXZDINX-NEXT:    and a0, a2, a0
 ; RV32IZFINXZDINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINXZDINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINXZDINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -837,12 +835,11 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ;
 ; RV64I-LABEL: fcvt_l_d_sat:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, -481
 ; RV64I-NEXT:    slli a1, a1, 53
@@ -851,10 +848,10 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixdfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    li s3, -1
 ; RV64I-NEXT:    bgez s2, .LBB12_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    slli s1, s3, 63
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    slli s1, s1, 63
 ; RV64I-NEXT:  .LBB12_2: # %start
 ; RV64I-NEXT:    li a0, 543
 ; RV64I-NEXT:    slli a0, a0, 53
@@ -863,7 +860,8 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    blez a0, .LBB12_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    srli s1, s3, 1
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    srli s1, s1, 1
 ; RV64I-NEXT:  .LBB12_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -871,12 +869,11 @@ define i64 @fcvt_l_d_sat(double %a) nounwind {
 ; RV64I-NEXT:    snez a0, a0
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i64 @llvm.fptosi.sat.i64.f64(double %a)
@@ -1873,40 +1870,39 @@ define zeroext i16 @fcvt_wu_s_sat_i16(double %a) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s1, a1
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv s0, a1
+; RV32I-NEXT:    mv s1, a0
 ; RV32I-NEXT:    lui a3, 265984
 ; RV32I-NEXT:    addi a3, a3, -32
 ; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    call __gtdf2
-; RV32I-NEXT:    mv s3, a0
-; RV32I-NEXT:    mv a0, s2
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __gedf2
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    mv a0, s2
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    call __fixunsdfsi
 ; RV32I-NEXT:    lui a1, 16
 ; RV32I-NEXT:    addi a1, a1, -1
-; RV32I-NEXT:    blez s3, .LBB28_2
+; RV32I-NEXT:    blez s2, .LBB28_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    mv a0, a1
 ; RV32I-NEXT:    j .LBB28_3
 ; RV32I-NEXT:  .LBB28_2:
-; RV32I-NEXT:    srli s0, s0, 31
-; RV32I-NEXT:    addi s0, s0, -1
-; RV32I-NEXT:    and a0, s0, a0
+; RV32I-NEXT:    lw a2, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a2, a2, 31
+; RV32I-NEXT:    addi a2, a2, -1
+; RV32I-NEXT:    and a0, a2, a0
 ; RV32I-NEXT:  .LBB28_3: # %start
 ; RV32I-NEXT:    and a0, a0, a1
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -1915,19 +1911,17 @@ define zeroext i16 @fcvt_wu_s_sat_i16(double %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gedf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunsdfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a0, 8312
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    slli a1, a0, 37
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    lui a1, 16
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -1936,15 +1930,15 @@ define zeroext i16 @fcvt_wu_s_sat_i16(double %a) nounwind {
 ; RV64I-NEXT:    mv a0, a1
 ; RV64I-NEXT:    j .LBB28_3
 ; RV64I-NEXT:  .LBB28_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a2, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a2, a0
 ; RV64I-NEXT:  .LBB28_3: # %start
 ; RV64I-NEXT:    and a0, a0, a1
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -2250,37 +2244,36 @@ define zeroext i8 @fcvt_wu_s_sat_i8(double %a) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s1, a1
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv s0, a1
+; RV32I-NEXT:    mv s1, a0
 ; RV32I-NEXT:    lui a3, 263934
 ; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    call __gtdf2
-; RV32I-NEXT:    mv s3, a0
-; RV32I-NEXT:    mv a0, s2
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    li a2, 0
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __gedf2
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    mv a0, s2
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    call __fixunsdfsi
-; RV32I-NEXT:    blez s3, .LBB32_2
+; RV32I-NEXT:    blez s2, .LBB32_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    li a0, 255
 ; RV32I-NEXT:    j .LBB32_3
 ; RV32I-NEXT:  .LBB32_2:
-; RV32I-NEXT:    srli s0, s0, 31
-; RV32I-NEXT:    addi s0, s0, -1
-; RV32I-NEXT:    and a0, s0, a0
+; RV32I-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a1, 31
+; RV32I-NEXT:    addi a1, a1, -1
+; RV32I-NEXT:    and a0, a1, a0
 ; RV32I-NEXT:  .LBB32_3: # %start
 ; RV32I-NEXT:    zext.b a0, a0
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -2289,33 +2282,31 @@ define zeroext i8 @fcvt_wu_s_sat_i8(double %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gedf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunsdfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 131967
 ; RV64I-NEXT:    slli a1, a1, 33
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    blez a0, .LBB32_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    li a0, 255
 ; RV64I-NEXT:    j .LBB32_3
 ; RV64I-NEXT:  .LBB32_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB32_3: # %start
 ; RV64I-NEXT:    zext.b a0, a0
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -2405,20 +2396,18 @@ define zeroext i32 @fcvt_wu_d_sat_zext(double %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gedf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunsdfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    li a0, 1055
 ; RV64I-NEXT:    slli a0, a0, 31
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    slli a1, a0, 21
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    blez a0, .LBB33_2
 ; RV64I-NEXT:  # %bb.1: # %start
@@ -2426,16 +2415,16 @@ define zeroext i32 @fcvt_wu_d_sat_zext(double %a) nounwind {
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    j .LBB33_3
 ; RV64I-NEXT:  .LBB33_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB33_3: # %start
 ; RV64I-NEXT:    slli a0, a0, 32
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -2525,12 +2514,11 @@ define signext i32 @fcvt_w_d_sat_sext(double %a) nounwind {
 ;
 ; RV64I-LABEL: fcvt_w_d_sat_sext:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, -497
 ; RV64I-NEXT:    slli a1, a1, 53
@@ -2539,7 +2527,6 @@ define signext i32 @fcvt_w_d_sat_sext(double %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixdfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    lui s3, 524288
 ; RV64I-NEXT:    bgez s2, .LBB34_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    lui s1, 524288
@@ -2552,7 +2539,8 @@ define signext i32 @fcvt_w_d_sat_sext(double %a) nounwind {
 ; RV64I-NEXT:    call __gtdf2
 ; RV64I-NEXT:    blez a0, .LBB34_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    addi s1, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addi s1, s1, -1
 ; RV64I-NEXT:  .LBB34_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -2561,12 +2549,11 @@ define signext i32 @fcvt_w_d_sat_sext(double %a) nounwind {
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i32 @llvm.fptosi.sat.i32.f64(double %a)
diff --git a/llvm/test/CodeGen/RISCV/double-round-conv-sat.ll b/llvm/test/CodeGen/RISCV/double-round-conv-sat.ll
index 3d084c8f4c09d7..788d09e0733ad3 100644
--- a/llvm/test/CodeGen/RISCV/double-round-conv-sat.ll
+++ b/llvm/test/CodeGen/RISCV/double-round-conv-sat.ll
@@ -53,27 +53,28 @@ define i64 @test_floor_si64(double %x) nounwind {
 ; RV32IFD-NEXT:    fmv.d fs0, fa0
 ; RV32IFD-NEXT:    fle.d s0, fa5, fa0
 ; RV32IFD-NEXT:    call __fixdfdi
+; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lui a4, 524288
 ; RV32IFD-NEXT:    lui a3, 524288
-; RV32IFD-NEXT:    lui a2, 524288
 ; RV32IFD-NEXT:    beqz s0, .LBB1_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    mv a3, a1
 ; RV32IFD-NEXT:  .LBB1_2:
 ; RV32IFD-NEXT:    lui a1, %hi(.LCPI1_1)
 ; RV32IFD-NEXT:    fld fa5, %lo(.LCPI1_1)(a1)
 ; RV32IFD-NEXT:    flt.d a1, fa5, fs0
 ; RV32IFD-NEXT:    beqz a1, .LBB1_4
 ; RV32IFD-NEXT:  # %bb.3:
-; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    addi a3, a4, -1
 ; RV32IFD-NEXT:  .LBB1_4:
-; RV32IFD-NEXT:    feq.d a3, fs0, fs0
-; RV32IFD-NEXT:    neg a4, s0
+; RV32IFD-NEXT:    feq.d a4, fs0, fs0
+; RV32IFD-NEXT:    neg a2, a2
 ; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a4, a0
-; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a0, a2, a0
+; RV32IFD-NEXT:    neg a2, a4
 ; RV32IFD-NEXT:    or a0, a1, a0
-; RV32IFD-NEXT:    and a1, a3, a2
-; RV32IFD-NEXT:    and a0, a3, a0
+; RV32IFD-NEXT:    and a1, a2, a3
+; RV32IFD-NEXT:    and a0, a2, a0
 ; RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -300,27 +301,28 @@ define i64 @test_ceil_si64(double %x) nounwind {
 ; RV32IFD-NEXT:    fmv.d fs0, fa0
 ; RV32IFD-NEXT:    fle.d s0, fa5, fa0
 ; RV32IFD-NEXT:    call __fixdfdi
+; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lui a4, 524288
 ; RV32IFD-NEXT:    lui a3, 524288
-; RV32IFD-NEXT:    lui a2, 524288
 ; RV32IFD-NEXT:    beqz s0, .LBB5_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    mv a3, a1
 ; RV32IFD-NEXT:  .LBB5_2:
 ; RV32IFD-NEXT:    lui a1, %hi(.LCPI5_1)
 ; RV32IFD-NEXT:    fld fa5, %lo(.LCPI5_1)(a1)
 ; RV32IFD-NEXT:    flt.d a1, fa5, fs0
 ; RV32IFD-NEXT:    beqz a1, .LBB5_4
 ; RV32IFD-NEXT:  # %bb.3:
-; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    addi a3, a4, -1
 ; RV32IFD-NEXT:  .LBB5_4:
-; RV32IFD-NEXT:    feq.d a3, fs0, fs0
-; RV32IFD-NEXT:    neg a4, s0
+; RV32IFD-NEXT:    feq.d a4, fs0, fs0
+; RV32IFD-NEXT:    neg a2, a2
 ; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a4, a0
-; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a0, a2, a0
+; RV32IFD-NEXT:    neg a2, a4
 ; RV32IFD-NEXT:    or a0, a1, a0
-; RV32IFD-NEXT:    and a1, a3, a2
-; RV32IFD-NEXT:    and a0, a3, a0
+; RV32IFD-NEXT:    and a1, a2, a3
+; RV32IFD-NEXT:    and a0, a2, a0
 ; RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -546,27 +548,28 @@ define i64 @test_trunc_si64(double %x) nounwind {
 ; RV32IFD-NEXT:    fmv.d fs0, fa0
 ; RV32IFD-NEXT:    fle.d s0, fa5, fa0
 ; RV32IFD-NEXT:    call __fixdfdi
+; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lui a4, 524288
 ; RV32IFD-NEXT:    lui a3, 524288
-; RV32IFD-NEXT:    lui a2, 524288
 ; RV32IFD-NEXT:    beqz s0, .LBB9_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    mv a3, a1
 ; RV32IFD-NEXT:  .LBB9_2:
 ; RV32IFD-NEXT:    lui a1, %hi(.LCPI9_1)
 ; RV32IFD-NEXT:    fld fa5, %lo(.LCPI9_1)(a1)
 ; RV32IFD-NEXT:    flt.d a1, fa5, fs0
 ; RV32IFD-NEXT:    beqz a1, .LBB9_4
 ; RV32IFD-NEXT:  # %bb.3:
-; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    addi a3, a4, -1
 ; RV32IFD-NEXT:  .LBB9_4:
-; RV32IFD-NEXT:    feq.d a3, fs0, fs0
-; RV32IFD-NEXT:    neg a4, s0
+; RV32IFD-NEXT:    feq.d a4, fs0, fs0
+; RV32IFD-NEXT:    neg a2, a2
 ; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a4, a0
-; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a0, a2, a0
+; RV32IFD-NEXT:    neg a2, a4
 ; RV32IFD-NEXT:    or a0, a1, a0
-; RV32IFD-NEXT:    and a1, a3, a2
-; RV32IFD-NEXT:    and a0, a3, a0
+; RV32IFD-NEXT:    and a1, a2, a3
+; RV32IFD-NEXT:    and a0, a2, a0
 ; RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -596,29 +599,30 @@ define i64 @test_trunc_si64(double %x) nounwind {
 ; RV32IZFINXZDINX-NEXT:    fmv.d s0, a0
 ; RV32IZFINXZDINX-NEXT:    fle.d s2, a4, s0
 ; RV32IZFINXZDINX-NEXT:    call __fixdfdi
+; RV32IZFINXZDINX-NEXT:    mv a2, s2
+; RV32IZFINXZDINX-NEXT:    lui a4, 524288
 ; RV32IZFINXZDINX-NEXT:    lui a3, 524288
-; RV32IZFINXZDINX-NEXT:    lui a2, 524288
 ; RV32IZFINXZDINX-NEXT:    beqz s2, .LBB9_2
 ; RV32IZFINXZDINX-NEXT:  # %bb.1:
-; RV32IZFINXZDINX-NEXT:    mv a2, a1
+; RV32IZFINXZDINX-NEXT:    mv a3, a1
 ; RV32IZFINXZDINX-NEXT:  .LBB9_2:
 ; RV32IZFINXZDINX-NEXT:    lui a1, %hi(.LCPI9_1)
-; RV32IZFINXZDINX-NEXT:    lw a4, %lo(.LCPI9_1)(a1)
+; RV32IZFINXZDINX-NEXT:    lw a6, %lo(.LCPI9_1)(a1)
 ; RV32IZFINXZDINX-NEXT:    addi a1, a1, %lo(.LCPI9_1)
-; RV32IZFINXZDINX-NEXT:    lw a5, 4(a1)
-; RV32IZFINXZDINX-NEXT:    flt.d a1, a4, s0
+; RV32IZFINXZDINX-NEXT:    lw a7, 4(a1)
+; RV32IZFINXZDINX-NEXT:    flt.d a1, a6, s0
 ; RV32IZFINXZDINX-NEXT:    beqz a1, .LBB9_4
 ; RV32IZFINXZDINX-NEXT:  # %bb.3:
-; RV32IZFINXZDINX-NEXT:    addi a2, a3, -1
+; RV32IZFINXZDINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINXZDINX-NEXT:  .LBB9_4:
-; RV32IZFINXZDINX-NEXT:    feq.d a3, s0, s0
-; RV32IZFINXZDINX-NEXT:    neg a4, s2
+; RV32IZFINXZDINX-NEXT:    feq.d a4, s0, s0
+; RV32IZFINXZDINX-NEXT:    neg a2, a2
 ; RV32IZFINXZDINX-NEXT:    neg a1, a1
-; RV32IZFINXZDINX-NEXT:    and a0, a4, a0
-; RV32IZFINXZDINX-NEXT:    neg a3, a3
+; RV32IZFINXZDINX-NEXT:    and a0, a2, a0
+; RV32IZFINXZDINX-NEXT:    neg a2, a4
 ; RV32IZFINXZDINX-NEXT:    or a0, a1, a0
-; RV32IZFINXZDINX-NEXT:    and a1, a3, a2
-; RV32IZFINXZDINX-NEXT:    and a0, a3, a0
+; RV32IZFINXZDINX-NEXT:    and a1, a2, a3
+; RV32IZFINXZDINX-NEXT:    and a0, a2, a0
 ; RV32IZFINXZDINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINXZDINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINXZDINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -799,27 +803,28 @@ define i64 @test_round_si64(double %x) nounwind {
 ; RV32IFD-NEXT:    fmv.d fs0, fa0
 ; RV32IFD-NEXT:    fle.d s0, fa5, fa0
 ; RV32IFD-NEXT:    call __fixdfdi
+; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lui a4, 524288
 ; RV32IFD-NEXT:    lui a3, 524288
-; RV32IFD-NEXT:    lui a2, 524288
 ; RV32IFD-NEXT:    beqz s0, .LBB13_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    mv a3, a1
 ; RV32IFD-NEXT:  .LBB13_2:
 ; RV32IFD-NEXT:    lui a1, %hi(.LCPI13_1)
 ; RV32IFD-NEXT:    fld fa5, %lo(.LCPI13_1)(a1)
 ; RV32IFD-NEXT:    flt.d a1, fa5, fs0
 ; RV32IFD-NEXT:    beqz a1, .LBB13_4
 ; RV32IFD-NEXT:  # %bb.3:
-; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    addi a3, a4, -1
 ; RV32IFD-NEXT:  .LBB13_4:
-; RV32IFD-NEXT:    feq.d a3, fs0, fs0
-; RV32IFD-NEXT:    neg a4, s0
+; RV32IFD-NEXT:    feq.d a4, fs0, fs0
+; RV32IFD-NEXT:    neg a2, a2
 ; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a4, a0
-; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a0, a2, a0
+; RV32IFD-NEXT:    neg a2, a4
 ; RV32IFD-NEXT:    or a0, a1, a0
-; RV32IFD-NEXT:    and a1, a3, a2
-; RV32IFD-NEXT:    and a0, a3, a0
+; RV32IFD-NEXT:    and a1, a2, a3
+; RV32IFD-NEXT:    and a0, a2, a0
 ; RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -1046,27 +1051,28 @@ define i64 @test_roundeven_si64(double %x) nounwind {
 ; RV32IFD-NEXT:    fmv.d fs0, fa0
 ; RV32IFD-NEXT:    fle.d s0, fa5, fa0
 ; RV32IFD-NEXT:    call __fixdfdi
+; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lui a4, 524288
 ; RV32IFD-NEXT:    lui a3, 524288
-; RV32IFD-NEXT:    lui a2, 524288
 ; RV32IFD-NEXT:    beqz s0, .LBB17_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    mv a3, a1
 ; RV32IFD-NEXT:  .LBB17_2:
 ; RV32IFD-NEXT:    lui a1, %hi(.LCPI17_1)
 ; RV32IFD-NEXT:    fld fa5, %lo(.LCPI17_1)(a1)
 ; RV32IFD-NEXT:    flt.d a1, fa5, fs0
 ; RV32IFD-NEXT:    beqz a1, .LBB17_4
 ; RV32IFD-NEXT:  # %bb.3:
-; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    addi a3, a4, -1
 ; RV32IFD-NEXT:  .LBB17_4:
-; RV32IFD-NEXT:    feq.d a3, fs0, fs0
-; RV32IFD-NEXT:    neg a4, s0
+; RV32IFD-NEXT:    feq.d a4, fs0, fs0
+; RV32IFD-NEXT:    neg a2, a2
 ; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a4, a0
-; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a0, a2, a0
+; RV32IFD-NEXT:    neg a2, a4
 ; RV32IFD-NEXT:    or a0, a1, a0
-; RV32IFD-NEXT:    and a1, a3, a2
-; RV32IFD-NEXT:    and a0, a3, a0
+; RV32IFD-NEXT:    and a1, a2, a3
+; RV32IFD-NEXT:    and a0, a2, a0
 ; RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -1293,27 +1299,28 @@ define i64 @test_rint_si64(double %x) nounwind {
 ; RV32IFD-NEXT:    fmv.d fs0, fa0
 ; RV32IFD-NEXT:    fle.d s0, fa5, fa0
 ; RV32IFD-NEXT:    call __fixdfdi
+; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lui a4, 524288
 ; RV32IFD-NEXT:    lui a3, 524288
-; RV32IFD-NEXT:    lui a2, 524288
 ; RV32IFD-NEXT:    beqz s0, .LBB21_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    mv a3, a1
 ; RV32IFD-NEXT:  .LBB21_2:
 ; RV32IFD-NEXT:    lui a1, %hi(.LCPI21_1)
 ; RV32IFD-NEXT:    fld fa5, %lo(.LCPI21_1)(a1)
 ; RV32IFD-NEXT:    flt.d a1, fa5, fs0
 ; RV32IFD-NEXT:    beqz a1, .LBB21_4
 ; RV32IFD-NEXT:  # %bb.3:
-; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    addi a3, a4, -1
 ; RV32IFD-NEXT:  .LBB21_4:
-; RV32IFD-NEXT:    feq.d a3, fs0, fs0
-; RV32IFD-NEXT:    neg a4, s0
+; RV32IFD-NEXT:    feq.d a4, fs0, fs0
+; RV32IFD-NEXT:    neg a2, a2
 ; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a4, a0
-; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a0, a2, a0
+; RV32IFD-NEXT:    neg a2, a4
 ; RV32IFD-NEXT:    or a0, a1, a0
-; RV32IFD-NEXT:    and a1, a3, a2
-; RV32IFD-NEXT:    and a0, a3, a0
+; RV32IFD-NEXT:    and a1, a2, a3
+; RV32IFD-NEXT:    and a0, a2, a0
 ; RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/double-stack-spill-restore.ll b/llvm/test/CodeGen/RISCV/double-stack-spill-restore.ll
index f46ab5a2561884..45a338b9cf4b3f 100644
--- a/llvm/test/CodeGen/RISCV/double-stack-spill-restore.ll
+++ b/llvm/test/CodeGen/RISCV/double-stack-spill-restore.ll
@@ -62,25 +62,20 @@ define double @func(double %d, i32 %n) nounwind {
 ;
 ; RV32IZFINXZDINX-LABEL: func:
 ; RV32IZFINXZDINX:       # %bb.0: # %entry
-; RV32IZFINXZDINX-NEXT:    addi sp, sp, -16
-; RV32IZFINXZDINX-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32IZFINXZDINX-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32IZFINXZDINX-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32IZFINXZDINX-NEXT:    fmv.d s0, a0
 ; RV32IZFINXZDINX-NEXT:    beqz a2, .LBB0_2
 ; RV32IZFINXZDINX-NEXT:  # %bb.1: # %if.else
+; RV32IZFINXZDINX-NEXT:    addi sp, sp, -16
+; RV32IZFINXZDINX-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32IZFINXZDINX-NEXT:    addi a2, a2, -1
-; RV32IZFINXZDINX-NEXT:    fmv.d a0, s0
+; RV32IZFINXZDINX-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IZFINXZDINX-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; RV32IZFINXZDINX-NEXT:    call func
-; RV32IZFINXZDINX-NEXT:    fadd.d a0, a0, s0
-; RV32IZFINXZDINX-NEXT:    j .LBB0_3
-; RV32IZFINXZDINX-NEXT:  .LBB0_2: # %return
-; RV32IZFINXZDINX-NEXT:    fmv.d a0, s0
-; RV32IZFINXZDINX-NEXT:  .LBB0_3: # %return
+; RV32IZFINXZDINX-NEXT:    lw a2, 4(sp) # 4-byte Folded Reload
+; RV32IZFINXZDINX-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
+; RV32IZFINXZDINX-NEXT:    fadd.d a0, a0, a2
 ; RV32IZFINXZDINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32IZFINXZDINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32IZFINXZDINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32IZFINXZDINX-NEXT:    addi sp, sp, 16
+; RV32IZFINXZDINX-NEXT:  .LBB0_2: # %return
 ; RV32IZFINXZDINX-NEXT:    ret
 ;
 ; RV64IZFINXZDINX-LABEL: func:
@@ -90,13 +85,13 @@ define double @func(double %d, i32 %n) nounwind {
 ; RV64IZFINXZDINX-NEXT:  # %bb.1: # %if.else
 ; RV64IZFINXZDINX-NEXT:    addi sp, sp, -16
 ; RV64IZFINXZDINX-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64IZFINXZDINX-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
 ; RV64IZFINXZDINX-NEXT:    addiw a1, a1, -1
-; RV64IZFINXZDINX-NEXT:    mv s0, a0
+; RV64IZFINXZDINX-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
+; RV64IZFINXZDINX-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64IZFINXZDINX-NEXT:    call func
-; RV64IZFINXZDINX-NEXT:    fadd.d a0, a0, s0
+; RV64IZFINXZDINX-NEXT:    ld a1, 0(sp) # 8-byte Folded Reload
+; RV64IZFINXZDINX-NEXT:    fadd.d a0, a0, a1
 ; RV64IZFINXZDINX-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64IZFINXZDINX-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64IZFINXZDINX-NEXT:    addi sp, sp, 16
 ; RV64IZFINXZDINX-NEXT:  .LBB0_2: # %return
 ; RV64IZFINXZDINX-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/exception-pointer-register.ll b/llvm/test/CodeGen/RISCV/exception-pointer-register.ll
index 6eb2227cebab9d..a9c656f9974f74 100644
--- a/llvm/test/CodeGen/RISCV/exception-pointer-register.ll
+++ b/llvm/test/CodeGen/RISCV/exception-pointer-register.ll
@@ -18,28 +18,22 @@ define void @caller(ptr %p) personality ptr @__gxx_personality_v0 {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
-; RV32I-NEXT:    .cfi_offset s0, -8
 ; RV32I-NEXT:    .cfi_remember_state
-; RV32I-NEXT:    mv s0, a0
+; RV32I-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    beqz a0, .LBB0_2
 ; RV32I-NEXT:  # %bb.1: # %bb2
 ; RV32I-NEXT:  .Ltmp0: # EH_LABEL
-; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call bar
 ; RV32I-NEXT:  .Ltmp1: # EH_LABEL
 ; RV32I-NEXT:    j .LBB0_3
 ; RV32I-NEXT:  .LBB0_2: # %bb1
 ; RV32I-NEXT:  .Ltmp2: # EH_LABEL
-; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call foo
 ; RV32I-NEXT:  .Ltmp3: # EH_LABEL
 ; RV32I-NEXT:  .LBB0_3: # %end2
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
-; RV32I-NEXT:    .cfi_restore s0
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
@@ -47,7 +41,7 @@ define void @caller(ptr %p) personality ptr @__gxx_personality_v0 {
 ; RV32I-NEXT:    .cfi_restore_state
 ; RV32I-NEXT:  .Ltmp4: # EH_LABEL
 ; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call callee
 ; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call _Unwind_Resume
@@ -57,28 +51,22 @@ define void @caller(ptr %p) personality ptr @__gxx_personality_v0 {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    .cfi_def_cfa_offset 32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    .cfi_offset ra, -8
-; RV64I-NEXT:    .cfi_offset s0, -16
 ; RV64I-NEXT:    .cfi_remember_state
-; RV64I-NEXT:    mv s0, a0
+; RV64I-NEXT:    sd a0, 16(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    beqz a0, .LBB0_2
 ; RV64I-NEXT:  # %bb.1: # %bb2
 ; RV64I-NEXT:  .Ltmp0: # EH_LABEL
-; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call bar
 ; RV64I-NEXT:  .Ltmp1: # EH_LABEL
 ; RV64I-NEXT:    j .LBB0_3
 ; RV64I-NEXT:  .LBB0_2: # %bb1
 ; RV64I-NEXT:  .Ltmp2: # EH_LABEL
-; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call foo
 ; RV64I-NEXT:  .Ltmp3: # EH_LABEL
 ; RV64I-NEXT:  .LBB0_3: # %end2
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    .cfi_restore ra
-; RV64I-NEXT:    .cfi_restore s0
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    .cfi_def_cfa_offset 0
 ; RV64I-NEXT:    ret
@@ -86,7 +74,7 @@ define void @caller(ptr %p) personality ptr @__gxx_personality_v0 {
 ; RV64I-NEXT:    .cfi_restore_state
 ; RV64I-NEXT:  .Ltmp4: # EH_LABEL
 ; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    call callee
 ; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    call _Unwind_Resume
diff --git a/llvm/test/CodeGen/RISCV/float-convert.ll b/llvm/test/CodeGen/RISCV/float-convert.ll
index b2a99d96dce620..e9189a8af1887d 100644
--- a/llvm/test/CodeGen/RISCV/float-convert.ll
+++ b/llvm/test/CodeGen/RISCV/float-convert.ll
@@ -65,12 +65,11 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ;
 ; RV32I-LABEL: fcvt_w_s_sat:
 ; RV32I:       # %bb.0: # %start
-; RV32I-NEXT:    addi sp, sp, -32
-; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    lui a1, 847872
 ; RV32I-NEXT:    call __gesf2
@@ -78,7 +77,6 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __fixsfsi
 ; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    lui s3, 524288
 ; RV32I-NEXT:    bgez s2, .LBB1_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    lui s1, 524288
@@ -89,7 +87,8 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    blez a0, .LBB1_4
 ; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    addi s1, s3, -1
+; RV32I-NEXT:    lui s1, 524288
+; RV32I-NEXT:    addi s1, s1, -1
 ; RV32I-NEXT:  .LBB1_4: # %start
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s0
@@ -97,22 +96,20 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    snez a0, a0
 ; RV32I-NEXT:    addi a0, a0, -1
 ; RV32I-NEXT:    and a0, a0, s1
-; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_w_s_sat:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    lui a1, 847872
 ; RV64I-NEXT:    call __gesf2
@@ -120,7 +117,6 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixsfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    lui s3, 524288
 ; RV64I-NEXT:    bgez s2, .LBB1_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    lui s1, 524288
@@ -131,7 +127,8 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB1_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    addiw s1, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addiw s1, s1, -1
 ; RV64I-NEXT:  .LBB1_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -139,12 +136,11 @@ define i32 @fcvt_w_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    snez a0, a0
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i32 @llvm.fptosi.sat.i32.f32(float %a)
@@ -300,18 +296,16 @@ define i32 @fcvt_wu_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 325632
 ; RV64I-NEXT:    addi a1, a1, -1
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB4_2
 ; RV64I-NEXT:  # %bb.1: # %start
@@ -319,14 +313,14 @@ define i32 @fcvt_wu_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    j .LBB4_3
 ; RV64I-NEXT:  .LBB4_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB4_3: # %start
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -615,11 +609,12 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32IF-NEXT:    fmv.s fs0, fa0
 ; RV32IF-NEXT:    fle.s s0, fa5, fa0
 ; RV32IF-NEXT:    call __fixsfdi
+; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lui a4, 524288
 ; RV32IF-NEXT:    lui a3, 524288
-; RV32IF-NEXT:    lui a2, 524288
 ; RV32IF-NEXT:    beqz s0, .LBB12_2
 ; RV32IF-NEXT:  # %bb.1: # %start
-; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    mv a3, a1
 ; RV32IF-NEXT:  .LBB12_2: # %start
 ; RV32IF-NEXT:    lui a1, 389120
 ; RV32IF-NEXT:    addi a1, a1, -1
@@ -627,16 +622,16 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32IF-NEXT:    flt.s a1, fa5, fs0
 ; RV32IF-NEXT:    beqz a1, .LBB12_4
 ; RV32IF-NEXT:  # %bb.3:
-; RV32IF-NEXT:    addi a2, a3, -1
+; RV32IF-NEXT:    addi a3, a4, -1
 ; RV32IF-NEXT:  .LBB12_4: # %start
-; RV32IF-NEXT:    feq.s a3, fs0, fs0
-; RV32IF-NEXT:    neg a4, s0
-; RV32IF-NEXT:    and a0, a4, a0
+; RV32IF-NEXT:    feq.s a4, fs0, fs0
+; RV32IF-NEXT:    neg a2, a2
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    neg a2, a4
 ; RV32IF-NEXT:    or a0, a1, a0
-; RV32IF-NEXT:    and a1, a3, a2
-; RV32IF-NEXT:    and a0, a3, a0
+; RV32IF-NEXT:    and a1, a2, a3
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -663,27 +658,28 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixsfdi
+; RV32IZFINX-NEXT:    mv a2, s1
+; RV32IZFINX-NEXT:    lui a4, 524288
 ; RV32IZFINX-NEXT:    lui a3, 524288
-; RV32IZFINX-NEXT:    lui a2, 524288
 ; RV32IZFINX-NEXT:    beqz s1, .LBB12_2
 ; RV32IZFINX-NEXT:  # %bb.1: # %start
-; RV32IZFINX-NEXT:    mv a2, a1
+; RV32IZFINX-NEXT:    mv a3, a1
 ; RV32IZFINX-NEXT:  .LBB12_2: # %start
 ; RV32IZFINX-NEXT:    lui a1, 389120
 ; RV32IZFINX-NEXT:    addi a1, a1, -1
 ; RV32IZFINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZFINX-NEXT:    beqz a1, .LBB12_4
 ; RV32IZFINX-NEXT:  # %bb.3:
-; RV32IZFINX-NEXT:    addi a2, a3, -1
+; RV32IZFINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINX-NEXT:  .LBB12_4: # %start
-; RV32IZFINX-NEXT:    feq.s a3, s0, s0
-; RV32IZFINX-NEXT:    neg a4, s1
-; RV32IZFINX-NEXT:    and a0, a4, a0
+; RV32IZFINX-NEXT:    feq.s a4, s0, s0
+; RV32IZFINX-NEXT:    neg a2, a2
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    neg a1, a1
-; RV32IZFINX-NEXT:    neg a3, a3
+; RV32IZFINX-NEXT:    neg a2, a4
 ; RV32IZFINX-NEXT:    or a0, a1, a0
-; RV32IZFINX-NEXT:    and a1, a3, a2
-; RV32IZFINX-NEXT:    and a0, a3, a0
+; RV32IZFINX-NEXT:    and a1, a2, a3
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -708,7 +704,6 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s1, a0
 ; RV32I-NEXT:    lui a1, 913408
 ; RV32I-NEXT:    call __gesf2
@@ -717,7 +712,6 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    call __fixsfdi
 ; RV32I-NEXT:    mv s2, a0
 ; RV32I-NEXT:    mv s0, a1
-; RV32I-NEXT:    lui s5, 524288
 ; RV32I-NEXT:    bgez s3, .LBB12_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    lui s0, 524288
@@ -729,7 +723,8 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    mv s4, a0
 ; RV32I-NEXT:    blez a0, .LBB12_4
 ; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    addi s0, s5, -1
+; RV32I-NEXT:    lui s0, 524288
+; RV32I-NEXT:    addi s0, s0, -1
 ; RV32I-NEXT:  .LBB12_4: # %start
 ; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    mv a1, s1
@@ -750,18 +745,16 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_l_s_sat:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    lui a1, 913408
 ; RV64I-NEXT:    call __gesf2
@@ -769,10 +762,10 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixsfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    li s3, -1
 ; RV64I-NEXT:    bgez s2, .LBB12_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    slli s1, s3, 63
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    slli s1, s1, 63
 ; RV64I-NEXT:  .LBB12_2: # %start
 ; RV64I-NEXT:    lui a1, 389120
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -780,7 +773,8 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB12_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    srli s1, s3, 1
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    srli s1, s1, 1
 ; RV64I-NEXT:  .LBB12_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -788,12 +782,11 @@ define i64 @fcvt_l_s_sat(float %a) nounwind {
 ; RV64I-NEXT:    snez a0, a0
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i64 @llvm.fptosi.sat.i64.f32(float %a)
@@ -1636,18 +1629,16 @@ define zeroext i16 @fcvt_wu_s_sat_i16(float %a) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __fixunssfsi
-; RV32I-NEXT:    mv s1, a0
+; RV32I-NEXT:    sw a0, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 292864
 ; RV32I-NEXT:    addi a1, a1, -256
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    lui a1, 16
 ; RV32I-NEXT:    addi a1, a1, -1
@@ -1656,15 +1647,15 @@ define zeroext i16 @fcvt_wu_s_sat_i16(float %a) nounwind {
 ; RV32I-NEXT:    mv a0, a1
 ; RV32I-NEXT:    j .LBB26_3
 ; RV32I-NEXT:  .LBB26_2:
-; RV32I-NEXT:    srli s0, s0, 31
-; RV32I-NEXT:    addi a0, s0, -1
-; RV32I-NEXT:    and a0, a0, s1
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a0, a0, 31
+; RV32I-NEXT:    addi a2, a0, -1
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a2, a0
 ; RV32I-NEXT:  .LBB26_3: # %start
 ; RV32I-NEXT:    and a0, a0, a1
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -1673,18 +1664,16 @@ define zeroext i16 @fcvt_wu_s_sat_i16(float %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 292864
 ; RV64I-NEXT:    addi a1, a1, -256
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    lui a1, 16
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -1693,15 +1682,15 @@ define zeroext i16 @fcvt_wu_s_sat_i16(float %a) nounwind {
 ; RV64I-NEXT:    mv a0, a1
 ; RV64I-NEXT:    j .LBB26_3
 ; RV64I-NEXT:  .LBB26_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a2, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a2, a0
 ; RV64I-NEXT:  .LBB26_3: # %start
 ; RV64I-NEXT:    and a0, a0, a1
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -1972,32 +1961,30 @@ define zeroext i8 @fcvt_wu_s_sat_i8(float %a) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __fixunssfsi
-; RV32I-NEXT:    mv s1, a0
+; RV32I-NEXT:    sw a0, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 276464
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    blez a0, .LBB30_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    li a0, 255
 ; RV32I-NEXT:    j .LBB30_3
 ; RV32I-NEXT:  .LBB30_2:
-; RV32I-NEXT:    srli s0, s0, 31
-; RV32I-NEXT:    addi a0, s0, -1
-; RV32I-NEXT:    and a0, a0, s1
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a0, a0, 31
+; RV32I-NEXT:    addi a1, a0, -1
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a1, a0
 ; RV32I-NEXT:  .LBB30_3: # %start
 ; RV32I-NEXT:    zext.b a0, a0
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -2006,32 +1993,30 @@ define zeroext i8 @fcvt_wu_s_sat_i8(float %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 276464
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB30_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    li a0, 255
 ; RV64I-NEXT:    j .LBB30_3
 ; RV64I-NEXT:  .LBB30_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB30_3: # %start
 ; RV64I-NEXT:    zext.b a0, a0
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -2112,18 +2097,16 @@ define zeroext i32 @fcvt_wu_s_sat_zext(float %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 325632
 ; RV64I-NEXT:    addi a1, a1, -1
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB31_2
 ; RV64I-NEXT:  # %bb.1: # %start
@@ -2131,16 +2114,16 @@ define zeroext i32 @fcvt_wu_s_sat_zext(float %a) nounwind {
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    j .LBB31_3
 ; RV64I-NEXT:  .LBB31_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB31_3: # %start
 ; RV64I-NEXT:    slli a0, a0, 32
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
@@ -2169,12 +2152,11 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ;
 ; RV32I-LABEL: fcvt_w_s_sat_sext:
 ; RV32I:       # %bb.0: # %start
-; RV32I-NEXT:    addi sp, sp, -32
-; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    lui a1, 847872
 ; RV32I-NEXT:    call __gesf2
@@ -2182,7 +2164,6 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __fixsfsi
 ; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    lui s3, 524288
 ; RV32I-NEXT:    bgez s2, .LBB32_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    lui s1, 524288
@@ -2193,7 +2174,8 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    blez a0, .LBB32_4
 ; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    addi s1, s3, -1
+; RV32I-NEXT:    lui s1, 524288
+; RV32I-NEXT:    addi s1, s1, -1
 ; RV32I-NEXT:  .LBB32_4: # %start
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s0
@@ -2201,22 +2183,20 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV32I-NEXT:    snez a0, a0
 ; RV32I-NEXT:    addi a0, a0, -1
 ; RV32I-NEXT:    and a0, a0, s1
-; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_w_s_sat_sext:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    lui a1, 847872
 ; RV64I-NEXT:    call __gesf2
@@ -2224,7 +2204,6 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixsfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    lui s3, 524288
 ; RV64I-NEXT:    bgez s2, .LBB32_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    lui s1, 524288
@@ -2235,7 +2214,8 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB32_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    addi s1, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addi s1, s1, -1
 ; RV64I-NEXT:  .LBB32_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -2244,12 +2224,11 @@ define signext i32 @fcvt_w_s_sat_sext(float %a) nounwind {
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 start:
   %0 = tail call i32 @llvm.fptosi.sat.i32.f32(float %a)
diff --git a/llvm/test/CodeGen/RISCV/float-round-conv-sat.ll b/llvm/test/CodeGen/RISCV/float-round-conv-sat.ll
index 7b103779f75bab..9ed23dcb69354f 100644
--- a/llvm/test/CodeGen/RISCV/float-round-conv-sat.ll
+++ b/llvm/test/CodeGen/RISCV/float-round-conv-sat.ll
@@ -54,11 +54,12 @@ define i64 @test_floor_si64(float %x) nounwind {
 ; RV32IF-NEXT:    fle.s s0, fa5, fs0
 ; RV32IF-NEXT:    fmv.s fa0, fs0
 ; RV32IF-NEXT:    call __fixsfdi
+; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lui a4, 524288
 ; RV32IF-NEXT:    lui a3, 524288
-; RV32IF-NEXT:    lui a2, 524288
 ; RV32IF-NEXT:    beqz s0, .LBB1_4
 ; RV32IF-NEXT:  # %bb.3:
-; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    mv a3, a1
 ; RV32IF-NEXT:  .LBB1_4:
 ; RV32IF-NEXT:    lui a1, 389120
 ; RV32IF-NEXT:    addi a1, a1, -1
@@ -66,16 +67,16 @@ define i64 @test_floor_si64(float %x) nounwind {
 ; RV32IF-NEXT:    flt.s a1, fa5, fs0
 ; RV32IF-NEXT:    beqz a1, .LBB1_6
 ; RV32IF-NEXT:  # %bb.5:
-; RV32IF-NEXT:    addi a2, a3, -1
+; RV32IF-NEXT:    addi a3, a4, -1
 ; RV32IF-NEXT:  .LBB1_6:
-; RV32IF-NEXT:    feq.s a3, fs0, fs0
-; RV32IF-NEXT:    neg a4, s0
-; RV32IF-NEXT:    and a0, a4, a0
+; RV32IF-NEXT:    feq.s a4, fs0, fs0
+; RV32IF-NEXT:    neg a2, a2
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    neg a2, a4
 ; RV32IF-NEXT:    or a0, a1, a0
-; RV32IF-NEXT:    and a1, a3, a2
-; RV32IF-NEXT:    and a0, a3, a0
+; RV32IF-NEXT:    and a1, a2, a3
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -111,27 +112,28 @@ define i64 @test_floor_si64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixsfdi
+; RV32IZFINX-NEXT:    mv a2, s1
+; RV32IZFINX-NEXT:    lui a4, 524288
 ; RV32IZFINX-NEXT:    lui a3, 524288
-; RV32IZFINX-NEXT:    lui a2, 524288
 ; RV32IZFINX-NEXT:    beqz s1, .LBB1_4
 ; RV32IZFINX-NEXT:  # %bb.3:
-; RV32IZFINX-NEXT:    mv a2, a1
+; RV32IZFINX-NEXT:    mv a3, a1
 ; RV32IZFINX-NEXT:  .LBB1_4:
 ; RV32IZFINX-NEXT:    lui a1, 389120
 ; RV32IZFINX-NEXT:    addi a1, a1, -1
 ; RV32IZFINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZFINX-NEXT:    beqz a1, .LBB1_6
 ; RV32IZFINX-NEXT:  # %bb.5:
-; RV32IZFINX-NEXT:    addi a2, a3, -1
+; RV32IZFINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINX-NEXT:  .LBB1_6:
-; RV32IZFINX-NEXT:    feq.s a3, s0, s0
-; RV32IZFINX-NEXT:    neg a4, s1
-; RV32IZFINX-NEXT:    and a0, a4, a0
+; RV32IZFINX-NEXT:    feq.s a4, s0, s0
+; RV32IZFINX-NEXT:    neg a2, a2
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    neg a1, a1
-; RV32IZFINX-NEXT:    neg a3, a3
+; RV32IZFINX-NEXT:    neg a2, a4
 ; RV32IZFINX-NEXT:    or a0, a1, a0
-; RV32IZFINX-NEXT:    and a1, a3, a2
-; RV32IZFINX-NEXT:    and a0, a3, a0
+; RV32IZFINX-NEXT:    and a1, a2, a3
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -228,17 +230,17 @@ define i64 @test_floor_ui64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    mv s0, a0
-; RV32IZFINX-NEXT:    fabs.s a0, a0
-; RV32IZFINX-NEXT:    lui a1, 307200
-; RV32IZFINX-NEXT:    flt.s a0, a0, a1
-; RV32IZFINX-NEXT:    beqz a0, .LBB3_2
+; RV32IZFINX-NEXT:    fabs.s a1, a0
+; RV32IZFINX-NEXT:    lui a2, 307200
+; RV32IZFINX-NEXT:    flt.s a1, a1, a2
+; RV32IZFINX-NEXT:    beqz a1, .LBB3_2
 ; RV32IZFINX-NEXT:  # %bb.1:
-; RV32IZFINX-NEXT:    fcvt.w.s a0, s0, rdn
-; RV32IZFINX-NEXT:    fcvt.s.w a0, a0, rdn
-; RV32IZFINX-NEXT:    fsgnj.s s0, a0, s0
+; RV32IZFINX-NEXT:    fcvt.w.s a1, s0, rdn
+; RV32IZFINX-NEXT:    fcvt.s.w a1, a1, rdn
+; RV32IZFINX-NEXT:    fsgnj.s s0, a1, s0
 ; RV32IZFINX-NEXT:  .LBB3_2:
-; RV32IZFINX-NEXT:    fle.s a0, zero, s0
-; RV32IZFINX-NEXT:    neg s1, a0
+; RV32IZFINX-NEXT:    fle.s a1, zero, s0
+; RV32IZFINX-NEXT:    neg s1, a1
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixunssfdi
 ; RV32IZFINX-NEXT:    lui a2, 391168
@@ -314,11 +316,12 @@ define i64 @test_ceil_si64(float %x) nounwind {
 ; RV32IF-NEXT:    fle.s s0, fa5, fs0
 ; RV32IF-NEXT:    fmv.s fa0, fs0
 ; RV32IF-NEXT:    call __fixsfdi
+; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lui a4, 524288
 ; RV32IF-NEXT:    lui a3, 524288
-; RV32IF-NEXT:    lui a2, 524288
 ; RV32IF-NEXT:    beqz s0, .LBB5_4
 ; RV32IF-NEXT:  # %bb.3:
-; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    mv a3, a1
 ; RV32IF-NEXT:  .LBB5_4:
 ; RV32IF-NEXT:    lui a1, 389120
 ; RV32IF-NEXT:    addi a1, a1, -1
@@ -326,16 +329,16 @@ define i64 @test_ceil_si64(float %x) nounwind {
 ; RV32IF-NEXT:    flt.s a1, fa5, fs0
 ; RV32IF-NEXT:    beqz a1, .LBB5_6
 ; RV32IF-NEXT:  # %bb.5:
-; RV32IF-NEXT:    addi a2, a3, -1
+; RV32IF-NEXT:    addi a3, a4, -1
 ; RV32IF-NEXT:  .LBB5_6:
-; RV32IF-NEXT:    feq.s a3, fs0, fs0
-; RV32IF-NEXT:    neg a4, s0
-; RV32IF-NEXT:    and a0, a4, a0
+; RV32IF-NEXT:    feq.s a4, fs0, fs0
+; RV32IF-NEXT:    neg a2, a2
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    neg a2, a4
 ; RV32IF-NEXT:    or a0, a1, a0
-; RV32IF-NEXT:    and a1, a3, a2
-; RV32IF-NEXT:    and a0, a3, a0
+; RV32IF-NEXT:    and a1, a2, a3
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -371,27 +374,28 @@ define i64 @test_ceil_si64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixsfdi
+; RV32IZFINX-NEXT:    mv a2, s1
+; RV32IZFINX-NEXT:    lui a4, 524288
 ; RV32IZFINX-NEXT:    lui a3, 524288
-; RV32IZFINX-NEXT:    lui a2, 524288
 ; RV32IZFINX-NEXT:    beqz s1, .LBB5_4
 ; RV32IZFINX-NEXT:  # %bb.3:
-; RV32IZFINX-NEXT:    mv a2, a1
+; RV32IZFINX-NEXT:    mv a3, a1
 ; RV32IZFINX-NEXT:  .LBB5_4:
 ; RV32IZFINX-NEXT:    lui a1, 389120
 ; RV32IZFINX-NEXT:    addi a1, a1, -1
 ; RV32IZFINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZFINX-NEXT:    beqz a1, .LBB5_6
 ; RV32IZFINX-NEXT:  # %bb.5:
-; RV32IZFINX-NEXT:    addi a2, a3, -1
+; RV32IZFINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINX-NEXT:  .LBB5_6:
-; RV32IZFINX-NEXT:    feq.s a3, s0, s0
-; RV32IZFINX-NEXT:    neg a4, s1
-; RV32IZFINX-NEXT:    and a0, a4, a0
+; RV32IZFINX-NEXT:    feq.s a4, s0, s0
+; RV32IZFINX-NEXT:    neg a2, a2
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    neg a1, a1
-; RV32IZFINX-NEXT:    neg a3, a3
+; RV32IZFINX-NEXT:    neg a2, a4
 ; RV32IZFINX-NEXT:    or a0, a1, a0
-; RV32IZFINX-NEXT:    and a1, a3, a2
-; RV32IZFINX-NEXT:    and a0, a3, a0
+; RV32IZFINX-NEXT:    and a1, a2, a3
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -488,17 +492,17 @@ define i64 @test_ceil_ui64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    mv s0, a0
-; RV32IZFINX-NEXT:    fabs.s a0, a0
-; RV32IZFINX-NEXT:    lui a1, 307200
-; RV32IZFINX-NEXT:    flt.s a0, a0, a1
-; RV32IZFINX-NEXT:    beqz a0, .LBB7_2
+; RV32IZFINX-NEXT:    fabs.s a1, a0
+; RV32IZFINX-NEXT:    lui a2, 307200
+; RV32IZFINX-NEXT:    flt.s a1, a1, a2
+; RV32IZFINX-NEXT:    beqz a1, .LBB7_2
 ; RV32IZFINX-NEXT:  # %bb.1:
-; RV32IZFINX-NEXT:    fcvt.w.s a0, s0, rup
-; RV32IZFINX-NEXT:    fcvt.s.w a0, a0, rup
-; RV32IZFINX-NEXT:    fsgnj.s s0, a0, s0
+; RV32IZFINX-NEXT:    fcvt.w.s a1, s0, rup
+; RV32IZFINX-NEXT:    fcvt.s.w a1, a1, rup
+; RV32IZFINX-NEXT:    fsgnj.s s0, a1, s0
 ; RV32IZFINX-NEXT:  .LBB7_2:
-; RV32IZFINX-NEXT:    fle.s a0, zero, s0
-; RV32IZFINX-NEXT:    neg s1, a0
+; RV32IZFINX-NEXT:    fle.s a1, zero, s0
+; RV32IZFINX-NEXT:    neg s1, a1
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixunssfdi
 ; RV32IZFINX-NEXT:    lui a2, 391168
@@ -563,11 +567,12 @@ define i64 @test_trunc_si64(float %x) nounwind {
 ; RV32IF-NEXT:    fmv.s fs0, fa0
 ; RV32IF-NEXT:    fle.s s0, fa5, fa0
 ; RV32IF-NEXT:    call __fixsfdi
+; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lui a4, 524288
 ; RV32IF-NEXT:    lui a3, 524288
-; RV32IF-NEXT:    lui a2, 524288
 ; RV32IF-NEXT:    beqz s0, .LBB9_2
 ; RV32IF-NEXT:  # %bb.1:
-; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    mv a3, a1
 ; RV32IF-NEXT:  .LBB9_2:
 ; RV32IF-NEXT:    lui a1, 389120
 ; RV32IF-NEXT:    addi a1, a1, -1
@@ -575,16 +580,16 @@ define i64 @test_trunc_si64(float %x) nounwind {
 ; RV32IF-NEXT:    flt.s a1, fa5, fs0
 ; RV32IF-NEXT:    beqz a1, .LBB9_4
 ; RV32IF-NEXT:  # %bb.3:
-; RV32IF-NEXT:    addi a2, a3, -1
+; RV32IF-NEXT:    addi a3, a4, -1
 ; RV32IF-NEXT:  .LBB9_4:
-; RV32IF-NEXT:    feq.s a3, fs0, fs0
-; RV32IF-NEXT:    neg a4, s0
-; RV32IF-NEXT:    and a0, a4, a0
+; RV32IF-NEXT:    feq.s a4, fs0, fs0
+; RV32IF-NEXT:    neg a2, a2
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    neg a2, a4
 ; RV32IF-NEXT:    or a0, a1, a0
-; RV32IF-NEXT:    and a1, a3, a2
-; RV32IF-NEXT:    and a0, a3, a0
+; RV32IF-NEXT:    and a1, a2, a3
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -611,27 +616,28 @@ define i64 @test_trunc_si64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixsfdi
+; RV32IZFINX-NEXT:    mv a2, s1
+; RV32IZFINX-NEXT:    lui a4, 524288
 ; RV32IZFINX-NEXT:    lui a3, 524288
-; RV32IZFINX-NEXT:    lui a2, 524288
 ; RV32IZFINX-NEXT:    beqz s1, .LBB9_2
 ; RV32IZFINX-NEXT:  # %bb.1:
-; RV32IZFINX-NEXT:    mv a2, a1
+; RV32IZFINX-NEXT:    mv a3, a1
 ; RV32IZFINX-NEXT:  .LBB9_2:
 ; RV32IZFINX-NEXT:    lui a1, 389120
 ; RV32IZFINX-NEXT:    addi a1, a1, -1
 ; RV32IZFINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZFINX-NEXT:    beqz a1, .LBB9_4
 ; RV32IZFINX-NEXT:  # %bb.3:
-; RV32IZFINX-NEXT:    addi a2, a3, -1
+; RV32IZFINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINX-NEXT:  .LBB9_4:
-; RV32IZFINX-NEXT:    feq.s a3, s0, s0
-; RV32IZFINX-NEXT:    neg a4, s1
-; RV32IZFINX-NEXT:    and a0, a4, a0
+; RV32IZFINX-NEXT:    feq.s a4, s0, s0
+; RV32IZFINX-NEXT:    neg a2, a2
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    neg a1, a1
-; RV32IZFINX-NEXT:    neg a3, a3
+; RV32IZFINX-NEXT:    neg a2, a4
 ; RV32IZFINX-NEXT:    or a0, a1, a0
-; RV32IZFINX-NEXT:    and a1, a3, a2
-; RV32IZFINX-NEXT:    and a0, a3, a0
+; RV32IZFINX-NEXT:    and a1, a2, a3
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -812,11 +818,12 @@ define i64 @test_round_si64(float %x) nounwind {
 ; RV32IF-NEXT:    fle.s s0, fa5, fs0
 ; RV32IF-NEXT:    fmv.s fa0, fs0
 ; RV32IF-NEXT:    call __fixsfdi
+; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lui a4, 524288
 ; RV32IF-NEXT:    lui a3, 524288
-; RV32IF-NEXT:    lui a2, 524288
 ; RV32IF-NEXT:    beqz s0, .LBB13_4
 ; RV32IF-NEXT:  # %bb.3:
-; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    mv a3, a1
 ; RV32IF-NEXT:  .LBB13_4:
 ; RV32IF-NEXT:    lui a1, 389120
 ; RV32IF-NEXT:    addi a1, a1, -1
@@ -824,16 +831,16 @@ define i64 @test_round_si64(float %x) nounwind {
 ; RV32IF-NEXT:    flt.s a1, fa5, fs0
 ; RV32IF-NEXT:    beqz a1, .LBB13_6
 ; RV32IF-NEXT:  # %bb.5:
-; RV32IF-NEXT:    addi a2, a3, -1
+; RV32IF-NEXT:    addi a3, a4, -1
 ; RV32IF-NEXT:  .LBB13_6:
-; RV32IF-NEXT:    feq.s a3, fs0, fs0
-; RV32IF-NEXT:    neg a4, s0
-; RV32IF-NEXT:    and a0, a4, a0
+; RV32IF-NEXT:    feq.s a4, fs0, fs0
+; RV32IF-NEXT:    neg a2, a2
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    neg a2, a4
 ; RV32IF-NEXT:    or a0, a1, a0
-; RV32IF-NEXT:    and a1, a3, a2
-; RV32IF-NEXT:    and a0, a3, a0
+; RV32IF-NEXT:    and a1, a2, a3
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -869,27 +876,28 @@ define i64 @test_round_si64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixsfdi
+; RV32IZFINX-NEXT:    mv a2, s1
+; RV32IZFINX-NEXT:    lui a4, 524288
 ; RV32IZFINX-NEXT:    lui a3, 524288
-; RV32IZFINX-NEXT:    lui a2, 524288
 ; RV32IZFINX-NEXT:    beqz s1, .LBB13_4
 ; RV32IZFINX-NEXT:  # %bb.3:
-; RV32IZFINX-NEXT:    mv a2, a1
+; RV32IZFINX-NEXT:    mv a3, a1
 ; RV32IZFINX-NEXT:  .LBB13_4:
 ; RV32IZFINX-NEXT:    lui a1, 389120
 ; RV32IZFINX-NEXT:    addi a1, a1, -1
 ; RV32IZFINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZFINX-NEXT:    beqz a1, .LBB13_6
 ; RV32IZFINX-NEXT:  # %bb.5:
-; RV32IZFINX-NEXT:    addi a2, a3, -1
+; RV32IZFINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINX-NEXT:  .LBB13_6:
-; RV32IZFINX-NEXT:    feq.s a3, s0, s0
-; RV32IZFINX-NEXT:    neg a4, s1
-; RV32IZFINX-NEXT:    and a0, a4, a0
+; RV32IZFINX-NEXT:    feq.s a4, s0, s0
+; RV32IZFINX-NEXT:    neg a2, a2
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    neg a1, a1
-; RV32IZFINX-NEXT:    neg a3, a3
+; RV32IZFINX-NEXT:    neg a2, a4
 ; RV32IZFINX-NEXT:    or a0, a1, a0
-; RV32IZFINX-NEXT:    and a1, a3, a2
-; RV32IZFINX-NEXT:    and a0, a3, a0
+; RV32IZFINX-NEXT:    and a1, a2, a3
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -986,17 +994,17 @@ define i64 @test_round_ui64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    mv s0, a0
-; RV32IZFINX-NEXT:    fabs.s a0, a0
-; RV32IZFINX-NEXT:    lui a1, 307200
-; RV32IZFINX-NEXT:    flt.s a0, a0, a1
-; RV32IZFINX-NEXT:    beqz a0, .LBB15_2
+; RV32IZFINX-NEXT:    fabs.s a1, a0
+; RV32IZFINX-NEXT:    lui a2, 307200
+; RV32IZFINX-NEXT:    flt.s a1, a1, a2
+; RV32IZFINX-NEXT:    beqz a1, .LBB15_2
 ; RV32IZFINX-NEXT:  # %bb.1:
-; RV32IZFINX-NEXT:    fcvt.w.s a0, s0, rmm
-; RV32IZFINX-NEXT:    fcvt.s.w a0, a0, rmm
-; RV32IZFINX-NEXT:    fsgnj.s s0, a0, s0
+; RV32IZFINX-NEXT:    fcvt.w.s a1, s0, rmm
+; RV32IZFINX-NEXT:    fcvt.s.w a1, a1, rmm
+; RV32IZFINX-NEXT:    fsgnj.s s0, a1, s0
 ; RV32IZFINX-NEXT:  .LBB15_2:
-; RV32IZFINX-NEXT:    fle.s a0, zero, s0
-; RV32IZFINX-NEXT:    neg s1, a0
+; RV32IZFINX-NEXT:    fle.s a1, zero, s0
+; RV32IZFINX-NEXT:    neg s1, a1
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixunssfdi
 ; RV32IZFINX-NEXT:    lui a2, 391168
@@ -1072,11 +1080,12 @@ define i64 @test_roundeven_si64(float %x) nounwind {
 ; RV32IF-NEXT:    fle.s s0, fa5, fs0
 ; RV32IF-NEXT:    fmv.s fa0, fs0
 ; RV32IF-NEXT:    call __fixsfdi
+; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lui a4, 524288
 ; RV32IF-NEXT:    lui a3, 524288
-; RV32IF-NEXT:    lui a2, 524288
 ; RV32IF-NEXT:    beqz s0, .LBB17_4
 ; RV32IF-NEXT:  # %bb.3:
-; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    mv a3, a1
 ; RV32IF-NEXT:  .LBB17_4:
 ; RV32IF-NEXT:    lui a1, 389120
 ; RV32IF-NEXT:    addi a1, a1, -1
@@ -1084,16 +1093,16 @@ define i64 @test_roundeven_si64(float %x) nounwind {
 ; RV32IF-NEXT:    flt.s a1, fa5, fs0
 ; RV32IF-NEXT:    beqz a1, .LBB17_6
 ; RV32IF-NEXT:  # %bb.5:
-; RV32IF-NEXT:    addi a2, a3, -1
+; RV32IF-NEXT:    addi a3, a4, -1
 ; RV32IF-NEXT:  .LBB17_6:
-; RV32IF-NEXT:    feq.s a3, fs0, fs0
-; RV32IF-NEXT:    neg a4, s0
-; RV32IF-NEXT:    and a0, a4, a0
+; RV32IF-NEXT:    feq.s a4, fs0, fs0
+; RV32IF-NEXT:    neg a2, a2
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    neg a2, a4
 ; RV32IF-NEXT:    or a0, a1, a0
-; RV32IF-NEXT:    and a1, a3, a2
-; RV32IF-NEXT:    and a0, a3, a0
+; RV32IF-NEXT:    and a1, a2, a3
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -1129,27 +1138,28 @@ define i64 @test_roundeven_si64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixsfdi
+; RV32IZFINX-NEXT:    mv a2, s1
+; RV32IZFINX-NEXT:    lui a4, 524288
 ; RV32IZFINX-NEXT:    lui a3, 524288
-; RV32IZFINX-NEXT:    lui a2, 524288
 ; RV32IZFINX-NEXT:    beqz s1, .LBB17_4
 ; RV32IZFINX-NEXT:  # %bb.3:
-; RV32IZFINX-NEXT:    mv a2, a1
+; RV32IZFINX-NEXT:    mv a3, a1
 ; RV32IZFINX-NEXT:  .LBB17_4:
 ; RV32IZFINX-NEXT:    lui a1, 389120
 ; RV32IZFINX-NEXT:    addi a1, a1, -1
 ; RV32IZFINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZFINX-NEXT:    beqz a1, .LBB17_6
 ; RV32IZFINX-NEXT:  # %bb.5:
-; RV32IZFINX-NEXT:    addi a2, a3, -1
+; RV32IZFINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINX-NEXT:  .LBB17_6:
-; RV32IZFINX-NEXT:    feq.s a3, s0, s0
-; RV32IZFINX-NEXT:    neg a4, s1
-; RV32IZFINX-NEXT:    and a0, a4, a0
+; RV32IZFINX-NEXT:    feq.s a4, s0, s0
+; RV32IZFINX-NEXT:    neg a2, a2
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    neg a1, a1
-; RV32IZFINX-NEXT:    neg a3, a3
+; RV32IZFINX-NEXT:    neg a2, a4
 ; RV32IZFINX-NEXT:    or a0, a1, a0
-; RV32IZFINX-NEXT:    and a1, a3, a2
-; RV32IZFINX-NEXT:    and a0, a3, a0
+; RV32IZFINX-NEXT:    and a1, a2, a3
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -1246,17 +1256,17 @@ define i64 @test_roundeven_ui64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    mv s0, a0
-; RV32IZFINX-NEXT:    fabs.s a0, a0
-; RV32IZFINX-NEXT:    lui a1, 307200
-; RV32IZFINX-NEXT:    flt.s a0, a0, a1
-; RV32IZFINX-NEXT:    beqz a0, .LBB19_2
+; RV32IZFINX-NEXT:    fabs.s a1, a0
+; RV32IZFINX-NEXT:    lui a2, 307200
+; RV32IZFINX-NEXT:    flt.s a1, a1, a2
+; RV32IZFINX-NEXT:    beqz a1, .LBB19_2
 ; RV32IZFINX-NEXT:  # %bb.1:
-; RV32IZFINX-NEXT:    fcvt.w.s a0, s0, rne
-; RV32IZFINX-NEXT:    fcvt.s.w a0, a0, rne
-; RV32IZFINX-NEXT:    fsgnj.s s0, a0, s0
+; RV32IZFINX-NEXT:    fcvt.w.s a1, s0, rne
+; RV32IZFINX-NEXT:    fcvt.s.w a1, a1, rne
+; RV32IZFINX-NEXT:    fsgnj.s s0, a1, s0
 ; RV32IZFINX-NEXT:  .LBB19_2:
-; RV32IZFINX-NEXT:    fle.s a0, zero, s0
-; RV32IZFINX-NEXT:    neg s1, a0
+; RV32IZFINX-NEXT:    fle.s a1, zero, s0
+; RV32IZFINX-NEXT:    neg s1, a1
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixunssfdi
 ; RV32IZFINX-NEXT:    lui a2, 391168
@@ -1332,11 +1342,12 @@ define i64 @test_rint_si64(float %x) nounwind {
 ; RV32IF-NEXT:    fle.s s0, fa5, fs0
 ; RV32IF-NEXT:    fmv.s fa0, fs0
 ; RV32IF-NEXT:    call __fixsfdi
+; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lui a4, 524288
 ; RV32IF-NEXT:    lui a3, 524288
-; RV32IF-NEXT:    lui a2, 524288
 ; RV32IF-NEXT:    beqz s0, .LBB21_4
 ; RV32IF-NEXT:  # %bb.3:
-; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    mv a3, a1
 ; RV32IF-NEXT:  .LBB21_4:
 ; RV32IF-NEXT:    lui a1, 389120
 ; RV32IF-NEXT:    addi a1, a1, -1
@@ -1344,16 +1355,16 @@ define i64 @test_rint_si64(float %x) nounwind {
 ; RV32IF-NEXT:    flt.s a1, fa5, fs0
 ; RV32IF-NEXT:    beqz a1, .LBB21_6
 ; RV32IF-NEXT:  # %bb.5:
-; RV32IF-NEXT:    addi a2, a3, -1
+; RV32IF-NEXT:    addi a3, a4, -1
 ; RV32IF-NEXT:  .LBB21_6:
-; RV32IF-NEXT:    feq.s a3, fs0, fs0
-; RV32IF-NEXT:    neg a4, s0
-; RV32IF-NEXT:    and a0, a4, a0
+; RV32IF-NEXT:    feq.s a4, fs0, fs0
+; RV32IF-NEXT:    neg a2, a2
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    neg a2, a4
 ; RV32IF-NEXT:    or a0, a1, a0
-; RV32IF-NEXT:    and a1, a3, a2
-; RV32IF-NEXT:    and a0, a3, a0
+; RV32IF-NEXT:    and a1, a2, a3
+; RV32IF-NEXT:    and a0, a2, a0
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -1389,27 +1400,28 @@ define i64 @test_rint_si64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixsfdi
+; RV32IZFINX-NEXT:    mv a2, s1
+; RV32IZFINX-NEXT:    lui a4, 524288
 ; RV32IZFINX-NEXT:    lui a3, 524288
-; RV32IZFINX-NEXT:    lui a2, 524288
 ; RV32IZFINX-NEXT:    beqz s1, .LBB21_4
 ; RV32IZFINX-NEXT:  # %bb.3:
-; RV32IZFINX-NEXT:    mv a2, a1
+; RV32IZFINX-NEXT:    mv a3, a1
 ; RV32IZFINX-NEXT:  .LBB21_4:
 ; RV32IZFINX-NEXT:    lui a1, 389120
 ; RV32IZFINX-NEXT:    addi a1, a1, -1
 ; RV32IZFINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZFINX-NEXT:    beqz a1, .LBB21_6
 ; RV32IZFINX-NEXT:  # %bb.5:
-; RV32IZFINX-NEXT:    addi a2, a3, -1
+; RV32IZFINX-NEXT:    addi a3, a4, -1
 ; RV32IZFINX-NEXT:  .LBB21_6:
-; RV32IZFINX-NEXT:    feq.s a3, s0, s0
-; RV32IZFINX-NEXT:    neg a4, s1
-; RV32IZFINX-NEXT:    and a0, a4, a0
+; RV32IZFINX-NEXT:    feq.s a4, s0, s0
+; RV32IZFINX-NEXT:    neg a2, a2
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    neg a1, a1
-; RV32IZFINX-NEXT:    neg a3, a3
+; RV32IZFINX-NEXT:    neg a2, a4
 ; RV32IZFINX-NEXT:    or a0, a1, a0
-; RV32IZFINX-NEXT:    and a1, a3, a2
-; RV32IZFINX-NEXT:    and a0, a3, a0
+; RV32IZFINX-NEXT:    and a1, a2, a3
+; RV32IZFINX-NEXT:    and a0, a2, a0
 ; RV32IZFINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -1506,17 +1518,17 @@ define i64 @test_rint_ui64(float %x) nounwind {
 ; RV32IZFINX-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32IZFINX-NEXT:    mv s0, a0
-; RV32IZFINX-NEXT:    fabs.s a0, a0
-; RV32IZFINX-NEXT:    lui a1, 307200
-; RV32IZFINX-NEXT:    flt.s a0, a0, a1
-; RV32IZFINX-NEXT:    beqz a0, .LBB23_2
+; RV32IZFINX-NEXT:    fabs.s a1, a0
+; RV32IZFINX-NEXT:    lui a2, 307200
+; RV32IZFINX-NEXT:    flt.s a1, a1, a2
+; RV32IZFINX-NEXT:    beqz a1, .LBB23_2
 ; RV32IZFINX-NEXT:  # %bb.1:
-; RV32IZFINX-NEXT:    fcvt.w.s a0, s0
-; RV32IZFINX-NEXT:    fcvt.s.w a0, a0
-; RV32IZFINX-NEXT:    fsgnj.s s0, a0, s0
+; RV32IZFINX-NEXT:    fcvt.w.s a1, s0
+; RV32IZFINX-NEXT:    fcvt.s.w a1, a1
+; RV32IZFINX-NEXT:    fsgnj.s s0, a1, s0
 ; RV32IZFINX-NEXT:  .LBB23_2:
-; RV32IZFINX-NEXT:    fle.s a0, zero, s0
-; RV32IZFINX-NEXT:    neg s1, a0
+; RV32IZFINX-NEXT:    fle.s a1, zero, s0
+; RV32IZFINX-NEXT:    neg s1, a1
 ; RV32IZFINX-NEXT:    mv a0, s0
 ; RV32IZFINX-NEXT:    call __fixunssfdi
 ; RV32IZFINX-NEXT:    lui a2, 391168
diff --git a/llvm/test/CodeGen/RISCV/fp128.ll b/llvm/test/CodeGen/RISCV/fp128.ll
index 19a4a507235135..8bbec91b4a73e4 100644
--- a/llvm/test/CodeGen/RISCV/fp128.ll
+++ b/llvm/test/CodeGen/RISCV/fp128.ll
@@ -318,212 +318,233 @@ define { fp128, fp128 } @modf(fp128 %a) nounwind {
 define i96 @fptosi_fp128_to_i96(fp128 %a) nounwind {
 ; RV32I-LABEL: fptosi_fp128_to_i96:
 ; RV32I:       # %bb.0: # %fp-to-i-entry
-; RV32I-NEXT:    addi sp, sp, -112
-; RV32I-NEXT:    sw ra, 108(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 104(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 100(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 96(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 92(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 88(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 84(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s6, 80(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s7, 76(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s8, 72(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s9, 68(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s10, 64(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    lui a6, 4
-; RV32I-NEXT:    lw a5, 12(a1)
-; RV32I-NEXT:    slli a4, a5, 1
-; RV32I-NEXT:    srli a4, a4, 17
-; RV32I-NEXT:    addi a2, a6, -1
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    bgeu a4, a2, .LBB5_2
+; RV32I-NEXT:    addi sp, sp, -128
+; RV32I-NEXT:    sw ra, 124(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lui a5, 4
+; RV32I-NEXT:    lw a7, 12(a1)
+; RV32I-NEXT:    slli a2, a7, 1
+; RV32I-NEXT:    srli a6, a2, 17
+; RV32I-NEXT:    addi a2, a5, -1
+; RV32I-NEXT:    bgeu a6, a2, .LBB5_2
 ; RV32I-NEXT:  # %bb.1:
-; RV32I-NEXT:    li s3, 0
-; RV32I-NEXT:    li s8, 0
-; RV32I-NEXT:    li a0, 0
-; RV32I-NEXT:    j .LBB5_5
+; RV32I-NEXT:    li a2, 0
+; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    li a3, 0
+; RV32I-NEXT:    j .LBB5_6
 ; RV32I-NEXT:  .LBB5_2: # %fp-to-i-if-check.exp.size
+; RV32I-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lw a3, 0(a1)
 ; RV32I-NEXT:    lw a2, 4(a1)
 ; RV32I-NEXT:    lw a0, 8(a1)
-; RV32I-NEXT:    srai s1, a5, 31
-; RV32I-NEXT:    addi a1, a6, 110
-; RV32I-NEXT:    ori s2, s1, 1
-; RV32I-NEXT:    bltu a1, a4, .LBB5_4
+; RV32I-NEXT:    srai a4, a7, 31
+; RV32I-NEXT:    addi a1, a5, 110
+; RV32I-NEXT:    ori a5, a4, 1
+; RV32I-NEXT:    sw a4, 52(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a5, 48(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    bltu a1, a6, .LBB5_4
 ; RV32I-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32I-NEXT:    slli a5, a5, 16
+; RV32I-NEXT:    slli a7, a7, 16
 ; RV32I-NEXT:    lui a1, 16
-; RV32I-NEXT:    srli a5, a5, 16
-; RV32I-NEXT:    addi a6, a6, 111
-; RV32I-NEXT:    or a1, a5, a1
-; RV32I-NEXT:    sub a4, a6, a4
-; RV32I-NEXT:    srli a5, a4, 3
-; RV32I-NEXT:    sw zero, 16(sp)
-; RV32I-NEXT:    sw zero, 20(sp)
-; RV32I-NEXT:    sw zero, 24(sp)
-; RV32I-NEXT:    sw zero, 28(sp)
-; RV32I-NEXT:    andi a5, a5, 12
-; RV32I-NEXT:    mv a6, sp
-; RV32I-NEXT:    sw a3, 0(sp)
-; RV32I-NEXT:    sw a2, 4(sp)
-; RV32I-NEXT:    sw a0, 8(sp)
-; RV32I-NEXT:    sw a1, 12(sp)
-; RV32I-NEXT:    add a5, a6, a5
-; RV32I-NEXT:    lw a0, 12(a5)
-; RV32I-NEXT:    lw a1, 8(a5)
-; RV32I-NEXT:    lw a2, 4(a5)
-; RV32I-NEXT:    lw a3, 0(a5)
-; RV32I-NEXT:    andi a5, a4, 31
-; RV32I-NEXT:    xori a5, a5, 31
-; RV32I-NEXT:    slli a6, a0, 1
-; RV32I-NEXT:    srl a7, a1, a4
-; RV32I-NEXT:    sll a6, a6, a5
-; RV32I-NEXT:    or s4, a7, a6
-; RV32I-NEXT:    slli a1, a1, 1
-; RV32I-NEXT:    srl a6, a2, a4
-; RV32I-NEXT:    sll a1, a1, a5
-; RV32I-NEXT:    or s5, a6, a1
+; RV32I-NEXT:    lui t0, 4
+; RV32I-NEXT:    srli a7, a7, 16
+; RV32I-NEXT:    addi t0, t0, 111
+; RV32I-NEXT:    or a1, a7, a1
+; RV32I-NEXT:    sub a6, t0, a6
+; RV32I-NEXT:    srli a7, a6, 3
+; RV32I-NEXT:    sw zero, 72(sp)
+; RV32I-NEXT:    sw zero, 76(sp)
+; RV32I-NEXT:    sw zero, 80(sp)
+; RV32I-NEXT:    sw zero, 84(sp)
+; RV32I-NEXT:    andi a7, a7, 12
+; RV32I-NEXT:    addi t0, sp, 56
+; RV32I-NEXT:    sw a3, 56(sp)
+; RV32I-NEXT:    sw a2, 60(sp)
+; RV32I-NEXT:    sw a0, 64(sp)
+; RV32I-NEXT:    sw a1, 68(sp)
+; RV32I-NEXT:    add a7, t0, a7
+; RV32I-NEXT:    lw a1, 12(a7)
+; RV32I-NEXT:    lw a2, 8(a7)
+; RV32I-NEXT:    lw a3, 4(a7)
+; RV32I-NEXT:    lw a7, 0(a7)
+; RV32I-NEXT:    andi a0, a6, 31
+; RV32I-NEXT:    xori t0, a0, 31
+; RV32I-NEXT:    slli a0, a1, 1
+; RV32I-NEXT:    srl t1, a2, a6
+; RV32I-NEXT:    sll a0, a0, t0
+; RV32I-NEXT:    or a0, t1, a0
+; RV32I-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    slli a2, a2, 1
-; RV32I-NEXT:    srl a1, a3, a4
-; RV32I-NEXT:    sll a2, a2, a5
-; RV32I-NEXT:    or s6, a1, a2
-; RV32I-NEXT:    srl s7, a0, a4
-; RV32I-NEXT:    mv a0, s6
+; RV32I-NEXT:    srl a0, a3, a6
+; RV32I-NEXT:    sll a2, a2, t0
+; RV32I-NEXT:    or a0, a0, a2
+; RV32I-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    slli a3, a3, 1
+; RV32I-NEXT:    srl a0, a7, a6
+; RV32I-NEXT:    sll a2, a3, t0
+; RV32I-NEXT:    or a0, a0, a2
+; RV32I-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    srl a0, a1, a6
+; RV32I-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s2
+; RV32I-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    mv s3, a0
-; RV32I-NEXT:    mv s8, a1
-; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a1, 44(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s2
+; RV32I-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    add s8, a0, s8
-; RV32I-NEXT:    sltu a0, s8, a0
-; RV32I-NEXT:    add s9, a1, a0
-; RV32I-NEXT:    mv a0, s6
+; RV32I-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a2, a0, a2
+; RV32I-NEXT:    sw a2, 44(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    sltu a0, a2, a0
+; RV32I-NEXT:    add a0, a1, a0
+; RV32I-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s1
+; RV32I-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    add s8, a0, s8
-; RV32I-NEXT:    sltu a0, s8, a0
-; RV32I-NEXT:    add s10, a1, a0
-; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a2, a0, a2
+; RV32I-NEXT:    sw a2, 44(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    sltu a0, a2, a0
+; RV32I-NEXT:    add a0, a1, a0
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s1
+; RV32I-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    add a0, a0, s9
-; RV32I-NEXT:    add s10, a0, s10
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a2, s6
-; RV32I-NEXT:    mv a3, s5
+; RV32I-NEXT:    lw a1, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a1, a0, a1
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a0, a1, a0
+; RV32I-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 52(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a3, 40(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    mv s5, a0
-; RV32I-NEXT:    mv a0, s2
-; RV32I-NEXT:    mv a1, s1
-; RV32I-NEXT:    mv a2, s4
-; RV32I-NEXT:    mv a3, s7
+; RV32I-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 48(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 52(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    add a0, s5, a0
-; RV32I-NEXT:    add a0, s10, a0
+; RV32I-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a3, 40(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a0, a3, a0
+; RV32I-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a3, a3, a0
 ; RV32I-NEXT:    j .LBB5_5
 ; RV32I-NEXT:  .LBB5_4: # %fp-to-i-if-exp.large
 ; RV32I-NEXT:    lui a1, 1048572
 ; RV32I-NEXT:    addi a1, a1, -111
-; RV32I-NEXT:    add a1, a4, a1
-; RV32I-NEXT:    sw zero, 32(sp)
-; RV32I-NEXT:    sw zero, 36(sp)
-; RV32I-NEXT:    sw zero, 40(sp)
-; RV32I-NEXT:    sw zero, 44(sp)
-; RV32I-NEXT:    srli a4, a1, 3
-; RV32I-NEXT:    addi a5, sp, 48
-; RV32I-NEXT:    andi a4, a4, 12
-; RV32I-NEXT:    sw a3, 48(sp)
-; RV32I-NEXT:    sw a2, 52(sp)
-; RV32I-NEXT:    sw a0, 56(sp)
-; RV32I-NEXT:    sub a5, a5, a4
-; RV32I-NEXT:    lw a0, 0(a5)
-; RV32I-NEXT:    lw a2, 4(a5)
+; RV32I-NEXT:    add a1, a6, a1
+; RV32I-NEXT:    sw zero, 88(sp)
+; RV32I-NEXT:    sw zero, 92(sp)
+; RV32I-NEXT:    sw zero, 96(sp)
+; RV32I-NEXT:    sw zero, 100(sp)
+; RV32I-NEXT:    srli a6, a1, 3
+; RV32I-NEXT:    addi a7, sp, 104
+; RV32I-NEXT:    andi a6, a6, 12
+; RV32I-NEXT:    sw a3, 104(sp)
+; RV32I-NEXT:    sw a2, 108(sp)
+; RV32I-NEXT:    sw a0, 112(sp)
+; RV32I-NEXT:    sub a0, a7, a6
+; RV32I-NEXT:    lw a6, 0(a0)
+; RV32I-NEXT:    lw a2, 4(a0)
 ; RV32I-NEXT:    andi a3, a1, 31
 ; RV32I-NEXT:    xori a3, a3, 31
-; RV32I-NEXT:    lw a4, 8(a5)
-; RV32I-NEXT:    lw a5, 12(a5)
-; RV32I-NEXT:    srli a6, a0, 1
-; RV32I-NEXT:    sll a7, a2, a1
-; RV32I-NEXT:    srl a6, a6, a3
+; RV32I-NEXT:    lw a7, 8(a0)
+; RV32I-NEXT:    lw t0, 12(a0)
+; RV32I-NEXT:    srli a0, a6, 1
+; RV32I-NEXT:    sll t1, a2, a1
+; RV32I-NEXT:    srl a0, a0, a3
 ; RV32I-NEXT:    srli a2, a2, 1
-; RV32I-NEXT:    or s4, a7, a6
-; RV32I-NEXT:    sll a6, a4, a1
+; RV32I-NEXT:    or a0, t1, a0
+; RV32I-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sll a0, a7, a1
 ; RV32I-NEXT:    srl a2, a2, a3
-; RV32I-NEXT:    or a2, a6, a2
-; RV32I-NEXT:    srli a4, a4, 1
-; RV32I-NEXT:    sll a5, a5, a1
-; RV32I-NEXT:    srl a3, a4, a3
-; RV32I-NEXT:    or a3, a5, a3
-; RV32I-NEXT:    sll s5, a0, a1
-; RV32I-NEXT:    mv a0, s2
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    or a2, a0, a2
+; RV32I-NEXT:    srli a0, a7, 1
+; RV32I-NEXT:    sll a7, t0, a1
+; RV32I-NEXT:    srl a0, a0, a3
+; RV32I-NEXT:    or a3, a7, a0
+; RV32I-NEXT:    sll a0, a6, a1
+; RV32I-NEXT:    sw a0, 44(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, a5
+; RV32I-NEXT:    mv a1, a4
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    mv s3, a0
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a2, s5
-; RV32I-NEXT:    mv a3, s4
+; RV32I-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 52(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a3, 40(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    add s9, a0, s3
-; RV32I-NEXT:    mv a0, s4
+; RV32I-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s2
+; RV32I-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    mv s6, a0
-; RV32I-NEXT:    mv s7, a1
-; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s2
+; RV32I-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    mv s3, a0
-; RV32I-NEXT:    add s8, s6, a1
-; RV32I-NEXT:    sltu a0, s8, s6
-; RV32I-NEXT:    add s7, s7, a0
-; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    sw a0, 48(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 48(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    sltu a1, a1, a0
+; RV32I-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s1
+; RV32I-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    add s8, a0, s8
-; RV32I-NEXT:    sltu a0, s8, a0
-; RV32I-NEXT:    add s2, a1, a0
-; RV32I-NEXT:    mv a0, s4
+; RV32I-NEXT:    lw a2, 48(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a2, a0, a2
+; RV32I-NEXT:    sw a2, 44(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    sltu a0, a2, a0
+; RV32I-NEXT:    add a0, a1, a0
+; RV32I-NEXT:    sw a0, 48(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a1, 0
-; RV32I-NEXT:    mv a2, s1
+; RV32I-NEXT:    lw a2, 52(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    li a3, 0
 ; RV32I-NEXT:    call __muldi3
-; RV32I-NEXT:    add a0, a0, s7
-; RV32I-NEXT:    add a0, a0, s2
-; RV32I-NEXT:    add a0, a0, s9
+; RV32I-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a3, 36(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a3, a0, a3
+; RV32I-NEXT:    lw a0, 48(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a0, a3, a0
+; RV32I-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    add a3, a0, a3
 ; RV32I-NEXT:  .LBB5_5: # %fp-to-i-cleanup
-; RV32I-NEXT:    sw s3, 0(s0)
-; RV32I-NEXT:    sw s8, 4(s0)
-; RV32I-NEXT:    sw a0, 8(s0)
-; RV32I-NEXT:    lw ra, 108(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 104(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 100(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 96(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 92(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 88(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 84(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s6, 80(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s7, 76(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s8, 72(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s9, 68(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s10, 64(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 112
+; RV32I-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
+; RV32I-NEXT:  .LBB5_6: # %fp-to-i-cleanup
+; RV32I-NEXT:    sw a2, 0(a0)
+; RV32I-NEXT:    sw a1, 4(a0)
+; RV32I-NEXT:    sw a3, 8(a0)
+; RV32I-NEXT:    lw ra, 124(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 128
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fptosi_fp128_to_i96:
diff --git a/llvm/test/CodeGen/RISCV/fpclamptosat.ll b/llvm/test/CodeGen/RISCV/fpclamptosat.ll
index cbf9b6a24b2e01..57da7689644280 100644
--- a/llvm/test/CodeGen/RISCV/fpclamptosat.ll
+++ b/llvm/test/CodeGen/RISCV/fpclamptosat.ll
@@ -1086,271 +1086,284 @@ define i64 @stest_f64i64(double %x) {
 ; RV32IF-NEXT:    addi sp, sp, -80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 80
 ; RV32IF-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s9, 36(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    .cfi_offset ra, -4
-; RV32IF-NEXT:    .cfi_offset s0, -8
-; RV32IF-NEXT:    .cfi_offset s1, -12
-; RV32IF-NEXT:    .cfi_offset s2, -16
-; RV32IF-NEXT:    .cfi_offset s3, -20
-; RV32IF-NEXT:    .cfi_offset s4, -24
-; RV32IF-NEXT:    .cfi_offset s5, -28
-; RV32IF-NEXT:    .cfi_offset s6, -32
-; RV32IF-NEXT:    .cfi_offset s7, -36
-; RV32IF-NEXT:    .cfi_offset s8, -40
-; RV32IF-NEXT:    .cfi_offset s9, -44
 ; RV32IF-NEXT:    li a3, 1023
 ; RV32IF-NEXT:    slli a2, a1, 1
 ; RV32IF-NEXT:    srli a2, a2, 21
 ; RV32IF-NEXT:    bgeu a2, a3, .LBB18_2
 ; RV32IF-NEXT:  # %bb.1:
-; RV32IF-NEXT:    li s1, 0
-; RV32IF-NEXT:    li s2, 0
-; RV32IF-NEXT:    li a0, 0
+; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    j .LBB18_8
+; RV32IF-NEXT:    li a0, 0
+; RV32IF-NEXT:    li a2, 0
+; RV32IF-NEXT:    j .LBB18_9
 ; RV32IF-NEXT:  .LBB18_2: # %fp-to-i-if-check.exp.size
-; RV32IF-NEXT:    srai s0, a1, 31
-; RV32IF-NEXT:    lui a3, 256
-; RV32IF-NEXT:    snez a4, a0
-; RV32IF-NEXT:    addi a3, a3, -1
+; RV32IF-NEXT:    mv a3, a1
+; RV32IF-NEXT:    srai a1, a1, 31
+; RV32IF-NEXT:    lui a4, 256
+; RV32IF-NEXT:    snez a5, a0
+; RV32IF-NEXT:    addi a4, a4, -1
 ; RV32IF-NEXT:    addi a0, a0, -1
-; RV32IF-NEXT:    and a1, a1, a3
+; RV32IF-NEXT:    and a3, a3, a4
 ; RV32IF-NEXT:    addi a0, a0, 1
-; RV32IF-NEXT:    add a1, a1, a4
-; RV32IF-NEXT:    seqz a4, a0
-; RV32IF-NEXT:    add a3, a4, a3
-; RV32IF-NEXT:    li a4, 1074
-; RV32IF-NEXT:    ori s2, s0, 1
-; RV32IF-NEXT:    add a1, a1, a3
-; RV32IF-NEXT:    bltu a4, a2, .LBB18_5
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    seqz a5, a0
+; RV32IF-NEXT:    add a5, a5, a4
+; RV32IF-NEXT:    li a6, 1074
+; RV32IF-NEXT:    ori a4, a1, 1
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    sw a1, 36(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a4, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    bltu a6, a2, .LBB18_5
 ; RV32IF-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    li a3, 1043
-; RV32IF-NEXT:    li a4, 1075
-; RV32IF-NEXT:    sub a3, a3, a2
-; RV32IF-NEXT:    sub a2, a4, a2
-; RV32IF-NEXT:    srai a4, a3, 31
-; RV32IF-NEXT:    srl a5, a1, a2
-; RV32IF-NEXT:    and s3, a4, a5
-; RV32IF-NEXT:    bltz a3, .LBB18_6
+; RV32IF-NEXT:    li a5, 1043
+; RV32IF-NEXT:    li a7, 1075
+; RV32IF-NEXT:    sub a6, a5, a2
+; RV32IF-NEXT:    sub a5, a7, a2
+; RV32IF-NEXT:    srai a2, a6, 31
+; RV32IF-NEXT:    srl a7, a3, a5
+; RV32IF-NEXT:    and a2, a2, a7
+; RV32IF-NEXT:    bltz a6, .LBB18_6
 ; RV32IF-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    srl s4, a1, a3
+; RV32IF-NEXT:    srl a0, a3, a6
 ; RV32IF-NEXT:    j .LBB18_7
 ; RV32IF-NEXT:  .LBB18_5: # %fp-to-i-if-exp.large
-; RV32IF-NEXT:    addi a4, a2, -1075
-; RV32IF-NEXT:    srli a2, a4, 3
-; RV32IF-NEXT:    sw a0, 16(sp)
-; RV32IF-NEXT:    sw a1, 20(sp)
-; RV32IF-NEXT:    sw zero, 24(sp)
-; RV32IF-NEXT:    sw zero, 28(sp)
+; RV32IF-NEXT:    addi a5, a2, -1075
+; RV32IF-NEXT:    srli a2, a5, 3
+; RV32IF-NEXT:    sw a0, 56(sp)
+; RV32IF-NEXT:    sw a3, 60(sp)
+; RV32IF-NEXT:    sw zero, 64(sp)
+; RV32IF-NEXT:    sw zero, 68(sp)
 ; RV32IF-NEXT:    andi a2, a2, 12
-; RV32IF-NEXT:    addi a0, sp, 16
-; RV32IF-NEXT:    sw zero, 0(sp)
-; RV32IF-NEXT:    sw zero, 4(sp)
-; RV32IF-NEXT:    sw zero, 8(sp)
-; RV32IF-NEXT:    sw zero, 12(sp)
+; RV32IF-NEXT:    addi a0, sp, 56
+; RV32IF-NEXT:    sw zero, 40(sp)
+; RV32IF-NEXT:    sw zero, 44(sp)
+; RV32IF-NEXT:    sw zero, 48(sp)
+; RV32IF-NEXT:    sw zero, 52(sp)
 ; RV32IF-NEXT:    sub a0, a0, a2
-; RV32IF-NEXT:    lw a1, 0(a0)
+; RV32IF-NEXT:    lw a6, 0(a0)
 ; RV32IF-NEXT:    lw a2, 4(a0)
-; RV32IF-NEXT:    andi a3, a4, 31
-; RV32IF-NEXT:    lw a5, 8(a0)
-; RV32IF-NEXT:    lw a0, 12(a0)
-; RV32IF-NEXT:    srli a6, a1, 1
+; RV32IF-NEXT:    andi a3, a5, 31
+; RV32IF-NEXT:    lw a7, 8(a0)
+; RV32IF-NEXT:    lw t0, 12(a0)
+; RV32IF-NEXT:    srli a0, a6, 1
 ; RV32IF-NEXT:    xori a3, a3, 31
-; RV32IF-NEXT:    sll a7, a2, a4
-; RV32IF-NEXT:    srl a6, a6, a3
-; RV32IF-NEXT:    or s3, a7, a6
+; RV32IF-NEXT:    sll t1, a2, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a0, t1, a0
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    srli a2, a2, 1
-; RV32IF-NEXT:    sll a6, a5, a4
+; RV32IF-NEXT:    sll a0, a7, a5
 ; RV32IF-NEXT:    srl a2, a2, a3
-; RV32IF-NEXT:    or a2, a6, a2
-; RV32IF-NEXT:    srli a5, a5, 1
-; RV32IF-NEXT:    sll a0, a0, a4
-; RV32IF-NEXT:    srl a3, a5, a3
-; RV32IF-NEXT:    or a3, a0, a3
-; RV32IF-NEXT:    sll s4, a1, a4
-; RV32IF-NEXT:    mv a0, s2
-; RV32IF-NEXT:    mv a1, s0
+; RV32IF-NEXT:    or a2, a0, a2
+; RV32IF-NEXT:    srli a0, a7, 1
+; RV32IF-NEXT:    sll a7, t0, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a3, a7, a0
+; RV32IF-NEXT:    sll a0, a6, a5
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    mv a0, a4
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s7, a0, s1
-; RV32IF-NEXT:    add a1, a1, s5
-; RV32IF-NEXT:    sltu a0, s7, a0
-; RV32IF-NEXT:    add s8, a1, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a1, a2
+; RV32IF-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    add a0, a2, a0
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s5, a0
-; RV32IF-NEXT:    mv s6, a1
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    add s2, s5, a1
-; RV32IF-NEXT:    sltu a0, s2, s5
-; RV32IF-NEXT:    add s6, s6, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s2
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s4, s6, a0
-; RV32IF-NEXT:    sltu s5, s4, s6
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s4, a0, s4
-; RV32IF-NEXT:    add a1, a1, s5
-; RV32IF-NEXT:    sltu a0, s4, a0
-; RV32IF-NEXT:    add a1, a1, a0
-; RV32IF-NEXT:    add a0, s4, s7
-; RV32IF-NEXT:    add a1, a1, s8
-; RV32IF-NEXT:    sltu a2, a0, s4
-; RV32IF-NEXT:    add a1, a1, a2
+; RV32IF-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a0, a4
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a2, a4
+; RV32IF-NEXT:    sltu a0, a5, a0
+; RV32IF-NEXT:    add a4, a2, a0
+; RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a5, a0
+; RV32IF-NEXT:    lw a2, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a4, a2
+; RV32IF-NEXT:    sltu a4, a0, a5
 ; RV32IF-NEXT:    j .LBB18_8
 ; RV32IF-NEXT:  .LBB18_6:
-; RV32IF-NEXT:    not a3, a2
-; RV32IF-NEXT:    slli a1, a1, 1
-; RV32IF-NEXT:    srl a0, a0, a2
-; RV32IF-NEXT:    sll a1, a1, a3
-; RV32IF-NEXT:    or s4, a0, a1
+; RV32IF-NEXT:    not a6, a5
+; RV32IF-NEXT:    slli a3, a3, 1
+; RV32IF-NEXT:    srl a0, a0, a5
+; RV32IF-NEXT:    sll a3, a3, a6
+; RV32IF-NEXT:    or a0, a0, a3
 ; RV32IF-NEXT:  .LBB18_7: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s5
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add s7, a1, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a0, a1, a0
+; RV32IF-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s2
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s8, s7, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s5, a0
-; RV32IF-NEXT:    mv s6, a1
-; RV32IF-NEXT:    add s9, a0, s8
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    sltu a2, s8, s7
-; RV32IF-NEXT:    sltu a3, s9, s5
-; RV32IF-NEXT:    add a2, s6, a2
-; RV32IF-NEXT:    add a0, s9, a0
-; RV32IF-NEXT:    add a2, a2, a3
-; RV32IF-NEXT:    sltu a3, a0, s9
-; RV32IF-NEXT:    add a1, a2, a1
-; RV32IF-NEXT:    add a1, a1, a3
+; RV32IF-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a5, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a6, a5, a4
+; RV32IF-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a5, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a5, a5, a4
+; RV32IF-NEXT:    lw a4, 0(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a6, a4, a6
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a4, a0
+; RV32IF-NEXT:    add a5, a6, a5
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a4, a0, a4
+; RV32IF-NEXT:    add a2, a5, a2
 ; RV32IF-NEXT:  .LBB18_8: # %fp-to-i-cleanup
-; RV32IF-NEXT:    lui a2, 524288
-; RV32IF-NEXT:    addi a3, a2, -1
-; RV32IF-NEXT:    beq s2, a3, .LBB18_10
-; RV32IF-NEXT:  # %bb.9: # %fp-to-i-cleanup
-; RV32IF-NEXT:    sltu a4, s2, a3
-; RV32IF-NEXT:    or a5, a0, a1
-; RV32IF-NEXT:    bnez a5, .LBB18_11
-; RV32IF-NEXT:    j .LBB18_12
-; RV32IF-NEXT:  .LBB18_10:
-; RV32IF-NEXT:    sltiu a4, s1, -1
-; RV32IF-NEXT:    or a5, a0, a1
-; RV32IF-NEXT:    beqz a5, .LBB18_12
-; RV32IF-NEXT:  .LBB18_11: # %fp-to-i-cleanup
-; RV32IF-NEXT:    srli a4, a1, 31
+; RV32IF-NEXT:    add a2, a2, a4
+; RV32IF-NEXT:  .LBB18_9: # %fp-to-i-cleanup
+; RV32IF-NEXT:    lui a4, 524288
+; RV32IF-NEXT:    addi a5, a4, -1
+; RV32IF-NEXT:    beq a1, a5, .LBB18_11
+; RV32IF-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IF-NEXT:    sltu a6, a1, a5
+; RV32IF-NEXT:    or a7, a0, a2
+; RV32IF-NEXT:    bnez a7, .LBB18_12
+; RV32IF-NEXT:    j .LBB18_13
+; RV32IF-NEXT:  .LBB18_11:
+; RV32IF-NEXT:    sltiu a6, a3, -1
+; RV32IF-NEXT:    or a7, a0, a2
+; RV32IF-NEXT:    beqz a7, .LBB18_13
 ; RV32IF-NEXT:  .LBB18_12: # %fp-to-i-cleanup
-; RV32IF-NEXT:    neg a5, a4
-; RV32IF-NEXT:    addi a6, a4, -1
-; RV32IF-NEXT:    bnez a4, .LBB18_14
-; RV32IF-NEXT:  # %bb.13: # %fp-to-i-cleanup
-; RV32IF-NEXT:    mv s2, a3
-; RV32IF-NEXT:  .LBB18_14: # %fp-to-i-cleanup
-; RV32IF-NEXT:    or a3, a6, s1
-; RV32IF-NEXT:    and a1, a5, a1
-; RV32IF-NEXT:    and a4, a5, a0
-; RV32IF-NEXT:    beq s2, a2, .LBB18_16
-; RV32IF-NEXT:  # %bb.15: # %fp-to-i-cleanup
-; RV32IF-NEXT:    sltu a0, a2, s2
-; RV32IF-NEXT:    j .LBB18_17
-; RV32IF-NEXT:  .LBB18_16:
+; RV32IF-NEXT:    srli a6, a2, 31
+; RV32IF-NEXT:  .LBB18_13: # %fp-to-i-cleanup
+; RV32IF-NEXT:    neg a7, a6
+; RV32IF-NEXT:    addi t0, a6, -1
+; RV32IF-NEXT:    bnez a6, .LBB18_15
+; RV32IF-NEXT:  # %bb.14: # %fp-to-i-cleanup
+; RV32IF-NEXT:    mv a1, a5
+; RV32IF-NEXT:  .LBB18_15: # %fp-to-i-cleanup
+; RV32IF-NEXT:    or a3, t0, a3
+; RV32IF-NEXT:    and a2, a7, a2
+; RV32IF-NEXT:    and a5, a7, a0
+; RV32IF-NEXT:    beq a1, a4, .LBB18_17
+; RV32IF-NEXT:  # %bb.16: # %fp-to-i-cleanup
+; RV32IF-NEXT:    sltu a0, a4, a1
+; RV32IF-NEXT:    j .LBB18_18
+; RV32IF-NEXT:  .LBB18_17:
 ; RV32IF-NEXT:    snez a0, a3
-; RV32IF-NEXT:  .LBB18_17: # %fp-to-i-cleanup
-; RV32IF-NEXT:    and a4, a4, a1
-; RV32IF-NEXT:    li a2, -1
-; RV32IF-NEXT:    beq a4, a2, .LBB18_19
-; RV32IF-NEXT:  # %bb.18: # %fp-to-i-cleanup
-; RV32IF-NEXT:    srli a1, a1, 31
-; RV32IF-NEXT:    xori a0, a1, 1
-; RV32IF-NEXT:  .LBB18_19: # %fp-to-i-cleanup
-; RV32IF-NEXT:    bnez a0, .LBB18_21
-; RV32IF-NEXT:  # %bb.20: # %fp-to-i-cleanup
-; RV32IF-NEXT:    lui s2, 524288
-; RV32IF-NEXT:  .LBB18_21: # %fp-to-i-cleanup
+; RV32IF-NEXT:  .LBB18_18: # %fp-to-i-cleanup
+; RV32IF-NEXT:    and a5, a5, a2
+; RV32IF-NEXT:    li a4, -1
+; RV32IF-NEXT:    beq a5, a4, .LBB18_20
+; RV32IF-NEXT:  # %bb.19: # %fp-to-i-cleanup
+; RV32IF-NEXT:    srli a2, a2, 31
+; RV32IF-NEXT:    xori a0, a2, 1
+; RV32IF-NEXT:  .LBB18_20: # %fp-to-i-cleanup
+; RV32IF-NEXT:    bnez a0, .LBB18_22
+; RV32IF-NEXT:  # %bb.21: # %fp-to-i-cleanup
+; RV32IF-NEXT:    lui a1, 524288
+; RV32IF-NEXT:  .LBB18_22: # %fp-to-i-cleanup
 ; RV32IF-NEXT:    neg a0, a0
 ; RV32IF-NEXT:    and a0, a0, a3
-; RV32IF-NEXT:    mv a1, s2
 ; RV32IF-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s9, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    .cfi_restore ra
-; RV32IF-NEXT:    .cfi_restore s0
-; RV32IF-NEXT:    .cfi_restore s1
-; RV32IF-NEXT:    .cfi_restore s2
-; RV32IF-NEXT:    .cfi_restore s3
-; RV32IF-NEXT:    .cfi_restore s4
-; RV32IF-NEXT:    .cfi_restore s5
-; RV32IF-NEXT:    .cfi_restore s6
-; RV32IF-NEXT:    .cfi_restore s7
-; RV32IF-NEXT:    .cfi_restore s8
-; RV32IF-NEXT:    .cfi_restore s9
 ; RV32IF-NEXT:    addi sp, sp, 80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IF-NEXT:    ret
@@ -1401,274 +1414,286 @@ define i64 @stest_f64i64(double %x) {
 ; RV32IFD-NEXT:    addi sp, sp, -96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 96
 ; RV32IFD-NEXT:    sw ra, 92(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s0, 88(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s1, 84(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s2, 80(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s3, 76(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s4, 72(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s5, 68(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s6, 64(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s7, 60(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s8, 56(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s9, 52(sp) # 4-byte Folded Spill
 ; RV32IFD-NEXT:    .cfi_offset ra, -4
-; RV32IFD-NEXT:    .cfi_offset s0, -8
-; RV32IFD-NEXT:    .cfi_offset s1, -12
-; RV32IFD-NEXT:    .cfi_offset s2, -16
-; RV32IFD-NEXT:    .cfi_offset s3, -20
-; RV32IFD-NEXT:    .cfi_offset s4, -24
-; RV32IFD-NEXT:    .cfi_offset s5, -28
-; RV32IFD-NEXT:    .cfi_offset s6, -32
-; RV32IFD-NEXT:    .cfi_offset s7, -36
-; RV32IFD-NEXT:    .cfi_offset s8, -40
-; RV32IFD-NEXT:    .cfi_offset s9, -44
-; RV32IFD-NEXT:    li a2, 1023
-; RV32IFD-NEXT:    fsd fa0, 8(sp)
-; RV32IFD-NEXT:    lw a1, 12(sp)
-; RV32IFD-NEXT:    slli a0, a1, 1
+; RV32IFD-NEXT:    li a1, 1023
+; RV32IFD-NEXT:    fsd fa0, 48(sp)
+; RV32IFD-NEXT:    lw a2, 52(sp)
+; RV32IFD-NEXT:    slli a0, a2, 1
 ; RV32IFD-NEXT:    srli a0, a0, 21
-; RV32IFD-NEXT:    bgeu a0, a2, .LBB18_2
+; RV32IFD-NEXT:    bgeu a0, a1, .LBB18_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    li s1, 0
-; RV32IFD-NEXT:    li s2, 0
-; RV32IFD-NEXT:    li a0, 0
+; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    j .LBB18_8
+; RV32IFD-NEXT:    li a0, 0
+; RV32IFD-NEXT:    li a2, 0
+; RV32IFD-NEXT:    j .LBB18_9
 ; RV32IFD-NEXT:  .LBB18_2: # %fp-to-i-if-check.exp.size
-; RV32IFD-NEXT:    lw a2, 8(sp)
-; RV32IFD-NEXT:    lui a3, 256
-; RV32IFD-NEXT:    srai s0, a1, 31
-; RV32IFD-NEXT:    addi a3, a3, -1
-; RV32IFD-NEXT:    and a4, a1, a3
-; RV32IFD-NEXT:    addi a1, a2, -1
-; RV32IFD-NEXT:    snez a2, a2
-; RV32IFD-NEXT:    addi a1, a1, 1
-; RV32IFD-NEXT:    add a2, a4, a2
-; RV32IFD-NEXT:    seqz a4, a1
-; RV32IFD-NEXT:    add a2, a2, a3
-; RV32IFD-NEXT:    li a3, 1074
-; RV32IFD-NEXT:    ori s2, s0, 1
-; RV32IFD-NEXT:    add a2, a2, a4
-; RV32IFD-NEXT:    bltu a3, a0, .LBB18_5
+; RV32IFD-NEXT:    lw a3, 48(sp)
+; RV32IFD-NEXT:    lui a4, 256
+; RV32IFD-NEXT:    srai a1, a2, 31
+; RV32IFD-NEXT:    addi a4, a4, -1
+; RV32IFD-NEXT:    and a5, a2, a4
+; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    snez a3, a3
+; RV32IFD-NEXT:    addi a2, a2, 1
+; RV32IFD-NEXT:    add a3, a5, a3
+; RV32IFD-NEXT:    seqz a5, a2
+; RV32IFD-NEXT:    add a3, a3, a4
+; RV32IFD-NEXT:    li a6, 1074
+; RV32IFD-NEXT:    ori a4, a1, 1
+; RV32IFD-NEXT:    add a3, a3, a5
+; RV32IFD-NEXT:    sw a1, 44(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a4, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    bltu a6, a0, .LBB18_5
 ; RV32IFD-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    li a3, 1043
-; RV32IFD-NEXT:    li a4, 1075
-; RV32IFD-NEXT:    sub a3, a3, a0
-; RV32IFD-NEXT:    sub a0, a4, a0
-; RV32IFD-NEXT:    srai a4, a3, 31
-; RV32IFD-NEXT:    srl a5, a2, a0
-; RV32IFD-NEXT:    and s3, a4, a5
-; RV32IFD-NEXT:    bltz a3, .LBB18_6
+; RV32IFD-NEXT:    li a5, 1043
+; RV32IFD-NEXT:    li a7, 1075
+; RV32IFD-NEXT:    sub a6, a5, a0
+; RV32IFD-NEXT:    sub a5, a7, a0
+; RV32IFD-NEXT:    srai a0, a6, 31
+; RV32IFD-NEXT:    srl a7, a3, a5
+; RV32IFD-NEXT:    and a0, a0, a7
+; RV32IFD-NEXT:    bltz a6, .LBB18_6
 ; RV32IFD-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    srl s4, a2, a3
+; RV32IFD-NEXT:    srl a2, a3, a6
 ; RV32IFD-NEXT:    j .LBB18_7
 ; RV32IFD-NEXT:  .LBB18_5: # %fp-to-i-if-exp.large
-; RV32IFD-NEXT:    addi a0, a0, -1075
-; RV32IFD-NEXT:    srli a3, a0, 3
-; RV32IFD-NEXT:    sw a1, 32(sp)
-; RV32IFD-NEXT:    sw a2, 36(sp)
-; RV32IFD-NEXT:    sw zero, 40(sp)
-; RV32IFD-NEXT:    sw zero, 44(sp)
-; RV32IFD-NEXT:    andi a3, a3, 12
-; RV32IFD-NEXT:    addi a1, sp, 32
-; RV32IFD-NEXT:    sw zero, 16(sp)
-; RV32IFD-NEXT:    sw zero, 20(sp)
-; RV32IFD-NEXT:    sw zero, 24(sp)
-; RV32IFD-NEXT:    sw zero, 28(sp)
-; RV32IFD-NEXT:    sub a1, a1, a3
-; RV32IFD-NEXT:    lw a4, 0(a1)
-; RV32IFD-NEXT:    lw a2, 4(a1)
-; RV32IFD-NEXT:    andi a3, a0, 31
-; RV32IFD-NEXT:    lw a5, 8(a1)
-; RV32IFD-NEXT:    lw a1, 12(a1)
-; RV32IFD-NEXT:    srli a6, a4, 1
-; RV32IFD-NEXT:    xori a3, a3, 31
-; RV32IFD-NEXT:    sll a7, a2, a0
-; RV32IFD-NEXT:    srl a6, a6, a3
-; RV32IFD-NEXT:    or s3, a7, a6
-; RV32IFD-NEXT:    srli a2, a2, 1
-; RV32IFD-NEXT:    sll a6, a5, a0
-; RV32IFD-NEXT:    srl a2, a2, a3
-; RV32IFD-NEXT:    or a2, a6, a2
-; RV32IFD-NEXT:    srli a5, a5, 1
-; RV32IFD-NEXT:    sll a1, a1, a0
-; RV32IFD-NEXT:    srl a3, a5, a3
-; RV32IFD-NEXT:    or a3, a1, a3
-; RV32IFD-NEXT:    sll s4, a4, a0
-; RV32IFD-NEXT:    mv a0, s2
-; RV32IFD-NEXT:    mv a1, s0
+; RV32IFD-NEXT:    addi a5, a0, -1075
+; RV32IFD-NEXT:    srli a0, a5, 3
+; RV32IFD-NEXT:    sw a2, 72(sp)
+; RV32IFD-NEXT:    sw a3, 76(sp)
+; RV32IFD-NEXT:    sw zero, 80(sp)
+; RV32IFD-NEXT:    sw zero, 84(sp)
+; RV32IFD-NEXT:    andi a0, a0, 12
+; RV32IFD-NEXT:    addi a2, sp, 72
+; RV32IFD-NEXT:    sw zero, 56(sp)
+; RV32IFD-NEXT:    sw zero, 60(sp)
+; RV32IFD-NEXT:    sw zero, 64(sp)
+; RV32IFD-NEXT:    sw zero, 68(sp)
+; RV32IFD-NEXT:    sub a2, a2, a0
+; RV32IFD-NEXT:    lw a6, 0(a2)
+; RV32IFD-NEXT:    lw a3, 4(a2)
+; RV32IFD-NEXT:    andi a0, a5, 31
+; RV32IFD-NEXT:    lw a7, 8(a2)
+; RV32IFD-NEXT:    lw t0, 12(a2)
+; RV32IFD-NEXT:    srli a2, a6, 1
+; RV32IFD-NEXT:    xori t1, a0, 31
+; RV32IFD-NEXT:    sll a0, a3, a5
+; RV32IFD-NEXT:    srl a2, a2, t1
+; RV32IFD-NEXT:    or a0, a0, a2
+; RV32IFD-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    srli a3, a3, 1
+; RV32IFD-NEXT:    sll a0, a7, a5
+; RV32IFD-NEXT:    srl a2, a3, t1
+; RV32IFD-NEXT:    or a2, a0, a2
+; RV32IFD-NEXT:    srli a0, a7, 1
+; RV32IFD-NEXT:    sll a3, t0, a5
+; RV32IFD-NEXT:    srl a0, a0, t1
+; RV32IFD-NEXT:    or a3, a3, a0
+; RV32IFD-NEXT:    sll a0, a6, a5
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    mv a0, a4
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s7, a0, s1
-; RV32IFD-NEXT:    add a1, a1, s5
-; RV32IFD-NEXT:    sltu a0, s7, a0
-; RV32IFD-NEXT:    add s8, a1, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a1, a2
+; RV32IFD-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    add a0, a2, a0
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s5, a0
-; RV32IFD-NEXT:    mv s6, a1
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    add s2, s5, a1
-; RV32IFD-NEXT:    sltu a0, s2, s5
-; RV32IFD-NEXT:    add s6, s6, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s2
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s4, s6, a0
-; RV32IFD-NEXT:    sltu s5, s4, s6
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s4, a0, s4
-; RV32IFD-NEXT:    add a1, a1, s5
-; RV32IFD-NEXT:    sltu a0, s4, a0
-; RV32IFD-NEXT:    add a1, a1, a0
-; RV32IFD-NEXT:    add a0, s4, s7
-; RV32IFD-NEXT:    add a1, a1, s8
-; RV32IFD-NEXT:    sltu a2, a0, s4
-; RV32IFD-NEXT:    add a1, a1, a2
+; RV32IFD-NEXT:    lw a3, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a0, a4
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a2, a4
+; RV32IFD-NEXT:    sltu a0, a5, a0
+; RV32IFD-NEXT:    add a4, a2, a0
+; RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a5, a0
+; RV32IFD-NEXT:    lw a2, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a4, a2
+; RV32IFD-NEXT:    sltu a4, a0, a5
 ; RV32IFD-NEXT:    j .LBB18_8
 ; RV32IFD-NEXT:  .LBB18_6:
-; RV32IFD-NEXT:    not a3, a0
-; RV32IFD-NEXT:    slli a2, a2, 1
-; RV32IFD-NEXT:    srl a0, a1, a0
-; RV32IFD-NEXT:    sll a1, a2, a3
-; RV32IFD-NEXT:    or s4, a0, a1
+; RV32IFD-NEXT:    not a6, a5
+; RV32IFD-NEXT:    slli a3, a3, 1
+; RV32IFD-NEXT:    srl a2, a2, a5
+; RV32IFD-NEXT:    sll a3, a3, a6
+; RV32IFD-NEXT:    or a2, a2, a3
 ; RV32IFD-NEXT:  .LBB18_7: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a2, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s5
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add s7, a1, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s2
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s8, s7, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s5, a0
-; RV32IFD-NEXT:    mv s6, a1
-; RV32IFD-NEXT:    add s9, a0, s8
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    sltu a2, s8, s7
-; RV32IFD-NEXT:    sltu a3, s9, s5
-; RV32IFD-NEXT:    add a2, s6, a2
-; RV32IFD-NEXT:    add a0, s9, a0
-; RV32IFD-NEXT:    add a2, a2, a3
-; RV32IFD-NEXT:    sltu a3, a0, s9
-; RV32IFD-NEXT:    add a1, a2, a1
-; RV32IFD-NEXT:    add a1, a1, a3
+; RV32IFD-NEXT:    lw a3, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a5, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a6, a5, a4
+; RV32IFD-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a5, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a5, a5, a4
+; RV32IFD-NEXT:    lw a4, 8(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a6, a4, a6
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a4, a0
+; RV32IFD-NEXT:    add a5, a6, a5
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a4, a0, a4
+; RV32IFD-NEXT:    add a2, a5, a2
 ; RV32IFD-NEXT:  .LBB18_8: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    lui a2, 524288
-; RV32IFD-NEXT:    addi a3, a2, -1
-; RV32IFD-NEXT:    beq s2, a3, .LBB18_10
-; RV32IFD-NEXT:  # %bb.9: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    sltu a4, s2, a3
-; RV32IFD-NEXT:    or a5, a0, a1
-; RV32IFD-NEXT:    bnez a5, .LBB18_11
-; RV32IFD-NEXT:    j .LBB18_12
-; RV32IFD-NEXT:  .LBB18_10:
-; RV32IFD-NEXT:    sltiu a4, s1, -1
-; RV32IFD-NEXT:    or a5, a0, a1
-; RV32IFD-NEXT:    beqz a5, .LBB18_12
-; RV32IFD-NEXT:  .LBB18_11: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    srli a4, a1, 31
+; RV32IFD-NEXT:    add a2, a2, a4
+; RV32IFD-NEXT:  .LBB18_9: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    lui a4, 524288
+; RV32IFD-NEXT:    addi a5, a4, -1
+; RV32IFD-NEXT:    beq a1, a5, .LBB18_11
+; RV32IFD-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    sltu a6, a1, a5
+; RV32IFD-NEXT:    or a7, a0, a2
+; RV32IFD-NEXT:    bnez a7, .LBB18_12
+; RV32IFD-NEXT:    j .LBB18_13
+; RV32IFD-NEXT:  .LBB18_11:
+; RV32IFD-NEXT:    sltiu a6, a3, -1
+; RV32IFD-NEXT:    or a7, a0, a2
+; RV32IFD-NEXT:    beqz a7, .LBB18_13
 ; RV32IFD-NEXT:  .LBB18_12: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    neg a5, a4
-; RV32IFD-NEXT:    addi a6, a4, -1
-; RV32IFD-NEXT:    bnez a4, .LBB18_14
-; RV32IFD-NEXT:  # %bb.13: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    mv s2, a3
-; RV32IFD-NEXT:  .LBB18_14: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    or a3, a6, s1
-; RV32IFD-NEXT:    and a1, a5, a1
-; RV32IFD-NEXT:    and a4, a5, a0
-; RV32IFD-NEXT:    beq s2, a2, .LBB18_16
-; RV32IFD-NEXT:  # %bb.15: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    sltu a0, a2, s2
-; RV32IFD-NEXT:    j .LBB18_17
-; RV32IFD-NEXT:  .LBB18_16:
+; RV32IFD-NEXT:    srli a6, a2, 31
+; RV32IFD-NEXT:  .LBB18_13: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    neg a7, a6
+; RV32IFD-NEXT:    addi t0, a6, -1
+; RV32IFD-NEXT:    bnez a6, .LBB18_15
+; RV32IFD-NEXT:  # %bb.14: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    mv a1, a5
+; RV32IFD-NEXT:  .LBB18_15: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    or a3, t0, a3
+; RV32IFD-NEXT:    and a2, a7, a2
+; RV32IFD-NEXT:    and a5, a7, a0
+; RV32IFD-NEXT:    beq a1, a4, .LBB18_17
+; RV32IFD-NEXT:  # %bb.16: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    sltu a0, a4, a1
+; RV32IFD-NEXT:    j .LBB18_18
+; RV32IFD-NEXT:  .LBB18_17:
 ; RV32IFD-NEXT:    snez a0, a3
-; RV32IFD-NEXT:  .LBB18_17: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    and a4, a4, a1
-; RV32IFD-NEXT:    li a2, -1
-; RV32IFD-NEXT:    beq a4, a2, .LBB18_19
-; RV32IFD-NEXT:  # %bb.18: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    srli a1, a1, 31
-; RV32IFD-NEXT:    xori a0, a1, 1
-; RV32IFD-NEXT:  .LBB18_19: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    bnez a0, .LBB18_21
-; RV32IFD-NEXT:  # %bb.20: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    lui s2, 524288
-; RV32IFD-NEXT:  .LBB18_21: # %fp-to-i-cleanup
+; RV32IFD-NEXT:  .LBB18_18: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    and a5, a5, a2
+; RV32IFD-NEXT:    li a4, -1
+; RV32IFD-NEXT:    beq a5, a4, .LBB18_20
+; RV32IFD-NEXT:  # %bb.19: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    srli a2, a2, 31
+; RV32IFD-NEXT:    xori a0, a2, 1
+; RV32IFD-NEXT:  .LBB18_20: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    bnez a0, .LBB18_22
+; RV32IFD-NEXT:  # %bb.21: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    lui a1, 524288
+; RV32IFD-NEXT:  .LBB18_22: # %fp-to-i-cleanup
 ; RV32IFD-NEXT:    neg a0, a0
 ; RV32IFD-NEXT:    and a0, a0, a3
-; RV32IFD-NEXT:    mv a1, s2
 ; RV32IFD-NEXT:    lw ra, 92(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s0, 88(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s1, 84(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s2, 80(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s3, 76(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s4, 72(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s5, 68(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s6, 64(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s7, 60(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s8, 56(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s9, 52(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    .cfi_restore ra
-; RV32IFD-NEXT:    .cfi_restore s0
-; RV32IFD-NEXT:    .cfi_restore s1
-; RV32IFD-NEXT:    .cfi_restore s2
-; RV32IFD-NEXT:    .cfi_restore s3
-; RV32IFD-NEXT:    .cfi_restore s4
-; RV32IFD-NEXT:    .cfi_restore s5
-; RV32IFD-NEXT:    .cfi_restore s6
-; RV32IFD-NEXT:    .cfi_restore s7
-; RV32IFD-NEXT:    .cfi_restore s8
-; RV32IFD-NEXT:    .cfi_restore s9
 ; RV32IFD-NEXT:    addi sp, sp, 96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IFD-NEXT:    ret
@@ -1918,261 +1943,272 @@ define i64 @ustest_f64i64(double %x) {
 ; RV32IF-NEXT:    addi sp, sp, -80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 80
 ; RV32IF-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s9, 36(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    .cfi_offset ra, -4
-; RV32IF-NEXT:    .cfi_offset s0, -8
-; RV32IF-NEXT:    .cfi_offset s1, -12
-; RV32IF-NEXT:    .cfi_offset s2, -16
-; RV32IF-NEXT:    .cfi_offset s3, -20
-; RV32IF-NEXT:    .cfi_offset s4, -24
-; RV32IF-NEXT:    .cfi_offset s5, -28
-; RV32IF-NEXT:    .cfi_offset s6, -32
-; RV32IF-NEXT:    .cfi_offset s7, -36
-; RV32IF-NEXT:    .cfi_offset s8, -40
-; RV32IF-NEXT:    .cfi_offset s9, -44
 ; RV32IF-NEXT:    li a3, 1023
 ; RV32IF-NEXT:    slli a2, a1, 1
 ; RV32IF-NEXT:    srli a2, a2, 21
-; RV32IF-NEXT:    bgeu a2, a3, .LBB20_2
+; RV32IF-NEXT:    bgeu a2, a3, .LBB20_3
 ; RV32IF-NEXT:  # %bb.1:
-; RV32IF-NEXT:    li s1, 0
-; RV32IF-NEXT:    li s6, 0
+; RV32IF-NEXT:    li a2, 0
 ; RV32IF-NEXT:    li a0, 0
+; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    j .LBB20_6
-; RV32IF-NEXT:  .LBB20_2: # %fp-to-i-if-check.exp.size
-; RV32IF-NEXT:    srai s0, a1, 31
-; RV32IF-NEXT:    lui a3, 256
-; RV32IF-NEXT:    snez a4, a0
-; RV32IF-NEXT:    addi a3, a3, -1
+; RV32IF-NEXT:  .LBB20_2:
+; RV32IF-NEXT:    seqz a4, a3
+; RV32IF-NEXT:    j .LBB20_11
+; RV32IF-NEXT:  .LBB20_3: # %fp-to-i-if-check.exp.size
+; RV32IF-NEXT:    mv a3, a1
+; RV32IF-NEXT:    srai a1, a1, 31
+; RV32IF-NEXT:    lui a4, 256
+; RV32IF-NEXT:    snez a5, a0
+; RV32IF-NEXT:    addi a4, a4, -1
 ; RV32IF-NEXT:    addi a0, a0, -1
-; RV32IF-NEXT:    and a1, a1, a3
+; RV32IF-NEXT:    and a3, a3, a4
 ; RV32IF-NEXT:    addi a0, a0, 1
-; RV32IF-NEXT:    add a1, a1, a4
-; RV32IF-NEXT:    seqz a4, a0
-; RV32IF-NEXT:    add a3, a4, a3
-; RV32IF-NEXT:    li a4, 1074
-; RV32IF-NEXT:    ori s2, s0, 1
-; RV32IF-NEXT:    add a1, a1, a3
-; RV32IF-NEXT:    bltu a4, a2, .LBB20_5
-; RV32IF-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    li a3, 1043
-; RV32IF-NEXT:    li a4, 1075
-; RV32IF-NEXT:    sub a3, a3, a2
-; RV32IF-NEXT:    sub a2, a4, a2
-; RV32IF-NEXT:    srai a4, a3, 31
-; RV32IF-NEXT:    srl a5, a1, a2
-; RV32IF-NEXT:    and s3, a4, a5
-; RV32IF-NEXT:    bltz a3, .LBB20_7
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    seqz a5, a0
+; RV32IF-NEXT:    add a5, a5, a4
+; RV32IF-NEXT:    li a6, 1074
+; RV32IF-NEXT:    ori a4, a1, 1
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    sw a1, 36(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a4, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    bltu a6, a2, .LBB20_6
 ; RV32IF-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    srl s4, a1, a3
+; RV32IF-NEXT:    li a5, 1043
+; RV32IF-NEXT:    li a7, 1075
+; RV32IF-NEXT:    sub a6, a5, a2
+; RV32IF-NEXT:    sub a5, a7, a2
+; RV32IF-NEXT:    srai a2, a6, 31
+; RV32IF-NEXT:    srl a7, a3, a5
+; RV32IF-NEXT:    and a2, a2, a7
+; RV32IF-NEXT:    bltz a6, .LBB20_7
+; RV32IF-NEXT:  # %bb.5: # %fp-to-i-if-exp.small
+; RV32IF-NEXT:    srl a0, a3, a6
 ; RV32IF-NEXT:    j .LBB20_8
-; RV32IF-NEXT:  .LBB20_5: # %fp-to-i-if-exp.large
-; RV32IF-NEXT:    addi a4, a2, -1075
-; RV32IF-NEXT:    srli a2, a4, 3
-; RV32IF-NEXT:    sw a0, 16(sp)
-; RV32IF-NEXT:    sw a1, 20(sp)
-; RV32IF-NEXT:    sw zero, 24(sp)
-; RV32IF-NEXT:    sw zero, 28(sp)
+; RV32IF-NEXT:  .LBB20_6: # %fp-to-i-if-exp.large
+; RV32IF-NEXT:    addi a5, a2, -1075
+; RV32IF-NEXT:    srli a2, a5, 3
+; RV32IF-NEXT:    sw a0, 56(sp)
+; RV32IF-NEXT:    sw a3, 60(sp)
+; RV32IF-NEXT:    sw zero, 64(sp)
+; RV32IF-NEXT:    sw zero, 68(sp)
 ; RV32IF-NEXT:    andi a2, a2, 12
-; RV32IF-NEXT:    addi a0, sp, 16
-; RV32IF-NEXT:    sw zero, 0(sp)
-; RV32IF-NEXT:    sw zero, 4(sp)
-; RV32IF-NEXT:    sw zero, 8(sp)
-; RV32IF-NEXT:    sw zero, 12(sp)
+; RV32IF-NEXT:    addi a0, sp, 56
+; RV32IF-NEXT:    sw zero, 40(sp)
+; RV32IF-NEXT:    sw zero, 44(sp)
+; RV32IF-NEXT:    sw zero, 48(sp)
+; RV32IF-NEXT:    sw zero, 52(sp)
 ; RV32IF-NEXT:    sub a0, a0, a2
-; RV32IF-NEXT:    lw a1, 0(a0)
+; RV32IF-NEXT:    lw a6, 0(a0)
 ; RV32IF-NEXT:    lw a2, 4(a0)
-; RV32IF-NEXT:    andi a3, a4, 31
-; RV32IF-NEXT:    lw a5, 8(a0)
-; RV32IF-NEXT:    lw a0, 12(a0)
-; RV32IF-NEXT:    srli a6, a1, 1
+; RV32IF-NEXT:    andi a3, a5, 31
+; RV32IF-NEXT:    lw a7, 8(a0)
+; RV32IF-NEXT:    lw t0, 12(a0)
+; RV32IF-NEXT:    srli a0, a6, 1
 ; RV32IF-NEXT:    xori a3, a3, 31
-; RV32IF-NEXT:    sll a7, a2, a4
-; RV32IF-NEXT:    srl a6, a6, a3
-; RV32IF-NEXT:    or s3, a7, a6
+; RV32IF-NEXT:    sll t1, a2, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a0, t1, a0
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    srli a2, a2, 1
-; RV32IF-NEXT:    sll a6, a5, a4
+; RV32IF-NEXT:    sll a0, a7, a5
 ; RV32IF-NEXT:    srl a2, a2, a3
-; RV32IF-NEXT:    or a2, a6, a2
-; RV32IF-NEXT:    srli a5, a5, 1
-; RV32IF-NEXT:    sll a0, a0, a4
-; RV32IF-NEXT:    srl a3, a5, a3
-; RV32IF-NEXT:    or a3, a0, a3
-; RV32IF-NEXT:    sll s4, a1, a4
-; RV32IF-NEXT:    mv a0, s2
-; RV32IF-NEXT:    mv a1, s0
+; RV32IF-NEXT:    or a2, a0, a2
+; RV32IF-NEXT:    srli a0, a7, 1
+; RV32IF-NEXT:    sll a7, t0, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a3, a7, a0
+; RV32IF-NEXT:    sll a0, a6, a5
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    mv a0, a4
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s7, a0, s1
-; RV32IF-NEXT:    add a1, a1, s5
-; RV32IF-NEXT:    sltu a0, s7, a0
-; RV32IF-NEXT:    add s8, a1, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a1, a2
+; RV32IF-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    add a0, a2, a0
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s5, a0
-; RV32IF-NEXT:    mv s6, a1
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    add s2, s5, a1
-; RV32IF-NEXT:    sltu a0, s2, s5
-; RV32IF-NEXT:    add s5, s6, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s6, a0, s2
-; RV32IF-NEXT:    sltu a0, s6, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s2, s5, a0
-; RV32IF-NEXT:    sltu s4, s2, s5
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s2
-; RV32IF-NEXT:    add a1, a1, s4
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add a1, a1, a0
-; RV32IF-NEXT:    add a0, s2, s7
-; RV32IF-NEXT:    add a1, a1, s8
-; RV32IF-NEXT:    sltu a2, a0, s2
-; RV32IF-NEXT:    add a1, a1, a2
-; RV32IF-NEXT:    bnez a1, .LBB20_9
-; RV32IF-NEXT:  .LBB20_6:
-; RV32IF-NEXT:    seqz a2, a0
-; RV32IF-NEXT:    j .LBB20_10
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a4, a0, a3
+; RV32IF-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a1, a1, a3
+; RV32IF-NEXT:    sltu a3, a4, a0
+; RV32IF-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a1, a3
+; RV32IF-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a3, a4, a3
+; RV32IF-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a1, a5, a1
+; RV32IF-NEXT:    sltu a4, a3, a4
+; RV32IF-NEXT:    j .LBB20_9
 ; RV32IF-NEXT:  .LBB20_7:
-; RV32IF-NEXT:    not a3, a2
-; RV32IF-NEXT:    slli a1, a1, 1
-; RV32IF-NEXT:    srl a0, a0, a2
-; RV32IF-NEXT:    sll a1, a1, a3
-; RV32IF-NEXT:    or s4, a0, a1
+; RV32IF-NEXT:    not a6, a5
+; RV32IF-NEXT:    slli a3, a3, 1
+; RV32IF-NEXT:    srl a0, a0, a5
+; RV32IF-NEXT:    sll a3, a3, a6
+; RV32IF-NEXT:    or a0, a0, a3
 ; RV32IF-NEXT:  .LBB20_8: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s5, a0, s5
-; RV32IF-NEXT:    sltu a0, s5, a0
-; RV32IF-NEXT:    add s7, a1, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a0, a1, a0
+; RV32IF-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s6, a0, s5
-; RV32IF-NEXT:    sltu a0, s6, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s8, s7, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s2, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    add s9, a0, s8
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    sltu a2, s8, s7
-; RV32IF-NEXT:    sltu a3, s9, s2
-; RV32IF-NEXT:    add a2, s5, a2
-; RV32IF-NEXT:    add a0, s9, a0
-; RV32IF-NEXT:    add a2, a2, a3
-; RV32IF-NEXT:    sltu a3, a0, s9
-; RV32IF-NEXT:    add a1, a2, a1
-; RV32IF-NEXT:    add a1, a1, a3
-; RV32IF-NEXT:    beqz a1, .LBB20_6
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a5, a4, a3
+; RV32IF-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a4, a4, a3
+; RV32IF-NEXT:    lw a3, 0(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a3, a5
+; RV32IF-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a3, a3, a0
+; RV32IF-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a5, a4
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a4, a3, a4
+; RV32IF-NEXT:    add a1, a5, a1
 ; RV32IF-NEXT:  .LBB20_9: # %fp-to-i-cleanup
-; RV32IF-NEXT:    srli a2, a1, 31
-; RV32IF-NEXT:  .LBB20_10: # %fp-to-i-cleanup
-; RV32IF-NEXT:    xori a3, a0, 1
-; RV32IF-NEXT:    or a3, a3, a1
-; RV32IF-NEXT:    seqz a3, a3
-; RV32IF-NEXT:    addi a3, a3, -1
-; RV32IF-NEXT:    and a3, a3, a2
-; RV32IF-NEXT:    neg a2, a3
-; RV32IF-NEXT:    bnez a3, .LBB20_12
-; RV32IF-NEXT:  # %bb.11: # %fp-to-i-cleanup
-; RV32IF-NEXT:    li a0, 1
-; RV32IF-NEXT:  .LBB20_12: # %fp-to-i-cleanup
-; RV32IF-NEXT:    and a3, a2, a1
-; RV32IF-NEXT:    beqz a3, .LBB20_14
-; RV32IF-NEXT:  # %bb.13: # %fp-to-i-cleanup
-; RV32IF-NEXT:    sgtz a1, a3
-; RV32IF-NEXT:    j .LBB20_15
-; RV32IF-NEXT:  .LBB20_14:
-; RV32IF-NEXT:    snez a1, a0
-; RV32IF-NEXT:  .LBB20_15: # %fp-to-i-cleanup
-; RV32IF-NEXT:    or a0, a0, a3
-; RV32IF-NEXT:    and a3, a2, s6
-; RV32IF-NEXT:    and a2, a2, s1
-; RV32IF-NEXT:    bnez a0, .LBB20_17
-; RV32IF-NEXT:  # %bb.16:
+; RV32IF-NEXT:    add a1, a1, a4
+; RV32IF-NEXT:    beqz a1, .LBB20_2
+; RV32IF-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IF-NEXT:    srli a4, a1, 31
+; RV32IF-NEXT:  .LBB20_11: # %fp-to-i-cleanup
+; RV32IF-NEXT:    xori a5, a3, 1
+; RV32IF-NEXT:    or a5, a5, a1
+; RV32IF-NEXT:    seqz a5, a5
+; RV32IF-NEXT:    addi a5, a5, -1
+; RV32IF-NEXT:    and a5, a5, a4
+; RV32IF-NEXT:    neg a4, a5
+; RV32IF-NEXT:    bnez a5, .LBB20_13
+; RV32IF-NEXT:  # %bb.12: # %fp-to-i-cleanup
+; RV32IF-NEXT:    li a3, 1
+; RV32IF-NEXT:  .LBB20_13: # %fp-to-i-cleanup
+; RV32IF-NEXT:    and a5, a4, a1
+; RV32IF-NEXT:    beqz a5, .LBB20_15
+; RV32IF-NEXT:  # %bb.14: # %fp-to-i-cleanup
+; RV32IF-NEXT:    sgtz a1, a5
+; RV32IF-NEXT:    j .LBB20_16
+; RV32IF-NEXT:  .LBB20_15:
+; RV32IF-NEXT:    snez a1, a3
+; RV32IF-NEXT:  .LBB20_16: # %fp-to-i-cleanup
+; RV32IF-NEXT:    or a5, a3, a5
+; RV32IF-NEXT:    and a3, a4, a0
+; RV32IF-NEXT:    and a2, a4, a2
+; RV32IF-NEXT:    bnez a5, .LBB20_18
+; RV32IF-NEXT:  # %bb.17:
 ; RV32IF-NEXT:    or a0, a2, a3
 ; RV32IF-NEXT:    snez a1, a0
-; RV32IF-NEXT:  .LBB20_17: # %fp-to-i-cleanup
+; RV32IF-NEXT:  .LBB20_18: # %fp-to-i-cleanup
 ; RV32IF-NEXT:    neg a1, a1
 ; RV32IF-NEXT:    and a0, a1, a2
 ; RV32IF-NEXT:    and a1, a1, a3
 ; RV32IF-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s9, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    .cfi_restore ra
-; RV32IF-NEXT:    .cfi_restore s0
-; RV32IF-NEXT:    .cfi_restore s1
-; RV32IF-NEXT:    .cfi_restore s2
-; RV32IF-NEXT:    .cfi_restore s3
-; RV32IF-NEXT:    .cfi_restore s4
-; RV32IF-NEXT:    .cfi_restore s5
-; RV32IF-NEXT:    .cfi_restore s6
-; RV32IF-NEXT:    .cfi_restore s7
-; RV32IF-NEXT:    .cfi_restore s8
-; RV32IF-NEXT:    .cfi_restore s9
 ; RV32IF-NEXT:    addi sp, sp, 80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IF-NEXT:    ret
@@ -2211,264 +2247,274 @@ define i64 @ustest_f64i64(double %x) {
 ; RV32IFD-NEXT:    addi sp, sp, -96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 96
 ; RV32IFD-NEXT:    sw ra, 92(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s0, 88(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s1, 84(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s2, 80(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s3, 76(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s4, 72(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s5, 68(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s6, 64(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s7, 60(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s8, 56(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s9, 52(sp) # 4-byte Folded Spill
 ; RV32IFD-NEXT:    .cfi_offset ra, -4
-; RV32IFD-NEXT:    .cfi_offset s0, -8
-; RV32IFD-NEXT:    .cfi_offset s1, -12
-; RV32IFD-NEXT:    .cfi_offset s2, -16
-; RV32IFD-NEXT:    .cfi_offset s3, -20
-; RV32IFD-NEXT:    .cfi_offset s4, -24
-; RV32IFD-NEXT:    .cfi_offset s5, -28
-; RV32IFD-NEXT:    .cfi_offset s6, -32
-; RV32IFD-NEXT:    .cfi_offset s7, -36
-; RV32IFD-NEXT:    .cfi_offset s8, -40
-; RV32IFD-NEXT:    .cfi_offset s9, -44
-; RV32IFD-NEXT:    li a2, 1023
-; RV32IFD-NEXT:    fsd fa0, 8(sp)
-; RV32IFD-NEXT:    lw a1, 12(sp)
-; RV32IFD-NEXT:    slli a0, a1, 1
+; RV32IFD-NEXT:    li a1, 1023
+; RV32IFD-NEXT:    fsd fa0, 48(sp)
+; RV32IFD-NEXT:    lw a2, 52(sp)
+; RV32IFD-NEXT:    slli a0, a2, 1
 ; RV32IFD-NEXT:    srli a0, a0, 21
-; RV32IFD-NEXT:    bgeu a0, a2, .LBB20_2
+; RV32IFD-NEXT:    bgeu a0, a1, .LBB20_3
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    li s1, 0
-; RV32IFD-NEXT:    li s6, 0
+; RV32IFD-NEXT:    li a2, 0
 ; RV32IFD-NEXT:    li a0, 0
+; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    j .LBB20_6
-; RV32IFD-NEXT:  .LBB20_2: # %fp-to-i-if-check.exp.size
-; RV32IFD-NEXT:    lw a2, 8(sp)
-; RV32IFD-NEXT:    lui a3, 256
-; RV32IFD-NEXT:    srai s0, a1, 31
-; RV32IFD-NEXT:    addi a3, a3, -1
-; RV32IFD-NEXT:    and a4, a1, a3
-; RV32IFD-NEXT:    addi a1, a2, -1
-; RV32IFD-NEXT:    snez a2, a2
-; RV32IFD-NEXT:    addi a1, a1, 1
-; RV32IFD-NEXT:    add a2, a4, a2
-; RV32IFD-NEXT:    seqz a4, a1
-; RV32IFD-NEXT:    add a2, a2, a3
-; RV32IFD-NEXT:    li a3, 1074
-; RV32IFD-NEXT:    ori s2, s0, 1
-; RV32IFD-NEXT:    add a2, a2, a4
-; RV32IFD-NEXT:    bltu a3, a0, .LBB20_5
-; RV32IFD-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    li a3, 1043
-; RV32IFD-NEXT:    li a4, 1075
-; RV32IFD-NEXT:    sub a3, a3, a0
-; RV32IFD-NEXT:    sub a0, a4, a0
-; RV32IFD-NEXT:    srai a4, a3, 31
-; RV32IFD-NEXT:    srl a5, a2, a0
-; RV32IFD-NEXT:    and s3, a4, a5
-; RV32IFD-NEXT:    bltz a3, .LBB20_7
+; RV32IFD-NEXT:  .LBB20_2:
+; RV32IFD-NEXT:    seqz a4, a3
+; RV32IFD-NEXT:    j .LBB20_11
+; RV32IFD-NEXT:  .LBB20_3: # %fp-to-i-if-check.exp.size
+; RV32IFD-NEXT:    lw a3, 48(sp)
+; RV32IFD-NEXT:    lui a4, 256
+; RV32IFD-NEXT:    srai a1, a2, 31
+; RV32IFD-NEXT:    addi a4, a4, -1
+; RV32IFD-NEXT:    and a5, a2, a4
+; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    snez a3, a3
+; RV32IFD-NEXT:    addi a2, a2, 1
+; RV32IFD-NEXT:    add a3, a5, a3
+; RV32IFD-NEXT:    seqz a5, a2
+; RV32IFD-NEXT:    add a3, a3, a4
+; RV32IFD-NEXT:    li a6, 1074
+; RV32IFD-NEXT:    ori a4, a1, 1
+; RV32IFD-NEXT:    add a3, a3, a5
+; RV32IFD-NEXT:    sw a1, 44(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a4, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    bltu a6, a0, .LBB20_6
 ; RV32IFD-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    srl s4, a2, a3
+; RV32IFD-NEXT:    li a5, 1043
+; RV32IFD-NEXT:    li a7, 1075
+; RV32IFD-NEXT:    sub a6, a5, a0
+; RV32IFD-NEXT:    sub a5, a7, a0
+; RV32IFD-NEXT:    srai a0, a6, 31
+; RV32IFD-NEXT:    srl a7, a3, a5
+; RV32IFD-NEXT:    and a0, a0, a7
+; RV32IFD-NEXT:    bltz a6, .LBB20_7
+; RV32IFD-NEXT:  # %bb.5: # %fp-to-i-if-exp.small
+; RV32IFD-NEXT:    srl a2, a3, a6
 ; RV32IFD-NEXT:    j .LBB20_8
-; RV32IFD-NEXT:  .LBB20_5: # %fp-to-i-if-exp.large
-; RV32IFD-NEXT:    addi a0, a0, -1075
-; RV32IFD-NEXT:    srli a3, a0, 3
-; RV32IFD-NEXT:    sw a1, 32(sp)
-; RV32IFD-NEXT:    sw a2, 36(sp)
-; RV32IFD-NEXT:    sw zero, 40(sp)
-; RV32IFD-NEXT:    sw zero, 44(sp)
-; RV32IFD-NEXT:    andi a3, a3, 12
-; RV32IFD-NEXT:    addi a1, sp, 32
-; RV32IFD-NEXT:    sw zero, 16(sp)
-; RV32IFD-NEXT:    sw zero, 20(sp)
-; RV32IFD-NEXT:    sw zero, 24(sp)
-; RV32IFD-NEXT:    sw zero, 28(sp)
-; RV32IFD-NEXT:    sub a1, a1, a3
-; RV32IFD-NEXT:    lw a4, 0(a1)
-; RV32IFD-NEXT:    lw a2, 4(a1)
-; RV32IFD-NEXT:    andi a3, a0, 31
-; RV32IFD-NEXT:    lw a5, 8(a1)
-; RV32IFD-NEXT:    lw a1, 12(a1)
-; RV32IFD-NEXT:    srli a6, a4, 1
-; RV32IFD-NEXT:    xori a3, a3, 31
-; RV32IFD-NEXT:    sll a7, a2, a0
-; RV32IFD-NEXT:    srl a6, a6, a3
-; RV32IFD-NEXT:    or s3, a7, a6
-; RV32IFD-NEXT:    srli a2, a2, 1
-; RV32IFD-NEXT:    sll a6, a5, a0
-; RV32IFD-NEXT:    srl a2, a2, a3
-; RV32IFD-NEXT:    or a2, a6, a2
-; RV32IFD-NEXT:    srli a5, a5, 1
-; RV32IFD-NEXT:    sll a1, a1, a0
-; RV32IFD-NEXT:    srl a3, a5, a3
-; RV32IFD-NEXT:    or a3, a1, a3
-; RV32IFD-NEXT:    sll s4, a4, a0
-; RV32IFD-NEXT:    mv a0, s2
-; RV32IFD-NEXT:    mv a1, s0
+; RV32IFD-NEXT:  .LBB20_6: # %fp-to-i-if-exp.large
+; RV32IFD-NEXT:    addi a5, a0, -1075
+; RV32IFD-NEXT:    srli a0, a5, 3
+; RV32IFD-NEXT:    sw a2, 72(sp)
+; RV32IFD-NEXT:    sw a3, 76(sp)
+; RV32IFD-NEXT:    sw zero, 80(sp)
+; RV32IFD-NEXT:    sw zero, 84(sp)
+; RV32IFD-NEXT:    andi a0, a0, 12
+; RV32IFD-NEXT:    addi a2, sp, 72
+; RV32IFD-NEXT:    sw zero, 56(sp)
+; RV32IFD-NEXT:    sw zero, 60(sp)
+; RV32IFD-NEXT:    sw zero, 64(sp)
+; RV32IFD-NEXT:    sw zero, 68(sp)
+; RV32IFD-NEXT:    sub a2, a2, a0
+; RV32IFD-NEXT:    lw a6, 0(a2)
+; RV32IFD-NEXT:    lw a3, 4(a2)
+; RV32IFD-NEXT:    andi a0, a5, 31
+; RV32IFD-NEXT:    lw a7, 8(a2)
+; RV32IFD-NEXT:    lw t0, 12(a2)
+; RV32IFD-NEXT:    srli a2, a6, 1
+; RV32IFD-NEXT:    xori t1, a0, 31
+; RV32IFD-NEXT:    sll a0, a3, a5
+; RV32IFD-NEXT:    srl a2, a2, t1
+; RV32IFD-NEXT:    or a0, a0, a2
+; RV32IFD-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    srli a3, a3, 1
+; RV32IFD-NEXT:    sll a0, a7, a5
+; RV32IFD-NEXT:    srl a2, a3, t1
+; RV32IFD-NEXT:    or a2, a0, a2
+; RV32IFD-NEXT:    srli a0, a7, 1
+; RV32IFD-NEXT:    sll a3, t0, a5
+; RV32IFD-NEXT:    srl a0, a0, t1
+; RV32IFD-NEXT:    or a3, a3, a0
+; RV32IFD-NEXT:    sll a0, a6, a5
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    mv a0, a4
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s7, a0, s1
-; RV32IFD-NEXT:    add a1, a1, s5
-; RV32IFD-NEXT:    sltu a0, s7, a0
-; RV32IFD-NEXT:    add s8, a1, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a1, a2
+; RV32IFD-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    add a0, a2, a0
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s5, a0
-; RV32IFD-NEXT:    mv s6, a1
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    add s2, s5, a1
-; RV32IFD-NEXT:    sltu a0, s2, s5
-; RV32IFD-NEXT:    add s5, s6, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s6, a0, s2
-; RV32IFD-NEXT:    sltu a0, s6, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s2, s5, a0
-; RV32IFD-NEXT:    sltu s4, s2, s5
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s2
-; RV32IFD-NEXT:    add a1, a1, s4
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add a1, a1, a0
-; RV32IFD-NEXT:    add a0, s2, s7
-; RV32IFD-NEXT:    add a1, a1, s8
-; RV32IFD-NEXT:    sltu a2, a0, s2
-; RV32IFD-NEXT:    add a1, a1, a2
-; RV32IFD-NEXT:    bnez a1, .LBB20_9
-; RV32IFD-NEXT:  .LBB20_6:
-; RV32IFD-NEXT:    seqz a2, a0
-; RV32IFD-NEXT:    j .LBB20_10
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a4, a0, a3
+; RV32IFD-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a1, a1, a3
+; RV32IFD-NEXT:    sltu a3, a4, a0
+; RV32IFD-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a1, a3
+; RV32IFD-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a3, a4, a3
+; RV32IFD-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a1, a5, a1
+; RV32IFD-NEXT:    sltu a4, a3, a4
+; RV32IFD-NEXT:    j .LBB20_9
 ; RV32IFD-NEXT:  .LBB20_7:
-; RV32IFD-NEXT:    not a3, a0
-; RV32IFD-NEXT:    slli a2, a2, 1
-; RV32IFD-NEXT:    srl a0, a1, a0
-; RV32IFD-NEXT:    sll a1, a2, a3
-; RV32IFD-NEXT:    or s4, a0, a1
+; RV32IFD-NEXT:    not a6, a5
+; RV32IFD-NEXT:    slli a3, a3, 1
+; RV32IFD-NEXT:    srl a2, a2, a5
+; RV32IFD-NEXT:    sll a3, a3, a6
+; RV32IFD-NEXT:    or a2, a2, a3
 ; RV32IFD-NEXT:  .LBB20_8: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a2, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s5, a0, s5
-; RV32IFD-NEXT:    sltu a0, s5, a0
-; RV32IFD-NEXT:    add s7, a1, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s6, a0, s5
-; RV32IFD-NEXT:    sltu a0, s6, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s8, s7, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s2, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    add s9, a0, s8
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    sltu a2, s8, s7
-; RV32IFD-NEXT:    sltu a3, s9, s2
-; RV32IFD-NEXT:    add a2, s5, a2
-; RV32IFD-NEXT:    add a0, s9, a0
-; RV32IFD-NEXT:    add a2, a2, a3
-; RV32IFD-NEXT:    sltu a3, a0, s9
-; RV32IFD-NEXT:    add a1, a2, a1
-; RV32IFD-NEXT:    add a1, a1, a3
-; RV32IFD-NEXT:    beqz a1, .LBB20_6
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a5, a4, a3
+; RV32IFD-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a4, a4, a3
+; RV32IFD-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a3, a5
+; RV32IFD-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a3, a3, a0
+; RV32IFD-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a5, a4
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a4, a3, a4
+; RV32IFD-NEXT:    add a1, a5, a1
 ; RV32IFD-NEXT:  .LBB20_9: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    srli a2, a1, 31
-; RV32IFD-NEXT:  .LBB20_10: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    xori a3, a0, 1
-; RV32IFD-NEXT:    or a3, a3, a1
-; RV32IFD-NEXT:    seqz a3, a3
-; RV32IFD-NEXT:    addi a3, a3, -1
-; RV32IFD-NEXT:    and a3, a3, a2
-; RV32IFD-NEXT:    neg a2, a3
-; RV32IFD-NEXT:    bnez a3, .LBB20_12
-; RV32IFD-NEXT:  # %bb.11: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    li a0, 1
-; RV32IFD-NEXT:  .LBB20_12: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    and a3, a2, a1
-; RV32IFD-NEXT:    beqz a3, .LBB20_14
-; RV32IFD-NEXT:  # %bb.13: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    sgtz a1, a3
-; RV32IFD-NEXT:    j .LBB20_15
-; RV32IFD-NEXT:  .LBB20_14:
-; RV32IFD-NEXT:    snez a1, a0
-; RV32IFD-NEXT:  .LBB20_15: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    or a0, a0, a3
-; RV32IFD-NEXT:    and a3, a2, s6
-; RV32IFD-NEXT:    and a2, a2, s1
-; RV32IFD-NEXT:    bnez a0, .LBB20_17
-; RV32IFD-NEXT:  # %bb.16:
+; RV32IFD-NEXT:    add a1, a1, a4
+; RV32IFD-NEXT:    beqz a1, .LBB20_2
+; RV32IFD-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    srli a4, a1, 31
+; RV32IFD-NEXT:  .LBB20_11: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    xori a5, a3, 1
+; RV32IFD-NEXT:    or a5, a5, a1
+; RV32IFD-NEXT:    seqz a5, a5
+; RV32IFD-NEXT:    addi a5, a5, -1
+; RV32IFD-NEXT:    and a5, a5, a4
+; RV32IFD-NEXT:    neg a4, a5
+; RV32IFD-NEXT:    bnez a5, .LBB20_13
+; RV32IFD-NEXT:  # %bb.12: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    li a3, 1
+; RV32IFD-NEXT:  .LBB20_13: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    and a5, a4, a1
+; RV32IFD-NEXT:    beqz a5, .LBB20_15
+; RV32IFD-NEXT:  # %bb.14: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    sgtz a1, a5
+; RV32IFD-NEXT:    j .LBB20_16
+; RV32IFD-NEXT:  .LBB20_15:
+; RV32IFD-NEXT:    snez a1, a3
+; RV32IFD-NEXT:  .LBB20_16: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    or a5, a3, a5
+; RV32IFD-NEXT:    and a3, a4, a0
+; RV32IFD-NEXT:    and a2, a4, a2
+; RV32IFD-NEXT:    bnez a5, .LBB20_18
+; RV32IFD-NEXT:  # %bb.17:
 ; RV32IFD-NEXT:    or a0, a2, a3
 ; RV32IFD-NEXT:    snez a1, a0
-; RV32IFD-NEXT:  .LBB20_17: # %fp-to-i-cleanup
+; RV32IFD-NEXT:  .LBB20_18: # %fp-to-i-cleanup
 ; RV32IFD-NEXT:    neg a1, a1
 ; RV32IFD-NEXT:    and a0, a1, a2
 ; RV32IFD-NEXT:    and a1, a1, a3
 ; RV32IFD-NEXT:    lw ra, 92(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s0, 88(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s1, 84(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s2, 80(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s3, 76(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s4, 72(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s5, 68(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s6, 64(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s7, 60(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s8, 56(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s9, 52(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    .cfi_restore ra
-; RV32IFD-NEXT:    .cfi_restore s0
-; RV32IFD-NEXT:    .cfi_restore s1
-; RV32IFD-NEXT:    .cfi_restore s2
-; RV32IFD-NEXT:    .cfi_restore s3
-; RV32IFD-NEXT:    .cfi_restore s4
-; RV32IFD-NEXT:    .cfi_restore s5
-; RV32IFD-NEXT:    .cfi_restore s6
-; RV32IFD-NEXT:    .cfi_restore s7
-; RV32IFD-NEXT:    .cfi_restore s8
-; RV32IFD-NEXT:    .cfi_restore s9
 ; RV32IFD-NEXT:    addi sp, sp, 96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IFD-NEXT:    ret
@@ -2488,247 +2534,263 @@ define i64 @stest_f32i64(float %x) {
 ; RV32-NEXT:    addi sp, sp, -80
 ; RV32-NEXT:    .cfi_def_cfa_offset 80
 ; RV32-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
-; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    .cfi_offset s1, -12
-; RV32-NEXT:    .cfi_offset s2, -16
-; RV32-NEXT:    .cfi_offset s3, -20
-; RV32-NEXT:    .cfi_offset s4, -24
-; RV32-NEXT:    .cfi_offset s5, -28
-; RV32-NEXT:    .cfi_offset s6, -32
-; RV32-NEXT:    .cfi_offset s7, -36
-; RV32-NEXT:    .cfi_offset s8, -40
 ; RV32-NEXT:    li a1, 127
-; RV32-NEXT:    fmv.x.w s0, fa0
-; RV32-NEXT:    slli a0, s0, 1
+; RV32-NEXT:    fmv.x.w a4, fa0
+; RV32-NEXT:    slli a0, a4, 1
 ; RV32-NEXT:    srli a0, a0, 24
 ; RV32-NEXT:    bgeu a0, a1, .LBB21_2
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    li s1, 0
-; RV32-NEXT:    li s2, 0
-; RV32-NEXT:    li a0, 0
+; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    j .LBB21_5
+; RV32-NEXT:    li a0, 0
+; RV32-NEXT:    li a2, 0
+; RV32-NEXT:    j .LBB21_6
 ; RV32-NEXT:  .LBB21_2: # %fp-to-i-if-check.exp.size
-; RV32-NEXT:    slli a1, s0, 9
-; RV32-NEXT:    srai s0, s0, 31
+; RV32-NEXT:    slli a1, a4, 9
+; RV32-NEXT:    srai a4, a4, 31
 ; RV32-NEXT:    srli a1, a1, 9
 ; RV32-NEXT:    lui a2, 2048
 ; RV32-NEXT:    li a3, 149
 ; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    ori s2, s0, 1
+; RV32-NEXT:    ori a5, a4, 1
+; RV32-NEXT:    sw a4, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a5, 32(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    bltu a3, a0, .LBB21_4
 ; RV32-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
 ; RV32-NEXT:    li a2, 150
 ; RV32-NEXT:    sub a2, a2, a0
-; RV32-NEXT:    srl s3, a1, a2
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    srl a0, a1, a2
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s4, a1
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s4
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add s6, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s7, s6, a0
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s4, a0
-; RV32-NEXT:    mv s5, a1
-; RV32-NEXT:    add s8, a0, s7
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s3
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sltu a2, s7, s6
-; RV32-NEXT:    sltu a3, s8, s4
-; RV32-NEXT:    add a2, s5, a2
-; RV32-NEXT:    add a0, s8, a0
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    sltu a3, a0, s8
-; RV32-NEXT:    add a1, a2, a1
-; RV32-NEXT:    add a1, a1, a3
+; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv a2, a1
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a6, a5, a4
+; RV32-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a5, a5, a4
+; RV32-NEXT:    lw a4, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a6, a4, a6
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a4, a0
+; RV32-NEXT:    add a5, a6, a5
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, a0, a4
+; RV32-NEXT:    add a2, a5, a2
 ; RV32-NEXT:    j .LBB21_5
 ; RV32-NEXT:  .LBB21_4: # %fp-to-i-if-exp.large
-; RV32-NEXT:    addi a0, a0, -150
-; RV32-NEXT:    srli a2, a0, 3
-; RV32-NEXT:    sw a1, 24(sp)
-; RV32-NEXT:    sw zero, 28(sp)
-; RV32-NEXT:    sw zero, 32(sp)
-; RV32-NEXT:    sw zero, 36(sp)
-; RV32-NEXT:    andi a2, a2, 12
-; RV32-NEXT:    addi a1, sp, 24
-; RV32-NEXT:    sw zero, 8(sp)
-; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    sw zero, 16(sp)
-; RV32-NEXT:    sw zero, 20(sp)
-; RV32-NEXT:    sub a1, a1, a2
-; RV32-NEXT:    lw a4, 0(a1)
+; RV32-NEXT:    addi a6, a0, -150
+; RV32-NEXT:    srli a0, a6, 3
+; RV32-NEXT:    sw a1, 56(sp)
+; RV32-NEXT:    sw zero, 60(sp)
+; RV32-NEXT:    sw zero, 64(sp)
+; RV32-NEXT:    sw zero, 68(sp)
+; RV32-NEXT:    andi a0, a0, 12
+; RV32-NEXT:    addi a1, sp, 56
+; RV32-NEXT:    sw zero, 40(sp)
+; RV32-NEXT:    sw zero, 44(sp)
+; RV32-NEXT:    sw zero, 48(sp)
+; RV32-NEXT:    sw zero, 52(sp)
+; RV32-NEXT:    sub a1, a1, a0
+; RV32-NEXT:    lw a7, 0(a1)
 ; RV32-NEXT:    lw a2, 4(a1)
-; RV32-NEXT:    andi a3, a0, 31
-; RV32-NEXT:    lw a5, 8(a1)
+; RV32-NEXT:    andi a0, a6, 31
+; RV32-NEXT:    lw a3, 8(a1)
 ; RV32-NEXT:    lw a1, 12(a1)
-; RV32-NEXT:    srli a6, a4, 1
-; RV32-NEXT:    xori a3, a3, 31
-; RV32-NEXT:    sll a7, a2, a0
-; RV32-NEXT:    srl a6, a6, a3
-; RV32-NEXT:    or s3, a7, a6
+; RV32-NEXT:    srli t0, a7, 1
+; RV32-NEXT:    xori t1, a0, 31
+; RV32-NEXT:    sll a0, a2, a6
+; RV32-NEXT:    srl t0, t0, t1
+; RV32-NEXT:    or a0, a0, t0
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    srli a2, a2, 1
-; RV32-NEXT:    sll a6, a5, a0
-; RV32-NEXT:    srl a2, a2, a3
-; RV32-NEXT:    or a2, a6, a2
-; RV32-NEXT:    srli a5, a5, 1
-; RV32-NEXT:    sll a1, a1, a0
-; RV32-NEXT:    srl a3, a5, a3
-; RV32-NEXT:    or a3, a1, a3
-; RV32-NEXT:    sll s4, a4, a0
-; RV32-NEXT:    mv a0, s2
-; RV32-NEXT:    mv a1, s0
+; RV32-NEXT:    sll a0, a3, a6
+; RV32-NEXT:    srl a2, a2, t1
+; RV32-NEXT:    or a2, a0, a2
+; RV32-NEXT:    srli a3, a3, 1
+; RV32-NEXT:    sll a0, a1, a6
+; RV32-NEXT:    srl a3, a3, t1
+; RV32-NEXT:    or a3, a0, a3
+; RV32-NEXT:    sll a0, a7, a6
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, a5
+; RV32-NEXT:    mv a1, a4
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s5, a1
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s4
-; RV32-NEXT:    mv a3, s3
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s7, a0, s1
-; RV32-NEXT:    add a1, a1, s5
-; RV32-NEXT:    sltu a0, s7, a0
-; RV32-NEXT:    add s8, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a1, a2
+; RV32-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    add a0, a2, a0
+; RV32-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s5, a0
-; RV32-NEXT:    mv s6, a1
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    add s2, s5, a1
-; RV32-NEXT:    sltu a0, s2, s5
-; RV32-NEXT:    add s6, s6, a0
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a1, a0
+; RV32-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s4, s6, a0
-; RV32-NEXT:    sltu s5, s4, s6
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    add a1, a1, s5
-; RV32-NEXT:    sltu a0, s4, a0
-; RV32-NEXT:    add a1, a1, a0
-; RV32-NEXT:    add a0, s4, s7
-; RV32-NEXT:    add a1, a1, s8
-; RV32-NEXT:    sltu a2, a0, s4
-; RV32-NEXT:    add a1, a1, a2
+; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv a2, a1
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a0, a4
+; RV32-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a2, a4
+; RV32-NEXT:    sltu a0, a5, a0
+; RV32-NEXT:    add a4, a2, a0
+; RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a5, a0
+; RV32-NEXT:    lw a2, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a4, a2
+; RV32-NEXT:    sltu a4, a0, a5
 ; RV32-NEXT:  .LBB21_5: # %fp-to-i-cleanup
-; RV32-NEXT:    lui a2, 524288
-; RV32-NEXT:    addi a3, a2, -1
-; RV32-NEXT:    beq s2, a3, .LBB21_7
-; RV32-NEXT:  # %bb.6: # %fp-to-i-cleanup
-; RV32-NEXT:    sltu a4, s2, a3
-; RV32-NEXT:    or a5, a0, a1
-; RV32-NEXT:    bnez a5, .LBB21_8
-; RV32-NEXT:    j .LBB21_9
-; RV32-NEXT:  .LBB21_7:
-; RV32-NEXT:    sltiu a4, s1, -1
-; RV32-NEXT:    or a5, a0, a1
-; RV32-NEXT:    beqz a5, .LBB21_9
-; RV32-NEXT:  .LBB21_8: # %fp-to-i-cleanup
-; RV32-NEXT:    srli a4, a1, 31
+; RV32-NEXT:    add a2, a2, a4
+; RV32-NEXT:  .LBB21_6: # %fp-to-i-cleanup
+; RV32-NEXT:    lui a4, 524288
+; RV32-NEXT:    addi a5, a4, -1
+; RV32-NEXT:    beq a1, a5, .LBB21_8
+; RV32-NEXT:  # %bb.7: # %fp-to-i-cleanup
+; RV32-NEXT:    sltu a6, a1, a5
+; RV32-NEXT:    or a7, a0, a2
+; RV32-NEXT:    bnez a7, .LBB21_9
+; RV32-NEXT:    j .LBB21_10
+; RV32-NEXT:  .LBB21_8:
+; RV32-NEXT:    sltiu a6, a3, -1
+; RV32-NEXT:    or a7, a0, a2
+; RV32-NEXT:    beqz a7, .LBB21_10
 ; RV32-NEXT:  .LBB21_9: # %fp-to-i-cleanup
-; RV32-NEXT:    neg a5, a4
-; RV32-NEXT:    addi a6, a4, -1
-; RV32-NEXT:    bnez a4, .LBB21_11
-; RV32-NEXT:  # %bb.10: # %fp-to-i-cleanup
-; RV32-NEXT:    mv s2, a3
-; RV32-NEXT:  .LBB21_11: # %fp-to-i-cleanup
-; RV32-NEXT:    or a3, a6, s1
-; RV32-NEXT:    and a1, a5, a1
-; RV32-NEXT:    and a4, a5, a0
-; RV32-NEXT:    beq s2, a2, .LBB21_13
-; RV32-NEXT:  # %bb.12: # %fp-to-i-cleanup
-; RV32-NEXT:    sltu a0, a2, s2
-; RV32-NEXT:    j .LBB21_14
-; RV32-NEXT:  .LBB21_13:
+; RV32-NEXT:    srli a6, a2, 31
+; RV32-NEXT:  .LBB21_10: # %fp-to-i-cleanup
+; RV32-NEXT:    neg a7, a6
+; RV32-NEXT:    addi t0, a6, -1
+; RV32-NEXT:    bnez a6, .LBB21_12
+; RV32-NEXT:  # %bb.11: # %fp-to-i-cleanup
+; RV32-NEXT:    mv a1, a5
+; RV32-NEXT:  .LBB21_12: # %fp-to-i-cleanup
+; RV32-NEXT:    or a3, t0, a3
+; RV32-NEXT:    and a2, a7, a2
+; RV32-NEXT:    and a5, a7, a0
+; RV32-NEXT:    beq a1, a4, .LBB21_14
+; RV32-NEXT:  # %bb.13: # %fp-to-i-cleanup
+; RV32-NEXT:    sltu a0, a4, a1
+; RV32-NEXT:    j .LBB21_15
+; RV32-NEXT:  .LBB21_14:
 ; RV32-NEXT:    snez a0, a3
-; RV32-NEXT:  .LBB21_14: # %fp-to-i-cleanup
-; RV32-NEXT:    and a4, a4, a1
-; RV32-NEXT:    li a2, -1
-; RV32-NEXT:    beq a4, a2, .LBB21_16
-; RV32-NEXT:  # %bb.15: # %fp-to-i-cleanup
-; RV32-NEXT:    srli a1, a1, 31
-; RV32-NEXT:    xori a0, a1, 1
-; RV32-NEXT:  .LBB21_16: # %fp-to-i-cleanup
-; RV32-NEXT:    bnez a0, .LBB21_18
-; RV32-NEXT:  # %bb.17: # %fp-to-i-cleanup
-; RV32-NEXT:    lui s2, 524288
-; RV32-NEXT:  .LBB21_18: # %fp-to-i-cleanup
+; RV32-NEXT:  .LBB21_15: # %fp-to-i-cleanup
+; RV32-NEXT:    and a5, a5, a2
+; RV32-NEXT:    li a4, -1
+; RV32-NEXT:    beq a5, a4, .LBB21_17
+; RV32-NEXT:  # %bb.16: # %fp-to-i-cleanup
+; RV32-NEXT:    srli a2, a2, 31
+; RV32-NEXT:    xori a0, a2, 1
+; RV32-NEXT:  .LBB21_17: # %fp-to-i-cleanup
+; RV32-NEXT:    bnez a0, .LBB21_19
+; RV32-NEXT:  # %bb.18: # %fp-to-i-cleanup
+; RV32-NEXT:    lui a1, 524288
+; RV32-NEXT:  .LBB21_19: # %fp-to-i-cleanup
 ; RV32-NEXT:    neg a0, a0
 ; RV32-NEXT:    and a0, a0, a3
-; RV32-NEXT:    mv a1, s2
 ; RV32-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
-; RV32-NEXT:    .cfi_restore s0
-; RV32-NEXT:    .cfi_restore s1
-; RV32-NEXT:    .cfi_restore s2
-; RV32-NEXT:    .cfi_restore s3
-; RV32-NEXT:    .cfi_restore s4
-; RV32-NEXT:    .cfi_restore s5
-; RV32-NEXT:    .cfi_restore s6
-; RV32-NEXT:    .cfi_restore s7
-; RV32-NEXT:    .cfi_restore s8
 ; RV32-NEXT:    addi sp, sp, 80
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
 ; RV32-NEXT:    ret
@@ -2859,237 +2921,251 @@ define i64 @ustest_f32i64(float %x) {
 ; RV32-NEXT:    addi sp, sp, -80
 ; RV32-NEXT:    .cfi_def_cfa_offset 80
 ; RV32-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
-; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    .cfi_offset s1, -12
-; RV32-NEXT:    .cfi_offset s2, -16
-; RV32-NEXT:    .cfi_offset s3, -20
-; RV32-NEXT:    .cfi_offset s4, -24
-; RV32-NEXT:    .cfi_offset s5, -28
-; RV32-NEXT:    .cfi_offset s6, -32
-; RV32-NEXT:    .cfi_offset s7, -36
-; RV32-NEXT:    .cfi_offset s8, -40
 ; RV32-NEXT:    li a1, 127
-; RV32-NEXT:    fmv.x.w s0, fa0
-; RV32-NEXT:    slli a0, s0, 1
+; RV32-NEXT:    fmv.x.w a4, fa0
+; RV32-NEXT:    slli a0, a4, 1
 ; RV32-NEXT:    srli a0, a0, 24
-; RV32-NEXT:    bgeu a0, a1, .LBB23_2
+; RV32-NEXT:    bgeu a0, a1, .LBB23_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    li s1, 0
-; RV32-NEXT:    li s5, 0
+; RV32-NEXT:    li a2, 0
 ; RV32-NEXT:    li a0, 0
+; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    j .LBB23_4
-; RV32-NEXT:  .LBB23_2: # %fp-to-i-if-check.exp.size
-; RV32-NEXT:    slli a1, s0, 9
-; RV32-NEXT:    srai s0, s0, 31
+; RV32-NEXT:  .LBB23_2:
+; RV32-NEXT:    seqz a4, a3
+; RV32-NEXT:    j .LBB23_8
+; RV32-NEXT:  .LBB23_3: # %fp-to-i-if-check.exp.size
+; RV32-NEXT:    slli a1, a4, 9
+; RV32-NEXT:    srai a4, a4, 31
 ; RV32-NEXT:    srli a1, a1, 9
 ; RV32-NEXT:    lui a2, 2048
 ; RV32-NEXT:    li a3, 149
 ; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    ori s2, s0, 1
+; RV32-NEXT:    ori a5, a4, 1
+; RV32-NEXT:    sw a4, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a5, 32(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    bltu a3, a0, .LBB23_5
-; RV32-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
+; RV32-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
 ; RV32-NEXT:    li a2, 150
 ; RV32-NEXT:    sub a2, a2, a0
-; RV32-NEXT:    srl s3, a1, a2
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    srl a0, a1, a2
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s4, a1
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    sltu a0, s4, a0
-; RV32-NEXT:    add s6, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s5, a0, s4
-; RV32-NEXT:    sltu a0, s5, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s7, s6, a0
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s2, a0
-; RV32-NEXT:    mv s4, a1
-; RV32-NEXT:    add s8, a0, s7
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s3
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sltu a2, s7, s6
-; RV32-NEXT:    sltu a3, s8, s2
-; RV32-NEXT:    add a2, s4, a2
-; RV32-NEXT:    add a0, s8, a0
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    sltu a3, a0, s8
-; RV32-NEXT:    add a1, a2, a1
-; RV32-NEXT:    add a1, a1, a3
-; RV32-NEXT:    bnez a1, .LBB23_6
-; RV32-NEXT:  .LBB23_4:
-; RV32-NEXT:    seqz a2, a0
-; RV32-NEXT:    j .LBB23_7
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a5, a4, a3
+; RV32-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, a4, a3
+; RV32-NEXT:    lw a3, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a3, a5
+; RV32-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a3, a0
+; RV32-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a5, a4
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, a3, a4
+; RV32-NEXT:    add a1, a5, a1
+; RV32-NEXT:    j .LBB23_6
 ; RV32-NEXT:  .LBB23_5: # %fp-to-i-if-exp.large
-; RV32-NEXT:    addi a0, a0, -150
-; RV32-NEXT:    srli a2, a0, 3
-; RV32-NEXT:    sw a1, 24(sp)
-; RV32-NEXT:    sw zero, 28(sp)
-; RV32-NEXT:    sw zero, 32(sp)
-; RV32-NEXT:    sw zero, 36(sp)
-; RV32-NEXT:    andi a2, a2, 12
-; RV32-NEXT:    addi a1, sp, 24
-; RV32-NEXT:    sw zero, 8(sp)
-; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    sw zero, 16(sp)
-; RV32-NEXT:    sw zero, 20(sp)
-; RV32-NEXT:    sub a1, a1, a2
-; RV32-NEXT:    lw a4, 0(a1)
+; RV32-NEXT:    addi a6, a0, -150
+; RV32-NEXT:    srli a0, a6, 3
+; RV32-NEXT:    sw a1, 56(sp)
+; RV32-NEXT:    sw zero, 60(sp)
+; RV32-NEXT:    sw zero, 64(sp)
+; RV32-NEXT:    sw zero, 68(sp)
+; RV32-NEXT:    andi a0, a0, 12
+; RV32-NEXT:    addi a1, sp, 56
+; RV32-NEXT:    sw zero, 40(sp)
+; RV32-NEXT:    sw zero, 44(sp)
+; RV32-NEXT:    sw zero, 48(sp)
+; RV32-NEXT:    sw zero, 52(sp)
+; RV32-NEXT:    sub a1, a1, a0
+; RV32-NEXT:    lw a7, 0(a1)
 ; RV32-NEXT:    lw a2, 4(a1)
-; RV32-NEXT:    andi a3, a0, 31
-; RV32-NEXT:    lw a5, 8(a1)
+; RV32-NEXT:    andi a0, a6, 31
+; RV32-NEXT:    lw a3, 8(a1)
 ; RV32-NEXT:    lw a1, 12(a1)
-; RV32-NEXT:    srli a6, a4, 1
-; RV32-NEXT:    xori a3, a3, 31
-; RV32-NEXT:    sll a7, a2, a0
-; RV32-NEXT:    srl a6, a6, a3
-; RV32-NEXT:    or s3, a7, a6
+; RV32-NEXT:    srli t0, a7, 1
+; RV32-NEXT:    xori t1, a0, 31
+; RV32-NEXT:    sll a0, a2, a6
+; RV32-NEXT:    srl t0, t0, t1
+; RV32-NEXT:    or a0, a0, t0
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    srli a2, a2, 1
-; RV32-NEXT:    sll a6, a5, a0
-; RV32-NEXT:    srl a2, a2, a3
-; RV32-NEXT:    or a2, a6, a2
-; RV32-NEXT:    srli a5, a5, 1
-; RV32-NEXT:    sll a1, a1, a0
-; RV32-NEXT:    srl a3, a5, a3
-; RV32-NEXT:    or a3, a1, a3
-; RV32-NEXT:    sll s4, a4, a0
-; RV32-NEXT:    mv a0, s2
-; RV32-NEXT:    mv a1, s0
+; RV32-NEXT:    sll a0, a3, a6
+; RV32-NEXT:    srl a2, a2, t1
+; RV32-NEXT:    or a2, a0, a2
+; RV32-NEXT:    srli a3, a3, 1
+; RV32-NEXT:    sll a0, a1, a6
+; RV32-NEXT:    srl a3, a3, t1
+; RV32-NEXT:    or a3, a0, a3
+; RV32-NEXT:    sll a0, a7, a6
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, a5
+; RV32-NEXT:    mv a1, a4
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s5, a1
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s4
-; RV32-NEXT:    mv a3, s3
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s7, a0, s1
-; RV32-NEXT:    add a1, a1, s5
-; RV32-NEXT:    sltu a0, s7, a0
-; RV32-NEXT:    add s8, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a1, a2
+; RV32-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    add a0, a2, a0
+; RV32-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s5, a0
-; RV32-NEXT:    mv s6, a1
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    add s2, s5, a1
-; RV32-NEXT:    sltu a0, s2, s5
-; RV32-NEXT:    add s6, s6, a0
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a1, a0
+; RV32-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s5, a0, s2
-; RV32-NEXT:    sltu a0, s5, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s2, s6, a0
-; RV32-NEXT:    sltu s4, s2, s6
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    add a1, a1, s4
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add a1, a1, a0
-; RV32-NEXT:    add a0, s2, s7
-; RV32-NEXT:    add a1, a1, s8
-; RV32-NEXT:    sltu a2, a0, s2
-; RV32-NEXT:    add a1, a1, a2
-; RV32-NEXT:    beqz a1, .LBB23_4
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a4, a0, a3
+; RV32-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a1, a3
+; RV32-NEXT:    sltu a3, a4, a0
+; RV32-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a1, a3
+; RV32-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a4, a3
+; RV32-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a5, a1
+; RV32-NEXT:    sltu a4, a3, a4
 ; RV32-NEXT:  .LBB23_6: # %fp-to-i-cleanup
-; RV32-NEXT:    srli a2, a1, 31
-; RV32-NEXT:  .LBB23_7: # %fp-to-i-cleanup
-; RV32-NEXT:    xori a3, a0, 1
-; RV32-NEXT:    or a3, a3, a1
-; RV32-NEXT:    seqz a3, a3
-; RV32-NEXT:    addi a3, a3, -1
-; RV32-NEXT:    and a3, a3, a2
-; RV32-NEXT:    neg a2, a3
-; RV32-NEXT:    bnez a3, .LBB23_9
-; RV32-NEXT:  # %bb.8: # %fp-to-i-cleanup
-; RV32-NEXT:    li a0, 1
-; RV32-NEXT:  .LBB23_9: # %fp-to-i-cleanup
-; RV32-NEXT:    and a3, a2, a1
-; RV32-NEXT:    beqz a3, .LBB23_11
-; RV32-NEXT:  # %bb.10: # %fp-to-i-cleanup
-; RV32-NEXT:    sgtz a1, a3
-; RV32-NEXT:    j .LBB23_12
-; RV32-NEXT:  .LBB23_11:
-; RV32-NEXT:    snez a1, a0
-; RV32-NEXT:  .LBB23_12: # %fp-to-i-cleanup
-; RV32-NEXT:    or a0, a0, a3
-; RV32-NEXT:    and a3, a2, s5
-; RV32-NEXT:    and a2, a2, s1
-; RV32-NEXT:    bnez a0, .LBB23_14
-; RV32-NEXT:  # %bb.13:
+; RV32-NEXT:    add a1, a1, a4
+; RV32-NEXT:    beqz a1, .LBB23_2
+; RV32-NEXT:  # %bb.7: # %fp-to-i-cleanup
+; RV32-NEXT:    srli a4, a1, 31
+; RV32-NEXT:  .LBB23_8: # %fp-to-i-cleanup
+; RV32-NEXT:    xori a5, a3, 1
+; RV32-NEXT:    or a5, a5, a1
+; RV32-NEXT:    seqz a5, a5
+; RV32-NEXT:    addi a5, a5, -1
+; RV32-NEXT:    and a5, a5, a4
+; RV32-NEXT:    neg a4, a5
+; RV32-NEXT:    bnez a5, .LBB23_10
+; RV32-NEXT:  # %bb.9: # %fp-to-i-cleanup
+; RV32-NEXT:    li a3, 1
+; RV32-NEXT:  .LBB23_10: # %fp-to-i-cleanup
+; RV32-NEXT:    and a5, a4, a1
+; RV32-NEXT:    beqz a5, .LBB23_12
+; RV32-NEXT:  # %bb.11: # %fp-to-i-cleanup
+; RV32-NEXT:    sgtz a1, a5
+; RV32-NEXT:    j .LBB23_13
+; RV32-NEXT:  .LBB23_12:
+; RV32-NEXT:    snez a1, a3
+; RV32-NEXT:  .LBB23_13: # %fp-to-i-cleanup
+; RV32-NEXT:    or a5, a3, a5
+; RV32-NEXT:    and a3, a4, a0
+; RV32-NEXT:    and a2, a4, a2
+; RV32-NEXT:    bnez a5, .LBB23_15
+; RV32-NEXT:  # %bb.14:
 ; RV32-NEXT:    or a0, a2, a3
 ; RV32-NEXT:    snez a1, a0
-; RV32-NEXT:  .LBB23_14: # %fp-to-i-cleanup
+; RV32-NEXT:  .LBB23_15: # %fp-to-i-cleanup
 ; RV32-NEXT:    neg a1, a1
 ; RV32-NEXT:    and a0, a1, a2
 ; RV32-NEXT:    and a1, a1, a3
 ; RV32-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
-; RV32-NEXT:    .cfi_restore s0
-; RV32-NEXT:    .cfi_restore s1
-; RV32-NEXT:    .cfi_restore s2
-; RV32-NEXT:    .cfi_restore s3
-; RV32-NEXT:    .cfi_restore s4
-; RV32-NEXT:    .cfi_restore s5
-; RV32-NEXT:    .cfi_restore s6
-; RV32-NEXT:    .cfi_restore s7
-; RV32-NEXT:    .cfi_restore s8
 ; RV32-NEXT:    addi sp, sp, 80
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
 ; RV32-NEXT:    ret
@@ -4399,271 +4475,284 @@ define i64 @stest_f64i64_mm(double %x) {
 ; RV32IF-NEXT:    addi sp, sp, -80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 80
 ; RV32IF-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s9, 36(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    .cfi_offset ra, -4
-; RV32IF-NEXT:    .cfi_offset s0, -8
-; RV32IF-NEXT:    .cfi_offset s1, -12
-; RV32IF-NEXT:    .cfi_offset s2, -16
-; RV32IF-NEXT:    .cfi_offset s3, -20
-; RV32IF-NEXT:    .cfi_offset s4, -24
-; RV32IF-NEXT:    .cfi_offset s5, -28
-; RV32IF-NEXT:    .cfi_offset s6, -32
-; RV32IF-NEXT:    .cfi_offset s7, -36
-; RV32IF-NEXT:    .cfi_offset s8, -40
-; RV32IF-NEXT:    .cfi_offset s9, -44
 ; RV32IF-NEXT:    li a3, 1023
 ; RV32IF-NEXT:    slli a2, a1, 1
 ; RV32IF-NEXT:    srli a2, a2, 21
 ; RV32IF-NEXT:    bgeu a2, a3, .LBB45_2
 ; RV32IF-NEXT:  # %bb.1:
-; RV32IF-NEXT:    li s1, 0
-; RV32IF-NEXT:    li s2, 0
-; RV32IF-NEXT:    li a0, 0
+; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    j .LBB45_8
+; RV32IF-NEXT:    li a0, 0
+; RV32IF-NEXT:    li a2, 0
+; RV32IF-NEXT:    j .LBB45_9
 ; RV32IF-NEXT:  .LBB45_2: # %fp-to-i-if-check.exp.size
-; RV32IF-NEXT:    srai s0, a1, 31
-; RV32IF-NEXT:    lui a3, 256
-; RV32IF-NEXT:    snez a4, a0
-; RV32IF-NEXT:    addi a3, a3, -1
+; RV32IF-NEXT:    mv a3, a1
+; RV32IF-NEXT:    srai a1, a1, 31
+; RV32IF-NEXT:    lui a4, 256
+; RV32IF-NEXT:    snez a5, a0
+; RV32IF-NEXT:    addi a4, a4, -1
 ; RV32IF-NEXT:    addi a0, a0, -1
-; RV32IF-NEXT:    and a1, a1, a3
+; RV32IF-NEXT:    and a3, a3, a4
 ; RV32IF-NEXT:    addi a0, a0, 1
-; RV32IF-NEXT:    add a1, a1, a4
-; RV32IF-NEXT:    seqz a4, a0
-; RV32IF-NEXT:    add a3, a4, a3
-; RV32IF-NEXT:    li a4, 1074
-; RV32IF-NEXT:    ori s2, s0, 1
-; RV32IF-NEXT:    add a1, a1, a3
-; RV32IF-NEXT:    bltu a4, a2, .LBB45_5
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    seqz a5, a0
+; RV32IF-NEXT:    add a5, a5, a4
+; RV32IF-NEXT:    li a6, 1074
+; RV32IF-NEXT:    ori a4, a1, 1
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    sw a1, 36(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a4, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    bltu a6, a2, .LBB45_5
 ; RV32IF-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    li a3, 1043
-; RV32IF-NEXT:    li a4, 1075
-; RV32IF-NEXT:    sub a3, a3, a2
-; RV32IF-NEXT:    sub a2, a4, a2
-; RV32IF-NEXT:    srai a4, a3, 31
-; RV32IF-NEXT:    srl a5, a1, a2
-; RV32IF-NEXT:    and s3, a4, a5
-; RV32IF-NEXT:    bltz a3, .LBB45_6
+; RV32IF-NEXT:    li a5, 1043
+; RV32IF-NEXT:    li a7, 1075
+; RV32IF-NEXT:    sub a6, a5, a2
+; RV32IF-NEXT:    sub a5, a7, a2
+; RV32IF-NEXT:    srai a2, a6, 31
+; RV32IF-NEXT:    srl a7, a3, a5
+; RV32IF-NEXT:    and a2, a2, a7
+; RV32IF-NEXT:    bltz a6, .LBB45_6
 ; RV32IF-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    srl s4, a1, a3
+; RV32IF-NEXT:    srl a0, a3, a6
 ; RV32IF-NEXT:    j .LBB45_7
 ; RV32IF-NEXT:  .LBB45_5: # %fp-to-i-if-exp.large
-; RV32IF-NEXT:    addi a4, a2, -1075
-; RV32IF-NEXT:    srli a2, a4, 3
-; RV32IF-NEXT:    sw a0, 16(sp)
-; RV32IF-NEXT:    sw a1, 20(sp)
-; RV32IF-NEXT:    sw zero, 24(sp)
-; RV32IF-NEXT:    sw zero, 28(sp)
+; RV32IF-NEXT:    addi a5, a2, -1075
+; RV32IF-NEXT:    srli a2, a5, 3
+; RV32IF-NEXT:    sw a0, 56(sp)
+; RV32IF-NEXT:    sw a3, 60(sp)
+; RV32IF-NEXT:    sw zero, 64(sp)
+; RV32IF-NEXT:    sw zero, 68(sp)
 ; RV32IF-NEXT:    andi a2, a2, 12
-; RV32IF-NEXT:    addi a0, sp, 16
-; RV32IF-NEXT:    sw zero, 0(sp)
-; RV32IF-NEXT:    sw zero, 4(sp)
-; RV32IF-NEXT:    sw zero, 8(sp)
-; RV32IF-NEXT:    sw zero, 12(sp)
+; RV32IF-NEXT:    addi a0, sp, 56
+; RV32IF-NEXT:    sw zero, 40(sp)
+; RV32IF-NEXT:    sw zero, 44(sp)
+; RV32IF-NEXT:    sw zero, 48(sp)
+; RV32IF-NEXT:    sw zero, 52(sp)
 ; RV32IF-NEXT:    sub a0, a0, a2
-; RV32IF-NEXT:    lw a1, 0(a0)
+; RV32IF-NEXT:    lw a6, 0(a0)
 ; RV32IF-NEXT:    lw a2, 4(a0)
-; RV32IF-NEXT:    andi a3, a4, 31
-; RV32IF-NEXT:    lw a5, 8(a0)
-; RV32IF-NEXT:    lw a0, 12(a0)
-; RV32IF-NEXT:    srli a6, a1, 1
+; RV32IF-NEXT:    andi a3, a5, 31
+; RV32IF-NEXT:    lw a7, 8(a0)
+; RV32IF-NEXT:    lw t0, 12(a0)
+; RV32IF-NEXT:    srli a0, a6, 1
 ; RV32IF-NEXT:    xori a3, a3, 31
-; RV32IF-NEXT:    sll a7, a2, a4
-; RV32IF-NEXT:    srl a6, a6, a3
-; RV32IF-NEXT:    or s3, a7, a6
+; RV32IF-NEXT:    sll t1, a2, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a0, t1, a0
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    srli a2, a2, 1
-; RV32IF-NEXT:    sll a6, a5, a4
+; RV32IF-NEXT:    sll a0, a7, a5
 ; RV32IF-NEXT:    srl a2, a2, a3
-; RV32IF-NEXT:    or a2, a6, a2
-; RV32IF-NEXT:    srli a5, a5, 1
-; RV32IF-NEXT:    sll a0, a0, a4
-; RV32IF-NEXT:    srl a3, a5, a3
-; RV32IF-NEXT:    or a3, a0, a3
-; RV32IF-NEXT:    sll s4, a1, a4
-; RV32IF-NEXT:    mv a0, s2
-; RV32IF-NEXT:    mv a1, s0
+; RV32IF-NEXT:    or a2, a0, a2
+; RV32IF-NEXT:    srli a0, a7, 1
+; RV32IF-NEXT:    sll a7, t0, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a3, a7, a0
+; RV32IF-NEXT:    sll a0, a6, a5
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    mv a0, a4
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s7, a0, s1
-; RV32IF-NEXT:    add a1, a1, s5
-; RV32IF-NEXT:    sltu a0, s7, a0
-; RV32IF-NEXT:    add s8, a1, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a1, a2
+; RV32IF-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    add a0, a2, a0
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s5, a0
-; RV32IF-NEXT:    mv s6, a1
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    add s2, s5, a1
-; RV32IF-NEXT:    sltu a0, s2, s5
-; RV32IF-NEXT:    add s6, s6, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s2
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s4, s6, a0
-; RV32IF-NEXT:    sltu s5, s4, s6
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s4, a0, s4
-; RV32IF-NEXT:    add a1, a1, s5
-; RV32IF-NEXT:    sltu a0, s4, a0
-; RV32IF-NEXT:    add a1, a1, a0
-; RV32IF-NEXT:    add a0, s4, s7
-; RV32IF-NEXT:    add a1, a1, s8
-; RV32IF-NEXT:    sltu a2, a0, s4
-; RV32IF-NEXT:    add a1, a1, a2
+; RV32IF-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a0, a4
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a2, a4
+; RV32IF-NEXT:    sltu a0, a5, a0
+; RV32IF-NEXT:    add a4, a2, a0
+; RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a5, a0
+; RV32IF-NEXT:    lw a2, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a4, a2
+; RV32IF-NEXT:    sltu a4, a0, a5
 ; RV32IF-NEXT:    j .LBB45_8
 ; RV32IF-NEXT:  .LBB45_6:
-; RV32IF-NEXT:    not a3, a2
-; RV32IF-NEXT:    slli a1, a1, 1
-; RV32IF-NEXT:    srl a0, a0, a2
-; RV32IF-NEXT:    sll a1, a1, a3
-; RV32IF-NEXT:    or s4, a0, a1
+; RV32IF-NEXT:    not a6, a5
+; RV32IF-NEXT:    slli a3, a3, 1
+; RV32IF-NEXT:    srl a0, a0, a5
+; RV32IF-NEXT:    sll a3, a3, a6
+; RV32IF-NEXT:    or a0, a0, a3
 ; RV32IF-NEXT:  .LBB45_7: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s5
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add s7, a1, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a0, a1, a0
+; RV32IF-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s2
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s8, s7, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s5, a0
-; RV32IF-NEXT:    mv s6, a1
-; RV32IF-NEXT:    add s9, a0, s8
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    sltu a2, s8, s7
-; RV32IF-NEXT:    sltu a3, s9, s5
-; RV32IF-NEXT:    add a2, s6, a2
-; RV32IF-NEXT:    add a0, s9, a0
-; RV32IF-NEXT:    add a2, a2, a3
-; RV32IF-NEXT:    sltu a3, a0, s9
-; RV32IF-NEXT:    add a1, a2, a1
-; RV32IF-NEXT:    add a1, a1, a3
+; RV32IF-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    mv a2, a1
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a5, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a6, a5, a4
+; RV32IF-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a5, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a5, a5, a4
+; RV32IF-NEXT:    lw a4, 0(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a6, a4, a6
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a4, a0
+; RV32IF-NEXT:    add a5, a6, a5
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a4, a0, a4
+; RV32IF-NEXT:    add a2, a5, a2
 ; RV32IF-NEXT:  .LBB45_8: # %fp-to-i-cleanup
-; RV32IF-NEXT:    lui a2, 524288
-; RV32IF-NEXT:    addi a3, a2, -1
-; RV32IF-NEXT:    beq s2, a3, .LBB45_10
-; RV32IF-NEXT:  # %bb.9: # %fp-to-i-cleanup
-; RV32IF-NEXT:    sltu a4, s2, a3
-; RV32IF-NEXT:    or a5, a0, a1
-; RV32IF-NEXT:    bnez a5, .LBB45_11
-; RV32IF-NEXT:    j .LBB45_12
-; RV32IF-NEXT:  .LBB45_10:
-; RV32IF-NEXT:    sltiu a4, s1, -1
-; RV32IF-NEXT:    or a5, a0, a1
-; RV32IF-NEXT:    beqz a5, .LBB45_12
-; RV32IF-NEXT:  .LBB45_11: # %fp-to-i-cleanup
-; RV32IF-NEXT:    srli a4, a1, 31
+; RV32IF-NEXT:    add a2, a2, a4
+; RV32IF-NEXT:  .LBB45_9: # %fp-to-i-cleanup
+; RV32IF-NEXT:    lui a4, 524288
+; RV32IF-NEXT:    addi a5, a4, -1
+; RV32IF-NEXT:    beq a1, a5, .LBB45_11
+; RV32IF-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IF-NEXT:    sltu a6, a1, a5
+; RV32IF-NEXT:    or a7, a0, a2
+; RV32IF-NEXT:    bnez a7, .LBB45_12
+; RV32IF-NEXT:    j .LBB45_13
+; RV32IF-NEXT:  .LBB45_11:
+; RV32IF-NEXT:    sltiu a6, a3, -1
+; RV32IF-NEXT:    or a7, a0, a2
+; RV32IF-NEXT:    beqz a7, .LBB45_13
 ; RV32IF-NEXT:  .LBB45_12: # %fp-to-i-cleanup
-; RV32IF-NEXT:    neg a5, a4
-; RV32IF-NEXT:    addi a6, a4, -1
-; RV32IF-NEXT:    bnez a4, .LBB45_14
-; RV32IF-NEXT:  # %bb.13: # %fp-to-i-cleanup
-; RV32IF-NEXT:    mv s2, a3
-; RV32IF-NEXT:  .LBB45_14: # %fp-to-i-cleanup
-; RV32IF-NEXT:    or a3, a6, s1
-; RV32IF-NEXT:    and a1, a5, a1
-; RV32IF-NEXT:    and a4, a5, a0
-; RV32IF-NEXT:    beq s2, a2, .LBB45_16
-; RV32IF-NEXT:  # %bb.15: # %fp-to-i-cleanup
-; RV32IF-NEXT:    sltu a0, a2, s2
-; RV32IF-NEXT:    j .LBB45_17
-; RV32IF-NEXT:  .LBB45_16:
+; RV32IF-NEXT:    srli a6, a2, 31
+; RV32IF-NEXT:  .LBB45_13: # %fp-to-i-cleanup
+; RV32IF-NEXT:    neg a7, a6
+; RV32IF-NEXT:    addi t0, a6, -1
+; RV32IF-NEXT:    bnez a6, .LBB45_15
+; RV32IF-NEXT:  # %bb.14: # %fp-to-i-cleanup
+; RV32IF-NEXT:    mv a1, a5
+; RV32IF-NEXT:  .LBB45_15: # %fp-to-i-cleanup
+; RV32IF-NEXT:    or a3, t0, a3
+; RV32IF-NEXT:    and a2, a7, a2
+; RV32IF-NEXT:    and a5, a7, a0
+; RV32IF-NEXT:    beq a1, a4, .LBB45_17
+; RV32IF-NEXT:  # %bb.16: # %fp-to-i-cleanup
+; RV32IF-NEXT:    sltu a0, a4, a1
+; RV32IF-NEXT:    j .LBB45_18
+; RV32IF-NEXT:  .LBB45_17:
 ; RV32IF-NEXT:    snez a0, a3
-; RV32IF-NEXT:  .LBB45_17: # %fp-to-i-cleanup
-; RV32IF-NEXT:    and a4, a4, a1
-; RV32IF-NEXT:    li a2, -1
-; RV32IF-NEXT:    beq a4, a2, .LBB45_19
-; RV32IF-NEXT:  # %bb.18: # %fp-to-i-cleanup
-; RV32IF-NEXT:    srli a1, a1, 31
-; RV32IF-NEXT:    xori a0, a1, 1
-; RV32IF-NEXT:  .LBB45_19: # %fp-to-i-cleanup
-; RV32IF-NEXT:    bnez a0, .LBB45_21
-; RV32IF-NEXT:  # %bb.20: # %fp-to-i-cleanup
-; RV32IF-NEXT:    lui s2, 524288
-; RV32IF-NEXT:  .LBB45_21: # %fp-to-i-cleanup
+; RV32IF-NEXT:  .LBB45_18: # %fp-to-i-cleanup
+; RV32IF-NEXT:    and a5, a5, a2
+; RV32IF-NEXT:    li a4, -1
+; RV32IF-NEXT:    beq a5, a4, .LBB45_20
+; RV32IF-NEXT:  # %bb.19: # %fp-to-i-cleanup
+; RV32IF-NEXT:    srli a2, a2, 31
+; RV32IF-NEXT:    xori a0, a2, 1
+; RV32IF-NEXT:  .LBB45_20: # %fp-to-i-cleanup
+; RV32IF-NEXT:    bnez a0, .LBB45_22
+; RV32IF-NEXT:  # %bb.21: # %fp-to-i-cleanup
+; RV32IF-NEXT:    lui a1, 524288
+; RV32IF-NEXT:  .LBB45_22: # %fp-to-i-cleanup
 ; RV32IF-NEXT:    neg a0, a0
 ; RV32IF-NEXT:    and a0, a0, a3
-; RV32IF-NEXT:    mv a1, s2
 ; RV32IF-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s9, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    .cfi_restore ra
-; RV32IF-NEXT:    .cfi_restore s0
-; RV32IF-NEXT:    .cfi_restore s1
-; RV32IF-NEXT:    .cfi_restore s2
-; RV32IF-NEXT:    .cfi_restore s3
-; RV32IF-NEXT:    .cfi_restore s4
-; RV32IF-NEXT:    .cfi_restore s5
-; RV32IF-NEXT:    .cfi_restore s6
-; RV32IF-NEXT:    .cfi_restore s7
-; RV32IF-NEXT:    .cfi_restore s8
-; RV32IF-NEXT:    .cfi_restore s9
 ; RV32IF-NEXT:    addi sp, sp, 80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IF-NEXT:    ret
@@ -4714,274 +4803,286 @@ define i64 @stest_f64i64_mm(double %x) {
 ; RV32IFD-NEXT:    addi sp, sp, -96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 96
 ; RV32IFD-NEXT:    sw ra, 92(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s0, 88(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s1, 84(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s2, 80(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s3, 76(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s4, 72(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s5, 68(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s6, 64(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s7, 60(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s8, 56(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s9, 52(sp) # 4-byte Folded Spill
 ; RV32IFD-NEXT:    .cfi_offset ra, -4
-; RV32IFD-NEXT:    .cfi_offset s0, -8
-; RV32IFD-NEXT:    .cfi_offset s1, -12
-; RV32IFD-NEXT:    .cfi_offset s2, -16
-; RV32IFD-NEXT:    .cfi_offset s3, -20
-; RV32IFD-NEXT:    .cfi_offset s4, -24
-; RV32IFD-NEXT:    .cfi_offset s5, -28
-; RV32IFD-NEXT:    .cfi_offset s6, -32
-; RV32IFD-NEXT:    .cfi_offset s7, -36
-; RV32IFD-NEXT:    .cfi_offset s8, -40
-; RV32IFD-NEXT:    .cfi_offset s9, -44
-; RV32IFD-NEXT:    li a2, 1023
-; RV32IFD-NEXT:    fsd fa0, 8(sp)
-; RV32IFD-NEXT:    lw a1, 12(sp)
-; RV32IFD-NEXT:    slli a0, a1, 1
+; RV32IFD-NEXT:    li a1, 1023
+; RV32IFD-NEXT:    fsd fa0, 48(sp)
+; RV32IFD-NEXT:    lw a2, 52(sp)
+; RV32IFD-NEXT:    slli a0, a2, 1
 ; RV32IFD-NEXT:    srli a0, a0, 21
-; RV32IFD-NEXT:    bgeu a0, a2, .LBB45_2
+; RV32IFD-NEXT:    bgeu a0, a1, .LBB45_2
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    li s1, 0
-; RV32IFD-NEXT:    li s2, 0
-; RV32IFD-NEXT:    li a0, 0
+; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    j .LBB45_8
+; RV32IFD-NEXT:    li a0, 0
+; RV32IFD-NEXT:    li a2, 0
+; RV32IFD-NEXT:    j .LBB45_9
 ; RV32IFD-NEXT:  .LBB45_2: # %fp-to-i-if-check.exp.size
-; RV32IFD-NEXT:    lw a2, 8(sp)
-; RV32IFD-NEXT:    lui a3, 256
-; RV32IFD-NEXT:    srai s0, a1, 31
-; RV32IFD-NEXT:    addi a3, a3, -1
-; RV32IFD-NEXT:    and a4, a1, a3
-; RV32IFD-NEXT:    addi a1, a2, -1
-; RV32IFD-NEXT:    snez a2, a2
-; RV32IFD-NEXT:    addi a1, a1, 1
-; RV32IFD-NEXT:    add a2, a4, a2
-; RV32IFD-NEXT:    seqz a4, a1
-; RV32IFD-NEXT:    add a2, a2, a3
-; RV32IFD-NEXT:    li a3, 1074
-; RV32IFD-NEXT:    ori s2, s0, 1
-; RV32IFD-NEXT:    add a2, a2, a4
-; RV32IFD-NEXT:    bltu a3, a0, .LBB45_5
+; RV32IFD-NEXT:    lw a3, 48(sp)
+; RV32IFD-NEXT:    lui a4, 256
+; RV32IFD-NEXT:    srai a1, a2, 31
+; RV32IFD-NEXT:    addi a4, a4, -1
+; RV32IFD-NEXT:    and a5, a2, a4
+; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    snez a3, a3
+; RV32IFD-NEXT:    addi a2, a2, 1
+; RV32IFD-NEXT:    add a3, a5, a3
+; RV32IFD-NEXT:    seqz a5, a2
+; RV32IFD-NEXT:    add a3, a3, a4
+; RV32IFD-NEXT:    li a6, 1074
+; RV32IFD-NEXT:    ori a4, a1, 1
+; RV32IFD-NEXT:    add a3, a3, a5
+; RV32IFD-NEXT:    sw a1, 44(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a4, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    bltu a6, a0, .LBB45_5
 ; RV32IFD-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    li a3, 1043
-; RV32IFD-NEXT:    li a4, 1075
-; RV32IFD-NEXT:    sub a3, a3, a0
-; RV32IFD-NEXT:    sub a0, a4, a0
-; RV32IFD-NEXT:    srai a4, a3, 31
-; RV32IFD-NEXT:    srl a5, a2, a0
-; RV32IFD-NEXT:    and s3, a4, a5
-; RV32IFD-NEXT:    bltz a3, .LBB45_6
+; RV32IFD-NEXT:    li a5, 1043
+; RV32IFD-NEXT:    li a7, 1075
+; RV32IFD-NEXT:    sub a6, a5, a0
+; RV32IFD-NEXT:    sub a5, a7, a0
+; RV32IFD-NEXT:    srai a0, a6, 31
+; RV32IFD-NEXT:    srl a7, a3, a5
+; RV32IFD-NEXT:    and a0, a0, a7
+; RV32IFD-NEXT:    bltz a6, .LBB45_6
 ; RV32IFD-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    srl s4, a2, a3
+; RV32IFD-NEXT:    srl a2, a3, a6
 ; RV32IFD-NEXT:    j .LBB45_7
 ; RV32IFD-NEXT:  .LBB45_5: # %fp-to-i-if-exp.large
-; RV32IFD-NEXT:    addi a0, a0, -1075
-; RV32IFD-NEXT:    srli a3, a0, 3
-; RV32IFD-NEXT:    sw a1, 32(sp)
-; RV32IFD-NEXT:    sw a2, 36(sp)
-; RV32IFD-NEXT:    sw zero, 40(sp)
-; RV32IFD-NEXT:    sw zero, 44(sp)
-; RV32IFD-NEXT:    andi a3, a3, 12
-; RV32IFD-NEXT:    addi a1, sp, 32
-; RV32IFD-NEXT:    sw zero, 16(sp)
-; RV32IFD-NEXT:    sw zero, 20(sp)
-; RV32IFD-NEXT:    sw zero, 24(sp)
-; RV32IFD-NEXT:    sw zero, 28(sp)
-; RV32IFD-NEXT:    sub a1, a1, a3
-; RV32IFD-NEXT:    lw a4, 0(a1)
-; RV32IFD-NEXT:    lw a2, 4(a1)
-; RV32IFD-NEXT:    andi a3, a0, 31
-; RV32IFD-NEXT:    lw a5, 8(a1)
-; RV32IFD-NEXT:    lw a1, 12(a1)
-; RV32IFD-NEXT:    srli a6, a4, 1
-; RV32IFD-NEXT:    xori a3, a3, 31
-; RV32IFD-NEXT:    sll a7, a2, a0
-; RV32IFD-NEXT:    srl a6, a6, a3
-; RV32IFD-NEXT:    or s3, a7, a6
-; RV32IFD-NEXT:    srli a2, a2, 1
-; RV32IFD-NEXT:    sll a6, a5, a0
-; RV32IFD-NEXT:    srl a2, a2, a3
-; RV32IFD-NEXT:    or a2, a6, a2
-; RV32IFD-NEXT:    srli a5, a5, 1
-; RV32IFD-NEXT:    sll a1, a1, a0
-; RV32IFD-NEXT:    srl a3, a5, a3
-; RV32IFD-NEXT:    or a3, a1, a3
-; RV32IFD-NEXT:    sll s4, a4, a0
-; RV32IFD-NEXT:    mv a0, s2
-; RV32IFD-NEXT:    mv a1, s0
+; RV32IFD-NEXT:    addi a5, a0, -1075
+; RV32IFD-NEXT:    srli a0, a5, 3
+; RV32IFD-NEXT:    sw a2, 72(sp)
+; RV32IFD-NEXT:    sw a3, 76(sp)
+; RV32IFD-NEXT:    sw zero, 80(sp)
+; RV32IFD-NEXT:    sw zero, 84(sp)
+; RV32IFD-NEXT:    andi a0, a0, 12
+; RV32IFD-NEXT:    addi a2, sp, 72
+; RV32IFD-NEXT:    sw zero, 56(sp)
+; RV32IFD-NEXT:    sw zero, 60(sp)
+; RV32IFD-NEXT:    sw zero, 64(sp)
+; RV32IFD-NEXT:    sw zero, 68(sp)
+; RV32IFD-NEXT:    sub a2, a2, a0
+; RV32IFD-NEXT:    lw a6, 0(a2)
+; RV32IFD-NEXT:    lw a3, 4(a2)
+; RV32IFD-NEXT:    andi a0, a5, 31
+; RV32IFD-NEXT:    lw a7, 8(a2)
+; RV32IFD-NEXT:    lw t0, 12(a2)
+; RV32IFD-NEXT:    srli a2, a6, 1
+; RV32IFD-NEXT:    xori t1, a0, 31
+; RV32IFD-NEXT:    sll a0, a3, a5
+; RV32IFD-NEXT:    srl a2, a2, t1
+; RV32IFD-NEXT:    or a0, a0, a2
+; RV32IFD-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    srli a3, a3, 1
+; RV32IFD-NEXT:    sll a0, a7, a5
+; RV32IFD-NEXT:    srl a2, a3, t1
+; RV32IFD-NEXT:    or a2, a0, a2
+; RV32IFD-NEXT:    srli a0, a7, 1
+; RV32IFD-NEXT:    sll a3, t0, a5
+; RV32IFD-NEXT:    srl a0, a0, t1
+; RV32IFD-NEXT:    or a3, a3, a0
+; RV32IFD-NEXT:    sll a0, a6, a5
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    mv a0, a4
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s7, a0, s1
-; RV32IFD-NEXT:    add a1, a1, s5
-; RV32IFD-NEXT:    sltu a0, s7, a0
-; RV32IFD-NEXT:    add s8, a1, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a1, a2
+; RV32IFD-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    add a0, a2, a0
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s5, a0
-; RV32IFD-NEXT:    mv s6, a1
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    add s2, s5, a1
-; RV32IFD-NEXT:    sltu a0, s2, s5
-; RV32IFD-NEXT:    add s6, s6, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s2
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s4, s6, a0
-; RV32IFD-NEXT:    sltu s5, s4, s6
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s4, a0, s4
-; RV32IFD-NEXT:    add a1, a1, s5
-; RV32IFD-NEXT:    sltu a0, s4, a0
-; RV32IFD-NEXT:    add a1, a1, a0
-; RV32IFD-NEXT:    add a0, s4, s7
-; RV32IFD-NEXT:    add a1, a1, s8
-; RV32IFD-NEXT:    sltu a2, a0, s4
-; RV32IFD-NEXT:    add a1, a1, a2
+; RV32IFD-NEXT:    lw a3, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a0, a4
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a2, a4
+; RV32IFD-NEXT:    sltu a0, a5, a0
+; RV32IFD-NEXT:    add a4, a2, a0
+; RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a5, a0
+; RV32IFD-NEXT:    lw a2, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a4, a2
+; RV32IFD-NEXT:    sltu a4, a0, a5
 ; RV32IFD-NEXT:    j .LBB45_8
 ; RV32IFD-NEXT:  .LBB45_6:
-; RV32IFD-NEXT:    not a3, a0
-; RV32IFD-NEXT:    slli a2, a2, 1
-; RV32IFD-NEXT:    srl a0, a1, a0
-; RV32IFD-NEXT:    sll a1, a2, a3
-; RV32IFD-NEXT:    or s4, a0, a1
+; RV32IFD-NEXT:    not a6, a5
+; RV32IFD-NEXT:    slli a3, a3, 1
+; RV32IFD-NEXT:    srl a2, a2, a5
+; RV32IFD-NEXT:    sll a3, a3, a6
+; RV32IFD-NEXT:    or a2, a2, a3
 ; RV32IFD-NEXT:  .LBB45_7: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a2, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s5
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add s7, a1, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s2
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s8, s7, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s5, a0
-; RV32IFD-NEXT:    mv s6, a1
-; RV32IFD-NEXT:    add s9, a0, s8
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    sltu a2, s8, s7
-; RV32IFD-NEXT:    sltu a3, s9, s5
-; RV32IFD-NEXT:    add a2, s6, a2
-; RV32IFD-NEXT:    add a0, s9, a0
-; RV32IFD-NEXT:    add a2, a2, a3
-; RV32IFD-NEXT:    sltu a3, a0, s9
-; RV32IFD-NEXT:    add a1, a2, a1
-; RV32IFD-NEXT:    add a1, a1, a3
+; RV32IFD-NEXT:    lw a3, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    mv a2, a1
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a5, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a6, a5, a4
+; RV32IFD-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a5, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a5, a5, a4
+; RV32IFD-NEXT:    lw a4, 8(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a6, a4, a6
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a4, a0
+; RV32IFD-NEXT:    add a5, a6, a5
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a4, a0, a4
+; RV32IFD-NEXT:    add a2, a5, a2
 ; RV32IFD-NEXT:  .LBB45_8: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    lui a2, 524288
-; RV32IFD-NEXT:    addi a3, a2, -1
-; RV32IFD-NEXT:    beq s2, a3, .LBB45_10
-; RV32IFD-NEXT:  # %bb.9: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    sltu a4, s2, a3
-; RV32IFD-NEXT:    or a5, a0, a1
-; RV32IFD-NEXT:    bnez a5, .LBB45_11
-; RV32IFD-NEXT:    j .LBB45_12
-; RV32IFD-NEXT:  .LBB45_10:
-; RV32IFD-NEXT:    sltiu a4, s1, -1
-; RV32IFD-NEXT:    or a5, a0, a1
-; RV32IFD-NEXT:    beqz a5, .LBB45_12
-; RV32IFD-NEXT:  .LBB45_11: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    srli a4, a1, 31
+; RV32IFD-NEXT:    add a2, a2, a4
+; RV32IFD-NEXT:  .LBB45_9: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    lui a4, 524288
+; RV32IFD-NEXT:    addi a5, a4, -1
+; RV32IFD-NEXT:    beq a1, a5, .LBB45_11
+; RV32IFD-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    sltu a6, a1, a5
+; RV32IFD-NEXT:    or a7, a0, a2
+; RV32IFD-NEXT:    bnez a7, .LBB45_12
+; RV32IFD-NEXT:    j .LBB45_13
+; RV32IFD-NEXT:  .LBB45_11:
+; RV32IFD-NEXT:    sltiu a6, a3, -1
+; RV32IFD-NEXT:    or a7, a0, a2
+; RV32IFD-NEXT:    beqz a7, .LBB45_13
 ; RV32IFD-NEXT:  .LBB45_12: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    neg a5, a4
-; RV32IFD-NEXT:    addi a6, a4, -1
-; RV32IFD-NEXT:    bnez a4, .LBB45_14
-; RV32IFD-NEXT:  # %bb.13: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    mv s2, a3
-; RV32IFD-NEXT:  .LBB45_14: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    or a3, a6, s1
-; RV32IFD-NEXT:    and a1, a5, a1
-; RV32IFD-NEXT:    and a4, a5, a0
-; RV32IFD-NEXT:    beq s2, a2, .LBB45_16
-; RV32IFD-NEXT:  # %bb.15: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    sltu a0, a2, s2
-; RV32IFD-NEXT:    j .LBB45_17
-; RV32IFD-NEXT:  .LBB45_16:
+; RV32IFD-NEXT:    srli a6, a2, 31
+; RV32IFD-NEXT:  .LBB45_13: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    neg a7, a6
+; RV32IFD-NEXT:    addi t0, a6, -1
+; RV32IFD-NEXT:    bnez a6, .LBB45_15
+; RV32IFD-NEXT:  # %bb.14: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    mv a1, a5
+; RV32IFD-NEXT:  .LBB45_15: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    or a3, t0, a3
+; RV32IFD-NEXT:    and a2, a7, a2
+; RV32IFD-NEXT:    and a5, a7, a0
+; RV32IFD-NEXT:    beq a1, a4, .LBB45_17
+; RV32IFD-NEXT:  # %bb.16: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    sltu a0, a4, a1
+; RV32IFD-NEXT:    j .LBB45_18
+; RV32IFD-NEXT:  .LBB45_17:
 ; RV32IFD-NEXT:    snez a0, a3
-; RV32IFD-NEXT:  .LBB45_17: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    and a4, a4, a1
-; RV32IFD-NEXT:    li a2, -1
-; RV32IFD-NEXT:    beq a4, a2, .LBB45_19
-; RV32IFD-NEXT:  # %bb.18: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    srli a1, a1, 31
-; RV32IFD-NEXT:    xori a0, a1, 1
-; RV32IFD-NEXT:  .LBB45_19: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    bnez a0, .LBB45_21
-; RV32IFD-NEXT:  # %bb.20: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    lui s2, 524288
-; RV32IFD-NEXT:  .LBB45_21: # %fp-to-i-cleanup
+; RV32IFD-NEXT:  .LBB45_18: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    and a5, a5, a2
+; RV32IFD-NEXT:    li a4, -1
+; RV32IFD-NEXT:    beq a5, a4, .LBB45_20
+; RV32IFD-NEXT:  # %bb.19: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    srli a2, a2, 31
+; RV32IFD-NEXT:    xori a0, a2, 1
+; RV32IFD-NEXT:  .LBB45_20: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    bnez a0, .LBB45_22
+; RV32IFD-NEXT:  # %bb.21: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    lui a1, 524288
+; RV32IFD-NEXT:  .LBB45_22: # %fp-to-i-cleanup
 ; RV32IFD-NEXT:    neg a0, a0
 ; RV32IFD-NEXT:    and a0, a0, a3
-; RV32IFD-NEXT:    mv a1, s2
 ; RV32IFD-NEXT:    lw ra, 92(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s0, 88(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s1, 84(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s2, 80(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s3, 76(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s4, 72(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s5, 68(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s6, 64(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s7, 60(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s8, 56(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s9, 52(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    .cfi_restore ra
-; RV32IFD-NEXT:    .cfi_restore s0
-; RV32IFD-NEXT:    .cfi_restore s1
-; RV32IFD-NEXT:    .cfi_restore s2
-; RV32IFD-NEXT:    .cfi_restore s3
-; RV32IFD-NEXT:    .cfi_restore s4
-; RV32IFD-NEXT:    .cfi_restore s5
-; RV32IFD-NEXT:    .cfi_restore s6
-; RV32IFD-NEXT:    .cfi_restore s7
-; RV32IFD-NEXT:    .cfi_restore s8
-; RV32IFD-NEXT:    .cfi_restore s9
 ; RV32IFD-NEXT:    addi sp, sp, 96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IFD-NEXT:    ret
@@ -5228,245 +5329,256 @@ define i64 @ustest_f64i64_mm(double %x) {
 ; RV32IF-NEXT:    addi sp, sp, -80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 80
 ; RV32IF-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s9, 36(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    .cfi_offset ra, -4
-; RV32IF-NEXT:    .cfi_offset s0, -8
-; RV32IF-NEXT:    .cfi_offset s1, -12
-; RV32IF-NEXT:    .cfi_offset s2, -16
-; RV32IF-NEXT:    .cfi_offset s3, -20
-; RV32IF-NEXT:    .cfi_offset s4, -24
-; RV32IF-NEXT:    .cfi_offset s5, -28
-; RV32IF-NEXT:    .cfi_offset s6, -32
-; RV32IF-NEXT:    .cfi_offset s7, -36
-; RV32IF-NEXT:    .cfi_offset s8, -40
-; RV32IF-NEXT:    .cfi_offset s9, -44
 ; RV32IF-NEXT:    li a3, 1023
 ; RV32IF-NEXT:    slli a2, a1, 1
 ; RV32IF-NEXT:    srli a2, a2, 21
-; RV32IF-NEXT:    bgeu a2, a3, .LBB47_2
+; RV32IF-NEXT:    bgeu a2, a3, .LBB47_3
 ; RV32IF-NEXT:  # %bb.1:
-; RV32IF-NEXT:    li s1, 0
-; RV32IF-NEXT:    li s6, 0
 ; RV32IF-NEXT:    li a2, 0
 ; RV32IF-NEXT:    li a0, 0
-; RV32IF-NEXT:    j .LBB47_6
-; RV32IF-NEXT:  .LBB47_2: # %fp-to-i-if-check.exp.size
-; RV32IF-NEXT:    srai s0, a1, 31
-; RV32IF-NEXT:    lui a3, 256
-; RV32IF-NEXT:    snez a4, a0
-; RV32IF-NEXT:    addi a3, a3, -1
+; RV32IF-NEXT:    li a3, 0
+; RV32IF-NEXT:    li a1, 0
+; RV32IF-NEXT:  .LBB47_2:
+; RV32IF-NEXT:    seqz a4, a3
+; RV32IF-NEXT:    j .LBB47_11
+; RV32IF-NEXT:  .LBB47_3: # %fp-to-i-if-check.exp.size
+; RV32IF-NEXT:    mv a3, a1
+; RV32IF-NEXT:    srai a1, a1, 31
+; RV32IF-NEXT:    lui a4, 256
+; RV32IF-NEXT:    snez a5, a0
+; RV32IF-NEXT:    addi a4, a4, -1
 ; RV32IF-NEXT:    addi a0, a0, -1
-; RV32IF-NEXT:    and a1, a1, a3
+; RV32IF-NEXT:    and a3, a3, a4
 ; RV32IF-NEXT:    addi a0, a0, 1
-; RV32IF-NEXT:    add a1, a1, a4
-; RV32IF-NEXT:    seqz a4, a0
-; RV32IF-NEXT:    add a3, a4, a3
-; RV32IF-NEXT:    li a4, 1074
-; RV32IF-NEXT:    ori s2, s0, 1
-; RV32IF-NEXT:    add a1, a1, a3
-; RV32IF-NEXT:    bltu a4, a2, .LBB47_5
-; RV32IF-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    li a3, 1043
-; RV32IF-NEXT:    li a4, 1075
-; RV32IF-NEXT:    sub a3, a3, a2
-; RV32IF-NEXT:    sub a2, a4, a2
-; RV32IF-NEXT:    srai a4, a3, 31
-; RV32IF-NEXT:    srl a5, a1, a2
-; RV32IF-NEXT:    and s3, a4, a5
-; RV32IF-NEXT:    bltz a3, .LBB47_7
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    seqz a5, a0
+; RV32IF-NEXT:    add a5, a5, a4
+; RV32IF-NEXT:    li a6, 1074
+; RV32IF-NEXT:    ori a4, a1, 1
+; RV32IF-NEXT:    add a3, a3, a5
+; RV32IF-NEXT:    sw a1, 36(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a4, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    bltu a6, a2, .LBB47_6
 ; RV32IF-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    srl s4, a1, a3
+; RV32IF-NEXT:    li a5, 1043
+; RV32IF-NEXT:    li a7, 1075
+; RV32IF-NEXT:    sub a6, a5, a2
+; RV32IF-NEXT:    sub a5, a7, a2
+; RV32IF-NEXT:    srai a2, a6, 31
+; RV32IF-NEXT:    srl a7, a3, a5
+; RV32IF-NEXT:    and a2, a2, a7
+; RV32IF-NEXT:    bltz a6, .LBB47_7
+; RV32IF-NEXT:  # %bb.5: # %fp-to-i-if-exp.small
+; RV32IF-NEXT:    srl a0, a3, a6
 ; RV32IF-NEXT:    j .LBB47_8
-; RV32IF-NEXT:  .LBB47_5: # %fp-to-i-if-exp.large
-; RV32IF-NEXT:    addi a4, a2, -1075
-; RV32IF-NEXT:    srli a2, a4, 3
-; RV32IF-NEXT:    sw a0, 16(sp)
-; RV32IF-NEXT:    sw a1, 20(sp)
-; RV32IF-NEXT:    sw zero, 24(sp)
-; RV32IF-NEXT:    sw zero, 28(sp)
+; RV32IF-NEXT:  .LBB47_6: # %fp-to-i-if-exp.large
+; RV32IF-NEXT:    addi a5, a2, -1075
+; RV32IF-NEXT:    srli a2, a5, 3
+; RV32IF-NEXT:    sw a0, 56(sp)
+; RV32IF-NEXT:    sw a3, 60(sp)
+; RV32IF-NEXT:    sw zero, 64(sp)
+; RV32IF-NEXT:    sw zero, 68(sp)
 ; RV32IF-NEXT:    andi a2, a2, 12
-; RV32IF-NEXT:    addi a0, sp, 16
-; RV32IF-NEXT:    sw zero, 0(sp)
-; RV32IF-NEXT:    sw zero, 4(sp)
-; RV32IF-NEXT:    sw zero, 8(sp)
-; RV32IF-NEXT:    sw zero, 12(sp)
+; RV32IF-NEXT:    addi a0, sp, 56
+; RV32IF-NEXT:    sw zero, 40(sp)
+; RV32IF-NEXT:    sw zero, 44(sp)
+; RV32IF-NEXT:    sw zero, 48(sp)
+; RV32IF-NEXT:    sw zero, 52(sp)
 ; RV32IF-NEXT:    sub a0, a0, a2
-; RV32IF-NEXT:    lw a1, 0(a0)
+; RV32IF-NEXT:    lw a6, 0(a0)
 ; RV32IF-NEXT:    lw a2, 4(a0)
-; RV32IF-NEXT:    andi a3, a4, 31
-; RV32IF-NEXT:    lw a5, 8(a0)
-; RV32IF-NEXT:    lw a0, 12(a0)
-; RV32IF-NEXT:    srli a6, a1, 1
+; RV32IF-NEXT:    andi a3, a5, 31
+; RV32IF-NEXT:    lw a7, 8(a0)
+; RV32IF-NEXT:    lw t0, 12(a0)
+; RV32IF-NEXT:    srli a0, a6, 1
 ; RV32IF-NEXT:    xori a3, a3, 31
-; RV32IF-NEXT:    sll a7, a2, a4
-; RV32IF-NEXT:    srl a6, a6, a3
-; RV32IF-NEXT:    or s3, a7, a6
+; RV32IF-NEXT:    sll t1, a2, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a0, t1, a0
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    srli a2, a2, 1
-; RV32IF-NEXT:    sll a6, a5, a4
+; RV32IF-NEXT:    sll a0, a7, a5
 ; RV32IF-NEXT:    srl a2, a2, a3
-; RV32IF-NEXT:    or a2, a6, a2
-; RV32IF-NEXT:    srli a5, a5, 1
-; RV32IF-NEXT:    sll a0, a0, a4
-; RV32IF-NEXT:    srl a3, a5, a3
-; RV32IF-NEXT:    or a3, a0, a3
-; RV32IF-NEXT:    sll s4, a1, a4
-; RV32IF-NEXT:    mv a0, s2
-; RV32IF-NEXT:    mv a1, s0
+; RV32IF-NEXT:    or a2, a0, a2
+; RV32IF-NEXT:    srli a0, a7, 1
+; RV32IF-NEXT:    sll a7, t0, a5
+; RV32IF-NEXT:    srl a0, a0, a3
+; RV32IF-NEXT:    or a3, a7, a0
+; RV32IF-NEXT:    sll a0, a6, a5
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    mv a0, a4
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s7, a0, s1
-; RV32IF-NEXT:    add a1, a1, s5
-; RV32IF-NEXT:    sltu a0, s7, a0
-; RV32IF-NEXT:    add s8, a1, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a1, a2
+; RV32IF-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    add a0, a2, a0
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s5, a0
-; RV32IF-NEXT:    mv s6, a1
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    add s2, s5, a1
-; RV32IF-NEXT:    sltu a0, s2, s5
-; RV32IF-NEXT:    add s5, s6, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s6, a0, s2
-; RV32IF-NEXT:    sltu a0, s6, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s2, s5, a0
-; RV32IF-NEXT:    sltu s4, s2, s5
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a1, a0
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s2, a0, s2
-; RV32IF-NEXT:    add a1, a1, s4
-; RV32IF-NEXT:    sltu a0, s2, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add a2, s2, s7
-; RV32IF-NEXT:    add a0, a0, s8
-; RV32IF-NEXT:    sltu a1, a2, s2
-; RV32IF-NEXT:    add a0, a0, a1
-; RV32IF-NEXT:    bnez a0, .LBB47_9
-; RV32IF-NEXT:  .LBB47_6:
-; RV32IF-NEXT:    seqz a1, a2
-; RV32IF-NEXT:    j .LBB47_10
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a4, a0, a3
+; RV32IF-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a1, a1, a3
+; RV32IF-NEXT:    sltu a3, a4, a0
+; RV32IF-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a1, a3
+; RV32IF-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a3, a4, a3
+; RV32IF-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a1, a5, a1
+; RV32IF-NEXT:    sltu a4, a3, a4
+; RV32IF-NEXT:    j .LBB47_9
 ; RV32IF-NEXT:  .LBB47_7:
-; RV32IF-NEXT:    not a3, a2
-; RV32IF-NEXT:    slli a1, a1, 1
-; RV32IF-NEXT:    srl a0, a0, a2
-; RV32IF-NEXT:    sll a1, a1, a3
-; RV32IF-NEXT:    or s4, a0, a1
+; RV32IF-NEXT:    not a6, a5
+; RV32IF-NEXT:    slli a3, a3, 1
+; RV32IF-NEXT:    srl a0, a0, a5
+; RV32IF-NEXT:    sll a3, a3, a6
+; RV32IF-NEXT:    or a0, a0, a3
 ; RV32IF-NEXT:  .LBB47_8: # %fp-to-i-if-exp.small
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s1, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s2
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s5, a0, s5
-; RV32IF-NEXT:    sltu a0, s5, a0
-; RV32IF-NEXT:    add s7, a1, a0
-; RV32IF-NEXT:    mv a0, s4
+; RV32IF-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a0, a1, a0
+; RV32IF-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    add s6, a0, s5
-; RV32IF-NEXT:    sltu a0, s6, a0
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    add s8, s7, a0
-; RV32IF-NEXT:    mv a0, s3
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a2, a0, a2
+; RV32IF-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a0, a2, a0
+; RV32IF-NEXT:    add a1, a1, a0
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a1, 0
-; RV32IF-NEXT:    mv a2, s0
+; RV32IF-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    li a3, 0
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    mv s2, a0
-; RV32IF-NEXT:    mv s5, a1
-; RV32IF-NEXT:    add s9, a0, s8
-; RV32IF-NEXT:    mv a0, s0
-; RV32IF-NEXT:    mv a1, s0
-; RV32IF-NEXT:    mv a2, s4
-; RV32IF-NEXT:    mv a3, s3
+; RV32IF-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a0, a0, a1
+; RV32IF-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    call __muldi3
-; RV32IF-NEXT:    sltu a2, s8, s7
-; RV32IF-NEXT:    sltu a3, s9, s2
-; RV32IF-NEXT:    add s5, s5, a2
-; RV32IF-NEXT:    add a2, s9, a0
-; RV32IF-NEXT:    add a3, s5, a3
-; RV32IF-NEXT:    sltu a0, a2, s9
-; RV32IF-NEXT:    add a1, a3, a1
-; RV32IF-NEXT:    add a0, a1, a0
-; RV32IF-NEXT:    beqz a0, .LBB47_6
+; RV32IF-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 4(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a5, a4, a3
+; RV32IF-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a4, a4, a3
+; RV32IF-NEXT:    lw a3, 0(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a3, a5
+; RV32IF-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a3, a3, a0
+; RV32IF-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    add a5, a5, a4
+; RV32IF-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    sltu a4, a3, a4
+; RV32IF-NEXT:    add a1, a5, a1
 ; RV32IF-NEXT:  .LBB47_9: # %fp-to-i-cleanup
-; RV32IF-NEXT:    srli a1, a0, 31
-; RV32IF-NEXT:  .LBB47_10: # %fp-to-i-cleanup
-; RV32IF-NEXT:    xori a2, a2, 1
-; RV32IF-NEXT:    or a2, a2, a0
-; RV32IF-NEXT:    seqz a2, a2
-; RV32IF-NEXT:    addi a2, a2, -1
-; RV32IF-NEXT:    and a1, a2, a1
-; RV32IF-NEXT:    neg a1, a1
-; RV32IF-NEXT:    and a0, a1, a0
-; RV32IF-NEXT:    and a2, a1, s6
-; RV32IF-NEXT:    srli a0, a0, 31
-; RV32IF-NEXT:    and a1, a1, s1
-; RV32IF-NEXT:    addi a3, a0, -1
-; RV32IF-NEXT:    and a0, a3, a1
-; RV32IF-NEXT:    and a1, a3, a2
+; RV32IF-NEXT:    add a1, a1, a4
+; RV32IF-NEXT:    beqz a1, .LBB47_2
+; RV32IF-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IF-NEXT:    srli a4, a1, 31
+; RV32IF-NEXT:  .LBB47_11: # %fp-to-i-cleanup
+; RV32IF-NEXT:    xori a3, a3, 1
+; RV32IF-NEXT:    or a3, a3, a1
+; RV32IF-NEXT:    seqz a3, a3
+; RV32IF-NEXT:    addi a3, a3, -1
+; RV32IF-NEXT:    and a3, a3, a4
+; RV32IF-NEXT:    neg a3, a3
+; RV32IF-NEXT:    and a1, a3, a1
+; RV32IF-NEXT:    and a4, a3, a0
+; RV32IF-NEXT:    srli a1, a1, 31
+; RV32IF-NEXT:    and a2, a3, a2
+; RV32IF-NEXT:    addi a1, a1, -1
+; RV32IF-NEXT:    and a0, a1, a2
+; RV32IF-NEXT:    and a1, a1, a4
 ; RV32IF-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s9, 36(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    .cfi_restore ra
-; RV32IF-NEXT:    .cfi_restore s0
-; RV32IF-NEXT:    .cfi_restore s1
-; RV32IF-NEXT:    .cfi_restore s2
-; RV32IF-NEXT:    .cfi_restore s3
-; RV32IF-NEXT:    .cfi_restore s4
-; RV32IF-NEXT:    .cfi_restore s5
-; RV32IF-NEXT:    .cfi_restore s6
-; RV32IF-NEXT:    .cfi_restore s7
-; RV32IF-NEXT:    .cfi_restore s8
-; RV32IF-NEXT:    .cfi_restore s9
 ; RV32IF-NEXT:    addi sp, sp, 80
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IF-NEXT:    ret
@@ -5500,248 +5612,258 @@ define i64 @ustest_f64i64_mm(double %x) {
 ; RV32IFD-NEXT:    addi sp, sp, -96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 96
 ; RV32IFD-NEXT:    sw ra, 92(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s0, 88(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s1, 84(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s2, 80(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s3, 76(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s4, 72(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s5, 68(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s6, 64(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s7, 60(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s8, 56(sp) # 4-byte Folded Spill
-; RV32IFD-NEXT:    sw s9, 52(sp) # 4-byte Folded Spill
 ; RV32IFD-NEXT:    .cfi_offset ra, -4
-; RV32IFD-NEXT:    .cfi_offset s0, -8
-; RV32IFD-NEXT:    .cfi_offset s1, -12
-; RV32IFD-NEXT:    .cfi_offset s2, -16
-; RV32IFD-NEXT:    .cfi_offset s3, -20
-; RV32IFD-NEXT:    .cfi_offset s4, -24
-; RV32IFD-NEXT:    .cfi_offset s5, -28
-; RV32IFD-NEXT:    .cfi_offset s6, -32
-; RV32IFD-NEXT:    .cfi_offset s7, -36
-; RV32IFD-NEXT:    .cfi_offset s8, -40
-; RV32IFD-NEXT:    .cfi_offset s9, -44
-; RV32IFD-NEXT:    li a2, 1023
-; RV32IFD-NEXT:    fsd fa0, 8(sp)
-; RV32IFD-NEXT:    lw a1, 12(sp)
-; RV32IFD-NEXT:    slli a0, a1, 1
+; RV32IFD-NEXT:    li a1, 1023
+; RV32IFD-NEXT:    fsd fa0, 48(sp)
+; RV32IFD-NEXT:    lw a2, 52(sp)
+; RV32IFD-NEXT:    slli a0, a2, 1
 ; RV32IFD-NEXT:    srli a0, a0, 21
-; RV32IFD-NEXT:    bgeu a0, a2, .LBB47_2
+; RV32IFD-NEXT:    bgeu a0, a1, .LBB47_3
 ; RV32IFD-NEXT:  # %bb.1:
-; RV32IFD-NEXT:    li s1, 0
-; RV32IFD-NEXT:    li s6, 0
 ; RV32IFD-NEXT:    li a2, 0
 ; RV32IFD-NEXT:    li a0, 0
-; RV32IFD-NEXT:    j .LBB47_6
-; RV32IFD-NEXT:  .LBB47_2: # %fp-to-i-if-check.exp.size
-; RV32IFD-NEXT:    lw a2, 8(sp)
-; RV32IFD-NEXT:    lui a3, 256
-; RV32IFD-NEXT:    srai s0, a1, 31
-; RV32IFD-NEXT:    addi a3, a3, -1
-; RV32IFD-NEXT:    and a4, a1, a3
-; RV32IFD-NEXT:    addi a1, a2, -1
-; RV32IFD-NEXT:    snez a2, a2
-; RV32IFD-NEXT:    addi a1, a1, 1
-; RV32IFD-NEXT:    add a2, a4, a2
-; RV32IFD-NEXT:    seqz a4, a1
-; RV32IFD-NEXT:    add a2, a2, a3
-; RV32IFD-NEXT:    li a3, 1074
-; RV32IFD-NEXT:    ori s2, s0, 1
-; RV32IFD-NEXT:    add a2, a2, a4
-; RV32IFD-NEXT:    bltu a3, a0, .LBB47_5
-; RV32IFD-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    li a3, 1043
-; RV32IFD-NEXT:    li a4, 1075
-; RV32IFD-NEXT:    sub a3, a3, a0
-; RV32IFD-NEXT:    sub a0, a4, a0
-; RV32IFD-NEXT:    srai a4, a3, 31
-; RV32IFD-NEXT:    srl a5, a2, a0
-; RV32IFD-NEXT:    and s3, a4, a5
-; RV32IFD-NEXT:    bltz a3, .LBB47_7
+; RV32IFD-NEXT:    li a3, 0
+; RV32IFD-NEXT:    li a1, 0
+; RV32IFD-NEXT:  .LBB47_2:
+; RV32IFD-NEXT:    seqz a4, a3
+; RV32IFD-NEXT:    j .LBB47_11
+; RV32IFD-NEXT:  .LBB47_3: # %fp-to-i-if-check.exp.size
+; RV32IFD-NEXT:    lw a3, 48(sp)
+; RV32IFD-NEXT:    lui a4, 256
+; RV32IFD-NEXT:    srai a1, a2, 31
+; RV32IFD-NEXT:    addi a4, a4, -1
+; RV32IFD-NEXT:    and a5, a2, a4
+; RV32IFD-NEXT:    addi a2, a3, -1
+; RV32IFD-NEXT:    snez a3, a3
+; RV32IFD-NEXT:    addi a2, a2, 1
+; RV32IFD-NEXT:    add a3, a5, a3
+; RV32IFD-NEXT:    seqz a5, a2
+; RV32IFD-NEXT:    add a3, a3, a4
+; RV32IFD-NEXT:    li a6, 1074
+; RV32IFD-NEXT:    ori a4, a1, 1
+; RV32IFD-NEXT:    add a3, a3, a5
+; RV32IFD-NEXT:    sw a1, 44(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a4, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    bltu a6, a0, .LBB47_6
 ; RV32IFD-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    srl s4, a2, a3
+; RV32IFD-NEXT:    li a5, 1043
+; RV32IFD-NEXT:    li a7, 1075
+; RV32IFD-NEXT:    sub a6, a5, a0
+; RV32IFD-NEXT:    sub a5, a7, a0
+; RV32IFD-NEXT:    srai a0, a6, 31
+; RV32IFD-NEXT:    srl a7, a3, a5
+; RV32IFD-NEXT:    and a0, a0, a7
+; RV32IFD-NEXT:    bltz a6, .LBB47_7
+; RV32IFD-NEXT:  # %bb.5: # %fp-to-i-if-exp.small
+; RV32IFD-NEXT:    srl a2, a3, a6
 ; RV32IFD-NEXT:    j .LBB47_8
-; RV32IFD-NEXT:  .LBB47_5: # %fp-to-i-if-exp.large
-; RV32IFD-NEXT:    addi a0, a0, -1075
-; RV32IFD-NEXT:    srli a3, a0, 3
-; RV32IFD-NEXT:    sw a1, 32(sp)
-; RV32IFD-NEXT:    sw a2, 36(sp)
-; RV32IFD-NEXT:    sw zero, 40(sp)
-; RV32IFD-NEXT:    sw zero, 44(sp)
-; RV32IFD-NEXT:    andi a3, a3, 12
-; RV32IFD-NEXT:    addi a1, sp, 32
-; RV32IFD-NEXT:    sw zero, 16(sp)
-; RV32IFD-NEXT:    sw zero, 20(sp)
-; RV32IFD-NEXT:    sw zero, 24(sp)
-; RV32IFD-NEXT:    sw zero, 28(sp)
-; RV32IFD-NEXT:    sub a1, a1, a3
-; RV32IFD-NEXT:    lw a4, 0(a1)
-; RV32IFD-NEXT:    lw a2, 4(a1)
-; RV32IFD-NEXT:    andi a3, a0, 31
-; RV32IFD-NEXT:    lw a5, 8(a1)
-; RV32IFD-NEXT:    lw a1, 12(a1)
-; RV32IFD-NEXT:    srli a6, a4, 1
-; RV32IFD-NEXT:    xori a3, a3, 31
-; RV32IFD-NEXT:    sll a7, a2, a0
-; RV32IFD-NEXT:    srl a6, a6, a3
-; RV32IFD-NEXT:    or s3, a7, a6
-; RV32IFD-NEXT:    srli a2, a2, 1
-; RV32IFD-NEXT:    sll a6, a5, a0
-; RV32IFD-NEXT:    srl a2, a2, a3
-; RV32IFD-NEXT:    or a2, a6, a2
-; RV32IFD-NEXT:    srli a5, a5, 1
-; RV32IFD-NEXT:    sll a1, a1, a0
-; RV32IFD-NEXT:    srl a3, a5, a3
-; RV32IFD-NEXT:    or a3, a1, a3
-; RV32IFD-NEXT:    sll s4, a4, a0
-; RV32IFD-NEXT:    mv a0, s2
-; RV32IFD-NEXT:    mv a1, s0
+; RV32IFD-NEXT:  .LBB47_6: # %fp-to-i-if-exp.large
+; RV32IFD-NEXT:    addi a5, a0, -1075
+; RV32IFD-NEXT:    srli a0, a5, 3
+; RV32IFD-NEXT:    sw a2, 72(sp)
+; RV32IFD-NEXT:    sw a3, 76(sp)
+; RV32IFD-NEXT:    sw zero, 80(sp)
+; RV32IFD-NEXT:    sw zero, 84(sp)
+; RV32IFD-NEXT:    andi a0, a0, 12
+; RV32IFD-NEXT:    addi a2, sp, 72
+; RV32IFD-NEXT:    sw zero, 56(sp)
+; RV32IFD-NEXT:    sw zero, 60(sp)
+; RV32IFD-NEXT:    sw zero, 64(sp)
+; RV32IFD-NEXT:    sw zero, 68(sp)
+; RV32IFD-NEXT:    sub a2, a2, a0
+; RV32IFD-NEXT:    lw a6, 0(a2)
+; RV32IFD-NEXT:    lw a3, 4(a2)
+; RV32IFD-NEXT:    andi a0, a5, 31
+; RV32IFD-NEXT:    lw a7, 8(a2)
+; RV32IFD-NEXT:    lw t0, 12(a2)
+; RV32IFD-NEXT:    srli a2, a6, 1
+; RV32IFD-NEXT:    xori t1, a0, 31
+; RV32IFD-NEXT:    sll a0, a3, a5
+; RV32IFD-NEXT:    srl a2, a2, t1
+; RV32IFD-NEXT:    or a0, a0, a2
+; RV32IFD-NEXT:    sw a0, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    srli a3, a3, 1
+; RV32IFD-NEXT:    sll a0, a7, a5
+; RV32IFD-NEXT:    srl a2, a3, t1
+; RV32IFD-NEXT:    or a2, a0, a2
+; RV32IFD-NEXT:    srli a0, a7, 1
+; RV32IFD-NEXT:    sll a3, t0, a5
+; RV32IFD-NEXT:    srl a0, a0, t1
+; RV32IFD-NEXT:    or a3, a3, a0
+; RV32IFD-NEXT:    sll a0, a6, a5
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    mv a0, a4
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s7, a0, s1
-; RV32IFD-NEXT:    add a1, a1, s5
-; RV32IFD-NEXT:    sltu a0, s7, a0
-; RV32IFD-NEXT:    add s8, a1, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a1, a2
+; RV32IFD-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    add a0, a2, a0
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s5, a0
-; RV32IFD-NEXT:    mv s6, a1
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    add s2, s5, a1
-; RV32IFD-NEXT:    sltu a0, s2, s5
-; RV32IFD-NEXT:    add s5, s6, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s6, a0, s2
-; RV32IFD-NEXT:    sltu a0, s6, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s2, s5, a0
-; RV32IFD-NEXT:    sltu s4, s2, s5
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s2, a0, s2
-; RV32IFD-NEXT:    add a1, a1, s4
-; RV32IFD-NEXT:    sltu a0, s2, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add a2, s2, s7
-; RV32IFD-NEXT:    add a0, a0, s8
-; RV32IFD-NEXT:    sltu a1, a2, s2
-; RV32IFD-NEXT:    add a0, a0, a1
-; RV32IFD-NEXT:    bnez a0, .LBB47_9
-; RV32IFD-NEXT:  .LBB47_6:
-; RV32IFD-NEXT:    seqz a1, a2
-; RV32IFD-NEXT:    j .LBB47_10
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a4, a0, a3
+; RV32IFD-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a1, a1, a3
+; RV32IFD-NEXT:    sltu a3, a4, a0
+; RV32IFD-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a1, a3
+; RV32IFD-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a3, a4, a3
+; RV32IFD-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a1, a5, a1
+; RV32IFD-NEXT:    sltu a4, a3, a4
+; RV32IFD-NEXT:    j .LBB47_9
 ; RV32IFD-NEXT:  .LBB47_7:
-; RV32IFD-NEXT:    not a3, a0
-; RV32IFD-NEXT:    slli a2, a2, 1
-; RV32IFD-NEXT:    srl a0, a1, a0
-; RV32IFD-NEXT:    sll a1, a2, a3
-; RV32IFD-NEXT:    or s4, a0, a1
+; RV32IFD-NEXT:    not a6, a5
+; RV32IFD-NEXT:    slli a3, a3, 1
+; RV32IFD-NEXT:    srl a2, a2, a5
+; RV32IFD-NEXT:    sll a3, a3, a6
+; RV32IFD-NEXT:    or a2, a2, a3
 ; RV32IFD-NEXT:  .LBB47_8: # %fp-to-i-if-exp.small
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a2, 36(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s1, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s2
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s5, a0, s5
-; RV32IFD-NEXT:    sltu a0, s5, a0
-; RV32IFD-NEXT:    add s7, a1, a0
-; RV32IFD-NEXT:    mv a0, s4
+; RV32IFD-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a0, a1, a0
+; RV32IFD-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    add s6, a0, s5
-; RV32IFD-NEXT:    sltu a0, s6, a0
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    add s8, s7, a0
-; RV32IFD-NEXT:    mv a0, s3
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a2, a0, a2
+; RV32IFD-NEXT:    sw a2, 40(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a2, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a0, a2, a0
+; RV32IFD-NEXT:    add a1, a1, a0
+; RV32IFD-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a1, 0
-; RV32IFD-NEXT:    mv a2, s0
+; RV32IFD-NEXT:    lw a2, 44(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    li a3, 0
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    mv s2, a0
-; RV32IFD-NEXT:    mv s5, a1
-; RV32IFD-NEXT:    add s9, a0, s8
-; RV32IFD-NEXT:    mv a0, s0
-; RV32IFD-NEXT:    mv a1, s0
-; RV32IFD-NEXT:    mv a2, s4
-; RV32IFD-NEXT:    mv a3, s3
+; RV32IFD-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a0, a0, a1
+; RV32IFD-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32IFD-NEXT:    lw a0, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a1, 44(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    call __muldi3
-; RV32IFD-NEXT:    sltu a2, s8, s7
-; RV32IFD-NEXT:    sltu a3, s9, s2
-; RV32IFD-NEXT:    add s5, s5, a2
-; RV32IFD-NEXT:    add a2, s9, a0
-; RV32IFD-NEXT:    add a3, s5, a3
-; RV32IFD-NEXT:    sltu a0, a2, s9
-; RV32IFD-NEXT:    add a1, a3, a1
-; RV32IFD-NEXT:    add a0, a1, a0
-; RV32IFD-NEXT:    beqz a0, .LBB47_6
+; RV32IFD-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a5, a4, a3
+; RV32IFD-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a4, a4, a3
+; RV32IFD-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a3, a5
+; RV32IFD-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a3, a3, a0
+; RV32IFD-NEXT:    lw a0, 40(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    add a5, a5, a4
+; RV32IFD-NEXT:    lw a4, 24(sp) # 4-byte Folded Reload
+; RV32IFD-NEXT:    sltu a4, a3, a4
+; RV32IFD-NEXT:    add a1, a5, a1
 ; RV32IFD-NEXT:  .LBB47_9: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    srli a1, a0, 31
-; RV32IFD-NEXT:  .LBB47_10: # %fp-to-i-cleanup
-; RV32IFD-NEXT:    xori a2, a2, 1
-; RV32IFD-NEXT:    or a2, a2, a0
-; RV32IFD-NEXT:    seqz a2, a2
-; RV32IFD-NEXT:    addi a2, a2, -1
-; RV32IFD-NEXT:    and a1, a2, a1
-; RV32IFD-NEXT:    neg a1, a1
-; RV32IFD-NEXT:    and a0, a1, a0
-; RV32IFD-NEXT:    and a2, a1, s6
-; RV32IFD-NEXT:    srli a0, a0, 31
-; RV32IFD-NEXT:    and a1, a1, s1
-; RV32IFD-NEXT:    addi a3, a0, -1
-; RV32IFD-NEXT:    and a0, a3, a1
-; RV32IFD-NEXT:    and a1, a3, a2
+; RV32IFD-NEXT:    add a1, a1, a4
+; RV32IFD-NEXT:    beqz a1, .LBB47_2
+; RV32IFD-NEXT:  # %bb.10: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    srli a4, a1, 31
+; RV32IFD-NEXT:  .LBB47_11: # %fp-to-i-cleanup
+; RV32IFD-NEXT:    xori a3, a3, 1
+; RV32IFD-NEXT:    or a3, a3, a1
+; RV32IFD-NEXT:    seqz a3, a3
+; RV32IFD-NEXT:    addi a3, a3, -1
+; RV32IFD-NEXT:    and a3, a3, a4
+; RV32IFD-NEXT:    neg a3, a3
+; RV32IFD-NEXT:    and a1, a3, a1
+; RV32IFD-NEXT:    and a4, a3, a0
+; RV32IFD-NEXT:    srli a1, a1, 31
+; RV32IFD-NEXT:    and a2, a3, a2
+; RV32IFD-NEXT:    addi a1, a1, -1
+; RV32IFD-NEXT:    and a0, a1, a2
+; RV32IFD-NEXT:    and a1, a1, a4
 ; RV32IFD-NEXT:    lw ra, 92(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s0, 88(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s1, 84(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s2, 80(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s3, 76(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s4, 72(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s5, 68(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s6, 64(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s7, 60(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s8, 56(sp) # 4-byte Folded Reload
-; RV32IFD-NEXT:    lw s9, 52(sp) # 4-byte Folded Reload
 ; RV32IFD-NEXT:    .cfi_restore ra
-; RV32IFD-NEXT:    .cfi_restore s0
-; RV32IFD-NEXT:    .cfi_restore s1
-; RV32IFD-NEXT:    .cfi_restore s2
-; RV32IFD-NEXT:    .cfi_restore s3
-; RV32IFD-NEXT:    .cfi_restore s4
-; RV32IFD-NEXT:    .cfi_restore s5
-; RV32IFD-NEXT:    .cfi_restore s6
-; RV32IFD-NEXT:    .cfi_restore s7
-; RV32IFD-NEXT:    .cfi_restore s8
-; RV32IFD-NEXT:    .cfi_restore s9
 ; RV32IFD-NEXT:    addi sp, sp, 96
 ; RV32IFD-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IFD-NEXT:    ret
@@ -5759,247 +5881,263 @@ define i64 @stest_f32i64_mm(float %x) {
 ; RV32-NEXT:    addi sp, sp, -80
 ; RV32-NEXT:    .cfi_def_cfa_offset 80
 ; RV32-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
-; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    .cfi_offset s1, -12
-; RV32-NEXT:    .cfi_offset s2, -16
-; RV32-NEXT:    .cfi_offset s3, -20
-; RV32-NEXT:    .cfi_offset s4, -24
-; RV32-NEXT:    .cfi_offset s5, -28
-; RV32-NEXT:    .cfi_offset s6, -32
-; RV32-NEXT:    .cfi_offset s7, -36
-; RV32-NEXT:    .cfi_offset s8, -40
 ; RV32-NEXT:    li a1, 127
-; RV32-NEXT:    fmv.x.w s0, fa0
-; RV32-NEXT:    slli a0, s0, 1
+; RV32-NEXT:    fmv.x.w a4, fa0
+; RV32-NEXT:    slli a0, a4, 1
 ; RV32-NEXT:    srli a0, a0, 24
 ; RV32-NEXT:    bgeu a0, a1, .LBB48_2
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    li s1, 0
-; RV32-NEXT:    li s2, 0
-; RV32-NEXT:    li a0, 0
+; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    j .LBB48_5
+; RV32-NEXT:    li a0, 0
+; RV32-NEXT:    li a2, 0
+; RV32-NEXT:    j .LBB48_6
 ; RV32-NEXT:  .LBB48_2: # %fp-to-i-if-check.exp.size
-; RV32-NEXT:    slli a1, s0, 9
-; RV32-NEXT:    srai s0, s0, 31
+; RV32-NEXT:    slli a1, a4, 9
+; RV32-NEXT:    srai a4, a4, 31
 ; RV32-NEXT:    srli a1, a1, 9
 ; RV32-NEXT:    lui a2, 2048
 ; RV32-NEXT:    li a3, 149
 ; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    ori s2, s0, 1
+; RV32-NEXT:    ori a5, a4, 1
+; RV32-NEXT:    sw a4, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a5, 32(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    bltu a3, a0, .LBB48_4
 ; RV32-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
 ; RV32-NEXT:    li a2, 150
 ; RV32-NEXT:    sub a2, a2, a0
-; RV32-NEXT:    srl s3, a1, a2
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    srl a0, a1, a2
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s4, a1
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s4
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add s6, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s7, s6, a0
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s4, a0
-; RV32-NEXT:    mv s5, a1
-; RV32-NEXT:    add s8, a0, s7
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s3
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sltu a2, s7, s6
-; RV32-NEXT:    sltu a3, s8, s4
-; RV32-NEXT:    add a2, s5, a2
-; RV32-NEXT:    add a0, s8, a0
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    sltu a3, a0, s8
-; RV32-NEXT:    add a1, a2, a1
-; RV32-NEXT:    add a1, a1, a3
+; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv a2, a1
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a6, a5, a4
+; RV32-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a5, a5, a4
+; RV32-NEXT:    lw a4, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a6, a4, a6
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a4, a0
+; RV32-NEXT:    add a5, a6, a5
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, a0, a4
+; RV32-NEXT:    add a2, a5, a2
 ; RV32-NEXT:    j .LBB48_5
 ; RV32-NEXT:  .LBB48_4: # %fp-to-i-if-exp.large
-; RV32-NEXT:    addi a0, a0, -150
-; RV32-NEXT:    srli a2, a0, 3
-; RV32-NEXT:    sw a1, 24(sp)
-; RV32-NEXT:    sw zero, 28(sp)
-; RV32-NEXT:    sw zero, 32(sp)
-; RV32-NEXT:    sw zero, 36(sp)
-; RV32-NEXT:    andi a2, a2, 12
-; RV32-NEXT:    addi a1, sp, 24
-; RV32-NEXT:    sw zero, 8(sp)
-; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    sw zero, 16(sp)
-; RV32-NEXT:    sw zero, 20(sp)
-; RV32-NEXT:    sub a1, a1, a2
-; RV32-NEXT:    lw a4, 0(a1)
+; RV32-NEXT:    addi a6, a0, -150
+; RV32-NEXT:    srli a0, a6, 3
+; RV32-NEXT:    sw a1, 56(sp)
+; RV32-NEXT:    sw zero, 60(sp)
+; RV32-NEXT:    sw zero, 64(sp)
+; RV32-NEXT:    sw zero, 68(sp)
+; RV32-NEXT:    andi a0, a0, 12
+; RV32-NEXT:    addi a1, sp, 56
+; RV32-NEXT:    sw zero, 40(sp)
+; RV32-NEXT:    sw zero, 44(sp)
+; RV32-NEXT:    sw zero, 48(sp)
+; RV32-NEXT:    sw zero, 52(sp)
+; RV32-NEXT:    sub a1, a1, a0
+; RV32-NEXT:    lw a7, 0(a1)
 ; RV32-NEXT:    lw a2, 4(a1)
-; RV32-NEXT:    andi a3, a0, 31
-; RV32-NEXT:    lw a5, 8(a1)
+; RV32-NEXT:    andi a0, a6, 31
+; RV32-NEXT:    lw a3, 8(a1)
 ; RV32-NEXT:    lw a1, 12(a1)
-; RV32-NEXT:    srli a6, a4, 1
-; RV32-NEXT:    xori a3, a3, 31
-; RV32-NEXT:    sll a7, a2, a0
-; RV32-NEXT:    srl a6, a6, a3
-; RV32-NEXT:    or s3, a7, a6
+; RV32-NEXT:    srli t0, a7, 1
+; RV32-NEXT:    xori t1, a0, 31
+; RV32-NEXT:    sll a0, a2, a6
+; RV32-NEXT:    srl t0, t0, t1
+; RV32-NEXT:    or a0, a0, t0
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    srli a2, a2, 1
-; RV32-NEXT:    sll a6, a5, a0
-; RV32-NEXT:    srl a2, a2, a3
-; RV32-NEXT:    or a2, a6, a2
-; RV32-NEXT:    srli a5, a5, 1
-; RV32-NEXT:    sll a1, a1, a0
-; RV32-NEXT:    srl a3, a5, a3
-; RV32-NEXT:    or a3, a1, a3
-; RV32-NEXT:    sll s4, a4, a0
-; RV32-NEXT:    mv a0, s2
-; RV32-NEXT:    mv a1, s0
+; RV32-NEXT:    sll a0, a3, a6
+; RV32-NEXT:    srl a2, a2, t1
+; RV32-NEXT:    or a2, a0, a2
+; RV32-NEXT:    srli a3, a3, 1
+; RV32-NEXT:    sll a0, a1, a6
+; RV32-NEXT:    srl a3, a3, t1
+; RV32-NEXT:    or a3, a0, a3
+; RV32-NEXT:    sll a0, a7, a6
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, a5
+; RV32-NEXT:    mv a1, a4
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s5, a1
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s4
-; RV32-NEXT:    mv a3, s3
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s7, a0, s1
-; RV32-NEXT:    add a1, a1, s5
-; RV32-NEXT:    sltu a0, s7, a0
-; RV32-NEXT:    add s8, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a1, a2
+; RV32-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    add a0, a2, a0
+; RV32-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s5, a0
-; RV32-NEXT:    mv s6, a1
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    add s2, s5, a1
-; RV32-NEXT:    sltu a0, s2, s5
-; RV32-NEXT:    add s6, s6, a0
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a1, a0
+; RV32-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s4, s6, a0
-; RV32-NEXT:    sltu s5, s4, s6
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    add a1, a1, s5
-; RV32-NEXT:    sltu a0, s4, a0
-; RV32-NEXT:    add a1, a1, a0
-; RV32-NEXT:    add a0, s4, s7
-; RV32-NEXT:    add a1, a1, s8
-; RV32-NEXT:    sltu a2, a0, s4
-; RV32-NEXT:    add a1, a1, a2
+; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    mv a2, a1
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a0, a4
+; RV32-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a2, a4
+; RV32-NEXT:    sltu a0, a5, a0
+; RV32-NEXT:    add a4, a2, a0
+; RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a5, a0
+; RV32-NEXT:    lw a2, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a4, a2
+; RV32-NEXT:    sltu a4, a0, a5
 ; RV32-NEXT:  .LBB48_5: # %fp-to-i-cleanup
-; RV32-NEXT:    lui a2, 524288
-; RV32-NEXT:    addi a3, a2, -1
-; RV32-NEXT:    beq s2, a3, .LBB48_7
-; RV32-NEXT:  # %bb.6: # %fp-to-i-cleanup
-; RV32-NEXT:    sltu a4, s2, a3
-; RV32-NEXT:    or a5, a0, a1
-; RV32-NEXT:    bnez a5, .LBB48_8
-; RV32-NEXT:    j .LBB48_9
-; RV32-NEXT:  .LBB48_7:
-; RV32-NEXT:    sltiu a4, s1, -1
-; RV32-NEXT:    or a5, a0, a1
-; RV32-NEXT:    beqz a5, .LBB48_9
-; RV32-NEXT:  .LBB48_8: # %fp-to-i-cleanup
-; RV32-NEXT:    srli a4, a1, 31
+; RV32-NEXT:    add a2, a2, a4
+; RV32-NEXT:  .LBB48_6: # %fp-to-i-cleanup
+; RV32-NEXT:    lui a4, 524288
+; RV32-NEXT:    addi a5, a4, -1
+; RV32-NEXT:    beq a1, a5, .LBB48_8
+; RV32-NEXT:  # %bb.7: # %fp-to-i-cleanup
+; RV32-NEXT:    sltu a6, a1, a5
+; RV32-NEXT:    or a7, a0, a2
+; RV32-NEXT:    bnez a7, .LBB48_9
+; RV32-NEXT:    j .LBB48_10
+; RV32-NEXT:  .LBB48_8:
+; RV32-NEXT:    sltiu a6, a3, -1
+; RV32-NEXT:    or a7, a0, a2
+; RV32-NEXT:    beqz a7, .LBB48_10
 ; RV32-NEXT:  .LBB48_9: # %fp-to-i-cleanup
-; RV32-NEXT:    neg a5, a4
-; RV32-NEXT:    addi a6, a4, -1
-; RV32-NEXT:    bnez a4, .LBB48_11
-; RV32-NEXT:  # %bb.10: # %fp-to-i-cleanup
-; RV32-NEXT:    mv s2, a3
-; RV32-NEXT:  .LBB48_11: # %fp-to-i-cleanup
-; RV32-NEXT:    or a3, a6, s1
-; RV32-NEXT:    and a1, a5, a1
-; RV32-NEXT:    and a4, a5, a0
-; RV32-NEXT:    beq s2, a2, .LBB48_13
-; RV32-NEXT:  # %bb.12: # %fp-to-i-cleanup
-; RV32-NEXT:    sltu a0, a2, s2
-; RV32-NEXT:    j .LBB48_14
-; RV32-NEXT:  .LBB48_13:
+; RV32-NEXT:    srli a6, a2, 31
+; RV32-NEXT:  .LBB48_10: # %fp-to-i-cleanup
+; RV32-NEXT:    neg a7, a6
+; RV32-NEXT:    addi t0, a6, -1
+; RV32-NEXT:    bnez a6, .LBB48_12
+; RV32-NEXT:  # %bb.11: # %fp-to-i-cleanup
+; RV32-NEXT:    mv a1, a5
+; RV32-NEXT:  .LBB48_12: # %fp-to-i-cleanup
+; RV32-NEXT:    or a3, t0, a3
+; RV32-NEXT:    and a2, a7, a2
+; RV32-NEXT:    and a5, a7, a0
+; RV32-NEXT:    beq a1, a4, .LBB48_14
+; RV32-NEXT:  # %bb.13: # %fp-to-i-cleanup
+; RV32-NEXT:    sltu a0, a4, a1
+; RV32-NEXT:    j .LBB48_15
+; RV32-NEXT:  .LBB48_14:
 ; RV32-NEXT:    snez a0, a3
-; RV32-NEXT:  .LBB48_14: # %fp-to-i-cleanup
-; RV32-NEXT:    and a4, a4, a1
-; RV32-NEXT:    li a2, -1
-; RV32-NEXT:    beq a4, a2, .LBB48_16
-; RV32-NEXT:  # %bb.15: # %fp-to-i-cleanup
-; RV32-NEXT:    srli a1, a1, 31
-; RV32-NEXT:    xori a0, a1, 1
-; RV32-NEXT:  .LBB48_16: # %fp-to-i-cleanup
-; RV32-NEXT:    bnez a0, .LBB48_18
-; RV32-NEXT:  # %bb.17: # %fp-to-i-cleanup
-; RV32-NEXT:    lui s2, 524288
-; RV32-NEXT:  .LBB48_18: # %fp-to-i-cleanup
+; RV32-NEXT:  .LBB48_15: # %fp-to-i-cleanup
+; RV32-NEXT:    and a5, a5, a2
+; RV32-NEXT:    li a4, -1
+; RV32-NEXT:    beq a5, a4, .LBB48_17
+; RV32-NEXT:  # %bb.16: # %fp-to-i-cleanup
+; RV32-NEXT:    srli a2, a2, 31
+; RV32-NEXT:    xori a0, a2, 1
+; RV32-NEXT:  .LBB48_17: # %fp-to-i-cleanup
+; RV32-NEXT:    bnez a0, .LBB48_19
+; RV32-NEXT:  # %bb.18: # %fp-to-i-cleanup
+; RV32-NEXT:    lui a1, 524288
+; RV32-NEXT:  .LBB48_19: # %fp-to-i-cleanup
 ; RV32-NEXT:    neg a0, a0
 ; RV32-NEXT:    and a0, a0, a3
-; RV32-NEXT:    mv a1, s2
 ; RV32-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
-; RV32-NEXT:    .cfi_restore s0
-; RV32-NEXT:    .cfi_restore s1
-; RV32-NEXT:    .cfi_restore s2
-; RV32-NEXT:    .cfi_restore s3
-; RV32-NEXT:    .cfi_restore s4
-; RV32-NEXT:    .cfi_restore s5
-; RV32-NEXT:    .cfi_restore s6
-; RV32-NEXT:    .cfi_restore s7
-; RV32-NEXT:    .cfi_restore s8
 ; RV32-NEXT:    addi sp, sp, 80
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
 ; RV32-NEXT:    ret
@@ -6127,221 +6265,235 @@ define i64 @ustest_f32i64_mm(float %x) {
 ; RV32-NEXT:    addi sp, sp, -80
 ; RV32-NEXT:    .cfi_def_cfa_offset 80
 ; RV32-NEXT:    sw ra, 76(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 72(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 68(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 64(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 60(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s4, 56(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s5, 52(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 48(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 44(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s8, 40(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
-; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    .cfi_offset s1, -12
-; RV32-NEXT:    .cfi_offset s2, -16
-; RV32-NEXT:    .cfi_offset s3, -20
-; RV32-NEXT:    .cfi_offset s4, -24
-; RV32-NEXT:    .cfi_offset s5, -28
-; RV32-NEXT:    .cfi_offset s6, -32
-; RV32-NEXT:    .cfi_offset s7, -36
-; RV32-NEXT:    .cfi_offset s8, -40
 ; RV32-NEXT:    li a1, 127
-; RV32-NEXT:    fmv.x.w s0, fa0
-; RV32-NEXT:    slli a0, s0, 1
+; RV32-NEXT:    fmv.x.w a4, fa0
+; RV32-NEXT:    slli a0, a4, 1
 ; RV32-NEXT:    srli a0, a0, 24
-; RV32-NEXT:    bgeu a0, a1, .LBB50_2
+; RV32-NEXT:    bgeu a0, a1, .LBB50_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    li s1, 0
-; RV32-NEXT:    li s5, 0
 ; RV32-NEXT:    li a2, 0
 ; RV32-NEXT:    li a0, 0
-; RV32-NEXT:    j .LBB50_4
-; RV32-NEXT:  .LBB50_2: # %fp-to-i-if-check.exp.size
-; RV32-NEXT:    slli a1, s0, 9
-; RV32-NEXT:    srai s0, s0, 31
+; RV32-NEXT:    li a3, 0
+; RV32-NEXT:    li a1, 0
+; RV32-NEXT:  .LBB50_2:
+; RV32-NEXT:    seqz a4, a3
+; RV32-NEXT:    j .LBB50_8
+; RV32-NEXT:  .LBB50_3: # %fp-to-i-if-check.exp.size
+; RV32-NEXT:    slli a1, a4, 9
+; RV32-NEXT:    srai a4, a4, 31
 ; RV32-NEXT:    srli a1, a1, 9
 ; RV32-NEXT:    lui a2, 2048
 ; RV32-NEXT:    li a3, 149
 ; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    ori s2, s0, 1
+; RV32-NEXT:    ori a5, a4, 1
+; RV32-NEXT:    sw a4, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a5, 32(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    bltu a3, a0, .LBB50_5
-; RV32-NEXT:  # %bb.3: # %fp-to-i-if-exp.small
+; RV32-NEXT:  # %bb.4: # %fp-to-i-if-exp.small
 ; RV32-NEXT:    li a2, 150
 ; RV32-NEXT:    sub a2, a2, a0
-; RV32-NEXT:    srl s3, a1, a2
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    srl a0, a1, a2
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s4, a1
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s4, a0, s4
-; RV32-NEXT:    sltu a0, s4, a0
-; RV32-NEXT:    add s6, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s5, a0, s4
-; RV32-NEXT:    sltu a0, s5, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s7, s6, a0
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s2, a0
-; RV32-NEXT:    mv s4, a1
-; RV32-NEXT:    add s8, a0, s7
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s3
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    sltu a2, s7, s6
-; RV32-NEXT:    sltu a3, s8, s2
-; RV32-NEXT:    add s4, s4, a2
-; RV32-NEXT:    add a2, s8, a0
-; RV32-NEXT:    add a3, s4, a3
-; RV32-NEXT:    sltu a0, a2, s8
-; RV32-NEXT:    add a1, a3, a1
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    bnez a0, .LBB50_6
-; RV32-NEXT:  .LBB50_4:
-; RV32-NEXT:    seqz a1, a2
-; RV32-NEXT:    j .LBB50_7
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a5, a4, a3
+; RV32-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, a4, a3
+; RV32-NEXT:    lw a3, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a3, a5
+; RV32-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a3, a0
+; RV32-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a5, a4
+; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a4, a3, a4
+; RV32-NEXT:    add a1, a5, a1
+; RV32-NEXT:    j .LBB50_6
 ; RV32-NEXT:  .LBB50_5: # %fp-to-i-if-exp.large
-; RV32-NEXT:    addi a0, a0, -150
-; RV32-NEXT:    srli a2, a0, 3
-; RV32-NEXT:    sw a1, 24(sp)
-; RV32-NEXT:    sw zero, 28(sp)
-; RV32-NEXT:    sw zero, 32(sp)
-; RV32-NEXT:    sw zero, 36(sp)
-; RV32-NEXT:    andi a2, a2, 12
-; RV32-NEXT:    addi a1, sp, 24
-; RV32-NEXT:    sw zero, 8(sp)
-; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    sw zero, 16(sp)
-; RV32-NEXT:    sw zero, 20(sp)
-; RV32-NEXT:    sub a1, a1, a2
-; RV32-NEXT:    lw a4, 0(a1)
+; RV32-NEXT:    addi a6, a0, -150
+; RV32-NEXT:    srli a0, a6, 3
+; RV32-NEXT:    sw a1, 56(sp)
+; RV32-NEXT:    sw zero, 60(sp)
+; RV32-NEXT:    sw zero, 64(sp)
+; RV32-NEXT:    sw zero, 68(sp)
+; RV32-NEXT:    andi a0, a0, 12
+; RV32-NEXT:    addi a1, sp, 56
+; RV32-NEXT:    sw zero, 40(sp)
+; RV32-NEXT:    sw zero, 44(sp)
+; RV32-NEXT:    sw zero, 48(sp)
+; RV32-NEXT:    sw zero, 52(sp)
+; RV32-NEXT:    sub a1, a1, a0
+; RV32-NEXT:    lw a7, 0(a1)
 ; RV32-NEXT:    lw a2, 4(a1)
-; RV32-NEXT:    andi a3, a0, 31
-; RV32-NEXT:    lw a5, 8(a1)
+; RV32-NEXT:    andi a0, a6, 31
+; RV32-NEXT:    lw a3, 8(a1)
 ; RV32-NEXT:    lw a1, 12(a1)
-; RV32-NEXT:    srli a6, a4, 1
-; RV32-NEXT:    xori a3, a3, 31
-; RV32-NEXT:    sll a7, a2, a0
-; RV32-NEXT:    srl a6, a6, a3
-; RV32-NEXT:    or s3, a7, a6
+; RV32-NEXT:    srli t0, a7, 1
+; RV32-NEXT:    xori t1, a0, 31
+; RV32-NEXT:    sll a0, a2, a6
+; RV32-NEXT:    srl t0, t0, t1
+; RV32-NEXT:    or a0, a0, t0
+; RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    srli a2, a2, 1
-; RV32-NEXT:    sll a6, a5, a0
-; RV32-NEXT:    srl a2, a2, a3
-; RV32-NEXT:    or a2, a6, a2
-; RV32-NEXT:    srli a5, a5, 1
-; RV32-NEXT:    sll a1, a1, a0
-; RV32-NEXT:    srl a3, a5, a3
-; RV32-NEXT:    or a3, a1, a3
-; RV32-NEXT:    sll s4, a4, a0
-; RV32-NEXT:    mv a0, s2
-; RV32-NEXT:    mv a1, s0
+; RV32-NEXT:    sll a0, a3, a6
+; RV32-NEXT:    srl a2, a2, t1
+; RV32-NEXT:    or a2, a0, a2
+; RV32-NEXT:    srli a3, a3, 1
+; RV32-NEXT:    sll a0, a1, a6
+; RV32-NEXT:    srl a3, a3, t1
+; RV32-NEXT:    or a3, a0, a3
+; RV32-NEXT:    sll a0, a7, a6
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv a0, a5
+; RV32-NEXT:    mv a1, a4
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    mv s5, a1
-; RV32-NEXT:    mv a0, s0
-; RV32-NEXT:    mv a1, s0
-; RV32-NEXT:    mv a2, s4
-; RV32-NEXT:    mv a3, s3
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s7, a0, s1
-; RV32-NEXT:    add a1, a1, s5
-; RV32-NEXT:    sltu a0, s7, a0
-; RV32-NEXT:    add s8, a1, a0
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a1, a2
+; RV32-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    add a0, a2, a0
+; RV32-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s5, a0
-; RV32-NEXT:    mv s6, a1
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s2
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    mv s1, a0
-; RV32-NEXT:    add s2, s5, a1
-; RV32-NEXT:    sltu a0, s2, s5
-; RV32-NEXT:    add s6, s6, a0
-; RV32-NEXT:    mv a0, s4
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a1, a0
+; RV32-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s5, a0, s2
-; RV32-NEXT:    sltu a0, s5, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add s2, s6, a0
-; RV32-NEXT:    sltu s4, s2, s6
-; RV32-NEXT:    mv a0, s3
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a2, a0, a2
+; RV32-NEXT:    sw a2, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    add a1, a1, a0
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a1
+; RV32-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a0, a1, a0
+; RV32-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    mv a2, s0
+; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __muldi3
-; RV32-NEXT:    add s2, a0, s2
-; RV32-NEXT:    add a1, a1, s4
-; RV32-NEXT:    sltu a0, s2, a0
-; RV32-NEXT:    add a0, a1, a0
-; RV32-NEXT:    add a2, s2, s7
-; RV32-NEXT:    add a0, a0, s8
-; RV32-NEXT:    sltu a1, a2, s2
-; RV32-NEXT:    add a0, a0, a1
-; RV32-NEXT:    beqz a0, .LBB50_4
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a4, a0, a3
+; RV32-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a1, a3
+; RV32-NEXT:    sltu a3, a4, a0
+; RV32-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a5, a1, a3
+; RV32-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a3, a4, a3
+; RV32-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a1, a5, a1
+; RV32-NEXT:    sltu a4, a3, a4
 ; RV32-NEXT:  .LBB50_6: # %fp-to-i-cleanup
-; RV32-NEXT:    srli a1, a0, 31
-; RV32-NEXT:  .LBB50_7: # %fp-to-i-cleanup
-; RV32-NEXT:    xori a2, a2, 1
-; RV32-NEXT:    or a2, a2, a0
-; RV32-NEXT:    seqz a2, a2
-; RV32-NEXT:    addi a2, a2, -1
-; RV32-NEXT:    and a1, a2, a1
-; RV32-NEXT:    neg a1, a1
-; RV32-NEXT:    and a0, a1, a0
-; RV32-NEXT:    and a2, a1, s5
-; RV32-NEXT:    srli a0, a0, 31
-; RV32-NEXT:    and a1, a1, s1
-; RV32-NEXT:    addi a3, a0, -1
-; RV32-NEXT:    and a0, a3, a1
-; RV32-NEXT:    and a1, a3, a2
+; RV32-NEXT:    add a1, a1, a4
+; RV32-NEXT:    beqz a1, .LBB50_2
+; RV32-NEXT:  # %bb.7: # %fp-to-i-cleanup
+; RV32-NEXT:    srli a4, a1, 31
+; RV32-NEXT:  .LBB50_8: # %fp-to-i-cleanup
+; RV32-NEXT:    xori a3, a3, 1
+; RV32-NEXT:    or a3, a3, a1
+; RV32-NEXT:    seqz a3, a3
+; RV32-NEXT:    addi a3, a3, -1
+; RV32-NEXT:    and a3, a3, a4
+; RV32-NEXT:    neg a3, a3
+; RV32-NEXT:    and a1, a3, a1
+; RV32-NEXT:    and a4, a3, a0
+; RV32-NEXT:    srli a1, a1, 31
+; RV32-NEXT:    and a2, a3, a2
+; RV32-NEXT:    addi a1, a1, -1
+; RV32-NEXT:    and a0, a1, a2
+; RV32-NEXT:    and a1, a1, a4
 ; RV32-NEXT:    lw ra, 76(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 72(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 68(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 64(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 60(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s4, 56(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s5, 52(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 48(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s7, 44(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s8, 40(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
-; RV32-NEXT:    .cfi_restore s0
-; RV32-NEXT:    .cfi_restore s1
-; RV32-NEXT:    .cfi_restore s2
-; RV32-NEXT:    .cfi_restore s3
-; RV32-NEXT:    .cfi_restore s4
-; RV32-NEXT:    .cfi_restore s5
-; RV32-NEXT:    .cfi_restore s6
-; RV32-NEXT:    .cfi_restore s7
-; RV32-NEXT:    .cfi_restore s8
 ; RV32-NEXT:    addi sp, sp, 80
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
 ; RV32-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/fpenv.ll b/llvm/test/CodeGen/RISCV/fpenv.ll
index cb7a94e9afeda4..40faa29257cf5e 100644
--- a/llvm/test/CodeGen/RISCV/fpenv.ll
+++ b/llvm/test/CodeGen/RISCV/fpenv.ll
@@ -32,34 +32,33 @@ define i1 @test_get_rounding_sideeffect() #0 {
 ; RV32IF-NEXT:    addi sp, sp, -16
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 16
 ; RV32IF-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32IF-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32IF-NEXT:    .cfi_offset ra, -4
-; RV32IF-NEXT:    .cfi_offset s0, -8
 ; RV32IF-NEXT:    li a0, 1
 ; RV32IF-NEXT:    call fesetround
 ; RV32IF-NEXT:    frrm a0
 ; RV32IF-NEXT:    lui a1, 66
-; RV32IF-NEXT:    slli a0, a0, 2
-; RV32IF-NEXT:    addi s0, a1, 769
-; RV32IF-NEXT:    srl a0, s0, a0
-; RV32IF-NEXT:    andi a0, a0, 7
-; RV32IF-NEXT:    beqz a0, .LBB1_2
+; RV32IF-NEXT:    slli a2, a0, 2
+; RV32IF-NEXT:    addi a0, a1, 769
+; RV32IF-NEXT:    srl a1, a0, a2
+; RV32IF-NEXT:    andi a1, a1, 7
+; RV32IF-NEXT:    beqz a1, .LBB1_2
 ; RV32IF-NEXT:  # %bb.1:
 ; RV32IF-NEXT:    li a0, 0
 ; RV32IF-NEXT:    j .LBB1_3
 ; RV32IF-NEXT:  .LBB1_2: # %if.end
+; RV32IF-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32IF-NEXT:    li a0, 0
 ; RV32IF-NEXT:    call fesetround
 ; RV32IF-NEXT:    frrm a0
-; RV32IF-NEXT:    slli a0, a0, 2
-; RV32IF-NEXT:    srl a0, s0, a0
+; RV32IF-NEXT:    slli a1, a0, 2
+; RV32IF-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32IF-NEXT:    srl a0, a0, a1
 ; RV32IF-NEXT:    andi a0, a0, 7
 ; RV32IF-NEXT:    addi a0, a0, -1
 ; RV32IF-NEXT:    seqz a0, a0
 ; RV32IF-NEXT:  .LBB1_3: # %return
 ; RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32IF-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IF-NEXT:    .cfi_restore ra
-; RV32IF-NEXT:    .cfi_restore s0
 ; RV32IF-NEXT:    addi sp, sp, 16
 ; RV32IF-NEXT:    .cfi_def_cfa_offset 0
 ; RV32IF-NEXT:    ret
@@ -69,34 +68,33 @@ define i1 @test_get_rounding_sideeffect() #0 {
 ; RV64IF-NEXT:    addi sp, sp, -16
 ; RV64IF-NEXT:    .cfi_def_cfa_offset 16
 ; RV64IF-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64IF-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
 ; RV64IF-NEXT:    .cfi_offset ra, -8
-; RV64IF-NEXT:    .cfi_offset s0, -16
 ; RV64IF-NEXT:    li a0, 1
 ; RV64IF-NEXT:    call fesetround
 ; RV64IF-NEXT:    frrm a0
 ; RV64IF-NEXT:    lui a1, 66
-; RV64IF-NEXT:    slli a0, a0, 2
-; RV64IF-NEXT:    addi s0, a1, 769
-; RV64IF-NEXT:    srl a0, s0, a0
-; RV64IF-NEXT:    andi a0, a0, 7
-; RV64IF-NEXT:    beqz a0, .LBB1_2
+; RV64IF-NEXT:    slli a2, a0, 2
+; RV64IF-NEXT:    addi a0, a1, 769
+; RV64IF-NEXT:    srl a1, a0, a2
+; RV64IF-NEXT:    andi a1, a1, 7
+; RV64IF-NEXT:    beqz a1, .LBB1_2
 ; RV64IF-NEXT:  # %bb.1:
 ; RV64IF-NEXT:    li a0, 0
 ; RV64IF-NEXT:    j .LBB1_3
 ; RV64IF-NEXT:  .LBB1_2: # %if.end
+; RV64IF-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
+; RV64IF-NEXT:    li a0, 0
 ; RV64IF-NEXT:    call fesetround
 ; RV64IF-NEXT:    frrm a0
-; RV64IF-NEXT:    slli a0, a0, 2
-; RV64IF-NEXT:    srl a0, s0, a0
+; RV64IF-NEXT:    slli a1, a0, 2
+; RV64IF-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64IF-NEXT:    srl a0, a0, a1
 ; RV64IF-NEXT:    andi a0, a0, 7
 ; RV64IF-NEXT:    addi a0, a0, -1
 ; RV64IF-NEXT:    seqz a0, a0
 ; RV64IF-NEXT:  .LBB1_3: # %return
 ; RV64IF-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64IF-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64IF-NEXT:    .cfi_restore ra
-; RV64IF-NEXT:    .cfi_restore s0
 ; RV64IF-NEXT:    addi sp, sp, 16
 ; RV64IF-NEXT:    .cfi_def_cfa_offset 0
 ; RV64IF-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/half-convert.ll b/llvm/test/CodeGen/RISCV/half-convert.ll
index 22d5e58886e677..1d8391dde4c3a9 100644
--- a/llvm/test/CodeGen/RISCV/half-convert.ll
+++ b/llvm/test/CodeGen/RISCV/half-convert.ll
@@ -811,36 +811,35 @@ define i16 @fcvt_ui_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -32
 ; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    lui s0, 16
-; RV32I-NEXT:    addi s0, s0, -1
-; RV32I-NEXT:    and a0, a0, s0
+; RV32I-NEXT:    lui a1, 16
+; RV32I-NEXT:    addi a1, a1, -1
+; RV32I-NEXT:    sw a1, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    and a0, a0, a1
 ; RV32I-NEXT:    call __extendhfsf2
-; RV32I-NEXT:    mv s3, a0
+; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    call __fixunssfsi
-; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    mv a0, s3
+; RV32I-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 292864
 ; RV32I-NEXT:    addi a1, a1, -256
-; RV32I-NEXT:    mv a0, s3
-; RV32I-NEXT:    call __gtsf2
-; RV32I-NEXT:    bgtz a0, .LBB3_2
-; RV32I-NEXT:  # %bb.1:
-; RV32I-NEXT:    srli a0, s2, 31
-; RV32I-NEXT:    addi a0, a0, -1
-; RV32I-NEXT:    and s0, a0, s1
-; RV32I-NEXT:  .LBB3_2: # %start
 ; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    call __gtsf2
+; RV32I-NEXT:    blez a0, .LBB3_2
+; RV32I-NEXT:  # %bb.1: # %start
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    j .LBB3_3
+; RV32I-NEXT:  .LBB3_2:
+; RV32I-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a0, a0, 31
+; RV32I-NEXT:    addi a1, a0, -1
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a1, a0
+; RV32I-NEXT:  .LBB3_3: # %start
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -849,36 +848,35 @@ define i16 @fcvt_ui_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -48
 ; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    lui s0, 16
-; RV64I-NEXT:    addi s0, s0, -1
-; RV64I-NEXT:    and a0, a0, s0
+; RV64I-NEXT:    lui a1, 16
+; RV64I-NEXT:    addi a1, a1, -1
+; RV64I-NEXT:    sd a1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    and a0, a0, a1
 ; RV64I-NEXT:    call __extendhfsf2
-; RV64I-NEXT:    mv s3, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    mv a0, s3
+; RV64I-NEXT:    sd a0, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    sd a0, 16(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 292864
 ; RV64I-NEXT:    addi a1, a1, -256
-; RV64I-NEXT:    mv a0, s3
-; RV64I-NEXT:    call __gtsf2
-; RV64I-NEXT:    bgtz a0, .LBB3_2
-; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    srli a0, s2, 63
-; RV64I-NEXT:    addi a0, a0, -1
-; RV64I-NEXT:    and s0, a0, s1
-; RV64I-NEXT:  .LBB3_2: # %start
 ; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    call __gtsf2
+; RV64I-NEXT:    blez a0, .LBB3_2
+; RV64I-NEXT:  # %bb.1: # %start
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    j .LBB3_3
+; RV64I-NEXT:  .LBB3_2:
+; RV64I-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
+; RV64I-NEXT:  .LBB3_3: # %start
 ; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 48
 ; RV64I-NEXT:    ret
 ;
@@ -1195,12 +1193,11 @@ define i32 @fcvt_w_h_sat(half %a) nounwind {
 ;
 ; RV32I-LABEL: fcvt_w_h_sat:
 ; RV32I:       # %bb.0: # %start
-; RV32I-NEXT:    addi sp, sp, -32
-; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    slli a0, a0, 16
 ; RV32I-NEXT:    srli a0, a0, 16
 ; RV32I-NEXT:    call __extendhfsf2
@@ -1211,7 +1208,6 @@ define i32 @fcvt_w_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __fixsfsi
 ; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    lui s3, 524288
 ; RV32I-NEXT:    bgez s2, .LBB5_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    lui s1, 524288
@@ -1222,7 +1218,8 @@ define i32 @fcvt_w_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    blez a0, .LBB5_4
 ; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    addi s1, s3, -1
+; RV32I-NEXT:    lui s1, 524288
+; RV32I-NEXT:    addi s1, s1, -1
 ; RV32I-NEXT:  .LBB5_4: # %start
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s0
@@ -1230,22 +1227,20 @@ define i32 @fcvt_w_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    snez a0, a0
 ; RV32I-NEXT:    addi a0, a0, -1
 ; RV32I-NEXT:    and a0, a0, s1
-; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_w_h_sat:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    slli a0, a0, 48
 ; RV64I-NEXT:    srli a0, a0, 48
 ; RV64I-NEXT:    call __extendhfsf2
@@ -1256,7 +1251,6 @@ define i32 @fcvt_w_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixsfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    lui s3, 524288
 ; RV64I-NEXT:    bgez s2, .LBB5_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    lui s1, 524288
@@ -1267,7 +1261,8 @@ define i32 @fcvt_w_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB5_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    addiw s1, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addiw s1, s1, -1
 ; RV64I-NEXT:  .LBB5_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -1275,12 +1270,11 @@ define i32 @fcvt_w_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    snez a0, a0
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
 ; RV32ID-ILP32-LABEL: fcvt_w_h_sat:
@@ -1823,21 +1817,19 @@ define i32 @fcvt_wu_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    slli a0, a0, 48
 ; RV64I-NEXT:    srli a0, a0, 48
 ; RV64I-NEXT:    call __extendhfsf2
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 325632
 ; RV64I-NEXT:    addi a1, a1, -1
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB8_2
 ; RV64I-NEXT:  # %bb.1: # %start
@@ -1845,14 +1837,14 @@ define i32 @fcvt_wu_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    j .LBB8_3
 ; RV64I-NEXT:  .LBB8_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB8_3: # %start
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
@@ -2151,11 +2143,12 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IZFH-NEXT:    call __fixsfdi
+; RV32IZFH-NEXT:    mv a2, s0
+; RV32IZFH-NEXT:    lui a4, 524288
 ; RV32IZFH-NEXT:    lui a3, 524288
-; RV32IZFH-NEXT:    lui a2, 524288
 ; RV32IZFH-NEXT:    beqz s0, .LBB10_2
 ; RV32IZFH-NEXT:  # %bb.1: # %start
-; RV32IZFH-NEXT:    mv a2, a1
+; RV32IZFH-NEXT:    mv a3, a1
 ; RV32IZFH-NEXT:  .LBB10_2: # %start
 ; RV32IZFH-NEXT:    lui a1, 389120
 ; RV32IZFH-NEXT:    addi a1, a1, -1
@@ -2163,16 +2156,16 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFH-NEXT:    beqz a1, .LBB10_4
 ; RV32IZFH-NEXT:  # %bb.3:
-; RV32IZFH-NEXT:    addi a2, a3, -1
+; RV32IZFH-NEXT:    addi a3, a4, -1
 ; RV32IZFH-NEXT:  .LBB10_4: # %start
-; RV32IZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFH-NEXT:    neg a4, s0
-; RV32IZFH-NEXT:    and a0, a4, a0
+; RV32IZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFH-NEXT:    neg a2, a2
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    neg a1, a1
-; RV32IZFH-NEXT:    neg a3, a3
+; RV32IZFH-NEXT:    neg a2, a4
 ; RV32IZFH-NEXT:    or a0, a1, a0
-; RV32IZFH-NEXT:    and a1, a3, a2
-; RV32IZFH-NEXT:    and a0, a3, a0
+; RV32IZFH-NEXT:    and a1, a2, a3
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -2200,11 +2193,12 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IDZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IDZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IDZFH-NEXT:    call __fixsfdi
+; RV32IDZFH-NEXT:    mv a2, s0
+; RV32IDZFH-NEXT:    lui a4, 524288
 ; RV32IDZFH-NEXT:    lui a3, 524288
-; RV32IDZFH-NEXT:    lui a2, 524288
 ; RV32IDZFH-NEXT:    beqz s0, .LBB10_2
 ; RV32IDZFH-NEXT:  # %bb.1: # %start
-; RV32IDZFH-NEXT:    mv a2, a1
+; RV32IDZFH-NEXT:    mv a3, a1
 ; RV32IDZFH-NEXT:  .LBB10_2: # %start
 ; RV32IDZFH-NEXT:    lui a1, 389120
 ; RV32IDZFH-NEXT:    addi a1, a1, -1
@@ -2212,16 +2206,16 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IDZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IDZFH-NEXT:    beqz a1, .LBB10_4
 ; RV32IDZFH-NEXT:  # %bb.3:
-; RV32IDZFH-NEXT:    addi a2, a3, -1
+; RV32IDZFH-NEXT:    addi a3, a4, -1
 ; RV32IDZFH-NEXT:  .LBB10_4: # %start
-; RV32IDZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IDZFH-NEXT:    neg a4, s0
-; RV32IDZFH-NEXT:    and a0, a4, a0
+; RV32IDZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IDZFH-NEXT:    neg a2, a2
+; RV32IDZFH-NEXT:    and a0, a2, a0
 ; RV32IDZFH-NEXT:    neg a1, a1
-; RV32IDZFH-NEXT:    neg a3, a3
+; RV32IDZFH-NEXT:    neg a2, a4
 ; RV32IDZFH-NEXT:    or a0, a1, a0
-; RV32IDZFH-NEXT:    and a1, a3, a2
-; RV32IDZFH-NEXT:    and a0, a3, a0
+; RV32IDZFH-NEXT:    and a1, a2, a3
+; RV32IDZFH-NEXT:    and a0, a2, a0
 ; RV32IDZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IDZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IDZFH-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -2248,27 +2242,28 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINX-NEXT:    mv a0, s0
 ; RV32IZHINX-NEXT:    call __fixsfdi
+; RV32IZHINX-NEXT:    mv a2, s1
+; RV32IZHINX-NEXT:    lui a4, 524288
 ; RV32IZHINX-NEXT:    lui a3, 524288
-; RV32IZHINX-NEXT:    lui a2, 524288
 ; RV32IZHINX-NEXT:    beqz s1, .LBB10_2
 ; RV32IZHINX-NEXT:  # %bb.1: # %start
-; RV32IZHINX-NEXT:    mv a2, a1
+; RV32IZHINX-NEXT:    mv a3, a1
 ; RV32IZHINX-NEXT:  .LBB10_2: # %start
 ; RV32IZHINX-NEXT:    lui a1, 389120
 ; RV32IZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINX-NEXT:    beqz a1, .LBB10_4
 ; RV32IZHINX-NEXT:  # %bb.3:
-; RV32IZHINX-NEXT:    addi a2, a3, -1
+; RV32IZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZHINX-NEXT:  .LBB10_4: # %start
-; RV32IZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZHINX-NEXT:    neg a4, s1
-; RV32IZHINX-NEXT:    and a0, a4, a0
+; RV32IZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZHINX-NEXT:    neg a2, a2
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    neg a1, a1
-; RV32IZHINX-NEXT:    neg a3, a3
+; RV32IZHINX-NEXT:    neg a2, a4
 ; RV32IZHINX-NEXT:    or a0, a1, a0
-; RV32IZHINX-NEXT:    and a1, a3, a2
-; RV32IZHINX-NEXT:    and a0, a3, a0
+; RV32IZHINX-NEXT:    and a1, a2, a3
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -2295,27 +2290,28 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IZDINXZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZDINXZHINX-NEXT:    mv a0, s0
 ; RV32IZDINXZHINX-NEXT:    call __fixsfdi
+; RV32IZDINXZHINX-NEXT:    mv a2, s1
+; RV32IZDINXZHINX-NEXT:    lui a4, 524288
 ; RV32IZDINXZHINX-NEXT:    lui a3, 524288
-; RV32IZDINXZHINX-NEXT:    lui a2, 524288
 ; RV32IZDINXZHINX-NEXT:    beqz s1, .LBB10_2
 ; RV32IZDINXZHINX-NEXT:  # %bb.1: # %start
-; RV32IZDINXZHINX-NEXT:    mv a2, a1
+; RV32IZDINXZHINX-NEXT:    mv a3, a1
 ; RV32IZDINXZHINX-NEXT:  .LBB10_2: # %start
 ; RV32IZDINXZHINX-NEXT:    lui a1, 389120
 ; RV32IZDINXZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZDINXZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZDINXZHINX-NEXT:    beqz a1, .LBB10_4
 ; RV32IZDINXZHINX-NEXT:  # %bb.3:
-; RV32IZDINXZHINX-NEXT:    addi a2, a3, -1
+; RV32IZDINXZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZDINXZHINX-NEXT:  .LBB10_4: # %start
-; RV32IZDINXZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZDINXZHINX-NEXT:    neg a4, s1
-; RV32IZDINXZHINX-NEXT:    and a0, a4, a0
+; RV32IZDINXZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZDINXZHINX-NEXT:    neg a2, a2
+; RV32IZDINXZHINX-NEXT:    and a0, a2, a0
 ; RV32IZDINXZHINX-NEXT:    neg a1, a1
-; RV32IZDINXZHINX-NEXT:    neg a3, a3
+; RV32IZDINXZHINX-NEXT:    neg a2, a4
 ; RV32IZDINXZHINX-NEXT:    or a0, a1, a0
-; RV32IZDINXZHINX-NEXT:    and a1, a3, a2
-; RV32IZDINXZHINX-NEXT:    and a0, a3, a0
+; RV32IZDINXZHINX-NEXT:    and a1, a2, a3
+; RV32IZDINXZHINX-NEXT:    and a0, a2, a0
 ; RV32IZDINXZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZDINXZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZDINXZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -2340,7 +2336,6 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __extendhfsf2
 ; RV32I-NEXT:    mv s2, a0
 ; RV32I-NEXT:    lui a1, 913408
@@ -2350,7 +2345,6 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    call __fixsfdi
 ; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    mv s1, a1
-; RV32I-NEXT:    lui s5, 524288
 ; RV32I-NEXT:    bgez s3, .LBB10_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    lui s1, 524288
@@ -2362,7 +2356,8 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    mv s4, a0
 ; RV32I-NEXT:    blez a0, .LBB10_4
 ; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    addi s1, s5, -1
+; RV32I-NEXT:    lui s1, 524288
+; RV32I-NEXT:    addi s1, s1, -1
 ; RV32I-NEXT:  .LBB10_4: # %start
 ; RV32I-NEXT:    mv a0, s2
 ; RV32I-NEXT:    mv a1, s2
@@ -2383,18 +2378,16 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_l_h_sat:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    slli a0, a0, 48
 ; RV64I-NEXT:    srli a0, a0, 48
 ; RV64I-NEXT:    call __extendhfsf2
@@ -2405,10 +2398,10 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixsfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    li s3, -1
 ; RV64I-NEXT:    bgez s2, .LBB10_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    slli s1, s3, 63
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    slli s1, s1, 63
 ; RV64I-NEXT:  .LBB10_2: # %start
 ; RV64I-NEXT:    lui a1, 389120
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -2416,7 +2409,8 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB10_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    srli s1, s3, 1
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    srli s1, s1, 1
 ; RV64I-NEXT:  .LBB10_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -2424,12 +2418,11 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV64I-NEXT:    snez a0, a0
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
 ; RV32ID-ILP32-LABEL: fcvt_l_h_sat:
@@ -2444,11 +2437,12 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32ID-ILP32-NEXT:    fsw fa4, 4(sp) # 4-byte Folded Spill
 ; RV32ID-ILP32-NEXT:    fle.s s0, fa5, fa4
 ; RV32ID-ILP32-NEXT:    call __fixsfdi
+; RV32ID-ILP32-NEXT:    mv a2, s0
+; RV32ID-ILP32-NEXT:    lui a4, 524288
 ; RV32ID-ILP32-NEXT:    lui a3, 524288
-; RV32ID-ILP32-NEXT:    lui a2, 524288
 ; RV32ID-ILP32-NEXT:    beqz s0, .LBB10_2
 ; RV32ID-ILP32-NEXT:  # %bb.1: # %start
-; RV32ID-ILP32-NEXT:    mv a2, a1
+; RV32ID-ILP32-NEXT:    mv a3, a1
 ; RV32ID-ILP32-NEXT:  .LBB10_2: # %start
 ; RV32ID-ILP32-NEXT:    lui a1, 389120
 ; RV32ID-ILP32-NEXT:    addi a1, a1, -1
@@ -2458,16 +2452,16 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32ID-ILP32-NEXT:    fmv.s fa5, fa4
 ; RV32ID-ILP32-NEXT:    beqz a1, .LBB10_4
 ; RV32ID-ILP32-NEXT:  # %bb.3:
-; RV32ID-ILP32-NEXT:    addi a2, a3, -1
+; RV32ID-ILP32-NEXT:    addi a3, a4, -1
 ; RV32ID-ILP32-NEXT:  .LBB10_4: # %start
-; RV32ID-ILP32-NEXT:    feq.s a3, fa5, fa5
-; RV32ID-ILP32-NEXT:    neg a4, s0
+; RV32ID-ILP32-NEXT:    feq.s a4, fa5, fa5
+; RV32ID-ILP32-NEXT:    neg a2, a2
 ; RV32ID-ILP32-NEXT:    neg a1, a1
-; RV32ID-ILP32-NEXT:    and a0, a4, a0
-; RV32ID-ILP32-NEXT:    neg a3, a3
+; RV32ID-ILP32-NEXT:    and a0, a2, a0
+; RV32ID-ILP32-NEXT:    neg a2, a4
 ; RV32ID-ILP32-NEXT:    or a0, a1, a0
-; RV32ID-ILP32-NEXT:    and a1, a3, a2
-; RV32ID-ILP32-NEXT:    and a0, a3, a0
+; RV32ID-ILP32-NEXT:    and a1, a2, a3
+; RV32ID-ILP32-NEXT:    and a0, a2, a0
 ; RV32ID-ILP32-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32ID-ILP32-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32ID-ILP32-NEXT:    addi sp, sp, 16
@@ -2500,11 +2494,12 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32ID-NEXT:    fmv.s fs0, fa0
 ; RV32ID-NEXT:    fle.s s0, fa5, fa0
 ; RV32ID-NEXT:    call __fixsfdi
+; RV32ID-NEXT:    mv a2, s0
+; RV32ID-NEXT:    lui a4, 524288
 ; RV32ID-NEXT:    lui a3, 524288
-; RV32ID-NEXT:    lui a2, 524288
 ; RV32ID-NEXT:    beqz s0, .LBB10_2
 ; RV32ID-NEXT:  # %bb.1: # %start
-; RV32ID-NEXT:    mv a2, a1
+; RV32ID-NEXT:    mv a3, a1
 ; RV32ID-NEXT:  .LBB10_2: # %start
 ; RV32ID-NEXT:    lui a1, 389120
 ; RV32ID-NEXT:    addi a1, a1, -1
@@ -2512,16 +2507,16 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32ID-NEXT:    flt.s a1, fa5, fs0
 ; RV32ID-NEXT:    beqz a1, .LBB10_4
 ; RV32ID-NEXT:  # %bb.3:
-; RV32ID-NEXT:    addi a2, a3, -1
+; RV32ID-NEXT:    addi a3, a4, -1
 ; RV32ID-NEXT:  .LBB10_4: # %start
-; RV32ID-NEXT:    feq.s a3, fs0, fs0
-; RV32ID-NEXT:    neg a4, s0
-; RV32ID-NEXT:    and a0, a4, a0
+; RV32ID-NEXT:    feq.s a4, fs0, fs0
+; RV32ID-NEXT:    neg a2, a2
+; RV32ID-NEXT:    and a0, a2, a0
 ; RV32ID-NEXT:    neg a1, a1
-; RV32ID-NEXT:    neg a3, a3
+; RV32ID-NEXT:    neg a2, a4
 ; RV32ID-NEXT:    or a0, a1, a0
-; RV32ID-NEXT:    and a1, a3, a2
-; RV32ID-NEXT:    and a0, a3, a0
+; RV32ID-NEXT:    and a1, a2, a3
+; RV32ID-NEXT:    and a0, a2, a0
 ; RV32ID-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32ID-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32ID-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -2554,11 +2549,12 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IFZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IFZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IFZFHMIN-NEXT:    call __fixsfdi
+; RV32IFZFHMIN-NEXT:    mv a2, s0
+; RV32IFZFHMIN-NEXT:    lui a4, 524288
 ; RV32IFZFHMIN-NEXT:    lui a3, 524288
-; RV32IFZFHMIN-NEXT:    lui a2, 524288
 ; RV32IFZFHMIN-NEXT:    beqz s0, .LBB10_2
 ; RV32IFZFHMIN-NEXT:  # %bb.1: # %start
-; RV32IFZFHMIN-NEXT:    mv a2, a1
+; RV32IFZFHMIN-NEXT:    mv a3, a1
 ; RV32IFZFHMIN-NEXT:  .LBB10_2: # %start
 ; RV32IFZFHMIN-NEXT:    lui a1, 389120
 ; RV32IFZFHMIN-NEXT:    addi a1, a1, -1
@@ -2566,16 +2562,16 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IFZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IFZFHMIN-NEXT:    beqz a1, .LBB10_4
 ; RV32IFZFHMIN-NEXT:  # %bb.3:
-; RV32IFZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IFZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IFZFHMIN-NEXT:  .LBB10_4: # %start
-; RV32IFZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IFZFHMIN-NEXT:    neg a4, s0
-; RV32IFZFHMIN-NEXT:    and a0, a4, a0
+; RV32IFZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IFZFHMIN-NEXT:    neg a2, a2
+; RV32IFZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IFZFHMIN-NEXT:    neg a1, a1
-; RV32IFZFHMIN-NEXT:    neg a3, a3
+; RV32IFZFHMIN-NEXT:    neg a2, a4
 ; RV32IFZFHMIN-NEXT:    or a0, a1, a0
-; RV32IFZFHMIN-NEXT:    and a1, a3, a2
-; RV32IFZFHMIN-NEXT:    and a0, a3, a0
+; RV32IFZFHMIN-NEXT:    and a1, a2, a3
+; RV32IFZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IFZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IFZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IFZFHMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -2604,11 +2600,12 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IDZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IDZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IDZFHMIN-NEXT:    call __fixsfdi
+; RV32IDZFHMIN-NEXT:    mv a2, s0
+; RV32IDZFHMIN-NEXT:    lui a4, 524288
 ; RV32IDZFHMIN-NEXT:    lui a3, 524288
-; RV32IDZFHMIN-NEXT:    lui a2, 524288
 ; RV32IDZFHMIN-NEXT:    beqz s0, .LBB10_2
 ; RV32IDZFHMIN-NEXT:  # %bb.1: # %start
-; RV32IDZFHMIN-NEXT:    mv a2, a1
+; RV32IDZFHMIN-NEXT:    mv a3, a1
 ; RV32IDZFHMIN-NEXT:  .LBB10_2: # %start
 ; RV32IDZFHMIN-NEXT:    lui a1, 389120
 ; RV32IDZFHMIN-NEXT:    addi a1, a1, -1
@@ -2616,16 +2613,16 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; RV32IDZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IDZFHMIN-NEXT:    beqz a1, .LBB10_4
 ; RV32IDZFHMIN-NEXT:  # %bb.3:
-; RV32IDZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IDZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IDZFHMIN-NEXT:  .LBB10_4: # %start
-; RV32IDZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IDZFHMIN-NEXT:    neg a4, s0
-; RV32IDZFHMIN-NEXT:    and a0, a4, a0
+; RV32IDZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IDZFHMIN-NEXT:    neg a2, a2
+; RV32IDZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IDZFHMIN-NEXT:    neg a1, a1
-; RV32IDZFHMIN-NEXT:    neg a3, a3
+; RV32IDZFHMIN-NEXT:    neg a2, a4
 ; RV32IDZFHMIN-NEXT:    or a0, a1, a0
-; RV32IDZFHMIN-NEXT:    and a1, a3, a2
-; RV32IDZFHMIN-NEXT:    and a0, a3, a0
+; RV32IDZFHMIN-NEXT:    and a1, a2, a3
+; RV32IDZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IDZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IDZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IDZFHMIN-NEXT:    fld fs0, 0(sp) # 8-byte Folded Reload
@@ -2643,27 +2640,28 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; CHECK32-IZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; CHECK32-IZHINXMIN-NEXT:    mv a0, s0
 ; CHECK32-IZHINXMIN-NEXT:    call __fixsfdi
+; CHECK32-IZHINXMIN-NEXT:    mv a2, s1
+; CHECK32-IZHINXMIN-NEXT:    lui a4, 524288
 ; CHECK32-IZHINXMIN-NEXT:    lui a3, 524288
-; CHECK32-IZHINXMIN-NEXT:    lui a2, 524288
 ; CHECK32-IZHINXMIN-NEXT:    beqz s1, .LBB10_2
 ; CHECK32-IZHINXMIN-NEXT:  # %bb.1: # %start
-; CHECK32-IZHINXMIN-NEXT:    mv a2, a1
+; CHECK32-IZHINXMIN-NEXT:    mv a3, a1
 ; CHECK32-IZHINXMIN-NEXT:  .LBB10_2: # %start
 ; CHECK32-IZHINXMIN-NEXT:    lui a1, 389120
 ; CHECK32-IZHINXMIN-NEXT:    addi a1, a1, -1
 ; CHECK32-IZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; CHECK32-IZHINXMIN-NEXT:    beqz a1, .LBB10_4
 ; CHECK32-IZHINXMIN-NEXT:  # %bb.3:
-; CHECK32-IZHINXMIN-NEXT:    addi a2, a3, -1
+; CHECK32-IZHINXMIN-NEXT:    addi a3, a4, -1
 ; CHECK32-IZHINXMIN-NEXT:  .LBB10_4: # %start
-; CHECK32-IZHINXMIN-NEXT:    feq.s a3, s0, s0
-; CHECK32-IZHINXMIN-NEXT:    neg a4, s1
-; CHECK32-IZHINXMIN-NEXT:    and a0, a4, a0
+; CHECK32-IZHINXMIN-NEXT:    feq.s a4, s0, s0
+; CHECK32-IZHINXMIN-NEXT:    neg a2, a2
+; CHECK32-IZHINXMIN-NEXT:    and a0, a2, a0
 ; CHECK32-IZHINXMIN-NEXT:    neg a1, a1
-; CHECK32-IZHINXMIN-NEXT:    neg a3, a3
+; CHECK32-IZHINXMIN-NEXT:    neg a2, a4
 ; CHECK32-IZHINXMIN-NEXT:    or a0, a1, a0
-; CHECK32-IZHINXMIN-NEXT:    and a1, a3, a2
-; CHECK32-IZHINXMIN-NEXT:    and a0, a3, a0
+; CHECK32-IZHINXMIN-NEXT:    and a1, a2, a3
+; CHECK32-IZHINXMIN-NEXT:    and a0, a2, a0
 ; CHECK32-IZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; CHECK32-IZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; CHECK32-IZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -2691,27 +2689,28 @@ define i64 @fcvt_l_h_sat(half %a) nounwind {
 ; CHECK32-IZDINXZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; CHECK32-IZDINXZHINXMIN-NEXT:    mv a0, s0
 ; CHECK32-IZDINXZHINXMIN-NEXT:    call __fixsfdi
+; CHECK32-IZDINXZHINXMIN-NEXT:    mv a2, s1
+; CHECK32-IZDINXZHINXMIN-NEXT:    lui a4, 524288
 ; CHECK32-IZDINXZHINXMIN-NEXT:    lui a3, 524288
-; CHECK32-IZDINXZHINXMIN-NEXT:    lui a2, 524288
 ; CHECK32-IZDINXZHINXMIN-NEXT:    beqz s1, .LBB10_2
 ; CHECK32-IZDINXZHINXMIN-NEXT:  # %bb.1: # %start
-; CHECK32-IZDINXZHINXMIN-NEXT:    mv a2, a1
+; CHECK32-IZDINXZHINXMIN-NEXT:    mv a3, a1
 ; CHECK32-IZDINXZHINXMIN-NEXT:  .LBB10_2: # %start
 ; CHECK32-IZDINXZHINXMIN-NEXT:    lui a1, 389120
 ; CHECK32-IZDINXZHINXMIN-NEXT:    addi a1, a1, -1
 ; CHECK32-IZDINXZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; CHECK32-IZDINXZHINXMIN-NEXT:    beqz a1, .LBB10_4
 ; CHECK32-IZDINXZHINXMIN-NEXT:  # %bb.3:
-; CHECK32-IZDINXZHINXMIN-NEXT:    addi a2, a3, -1
+; CHECK32-IZDINXZHINXMIN-NEXT:    addi a3, a4, -1
 ; CHECK32-IZDINXZHINXMIN-NEXT:  .LBB10_4: # %start
-; CHECK32-IZDINXZHINXMIN-NEXT:    feq.s a3, s0, s0
-; CHECK32-IZDINXZHINXMIN-NEXT:    neg a4, s1
-; CHECK32-IZDINXZHINXMIN-NEXT:    and a0, a4, a0
+; CHECK32-IZDINXZHINXMIN-NEXT:    feq.s a4, s0, s0
+; CHECK32-IZDINXZHINXMIN-NEXT:    neg a2, a2
+; CHECK32-IZDINXZHINXMIN-NEXT:    and a0, a2, a0
 ; CHECK32-IZDINXZHINXMIN-NEXT:    neg a1, a1
-; CHECK32-IZDINXZHINXMIN-NEXT:    neg a3, a3
+; CHECK32-IZDINXZHINXMIN-NEXT:    neg a2, a4
 ; CHECK32-IZDINXZHINXMIN-NEXT:    or a0, a1, a0
-; CHECK32-IZDINXZHINXMIN-NEXT:    and a1, a3, a2
-; CHECK32-IZDINXZHINXMIN-NEXT:    and a0, a3, a0
+; CHECK32-IZDINXZHINXMIN-NEXT:    and a1, a2, a3
+; CHECK32-IZDINXZHINXMIN-NEXT:    and a0, a2, a0
 ; CHECK32-IZDINXZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; CHECK32-IZDINXZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; CHECK32-IZDINXZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -6932,38 +6931,36 @@ define zeroext i16 @fcvt_wu_s_sat_i16(half %a) nounwind {
 ; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    lui s3, 16
-; RV32I-NEXT:    addi s3, s3, -1
-; RV32I-NEXT:    and a0, a0, s3
+; RV32I-NEXT:    lui s1, 16
+; RV32I-NEXT:    addi s1, s1, -1
+; RV32I-NEXT:    and a0, a0, s1
 ; RV32I-NEXT:    call __extendhfsf2
-; RV32I-NEXT:    mv s2, a0
-; RV32I-NEXT:    call __fixunssfsi
 ; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    call __fixunssfsi
+; RV32I-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    mv s1, a0
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 292864
 ; RV32I-NEXT:    addi a1, a1, -256
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    blez a0, .LBB34_2
 ; RV32I-NEXT:  # %bb.1: # %start
-; RV32I-NEXT:    mv a0, s3
+; RV32I-NEXT:    mv a0, s1
 ; RV32I-NEXT:    j .LBB34_3
 ; RV32I-NEXT:  .LBB34_2:
-; RV32I-NEXT:    srli s1, s1, 31
-; RV32I-NEXT:    addi a0, s1, -1
-; RV32I-NEXT:    and a0, a0, s0
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a0, a0, 31
+; RV32I-NEXT:    addi a1, a0, -1
+; RV32I-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a1, a0
 ; RV32I-NEXT:  .LBB34_3: # %start
-; RV32I-NEXT:    and a0, a0, s3
+; RV32I-NEXT:    and a0, a0, s1
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -6973,38 +6970,36 @@ define zeroext i16 @fcvt_wu_s_sat_i16(half %a) nounwind {
 ; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    lui s3, 16
-; RV64I-NEXT:    addi s3, s3, -1
-; RV64I-NEXT:    and a0, a0, s3
+; RV64I-NEXT:    lui s1, 16
+; RV64I-NEXT:    addi s1, s1, -1
+; RV64I-NEXT:    and a0, a0, s1
 ; RV64I-NEXT:    call __extendhfsf2
-; RV64I-NEXT:    mv s2, a0
-; RV64I-NEXT:    call __fixunssfdi
 ; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    call __fixunssfdi
+; RV64I-NEXT:    sd a0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 292864
 ; RV64I-NEXT:    addi a1, a1, -256
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB34_2
 ; RV64I-NEXT:  # %bb.1: # %start
-; RV64I-NEXT:    mv a0, s3
+; RV64I-NEXT:    mv a0, s1
 ; RV64I-NEXT:    j .LBB34_3
 ; RV64I-NEXT:  .LBB34_2:
-; RV64I-NEXT:    srli s1, s1, 63
-; RV64I-NEXT:    addi a0, s1, -1
-; RV64I-NEXT:    and a0, a0, s0
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB34_3: # %start
-; RV64I-NEXT:    and a0, a0, s3
+; RV64I-NEXT:    and a0, a0, s1
 ; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 48
 ; RV64I-NEXT:    ret
 ;
@@ -7883,35 +7878,33 @@ define zeroext i8 @fcvt_wu_s_sat_i8(half %a) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    slli a0, a0, 16
 ; RV32I-NEXT:    srli a0, a0, 16
 ; RV32I-NEXT:    call __extendhfsf2
-; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    mv s0, a0
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __gesf2
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    sw a0, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __fixunssfsi
-; RV32I-NEXT:    mv s1, a0
+; RV32I-NEXT:    sw a0, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    lui a1, 276464
-; RV32I-NEXT:    mv a0, s2
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    blez a0, .LBB38_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    li a0, 255
 ; RV32I-NEXT:    j .LBB38_3
 ; RV32I-NEXT:  .LBB38_2:
-; RV32I-NEXT:    srli s0, s0, 31
-; RV32I-NEXT:    addi a0, s0, -1
-; RV32I-NEXT:    and a0, a0, s1
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a0, a0, 31
+; RV32I-NEXT:    addi a1, a0, -1
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a1, a0
 ; RV32I-NEXT:  .LBB38_3: # %start
 ; RV32I-NEXT:    zext.b a0, a0
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -7920,35 +7913,33 @@ define zeroext i8 @fcvt_wu_s_sat_i8(half %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    slli a0, a0, 48
 ; RV64I-NEXT:    srli a0, a0, 48
 ; RV64I-NEXT:    call __extendhfsf2
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 276464
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB38_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    li a0, 255
 ; RV64I-NEXT:    j .LBB38_3
 ; RV64I-NEXT:  .LBB38_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB38_3: # %start
 ; RV64I-NEXT:    zext.b a0, a0
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
@@ -8194,21 +8185,19 @@ define zeroext i32 @fcvt_wu_h_sat_zext(half %a) nounwind {
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    slli a0, a0, 48
 ; RV64I-NEXT:    srli a0, a0, 48
 ; RV64I-NEXT:    call __extendhfsf2
-; RV64I-NEXT:    mv s2, a0
+; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, 0
 ; RV64I-NEXT:    call __gesf2
-; RV64I-NEXT:    mv s0, a0
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixunssfdi
-; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    lui a1, 325632
 ; RV64I-NEXT:    addi a1, a1, -1
-; RV64I-NEXT:    mv a0, s2
+; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB39_2
 ; RV64I-NEXT:  # %bb.1: # %start
@@ -8216,16 +8205,16 @@ define zeroext i32 @fcvt_wu_h_sat_zext(half %a) nounwind {
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    j .LBB39_3
 ; RV64I-NEXT:  .LBB39_2:
-; RV64I-NEXT:    srli s0, s0, 63
-; RV64I-NEXT:    addi a0, s0, -1
-; RV64I-NEXT:    and a0, a0, s1
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    srli a0, a0, 63
+; RV64I-NEXT:    addi a1, a0, -1
+; RV64I-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    and a0, a1, a0
 ; RV64I-NEXT:  .LBB39_3: # %start
 ; RV64I-NEXT:    slli a0, a0, 32
 ; RV64I-NEXT:    srli a0, a0, 32
 ; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
@@ -8409,12 +8398,11 @@ define signext i32 @fcvt_w_h_sat_sext(half %a) nounwind {
 ;
 ; RV32I-LABEL: fcvt_w_h_sat_sext:
 ; RV32I:       # %bb.0: # %start
-; RV32I-NEXT:    addi sp, sp, -32
-; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    slli a0, a0, 16
 ; RV32I-NEXT:    srli a0, a0, 16
 ; RV32I-NEXT:    call __extendhfsf2
@@ -8425,7 +8413,6 @@ define signext i32 @fcvt_w_h_sat_sext(half %a) nounwind {
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call __fixsfsi
 ; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    lui s3, 524288
 ; RV32I-NEXT:    bgez s2, .LBB40_2
 ; RV32I-NEXT:  # %bb.1: # %start
 ; RV32I-NEXT:    lui s1, 524288
@@ -8436,7 +8423,8 @@ define signext i32 @fcvt_w_h_sat_sext(half %a) nounwind {
 ; RV32I-NEXT:    call __gtsf2
 ; RV32I-NEXT:    blez a0, .LBB40_4
 ; RV32I-NEXT:  # %bb.3: # %start
-; RV32I-NEXT:    addi s1, s3, -1
+; RV32I-NEXT:    lui s1, 524288
+; RV32I-NEXT:    addi s1, s1, -1
 ; RV32I-NEXT:  .LBB40_4: # %start
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    mv a1, s0
@@ -8444,22 +8432,20 @@ define signext i32 @fcvt_w_h_sat_sext(half %a) nounwind {
 ; RV32I-NEXT:    snez a0, a0
 ; RV32I-NEXT:    addi a0, a0, -1
 ; RV32I-NEXT:    and a0, a0, s1
-; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: fcvt_w_h_sat_sext:
 ; RV64I:       # %bb.0: # %start
-; RV64I-NEXT:    addi sp, sp, -48
-; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -32
+; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    slli a0, a0, 48
 ; RV64I-NEXT:    srli a0, a0, 48
 ; RV64I-NEXT:    call __extendhfsf2
@@ -8470,7 +8456,6 @@ define signext i32 @fcvt_w_h_sat_sext(half %a) nounwind {
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call __fixsfdi
 ; RV64I-NEXT:    mv s1, a0
-; RV64I-NEXT:    lui s3, 524288
 ; RV64I-NEXT:    bgez s2, .LBB40_2
 ; RV64I-NEXT:  # %bb.1: # %start
 ; RV64I-NEXT:    lui s1, 524288
@@ -8481,7 +8466,8 @@ define signext i32 @fcvt_w_h_sat_sext(half %a) nounwind {
 ; RV64I-NEXT:    call __gtsf2
 ; RV64I-NEXT:    blez a0, .LBB40_4
 ; RV64I-NEXT:  # %bb.3: # %start
-; RV64I-NEXT:    addi s1, s3, -1
+; RV64I-NEXT:    lui s1, 524288
+; RV64I-NEXT:    addi s1, s1, -1
 ; RV64I-NEXT:  .LBB40_4: # %start
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -8490,12 +8476,11 @@ define signext i32 @fcvt_w_h_sat_sext(half %a) nounwind {
 ; RV64I-NEXT:    addi a0, a0, -1
 ; RV64I-NEXT:    and a0, a0, s1
 ; RV64I-NEXT:    sext.w a0, a0
-; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 32
 ; RV64I-NEXT:    ret
 ;
 ; RV32ID-ILP32-LABEL: fcvt_w_h_sat_sext:
diff --git a/llvm/test/CodeGen/RISCV/half-round-conv-sat.ll b/llvm/test/CodeGen/RISCV/half-round-conv-sat.ll
index b74750a86ff3c2..4ebc96d1eca524 100644
--- a/llvm/test/CodeGen/RISCV/half-round-conv-sat.ll
+++ b/llvm/test/CodeGen/RISCV/half-round-conv-sat.ll
@@ -116,11 +116,12 @@ define i64 @test_floor_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IZFH-NEXT:    call __fixsfdi
+; RV32IZFH-NEXT:    mv a2, s0
+; RV32IZFH-NEXT:    lui a4, 524288
 ; RV32IZFH-NEXT:    lui a3, 524288
-; RV32IZFH-NEXT:    lui a2, 524288
 ; RV32IZFH-NEXT:    beqz s0, .LBB1_4
 ; RV32IZFH-NEXT:  # %bb.3:
-; RV32IZFH-NEXT:    mv a2, a1
+; RV32IZFH-NEXT:    mv a3, a1
 ; RV32IZFH-NEXT:  .LBB1_4:
 ; RV32IZFH-NEXT:    lui a1, 389120
 ; RV32IZFH-NEXT:    addi a1, a1, -1
@@ -128,16 +129,16 @@ define i64 @test_floor_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFH-NEXT:    beqz a1, .LBB1_6
 ; RV32IZFH-NEXT:  # %bb.5:
-; RV32IZFH-NEXT:    addi a2, a3, -1
+; RV32IZFH-NEXT:    addi a3, a4, -1
 ; RV32IZFH-NEXT:  .LBB1_6:
-; RV32IZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFH-NEXT:    neg a4, s0
-; RV32IZFH-NEXT:    and a0, a4, a0
+; RV32IZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFH-NEXT:    neg a2, a2
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    neg a1, a1
-; RV32IZFH-NEXT:    neg a3, a3
+; RV32IZFH-NEXT:    neg a2, a4
 ; RV32IZFH-NEXT:    or a0, a1, a0
-; RV32IZFH-NEXT:    and a1, a3, a2
-; RV32IZFH-NEXT:    and a0, a3, a0
+; RV32IZFH-NEXT:    and a1, a2, a3
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -174,27 +175,28 @@ define i64 @test_floor_si64(half %x) nounwind {
 ; RV32IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINX-NEXT:    mv a0, s0
 ; RV32IZHINX-NEXT:    call __fixsfdi
+; RV32IZHINX-NEXT:    mv a2, s1
+; RV32IZHINX-NEXT:    lui a4, 524288
 ; RV32IZHINX-NEXT:    lui a3, 524288
-; RV32IZHINX-NEXT:    lui a2, 524288
 ; RV32IZHINX-NEXT:    beqz s1, .LBB1_4
 ; RV32IZHINX-NEXT:  # %bb.3:
-; RV32IZHINX-NEXT:    mv a2, a1
+; RV32IZHINX-NEXT:    mv a3, a1
 ; RV32IZHINX-NEXT:  .LBB1_4:
 ; RV32IZHINX-NEXT:    lui a1, 389120
 ; RV32IZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINX-NEXT:    beqz a1, .LBB1_6
 ; RV32IZHINX-NEXT:  # %bb.5:
-; RV32IZHINX-NEXT:    addi a2, a3, -1
+; RV32IZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZHINX-NEXT:  .LBB1_6:
-; RV32IZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZHINX-NEXT:    neg a4, s1
-; RV32IZHINX-NEXT:    and a0, a4, a0
+; RV32IZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZHINX-NEXT:    neg a2, a2
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    neg a1, a1
-; RV32IZHINX-NEXT:    neg a3, a3
+; RV32IZHINX-NEXT:    neg a2, a4
 ; RV32IZHINX-NEXT:    or a0, a1, a0
-; RV32IZHINX-NEXT:    and a1, a3, a2
-; RV32IZHINX-NEXT:    and a0, a3, a0
+; RV32IZHINX-NEXT:    and a1, a2, a3
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -244,11 +246,12 @@ define i64 @test_floor_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IZFHMIN-NEXT:    call __fixsfdi
+; RV32IZFHMIN-NEXT:    mv a2, s0
+; RV32IZFHMIN-NEXT:    lui a4, 524288
 ; RV32IZFHMIN-NEXT:    lui a3, 524288
-; RV32IZFHMIN-NEXT:    lui a2, 524288
 ; RV32IZFHMIN-NEXT:    beqz s0, .LBB1_4
 ; RV32IZFHMIN-NEXT:  # %bb.3:
-; RV32IZFHMIN-NEXT:    mv a2, a1
+; RV32IZFHMIN-NEXT:    mv a3, a1
 ; RV32IZFHMIN-NEXT:  .LBB1_4:
 ; RV32IZFHMIN-NEXT:    lui a1, 389120
 ; RV32IZFHMIN-NEXT:    addi a1, a1, -1
@@ -256,16 +259,16 @@ define i64 @test_floor_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFHMIN-NEXT:    beqz a1, .LBB1_6
 ; RV32IZFHMIN-NEXT:  # %bb.5:
-; RV32IZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IZFHMIN-NEXT:  .LBB1_6:
-; RV32IZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFHMIN-NEXT:    neg a4, s0
-; RV32IZFHMIN-NEXT:    and a0, a4, a0
+; RV32IZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFHMIN-NEXT:    neg a2, a2
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    neg a1, a1
-; RV32IZFHMIN-NEXT:    neg a3, a3
+; RV32IZFHMIN-NEXT:    neg a2, a4
 ; RV32IZFHMIN-NEXT:    or a0, a1, a0
-; RV32IZFHMIN-NEXT:    and a1, a3, a2
-; RV32IZFHMIN-NEXT:    and a0, a3, a0
+; RV32IZFHMIN-NEXT:    and a1, a2, a3
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -316,27 +319,28 @@ define i64 @test_floor_si64(half %x) nounwind {
 ; RV32IZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINXMIN-NEXT:    mv a0, s0
 ; RV32IZHINXMIN-NEXT:    call __fixsfdi
+; RV32IZHINXMIN-NEXT:    mv a2, s1
+; RV32IZHINXMIN-NEXT:    lui a4, 524288
 ; RV32IZHINXMIN-NEXT:    lui a3, 524288
-; RV32IZHINXMIN-NEXT:    lui a2, 524288
 ; RV32IZHINXMIN-NEXT:    beqz s1, .LBB1_4
 ; RV32IZHINXMIN-NEXT:  # %bb.3:
-; RV32IZHINXMIN-NEXT:    mv a2, a1
+; RV32IZHINXMIN-NEXT:    mv a3, a1
 ; RV32IZHINXMIN-NEXT:  .LBB1_4:
 ; RV32IZHINXMIN-NEXT:    lui a1, 389120
 ; RV32IZHINXMIN-NEXT:    addi a1, a1, -1
 ; RV32IZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINXMIN-NEXT:    beqz a1, .LBB1_6
 ; RV32IZHINXMIN-NEXT:  # %bb.5:
-; RV32IZHINXMIN-NEXT:    addi a2, a3, -1
+; RV32IZHINXMIN-NEXT:    addi a3, a4, -1
 ; RV32IZHINXMIN-NEXT:  .LBB1_6:
-; RV32IZHINXMIN-NEXT:    feq.s a3, s0, s0
-; RV32IZHINXMIN-NEXT:    neg a4, s1
-; RV32IZHINXMIN-NEXT:    and a0, a4, a0
+; RV32IZHINXMIN-NEXT:    feq.s a4, s0, s0
+; RV32IZHINXMIN-NEXT:    neg a2, a2
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    neg a1, a1
-; RV32IZHINXMIN-NEXT:    neg a3, a3
+; RV32IZHINXMIN-NEXT:    neg a2, a4
 ; RV32IZHINXMIN-NEXT:    or a0, a1, a0
-; RV32IZHINXMIN-NEXT:    and a1, a3, a2
-; RV32IZHINXMIN-NEXT:    and a0, a3, a0
+; RV32IZHINXMIN-NEXT:    and a1, a2, a3
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -830,11 +834,12 @@ define i64 @test_ceil_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IZFH-NEXT:    call __fixsfdi
+; RV32IZFH-NEXT:    mv a2, s0
+; RV32IZFH-NEXT:    lui a4, 524288
 ; RV32IZFH-NEXT:    lui a3, 524288
-; RV32IZFH-NEXT:    lui a2, 524288
 ; RV32IZFH-NEXT:    beqz s0, .LBB5_4
 ; RV32IZFH-NEXT:  # %bb.3:
-; RV32IZFH-NEXT:    mv a2, a1
+; RV32IZFH-NEXT:    mv a3, a1
 ; RV32IZFH-NEXT:  .LBB5_4:
 ; RV32IZFH-NEXT:    lui a1, 389120
 ; RV32IZFH-NEXT:    addi a1, a1, -1
@@ -842,16 +847,16 @@ define i64 @test_ceil_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFH-NEXT:    beqz a1, .LBB5_6
 ; RV32IZFH-NEXT:  # %bb.5:
-; RV32IZFH-NEXT:    addi a2, a3, -1
+; RV32IZFH-NEXT:    addi a3, a4, -1
 ; RV32IZFH-NEXT:  .LBB5_6:
-; RV32IZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFH-NEXT:    neg a4, s0
-; RV32IZFH-NEXT:    and a0, a4, a0
+; RV32IZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFH-NEXT:    neg a2, a2
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    neg a1, a1
-; RV32IZFH-NEXT:    neg a3, a3
+; RV32IZFH-NEXT:    neg a2, a4
 ; RV32IZFH-NEXT:    or a0, a1, a0
-; RV32IZFH-NEXT:    and a1, a3, a2
-; RV32IZFH-NEXT:    and a0, a3, a0
+; RV32IZFH-NEXT:    and a1, a2, a3
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -888,27 +893,28 @@ define i64 @test_ceil_si64(half %x) nounwind {
 ; RV32IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINX-NEXT:    mv a0, s0
 ; RV32IZHINX-NEXT:    call __fixsfdi
+; RV32IZHINX-NEXT:    mv a2, s1
+; RV32IZHINX-NEXT:    lui a4, 524288
 ; RV32IZHINX-NEXT:    lui a3, 524288
-; RV32IZHINX-NEXT:    lui a2, 524288
 ; RV32IZHINX-NEXT:    beqz s1, .LBB5_4
 ; RV32IZHINX-NEXT:  # %bb.3:
-; RV32IZHINX-NEXT:    mv a2, a1
+; RV32IZHINX-NEXT:    mv a3, a1
 ; RV32IZHINX-NEXT:  .LBB5_4:
 ; RV32IZHINX-NEXT:    lui a1, 389120
 ; RV32IZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINX-NEXT:    beqz a1, .LBB5_6
 ; RV32IZHINX-NEXT:  # %bb.5:
-; RV32IZHINX-NEXT:    addi a2, a3, -1
+; RV32IZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZHINX-NEXT:  .LBB5_6:
-; RV32IZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZHINX-NEXT:    neg a4, s1
-; RV32IZHINX-NEXT:    and a0, a4, a0
+; RV32IZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZHINX-NEXT:    neg a2, a2
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    neg a1, a1
-; RV32IZHINX-NEXT:    neg a3, a3
+; RV32IZHINX-NEXT:    neg a2, a4
 ; RV32IZHINX-NEXT:    or a0, a1, a0
-; RV32IZHINX-NEXT:    and a1, a3, a2
-; RV32IZHINX-NEXT:    and a0, a3, a0
+; RV32IZHINX-NEXT:    and a1, a2, a3
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -958,11 +964,12 @@ define i64 @test_ceil_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IZFHMIN-NEXT:    call __fixsfdi
+; RV32IZFHMIN-NEXT:    mv a2, s0
+; RV32IZFHMIN-NEXT:    lui a4, 524288
 ; RV32IZFHMIN-NEXT:    lui a3, 524288
-; RV32IZFHMIN-NEXT:    lui a2, 524288
 ; RV32IZFHMIN-NEXT:    beqz s0, .LBB5_4
 ; RV32IZFHMIN-NEXT:  # %bb.3:
-; RV32IZFHMIN-NEXT:    mv a2, a1
+; RV32IZFHMIN-NEXT:    mv a3, a1
 ; RV32IZFHMIN-NEXT:  .LBB5_4:
 ; RV32IZFHMIN-NEXT:    lui a1, 389120
 ; RV32IZFHMIN-NEXT:    addi a1, a1, -1
@@ -970,16 +977,16 @@ define i64 @test_ceil_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFHMIN-NEXT:    beqz a1, .LBB5_6
 ; RV32IZFHMIN-NEXT:  # %bb.5:
-; RV32IZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IZFHMIN-NEXT:  .LBB5_6:
-; RV32IZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFHMIN-NEXT:    neg a4, s0
-; RV32IZFHMIN-NEXT:    and a0, a4, a0
+; RV32IZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFHMIN-NEXT:    neg a2, a2
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    neg a1, a1
-; RV32IZFHMIN-NEXT:    neg a3, a3
+; RV32IZFHMIN-NEXT:    neg a2, a4
 ; RV32IZFHMIN-NEXT:    or a0, a1, a0
-; RV32IZFHMIN-NEXT:    and a1, a3, a2
-; RV32IZFHMIN-NEXT:    and a0, a3, a0
+; RV32IZFHMIN-NEXT:    and a1, a2, a3
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -1030,27 +1037,28 @@ define i64 @test_ceil_si64(half %x) nounwind {
 ; RV32IZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINXMIN-NEXT:    mv a0, s0
 ; RV32IZHINXMIN-NEXT:    call __fixsfdi
+; RV32IZHINXMIN-NEXT:    mv a2, s1
+; RV32IZHINXMIN-NEXT:    lui a4, 524288
 ; RV32IZHINXMIN-NEXT:    lui a3, 524288
-; RV32IZHINXMIN-NEXT:    lui a2, 524288
 ; RV32IZHINXMIN-NEXT:    beqz s1, .LBB5_4
 ; RV32IZHINXMIN-NEXT:  # %bb.3:
-; RV32IZHINXMIN-NEXT:    mv a2, a1
+; RV32IZHINXMIN-NEXT:    mv a3, a1
 ; RV32IZHINXMIN-NEXT:  .LBB5_4:
 ; RV32IZHINXMIN-NEXT:    lui a1, 389120
 ; RV32IZHINXMIN-NEXT:    addi a1, a1, -1
 ; RV32IZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINXMIN-NEXT:    beqz a1, .LBB5_6
 ; RV32IZHINXMIN-NEXT:  # %bb.5:
-; RV32IZHINXMIN-NEXT:    addi a2, a3, -1
+; RV32IZHINXMIN-NEXT:    addi a3, a4, -1
 ; RV32IZHINXMIN-NEXT:  .LBB5_6:
-; RV32IZHINXMIN-NEXT:    feq.s a3, s0, s0
-; RV32IZHINXMIN-NEXT:    neg a4, s1
-; RV32IZHINXMIN-NEXT:    and a0, a4, a0
+; RV32IZHINXMIN-NEXT:    feq.s a4, s0, s0
+; RV32IZHINXMIN-NEXT:    neg a2, a2
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    neg a1, a1
-; RV32IZHINXMIN-NEXT:    neg a3, a3
+; RV32IZHINXMIN-NEXT:    neg a2, a4
 ; RV32IZHINXMIN-NEXT:    or a0, a1, a0
-; RV32IZHINXMIN-NEXT:    and a1, a3, a2
-; RV32IZHINXMIN-NEXT:    and a0, a3, a0
+; RV32IZHINXMIN-NEXT:    and a1, a2, a3
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -1500,11 +1508,12 @@ define i64 @test_trunc_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IZFH-NEXT:    call __fixsfdi
+; RV32IZFH-NEXT:    mv a2, s0
+; RV32IZFH-NEXT:    lui a4, 524288
 ; RV32IZFH-NEXT:    lui a3, 524288
-; RV32IZFH-NEXT:    lui a2, 524288
 ; RV32IZFH-NEXT:    beqz s0, .LBB9_2
 ; RV32IZFH-NEXT:  # %bb.1:
-; RV32IZFH-NEXT:    mv a2, a1
+; RV32IZFH-NEXT:    mv a3, a1
 ; RV32IZFH-NEXT:  .LBB9_2:
 ; RV32IZFH-NEXT:    lui a1, 389120
 ; RV32IZFH-NEXT:    addi a1, a1, -1
@@ -1512,16 +1521,16 @@ define i64 @test_trunc_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFH-NEXT:    beqz a1, .LBB9_4
 ; RV32IZFH-NEXT:  # %bb.3:
-; RV32IZFH-NEXT:    addi a2, a3, -1
+; RV32IZFH-NEXT:    addi a3, a4, -1
 ; RV32IZFH-NEXT:  .LBB9_4:
-; RV32IZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFH-NEXT:    neg a4, s0
-; RV32IZFH-NEXT:    and a0, a4, a0
+; RV32IZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFH-NEXT:    neg a2, a2
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    neg a1, a1
-; RV32IZFH-NEXT:    neg a3, a3
+; RV32IZFH-NEXT:    neg a2, a4
 ; RV32IZFH-NEXT:    or a0, a1, a0
-; RV32IZFH-NEXT:    and a1, a3, a2
-; RV32IZFH-NEXT:    and a0, a3, a0
+; RV32IZFH-NEXT:    and a1, a2, a3
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -1548,27 +1557,28 @@ define i64 @test_trunc_si64(half %x) nounwind {
 ; RV32IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINX-NEXT:    mv a0, s0
 ; RV32IZHINX-NEXT:    call __fixsfdi
+; RV32IZHINX-NEXT:    mv a2, s1
+; RV32IZHINX-NEXT:    lui a4, 524288
 ; RV32IZHINX-NEXT:    lui a3, 524288
-; RV32IZHINX-NEXT:    lui a2, 524288
 ; RV32IZHINX-NEXT:    beqz s1, .LBB9_2
 ; RV32IZHINX-NEXT:  # %bb.1:
-; RV32IZHINX-NEXT:    mv a2, a1
+; RV32IZHINX-NEXT:    mv a3, a1
 ; RV32IZHINX-NEXT:  .LBB9_2:
 ; RV32IZHINX-NEXT:    lui a1, 389120
 ; RV32IZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINX-NEXT:    beqz a1, .LBB9_4
 ; RV32IZHINX-NEXT:  # %bb.3:
-; RV32IZHINX-NEXT:    addi a2, a3, -1
+; RV32IZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZHINX-NEXT:  .LBB9_4:
-; RV32IZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZHINX-NEXT:    neg a4, s1
-; RV32IZHINX-NEXT:    and a0, a4, a0
+; RV32IZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZHINX-NEXT:    neg a2, a2
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    neg a1, a1
-; RV32IZHINX-NEXT:    neg a3, a3
+; RV32IZHINX-NEXT:    neg a2, a4
 ; RV32IZHINX-NEXT:    or a0, a1, a0
-; RV32IZHINX-NEXT:    and a1, a3, a2
-; RV32IZHINX-NEXT:    and a0, a3, a0
+; RV32IZHINX-NEXT:    and a1, a2, a3
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -1596,11 +1606,12 @@ define i64 @test_trunc_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IZFHMIN-NEXT:    call __fixsfdi
+; RV32IZFHMIN-NEXT:    mv a2, s0
+; RV32IZFHMIN-NEXT:    lui a4, 524288
 ; RV32IZFHMIN-NEXT:    lui a3, 524288
-; RV32IZFHMIN-NEXT:    lui a2, 524288
 ; RV32IZFHMIN-NEXT:    beqz s0, .LBB9_2
 ; RV32IZFHMIN-NEXT:  # %bb.1:
-; RV32IZFHMIN-NEXT:    mv a2, a1
+; RV32IZFHMIN-NEXT:    mv a3, a1
 ; RV32IZFHMIN-NEXT:  .LBB9_2:
 ; RV32IZFHMIN-NEXT:    lui a1, 389120
 ; RV32IZFHMIN-NEXT:    addi a1, a1, -1
@@ -1608,16 +1619,16 @@ define i64 @test_trunc_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFHMIN-NEXT:    beqz a1, .LBB9_4
 ; RV32IZFHMIN-NEXT:  # %bb.3:
-; RV32IZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IZFHMIN-NEXT:  .LBB9_4:
-; RV32IZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFHMIN-NEXT:    neg a4, s0
-; RV32IZFHMIN-NEXT:    and a0, a4, a0
+; RV32IZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFHMIN-NEXT:    neg a2, a2
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    neg a1, a1
-; RV32IZFHMIN-NEXT:    neg a3, a3
+; RV32IZFHMIN-NEXT:    neg a2, a4
 ; RV32IZFHMIN-NEXT:    or a0, a1, a0
-; RV32IZFHMIN-NEXT:    and a1, a3, a2
-; RV32IZFHMIN-NEXT:    and a0, a3, a0
+; RV32IZFHMIN-NEXT:    and a1, a2, a3
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -1645,27 +1656,28 @@ define i64 @test_trunc_si64(half %x) nounwind {
 ; RV32IZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINXMIN-NEXT:    mv a0, s0
 ; RV32IZHINXMIN-NEXT:    call __fixsfdi
+; RV32IZHINXMIN-NEXT:    mv a2, s1
+; RV32IZHINXMIN-NEXT:    lui a4, 524288
 ; RV32IZHINXMIN-NEXT:    lui a3, 524288
-; RV32IZHINXMIN-NEXT:    lui a2, 524288
 ; RV32IZHINXMIN-NEXT:    beqz s1, .LBB9_2
 ; RV32IZHINXMIN-NEXT:  # %bb.1:
-; RV32IZHINXMIN-NEXT:    mv a2, a1
+; RV32IZHINXMIN-NEXT:    mv a3, a1
 ; RV32IZHINXMIN-NEXT:  .LBB9_2:
 ; RV32IZHINXMIN-NEXT:    lui a1, 389120
 ; RV32IZHINXMIN-NEXT:    addi a1, a1, -1
 ; RV32IZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINXMIN-NEXT:    beqz a1, .LBB9_4
 ; RV32IZHINXMIN-NEXT:  # %bb.3:
-; RV32IZHINXMIN-NEXT:    addi a2, a3, -1
+; RV32IZHINXMIN-NEXT:    addi a3, a4, -1
 ; RV32IZHINXMIN-NEXT:  .LBB9_4:
-; RV32IZHINXMIN-NEXT:    feq.s a3, s0, s0
-; RV32IZHINXMIN-NEXT:    neg a4, s1
-; RV32IZHINXMIN-NEXT:    and a0, a4, a0
+; RV32IZHINXMIN-NEXT:    feq.s a4, s0, s0
+; RV32IZHINXMIN-NEXT:    neg a2, a2
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    neg a1, a1
-; RV32IZHINXMIN-NEXT:    neg a3, a3
+; RV32IZHINXMIN-NEXT:    neg a2, a4
 ; RV32IZHINXMIN-NEXT:    or a0, a1, a0
-; RV32IZHINXMIN-NEXT:    and a1, a3, a2
-; RV32IZHINXMIN-NEXT:    and a0, a3, a0
+; RV32IZHINXMIN-NEXT:    and a1, a2, a3
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -2014,11 +2026,12 @@ define i64 @test_round_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IZFH-NEXT:    call __fixsfdi
+; RV32IZFH-NEXT:    mv a2, s0
+; RV32IZFH-NEXT:    lui a4, 524288
 ; RV32IZFH-NEXT:    lui a3, 524288
-; RV32IZFH-NEXT:    lui a2, 524288
 ; RV32IZFH-NEXT:    beqz s0, .LBB13_4
 ; RV32IZFH-NEXT:  # %bb.3:
-; RV32IZFH-NEXT:    mv a2, a1
+; RV32IZFH-NEXT:    mv a3, a1
 ; RV32IZFH-NEXT:  .LBB13_4:
 ; RV32IZFH-NEXT:    lui a1, 389120
 ; RV32IZFH-NEXT:    addi a1, a1, -1
@@ -2026,16 +2039,16 @@ define i64 @test_round_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFH-NEXT:    beqz a1, .LBB13_6
 ; RV32IZFH-NEXT:  # %bb.5:
-; RV32IZFH-NEXT:    addi a2, a3, -1
+; RV32IZFH-NEXT:    addi a3, a4, -1
 ; RV32IZFH-NEXT:  .LBB13_6:
-; RV32IZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFH-NEXT:    neg a4, s0
-; RV32IZFH-NEXT:    and a0, a4, a0
+; RV32IZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFH-NEXT:    neg a2, a2
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    neg a1, a1
-; RV32IZFH-NEXT:    neg a3, a3
+; RV32IZFH-NEXT:    neg a2, a4
 ; RV32IZFH-NEXT:    or a0, a1, a0
-; RV32IZFH-NEXT:    and a1, a3, a2
-; RV32IZFH-NEXT:    and a0, a3, a0
+; RV32IZFH-NEXT:    and a1, a2, a3
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -2072,27 +2085,28 @@ define i64 @test_round_si64(half %x) nounwind {
 ; RV32IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINX-NEXT:    mv a0, s0
 ; RV32IZHINX-NEXT:    call __fixsfdi
+; RV32IZHINX-NEXT:    mv a2, s1
+; RV32IZHINX-NEXT:    lui a4, 524288
 ; RV32IZHINX-NEXT:    lui a3, 524288
-; RV32IZHINX-NEXT:    lui a2, 524288
 ; RV32IZHINX-NEXT:    beqz s1, .LBB13_4
 ; RV32IZHINX-NEXT:  # %bb.3:
-; RV32IZHINX-NEXT:    mv a2, a1
+; RV32IZHINX-NEXT:    mv a3, a1
 ; RV32IZHINX-NEXT:  .LBB13_4:
 ; RV32IZHINX-NEXT:    lui a1, 389120
 ; RV32IZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINX-NEXT:    beqz a1, .LBB13_6
 ; RV32IZHINX-NEXT:  # %bb.5:
-; RV32IZHINX-NEXT:    addi a2, a3, -1
+; RV32IZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZHINX-NEXT:  .LBB13_6:
-; RV32IZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZHINX-NEXT:    neg a4, s1
-; RV32IZHINX-NEXT:    and a0, a4, a0
+; RV32IZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZHINX-NEXT:    neg a2, a2
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    neg a1, a1
-; RV32IZHINX-NEXT:    neg a3, a3
+; RV32IZHINX-NEXT:    neg a2, a4
 ; RV32IZHINX-NEXT:    or a0, a1, a0
-; RV32IZHINX-NEXT:    and a1, a3, a2
-; RV32IZHINX-NEXT:    and a0, a3, a0
+; RV32IZHINX-NEXT:    and a1, a2, a3
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -2142,11 +2156,12 @@ define i64 @test_round_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IZFHMIN-NEXT:    call __fixsfdi
+; RV32IZFHMIN-NEXT:    mv a2, s0
+; RV32IZFHMIN-NEXT:    lui a4, 524288
 ; RV32IZFHMIN-NEXT:    lui a3, 524288
-; RV32IZFHMIN-NEXT:    lui a2, 524288
 ; RV32IZFHMIN-NEXT:    beqz s0, .LBB13_4
 ; RV32IZFHMIN-NEXT:  # %bb.3:
-; RV32IZFHMIN-NEXT:    mv a2, a1
+; RV32IZFHMIN-NEXT:    mv a3, a1
 ; RV32IZFHMIN-NEXT:  .LBB13_4:
 ; RV32IZFHMIN-NEXT:    lui a1, 389120
 ; RV32IZFHMIN-NEXT:    addi a1, a1, -1
@@ -2154,16 +2169,16 @@ define i64 @test_round_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFHMIN-NEXT:    beqz a1, .LBB13_6
 ; RV32IZFHMIN-NEXT:  # %bb.5:
-; RV32IZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IZFHMIN-NEXT:  .LBB13_6:
-; RV32IZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFHMIN-NEXT:    neg a4, s0
-; RV32IZFHMIN-NEXT:    and a0, a4, a0
+; RV32IZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFHMIN-NEXT:    neg a2, a2
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    neg a1, a1
-; RV32IZFHMIN-NEXT:    neg a3, a3
+; RV32IZFHMIN-NEXT:    neg a2, a4
 ; RV32IZFHMIN-NEXT:    or a0, a1, a0
-; RV32IZFHMIN-NEXT:    and a1, a3, a2
-; RV32IZFHMIN-NEXT:    and a0, a3, a0
+; RV32IZFHMIN-NEXT:    and a1, a2, a3
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -2214,27 +2229,28 @@ define i64 @test_round_si64(half %x) nounwind {
 ; RV32IZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINXMIN-NEXT:    mv a0, s0
 ; RV32IZHINXMIN-NEXT:    call __fixsfdi
+; RV32IZHINXMIN-NEXT:    mv a2, s1
+; RV32IZHINXMIN-NEXT:    lui a4, 524288
 ; RV32IZHINXMIN-NEXT:    lui a3, 524288
-; RV32IZHINXMIN-NEXT:    lui a2, 524288
 ; RV32IZHINXMIN-NEXT:    beqz s1, .LBB13_4
 ; RV32IZHINXMIN-NEXT:  # %bb.3:
-; RV32IZHINXMIN-NEXT:    mv a2, a1
+; RV32IZHINXMIN-NEXT:    mv a3, a1
 ; RV32IZHINXMIN-NEXT:  .LBB13_4:
 ; RV32IZHINXMIN-NEXT:    lui a1, 389120
 ; RV32IZHINXMIN-NEXT:    addi a1, a1, -1
 ; RV32IZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINXMIN-NEXT:    beqz a1, .LBB13_6
 ; RV32IZHINXMIN-NEXT:  # %bb.5:
-; RV32IZHINXMIN-NEXT:    addi a2, a3, -1
+; RV32IZHINXMIN-NEXT:    addi a3, a4, -1
 ; RV32IZHINXMIN-NEXT:  .LBB13_6:
-; RV32IZHINXMIN-NEXT:    feq.s a3, s0, s0
-; RV32IZHINXMIN-NEXT:    neg a4, s1
-; RV32IZHINXMIN-NEXT:    and a0, a4, a0
+; RV32IZHINXMIN-NEXT:    feq.s a4, s0, s0
+; RV32IZHINXMIN-NEXT:    neg a2, a2
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    neg a1, a1
-; RV32IZHINXMIN-NEXT:    neg a3, a3
+; RV32IZHINXMIN-NEXT:    neg a2, a4
 ; RV32IZHINXMIN-NEXT:    or a0, a1, a0
-; RV32IZHINXMIN-NEXT:    and a1, a3, a2
-; RV32IZHINXMIN-NEXT:    and a0, a3, a0
+; RV32IZHINXMIN-NEXT:    and a1, a2, a3
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -2728,11 +2744,12 @@ define i64 @test_roundeven_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IZFH-NEXT:    call __fixsfdi
+; RV32IZFH-NEXT:    mv a2, s0
+; RV32IZFH-NEXT:    lui a4, 524288
 ; RV32IZFH-NEXT:    lui a3, 524288
-; RV32IZFH-NEXT:    lui a2, 524288
 ; RV32IZFH-NEXT:    beqz s0, .LBB17_4
 ; RV32IZFH-NEXT:  # %bb.3:
-; RV32IZFH-NEXT:    mv a2, a1
+; RV32IZFH-NEXT:    mv a3, a1
 ; RV32IZFH-NEXT:  .LBB17_4:
 ; RV32IZFH-NEXT:    lui a1, 389120
 ; RV32IZFH-NEXT:    addi a1, a1, -1
@@ -2740,16 +2757,16 @@ define i64 @test_roundeven_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFH-NEXT:    beqz a1, .LBB17_6
 ; RV32IZFH-NEXT:  # %bb.5:
-; RV32IZFH-NEXT:    addi a2, a3, -1
+; RV32IZFH-NEXT:    addi a3, a4, -1
 ; RV32IZFH-NEXT:  .LBB17_6:
-; RV32IZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFH-NEXT:    neg a4, s0
-; RV32IZFH-NEXT:    and a0, a4, a0
+; RV32IZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFH-NEXT:    neg a2, a2
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    neg a1, a1
-; RV32IZFH-NEXT:    neg a3, a3
+; RV32IZFH-NEXT:    neg a2, a4
 ; RV32IZFH-NEXT:    or a0, a1, a0
-; RV32IZFH-NEXT:    and a1, a3, a2
-; RV32IZFH-NEXT:    and a0, a3, a0
+; RV32IZFH-NEXT:    and a1, a2, a3
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -2786,27 +2803,28 @@ define i64 @test_roundeven_si64(half %x) nounwind {
 ; RV32IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINX-NEXT:    mv a0, s0
 ; RV32IZHINX-NEXT:    call __fixsfdi
+; RV32IZHINX-NEXT:    mv a2, s1
+; RV32IZHINX-NEXT:    lui a4, 524288
 ; RV32IZHINX-NEXT:    lui a3, 524288
-; RV32IZHINX-NEXT:    lui a2, 524288
 ; RV32IZHINX-NEXT:    beqz s1, .LBB17_4
 ; RV32IZHINX-NEXT:  # %bb.3:
-; RV32IZHINX-NEXT:    mv a2, a1
+; RV32IZHINX-NEXT:    mv a3, a1
 ; RV32IZHINX-NEXT:  .LBB17_4:
 ; RV32IZHINX-NEXT:    lui a1, 389120
 ; RV32IZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINX-NEXT:    beqz a1, .LBB17_6
 ; RV32IZHINX-NEXT:  # %bb.5:
-; RV32IZHINX-NEXT:    addi a2, a3, -1
+; RV32IZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZHINX-NEXT:  .LBB17_6:
-; RV32IZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZHINX-NEXT:    neg a4, s1
-; RV32IZHINX-NEXT:    and a0, a4, a0
+; RV32IZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZHINX-NEXT:    neg a2, a2
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    neg a1, a1
-; RV32IZHINX-NEXT:    neg a3, a3
+; RV32IZHINX-NEXT:    neg a2, a4
 ; RV32IZHINX-NEXT:    or a0, a1, a0
-; RV32IZHINX-NEXT:    and a1, a3, a2
-; RV32IZHINX-NEXT:    and a0, a3, a0
+; RV32IZHINX-NEXT:    and a1, a2, a3
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -2856,11 +2874,12 @@ define i64 @test_roundeven_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IZFHMIN-NEXT:    call __fixsfdi
+; RV32IZFHMIN-NEXT:    mv a2, s0
+; RV32IZFHMIN-NEXT:    lui a4, 524288
 ; RV32IZFHMIN-NEXT:    lui a3, 524288
-; RV32IZFHMIN-NEXT:    lui a2, 524288
 ; RV32IZFHMIN-NEXT:    beqz s0, .LBB17_4
 ; RV32IZFHMIN-NEXT:  # %bb.3:
-; RV32IZFHMIN-NEXT:    mv a2, a1
+; RV32IZFHMIN-NEXT:    mv a3, a1
 ; RV32IZFHMIN-NEXT:  .LBB17_4:
 ; RV32IZFHMIN-NEXT:    lui a1, 389120
 ; RV32IZFHMIN-NEXT:    addi a1, a1, -1
@@ -2868,16 +2887,16 @@ define i64 @test_roundeven_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFHMIN-NEXT:    beqz a1, .LBB17_6
 ; RV32IZFHMIN-NEXT:  # %bb.5:
-; RV32IZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IZFHMIN-NEXT:  .LBB17_6:
-; RV32IZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFHMIN-NEXT:    neg a4, s0
-; RV32IZFHMIN-NEXT:    and a0, a4, a0
+; RV32IZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFHMIN-NEXT:    neg a2, a2
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    neg a1, a1
-; RV32IZFHMIN-NEXT:    neg a3, a3
+; RV32IZFHMIN-NEXT:    neg a2, a4
 ; RV32IZFHMIN-NEXT:    or a0, a1, a0
-; RV32IZFHMIN-NEXT:    and a1, a3, a2
-; RV32IZFHMIN-NEXT:    and a0, a3, a0
+; RV32IZFHMIN-NEXT:    and a1, a2, a3
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -2928,27 +2947,28 @@ define i64 @test_roundeven_si64(half %x) nounwind {
 ; RV32IZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINXMIN-NEXT:    mv a0, s0
 ; RV32IZHINXMIN-NEXT:    call __fixsfdi
+; RV32IZHINXMIN-NEXT:    mv a2, s1
+; RV32IZHINXMIN-NEXT:    lui a4, 524288
 ; RV32IZHINXMIN-NEXT:    lui a3, 524288
-; RV32IZHINXMIN-NEXT:    lui a2, 524288
 ; RV32IZHINXMIN-NEXT:    beqz s1, .LBB17_4
 ; RV32IZHINXMIN-NEXT:  # %bb.3:
-; RV32IZHINXMIN-NEXT:    mv a2, a1
+; RV32IZHINXMIN-NEXT:    mv a3, a1
 ; RV32IZHINXMIN-NEXT:  .LBB17_4:
 ; RV32IZHINXMIN-NEXT:    lui a1, 389120
 ; RV32IZHINXMIN-NEXT:    addi a1, a1, -1
 ; RV32IZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINXMIN-NEXT:    beqz a1, .LBB17_6
 ; RV32IZHINXMIN-NEXT:  # %bb.5:
-; RV32IZHINXMIN-NEXT:    addi a2, a3, -1
+; RV32IZHINXMIN-NEXT:    addi a3, a4, -1
 ; RV32IZHINXMIN-NEXT:  .LBB17_6:
-; RV32IZHINXMIN-NEXT:    feq.s a3, s0, s0
-; RV32IZHINXMIN-NEXT:    neg a4, s1
-; RV32IZHINXMIN-NEXT:    and a0, a4, a0
+; RV32IZHINXMIN-NEXT:    feq.s a4, s0, s0
+; RV32IZHINXMIN-NEXT:    neg a2, a2
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    neg a1, a1
-; RV32IZHINXMIN-NEXT:    neg a3, a3
+; RV32IZHINXMIN-NEXT:    neg a2, a4
 ; RV32IZHINXMIN-NEXT:    or a0, a1, a0
-; RV32IZHINXMIN-NEXT:    and a1, a3, a2
-; RV32IZHINXMIN-NEXT:    and a0, a3, a0
+; RV32IZHINXMIN-NEXT:    and a1, a2, a3
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -3442,11 +3462,12 @@ define i64 @test_rint_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFH-NEXT:    fmv.s fa0, fs0
 ; RV32IZFH-NEXT:    call __fixsfdi
+; RV32IZFH-NEXT:    mv a2, s0
+; RV32IZFH-NEXT:    lui a4, 524288
 ; RV32IZFH-NEXT:    lui a3, 524288
-; RV32IZFH-NEXT:    lui a2, 524288
 ; RV32IZFH-NEXT:    beqz s0, .LBB21_4
 ; RV32IZFH-NEXT:  # %bb.3:
-; RV32IZFH-NEXT:    mv a2, a1
+; RV32IZFH-NEXT:    mv a3, a1
 ; RV32IZFH-NEXT:  .LBB21_4:
 ; RV32IZFH-NEXT:    lui a1, 389120
 ; RV32IZFH-NEXT:    addi a1, a1, -1
@@ -3454,16 +3475,16 @@ define i64 @test_rint_si64(half %x) nounwind {
 ; RV32IZFH-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFH-NEXT:    beqz a1, .LBB21_6
 ; RV32IZFH-NEXT:  # %bb.5:
-; RV32IZFH-NEXT:    addi a2, a3, -1
+; RV32IZFH-NEXT:    addi a3, a4, -1
 ; RV32IZFH-NEXT:  .LBB21_6:
-; RV32IZFH-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFH-NEXT:    neg a4, s0
-; RV32IZFH-NEXT:    and a0, a4, a0
+; RV32IZFH-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFH-NEXT:    neg a2, a2
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    neg a1, a1
-; RV32IZFH-NEXT:    neg a3, a3
+; RV32IZFH-NEXT:    neg a2, a4
 ; RV32IZFH-NEXT:    or a0, a1, a0
-; RV32IZFH-NEXT:    and a1, a3, a2
-; RV32IZFH-NEXT:    and a0, a3, a0
+; RV32IZFH-NEXT:    and a1, a2, a3
+; RV32IZFH-NEXT:    and a0, a2, a0
 ; RV32IZFH-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFH-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -3500,27 +3521,28 @@ define i64 @test_rint_si64(half %x) nounwind {
 ; RV32IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINX-NEXT:    mv a0, s0
 ; RV32IZHINX-NEXT:    call __fixsfdi
+; RV32IZHINX-NEXT:    mv a2, s1
+; RV32IZHINX-NEXT:    lui a4, 524288
 ; RV32IZHINX-NEXT:    lui a3, 524288
-; RV32IZHINX-NEXT:    lui a2, 524288
 ; RV32IZHINX-NEXT:    beqz s1, .LBB21_4
 ; RV32IZHINX-NEXT:  # %bb.3:
-; RV32IZHINX-NEXT:    mv a2, a1
+; RV32IZHINX-NEXT:    mv a3, a1
 ; RV32IZHINX-NEXT:  .LBB21_4:
 ; RV32IZHINX-NEXT:    lui a1, 389120
 ; RV32IZHINX-NEXT:    addi a1, a1, -1
 ; RV32IZHINX-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINX-NEXT:    beqz a1, .LBB21_6
 ; RV32IZHINX-NEXT:  # %bb.5:
-; RV32IZHINX-NEXT:    addi a2, a3, -1
+; RV32IZHINX-NEXT:    addi a3, a4, -1
 ; RV32IZHINX-NEXT:  .LBB21_6:
-; RV32IZHINX-NEXT:    feq.s a3, s0, s0
-; RV32IZHINX-NEXT:    neg a4, s1
-; RV32IZHINX-NEXT:    and a0, a4, a0
+; RV32IZHINX-NEXT:    feq.s a4, s0, s0
+; RV32IZHINX-NEXT:    neg a2, a2
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    neg a1, a1
-; RV32IZHINX-NEXT:    neg a3, a3
+; RV32IZHINX-NEXT:    neg a2, a4
 ; RV32IZHINX-NEXT:    or a0, a1, a0
-; RV32IZHINX-NEXT:    and a1, a3, a2
-; RV32IZHINX-NEXT:    and a0, a3, a0
+; RV32IZHINX-NEXT:    and a1, a2, a3
+; RV32IZHINX-NEXT:    and a0, a2, a0
 ; RV32IZHINX-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINX-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
@@ -3570,11 +3592,12 @@ define i64 @test_rint_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    fle.s s0, fa5, fs0
 ; RV32IZFHMIN-NEXT:    fmv.s fa0, fs0
 ; RV32IZFHMIN-NEXT:    call __fixsfdi
+; RV32IZFHMIN-NEXT:    mv a2, s0
+; RV32IZFHMIN-NEXT:    lui a4, 524288
 ; RV32IZFHMIN-NEXT:    lui a3, 524288
-; RV32IZFHMIN-NEXT:    lui a2, 524288
 ; RV32IZFHMIN-NEXT:    beqz s0, .LBB21_4
 ; RV32IZFHMIN-NEXT:  # %bb.3:
-; RV32IZFHMIN-NEXT:    mv a2, a1
+; RV32IZFHMIN-NEXT:    mv a3, a1
 ; RV32IZFHMIN-NEXT:  .LBB21_4:
 ; RV32IZFHMIN-NEXT:    lui a1, 389120
 ; RV32IZFHMIN-NEXT:    addi a1, a1, -1
@@ -3582,16 +3605,16 @@ define i64 @test_rint_si64(half %x) nounwind {
 ; RV32IZFHMIN-NEXT:    flt.s a1, fa5, fs0
 ; RV32IZFHMIN-NEXT:    beqz a1, .LBB21_6
 ; RV32IZFHMIN-NEXT:  # %bb.5:
-; RV32IZFHMIN-NEXT:    addi a2, a3, -1
+; RV32IZFHMIN-NEXT:    addi a3, a4, -1
 ; RV32IZFHMIN-NEXT:  .LBB21_6:
-; RV32IZFHMIN-NEXT:    feq.s a3, fs0, fs0
-; RV32IZFHMIN-NEXT:    neg a4, s0
-; RV32IZFHMIN-NEXT:    and a0, a4, a0
+; RV32IZFHMIN-NEXT:    feq.s a4, fs0, fs0
+; RV32IZFHMIN-NEXT:    neg a2, a2
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    neg a1, a1
-; RV32IZFHMIN-NEXT:    neg a3, a3
+; RV32IZFHMIN-NEXT:    neg a2, a4
 ; RV32IZFHMIN-NEXT:    or a0, a1, a0
-; RV32IZFHMIN-NEXT:    and a1, a3, a2
-; RV32IZFHMIN-NEXT:    and a0, a3, a0
+; RV32IZFHMIN-NEXT:    and a1, a2, a3
+; RV32IZFHMIN-NEXT:    and a0, a2, a0
 ; RV32IZFHMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZFHMIN-NEXT:    flw fs0, 4(sp) # 4-byte Folded Reload
@@ -3642,27 +3665,28 @@ define i64 @test_rint_si64(half %x) nounwind {
 ; RV32IZHINXMIN-NEXT:    fle.s s1, a0, s0
 ; RV32IZHINXMIN-NEXT:    mv a0, s0
 ; RV32IZHINXMIN-NEXT:    call __fixsfdi
+; RV32IZHINXMIN-NEXT:    mv a2, s1
+; RV32IZHINXMIN-NEXT:    lui a4, 524288
 ; RV32IZHINXMIN-NEXT:    lui a3, 524288
-; RV32IZHINXMIN-NEXT:    lui a2, 524288
 ; RV32IZHINXMIN-NEXT:    beqz s1, .LBB21_4
 ; RV32IZHINXMIN-NEXT:  # %bb.3:
-; RV32IZHINXMIN-NEXT:    mv a2, a1
+; RV32IZHINXMIN-NEXT:    mv a3, a1
 ; RV32IZHINXMIN-NEXT:  .LBB21_4:
 ; RV32IZHINXMIN-NEXT:    lui a1, 389120
 ; RV32IZHINXMIN-NEXT:    addi a1, a1, -1
 ; RV32IZHINXMIN-NEXT:    flt.s a1, a1, s0
 ; RV32IZHINXMIN-NEXT:    beqz a1, .LBB21_6
 ; RV32IZHINXMIN-NEXT:  # %bb.5:
-; RV32IZHINXMIN-NEXT:    addi a2, a3, -1
+; RV32IZHINXMIN-NEXT:    addi a3, a4, -1
 ; RV32IZHINXMIN-NEXT:  .LBB21_6:
-; RV32IZHINXMIN-NEXT:    feq.s a3, s0, s0
-; RV32IZHINXMIN-NEXT:    neg a4, s1
-; RV32IZHINXMIN-NEXT:    and a0, a4, a0
+; RV32IZHINXMIN-NEXT:    feq.s a4, s0, s0
+; RV32IZHINXMIN-NEXT:    neg a2, a2
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    neg a1, a1
-; RV32IZHINXMIN-NEXT:    neg a3, a3
+; RV32IZHINXMIN-NEXT:    neg a2, a4
 ; RV32IZHINXMIN-NEXT:    or a0, a1, a0
-; RV32IZHINXMIN-NEXT:    and a1, a3, a2
-; RV32IZHINXMIN-NEXT:    and a0, a3, a0
+; RV32IZHINXMIN-NEXT:    and a1, a2, a3
+; RV32IZHINXMIN-NEXT:    and a0, a2, a0
 ; RV32IZHINXMIN-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IZHINXMIN-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/idiv_large.ll b/llvm/test/CodeGen/RISCV/idiv_large.ll
index 82042aa44d6fcf..86ffbc257f824e 100644
--- a/llvm/test/CodeGen/RISCV/idiv_large.ll
+++ b/llvm/test/CodeGen/RISCV/idiv_large.ll
@@ -102,7 +102,6 @@ define i65 @udiv_i65(i65 %x, i65 %y) nounwind {
 ; RV32-NEXT:    sw s3, 80(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s4, 76(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s5, 72(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 68(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    slli a4, a2, 31
 ; RV32-NEXT:    li t5, 64
 ; RV32-NEXT:    bnez a4, .LBB1_5
@@ -138,20 +137,19 @@ define i65 @udiv_i65(i65 %x, i65 %y) nounwind {
 ; RV32-NEXT:    srli s0, a4, 24
 ; RV32-NEXT:  .LBB1_6: # %_udiv-special-cases
 ; RV32-NEXT:    lw a5, 0(a1)
-; RV32-NEXT:    lw a4, 4(a1)
-; RV32-NEXT:    lw s3, 8(a1)
+; RV32-NEXT:    lw s5, 4(a1)
+; RV32-NEXT:    lw s2, 8(a1)
 ; RV32-NEXT:    or t0, t4, t0
-; RV32-NEXT:    addi s2, s0, 64
 ; RV32-NEXT:    bnez t0, .LBB1_8
 ; RV32-NEXT:  # %bb.7: # %_udiv-special-cases
-; RV32-NEXT:    mv t3, s2
+; RV32-NEXT:    addi t3, s0, 64
 ; RV32-NEXT:  .LBB1_8: # %_udiv-special-cases
-; RV32-NEXT:    srli a1, a4, 1
-; RV32-NEXT:    slli t4, s3, 31
+; RV32-NEXT:    srli a1, s5, 1
+; RV32-NEXT:    slli t4, s2, 31
 ; RV32-NEXT:    or a1, t4, a1
-; RV32-NEXT:    slli t4, a4, 31
+; RV32-NEXT:    slli t4, s5, 31
 ; RV32-NEXT:    srli t6, a5, 1
-; RV32-NEXT:    snez s4, t0
+; RV32-NEXT:    snez s3, t0
 ; RV32-NEXT:    or t6, t6, t4
 ; RV32-NEXT:    bnez a1, .LBB1_10
 ; RV32-NEXT:  # %bb.9: # %_udiv-special-cases
@@ -211,28 +209,29 @@ define i65 @udiv_i65(i65 %x, i65 %y) nounwind {
 ; RV32-NEXT:    add t0, t0, t4
 ; RV32-NEXT:    srli s1, t0, 24
 ; RV32-NEXT:  .LBB1_11: # %_udiv-special-cases
-; RV32-NEXT:    andi t4, s3, 1
+; RV32-NEXT:    andi t4, s2, 1
 ; RV32-NEXT:    andi a6, a6, 1
 ; RV32-NEXT:    or t0, a2, a3
+; RV32-NEXT:    addi s2, s0, 64
 ; RV32-NEXT:    sltu s0, s2, s0
-; RV32-NEXT:    slli s3, a5, 31
-; RV32-NEXT:    or s2, a5, a4
-; RV32-NEXT:    addi s4, s4, -1
-; RV32-NEXT:    beqz s3, .LBB1_13
+; RV32-NEXT:    slli s4, a5, 31
+; RV32-NEXT:    or s2, a5, s5
+; RV32-NEXT:    addi s3, s3, -1
+; RV32-NEXT:    beqz s4, .LBB1_13
 ; RV32-NEXT:  # %bb.12:
-; RV32-NEXT:    srli t5, s3, 1
-; RV32-NEXT:    or t5, s3, t5
-; RV32-NEXT:    srli s3, t5, 2
-; RV32-NEXT:    or t5, t5, s3
-; RV32-NEXT:    srli s3, t5, 4
-; RV32-NEXT:    or t5, t5, s3
-; RV32-NEXT:    srli s3, t5, 8
-; RV32-NEXT:    or t5, t5, s3
-; RV32-NEXT:    srli s3, t5, 16
-; RV32-NEXT:    or t5, t5, s3
+; RV32-NEXT:    srli t5, s4, 1
+; RV32-NEXT:    or t5, s4, t5
+; RV32-NEXT:    srli s4, t5, 2
+; RV32-NEXT:    or t5, t5, s4
+; RV32-NEXT:    srli s4, t5, 4
+; RV32-NEXT:    or t5, t5, s4
+; RV32-NEXT:    srli s4, t5, 8
+; RV32-NEXT:    or t5, t5, s4
+; RV32-NEXT:    srli s4, t5, 16
+; RV32-NEXT:    or t5, t5, s4
 ; RV32-NEXT:    not t5, t5
-; RV32-NEXT:    srli s3, t5, 1
-; RV32-NEXT:    and t2, s3, t2
+; RV32-NEXT:    srli s4, t5, 1
+; RV32-NEXT:    and t2, s4, t2
 ; RV32-NEXT:    sub t2, t5, t2
 ; RV32-NEXT:    srli t5, t2, 2
 ; RV32-NEXT:    and t2, t2, a7
@@ -250,7 +249,7 @@ define i65 @udiv_i65(i65 %x, i65 %y) nounwind {
 ; RV32-NEXT:    or t1, t0, a6
 ; RV32-NEXT:    or a7, s2, t4
 ; RV32-NEXT:    or t2, t6, a1
-; RV32-NEXT:    and a1, s4, s0
+; RV32-NEXT:    and a1, s3, s0
 ; RV32-NEXT:    addi t6, t5, 64
 ; RV32-NEXT:    bnez t2, .LBB1_15
 ; RV32-NEXT:  # %bb.14: # %_udiv-special-cases
@@ -286,7 +285,7 @@ define i65 @udiv_i65(i65 %x, i65 %y) nounwind {
 ; RV32-NEXT:    or t6, t5, a1
 ; RV32-NEXT:    addi t5, t6, -1
 ; RV32-NEXT:    and a1, t4, t5
-; RV32-NEXT:    and t3, t5, a4
+; RV32-NEXT:    and t3, t5, s5
 ; RV32-NEXT:    and t5, t5, a5
 ; RV32-NEXT:    bnez t6, .LBB1_29
 ; RV32-NEXT:  # %bb.21: # %_udiv-special-cases
@@ -295,131 +294,135 @@ define i65 @udiv_i65(i65 %x, i65 %y) nounwind {
 ; RV32-NEXT:    or t6, t6, s0
 ; RV32-NEXT:    beqz t6, .LBB1_29
 ; RV32-NEXT:  # %bb.22: # %udiv-bb1
+; RV32-NEXT:    sw a6, 4(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    addi a1, a7, 1
-; RV32-NEXT:    li t3, 64
-; RV32-NEXT:    seqz t5, a1
-; RV32-NEXT:    sub t6, t3, a7
-; RV32-NEXT:    add t1, t1, t5
-; RV32-NEXT:    sw zero, 32(sp)
-; RV32-NEXT:    sw zero, 36(sp)
+; RV32-NEXT:    li a4, 64
+; RV32-NEXT:    seqz a6, a1
+; RV32-NEXT:    sub t5, a4, a7
+; RV32-NEXT:    add t1, t1, a6
 ; RV32-NEXT:    sw zero, 40(sp)
 ; RV32-NEXT:    sw zero, 44(sp)
-; RV32-NEXT:    srli t3, t6, 3
-; RV32-NEXT:    andi t3, t3, 12
-; RV32-NEXT:    addi t5, sp, 48
-; RV32-NEXT:    sw a5, 48(sp)
-; RV32-NEXT:    sw a4, 52(sp)
-; RV32-NEXT:    sw t4, 56(sp)
-; RV32-NEXT:    sub s4, t5, t3
-; RV32-NEXT:    or t5, a1, t1
-; RV32-NEXT:    lw s0, 0(s4)
-; RV32-NEXT:    seqz t3, t5
+; RV32-NEXT:    sw zero, 48(sp)
+; RV32-NEXT:    sw zero, 52(sp)
+; RV32-NEXT:    srli a4, t5, 3
+; RV32-NEXT:    andi a6, a4, 12
+; RV32-NEXT:    addi t3, sp, 56
+; RV32-NEXT:    mv a4, a5
+; RV32-NEXT:    sw a5, 56(sp)
+; RV32-NEXT:    sw s5, 60(sp)
+; RV32-NEXT:    mv a5, t4
+; RV32-NEXT:    sw t4, 64(sp)
+; RV32-NEXT:    sub s3, t3, a6
+; RV32-NEXT:    or a6, a1, t1
+; RV32-NEXT:    lw t4, 0(s3)
+; RV32-NEXT:    seqz t3, a6
 ; RV32-NEXT:    add t2, t2, t3
-; RV32-NEXT:    lw s2, 4(s4)
+; RV32-NEXT:    lw s1, 4(s3)
 ; RV32-NEXT:    andi t3, t2, 1
-; RV32-NEXT:    andi t2, t6, 31
-; RV32-NEXT:    xori s1, t2, 31
-; RV32-NEXT:    srli t2, s0, 1
-; RV32-NEXT:    srl t2, t2, s1
-; RV32-NEXT:    neg s3, a7
-; RV32-NEXT:    sll a7, s2, s3
-; RV32-NEXT:    or t5, t5, t3
+; RV32-NEXT:    andi t2, t5, 31
+; RV32-NEXT:    xori s0, t2, 31
+; RV32-NEXT:    srli t2, t4, 1
+; RV32-NEXT:    srl t2, t2, s0
+; RV32-NEXT:    neg s2, a7
+; RV32-NEXT:    sll a7, s1, s2
+; RV32-NEXT:    or a6, a6, t3
 ; RV32-NEXT:    or a7, a7, t2
-; RV32-NEXT:    sll t2, s0, s3
-; RV32-NEXT:    beqz t5, .LBB1_28
-; RV32-NEXT:  # %bb.23: # %udiv-preheader
+; RV32-NEXT:    sll t2, t4, s2
 ; RV32-NEXT:    li t5, 0
+; RV32-NEXT:    beqz a6, .LBB1_28
+; RV32-NEXT:  # %bb.23: # %udiv-preheader
 ; RV32-NEXT:    li t6, 0
-; RV32-NEXT:    li s0, 0
-; RV32-NEXT:    lw s4, 8(s4)
-; RV32-NEXT:    srli s5, a1, 3
-; RV32-NEXT:    sw zero, 16(sp)
-; RV32-NEXT:    sw zero, 20(sp)
+; RV32-NEXT:    lw s3, 8(s3)
+; RV32-NEXT:    srli a6, a1, 3
 ; RV32-NEXT:    sw zero, 24(sp)
 ; RV32-NEXT:    sw zero, 28(sp)
-; RV32-NEXT:    andi s5, s5, 12
-; RV32-NEXT:    mv s6, sp
-; RV32-NEXT:    sw a5, 0(sp)
-; RV32-NEXT:    sw a4, 4(sp)
-; RV32-NEXT:    sw t4, 8(sp)
-; RV32-NEXT:    sw zero, 12(sp)
-; RV32-NEXT:    add s5, s6, s5
-; RV32-NEXT:    lw a4, 4(s5)
-; RV32-NEXT:    srli a5, s2, 1
-; RV32-NEXT:    lw t4, 8(s5)
-; RV32-NEXT:    srl a5, a5, s1
-; RV32-NEXT:    sll s1, s4, s3
-; RV32-NEXT:    lw s2, 0(s5)
-; RV32-NEXT:    or a5, s1, a5
-; RV32-NEXT:    srl s1, a4, a1
-; RV32-NEXT:    andi s3, a1, 31
+; RV32-NEXT:    sw zero, 32(sp)
+; RV32-NEXT:    sw zero, 36(sp)
+; RV32-NEXT:    andi a6, a6, 12
+; RV32-NEXT:    addi t4, sp, 8
+; RV32-NEXT:    sw a4, 8(sp)
+; RV32-NEXT:    sw s5, 12(sp)
+; RV32-NEXT:    li a4, 0
+; RV32-NEXT:    sw a5, 16(sp)
+; RV32-NEXT:    sw zero, 20(sp)
+; RV32-NEXT:    add a6, t4, a6
+; RV32-NEXT:    lw a5, 4(a6)
+; RV32-NEXT:    srli s1, s1, 1
+; RV32-NEXT:    lw t4, 8(a6)
+; RV32-NEXT:    srl s0, s1, s0
+; RV32-NEXT:    sll s1, s3, s2
+; RV32-NEXT:    lw a6, 0(a6)
+; RV32-NEXT:    or s0, s1, s0
+; RV32-NEXT:    srl s1, a5, a1
+; RV32-NEXT:    andi s2, a1, 31
 ; RV32-NEXT:    slli t4, t4, 1
-; RV32-NEXT:    xori s3, s3, 31
-; RV32-NEXT:    sll t4, t4, s3
-; RV32-NEXT:    slli a4, a4, 1
+; RV32-NEXT:    xori s2, s2, 31
+; RV32-NEXT:    sll t4, t4, s2
+; RV32-NEXT:    slli a5, a5, 1
 ; RV32-NEXT:    or t4, s1, t4
-; RV32-NEXT:    sll a4, a4, s3
-; RV32-NEXT:    andi s1, a5, 1
-; RV32-NEXT:    srl a5, s2, a1
-; RV32-NEXT:    or s2, a5, a4
-; RV32-NEXT:    snez a4, t0
-; RV32-NEXT:    seqz a5, a2
-; RV32-NEXT:    add a6, a6, a4
-; RV32-NEXT:    sub a4, a3, a5
-; RV32-NEXT:    not a5, a6
-; RV32-NEXT:    andi a5, a5, 1
-; RV32-NEXT:    addi a6, a2, -1
+; RV32-NEXT:    sll a5, a5, s2
+; RV32-NEXT:    andi s0, s0, 1
+; RV32-NEXT:    srl s1, a6, a1
+; RV32-NEXT:    or s1, s1, a5
+; RV32-NEXT:    snez a6, t0
+; RV32-NEXT:    seqz t0, a2
+; RV32-NEXT:    lw a5, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a6, a5, a6
+; RV32-NEXT:    sub a5, a3, t0
+; RV32-NEXT:    not a6, a6
+; RV32-NEXT:    andi a6, a6, 1
+; RV32-NEXT:    addi t0, a2, -1
 ; RV32-NEXT:    j .LBB1_26
 ; RV32-NEXT:  .LBB1_24: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB1_26 Depth=1
-; RV32-NEXT:    sltu s1, a4, t0
+; RV32-NEXT:    sltu s0, a5, s2
 ; RV32-NEXT:  .LBB1_25: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB1_26 Depth=1
 ; RV32-NEXT:    srli t4, t4, 31
-; RV32-NEXT:    sub t4, a5, t4
-; RV32-NEXT:    sub t4, t4, s1
+; RV32-NEXT:    sub t4, a6, t4
+; RV32-NEXT:    sub t4, t4, s0
 ; RV32-NEXT:    slli t4, t4, 31
 ; RV32-NEXT:    srai t4, t4, 31
 ; RV32-NEXT:    and s3, t4, a3
-; RV32-NEXT:    srli s1, t2, 31
+; RV32-NEXT:    srli s0, t2, 31
 ; RV32-NEXT:    slli s4, a7, 1
 ; RV32-NEXT:    slli t2, t2, 1
-; RV32-NEXT:    or s1, s4, s1
+; RV32-NEXT:    or s0, s4, s0
 ; RV32-NEXT:    srli s4, a7, 31
 ; RV32-NEXT:    or t2, t5, t2
-; RV32-NEXT:    or a7, t6, s1
-; RV32-NEXT:    or s1, s0, s4
-; RV32-NEXT:    andi s1, s1, 1
-; RV32-NEXT:    and t6, t4, a2
-; RV32-NEXT:    sltu s0, s2, t6
-; RV32-NEXT:    sub t0, t0, s3
+; RV32-NEXT:    or a7, t6, s0
+; RV32-NEXT:    or s0, a4, s4
+; RV32-NEXT:    andi s0, s0, 1
+; RV32-NEXT:    and a4, t4, a2
+; RV32-NEXT:    sltu t6, s1, a4
+; RV32-NEXT:    sub s2, s2, s3
 ; RV32-NEXT:    andi t5, t4, 1
-; RV32-NEXT:    sub t4, t0, s0
-; RV32-NEXT:    sub s2, s2, t6
-; RV32-NEXT:    or t0, a1, t1
+; RV32-NEXT:    sub t4, s2, t6
+; RV32-NEXT:    sub s1, s1, a4
+; RV32-NEXT:    or a4, a1, t1
 ; RV32-NEXT:    seqz t6, a1
-; RV32-NEXT:    snez t0, t0
+; RV32-NEXT:    snez a4, a4
 ; RV32-NEXT:    sub t1, t1, t6
-; RV32-NEXT:    add t0, t3, t0
-; RV32-NEXT:    not t0, t0
+; RV32-NEXT:    add a4, t3, a4
+; RV32-NEXT:    not a4, a4
 ; RV32-NEXT:    addi a1, a1, -1
-; RV32-NEXT:    andi t3, t0, 1
-; RV32-NEXT:    or t0, a1, t1
-; RV32-NEXT:    or t0, t0, t3
+; RV32-NEXT:    andi t3, a4, 1
+; RV32-NEXT:    or a4, a1, t1
+; RV32-NEXT:    or s2, a4, t3
 ; RV32-NEXT:    li t6, 0
-; RV32-NEXT:    li s0, 0
-; RV32-NEXT:    beqz t0, .LBB1_28
+; RV32-NEXT:    li a4, 0
+; RV32-NEXT:    beqz s2, .LBB1_28
 ; RV32-NEXT:  .LBB1_26: # %udiv-do-while
 ; RV32-NEXT:    # =>This Inner Loop Header: Depth=1
-; RV32-NEXT:    srli t0, s2, 31
+; RV32-NEXT:    srli s2, s1, 31
 ; RV32-NEXT:    slli s3, t4, 1
-; RV32-NEXT:    or t0, s3, t0
-; RV32-NEXT:    slli s2, s2, 1
-; RV32-NEXT:    andi s1, s1, 1
-; RV32-NEXT:    or s2, s2, s1
-; RV32-NEXT:    bne a4, t0, .LBB1_24
+; RV32-NEXT:    or s2, s3, s2
+; RV32-NEXT:    slli s1, s1, 1
+; RV32-NEXT:    andi s0, s0, 1
+; RV32-NEXT:    or s1, s1, s0
+; RV32-NEXT:    bne a5, s2, .LBB1_24
 ; RV32-NEXT:  # %bb.27: # in Loop: Header=BB1_26 Depth=1
-; RV32-NEXT:    sltu s1, a6, s2
+; RV32-NEXT:    sltu s0, t0, s1
 ; RV32-NEXT:    j .LBB1_25
 ; RV32-NEXT:  .LBB1_28: # %udiv-loop-exit
 ; RV32-NEXT:    srli a1, t2, 31
@@ -438,7 +441,6 @@ define i65 @udiv_i65(i65 %x, i65 %y) nounwind {
 ; RV32-NEXT:    lw s3, 80(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s4, 76(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s5, 72(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 68(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    addi sp, sp, 96
 ; RV32-NEXT:    ret
 ;
@@ -810,162 +812,162 @@ define i128 @udiv_i128(i128 %x, i128 %y) nounwind {
 ; RV32-NEXT:    sw a1, 80(sp)
 ; RV32-NEXT:    sw a7, 84(sp)
 ; RV32-NEXT:    sub a4, a4, a3
-; RV32-NEXT:    lw s0, 4(a4)
-; RV32-NEXT:    lw a3, 8(a4)
+; RV32-NEXT:    lw a3, 4(a4)
+; RV32-NEXT:    lw s0, 8(a4)
 ; RV32-NEXT:    lw s1, 0(a4)
 ; RV32-NEXT:    lw a4, 12(a4)
 ; RV32-NEXT:    or s2, t1, t5
 ; RV32-NEXT:    andi a0, a0, 31
-; RV32-NEXT:    or s3, s2, t6
+; RV32-NEXT:    or s2, s2, t6
 ; RV32-NEXT:    xori a0, a0, 31
-; RV32-NEXT:    srli t6, a3, 1
+; RV32-NEXT:    srli t6, s0, 1
 ; RV32-NEXT:    not t4, t4
 ; RV32-NEXT:    srl t6, t6, a0
 ; RV32-NEXT:    sll a4, a4, t4
 ; RV32-NEXT:    or t6, a4, t6
-; RV32-NEXT:    srli a4, s0, 1
+; RV32-NEXT:    srli a4, a3, 1
 ; RV32-NEXT:    srl a4, a4, a0
-; RV32-NEXT:    sll a3, a3, t4
-; RV32-NEXT:    or a3, a3, a4
+; RV32-NEXT:    sll s0, s0, t4
+; RV32-NEXT:    or s5, s0, a4
 ; RV32-NEXT:    srli a4, s1, 1
 ; RV32-NEXT:    srl a0, a4, a0
-; RV32-NEXT:    sll a4, s0, t4
-; RV32-NEXT:    or s2, a4, a0
+; RV32-NEXT:    sll a3, a3, t4
+; RV32-NEXT:    or s4, a3, a0
 ; RV32-NEXT:    sll t4, s1, t4
 ; RV32-NEXT:    li a0, 0
-; RV32-NEXT:    beqz s3, .LBB2_31
+; RV32-NEXT:    beqz s2, .LBB2_31
 ; RV32-NEXT:  # %bb.24: # %udiv-preheader
+; RV32-NEXT:    li s2, 0
+; RV32-NEXT:    li s1, 0
 ; RV32-NEXT:    li s3, 0
-; RV32-NEXT:    li s4, 0
-; RV32-NEXT:    li s5, 0
-; RV32-NEXT:    srli a4, t1, 3
+; RV32-NEXT:    srli a3, t1, 3
 ; RV32-NEXT:    sw zero, 40(sp)
 ; RV32-NEXT:    sw zero, 44(sp)
 ; RV32-NEXT:    sw zero, 48(sp)
 ; RV32-NEXT:    sw zero, 52(sp)
-; RV32-NEXT:    andi a4, a4, 12
-; RV32-NEXT:    addi s0, sp, 24
+; RV32-NEXT:    andi a3, a3, 12
+; RV32-NEXT:    addi a4, sp, 24
 ; RV32-NEXT:    sw t0, 24(sp)
 ; RV32-NEXT:    sw a6, 28(sp)
 ; RV32-NEXT:    sw a1, 32(sp)
 ; RV32-NEXT:    sw a7, 36(sp)
-; RV32-NEXT:    add a4, s0, a4
-; RV32-NEXT:    lw a1, 12(a4)
-; RV32-NEXT:    lw a6, 8(a4)
-; RV32-NEXT:    lw a7, 4(a4)
-; RV32-NEXT:    lw a4, 0(a4)
-; RV32-NEXT:    andi t0, t1, 31
-; RV32-NEXT:    slli s0, a1, 1
-; RV32-NEXT:    xori t0, t0, 31
-; RV32-NEXT:    srl s1, a6, t1
-; RV32-NEXT:    sll s0, s0, t0
-; RV32-NEXT:    or s6, s1, s0
+; RV32-NEXT:    add a3, a4, a3
+; RV32-NEXT:    lw a1, 12(a3)
+; RV32-NEXT:    lw a4, 8(a3)
+; RV32-NEXT:    lw a6, 4(a3)
+; RV32-NEXT:    lw a3, 0(a3)
+; RV32-NEXT:    andi a7, t1, 31
+; RV32-NEXT:    slli t0, a1, 1
+; RV32-NEXT:    xori a7, a7, 31
+; RV32-NEXT:    srl s0, a4, t1
+; RV32-NEXT:    sll t0, t0, a7
+; RV32-NEXT:    or s6, s0, t0
+; RV32-NEXT:    slli a4, a4, 1
+; RV32-NEXT:    srl t0, a6, t1
+; RV32-NEXT:    sll a4, a4, a7
+; RV32-NEXT:    or s8, t0, a4
 ; RV32-NEXT:    slli a6, a6, 1
-; RV32-NEXT:    srl s0, a7, t1
-; RV32-NEXT:    sll a6, a6, t0
-; RV32-NEXT:    or s8, s0, a6
-; RV32-NEXT:    slli a7, a7, 1
-; RV32-NEXT:    srl a4, a4, t1
-; RV32-NEXT:    sll a6, a7, t0
-; RV32-NEXT:    or s9, a4, a6
+; RV32-NEXT:    srl a3, a3, t1
+; RV32-NEXT:    sll a4, a6, a7
+; RV32-NEXT:    or s9, a3, a4
 ; RV32-NEXT:    srl s7, a1, t1
 ; RV32-NEXT:    seqz a1, ra
-; RV32-NEXT:    or a4, ra, a5
+; RV32-NEXT:    or a3, ra, a5
 ; RV32-NEXT:    sw a5, 20(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sub a1, a5, a1
-; RV32-NEXT:    seqz a4, a4
-; RV32-NEXT:    sub a6, s10, a4
+; RV32-NEXT:    sub a6, a5, a1
+; RV32-NEXT:    seqz a1, a3
+; RV32-NEXT:    sub a7, s10, a1
 ; RV32-NEXT:    mv a2, s10
-; RV32-NEXT:    sltu a4, s10, a4
-; RV32-NEXT:    mv a7, s11
-; RV32-NEXT:    sub a4, s11, a4
-; RV32-NEXT:    sw a4, 16(sp) # 4-byte Folded Spill
-; RV32-NEXT:    addi a4, ra, -1
-; RV32-NEXT:    sw a4, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a1, s10, a1
+; RV32-NEXT:    mv t0, s11
+; RV32-NEXT:    sub a1, s11, a1
+; RV32-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    addi a1, ra, -1
+; RV32-NEXT:    sw a1, 12(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    j .LBB2_26
 ; RV32-NEXT:  .LBB2_25: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB2_26 Depth=1
-; RV32-NEXT:    srli s0, a3, 31
+; RV32-NEXT:    srli a3, s5, 31
 ; RV32-NEXT:    slli t6, t6, 1
-; RV32-NEXT:    srli a4, s2, 31
-; RV32-NEXT:    slli a3, a3, 1
-; RV32-NEXT:    or t6, t6, s0
-; RV32-NEXT:    or a3, a3, a4
-; RV32-NEXT:    srli a4, t4, 31
-; RV32-NEXT:    slli s2, s2, 1
-; RV32-NEXT:    or a4, s2, a4
+; RV32-NEXT:    srli a4, s4, 31
+; RV32-NEXT:    slli s5, s5, 1
+; RV32-NEXT:    or a3, t6, a3
+; RV32-NEXT:    or a4, s5, a4
+; RV32-NEXT:    srli t6, t4, 31
+; RV32-NEXT:    slli s4, s4, 1
+; RV32-NEXT:    or t6, s4, t6
 ; RV32-NEXT:    slli t4, t4, 1
 ; RV32-NEXT:    or t4, a0, t4
-; RV32-NEXT:    or s2, s3, a4
-; RV32-NEXT:    or a3, s4, a3
-; RV32-NEXT:    or t6, s5, t6
-; RV32-NEXT:    andi a0, s1, 1
-; RV32-NEXT:    and a4, s1, a2
-; RV32-NEXT:    and s1, s1, a7
-; RV32-NEXT:    sub s0, s7, a4
-; RV32-NEXT:    sltu a4, s7, a4
-; RV32-NEXT:    sub s1, s6, s1
-; RV32-NEXT:    sltu s3, s0, s8
-; RV32-NEXT:    sub s1, s1, a4
-; RV32-NEXT:    sub s6, s0, s8
-; RV32-NEXT:    sub s7, s1, s3
-; RV32-NEXT:    sub a4, s10, ra
-; RV32-NEXT:    or s0, t1, t2
-; RV32-NEXT:    sub s8, a4, t0
-; RV32-NEXT:    seqz a4, s0
-; RV32-NEXT:    seqz t0, t1
-; RV32-NEXT:    sltu s0, t5, a4
-; RV32-NEXT:    sub t5, t5, a4
-; RV32-NEXT:    sub t2, t2, t0
-; RV32-NEXT:    sub t3, t3, s0
+; RV32-NEXT:    or s4, s2, t6
+; RV32-NEXT:    or s5, s1, a4
+; RV32-NEXT:    or t6, s3, a3
+; RV32-NEXT:    andi a0, a1, 1
+; RV32-NEXT:    and a3, a1, a2
+; RV32-NEXT:    and a1, a1, t0
+; RV32-NEXT:    sub a4, s7, a3
+; RV32-NEXT:    sltu a3, s7, a3
+; RV32-NEXT:    sub a1, s6, a1
+; RV32-NEXT:    sltu s1, a4, s8
+; RV32-NEXT:    sub a1, a1, a3
+; RV32-NEXT:    sub s6, a4, s8
+; RV32-NEXT:    sub s7, a1, s1
+; RV32-NEXT:    sub a1, s10, ra
+; RV32-NEXT:    or a3, t1, t2
+; RV32-NEXT:    sub s8, a1, s0
+; RV32-NEXT:    seqz a1, a3
+; RV32-NEXT:    seqz a3, t1
+; RV32-NEXT:    sltu a4, t5, a1
+; RV32-NEXT:    sub t5, t5, a1
+; RV32-NEXT:    sub t2, t2, a3
+; RV32-NEXT:    sub t3, t3, a4
 ; RV32-NEXT:    addi t1, t1, -1
-; RV32-NEXT:    or a4, t2, t3
-; RV32-NEXT:    or t0, t1, t5
+; RV32-NEXT:    or a1, t2, t3
+; RV32-NEXT:    or a3, t1, t5
 ; RV32-NEXT:    sub s9, s9, s11
-; RV32-NEXT:    or a4, t0, a4
+; RV32-NEXT:    or a1, a3, a1
+; RV32-NEXT:    li s2, 0
+; RV32-NEXT:    li s1, 0
 ; RV32-NEXT:    li s3, 0
-; RV32-NEXT:    li s4, 0
-; RV32-NEXT:    li s5, 0
 ; RV32-NEXT:    mv ra, a5
-; RV32-NEXT:    beqz a4, .LBB2_31
+; RV32-NEXT:    beqz a1, .LBB2_31
 ; RV32-NEXT:  .LBB2_26: # %udiv-do-while
 ; RV32-NEXT:    # =>This Inner Loop Header: Depth=1
-; RV32-NEXT:    srli t0, s9, 31
-; RV32-NEXT:    slli s1, s8, 1
-; RV32-NEXT:    or s10, s1, t0
+; RV32-NEXT:    srli a1, s9, 31
+; RV32-NEXT:    slli s0, s8, 1
+; RV32-NEXT:    or s10, s0, a1
 ; RV32-NEXT:    slli s9, s9, 1
-; RV32-NEXT:    srli t0, t6, 31
-; RV32-NEXT:    or s9, s9, t0
-; RV32-NEXT:    beq a1, s10, .LBB2_28
+; RV32-NEXT:    srli a1, t6, 31
+; RV32-NEXT:    or s9, s9, a1
+; RV32-NEXT:    beq a6, s10, .LBB2_28
 ; RV32-NEXT:  # %bb.27: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB2_26 Depth=1
-; RV32-NEXT:    sltu t0, a1, s10
+; RV32-NEXT:    sltu a1, a6, s10
 ; RV32-NEXT:    j .LBB2_29
 ; RV32-NEXT:  .LBB2_28: # in Loop: Header=BB2_26 Depth=1
-; RV32-NEXT:    lw a4, 12(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu t0, a4, s9
+; RV32-NEXT:    lw a1, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a1, a1, s9
 ; RV32-NEXT:  .LBB2_29: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB2_26 Depth=1
-; RV32-NEXT:    srli s1, s6, 31
+; RV32-NEXT:    srli s0, s6, 31
 ; RV32-NEXT:    srli s8, s8, 31
 ; RV32-NEXT:    slli s6, s6, 1
 ; RV32-NEXT:    slli s11, s7, 1
 ; RV32-NEXT:    or s7, s6, s8
-; RV32-NEXT:    or s6, s11, s1
-; RV32-NEXT:    sub s1, a6, s7
-; RV32-NEXT:    sltu s8, a6, s7
-; RV32-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sub s11, a4, s6
-; RV32-NEXT:    sltu t0, s1, t0
-; RV32-NEXT:    sub s1, s11, s8
-; RV32-NEXT:    sub s1, s1, t0
-; RV32-NEXT:    srai s1, s1, 31
+; RV32-NEXT:    or s6, s11, s0
+; RV32-NEXT:    sub s0, a7, s7
+; RV32-NEXT:    sltu s8, a7, s7
+; RV32-NEXT:    lw a3, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sub s11, a3, s6
+; RV32-NEXT:    sltu a1, s0, a1
+; RV32-NEXT:    sub s0, s11, s8
+; RV32-NEXT:    sub a1, s0, a1
+; RV32-NEXT:    srai a1, a1, 31
 ; RV32-NEXT:    mv a5, ra
-; RV32-NEXT:    and s11, s1, ra
-; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
-; RV32-NEXT:    and ra, s1, a4
-; RV32-NEXT:    sltu t0, s9, s11
-; RV32-NEXT:    mv s8, t0
+; RV32-NEXT:    and s11, a1, ra
+; RV32-NEXT:    lw a3, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    and ra, a1, a3
+; RV32-NEXT:    sltu s0, s9, s11
+; RV32-NEXT:    mv s8, s0
 ; RV32-NEXT:    beq s10, ra, .LBB2_25
 ; RV32-NEXT:  # %bb.30: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB2_26 Depth=1
@@ -973,15 +975,15 @@ define i128 @udiv_i128(i128 %x, i128 %y) nounwind {
 ; RV32-NEXT:    j .LBB2_25
 ; RV32-NEXT:  .LBB2_31: # %udiv-loop-exit
 ; RV32-NEXT:    srli a1, t4, 31
-; RV32-NEXT:    slli a2, s2, 1
-; RV32-NEXT:    srli a4, s2, 31
+; RV32-NEXT:    slli a2, s4, 1
+; RV32-NEXT:    srli a3, s4, 31
 ; RV32-NEXT:    or t1, a2, a1
-; RV32-NEXT:    slli a1, a3, 1
-; RV32-NEXT:    or t5, a1, a4
-; RV32-NEXT:    srli s0, a3, 31
+; RV32-NEXT:    slli a1, s5, 1
+; RV32-NEXT:    or t5, a1, a3
+; RV32-NEXT:    srli a1, s5, 31
 ; RV32-NEXT:    slli t6, t6, 1
 ; RV32-NEXT:    slli t4, t4, 1
-; RV32-NEXT:    or s0, t6, s0
+; RV32-NEXT:    or s0, t6, a1
 ; RV32-NEXT:    or s1, a0, t4
 ; RV32-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
 ; RV32-NEXT:  .LBB2_32: # %udiv-end
@@ -1034,102 +1036,102 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:    sw s9, 180(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s10, 176(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s11, 172(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv s7, a0
+; RV32-NEXT:    mv s6, a0
 ; RV32-NEXT:    lw a5, 4(a2)
-; RV32-NEXT:    lw a4, 8(a2)
+; RV32-NEXT:    lw a3, 8(a2)
 ; RV32-NEXT:    lui a0, 349525
-; RV32-NEXT:    lui a3, 209715
+; RV32-NEXT:    lui a4, 209715
 ; RV32-NEXT:    lw t2, 16(a2)
 ; RV32-NEXT:    lw a6, 0(a2)
 ; RV32-NEXT:    lw a2, 12(a2)
-; RV32-NEXT:    sw a2, 24(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a4, 32(sp) # 4-byte Folded Spill
-; RV32-NEXT:    slli a2, a4, 31
-; RV32-NEXT:    srli a4, a5, 1
+; RV32-NEXT:    sw a2, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a3, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    slli a2, a3, 31
+; RV32-NEXT:    srli a3, a5, 1
 ; RV32-NEXT:    addi t5, a0, 1365
-; RV32-NEXT:    or a0, a4, a2
-; RV32-NEXT:    addi t4, a3, 819
-; RV32-NEXT:    lui a2, 61681
-; RV32-NEXT:    sw a5, 20(sp) # 4-byte Folded Spill
-; RV32-NEXT:    slli a3, a5, 31
-; RV32-NEXT:    sw a6, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    or a3, a3, a2
+; RV32-NEXT:    addi t4, a4, 819
+; RV32-NEXT:    lui a0, 61681
+; RV32-NEXT:    sw a5, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    slli a2, a5, 31
+; RV32-NEXT:    sw a6, 24(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    srli a4, a6, 1
-; RV32-NEXT:    addi t3, a2, -241
-; RV32-NEXT:    or a6, a4, a3
-; RV32-NEXT:    bnez a0, .LBB3_2
+; RV32-NEXT:    addi t3, a0, -241
+; RV32-NEXT:    or a6, a4, a2
+; RV32-NEXT:    bnez a3, .LBB3_2
 ; RV32-NEXT:  # %bb.1: # %_udiv-special-cases
-; RV32-NEXT:    srli a2, a6, 1
-; RV32-NEXT:    or a2, a6, a2
-; RV32-NEXT:    srli a3, a2, 2
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 8
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 16
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    not a2, a2
-; RV32-NEXT:    srli a3, a2, 1
-; RV32-NEXT:    and a3, a3, t5
-; RV32-NEXT:    sub a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 2
+; RV32-NEXT:    srli a0, a6, 1
+; RV32-NEXT:    or a0, a6, a0
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 8
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 16
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    not a0, a0
+; RV32-NEXT:    srli a2, a0, 1
+; RV32-NEXT:    and a2, a2, t5
+; RV32-NEXT:    sub a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    and a0, a0, t4
 ; RV32-NEXT:    and a2, a2, t4
-; RV32-NEXT:    and a3, a3, t4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    and a2, a2, t3
-; RV32-NEXT:    slli a3, a2, 8
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    slli a3, a2, 16
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli a2, a2, 24
-; RV32-NEXT:    addi t0, a2, 32
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    and a0, a0, t3
+; RV32-NEXT:    slli a2, a0, 8
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    slli a2, a0, 16
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli a0, a0, 24
+; RV32-NEXT:    addi t0, a0, 32
 ; RV32-NEXT:    j .LBB3_3
 ; RV32-NEXT:  .LBB3_2:
+; RV32-NEXT:    srli a0, a3, 1
+; RV32-NEXT:    or a0, a3, a0
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 8
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 16
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    not a0, a0
 ; RV32-NEXT:    srli a2, a0, 1
-; RV32-NEXT:    or a2, a0, a2
-; RV32-NEXT:    srli a3, a2, 2
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 8
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 16
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    not a2, a2
-; RV32-NEXT:    srli a3, a2, 1
-; RV32-NEXT:    and a3, a3, t5
-; RV32-NEXT:    sub a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 2
+; RV32-NEXT:    and a2, a2, t5
+; RV32-NEXT:    sub a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    and a0, a0, t4
 ; RV32-NEXT:    and a2, a2, t4
-; RV32-NEXT:    and a3, a3, t4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    and a2, a2, t3
-; RV32-NEXT:    slli a3, a2, 8
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    slli a3, a2, 16
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli t0, a2, 24
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    and a0, a0, t3
+; RV32-NEXT:    slli a2, a0, 8
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    slli a2, a0, 16
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli t0, a0, 24
 ; RV32-NEXT:  .LBB3_3: # %_udiv-special-cases
-; RV32-NEXT:    lw a5, 24(sp) # 4-byte Folded Reload
-; RV32-NEXT:    srli a2, a5, 1
+; RV32-NEXT:    lw a5, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    srli a0, a5, 1
 ; RV32-NEXT:    slli a4, t2, 31
 ; RV32-NEXT:    slli a5, a5, 31
-; RV32-NEXT:    lw a3, 28(sp) # 4-byte Folded Reload
-; RV32-NEXT:    slli a3, a3, 31
-; RV32-NEXT:    lw a7, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    slli a2, a2, 31
+; RV32-NEXT:    lw a7, 28(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    srli a7, a7, 1
 ; RV32-NEXT:    li s1, 64
-; RV32-NEXT:    bnez a3, .LBB3_5
+; RV32-NEXT:    bnez a2, .LBB3_5
 ; RV32-NEXT:  # %bb.4: # %_udiv-special-cases
 ; RV32-NEXT:    li t1, 64
 ; RV32-NEXT:    j .LBB3_6
 ; RV32-NEXT:  .LBB3_5:
-; RV32-NEXT:    srli t1, a3, 1
-; RV32-NEXT:    or t1, a3, t1
+; RV32-NEXT:    srli t1, a2, 1
+; RV32-NEXT:    or t1, a2, t1
 ; RV32-NEXT:    srli t6, t1, 2
 ; RV32-NEXT:    or t1, t1, t6
 ; RV32-NEXT:    srli t6, t1, 4
@@ -1155,25 +1157,25 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:    add t1, t1, t6
 ; RV32-NEXT:    srli t1, t1, 24
 ; RV32-NEXT:  .LBB3_6: # %_udiv-special-cases
-; RV32-NEXT:    or a2, a4, a2
-; RV32-NEXT:    or a4, a7, a5
-; RV32-NEXT:    bnez a3, .LBB3_8
+; RV32-NEXT:    or a4, a4, a0
+; RV32-NEXT:    or a5, a7, a5
+; RV32-NEXT:    bnez a2, .LBB3_8
 ; RV32-NEXT:  # %bb.7: # %_udiv-special-cases
 ; RV32-NEXT:    li t1, 128
 ; RV32-NEXT:  .LBB3_8: # %_udiv-special-cases
-; RV32-NEXT:    or a0, a0, a2
-; RV32-NEXT:    or a3, a6, a4
-; RV32-NEXT:    or a7, a4, a2
-; RV32-NEXT:    or a3, a3, a0
+; RV32-NEXT:    or a3, a3, a4
+; RV32-NEXT:    or a0, a6, a5
+; RV32-NEXT:    or a7, a5, a4
+; RV32-NEXT:    or a3, a0, a3
 ; RV32-NEXT:    addi a6, t0, 64
 ; RV32-NEXT:    addi a0, t1, 128
 ; RV32-NEXT:    bnez a3, .LBB3_11
 ; RV32-NEXT:  # %bb.9: # %_udiv-special-cases
 ; RV32-NEXT:    sltu t6, a0, t1
-; RV32-NEXT:    beqz a2, .LBB3_12
+; RV32-NEXT:    beqz a4, .LBB3_12
 ; RV32-NEXT:  .LBB3_10:
-; RV32-NEXT:    srli a4, a2, 1
-; RV32-NEXT:    or a2, a2, a4
+; RV32-NEXT:    srli a2, a4, 1
+; RV32-NEXT:    or a2, a4, a2
 ; RV32-NEXT:    srli a4, a2, 2
 ; RV32-NEXT:    or a2, a2, a4
 ; RV32-NEXT:    srli a4, a2, 4
@@ -1201,14 +1203,14 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:    beqz a7, .LBB3_13
 ; RV32-NEXT:    j .LBB3_14
 ; RV32-NEXT:  .LBB3_11:
-; RV32-NEXT:    snez a5, a7
+; RV32-NEXT:    snez a2, a7
 ; RV32-NEXT:    sltu t0, a6, t0
-; RV32-NEXT:    addi a5, a5, -1
-; RV32-NEXT:    and t6, a5, t0
-; RV32-NEXT:    bnez a2, .LBB3_10
+; RV32-NEXT:    addi a2, a2, -1
+; RV32-NEXT:    and t6, a2, t0
+; RV32-NEXT:    bnez a4, .LBB3_10
 ; RV32-NEXT:  .LBB3_12: # %_udiv-special-cases
-; RV32-NEXT:    srli a2, a4, 1
-; RV32-NEXT:    or a2, a4, a2
+; RV32-NEXT:    srli a2, a5, 1
+; RV32-NEXT:    or a2, a5, a2
 ; RV32-NEXT:    srli a4, a2, 2
 ; RV32-NEXT:    or a2, a2, a4
 ; RV32-NEXT:    srli a4, a2, 4
@@ -1248,71 +1250,71 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:    lw t1, 12(a1)
 ; RV32-NEXT:    lw a1, 16(a1)
 ; RV32-NEXT:    slli a0, a6, 31
-; RV32-NEXT:    srli a2, t0, 1
-; RV32-NEXT:    or a0, a2, a0
-; RV32-NEXT:    slli a2, t0, 31
-; RV32-NEXT:    srli a3, a7, 1
-; RV32-NEXT:    or s2, a3, a2
-; RV32-NEXT:    bnez a0, .LBB3_18
+; RV32-NEXT:    srli a3, t0, 1
+; RV32-NEXT:    or a3, a3, a0
+; RV32-NEXT:    slli a0, t0, 31
+; RV32-NEXT:    srli a2, a7, 1
+; RV32-NEXT:    or s2, a2, a0
+; RV32-NEXT:    bnez a3, .LBB3_18
 ; RV32-NEXT:  # %bb.17: # %_udiv-special-cases
-; RV32-NEXT:    srli a2, s2, 1
-; RV32-NEXT:    or a2, s2, a2
-; RV32-NEXT:    srli a3, a2, 2
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 8
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 16
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    not a2, a2
-; RV32-NEXT:    srli a3, a2, 1
-; RV32-NEXT:    and a3, a3, t5
-; RV32-NEXT:    sub a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 2
+; RV32-NEXT:    srli a0, s2, 1
+; RV32-NEXT:    or a0, s2, a0
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 8
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 16
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    not a0, a0
+; RV32-NEXT:    srli a2, a0, 1
+; RV32-NEXT:    and a2, a2, t5
+; RV32-NEXT:    sub a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    and a0, a0, t4
 ; RV32-NEXT:    and a2, a2, t4
-; RV32-NEXT:    and a3, a3, t4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    and a2, a2, t3
-; RV32-NEXT:    slli a3, a2, 8
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    slli a3, a2, 16
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli a2, a2, 24
-; RV32-NEXT:    addi s4, a2, 32
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    and a0, a0, t3
+; RV32-NEXT:    slli a2, a0, 8
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    slli a2, a0, 16
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli a0, a0, 24
+; RV32-NEXT:    addi s5, a0, 32
 ; RV32-NEXT:    j .LBB3_19
 ; RV32-NEXT:  .LBB3_18:
+; RV32-NEXT:    srli a0, a3, 1
+; RV32-NEXT:    or a0, a3, a0
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 8
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 16
+; RV32-NEXT:    or a0, a0, a2
+; RV32-NEXT:    not a0, a0
 ; RV32-NEXT:    srli a2, a0, 1
-; RV32-NEXT:    or a2, a0, a2
-; RV32-NEXT:    srli a3, a2, 2
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 8
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 16
-; RV32-NEXT:    or a2, a2, a3
-; RV32-NEXT:    not a2, a2
-; RV32-NEXT:    srli a3, a2, 1
-; RV32-NEXT:    and a3, a3, t5
-; RV32-NEXT:    sub a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 2
+; RV32-NEXT:    and a2, a2, t5
+; RV32-NEXT:    sub a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 2
+; RV32-NEXT:    and a0, a0, t4
 ; RV32-NEXT:    and a2, a2, t4
-; RV32-NEXT:    and a3, a3, t4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli a3, a2, 4
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    and a2, a2, t3
-; RV32-NEXT:    slli a3, a2, 8
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    slli a3, a2, 16
-; RV32-NEXT:    add a2, a2, a3
-; RV32-NEXT:    srli s4, a2, 24
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli a2, a0, 4
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    and a0, a0, t3
+; RV32-NEXT:    slli a2, a0, 8
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    slli a2, a0, 16
+; RV32-NEXT:    add a0, a0, a2
+; RV32-NEXT:    srli s5, a0, 24
 ; RV32-NEXT:  .LBB3_19: # %_udiv-special-cases
-; RV32-NEXT:    srli a2, t1, 1
-; RV32-NEXT:    slli a3, a1, 31
+; RV32-NEXT:    srli a0, t1, 1
+; RV32-NEXT:    slli a2, a1, 31
 ; RV32-NEXT:    slli a4, a7, 31
 ; RV32-NEXT:    slli a5, t1, 31
 ; RV32-NEXT:    srli s3, a6, 1
@@ -1320,46 +1322,46 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:  # %bb.20:
 ; RV32-NEXT:    srli s1, a4, 1
 ; RV32-NEXT:    or s1, a4, s1
-; RV32-NEXT:    srli s5, s1, 2
-; RV32-NEXT:    or s1, s1, s5
-; RV32-NEXT:    srli s5, s1, 4
-; RV32-NEXT:    or s1, s1, s5
-; RV32-NEXT:    srli s5, s1, 8
-; RV32-NEXT:    or s1, s1, s5
-; RV32-NEXT:    srli s5, s1, 16
-; RV32-NEXT:    or s1, s1, s5
+; RV32-NEXT:    srli s4, s1, 2
+; RV32-NEXT:    or s1, s1, s4
+; RV32-NEXT:    srli s4, s1, 4
+; RV32-NEXT:    or s1, s1, s4
+; RV32-NEXT:    srli s4, s1, 8
+; RV32-NEXT:    or s1, s1, s4
+; RV32-NEXT:    srli s4, s1, 16
+; RV32-NEXT:    or s1, s1, s4
 ; RV32-NEXT:    not s1, s1
-; RV32-NEXT:    srli s5, s1, 1
-; RV32-NEXT:    and s5, s5, t5
-; RV32-NEXT:    sub s1, s1, s5
-; RV32-NEXT:    srli s5, s1, 2
+; RV32-NEXT:    srli s4, s1, 1
+; RV32-NEXT:    and s4, s4, t5
+; RV32-NEXT:    sub s1, s1, s4
+; RV32-NEXT:    srli s4, s1, 2
 ; RV32-NEXT:    and s1, s1, t4
-; RV32-NEXT:    and s5, s5, t4
-; RV32-NEXT:    add s1, s1, s5
-; RV32-NEXT:    srli s5, s1, 4
-; RV32-NEXT:    add s1, s1, s5
+; RV32-NEXT:    and s4, s4, t4
+; RV32-NEXT:    add s1, s1, s4
+; RV32-NEXT:    srli s4, s1, 4
+; RV32-NEXT:    add s1, s1, s4
 ; RV32-NEXT:    and s1, s1, t3
-; RV32-NEXT:    slli s5, s1, 8
-; RV32-NEXT:    add s1, s1, s5
-; RV32-NEXT:    slli s5, s1, 16
-; RV32-NEXT:    add s1, s1, s5
+; RV32-NEXT:    slli s4, s1, 8
+; RV32-NEXT:    add s1, s1, s4
+; RV32-NEXT:    slli s4, s1, 16
+; RV32-NEXT:    add s1, s1, s4
 ; RV32-NEXT:    srli s1, s1, 24
 ; RV32-NEXT:  .LBB3_21: # %_udiv-special-cases
-; RV32-NEXT:    or a2, a3, a2
+; RV32-NEXT:    or a2, a2, a0
 ; RV32-NEXT:    or a5, s3, a5
 ; RV32-NEXT:    bnez a4, .LBB3_23
 ; RV32-NEXT:  # %bb.22: # %_udiv-special-cases
 ; RV32-NEXT:    li s1, 128
 ; RV32-NEXT:  .LBB3_23: # %_udiv-special-cases
-; RV32-NEXT:    or a0, a0, a2
-; RV32-NEXT:    or a3, s2, a5
-; RV32-NEXT:    or s5, a5, a2
-; RV32-NEXT:    or s3, a3, a0
-; RV32-NEXT:    addi a3, s4, 64
+; RV32-NEXT:    or a3, a3, a2
+; RV32-NEXT:    or a0, s2, a5
+; RV32-NEXT:    or s4, a5, a2
+; RV32-NEXT:    or s3, a0, a3
+; RV32-NEXT:    addi a0, s5, 64
 ; RV32-NEXT:    addi s2, s1, 128
 ; RV32-NEXT:    bnez s3, .LBB3_26
 ; RV32-NEXT:  # %bb.24: # %_udiv-special-cases
-; RV32-NEXT:    sltu a0, s2, s1
+; RV32-NEXT:    sltu a3, s2, s1
 ; RV32-NEXT:    beqz a2, .LBB3_27
 ; RV32-NEXT:  .LBB3_25:
 ; RV32-NEXT:    srli a4, a2, 1
@@ -1390,10 +1392,10 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:    srli a4, a2, 24
 ; RV32-NEXT:    j .LBB3_28
 ; RV32-NEXT:  .LBB3_26:
-; RV32-NEXT:    snez a0, s5
-; RV32-NEXT:    sltu a4, a3, s4
-; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    and a0, a0, a4
+; RV32-NEXT:    snez a3, s4
+; RV32-NEXT:    sltu a4, a0, s5
+; RV32-NEXT:    addi a3, a3, -1
+; RV32-NEXT:    and a3, a3, a4
 ; RV32-NEXT:    bnez a2, .LBB3_25
 ; RV32-NEXT:  .LBB3_27: # %_udiv-special-cases
 ; RV32-NEXT:    srli a2, a5, 1
@@ -1424,140 +1426,142 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:    srli a2, a2, 24
 ; RV32-NEXT:    addi a4, a2, 32
 ; RV32-NEXT:  .LBB3_28: # %_udiv-special-cases
-; RV32-NEXT:    andi s11, a1, 1
-; RV32-NEXT:    andi s4, t2, 1
-; RV32-NEXT:    lw a1, 32(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a5, 28(sp) # 4-byte Folded Reload
-; RV32-NEXT:    or a5, a5, a1
-; RV32-NEXT:    or a1, a7, a6
-; RV32-NEXT:    bnez s5, .LBB3_30
+; RV32-NEXT:    andi s1, a1, 1
+; RV32-NEXT:    andi a1, t2, 1
+; RV32-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    or s5, a5, a2
+; RV32-NEXT:    or a5, a7, a6
+; RV32-NEXT:    bnez s4, .LBB3_30
 ; RV32-NEXT:  # %bb.29: # %_udiv-special-cases
-; RV32-NEXT:    mv a4, a3
+; RV32-NEXT:    mv a4, a0
 ; RV32-NEXT:  .LBB3_30: # %_udiv-special-cases
+; RV32-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
-; RV32-NEXT:    or s6, a2, a3
-; RV32-NEXT:    or t2, a5, s4
+; RV32-NEXT:    or t2, a0, a2
+; RV32-NEXT:    or a0, s5, a1
 ; RV32-NEXT:    or a2, t0, t1
-; RV32-NEXT:    or a3, a1, s11
+; RV32-NEXT:    or a5, a5, s1
 ; RV32-NEXT:    bnez s3, .LBB3_32
 ; RV32-NEXT:  # %bb.31: # %_udiv-special-cases
 ; RV32-NEXT:    mv a4, s2
 ; RV32-NEXT:  .LBB3_32: # %_udiv-special-cases
-; RV32-NEXT:    or a1, t2, s6
-; RV32-NEXT:    sltu t2, s0, a4
-; RV32-NEXT:    or a2, a3, a2
-; RV32-NEXT:    mv t4, t2
-; RV32-NEXT:    beq t6, a0, .LBB3_34
+; RV32-NEXT:    or a0, a0, t2
+; RV32-NEXT:    sltu t4, s0, a4
+; RV32-NEXT:    or a5, a5, a2
+; RV32-NEXT:    mv t5, t4
+; RV32-NEXT:    beq t6, a3, .LBB3_34
 ; RV32-NEXT:  # %bb.33: # %_udiv-special-cases
-; RV32-NEXT:    sltu t4, t6, a0
+; RV32-NEXT:    sltu t5, t6, a3
 ; RV32-NEXT:  .LBB3_34: # %_udiv-special-cases
-; RV32-NEXT:    seqz a1, a1
-; RV32-NEXT:    seqz a2, a2
-; RV32-NEXT:    sub a0, t6, a0
-; RV32-NEXT:    snez t3, t4
-; RV32-NEXT:    neg a3, t4
+; RV32-NEXT:    seqz a2, a0
+; RV32-NEXT:    seqz a5, a5
+; RV32-NEXT:    sub a3, t6, a3
+; RV32-NEXT:    snez t3, t5
+; RV32-NEXT:    neg a0, t5
 ; RV32-NEXT:    neg t3, t3
-; RV32-NEXT:    sub t2, a0, t2
-; RV32-NEXT:    or t6, a3, t3
-; RV32-NEXT:    andi t5, t4, 1
-; RV32-NEXT:    sub a0, s0, a4
-; RV32-NEXT:    beqz t6, .LBB3_36
+; RV32-NEXT:    sub t4, a3, t4
+; RV32-NEXT:    or a3, a0, t3
+; RV32-NEXT:    andi t6, t5, 1
+; RV32-NEXT:    sub t5, s0, a4
+; RV32-NEXT:    beqz a3, .LBB3_36
 ; RV32-NEXT:  # %bb.35: # %_udiv-special-cases
-; RV32-NEXT:    snez a4, t6
-; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    bnez t5, .LBB3_37
+; RV32-NEXT:    snez a3, a3
+; RV32-NEXT:    or a2, a2, a5
+; RV32-NEXT:    bnez t6, .LBB3_37
 ; RV32-NEXT:    j .LBB3_38
 ; RV32-NEXT:  .LBB3_36:
-; RV32-NEXT:    sltiu a4, a0, 129
-; RV32-NEXT:    snez t4, t2
-; RV32-NEXT:    xori a4, a4, 1
-; RV32-NEXT:    or a4, a4, t4
-; RV32-NEXT:    or a1, a1, a2
-; RV32-NEXT:    beqz t5, .LBB3_38
+; RV32-NEXT:    sltiu a3, t5, 129
+; RV32-NEXT:    snez a4, t4
+; RV32-NEXT:    xori a3, a3, 1
+; RV32-NEXT:    or a3, a3, a4
+; RV32-NEXT:    or a2, a2, a5
+; RV32-NEXT:    beqz t6, .LBB3_38
 ; RV32-NEXT:  .LBB3_37: # %_udiv-special-cases
-; RV32-NEXT:    mv a4, t5
+; RV32-NEXT:    mv a3, t6
 ; RV32-NEXT:  .LBB3_38: # %_udiv-special-cases
-; RV32-NEXT:    or s0, a1, a4
-; RV32-NEXT:    addi a2, s0, -1
-; RV32-NEXT:    and t6, s11, a2
-; RV32-NEXT:    and t4, a2, t1
-; RV32-NEXT:    and a1, a2, a6
-; RV32-NEXT:    and a4, a2, t0
+; RV32-NEXT:    or a3, a2, a3
+; RV32-NEXT:    addi a2, a3, -1
+; RV32-NEXT:    and s3, s1, a2
+; RV32-NEXT:    and s2, a2, t1
+; RV32-NEXT:    and a4, a2, a6
+; RV32-NEXT:    and s0, a2, t0
 ; RV32-NEXT:    and a2, a2, a7
-; RV32-NEXT:    bnez s0, .LBB3_55
+; RV32-NEXT:    bnez a3, .LBB3_55
 ; RV32-NEXT:  # %bb.39: # %_udiv-special-cases
-; RV32-NEXT:    xori s0, a0, 128
-; RV32-NEXT:    or s1, t5, a3
-; RV32-NEXT:    or s2, t2, t3
-; RV32-NEXT:    or s0, s0, s1
-; RV32-NEXT:    or s0, s0, s2
-; RV32-NEXT:    beqz s0, .LBB3_55
+; RV32-NEXT:    mv a3, t2
+; RV32-NEXT:    sw a1, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    xori a1, t5, 128
+; RV32-NEXT:    or a5, t6, a0
+; RV32-NEXT:    or t2, t4, t3
+; RV32-NEXT:    or a1, a1, a5
+; RV32-NEXT:    or a1, a1, t2
+; RV32-NEXT:    beqz a1, .LBB3_55
 ; RV32-NEXT:  # %bb.40: # %udiv-bb1
-; RV32-NEXT:    sw s4, 36(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s7, 0(sp) # 4-byte Folded Spill
-; RV32-NEXT:    addi a1, a0, 1
-; RV32-NEXT:    seqz a2, a1
-; RV32-NEXT:    add t2, t2, a2
+; RV32-NEXT:    sw s6, 0(sp) # 4-byte Folded Spill
+; RV32-NEXT:    addi a1, t5, 1
+; RV32-NEXT:    seqz t2, a1
+; RV32-NEXT:    add t2, t4, t2
 ; RV32-NEXT:    or a2, a1, t2
 ; RV32-NEXT:    seqz t4, a2
 ; RV32-NEXT:    sw zero, 120(sp)
 ; RV32-NEXT:    sw zero, 124(sp)
 ; RV32-NEXT:    sw zero, 128(sp)
 ; RV32-NEXT:    sw zero, 132(sp)
-; RV32-NEXT:    add t4, a3, t4
-; RV32-NEXT:    sltu a2, t4, a3
+; RV32-NEXT:    add t4, a0, t4
+; RV32-NEXT:    sltu a0, t4, a0
 ; RV32-NEXT:    sw zero, 104(sp)
 ; RV32-NEXT:    sw zero, 108(sp)
 ; RV32-NEXT:    sw zero, 112(sp)
 ; RV32-NEXT:    sw zero, 116(sp)
-; RV32-NEXT:    add t3, t3, a2
-; RV32-NEXT:    or a2, a1, t4
-; RV32-NEXT:    or a3, t2, t3
+; RV32-NEXT:    add t3, t3, a0
+; RV32-NEXT:    or a0, a1, t4
+; RV32-NEXT:    or a2, t2, t3
 ; RV32-NEXT:    li a4, 128
-; RV32-NEXT:    or t6, a2, a3
-; RV32-NEXT:    sub a3, a4, a0
+; RV32-NEXT:    or a5, a0, a2
+; RV32-NEXT:    sub a0, a4, t5
 ; RV32-NEXT:    sw a7, 136(sp)
 ; RV32-NEXT:    sw t0, 140(sp)
 ; RV32-NEXT:    sw a6, 144(sp)
 ; RV32-NEXT:    sw t1, 148(sp)
-; RV32-NEXT:    srli a2, a3, 3
+; RV32-NEXT:    srli a2, a0, 3
 ; RV32-NEXT:    andi a2, a2, 28
 ; RV32-NEXT:    addi a4, sp, 136
-; RV32-NEXT:    sw s11, 152(sp)
+; RV32-NEXT:    sw s1, 152(sp)
 ; RV32-NEXT:    sub a4, a4, a2
-; RV32-NEXT:    lw s1, 4(a4)
-; RV32-NEXT:    lw s0, 8(a4)
-; RV32-NEXT:    lw s4, 0(a4)
+; RV32-NEXT:    lw s0, 4(a4)
+; RV32-NEXT:    lw s2, 8(a4)
+; RV32-NEXT:    lw s3, 0(a4)
 ; RV32-NEXT:    lw a2, 12(a4)
-; RV32-NEXT:    seqz s2, t6
-; RV32-NEXT:    andi a3, a3, 31
-; RV32-NEXT:    add s2, t5, s2
-; RV32-NEXT:    xori a3, a3, 31
-; RV32-NEXT:    srli t5, s0, 1
-; RV32-NEXT:    neg a0, a0
-; RV32-NEXT:    srl t5, t5, a3
-; RV32-NEXT:    sll s3, a2, a0
-; RV32-NEXT:    or t5, s3, t5
-; RV32-NEXT:    srli s3, s1, 1
-; RV32-NEXT:    srl s3, s3, a3
-; RV32-NEXT:    sll s0, s0, a0
-; RV32-NEXT:    or s0, s0, s3
-; RV32-NEXT:    srli s5, s4, 1
-; RV32-NEXT:    andi s3, s2, 1
-; RV32-NEXT:    srl s2, s5, a3
-; RV32-NEXT:    sll s1, s1, a0
-; RV32-NEXT:    or s5, t6, s3
-; RV32-NEXT:    or s2, s1, s2
-; RV32-NEXT:    sll t6, s4, a0
-; RV32-NEXT:    beqz s5, .LBB3_53
+; RV32-NEXT:    seqz s4, a5
+; RV32-NEXT:    andi a0, a0, 31
+; RV32-NEXT:    add t6, t6, s4
+; RV32-NEXT:    xori a0, a0, 31
+; RV32-NEXT:    srli s4, s2, 1
+; RV32-NEXT:    neg s8, t5
+; RV32-NEXT:    srl t5, s4, a0
+; RV32-NEXT:    sll s4, a2, s8
+; RV32-NEXT:    or t5, s4, t5
+; RV32-NEXT:    srli s4, s0, 1
+; RV32-NEXT:    srl s4, s4, a0
+; RV32-NEXT:    sll s2, s2, s8
+; RV32-NEXT:    or s7, s2, s4
+; RV32-NEXT:    srli s2, s3, 1
+; RV32-NEXT:    andi t6, t6, 1
+; RV32-NEXT:    srl s2, s2, a0
+; RV32-NEXT:    sll s0, s0, s8
+; RV32-NEXT:    mv s9, t6
+; RV32-NEXT:    or a5, a5, t6
+; RV32-NEXT:    or s6, s0, s2
+; RV32-NEXT:    sll t6, s3, s8
+; RV32-NEXT:    beqz a5, .LBB3_53
 ; RV32-NEXT:  # %bb.41: # %udiv-preheader
-; RV32-NEXT:    li s7, 0
-; RV32-NEXT:    li s8, 0
-; RV32-NEXT:    li s9, 0
-; RV32-NEXT:    li s10, 0
-; RV32-NEXT:    srli s5, a2, 1
-; RV32-NEXT:    lw ra, 16(a4)
+; RV32-NEXT:    sw zero, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    li s0, 0
+; RV32-NEXT:    li s2, 0
+; RV32-NEXT:    li s3, 0
+; RV32-NEXT:    srli s10, a2, 1
+; RV32-NEXT:    lw s11, 16(a4)
 ; RV32-NEXT:    sw zero, 88(sp)
 ; RV32-NEXT:    sw zero, 92(sp)
 ; RV32-NEXT:    sw zero, 96(sp)
@@ -1567,224 +1571,221 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV32-NEXT:    sw zero, 80(sp)
 ; RV32-NEXT:    sw zero, 84(sp)
 ; RV32-NEXT:    srli a2, a1, 3
-; RV32-NEXT:    sw s11, 56(sp)
+; RV32-NEXT:    sw s1, 56(sp)
 ; RV32-NEXT:    sw zero, 60(sp)
 ; RV32-NEXT:    sw zero, 64(sp)
 ; RV32-NEXT:    sw zero, 68(sp)
 ; RV32-NEXT:    andi a2, a2, 28
 ; RV32-NEXT:    addi a4, sp, 40
 ; RV32-NEXT:    sw a7, 40(sp)
+; RV32-NEXT:    li s1, 0
 ; RV32-NEXT:    sw t0, 44(sp)
 ; RV32-NEXT:    sw a6, 48(sp)
 ; RV32-NEXT:    sw t1, 52(sp)
 ; RV32-NEXT:    add a2, a4, a2
 ; RV32-NEXT:    lw a4, 16(a2)
-; RV32-NEXT:    lw a6, 4(a2)
+; RV32-NEXT:    lw a5, 4(a2)
 ; RV32-NEXT:    lw a7, 8(a2)
 ; RV32-NEXT:    lw t0, 12(a2)
 ; RV32-NEXT:    lw a2, 0(a2)
-; RV32-NEXT:    srl a3, s5, a3
-; RV32-NEXT:    sll a0, ra, a0
-; RV32-NEXT:    andi t1, a1, 31
+; RV32-NEXT:    srl a0, s10, a0
+; RV32-NEXT:    sll t1, s11, s8
+; RV32-NEXT:    andi s8, a1, 31
 ; RV32-NEXT:    slli a4, a4, 1
-; RV32-NEXT:    xori t1, t1, 31
-; RV32-NEXT:    srl s1, t0, a1
-; RV32-NEXT:    sll a4, a4, t1
-; RV32-NEXT:    or a3, a0, a3
-; RV32-NEXT:    or s11, s1, a4
-; RV32-NEXT:    li s1, 0
-; RV32-NEXT:    srl a0, a7, a1
+; RV32-NEXT:    xori s4, s8, 31
+; RV32-NEXT:    srl s8, t0, a1
+; RV32-NEXT:    sll a4, a4, s4
+; RV32-NEXT:    or a0, t1, a0
+; RV32-NEXT:    or s8, s8, a4
+; RV32-NEXT:    srl a4, a7, a1
 ; RV32-NEXT:    slli t0, t0, 1
-; RV32-NEXT:    sll a4, t0, t1
+; RV32-NEXT:    sll t0, t0, s4
 ; RV32-NEXT:    slli a7, a7, 1
-; RV32-NEXT:    srl t0, a6, a1
-; RV32-NEXT:    sll a7, a7, t1
-; RV32-NEXT:    or a0, a0, a4
-; RV32-NEXT:    or s5, t0, a7
-; RV32-NEXT:    andi ra, a3, 1
-; RV32-NEXT:    or a3, a5, s6
+; RV32-NEXT:    srl t1, a5, a1
+; RV32-NEXT:    sll a7, a7, s4
+; RV32-NEXT:    or ra, a4, t0
+; RV32-NEXT:    or s11, t1, a7
+; RV32-NEXT:    andi s10, a0, 1
+; RV32-NEXT:    or a0, s5, a3
 ; RV32-NEXT:    srl a2, a2, a1
-; RV32-NEXT:    snez a3, a3
-; RV32-NEXT:    slli a6, a6, 1
-; RV32-NEXT:    lw a4, 36(sp) # 4-byte Folded Reload
-; RV32-NEXT:    add a3, a4, a3
-; RV32-NEXT:    sll a4, a6, t1
-; RV32-NEXT:    not a3, a3
-; RV32-NEXT:    or s6, a2, a4
-; RV32-NEXT:    andi a3, a3, 1
-; RV32-NEXT:    sw a3, 16(sp) # 4-byte Folded Spill
-; RV32-NEXT:    lw a5, 28(sp) # 4-byte Folded Reload
-; RV32-NEXT:    seqz a2, a5
-; RV32-NEXT:    lw a4, 20(sp) # 4-byte Folded Reload
-; RV32-NEXT:    or a3, a5, a4
-; RV32-NEXT:    sub a6, a4, a2
-; RV32-NEXT:    seqz a2, a3
-; RV32-NEXT:    lw a3, 32(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sub a4, a3, a2
-; RV32-NEXT:    sw a4, 36(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sltu a2, a3, a2
-; RV32-NEXT:    lw a3, 24(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sub a7, a3, a2
+; RV32-NEXT:    snez a3, a0
+; RV32-NEXT:    slli a5, a5, 1
+; RV32-NEXT:    lw a0, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    add a0, a0, a3
+; RV32-NEXT:    sll a3, a5, s4
+; RV32-NEXT:    not a0, a0
+; RV32-NEXT:    or a3, a2, a3
+; RV32-NEXT:    andi a0, a0, 1
+; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a5, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    seqz a0, a5
+; RV32-NEXT:    lw a4, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    or a2, a5, a4
+; RV32-NEXT:    sub a7, a4, a0
+; RV32-NEXT:    seqz a0, a2
+; RV32-NEXT:    lw a2, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sub a4, a2, a0
+; RV32-NEXT:    sw a4, 32(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu a0, a2, a0
+; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sub t0, a2, a0
 ; RV32-NEXT:    addi a5, a5, -1
 ; RV32-NEXT:    sw a5, 4(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw a6, 12(sp) # 4-byte Folded Spill
-; RV32-NEXT:    mv t1, t5
 ; RV32-NEXT:    sw a7, 8(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv s4, t5
 ; RV32-NEXT:    j .LBB3_43
 ; RV32-NEXT:  .LBB3_42: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    mv s8, s7
-; RV32-NEXT:    lw a7, 32(sp) # 4-byte Folded Reload
-; RV32-NEXT:    and a7, s1, a7
-; RV32-NEXT:    xor ra, a3, s11
-; RV32-NEXT:    xor t0, a0, a7
-; RV32-NEXT:    or t0, t0, ra
-; RV32-NEXT:    srli t0, s0, 31
-; RV32-NEXT:    slli ra, t1, 1
-; RV32-NEXT:    mv s7, t1
-; RV32-NEXT:    srli t1, s2, 31
-; RV32-NEXT:    slli s0, s0, 1
-; RV32-NEXT:    or t0, ra, t0
-; RV32-NEXT:    or t1, s0, t1
-; RV32-NEXT:    srli s0, t6, 31
-; RV32-NEXT:    slli s2, s2, 1
-; RV32-NEXT:    or s0, s2, s0
+; RV32-NEXT:    lw t0, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    and t0, s5, t0
+; RV32-NEXT:    xor s10, a0, s8
+; RV32-NEXT:    xor t1, ra, t0
+; RV32-NEXT:    or t1, t1, s10
+; RV32-NEXT:    srli t1, s7, 31
+; RV32-NEXT:    slli s10, s4, 1
+; RV32-NEXT:    mv a6, s4
+; RV32-NEXT:    srli s4, s6, 31
+; RV32-NEXT:    slli s7, s7, 1
+; RV32-NEXT:    or t1, s10, t1
+; RV32-NEXT:    or s4, s7, s4
+; RV32-NEXT:    srli s7, t6, 31
+; RV32-NEXT:    slli s6, s6, 1
+; RV32-NEXT:    or s6, s6, s7
 ; RV32-NEXT:    slli t6, t6, 1
-; RV32-NEXT:    srli ra, s7, 31
-; RV32-NEXT:    or t6, t5, t6
-; RV32-NEXT:    or s2, s8, s0
-; RV32-NEXT:    or s0, s9, t1
-; RV32-NEXT:    or t1, s10, t0
-; RV32-NEXT:    or t0, s3, ra
-; RV32-NEXT:    andi ra, t0, 1
-; RV32-NEXT:    andi s1, s1, 1
-; RV32-NEXT:    sltu t0, a0, a7
-; RV32-NEXT:    sub a7, a0, a7
-; RV32-NEXT:    sub a0, a7, a5
-; RV32-NEXT:    sub a3, a3, s11
-; RV32-NEXT:    sltu a5, a7, a5
-; RV32-NEXT:    sub a3, a3, t0
-; RV32-NEXT:    sub s11, a3, a5
-; RV32-NEXT:    sub a2, s5, a2
-; RV32-NEXT:    or a3, a1, t4
+; RV32-NEXT:    srli s10, a6, 31
+; RV32-NEXT:    lw a6, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    or t6, a6, t6
+; RV32-NEXT:    or s6, s0, s6
+; RV32-NEXT:    or s7, s2, s4
+; RV32-NEXT:    or s4, s1, t1
+; RV32-NEXT:    or a6, s3, s10
+; RV32-NEXT:    andi s10, a6, 1
+; RV32-NEXT:    andi a6, s5, 1
+; RV32-NEXT:    sw a6, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sltu t1, ra, t0
+; RV32-NEXT:    sub t0, ra, t0
+; RV32-NEXT:    sub ra, t0, a5
+; RV32-NEXT:    sub a0, a0, s8
+; RV32-NEXT:    sltu a5, t0, a5
+; RV32-NEXT:    sub a0, a0, t1
+; RV32-NEXT:    sub s8, a0, a5
+; RV32-NEXT:    sub a0, s11, a2
+; RV32-NEXT:    or a2, a1, t4
 ; RV32-NEXT:    or a5, t2, t3
-; RV32-NEXT:    sub s5, a2, a6
-; RV32-NEXT:    or a3, a3, a5
-; RV32-NEXT:    sub s6, s6, a4
-; RV32-NEXT:    snez a2, a3
-; RV32-NEXT:    add a2, s4, a2
-; RV32-NEXT:    or a3, a1, t2
-; RV32-NEXT:    not a2, a2
-; RV32-NEXT:    seqz a3, a3
+; RV32-NEXT:    sub s11, a0, a7
+; RV32-NEXT:    or a2, a2, a5
+; RV32-NEXT:    sub a3, a3, a4
+; RV32-NEXT:    snez a0, a2
+; RV32-NEXT:    add a0, s9, a0
+; RV32-NEXT:    or a2, a1, t2
+; RV32-NEXT:    not a0, a0
+; RV32-NEXT:    seqz a2, a2
 ; RV32-NEXT:    seqz a4, a1
-; RV32-NEXT:    sltu a5, t4, a3
-; RV32-NEXT:    sub t4, t4, a3
+; RV32-NEXT:    sltu a5, t4, a2
+; RV32-NEXT:    sub t4, t4, a2
 ; RV32-NEXT:    sub t2, t2, a4
 ; RV32-NEXT:    sub t3, t3, a5
 ; RV32-NEXT:    addi a1, a1, -1
-; RV32-NEXT:    or a3, t2, t3
+; RV32-NEXT:    or a2, t2, t3
 ; RV32-NEXT:    or a4, a1, t4
-; RV32-NEXT:    andi s3, a2, 1
-; RV32-NEXT:    or a3, a4, a3
-; RV32-NEXT:    or a2, a3, s3
-; RV32-NEXT:    li s7, 0
-; RV32-NEXT:    li s8, 0
-; RV32-NEXT:    li s9, 0
-; RV32-NEXT:    li s10, 0
-; RV32-NEXT:    lw a6, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    andi s9, a0, 1
+; RV32-NEXT:    or a2, a4, a2
+; RV32-NEXT:    or a0, a2, s9
+; RV32-NEXT:    li s0, 0
+; RV32-NEXT:    li s2, 0
+; RV32-NEXT:    li s1, 0
+; RV32-NEXT:    li s3, 0
 ; RV32-NEXT:    lw a7, 8(sp) # 4-byte Folded Reload
-; RV32-NEXT:    beqz a2, .LBB3_54
+; RV32-NEXT:    mv t0, t5
+; RV32-NEXT:    beqz a0, .LBB3_54
 ; RV32-NEXT:  .LBB3_43: # %udiv-do-while
 ; RV32-NEXT:    # =>This Inner Loop Header: Depth=1
-; RV32-NEXT:    srli a2, a0, 31
-; RV32-NEXT:    slli a3, s11, 1
-; RV32-NEXT:    or a3, a3, a2
-; RV32-NEXT:    srli a2, s5, 31
-; RV32-NEXT:    slli a0, a0, 1
-; RV32-NEXT:    or a0, a0, a2
-; RV32-NEXT:    beq a7, a3, .LBB3_45
+; RV32-NEXT:    srli a0, ra, 31
+; RV32-NEXT:    slli a2, s8, 1
+; RV32-NEXT:    or a0, a2, a0
+; RV32-NEXT:    srli a2, s11, 31
+; RV32-NEXT:    slli ra, ra, 1
+; RV32-NEXT:    or ra, ra, a2
+; RV32-NEXT:    beq t0, a0, .LBB3_45
 ; RV32-NEXT:  # %bb.44: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    sltu a2, a7, a3
+; RV32-NEXT:    sltu a2, t0, a0
 ; RV32-NEXT:    j .LBB3_46
 ; RV32-NEXT:  .LBB3_45: # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    lw a2, 36(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a2, a2, a0
+; RV32-NEXT:    lw a2, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    sltu a2, a2, ra
 ; RV32-NEXT:  .LBB3_46: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    srli a4, s6, 31
-; RV32-NEXT:    slli s5, s5, 1
-; RV32-NEXT:    or s5, s5, a4
-; RV32-NEXT:    slli s6, s6, 1
-; RV32-NEXT:    andi a4, ra, 1
-; RV32-NEXT:    or s6, s6, a4
-; RV32-NEXT:    mv s4, s3
-; RV32-NEXT:    beq a6, s5, .LBB3_48
+; RV32-NEXT:    srli a4, a3, 31
+; RV32-NEXT:    slli s11, s11, 1
+; RV32-NEXT:    or s11, s11, a4
+; RV32-NEXT:    slli a3, a3, 1
+; RV32-NEXT:    andi a4, s10, 1
+; RV32-NEXT:    or a3, a3, a4
+; RV32-NEXT:    beq a7, s11, .LBB3_48
 ; RV32-NEXT:  # %bb.47: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    sltu a4, a6, s5
+; RV32-NEXT:    sltu a4, a7, s11
 ; RV32-NEXT:    j .LBB3_49
 ; RV32-NEXT:  .LBB3_48: # in Loop: Header=BB3_43 Depth=1
 ; RV32-NEXT:    lw a4, 4(sp) # 4-byte Folded Reload
-; RV32-NEXT:    sltu a4, a4, s6
+; RV32-NEXT:    sltu a4, a4, a3
 ; RV32-NEXT:  .LBB3_49: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    mv s3, s10
-; RV32-NEXT:    mv t5, s1
-; RV32-NEXT:    lw a5, 36(sp) # 4-byte Folded Reload
-; RV32-NEXT:    xor a5, a5, a0
-; RV32-NEXT:    xor a6, a7, a3
-; RV32-NEXT:    or a5, a5, a6
+; RV32-NEXT:    lw a5, 32(sp) # 4-byte Folded Reload
+; RV32-NEXT:    xor a5, a5, ra
+; RV32-NEXT:    xor a7, t0, a0
+; RV32-NEXT:    or a5, a5, a7
 ; RV32-NEXT:    beqz a5, .LBB3_51
 ; RV32-NEXT:  # %bb.50: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
 ; RV32-NEXT:    mv a4, a2
 ; RV32-NEXT:  .LBB3_51: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    mv s10, s9
-; RV32-NEXT:    srli a2, s11, 31
-; RV32-NEXT:    lw a5, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    srli a2, s8, 31
+; RV32-NEXT:    lw a5, 12(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    sub a2, a5, a2
 ; RV32-NEXT:    sub a2, a2, a4
 ; RV32-NEXT:    slli a2, a2, 31
-; RV32-NEXT:    srai s1, a2, 31
-; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
-; RV32-NEXT:    and s11, s1, a2
-; RV32-NEXT:    mv s9, s8
-; RV32-NEXT:    lw a4, 28(sp) # 4-byte Folded Reload
-; RV32-NEXT:    and a4, s1, a4
+; RV32-NEXT:    srai s5, a2, 31
 ; RV32-NEXT:    lw a2, 20(sp) # 4-byte Folded Reload
-; RV32-NEXT:    and a2, s1, a2
-; RV32-NEXT:    sltu a6, s6, a4
-; RV32-NEXT:    mv a5, a6
-; RV32-NEXT:    beq s5, a2, .LBB3_42
+; RV32-NEXT:    and s8, s5, a2
+; RV32-NEXT:    mv t5, t0
+; RV32-NEXT:    lw a2, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    and a4, s5, a2
+; RV32-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    and a2, s5, a2
+; RV32-NEXT:    sltu a7, a3, a4
+; RV32-NEXT:    mv a5, a7
+; RV32-NEXT:    beq s11, a2, .LBB3_42
 ; RV32-NEXT:  # %bb.52: # %udiv-do-while
 ; RV32-NEXT:    # in Loop: Header=BB3_43 Depth=1
-; RV32-NEXT:    sltu a5, s5, a2
+; RV32-NEXT:    sltu a5, s11, a2
 ; RV32-NEXT:    j .LBB3_42
 ; RV32-NEXT:  .LBB3_53:
-; RV32-NEXT:    li s1, 0
-; RV32-NEXT:    mv t1, t5
+; RV32-NEXT:    sw zero, 36(sp) # 4-byte Folded Spill
+; RV32-NEXT:    mv s4, t5
 ; RV32-NEXT:  .LBB3_54: # %udiv-loop-exit
 ; RV32-NEXT:    srli a0, t6, 31
-; RV32-NEXT:    slli a4, s2, 1
-; RV32-NEXT:    srli a1, s2, 31
-; RV32-NEXT:    slli a2, s0, 1
-; RV32-NEXT:    or a4, a4, a0
-; RV32-NEXT:    or a1, a2, a1
-; RV32-NEXT:    srli s0, s0, 31
-; RV32-NEXT:    slli a0, t1, 1
-; RV32-NEXT:    or t4, a0, s0
-; RV32-NEXT:    slli a0, t6, 1
-; RV32-NEXT:    srli t6, t1, 31
-; RV32-NEXT:    or a2, s1, a0
-; RV32-NEXT:    lw s7, 0(sp) # 4-byte Folded Reload
+; RV32-NEXT:    slli s0, s6, 1
+; RV32-NEXT:    srli a1, s6, 31
+; RV32-NEXT:    slli a4, s7, 1
+; RV32-NEXT:    or s0, s0, a0
+; RV32-NEXT:    or a4, a4, a1
+; RV32-NEXT:    srli a0, s7, 31
+; RV32-NEXT:    slli a1, s4, 1
+; RV32-NEXT:    or s2, a1, a0
+; RV32-NEXT:    slli t6, t6, 1
+; RV32-NEXT:    srli s3, s4, 31
+; RV32-NEXT:    lw a0, 36(sp) # 4-byte Folded Reload
+; RV32-NEXT:    or a2, a0, t6
+; RV32-NEXT:    lw s6, 0(sp) # 4-byte Folded Reload
 ; RV32-NEXT:  .LBB3_55: # %udiv-end
-; RV32-NEXT:    sw a2, 0(s7)
-; RV32-NEXT:    sw a4, 4(s7)
-; RV32-NEXT:    sw a1, 8(s7)
-; RV32-NEXT:    sw t4, 12(s7)
-; RV32-NEXT:    sb t6, 16(s7)
+; RV32-NEXT:    sw a2, 0(s6)
+; RV32-NEXT:    sw s0, 4(s6)
+; RV32-NEXT:    sw a4, 8(s6)
+; RV32-NEXT:    sw s2, 12(s6)
+; RV32-NEXT:    sb s3, 16(s6)
 ; RV32-NEXT:    lw ra, 220(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s0, 216(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s1, 212(sp) # 4-byte Folded Reload
@@ -1896,8 +1897,6 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    sd s2, 168(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    sd s3, 160(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    sd s4, 152(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s5, 144(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s6, 136(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    slli a4, a2, 63
 ; RV64-NEXT:    li t4, 128
 ; RV64-NEXT:    bnez a4, .LBB3_5
@@ -1940,91 +1939,91 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    ld a4, 8(a1)
 ; RV64-NEXT:    ld a1, 16(a1)
 ; RV64-NEXT:    or t3, t5, t3
-; RV64-NEXT:    addi s1, s0, 128
 ; RV64-NEXT:    bnez t3, .LBB3_8
 ; RV64-NEXT:  # %bb.7: # %_udiv-special-cases
-; RV64-NEXT:    mv t2, s1
+; RV64-NEXT:    addi t2, s0, 128
 ; RV64-NEXT:  .LBB3_8: # %_udiv-special-cases
 ; RV64-NEXT:    srli t5, a4, 1
 ; RV64-NEXT:    slli t6, a1, 63
 ; RV64-NEXT:    or t5, t6, t5
 ; RV64-NEXT:    slli t6, a4, 63
-; RV64-NEXT:    srli s2, a5, 1
-; RV64-NEXT:    snez s3, t3
-; RV64-NEXT:    or t6, s2, t6
+; RV64-NEXT:    srli s1, a5, 1
+; RV64-NEXT:    snez s2, t3
+; RV64-NEXT:    or t6, s1, t6
 ; RV64-NEXT:    bnez t5, .LBB3_10
 ; RV64-NEXT:  # %bb.9: # %_udiv-special-cases
 ; RV64-NEXT:    srli t3, t6, 1
 ; RV64-NEXT:    or t3, t6, t3
-; RV64-NEXT:    srli s2, t3, 2
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 4
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 8
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 16
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 32
-; RV64-NEXT:    or t3, t3, s2
+; RV64-NEXT:    srli s1, t3, 2
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 4
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 8
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 16
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 32
+; RV64-NEXT:    or t3, t3, s1
 ; RV64-NEXT:    not t3, t3
-; RV64-NEXT:    srli s2, t3, 1
-; RV64-NEXT:    and s2, s2, a6
-; RV64-NEXT:    sub t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 2
+; RV64-NEXT:    srli s1, t3, 1
+; RV64-NEXT:    and s1, s1, a6
+; RV64-NEXT:    sub t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 2
 ; RV64-NEXT:    and t3, t3, a7
-; RV64-NEXT:    and s2, s2, a7
-; RV64-NEXT:    add t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 4
-; RV64-NEXT:    add t3, t3, s2
+; RV64-NEXT:    and s1, s1, a7
+; RV64-NEXT:    add t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 4
+; RV64-NEXT:    add t3, t3, s1
 ; RV64-NEXT:    and t3, t3, t0
-; RV64-NEXT:    slli s2, t3, 8
-; RV64-NEXT:    add t3, t3, s2
-; RV64-NEXT:    slli s2, t3, 16
-; RV64-NEXT:    add t3, t3, s2
-; RV64-NEXT:    slli s2, t3, 32
-; RV64-NEXT:    add t3, t3, s2
+; RV64-NEXT:    slli s1, t3, 8
+; RV64-NEXT:    add t3, t3, s1
+; RV64-NEXT:    slli s1, t3, 16
+; RV64-NEXT:    add t3, t3, s1
+; RV64-NEXT:    slli s1, t3, 32
+; RV64-NEXT:    add t3, t3, s1
 ; RV64-NEXT:    srli t3, t3, 56
-; RV64-NEXT:    addi s2, t3, 64
+; RV64-NEXT:    addi s1, t3, 64
 ; RV64-NEXT:    j .LBB3_11
 ; RV64-NEXT:  .LBB3_10:
 ; RV64-NEXT:    srli t3, t5, 1
 ; RV64-NEXT:    or t3, t5, t3
-; RV64-NEXT:    srli s2, t3, 2
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 4
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 8
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 16
-; RV64-NEXT:    or t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 32
-; RV64-NEXT:    or t3, t3, s2
+; RV64-NEXT:    srli s1, t3, 2
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 4
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 8
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 16
+; RV64-NEXT:    or t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 32
+; RV64-NEXT:    or t3, t3, s1
 ; RV64-NEXT:    not t3, t3
-; RV64-NEXT:    srli s2, t3, 1
-; RV64-NEXT:    and s2, s2, a6
-; RV64-NEXT:    sub t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 2
+; RV64-NEXT:    srli s1, t3, 1
+; RV64-NEXT:    and s1, s1, a6
+; RV64-NEXT:    sub t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 2
 ; RV64-NEXT:    and t3, t3, a7
-; RV64-NEXT:    and s2, s2, a7
-; RV64-NEXT:    add t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 4
-; RV64-NEXT:    add t3, t3, s2
+; RV64-NEXT:    and s1, s1, a7
+; RV64-NEXT:    add t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 4
+; RV64-NEXT:    add t3, t3, s1
 ; RV64-NEXT:    and t3, t3, t0
-; RV64-NEXT:    slli s2, t3, 8
-; RV64-NEXT:    add t3, t3, s2
-; RV64-NEXT:    slli s2, t3, 16
-; RV64-NEXT:    add t3, t3, s2
-; RV64-NEXT:    slli s2, t3, 32
-; RV64-NEXT:    add t3, t3, s2
-; RV64-NEXT:    srli s2, t3, 56
+; RV64-NEXT:    slli s1, t3, 8
+; RV64-NEXT:    add t3, t3, s1
+; RV64-NEXT:    slli s1, t3, 16
+; RV64-NEXT:    add t3, t3, s1
+; RV64-NEXT:    slli s1, t3, 32
+; RV64-NEXT:    add t3, t3, s1
+; RV64-NEXT:    srli s1, t3, 56
 ; RV64-NEXT:  .LBB3_11: # %_udiv-special-cases
 ; RV64-NEXT:    andi t3, a1, 1
 ; RV64-NEXT:    andi a1, t1, 1
 ; RV64-NEXT:    or t1, a2, a3
-; RV64-NEXT:    sltu s0, s1, s0
+; RV64-NEXT:    addi s3, s0, 128
+; RV64-NEXT:    sltu s0, s3, s0
 ; RV64-NEXT:    slli s4, a5, 63
-; RV64-NEXT:    or s1, a5, a4
-; RV64-NEXT:    addi s3, s3, -1
+; RV64-NEXT:    or s3, a5, a4
+; RV64-NEXT:    addi s2, s2, -1
 ; RV64-NEXT:    beqz s4, .LBB3_13
 ; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    srli t4, s4, 1
@@ -2059,20 +2058,20 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    srli t4, a6, 56
 ; RV64-NEXT:  .LBB3_13: # %_udiv-special-cases
 ; RV64-NEXT:    or t0, t1, a1
-; RV64-NEXT:    or a7, s1, t3
+; RV64-NEXT:    or a7, s3, t3
 ; RV64-NEXT:    or t5, t6, t5
-; RV64-NEXT:    and a6, s3, s0
+; RV64-NEXT:    and a6, s2, s0
 ; RV64-NEXT:    addi t6, t4, 128
 ; RV64-NEXT:    bnez t5, .LBB3_15
 ; RV64-NEXT:  # %bb.14: # %_udiv-special-cases
-; RV64-NEXT:    mv s2, t6
+; RV64-NEXT:    mv s1, t6
 ; RV64-NEXT:  .LBB3_15: # %_udiv-special-cases
 ; RV64-NEXT:    seqz t0, t0
 ; RV64-NEXT:    snez t5, t5
 ; RV64-NEXT:    sltu t4, t6, t4
 ; RV64-NEXT:    addi t5, t5, -1
 ; RV64-NEXT:    and t6, t5, t4
-; RV64-NEXT:    sltu s0, t2, s2
+; RV64-NEXT:    sltu s0, t2, s1
 ; RV64-NEXT:    seqz a7, a7
 ; RV64-NEXT:    mv t4, s0
 ; RV64-NEXT:    beq a6, t6, .LBB3_17
@@ -2083,7 +2082,7 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    andi t4, t4, 1
 ; RV64-NEXT:    sub a6, a6, t6
 ; RV64-NEXT:    sub t0, a6, s0
-; RV64-NEXT:    sub a7, t2, s2
+; RV64-NEXT:    sub a7, t2, s1
 ; RV64-NEXT:    beqz t4, .LBB3_19
 ; RV64-NEXT:  # %bb.18: # %_udiv-special-cases
 ; RV64-NEXT:    mv a6, t4
@@ -2101,81 +2100,85 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    and t5, t5, a5
 ; RV64-NEXT:    bnez t6, .LBB3_29
 ; RV64-NEXT:  # %bb.21: # %_udiv-special-cases
+; RV64-NEXT:    sd t1, 8(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    xori t6, a7, 128
 ; RV64-NEXT:    or s0, t4, t0
 ; RV64-NEXT:    or t6, t6, s0
 ; RV64-NEXT:    beqz t6, .LBB3_29
 ; RV64-NEXT:  # %bb.22: # %udiv-bb1
+; RV64-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    addi a6, a7, 1
-; RV64-NEXT:    li t2, 128
-; RV64-NEXT:    seqz t5, a6
-; RV64-NEXT:    sub t2, t2, a7
-; RV64-NEXT:    add t0, t0, t5
-; RV64-NEXT:    sd zero, 64(sp)
-; RV64-NEXT:    sd zero, 72(sp)
+; RV64-NEXT:    li a1, 128
+; RV64-NEXT:    seqz t1, a6
+; RV64-NEXT:    sub a1, a1, a7
+; RV64-NEXT:    add t0, t0, t1
 ; RV64-NEXT:    sd zero, 80(sp)
 ; RV64-NEXT:    sd zero, 88(sp)
-; RV64-NEXT:    srli t5, t2, 3
-; RV64-NEXT:    andi t5, t5, 24
-; RV64-NEXT:    addi t6, sp, 96
-; RV64-NEXT:    sd a5, 96(sp)
-; RV64-NEXT:    sd a4, 104(sp)
-; RV64-NEXT:    sd t3, 112(sp)
-; RV64-NEXT:    sub s4, t6, t5
-; RV64-NEXT:    or t5, a6, t0
-; RV64-NEXT:    ld t6, 0(s4)
-; RV64-NEXT:    seqz s0, t5
-; RV64-NEXT:    add t4, t4, s0
-; RV64-NEXT:    ld s2, 8(s4)
+; RV64-NEXT:    sd zero, 96(sp)
+; RV64-NEXT:    sd zero, 104(sp)
+; RV64-NEXT:    srli t1, a1, 3
+; RV64-NEXT:    andi t1, t1, 24
+; RV64-NEXT:    addi t2, sp, 112
+; RV64-NEXT:    sd a5, 112(sp)
+; RV64-NEXT:    sd a4, 120(sp)
+; RV64-NEXT:    sd t3, 128(sp)
+; RV64-NEXT:    sub s4, t2, t1
+; RV64-NEXT:    or t1, a6, t0
+; RV64-NEXT:    ld t2, 0(s4)
+; RV64-NEXT:    seqz t5, t1
+; RV64-NEXT:    add t4, t4, t5
+; RV64-NEXT:    ld s1, 8(s4)
 ; RV64-NEXT:    andi t4, t4, 1
-; RV64-NEXT:    andi t2, t2, 63
-; RV64-NEXT:    xori s1, t2, 63
-; RV64-NEXT:    srli t2, t6, 1
-; RV64-NEXT:    srl t2, t2, s1
+; RV64-NEXT:    andi a1, a1, 63
+; RV64-NEXT:    xori s0, a1, 63
+; RV64-NEXT:    srli a1, t2, 1
+; RV64-NEXT:    srl a1, a1, s0
 ; RV64-NEXT:    neg s3, a7
-; RV64-NEXT:    sll a7, s2, s3
-; RV64-NEXT:    or t5, t5, t4
-; RV64-NEXT:    or a7, a7, t2
-; RV64-NEXT:    sll t2, t6, s3
-; RV64-NEXT:    beqz t5, .LBB3_28
-; RV64-NEXT:  # %bb.23: # %udiv-preheader
+; RV64-NEXT:    sll a7, s1, s3
+; RV64-NEXT:    or t1, t1, t4
+; RV64-NEXT:    or a7, a7, a1
+; RV64-NEXT:    sll t2, t2, s3
 ; RV64-NEXT:    li t5, 0
+; RV64-NEXT:    beqz t1, .LBB3_28
+; RV64-NEXT:  # %bb.23: # %udiv-preheader
 ; RV64-NEXT:    li t6, 0
-; RV64-NEXT:    li s0, 0
 ; RV64-NEXT:    ld s4, 16(s4)
-; RV64-NEXT:    srli s5, a6, 3
-; RV64-NEXT:    sd zero, 32(sp)
-; RV64-NEXT:    sd zero, 40(sp)
+; RV64-NEXT:    srli a1, a6, 3
 ; RV64-NEXT:    sd zero, 48(sp)
 ; RV64-NEXT:    sd zero, 56(sp)
-; RV64-NEXT:    andi s5, s5, 24
-; RV64-NEXT:    mv s6, sp
-; RV64-NEXT:    sd a5, 0(sp)
-; RV64-NEXT:    sd a4, 8(sp)
-; RV64-NEXT:    sd t3, 16(sp)
-; RV64-NEXT:    sd zero, 24(sp)
-; RV64-NEXT:    add s5, s6, s5
-; RV64-NEXT:    ld a4, 8(s5)
-; RV64-NEXT:    srli a5, s2, 1
-; RV64-NEXT:    ld t3, 16(s5)
-; RV64-NEXT:    srl a5, a5, s1
-; RV64-NEXT:    sll s1, s4, s3
-; RV64-NEXT:    ld s2, 0(s5)
-; RV64-NEXT:    or a5, s1, a5
-; RV64-NEXT:    srl s1, a4, a6
-; RV64-NEXT:    andi s3, a6, 63
-; RV64-NEXT:    slli t3, t3, 1
-; RV64-NEXT:    xori s3, s3, 63
-; RV64-NEXT:    sll t3, t3, s3
+; RV64-NEXT:    sd zero, 64(sp)
+; RV64-NEXT:    sd zero, 72(sp)
+; RV64-NEXT:    andi a1, a1, 24
+; RV64-NEXT:    addi t1, sp, 16
+; RV64-NEXT:    sd a5, 16(sp)
+; RV64-NEXT:    sd a4, 24(sp)
+; RV64-NEXT:    sd t3, 32(sp)
+; RV64-NEXT:    sd zero, 40(sp)
+; RV64-NEXT:    add a1, t1, a1
+; RV64-NEXT:    ld a4, 8(a1)
+; RV64-NEXT:    srli s1, s1, 1
+; RV64-NEXT:    ld a5, 16(a1)
+; RV64-NEXT:    srl t1, s1, s0
+; RV64-NEXT:    sll t3, s4, s3
+; RV64-NEXT:    ld a1, 0(a1)
+; RV64-NEXT:    or s0, t3, t1
+; RV64-NEXT:    srl t1, a4, a6
+; RV64-NEXT:    andi t3, a6, 63
+; RV64-NEXT:    slli a5, a5, 1
+; RV64-NEXT:    xori s1, t3, 63
+; RV64-NEXT:    sll a5, a5, s1
 ; RV64-NEXT:    slli a4, a4, 1
-; RV64-NEXT:    or t3, s1, t3
-; RV64-NEXT:    sll a4, a4, s3
-; RV64-NEXT:    andi s1, a5, 1
-; RV64-NEXT:    srl a5, s2, a6
-; RV64-NEXT:    or s2, a5, a4
-; RV64-NEXT:    snez a4, t1
+; RV64-NEXT:    or t3, t1, a5
+; RV64-NEXT:    sll a4, a4, s1
+; RV64-NEXT:    andi s0, s0, 1
+; RV64-NEXT:    srl s1, a1, a6
+; RV64-NEXT:    or s1, s1, a4
+; RV64-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    snez a4, a1
 ; RV64-NEXT:    seqz a5, a2
+; RV64-NEXT:    ld a1, 0(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    add a4, a1, a4
+; RV64-NEXT:    li t1, 0
 ; RV64-NEXT:    sub a1, a3, a5
 ; RV64-NEXT:    not a4, a4
 ; RV64-NEXT:    andi a4, a4, 1
@@ -2183,30 +2186,30 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    j .LBB3_26
 ; RV64-NEXT:  .LBB3_24: # %udiv-do-while
 ; RV64-NEXT:    # in Loop: Header=BB3_26 Depth=1
-; RV64-NEXT:    sltu s1, a1, t1
+; RV64-NEXT:    sltu s0, a1, s2
 ; RV64-NEXT:  .LBB3_25: # %udiv-do-while
 ; RV64-NEXT:    # in Loop: Header=BB3_26 Depth=1
 ; RV64-NEXT:    srli t3, t3, 63
 ; RV64-NEXT:    sub t3, a4, t3
-; RV64-NEXT:    sub t3, t3, s1
+; RV64-NEXT:    sub t3, t3, s0
 ; RV64-NEXT:    slli t3, t3, 63
 ; RV64-NEXT:    srai t3, t3, 63
 ; RV64-NEXT:    and s3, t3, a3
-; RV64-NEXT:    srli s1, t2, 63
+; RV64-NEXT:    srli s0, t2, 63
 ; RV64-NEXT:    slli s4, a7, 1
 ; RV64-NEXT:    slli t2, t2, 1
-; RV64-NEXT:    or s1, s4, s1
+; RV64-NEXT:    or s0, s4, s0
 ; RV64-NEXT:    srli s4, a7, 63
 ; RV64-NEXT:    or t2, t5, t2
-; RV64-NEXT:    or a7, t6, s1
-; RV64-NEXT:    or s1, s0, s4
-; RV64-NEXT:    andi s1, s1, 1
-; RV64-NEXT:    and t6, t3, a2
-; RV64-NEXT:    sltu s0, s2, t6
-; RV64-NEXT:    sub t1, t1, s3
+; RV64-NEXT:    or a7, t6, s0
+; RV64-NEXT:    or s0, t1, s4
+; RV64-NEXT:    andi s0, s0, 1
+; RV64-NEXT:    and t1, t3, a2
+; RV64-NEXT:    sltu t6, s1, t1
+; RV64-NEXT:    sub s2, s2, s3
 ; RV64-NEXT:    andi t5, t3, 1
-; RV64-NEXT:    sub t3, t1, s0
-; RV64-NEXT:    sub s2, s2, t6
+; RV64-NEXT:    sub t3, s2, t6
+; RV64-NEXT:    sub s1, s1, t1
 ; RV64-NEXT:    or t1, a6, t0
 ; RV64-NEXT:    seqz t6, a6
 ; RV64-NEXT:    snez t1, t1
@@ -2216,21 +2219,21 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    addi a6, a6, -1
 ; RV64-NEXT:    andi t4, t1, 1
 ; RV64-NEXT:    or t1, a6, t0
-; RV64-NEXT:    or t1, t1, t4
+; RV64-NEXT:    or s2, t1, t4
 ; RV64-NEXT:    li t6, 0
-; RV64-NEXT:    li s0, 0
-; RV64-NEXT:    beqz t1, .LBB3_28
+; RV64-NEXT:    li t1, 0
+; RV64-NEXT:    beqz s2, .LBB3_28
 ; RV64-NEXT:  .LBB3_26: # %udiv-do-while
 ; RV64-NEXT:    # =>This Inner Loop Header: Depth=1
-; RV64-NEXT:    srli t1, s2, 63
+; RV64-NEXT:    srli s2, s1, 63
 ; RV64-NEXT:    slli s3, t3, 1
-; RV64-NEXT:    or t1, s3, t1
-; RV64-NEXT:    slli s2, s2, 1
-; RV64-NEXT:    andi s1, s1, 1
-; RV64-NEXT:    or s2, s2, s1
-; RV64-NEXT:    bne a1, t1, .LBB3_24
+; RV64-NEXT:    or s2, s3, s2
+; RV64-NEXT:    slli s1, s1, 1
+; RV64-NEXT:    andi s0, s0, 1
+; RV64-NEXT:    or s1, s1, s0
+; RV64-NEXT:    bne a1, s2, .LBB3_24
 ; RV64-NEXT:  # %bb.27: # in Loop: Header=BB3_26 Depth=1
-; RV64-NEXT:    sltu s1, a5, s2
+; RV64-NEXT:    sltu s0, a5, s1
 ; RV64-NEXT:    j .LBB3_25
 ; RV64-NEXT:  .LBB3_28: # %udiv-loop-exit
 ; RV64-NEXT:    srli a1, t2, 63
@@ -2248,8 +2251,6 @@ define i129 @udiv_i129(i129 %x, i129 %y) nounwind {
 ; RV64-NEXT:    ld s2, 168(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    ld s3, 160(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    ld s4, 152(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s5, 144(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s6, 136(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    addi sp, sp, 192
 ; RV64-NEXT:    ret
   %res = udiv i129 %x, %y
diff --git a/llvm/test/CodeGen/RISCV/interrupt-attr-nocall.ll b/llvm/test/CodeGen/RISCV/interrupt-attr-nocall.ll
index 33f42e61a76905..db5db47dd47521 100644
--- a/llvm/test/CodeGen/RISCV/interrupt-attr-nocall.ll
+++ b/llvm/test/CodeGen/RISCV/interrupt-attr-nocall.ll
@@ -29,10 +29,10 @@ define void @foo_i32() nounwind #0 {
 ; CHECK-RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; CHECK-RV32-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; CHECK-RV32-NEXT:    lui a0, %hi(a)
-; CHECK-RV32-NEXT:    lui a1, %hi(b)
-; CHECK-RV32-NEXT:    lw a0, %lo(a)(a0)
-; CHECK-RV32-NEXT:    lw a1, %lo(b)(a1)
-; CHECK-RV32-NEXT:    add a0, a1, a0
+; CHECK-RV32-NEXT:    lw a1, %lo(a)(a0)
+; CHECK-RV32-NEXT:    lui a0, %hi(b)
+; CHECK-RV32-NEXT:    lw a0, %lo(b)(a0)
+; CHECK-RV32-NEXT:    add a0, a0, a1
 ; CHECK-RV32-NEXT:    lui a1, %hi(c)
 ; CHECK-RV32-NEXT:    sw a0, %lo(c)(a1)
 ; CHECK-RV32-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
@@ -46,10 +46,10 @@ define void @foo_i32() nounwind #0 {
 ; CHECK-RV32IF-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; CHECK-RV32IF-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; CHECK-RV32IF-NEXT:    lui a0, %hi(a)
-; CHECK-RV32IF-NEXT:    lui a1, %hi(b)
-; CHECK-RV32IF-NEXT:    lw a0, %lo(a)(a0)
-; CHECK-RV32IF-NEXT:    lw a1, %lo(b)(a1)
-; CHECK-RV32IF-NEXT:    add a0, a1, a0
+; CHECK-RV32IF-NEXT:    lw a1, %lo(a)(a0)
+; CHECK-RV32IF-NEXT:    lui a0, %hi(b)
+; CHECK-RV32IF-NEXT:    lw a0, %lo(b)(a0)
+; CHECK-RV32IF-NEXT:    add a0, a0, a1
 ; CHECK-RV32IF-NEXT:    lui a1, %hi(c)
 ; CHECK-RV32IF-NEXT:    sw a0, %lo(c)(a1)
 ; CHECK-RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
@@ -63,10 +63,10 @@ define void @foo_i32() nounwind #0 {
 ; CHECK-RV32IFD-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(a)
-; CHECK-RV32IFD-NEXT:    lui a1, %hi(b)
-; CHECK-RV32IFD-NEXT:    lw a0, %lo(a)(a0)
-; CHECK-RV32IFD-NEXT:    lw a1, %lo(b)(a1)
-; CHECK-RV32IFD-NEXT:    add a0, a1, a0
+; CHECK-RV32IFD-NEXT:    lw a1, %lo(a)(a0)
+; CHECK-RV32IFD-NEXT:    lui a0, %hi(b)
+; CHECK-RV32IFD-NEXT:    lw a0, %lo(b)(a0)
+; CHECK-RV32IFD-NEXT:    add a0, a0, a1
 ; CHECK-RV32IFD-NEXT:    lui a1, %hi(c)
 ; CHECK-RV32IFD-NEXT:    sw a0, %lo(c)(a1)
 ; CHECK-RV32IFD-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
@@ -94,10 +94,10 @@ define void @foo_fp_i32() nounwind #1 {
 ; CHECK-RV32-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
 ; CHECK-RV32-NEXT:    addi s0, sp, 16
 ; CHECK-RV32-NEXT:    lui a0, %hi(a)
-; CHECK-RV32-NEXT:    lui a1, %hi(b)
-; CHECK-RV32-NEXT:    lw a0, %lo(a)(a0)
-; CHECK-RV32-NEXT:    lw a1, %lo(b)(a1)
-; CHECK-RV32-NEXT:    add a0, a1, a0
+; CHECK-RV32-NEXT:    lw a1, %lo(a)(a0)
+; CHECK-RV32-NEXT:    lui a0, %hi(b)
+; CHECK-RV32-NEXT:    lw a0, %lo(b)(a0)
+; CHECK-RV32-NEXT:    add a0, a0, a1
 ; CHECK-RV32-NEXT:    lui a1, %hi(c)
 ; CHECK-RV32-NEXT:    sw a0, %lo(c)(a1)
 ; CHECK-RV32-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
@@ -116,10 +116,10 @@ define void @foo_fp_i32() nounwind #1 {
 ; CHECK-RV32IF-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
 ; CHECK-RV32IF-NEXT:    addi s0, sp, 16
 ; CHECK-RV32IF-NEXT:    lui a0, %hi(a)
-; CHECK-RV32IF-NEXT:    lui a1, %hi(b)
-; CHECK-RV32IF-NEXT:    lw a0, %lo(a)(a0)
-; CHECK-RV32IF-NEXT:    lw a1, %lo(b)(a1)
-; CHECK-RV32IF-NEXT:    add a0, a1, a0
+; CHECK-RV32IF-NEXT:    lw a1, %lo(a)(a0)
+; CHECK-RV32IF-NEXT:    lui a0, %hi(b)
+; CHECK-RV32IF-NEXT:    lw a0, %lo(b)(a0)
+; CHECK-RV32IF-NEXT:    add a0, a0, a1
 ; CHECK-RV32IF-NEXT:    lui a1, %hi(c)
 ; CHECK-RV32IF-NEXT:    sw a0, %lo(c)(a1)
 ; CHECK-RV32IF-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
@@ -138,10 +138,10 @@ define void @foo_fp_i32() nounwind #1 {
 ; CHECK-RV32IFD-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    addi s0, sp, 16
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(a)
-; CHECK-RV32IFD-NEXT:    lui a1, %hi(b)
-; CHECK-RV32IFD-NEXT:    lw a0, %lo(a)(a0)
-; CHECK-RV32IFD-NEXT:    lw a1, %lo(b)(a1)
-; CHECK-RV32IFD-NEXT:    add a0, a1, a0
+; CHECK-RV32IFD-NEXT:    lw a1, %lo(a)(a0)
+; CHECK-RV32IFD-NEXT:    lui a0, %hi(b)
+; CHECK-RV32IFD-NEXT:    lw a0, %lo(b)(a0)
+; CHECK-RV32IFD-NEXT:    add a0, a0, a1
 ; CHECK-RV32IFD-NEXT:    lui a1, %hi(c)
 ; CHECK-RV32IFD-NEXT:    sw a0, %lo(c)(a1)
 ; CHECK-RV32IFD-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
@@ -211,20 +211,18 @@ define void @foo_float() nounwind #0 {
 ; CHECK-RV32IF:       # %bb.0:
 ; CHECK-RV32IF-NEXT:    addi sp, sp, -16
 ; CHECK-RV32IF-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
-; CHECK-RV32IF-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
-; CHECK-RV32IF-NEXT:    fsw fa4, 4(sp) # 4-byte Folded Spill
-; CHECK-RV32IF-NEXT:    fsw fa5, 0(sp) # 4-byte Folded Spill
+; CHECK-RV32IF-NEXT:    fsw fa4, 8(sp) # 4-byte Folded Spill
+; CHECK-RV32IF-NEXT:    fsw fa5, 4(sp) # 4-byte Folded Spill
 ; CHECK-RV32IF-NEXT:    lui a0, %hi(e)
-; CHECK-RV32IF-NEXT:    lui a1, %hi(f)
 ; CHECK-RV32IF-NEXT:    flw fa5, %lo(e)(a0)
-; CHECK-RV32IF-NEXT:    flw fa4, %lo(f)(a1)
+; CHECK-RV32IF-NEXT:    lui a0, %hi(f)
+; CHECK-RV32IF-NEXT:    flw fa4, %lo(f)(a0)
 ; CHECK-RV32IF-NEXT:    fadd.s fa5, fa5, fa4
 ; CHECK-RV32IF-NEXT:    lui a0, %hi(d)
 ; CHECK-RV32IF-NEXT:    fsw fa5, %lo(d)(a0)
 ; CHECK-RV32IF-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
-; CHECK-RV32IF-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
-; CHECK-RV32IF-NEXT:    flw fa4, 4(sp) # 4-byte Folded Reload
-; CHECK-RV32IF-NEXT:    flw fa5, 0(sp) # 4-byte Folded Reload
+; CHECK-RV32IF-NEXT:    flw fa4, 8(sp) # 4-byte Folded Reload
+; CHECK-RV32IF-NEXT:    flw fa5, 4(sp) # 4-byte Folded Reload
 ; CHECK-RV32IF-NEXT:    addi sp, sp, 16
 ; CHECK-RV32IF-NEXT:    mret
 ;
@@ -232,18 +230,16 @@ define void @foo_float() nounwind #0 {
 ; CHECK-RV32IFD:       # %bb.0:
 ; CHECK-RV32IFD-NEXT:    addi sp, sp, -32
 ; CHECK-RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
-; CHECK-RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa4, 16(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa5, 8(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(e)
-; CHECK-RV32IFD-NEXT:    lui a1, %hi(f)
 ; CHECK-RV32IFD-NEXT:    flw fa5, %lo(e)(a0)
-; CHECK-RV32IFD-NEXT:    flw fa4, %lo(f)(a1)
+; CHECK-RV32IFD-NEXT:    lui a0, %hi(f)
+; CHECK-RV32IFD-NEXT:    flw fa4, %lo(f)(a0)
 ; CHECK-RV32IFD-NEXT:    fadd.s fa5, fa5, fa4
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(d)
 ; CHECK-RV32IFD-NEXT:    fsw fa5, %lo(d)(a0)
 ; CHECK-RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
-; CHECK-RV32IFD-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa4, 16(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa5, 8(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    addi sp, sp, 32
@@ -313,23 +309,21 @@ define void @foo_fp_float() nounwind #1 {
 ; CHECK-RV32IF-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
 ; CHECK-RV32IF-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; CHECK-RV32IF-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
-; CHECK-RV32IF-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
-; CHECK-RV32IF-NEXT:    fsw fa4, 12(sp) # 4-byte Folded Spill
-; CHECK-RV32IF-NEXT:    fsw fa5, 8(sp) # 4-byte Folded Spill
+; CHECK-RV32IF-NEXT:    fsw fa4, 16(sp) # 4-byte Folded Spill
+; CHECK-RV32IF-NEXT:    fsw fa5, 12(sp) # 4-byte Folded Spill
 ; CHECK-RV32IF-NEXT:    addi s0, sp, 32
 ; CHECK-RV32IF-NEXT:    lui a0, %hi(e)
-; CHECK-RV32IF-NEXT:    lui a1, %hi(f)
 ; CHECK-RV32IF-NEXT:    flw fa5, %lo(e)(a0)
-; CHECK-RV32IF-NEXT:    flw fa4, %lo(f)(a1)
+; CHECK-RV32IF-NEXT:    lui a0, %hi(f)
+; CHECK-RV32IF-NEXT:    flw fa4, %lo(f)(a0)
 ; CHECK-RV32IF-NEXT:    fadd.s fa5, fa5, fa4
 ; CHECK-RV32IF-NEXT:    lui a0, %hi(d)
 ; CHECK-RV32IF-NEXT:    fsw fa5, %lo(d)(a0)
 ; CHECK-RV32IF-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; CHECK-RV32IF-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; CHECK-RV32IF-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
-; CHECK-RV32IF-NEXT:    lw a1, 16(sp) # 4-byte Folded Reload
-; CHECK-RV32IF-NEXT:    flw fa4, 12(sp) # 4-byte Folded Reload
-; CHECK-RV32IF-NEXT:    flw fa5, 8(sp) # 4-byte Folded Reload
+; CHECK-RV32IF-NEXT:    flw fa4, 16(sp) # 4-byte Folded Reload
+; CHECK-RV32IF-NEXT:    flw fa5, 12(sp) # 4-byte Folded Reload
 ; CHECK-RV32IF-NEXT:    addi sp, sp, 32
 ; CHECK-RV32IF-NEXT:    mret
 ;
@@ -339,21 +333,19 @@ define void @foo_fp_float() nounwind #1 {
 ; CHECK-RV32IFD-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
-; CHECK-RV32IFD-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa4, 8(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa5, 0(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    addi s0, sp, 32
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(e)
-; CHECK-RV32IFD-NEXT:    lui a1, %hi(f)
 ; CHECK-RV32IFD-NEXT:    flw fa5, %lo(e)(a0)
-; CHECK-RV32IFD-NEXT:    flw fa4, %lo(f)(a1)
+; CHECK-RV32IFD-NEXT:    lui a0, %hi(f)
+; CHECK-RV32IFD-NEXT:    flw fa4, %lo(f)(a0)
 ; CHECK-RV32IFD-NEXT:    fadd.s fa5, fa5, fa4
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(d)
 ; CHECK-RV32IFD-NEXT:    fsw fa5, %lo(d)(a0)
 ; CHECK-RV32IFD-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
-; CHECK-RV32IFD-NEXT:    lw a1, 16(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa4, 8(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa5, 0(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    addi sp, sp, 32
@@ -510,18 +502,16 @@ define void @foo_double() nounwind #0 {
 ; CHECK-RV32IFD:       # %bb.0:
 ; CHECK-RV32IFD-NEXT:    addi sp, sp, -32
 ; CHECK-RV32IFD-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
-; CHECK-RV32IFD-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa4, 16(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa5, 8(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(h)
-; CHECK-RV32IFD-NEXT:    lui a1, %hi(i)
 ; CHECK-RV32IFD-NEXT:    fld fa5, %lo(h)(a0)
-; CHECK-RV32IFD-NEXT:    fld fa4, %lo(i)(a1)
+; CHECK-RV32IFD-NEXT:    lui a0, %hi(i)
+; CHECK-RV32IFD-NEXT:    fld fa4, %lo(i)(a0)
 ; CHECK-RV32IFD-NEXT:    fadd.d fa5, fa5, fa4
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(g)
 ; CHECK-RV32IFD-NEXT:    fsd fa5, %lo(g)(a0)
 ; CHECK-RV32IFD-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
-; CHECK-RV32IFD-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa4, 16(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa5, 8(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    addi sp, sp, 32
@@ -685,21 +675,19 @@ define void @foo_fp_double() nounwind #1 {
 ; CHECK-RV32IFD-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
-; CHECK-RV32IFD-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa4, 8(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    fsd fa5, 0(sp) # 8-byte Folded Spill
 ; CHECK-RV32IFD-NEXT:    addi s0, sp, 32
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(h)
-; CHECK-RV32IFD-NEXT:    lui a1, %hi(i)
 ; CHECK-RV32IFD-NEXT:    fld fa5, %lo(h)(a0)
-; CHECK-RV32IFD-NEXT:    fld fa4, %lo(i)(a1)
+; CHECK-RV32IFD-NEXT:    lui a0, %hi(i)
+; CHECK-RV32IFD-NEXT:    fld fa4, %lo(i)(a0)
 ; CHECK-RV32IFD-NEXT:    fadd.d fa5, fa5, fa4
 ; CHECK-RV32IFD-NEXT:    lui a0, %hi(g)
 ; CHECK-RV32IFD-NEXT:    fsd fa5, %lo(g)(a0)
 ; CHECK-RV32IFD-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
-; CHECK-RV32IFD-NEXT:    lw a1, 16(sp) # 4-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa4, 8(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    fld fa5, 0(sp) # 8-byte Folded Reload
 ; CHECK-RV32IFD-NEXT:    addi sp, sp, 32
diff --git a/llvm/test/CodeGen/RISCV/machine-outliner-and-machine-copy-propagation.ll b/llvm/test/CodeGen/RISCV/machine-outliner-and-machine-copy-propagation.ll
index a1a6fa1f62a549..da98bc394b4ecd 100644
--- a/llvm/test/CodeGen/RISCV/machine-outliner-and-machine-copy-propagation.ll
+++ b/llvm/test/CodeGen/RISCV/machine-outliner-and-machine-copy-propagation.ll
@@ -145,7 +145,6 @@ declare void @exit(i32 signext) noreturn
 ; RV64I-NEXT:    addi sp, sp, -32
 ; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    call t0, OUTLINED_FUNCTION_0
 ; RV64I-NEXT:    call nge
 ; RV64I-NEXT:    bnez a0, .LBB4_9
@@ -163,25 +162,30 @@ declare void @exit(i32 signext) noreturn
 ; RV64I-NEXT:    lui a1, 524288
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call ngt
-; RV64I-NEXT:    li s1, -1
-; RV64I-NEXT:    bne a0, s1, .LBB4_9
+; RV64I-NEXT:    li a1, -1
+; RV64I-NEXT:    bne a0, a1, .LBB4_9
 ; RV64I-NEXT:  # %bb.4: # %if.end12
 ; RV64I-NEXT:    call t0, OUTLINED_FUNCTION_0
 ; RV64I-NEXT:    call nle
-; RV64I-NEXT:    bne a0, s1, .LBB4_9
+; RV64I-NEXT:    li a1, -1
+; RV64I-NEXT:    bne a0, a1, .LBB4_9
 ; RV64I-NEXT:  # %bb.5: # %if.end16
 ; RV64I-NEXT:    lui a1, 524288
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    call nle
 ; RV64I-NEXT:    bnez a0, .LBB4_9
 ; RV64I-NEXT:  # %bb.6: # %if.end20
-; RV64I-NEXT:    call t0, OUTLINED_FUNCTION_0
+; RV64I-NEXT:    lui a0, 524288
+; RV64I-NEXT:    addiw a0, a0, -1
+; RV64I-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    lui a0, 524288
+; RV64I-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    call nlt
 ; RV64I-NEXT:    li a1, -1
 ; RV64I-NEXT:    bne a0, a1, .LBB4_9
 ; RV64I-NEXT:  # %bb.7: # %if.end24
 ; RV64I-NEXT:    lui a1, 524288
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    call nlt
 ; RV64I-NEXT:    bnez a0, .LBB4_9
 ; RV64I-NEXT:  # %bb.8: # %if.end28
diff --git a/llvm/test/CodeGen/RISCV/machine-sink-load-immediate.ll b/llvm/test/CodeGen/RISCV/machine-sink-load-immediate.ll
index dff2293c1f941c..e289f97adcee20 100644
--- a/llvm/test/CodeGen/RISCV/machine-sink-load-immediate.ll
+++ b/llvm/test/CodeGen/RISCV/machine-sink-load-immediate.ll
@@ -7,82 +7,82 @@ define i1 @sink_li(ptr %text, ptr %text.addr.0) nounwind {
 ; CHECK-NEXT:    addi sp, sp, -32
 ; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    mv s1, a1
+; CHECK-NEXT:    sd a1, 8(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    mv s0, a0
 ; CHECK-NEXT:    call toupper
-; CHECK-NEXT:    li a1, 0
-; CHECK-NEXT:    beqz s0, .LBB0_25
-; CHECK-NEXT:  .LBB0_1: # %while.body
+; CHECK-NEXT:    beqz s0, .LBB0_24
+; CHECK-NEXT:  # %bb.1: # %while.body.preheader
+; CHECK-NEXT:    mv a1, s0
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:  .LBB0_2: # %while.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    j .LBB0_3
-; CHECK-NEXT:  # %bb.2: # %while.body
-; CHECK-NEXT:    j .LBB0_15
-; CHECK-NEXT:  .LBB0_3: # %while.body.1
-; CHECK-NEXT:    # in Loop: Header=BB0_1 Depth=1
-; CHECK-NEXT:    j .LBB0_5
-; CHECK-NEXT:  # %bb.4: # %while.body.1
+; CHECK-NEXT:    j .LBB0_4
+; CHECK-NEXT:  # %bb.3: # %while.body
 ; CHECK-NEXT:    j .LBB0_16
-; CHECK-NEXT:  .LBB0_5: # %while.body.3
-; CHECK-NEXT:    # in Loop: Header=BB0_1 Depth=1
-; CHECK-NEXT:    j .LBB0_7
-; CHECK-NEXT:  # %bb.6: # %while.body.3
-; CHECK-NEXT:    j .LBB0_18
-; CHECK-NEXT:  .LBB0_7: # %while.body.4
-; CHECK-NEXT:    # in Loop: Header=BB0_1 Depth=1
-; CHECK-NEXT:    j .LBB0_9
-; CHECK-NEXT:  # %bb.8: # %while.body.4
-; CHECK-NEXT:    j .LBB0_20
-; CHECK-NEXT:  .LBB0_9: # %while.body.5
-; CHECK-NEXT:    # in Loop: Header=BB0_1 Depth=1
-; CHECK-NEXT:    j .LBB0_11
-; CHECK-NEXT:  # %bb.10: # %while.body.5
-; CHECK-NEXT:    j .LBB0_22
-; CHECK-NEXT:  .LBB0_11: # %while.body.6
-; CHECK-NEXT:    # in Loop: Header=BB0_1 Depth=1
-; CHECK-NEXT:    j .LBB0_1
-; CHECK-NEXT:  # %bb.12: # %while.body.6
-; CHECK-NEXT:  # %bb.13: # %while.body.6
-; CHECK-NEXT:  # %bb.14: # %strdup.exit.split.loop.exit126
-; CHECK-NEXT:    addi s0, s1, 7
-; CHECK-NEXT:    j .LBB0_24
-; CHECK-NEXT:  .LBB0_15: # %while.body
+; CHECK-NEXT:  .LBB0_4: # %while.body.1
+; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
+; CHECK-NEXT:    j .LBB0_6
+; CHECK-NEXT:  # %bb.5: # %while.body.1
 ; CHECK-NEXT:    j .LBB0_17
-; CHECK-NEXT:  .LBB0_16: # %while.body.1
-; CHECK-NEXT:  .LBB0_17: # %strdup.exit.loopexit
-; CHECK-NEXT:    li s0, 0
-; CHECK-NEXT:    j .LBB0_24
-; CHECK-NEXT:  .LBB0_18: # %while.body.3
-; CHECK-NEXT:  # %bb.19: # %strdup.exit.split.loop.exit120
-; CHECK-NEXT:    addi s0, s1, 4
-; CHECK-NEXT:    j .LBB0_24
-; CHECK-NEXT:  .LBB0_20: # %while.body.4
-; CHECK-NEXT:  # %bb.21: # %strdup.exit.split.loop.exit122
-; CHECK-NEXT:    addi s0, s1, 5
-; CHECK-NEXT:    j .LBB0_24
-; CHECK-NEXT:  .LBB0_22: # %while.body.5
-; CHECK-NEXT:    j .LBB0_24
-; CHECK-NEXT:  # %bb.23:
+; CHECK-NEXT:  .LBB0_6: # %while.body.3
+; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
+; CHECK-NEXT:    j .LBB0_8
+; CHECK-NEXT:  # %bb.7: # %while.body.3
+; CHECK-NEXT:    j .LBB0_19
+; CHECK-NEXT:  .LBB0_8: # %while.body.4
+; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
+; CHECK-NEXT:    j .LBB0_10
+; CHECK-NEXT:  # %bb.9: # %while.body.4
+; CHECK-NEXT:    j .LBB0_21
+; CHECK-NEXT:  .LBB0_10: # %while.body.5
+; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
+; CHECK-NEXT:    j .LBB0_12
+; CHECK-NEXT:  # %bb.11: # %while.body.5
+; CHECK-NEXT:    j .LBB0_23
+; CHECK-NEXT:  .LBB0_12: # %while.body.6
+; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
+; CHECK-NEXT:    j .LBB0_2
+; CHECK-NEXT:  # %bb.13: # %while.body.6
+; CHECK-NEXT:  # %bb.14: # %while.body.6
+; CHECK-NEXT:  # %bb.15: # %strdup.exit.split.loop.exit126
+; CHECK-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi a1, a1, 7
+; CHECK-NEXT:    j .LBB0_25
+; CHECK-NEXT:  .LBB0_16: # %while.body
+; CHECK-NEXT:    j .LBB0_18
+; CHECK-NEXT:  .LBB0_17: # %while.body.1
+; CHECK-NEXT:  .LBB0_18: # %strdup.exit.loopexit
 ; CHECK-NEXT:    li a1, 0
 ; CHECK-NEXT:    j .LBB0_25
-; CHECK-NEXT:  .LBB0_24: # %strdup.exit
-; CHECK-NEXT:    li s1, 0
-; CHECK-NEXT:    mv s2, a0
+; CHECK-NEXT:  .LBB0_19: # %while.body.3
+; CHECK-NEXT:  # %bb.20: # %strdup.exit.split.loop.exit120
+; CHECK-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi a1, a1, 4
+; CHECK-NEXT:    j .LBB0_25
+; CHECK-NEXT:  .LBB0_21: # %while.body.4
+; CHECK-NEXT:  # %bb.22: # %strdup.exit.split.loop.exit122
+; CHECK-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi a1, a1, 5
+; CHECK-NEXT:    j .LBB0_25
+; CHECK-NEXT:  .LBB0_23: # %while.body.5
+; CHECK-NEXT:    j .LBB0_25
+; CHECK-NEXT:  .LBB0_24:
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    j .LBB0_26
+; CHECK-NEXT:  .LBB0_25: # %strdup.exit
+; CHECK-NEXT:    li s0, 0
+; CHECK-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    li a0, 0
-; CHECK-NEXT:    mv a1, s0
-; CHECK-NEXT:    jalr s1
+; CHECK-NEXT:    jalr s0
 ; CHECK-NEXT:    li a0, 0
-; CHECK-NEXT:    mv a1, s2
+; CHECK-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    li a2, 0
-; CHECK-NEXT:    jalr s1
-; CHECK-NEXT:    li a1, 1
-; CHECK-NEXT:  .LBB0_25: # %return
-; CHECK-NEXT:    mv a0, a1
+; CHECK-NEXT:    jalr s0
+; CHECK-NEXT:    li a2, 1
+; CHECK-NEXT:  .LBB0_26: # %return
+; CHECK-NEXT:    mv a0, a2
 ; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    ret
 entry:
diff --git a/llvm/test/CodeGen/RISCV/miss-sp-restore-eh.ll b/llvm/test/CodeGen/RISCV/miss-sp-restore-eh.ll
index 4bec4504ec79da..77bbd5437569dd 100644
--- a/llvm/test/CodeGen/RISCV/miss-sp-restore-eh.ll
+++ b/llvm/test/CodeGen/RISCV/miss-sp-restore-eh.ll
@@ -17,10 +17,8 @@ define signext i32 @foo() #1 personality ptr @__gxx_personality_v0 {
 ; CHECK-NEXT:    .cfi_def_cfa_offset 32
 ; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    .cfi_offset ra, -8
 ; CHECK-NEXT:    .cfi_offset s0, -16
-; CHECK-NEXT:    .cfi_offset s1, -24
 ; CHECK-NEXT:    addi s0, sp, 32
 ; CHECK-NEXT:    .cfi_def_cfa s0, 0
 ; CHECK-NEXT:    .cfi_remember_state
@@ -39,17 +37,16 @@ define signext i32 @foo() #1 personality ptr @__gxx_personality_v0 {
 ; CHECK-NEXT:    bne a1, a2, .LBB0_4
 ; CHECK-NEXT:  # %bb.3: # %catch
 ; CHECK-NEXT:    call __cxa_begin_catch
-; CHECK-NEXT:    lw s1, 0(a0)
+; CHECK-NEXT:    lw a0, 0(a0)
+; CHECK-NEXT:    sd a0, -24(s0) # 8-byte Folded Spill
 ; CHECK-NEXT:    call __cxa_end_catch
-; CHECK-NEXT:    mv a0, s1
+; CHECK-NEXT:    ld a0, -24(s0) # 8-byte Folded Reload
 ; CHECK-NEXT:    addi sp, s0, -32
 ; CHECK-NEXT:    .cfi_def_cfa sp, 32
 ; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    .cfi_restore ra
 ; CHECK-NEXT:    .cfi_restore s0
-; CHECK-NEXT:    .cfi_restore s1
 ; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    .cfi_def_cfa_offset 0
 ; CHECK-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/overflow-intrinsics.ll b/llvm/test/CodeGen/RISCV/overflow-intrinsics.ll
index aef7b82eb78b45..a51747c9014d4f 100644
--- a/llvm/test/CodeGen/RISCV/overflow-intrinsics.ll
+++ b/llvm/test/CodeGen/RISCV/overflow-intrinsics.ll
@@ -1071,73 +1071,51 @@ end:
 define i1 @usubo_ult_cmp_dominates_i64(i64 %x, i64 %y, ptr %p, i1 %cond) {
 ; RV32-LABEL: usubo_ult_cmp_dominates_i64:
 ; RV32:       # %bb.0: # %entry
+; RV32-NEXT:    andi a6, a5, 1
+; RV32-NEXT:    beqz a6, .LBB32_6
+; RV32-NEXT:  # %bb.1: # %t
 ; RV32-NEXT:    addi sp, sp, -32
 ; RV32-NEXT:    .cfi_def_cfa_offset 32
 ; RV32-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s6, 0(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
-; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    .cfi_offset s1, -12
-; RV32-NEXT:    .cfi_offset s2, -16
-; RV32-NEXT:    .cfi_offset s3, -20
-; RV32-NEXT:    .cfi_offset s4, -24
-; RV32-NEXT:    .cfi_offset s5, -28
-; RV32-NEXT:    .cfi_offset s6, -32
-; RV32-NEXT:    mv s1, a5
-; RV32-NEXT:    mv s4, a1
-; RV32-NEXT:    andi a1, a5, 1
-; RV32-NEXT:    beqz a1, .LBB32_6
-; RV32-NEXT:  # %bb.1: # %t
-; RV32-NEXT:    mv s0, a4
-; RV32-NEXT:    mv s3, a3
-; RV32-NEXT:    mv s2, a2
-; RV32-NEXT:    mv s5, a0
-; RV32-NEXT:    beq s4, a3, .LBB32_3
+; RV32-NEXT:    sw a5, 20(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a2, 16(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a4, 8(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw a3, 0(sp) # 4-byte Folded Spill
+; RV32-NEXT:    beq a1, a3, .LBB32_3
 ; RV32-NEXT:  # %bb.2: # %t
-; RV32-NEXT:    sltu s6, s4, s3
+; RV32-NEXT:    sltu a0, a1, a3
 ; RV32-NEXT:    j .LBB32_4
 ; RV32-NEXT:  .LBB32_3:
-; RV32-NEXT:    sltu s6, s5, s2
+; RV32-NEXT:    sltu a0, a0, a2
 ; RV32-NEXT:  .LBB32_4: # %t
-; RV32-NEXT:    mv a0, s6
+; RV32-NEXT:    sw a0, 24(sp) # 4-byte Folded Spill
+; RV32-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    call call
-; RV32-NEXT:    beqz s6, .LBB32_6
-; RV32-NEXT:  # %bb.5: # %end
-; RV32-NEXT:    sltu a0, s5, s2
-; RV32-NEXT:    sub a1, s4, s3
-; RV32-NEXT:    sub a1, a1, a0
-; RV32-NEXT:    sub a2, s5, s2
-; RV32-NEXT:    mv a0, s6
-; RV32-NEXT:    sw a2, 0(s0)
-; RV32-NEXT:    sw a1, 4(s0)
-; RV32-NEXT:    j .LBB32_7
-; RV32-NEXT:  .LBB32_6: # %f
-; RV32-NEXT:    mv a0, s1
-; RV32-NEXT:  .LBB32_7: # %f
+; RV32-NEXT:    lw a0, 24(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a5, 20(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a2, 16(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a4, 0(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a6, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw a3, 12(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s6, 0(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
-; RV32-NEXT:    .cfi_restore s0
-; RV32-NEXT:    .cfi_restore s1
-; RV32-NEXT:    .cfi_restore s2
-; RV32-NEXT:    .cfi_restore s3
-; RV32-NEXT:    .cfi_restore s4
-; RV32-NEXT:    .cfi_restore s5
-; RV32-NEXT:    .cfi_restore s6
 ; RV32-NEXT:    addi sp, sp, 32
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    beqz a0, .LBB32_6
+; RV32-NEXT:  # %bb.5: # %end
+; RV32-NEXT:    sltu a5, a3, a2
+; RV32-NEXT:    sub a4, a6, a4
+; RV32-NEXT:    sub a4, a4, a5
+; RV32-NEXT:    sub a3, a3, a2
+; RV32-NEXT:    sw a3, 0(a1)
+; RV32-NEXT:    sw a4, 4(a1)
+; RV32-NEXT:    ret
+; RV32-NEXT:  .LBB32_6: # %f
+; RV32-NEXT:    mv a0, a5
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: usubo_ult_cmp_dominates_i64:
@@ -1145,48 +1123,33 @@ define i1 @usubo_ult_cmp_dominates_i64(i64 %x, i64 %y, ptr %p, i1 %cond) {
 ; RV64-NEXT:    addi sp, sp, -48
 ; RV64-NEXT:    .cfi_def_cfa_offset 48
 ; RV64-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    .cfi_offset ra, -8
-; RV64-NEXT:    .cfi_offset s0, -16
-; RV64-NEXT:    .cfi_offset s1, -24
-; RV64-NEXT:    .cfi_offset s2, -32
-; RV64-NEXT:    .cfi_offset s3, -40
-; RV64-NEXT:    .cfi_offset s4, -48
-; RV64-NEXT:    mv s1, a3
-; RV64-NEXT:    mv s2, a1
-; RV64-NEXT:    andi a1, a3, 1
-; RV64-NEXT:    beqz a1, .LBB32_3
+; RV64-NEXT:    andi a4, a3, 1
+; RV64-NEXT:    beqz a4, .LBB32_3
 ; RV64-NEXT:  # %bb.1: # %t
-; RV64-NEXT:    mv s0, a2
-; RV64-NEXT:    mv s3, a0
-; RV64-NEXT:    sltu s4, a0, s2
-; RV64-NEXT:    mv a0, s4
+; RV64-NEXT:    sltu a4, a0, a1
+; RV64-NEXT:    sd a4, 32(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a0, 24(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 32(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sd a2, 16(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a3, 0(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd a1, 8(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    call call
-; RV64-NEXT:    bgeu s3, s2, .LBB32_3
+; RV64-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a2, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a3, 0(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld a1, 16(sp) # 8-byte Folded Reload
+; RV64-NEXT:    bgeu a0, a2, .LBB32_3
 ; RV64-NEXT:  # %bb.2: # %end
-; RV64-NEXT:    sub a1, s3, s2
-; RV64-NEXT:    mv a0, s4
-; RV64-NEXT:    sd a1, 0(s0)
+; RV64-NEXT:    sub a2, a0, a2
+; RV64-NEXT:    ld a0, 32(sp) # 8-byte Folded Reload
+; RV64-NEXT:    sd a2, 0(a1)
 ; RV64-NEXT:    j .LBB32_4
 ; RV64-NEXT:  .LBB32_3: # %f
-; RV64-NEXT:    mv a0, s1
+; RV64-NEXT:    mv a0, a3
 ; RV64-NEXT:  .LBB32_4: # %f
 ; RV64-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    .cfi_restore ra
-; RV64-NEXT:    .cfi_restore s0
-; RV64-NEXT:    .cfi_restore s1
-; RV64-NEXT:    .cfi_restore s2
-; RV64-NEXT:    .cfi_restore s3
-; RV64-NEXT:    .cfi_restore s4
 ; RV64-NEXT:    addi sp, sp, 48
 ; RV64-NEXT:    .cfi_def_cfa_offset 0
 ; RV64-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/qci-interrupt-attr-fpr.ll b/llvm/test/CodeGen/RISCV/qci-interrupt-attr-fpr.ll
index a16cc9ef871779..b753980839e080 100644
--- a/llvm/test/CodeGen/RISCV/qci-interrupt-attr-fpr.ll
+++ b/llvm/test/CodeGen/RISCV/qci-interrupt-attr-fpr.ll
@@ -544,29 +544,29 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-F-NEXT:    li a3, 0
 ; QCI-F-NEXT:    li a5, 0
 ; QCI-F-NEXT:    li a7, 0
-; QCI-F-NEXT:    lui t0, 266496
-; QCI-F-NEXT:    li t1, 5
-; QCI-F-NEXT:    lui a0, 260096
-; QCI-F-NEXT:    lui a2, 262144
-; QCI-F-NEXT:    fmv.w.x fa0, a0
-; QCI-F-NEXT:    fmv.w.x fa1, a2
+; QCI-F-NEXT:    lui t0, 260096
+; QCI-F-NEXT:    fmv.w.x fa0, t0
+; QCI-F-NEXT:    lui a0, 262144
+; QCI-F-NEXT:    fmv.w.x fa1, a0
 ; QCI-F-NEXT:    lui a0, 263168
-; QCI-F-NEXT:    lui a2, 264192
+; QCI-F-NEXT:    lui t0, 264192
 ; QCI-F-NEXT:    fmv.w.x fa2, a0
-; QCI-F-NEXT:    fmv.w.x fa3, a2
+; QCI-F-NEXT:    fmv.w.x fa3, t0
 ; QCI-F-NEXT:    lui a0, 264704
-; QCI-F-NEXT:    lui a2, 265216
+; QCI-F-NEXT:    lui t0, 265216
 ; QCI-F-NEXT:    fmv.w.x fa4, a0
-; QCI-F-NEXT:    fmv.w.x fa5, a2
+; QCI-F-NEXT:    fmv.w.x fa5, t0
 ; QCI-F-NEXT:    lui a0, 265728
-; QCI-F-NEXT:    lui a2, 266240
+; QCI-F-NEXT:    lui t0, 266240
 ; QCI-F-NEXT:    fmv.w.x fa6, a0
-; QCI-F-NEXT:    fmv.w.x fa7, a2
+; QCI-F-NEXT:    fmv.w.x fa7, t0
 ; QCI-F-NEXT:    li a0, 1
 ; QCI-F-NEXT:    li a2, 2
 ; QCI-F-NEXT:    li a4, 3
 ; QCI-F-NEXT:    li a6, 4
-; QCI-F-NEXT:    sw t1, 0(sp)
+; QCI-F-NEXT:    li t0, 5
+; QCI-F-NEXT:    sw t0, 0(sp)
+; QCI-F-NEXT:    lui t0, 266496
 ; QCI-F-NEXT:    sw t0, 4(sp)
 ; QCI-F-NEXT:    call function_with_one_stack_arg
 ; QCI-F-NEXT:    call use_float
@@ -681,29 +681,29 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-D-NEXT:    li a3, 0
 ; QCI-D-NEXT:    li a5, 0
 ; QCI-D-NEXT:    li a7, 0
-; QCI-D-NEXT:    lui t0, 266496
-; QCI-D-NEXT:    li t1, 5
-; QCI-D-NEXT:    lui a0, 260096
-; QCI-D-NEXT:    lui a2, 262144
-; QCI-D-NEXT:    fmv.w.x fa0, a0
-; QCI-D-NEXT:    fmv.w.x fa1, a2
+; QCI-D-NEXT:    lui t0, 260096
+; QCI-D-NEXT:    fmv.w.x fa0, t0
+; QCI-D-NEXT:    lui a0, 262144
+; QCI-D-NEXT:    fmv.w.x fa1, a0
 ; QCI-D-NEXT:    lui a0, 263168
-; QCI-D-NEXT:    lui a2, 264192
+; QCI-D-NEXT:    lui t0, 264192
 ; QCI-D-NEXT:    fmv.w.x fa2, a0
-; QCI-D-NEXT:    fmv.w.x fa3, a2
+; QCI-D-NEXT:    fmv.w.x fa3, t0
 ; QCI-D-NEXT:    lui a0, 264704
-; QCI-D-NEXT:    lui a2, 265216
+; QCI-D-NEXT:    lui t0, 265216
 ; QCI-D-NEXT:    fmv.w.x fa4, a0
-; QCI-D-NEXT:    fmv.w.x fa5, a2
+; QCI-D-NEXT:    fmv.w.x fa5, t0
 ; QCI-D-NEXT:    lui a0, 265728
-; QCI-D-NEXT:    lui a2, 266240
+; QCI-D-NEXT:    lui t0, 266240
 ; QCI-D-NEXT:    fmv.w.x fa6, a0
-; QCI-D-NEXT:    fmv.w.x fa7, a2
+; QCI-D-NEXT:    fmv.w.x fa7, t0
 ; QCI-D-NEXT:    li a0, 1
 ; QCI-D-NEXT:    li a2, 2
 ; QCI-D-NEXT:    li a4, 3
 ; QCI-D-NEXT:    li a6, 4
-; QCI-D-NEXT:    sw t1, 0(sp)
+; QCI-D-NEXT:    li t0, 5
+; QCI-D-NEXT:    sw t0, 0(sp)
+; QCI-D-NEXT:    lui t0, 266496
 ; QCI-D-NEXT:    sw t0, 4(sp)
 ; QCI-D-NEXT:    call function_with_one_stack_arg
 ; QCI-D-NEXT:    call use_float
@@ -823,29 +823,29 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-F-NEXT:    li a3, 0
 ; QCI-F-NEXT:    li a5, 0
 ; QCI-F-NEXT:    li a7, 0
-; QCI-F-NEXT:    lui t0, 266496
-; QCI-F-NEXT:    li t1, 5
-; QCI-F-NEXT:    lui a0, 260096
-; QCI-F-NEXT:    lui a2, 262144
-; QCI-F-NEXT:    fmv.w.x fa0, a0
-; QCI-F-NEXT:    fmv.w.x fa1, a2
+; QCI-F-NEXT:    lui t0, 260096
+; QCI-F-NEXT:    fmv.w.x fa0, t0
+; QCI-F-NEXT:    lui a0, 262144
+; QCI-F-NEXT:    fmv.w.x fa1, a0
 ; QCI-F-NEXT:    lui a0, 263168
-; QCI-F-NEXT:    lui a2, 264192
+; QCI-F-NEXT:    lui t0, 264192
 ; QCI-F-NEXT:    fmv.w.x fa2, a0
-; QCI-F-NEXT:    fmv.w.x fa3, a2
+; QCI-F-NEXT:    fmv.w.x fa3, t0
 ; QCI-F-NEXT:    lui a0, 264704
-; QCI-F-NEXT:    lui a2, 265216
+; QCI-F-NEXT:    lui t0, 265216
 ; QCI-F-NEXT:    fmv.w.x fa4, a0
-; QCI-F-NEXT:    fmv.w.x fa5, a2
+; QCI-F-NEXT:    fmv.w.x fa5, t0
 ; QCI-F-NEXT:    lui a0, 265728
-; QCI-F-NEXT:    lui a2, 266240
+; QCI-F-NEXT:    lui t0, 266240
 ; QCI-F-NEXT:    fmv.w.x fa6, a0
-; QCI-F-NEXT:    fmv.w.x fa7, a2
+; QCI-F-NEXT:    fmv.w.x fa7, t0
 ; QCI-F-NEXT:    li a0, 1
 ; QCI-F-NEXT:    li a2, 2
 ; QCI-F-NEXT:    li a4, 3
 ; QCI-F-NEXT:    li a6, 4
-; QCI-F-NEXT:    sw t1, 0(sp)
+; QCI-F-NEXT:    li t0, 5
+; QCI-F-NEXT:    sw t0, 0(sp)
+; QCI-F-NEXT:    lui t0, 266496
 ; QCI-F-NEXT:    sw t0, 4(sp)
 ; QCI-F-NEXT:    call function_with_one_stack_arg
 ; QCI-F-NEXT:    call use_float
@@ -960,29 +960,29 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-D-NEXT:    li a3, 0
 ; QCI-D-NEXT:    li a5, 0
 ; QCI-D-NEXT:    li a7, 0
-; QCI-D-NEXT:    lui t0, 266496
-; QCI-D-NEXT:    li t1, 5
-; QCI-D-NEXT:    lui a0, 260096
-; QCI-D-NEXT:    lui a2, 262144
-; QCI-D-NEXT:    fmv.w.x fa0, a0
-; QCI-D-NEXT:    fmv.w.x fa1, a2
+; QCI-D-NEXT:    lui t0, 260096
+; QCI-D-NEXT:    fmv.w.x fa0, t0
+; QCI-D-NEXT:    lui a0, 262144
+; QCI-D-NEXT:    fmv.w.x fa1, a0
 ; QCI-D-NEXT:    lui a0, 263168
-; QCI-D-NEXT:    lui a2, 264192
+; QCI-D-NEXT:    lui t0, 264192
 ; QCI-D-NEXT:    fmv.w.x fa2, a0
-; QCI-D-NEXT:    fmv.w.x fa3, a2
+; QCI-D-NEXT:    fmv.w.x fa3, t0
 ; QCI-D-NEXT:    lui a0, 264704
-; QCI-D-NEXT:    lui a2, 265216
+; QCI-D-NEXT:    lui t0, 265216
 ; QCI-D-NEXT:    fmv.w.x fa4, a0
-; QCI-D-NEXT:    fmv.w.x fa5, a2
+; QCI-D-NEXT:    fmv.w.x fa5, t0
 ; QCI-D-NEXT:    lui a0, 265728
-; QCI-D-NEXT:    lui a2, 266240
+; QCI-D-NEXT:    lui t0, 266240
 ; QCI-D-NEXT:    fmv.w.x fa6, a0
-; QCI-D-NEXT:    fmv.w.x fa7, a2
+; QCI-D-NEXT:    fmv.w.x fa7, t0
 ; QCI-D-NEXT:    li a0, 1
 ; QCI-D-NEXT:    li a2, 2
 ; QCI-D-NEXT:    li a4, 3
 ; QCI-D-NEXT:    li a6, 4
-; QCI-D-NEXT:    sw t1, 0(sp)
+; QCI-D-NEXT:    li t0, 5
+; QCI-D-NEXT:    sw t0, 0(sp)
+; QCI-D-NEXT:    lui t0, 266496
 ; QCI-D-NEXT:    sw t0, 4(sp)
 ; QCI-D-NEXT:    call function_with_one_stack_arg
 ; QCI-D-NEXT:    call use_float
@@ -1764,7 +1764,8 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-F-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
 ; QCI-F-NEXT:    lw a0, 80(s0)
 ; QCI-F-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
-; QCI-F-NEXT:    lw s11, 84(s0)
+; QCI-F-NEXT:    lw a0, 84(s0)
+; QCI-F-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; QCI-F-NEXT:    lw s1, 88(s0)
 ; QCI-F-NEXT:    lw s2, 92(s0)
 ; QCI-F-NEXT:    lw s3, 96(s0)
@@ -1779,29 +1780,29 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-F-NEXT:    li a3, 0
 ; QCI-F-NEXT:    li a5, 0
 ; QCI-F-NEXT:    li a7, 0
-; QCI-F-NEXT:    lui t0, 266496
-; QCI-F-NEXT:    li t1, 5
-; QCI-F-NEXT:    lui a0, 260096
-; QCI-F-NEXT:    lui a2, 262144
-; QCI-F-NEXT:    fmv.w.x fa0, a0
-; QCI-F-NEXT:    fmv.w.x fa1, a2
-; QCI-F-NEXT:    lui a0, 263168
-; QCI-F-NEXT:    lui a2, 264192
-; QCI-F-NEXT:    fmv.w.x fa2, a0
-; QCI-F-NEXT:    fmv.w.x fa3, a2
-; QCI-F-NEXT:    lui a0, 264704
-; QCI-F-NEXT:    lui a2, 265216
-; QCI-F-NEXT:    fmv.w.x fa4, a0
-; QCI-F-NEXT:    fmv.w.x fa5, a2
-; QCI-F-NEXT:    lui a0, 265728
-; QCI-F-NEXT:    lui a2, 266240
-; QCI-F-NEXT:    fmv.w.x fa6, a0
-; QCI-F-NEXT:    fmv.w.x fa7, a2
+; QCI-F-NEXT:    lui t0, 260096
+; QCI-F-NEXT:    lui s11, 262144
+; QCI-F-NEXT:    fmv.w.x fa0, t0
+; QCI-F-NEXT:    fmv.w.x fa1, s11
+; QCI-F-NEXT:    lui s11, 263168
+; QCI-F-NEXT:    lui t0, 264192
+; QCI-F-NEXT:    fmv.w.x fa2, s11
+; QCI-F-NEXT:    fmv.w.x fa3, t0
+; QCI-F-NEXT:    lui s11, 264704
+; QCI-F-NEXT:    lui t0, 265216
+; QCI-F-NEXT:    fmv.w.x fa4, s11
+; QCI-F-NEXT:    fmv.w.x fa5, t0
+; QCI-F-NEXT:    lui s11, 265728
+; QCI-F-NEXT:    lui t0, 266240
+; QCI-F-NEXT:    fmv.w.x fa6, s11
+; QCI-F-NEXT:    fmv.w.x fa7, t0
 ; QCI-F-NEXT:    li a0, 1
 ; QCI-F-NEXT:    li a2, 2
 ; QCI-F-NEXT:    li a4, 3
 ; QCI-F-NEXT:    li a6, 4
-; QCI-F-NEXT:    sw t1, 0(sp)
+; QCI-F-NEXT:    li t0, 5
+; QCI-F-NEXT:    sw t0, 0(sp)
+; QCI-F-NEXT:    lui t0, 266496
 ; QCI-F-NEXT:    sw t0, 4(sp)
 ; QCI-F-NEXT:    call function_with_one_stack_arg
 ; QCI-F-NEXT:    call use_float
@@ -1815,7 +1816,8 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-F-NEXT:    sw s3, 96(s0)
 ; QCI-F-NEXT:    sw s2, 92(s0)
 ; QCI-F-NEXT:    sw s1, 88(s0)
-; QCI-F-NEXT:    sw s11, 84(s0)
+; QCI-F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; QCI-F-NEXT:    sw a0, 84(s0)
 ; QCI-F-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
 ; QCI-F-NEXT:    sw a0, 80(s0)
 ; QCI-F-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
@@ -2053,7 +2055,8 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-D-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
 ; QCI-D-NEXT:    lw a0, 80(s0)
 ; QCI-D-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
-; QCI-D-NEXT:    lw s11, 84(s0)
+; QCI-D-NEXT:    lw a0, 84(s0)
+; QCI-D-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
 ; QCI-D-NEXT:    lw s1, 88(s0)
 ; QCI-D-NEXT:    lw s2, 92(s0)
 ; QCI-D-NEXT:    lw s3, 96(s0)
@@ -2068,29 +2071,29 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-D-NEXT:    li a3, 0
 ; QCI-D-NEXT:    li a5, 0
 ; QCI-D-NEXT:    li a7, 0
-; QCI-D-NEXT:    lui t0, 266496
-; QCI-D-NEXT:    li t1, 5
-; QCI-D-NEXT:    lui a0, 260096
-; QCI-D-NEXT:    lui a2, 262144
-; QCI-D-NEXT:    fmv.w.x fa0, a0
-; QCI-D-NEXT:    fmv.w.x fa1, a2
-; QCI-D-NEXT:    lui a0, 263168
-; QCI-D-NEXT:    lui a2, 264192
-; QCI-D-NEXT:    fmv.w.x fa2, a0
-; QCI-D-NEXT:    fmv.w.x fa3, a2
-; QCI-D-NEXT:    lui a0, 264704
-; QCI-D-NEXT:    lui a2, 265216
-; QCI-D-NEXT:    fmv.w.x fa4, a0
-; QCI-D-NEXT:    fmv.w.x fa5, a2
-; QCI-D-NEXT:    lui a0, 265728
-; QCI-D-NEXT:    lui a2, 266240
-; QCI-D-NEXT:    fmv.w.x fa6, a0
-; QCI-D-NEXT:    fmv.w.x fa7, a2
+; QCI-D-NEXT:    lui t0, 260096
+; QCI-D-NEXT:    lui s11, 262144
+; QCI-D-NEXT:    fmv.w.x fa0, t0
+; QCI-D-NEXT:    fmv.w.x fa1, s11
+; QCI-D-NEXT:    lui s11, 263168
+; QCI-D-NEXT:    lui t0, 264192
+; QCI-D-NEXT:    fmv.w.x fa2, s11
+; QCI-D-NEXT:    fmv.w.x fa3, t0
+; QCI-D-NEXT:    lui s11, 264704
+; QCI-D-NEXT:    lui t0, 265216
+; QCI-D-NEXT:    fmv.w.x fa4, s11
+; QCI-D-NEXT:    fmv.w.x fa5, t0
+; QCI-D-NEXT:    lui s11, 265728
+; QCI-D-NEXT:    lui t0, 266240
+; QCI-D-NEXT:    fmv.w.x fa6, s11
+; QCI-D-NEXT:    fmv.w.x fa7, t0
 ; QCI-D-NEXT:    li a0, 1
 ; QCI-D-NEXT:    li a2, 2
 ; QCI-D-NEXT:    li a4, 3
 ; QCI-D-NEXT:    li a6, 4
-; QCI-D-NEXT:    sw t1, 0(sp)
+; QCI-D-NEXT:    li t0, 5
+; QCI-D-NEXT:    sw t0, 0(sp)
+; QCI-D-NEXT:    lui t0, 266496
 ; QCI-D-NEXT:    sw t0, 4(sp)
 ; QCI-D-NEXT:    call function_with_one_stack_arg
 ; QCI-D-NEXT:    call use_float
@@ -2104,7 +2107,8 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-D-NEXT:    sw s3, 96(s0)
 ; QCI-D-NEXT:    sw s2, 92(s0)
 ; QCI-D-NEXT:    sw s1, 88(s0)
-; QCI-D-NEXT:    sw s11, 84(s0)
+; QCI-D-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; QCI-D-NEXT:    sw a0, 84(s0)
 ; QCI-D-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
 ; QCI-D-NEXT:    sw a0, 80(s0)
 ; QCI-D-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
@@ -2349,7 +2353,8 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-F-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
 ; QCI-F-NEXT:    lw a0, 80(s0)
 ; QCI-F-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
-; QCI-F-NEXT:    lw s11, 84(s0)
+; QCI-F-NEXT:    lw a0, 84(s0)
+; QCI-F-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
 ; QCI-F-NEXT:    lw s1, 88(s0)
 ; QCI-F-NEXT:    lw s2, 92(s0)
 ; QCI-F-NEXT:    lw s3, 96(s0)
@@ -2364,29 +2369,29 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-F-NEXT:    li a3, 0
 ; QCI-F-NEXT:    li a5, 0
 ; QCI-F-NEXT:    li a7, 0
-; QCI-F-NEXT:    lui t0, 266496
-; QCI-F-NEXT:    li t1, 5
-; QCI-F-NEXT:    lui a0, 260096
-; QCI-F-NEXT:    lui a2, 262144
-; QCI-F-NEXT:    fmv.w.x fa0, a0
-; QCI-F-NEXT:    fmv.w.x fa1, a2
-; QCI-F-NEXT:    lui a0, 263168
-; QCI-F-NEXT:    lui a2, 264192
-; QCI-F-NEXT:    fmv.w.x fa2, a0
-; QCI-F-NEXT:    fmv.w.x fa3, a2
-; QCI-F-NEXT:    lui a0, 264704
-; QCI-F-NEXT:    lui a2, 265216
-; QCI-F-NEXT:    fmv.w.x fa4, a0
-; QCI-F-NEXT:    fmv.w.x fa5, a2
-; QCI-F-NEXT:    lui a0, 265728
-; QCI-F-NEXT:    lui a2, 266240
-; QCI-F-NEXT:    fmv.w.x fa6, a0
-; QCI-F-NEXT:    fmv.w.x fa7, a2
+; QCI-F-NEXT:    lui t0, 260096
+; QCI-F-NEXT:    lui s11, 262144
+; QCI-F-NEXT:    fmv.w.x fa0, t0
+; QCI-F-NEXT:    fmv.w.x fa1, s11
+; QCI-F-NEXT:    lui s11, 263168
+; QCI-F-NEXT:    lui t0, 264192
+; QCI-F-NEXT:    fmv.w.x fa2, s11
+; QCI-F-NEXT:    fmv.w.x fa3, t0
+; QCI-F-NEXT:    lui s11, 264704
+; QCI-F-NEXT:    lui t0, 265216
+; QCI-F-NEXT:    fmv.w.x fa4, s11
+; QCI-F-NEXT:    fmv.w.x fa5, t0
+; QCI-F-NEXT:    lui s11, 265728
+; QCI-F-NEXT:    lui t0, 266240
+; QCI-F-NEXT:    fmv.w.x fa6, s11
+; QCI-F-NEXT:    fmv.w.x fa7, t0
 ; QCI-F-NEXT:    li a0, 1
 ; QCI-F-NEXT:    li a2, 2
 ; QCI-F-NEXT:    li a4, 3
 ; QCI-F-NEXT:    li a6, 4
-; QCI-F-NEXT:    sw t1, 0(sp)
+; QCI-F-NEXT:    li t0, 5
+; QCI-F-NEXT:    sw t0, 0(sp)
+; QCI-F-NEXT:    lui t0, 266496
 ; QCI-F-NEXT:    sw t0, 4(sp)
 ; QCI-F-NEXT:    call function_with_one_stack_arg
 ; QCI-F-NEXT:    call use_float
@@ -2400,7 +2405,8 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-F-NEXT:    sw s3, 96(s0)
 ; QCI-F-NEXT:    sw s2, 92(s0)
 ; QCI-F-NEXT:    sw s1, 88(s0)
-; QCI-F-NEXT:    sw s11, 84(s0)
+; QCI-F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; QCI-F-NEXT:    sw a0, 84(s0)
 ; QCI-F-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
 ; QCI-F-NEXT:    sw a0, 80(s0)
 ; QCI-F-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
@@ -2638,7 +2644,8 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-D-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
 ; QCI-D-NEXT:    lw a0, 80(s0)
 ; QCI-D-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
-; QCI-D-NEXT:    lw s11, 84(s0)
+; QCI-D-NEXT:    lw a0, 84(s0)
+; QCI-D-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
 ; QCI-D-NEXT:    lw s1, 88(s0)
 ; QCI-D-NEXT:    lw s2, 92(s0)
 ; QCI-D-NEXT:    lw s3, 96(s0)
@@ -2653,29 +2660,29 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-D-NEXT:    li a3, 0
 ; QCI-D-NEXT:    li a5, 0
 ; QCI-D-NEXT:    li a7, 0
-; QCI-D-NEXT:    lui t0, 266496
-; QCI-D-NEXT:    li t1, 5
-; QCI-D-NEXT:    lui a0, 260096
-; QCI-D-NEXT:    lui a2, 262144
-; QCI-D-NEXT:    fmv.w.x fa0, a0
-; QCI-D-NEXT:    fmv.w.x fa1, a2
-; QCI-D-NEXT:    lui a0, 263168
-; QCI-D-NEXT:    lui a2, 264192
-; QCI-D-NEXT:    fmv.w.x fa2, a0
-; QCI-D-NEXT:    fmv.w.x fa3, a2
-; QCI-D-NEXT:    lui a0, 264704
-; QCI-D-NEXT:    lui a2, 265216
-; QCI-D-NEXT:    fmv.w.x fa4, a0
-; QCI-D-NEXT:    fmv.w.x fa5, a2
-; QCI-D-NEXT:    lui a0, 265728
-; QCI-D-NEXT:    lui a2, 266240
-; QCI-D-NEXT:    fmv.w.x fa6, a0
-; QCI-D-NEXT:    fmv.w.x fa7, a2
+; QCI-D-NEXT:    lui t0, 260096
+; QCI-D-NEXT:    lui s11, 262144
+; QCI-D-NEXT:    fmv.w.x fa0, t0
+; QCI-D-NEXT:    fmv.w.x fa1, s11
+; QCI-D-NEXT:    lui s11, 263168
+; QCI-D-NEXT:    lui t0, 264192
+; QCI-D-NEXT:    fmv.w.x fa2, s11
+; QCI-D-NEXT:    fmv.w.x fa3, t0
+; QCI-D-NEXT:    lui s11, 264704
+; QCI-D-NEXT:    lui t0, 265216
+; QCI-D-NEXT:    fmv.w.x fa4, s11
+; QCI-D-NEXT:    fmv.w.x fa5, t0
+; QCI-D-NEXT:    lui s11, 265728
+; QCI-D-NEXT:    lui t0, 266240
+; QCI-D-NEXT:    fmv.w.x fa6, s11
+; QCI-D-NEXT:    fmv.w.x fa7, t0
 ; QCI-D-NEXT:    li a0, 1
 ; QCI-D-NEXT:    li a2, 2
 ; QCI-D-NEXT:    li a4, 3
 ; QCI-D-NEXT:    li a6, 4
-; QCI-D-NEXT:    sw t1, 0(sp)
+; QCI-D-NEXT:    li t0, 5
+; QCI-D-NEXT:    sw t0, 0(sp)
+; QCI-D-NEXT:    lui t0, 266496
 ; QCI-D-NEXT:    sw t0, 4(sp)
 ; QCI-D-NEXT:    call function_with_one_stack_arg
 ; QCI-D-NEXT:    call use_float
@@ -2689,7 +2696,8 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-D-NEXT:    sw s3, 96(s0)
 ; QCI-D-NEXT:    sw s2, 92(s0)
 ; QCI-D-NEXT:    sw s1, 88(s0)
-; QCI-D-NEXT:    sw s11, 84(s0)
+; QCI-D-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; QCI-D-NEXT:    sw a0, 84(s0)
 ; QCI-D-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
 ; QCI-D-NEXT:    sw a0, 80(s0)
 ; QCI-D-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/qci-interrupt-attr.ll b/llvm/test/CodeGen/RISCV/qci-interrupt-attr.ll
index 8551b7813862f0..3e5a94274dda48 100644
--- a/llvm/test/CodeGen/RISCV/qci-interrupt-attr.ll
+++ b/llvm/test/CodeGen/RISCV/qci-interrupt-attr.ll
@@ -601,7 +601,6 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-NEXT:    .cfi_offset t6, -80
 ; QCI-NEXT:    addi sp, sp, -16
 ; QCI-NEXT:    .cfi_def_cfa_offset 112
-; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    li a2, 1
 ; QCI-NEXT:    li a4, 2
 ; QCI-NEXT:    li a6, 3
@@ -610,6 +609,7 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-NEXT:    li a3, 0
 ; QCI-NEXT:    li a5, 0
 ; QCI-NEXT:    li a7, 0
+; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    sw t0, 0(sp)
 ; QCI-NEXT:    call function_with_one_stack_arg
 ; QCI-NEXT:    call use_i64
@@ -641,7 +641,6 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-FP-NEXT:    addi sp, sp, -16
 ; QCI-FP-NEXT:    .cfi_def_cfa_offset 112
 ; QCI-FP-NEXT:    .cfi_def_cfa s0, 0
-; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    li a2, 1
 ; QCI-FP-NEXT:    li a4, 2
 ; QCI-FP-NEXT:    li a6, 3
@@ -650,6 +649,7 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-FP-NEXT:    li a3, 0
 ; QCI-FP-NEXT:    li a5, 0
 ; QCI-FP-NEXT:    li a7, 0
+; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    sw t0, 0(sp)
 ; QCI-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-FP-NEXT:    call use_i64
@@ -681,7 +681,6 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-PUSH-POP-NEXT:    .cfi_offset t6, -80
 ; QCI-PUSH-POP-NEXT:    addi sp, sp, -16
 ; QCI-PUSH-POP-NEXT:    .cfi_def_cfa_offset 112
-; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    li a2, 1
 ; QCI-PUSH-POP-NEXT:    li a4, 2
 ; QCI-PUSH-POP-NEXT:    li a6, 3
@@ -690,6 +689,7 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-PUSH-POP-NEXT:    li a3, 0
 ; QCI-PUSH-POP-NEXT:    li a5, 0
 ; QCI-PUSH-POP-NEXT:    li a7, 0
+; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-PUSH-POP-NEXT:    call use_i64
@@ -720,7 +720,6 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    .cfi_offset t6, -80
 ; QCI-QCCMP-PUSH-POP-NEXT:    addi sp, sp, -16
 ; QCI-QCCMP-PUSH-POP-NEXT:    .cfi_def_cfa_offset 112
-; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a6, 3
@@ -729,6 +728,7 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-NEXT:    call use_i64
@@ -760,7 +760,6 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    addi sp, sp, -16
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    .cfi_def_cfa_offset 112
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    .cfi_def_cfa s0, 0
-; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a6, 3
@@ -769,6 +768,7 @@ define void @test_nest_call() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call use_i64
@@ -805,7 +805,6 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-NEXT:    .cfi_offset t6, -80
 ; QCI-NEXT:    addi sp, sp, -16
 ; QCI-NEXT:    .cfi_def_cfa_offset 112
-; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    li a2, 1
 ; QCI-NEXT:    li a4, 2
 ; QCI-NEXT:    li a6, 3
@@ -814,6 +813,7 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-NEXT:    li a3, 0
 ; QCI-NEXT:    li a5, 0
 ; QCI-NEXT:    li a7, 0
+; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    sw t0, 0(sp)
 ; QCI-NEXT:    call function_with_one_stack_arg
 ; QCI-NEXT:    call use_i64
@@ -845,7 +845,6 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-FP-NEXT:    addi sp, sp, -16
 ; QCI-FP-NEXT:    .cfi_def_cfa_offset 112
 ; QCI-FP-NEXT:    .cfi_def_cfa s0, 0
-; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    li a2, 1
 ; QCI-FP-NEXT:    li a4, 2
 ; QCI-FP-NEXT:    li a6, 3
@@ -854,6 +853,7 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-FP-NEXT:    li a3, 0
 ; QCI-FP-NEXT:    li a5, 0
 ; QCI-FP-NEXT:    li a7, 0
+; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    sw t0, 0(sp)
 ; QCI-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-FP-NEXT:    call use_i64
@@ -885,7 +885,6 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-PUSH-POP-NEXT:    .cfi_offset t6, -80
 ; QCI-PUSH-POP-NEXT:    addi sp, sp, -16
 ; QCI-PUSH-POP-NEXT:    .cfi_def_cfa_offset 112
-; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    li a2, 1
 ; QCI-PUSH-POP-NEXT:    li a4, 2
 ; QCI-PUSH-POP-NEXT:    li a6, 3
@@ -894,6 +893,7 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-PUSH-POP-NEXT:    li a3, 0
 ; QCI-PUSH-POP-NEXT:    li a5, 0
 ; QCI-PUSH-POP-NEXT:    li a7, 0
+; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-PUSH-POP-NEXT:    call use_i64
@@ -924,7 +924,6 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    .cfi_offset t6, -80
 ; QCI-QCCMP-PUSH-POP-NEXT:    addi sp, sp, -16
 ; QCI-QCCMP-PUSH-POP-NEXT:    .cfi_def_cfa_offset 112
-; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a6, 3
@@ -933,6 +932,7 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-NEXT:    call use_i64
@@ -964,7 +964,6 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    addi sp, sp, -16
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    .cfi_def_cfa_offset 112
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    .cfi_def_cfa s0, 0
-; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a6, 3
@@ -973,6 +972,7 @@ define void @test_nonest_call() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call use_i64
@@ -2452,7 +2452,6 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-NEXT:    lw s8, 116(s0)
 ; QCI-NEXT:    lw s9, 120(s0)
 ; QCI-NEXT:    lw s10, 124(s0)
-; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    li a2, 1
 ; QCI-NEXT:    li a4, 2
 ; QCI-NEXT:    li a6, 3
@@ -2461,6 +2460,7 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-NEXT:    li a3, 0
 ; QCI-NEXT:    li a5, 0
 ; QCI-NEXT:    li a7, 0
+; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    sw t0, 0(sp)
 ; QCI-NEXT:    call function_with_one_stack_arg
 ; QCI-NEXT:    call use_i64
@@ -2645,7 +2645,6 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-FP-NEXT:    lw s11, 116(s1)
 ; QCI-FP-NEXT:    lw s2, 120(s1)
 ; QCI-FP-NEXT:    lw s3, 124(s1)
-; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    li a2, 1
 ; QCI-FP-NEXT:    li a4, 2
 ; QCI-FP-NEXT:    li a6, 3
@@ -2654,6 +2653,7 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-FP-NEXT:    li a3, 0
 ; QCI-FP-NEXT:    li a5, 0
 ; QCI-FP-NEXT:    li a7, 0
+; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    sw t0, 0(sp)
 ; QCI-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-FP-NEXT:    call use_i64
@@ -2829,7 +2829,6 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-PUSH-POP-NEXT:    lw s8, 116(s0)
 ; QCI-PUSH-POP-NEXT:    lw s9, 120(s0)
 ; QCI-PUSH-POP-NEXT:    lw s10, 124(s0)
-; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    li a2, 1
 ; QCI-PUSH-POP-NEXT:    li a4, 2
 ; QCI-PUSH-POP-NEXT:    li a6, 3
@@ -2838,6 +2837,7 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-PUSH-POP-NEXT:    li a3, 0
 ; QCI-PUSH-POP-NEXT:    li a5, 0
 ; QCI-PUSH-POP-NEXT:    li a7, 0
+; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-PUSH-POP-NEXT:    call use_i64
@@ -3002,7 +3002,6 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    lw s8, 116(s0)
 ; QCI-QCCMP-PUSH-POP-NEXT:    lw s9, 120(s0)
 ; QCI-QCCMP-PUSH-POP-NEXT:    lw s10, 124(s0)
-; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a6, 3
@@ -3011,6 +3010,7 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-NEXT:    call use_i64
@@ -3177,7 +3177,6 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    lw s11, 116(s1)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    lw s2, 120(s1)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    lw s3, 124(s1)
-; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a6, 3
@@ -3186,6 +3185,7 @@ define void @test_spill_call_nest() "interrupt"="qci-nest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call use_i64
@@ -3368,7 +3368,6 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-NEXT:    lw s8, 116(s0)
 ; QCI-NEXT:    lw s9, 120(s0)
 ; QCI-NEXT:    lw s10, 124(s0)
-; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    li a2, 1
 ; QCI-NEXT:    li a4, 2
 ; QCI-NEXT:    li a6, 3
@@ -3377,6 +3376,7 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-NEXT:    li a3, 0
 ; QCI-NEXT:    li a5, 0
 ; QCI-NEXT:    li a7, 0
+; QCI-NEXT:    li t0, 4
 ; QCI-NEXT:    sw t0, 0(sp)
 ; QCI-NEXT:    call function_with_one_stack_arg
 ; QCI-NEXT:    call use_i64
@@ -3561,7 +3561,6 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-FP-NEXT:    lw s11, 116(s1)
 ; QCI-FP-NEXT:    lw s2, 120(s1)
 ; QCI-FP-NEXT:    lw s3, 124(s1)
-; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    li a2, 1
 ; QCI-FP-NEXT:    li a4, 2
 ; QCI-FP-NEXT:    li a6, 3
@@ -3570,6 +3569,7 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-FP-NEXT:    li a3, 0
 ; QCI-FP-NEXT:    li a5, 0
 ; QCI-FP-NEXT:    li a7, 0
+; QCI-FP-NEXT:    li t0, 4
 ; QCI-FP-NEXT:    sw t0, 0(sp)
 ; QCI-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-FP-NEXT:    call use_i64
@@ -3745,7 +3745,6 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-PUSH-POP-NEXT:    lw s8, 116(s0)
 ; QCI-PUSH-POP-NEXT:    lw s9, 120(s0)
 ; QCI-PUSH-POP-NEXT:    lw s10, 124(s0)
-; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    li a2, 1
 ; QCI-PUSH-POP-NEXT:    li a4, 2
 ; QCI-PUSH-POP-NEXT:    li a6, 3
@@ -3754,6 +3753,7 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-PUSH-POP-NEXT:    li a3, 0
 ; QCI-PUSH-POP-NEXT:    li a5, 0
 ; QCI-PUSH-POP-NEXT:    li a7, 0
+; QCI-PUSH-POP-NEXT:    li t0, 4
 ; QCI-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-PUSH-POP-NEXT:    call use_i64
@@ -3918,7 +3918,6 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    lw s8, 116(s0)
 ; QCI-QCCMP-PUSH-POP-NEXT:    lw s9, 120(s0)
 ; QCI-QCCMP-PUSH-POP-NEXT:    lw s10, 124(s0)
-; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a6, 3
@@ -3927,6 +3926,7 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-NEXT:    call use_i64
@@ -4093,7 +4093,6 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    lw s11, 116(s1)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    lw s2, 120(s1)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    lw s3, 124(s1)
-; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a2, 1
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a4, 2
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a6, 3
@@ -4102,6 +4101,7 @@ define void @test_spill_call_nonest() "interrupt"="qci-nonest" {
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a3, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a5, 0
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    li a7, 0
+; QCI-QCCMP-PUSH-POP-FP-NEXT:    li t0, 4
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    sw t0, 0(sp)
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call function_with_one_stack_arg
 ; QCI-QCCMP-PUSH-POP-FP-NEXT:    call use_i64
diff --git a/llvm/test/CodeGen/RISCV/regalloc-csr-cost-patterns.ll b/llvm/test/CodeGen/RISCV/regalloc-csr-cost-patterns.ll
index 97ba7f1391386f..2347cc420d02a9 100644
--- a/llvm/test/CodeGen/RISCV/regalloc-csr-cost-patterns.ll
+++ b/llvm/test/CodeGen/RISCV/regalloc-csr-cost-patterns.ll
@@ -4,81 +4,70 @@
 define signext i32 @remat_const_across_call_in_loop(ptr %toks, ptr %pat, i32 signext %varid) nounwind {
 ; CHECK-LABEL: remat_const_across_call_in_loop:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -64
-; CHECK-NEXT:    sd ra, 56(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 48(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 40(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 32(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s3, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s4, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s5, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s6, 0(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    mv s0, a2
-; CHECK-NEXT:    mv a2, a0
+; CHECK-NEXT:    mv a3, a0
 ; CHECK-NEXT:    ld a0, 0(a0)
-; CHECK-NEXT:    beqz a0, .LBB0_11
+; CHECK-NEXT:    beqz a0, .LBB0_13
 ; CHECK-NEXT:  # %bb.1: # %if.end.preheader
-; CHECK-NEXT:    mv s1, a1
-; CHECK-NEXT:    addi s2, a2, 8
-; CHECK-NEXT:    li s3, 37
-; CHECK-NEXT:    li s4, 1
-; CHECK-NEXT:    li s5, -1
-; CHECK-NEXT:    li s6, 2
+; CHECK-NEXT:    addi sp, sp, -32
+; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    addi s0, a3, 8
+; CHECK-NEXT:    li a3, 37
+; CHECK-NEXT:    sd a2, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:  .LBB0_2: # %if.end
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    lbu a1, 0(s1)
-; CHECK-NEXT:    bne a1, s3, .LBB0_7
+; CHECK-NEXT:    lbu a4, 0(a1)
+; CHECK-NEXT:    bne a4, a3, .LBB0_7
 ; CHECK-NEXT:  # %bb.3: # %if.then2
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT:    mv a1, s1
-; CHECK-NEXT:    mv a2, s0
+; CHECK-NEXT:    sd a1, 8(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    call multi_compare
-; CHECK-NEXT:    beq a0, s4, .LBB0_9
+; CHECK-NEXT:    li a1, 1
+; CHECK-NEXT:    beq a0, a1, .LBB0_9
 ; CHECK-NEXT:  # %bb.4: # %if.then2
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT:    beq a0, s5, .LBB0_11
+; CHECK-NEXT:    li a1, -1
+; CHECK-NEXT:    beq a0, a1, .LBB0_12
 ; CHECK-NEXT:  # %bb.5: # %if.then2
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT:    bne a0, s6, .LBB0_10
+; CHECK-NEXT:    li a1, 2
+; CHECK-NEXT:    bne a0, a1, .LBB0_11
 ; CHECK-NEXT:  # %bb.6: # %if.then14
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT:    addi s1, s1, 2
-; CHECK-NEXT:    ld a0, 0(s2)
-; CHECK-NEXT:    addi s1, s1, 1
-; CHECK-NEXT:    addi s2, s2, 8
-; CHECK-NEXT:    bnez a0, .LBB0_2
-; CHECK-NEXT:    j .LBB0_11
+; CHECK-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi a1, a1, 2
+; CHECK-NEXT:    j .LBB0_10
 ; CHECK-NEXT:  .LBB0_7: # %if.else22
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
 ; CHECK-NEXT:    lbu a0, 0(a0)
-; CHECK-NEXT:    bne a0, a1, .LBB0_10
+; CHECK-NEXT:    bne a0, a4, .LBB0_11
 ; CHECK-NEXT:  # %bb.8: # %cleanup31
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT:    ld a0, 0(s2)
-; CHECK-NEXT:    addi s1, s1, 1
-; CHECK-NEXT:    addi s2, s2, 8
+; CHECK-NEXT:    ld a0, 0(s0)
+; CHECK-NEXT:    addi a1, a1, 1
+; CHECK-NEXT:    addi s0, s0, 8
 ; CHECK-NEXT:    bnez a0, .LBB0_2
-; CHECK-NEXT:    j .LBB0_11
+; CHECK-NEXT:    j .LBB0_12
 ; CHECK-NEXT:  .LBB0_9: # %if.then17
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT:    addi s1, s1, 1
-; CHECK-NEXT:    ld a0, 0(s2)
-; CHECK-NEXT:    addi s1, s1, 1
-; CHECK-NEXT:    addi s2, s2, 8
+; CHECK-NEXT:    ld a1, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi a1, a1, 1
+; CHECK-NEXT:  .LBB0_10: # %cleanup31
+; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
+; CHECK-NEXT:    ld a2, 0(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    li a3, 37
+; CHECK-NEXT:    ld a0, 0(s0)
+; CHECK-NEXT:    addi a1, a1, 1
+; CHECK-NEXT:    addi s0, s0, 8
 ; CHECK-NEXT:    bnez a0, .LBB0_2
-; CHECK-NEXT:    j .LBB0_11
-; CHECK-NEXT:  .LBB0_10:
+; CHECK-NEXT:    j .LBB0_12
+; CHECK-NEXT:  .LBB0_11:
 ; CHECK-NEXT:    li a0, 0
-; CHECK-NEXT:  .LBB0_11: # %return
-; CHECK-NEXT:    ld ra, 56(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 48(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 40(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 32(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s3, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s4, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s5, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s6, 0(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 64
+; CHECK-NEXT:  .LBB0_12:
+; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 32
+; CHECK-NEXT:  .LBB0_13: # %return
 ; CHECK-NEXT:    ret
 entry:
   %0 = load ptr, ptr %toks, align 8
@@ -135,77 +124,70 @@ declare signext i32 @multi_compare(ptr, ptr, i32 signext)
 define i32 @evict_hoisted_consts(ptr %tok, ptr %p, i8 %c0, i1 %c1, i8 %c2, i1 %c3, i8 %c4, i1 %c5, i1 %c6) nounwind {
 ; CHECK-LABEL: evict_hoisted_consts:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -16
-; CHECK-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 0(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    ld t3, 16(sp)
+; CHECK-NEXT:    ld t1, 0(sp)
 ; CHECK-NEXT:    zext.b a2, a2
 ; CHECK-NEXT:    li t0, 111
 ; CHECK-NEXT:    andi a3, a3, 1
 ; CHECK-NEXT:    andi a5, a5, 1
 ; CHECK-NEXT:    zext.b a6, a6
-; CHECK-NEXT:    li t1, 6
 ; CHECK-NEXT:    zext.b a4, a4
-; CHECK-NEXT:    li t2, 1
 ; CHECK-NEXT:    andi a7, a7, 1
-; CHECK-NEXT:    andi t3, t3, 1
+; CHECK-NEXT:    andi t3, t1, 1
 ; CHECK-NEXT:    li t4, 116
 ; CHECK-NEXT:    li t5, 118
 ; CHECK-NEXT:    li t6, 124
-; CHECK-NEXT:    li s0, 32
+; CHECK-NEXT:    li t1, 32
 ; CHECK-NEXT:  .LBB1_1: # %for.cond
 ; CHECK-NEXT:    # =>This Loop Header: Depth=1
-; CHECK-NEXT:    # Child Loop BB1_4 Depth 2
-; CHECK-NEXT:    beq a2, t0, .LBB1_7
+; CHECK-NEXT:    # Child Loop BB1_5 Depth 2
+; CHECK-NEXT:    beq a2, t0, .LBB1_8
 ; CHECK-NEXT:  # %bb.2: # %for.cond
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
-; CHECK-NEXT:    beq a2, t4, .LBB1_14
+; CHECK-NEXT:    beq a2, t4, .LBB1_15
 ; CHECK-NEXT:  # %bb.3: # %for.cond
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
-; CHECK-NEXT:    beq a2, t5, .LBB1_14
-; CHECK-NEXT:  .LBB1_4: # %do.body
+; CHECK-NEXT:    beq a2, t5, .LBB1_15
+; CHECK-NEXT:  # %bb.4: # %for.cond.do.body_crit_edge
+; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
+; CHECK-NEXT:    lbu t2, 0(a0)
+; CHECK-NEXT:  .LBB1_5: # %do.body
 ; CHECK-NEXT:    # Parent Loop BB1_1 Depth=1
 ; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
-; CHECK-NEXT:    lbu s1, 0(a0)
-; CHECK-NEXT:    beqz s1, .LBB1_14
-; CHECK-NEXT:  # %bb.5: # %do.body
-; CHECK-NEXT:    # in Loop: Header=BB1_4 Depth=2
-; CHECK-NEXT:    beq s1, t6, .LBB1_1
+; CHECK-NEXT:    beqz t2, .LBB1_15
 ; CHECK-NEXT:  # %bb.6: # %do.body
-; CHECK-NEXT:    # in Loop: Header=BB1_4 Depth=2
-; CHECK-NEXT:    bne s1, s0, .LBB1_4
-; CHECK-NEXT:    j .LBB1_14
-; CHECK-NEXT:  .LBB1_7: # %sw.o
+; CHECK-NEXT:    # in Loop: Header=BB1_5 Depth=2
+; CHECK-NEXT:    beq t2, t6, .LBB1_1
+; CHECK-NEXT:  # %bb.7: # %do.body
+; CHECK-NEXT:    # in Loop: Header=BB1_5 Depth=2
+; CHECK-NEXT:    bne t2, t1, .LBB1_5
+; CHECK-NEXT:    j .LBB1_15
+; CHECK-NEXT:  .LBB1_8: # %sw.o
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
-; CHECK-NEXT:    beqz a3, .LBB1_10
-; CHECK-NEXT:  # %bb.8: # %if.then
+; CHECK-NEXT:    beqz a3, .LBB1_11
+; CHECK-NEXT:  # %bb.9: # %if.then
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
-; CHECK-NEXT:    beqz a5, .LBB1_12
-; CHECK-NEXT:  # %bb.9: # %if.then2
+; CHECK-NEXT:    beqz a5, .LBB1_13
+; CHECK-NEXT:  # %bb.10: # %if.then2
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
-; CHECK-NEXT:    bgeu a6, t1, .LBB1_1
-; CHECK-NEXT:    j .LBB1_14
-; CHECK-NEXT:  .LBB1_10: # %if.else
+; CHECK-NEXT:    li t2, 6
+; CHECK-NEXT:    bgeu a6, t2, .LBB1_1
+; CHECK-NEXT:    j .LBB1_15
+; CHECK-NEXT:  .LBB1_11: # %if.else
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
 ; CHECK-NEXT:    beqz a7, .LBB1_1
-; CHECK-NEXT:  # %bb.11: # %if.else3
+; CHECK-NEXT:  # %bb.12: # %if.else3
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
 ; CHECK-NEXT:    beqz t3, .LBB1_1
-; CHECK-NEXT:    j .LBB1_14
-; CHECK-NEXT:  .LBB1_12: # %if.else2
+; CHECK-NEXT:    j .LBB1_15
+; CHECK-NEXT:  .LBB1_13: # %if.else2
 ; CHECK-NEXT:    # in Loop: Header=BB1_1 Depth=1
+; CHECK-NEXT:    li t2, 1
 ; CHECK-NEXT:    bne a4, t2, .LBB1_1
-; CHECK-NEXT:  # %bb.13: # %cold
+; CHECK-NEXT:  # %bb.14: # %cold
 ; CHECK-NEXT:    mv a0, a1
-; CHECK-NEXT:    ld s0, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 0(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-NEXT:    tail internal_error
-; CHECK-NEXT:  .LBB1_14: # %ret0
+; CHECK-NEXT:  .LBB1_15: # %ret0
 ; CHECK-NEXT:    li a0, 0
-; CHECK-NEXT:    ld s0, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 0(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-NEXT:    ret
 entry:
   br label %for.cond
@@ -261,13 +243,9 @@ declare i32 @internal_error(ptr)
 define i64 @csr_only_in_cold_case(ptr %a, ptr %b, i64 %n, ptr %m) nounwind {
 ; CHECK-LABEL: csr_only_in_cold_case:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -48
-; CHECK-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    addi sp, sp, -64
+; CHECK-NEXT:    sd ra, 56(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s0, 48(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    li a4, 6
 ; CHECK-NEXT:    bltu a4, a2, .LBB2_3
 ; CHECK-NEXT:  # %bb.1: # %entry
@@ -286,16 +264,17 @@ define i64 @csr_only_in_cold_case(ptr %a, ptr %b, i64 %n, ptr %m) nounwind {
 ; CHECK-NEXT:    addi a4, a0, 48
 ; CHECK-NEXT:    addi a5, a1, 48
 ; CHECK-NEXT:    addi a2, a2, -6
-; CHECK-NEXT:    mv s1, a0
+; CHECK-NEXT:    sd a0, 32(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    mv a0, a4
-; CHECK-NEXT:    mv s2, a1
+; CHECK-NEXT:    sd a1, 24(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    mv a1, a5
-; CHECK-NEXT:    mv s0, a3
+; CHECK-NEXT:    sd a3, 40(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a3, 40(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    call csr_only_in_cold_case
-; CHECK-NEXT:    mv a1, s2
-; CHECK-NEXT:    mv a3, s0
-; CHECK-NEXT:    mv a2, a0
-; CHECK-NEXT:    mv a0, s1
+; CHECK-NEXT:    ld a1, 24(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a3, 40(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a0, 32(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    j .LBB2_7
 ; CHECK-NEXT:  .LBB2_4: # %sw.bb3
 ; CHECK-NEXT:    ld a2, 0(a0)
@@ -324,50 +303,63 @@ define i64 @csr_only_in_cold_case(ptr %a, ptr %b, i64 %n, ptr %m) nounwind {
 ; CHECK-NEXT:    add a0, a2, a0
 ; CHECK-NEXT:    j .LBB2_8
 ; CHECK-NEXT:  .LBB2_6:
-; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    sd zero, 8(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:  .LBB2_7: # %big
-; CHECK-NEXT:    ld a4, 0(a0)
-; CHECK-NEXT:    ld a5, 8(a0)
-; CHECK-NEXT:    ld a6, 16(a0)
+; CHECK-NEXT:    ld a2, 0(a0)
+; CHECK-NEXT:    sd a2, 32(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a2, 8(a0)
+; CHECK-NEXT:    sd a2, 24(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a2, 16(a0)
+; CHECK-NEXT:    sd a2, 16(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    ld a7, 24(a0)
-; CHECK-NEXT:    ld t0, 0(a1)
+; CHECK-NEXT:    ld a2, 0(a1)
+; CHECK-NEXT:    sd a2, 40(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    ld t1, 8(a1)
 ; CHECK-NEXT:    ld t2, 16(a1)
 ; CHECK-NEXT:    ld t3, 24(a1)
 ; CHECK-NEXT:    ld t4, 32(a1)
 ; CHECK-NEXT:    ld t5, 32(a0)
-; CHECK-NEXT:    ld a0, 40(a0)
+; CHECK-NEXT:    ld a6, 40(a0)
 ; CHECK-NEXT:    ld a1, 40(a1)
-; CHECK-NEXT:    xor t6, t0, a4
-; CHECK-NEXT:    xor s0, t1, a5
-; CHECK-NEXT:    xor s1, t2, a6
-; CHECK-NEXT:    xor s2, t3, a7
-; CHECK-NEXT:    xor s3, t4, t5
-; CHECK-NEXT:    xor s4, a1, a0
-; CHECK-NEXT:    add a2, t6, a2
-; CHECK-NEXT:    add s0, s0, s1
-; CHECK-NEXT:    add s2, s2, s3
-; CHECK-NEXT:    add a2, a2, s0
-; CHECK-NEXT:    add s2, s2, s4
-; CHECK-NEXT:    add a1, a1, a4
-; CHECK-NEXT:    add a5, t4, a5
-; CHECK-NEXT:    srli a1, a1, 1
-; CHECK-NEXT:    srli a5, a5, 2
-; CHECK-NEXT:    add a6, t3, a6
+; CHECK-NEXT:    ld a2, 40(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a4, 32(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    xor t6, a2, a4
+; CHECK-NEXT:    ld a2, 24(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    xor a2, t1, a2
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    xor t0, t2, a0
+; CHECK-NEXT:    xor a4, t3, a7
+; CHECK-NEXT:    xor a5, t4, t5
+; CHECK-NEXT:    xor a0, a1, a6
+; CHECK-NEXT:    ld s0, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    add t6, t6, s0
+; CHECK-NEXT:    add a2, a2, t0
+; CHECK-NEXT:    add a4, a4, a5
+; CHECK-NEXT:    add t6, t6, a2
+; CHECK-NEXT:    add a4, a4, a0
+; CHECK-NEXT:    ld a2, 32(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    add a2, a1, a2
+; CHECK-NEXT:    ld a1, 24(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    add a1, t4, a1
+; CHECK-NEXT:    srli a2, a2, 1
+; CHECK-NEXT:    srli a5, a1, 2
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    add a0, t3, a0
 ; CHECK-NEXT:    add a7, a7, t2
-; CHECK-NEXT:    srli a4, a6, 3
-; CHECK-NEXT:    srli a6, a7, 4
+; CHECK-NEXT:    srli a0, a0, 3
+; CHECK-NEXT:    srli a7, a7, 4
 ; CHECK-NEXT:    add t1, t5, t1
-; CHECK-NEXT:    add a0, a0, t0
-; CHECK-NEXT:    srli a7, t1, 5
-; CHECK-NEXT:    srli a0, a0, 6
-; CHECK-NEXT:    add a4, a4, a6
-; CHECK-NEXT:    add a5, a7, a5
-; CHECK-NEXT:    add a4, a4, a5
+; CHECK-NEXT:    ld a1, 40(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    add a1, a6, a1
+; CHECK-NEXT:    srli a6, t1, 5
+; CHECK-NEXT:    srli a1, a1, 6
+; CHECK-NEXT:    add a0, a0, a7
+; CHECK-NEXT:    add a5, a6, a5
+; CHECK-NEXT:    add a0, a0, a5
+; CHECK-NEXT:    add a1, a1, a2
+; CHECK-NEXT:    add a4, t6, a4
 ; CHECK-NEXT:    add a0, a0, a1
-; CHECK-NEXT:    add a2, a2, s2
-; CHECK-NEXT:    add a0, a4, a0
-; CHECK-NEXT:    xor a0, a0, a2
+; CHECK-NEXT:    xor a0, a0, a4
 ; CHECK-NEXT:  .LBB2_8: # %sw.epilog
 ; CHECK-NEXT:    ld a1, 8(a3)
 ; CHECK-NEXT:    ld a2, 0(a3)
@@ -378,13 +370,9 @@ define i64 @csr_only_in_cold_case(ptr %a, ptr %b, i64 %n, ptr %m) nounwind {
 ; CHECK-NEXT:    sub a2, a0, a1
 ; CHECK-NEXT:  .LBB2_9: # %cleanup
 ; CHECK-NEXT:    mv a0, a2
-; CHECK-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 48
+; CHECK-NEXT:    ld ra, 56(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s0, 48(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 64
 ; CHECK-NEXT:    ret
 entry:
   switch i64 %n, label %sw.default [
@@ -521,9 +509,6 @@ define i64 @spill_on_warm_call_path(ptr %c, i32 signext %idx) nounwind {
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    addi sp, sp, -32
 ; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    li a2, 6
 ; CHECK-NEXT:    beq a1, a2, .LBB3_3
 ; CHECK-NEXT:  # %bb.1: # %entry
@@ -535,18 +520,17 @@ define i64 @spill_on_warm_call_path(ptr %c, i32 signext %idx) nounwind {
 ; CHECK-NEXT:    add a0, a0, a1
 ; CHECK-NEXT:    ld a0, 8(a0)
 ; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    ret
 ; CHECK-NEXT:  .LBB3_3: # %if.then
 ; CHECK-NEXT:    ld a1, 0(a0)
 ; CHECK-NEXT:    ld a2, 0(a1)
 ; CHECK-NEXT:    li a1, 6
-; CHECK-NEXT:    mv s0, a0
+; CHECK-NEXT:    sd a0, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    jalr a2
-; CHECK-NEXT:    ld a1, 0(s0)
+; CHECK-NEXT:    ld a1, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a1, 0(a1)
 ; CHECK-NEXT:    lui a2, 24
 ; CHECK-NEXT:    and a3, a0, a2
 ; CHECK-NEXT:    xor a2, a3, a2
@@ -555,50 +539,48 @@ define i64 @spill_on_warm_call_path(ptr %c, i32 signext %idx) nounwind {
 ; CHECK-NEXT:    slli a1, a1, 63
 ; CHECK-NEXT:    or a2, a0, a1
 ; CHECK-NEXT:    li a1, 6
-; CHECK-NEXT:    mv a0, s0
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    jalr a3
-; CHECK-NEXT:    ld a0, 0(s0)
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a0, 0(a0)
 ; CHECK-NEXT:    ld t1, 0(a0)
 ; CHECK-NEXT:    li a1, 6
-; CHECK-NEXT:    mv a0, s0
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    jr t1
 ; CHECK-NEXT:  .LBB3_4: # %if.then13
 ; CHECK-NEXT:    ld a1, 0(a0)
 ; CHECK-NEXT:    ld a2, 16(a1)
 ; CHECK-NEXT:    li a1, 1
-; CHECK-NEXT:    mv s2, a0
+; CHECK-NEXT:    sd a0, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    jalr a2
-; CHECK-NEXT:    ld a1, 0(s2)
+; CHECK-NEXT:    ld a1, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a1, 0(a1)
 ; CHECK-NEXT:    ld a2, 16(a1)
-; CHECK-NEXT:    mv s0, a0
-; CHECK-NEXT:    sext.w a1, a0
-; CHECK-NEXT:    mv a0, s2
+; CHECK-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    lw a1, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    jalr a2
-; CHECK-NEXT:    mv s1, a0
+; CHECK-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    li a1, 7
-; CHECK-NEXT:    mv a0, s2
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    call hpm
 ; CHECK-NEXT:    beqz a0, .LBB3_6
 ; CHECK-NEXT:  # %bb.5: # %if.then13
-; CHECK-NEXT:    add a0, s0, s1
+; CHECK-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a1, 0(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    add a0, a0, a1
 ; CHECK-NEXT:    addi a0, a0, 7
 ; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    ret
 ; CHECK-NEXT:  .LBB3_6:
-; CHECK-NEXT:    xor a0, s1, s0
+; CHECK-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a1, 0(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    xor a0, a1, a0
 ; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    ret
 entry:
@@ -660,45 +642,48 @@ declare i64 @hpm(ptr, i32 signext)
 define signext i32 @frame_leaves_hot_path(ptr %b, i64 %now) nounwind {
 ; CHECK-LABEL: frame_leaves_hot_path:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -32
-; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    ld a3, 32(a0)
+; CHECK-NEXT:    ld a2, 32(a0)
 ; CHECK-NEXT:    lw a4, 24(a0)
 ; CHECK-NEXT:    lw a5, 28(a0)
-; CHECK-NEXT:    ld a2, 8(a0)
+; CHECK-NEXT:    ld a3, 8(a0)
 ; CHECK-NEXT:    ld a6, 16(a0)
 ; CHECK-NEXT:    slt a5, a5, a4
-; CHECK-NEXT:    slt a3, a3, a1
+; CHECK-NEXT:    slt a2, a2, a1
 ; CHECK-NEXT:    not a5, a5
-; CHECK-NEXT:    not a3, a3
-; CHECK-NEXT:    and a3, a5, a3
-; CHECK-NEXT:    snez s0, a4
-; CHECK-NEXT:    and s0, a3, s0
-; CHECK-NEXT:    beq a2, a6, .LBB4_3
+; CHECK-NEXT:    not a2, a2
+; CHECK-NEXT:    and a2, a5, a2
+; CHECK-NEXT:    snez a4, a4
+; CHECK-NEXT:    and a2, a2, a4
+; CHECK-NEXT:    beq a3, a6, .LBB4_3
 ; CHECK-NEXT:  # %bb.1: # %if.then
-; CHECK-NEXT:    ld a2, 0(a2)
-; CHECK-NEXT:    ld a2, 0(a2)
-; CHECK-NEXT:    slt a1, a1, a2
-; CHECK-NEXT:    xori s1, a1, 1
-; CHECK-NEXT:    and a1, s1, s0
-; CHECK-NEXT:    beqz a1, .LBB4_4
-; CHECK-NEXT:  # %bb.2: # %if.then15
+; CHECK-NEXT:    ld a3, 0(a3)
+; CHECK-NEXT:    ld a3, 0(a3)
+; CHECK-NEXT:    slt a1, a1, a3
+; CHECK-NEXT:    xori a1, a1, 1
+; CHECK-NEXT:    and a3, a1, a2
+; CHECK-NEXT:    bnez a3, .LBB4_4
+; CHECK-NEXT:  # %bb.2: # %if.end16
+; CHECK-NEXT:    addi a0, a2, -1
+; CHECK-NEXT:    and a0, a0, a1
+; CHECK-NEXT:    ret
+; CHECK-NEXT:  .LBB4_3:
+; CHECK-NEXT:    li a1, 1
+; CHECK-NEXT:    addi a0, a2, -1
+; CHECK-NEXT:    and a0, a0, a1
+; CHECK-NEXT:    ret
+; CHECK-NEXT:  .LBB4_4: # %if.then15
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    ld a0, 0(a0)
 ; CHECK-NEXT:    li a1, 1
-; CHECK-NEXT:    li s1, 1
+; CHECK-NEXT:    sd a2, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    call schedule_event
-; CHECK-NEXT:    j .LBB4_4
-; CHECK-NEXT:  .LBB4_3:
-; CHECK-NEXT:    li s1, 1
-; CHECK-NEXT:  .LBB4_4: # %if.end16
-; CHECK-NEXT:    addi a0, s0, -1
-; CHECK-NEXT:    and a0, a0, s1
-; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 32
+; CHECK-NEXT:    li a1, 1
+; CHECK-NEXT:    ld a2, 0(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    addi a0, a2, -1
+; CHECK-NEXT:    and a0, a0, a1
 ; CHECK-NEXT:    ret
 entry:
   %size = getelementptr inbounds nuw i8, ptr %b, i64 24
@@ -747,81 +732,66 @@ declare void @schedule_event(ptr, i64)
 define void @spill_on_throw_path(ptr %self, ptr %obj) personality ptr @__gxx_personality_v0 {
 ; CHECK-LABEL: spill_on_throw_path:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -48
-; CHECK-NEXT:    .cfi_def_cfa_offset 48
-; CHECK-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    addi sp, sp, -32
+; CHECK-NEXT:    .cfi_def_cfa_offset 32
+; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    .cfi_offset ra, -8
-; CHECK-NEXT:    .cfi_offset s0, -16
-; CHECK-NEXT:    .cfi_offset s1, -24
-; CHECK-NEXT:    .cfi_offset s2, -32
-; CHECK-NEXT:    .cfi_offset s3, -40
 ; CHECK-NEXT:    .cfi_remember_state
 ; CHECK-NEXT:    ld a2, 8(a1)
 ; CHECK-NEXT:    bne a2, a0, .LBB5_2
 ; CHECK-NEXT:  # %bb.1: # %if.end
 ; CHECK-NEXT:    lui a0, %hi(default_owner)
 ; CHECK-NEXT:    ld a0, %lo(default_owner)(a0)
-; CHECK-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    .cfi_restore ra
-; CHECK-NEXT:    .cfi_restore s0
-; CHECK-NEXT:    .cfi_restore s1
-; CHECK-NEXT:    .cfi_restore s2
-; CHECK-NEXT:    .cfi_restore s3
-; CHECK-NEXT:    addi sp, sp, 48
+; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    .cfi_def_cfa_offset 0
 ; CHECK-NEXT:    tail owner_insert
 ; CHECK-NEXT:  .LBB5_2: # %if.then
 ; CHECK-NEXT:    .cfi_restore_state
-; CHECK-NEXT:    mv s1, a0
+; CHECK-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    li a0, 1
-; CHECK-NEXT:    mv s2, a1
+; CHECK-NEXT:    sd a1, 16(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    call __cxa_allocate_exception
-; CHECK-NEXT:    ld a1, 0(s2)
+; CHECK-NEXT:    ld a1, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a1, 0(a1)
 ; CHECK-NEXT:    ld a1, 0(a1)
-; CHECK-NEXT:    mv s0, a0
+; CHECK-NEXT:    sd a0, 8(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:  .Ltmp0: # EH_LABEL
-; CHECK-NEXT:    mv s3, s2
-; CHECK-NEXT:    mv a0, s2
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    jalr a1
+; CHECK-NEXT:    mv a1, a0
 ; CHECK-NEXT:  .Ltmp1: # EH_LABEL
 ; CHECK-NEXT:  # %bb.3: # %invoke.cont
-; CHECK-NEXT:    mv s2, a0
-; CHECK-NEXT:    mv a0, s3
-; CHECK-NEXT:    ld a1, 0(s3)
-; CHECK-NEXT:    ld a1, 8(a1)
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a2, 0(a0)
+; CHECK-NEXT:    ld a2, 8(a2)
 ; CHECK-NEXT:  .Ltmp2: # EH_LABEL
-; CHECK-NEXT:    jalr a1
+; CHECK-NEXT:    sd a1, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    jalr a2
+; CHECK-NEXT:    ld a3, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:  .Ltmp3: # EH_LABEL
 ; CHECK-NEXT:  # %bb.4: # %invoke.cont3
 ; CHECK-NEXT:  .Ltmp4: # EH_LABEL
 ; CHECK-NEXT:    mv a4, a0
 ; CHECK-NEXT:    lui a2, %hi(.L.str)
 ; CHECK-NEXT:    addi a2, a2, %lo(.L.str)
-; CHECK-NEXT:    mv a0, s0
-; CHECK-NEXT:    mv a1, s1
-; CHECK-NEXT:    mv a3, s2
+; CHECK-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a1, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    call error_init
 ; CHECK-NEXT:  .Ltmp5: # EH_LABEL
 ; CHECK-NEXT:  # %bb.5: # %invoke.cont5
 ; CHECK-NEXT:    lui a1, %hi(_ZTI5Error)
 ; CHECK-NEXT:    addi a1, a1, %lo(_ZTI5Error)
-; CHECK-NEXT:    mv a0, s0
+; CHECK-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    li a2, 0
 ; CHECK-NEXT:    call __cxa_throw
 ; CHECK-NEXT:  .LBB5_6: # %lpad
 ; CHECK-NEXT:  .Ltmp6: # EH_LABEL
-; CHECK-NEXT:    mv s1, a0
-; CHECK-NEXT:    mv a0, s0
+; CHECK-NEXT:    sd a0, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    call __cxa_free_exception
-; CHECK-NEXT:    mv a0, s1
+; CHECK-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    call _Unwind_Resume
 entry:
   %owner = getelementptr inbounds nuw i8, ptr %obj, i64 8
@@ -873,131 +843,132 @@ declare void @owner_insert(ptr, ptr)
 define fastcc signext i32 @no_presplit_around_call_in_loop(i32 signext %nKey1, ptr %pKey1, ptr %pPKey2, i1 %tobool.not.i, i8 %retval.0.i, i8 %narrow, i8 %0, i1 %cmp24, i64 %indvars.iv.next, i32 %add306, i32 %1, i32 %add309, i32 %idx1.1, i1 %cmp35, i1 %cmp.i137, i1 %cmp4.i, i32 %conv9) nounwind {
 ; CHECK-LABEL: no_presplit_around_call_in_loop:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -128
-; CHECK-NEXT:    sd ra, 120(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 112(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 104(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 96(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s3, 88(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s4, 80(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s5, 72(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s6, 64(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s7, 56(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s8, 48(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s9, 40(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s10, 32(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s11, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    mv s0, t5
-; CHECK-NEXT:    mv s1, t4
-; CHECK-NEXT:    mv s6, a7
-; CHECK-NEXT:    mv s5, a4
-; CHECK-NEXT:    mv s2, a2
-; CHECK-NEXT:    mv s3, a1
-; CHECK-NEXT:    andi s8, a3, 1
-; CHECK-NEXT:    mv s7, a0
-; CHECK-NEXT:    bnez s8, .LBB6_2
+; CHECK-NEXT:    addi sp, sp, -112
+; CHECK-NEXT:    sd ra, 104(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd t5, 56(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a2, 80(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a1, 72(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    andi a3, a3, 1
+; CHECK-NEXT:    sd t4, 96(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a3, 88(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    bnez a3, .LBB6_2
 ; CHECK-NEXT:  # %bb.1: # %if.end.i
-; CHECK-NEXT:    li a0, 0
 ; CHECK-NEXT:    li a1, 0
-; CHECK-NEXT:    jalr a0
+; CHECK-NEXT:    sd a0, 64(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    li a0, 0
+; CHECK-NEXT:    sd a4, 40(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a7, 48(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    jalr a1
+; CHECK-NEXT:    ld a3, 88(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a0, 64(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a7, 48(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a4, 40(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld t4, 96(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:  .LBB6_2: # %cond.end
-; CHECK-NEXT:    beqz s8, .LBB6_4
+; CHECK-NEXT:    beqz a3, .LBB6_5
 ; CHECK-NEXT:  .LBB6_3: # %if.then26
 ; CHECK-NEXT:    li a0, 0
-; CHECK-NEXT:    j .LBB6_21
-; CHECK-NEXT:  .LBB6_4: # %do.body.preheader
-; CHECK-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    li s4, 0
-; CHECK-NEXT:    lw a0, 160(sp)
-; CHECK-NEXT:    lw a1, 128(sp)
-; CHECK-NEXT:    ld a2, 152(sp)
-; CHECK-NEXT:    ld a3, 136(sp)
-; CHECK-NEXT:    ld s0, 144(sp)
-; CHECK-NEXT:    sext.w a4, s1
-; CHECK-NEXT:    andi s6, s6, 1
-; CHECK-NEXT:    sltu a4, s7, a4
-; CHECK-NEXT:    sltu a0, a0, a1
-; CHECK-NEXT:    not a1, a4
-; CHECK-NEXT:    andi s7, a3, 1
-; CHECK-NEXT:    and s9, a0, a1
-; CHECK-NEXT:    zext.b s5, s5
-; CHECK-NEXT:    li s10, 7
-; CHECK-NEXT:    li s11, 1
-; CHECK-NEXT:    andi s0, s0, 1
-; CHECK-NEXT:    andi s2, a2, 1
-; CHECK-NEXT:    li a0, 0
-; CHECK-NEXT:    j .LBB6_6
-; CHECK-NEXT:  .LBB6_5: # %if.end270
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    mv a0, s1
-; CHECK-NEXT:    beqz s9, .LBB6_3
-; CHECK-NEXT:  .LBB6_6: # %do.body
+; CHECK-NEXT:  .LBB6_4: # %do.end
+; CHECK-NEXT:    ld ra, 104(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 112
+; CHECK-NEXT:    ret
+; CHECK-NEXT:  .LBB6_5: # %do.body.preheader
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    lw a5, 144(sp)
+; CHECK-NEXT:    lw a6, 112(sp)
+; CHECK-NEXT:    ld t2, 136(sp)
+; CHECK-NEXT:    ld t0, 120(sp)
+; CHECK-NEXT:    ld t3, 128(sp)
+; CHECK-NEXT:    sext.w t1, t4
+; CHECK-NEXT:    andi a1, a7, 1
+; CHECK-NEXT:    sltu a0, a0, t1
+; CHECK-NEXT:    sltu a6, a5, a6
+; CHECK-NEXT:    not a0, a0
+; CHECK-NEXT:    andi a5, t0, 1
+; CHECK-NEXT:    and a0, a6, a0
+; CHECK-NEXT:    zext.b a6, a4
+; CHECK-NEXT:    li t0, 7
+; CHECK-NEXT:    li t1, 1
+; CHECK-NEXT:    andi a7, t3, 1
+; CHECK-NEXT:    andi a4, t2, 1
+; CHECK-NEXT:    li t2, 0
+; CHECK-NEXT:    j .LBB6_7
+; CHECK-NEXT:  .LBB6_6: # %if.end270
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    mv t2, t4
+; CHECK-NEXT:    beqz a0, .LBB6_4
+; CHECK-NEXT:  .LBB6_7: # %do.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    bnez s4, .LBB6_13
-; CHECK-NEXT:  # %bb.7: # %if.then32
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bnez s8, .LBB6_20
-; CHECK-NEXT:  # %bb.8: # %if.else38
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bne s5, s10, .LBB6_19
-; CHECK-NEXT:  # %bb.9: # %if.then45
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    sh s11, 0(zero)
-; CHECK-NEXT:    bnez s0, .LBB6_20
-; CHECK-NEXT:  # %bb.10: # %if.end.i138
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bnez s2, .LBB6_20
-; CHECK-NEXT:  # %bb.11: # %if.end7.i
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bnez s6, .LBB6_20
-; CHECK-NEXT:  # %bb.12: # %if.end11.i
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    beqz s8, .LBB6_5
-; CHECK-NEXT:    j .LBB6_20
-; CHECK-NEXT:  .LBB6_13: # %if.else68
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bnez s8, .LBB6_5
-; CHECK-NEXT:  # %bb.14: # %if.then116
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bnez s7, .LBB6_20
-; CHECK-NEXT:  # %bb.15: # %if.else130
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bnez s6, .LBB6_20
-; CHECK-NEXT:  # %bb.16: # %if.else134
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    bnez s8, .LBB6_3
-; CHECK-NEXT:  # %bb.17: # %lor.lhs.false
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    beqz s8, .LBB6_3
-; CHECK-NEXT:  # %bb.18: # %if.else148
-; CHECK-NEXT:    # in Loop: Header=BB6_6 Depth=1
-; CHECK-NEXT:    slli a0, a0, 32
-; CHECK-NEXT:    srli a0, a0, 32
-; CHECK-NEXT:    add a0, s3, a0
+; CHECK-NEXT:    bnez a2, .LBB6_14
+; CHECK-NEXT:  # %bb.8: # %if.then32
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bnez a3, .LBB6_21
+; CHECK-NEXT:  # %bb.9: # %if.else38
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bne a6, t0, .LBB6_20
+; CHECK-NEXT:  # %bb.10: # %if.then45
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    sh t1, 0(zero)
+; CHECK-NEXT:    bnez a7, .LBB6_21
+; CHECK-NEXT:  # %bb.11: # %if.end.i138
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bnez a4, .LBB6_21
+; CHECK-NEXT:  # %bb.12: # %if.end7.i
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bnez a1, .LBB6_21
+; CHECK-NEXT:  # %bb.13: # %if.end11.i
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    beqz a3, .LBB6_6
+; CHECK-NEXT:    j .LBB6_21
+; CHECK-NEXT:  .LBB6_14: # %if.else68
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bnez a3, .LBB6_6
+; CHECK-NEXT:  # %bb.15: # %if.then116
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bnez a5, .LBB6_21
+; CHECK-NEXT:  # %bb.16: # %if.else130
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bnez a1, .LBB6_21
+; CHECK-NEXT:  # %bb.17: # %if.else134
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    bnez a3, .LBB6_3
+; CHECK-NEXT:  # %bb.18: # %lor.lhs.false
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    beqz a3, .LBB6_3
+; CHECK-NEXT:  # %bb.19: # %if.else148
+; CHECK-NEXT:    # in Loop: Header=BB6_7 Depth=1
+; CHECK-NEXT:    slli t2, t2, 32
+; CHECK-NEXT:    srli t1, t2, 32
+; CHECK-NEXT:    ld t0, 72(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    sd a5, 48(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a6, 32(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a0, 24(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a7, 40(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    add a0, t0, t1
+; CHECK-NEXT:    sd a1, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a1, 80(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    sd a2, 64(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd a4, 8(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    ld a3, 64(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    jalr a3
+; CHECK-NEXT:    ld a4, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a7, 40(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    li t1, 1
+; CHECK-NEXT:    li t0, 7
+; CHECK-NEXT:    ld a0, 24(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a6, 32(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a5, 48(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    ld a1, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    mv a2, s4
-; CHECK-NEXT:    jalr s4
-; CHECK-NEXT:    j .LBB6_5
-; CHECK-NEXT:  .LBB6_19: # %if.else38
-; CHECK-NEXT:    bnez s5, .LBB6_3
-; CHECK-NEXT:  .LBB6_20: # %if.then273.loopexit372
-; CHECK-NEXT:    lw a0, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:  .LBB6_21: # %if.then26
-; CHECK-NEXT:    ld ra, 120(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 112(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 104(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 96(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s3, 88(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s4, 80(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s5, 72(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s6, 64(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s7, 56(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s8, 48(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s9, 40(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s10, 32(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s11, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 128
+; CHECK-NEXT:    ld a3, 88(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld a2, 64(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld t4, 96(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    j .LBB6_6
+; CHECK-NEXT:  .LBB6_20: # %if.else38
+; CHECK-NEXT:    bnez a6, .LBB6_3
+; CHECK-NEXT:  .LBB6_21: # %if.then273.loopexit372
+; CHECK-NEXT:    lw a0, 56(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld ra, 104(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 112
 ; CHECK-NEXT:    ret
 entry:
   br i1 %tobool.not.i, label %if.then.i, label %if.end.i
diff --git a/llvm/test/CodeGen/RISCV/rv32xtheadbb.ll b/llvm/test/CodeGen/RISCV/rv32xtheadbb.ll
index 3ba5312f976263..aaeb850dd5fdd6 100644
--- a/llvm/test/CodeGen/RISCV/rv32xtheadbb.ll
+++ b/llvm/test/CodeGen/RISCV/rv32xtheadbb.ll
@@ -215,55 +215,53 @@ define i32 @cttz_i32(i32 %a) nounwind {
 define i64 @cttz_i64(i64 %a) nounwind {
 ; RV32I-LABEL: cttz_i64:
 ; RV32I:       # %bb.0:
+; RV32I-NEXT:    or a2, a0, a1
+; RV32I-NEXT:    beqz a2, .LBB3_3
+; RV32I-NEXT:  # %bb.1: # %cond.false
 ; RV32I-NEXT:    addi sp, sp, -32
 ; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a1
-; RV32I-NEXT:    or a1, a0, a1
-; RV32I-NEXT:    beqz a1, .LBB3_3
-; RV32I-NEXT:  # %bb.1: # %cond.false
-; RV32I-NEXT:    neg a1, a0
-; RV32I-NEXT:    and a1, a0, a1
-; RV32I-NEXT:    lui s2, 30667
-; RV32I-NEXT:    addi s2, s2, 1329
-; RV32I-NEXT:    mv s4, a0
-; RV32I-NEXT:    mv a0, a1
-; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    neg a2, a0
+; RV32I-NEXT:    and a2, a0, a2
+; RV32I-NEXT:    lui a3, 30667
+; RV32I-NEXT:    addi a3, a3, 1329
+; RV32I-NEXT:    sw a3, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, a2
+; RV32I-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    neg a0, s0
-; RV32I-NEXT:    lui s3, %hi(.LCPI3_0)
-; RV32I-NEXT:    addi s3, s3, %lo(.LCPI3_0)
-; RV32I-NEXT:    and a0, s0, a0
-; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    neg a1, a0
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a0, a1
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s4, .LBB3_4
+; RV32I-NEXT:    lw a1, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    bnez a1, .LBB3_4
 ; RV32I-NEXT:  # %bb.2: # %cond.false
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s3, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    j .LBB3_5
 ; RV32I-NEXT:  .LBB3_3:
 ; RV32I-NEXT:    li a0, 64
-; RV32I-NEXT:    j .LBB3_5
-; RV32I-NEXT:  .LBB3_4:
-; RV32I-NEXT:    srli s1, s1, 27
-; RV32I-NEXT:    add s1, s3, s1
-; RV32I-NEXT:    lbu a0, 0(s1)
-; RV32I-NEXT:  .LBB3_5: # %cond.end
 ; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    ret
+; RV32I-NEXT:  .LBB3_4:
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
+; RV32I-NEXT:  .LBB3_5: # %cond.false
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    ret
 ;
 ; RV32XTHEADBB-NOB-LABEL: cttz_i64:
diff --git a/llvm/test/CodeGen/RISCV/rv32zbb.ll b/llvm/test/CodeGen/RISCV/rv32zbb.ll
index 6164065d02d94b..b3babadbdb7ee7 100644
--- a/llvm/test/CodeGen/RISCV/rv32zbb.ll
+++ b/llvm/test/CodeGen/RISCV/rv32zbb.ll
@@ -182,55 +182,53 @@ define i32 @cttz_i32(i32 %a) nounwind {
 define i64 @cttz_i64(i64 %a) nounwind {
 ; RV32I-LABEL: cttz_i64:
 ; RV32I:       # %bb.0:
+; RV32I-NEXT:    or a2, a0, a1
+; RV32I-NEXT:    beqz a2, .LBB3_3
+; RV32I-NEXT:  # %bb.1: # %cond.false
 ; RV32I-NEXT:    addi sp, sp, -32
 ; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a1
-; RV32I-NEXT:    or a1, a0, a1
-; RV32I-NEXT:    beqz a1, .LBB3_3
-; RV32I-NEXT:  # %bb.1: # %cond.false
-; RV32I-NEXT:    neg a1, a0
-; RV32I-NEXT:    and a1, a0, a1
-; RV32I-NEXT:    lui s2, 30667
-; RV32I-NEXT:    addi s2, s2, 1329
-; RV32I-NEXT:    mv s4, a0
-; RV32I-NEXT:    mv a0, a1
-; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    neg a2, a0
+; RV32I-NEXT:    and a2, a0, a2
+; RV32I-NEXT:    lui a3, 30667
+; RV32I-NEXT:    addi a3, a3, 1329
+; RV32I-NEXT:    sw a3, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a0, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, a2
+; RV32I-NEXT:    sw a1, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s1, a0
-; RV32I-NEXT:    neg a0, s0
-; RV32I-NEXT:    lui s3, %hi(.LCPI3_0)
-; RV32I-NEXT:    addi s3, s3, %lo(.LCPI3_0)
-; RV32I-NEXT:    and a0, s0, a0
-; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    neg a1, a0
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a0, a1
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s4, .LBB3_4
+; RV32I-NEXT:    lw a1, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    bnez a1, .LBB3_4
 ; RV32I-NEXT:  # %bb.2: # %cond.false
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s3, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    j .LBB3_5
 ; RV32I-NEXT:  .LBB3_3:
 ; RV32I-NEXT:    li a0, 64
-; RV32I-NEXT:    j .LBB3_5
-; RV32I-NEXT:  .LBB3_4:
-; RV32I-NEXT:    srli s1, s1, 27
-; RV32I-NEXT:    add s1, s3, s1
-; RV32I-NEXT:    lbu a0, 0(s1)
-; RV32I-NEXT:  .LBB3_5: # %cond.end
 ; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    ret
+; RV32I-NEXT:  .LBB3_4:
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
+; RV32I-NEXT:  .LBB3_5: # %cond.false
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: cttz_i64:
diff --git a/llvm/test/CodeGen/RISCV/rv64-double-convert.ll b/llvm/test/CodeGen/RISCV/rv64-double-convert.ll
index bc0a5af35208ed..9d74e97eaf44bf 100644
--- a/llvm/test/CodeGen/RISCV/rv64-double-convert.ll
+++ b/llvm/test/CodeGen/RISCV/rv64-double-convert.ll
@@ -61,14 +61,13 @@ define double @uitofp_i128_to_f64(i128 %a) nounwind {
 define i128 @fptosi_sat_f64_to_i128(double %a) nounwind {
 ; RV64I-LABEL: fptosi_sat_f64_to_i128:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -64
-; RV64I-NEXT:    sd ra, 56(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 48(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s4, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s5, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -48
+; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    li a1, -449
 ; RV64I-NEXT:    slli a1, a1, 53
@@ -78,10 +77,10 @@ define i128 @fptosi_sat_f64_to_i128(double %a) nounwind {
 ; RV64I-NEXT:    call __fixdfti
 ; RV64I-NEXT:    mv s2, a0
 ; RV64I-NEXT:    mv s1, a1
-; RV64I-NEXT:    li s5, -1
 ; RV64I-NEXT:    bgez s3, .LBB4_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    slli s1, s5, 63
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    slli s1, s1, 63
 ; RV64I-NEXT:  .LBB4_2:
 ; RV64I-NEXT:    li a0, 575
 ; RV64I-NEXT:    slli a0, a0, 53
@@ -91,7 +90,8 @@ define i128 @fptosi_sat_f64_to_i128(double %a) nounwind {
 ; RV64I-NEXT:    mv s4, a0
 ; RV64I-NEXT:    blez a0, .LBB4_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    srli s1, s5, 1
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    srli s1, s1, 1
 ; RV64I-NEXT:  .LBB4_4:
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -106,14 +106,13 @@ define i128 @fptosi_sat_f64_to_i128(double %a) nounwind {
 ; RV64I-NEXT:    or a2, a2, a1
 ; RV64I-NEXT:    and a1, a0, s1
 ; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    ld ra, 56(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 48(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s4, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s5, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 64
+; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 48
 ; RV64I-NEXT:    ret
 ;
 ; RV64ID-LABEL: fptosi_sat_f64_to_i128:
@@ -128,26 +127,27 @@ define i128 @fptosi_sat_f64_to_i128(double %a) nounwind {
 ; RV64ID-NEXT:    fmv.d fs0, fa0
 ; RV64ID-NEXT:    fle.d s0, fa5, fa0
 ; RV64ID-NEXT:    call __fixdfti
-; RV64ID-NEXT:    li a3, -1
+; RV64ID-NEXT:    mv a2, s0
+; RV64ID-NEXT:    li a4, -1
 ; RV64ID-NEXT:    bnez s0, .LBB4_2
 ; RV64ID-NEXT:  # %bb.1:
-; RV64ID-NEXT:    slli a1, a3, 63
+; RV64ID-NEXT:    slli a1, a4, 63
 ; RV64ID-NEXT:  .LBB4_2:
-; RV64ID-NEXT:    lui a2, %hi(.LCPI4_0)
-; RV64ID-NEXT:    fld fa5, %lo(.LCPI4_0)(a2)
-; RV64ID-NEXT:    flt.d a2, fa5, fs0
-; RV64ID-NEXT:    beqz a2, .LBB4_4
+; RV64ID-NEXT:    lui a3, %hi(.LCPI4_0)
+; RV64ID-NEXT:    fld fa5, %lo(.LCPI4_0)(a3)
+; RV64ID-NEXT:    flt.d a3, fa5, fs0
+; RV64ID-NEXT:    beqz a3, .LBB4_4
 ; RV64ID-NEXT:  # %bb.3:
-; RV64ID-NEXT:    srli a1, a3, 1
+; RV64ID-NEXT:    srli a1, a4, 1
 ; RV64ID-NEXT:  .LBB4_4:
-; RV64ID-NEXT:    feq.d a3, fs0, fs0
-; RV64ID-NEXT:    neg a4, s0
+; RV64ID-NEXT:    feq.d a4, fs0, fs0
 ; RV64ID-NEXT:    neg a2, a2
-; RV64ID-NEXT:    and a0, a4, a0
 ; RV64ID-NEXT:    neg a3, a3
-; RV64ID-NEXT:    or a0, a2, a0
-; RV64ID-NEXT:    and a1, a3, a1
-; RV64ID-NEXT:    and a0, a3, a0
+; RV64ID-NEXT:    and a0, a2, a0
+; RV64ID-NEXT:    neg a2, a4
+; RV64ID-NEXT:    or a0, a3, a0
+; RV64ID-NEXT:    and a1, a2, a1
+; RV64ID-NEXT:    and a0, a2, a0
 ; RV64ID-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64ID-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64ID-NEXT:    fld fs0, 8(sp) # 8-byte Folded Reload
@@ -166,27 +166,28 @@ define i128 @fptosi_sat_f64_to_i128(double %a) nounwind {
 ; RV64IDINX-NEXT:    fle.d s1, a0, s0
 ; RV64IDINX-NEXT:    mv a0, s0
 ; RV64IDINX-NEXT:    call __fixdfti
-; RV64IDINX-NEXT:    li a2, -1
+; RV64IDINX-NEXT:    mv a2, s1
+; RV64IDINX-NEXT:    li a3, -1
 ; RV64IDINX-NEXT:    bnez s1, .LBB4_2
 ; RV64IDINX-NEXT:  # %bb.1:
-; RV64IDINX-NEXT:    slli a1, a2, 63
+; RV64IDINX-NEXT:    slli a1, a3, 63
 ; RV64IDINX-NEXT:  .LBB4_2:
-; RV64IDINX-NEXT:    li a3, 575
-; RV64IDINX-NEXT:    slli a3, a3, 53
-; RV64IDINX-NEXT:    addi a3, a3, -1
-; RV64IDINX-NEXT:    flt.d a3, a3, s0
-; RV64IDINX-NEXT:    beqz a3, .LBB4_4
+; RV64IDINX-NEXT:    li a4, 575
+; RV64IDINX-NEXT:    slli a4, a4, 53
+; RV64IDINX-NEXT:    addi a4, a4, -1
+; RV64IDINX-NEXT:    flt.d a4, a4, s0
+; RV64IDINX-NEXT:    beqz a4, .LBB4_4
 ; RV64IDINX-NEXT:  # %bb.3:
-; RV64IDINX-NEXT:    srli a1, a2, 1
+; RV64IDINX-NEXT:    srli a1, a3, 1
 ; RV64IDINX-NEXT:  .LBB4_4:
-; RV64IDINX-NEXT:    feq.d a2, s0, s0
-; RV64IDINX-NEXT:    neg a4, s1
-; RV64IDINX-NEXT:    and a0, a4, a0
-; RV64IDINX-NEXT:    neg a3, a3
+; RV64IDINX-NEXT:    feq.d a3, s0, s0
 ; RV64IDINX-NEXT:    neg a2, a2
-; RV64IDINX-NEXT:    or a0, a3, a0
-; RV64IDINX-NEXT:    and a1, a2, a1
 ; RV64IDINX-NEXT:    and a0, a2, a0
+; RV64IDINX-NEXT:    neg a2, a4
+; RV64IDINX-NEXT:    neg a3, a3
+; RV64IDINX-NEXT:    or a0, a2, a0
+; RV64IDINX-NEXT:    and a1, a3, a1
+; RV64IDINX-NEXT:    and a0, a3, a0
 ; RV64IDINX-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64IDINX-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64IDINX-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/rv64-float-convert.ll b/llvm/test/CodeGen/RISCV/rv64-float-convert.ll
index df9a3da0957299..f1523860c607c3 100644
--- a/llvm/test/CodeGen/RISCV/rv64-float-convert.ll
+++ b/llvm/test/CodeGen/RISCV/rv64-float-convert.ll
@@ -61,14 +61,13 @@ define float @uitofp_i128_to_f32(i128 %a) nounwind {
 define i128 @fptosi_sat_f32_to_i128(float %a) nounwind {
 ; RV64I-LABEL: fptosi_sat_f32_to_i128:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -64
-; RV64I-NEXT:    sd ra, 56(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 48(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s4, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s5, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -48
+; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    lui a1, 1044480
 ; RV64I-NEXT:    call __gesf2
@@ -77,10 +76,10 @@ define i128 @fptosi_sat_f32_to_i128(float %a) nounwind {
 ; RV64I-NEXT:    call __fixsfti
 ; RV64I-NEXT:    mv s2, a0
 ; RV64I-NEXT:    mv s1, a1
-; RV64I-NEXT:    li s5, -1
 ; RV64I-NEXT:    bgez s3, .LBB4_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    slli s1, s5, 63
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    slli s1, s1, 63
 ; RV64I-NEXT:  .LBB4_2:
 ; RV64I-NEXT:    lui a1, 520192
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -89,7 +88,8 @@ define i128 @fptosi_sat_f32_to_i128(float %a) nounwind {
 ; RV64I-NEXT:    mv s4, a0
 ; RV64I-NEXT:    blez a0, .LBB4_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    srli s1, s5, 1
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    srli s1, s1, 1
 ; RV64I-NEXT:  .LBB4_4:
 ; RV64I-NEXT:    mv a0, s0
 ; RV64I-NEXT:    mv a1, s0
@@ -104,14 +104,13 @@ define i128 @fptosi_sat_f32_to_i128(float %a) nounwind {
 ; RV64I-NEXT:    or a2, a2, a1
 ; RV64I-NEXT:    and a1, a0, s1
 ; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    ld ra, 56(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 48(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s4, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s5, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 64
+; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 48
 ; RV64I-NEXT:    ret
 ;
 ; RV64IF-LABEL: fptosi_sat_f32_to_i128:
@@ -125,27 +124,28 @@ define i128 @fptosi_sat_f32_to_i128(float %a) nounwind {
 ; RV64IF-NEXT:    fmv.s fs0, fa0
 ; RV64IF-NEXT:    fle.s s0, fa5, fa0
 ; RV64IF-NEXT:    call __fixsfti
-; RV64IF-NEXT:    li a2, -1
+; RV64IF-NEXT:    mv a2, s0
+; RV64IF-NEXT:    li a3, -1
 ; RV64IF-NEXT:    bnez s0, .LBB4_2
 ; RV64IF-NEXT:  # %bb.1:
-; RV64IF-NEXT:    slli a1, a2, 63
+; RV64IF-NEXT:    slli a1, a3, 63
 ; RV64IF-NEXT:  .LBB4_2:
-; RV64IF-NEXT:    lui a3, 520192
-; RV64IF-NEXT:    addi a3, a3, -1
-; RV64IF-NEXT:    fmv.w.x fa5, a3
-; RV64IF-NEXT:    flt.s a3, fa5, fs0
-; RV64IF-NEXT:    beqz a3, .LBB4_4
+; RV64IF-NEXT:    lui a4, 520192
+; RV64IF-NEXT:    addi a4, a4, -1
+; RV64IF-NEXT:    fmv.w.x fa5, a4
+; RV64IF-NEXT:    flt.s a4, fa5, fs0
+; RV64IF-NEXT:    beqz a4, .LBB4_4
 ; RV64IF-NEXT:  # %bb.3:
-; RV64IF-NEXT:    srli a1, a2, 1
+; RV64IF-NEXT:    srli a1, a3, 1
 ; RV64IF-NEXT:  .LBB4_4:
-; RV64IF-NEXT:    feq.s a2, fs0, fs0
-; RV64IF-NEXT:    neg a4, s0
-; RV64IF-NEXT:    and a0, a4, a0
-; RV64IF-NEXT:    neg a3, a3
+; RV64IF-NEXT:    feq.s a3, fs0, fs0
 ; RV64IF-NEXT:    neg a2, a2
-; RV64IF-NEXT:    or a0, a3, a0
-; RV64IF-NEXT:    and a1, a2, a1
 ; RV64IF-NEXT:    and a0, a2, a0
+; RV64IF-NEXT:    neg a2, a4
+; RV64IF-NEXT:    neg a3, a3
+; RV64IF-NEXT:    or a0, a2, a0
+; RV64IF-NEXT:    and a1, a3, a1
+; RV64IF-NEXT:    and a0, a3, a0
 ; RV64IF-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64IF-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64IF-NEXT:    flw fs0, 12(sp) # 4-byte Folded Reload
@@ -163,26 +163,27 @@ define i128 @fptosi_sat_f32_to_i128(float %a) nounwind {
 ; RV64IZFINX-NEXT:    fle.s s1, a0, s0
 ; RV64IZFINX-NEXT:    mv a0, s0
 ; RV64IZFINX-NEXT:    call __fixsfti
-; RV64IZFINX-NEXT:    li a2, -1
+; RV64IZFINX-NEXT:    mv a2, s1
+; RV64IZFINX-NEXT:    li a3, -1
 ; RV64IZFINX-NEXT:    bnez s1, .LBB4_2
 ; RV64IZFINX-NEXT:  # %bb.1:
-; RV64IZFINX-NEXT:    slli a1, a2, 63
+; RV64IZFINX-NEXT:    slli a1, a3, 63
 ; RV64IZFINX-NEXT:  .LBB4_2:
-; RV64IZFINX-NEXT:    lui a3, 520192
-; RV64IZFINX-NEXT:    addi a3, a3, -1
-; RV64IZFINX-NEXT:    flt.s a3, a3, s0
-; RV64IZFINX-NEXT:    beqz a3, .LBB4_4
+; RV64IZFINX-NEXT:    lui a4, 520192
+; RV64IZFINX-NEXT:    addi a4, a4, -1
+; RV64IZFINX-NEXT:    flt.s a4, a4, s0
+; RV64IZFINX-NEXT:    beqz a4, .LBB4_4
 ; RV64IZFINX-NEXT:  # %bb.3:
-; RV64IZFINX-NEXT:    srli a1, a2, 1
+; RV64IZFINX-NEXT:    srli a1, a3, 1
 ; RV64IZFINX-NEXT:  .LBB4_4:
-; RV64IZFINX-NEXT:    feq.s a2, s0, s0
-; RV64IZFINX-NEXT:    neg a4, s1
-; RV64IZFINX-NEXT:    and a0, a4, a0
-; RV64IZFINX-NEXT:    neg a3, a3
+; RV64IZFINX-NEXT:    feq.s a3, s0, s0
 ; RV64IZFINX-NEXT:    neg a2, a2
-; RV64IZFINX-NEXT:    or a0, a3, a0
-; RV64IZFINX-NEXT:    and a1, a2, a1
 ; RV64IZFINX-NEXT:    and a0, a2, a0
+; RV64IZFINX-NEXT:    neg a2, a4
+; RV64IZFINX-NEXT:    neg a3, a3
+; RV64IZFINX-NEXT:    or a0, a2, a0
+; RV64IZFINX-NEXT:    and a1, a3, a1
+; RV64IZFINX-NEXT:    and a0, a3, a0
 ; RV64IZFINX-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64IZFINX-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64IZFINX-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/rv64-half-convert.ll b/llvm/test/CodeGen/RISCV/rv64-half-convert.ll
index 0f724e22651dd6..eefbdf486aa159 100644
--- a/llvm/test/CodeGen/RISCV/rv64-half-convert.ll
+++ b/llvm/test/CodeGen/RISCV/rv64-half-convert.ll
@@ -125,14 +125,13 @@ define i128 @fptoui_f16_to_i128(half %a) nounwind {
 define i128 @fptosi_sat_f16_to_i128(half %a) nounwind {
 ; RV64I-LABEL: fptosi_sat_f16_to_i128:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -64
-; RV64I-NEXT:    sd ra, 56(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 48(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s1, 40(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s2, 32(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s3, 24(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s4, 16(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s5, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    addi sp, sp, -48
+; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    call __extendhfsf2
 ; RV64I-NEXT:    mv s2, a0
 ; RV64I-NEXT:    lui a1, 1044480
@@ -142,10 +141,10 @@ define i128 @fptosi_sat_f16_to_i128(half %a) nounwind {
 ; RV64I-NEXT:    call __fixsfti
 ; RV64I-NEXT:    mv s0, a0
 ; RV64I-NEXT:    mv s1, a1
-; RV64I-NEXT:    li s5, -1
 ; RV64I-NEXT:    bgez s3, .LBB4_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    slli s1, s5, 63
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    slli s1, s1, 63
 ; RV64I-NEXT:  .LBB4_2:
 ; RV64I-NEXT:    lui a1, 520192
 ; RV64I-NEXT:    addi a1, a1, -1
@@ -154,7 +153,8 @@ define i128 @fptosi_sat_f16_to_i128(half %a) nounwind {
 ; RV64I-NEXT:    mv s4, a0
 ; RV64I-NEXT:    blez a0, .LBB4_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    srli s1, s5, 1
+; RV64I-NEXT:    li s1, -1
+; RV64I-NEXT:    srli s1, s1, 1
 ; RV64I-NEXT:  .LBB4_4:
 ; RV64I-NEXT:    mv a0, s2
 ; RV64I-NEXT:    mv a1, s2
@@ -169,14 +169,13 @@ define i128 @fptosi_sat_f16_to_i128(half %a) nounwind {
 ; RV64I-NEXT:    or a2, a2, a1
 ; RV64I-NEXT:    and a1, a0, s1
 ; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    ld ra, 56(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 48(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s1, 40(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s2, 32(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s3, 24(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s4, 16(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s5, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 64
+; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 48
 ; RV64I-NEXT:    ret
 ;
 ; RV64IZFH-LABEL: fptosi_sat_f16_to_i128:
@@ -191,27 +190,28 @@ define i128 @fptosi_sat_f16_to_i128(half %a) nounwind {
 ; RV64IZFH-NEXT:    fle.s s0, fa5, fs0
 ; RV64IZFH-NEXT:    fmv.s fa0, fs0
 ; RV64IZFH-NEXT:    call __fixsfti
-; RV64IZFH-NEXT:    li a2, -1
+; RV64IZFH-NEXT:    mv a2, s0
+; RV64IZFH-NEXT:    li a3, -1
 ; RV64IZFH-NEXT:    bnez s0, .LBB4_2
 ; RV64IZFH-NEXT:  # %bb.1:
-; RV64IZFH-NEXT:    slli a1, a2, 63
+; RV64IZFH-NEXT:    slli a1, a3, 63
 ; RV64IZFH-NEXT:  .LBB4_2:
-; RV64IZFH-NEXT:    lui a3, 520192
-; RV64IZFH-NEXT:    addi a3, a3, -1
-; RV64IZFH-NEXT:    fmv.w.x fa5, a3
-; RV64IZFH-NEXT:    flt.s a3, fa5, fs0
-; RV64IZFH-NEXT:    beqz a3, .LBB4_4
+; RV64IZFH-NEXT:    lui a4, 520192
+; RV64IZFH-NEXT:    addi a4, a4, -1
+; RV64IZFH-NEXT:    fmv.w.x fa5, a4
+; RV64IZFH-NEXT:    flt.s a4, fa5, fs0
+; RV64IZFH-NEXT:    beqz a4, .LBB4_4
 ; RV64IZFH-NEXT:  # %bb.3:
-; RV64IZFH-NEXT:    srli a1, a2, 1
+; RV64IZFH-NEXT:    srli a1, a3, 1
 ; RV64IZFH-NEXT:  .LBB4_4:
-; RV64IZFH-NEXT:    feq.s a2, fs0, fs0
-; RV64IZFH-NEXT:    neg a4, s0
-; RV64IZFH-NEXT:    and a0, a4, a0
-; RV64IZFH-NEXT:    neg a3, a3
+; RV64IZFH-NEXT:    feq.s a3, fs0, fs0
 ; RV64IZFH-NEXT:    neg a2, a2
-; RV64IZFH-NEXT:    or a0, a3, a0
-; RV64IZFH-NEXT:    and a1, a2, a1
 ; RV64IZFH-NEXT:    and a0, a2, a0
+; RV64IZFH-NEXT:    neg a2, a4
+; RV64IZFH-NEXT:    neg a3, a3
+; RV64IZFH-NEXT:    or a0, a2, a0
+; RV64IZFH-NEXT:    and a1, a3, a1
+; RV64IZFH-NEXT:    and a0, a3, a0
 ; RV64IZFH-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64IZFH-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64IZFH-NEXT:    flw fs0, 12(sp) # 4-byte Folded Reload
@@ -229,26 +229,27 @@ define i128 @fptosi_sat_f16_to_i128(half %a) nounwind {
 ; RV64IZHINX-NEXT:    fle.s s1, a0, s0
 ; RV64IZHINX-NEXT:    mv a0, s0
 ; RV64IZHINX-NEXT:    call __fixsfti
-; RV64IZHINX-NEXT:    li a2, -1
+; RV64IZHINX-NEXT:    mv a2, s1
+; RV64IZHINX-NEXT:    li a3, -1
 ; RV64IZHINX-NEXT:    bnez s1, .LBB4_2
 ; RV64IZHINX-NEXT:  # %bb.1:
-; RV64IZHINX-NEXT:    slli a1, a2, 63
+; RV64IZHINX-NEXT:    slli a1, a3, 63
 ; RV64IZHINX-NEXT:  .LBB4_2:
-; RV64IZHINX-NEXT:    lui a3, 520192
-; RV64IZHINX-NEXT:    addi a3, a3, -1
-; RV64IZHINX-NEXT:    flt.s a3, a3, s0
-; RV64IZHINX-NEXT:    beqz a3, .LBB4_4
+; RV64IZHINX-NEXT:    lui a4, 520192
+; RV64IZHINX-NEXT:    addi a4, a4, -1
+; RV64IZHINX-NEXT:    flt.s a4, a4, s0
+; RV64IZHINX-NEXT:    beqz a4, .LBB4_4
 ; RV64IZHINX-NEXT:  # %bb.3:
-; RV64IZHINX-NEXT:    srli a1, a2, 1
+; RV64IZHINX-NEXT:    srli a1, a3, 1
 ; RV64IZHINX-NEXT:  .LBB4_4:
-; RV64IZHINX-NEXT:    feq.s a2, s0, s0
-; RV64IZHINX-NEXT:    neg a4, s1
-; RV64IZHINX-NEXT:    and a0, a4, a0
-; RV64IZHINX-NEXT:    neg a3, a3
+; RV64IZHINX-NEXT:    feq.s a3, s0, s0
 ; RV64IZHINX-NEXT:    neg a2, a2
-; RV64IZHINX-NEXT:    or a0, a3, a0
-; RV64IZHINX-NEXT:    and a1, a2, a1
 ; RV64IZHINX-NEXT:    and a0, a2, a0
+; RV64IZHINX-NEXT:    neg a2, a4
+; RV64IZHINX-NEXT:    neg a3, a3
+; RV64IZHINX-NEXT:    or a0, a2, a0
+; RV64IZHINX-NEXT:    and a1, a3, a1
+; RV64IZHINX-NEXT:    and a0, a3, a0
 ; RV64IZHINX-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64IZHINX-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; RV64IZHINX-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/rv64-statepoint-call-lowering.ll b/llvm/test/CodeGen/RISCV/rv64-statepoint-call-lowering.ll
index 53240c5cdb24d3..919d983c0f2a60 100644
--- a/llvm/test/CodeGen/RISCV/rv64-statepoint-call-lowering.ll
+++ b/llvm/test/CodeGen/RISCV/rv64-statepoint-call-lowering.ll
@@ -203,10 +203,10 @@ define i1 @test_cross_bb(ptr addrspace(1) %a, i1 %external_cond) gc "statepoint-
 ; CHECK-NEXT:    beqz s0, .LBB8_2
 ; CHECK-NEXT:  # %bb.1: # %left
 ; CHECK-NEXT:    ld a1, 8(sp)
-; CHECK-NEXT:    mv s0, a0
+; CHECK-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    mv a0, a1
 ; CHECK-NEXT:    call consume
-; CHECK-NEXT:    mv a0, s0
+; CHECK-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    j .LBB8_3
 ; CHECK-NEXT:  .LBB8_2: # %right
 ; CHECK-NEXT:    li a0, 1
diff --git a/llvm/test/CodeGen/RISCV/rvv/clmulh-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/clmulh-sdnode.ll
index 795be4ba73e6e6..2fad30c93ff49e 100644
--- a/llvm/test/CodeGen/RISCV/rvv/clmulh-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/clmulh-sdnode.ll
@@ -22341,9 +22341,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64V:       # %bb.0:
 ; RV64V-NEXT:    addi sp, sp, -16
 ; RV64V-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
-; RV64V-NEXT:    li a1, 56
+; RV64V-NEXT:    li s0, 56
 ; RV64V-NEXT:    vsetvli a2, zero, e64, m1, ta, ma
-; RV64V-NEXT:    vsrl.vx v9, v8, a1
+; RV64V-NEXT:    vsrl.vx v9, v8, s0
 ; RV64V-NEXT:    li a2, 40
 ; RV64V-NEXT:    vsrl.vx v10, v8, a2
 ; RV64V-NEXT:    lui a3, 16
@@ -22363,7 +22363,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vsll.vi v11, v11, 24
 ; RV64V-NEXT:    vsll.vi v12, v12, 8
 ; RV64V-NEXT:    vand.vx v13, v8, a3
-; RV64V-NEXT:    vsll.vx v8, v8, a1
+; RV64V-NEXT:    vsll.vx v8, v8, s0
 ; RV64V-NEXT:    vsll.vx v13, v13, a2
 ; RV64V-NEXT:    vor.vv v11, v11, v12
 ; RV64V-NEXT:    vor.vv v8, v8, v13
@@ -22450,15 +22450,15 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vmul.vx v12, v10, t4
 ; RV64V-NEXT:    addi t5, t5, 1092
 ; RV64V-NEXT:    slli t6, t5, 32
-; RV64V-NEXT:    lui s0, %hi(.LCPI37_0)
+; RV64V-NEXT:    lui a1, %hi(.LCPI37_0)
 ; RV64V-NEXT:    add t5, t5, t6
-; RV64V-NEXT:    ld t6, %lo(.LCPI37_0)(s0)
+; RV64V-NEXT:    ld a1, %lo(.LCPI37_0)(a1)
 ; RV64V-NEXT:    vand.vx v13, v8, t5
-; RV64V-NEXT:    and s0, t2, t6
-; RV64V-NEXT:    vmul.vx v14, v13, s0
+; RV64V-NEXT:    and t6, t2, a1
+; RV64V-NEXT:    vmul.vx v14, v13, t6
 ; RV64V-NEXT:    vxor.vv v11, v12, v11
-; RV64V-NEXT:    vand.vx v8, v8, t6
-; RV64V-NEXT:    vmul.vx v12, v9, s0
+; RV64V-NEXT:    vand.vx v8, v8, a1
+; RV64V-NEXT:    vmul.vx v12, v9, t6
 ; RV64V-NEXT:    vmul.vx v15, v10, t1
 ; RV64V-NEXT:    vxor.vv v11, v11, v14
 ; RV64V-NEXT:    and t2, t2, t5
@@ -22477,9 +22477,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vor.vv v11, v12, v11
 ; RV64V-NEXT:    vxor.vv v12, v15, v14
 ; RV64V-NEXT:    vmul.vx v9, v9, t2
-; RV64V-NEXT:    vmul.vx v10, v10, s0
+; RV64V-NEXT:    vmul.vx v10, v10, t6
 ; RV64V-NEXT:    vxor.vv v12, v12, v16
-; RV64V-NEXT:    vmul.vx v14, v8, s0
+; RV64V-NEXT:    vmul.vx v14, v8, t6
 ; RV64V-NEXT:    vmul.vx v13, v13, t4
 ; RV64V-NEXT:    vxor.vv v9, v10, v9
 ; RV64V-NEXT:    vmul.vx v8, v8, t1
@@ -22488,10 +22488,10 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vand.vx v10, v10, t5
 ; RV64V-NEXT:    vxor.vv v8, v9, v8
 ; RV64V-NEXT:    vor.vv v9, v11, v10
-; RV64V-NEXT:    vand.vx v8, v8, t6
+; RV64V-NEXT:    vand.vx v8, v8, a1
 ; RV64V-NEXT:    vor.vv v8, v9, v8
 ; RV64V-NEXT:    vsrl.vx v9, v8, a2
-; RV64V-NEXT:    vsrl.vx v10, v8, a1
+; RV64V-NEXT:    vsrl.vx v10, v8, s0
 ; RV64V-NEXT:    vand.vx v9, v9, a3
 ; RV64V-NEXT:    vsrl.vi v11, v8, 24
 ; RV64V-NEXT:    vsrl.vi v12, v8, 8
@@ -22504,7 +22504,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vsll.vi v11, v11, 8
 ; RV64V-NEXT:    vsll.vi v12, v12, 24
 ; RV64V-NEXT:    vand.vx v13, v8, a3
-; RV64V-NEXT:    vsll.vx v8, v8, a1
+; RV64V-NEXT:    vsll.vx v8, v8, s0
 ; RV64V-NEXT:    vsll.vx v13, v13, a2
 ; RV64V-NEXT:    vor.vv v11, v12, v11
 ; RV64V-NEXT:    vor.vv v8, v8, v13
@@ -22761,9 +22761,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64ZVBC32:       # %bb.0:
 ; RV64ZVBC32-NEXT:    addi sp, sp, -16
 ; RV64ZVBC32-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
-; RV64ZVBC32-NEXT:    li a1, 56
+; RV64ZVBC32-NEXT:    li s0, 56
 ; RV64ZVBC32-NEXT:    vsetvli a2, zero, e64, m1, ta, ma
-; RV64ZVBC32-NEXT:    vsrl.vx v9, v8, a1
+; RV64ZVBC32-NEXT:    vsrl.vx v9, v8, s0
 ; RV64ZVBC32-NEXT:    li a2, 40
 ; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, a2
 ; RV64ZVBC32-NEXT:    lui a3, 16
@@ -22783,7 +22783,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vsll.vi v11, v11, 24
 ; RV64ZVBC32-NEXT:    vsll.vi v12, v12, 8
 ; RV64ZVBC32-NEXT:    vand.vx v13, v8, a3
-; RV64ZVBC32-NEXT:    vsll.vx v8, v8, a1
+; RV64ZVBC32-NEXT:    vsll.vx v8, v8, s0
 ; RV64ZVBC32-NEXT:    vsll.vx v13, v13, a2
 ; RV64ZVBC32-NEXT:    vor.vv v11, v11, v12
 ; RV64ZVBC32-NEXT:    vor.vv v8, v8, v13
@@ -22870,15 +22870,15 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vmul.vx v12, v10, t4
 ; RV64ZVBC32-NEXT:    addi t5, t5, 1092
 ; RV64ZVBC32-NEXT:    slli t6, t5, 32
-; RV64ZVBC32-NEXT:    lui s0, %hi(.LCPI37_0)
+; RV64ZVBC32-NEXT:    lui a1, %hi(.LCPI37_0)
 ; RV64ZVBC32-NEXT:    add t5, t5, t6
-; RV64ZVBC32-NEXT:    ld t6, %lo(.LCPI37_0)(s0)
+; RV64ZVBC32-NEXT:    ld a1, %lo(.LCPI37_0)(a1)
 ; RV64ZVBC32-NEXT:    vand.vx v13, v8, t5
-; RV64ZVBC32-NEXT:    and s0, t2, t6
-; RV64ZVBC32-NEXT:    vmul.vx v14, v13, s0
+; RV64ZVBC32-NEXT:    and t6, t2, a1
+; RV64ZVBC32-NEXT:    vmul.vx v14, v13, t6
 ; RV64ZVBC32-NEXT:    vxor.vv v11, v12, v11
-; RV64ZVBC32-NEXT:    vand.vx v8, v8, t6
-; RV64ZVBC32-NEXT:    vmul.vx v12, v9, s0
+; RV64ZVBC32-NEXT:    vand.vx v8, v8, a1
+; RV64ZVBC32-NEXT:    vmul.vx v12, v9, t6
 ; RV64ZVBC32-NEXT:    vmul.vx v15, v10, t1
 ; RV64ZVBC32-NEXT:    vxor.vv v11, v11, v14
 ; RV64ZVBC32-NEXT:    and t2, t2, t5
@@ -22897,9 +22897,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vor.vv v11, v12, v11
 ; RV64ZVBC32-NEXT:    vxor.vv v12, v15, v14
 ; RV64ZVBC32-NEXT:    vmul.vx v9, v9, t2
-; RV64ZVBC32-NEXT:    vmul.vx v10, v10, s0
+; RV64ZVBC32-NEXT:    vmul.vx v10, v10, t6
 ; RV64ZVBC32-NEXT:    vxor.vv v12, v12, v16
-; RV64ZVBC32-NEXT:    vmul.vx v14, v8, s0
+; RV64ZVBC32-NEXT:    vmul.vx v14, v8, t6
 ; RV64ZVBC32-NEXT:    vmul.vx v13, v13, t4
 ; RV64ZVBC32-NEXT:    vxor.vv v9, v10, v9
 ; RV64ZVBC32-NEXT:    vmul.vx v8, v8, t1
@@ -22908,10 +22908,10 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vand.vx v10, v10, t5
 ; RV64ZVBC32-NEXT:    vxor.vv v8, v9, v8
 ; RV64ZVBC32-NEXT:    vor.vv v9, v11, v10
-; RV64ZVBC32-NEXT:    vand.vx v8, v8, t6
+; RV64ZVBC32-NEXT:    vand.vx v8, v8, a1
 ; RV64ZVBC32-NEXT:    vor.vv v8, v9, v8
 ; RV64ZVBC32-NEXT:    vsrl.vx v9, v8, a2
-; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, a1
+; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, s0
 ; RV64ZVBC32-NEXT:    vand.vx v9, v9, a3
 ; RV64ZVBC32-NEXT:    vsrl.vi v11, v8, 24
 ; RV64ZVBC32-NEXT:    vsrl.vi v12, v8, 8
@@ -22924,7 +22924,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx(<vscale x 1 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vsll.vi v11, v11, 8
 ; RV64ZVBC32-NEXT:    vsll.vi v12, v12, 24
 ; RV64ZVBC32-NEXT:    vand.vx v13, v8, a3
-; RV64ZVBC32-NEXT:    vsll.vx v8, v8, a1
+; RV64ZVBC32-NEXT:    vsll.vx v8, v8, s0
 ; RV64ZVBC32-NEXT:    vsll.vx v13, v13, a2
 ; RV64ZVBC32-NEXT:    vor.vv v11, v12, v11
 ; RV64ZVBC32-NEXT:    vor.vv v8, v8, v13
@@ -24089,9 +24089,9 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64V:       # %bb.0:
 ; RV64V-NEXT:    addi sp, sp, -16
 ; RV64V-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
-; RV64V-NEXT:    li a1, 56
+; RV64V-NEXT:    li s0, 56
 ; RV64V-NEXT:    vsetvli a2, zero, e64, m2, ta, ma
-; RV64V-NEXT:    vsrl.vx v10, v8, a1
+; RV64V-NEXT:    vsrl.vx v10, v8, s0
 ; RV64V-NEXT:    li a2, 40
 ; RV64V-NEXT:    vsrl.vx v12, v8, a2
 ; RV64V-NEXT:    lui a3, 16
@@ -24112,7 +24112,7 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vand.vx v14, v8, a5
 ; RV64V-NEXT:    vsll.vi v14, v14, 8
 ; RV64V-NEXT:    vor.vv v12, v12, v14
-; RV64V-NEXT:    vsll.vx v14, v8, a1
+; RV64V-NEXT:    vsll.vx v14, v8, s0
 ; RV64V-NEXT:    vand.vx v8, v8, a3
 ; RV64V-NEXT:    vsll.vx v8, v8, a2
 ; RV64V-NEXT:    vor.vv v8, v14, v8
@@ -24199,18 +24199,18 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    lui t5, 279620
 ; RV64V-NEXT:    addi t5, t5, 1092
 ; RV64V-NEXT:    slli t6, t5, 32
-; RV64V-NEXT:    lui s0, %hi(.LCPI39_0)
+; RV64V-NEXT:    lui a1, %hi(.LCPI39_0)
 ; RV64V-NEXT:    add t5, t5, t6
-; RV64V-NEXT:    ld t6, %lo(.LCPI39_0)(s0)
+; RV64V-NEXT:    ld a1, %lo(.LCPI39_0)(a1)
 ; RV64V-NEXT:    vand.vx v14, v8, t5
-; RV64V-NEXT:    and s0, a0, t6
-; RV64V-NEXT:    vmul.vx v18, v14, s0
-; RV64V-NEXT:    vand.vx v8, v8, t6
+; RV64V-NEXT:    and t6, a0, a1
+; RV64V-NEXT:    vmul.vx v18, v14, t6
+; RV64V-NEXT:    vand.vx v8, v8, a1
 ; RV64V-NEXT:    and a0, a0, t5
 ; RV64V-NEXT:    vmul.vx v20, v8, a0
 ; RV64V-NEXT:    vxor.vv v16, v16, v18
 ; RV64V-NEXT:    vxor.vv v16, v16, v20
-; RV64V-NEXT:    vmul.vx v18, v10, s0
+; RV64V-NEXT:    vmul.vx v18, v10, t6
 ; RV64V-NEXT:    vmul.vx v20, v12, t3
 ; RV64V-NEXT:    vand.vx v16, v16, t1
 ; RV64V-NEXT:    vxor.vv v18, v20, v18
@@ -24224,21 +24224,21 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vor.vv v16, v18, v16
 ; RV64V-NEXT:    vxor.vv v18, v22, v20
 ; RV64V-NEXT:    vmul.vx v20, v14, t3
-; RV64V-NEXT:    vmul.vx v22, v8, s0
+; RV64V-NEXT:    vmul.vx v22, v8, t6
 ; RV64V-NEXT:    vxor.vv v18, v18, v20
 ; RV64V-NEXT:    vxor.vv v18, v18, v22
 ; RV64V-NEXT:    vand.vx v18, v18, t5
 ; RV64V-NEXT:    vmul.vx v10, v10, a0
-; RV64V-NEXT:    vmul.vx v12, v12, s0
+; RV64V-NEXT:    vmul.vx v12, v12, t6
 ; RV64V-NEXT:    vor.vv v16, v16, v18
 ; RV64V-NEXT:    vxor.vv v10, v12, v10
 ; RV64V-NEXT:    vmul.vx v12, v14, t4
 ; RV64V-NEXT:    vmul.vx v8, v8, t3
 ; RV64V-NEXT:    vxor.vv v10, v10, v12
 ; RV64V-NEXT:    vxor.vv v8, v10, v8
-; RV64V-NEXT:    vand.vx v8, v8, t6
+; RV64V-NEXT:    vand.vx v8, v8, a1
 ; RV64V-NEXT:    vor.vv v8, v16, v8
-; RV64V-NEXT:    vsrl.vx v10, v8, a1
+; RV64V-NEXT:    vsrl.vx v10, v8, s0
 ; RV64V-NEXT:    vsrl.vx v12, v8, a2
 ; RV64V-NEXT:    vand.vx v12, v12, a3
 ; RV64V-NEXT:    vor.vv v10, v12, v10
@@ -24253,7 +24253,7 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64V-NEXT:    vand.vx v14, v8, a4
 ; RV64V-NEXT:    vsll.vi v14, v14, 24
 ; RV64V-NEXT:    vor.vv v12, v14, v12
-; RV64V-NEXT:    vsll.vx v14, v8, a1
+; RV64V-NEXT:    vsll.vx v14, v8, s0
 ; RV64V-NEXT:    vand.vx v8, v8, a3
 ; RV64V-NEXT:    vsll.vx v8, v8, a2
 ; RV64V-NEXT:    vor.vv v8, v14, v8
@@ -24545,9 +24545,9 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64ZVBC32:       # %bb.0:
 ; RV64ZVBC32-NEXT:    addi sp, sp, -16
 ; RV64ZVBC32-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
-; RV64ZVBC32-NEXT:    li a1, 56
+; RV64ZVBC32-NEXT:    li s0, 56
 ; RV64ZVBC32-NEXT:    vsetvli a2, zero, e64, m2, ta, ma
-; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, a1
+; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, s0
 ; RV64ZVBC32-NEXT:    li a2, 40
 ; RV64ZVBC32-NEXT:    vsrl.vx v12, v8, a2
 ; RV64ZVBC32-NEXT:    lui a3, 16
@@ -24568,7 +24568,7 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vand.vx v14, v8, a5
 ; RV64ZVBC32-NEXT:    vsll.vi v14, v14, 8
 ; RV64ZVBC32-NEXT:    vor.vv v12, v12, v14
-; RV64ZVBC32-NEXT:    vsll.vx v14, v8, a1
+; RV64ZVBC32-NEXT:    vsll.vx v14, v8, s0
 ; RV64ZVBC32-NEXT:    vand.vx v8, v8, a3
 ; RV64ZVBC32-NEXT:    vsll.vx v8, v8, a2
 ; RV64ZVBC32-NEXT:    vor.vv v8, v14, v8
@@ -24655,18 +24655,18 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    lui t5, 279620
 ; RV64ZVBC32-NEXT:    addi t5, t5, 1092
 ; RV64ZVBC32-NEXT:    slli t6, t5, 32
-; RV64ZVBC32-NEXT:    lui s0, %hi(.LCPI39_0)
+; RV64ZVBC32-NEXT:    lui a1, %hi(.LCPI39_0)
 ; RV64ZVBC32-NEXT:    add t5, t5, t6
-; RV64ZVBC32-NEXT:    ld t6, %lo(.LCPI39_0)(s0)
+; RV64ZVBC32-NEXT:    ld a1, %lo(.LCPI39_0)(a1)
 ; RV64ZVBC32-NEXT:    vand.vx v14, v8, t5
-; RV64ZVBC32-NEXT:    and s0, a0, t6
-; RV64ZVBC32-NEXT:    vmul.vx v18, v14, s0
-; RV64ZVBC32-NEXT:    vand.vx v8, v8, t6
+; RV64ZVBC32-NEXT:    and t6, a0, a1
+; RV64ZVBC32-NEXT:    vmul.vx v18, v14, t6
+; RV64ZVBC32-NEXT:    vand.vx v8, v8, a1
 ; RV64ZVBC32-NEXT:    and a0, a0, t5
 ; RV64ZVBC32-NEXT:    vmul.vx v20, v8, a0
 ; RV64ZVBC32-NEXT:    vxor.vv v16, v16, v18
 ; RV64ZVBC32-NEXT:    vxor.vv v16, v16, v20
-; RV64ZVBC32-NEXT:    vmul.vx v18, v10, s0
+; RV64ZVBC32-NEXT:    vmul.vx v18, v10, t6
 ; RV64ZVBC32-NEXT:    vmul.vx v20, v12, t3
 ; RV64ZVBC32-NEXT:    vand.vx v16, v16, t1
 ; RV64ZVBC32-NEXT:    vxor.vv v18, v20, v18
@@ -24680,21 +24680,21 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vor.vv v16, v18, v16
 ; RV64ZVBC32-NEXT:    vxor.vv v18, v22, v20
 ; RV64ZVBC32-NEXT:    vmul.vx v20, v14, t3
-; RV64ZVBC32-NEXT:    vmul.vx v22, v8, s0
+; RV64ZVBC32-NEXT:    vmul.vx v22, v8, t6
 ; RV64ZVBC32-NEXT:    vxor.vv v18, v18, v20
 ; RV64ZVBC32-NEXT:    vxor.vv v18, v18, v22
 ; RV64ZVBC32-NEXT:    vand.vx v18, v18, t5
 ; RV64ZVBC32-NEXT:    vmul.vx v10, v10, a0
-; RV64ZVBC32-NEXT:    vmul.vx v12, v12, s0
+; RV64ZVBC32-NEXT:    vmul.vx v12, v12, t6
 ; RV64ZVBC32-NEXT:    vor.vv v16, v16, v18
 ; RV64ZVBC32-NEXT:    vxor.vv v10, v12, v10
 ; RV64ZVBC32-NEXT:    vmul.vx v12, v14, t4
 ; RV64ZVBC32-NEXT:    vmul.vx v8, v8, t3
 ; RV64ZVBC32-NEXT:    vxor.vv v10, v10, v12
 ; RV64ZVBC32-NEXT:    vxor.vv v8, v10, v8
-; RV64ZVBC32-NEXT:    vand.vx v8, v8, t6
+; RV64ZVBC32-NEXT:    vand.vx v8, v8, a1
 ; RV64ZVBC32-NEXT:    vor.vv v8, v16, v8
-; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, a1
+; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, s0
 ; RV64ZVBC32-NEXT:    vsrl.vx v12, v8, a2
 ; RV64ZVBC32-NEXT:    vand.vx v12, v12, a3
 ; RV64ZVBC32-NEXT:    vor.vv v10, v12, v10
@@ -24709,7 +24709,7 @@ define <vscale x 2 x i64> @clmulh_nxv2i64_vx(<vscale x 2 x i64> %va, i64 %b) nou
 ; RV64ZVBC32-NEXT:    vand.vx v14, v8, a4
 ; RV64ZVBC32-NEXT:    vsll.vi v14, v14, 24
 ; RV64ZVBC32-NEXT:    vor.vv v12, v14, v12
-; RV64ZVBC32-NEXT:    vsll.vx v14, v8, a1
+; RV64ZVBC32-NEXT:    vsll.vx v14, v8, s0
 ; RV64ZVBC32-NEXT:    vand.vx v8, v8, a3
 ; RV64ZVBC32-NEXT:    vsll.vx v8, v8, a2
 ; RV64ZVBC32-NEXT:    vor.vv v8, v14, v8
@@ -35698,9 +35698,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64V-NEXT:    .cfi_def_cfa_offset 16
 ; RV64V-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
 ; RV64V-NEXT:    .cfi_offset s0, -8
-; RV64V-NEXT:    li a1, 56
+; RV64V-NEXT:    li s0, 56
 ; RV64V-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
-; RV64V-NEXT:    vsrl.vx v9, v8, a1
+; RV64V-NEXT:    vsrl.vx v9, v8, s0
 ; RV64V-NEXT:    li a2, 40
 ; RV64V-NEXT:    vsrl.vx v10, v8, a2
 ; RV64V-NEXT:    lui a3, 16
@@ -35720,7 +35720,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64V-NEXT:    vsll.vi v11, v11, 24
 ; RV64V-NEXT:    vsll.vi v12, v12, 8
 ; RV64V-NEXT:    vand.vx v13, v8, a3
-; RV64V-NEXT:    vsll.vx v14, v8, a1
+; RV64V-NEXT:    vsll.vx v14, v8, s0
 ; RV64V-NEXT:    vsll.vx v13, v13, a2
 ; RV64V-NEXT:    vor.vv v11, v11, v12
 ; RV64V-NEXT:    vor.vv v12, v14, v13
@@ -35807,15 +35807,15 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64V-NEXT:    vmul.vx v13, v11, t4
 ; RV64V-NEXT:    addi t5, t5, 1092
 ; RV64V-NEXT:    slli t6, t5, 32
-; RV64V-NEXT:    lui s0, %hi(.LCPI45_0)
+; RV64V-NEXT:    lui a1, %hi(.LCPI45_0)
 ; RV64V-NEXT:    add t5, t5, t6
-; RV64V-NEXT:    ld t6, %lo(.LCPI45_0)(s0)
+; RV64V-NEXT:    ld a1, %lo(.LCPI45_0)(a1)
 ; RV64V-NEXT:    vand.vx v14, v9, t5
-; RV64V-NEXT:    and s0, t2, t6
-; RV64V-NEXT:    vmul.vx v15, v14, s0
+; RV64V-NEXT:    and t6, t2, a1
+; RV64V-NEXT:    vmul.vx v15, v14, t6
 ; RV64V-NEXT:    vxor.vv v12, v13, v12
-; RV64V-NEXT:    vand.vx v9, v9, t6
-; RV64V-NEXT:    vmul.vx v13, v10, s0
+; RV64V-NEXT:    vand.vx v9, v9, a1
+; RV64V-NEXT:    vmul.vx v13, v10, t6
 ; RV64V-NEXT:    vmul.vx v16, v11, t1
 ; RV64V-NEXT:    vxor.vv v12, v12, v15
 ; RV64V-NEXT:    and t2, t2, t5
@@ -35834,9 +35834,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64V-NEXT:    vor.vv v12, v13, v12
 ; RV64V-NEXT:    vxor.vv v13, v16, v15
 ; RV64V-NEXT:    vmul.vx v10, v10, t2
-; RV64V-NEXT:    vmul.vx v11, v11, s0
+; RV64V-NEXT:    vmul.vx v11, v11, t6
 ; RV64V-NEXT:    vxor.vv v13, v13, v17
-; RV64V-NEXT:    vmul.vx v15, v9, s0
+; RV64V-NEXT:    vmul.vx v15, v9, t6
 ; RV64V-NEXT:    vmul.vx v14, v14, t4
 ; RV64V-NEXT:    vxor.vv v10, v11, v10
 ; RV64V-NEXT:    vmul.vx v9, v9, t1
@@ -35845,10 +35845,10 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64V-NEXT:    vand.vx v11, v11, t5
 ; RV64V-NEXT:    vxor.vv v9, v10, v9
 ; RV64V-NEXT:    vor.vv v10, v12, v11
-; RV64V-NEXT:    vand.vx v9, v9, t6
+; RV64V-NEXT:    vand.vx v9, v9, a1
 ; RV64V-NEXT:    vor.vv v9, v10, v9
 ; RV64V-NEXT:    vsrl.vx v10, v9, a2
-; RV64V-NEXT:    vsrl.vx v11, v9, a1
+; RV64V-NEXT:    vsrl.vx v11, v9, s0
 ; RV64V-NEXT:    vand.vx v10, v10, a3
 ; RV64V-NEXT:    vsrl.vi v12, v9, 24
 ; RV64V-NEXT:    vsrl.vi v13, v9, 8
@@ -35861,7 +35861,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64V-NEXT:    vsll.vi v12, v12, 8
 ; RV64V-NEXT:    vsll.vi v13, v13, 24
 ; RV64V-NEXT:    vand.vx v14, v9, a3
-; RV64V-NEXT:    vsll.vx v9, v9, a1
+; RV64V-NEXT:    vsll.vx v9, v9, s0
 ; RV64V-NEXT:    vsll.vx v14, v14, a2
 ; RV64V-NEXT:    vor.vv v12, v13, v12
 ; RV64V-NEXT:    vor.vv v9, v9, v14
@@ -36126,9 +36126,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64ZVBC32-NEXT:    .cfi_def_cfa_offset 16
 ; RV64ZVBC32-NEXT:    sd s0, 8(sp) # 8-byte Folded Spill
 ; RV64ZVBC32-NEXT:    .cfi_offset s0, -8
-; RV64ZVBC32-NEXT:    li a1, 56
+; RV64ZVBC32-NEXT:    li s0, 56
 ; RV64ZVBC32-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
-; RV64ZVBC32-NEXT:    vsrl.vx v9, v8, a1
+; RV64ZVBC32-NEXT:    vsrl.vx v9, v8, s0
 ; RV64ZVBC32-NEXT:    li a2, 40
 ; RV64ZVBC32-NEXT:    vsrl.vx v10, v8, a2
 ; RV64ZVBC32-NEXT:    lui a3, 16
@@ -36148,7 +36148,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64ZVBC32-NEXT:    vsll.vi v11, v11, 24
 ; RV64ZVBC32-NEXT:    vsll.vi v12, v12, 8
 ; RV64ZVBC32-NEXT:    vand.vx v13, v8, a3
-; RV64ZVBC32-NEXT:    vsll.vx v14, v8, a1
+; RV64ZVBC32-NEXT:    vsll.vx v14, v8, s0
 ; RV64ZVBC32-NEXT:    vsll.vx v13, v13, a2
 ; RV64ZVBC32-NEXT:    vor.vv v11, v11, v12
 ; RV64ZVBC32-NEXT:    vor.vv v12, v14, v13
@@ -36235,15 +36235,15 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64ZVBC32-NEXT:    vmul.vx v13, v11, t4
 ; RV64ZVBC32-NEXT:    addi t5, t5, 1092
 ; RV64ZVBC32-NEXT:    slli t6, t5, 32
-; RV64ZVBC32-NEXT:    lui s0, %hi(.LCPI45_0)
+; RV64ZVBC32-NEXT:    lui a1, %hi(.LCPI45_0)
 ; RV64ZVBC32-NEXT:    add t5, t5, t6
-; RV64ZVBC32-NEXT:    ld t6, %lo(.LCPI45_0)(s0)
+; RV64ZVBC32-NEXT:    ld a1, %lo(.LCPI45_0)(a1)
 ; RV64ZVBC32-NEXT:    vand.vx v14, v9, t5
-; RV64ZVBC32-NEXT:    and s0, t2, t6
-; RV64ZVBC32-NEXT:    vmul.vx v15, v14, s0
+; RV64ZVBC32-NEXT:    and t6, t2, a1
+; RV64ZVBC32-NEXT:    vmul.vx v15, v14, t6
 ; RV64ZVBC32-NEXT:    vxor.vv v12, v13, v12
-; RV64ZVBC32-NEXT:    vand.vx v9, v9, t6
-; RV64ZVBC32-NEXT:    vmul.vx v13, v10, s0
+; RV64ZVBC32-NEXT:    vand.vx v9, v9, a1
+; RV64ZVBC32-NEXT:    vmul.vx v13, v10, t6
 ; RV64ZVBC32-NEXT:    vmul.vx v16, v11, t1
 ; RV64ZVBC32-NEXT:    vxor.vv v12, v12, v15
 ; RV64ZVBC32-NEXT:    and t2, t2, t5
@@ -36262,9 +36262,9 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64ZVBC32-NEXT:    vor.vv v12, v13, v12
 ; RV64ZVBC32-NEXT:    vxor.vv v13, v16, v15
 ; RV64ZVBC32-NEXT:    vmul.vx v10, v10, t2
-; RV64ZVBC32-NEXT:    vmul.vx v11, v11, s0
+; RV64ZVBC32-NEXT:    vmul.vx v11, v11, t6
 ; RV64ZVBC32-NEXT:    vxor.vv v13, v13, v17
-; RV64ZVBC32-NEXT:    vmul.vx v15, v9, s0
+; RV64ZVBC32-NEXT:    vmul.vx v15, v9, t6
 ; RV64ZVBC32-NEXT:    vmul.vx v14, v14, t4
 ; RV64ZVBC32-NEXT:    vxor.vv v10, v11, v10
 ; RV64ZVBC32-NEXT:    vmul.vx v9, v9, t1
@@ -36273,10 +36273,10 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64ZVBC32-NEXT:    vand.vx v11, v11, t5
 ; RV64ZVBC32-NEXT:    vxor.vv v9, v10, v9
 ; RV64ZVBC32-NEXT:    vor.vv v10, v12, v11
-; RV64ZVBC32-NEXT:    vand.vx v9, v9, t6
+; RV64ZVBC32-NEXT:    vand.vx v9, v9, a1
 ; RV64ZVBC32-NEXT:    vor.vv v9, v10, v9
 ; RV64ZVBC32-NEXT:    vsrl.vx v10, v9, a2
-; RV64ZVBC32-NEXT:    vsrl.vx v11, v9, a1
+; RV64ZVBC32-NEXT:    vsrl.vx v11, v9, s0
 ; RV64ZVBC32-NEXT:    vand.vx v10, v10, a3
 ; RV64ZVBC32-NEXT:    vsrl.vi v12, v9, 24
 ; RV64ZVBC32-NEXT:    vsrl.vi v13, v9, 8
@@ -36289,7 +36289,7 @@ define <vscale x 1 x i64> @clmulh_nxv1i64_vx_mask(<vscale x 1 x i64> %va, i64 %b
 ; RV64ZVBC32-NEXT:    vsll.vi v12, v12, 8
 ; RV64ZVBC32-NEXT:    vsll.vi v13, v13, 24
 ; RV64ZVBC32-NEXT:    vand.vx v14, v9, a3
-; RV64ZVBC32-NEXT:    vsll.vx v9, v9, a1
+; RV64ZVBC32-NEXT:    vsll.vx v9, v9, s0
 ; RV64ZVBC32-NEXT:    vsll.vx v14, v14, a2
 ; RV64ZVBC32-NEXT:    vor.vv v12, v13, v12
 ; RV64ZVBC32-NEXT:    vor.vv v9, v9, v14
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-buildvec.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-buildvec.ll
index a0b8102ce0f929..af4e4547c24dc2 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-buildvec.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-buildvec.ll
@@ -565,23 +565,11 @@ define <32 x float> @buildvec_v32f32(float %e0, float %e1, float %e2, float %e3,
 ; RV32-NEXT:    .cfi_def_cfa_offset 256
 ; RV32-NEXT:    sw ra, 252(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s0, 248(sp) # 4-byte Folded Spill
-; RV32-NEXT:    fsd fs0, 240(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs1, 232(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs2, 224(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs3, 216(sp) # 8-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
 ; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    .cfi_offset fs0, -16
-; RV32-NEXT:    .cfi_offset fs1, -24
-; RV32-NEXT:    .cfi_offset fs2, -32
-; RV32-NEXT:    .cfi_offset fs3, -40
 ; RV32-NEXT:    addi s0, sp, 256
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -128
-; RV32-NEXT:    flw ft0, 0(s0)
-; RV32-NEXT:    flw ft1, 4(s0)
-; RV32-NEXT:    flw ft2, 8(s0)
-; RV32-NEXT:    flw ft3, 12(s0)
 ; RV32-NEXT:    flw ft4, 16(s0)
 ; RV32-NEXT:    flw ft5, 20(s0)
 ; RV32-NEXT:    flw ft6, 24(s0)
@@ -590,26 +578,30 @@ define <32 x float> @buildvec_v32f32(float %e0, float %e1, float %e2, float %e3,
 ; RV32-NEXT:    flw ft9, 52(s0)
 ; RV32-NEXT:    flw ft10, 56(s0)
 ; RV32-NEXT:    flw ft11, 60(s0)
-; RV32-NEXT:    flw fs0, 32(s0)
-; RV32-NEXT:    flw fs1, 36(s0)
-; RV32-NEXT:    flw fs2, 40(s0)
-; RV32-NEXT:    flw fs3, 44(s0)
+; RV32-NEXT:    flw ft0, 32(s0)
+; RV32-NEXT:    flw ft1, 36(s0)
+; RV32-NEXT:    flw ft2, 40(s0)
+; RV32-NEXT:    flw ft3, 44(s0)
 ; RV32-NEXT:    fsw ft8, 112(sp)
 ; RV32-NEXT:    fsw ft9, 116(sp)
 ; RV32-NEXT:    fsw ft10, 120(sp)
 ; RV32-NEXT:    fsw ft11, 124(sp)
-; RV32-NEXT:    fsw fs0, 96(sp)
-; RV32-NEXT:    fsw fs1, 100(sp)
-; RV32-NEXT:    fsw fs2, 104(sp)
-; RV32-NEXT:    fsw fs3, 108(sp)
+; RV32-NEXT:    fsw ft0, 96(sp)
+; RV32-NEXT:    fsw ft1, 100(sp)
+; RV32-NEXT:    fsw ft2, 104(sp)
+; RV32-NEXT:    fsw ft3, 108(sp)
 ; RV32-NEXT:    fsw ft4, 80(sp)
 ; RV32-NEXT:    fsw ft5, 84(sp)
 ; RV32-NEXT:    fsw ft6, 88(sp)
 ; RV32-NEXT:    fsw ft7, 92(sp)
+; RV32-NEXT:    flw ft0, 0(s0)
 ; RV32-NEXT:    fsw ft0, 64(sp)
-; RV32-NEXT:    fsw ft1, 68(sp)
-; RV32-NEXT:    fsw ft2, 72(sp)
-; RV32-NEXT:    fsw ft3, 76(sp)
+; RV32-NEXT:    flw ft0, 4(s0)
+; RV32-NEXT:    fsw ft0, 68(sp)
+; RV32-NEXT:    flw ft0, 8(s0)
+; RV32-NEXT:    fsw ft0, 72(sp)
+; RV32-NEXT:    flw ft0, 12(s0)
+; RV32-NEXT:    fsw ft0, 76(sp)
 ; RV32-NEXT:    sw a4, 48(sp)
 ; RV32-NEXT:    sw a5, 52(sp)
 ; RV32-NEXT:    sw a6, 56(sp)
@@ -634,16 +626,8 @@ define <32 x float> @buildvec_v32f32(float %e0, float %e1, float %e2, float %e3,
 ; RV32-NEXT:    .cfi_def_cfa sp, 256
 ; RV32-NEXT:    lw ra, 252(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s0, 248(sp) # 4-byte Folded Reload
-; RV32-NEXT:    fld fs0, 240(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs1, 232(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs2, 224(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs3, 216(sp) # 8-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
 ; RV32-NEXT:    .cfi_restore s0
-; RV32-NEXT:    .cfi_restore fs0
-; RV32-NEXT:    .cfi_restore fs1
-; RV32-NEXT:    .cfi_restore fs2
-; RV32-NEXT:    .cfi_restore fs3
 ; RV32-NEXT:    addi sp, sp, 256
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
 ; RV32-NEXT:    ret
@@ -658,28 +642,12 @@ define <32 x float> @buildvec_v32f32(float %e0, float %e1, float %e2, float %e3,
 ; RV64-NEXT:    fsd fs1, 224(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    fsd fs2, 216(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    fsd fs3, 208(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs4, 200(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs5, 192(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs6, 184(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs7, 176(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs8, 168(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs9, 160(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs10, 152(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs11, 144(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    .cfi_offset ra, -8
 ; RV64-NEXT:    .cfi_offset s0, -16
 ; RV64-NEXT:    .cfi_offset fs0, -24
 ; RV64-NEXT:    .cfi_offset fs1, -32
 ; RV64-NEXT:    .cfi_offset fs2, -40
 ; RV64-NEXT:    .cfi_offset fs3, -48
-; RV64-NEXT:    .cfi_offset fs4, -56
-; RV64-NEXT:    .cfi_offset fs5, -64
-; RV64-NEXT:    .cfi_offset fs6, -72
-; RV64-NEXT:    .cfi_offset fs7, -80
-; RV64-NEXT:    .cfi_offset fs8, -88
-; RV64-NEXT:    .cfi_offset fs9, -96
-; RV64-NEXT:    .cfi_offset fs10, -104
-; RV64-NEXT:    .cfi_offset fs11, -112
 ; RV64-NEXT:    addi s0, sp, 256
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -128
@@ -687,42 +655,42 @@ define <32 x float> @buildvec_v32f32(float %e0, float %e1, float %e2, float %e3,
 ; RV64-NEXT:    fmv.w.x ft1, a1
 ; RV64-NEXT:    fmv.w.x ft2, a2
 ; RV64-NEXT:    fmv.w.x ft3, a3
-; RV64-NEXT:    fmv.w.x ft4, a4
-; RV64-NEXT:    fmv.w.x ft5, a5
-; RV64-NEXT:    fmv.w.x ft6, a6
-; RV64-NEXT:    fmv.w.x ft7, a7
+; RV64-NEXT:    fmv.w.x fs3, a4
+; RV64-NEXT:    fmv.w.x ft6, a5
+; RV64-NEXT:    fmv.w.x fs2, a6
+; RV64-NEXT:    fmv.w.x ft5, a7
+; RV64-NEXT:    flw ft8, 96(s0)
+; RV64-NEXT:    flw fs0, 104(s0)
+; RV64-NEXT:    flw fs1, 112(s0)
+; RV64-NEXT:    flw ft10, 120(s0)
+; RV64-NEXT:    flw ft7, 64(s0)
+; RV64-NEXT:    flw ft11, 72(s0)
+; RV64-NEXT:    flw ft9, 80(s0)
+; RV64-NEXT:    flw ft4, 88(s0)
+; RV64-NEXT:    fsw ft8, 112(sp)
+; RV64-NEXT:    fsw fs0, 116(sp)
+; RV64-NEXT:    fsw fs1, 120(sp)
+; RV64-NEXT:    fsw ft10, 124(sp)
+; RV64-NEXT:    fsw ft7, 96(sp)
+; RV64-NEXT:    fsw ft11, 100(sp)
+; RV64-NEXT:    fsw ft9, 104(sp)
+; RV64-NEXT:    fsw ft4, 108(sp)
+; RV64-NEXT:    flw ft8, 32(s0)
+; RV64-NEXT:    fsw ft8, 80(sp)
+; RV64-NEXT:    flw ft8, 40(s0)
+; RV64-NEXT:    fsw ft8, 84(sp)
+; RV64-NEXT:    flw ft8, 48(s0)
+; RV64-NEXT:    fsw ft8, 88(sp)
+; RV64-NEXT:    flw ft8, 56(s0)
+; RV64-NEXT:    fsw ft8, 92(sp)
 ; RV64-NEXT:    flw ft8, 0(s0)
-; RV64-NEXT:    flw ft9, 8(s0)
-; RV64-NEXT:    flw ft10, 16(s0)
-; RV64-NEXT:    flw ft11, 24(s0)
-; RV64-NEXT:    flw fs0, 32(s0)
-; RV64-NEXT:    flw fs1, 40(s0)
-; RV64-NEXT:    flw fs2, 48(s0)
-; RV64-NEXT:    flw fs3, 56(s0)
-; RV64-NEXT:    flw fs4, 96(s0)
-; RV64-NEXT:    flw fs5, 104(s0)
-; RV64-NEXT:    flw fs6, 112(s0)
-; RV64-NEXT:    flw fs7, 120(s0)
-; RV64-NEXT:    flw fs8, 64(s0)
-; RV64-NEXT:    flw fs9, 72(s0)
-; RV64-NEXT:    flw fs10, 80(s0)
-; RV64-NEXT:    flw fs11, 88(s0)
-; RV64-NEXT:    fsw fs4, 112(sp)
-; RV64-NEXT:    fsw fs5, 116(sp)
-; RV64-NEXT:    fsw fs6, 120(sp)
-; RV64-NEXT:    fsw fs7, 124(sp)
-; RV64-NEXT:    fsw fs8, 96(sp)
-; RV64-NEXT:    fsw fs9, 100(sp)
-; RV64-NEXT:    fsw fs10, 104(sp)
-; RV64-NEXT:    fsw fs11, 108(sp)
-; RV64-NEXT:    fsw fs0, 80(sp)
-; RV64-NEXT:    fsw fs1, 84(sp)
-; RV64-NEXT:    fsw fs2, 88(sp)
-; RV64-NEXT:    fsw fs3, 92(sp)
 ; RV64-NEXT:    fsw ft8, 64(sp)
-; RV64-NEXT:    fsw ft9, 68(sp)
-; RV64-NEXT:    fsw ft10, 72(sp)
-; RV64-NEXT:    fsw ft11, 76(sp)
+; RV64-NEXT:    flw ft8, 8(s0)
+; RV64-NEXT:    fsw ft8, 68(sp)
+; RV64-NEXT:    flw ft8, 16(s0)
+; RV64-NEXT:    fsw ft8, 72(sp)
+; RV64-NEXT:    flw ft8, 24(s0)
+; RV64-NEXT:    fsw ft8, 76(sp)
 ; RV64-NEXT:    fsw fa4, 16(sp)
 ; RV64-NEXT:    fsw fa5, 20(sp)
 ; RV64-NEXT:    fsw fa6, 24(sp)
@@ -731,10 +699,10 @@ define <32 x float> @buildvec_v32f32(float %e0, float %e1, float %e2, float %e3,
 ; RV64-NEXT:    fsw fa1, 4(sp)
 ; RV64-NEXT:    fsw fa2, 8(sp)
 ; RV64-NEXT:    fsw fa3, 12(sp)
-; RV64-NEXT:    fsw ft4, 48(sp)
-; RV64-NEXT:    fsw ft5, 52(sp)
-; RV64-NEXT:    fsw ft6, 56(sp)
-; RV64-NEXT:    fsw ft7, 60(sp)
+; RV64-NEXT:    fsw fs3, 48(sp)
+; RV64-NEXT:    fsw ft6, 52(sp)
+; RV64-NEXT:    fsw fs2, 56(sp)
+; RV64-NEXT:    fsw ft5, 60(sp)
 ; RV64-NEXT:    fsw ft0, 32(sp)
 ; RV64-NEXT:    fsw ft1, 36(sp)
 ; RV64-NEXT:    fsw ft2, 40(sp)
@@ -751,28 +719,12 @@ define <32 x float> @buildvec_v32f32(float %e0, float %e1, float %e2, float %e3,
 ; RV64-NEXT:    fld fs1, 224(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    fld fs2, 216(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    fld fs3, 208(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs4, 200(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs5, 192(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs6, 184(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs7, 176(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs8, 168(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs9, 160(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs10, 152(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs11, 144(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    .cfi_restore ra
 ; RV64-NEXT:    .cfi_restore s0
 ; RV64-NEXT:    .cfi_restore fs0
 ; RV64-NEXT:    .cfi_restore fs1
 ; RV64-NEXT:    .cfi_restore fs2
 ; RV64-NEXT:    .cfi_restore fs3
-; RV64-NEXT:    .cfi_restore fs4
-; RV64-NEXT:    .cfi_restore fs5
-; RV64-NEXT:    .cfi_restore fs6
-; RV64-NEXT:    .cfi_restore fs7
-; RV64-NEXT:    .cfi_restore fs8
-; RV64-NEXT:    .cfi_restore fs9
-; RV64-NEXT:    .cfi_restore fs10
-; RV64-NEXT:    .cfi_restore fs11
 ; RV64-NEXT:    addi sp, sp, 256
 ; RV64-NEXT:    .cfi_def_cfa_offset 0
 ; RV64-NEXT:    ret
@@ -1008,32 +960,8 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV32-NEXT:    .cfi_def_cfa_offset 512
 ; RV32-NEXT:    sw ra, 508(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s0, 504(sp) # 4-byte Folded Spill
-; RV32-NEXT:    fsd fs0, 496(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs1, 488(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs2, 480(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs3, 472(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs4, 464(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs5, 456(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs6, 448(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs7, 440(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs8, 432(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs9, 424(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs10, 416(sp) # 8-byte Folded Spill
-; RV32-NEXT:    fsd fs11, 408(sp) # 8-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
 ; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    .cfi_offset fs0, -16
-; RV32-NEXT:    .cfi_offset fs1, -24
-; RV32-NEXT:    .cfi_offset fs2, -32
-; RV32-NEXT:    .cfi_offset fs3, -40
-; RV32-NEXT:    .cfi_offset fs4, -48
-; RV32-NEXT:    .cfi_offset fs5, -56
-; RV32-NEXT:    .cfi_offset fs6, -64
-; RV32-NEXT:    .cfi_offset fs7, -72
-; RV32-NEXT:    .cfi_offset fs8, -80
-; RV32-NEXT:    .cfi_offset fs9, -88
-; RV32-NEXT:    .cfi_offset fs10, -96
-; RV32-NEXT:    .cfi_offset fs11, -104
 ; RV32-NEXT:    addi s0, sp, 512
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -128
@@ -1049,46 +977,46 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV32-NEXT:    sw a6, 120(sp)
 ; RV32-NEXT:    sw a7, 124(sp)
 ; RV32-NEXT:    fld ft3, 120(sp)
+; RV32-NEXT:    fld ft8, 128(s0)
+; RV32-NEXT:    fld ft9, 136(s0)
+; RV32-NEXT:    fld ft10, 144(s0)
+; RV32-NEXT:    fld ft11, 96(s0)
+; RV32-NEXT:    fld ft4, 152(s0)
+; RV32-NEXT:    fld ft5, 104(s0)
+; RV32-NEXT:    fld ft6, 112(s0)
+; RV32-NEXT:    fld ft7, 120(s0)
+; RV32-NEXT:    fsd ft8, 224(sp)
+; RV32-NEXT:    fsd ft9, 232(sp)
+; RV32-NEXT:    fsd ft10, 240(sp)
+; RV32-NEXT:    fsd ft4, 248(sp)
+; RV32-NEXT:    fsd ft11, 192(sp)
+; RV32-NEXT:    fsd ft5, 200(sp)
+; RV32-NEXT:    fsd ft6, 208(sp)
+; RV32-NEXT:    fsd ft7, 216(sp)
+; RV32-NEXT:    fld ft4, 64(s0)
+; RV32-NEXT:    fsd ft4, 160(sp)
+; RV32-NEXT:    fld ft4, 72(s0)
+; RV32-NEXT:    fsd ft4, 168(sp)
+; RV32-NEXT:    fld ft4, 80(s0)
+; RV32-NEXT:    fsd ft4, 176(sp)
+; RV32-NEXT:    fld ft4, 88(s0)
+; RV32-NEXT:    fsd ft4, 184(sp)
+; RV32-NEXT:    fld ft4, 32(s0)
+; RV32-NEXT:    fsd ft4, 128(sp)
+; RV32-NEXT:    fld ft4, 40(s0)
+; RV32-NEXT:    fsd ft4, 136(sp)
+; RV32-NEXT:    fld ft4, 48(s0)
+; RV32-NEXT:    fsd ft4, 144(sp)
+; RV32-NEXT:    fld ft4, 56(s0)
+; RV32-NEXT:    fsd ft4, 152(sp)
 ; RV32-NEXT:    fld ft4, 0(s0)
-; RV32-NEXT:    fld ft5, 8(s0)
-; RV32-NEXT:    fld ft6, 16(s0)
-; RV32-NEXT:    fld ft7, 24(s0)
-; RV32-NEXT:    fld ft8, 32(s0)
-; RV32-NEXT:    fld ft9, 40(s0)
-; RV32-NEXT:    fld ft10, 48(s0)
-; RV32-NEXT:    fld ft11, 56(s0)
-; RV32-NEXT:    fld fs0, 64(s0)
-; RV32-NEXT:    fld fs1, 72(s0)
-; RV32-NEXT:    fld fs2, 80(s0)
-; RV32-NEXT:    fld fs3, 88(s0)
-; RV32-NEXT:    fld fs4, 128(s0)
-; RV32-NEXT:    fld fs5, 136(s0)
-; RV32-NEXT:    fld fs6, 144(s0)
-; RV32-NEXT:    fld fs7, 96(s0)
-; RV32-NEXT:    fld fs8, 152(s0)
-; RV32-NEXT:    fld fs9, 104(s0)
-; RV32-NEXT:    fld fs10, 112(s0)
-; RV32-NEXT:    fld fs11, 120(s0)
-; RV32-NEXT:    fsd fs4, 224(sp)
-; RV32-NEXT:    fsd fs5, 232(sp)
-; RV32-NEXT:    fsd fs6, 240(sp)
-; RV32-NEXT:    fsd fs8, 248(sp)
-; RV32-NEXT:    fsd fs7, 192(sp)
-; RV32-NEXT:    fsd fs9, 200(sp)
-; RV32-NEXT:    fsd fs10, 208(sp)
-; RV32-NEXT:    fsd fs11, 216(sp)
-; RV32-NEXT:    fsd fs0, 160(sp)
-; RV32-NEXT:    fsd fs1, 168(sp)
-; RV32-NEXT:    fsd fs2, 176(sp)
-; RV32-NEXT:    fsd fs3, 184(sp)
-; RV32-NEXT:    fsd ft8, 128(sp)
-; RV32-NEXT:    fsd ft9, 136(sp)
-; RV32-NEXT:    fsd ft10, 144(sp)
-; RV32-NEXT:    fsd ft11, 152(sp)
 ; RV32-NEXT:    fsd ft4, 352(sp)
-; RV32-NEXT:    fsd ft5, 360(sp)
-; RV32-NEXT:    fsd ft6, 368(sp)
-; RV32-NEXT:    fsd ft7, 376(sp)
+; RV32-NEXT:    fld ft4, 8(s0)
+; RV32-NEXT:    fsd ft4, 360(sp)
+; RV32-NEXT:    fld ft4, 16(s0)
+; RV32-NEXT:    fsd ft4, 368(sp)
+; RV32-NEXT:    fld ft4, 24(s0)
+; RV32-NEXT:    fsd ft4, 376(sp)
 ; RV32-NEXT:    fsd fa4, 288(sp)
 ; RV32-NEXT:    fsd fa5, 296(sp)
 ; RV32-NEXT:    fsd fa6, 304(sp)
@@ -1110,32 +1038,8 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV32-NEXT:    .cfi_def_cfa sp, 512
 ; RV32-NEXT:    lw ra, 508(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s0, 504(sp) # 4-byte Folded Reload
-; RV32-NEXT:    fld fs0, 496(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs1, 488(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs2, 480(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs3, 472(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs4, 464(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs5, 456(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs6, 448(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs7, 440(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs8, 432(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs9, 424(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs10, 416(sp) # 8-byte Folded Reload
-; RV32-NEXT:    fld fs11, 408(sp) # 8-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
 ; RV32-NEXT:    .cfi_restore s0
-; RV32-NEXT:    .cfi_restore fs0
-; RV32-NEXT:    .cfi_restore fs1
-; RV32-NEXT:    .cfi_restore fs2
-; RV32-NEXT:    .cfi_restore fs3
-; RV32-NEXT:    .cfi_restore fs4
-; RV32-NEXT:    .cfi_restore fs5
-; RV32-NEXT:    .cfi_restore fs6
-; RV32-NEXT:    .cfi_restore fs7
-; RV32-NEXT:    .cfi_restore fs8
-; RV32-NEXT:    .cfi_restore fs9
-; RV32-NEXT:    .cfi_restore fs10
-; RV32-NEXT:    .cfi_restore fs11
 ; RV32-NEXT:    addi sp, sp, 512
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
 ; RV32-NEXT:    ret
@@ -1146,23 +1050,11 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV64-NEXT:    .cfi_def_cfa_offset 384
 ; RV64-NEXT:    sd ra, 376(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    sd s0, 368(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs0, 360(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs1, 352(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs2, 344(sp) # 8-byte Folded Spill
-; RV64-NEXT:    fsd fs3, 336(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    .cfi_offset ra, -8
 ; RV64-NEXT:    .cfi_offset s0, -16
-; RV64-NEXT:    .cfi_offset fs0, -24
-; RV64-NEXT:    .cfi_offset fs1, -32
-; RV64-NEXT:    .cfi_offset fs2, -40
-; RV64-NEXT:    .cfi_offset fs3, -48
 ; RV64-NEXT:    addi s0, sp, 384
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -128
-; RV64-NEXT:    fld ft0, 0(s0)
-; RV64-NEXT:    fld ft1, 8(s0)
-; RV64-NEXT:    fld ft2, 16(s0)
-; RV64-NEXT:    fld ft3, 24(s0)
 ; RV64-NEXT:    fld ft4, 32(s0)
 ; RV64-NEXT:    fld ft5, 40(s0)
 ; RV64-NEXT:    fld ft6, 48(s0)
@@ -1171,10 +1063,10 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV64-NEXT:    fld ft9, 72(s0)
 ; RV64-NEXT:    fld ft10, 80(s0)
 ; RV64-NEXT:    fld ft11, 88(s0)
-; RV64-NEXT:    fld fs0, 96(s0)
-; RV64-NEXT:    fld fs1, 104(s0)
-; RV64-NEXT:    fld fs2, 112(s0)
-; RV64-NEXT:    fld fs3, 120(s0)
+; RV64-NEXT:    fld ft0, 96(s0)
+; RV64-NEXT:    fld ft1, 104(s0)
+; RV64-NEXT:    fld ft2, 112(s0)
+; RV64-NEXT:    fld ft3, 120(s0)
 ; RV64-NEXT:    sd a4, 224(sp)
 ; RV64-NEXT:    sd a5, 232(sp)
 ; RV64-NEXT:    sd a6, 240(sp)
@@ -1191,10 +1083,10 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV64-NEXT:    fsd fa1, 136(sp)
 ; RV64-NEXT:    fsd fa2, 144(sp)
 ; RV64-NEXT:    fsd fa3, 152(sp)
-; RV64-NEXT:    fsd fs0, 96(sp)
-; RV64-NEXT:    fsd fs1, 104(sp)
-; RV64-NEXT:    fsd fs2, 112(sp)
-; RV64-NEXT:    fsd fs3, 120(sp)
+; RV64-NEXT:    fsd ft0, 96(sp)
+; RV64-NEXT:    fsd ft1, 104(sp)
+; RV64-NEXT:    fsd ft2, 112(sp)
+; RV64-NEXT:    fsd ft3, 120(sp)
 ; RV64-NEXT:    fsd ft8, 64(sp)
 ; RV64-NEXT:    fsd ft9, 72(sp)
 ; RV64-NEXT:    fsd ft10, 80(sp)
@@ -1203,10 +1095,14 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV64-NEXT:    fsd ft5, 40(sp)
 ; RV64-NEXT:    fsd ft6, 48(sp)
 ; RV64-NEXT:    fsd ft7, 56(sp)
-; RV64-NEXT:    fsd ft0, 0(sp)
-; RV64-NEXT:    fsd ft1, 8(sp)
-; RV64-NEXT:    fsd ft2, 16(sp)
-; RV64-NEXT:    fsd ft3, 24(sp)
+; RV64-NEXT:    fld fa5, 0(s0)
+; RV64-NEXT:    fsd fa5, 0(sp)
+; RV64-NEXT:    fld fa5, 8(s0)
+; RV64-NEXT:    fsd fa5, 8(sp)
+; RV64-NEXT:    fld fa5, 16(s0)
+; RV64-NEXT:    fsd fa5, 16(sp)
+; RV64-NEXT:    fld fa5, 24(s0)
+; RV64-NEXT:    fsd fa5, 24(sp)
 ; RV64-NEXT:    addi a0, sp, 128
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
@@ -1216,16 +1112,8 @@ define <32 x double> @buildvec_v32f64(double %e0, double %e1, double %e2, double
 ; RV64-NEXT:    .cfi_def_cfa sp, 384
 ; RV64-NEXT:    ld ra, 376(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    ld s0, 368(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs0, 360(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs1, 352(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs2, 344(sp) # 8-byte Folded Reload
-; RV64-NEXT:    fld fs3, 336(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    .cfi_restore ra
 ; RV64-NEXT:    .cfi_restore s0
-; RV64-NEXT:    .cfi_restore fs0
-; RV64-NEXT:    .cfi_restore fs1
-; RV64-NEXT:    .cfi_restore fs2
-; RV64-NEXT:    .cfi_restore fs3
 ; RV64-NEXT:    addi sp, sp, 384
 ; RV64-NEXT:    .cfi_def_cfa_offset 0
 ; RV64-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-scatter.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-scatter.ll
index 366d2e881d9198..c0eda208ffd835 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-scatter.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-scatter.ll
@@ -3171,13 +3171,10 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
 ; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    lw a1, 56(a0)
+; RV32ZVE32F-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a2, 60(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 40(a0)
 ; RV32ZVE32F-NEXT:    lw a6, 44(a0)
@@ -3190,11 +3187,11 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV32ZVE32F-NEXT:    lw t0, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 36(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 8(a0)
-; RV32ZVE32F-NEXT:    lw s0, 12(a0)
+; RV32ZVE32F-NEXT:    lw a1, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t4, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 20(a0)
-; RV32ZVE32F-NEXT:    andi s1, a7, 1
-; RV32ZVE32F-NEXT:    bnez s1, .LBB41_10
+; RV32ZVE32F-NEXT:    andi s0, a7, 1
+; RV32ZVE32F-NEXT:    bnez s0, .LBB41_10
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a7, 2
 ; RV32ZVE32F-NEXT:    bnez a0, .LBB41_11
@@ -3219,27 +3216,46 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV32ZVE32F-NEXT:  .LBB41_8: # %cond.store13
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a1, 0(a0)
-; RV32ZVE32F-NEXT:    sw a2, 4(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    sw a2, 4(a1)
 ; RV32ZVE32F-NEXT:  .LBB41_9: # %else14
 ; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
 ; RV32ZVE32F-NEXT:  .LBB41_10: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vsetvli zero, zero, e32, m4, ta, ma
-; RV32ZVE32F-NEXT:    vmv.x.s s1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s s0, v8
+; RV32ZVE32F-NEXT:    sw t5, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    mv a5, a4
+; RV32ZVE32F-NEXT:    mv a4, a3
+; RV32ZVE32F-NEXT:    mv a3, a2
+; RV32ZVE32F-NEXT:    lw a2, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(s1)
-; RV32ZVE32F-NEXT:    sw a0, 4(s1)
+; RV32ZVE32F-NEXT:    sw a2, 0(s0)
+; RV32ZVE32F-NEXT:    mv a2, a3
+; RV32ZVE32F-NEXT:    mv a3, a4
+; RV32ZVE32F-NEXT:    mv a4, a5
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    lw t5, 4(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(s0)
 ; RV32ZVE32F-NEXT:    andi a0, a7, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB41_2
 ; RV32ZVE32F-NEXT:  .LBB41_11: # %cond.store1
@@ -3247,7 +3263,7 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw t6, 0(a0)
-; RV32ZVE32F-NEXT:    sw s0, 4(a0)
+; RV32ZVE32F-NEXT:    sw a1, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a7, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB41_3
 ; RV32ZVE32F-NEXT:  .LBB41_12: # %cond.store3
@@ -3297,15 +3313,13 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV64ZVE32F-NEXT:    addi sp, sp, -32
 ; RV64ZVE32F-NEXT:    .cfi_def_cfa_offset 32
 ; RV64ZVE32F-NEXT:    sd s0, 24(sp) # 8-byte Folded Spill
-; RV64ZVE32F-NEXT:    sd s1, 16(sp) # 8-byte Folded Spill
-; RV64ZVE32F-NEXT:    sd s2, 8(sp) # 8-byte Folded Spill
 ; RV64ZVE32F-NEXT:    .cfi_offset s0, -8
-; RV64ZVE32F-NEXT:    .cfi_offset s1, -16
-; RV64ZVE32F-NEXT:    .cfi_offset s2, -24
 ; RV64ZVE32F-NEXT:    .cfi_remember_state
 ; RV64ZVE32F-NEXT:    ld a4, 40(a1)
-; RV64ZVE32F-NEXT:    ld a3, 48(a1)
+; RV64ZVE32F-NEXT:    ld a2, 48(a1)
+; RV64ZVE32F-NEXT:    sd a2, 8(sp) # 8-byte Folded Spill
 ; RV64ZVE32F-NEXT:    ld a2, 56(a1)
+; RV64ZVE32F-NEXT:    sd a2, 16(sp) # 8-byte Folded Spill
 ; RV64ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV64ZVE32F-NEXT:    vmv.x.s a5, v0
 ; RV64ZVE32F-NEXT:    ld t6, 8(a1)
@@ -3315,12 +3329,12 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV64ZVE32F-NEXT:    ld t0, 40(a0)
 ; RV64ZVE32F-NEXT:    ld a7, 48(a0)
 ; RV64ZVE32F-NEXT:    ld a6, 56(a0)
-; RV64ZVE32F-NEXT:    ld s1, 8(a0)
-; RV64ZVE32F-NEXT:    ld s0, 16(a0)
+; RV64ZVE32F-NEXT:    ld a3, 8(a0)
+; RV64ZVE32F-NEXT:    ld a2, 16(a0)
 ; RV64ZVE32F-NEXT:    ld t5, 24(a0)
 ; RV64ZVE32F-NEXT:    ld t3, 32(a0)
-; RV64ZVE32F-NEXT:    andi s2, a5, 1
-; RV64ZVE32F-NEXT:    bnez s2, .LBB41_10
+; RV64ZVE32F-NEXT:    andi s0, a5, 1
+; RV64ZVE32F-NEXT:    bnez s0, .LBB41_10
 ; RV64ZVE32F-NEXT:  # %bb.1: # %else
 ; RV64ZVE32F-NEXT:    andi a0, a5, 2
 ; RV64ZVE32F-NEXT:    bnez a0, .LBB41_11
@@ -3343,14 +3357,11 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV64ZVE32F-NEXT:    andi a0, a5, -128
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB41_9
 ; RV64ZVE32F-NEXT:  .LBB41_8: # %cond.store13
-; RV64ZVE32F-NEXT:    sd a6, 0(a2)
+; RV64ZVE32F-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; RV64ZVE32F-NEXT:    sd a6, 0(a0)
 ; RV64ZVE32F-NEXT:  .LBB41_9: # %else14
 ; RV64ZVE32F-NEXT:    ld s0, 24(sp) # 8-byte Folded Reload
-; RV64ZVE32F-NEXT:    ld s1, 16(sp) # 8-byte Folded Reload
-; RV64ZVE32F-NEXT:    ld s2, 8(sp) # 8-byte Folded Reload
 ; RV64ZVE32F-NEXT:    .cfi_restore s0
-; RV64ZVE32F-NEXT:    .cfi_restore s1
-; RV64ZVE32F-NEXT:    .cfi_restore s2
 ; RV64ZVE32F-NEXT:    addi sp, sp, 32
 ; RV64ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV64ZVE32F-NEXT:    ret
@@ -3362,11 +3373,11 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV64ZVE32F-NEXT:    andi a0, a5, 2
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB41_2
 ; RV64ZVE32F-NEXT:  .LBB41_11: # %cond.store1
-; RV64ZVE32F-NEXT:    sd s1, 0(t6)
+; RV64ZVE32F-NEXT:    sd a3, 0(t6)
 ; RV64ZVE32F-NEXT:    andi a0, a5, 4
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB41_3
 ; RV64ZVE32F-NEXT:  .LBB41_12: # %cond.store3
-; RV64ZVE32F-NEXT:    sd s0, 0(t4)
+; RV64ZVE32F-NEXT:    sd a2, 0(t4)
 ; RV64ZVE32F-NEXT:    andi a0, a5, 8
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB41_4
 ; RV64ZVE32F-NEXT:  .LBB41_13: # %cond.store5
@@ -3382,7 +3393,8 @@ define void @mscatter_v8i64(<8 x i64> %val, <8 x ptr> %ptrs, <8 x i1> %m) {
 ; RV64ZVE32F-NEXT:    andi a0, a5, 64
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB41_7
 ; RV64ZVE32F-NEXT:  .LBB41_16: # %cond.store11
-; RV64ZVE32F-NEXT:    sd a7, 0(a3)
+; RV64ZVE32F-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64ZVE32F-NEXT:    sd a7, 0(a0)
 ; RV64ZVE32F-NEXT:    andi a0, a5, -128
 ; RV64ZVE32F-NEXT:    bnez a0, .LBB41_8
 ; RV64ZVE32F-NEXT:    j .LBB41_9
@@ -3412,19 +3424,15 @@ define void @mscatter_baseidx_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8> %id
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vsext.vf4 v10, v8
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -3434,68 +3442,76 @@ define void @mscatter_baseidx_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8> %id
 ; RV32ZVE32F-NEXT:    lw t4, 28(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t2, 36(a0)
-; RV32ZVE32F-NEXT:    lw s0, 8(a0)
-; RV32ZVE32F-NEXT:    lw s1, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 8(a0)
+; RV32ZVE32F-NEXT:    lw a3, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 20(a0)
 ; RV32ZVE32F-NEXT:    vadd.vx v8, v8, a1
 ; RV32ZVE32F-NEXT:    andi a1, a4, 1
-; RV32ZVE32F-NEXT:    bnez a1, .LBB42_10
+; RV32ZVE32F-NEXT:    bnez a1, .LBB42_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB42_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB42_10
 ; RV32ZVE32F-NEXT:  .LBB42_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB42_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB42_11
 ; RV32ZVE32F-NEXT:  .LBB42_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB42_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB42_12
 ; RV32ZVE32F-NEXT:  .LBB42_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB42_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB42_13
 ; RV32ZVE32F-NEXT:  .LBB42_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB42_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB42_14
 ; RV32ZVE32F-NEXT:  .LBB42_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB42_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB42_15
 ; RV32ZVE32F-NEXT:  .LBB42_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB42_9
-; RV32ZVE32F-NEXT:  .LBB42_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB42_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB42_16
+; RV32ZVE32F-NEXT:  .LBB42_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB42_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB42_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    sw t6, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t6, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(a1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a1)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, t6
+; RV32ZVE32F-NEXT:    lw t6, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    sw a0, 4(a1)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB42_2
-; RV32ZVE32F-NEXT:  .LBB42_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB42_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
-; RV32ZVE32F-NEXT:    sw s0, 0(a0)
-; RV32ZVE32F-NEXT:    sw s1, 4(a0)
+; RV32ZVE32F-NEXT:    sw a2, 0(a0)
+; RV32ZVE32F-NEXT:    sw a3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB42_3
-; RV32ZVE32F-NEXT:  .LBB42_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB42_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3503,7 +3519,7 @@ define void @mscatter_baseidx_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8> %id
 ; RV32ZVE32F-NEXT:    sw t6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB42_4
-; RV32ZVE32F-NEXT:  .LBB42_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB42_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3511,7 +3527,7 @@ define void @mscatter_baseidx_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8> %id
 ; RV32ZVE32F-NEXT:    sw t4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB42_5
-; RV32ZVE32F-NEXT:  .LBB42_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB42_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3519,7 +3535,7 @@ define void @mscatter_baseidx_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8> %id
 ; RV32ZVE32F-NEXT:    sw t2, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB42_6
-; RV32ZVE32F-NEXT:  .LBB42_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB42_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3527,15 +3543,25 @@ define void @mscatter_baseidx_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8> %id
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB42_7
-; RV32ZVE32F-NEXT:  .LBB42_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB42_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB42_8
-; RV32ZVE32F-NEXT:    j .LBB42_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB42_8
+; RV32ZVE32F-NEXT:  .LBB42_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_v8i8_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -3669,19 +3695,15 @@ define void @mscatter_baseidx_sext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vsext.vf4 v10, v8
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -3691,68 +3713,76 @@ define void @mscatter_baseidx_sext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    lw t4, 28(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t2, 36(a0)
-; RV32ZVE32F-NEXT:    lw s0, 8(a0)
-; RV32ZVE32F-NEXT:    lw s1, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 8(a0)
+; RV32ZVE32F-NEXT:    lw a3, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 20(a0)
 ; RV32ZVE32F-NEXT:    vadd.vx v8, v8, a1
 ; RV32ZVE32F-NEXT:    andi a1, a4, 1
-; RV32ZVE32F-NEXT:    bnez a1, .LBB43_10
+; RV32ZVE32F-NEXT:    bnez a1, .LBB43_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB43_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB43_10
 ; RV32ZVE32F-NEXT:  .LBB43_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB43_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB43_11
 ; RV32ZVE32F-NEXT:  .LBB43_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB43_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB43_12
 ; RV32ZVE32F-NEXT:  .LBB43_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB43_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB43_13
 ; RV32ZVE32F-NEXT:  .LBB43_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB43_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB43_14
 ; RV32ZVE32F-NEXT:  .LBB43_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB43_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB43_15
 ; RV32ZVE32F-NEXT:  .LBB43_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB43_9
-; RV32ZVE32F-NEXT:  .LBB43_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB43_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB43_16
+; RV32ZVE32F-NEXT:  .LBB43_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB43_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB43_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    sw t6, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t6, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(a1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a1)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, t6
+; RV32ZVE32F-NEXT:    lw t6, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    sw a0, 4(a1)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB43_2
-; RV32ZVE32F-NEXT:  .LBB43_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB43_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
-; RV32ZVE32F-NEXT:    sw s0, 0(a0)
-; RV32ZVE32F-NEXT:    sw s1, 4(a0)
+; RV32ZVE32F-NEXT:    sw a2, 0(a0)
+; RV32ZVE32F-NEXT:    sw a3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB43_3
-; RV32ZVE32F-NEXT:  .LBB43_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB43_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3760,7 +3790,7 @@ define void @mscatter_baseidx_sext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB43_4
-; RV32ZVE32F-NEXT:  .LBB43_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB43_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3768,7 +3798,7 @@ define void @mscatter_baseidx_sext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB43_5
-; RV32ZVE32F-NEXT:  .LBB43_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB43_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3776,7 +3806,7 @@ define void @mscatter_baseidx_sext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t2, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB43_6
-; RV32ZVE32F-NEXT:  .LBB43_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB43_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -3784,15 +3814,25 @@ define void @mscatter_baseidx_sext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB43_7
-; RV32ZVE32F-NEXT:  .LBB43_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB43_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB43_8
-; RV32ZVE32F-NEXT:    j .LBB43_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB43_8
+; RV32ZVE32F-NEXT:  .LBB43_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_sext_v8i8_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -3928,19 +3968,15 @@ define void @mscatter_baseidx_zext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vzext.vf4 v10, v8
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -3950,68 +3986,76 @@ define void @mscatter_baseidx_zext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    lw t4, 28(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t2, 36(a0)
-; RV32ZVE32F-NEXT:    lw s0, 8(a0)
-; RV32ZVE32F-NEXT:    lw s1, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 8(a0)
+; RV32ZVE32F-NEXT:    lw a3, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 20(a0)
 ; RV32ZVE32F-NEXT:    vadd.vx v8, v8, a1
 ; RV32ZVE32F-NEXT:    andi a1, a4, 1
-; RV32ZVE32F-NEXT:    bnez a1, .LBB44_10
+; RV32ZVE32F-NEXT:    bnez a1, .LBB44_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB44_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB44_10
 ; RV32ZVE32F-NEXT:  .LBB44_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB44_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB44_11
 ; RV32ZVE32F-NEXT:  .LBB44_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB44_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB44_12
 ; RV32ZVE32F-NEXT:  .LBB44_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB44_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB44_13
 ; RV32ZVE32F-NEXT:  .LBB44_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB44_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB44_14
 ; RV32ZVE32F-NEXT:  .LBB44_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB44_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB44_15
 ; RV32ZVE32F-NEXT:  .LBB44_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB44_9
-; RV32ZVE32F-NEXT:  .LBB44_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB44_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB44_16
+; RV32ZVE32F-NEXT:  .LBB44_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB44_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB44_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    sw t6, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t6, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(a1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a1)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, t6
+; RV32ZVE32F-NEXT:    lw t6, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    sw a0, 4(a1)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB44_2
-; RV32ZVE32F-NEXT:  .LBB44_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB44_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
-; RV32ZVE32F-NEXT:    sw s0, 0(a0)
-; RV32ZVE32F-NEXT:    sw s1, 4(a0)
+; RV32ZVE32F-NEXT:    sw a2, 0(a0)
+; RV32ZVE32F-NEXT:    sw a3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB44_3
-; RV32ZVE32F-NEXT:  .LBB44_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB44_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -4019,7 +4063,7 @@ define void @mscatter_baseidx_zext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB44_4
-; RV32ZVE32F-NEXT:  .LBB44_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB44_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -4027,7 +4071,7 @@ define void @mscatter_baseidx_zext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB44_5
-; RV32ZVE32F-NEXT:  .LBB44_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB44_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -4035,7 +4079,7 @@ define void @mscatter_baseidx_zext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t2, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB44_6
-; RV32ZVE32F-NEXT:  .LBB44_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB44_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -4043,15 +4087,25 @@ define void @mscatter_baseidx_zext_v8i8_v8i64(<8 x i64> %val, ptr %base, <8 x i8
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB44_7
-; RV32ZVE32F-NEXT:  .LBB44_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB44_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB44_8
-; RV32ZVE32F-NEXT:    j .LBB44_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB44_8
+; RV32ZVE32F-NEXT:  .LBB44_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_zext_v8i8_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -4194,17 +4248,13 @@ define void @mscatter_baseidx_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i16> %
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -4215,71 +4265,81 @@ define void @mscatter_baseidx_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i16> %
 ; RV32ZVE32F-NEXT:    lw t3, 28(a0)
 ; RV32ZVE32F-NEXT:    lw a1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 36(a0)
-; RV32ZVE32F-NEXT:    li s1, 8
+; RV32ZVE32F-NEXT:    li a3, 8
 ; RV32ZVE32F-NEXT:    lw t6, 8(a0)
-; RV32ZVE32F-NEXT:    lw s0, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 12(a0)
+; RV32ZVE32F-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw t4, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 20(a0)
 ; RV32ZVE32F-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
-; RV32ZVE32F-NEXT:    vwmaccus.vx v10, s1, v8
-; RV32ZVE32F-NEXT:    andi s1, a4, 1
-; RV32ZVE32F-NEXT:    bnez s1, .LBB45_10
+; RV32ZVE32F-NEXT:    vwmaccus.vx v10, a3, v8
+; RV32ZVE32F-NEXT:    andi a3, a4, 1
+; RV32ZVE32F-NEXT:    bnez a3, .LBB45_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB45_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB45_10
 ; RV32ZVE32F-NEXT:  .LBB45_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB45_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB45_11
 ; RV32ZVE32F-NEXT:  .LBB45_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB45_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB45_12
 ; RV32ZVE32F-NEXT:  .LBB45_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB45_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB45_13
 ; RV32ZVE32F-NEXT:  .LBB45_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB45_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB45_14
 ; RV32ZVE32F-NEXT:  .LBB45_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB45_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB45_15
 ; RV32ZVE32F-NEXT:  .LBB45_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB45_9
-; RV32ZVE32F-NEXT:  .LBB45_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB45_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB45_16
+; RV32ZVE32F-NEXT:  .LBB45_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB45_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB45_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vmv.x.s s1, v10
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s a3, v10
+; RV32ZVE32F-NEXT:    mv a2, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, a1
+; RV32ZVE32F-NEXT:    mv a1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(s1)
-; RV32ZVE32F-NEXT:    sw a0, 4(s1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a3)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, a1
+; RV32ZVE32F-NEXT:    mv a1, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, a2
+; RV32ZVE32F-NEXT:    sw a0, 4(a3)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB45_2
-; RV32ZVE32F-NEXT:  .LBB45_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB45_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 1
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw t6, 0(a0)
-; RV32ZVE32F-NEXT:    sw s0, 4(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s a2, v8
+; RV32ZVE32F-NEXT:    sw t6, 0(a2)
+; RV32ZVE32F-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a2)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB45_3
-; RV32ZVE32F-NEXT:  .LBB45_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB45_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4287,7 +4347,7 @@ define void @mscatter_baseidx_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i16> %
 ; RV32ZVE32F-NEXT:    sw t5, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB45_4
-; RV32ZVE32F-NEXT:  .LBB45_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB45_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4295,7 +4355,7 @@ define void @mscatter_baseidx_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i16> %
 ; RV32ZVE32F-NEXT:    sw t3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB45_5
-; RV32ZVE32F-NEXT:  .LBB45_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB45_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4303,7 +4363,7 @@ define void @mscatter_baseidx_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i16> %
 ; RV32ZVE32F-NEXT:    sw t1, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB45_6
-; RV32ZVE32F-NEXT:  .LBB45_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB45_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4311,15 +4371,25 @@ define void @mscatter_baseidx_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i16> %
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB45_7
-; RV32ZVE32F-NEXT:  .LBB45_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB45_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB45_8
-; RV32ZVE32F-NEXT:    j .LBB45_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB45_8
+; RV32ZVE32F-NEXT:  .LBB45_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_v8i16_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -4454,17 +4524,13 @@ define void @mscatter_baseidx_sext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -4475,71 +4541,81 @@ define void @mscatter_baseidx_sext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    lw t3, 28(a0)
 ; RV32ZVE32F-NEXT:    lw a1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 36(a0)
-; RV32ZVE32F-NEXT:    li s1, 8
+; RV32ZVE32F-NEXT:    li a3, 8
 ; RV32ZVE32F-NEXT:    lw t6, 8(a0)
-; RV32ZVE32F-NEXT:    lw s0, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 12(a0)
+; RV32ZVE32F-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw t4, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 20(a0)
 ; RV32ZVE32F-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
-; RV32ZVE32F-NEXT:    vwmaccus.vx v10, s1, v8
-; RV32ZVE32F-NEXT:    andi s1, a4, 1
-; RV32ZVE32F-NEXT:    bnez s1, .LBB46_10
+; RV32ZVE32F-NEXT:    vwmaccus.vx v10, a3, v8
+; RV32ZVE32F-NEXT:    andi a3, a4, 1
+; RV32ZVE32F-NEXT:    bnez a3, .LBB46_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB46_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB46_10
 ; RV32ZVE32F-NEXT:  .LBB46_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB46_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB46_11
 ; RV32ZVE32F-NEXT:  .LBB46_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB46_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB46_12
 ; RV32ZVE32F-NEXT:  .LBB46_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB46_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB46_13
 ; RV32ZVE32F-NEXT:  .LBB46_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB46_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB46_14
 ; RV32ZVE32F-NEXT:  .LBB46_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB46_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB46_15
 ; RV32ZVE32F-NEXT:  .LBB46_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB46_9
-; RV32ZVE32F-NEXT:  .LBB46_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB46_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB46_16
+; RV32ZVE32F-NEXT:  .LBB46_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB46_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB46_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vmv.x.s s1, v10
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s a3, v10
+; RV32ZVE32F-NEXT:    mv a2, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, a1
+; RV32ZVE32F-NEXT:    mv a1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(s1)
-; RV32ZVE32F-NEXT:    sw a0, 4(s1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a3)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, a1
+; RV32ZVE32F-NEXT:    mv a1, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, a2
+; RV32ZVE32F-NEXT:    sw a0, 4(a3)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB46_2
-; RV32ZVE32F-NEXT:  .LBB46_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB46_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 1
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw t6, 0(a0)
-; RV32ZVE32F-NEXT:    sw s0, 4(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s a2, v8
+; RV32ZVE32F-NEXT:    sw t6, 0(a2)
+; RV32ZVE32F-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a2)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB46_3
-; RV32ZVE32F-NEXT:  .LBB46_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB46_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4547,7 +4623,7 @@ define void @mscatter_baseidx_sext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t5, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB46_4
-; RV32ZVE32F-NEXT:  .LBB46_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB46_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4555,7 +4631,7 @@ define void @mscatter_baseidx_sext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB46_5
-; RV32ZVE32F-NEXT:  .LBB46_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB46_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4563,7 +4639,7 @@ define void @mscatter_baseidx_sext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t1, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB46_6
-; RV32ZVE32F-NEXT:  .LBB46_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB46_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4571,15 +4647,25 @@ define void @mscatter_baseidx_sext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB46_7
-; RV32ZVE32F-NEXT:  .LBB46_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB46_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB46_8
-; RV32ZVE32F-NEXT:    j .LBB46_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB46_8
+; RV32ZVE32F-NEXT:  .LBB46_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_sext_v8i16_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -4716,17 +4802,13 @@ define void @mscatter_baseidx_zext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -4737,71 +4819,81 @@ define void @mscatter_baseidx_zext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    lw t3, 28(a0)
 ; RV32ZVE32F-NEXT:    lw a1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 36(a0)
-; RV32ZVE32F-NEXT:    li s1, 8
+; RV32ZVE32F-NEXT:    li a3, 8
 ; RV32ZVE32F-NEXT:    lw t6, 8(a0)
-; RV32ZVE32F-NEXT:    lw s0, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 12(a0)
+; RV32ZVE32F-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw t4, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 20(a0)
 ; RV32ZVE32F-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
-; RV32ZVE32F-NEXT:    vwmaccu.vx v10, s1, v8
-; RV32ZVE32F-NEXT:    andi s1, a4, 1
-; RV32ZVE32F-NEXT:    bnez s1, .LBB47_10
+; RV32ZVE32F-NEXT:    vwmaccu.vx v10, a3, v8
+; RV32ZVE32F-NEXT:    andi a3, a4, 1
+; RV32ZVE32F-NEXT:    bnez a3, .LBB47_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB47_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB47_10
 ; RV32ZVE32F-NEXT:  .LBB47_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB47_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB47_11
 ; RV32ZVE32F-NEXT:  .LBB47_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB47_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB47_12
 ; RV32ZVE32F-NEXT:  .LBB47_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB47_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB47_13
 ; RV32ZVE32F-NEXT:  .LBB47_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB47_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB47_14
 ; RV32ZVE32F-NEXT:  .LBB47_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB47_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB47_15
 ; RV32ZVE32F-NEXT:  .LBB47_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB47_9
-; RV32ZVE32F-NEXT:  .LBB47_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB47_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB47_16
+; RV32ZVE32F-NEXT:  .LBB47_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB47_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB47_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vmv.x.s s1, v10
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s a3, v10
+; RV32ZVE32F-NEXT:    mv a2, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, a1
+; RV32ZVE32F-NEXT:    mv a1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(s1)
-; RV32ZVE32F-NEXT:    sw a0, 4(s1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a3)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, a1
+; RV32ZVE32F-NEXT:    mv a1, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, a2
+; RV32ZVE32F-NEXT:    sw a0, 4(a3)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB47_2
-; RV32ZVE32F-NEXT:  .LBB47_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB47_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 1
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw t6, 0(a0)
-; RV32ZVE32F-NEXT:    sw s0, 4(a0)
+; RV32ZVE32F-NEXT:    vmv.x.s a2, v8
+; RV32ZVE32F-NEXT:    sw t6, 0(a2)
+; RV32ZVE32F-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a2)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB47_3
-; RV32ZVE32F-NEXT:  .LBB47_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB47_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4809,7 +4901,7 @@ define void @mscatter_baseidx_zext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t5, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB47_4
-; RV32ZVE32F-NEXT:  .LBB47_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB47_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4817,7 +4909,7 @@ define void @mscatter_baseidx_zext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB47_5
-; RV32ZVE32F-NEXT:  .LBB47_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB47_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4825,7 +4917,7 @@ define void @mscatter_baseidx_zext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t1, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB47_6
-; RV32ZVE32F-NEXT:  .LBB47_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB47_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
@@ -4833,15 +4925,25 @@ define void @mscatter_baseidx_zext_v8i16_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB47_7
-; RV32ZVE32F-NEXT:  .LBB47_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB47_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB47_8
-; RV32ZVE32F-NEXT:    j .LBB47_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB47_8
+; RV32ZVE32F-NEXT:  .LBB47_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v10, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_zext_v8i16_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -4985,17 +5087,13 @@ define void @mscatter_baseidx_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i32> %
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -5006,68 +5104,76 @@ define void @mscatter_baseidx_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i32> %
 ; RV32ZVE32F-NEXT:    lw t4, 28(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t2, 36(a0)
-; RV32ZVE32F-NEXT:    lw s0, 8(a0)
-; RV32ZVE32F-NEXT:    lw s1, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 8(a0)
+; RV32ZVE32F-NEXT:    lw a3, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 20(a0)
 ; RV32ZVE32F-NEXT:    vadd.vx v8, v8, a1
 ; RV32ZVE32F-NEXT:    andi a1, a4, 1
-; RV32ZVE32F-NEXT:    bnez a1, .LBB48_10
+; RV32ZVE32F-NEXT:    bnez a1, .LBB48_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB48_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB48_10
 ; RV32ZVE32F-NEXT:  .LBB48_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB48_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB48_11
 ; RV32ZVE32F-NEXT:  .LBB48_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB48_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB48_12
 ; RV32ZVE32F-NEXT:  .LBB48_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB48_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB48_13
 ; RV32ZVE32F-NEXT:  .LBB48_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB48_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB48_14
 ; RV32ZVE32F-NEXT:  .LBB48_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB48_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB48_15
 ; RV32ZVE32F-NEXT:  .LBB48_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB48_9
-; RV32ZVE32F-NEXT:  .LBB48_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB48_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB48_16
+; RV32ZVE32F-NEXT:  .LBB48_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB48_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB48_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    sw t6, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t6, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(a1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a1)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, t6
+; RV32ZVE32F-NEXT:    lw t6, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    sw a0, 4(a1)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB48_2
-; RV32ZVE32F-NEXT:  .LBB48_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB48_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
-; RV32ZVE32F-NEXT:    sw s0, 0(a0)
-; RV32ZVE32F-NEXT:    sw s1, 4(a0)
+; RV32ZVE32F-NEXT:    sw a2, 0(a0)
+; RV32ZVE32F-NEXT:    sw a3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB48_3
-; RV32ZVE32F-NEXT:  .LBB48_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB48_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5075,7 +5181,7 @@ define void @mscatter_baseidx_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i32> %
 ; RV32ZVE32F-NEXT:    sw t6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB48_4
-; RV32ZVE32F-NEXT:  .LBB48_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB48_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5083,7 +5189,7 @@ define void @mscatter_baseidx_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i32> %
 ; RV32ZVE32F-NEXT:    sw t4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB48_5
-; RV32ZVE32F-NEXT:  .LBB48_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB48_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5091,7 +5197,7 @@ define void @mscatter_baseidx_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i32> %
 ; RV32ZVE32F-NEXT:    sw t2, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB48_6
-; RV32ZVE32F-NEXT:  .LBB48_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB48_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5099,15 +5205,25 @@ define void @mscatter_baseidx_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i32> %
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB48_7
-; RV32ZVE32F-NEXT:  .LBB48_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB48_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB48_8
-; RV32ZVE32F-NEXT:    j .LBB48_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB48_8
+; RV32ZVE32F-NEXT:  .LBB48_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_v8i32_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -5243,17 +5359,13 @@ define void @mscatter_baseidx_sext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -5264,68 +5376,76 @@ define void @mscatter_baseidx_sext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    lw t4, 28(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t2, 36(a0)
-; RV32ZVE32F-NEXT:    lw s0, 8(a0)
-; RV32ZVE32F-NEXT:    lw s1, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 8(a0)
+; RV32ZVE32F-NEXT:    lw a3, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 20(a0)
 ; RV32ZVE32F-NEXT:    vadd.vx v8, v8, a1
 ; RV32ZVE32F-NEXT:    andi a1, a4, 1
-; RV32ZVE32F-NEXT:    bnez a1, .LBB49_10
+; RV32ZVE32F-NEXT:    bnez a1, .LBB49_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB49_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB49_10
 ; RV32ZVE32F-NEXT:  .LBB49_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB49_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB49_11
 ; RV32ZVE32F-NEXT:  .LBB49_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB49_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB49_12
 ; RV32ZVE32F-NEXT:  .LBB49_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB49_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB49_13
 ; RV32ZVE32F-NEXT:  .LBB49_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB49_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB49_14
 ; RV32ZVE32F-NEXT:  .LBB49_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB49_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB49_15
 ; RV32ZVE32F-NEXT:  .LBB49_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB49_9
-; RV32ZVE32F-NEXT:  .LBB49_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB49_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB49_16
+; RV32ZVE32F-NEXT:  .LBB49_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB49_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB49_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    sw t6, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t6, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(a1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a1)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, t6
+; RV32ZVE32F-NEXT:    lw t6, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    sw a0, 4(a1)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB49_2
-; RV32ZVE32F-NEXT:  .LBB49_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB49_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
-; RV32ZVE32F-NEXT:    sw s0, 0(a0)
-; RV32ZVE32F-NEXT:    sw s1, 4(a0)
+; RV32ZVE32F-NEXT:    sw a2, 0(a0)
+; RV32ZVE32F-NEXT:    sw a3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB49_3
-; RV32ZVE32F-NEXT:  .LBB49_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB49_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5333,7 +5453,7 @@ define void @mscatter_baseidx_sext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB49_4
-; RV32ZVE32F-NEXT:  .LBB49_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB49_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5341,7 +5461,7 @@ define void @mscatter_baseidx_sext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB49_5
-; RV32ZVE32F-NEXT:  .LBB49_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB49_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5349,7 +5469,7 @@ define void @mscatter_baseidx_sext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t2, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB49_6
-; RV32ZVE32F-NEXT:  .LBB49_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB49_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5357,15 +5477,25 @@ define void @mscatter_baseidx_sext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB49_7
-; RV32ZVE32F-NEXT:  .LBB49_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB49_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB49_8
-; RV32ZVE32F-NEXT:    j .LBB49_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB49_8
+; RV32ZVE32F-NEXT:  .LBB49_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_sext_v8i32_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -5502,17 +5632,13 @@ define void @mscatter_baseidx_zext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a4, v0
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a3, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -5523,68 +5649,76 @@ define void @mscatter_baseidx_zext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    lw t4, 28(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t2, 36(a0)
-; RV32ZVE32F-NEXT:    lw s0, 8(a0)
-; RV32ZVE32F-NEXT:    lw s1, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 8(a0)
+; RV32ZVE32F-NEXT:    lw a3, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 20(a0)
 ; RV32ZVE32F-NEXT:    vadd.vx v8, v8, a1
 ; RV32ZVE32F-NEXT:    andi a1, a4, 1
-; RV32ZVE32F-NEXT:    bnez a1, .LBB50_10
+; RV32ZVE32F-NEXT:    bnez a1, .LBB50_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB50_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB50_10
 ; RV32ZVE32F-NEXT:  .LBB50_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB50_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB50_11
 ; RV32ZVE32F-NEXT:  .LBB50_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB50_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB50_12
 ; RV32ZVE32F-NEXT:  .LBB50_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB50_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB50_13
 ; RV32ZVE32F-NEXT:  .LBB50_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB50_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB50_14
 ; RV32ZVE32F-NEXT:  .LBB50_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB50_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB50_15
 ; RV32ZVE32F-NEXT:  .LBB50_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB50_9
-; RV32ZVE32F-NEXT:  .LBB50_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a3, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB50_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB50_16
+; RV32ZVE32F-NEXT:  .LBB50_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB50_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB50_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    sw t6, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t6, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(a1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a1)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, t6
+; RV32ZVE32F-NEXT:    lw t6, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    sw a0, 4(a1)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB50_2
-; RV32ZVE32F-NEXT:  .LBB50_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB50_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
-; RV32ZVE32F-NEXT:    sw s0, 0(a0)
-; RV32ZVE32F-NEXT:    sw s1, 4(a0)
+; RV32ZVE32F-NEXT:    sw a2, 0(a0)
+; RV32ZVE32F-NEXT:    sw a3, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB50_3
-; RV32ZVE32F-NEXT:  .LBB50_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB50_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5592,7 +5726,7 @@ define void @mscatter_baseidx_zext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB50_4
-; RV32ZVE32F-NEXT:  .LBB50_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB50_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5600,7 +5734,7 @@ define void @mscatter_baseidx_zext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB50_5
-; RV32ZVE32F-NEXT:  .LBB50_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB50_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5608,7 +5742,7 @@ define void @mscatter_baseidx_zext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t2, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB50_6
-; RV32ZVE32F-NEXT:  .LBB50_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB50_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5616,15 +5750,25 @@ define void @mscatter_baseidx_zext_v8i32_v8i64(<8 x i64> %val, ptr %base, <8 x i
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB50_7
-; RV32ZVE32F-NEXT:  .LBB50_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB50_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a4, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB50_8
-; RV32ZVE32F-NEXT:    j .LBB50_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB50_8
+; RV32ZVE32F-NEXT:  .LBB50_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_zext_v8i32_v8i64:
 ; RV64ZVE32F:       # %bb.0:
@@ -5768,12 +5912,6 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV32ZVE32F:       # %bb.0:
 ; RV32ZVE32F-NEXT:    addi sp, sp, -16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 16
-; RV32ZVE32F-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
-; RV32ZVE32F-NEXT:    .cfi_offset s0, -4
-; RV32ZVE32F-NEXT:    .cfi_offset s1, -8
-; RV32ZVE32F-NEXT:    .cfi_offset s2, -12
 ; RV32ZVE32F-NEXT:    .cfi_remember_state
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vmv.x.s a3, v0
@@ -5795,7 +5933,9 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV32ZVE32F-NEXT:    vslide1down.vx v8, v8, a6
 ; RV32ZVE32F-NEXT:    vslide1down.vx v8, v8, a2
 ; RV32ZVE32F-NEXT:    lw a2, 56(a0)
-; RV32ZVE32F-NEXT:    lw a4, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 12(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    lw a2, 60(a0)
+; RV32ZVE32F-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32ZVE32F-NEXT:    lw a7, 40(a0)
 ; RV32ZVE32F-NEXT:    lw t0, 44(a0)
 ; RV32ZVE32F-NEXT:    lw a5, 48(a0)
@@ -5805,68 +5945,76 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV32ZVE32F-NEXT:    lw t4, 28(a0)
 ; RV32ZVE32F-NEXT:    lw t1, 32(a0)
 ; RV32ZVE32F-NEXT:    lw t2, 36(a0)
-; RV32ZVE32F-NEXT:    lw s0, 8(a0)
-; RV32ZVE32F-NEXT:    lw s1, 12(a0)
+; RV32ZVE32F-NEXT:    lw a2, 8(a0)
+; RV32ZVE32F-NEXT:    lw a4, 12(a0)
 ; RV32ZVE32F-NEXT:    lw t5, 16(a0)
 ; RV32ZVE32F-NEXT:    lw t6, 20(a0)
 ; RV32ZVE32F-NEXT:    vadd.vx v8, v8, a1
 ; RV32ZVE32F-NEXT:    andi a1, a3, 1
-; RV32ZVE32F-NEXT:    bnez a1, .LBB51_10
+; RV32ZVE32F-NEXT:    bnez a1, .LBB51_9
 ; RV32ZVE32F-NEXT:  # %bb.1: # %else
 ; RV32ZVE32F-NEXT:    andi a0, a3, 2
-; RV32ZVE32F-NEXT:    bnez a0, .LBB51_11
+; RV32ZVE32F-NEXT:    bnez a0, .LBB51_10
 ; RV32ZVE32F-NEXT:  .LBB51_2: # %else2
 ; RV32ZVE32F-NEXT:    andi a0, a3, 4
-; RV32ZVE32F-NEXT:    bnez a0, .LBB51_12
+; RV32ZVE32F-NEXT:    bnez a0, .LBB51_11
 ; RV32ZVE32F-NEXT:  .LBB51_3: # %else4
 ; RV32ZVE32F-NEXT:    andi a0, a3, 8
-; RV32ZVE32F-NEXT:    bnez a0, .LBB51_13
+; RV32ZVE32F-NEXT:    bnez a0, .LBB51_12
 ; RV32ZVE32F-NEXT:  .LBB51_4: # %else6
 ; RV32ZVE32F-NEXT:    andi a0, a3, 16
-; RV32ZVE32F-NEXT:    bnez a0, .LBB51_14
+; RV32ZVE32F-NEXT:    bnez a0, .LBB51_13
 ; RV32ZVE32F-NEXT:  .LBB51_5: # %else8
 ; RV32ZVE32F-NEXT:    andi a0, a3, 32
-; RV32ZVE32F-NEXT:    bnez a0, .LBB51_15
+; RV32ZVE32F-NEXT:    bnez a0, .LBB51_14
 ; RV32ZVE32F-NEXT:  .LBB51_6: # %else10
 ; RV32ZVE32F-NEXT:    andi a0, a3, 64
-; RV32ZVE32F-NEXT:    bnez a0, .LBB51_16
+; RV32ZVE32F-NEXT:    bnez a0, .LBB51_15
 ; RV32ZVE32F-NEXT:  .LBB51_7: # %else12
 ; RV32ZVE32F-NEXT:    andi a0, a3, -128
-; RV32ZVE32F-NEXT:    beqz a0, .LBB51_9
-; RV32ZVE32F-NEXT:  .LBB51_8: # %cond.store13
-; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
-; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
-; RV32ZVE32F-NEXT:    vmv.x.s a0, v8
-; RV32ZVE32F-NEXT:    sw a2, 0(a0)
-; RV32ZVE32F-NEXT:    sw a4, 4(a0)
-; RV32ZVE32F-NEXT:  .LBB51_9: # %else14
-; RV32ZVE32F-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
-; RV32ZVE32F-NEXT:    .cfi_restore s0
-; RV32ZVE32F-NEXT:    .cfi_restore s1
-; RV32ZVE32F-NEXT:    .cfi_restore s2
+; RV32ZVE32F-NEXT:    bnez a0, .LBB51_16
+; RV32ZVE32F-NEXT:  .LBB51_8: # %else14
 ; RV32ZVE32F-NEXT:    addi sp, sp, 16
 ; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVE32F-NEXT:    ret
-; RV32ZVE32F-NEXT:  .LBB51_10: # %cond.store
+; RV32ZVE32F-NEXT:  .LBB51_9: # %cond.store
 ; RV32ZVE32F-NEXT:    .cfi_restore_state
 ; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
-; RV32ZVE32F-NEXT:    lw s2, 0(a0)
+; RV32ZVE32F-NEXT:    sw t6, 4(sp) # 4-byte Folded Spill
+; RV32ZVE32F-NEXT:    mv t6, t5
+; RV32ZVE32F-NEXT:    mv t5, t4
+; RV32ZVE32F-NEXT:    mv t4, t3
+; RV32ZVE32F-NEXT:    mv t3, t2
+; RV32ZVE32F-NEXT:    mv t2, t1
+; RV32ZVE32F-NEXT:    mv t1, t0
+; RV32ZVE32F-NEXT:    mv t0, a7
+; RV32ZVE32F-NEXT:    mv a7, a6
+; RV32ZVE32F-NEXT:    mv a6, a5
+; RV32ZVE32F-NEXT:    lw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    lw a0, 4(a0)
-; RV32ZVE32F-NEXT:    sw s2, 0(a1)
+; RV32ZVE32F-NEXT:    sw a5, 0(a1)
+; RV32ZVE32F-NEXT:    mv a5, a6
+; RV32ZVE32F-NEXT:    mv a6, a7
+; RV32ZVE32F-NEXT:    mv a7, t0
+; RV32ZVE32F-NEXT:    mv t0, t1
+; RV32ZVE32F-NEXT:    mv t1, t2
+; RV32ZVE32F-NEXT:    mv t2, t3
+; RV32ZVE32F-NEXT:    mv t3, t4
+; RV32ZVE32F-NEXT:    mv t4, t5
+; RV32ZVE32F-NEXT:    mv t5, t6
+; RV32ZVE32F-NEXT:    lw t6, 4(sp) # 4-byte Folded Reload
 ; RV32ZVE32F-NEXT:    sw a0, 4(a1)
 ; RV32ZVE32F-NEXT:    andi a0, a3, 2
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB51_2
-; RV32ZVE32F-NEXT:  .LBB51_11: # %cond.store1
+; RV32ZVE32F-NEXT:  .LBB51_10: # %cond.store1
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 1
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
-; RV32ZVE32F-NEXT:    sw s0, 0(a0)
-; RV32ZVE32F-NEXT:    sw s1, 4(a0)
+; RV32ZVE32F-NEXT:    sw a2, 0(a0)
+; RV32ZVE32F-NEXT:    sw a4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a3, 4
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB51_3
-; RV32ZVE32F-NEXT:  .LBB51_12: # %cond.store3
+; RV32ZVE32F-NEXT:  .LBB51_11: # %cond.store3
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 2
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5874,7 +6022,7 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV32ZVE32F-NEXT:    sw t6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a3, 8
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB51_4
-; RV32ZVE32F-NEXT:  .LBB51_13: # %cond.store5
+; RV32ZVE32F-NEXT:  .LBB51_12: # %cond.store5
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 3
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5882,7 +6030,7 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV32ZVE32F-NEXT:    sw t4, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a3, 16
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB51_5
-; RV32ZVE32F-NEXT:  .LBB51_14: # %cond.store7
+; RV32ZVE32F-NEXT:  .LBB51_13: # %cond.store7
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 4
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5890,7 +6038,7 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV32ZVE32F-NEXT:    sw t2, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a3, 32
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB51_6
-; RV32ZVE32F-NEXT:  .LBB51_15: # %cond.store9
+; RV32ZVE32F-NEXT:  .LBB51_14: # %cond.store9
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 5
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
@@ -5898,47 +6046,54 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV32ZVE32F-NEXT:    sw t0, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a3, 64
 ; RV32ZVE32F-NEXT:    beqz a0, .LBB51_7
-; RV32ZVE32F-NEXT:  .LBB51_16: # %cond.store11
+; RV32ZVE32F-NEXT:  .LBB51_15: # %cond.store11
 ; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
 ; RV32ZVE32F-NEXT:    vslidedown.vi v10, v8, 6
 ; RV32ZVE32F-NEXT:    vmv.x.s a0, v10
 ; RV32ZVE32F-NEXT:    sw a5, 0(a0)
 ; RV32ZVE32F-NEXT:    sw a6, 4(a0)
 ; RV32ZVE32F-NEXT:    andi a0, a3, -128
-; RV32ZVE32F-NEXT:    bnez a0, .LBB51_8
-; RV32ZVE32F-NEXT:    j .LBB51_9
+; RV32ZVE32F-NEXT:    beqz a0, .LBB51_8
+; RV32ZVE32F-NEXT:  .LBB51_16: # %cond.store13
+; RV32ZVE32F-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32ZVE32F-NEXT:    vslidedown.vi v8, v8, 7
+; RV32ZVE32F-NEXT:    vmv.x.s a1, v8
+; RV32ZVE32F-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 0(a1)
+; RV32ZVE32F-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32ZVE32F-NEXT:    sw a0, 4(a1)
+; RV32ZVE32F-NEXT:    addi sp, sp, 16
+; RV32ZVE32F-NEXT:    .cfi_def_cfa_offset 0
+; RV32ZVE32F-NEXT:    ret
 ;
 ; RV64ZVE32F-LABEL: mscatter_baseidx_v8i64:
 ; RV64ZVE32F:       # %bb.0:
 ; RV64ZVE32F-NEXT:    addi sp, sp, -32
 ; RV64ZVE32F-NEXT:    .cfi_def_cfa_offset 32
 ; RV64ZVE32F-NEXT:    sd s0, 24(sp) # 8-byte Folded Spill
-; RV64ZVE32F-NEXT:    sd s1, 16(sp) # 8-byte Folded Spill
-; RV64ZVE32F-NEXT:    sd s2, 8(sp) # 8-byte Folded Spill
-; RV64ZVE32F-NEXT:    sd s3, 0(sp) # 8-byte Folded Spill
 ; RV64ZVE32F-NEXT:    .cfi_offset s0, -8
-; RV64ZVE32F-NEXT:    .cfi_offset s1, -16
-; RV64ZVE32F-NEXT:    .cfi_offset s2, -24
-; RV64ZVE32F-NEXT:    .cfi_offset s3, -32
 ; RV64ZVE32F-NEXT:    .cfi_remember_state
 ; RV64ZVE32F-NEXT:    ld t3, 8(a2)
-; RV64ZVE32F-NEXT:    ld a5, 40(a0)
-; RV64ZVE32F-NEXT:    ld a4, 48(a0)
+; RV64ZVE32F-NEXT:    ld a3, 40(a0)
+; RV64ZVE32F-NEXT:    sd a3, 0(sp) # 8-byte Folded Spill
+; RV64ZVE32F-NEXT:    ld a3, 48(a0)
+; RV64ZVE32F-NEXT:    sd a3, 8(sp) # 8-byte Folded Spill
 ; RV64ZVE32F-NEXT:    ld a3, 56(a0)
+; RV64ZVE32F-NEXT:    sd a3, 16(sp) # 8-byte Folded Spill
 ; RV64ZVE32F-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; RV64ZVE32F-NEXT:    vmv.x.s a6, v0
-; RV64ZVE32F-NEXT:    ld s2, 8(a0)
-; RV64ZVE32F-NEXT:    ld s0, 16(a0)
+; RV64ZVE32F-NEXT:    ld a5, 8(a0)
+; RV64ZVE32F-NEXT:    ld a3, 16(a0)
 ; RV64ZVE32F-NEXT:    ld t5, 24(a0)
 ; RV64ZVE32F-NEXT:    ld t2, 32(a0)
-; RV64ZVE32F-NEXT:    ld s1, 16(a2)
+; RV64ZVE32F-NEXT:    ld a4, 16(a2)
 ; RV64ZVE32F-NEXT:    ld t6, 24(a2)
 ; RV64ZVE32F-NEXT:    ld t4, 32(a2)
 ; RV64ZVE32F-NEXT:    ld t1, 40(a2)
 ; RV64ZVE32F-NEXT:    ld t0, 48(a2)
 ; RV64ZVE32F-NEXT:    ld a7, 56(a2)
-; RV64ZVE32F-NEXT:    andi s3, a6, 1
-; RV64ZVE32F-NEXT:    bnez s3, .LBB51_10
+; RV64ZVE32F-NEXT:    andi s0, a6, 1
+; RV64ZVE32F-NEXT:    bnez s0, .LBB51_10
 ; RV64ZVE32F-NEXT:  # %bb.1: # %else
 ; RV64ZVE32F-NEXT:    andi a0, a6, 2
 ; RV64ZVE32F-NEXT:    bnez a0, .LBB51_11
@@ -5963,16 +6118,11 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV64ZVE32F-NEXT:  .LBB51_8: # %cond.store13
 ; RV64ZVE32F-NEXT:    slli a7, a7, 3
 ; RV64ZVE32F-NEXT:    add a1, a1, a7
-; RV64ZVE32F-NEXT:    sd a3, 0(a1)
+; RV64ZVE32F-NEXT:    ld a0, 16(sp) # 8-byte Folded Reload
+; RV64ZVE32F-NEXT:    sd a0, 0(a1)
 ; RV64ZVE32F-NEXT:  .LBB51_9: # %else14
 ; RV64ZVE32F-NEXT:    ld s0, 24(sp) # 8-byte Folded Reload
-; RV64ZVE32F-NEXT:    ld s1, 16(sp) # 8-byte Folded Reload
-; RV64ZVE32F-NEXT:    ld s2, 8(sp) # 8-byte Folded Reload
-; RV64ZVE32F-NEXT:    ld s3, 0(sp) # 8-byte Folded Reload
 ; RV64ZVE32F-NEXT:    .cfi_restore s0
-; RV64ZVE32F-NEXT:    .cfi_restore s1
-; RV64ZVE32F-NEXT:    .cfi_restore s2
-; RV64ZVE32F-NEXT:    .cfi_restore s3
 ; RV64ZVE32F-NEXT:    addi sp, sp, 32
 ; RV64ZVE32F-NEXT:    .cfi_def_cfa_offset 0
 ; RV64ZVE32F-NEXT:    ret
@@ -5988,13 +6138,13 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV64ZVE32F-NEXT:  .LBB51_11: # %cond.store1
 ; RV64ZVE32F-NEXT:    slli t3, t3, 3
 ; RV64ZVE32F-NEXT:    add t3, a1, t3
-; RV64ZVE32F-NEXT:    sd s2, 0(t3)
+; RV64ZVE32F-NEXT:    sd a5, 0(t3)
 ; RV64ZVE32F-NEXT:    andi a0, a6, 4
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB51_3
 ; RV64ZVE32F-NEXT:  .LBB51_12: # %cond.store3
-; RV64ZVE32F-NEXT:    slli s1, s1, 3
-; RV64ZVE32F-NEXT:    add s1, a1, s1
-; RV64ZVE32F-NEXT:    sd s0, 0(s1)
+; RV64ZVE32F-NEXT:    slli a4, a4, 3
+; RV64ZVE32F-NEXT:    add a4, a1, a4
+; RV64ZVE32F-NEXT:    sd a3, 0(a4)
 ; RV64ZVE32F-NEXT:    andi a0, a6, 8
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB51_4
 ; RV64ZVE32F-NEXT:  .LBB51_13: # %cond.store5
@@ -6012,13 +6162,15 @@ define void @mscatter_baseidx_v8i64(<8 x i64> %val, ptr %base, <8 x i64> %idxs,
 ; RV64ZVE32F-NEXT:  .LBB51_15: # %cond.store9
 ; RV64ZVE32F-NEXT:    slli t1, t1, 3
 ; RV64ZVE32F-NEXT:    add t1, a1, t1
-; RV64ZVE32F-NEXT:    sd a5, 0(t1)
+; RV64ZVE32F-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64ZVE32F-NEXT:    sd a0, 0(t1)
 ; RV64ZVE32F-NEXT:    andi a0, a6, 64
 ; RV64ZVE32F-NEXT:    beqz a0, .LBB51_7
 ; RV64ZVE32F-NEXT:  .LBB51_16: # %cond.store11
 ; RV64ZVE32F-NEXT:    slli t0, t0, 3
 ; RV64ZVE32F-NEXT:    add t0, a1, t0
-; RV64ZVE32F-NEXT:    sd a4, 0(t0)
+; RV64ZVE32F-NEXT:    ld a0, 8(sp) # 8-byte Folded Reload
+; RV64ZVE32F-NEXT:    sd a0, 0(t0)
 ; RV64ZVE32F-NEXT:    andi a0, a6, -128
 ; RV64ZVE32F-NEXT:    bnez a0, .LBB51_8
 ; RV64ZVE32F-NEXT:    j .LBB51_9
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-shuffle-rotate.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-shuffle-rotate.ll
index 9e949c1400cd9f..b855221ca8fa84 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-shuffle-rotate.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-shuffle-rotate.ll
@@ -935,16 +935,13 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    sw s0, 120(sp) # 4-byte Folded Spill
 ; RV32ZVKB-ZVE32X-NEXT:    sw s1, 116(sp) # 4-byte Folded Spill
 ; RV32ZVKB-ZVE32X-NEXT:    sw s2, 112(sp) # 4-byte Folded Spill
-; RV32ZVKB-ZVE32X-NEXT:    sw s3, 108(sp) # 4-byte Folded Spill
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset ra, -4
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s0, -8
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s1, -12
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s2, -16
-; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s3, -20
 ; RV32ZVKB-ZVE32X-NEXT:    addi s0, sp, 128
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_def_cfa s0, 0
 ; RV32ZVKB-ZVE32X-NEXT:    andi sp, sp, -64
-; RV32ZVKB-ZVE32X-NEXT:    mv a2, sp
 ; RV32ZVKB-ZVE32X-NEXT:    lw a3, 0(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw a4, 4(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw a5, 8(a1)
@@ -957,9 +954,9 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    lw t4, 36(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw t5, 40(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw t6, 44(a1)
-; RV32ZVKB-ZVE32X-NEXT:    lw s1, 16(a1)
-; RV32ZVKB-ZVE32X-NEXT:    lw s2, 20(a1)
-; RV32ZVKB-ZVE32X-NEXT:    lw s3, 24(a1)
+; RV32ZVKB-ZVE32X-NEXT:    lw a2, 16(a1)
+; RV32ZVKB-ZVE32X-NEXT:    lw s1, 20(a1)
+; RV32ZVKB-ZVE32X-NEXT:    lw s2, 24(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw a1, 28(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t1, 48(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t2, 52(sp)
@@ -969,16 +966,17 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    sw t6, 36(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t3, 40(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t4, 44(sp)
-; RV32ZVKB-ZVE32X-NEXT:    sw s3, 16(sp)
+; RV32ZVKB-ZVE32X-NEXT:    sw s2, 16(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a1, 20(sp)
-; RV32ZVKB-ZVE32X-NEXT:    sw s1, 24(sp)
-; RV32ZVKB-ZVE32X-NEXT:    sw s2, 28(sp)
+; RV32ZVKB-ZVE32X-NEXT:    sw a2, 24(sp)
+; RV32ZVKB-ZVE32X-NEXT:    sw s1, 28(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a5, 0(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a6, 4(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a3, 8(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a4, 12(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    vsetivli zero, 16, e32, m8, ta, ma
-; RV32ZVKB-ZVE32X-NEXT:    vle32.v v8, (a2)
+; RV32ZVKB-ZVE32X-NEXT:    mv a1, sp
+; RV32ZVKB-ZVE32X-NEXT:    vle32.v v8, (a1)
 ; RV32ZVKB-ZVE32X-NEXT:    vse32.v v8, (a0)
 ; RV32ZVKB-ZVE32X-NEXT:    addi sp, s0, -128
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_def_cfa sp, 128
@@ -986,12 +984,10 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    lw s0, 120(sp) # 4-byte Folded Reload
 ; RV32ZVKB-ZVE32X-NEXT:    lw s1, 116(sp) # 4-byte Folded Reload
 ; RV32ZVKB-ZVE32X-NEXT:    lw s2, 112(sp) # 4-byte Folded Reload
-; RV32ZVKB-ZVE32X-NEXT:    lw s3, 108(sp) # 4-byte Folded Reload
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore ra
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s0
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s1
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s2
-; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s3
 ; RV32ZVKB-ZVE32X-NEXT:    addi sp, sp, 128
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVKB-ZVE32X-NEXT:    ret
@@ -1004,16 +1000,13 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    sd s0, 112(sp) # 8-byte Folded Spill
 ; RV64ZVKB-ZVE32X-NEXT:    sd s1, 104(sp) # 8-byte Folded Spill
 ; RV64ZVKB-ZVE32X-NEXT:    sd s2, 96(sp) # 8-byte Folded Spill
-; RV64ZVKB-ZVE32X-NEXT:    sd s3, 88(sp) # 8-byte Folded Spill
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset ra, -8
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s0, -16
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s1, -24
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s2, -32
-; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s3, -40
 ; RV64ZVKB-ZVE32X-NEXT:    addi s0, sp, 128
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_def_cfa s0, 0
 ; RV64ZVKB-ZVE32X-NEXT:    andi sp, sp, -64
-; RV64ZVKB-ZVE32X-NEXT:    mv a2, sp
 ; RV64ZVKB-ZVE32X-NEXT:    ld a3, 0(a1)
 ; RV64ZVKB-ZVE32X-NEXT:    ld a4, 8(a1)
 ; RV64ZVKB-ZVE32X-NEXT:    ld a5, 16(a1)
@@ -1027,17 +1020,17 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    srli t4, a6, 32
 ; RV64ZVKB-ZVE32X-NEXT:    srli t5, a5, 32
 ; RV64ZVKB-ZVE32X-NEXT:    srli t6, t0, 32
-; RV64ZVKB-ZVE32X-NEXT:    srli s1, a7, 32
-; RV64ZVKB-ZVE32X-NEXT:    srli s2, a1, 32
-; RV64ZVKB-ZVE32X-NEXT:    srli s3, t1, 32
+; RV64ZVKB-ZVE32X-NEXT:    srli a2, a7, 32
+; RV64ZVKB-ZVE32X-NEXT:    srli s1, a1, 32
+; RV64ZVKB-ZVE32X-NEXT:    srli s2, t1, 32
 ; RV64ZVKB-ZVE32X-NEXT:    sw a1, 48(sp)
-; RV64ZVKB-ZVE32X-NEXT:    sw s2, 52(sp)
+; RV64ZVKB-ZVE32X-NEXT:    sw s1, 52(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t1, 56(sp)
-; RV64ZVKB-ZVE32X-NEXT:    sw s3, 60(sp)
+; RV64ZVKB-ZVE32X-NEXT:    sw s2, 60(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t0, 32(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t6, 36(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw a7, 40(sp)
-; RV64ZVKB-ZVE32X-NEXT:    sw s1, 44(sp)
+; RV64ZVKB-ZVE32X-NEXT:    sw a2, 44(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw a6, 16(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t4, 20(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw a5, 24(sp)
@@ -1047,7 +1040,8 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    sw a3, 8(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t3, 12(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    vsetivli zero, 16, e32, m8, ta, ma
-; RV64ZVKB-ZVE32X-NEXT:    vle32.v v8, (a2)
+; RV64ZVKB-ZVE32X-NEXT:    mv a1, sp
+; RV64ZVKB-ZVE32X-NEXT:    vle32.v v8, (a1)
 ; RV64ZVKB-ZVE32X-NEXT:    vse32.v v8, (a0)
 ; RV64ZVKB-ZVE32X-NEXT:    addi sp, s0, -128
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_def_cfa sp, 128
@@ -1055,12 +1049,10 @@ define <8 x i64> @shuffle_v8i64_as_i128(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    ld s0, 112(sp) # 8-byte Folded Reload
 ; RV64ZVKB-ZVE32X-NEXT:    ld s1, 104(sp) # 8-byte Folded Reload
 ; RV64ZVKB-ZVE32X-NEXT:    ld s2, 96(sp) # 8-byte Folded Reload
-; RV64ZVKB-ZVE32X-NEXT:    ld s3, 88(sp) # 8-byte Folded Reload
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore ra
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s0
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s1
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s2
-; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s3
 ; RV64ZVKB-ZVE32X-NEXT:    addi sp, sp, 128
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_def_cfa_offset 0
 ; RV64ZVKB-ZVE32X-NEXT:    ret
@@ -1220,16 +1212,13 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    sw s0, 120(sp) # 4-byte Folded Spill
 ; RV32ZVKB-ZVE32X-NEXT:    sw s1, 116(sp) # 4-byte Folded Spill
 ; RV32ZVKB-ZVE32X-NEXT:    sw s2, 112(sp) # 4-byte Folded Spill
-; RV32ZVKB-ZVE32X-NEXT:    sw s3, 108(sp) # 4-byte Folded Spill
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset ra, -4
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s0, -8
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s1, -12
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s2, -16
-; RV32ZVKB-ZVE32X-NEXT:    .cfi_offset s3, -20
 ; RV32ZVKB-ZVE32X-NEXT:    addi s0, sp, 128
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_def_cfa s0, 0
 ; RV32ZVKB-ZVE32X-NEXT:    andi sp, sp, -64
-; RV32ZVKB-ZVE32X-NEXT:    mv a2, sp
 ; RV32ZVKB-ZVE32X-NEXT:    lw a3, 0(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw a4, 4(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw a5, 8(a1)
@@ -1242,9 +1231,9 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    lw t4, 60(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw t5, 16(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw t6, 20(a1)
-; RV32ZVKB-ZVE32X-NEXT:    lw s1, 24(a1)
-; RV32ZVKB-ZVE32X-NEXT:    lw s2, 28(a1)
-; RV32ZVKB-ZVE32X-NEXT:    lw s3, 40(a1)
+; RV32ZVKB-ZVE32X-NEXT:    lw a2, 24(a1)
+; RV32ZVKB-ZVE32X-NEXT:    lw s1, 28(a1)
+; RV32ZVKB-ZVE32X-NEXT:    lw s2, 40(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    lw a1, 44(a1)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a7, 48(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t0, 52(sp)
@@ -1252,18 +1241,19 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    sw t4, 60(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t1, 32(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t2, 36(sp)
-; RV32ZVKB-ZVE32X-NEXT:    sw s3, 40(sp)
+; RV32ZVKB-ZVE32X-NEXT:    sw s2, 40(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a1, 44(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a3, 16(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a4, 20(sp)
-; RV32ZVKB-ZVE32X-NEXT:    sw s1, 24(sp)
-; RV32ZVKB-ZVE32X-NEXT:    sw s2, 28(sp)
+; RV32ZVKB-ZVE32X-NEXT:    sw a2, 24(sp)
+; RV32ZVKB-ZVE32X-NEXT:    sw s1, 28(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t5, 0(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw t6, 4(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a5, 8(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    sw a6, 12(sp)
 ; RV32ZVKB-ZVE32X-NEXT:    vsetivli zero, 16, e32, m8, ta, ma
-; RV32ZVKB-ZVE32X-NEXT:    vle32.v v8, (a2)
+; RV32ZVKB-ZVE32X-NEXT:    mv a1, sp
+; RV32ZVKB-ZVE32X-NEXT:    vle32.v v8, (a1)
 ; RV32ZVKB-ZVE32X-NEXT:    vse32.v v8, (a0)
 ; RV32ZVKB-ZVE32X-NEXT:    addi sp, s0, -128
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_def_cfa sp, 128
@@ -1271,12 +1261,10 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV32ZVKB-ZVE32X-NEXT:    lw s0, 120(sp) # 4-byte Folded Reload
 ; RV32ZVKB-ZVE32X-NEXT:    lw s1, 116(sp) # 4-byte Folded Reload
 ; RV32ZVKB-ZVE32X-NEXT:    lw s2, 112(sp) # 4-byte Folded Reload
-; RV32ZVKB-ZVE32X-NEXT:    lw s3, 108(sp) # 4-byte Folded Reload
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore ra
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s0
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s1
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s2
-; RV32ZVKB-ZVE32X-NEXT:    .cfi_restore s3
 ; RV32ZVKB-ZVE32X-NEXT:    addi sp, sp, 128
 ; RV32ZVKB-ZVE32X-NEXT:    .cfi_def_cfa_offset 0
 ; RV32ZVKB-ZVE32X-NEXT:    ret
@@ -1289,16 +1277,13 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    sd s0, 112(sp) # 8-byte Folded Spill
 ; RV64ZVKB-ZVE32X-NEXT:    sd s1, 104(sp) # 8-byte Folded Spill
 ; RV64ZVKB-ZVE32X-NEXT:    sd s2, 96(sp) # 8-byte Folded Spill
-; RV64ZVKB-ZVE32X-NEXT:    sd s3, 88(sp) # 8-byte Folded Spill
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset ra, -8
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s0, -16
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s1, -24
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s2, -32
-; RV64ZVKB-ZVE32X-NEXT:    .cfi_offset s3, -40
 ; RV64ZVKB-ZVE32X-NEXT:    addi s0, sp, 128
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_def_cfa s0, 0
 ; RV64ZVKB-ZVE32X-NEXT:    andi sp, sp, -64
-; RV64ZVKB-ZVE32X-NEXT:    mv a2, sp
 ; RV64ZVKB-ZVE32X-NEXT:    ld a3, 8(a1)
 ; RV64ZVKB-ZVE32X-NEXT:    ld a4, 16(a1)
 ; RV64ZVKB-ZVE32X-NEXT:    ld a5, 0(a1)
@@ -1312,17 +1297,17 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    srli t4, a5, 32
 ; RV64ZVKB-ZVE32X-NEXT:    srli t5, a6, 32
 ; RV64ZVKB-ZVE32X-NEXT:    srli t6, t0, 32
-; RV64ZVKB-ZVE32X-NEXT:    srli s1, a7, 32
-; RV64ZVKB-ZVE32X-NEXT:    srli s2, t1, 32
-; RV64ZVKB-ZVE32X-NEXT:    srli s3, a1, 32
+; RV64ZVKB-ZVE32X-NEXT:    srli a2, a7, 32
+; RV64ZVKB-ZVE32X-NEXT:    srli s1, t1, 32
+; RV64ZVKB-ZVE32X-NEXT:    srli s2, a1, 32
 ; RV64ZVKB-ZVE32X-NEXT:    sw t1, 48(sp)
-; RV64ZVKB-ZVE32X-NEXT:    sw s2, 52(sp)
+; RV64ZVKB-ZVE32X-NEXT:    sw s1, 52(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw a1, 56(sp)
-; RV64ZVKB-ZVE32X-NEXT:    sw s3, 60(sp)
+; RV64ZVKB-ZVE32X-NEXT:    sw s2, 60(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t0, 32(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t6, 36(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw a7, 40(sp)
-; RV64ZVKB-ZVE32X-NEXT:    sw s1, 44(sp)
+; RV64ZVKB-ZVE32X-NEXT:    sw a2, 44(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw a5, 16(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t4, 20(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw a6, 24(sp)
@@ -1332,7 +1317,8 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    sw a3, 8(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    sw t3, 12(sp)
 ; RV64ZVKB-ZVE32X-NEXT:    vsetivli zero, 16, e32, m8, ta, ma
-; RV64ZVKB-ZVE32X-NEXT:    vle32.v v8, (a2)
+; RV64ZVKB-ZVE32X-NEXT:    mv a1, sp
+; RV64ZVKB-ZVE32X-NEXT:    vle32.v v8, (a1)
 ; RV64ZVKB-ZVE32X-NEXT:    vse32.v v8, (a0)
 ; RV64ZVKB-ZVE32X-NEXT:    addi sp, s0, -128
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_def_cfa sp, 128
@@ -1340,12 +1326,10 @@ define <8 x i64> @shuffle_v8i64_as_i256(<8 x i64> %v) {
 ; RV64ZVKB-ZVE32X-NEXT:    ld s0, 112(sp) # 8-byte Folded Reload
 ; RV64ZVKB-ZVE32X-NEXT:    ld s1, 104(sp) # 8-byte Folded Reload
 ; RV64ZVKB-ZVE32X-NEXT:    ld s2, 96(sp) # 8-byte Folded Reload
-; RV64ZVKB-ZVE32X-NEXT:    ld s3, 88(sp) # 8-byte Folded Reload
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore ra
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s0
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s1
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s2
-; RV64ZVKB-ZVE32X-NEXT:    .cfi_restore s3
 ; RV64ZVKB-ZVE32X-NEXT:    addi sp, sp, 128
 ; RV64ZVKB-ZVE32X-NEXT:    .cfi_def_cfa_offset 0
 ; RV64ZVKB-ZVE32X-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll b/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll
index 30648f371c28f4..ca4e41c22d1da5 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll
@@ -719,55 +719,57 @@ define <4 x i32> @ustest_f16i32(<4 x half> %x) {
 ; CHECK-NOV-NEXT:    .cfi_offset fs0, -56
 ; CHECK-NOV-NEXT:    .cfi_offset fs1, -64
 ; CHECK-NOV-NEXT:    .cfi_remember_state
-; CHECK-NOV-NEXT:    lhu s3, 0(a1)
+; CHECK-NOV-NEXT:    lhu s1, 0(a1)
 ; CHECK-NOV-NEXT:    lhu a2, 8(a1)
-; CHECK-NOV-NEXT:    lhu s1, 16(a1)
-; CHECK-NOV-NEXT:    lhu s2, 24(a1)
+; CHECK-NOV-NEXT:    lhu s2, 16(a1)
+; CHECK-NOV-NEXT:    lhu s3, 24(a1)
 ; CHECK-NOV-NEXT:    mv s0, a0
 ; CHECK-NOV-NEXT:    fmv.w.x fa0, a2
 ; CHECK-NOV-NEXT:    call __extendhfsf2
 ; CHECK-NOV-NEXT:    fmv.s fs0, fa0
-; CHECK-NOV-NEXT:    fmv.w.x fa0, s1
+; CHECK-NOV-NEXT:    fmv.w.x fa0, s2
 ; CHECK-NOV-NEXT:    call __extendhfsf2
 ; CHECK-NOV-NEXT:    fmv.s fs1, fa0
-; CHECK-NOV-NEXT:    fmv.w.x fa0, s2
+; CHECK-NOV-NEXT:    fmv.w.x fa0, s3
 ; CHECK-NOV-NEXT:    call __extendhfsf2
-; CHECK-NOV-NEXT:    fcvt.l.s s1, fa0, rtz
-; CHECK-NOV-NEXT:    fcvt.l.s s2, fs1, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s2, fa0, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s3, fs1, rtz
 ; CHECK-NOV-NEXT:    fcvt.l.s s4, fs0, rtz
-; CHECK-NOV-NEXT:    fmv.w.x fa0, s3
+; CHECK-NOV-NEXT:    fmv.w.x fa0, s1
 ; CHECK-NOV-NEXT:    call __extendhfsf2
+; CHECK-NOV-NEXT:    mv a4, s3
+; CHECK-NOV-NEXT:    mv a2, s2
 ; CHECK-NOV-NEXT:    fcvt.l.s a0, fa0, rtz
 ; CHECK-NOV-NEXT:    sgtz a1, s4
-; CHECK-NOV-NEXT:    sgtz a4, s2
-; CHECK-NOV-NEXT:    sgtz a2, a0
-; CHECK-NOV-NEXT:    sgtz a3, s1
-; CHECK-NOV-NEXT:    neg a2, a2
-; CHECK-NOV-NEXT:    and a0, a2, a0
-; CHECK-NOV-NEXT:    li a2, -1
+; CHECK-NOV-NEXT:    sgtz a6, s3
+; CHECK-NOV-NEXT:    sgtz a3, a0
+; CHECK-NOV-NEXT:    sgtz a5, s2
+; CHECK-NOV-NEXT:    neg a3, a3
+; CHECK-NOV-NEXT:    and a0, a3, a0
+; CHECK-NOV-NEXT:    li a3, -1
 ; CHECK-NOV-NEXT:    neg a1, a1
-; CHECK-NOV-NEXT:    srli a2, a2, 32
-; CHECK-NOV-NEXT:    neg a4, a4
+; CHECK-NOV-NEXT:    srli a3, a3, 32
+; CHECK-NOV-NEXT:    neg a6, a6
 ; CHECK-NOV-NEXT:    and a1, a1, s4
-; CHECK-NOV-NEXT:    bltu a0, a2, .LBB8_2
+; CHECK-NOV-NEXT:    bltu a0, a3, .LBB8_2
 ; CHECK-NOV-NEXT:  # %bb.1: # %entry
-; CHECK-NOV-NEXT:    mv a0, a2
+; CHECK-NOV-NEXT:    mv a0, a3
 ; CHECK-NOV-NEXT:  .LBB8_2: # %entry
-; CHECK-NOV-NEXT:    neg a5, a3
-; CHECK-NOV-NEXT:    and a3, a4, s2
-; CHECK-NOV-NEXT:    bgeu a1, a2, .LBB8_7
+; CHECK-NOV-NEXT:    neg a5, a5
+; CHECK-NOV-NEXT:    and a4, a6, a4
+; CHECK-NOV-NEXT:    bgeu a1, a3, .LBB8_7
 ; CHECK-NOV-NEXT:  # %bb.3: # %entry
-; CHECK-NOV-NEXT:    and a4, a5, s1
-; CHECK-NOV-NEXT:    bgeu a3, a2, .LBB8_8
+; CHECK-NOV-NEXT:    and a2, a5, a2
+; CHECK-NOV-NEXT:    bgeu a4, a3, .LBB8_8
 ; CHECK-NOV-NEXT:  .LBB8_4: # %entry
-; CHECK-NOV-NEXT:    bltu a4, a2, .LBB8_6
+; CHECK-NOV-NEXT:    bltu a2, a3, .LBB8_6
 ; CHECK-NOV-NEXT:  .LBB8_5: # %entry
-; CHECK-NOV-NEXT:    mv a4, a2
+; CHECK-NOV-NEXT:    mv a2, a3
 ; CHECK-NOV-NEXT:  .LBB8_6: # %entry
 ; CHECK-NOV-NEXT:    sw a0, 0(s0)
 ; CHECK-NOV-NEXT:    sw a1, 4(s0)
-; CHECK-NOV-NEXT:    sw a3, 8(s0)
-; CHECK-NOV-NEXT:    sw a4, 12(s0)
+; CHECK-NOV-NEXT:    sw a4, 8(s0)
+; CHECK-NOV-NEXT:    sw a2, 12(s0)
 ; CHECK-NOV-NEXT:    ld ra, 56(sp) # 8-byte Folded Reload
 ; CHECK-NOV-NEXT:    ld s0, 48(sp) # 8-byte Folded Reload
 ; CHECK-NOV-NEXT:    ld s1, 40(sp) # 8-byte Folded Reload
@@ -789,12 +791,12 @@ define <4 x i32> @ustest_f16i32(<4 x half> %x) {
 ; CHECK-NOV-NEXT:    ret
 ; CHECK-NOV-NEXT:  .LBB8_7: # %entry
 ; CHECK-NOV-NEXT:    .cfi_restore_state
-; CHECK-NOV-NEXT:    mv a1, a2
-; CHECK-NOV-NEXT:    and a4, a5, s1
-; CHECK-NOV-NEXT:    bltu a3, a2, .LBB8_4
+; CHECK-NOV-NEXT:    mv a1, a3
+; CHECK-NOV-NEXT:    and a2, a5, a2
+; CHECK-NOV-NEXT:    bltu a4, a3, .LBB8_4
 ; CHECK-NOV-NEXT:  .LBB8_8: # %entry
-; CHECK-NOV-NEXT:    mv a3, a2
-; CHECK-NOV-NEXT:    bgeu a4, a2, .LBB8_5
+; CHECK-NOV-NEXT:    mv a4, a3
+; CHECK-NOV-NEXT:    bgeu a2, a3, .LBB8_5
 ; CHECK-NOV-NEXT:    j .LBB8_6
 ;
 ; CHECK-V-LABEL: ustest_f16i32:
@@ -2005,34 +2007,36 @@ define <8 x i16> @ustest_f16i16(<8 x half> %x) {
 ; CHECK-NOV-NEXT:    fcvt.l.s s4, fs3, rtz
 ; CHECK-NOV-NEXT:    fmv.w.x fa0, s5
 ; CHECK-NOV-NEXT:    call __extendhfsf2
+; CHECK-NOV-NEXT:    mv a2, s3
+; CHECK-NOV-NEXT:    mv a4, s2
 ; CHECK-NOV-NEXT:    fcvt.l.s a0, fa0, rtz
 ; CHECK-NOV-NEXT:    sgtz a1, s4
-; CHECK-NOV-NEXT:    sgtz a4, s3
+; CHECK-NOV-NEXT:    sgtz a6, s3
 ; CHECK-NOV-NEXT:    sgtz a3, a0
-; CHECK-NOV-NEXT:    sgtz a2, s2
+; CHECK-NOV-NEXT:    sgtz a5, s2
 ; CHECK-NOV-NEXT:    neg a3, a3
 ; CHECK-NOV-NEXT:    and a0, a3, a0
 ; CHECK-NOV-NEXT:    lui a3, 16
 ; CHECK-NOV-NEXT:    neg a1, a1
 ; CHECK-NOV-NEXT:    addi a3, a3, -1
-; CHECK-NOV-NEXT:    neg a5, a4
+; CHECK-NOV-NEXT:    neg a7, a6
 ; CHECK-NOV-NEXT:    and a1, a1, s4
 ; CHECK-NOV-NEXT:    bltu a0, a3, .LBB17_2
 ; CHECK-NOV-NEXT:  # %bb.1: # %entry
 ; CHECK-NOV-NEXT:    mv a0, a3
 ; CHECK-NOV-NEXT:  .LBB17_2: # %entry
 ; CHECK-NOV-NEXT:    fcvt.l.s a6, fs2, rtz
-; CHECK-NOV-NEXT:    sgtz a4, s1
-; CHECK-NOV-NEXT:    neg t0, a2
-; CHECK-NOV-NEXT:    and a2, a5, s3
+; CHECK-NOV-NEXT:    sgtz t0, s1
+; CHECK-NOV-NEXT:    neg t2, a5
+; CHECK-NOV-NEXT:    and a2, a7, a2
 ; CHECK-NOV-NEXT:    bltu a1, a3, .LBB17_4
 ; CHECK-NOV-NEXT:  # %bb.3: # %entry
 ; CHECK-NOV-NEXT:    mv a1, a3
 ; CHECK-NOV-NEXT:  .LBB17_4: # %entry
 ; CHECK-NOV-NEXT:    fcvt.l.s a7, fs1, rtz
 ; CHECK-NOV-NEXT:    sgtz a5, a6
-; CHECK-NOV-NEXT:    neg t1, a4
-; CHECK-NOV-NEXT:    and a4, t0, s2
+; CHECK-NOV-NEXT:    neg t1, t0
+; CHECK-NOV-NEXT:    and a4, t2, a4
 ; CHECK-NOV-NEXT:    bltu a2, a3, .LBB17_6
 ; CHECK-NOV-NEXT:  # %bb.5: # %entry
 ; CHECK-NOV-NEXT:    mv a2, a3
@@ -3423,22 +3427,22 @@ define <2 x i64> @ustest_f16i64(<2 x half> %x) {
 ; CHECK-NOV-NEXT:    .cfi_offset s0, -16
 ; CHECK-NOV-NEXT:    .cfi_offset s1, -24
 ; CHECK-NOV-NEXT:    .cfi_offset s2, -32
-; CHECK-NOV-NEXT:    mv s2, a1
+; CHECK-NOV-NEXT:    mv s1, a1
 ; CHECK-NOV-NEXT:    fmv.w.x fa0, a0
 ; CHECK-NOV-NEXT:    call __extendhfsf2
 ; CHECK-NOV-NEXT:    call __fixsfti
 ; CHECK-NOV-NEXT:    mv s0, a0
-; CHECK-NOV-NEXT:    mv s1, a1
-; CHECK-NOV-NEXT:    fmv.w.x fa0, s2
+; CHECK-NOV-NEXT:    mv s2, a1
+; CHECK-NOV-NEXT:    fmv.w.x fa0, s1
 ; CHECK-NOV-NEXT:    call __extendhfsf2
 ; CHECK-NOV-NEXT:    call __fixsfti
-; CHECK-NOV-NEXT:    mv a2, s1
-; CHECK-NOV-NEXT:    blez s1, .LBB26_2
+; CHECK-NOV-NEXT:    mv a2, s2
+; CHECK-NOV-NEXT:    blez s2, .LBB26_2
 ; CHECK-NOV-NEXT:  # %bb.1: # %entry
 ; CHECK-NOV-NEXT:    li a2, 1
 ; CHECK-NOV-NEXT:  .LBB26_2: # %entry
 ; CHECK-NOV-NEXT:    slti a3, a1, 1
-; CHECK-NOV-NEXT:    slti a4, s1, 1
+; CHECK-NOV-NEXT:    slti a4, s2, 1
 ; CHECK-NOV-NEXT:    blez a1, .LBB26_4
 ; CHECK-NOV-NEXT:  # %bb.3: # %entry
 ; CHECK-NOV-NEXT:    li a1, 1
@@ -3489,22 +3493,22 @@ define <2 x i64> @ustest_f16i64(<2 x half> %x) {
 ; CHECK-V-NEXT:    .cfi_offset s0, -16
 ; CHECK-V-NEXT:    .cfi_offset s1, -24
 ; CHECK-V-NEXT:    .cfi_offset s2, -32
-; CHECK-V-NEXT:    mv s2, a1
+; CHECK-V-NEXT:    mv s1, a1
 ; CHECK-V-NEXT:    fmv.w.x fa0, a0
 ; CHECK-V-NEXT:    call __extendhfsf2
 ; CHECK-V-NEXT:    call __fixsfti
 ; CHECK-V-NEXT:    mv s0, a0
-; CHECK-V-NEXT:    mv s1, a1
-; CHECK-V-NEXT:    fmv.w.x fa0, s2
+; CHECK-V-NEXT:    mv s2, a1
+; CHECK-V-NEXT:    fmv.w.x fa0, s1
 ; CHECK-V-NEXT:    call __extendhfsf2
 ; CHECK-V-NEXT:    call __fixsfti
-; CHECK-V-NEXT:    mv a2, s1
-; CHECK-V-NEXT:    blez s1, .LBB26_2
+; CHECK-V-NEXT:    mv a2, s2
+; CHECK-V-NEXT:    blez s2, .LBB26_2
 ; CHECK-V-NEXT:  # %bb.1: # %entry
 ; CHECK-V-NEXT:    li a2, 1
 ; CHECK-V-NEXT:  .LBB26_2: # %entry
 ; CHECK-V-NEXT:    slti a3, a1, 1
-; CHECK-V-NEXT:    slti a4, s1, 1
+; CHECK-V-NEXT:    slti a4, s2, 1
 ; CHECK-V-NEXT:    blez a1, .LBB26_4
 ; CHECK-V-NEXT:  # %bb.3: # %entry
 ; CHECK-V-NEXT:    li a1, 1
@@ -4271,29 +4275,30 @@ define <4 x i32> @ustest_f16i32_mm(<4 x half> %x) {
 ; CHECK-NOV-NEXT:    fmv.s fs1, fa0
 ; CHECK-NOV-NEXT:    fmv.w.x fa0, s1
 ; CHECK-NOV-NEXT:    call __extendhfsf2
-; CHECK-NOV-NEXT:    fcvt.l.s s2, fa0, rtz
-; CHECK-NOV-NEXT:    fcvt.l.s s4, fs1, rtz
-; CHECK-NOV-NEXT:    fcvt.l.s s1, fs0, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s1, fa0, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s2, fs1, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s4, fs0, rtz
 ; CHECK-NOV-NEXT:    fmv.w.x fa0, s3
 ; CHECK-NOV-NEXT:    call __extendhfsf2
+; CHECK-NOV-NEXT:    mv a4, s4
 ; CHECK-NOV-NEXT:    fcvt.l.s a2, fa0, rtz
-; CHECK-NOV-NEXT:    sgtz a0, s4
-; CHECK-NOV-NEXT:    sgtz a5, s1
+; CHECK-NOV-NEXT:    sgtz a0, s2
+; CHECK-NOV-NEXT:    sgtz a6, s4
 ; CHECK-NOV-NEXT:    neg a0, a0
-; CHECK-NOV-NEXT:    and a0, a0, s4
-; CHECK-NOV-NEXT:    sgtz a1, s2
+; CHECK-NOV-NEXT:    and a0, a0, s2
+; CHECK-NOV-NEXT:    sgtz a1, s1
 ; CHECK-NOV-NEXT:    neg a1, a1
 ; CHECK-NOV-NEXT:    li a3, -1
-; CHECK-NOV-NEXT:    and a1, a1, s2
+; CHECK-NOV-NEXT:    and a1, a1, s1
 ; CHECK-NOV-NEXT:    srli a3, a3, 32
-; CHECK-NOV-NEXT:    sgtz a4, a2
-; CHECK-NOV-NEXT:    neg a6, a5
+; CHECK-NOV-NEXT:    sgtz a5, a2
+; CHECK-NOV-NEXT:    neg a6, a6
 ; CHECK-NOV-NEXT:    blt a1, a3, .LBB35_2
 ; CHECK-NOV-NEXT:  # %bb.1: # %entry
 ; CHECK-NOV-NEXT:    mv a1, a3
 ; CHECK-NOV-NEXT:  .LBB35_2: # %entry
-; CHECK-NOV-NEXT:    neg a5, a4
-; CHECK-NOV-NEXT:    and a4, a6, s1
+; CHECK-NOV-NEXT:    neg a5, a5
+; CHECK-NOV-NEXT:    and a4, a6, a4
 ; CHECK-NOV-NEXT:    bge a0, a3, .LBB35_7
 ; CHECK-NOV-NEXT:  # %bb.3: # %entry
 ; CHECK-NOV-NEXT:    and a2, a5, a2
@@ -5494,8 +5499,8 @@ define <8 x i16> @ustest_f16i16_mm(<8 x half> %x) {
 ; CHECK-NOV-NEXT:    lhu s1, 32(a1)
 ; CHECK-NOV-NEXT:    lhu s2, 40(a1)
 ; CHECK-NOV-NEXT:    lhu a2, 48(a1)
-; CHECK-NOV-NEXT:    lhu s4, 56(a1)
-; CHECK-NOV-NEXT:    lhu s3, 0(a1)
+; CHECK-NOV-NEXT:    lhu s3, 56(a1)
+; CHECK-NOV-NEXT:    lhu s4, 0(a1)
 ; CHECK-NOV-NEXT:    lhu s5, 8(a1)
 ; CHECK-NOV-NEXT:    lhu s6, 16(a1)
 ; CHECK-NOV-NEXT:    lhu s7, 24(a1)
@@ -5518,42 +5523,44 @@ define <8 x i16> @ustest_f16i16_mm(<8 x half> %x) {
 ; CHECK-NOV-NEXT:    fmv.w.x fa0, s5
 ; CHECK-NOV-NEXT:    call __extendhfsf2
 ; CHECK-NOV-NEXT:    fmv.s fs5, fa0
-; CHECK-NOV-NEXT:    fmv.w.x fa0, s3
+; CHECK-NOV-NEXT:    fmv.w.x fa0, s4
 ; CHECK-NOV-NEXT:    call __extendhfsf2
-; CHECK-NOV-NEXT:    fcvt.l.s s5, fa0, rtz
-; CHECK-NOV-NEXT:    fcvt.l.s s6, fs5, rtz
-; CHECK-NOV-NEXT:    fcvt.l.s s3, fs4, rtz
-; CHECK-NOV-NEXT:    fcvt.l.s s2, fs3, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s2, fa0, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s4, fs5, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s5, fs4, rtz
+; CHECK-NOV-NEXT:    fcvt.l.s s6, fs3, rtz
 ; CHECK-NOV-NEXT:    fcvt.l.s s1, fs2, rtz
-; CHECK-NOV-NEXT:    fmv.w.x fa0, s4
+; CHECK-NOV-NEXT:    fmv.w.x fa0, s3
 ; CHECK-NOV-NEXT:    call __extendhfsf2
-; CHECK-NOV-NEXT:    sgtz a0, s6
-; CHECK-NOV-NEXT:    sgtz a4, s3
+; CHECK-NOV-NEXT:    mv a2, s5
+; CHECK-NOV-NEXT:    mv a4, s6
+; CHECK-NOV-NEXT:    sgtz a0, s4
+; CHECK-NOV-NEXT:    sgtz a6, s5
 ; CHECK-NOV-NEXT:    neg a0, a0
-; CHECK-NOV-NEXT:    sgtz a1, s5
-; CHECK-NOV-NEXT:    and a0, a0, s6
+; CHECK-NOV-NEXT:    sgtz a1, s2
+; CHECK-NOV-NEXT:    and a0, a0, s4
 ; CHECK-NOV-NEXT:    neg a1, a1
 ; CHECK-NOV-NEXT:    lui a3, 16
-; CHECK-NOV-NEXT:    and a1, a1, s5
+; CHECK-NOV-NEXT:    and a1, a1, s2
 ; CHECK-NOV-NEXT:    addi a3, a3, -1
-; CHECK-NOV-NEXT:    sgtz a2, s2
-; CHECK-NOV-NEXT:    neg a5, a4
+; CHECK-NOV-NEXT:    sgtz a5, s6
+; CHECK-NOV-NEXT:    neg a7, a6
 ; CHECK-NOV-NEXT:    blt a1, a3, .LBB44_2
 ; CHECK-NOV-NEXT:  # %bb.1: # %entry
 ; CHECK-NOV-NEXT:    mv a1, a3
 ; CHECK-NOV-NEXT:  .LBB44_2: # %entry
 ; CHECK-NOV-NEXT:    fcvt.l.s a6, fs1, rtz
-; CHECK-NOV-NEXT:    sgtz a4, s1
-; CHECK-NOV-NEXT:    neg t0, a2
-; CHECK-NOV-NEXT:    and a2, a5, s3
+; CHECK-NOV-NEXT:    sgtz t0, s1
+; CHECK-NOV-NEXT:    neg t2, a5
+; CHECK-NOV-NEXT:    and a2, a7, a2
 ; CHECK-NOV-NEXT:    blt a0, a3, .LBB44_4
 ; CHECK-NOV-NEXT:  # %bb.3: # %entry
 ; CHECK-NOV-NEXT:    mv a0, a3
 ; CHECK-NOV-NEXT:  .LBB44_4: # %entry
 ; CHECK-NOV-NEXT:    fcvt.l.s a7, fs0, rtz
 ; CHECK-NOV-NEXT:    sgtz a5, a6
-; CHECK-NOV-NEXT:    neg t1, a4
-; CHECK-NOV-NEXT:    and a4, t0, s2
+; CHECK-NOV-NEXT:    neg t1, t0
+; CHECK-NOV-NEXT:    and a4, t2, a4
 ; CHECK-NOV-NEXT:    blt a2, a3, .LBB44_6
 ; CHECK-NOV-NEXT:  # %bb.5: # %entry
 ; CHECK-NOV-NEXT:    mv a2, a3
diff --git a/llvm/test/CodeGen/RISCV/rvv/interrupt-attr-nocall.ll b/llvm/test/CodeGen/RISCV/rvv/interrupt-attr-nocall.ll
index 42c2556f2157db..4d0a82680a4503 100644
--- a/llvm/test/CodeGen/RISCV/rvv/interrupt-attr-nocall.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/interrupt-attr-nocall.ll
@@ -131,7 +131,6 @@ define void @foo_lmul8() nounwind #0 {
 ; CHECK-RV32:       # %bb.0:
 ; CHECK-RV32-NEXT:    addi sp, sp, -32
 ; CHECK-RV32-NEXT:    sw a0, 28(sp) # 4-byte Folded Spill
-; CHECK-RV32-NEXT:    sw a1, 24(sp) # 4-byte Folded Spill
 ; CHECK-RV32-NEXT:    csrr a0, vlenb
 ; CHECK-RV32-NEXT:    slli a0, a0, 4
 ; CHECK-RV32-NEXT:    sub sp, sp, a0
@@ -142,10 +141,10 @@ define void @foo_lmul8() nounwind #0 {
 ; CHECK-RV32-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV32-NEXT:    addi a0, sp, 16
 ; CHECK-RV32-NEXT:    vs8r.v v16, (a0) # vscale x 64-byte Folded Spill
+; CHECK-RV32-NEXT:    li a0, 32
+; CHECK-RV32-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
 ; CHECK-RV32-NEXT:    lui a0, %hi(j)
 ; CHECK-RV32-NEXT:    addi a0, a0, %lo(j)
-; CHECK-RV32-NEXT:    li a1, 32
-; CHECK-RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, ma
 ; CHECK-RV32-NEXT:    vle32.v v8, (a0)
 ; CHECK-RV32-NEXT:    lui a0, %hi(k)
 ; CHECK-RV32-NEXT:    addi a0, a0, %lo(k)
@@ -165,7 +164,6 @@ define void @foo_lmul8() nounwind #0 {
 ; CHECK-RV32-NEXT:    slli a0, a0, 4
 ; CHECK-RV32-NEXT:    add sp, sp, a0
 ; CHECK-RV32-NEXT:    lw a0, 28(sp) # 4-byte Folded Reload
-; CHECK-RV32-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; CHECK-RV32-NEXT:    addi sp, sp, 32
 ; CHECK-RV32-NEXT:    mret
   %1 = load <32 x i32>, ptr @j
diff --git a/llvm/test/CodeGen/RISCV/rvv/localvar.ll b/llvm/test/CodeGen/RISCV/rvv/localvar.ll
index f7c921c487e637..88b6e05bba22a5 100644
--- a/llvm/test/CodeGen/RISCV/rvv/localvar.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/localvar.ll
@@ -210,10 +210,8 @@ define void @local_var_m2_with_varsize_object(i64 %n) {
 ; RV64IV-NEXT:    .cfi_def_cfa_offset 32
 ; RV64IV-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; RV64IV-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64IV-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
 ; RV64IV-NEXT:    .cfi_offset ra, -8
 ; RV64IV-NEXT:    .cfi_offset s0, -16
-; RV64IV-NEXT:    .cfi_offset s1, -24
 ; RV64IV-NEXT:    addi s0, sp, 32
 ; RV64IV-NEXT:    .cfi_def_cfa s0, 0
 ; RV64IV-NEXT:    csrr a1, vlenb
@@ -227,12 +225,12 @@ define void @local_var_m2_with_varsize_object(i64 %n) {
 ; RV64IV-NEXT:    slli a1, a1, 1
 ; RV64IV-NEXT:    sub a1, s0, a1
 ; RV64IV-NEXT:    addi a1, a1, -32
-; RV64IV-NEXT:    csrr s1, vlenb
-; RV64IV-NEXT:    slli s1, s1, 1
-; RV64IV-NEXT:    sub s1, s0, s1
-; RV64IV-NEXT:    addi s1, s1, -32
 ; RV64IV-NEXT:    call notdead
-; RV64IV-NEXT:    vl2r.v v8, (s1)
+; RV64IV-NEXT:    csrr a0, vlenb
+; RV64IV-NEXT:    slli a0, a0, 1
+; RV64IV-NEXT:    sub a0, s0, a0
+; RV64IV-NEXT:    addi a0, a0, -32
+; RV64IV-NEXT:    vl2r.v v8, (a0)
 ; RV64IV-NEXT:    csrr a0, vlenb
 ; RV64IV-NEXT:    slli a0, a0, 2
 ; RV64IV-NEXT:    sub a0, s0, a0
@@ -242,10 +240,8 @@ define void @local_var_m2_with_varsize_object(i64 %n) {
 ; RV64IV-NEXT:    .cfi_def_cfa sp, 32
 ; RV64IV-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; RV64IV-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64IV-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
 ; RV64IV-NEXT:    .cfi_restore ra
 ; RV64IV-NEXT:    .cfi_restore s0
-; RV64IV-NEXT:    .cfi_restore s1
 ; RV64IV-NEXT:    addi sp, sp, 32
 ; RV64IV-NEXT:    .cfi_def_cfa_offset 0
 ; RV64IV-NEXT:    ret
@@ -261,17 +257,15 @@ define void @local_var_m2_with_varsize_object(i64 %n) {
 define void @local_var_m2_with_bp(i64 %n) {
 ; RV64IV-LABEL: local_var_m2_with_bp:
 ; RV64IV:       # %bb.0:
-; RV64IV-NEXT:    addi sp, sp, -256
-; RV64IV-NEXT:    .cfi_def_cfa_offset 256
-; RV64IV-NEXT:    sd ra, 248(sp) # 8-byte Folded Spill
-; RV64IV-NEXT:    sd s0, 240(sp) # 8-byte Folded Spill
-; RV64IV-NEXT:    sd s1, 232(sp) # 8-byte Folded Spill
-; RV64IV-NEXT:    sd s2, 224(sp) # 8-byte Folded Spill
+; RV64IV-NEXT:    addi sp, sp, -272
+; RV64IV-NEXT:    .cfi_def_cfa_offset 272
+; RV64IV-NEXT:    sd ra, 264(sp) # 8-byte Folded Spill
+; RV64IV-NEXT:    sd s0, 256(sp) # 8-byte Folded Spill
+; RV64IV-NEXT:    sd s1, 248(sp) # 8-byte Folded Spill
 ; RV64IV-NEXT:    .cfi_offset ra, -8
 ; RV64IV-NEXT:    .cfi_offset s0, -16
 ; RV64IV-NEXT:    .cfi_offset s1, -24
-; RV64IV-NEXT:    .cfi_offset s2, -32
-; RV64IV-NEXT:    addi s0, sp, 256
+; RV64IV-NEXT:    addi s0, sp, 272
 ; RV64IV-NEXT:    .cfi_def_cfa s0, 0
 ; RV64IV-NEXT:    csrr a1, vlenb
 ; RV64IV-NEXT:    slli a1, a1, 2
@@ -286,28 +280,26 @@ define void @local_var_m2_with_bp(i64 %n) {
 ; RV64IV-NEXT:    csrr a2, vlenb
 ; RV64IV-NEXT:    slli a2, a2, 1
 ; RV64IV-NEXT:    add a2, s1, a2
-; RV64IV-NEXT:    addi a2, a2, 224
-; RV64IV-NEXT:    csrr s2, vlenb
-; RV64IV-NEXT:    slli s2, s2, 1
-; RV64IV-NEXT:    add s2, s1, s2
-; RV64IV-NEXT:    addi s2, s2, 224
+; RV64IV-NEXT:    addi a2, a2, 240
 ; RV64IV-NEXT:    call notdead2
 ; RV64IV-NEXT:    lw zero, 124(s1)
-; RV64IV-NEXT:    vl2r.v v8, (s2)
-; RV64IV-NEXT:    addi a0, s1, 224
+; RV64IV-NEXT:    csrr a0, vlenb
+; RV64IV-NEXT:    slli a0, a0, 1
+; RV64IV-NEXT:    add a0, s1, a0
+; RV64IV-NEXT:    addi a0, a0, 240
+; RV64IV-NEXT:    vl2r.v v8, (a0)
+; RV64IV-NEXT:    addi a0, s1, 240
 ; RV64IV-NEXT:    vl2r.v v8, (a0)
 ; RV64IV-NEXT:    lw zero, 120(s1)
-; RV64IV-NEXT:    addi sp, s0, -256
-; RV64IV-NEXT:    .cfi_def_cfa sp, 256
-; RV64IV-NEXT:    ld ra, 248(sp) # 8-byte Folded Reload
-; RV64IV-NEXT:    ld s0, 240(sp) # 8-byte Folded Reload
-; RV64IV-NEXT:    ld s1, 232(sp) # 8-byte Folded Reload
-; RV64IV-NEXT:    ld s2, 224(sp) # 8-byte Folded Reload
+; RV64IV-NEXT:    addi sp, s0, -272
+; RV64IV-NEXT:    .cfi_def_cfa sp, 272
+; RV64IV-NEXT:    ld ra, 264(sp) # 8-byte Folded Reload
+; RV64IV-NEXT:    ld s0, 256(sp) # 8-byte Folded Reload
+; RV64IV-NEXT:    ld s1, 248(sp) # 8-byte Folded Reload
 ; RV64IV-NEXT:    .cfi_restore ra
 ; RV64IV-NEXT:    .cfi_restore s0
 ; RV64IV-NEXT:    .cfi_restore s1
-; RV64IV-NEXT:    .cfi_restore s2
-; RV64IV-NEXT:    addi sp, sp, 256
+; RV64IV-NEXT:    addi sp, sp, 272
 ; RV64IV-NEXT:    .cfi_def_cfa_offset 0
 ; RV64IV-NEXT:    ret
   %1 = alloca i8, i64 %n
diff --git a/llvm/test/CodeGen/RISCV/rvv/nontemporal-vp-scalable.ll b/llvm/test/CodeGen/RISCV/rvv/nontemporal-vp-scalable.ll
index fbdfcea6cc5113..6e5c4775335147 100644
--- a/llvm/test/CodeGen/RISCV/rvv/nontemporal-vp-scalable.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/nontemporal-vp-scalable.ll
@@ -34473,7 +34473,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a1, a1, a2
 ; CHECK-RV64V-NEXT:    sub sp, sp, a1
 ; CHECK-RV64V-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64V-NEXT:    mv s2, a0
+; CHECK-RV64V-NEXT:    mv s0, a0
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
@@ -34487,9 +34487,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
-; CHECK-RV64V-NEXT:    slli a0, s0, 3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    csrr s1, vlenb
+; CHECK-RV64V-NEXT:    slli a0, s1, 3
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -34499,63 +34499,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    slli a0, s0, 4
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a0, s1, 4
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 4
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s1, a6
+; CHECK-RV64V-NEXT:    mv s2, a6
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s1, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    addi a0, sp, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
-; CHECK-RV64V-NEXT:    sub a1, s1, a2
-; CHECK-RV64V-NEXT:    sltu a3, s1, a1
+; CHECK-RV64V-NEXT:    slli a2, s1, 2
+; CHECK-RV64V-NEXT:    sub a1, s2, a2
+; CHECK-RV64V-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a3, a3, a1
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
+; CHECK-RV64V-NEXT:    slli a1, s1, 1
 ; CHECK-RV64V-NEXT:    sub a4, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64V-NEXT:    addi a5, a5, -1
 ; CHECK-RV64V-NEXT:    and a4, a5, a4
-; CHECK-RV64V-NEXT:    sub a5, a4, s0
+; CHECK-RV64V-NEXT:    sub a5, a4, s1
 ; CHECK-RV64V-NEXT:    sltu a6, a4, a5
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB910_2
+; CHECK-RV64V-NEXT:    bltu a4, s1, .LBB910_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
-; CHECK-RV64V-NEXT:    mv a4, s0
+; CHECK-RV64V-NEXT:    mv a4, s1
 ; CHECK-RV64V-NEXT:  .LBB910_2:
 ; CHECK-RV64V-NEXT:    addi a6, a6, -1
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB910_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a3, a1
 ; CHECK-RV64V-NEXT:  .LBB910_4:
-; CHECK-RV64V-NEXT:    sub a7, a3, s0
+; CHECK-RV64V-NEXT:    sub a7, a3, s1
 ; CHECK-RV64V-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64V-NEXT:    and a5, a6, a5
 ; CHECK-RV64V-NEXT:    addi t0, t0, -1
-; CHECK-RV64V-NEXT:    add a6, s2, a0
+; CHECK-RV64V-NEXT:    add a6, s0, a0
 ; CHECK-RV64V-NEXT:    and a0, t0, a7
-; CHECK-RV64V-NEXT:    bltu a3, s0, .LBB910_6
+; CHECK-RV64V-NEXT:    bltu a3, s1, .LBB910_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:  .LBB910_6:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64V-NEXT:    vle64.v v16, (a6)
@@ -34580,25 +34580,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB910_8
+; CHECK-RV64V-NEXT:    bltu s2, a2, .LBB910_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv s1, a2
+; CHECK-RV64V-NEXT:    mv s2, a2
 ; CHECK-RV64V-NEXT:  .LBB910_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64V-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64V-NEXT:    sub a0, s1, a1
-; CHECK-RV64V-NEXT:    sltu a2, s1, a0
+; CHECK-RV64V-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64V-NEXT:    sub a0, s2, a1
+; CHECK-RV64V-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64V-NEXT:    addi a2, a2, -1
 ; CHECK-RV64V-NEXT:    and a0, a2, a0
-; CHECK-RV64V-NEXT:    sub a2, a0, s0
+; CHECK-RV64V-NEXT:    sub a2, a0, s1
 ; CHECK-RV64V-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a2, a3, a2
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB910_10
+; CHECK-RV64V-NEXT:    bltu a0, s1, .LBB910_10
 ; CHECK-RV64V-NEXT:  # %bb.9:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:  .LBB910_10:
 ; CHECK-RV64V-NEXT:    csrr a3, vlenb
 ; CHECK-RV64V-NEXT:    slli a3, a3, 3
@@ -34614,12 +34614,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, a1, .LBB910_12
+; CHECK-RV64V-NEXT:    bltu s2, a1, .LBB910_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv s1, a1
+; CHECK-RV64V-NEXT:    mv s2, a1
 ; CHECK-RV64V-NEXT:  .LBB910_12:
-; CHECK-RV64V-NEXT:    sub a0, s1, s0
-; CHECK-RV64V-NEXT:    sltu a1, s1, a0
+; CHECK-RV64V-NEXT:    sub a0, s2, s1
+; CHECK-RV64V-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64V-NEXT:    addi a1, a1, -1
 ; CHECK-RV64V-NEXT:    and a0, a1, a0
 ; CHECK-RV64V-NEXT:    csrr a1, vlenb
@@ -34630,9 +34630,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, s0, .LBB910_14
+; CHECK-RV64V-NEXT:    bltu s2, s1, .LBB910_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv s1, s0
+; CHECK-RV64V-NEXT:    mv s2, s1
 ; CHECK-RV64V-NEXT:  .LBB910_14:
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -34642,7 +34642,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64V-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64V-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
@@ -34748,7 +34748,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a1, a1, a2
 ; CHECK-RV64VC-NEXT:    sub sp, sp, a1
 ; CHECK-RV64VC-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64VC-NEXT:    mv s2, a0
+; CHECK-RV64VC-NEXT:    mv s0, a0
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
@@ -34762,9 +34762,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
-; CHECK-RV64VC-NEXT:    slli a0, s0, 3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    csrr s1, vlenb
+; CHECK-RV64VC-NEXT:    slli a0, s1, 3
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -34774,63 +34774,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    slli a0, s0, 4
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a0, s1, 4
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 4
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s1, a6
+; CHECK-RV64VC-NEXT:    mv s2, a6
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s1, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    addi a0, sp, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
-; CHECK-RV64VC-NEXT:    sub a1, s1, a2
-; CHECK-RV64VC-NEXT:    sltu a3, s1, a1
+; CHECK-RV64VC-NEXT:    slli a2, s1, 2
+; CHECK-RV64VC-NEXT:    sub a1, s2, a2
+; CHECK-RV64VC-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a3, a3, a1
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
+; CHECK-RV64VC-NEXT:    slli a1, s1, 1
 ; CHECK-RV64VC-NEXT:    sub a4, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64VC-NEXT:    addi a5, a5, -1
 ; CHECK-RV64VC-NEXT:    and a4, a4, a5
-; CHECK-RV64VC-NEXT:    sub a5, a4, s0
+; CHECK-RV64VC-NEXT:    sub a5, a4, s1
 ; CHECK-RV64VC-NEXT:    sltu a6, a4, a5
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB910_2
+; CHECK-RV64VC-NEXT:    bltu a4, s1, .LBB910_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
-; CHECK-RV64VC-NEXT:    mv a4, s0
+; CHECK-RV64VC-NEXT:    mv a4, s1
 ; CHECK-RV64VC-NEXT:  .LBB910_2:
 ; CHECK-RV64VC-NEXT:    addi a6, a6, -1
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB910_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a3, a1
 ; CHECK-RV64VC-NEXT:  .LBB910_4:
-; CHECK-RV64VC-NEXT:    sub a7, a3, s0
+; CHECK-RV64VC-NEXT:    sub a7, a3, s1
 ; CHECK-RV64VC-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64VC-NEXT:    and a5, a6, a5
 ; CHECK-RV64VC-NEXT:    addi t0, t0, -1
-; CHECK-RV64VC-NEXT:    add a6, s2, a0
+; CHECK-RV64VC-NEXT:    add a6, s0, a0
 ; CHECK-RV64VC-NEXT:    and a0, t0, a7
-; CHECK-RV64VC-NEXT:    bltu a3, s0, .LBB910_6
+; CHECK-RV64VC-NEXT:    bltu a3, s1, .LBB910_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:  .LBB910_6:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64VC-NEXT:    vle64.v v16, (a6)
@@ -34855,25 +34855,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB910_8
+; CHECK-RV64VC-NEXT:    bltu s2, a2, .LBB910_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv s1, a2
+; CHECK-RV64VC-NEXT:    mv s2, a2
 ; CHECK-RV64VC-NEXT:  .LBB910_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64VC-NEXT:    sub a0, s1, a1
-; CHECK-RV64VC-NEXT:    sltu a2, s1, a0
+; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64VC-NEXT:    sub a0, s2, a1
+; CHECK-RV64VC-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a2, a2, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a2
-; CHECK-RV64VC-NEXT:    sub a2, a0, s0
+; CHECK-RV64VC-NEXT:    sub a2, a0, s1
 ; CHECK-RV64VC-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a2, a2, a3
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB910_10
+; CHECK-RV64VC-NEXT:    bltu a0, s1, .LBB910_10
 ; CHECK-RV64VC-NEXT:  # %bb.9:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:  .LBB910_10:
 ; CHECK-RV64VC-NEXT:    csrr a3, vlenb
 ; CHECK-RV64VC-NEXT:    slli a3, a3, 3
@@ -34889,12 +34889,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, a1, .LBB910_12
+; CHECK-RV64VC-NEXT:    bltu s2, a1, .LBB910_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv s1, a1
+; CHECK-RV64VC-NEXT:    mv s2, a1
 ; CHECK-RV64VC-NEXT:  .LBB910_12:
-; CHECK-RV64VC-NEXT:    sub a0, s1, s0
-; CHECK-RV64VC-NEXT:    sltu a1, s1, a0
+; CHECK-RV64VC-NEXT:    sub a0, s2, s1
+; CHECK-RV64VC-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a1, a1, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
 ; CHECK-RV64VC-NEXT:    csrr a1, vlenb
@@ -34905,9 +34905,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, s0, .LBB910_14
+; CHECK-RV64VC-NEXT:    bltu s2, s1, .LBB910_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv s1, s0
+; CHECK-RV64VC-NEXT:    mv s2, s1
 ; CHECK-RV64VC-NEXT:  .LBB910_14:
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -34917,7 +34917,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_P1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64VC-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64VC-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
@@ -35028,7 +35028,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64V-NEXT:    add a1, a1, a2
 ; CHECK-RV64V-NEXT:    sub sp, sp, a1
 ; CHECK-RV64V-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64V-NEXT:    mv s2, a0
+; CHECK-RV64V-NEXT:    mv s0, a0
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
@@ -35042,9 +35042,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
-; CHECK-RV64V-NEXT:    slli a0, s0, 3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    csrr s1, vlenb
+; CHECK-RV64V-NEXT:    slli a0, s1, 3
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -35054,63 +35054,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    slli a0, s0, 4
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a0, s1, 4
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 4
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s1, a6
+; CHECK-RV64V-NEXT:    mv s2, a6
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s1, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    addi a0, sp, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
-; CHECK-RV64V-NEXT:    sub a1, s1, a2
-; CHECK-RV64V-NEXT:    sltu a3, s1, a1
+; CHECK-RV64V-NEXT:    slli a2, s1, 2
+; CHECK-RV64V-NEXT:    sub a1, s2, a2
+; CHECK-RV64V-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a3, a3, a1
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
+; CHECK-RV64V-NEXT:    slli a1, s1, 1
 ; CHECK-RV64V-NEXT:    sub a4, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64V-NEXT:    addi a5, a5, -1
 ; CHECK-RV64V-NEXT:    and a4, a5, a4
-; CHECK-RV64V-NEXT:    sub a5, a4, s0
+; CHECK-RV64V-NEXT:    sub a5, a4, s1
 ; CHECK-RV64V-NEXT:    sltu a6, a4, a5
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB911_2
+; CHECK-RV64V-NEXT:    bltu a4, s1, .LBB911_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
-; CHECK-RV64V-NEXT:    mv a4, s0
+; CHECK-RV64V-NEXT:    mv a4, s1
 ; CHECK-RV64V-NEXT:  .LBB911_2:
 ; CHECK-RV64V-NEXT:    addi a6, a6, -1
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB911_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a3, a1
 ; CHECK-RV64V-NEXT:  .LBB911_4:
-; CHECK-RV64V-NEXT:    sub a7, a3, s0
+; CHECK-RV64V-NEXT:    sub a7, a3, s1
 ; CHECK-RV64V-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64V-NEXT:    and a5, a6, a5
 ; CHECK-RV64V-NEXT:    addi t0, t0, -1
-; CHECK-RV64V-NEXT:    add a6, s2, a0
+; CHECK-RV64V-NEXT:    add a6, s0, a0
 ; CHECK-RV64V-NEXT:    and a0, t0, a7
-; CHECK-RV64V-NEXT:    bltu a3, s0, .LBB911_6
+; CHECK-RV64V-NEXT:    bltu a3, s1, .LBB911_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:  .LBB911_6:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64V-NEXT:    vle64.v v16, (a6)
@@ -35135,25 +35135,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB911_8
+; CHECK-RV64V-NEXT:    bltu s2, a2, .LBB911_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv s1, a2
+; CHECK-RV64V-NEXT:    mv s2, a2
 ; CHECK-RV64V-NEXT:  .LBB911_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64V-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64V-NEXT:    sub a0, s1, a1
-; CHECK-RV64V-NEXT:    sltu a2, s1, a0
+; CHECK-RV64V-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64V-NEXT:    sub a0, s2, a1
+; CHECK-RV64V-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64V-NEXT:    addi a2, a2, -1
 ; CHECK-RV64V-NEXT:    and a0, a2, a0
-; CHECK-RV64V-NEXT:    sub a2, a0, s0
+; CHECK-RV64V-NEXT:    sub a2, a0, s1
 ; CHECK-RV64V-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a2, a3, a2
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB911_10
+; CHECK-RV64V-NEXT:    bltu a0, s1, .LBB911_10
 ; CHECK-RV64V-NEXT:  # %bb.9:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:  .LBB911_10:
 ; CHECK-RV64V-NEXT:    csrr a3, vlenb
 ; CHECK-RV64V-NEXT:    slli a3, a3, 3
@@ -35169,12 +35169,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, a1, .LBB911_12
+; CHECK-RV64V-NEXT:    bltu s2, a1, .LBB911_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv s1, a1
+; CHECK-RV64V-NEXT:    mv s2, a1
 ; CHECK-RV64V-NEXT:  .LBB911_12:
-; CHECK-RV64V-NEXT:    sub a0, s1, s0
-; CHECK-RV64V-NEXT:    sltu a1, s1, a0
+; CHECK-RV64V-NEXT:    sub a0, s2, s1
+; CHECK-RV64V-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64V-NEXT:    addi a1, a1, -1
 ; CHECK-RV64V-NEXT:    and a0, a1, a0
 ; CHECK-RV64V-NEXT:    csrr a1, vlenb
@@ -35185,9 +35185,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, s0, .LBB911_14
+; CHECK-RV64V-NEXT:    bltu s2, s1, .LBB911_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv s1, s0
+; CHECK-RV64V-NEXT:    mv s2, s1
 ; CHECK-RV64V-NEXT:  .LBB911_14:
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -35197,7 +35197,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64V-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64V-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
@@ -35303,7 +35303,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64VC-NEXT:    add a1, a1, a2
 ; CHECK-RV64VC-NEXT:    sub sp, sp, a1
 ; CHECK-RV64VC-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64VC-NEXT:    mv s2, a0
+; CHECK-RV64VC-NEXT:    mv s0, a0
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
@@ -35317,9 +35317,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
-; CHECK-RV64VC-NEXT:    slli a0, s0, 3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    csrr s1, vlenb
+; CHECK-RV64VC-NEXT:    slli a0, s1, 3
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -35329,63 +35329,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    slli a0, s0, 4
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a0, s1, 4
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 4
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s1, a6
+; CHECK-RV64VC-NEXT:    mv s2, a6
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s1, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    addi a0, sp, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
-; CHECK-RV64VC-NEXT:    sub a1, s1, a2
-; CHECK-RV64VC-NEXT:    sltu a3, s1, a1
+; CHECK-RV64VC-NEXT:    slli a2, s1, 2
+; CHECK-RV64VC-NEXT:    sub a1, s2, a2
+; CHECK-RV64VC-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a3, a3, a1
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
+; CHECK-RV64VC-NEXT:    slli a1, s1, 1
 ; CHECK-RV64VC-NEXT:    sub a4, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64VC-NEXT:    addi a5, a5, -1
 ; CHECK-RV64VC-NEXT:    and a4, a4, a5
-; CHECK-RV64VC-NEXT:    sub a5, a4, s0
+; CHECK-RV64VC-NEXT:    sub a5, a4, s1
 ; CHECK-RV64VC-NEXT:    sltu a6, a4, a5
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB911_2
+; CHECK-RV64VC-NEXT:    bltu a4, s1, .LBB911_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
-; CHECK-RV64VC-NEXT:    mv a4, s0
+; CHECK-RV64VC-NEXT:    mv a4, s1
 ; CHECK-RV64VC-NEXT:  .LBB911_2:
 ; CHECK-RV64VC-NEXT:    addi a6, a6, -1
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB911_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a3, a1
 ; CHECK-RV64VC-NEXT:  .LBB911_4:
-; CHECK-RV64VC-NEXT:    sub a7, a3, s0
+; CHECK-RV64VC-NEXT:    sub a7, a3, s1
 ; CHECK-RV64VC-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64VC-NEXT:    and a5, a6, a5
 ; CHECK-RV64VC-NEXT:    addi t0, t0, -1
-; CHECK-RV64VC-NEXT:    add a6, s2, a0
+; CHECK-RV64VC-NEXT:    add a6, s0, a0
 ; CHECK-RV64VC-NEXT:    and a0, t0, a7
-; CHECK-RV64VC-NEXT:    bltu a3, s0, .LBB911_6
+; CHECK-RV64VC-NEXT:    bltu a3, s1, .LBB911_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:  .LBB911_6:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64VC-NEXT:    vle64.v v16, (a6)
@@ -35410,25 +35410,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB911_8
+; CHECK-RV64VC-NEXT:    bltu s2, a2, .LBB911_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv s1, a2
+; CHECK-RV64VC-NEXT:    mv s2, a2
 ; CHECK-RV64VC-NEXT:  .LBB911_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64VC-NEXT:    sub a0, s1, a1
-; CHECK-RV64VC-NEXT:    sltu a2, s1, a0
+; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64VC-NEXT:    sub a0, s2, a1
+; CHECK-RV64VC-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a2, a2, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a2
-; CHECK-RV64VC-NEXT:    sub a2, a0, s0
+; CHECK-RV64VC-NEXT:    sub a2, a0, s1
 ; CHECK-RV64VC-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a2, a2, a3
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB911_10
+; CHECK-RV64VC-NEXT:    bltu a0, s1, .LBB911_10
 ; CHECK-RV64VC-NEXT:  # %bb.9:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:  .LBB911_10:
 ; CHECK-RV64VC-NEXT:    csrr a3, vlenb
 ; CHECK-RV64VC-NEXT:    slli a3, a3, 3
@@ -35444,12 +35444,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, a1, .LBB911_12
+; CHECK-RV64VC-NEXT:    bltu s2, a1, .LBB911_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv s1, a1
+; CHECK-RV64VC-NEXT:    mv s2, a1
 ; CHECK-RV64VC-NEXT:  .LBB911_12:
-; CHECK-RV64VC-NEXT:    sub a0, s1, s0
-; CHECK-RV64VC-NEXT:    sltu a1, s1, a0
+; CHECK-RV64VC-NEXT:    sub a0, s2, s1
+; CHECK-RV64VC-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a1, a1, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
 ; CHECK-RV64VC-NEXT:    csrr a1, vlenb
@@ -35460,9 +35460,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, s0, .LBB911_14
+; CHECK-RV64VC-NEXT:    bltu s2, s1, .LBB911_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv s1, s0
+; CHECK-RV64VC-NEXT:    mv s2, s1
 ; CHECK-RV64VC-NEXT:  .LBB911_14:
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -35472,7 +35472,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_PALL(<vscale x 64
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64VC-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64VC-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
@@ -35583,7 +35583,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a1, a1, a2
 ; CHECK-RV64V-NEXT:    sub sp, sp, a1
 ; CHECK-RV64V-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64V-NEXT:    mv s2, a0
+; CHECK-RV64V-NEXT:    mv s0, a0
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
@@ -35597,9 +35597,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
-; CHECK-RV64V-NEXT:    slli a0, s0, 3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    csrr s1, vlenb
+; CHECK-RV64V-NEXT:    slli a0, s1, 3
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -35609,63 +35609,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    slli a0, s0, 4
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a0, s1, 4
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 4
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s1, a6
+; CHECK-RV64V-NEXT:    mv s2, a6
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s1, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    addi a0, sp, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
-; CHECK-RV64V-NEXT:    sub a1, s1, a2
-; CHECK-RV64V-NEXT:    sltu a3, s1, a1
+; CHECK-RV64V-NEXT:    slli a2, s1, 2
+; CHECK-RV64V-NEXT:    sub a1, s2, a2
+; CHECK-RV64V-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a3, a3, a1
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
+; CHECK-RV64V-NEXT:    slli a1, s1, 1
 ; CHECK-RV64V-NEXT:    sub a4, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64V-NEXT:    addi a5, a5, -1
 ; CHECK-RV64V-NEXT:    and a4, a5, a4
-; CHECK-RV64V-NEXT:    sub a5, a4, s0
+; CHECK-RV64V-NEXT:    sub a5, a4, s1
 ; CHECK-RV64V-NEXT:    sltu a6, a4, a5
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB912_2
+; CHECK-RV64V-NEXT:    bltu a4, s1, .LBB912_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
-; CHECK-RV64V-NEXT:    mv a4, s0
+; CHECK-RV64V-NEXT:    mv a4, s1
 ; CHECK-RV64V-NEXT:  .LBB912_2:
 ; CHECK-RV64V-NEXT:    addi a6, a6, -1
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB912_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a3, a1
 ; CHECK-RV64V-NEXT:  .LBB912_4:
-; CHECK-RV64V-NEXT:    sub a7, a3, s0
+; CHECK-RV64V-NEXT:    sub a7, a3, s1
 ; CHECK-RV64V-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64V-NEXT:    and a5, a6, a5
 ; CHECK-RV64V-NEXT:    addi t0, t0, -1
-; CHECK-RV64V-NEXT:    add a6, s2, a0
+; CHECK-RV64V-NEXT:    add a6, s0, a0
 ; CHECK-RV64V-NEXT:    and a0, t0, a7
-; CHECK-RV64V-NEXT:    bltu a3, s0, .LBB912_6
+; CHECK-RV64V-NEXT:    bltu a3, s1, .LBB912_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:  .LBB912_6:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64V-NEXT:    vle64.v v16, (a6)
@@ -35690,25 +35690,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB912_8
+; CHECK-RV64V-NEXT:    bltu s2, a2, .LBB912_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv s1, a2
+; CHECK-RV64V-NEXT:    mv s2, a2
 ; CHECK-RV64V-NEXT:  .LBB912_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64V-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64V-NEXT:    sub a0, s1, a1
-; CHECK-RV64V-NEXT:    sltu a2, s1, a0
+; CHECK-RV64V-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64V-NEXT:    sub a0, s2, a1
+; CHECK-RV64V-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64V-NEXT:    addi a2, a2, -1
 ; CHECK-RV64V-NEXT:    and a0, a2, a0
-; CHECK-RV64V-NEXT:    sub a2, a0, s0
+; CHECK-RV64V-NEXT:    sub a2, a0, s1
 ; CHECK-RV64V-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a2, a3, a2
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB912_10
+; CHECK-RV64V-NEXT:    bltu a0, s1, .LBB912_10
 ; CHECK-RV64V-NEXT:  # %bb.9:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:  .LBB912_10:
 ; CHECK-RV64V-NEXT:    csrr a3, vlenb
 ; CHECK-RV64V-NEXT:    slli a3, a3, 3
@@ -35724,12 +35724,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, a1, .LBB912_12
+; CHECK-RV64V-NEXT:    bltu s2, a1, .LBB912_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv s1, a1
+; CHECK-RV64V-NEXT:    mv s2, a1
 ; CHECK-RV64V-NEXT:  .LBB912_12:
-; CHECK-RV64V-NEXT:    sub a0, s1, s0
-; CHECK-RV64V-NEXT:    sltu a1, s1, a0
+; CHECK-RV64V-NEXT:    sub a0, s2, s1
+; CHECK-RV64V-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64V-NEXT:    addi a1, a1, -1
 ; CHECK-RV64V-NEXT:    and a0, a1, a0
 ; CHECK-RV64V-NEXT:    csrr a1, vlenb
@@ -35740,9 +35740,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, s0, .LBB912_14
+; CHECK-RV64V-NEXT:    bltu s2, s1, .LBB912_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv s1, s0
+; CHECK-RV64V-NEXT:    mv s2, s1
 ; CHECK-RV64V-NEXT:  .LBB912_14:
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -35752,7 +35752,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64V-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64V-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
@@ -35858,7 +35858,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a1, a1, a2
 ; CHECK-RV64VC-NEXT:    sub sp, sp, a1
 ; CHECK-RV64VC-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64VC-NEXT:    mv s2, a0
+; CHECK-RV64VC-NEXT:    mv s0, a0
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
@@ -35872,9 +35872,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
-; CHECK-RV64VC-NEXT:    slli a0, s0, 3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    csrr s1, vlenb
+; CHECK-RV64VC-NEXT:    slli a0, s1, 3
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -35884,63 +35884,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    slli a0, s0, 4
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a0, s1, 4
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 4
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s1, a6
+; CHECK-RV64VC-NEXT:    mv s2, a6
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s1, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    addi a0, sp, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
-; CHECK-RV64VC-NEXT:    sub a1, s1, a2
-; CHECK-RV64VC-NEXT:    sltu a3, s1, a1
+; CHECK-RV64VC-NEXT:    slli a2, s1, 2
+; CHECK-RV64VC-NEXT:    sub a1, s2, a2
+; CHECK-RV64VC-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a3, a3, a1
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
+; CHECK-RV64VC-NEXT:    slli a1, s1, 1
 ; CHECK-RV64VC-NEXT:    sub a4, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64VC-NEXT:    addi a5, a5, -1
 ; CHECK-RV64VC-NEXT:    and a4, a4, a5
-; CHECK-RV64VC-NEXT:    sub a5, a4, s0
+; CHECK-RV64VC-NEXT:    sub a5, a4, s1
 ; CHECK-RV64VC-NEXT:    sltu a6, a4, a5
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB912_2
+; CHECK-RV64VC-NEXT:    bltu a4, s1, .LBB912_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
-; CHECK-RV64VC-NEXT:    mv a4, s0
+; CHECK-RV64VC-NEXT:    mv a4, s1
 ; CHECK-RV64VC-NEXT:  .LBB912_2:
 ; CHECK-RV64VC-NEXT:    addi a6, a6, -1
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB912_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a3, a1
 ; CHECK-RV64VC-NEXT:  .LBB912_4:
-; CHECK-RV64VC-NEXT:    sub a7, a3, s0
+; CHECK-RV64VC-NEXT:    sub a7, a3, s1
 ; CHECK-RV64VC-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64VC-NEXT:    and a5, a6, a5
 ; CHECK-RV64VC-NEXT:    addi t0, t0, -1
-; CHECK-RV64VC-NEXT:    add a6, s2, a0
+; CHECK-RV64VC-NEXT:    add a6, s0, a0
 ; CHECK-RV64VC-NEXT:    and a0, t0, a7
-; CHECK-RV64VC-NEXT:    bltu a3, s0, .LBB912_6
+; CHECK-RV64VC-NEXT:    bltu a3, s1, .LBB912_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:  .LBB912_6:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64VC-NEXT:    vle64.v v16, (a6)
@@ -35965,25 +35965,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB912_8
+; CHECK-RV64VC-NEXT:    bltu s2, a2, .LBB912_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv s1, a2
+; CHECK-RV64VC-NEXT:    mv s2, a2
 ; CHECK-RV64VC-NEXT:  .LBB912_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64VC-NEXT:    sub a0, s1, a1
-; CHECK-RV64VC-NEXT:    sltu a2, s1, a0
+; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64VC-NEXT:    sub a0, s2, a1
+; CHECK-RV64VC-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a2, a2, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a2
-; CHECK-RV64VC-NEXT:    sub a2, a0, s0
+; CHECK-RV64VC-NEXT:    sub a2, a0, s1
 ; CHECK-RV64VC-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a2, a2, a3
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB912_10
+; CHECK-RV64VC-NEXT:    bltu a0, s1, .LBB912_10
 ; CHECK-RV64VC-NEXT:  # %bb.9:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:  .LBB912_10:
 ; CHECK-RV64VC-NEXT:    csrr a3, vlenb
 ; CHECK-RV64VC-NEXT:    slli a3, a3, 3
@@ -35999,12 +35999,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, a1, .LBB912_12
+; CHECK-RV64VC-NEXT:    bltu s2, a1, .LBB912_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv s1, a1
+; CHECK-RV64VC-NEXT:    mv s2, a1
 ; CHECK-RV64VC-NEXT:  .LBB912_12:
-; CHECK-RV64VC-NEXT:    sub a0, s1, s0
-; CHECK-RV64VC-NEXT:    sltu a1, s1, a0
+; CHECK-RV64VC-NEXT:    sub a0, s2, s1
+; CHECK-RV64VC-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a1, a1, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
 ; CHECK-RV64VC-NEXT:    csrr a1, vlenb
@@ -36015,9 +36015,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, s0, .LBB912_14
+; CHECK-RV64VC-NEXT:    bltu s2, s1, .LBB912_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv s1, s0
+; CHECK-RV64VC-NEXT:    mv s2, s1
 ; CHECK-RV64VC-NEXT:  .LBB912_14:
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -36027,7 +36027,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_S1(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64VC-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64VC-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
@@ -36138,7 +36138,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a1, a1, a2
 ; CHECK-RV64V-NEXT:    sub sp, sp, a1
 ; CHECK-RV64V-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64V-NEXT:    mv s2, a0
+; CHECK-RV64V-NEXT:    mv s0, a0
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
@@ -36152,9 +36152,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
-; CHECK-RV64V-NEXT:    slli a0, s0, 3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    csrr s1, vlenb
+; CHECK-RV64V-NEXT:    slli a0, s1, 3
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -36164,63 +36164,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    slli a0, s0, 4
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a0, s1, 4
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 4
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s1, a6
+; CHECK-RV64V-NEXT:    mv s2, a6
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s1, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    addi a0, sp, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
-; CHECK-RV64V-NEXT:    sub a1, s1, a2
-; CHECK-RV64V-NEXT:    sltu a3, s1, a1
+; CHECK-RV64V-NEXT:    slli a2, s1, 2
+; CHECK-RV64V-NEXT:    sub a1, s2, a2
+; CHECK-RV64V-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a3, a3, a1
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
+; CHECK-RV64V-NEXT:    slli a1, s1, 1
 ; CHECK-RV64V-NEXT:    sub a4, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64V-NEXT:    addi a5, a5, -1
 ; CHECK-RV64V-NEXT:    and a4, a5, a4
-; CHECK-RV64V-NEXT:    sub a5, a4, s0
+; CHECK-RV64V-NEXT:    sub a5, a4, s1
 ; CHECK-RV64V-NEXT:    sltu a6, a4, a5
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB913_2
+; CHECK-RV64V-NEXT:    bltu a4, s1, .LBB913_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
-; CHECK-RV64V-NEXT:    mv a4, s0
+; CHECK-RV64V-NEXT:    mv a4, s1
 ; CHECK-RV64V-NEXT:  .LBB913_2:
 ; CHECK-RV64V-NEXT:    addi a6, a6, -1
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB913_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a3, a1
 ; CHECK-RV64V-NEXT:  .LBB913_4:
-; CHECK-RV64V-NEXT:    sub a7, a3, s0
+; CHECK-RV64V-NEXT:    sub a7, a3, s1
 ; CHECK-RV64V-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64V-NEXT:    and a5, a6, a5
 ; CHECK-RV64V-NEXT:    addi t0, t0, -1
-; CHECK-RV64V-NEXT:    add a6, s2, a0
+; CHECK-RV64V-NEXT:    add a6, s0, a0
 ; CHECK-RV64V-NEXT:    and a0, t0, a7
-; CHECK-RV64V-NEXT:    bltu a3, s0, .LBB913_6
+; CHECK-RV64V-NEXT:    bltu a3, s1, .LBB913_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:  .LBB913_6:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64V-NEXT:    vle64.v v16, (a6)
@@ -36245,25 +36245,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB913_8
+; CHECK-RV64V-NEXT:    bltu s2, a2, .LBB913_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv s1, a2
+; CHECK-RV64V-NEXT:    mv s2, a2
 ; CHECK-RV64V-NEXT:  .LBB913_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64V-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64V-NEXT:    sub a0, s1, a1
-; CHECK-RV64V-NEXT:    sltu a2, s1, a0
+; CHECK-RV64V-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64V-NEXT:    sub a0, s2, a1
+; CHECK-RV64V-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64V-NEXT:    addi a2, a2, -1
 ; CHECK-RV64V-NEXT:    and a0, a2, a0
-; CHECK-RV64V-NEXT:    sub a2, a0, s0
+; CHECK-RV64V-NEXT:    sub a2, a0, s1
 ; CHECK-RV64V-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a2, a3, a2
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB913_10
+; CHECK-RV64V-NEXT:    bltu a0, s1, .LBB913_10
 ; CHECK-RV64V-NEXT:  # %bb.9:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:  .LBB913_10:
 ; CHECK-RV64V-NEXT:    csrr a3, vlenb
 ; CHECK-RV64V-NEXT:    slli a3, a3, 3
@@ -36279,12 +36279,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, a1, .LBB913_12
+; CHECK-RV64V-NEXT:    bltu s2, a1, .LBB913_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv s1, a1
+; CHECK-RV64V-NEXT:    mv s2, a1
 ; CHECK-RV64V-NEXT:  .LBB913_12:
-; CHECK-RV64V-NEXT:    sub a0, s1, s0
-; CHECK-RV64V-NEXT:    sltu a1, s1, a0
+; CHECK-RV64V-NEXT:    sub a0, s2, s1
+; CHECK-RV64V-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64V-NEXT:    addi a1, a1, -1
 ; CHECK-RV64V-NEXT:    and a0, a1, a0
 ; CHECK-RV64V-NEXT:    csrr a1, vlenb
@@ -36295,9 +36295,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, s0, .LBB913_14
+; CHECK-RV64V-NEXT:    bltu s2, s1, .LBB913_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv s1, s0
+; CHECK-RV64V-NEXT:    mv s2, s1
 ; CHECK-RV64V-NEXT:  .LBB913_14:
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -36307,7 +36307,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64V-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64V-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
@@ -36413,7 +36413,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a1, a1, a2
 ; CHECK-RV64VC-NEXT:    sub sp, sp, a1
 ; CHECK-RV64VC-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64VC-NEXT:    mv s2, a0
+; CHECK-RV64VC-NEXT:    mv s0, a0
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
@@ -36427,9 +36427,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
-; CHECK-RV64VC-NEXT:    slli a0, s0, 3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    csrr s1, vlenb
+; CHECK-RV64VC-NEXT:    slli a0, s1, 3
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -36439,63 +36439,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    slli a0, s0, 4
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a0, s1, 4
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 4
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s1, a6
+; CHECK-RV64VC-NEXT:    mv s2, a6
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s1, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    addi a0, sp, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
-; CHECK-RV64VC-NEXT:    sub a1, s1, a2
-; CHECK-RV64VC-NEXT:    sltu a3, s1, a1
+; CHECK-RV64VC-NEXT:    slli a2, s1, 2
+; CHECK-RV64VC-NEXT:    sub a1, s2, a2
+; CHECK-RV64VC-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a3, a3, a1
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
+; CHECK-RV64VC-NEXT:    slli a1, s1, 1
 ; CHECK-RV64VC-NEXT:    sub a4, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64VC-NEXT:    addi a5, a5, -1
 ; CHECK-RV64VC-NEXT:    and a4, a4, a5
-; CHECK-RV64VC-NEXT:    sub a5, a4, s0
+; CHECK-RV64VC-NEXT:    sub a5, a4, s1
 ; CHECK-RV64VC-NEXT:    sltu a6, a4, a5
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB913_2
+; CHECK-RV64VC-NEXT:    bltu a4, s1, .LBB913_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
-; CHECK-RV64VC-NEXT:    mv a4, s0
+; CHECK-RV64VC-NEXT:    mv a4, s1
 ; CHECK-RV64VC-NEXT:  .LBB913_2:
 ; CHECK-RV64VC-NEXT:    addi a6, a6, -1
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB913_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a3, a1
 ; CHECK-RV64VC-NEXT:  .LBB913_4:
-; CHECK-RV64VC-NEXT:    sub a7, a3, s0
+; CHECK-RV64VC-NEXT:    sub a7, a3, s1
 ; CHECK-RV64VC-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64VC-NEXT:    and a5, a6, a5
 ; CHECK-RV64VC-NEXT:    addi t0, t0, -1
-; CHECK-RV64VC-NEXT:    add a6, s2, a0
+; CHECK-RV64VC-NEXT:    add a6, s0, a0
 ; CHECK-RV64VC-NEXT:    and a0, t0, a7
-; CHECK-RV64VC-NEXT:    bltu a3, s0, .LBB913_6
+; CHECK-RV64VC-NEXT:    bltu a3, s1, .LBB913_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:  .LBB913_6:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64VC-NEXT:    vle64.v v16, (a6)
@@ -36520,25 +36520,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB913_8
+; CHECK-RV64VC-NEXT:    bltu s2, a2, .LBB913_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv s1, a2
+; CHECK-RV64VC-NEXT:    mv s2, a2
 ; CHECK-RV64VC-NEXT:  .LBB913_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64VC-NEXT:    sub a0, s1, a1
-; CHECK-RV64VC-NEXT:    sltu a2, s1, a0
+; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64VC-NEXT:    sub a0, s2, a1
+; CHECK-RV64VC-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a2, a2, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a2
-; CHECK-RV64VC-NEXT:    sub a2, a0, s0
+; CHECK-RV64VC-NEXT:    sub a2, a0, s1
 ; CHECK-RV64VC-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a2, a2, a3
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB913_10
+; CHECK-RV64VC-NEXT:    bltu a0, s1, .LBB913_10
 ; CHECK-RV64VC-NEXT:  # %bb.9:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:  .LBB913_10:
 ; CHECK-RV64VC-NEXT:    csrr a3, vlenb
 ; CHECK-RV64VC-NEXT:    slli a3, a3, 3
@@ -36554,12 +36554,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, a1, .LBB913_12
+; CHECK-RV64VC-NEXT:    bltu s2, a1, .LBB913_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv s1, a1
+; CHECK-RV64VC-NEXT:    mv s2, a1
 ; CHECK-RV64VC-NEXT:  .LBB913_12:
-; CHECK-RV64VC-NEXT:    sub a0, s1, s0
-; CHECK-RV64VC-NEXT:    sltu a1, s1, a0
+; CHECK-RV64VC-NEXT:    sub a0, s2, s1
+; CHECK-RV64VC-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a1, a1, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
 ; CHECK-RV64VC-NEXT:    csrr a1, vlenb
@@ -36570,9 +36570,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, s0, .LBB913_14
+; CHECK-RV64VC-NEXT:    bltu s2, s1, .LBB913_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv s1, s0
+; CHECK-RV64VC-NEXT:    mv s2, s1
 ; CHECK-RV64VC-NEXT:  .LBB913_14:
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -36582,7 +36582,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_ALL(<vscale x 64 x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64VC-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64VC-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
@@ -36692,7 +36692,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64V-NEXT:    add a1, a1, a2
 ; CHECK-RV64V-NEXT:    sub sp, sp, a1
 ; CHECK-RV64V-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64V-NEXT:    mv s2, a0
+; CHECK-RV64V-NEXT:    mv s0, a0
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
@@ -36706,9 +36706,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
-; CHECK-RV64V-NEXT:    slli a0, s0, 3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    csrr s1, vlenb
+; CHECK-RV64V-NEXT:    slli a0, s1, 3
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -36718,63 +36718,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    slli a0, s0, 4
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a0, s1, 4
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 4
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s1, a6
+; CHECK-RV64V-NEXT:    mv s2, a6
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s1, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    addi a0, sp, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
-; CHECK-RV64V-NEXT:    sub a1, s1, a2
-; CHECK-RV64V-NEXT:    sltu a3, s1, a1
+; CHECK-RV64V-NEXT:    slli a2, s1, 2
+; CHECK-RV64V-NEXT:    sub a1, s2, a2
+; CHECK-RV64V-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a3, a3, a1
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
+; CHECK-RV64V-NEXT:    slli a1, s1, 1
 ; CHECK-RV64V-NEXT:    sub a4, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64V-NEXT:    addi a5, a5, -1
 ; CHECK-RV64V-NEXT:    and a4, a5, a4
-; CHECK-RV64V-NEXT:    sub a5, a4, s0
+; CHECK-RV64V-NEXT:    sub a5, a4, s1
 ; CHECK-RV64V-NEXT:    sltu a6, a4, a5
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB914_2
+; CHECK-RV64V-NEXT:    bltu a4, s1, .LBB914_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
-; CHECK-RV64V-NEXT:    mv a4, s0
+; CHECK-RV64V-NEXT:    mv a4, s1
 ; CHECK-RV64V-NEXT:  .LBB914_2:
 ; CHECK-RV64V-NEXT:    addi a6, a6, -1
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB914_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a3, a1
 ; CHECK-RV64V-NEXT:  .LBB914_4:
-; CHECK-RV64V-NEXT:    sub a7, a3, s0
+; CHECK-RV64V-NEXT:    sub a7, a3, s1
 ; CHECK-RV64V-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64V-NEXT:    and a5, a6, a5
 ; CHECK-RV64V-NEXT:    addi t0, t0, -1
-; CHECK-RV64V-NEXT:    add a6, s2, a0
+; CHECK-RV64V-NEXT:    add a6, s0, a0
 ; CHECK-RV64V-NEXT:    and a0, t0, a7
-; CHECK-RV64V-NEXT:    bltu a3, s0, .LBB914_6
+; CHECK-RV64V-NEXT:    bltu a3, s1, .LBB914_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:  .LBB914_6:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64V-NEXT:    vle64.v v16, (a6)
@@ -36799,25 +36799,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB914_8
+; CHECK-RV64V-NEXT:    bltu s2, a2, .LBB914_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv s1, a2
+; CHECK-RV64V-NEXT:    mv s2, a2
 ; CHECK-RV64V-NEXT:  .LBB914_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64V-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64V-NEXT:    sub a0, s1, a1
-; CHECK-RV64V-NEXT:    sltu a2, s1, a0
+; CHECK-RV64V-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64V-NEXT:    sub a0, s2, a1
+; CHECK-RV64V-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64V-NEXT:    addi a2, a2, -1
 ; CHECK-RV64V-NEXT:    and a0, a2, a0
-; CHECK-RV64V-NEXT:    sub a2, a0, s0
+; CHECK-RV64V-NEXT:    sub a2, a0, s1
 ; CHECK-RV64V-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64V-NEXT:    addi a3, a3, -1
 ; CHECK-RV64V-NEXT:    and a2, a3, a2
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB914_10
+; CHECK-RV64V-NEXT:    bltu a0, s1, .LBB914_10
 ; CHECK-RV64V-NEXT:  # %bb.9:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s1
 ; CHECK-RV64V-NEXT:  .LBB914_10:
 ; CHECK-RV64V-NEXT:    csrr a3, vlenb
 ; CHECK-RV64V-NEXT:    slli a3, a3, 3
@@ -36833,12 +36833,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, a1, .LBB914_12
+; CHECK-RV64V-NEXT:    bltu s2, a1, .LBB914_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv s1, a1
+; CHECK-RV64V-NEXT:    mv s2, a1
 ; CHECK-RV64V-NEXT:  .LBB914_12:
-; CHECK-RV64V-NEXT:    sub a0, s1, s0
-; CHECK-RV64V-NEXT:    sltu a1, s1, a0
+; CHECK-RV64V-NEXT:    sub a0, s2, s1
+; CHECK-RV64V-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64V-NEXT:    addi a1, a1, -1
 ; CHECK-RV64V-NEXT:    and a0, a1, a0
 ; CHECK-RV64V-NEXT:    csrr a1, vlenb
@@ -36849,9 +36849,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64V-NEXT:    bltu s1, s0, .LBB914_14
+; CHECK-RV64V-NEXT:    bltu s2, s1, .LBB914_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv s1, s0
+; CHECK-RV64V-NEXT:    mv s2, s1
 ; CHECK-RV64V-NEXT:  .LBB914_14:
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -36861,7 +36861,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64V-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64V-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
@@ -36967,7 +36967,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64VC-NEXT:    add a1, a1, a2
 ; CHECK-RV64VC-NEXT:    sub sp, sp, a1
 ; CHECK-RV64VC-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x30, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 48 * vlenb
-; CHECK-RV64VC-NEXT:    mv s2, a0
+; CHECK-RV64VC-NEXT:    mv s0, a0
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
@@ -36981,9 +36981,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
-; CHECK-RV64VC-NEXT:    slli a0, s0, 3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    csrr s1, vlenb
+; CHECK-RV64VC-NEXT:    slli a0, s1, 3
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -36993,63 +36993,63 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    slli a0, s0, 4
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a0, s1, 4
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 4
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s1, a6
+; CHECK-RV64VC-NEXT:    mv s2, a6
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s1, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    addi a0, sp, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
-; CHECK-RV64VC-NEXT:    sub a1, s1, a2
-; CHECK-RV64VC-NEXT:    sltu a3, s1, a1
+; CHECK-RV64VC-NEXT:    slli a2, s1, 2
+; CHECK-RV64VC-NEXT:    sub a1, s2, a2
+; CHECK-RV64VC-NEXT:    sltu a3, s2, a1
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a3, a3, a1
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
+; CHECK-RV64VC-NEXT:    slli a1, s1, 1
 ; CHECK-RV64VC-NEXT:    sub a4, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a5, a3, a4
 ; CHECK-RV64VC-NEXT:    addi a5, a5, -1
 ; CHECK-RV64VC-NEXT:    and a4, a4, a5
-; CHECK-RV64VC-NEXT:    sub a5, a4, s0
+; CHECK-RV64VC-NEXT:    sub a5, a4, s1
 ; CHECK-RV64VC-NEXT:    sltu a6, a4, a5
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB914_2
+; CHECK-RV64VC-NEXT:    bltu a4, s1, .LBB914_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
-; CHECK-RV64VC-NEXT:    mv a4, s0
+; CHECK-RV64VC-NEXT:    mv a4, s1
 ; CHECK-RV64VC-NEXT:  .LBB914_2:
 ; CHECK-RV64VC-NEXT:    addi a6, a6, -1
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB914_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a3, a1
 ; CHECK-RV64VC-NEXT:  .LBB914_4:
-; CHECK-RV64VC-NEXT:    sub a7, a3, s0
+; CHECK-RV64VC-NEXT:    sub a7, a3, s1
 ; CHECK-RV64VC-NEXT:    sltu t0, a3, a7
 ; CHECK-RV64VC-NEXT:    and a5, a6, a5
 ; CHECK-RV64VC-NEXT:    addi t0, t0, -1
-; CHECK-RV64VC-NEXT:    add a6, s2, a0
+; CHECK-RV64VC-NEXT:    add a6, s0, a0
 ; CHECK-RV64VC-NEXT:    and a0, t0, a7
-; CHECK-RV64VC-NEXT:    bltu a3, s0, .LBB914_6
+; CHECK-RV64VC-NEXT:    bltu a3, s1, .LBB914_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:  .LBB914_6:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e64, m8, ta, ma
 ; CHECK-RV64VC-NEXT:    vle64.v v16, (a6)
@@ -37074,25 +37074,25 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v12, (zero), v24
-; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB914_8
+; CHECK-RV64VC-NEXT:    bltu s2, a2, .LBB914_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv s1, a2
+; CHECK-RV64VC-NEXT:    mv s2, a2
 ; CHECK-RV64VC-NEXT:  .LBB914_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a5, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v15, (zero), v16
-; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s2)
-; CHECK-RV64VC-NEXT:    sub a0, s1, a1
-; CHECK-RV64VC-NEXT:    sltu a2, s1, a0
+; CHECK-RV64VC-NEXT:    vl8re64.v v16, (s0)
+; CHECK-RV64VC-NEXT:    sub a0, s2, a1
+; CHECK-RV64VC-NEXT:    sltu a2, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a2, a2, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a2
-; CHECK-RV64VC-NEXT:    sub a2, a0, s0
+; CHECK-RV64VC-NEXT:    sub a2, a0, s1
 ; CHECK-RV64VC-NEXT:    sltu a3, a0, a2
 ; CHECK-RV64VC-NEXT:    addi a3, a3, -1
 ; CHECK-RV64VC-NEXT:    and a2, a2, a3
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB914_10
+; CHECK-RV64VC-NEXT:    bltu a0, s1, .LBB914_10
 ; CHECK-RV64VC-NEXT:  # %bb.9:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s1
 ; CHECK-RV64VC-NEXT:  .LBB914_10:
 ; CHECK-RV64VC-NEXT:    csrr a3, vlenb
 ; CHECK-RV64VC-NEXT:    slli a3, a3, 3
@@ -37108,12 +37108,12 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v10, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, a1, .LBB914_12
+; CHECK-RV64VC-NEXT:    bltu s2, a1, .LBB914_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv s1, a1
+; CHECK-RV64VC-NEXT:    mv s2, a1
 ; CHECK-RV64VC-NEXT:  .LBB914_12:
-; CHECK-RV64VC-NEXT:    sub a0, s1, s0
-; CHECK-RV64VC-NEXT:    sltu a1, s1, a0
+; CHECK-RV64VC-NEXT:    sub a0, s2, s1
+; CHECK-RV64VC-NEXT:    sltu a1, s2, a0
 ; CHECK-RV64VC-NEXT:    addi a1, a1, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
 ; CHECK-RV64VC-NEXT:    csrr a1, vlenb
@@ -37124,9 +37124,9 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v9, (zero), v16
-; CHECK-RV64VC-NEXT:    bltu s1, s0, .LBB914_14
+; CHECK-RV64VC-NEXT:    bltu s2, s1, .LBB914_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv s1, s0
+; CHECK-RV64VC-NEXT:    mv s2, s1
 ; CHECK-RV64VC-NEXT:  .LBB914_14:
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -37136,7 +37136,7 @@ define <vscale x 64 x i8> @test_nontemporal_vp_gather_nxv64i8_DEFAULT(<vscale x
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vl8r.v v16, (a0) # vscale x 64-byte Folded Reload
-; CHECK-RV64VC-NEXT:    vsetvli zero, s1, e8, m1, ta, ma
+; CHECK-RV64VC-NEXT:    vsetvli zero, s2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vluxei64.v v8, (zero), v16
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
@@ -37263,12 +37263,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    add a1, sp, a1
 ; CHECK-RV64V-NEXT:    addi a1, a1, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s2, a0
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
+; CHECK-RV64V-NEXT:    mv s0, a0
+; CHECK-RV64V-NEXT:    csrr s2, vlenb
 ; CHECK-RV64V-NEXT:    li a1, 48
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -37279,17 +37279,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s2, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -37300,30 +37300,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
+; CHECK-RV64V-NEXT:    slli a2, s2, 2
 ; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB915_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
 ; CHECK-RV64V-NEXT:    mv a3, a2
 ; CHECK-RV64V-NEXT:  .LBB915_2:
-; CHECK-RV64V-NEXT:    slli a5, s0, 4
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
-; CHECK-RV64V-NEXT:    slli a6, s0, 3
+; CHECK-RV64V-NEXT:    slli a5, s2, 4
+; CHECK-RV64V-NEXT:    slli a1, s2, 1
+; CHECK-RV64V-NEXT:    slli a6, s2, 3
 ; CHECK-RV64V-NEXT:    mv a4, a3
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB915_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a4, a1
 ; CHECK-RV64V-NEXT:  .LBB915_4:
-; CHECK-RV64V-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64V-NEXT:    add a7, s2, a0
-; CHECK-RV64V-NEXT:    add a5, s2, a5
-; CHECK-RV64V-NEXT:    add a6, s2, a6
+; CHECK-RV64V-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64V-NEXT:    add a7, s0, a0
+; CHECK-RV64V-NEXT:    add a5, s0, a5
+; CHECK-RV64V-NEXT:    add a6, s0, a6
 ; CHECK-RV64V-NEXT:    mv a0, a4
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB915_6
+; CHECK-RV64V-NEXT:    bltu a4, s2, .LBB915_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:  .LBB915_6:
 ; CHECK-RV64V-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64V-NEXT:    csrr a7, vlenb
@@ -37351,7 +37351,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64V-NEXT:    sub a0, a4, s0
+; CHECK-RV64V-NEXT:    sub a0, a4, s2
 ; CHECK-RV64V-NEXT:    sub a5, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64V-NEXT:    sltu a3, a3, a5
@@ -37363,14 +37363,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64V-NEXT:    mv a3, a0
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB915_8
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB915_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB915_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64V-NEXT:    sub a3, a0, s0
+; CHECK-RV64V-NEXT:    sub a3, a0, s2
 ; CHECK-RV64V-NEXT:    sub a2, s1, a2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64V-NEXT:    sltu a4, s1, a2
@@ -37389,9 +37389,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    mv a2, a1
 ; CHECK-RV64V-NEXT:  .LBB915_10:
 ; CHECK-RV64V-NEXT:    mv a3, a2
-; CHECK-RV64V-NEXT:    bltu a2, s0, .LBB915_12
+; CHECK-RV64V-NEXT:    bltu a2, s2, .LBB915_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB915_12:
 ; CHECK-RV64V-NEXT:    csrr a4, vlenb
 ; CHECK-RV64V-NEXT:    slli a4, a4, 3
@@ -37401,7 +37401,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64V-NEXT:    sub a3, a2, s0
+; CHECK-RV64V-NEXT:    sub a3, a2, s2
 ; CHECK-RV64V-NEXT:    sub a1, a0, a1
 ; CHECK-RV64V-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
@@ -37421,9 +37421,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB915_14
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB915_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv a1, s0
+; CHECK-RV64V-NEXT:    mv a1, s2
 ; CHECK-RV64V-NEXT:  .LBB915_14:
 ; CHECK-RV64V-NEXT:    csrr a2, vlenb
 ; CHECK-RV64V-NEXT:    slli a2, a2, 5
@@ -37433,7 +37433,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.p1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64V-NEXT:    sub a1, a0, s0
+; CHECK-RV64V-NEXT:    sub a1, a0, s2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64V-NEXT:    addi a0, a0, -1
 ; CHECK-RV64V-NEXT:    and a0, a0, a1
@@ -37573,12 +37573,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    add a1, a1, sp
 ; CHECK-RV64VC-NEXT:    addi a1, a1, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s2, a0
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
+; CHECK-RV64VC-NEXT:    mv s0, a0
+; CHECK-RV64VC-NEXT:    csrr s2, vlenb
 ; CHECK-RV64VC-NEXT:    li a1, 48
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -37589,17 +37589,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s2, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -37610,30 +37610,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
+; CHECK-RV64VC-NEXT:    slli a2, s2, 2
 ; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB915_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
 ; CHECK-RV64VC-NEXT:  .LBB915_2:
-; CHECK-RV64VC-NEXT:    slli a5, s0, 4
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
-; CHECK-RV64VC-NEXT:    slli a6, s0, 3
+; CHECK-RV64VC-NEXT:    slli a5, s2, 4
+; CHECK-RV64VC-NEXT:    slli a1, s2, 1
+; CHECK-RV64VC-NEXT:    slli a6, s2, 3
 ; CHECK-RV64VC-NEXT:    mv a4, a3
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB915_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a4, a1
 ; CHECK-RV64VC-NEXT:  .LBB915_4:
-; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64VC-NEXT:    add a7, s2, a0
-; CHECK-RV64VC-NEXT:    add a5, a5, s2
-; CHECK-RV64VC-NEXT:    add a6, a6, s2
+; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64VC-NEXT:    add a7, s0, a0
+; CHECK-RV64VC-NEXT:    add a5, a5, s0
+; CHECK-RV64VC-NEXT:    add a6, a6, s0
 ; CHECK-RV64VC-NEXT:    mv a0, a4
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB915_6
+; CHECK-RV64VC-NEXT:    bltu a4, s2, .LBB915_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:  .LBB915_6:
 ; CHECK-RV64VC-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64VC-NEXT:    csrr a7, vlenb
@@ -37661,7 +37661,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64VC-NEXT:    sub a0, a4, s0
+; CHECK-RV64VC-NEXT:    sub a0, a4, s2
 ; CHECK-RV64VC-NEXT:    sub a5, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64VC-NEXT:    sltu a3, a3, a5
@@ -37673,14 +37673,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64VC-NEXT:    mv a3, a0
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB915_8
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB915_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB915_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64VC-NEXT:    sub a3, a0, s0
+; CHECK-RV64VC-NEXT:    sub a3, a0, s2
 ; CHECK-RV64VC-NEXT:    sub a2, s1, a2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64VC-NEXT:    sltu a4, s1, a2
@@ -37699,9 +37699,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    mv a2, a1
 ; CHECK-RV64VC-NEXT:  .LBB915_10:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
-; CHECK-RV64VC-NEXT:    bltu a2, s0, .LBB915_12
+; CHECK-RV64VC-NEXT:    bltu a2, s2, .LBB915_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB915_12:
 ; CHECK-RV64VC-NEXT:    csrr a4, vlenb
 ; CHECK-RV64VC-NEXT:    slli a4, a4, 3
@@ -37711,7 +37711,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a3, a2, s0
+; CHECK-RV64VC-NEXT:    sub a3, a2, s2
 ; CHECK-RV64VC-NEXT:    sub a1, a0, a1
 ; CHECK-RV64VC-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
@@ -37731,9 +37731,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB915_14
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB915_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv a1, s0
+; CHECK-RV64VC-NEXT:    mv a1, s2
 ; CHECK-RV64VC-NEXT:  .LBB915_14:
 ; CHECK-RV64VC-NEXT:    csrr a2, vlenb
 ; CHECK-RV64VC-NEXT:    slli a2, a2, 5
@@ -37743,7 +37743,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_P1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.p1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a1, a0, s0
+; CHECK-RV64VC-NEXT:    sub a1, a0, s2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64VC-NEXT:    addi a0, a0, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
@@ -37888,12 +37888,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    add a1, sp, a1
 ; CHECK-RV64V-NEXT:    addi a1, a1, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s2, a0
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
+; CHECK-RV64V-NEXT:    mv s0, a0
+; CHECK-RV64V-NEXT:    csrr s2, vlenb
 ; CHECK-RV64V-NEXT:    li a1, 48
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -37904,17 +37904,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s2, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -37925,30 +37925,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
+; CHECK-RV64V-NEXT:    slli a2, s2, 2
 ; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB916_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
 ; CHECK-RV64V-NEXT:    mv a3, a2
 ; CHECK-RV64V-NEXT:  .LBB916_2:
-; CHECK-RV64V-NEXT:    slli a5, s0, 4
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
-; CHECK-RV64V-NEXT:    slli a6, s0, 3
+; CHECK-RV64V-NEXT:    slli a5, s2, 4
+; CHECK-RV64V-NEXT:    slli a1, s2, 1
+; CHECK-RV64V-NEXT:    slli a6, s2, 3
 ; CHECK-RV64V-NEXT:    mv a4, a3
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB916_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a4, a1
 ; CHECK-RV64V-NEXT:  .LBB916_4:
-; CHECK-RV64V-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64V-NEXT:    add a7, s2, a0
-; CHECK-RV64V-NEXT:    add a5, s2, a5
-; CHECK-RV64V-NEXT:    add a6, s2, a6
+; CHECK-RV64V-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64V-NEXT:    add a7, s0, a0
+; CHECK-RV64V-NEXT:    add a5, s0, a5
+; CHECK-RV64V-NEXT:    add a6, s0, a6
 ; CHECK-RV64V-NEXT:    mv a0, a4
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB916_6
+; CHECK-RV64V-NEXT:    bltu a4, s2, .LBB916_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:  .LBB916_6:
 ; CHECK-RV64V-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64V-NEXT:    csrr a7, vlenb
@@ -37976,7 +37976,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64V-NEXT:    sub a0, a4, s0
+; CHECK-RV64V-NEXT:    sub a0, a4, s2
 ; CHECK-RV64V-NEXT:    sub a5, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64V-NEXT:    sltu a3, a3, a5
@@ -37988,14 +37988,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64V-NEXT:    mv a3, a0
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB916_8
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB916_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB916_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64V-NEXT:    sub a3, a0, s0
+; CHECK-RV64V-NEXT:    sub a3, a0, s2
 ; CHECK-RV64V-NEXT:    sub a2, s1, a2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64V-NEXT:    sltu a4, s1, a2
@@ -38014,9 +38014,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    mv a2, a1
 ; CHECK-RV64V-NEXT:  .LBB916_10:
 ; CHECK-RV64V-NEXT:    mv a3, a2
-; CHECK-RV64V-NEXT:    bltu a2, s0, .LBB916_12
+; CHECK-RV64V-NEXT:    bltu a2, s2, .LBB916_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB916_12:
 ; CHECK-RV64V-NEXT:    csrr a4, vlenb
 ; CHECK-RV64V-NEXT:    slli a4, a4, 3
@@ -38026,7 +38026,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64V-NEXT:    sub a3, a2, s0
+; CHECK-RV64V-NEXT:    sub a3, a2, s2
 ; CHECK-RV64V-NEXT:    sub a1, a0, a1
 ; CHECK-RV64V-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
@@ -38046,9 +38046,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB916_14
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB916_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv a1, s0
+; CHECK-RV64V-NEXT:    mv a1, s2
 ; CHECK-RV64V-NEXT:  .LBB916_14:
 ; CHECK-RV64V-NEXT:    csrr a2, vlenb
 ; CHECK-RV64V-NEXT:    slli a2, a2, 5
@@ -38058,7 +38058,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64V-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.pall
 ; CHECK-RV64V-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64V-NEXT:    sub a1, a0, s0
+; CHECK-RV64V-NEXT:    sub a1, a0, s2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64V-NEXT:    addi a0, a0, -1
 ; CHECK-RV64V-NEXT:    and a0, a0, a1
@@ -38198,12 +38198,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    add a1, a1, sp
 ; CHECK-RV64VC-NEXT:    addi a1, a1, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s2, a0
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
+; CHECK-RV64VC-NEXT:    mv s0, a0
+; CHECK-RV64VC-NEXT:    csrr s2, vlenb
 ; CHECK-RV64VC-NEXT:    li a1, 48
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -38214,17 +38214,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s2, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -38235,30 +38235,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
+; CHECK-RV64VC-NEXT:    slli a2, s2, 2
 ; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB916_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
 ; CHECK-RV64VC-NEXT:  .LBB916_2:
-; CHECK-RV64VC-NEXT:    slli a5, s0, 4
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
-; CHECK-RV64VC-NEXT:    slli a6, s0, 3
+; CHECK-RV64VC-NEXT:    slli a5, s2, 4
+; CHECK-RV64VC-NEXT:    slli a1, s2, 1
+; CHECK-RV64VC-NEXT:    slli a6, s2, 3
 ; CHECK-RV64VC-NEXT:    mv a4, a3
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB916_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a4, a1
 ; CHECK-RV64VC-NEXT:  .LBB916_4:
-; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64VC-NEXT:    add a7, s2, a0
-; CHECK-RV64VC-NEXT:    add a5, a5, s2
-; CHECK-RV64VC-NEXT:    add a6, a6, s2
+; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64VC-NEXT:    add a7, s0, a0
+; CHECK-RV64VC-NEXT:    add a5, a5, s0
+; CHECK-RV64VC-NEXT:    add a6, a6, s0
 ; CHECK-RV64VC-NEXT:    mv a0, a4
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB916_6
+; CHECK-RV64VC-NEXT:    bltu a4, s2, .LBB916_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:  .LBB916_6:
 ; CHECK-RV64VC-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64VC-NEXT:    csrr a7, vlenb
@@ -38286,7 +38286,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64VC-NEXT:    sub a0, a4, s0
+; CHECK-RV64VC-NEXT:    sub a0, a4, s2
 ; CHECK-RV64VC-NEXT:    sub a5, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64VC-NEXT:    sltu a3, a3, a5
@@ -38298,14 +38298,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64VC-NEXT:    mv a3, a0
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB916_8
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB916_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB916_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64VC-NEXT:    sub a3, a0, s0
+; CHECK-RV64VC-NEXT:    sub a3, a0, s2
 ; CHECK-RV64VC-NEXT:    sub a2, s1, a2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64VC-NEXT:    sltu a4, s1, a2
@@ -38324,9 +38324,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    mv a2, a1
 ; CHECK-RV64VC-NEXT:  .LBB916_10:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
-; CHECK-RV64VC-NEXT:    bltu a2, s0, .LBB916_12
+; CHECK-RV64VC-NEXT:    bltu a2, s2, .LBB916_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB916_12:
 ; CHECK-RV64VC-NEXT:    csrr a4, vlenb
 ; CHECK-RV64VC-NEXT:    slli a4, a4, 3
@@ -38336,7 +38336,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a3, a2, s0
+; CHECK-RV64VC-NEXT:    sub a3, a2, s2
 ; CHECK-RV64VC-NEXT:    sub a1, a0, a1
 ; CHECK-RV64VC-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
@@ -38356,9 +38356,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB916_14
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB916_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv a1, s0
+; CHECK-RV64VC-NEXT:    mv a1, s2
 ; CHECK-RV64VC-NEXT:  .LBB916_14:
 ; CHECK-RV64VC-NEXT:    csrr a2, vlenb
 ; CHECK-RV64VC-NEXT:    slli a2, a2, 5
@@ -38368,7 +38368,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_PALL(<vscale x 64 x i8> %val, <
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.pall
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a1, a0, s0
+; CHECK-RV64VC-NEXT:    sub a1, a0, s2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64VC-NEXT:    addi a0, a0, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
@@ -38513,12 +38513,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    add a1, sp, a1
 ; CHECK-RV64V-NEXT:    addi a1, a1, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s2, a0
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
+; CHECK-RV64V-NEXT:    mv s0, a0
+; CHECK-RV64V-NEXT:    csrr s2, vlenb
 ; CHECK-RV64V-NEXT:    li a1, 48
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -38529,17 +38529,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s2, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -38550,30 +38550,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
+; CHECK-RV64V-NEXT:    slli a2, s2, 2
 ; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB917_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
 ; CHECK-RV64V-NEXT:    mv a3, a2
 ; CHECK-RV64V-NEXT:  .LBB917_2:
-; CHECK-RV64V-NEXT:    slli a5, s0, 4
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
-; CHECK-RV64V-NEXT:    slli a6, s0, 3
+; CHECK-RV64V-NEXT:    slli a5, s2, 4
+; CHECK-RV64V-NEXT:    slli a1, s2, 1
+; CHECK-RV64V-NEXT:    slli a6, s2, 3
 ; CHECK-RV64V-NEXT:    mv a4, a3
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB917_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a4, a1
 ; CHECK-RV64V-NEXT:  .LBB917_4:
-; CHECK-RV64V-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64V-NEXT:    add a7, s2, a0
-; CHECK-RV64V-NEXT:    add a5, s2, a5
-; CHECK-RV64V-NEXT:    add a6, s2, a6
+; CHECK-RV64V-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64V-NEXT:    add a7, s0, a0
+; CHECK-RV64V-NEXT:    add a5, s0, a5
+; CHECK-RV64V-NEXT:    add a6, s0, a6
 ; CHECK-RV64V-NEXT:    mv a0, a4
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB917_6
+; CHECK-RV64V-NEXT:    bltu a4, s2, .LBB917_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:  .LBB917_6:
 ; CHECK-RV64V-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64V-NEXT:    csrr a7, vlenb
@@ -38601,7 +38601,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64V-NEXT:    sub a0, a4, s0
+; CHECK-RV64V-NEXT:    sub a0, a4, s2
 ; CHECK-RV64V-NEXT:    sub a5, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64V-NEXT:    sltu a3, a3, a5
@@ -38613,14 +38613,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64V-NEXT:    mv a3, a0
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB917_8
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB917_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB917_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64V-NEXT:    sub a3, a0, s0
+; CHECK-RV64V-NEXT:    sub a3, a0, s2
 ; CHECK-RV64V-NEXT:    sub a2, s1, a2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64V-NEXT:    sltu a4, s1, a2
@@ -38639,9 +38639,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    mv a2, a1
 ; CHECK-RV64V-NEXT:  .LBB917_10:
 ; CHECK-RV64V-NEXT:    mv a3, a2
-; CHECK-RV64V-NEXT:    bltu a2, s0, .LBB917_12
+; CHECK-RV64V-NEXT:    bltu a2, s2, .LBB917_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB917_12:
 ; CHECK-RV64V-NEXT:    csrr a4, vlenb
 ; CHECK-RV64V-NEXT:    slli a4, a4, 3
@@ -38651,7 +38651,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64V-NEXT:    sub a3, a2, s0
+; CHECK-RV64V-NEXT:    sub a3, a2, s2
 ; CHECK-RV64V-NEXT:    sub a1, a0, a1
 ; CHECK-RV64V-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
@@ -38671,9 +38671,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB917_14
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB917_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv a1, s0
+; CHECK-RV64V-NEXT:    mv a1, s2
 ; CHECK-RV64V-NEXT:  .LBB917_14:
 ; CHECK-RV64V-NEXT:    csrr a2, vlenb
 ; CHECK-RV64V-NEXT:    slli a2, a2, 5
@@ -38683,7 +38683,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64V-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.s1
 ; CHECK-RV64V-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64V-NEXT:    sub a1, a0, s0
+; CHECK-RV64V-NEXT:    sub a1, a0, s2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64V-NEXT:    addi a0, a0, -1
 ; CHECK-RV64V-NEXT:    and a0, a0, a1
@@ -38823,12 +38823,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    add a1, a1, sp
 ; CHECK-RV64VC-NEXT:    addi a1, a1, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s2, a0
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
+; CHECK-RV64VC-NEXT:    mv s0, a0
+; CHECK-RV64VC-NEXT:    csrr s2, vlenb
 ; CHECK-RV64VC-NEXT:    li a1, 48
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -38839,17 +38839,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s2, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -38860,30 +38860,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
+; CHECK-RV64VC-NEXT:    slli a2, s2, 2
 ; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB917_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
 ; CHECK-RV64VC-NEXT:  .LBB917_2:
-; CHECK-RV64VC-NEXT:    slli a5, s0, 4
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
-; CHECK-RV64VC-NEXT:    slli a6, s0, 3
+; CHECK-RV64VC-NEXT:    slli a5, s2, 4
+; CHECK-RV64VC-NEXT:    slli a1, s2, 1
+; CHECK-RV64VC-NEXT:    slli a6, s2, 3
 ; CHECK-RV64VC-NEXT:    mv a4, a3
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB917_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a4, a1
 ; CHECK-RV64VC-NEXT:  .LBB917_4:
-; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64VC-NEXT:    add a7, s2, a0
-; CHECK-RV64VC-NEXT:    add a5, a5, s2
-; CHECK-RV64VC-NEXT:    add a6, a6, s2
+; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64VC-NEXT:    add a7, s0, a0
+; CHECK-RV64VC-NEXT:    add a5, a5, s0
+; CHECK-RV64VC-NEXT:    add a6, a6, s0
 ; CHECK-RV64VC-NEXT:    mv a0, a4
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB917_6
+; CHECK-RV64VC-NEXT:    bltu a4, s2, .LBB917_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:  .LBB917_6:
 ; CHECK-RV64VC-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64VC-NEXT:    csrr a7, vlenb
@@ -38911,7 +38911,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64VC-NEXT:    sub a0, a4, s0
+; CHECK-RV64VC-NEXT:    sub a0, a4, s2
 ; CHECK-RV64VC-NEXT:    sub a5, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64VC-NEXT:    sltu a3, a3, a5
@@ -38923,14 +38923,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64VC-NEXT:    mv a3, a0
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB917_8
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB917_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB917_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64VC-NEXT:    sub a3, a0, s0
+; CHECK-RV64VC-NEXT:    sub a3, a0, s2
 ; CHECK-RV64VC-NEXT:    sub a2, s1, a2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64VC-NEXT:    sltu a4, s1, a2
@@ -38949,9 +38949,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    mv a2, a1
 ; CHECK-RV64VC-NEXT:  .LBB917_10:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
-; CHECK-RV64VC-NEXT:    bltu a2, s0, .LBB917_12
+; CHECK-RV64VC-NEXT:    bltu a2, s2, .LBB917_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB917_12:
 ; CHECK-RV64VC-NEXT:    csrr a4, vlenb
 ; CHECK-RV64VC-NEXT:    slli a4, a4, 3
@@ -38961,7 +38961,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a3, a2, s0
+; CHECK-RV64VC-NEXT:    sub a3, a2, s2
 ; CHECK-RV64VC-NEXT:    sub a1, a0, a1
 ; CHECK-RV64VC-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
@@ -38981,9 +38981,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB917_14
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB917_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv a1, s0
+; CHECK-RV64VC-NEXT:    mv a1, s2
 ; CHECK-RV64VC-NEXT:  .LBB917_14:
 ; CHECK-RV64VC-NEXT:    csrr a2, vlenb
 ; CHECK-RV64VC-NEXT:    slli a2, a2, 5
@@ -38993,7 +38993,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_S1(<vscale x 64 x i8> %val, <vs
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.s1
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a1, a0, s0
+; CHECK-RV64VC-NEXT:    sub a1, a0, s2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64VC-NEXT:    addi a0, a0, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
@@ -39138,12 +39138,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    add a1, sp, a1
 ; CHECK-RV64V-NEXT:    addi a1, a1, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s2, a0
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
+; CHECK-RV64V-NEXT:    mv s0, a0
+; CHECK-RV64V-NEXT:    csrr s2, vlenb
 ; CHECK-RV64V-NEXT:    li a1, 48
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -39154,17 +39154,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s2, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -39175,30 +39175,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
+; CHECK-RV64V-NEXT:    slli a2, s2, 2
 ; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB918_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
 ; CHECK-RV64V-NEXT:    mv a3, a2
 ; CHECK-RV64V-NEXT:  .LBB918_2:
-; CHECK-RV64V-NEXT:    slli a5, s0, 4
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
-; CHECK-RV64V-NEXT:    slli a6, s0, 3
+; CHECK-RV64V-NEXT:    slli a5, s2, 4
+; CHECK-RV64V-NEXT:    slli a1, s2, 1
+; CHECK-RV64V-NEXT:    slli a6, s2, 3
 ; CHECK-RV64V-NEXT:    mv a4, a3
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB918_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a4, a1
 ; CHECK-RV64V-NEXT:  .LBB918_4:
-; CHECK-RV64V-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64V-NEXT:    add a7, s2, a0
-; CHECK-RV64V-NEXT:    add a5, s2, a5
-; CHECK-RV64V-NEXT:    add a6, s2, a6
+; CHECK-RV64V-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64V-NEXT:    add a7, s0, a0
+; CHECK-RV64V-NEXT:    add a5, s0, a5
+; CHECK-RV64V-NEXT:    add a6, s0, a6
 ; CHECK-RV64V-NEXT:    mv a0, a4
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB918_6
+; CHECK-RV64V-NEXT:    bltu a4, s2, .LBB918_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:  .LBB918_6:
 ; CHECK-RV64V-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64V-NEXT:    csrr a7, vlenb
@@ -39226,7 +39226,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64V-NEXT:    sub a0, a4, s0
+; CHECK-RV64V-NEXT:    sub a0, a4, s2
 ; CHECK-RV64V-NEXT:    sub a5, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64V-NEXT:    sltu a3, a3, a5
@@ -39238,14 +39238,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64V-NEXT:    mv a3, a0
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB918_8
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB918_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB918_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64V-NEXT:    sub a3, a0, s0
+; CHECK-RV64V-NEXT:    sub a3, a0, s2
 ; CHECK-RV64V-NEXT:    sub a2, s1, a2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64V-NEXT:    sltu a4, s1, a2
@@ -39264,9 +39264,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    mv a2, a1
 ; CHECK-RV64V-NEXT:  .LBB918_10:
 ; CHECK-RV64V-NEXT:    mv a3, a2
-; CHECK-RV64V-NEXT:    bltu a2, s0, .LBB918_12
+; CHECK-RV64V-NEXT:    bltu a2, s2, .LBB918_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB918_12:
 ; CHECK-RV64V-NEXT:    csrr a4, vlenb
 ; CHECK-RV64V-NEXT:    slli a4, a4, 3
@@ -39276,7 +39276,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64V-NEXT:    sub a3, a2, s0
+; CHECK-RV64V-NEXT:    sub a3, a2, s2
 ; CHECK-RV64V-NEXT:    sub a1, a0, a1
 ; CHECK-RV64V-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
@@ -39296,9 +39296,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB918_14
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB918_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv a1, s0
+; CHECK-RV64V-NEXT:    mv a1, s2
 ; CHECK-RV64V-NEXT:  .LBB918_14:
 ; CHECK-RV64V-NEXT:    csrr a2, vlenb
 ; CHECK-RV64V-NEXT:    slli a2, a2, 5
@@ -39308,7 +39308,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64V-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64V-NEXT:    sub a1, a0, s0
+; CHECK-RV64V-NEXT:    sub a1, a0, s2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64V-NEXT:    addi a0, a0, -1
 ; CHECK-RV64V-NEXT:    and a0, a0, a1
@@ -39448,12 +39448,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    add a1, a1, sp
 ; CHECK-RV64VC-NEXT:    addi a1, a1, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s2, a0
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
+; CHECK-RV64VC-NEXT:    mv s0, a0
+; CHECK-RV64VC-NEXT:    csrr s2, vlenb
 ; CHECK-RV64VC-NEXT:    li a1, 48
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -39464,17 +39464,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s2, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -39485,30 +39485,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
+; CHECK-RV64VC-NEXT:    slli a2, s2, 2
 ; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB918_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
 ; CHECK-RV64VC-NEXT:  .LBB918_2:
-; CHECK-RV64VC-NEXT:    slli a5, s0, 4
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
-; CHECK-RV64VC-NEXT:    slli a6, s0, 3
+; CHECK-RV64VC-NEXT:    slli a5, s2, 4
+; CHECK-RV64VC-NEXT:    slli a1, s2, 1
+; CHECK-RV64VC-NEXT:    slli a6, s2, 3
 ; CHECK-RV64VC-NEXT:    mv a4, a3
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB918_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a4, a1
 ; CHECK-RV64VC-NEXT:  .LBB918_4:
-; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64VC-NEXT:    add a7, s2, a0
-; CHECK-RV64VC-NEXT:    add a5, a5, s2
-; CHECK-RV64VC-NEXT:    add a6, a6, s2
+; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64VC-NEXT:    add a7, s0, a0
+; CHECK-RV64VC-NEXT:    add a5, a5, s0
+; CHECK-RV64VC-NEXT:    add a6, a6, s0
 ; CHECK-RV64VC-NEXT:    mv a0, a4
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB918_6
+; CHECK-RV64VC-NEXT:    bltu a4, s2, .LBB918_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:  .LBB918_6:
 ; CHECK-RV64VC-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64VC-NEXT:    csrr a7, vlenb
@@ -39536,7 +39536,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64VC-NEXT:    sub a0, a4, s0
+; CHECK-RV64VC-NEXT:    sub a0, a4, s2
 ; CHECK-RV64VC-NEXT:    sub a5, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64VC-NEXT:    sltu a3, a3, a5
@@ -39548,14 +39548,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64VC-NEXT:    mv a3, a0
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB918_8
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB918_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB918_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64VC-NEXT:    sub a3, a0, s0
+; CHECK-RV64VC-NEXT:    sub a3, a0, s2
 ; CHECK-RV64VC-NEXT:    sub a2, s1, a2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64VC-NEXT:    sltu a4, s1, a2
@@ -39574,9 +39574,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    mv a2, a1
 ; CHECK-RV64VC-NEXT:  .LBB918_10:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
-; CHECK-RV64VC-NEXT:    bltu a2, s0, .LBB918_12
+; CHECK-RV64VC-NEXT:    bltu a2, s2, .LBB918_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB918_12:
 ; CHECK-RV64VC-NEXT:    csrr a4, vlenb
 ; CHECK-RV64VC-NEXT:    slli a4, a4, 3
@@ -39586,7 +39586,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a3, a2, s0
+; CHECK-RV64VC-NEXT:    sub a3, a2, s2
 ; CHECK-RV64VC-NEXT:    sub a1, a0, a1
 ; CHECK-RV64VC-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
@@ -39606,9 +39606,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB918_14
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB918_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv a1, s0
+; CHECK-RV64VC-NEXT:    mv a1, s2
 ; CHECK-RV64VC-NEXT:  .LBB918_14:
 ; CHECK-RV64VC-NEXT:    csrr a2, vlenb
 ; CHECK-RV64VC-NEXT:    slli a2, a2, 5
@@ -39618,7 +39618,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_ALL(<vscale x 64 x i8> %val, <v
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a1, a0, s0
+; CHECK-RV64VC-NEXT:    sub a1, a0, s2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64VC-NEXT:    addi a0, a0, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
@@ -39762,12 +39762,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    add a1, sp, a1
 ; CHECK-RV64V-NEXT:    addi a1, a1, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    mv s2, a0
-; CHECK-RV64V-NEXT:    csrr s0, vlenb
+; CHECK-RV64V-NEXT:    mv s0, a0
+; CHECK-RV64V-NEXT:    csrr s2, vlenb
 ; CHECK-RV64V-NEXT:    li a1, 48
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -39778,17 +39778,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 40
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a1, s0, 5
-; CHECK-RV64V-NEXT:    add a0, s2, a0
+; CHECK-RV64V-NEXT:    slli a1, s2, 5
+; CHECK-RV64V-NEXT:    add a0, s0, a0
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 5
 ; CHECK-RV64V-NEXT:    add a0, sp, a0
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64V-NEXT:    add a1, s2, a1
+; CHECK-RV64V-NEXT:    add a1, s0, a1
 ; CHECK-RV64V-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64V-NEXT:    csrr a0, vlenb
 ; CHECK-RV64V-NEXT:    slli a0, a0, 3
@@ -39799,30 +39799,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    addi a0, a0, 16
 ; CHECK-RV64V-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64V-NEXT:    li a1, 24
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:    call __muldi3
-; CHECK-RV64V-NEXT:    slli a2, s0, 2
+; CHECK-RV64V-NEXT:    slli a2, s2, 2
 ; CHECK-RV64V-NEXT:    mv a3, s1
 ; CHECK-RV64V-NEXT:    bltu s1, a2, .LBB919_2
 ; CHECK-RV64V-NEXT:  # %bb.1:
 ; CHECK-RV64V-NEXT:    mv a3, a2
 ; CHECK-RV64V-NEXT:  .LBB919_2:
-; CHECK-RV64V-NEXT:    slli a5, s0, 4
-; CHECK-RV64V-NEXT:    slli a1, s0, 1
-; CHECK-RV64V-NEXT:    slli a6, s0, 3
+; CHECK-RV64V-NEXT:    slli a5, s2, 4
+; CHECK-RV64V-NEXT:    slli a1, s2, 1
+; CHECK-RV64V-NEXT:    slli a6, s2, 3
 ; CHECK-RV64V-NEXT:    mv a4, a3
 ; CHECK-RV64V-NEXT:    bltu a3, a1, .LBB919_4
 ; CHECK-RV64V-NEXT:  # %bb.3:
 ; CHECK-RV64V-NEXT:    mv a4, a1
 ; CHECK-RV64V-NEXT:  .LBB919_4:
-; CHECK-RV64V-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64V-NEXT:    add a7, s2, a0
-; CHECK-RV64V-NEXT:    add a5, s2, a5
-; CHECK-RV64V-NEXT:    add a6, s2, a6
+; CHECK-RV64V-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64V-NEXT:    add a7, s0, a0
+; CHECK-RV64V-NEXT:    add a5, s0, a5
+; CHECK-RV64V-NEXT:    add a6, s0, a6
 ; CHECK-RV64V-NEXT:    mv a0, a4
-; CHECK-RV64V-NEXT:    bltu a4, s0, .LBB919_6
+; CHECK-RV64V-NEXT:    bltu a4, s2, .LBB919_6
 ; CHECK-RV64V-NEXT:  # %bb.5:
-; CHECK-RV64V-NEXT:    mv a0, s0
+; CHECK-RV64V-NEXT:    mv a0, s2
 ; CHECK-RV64V-NEXT:  .LBB919_6:
 ; CHECK-RV64V-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64V-NEXT:    csrr a7, vlenb
@@ -39850,7 +39850,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64V-NEXT:    sub a0, a4, s0
+; CHECK-RV64V-NEXT:    sub a0, a4, s2
 ; CHECK-RV64V-NEXT:    sub a5, a3, a1
 ; CHECK-RV64V-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64V-NEXT:    sltu a3, a3, a5
@@ -39862,14 +39862,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64V-NEXT:    mv a3, a0
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB919_8
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB919_8
 ; CHECK-RV64V-NEXT:  # %bb.7:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB919_8:
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64V-NEXT:    sub a3, a0, s0
+; CHECK-RV64V-NEXT:    sub a3, a0, s2
 ; CHECK-RV64V-NEXT:    sub a2, s1, a2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64V-NEXT:    sltu a4, s1, a2
@@ -39888,9 +39888,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    mv a2, a1
 ; CHECK-RV64V-NEXT:  .LBB919_10:
 ; CHECK-RV64V-NEXT:    mv a3, a2
-; CHECK-RV64V-NEXT:    bltu a2, s0, .LBB919_12
+; CHECK-RV64V-NEXT:    bltu a2, s2, .LBB919_12
 ; CHECK-RV64V-NEXT:  # %bb.11:
-; CHECK-RV64V-NEXT:    mv a3, s0
+; CHECK-RV64V-NEXT:    mv a3, s2
 ; CHECK-RV64V-NEXT:  .LBB919_12:
 ; CHECK-RV64V-NEXT:    csrr a4, vlenb
 ; CHECK-RV64V-NEXT:    slli a4, a4, 3
@@ -39900,7 +39900,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64V-NEXT:    sub a3, a2, s0
+; CHECK-RV64V-NEXT:    sub a3, a2, s2
 ; CHECK-RV64V-NEXT:    sub a1, a0, a1
 ; CHECK-RV64V-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
@@ -39920,9 +39920,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64V-NEXT:    bltu a0, s0, .LBB919_14
+; CHECK-RV64V-NEXT:    bltu a0, s2, .LBB919_14
 ; CHECK-RV64V-NEXT:  # %bb.13:
-; CHECK-RV64V-NEXT:    mv a1, s0
+; CHECK-RV64V-NEXT:    mv a1, s2
 ; CHECK-RV64V-NEXT:  .LBB919_14:
 ; CHECK-RV64V-NEXT:    csrr a2, vlenb
 ; CHECK-RV64V-NEXT:    slli a2, a2, 5
@@ -39932,7 +39932,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64V-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64V-NEXT:    ntl.all
 ; CHECK-RV64V-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64V-NEXT:    sub a1, a0, s0
+; CHECK-RV64V-NEXT:    sub a1, a0, s2
 ; CHECK-RV64V-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64V-NEXT:    addi a0, a0, -1
 ; CHECK-RV64V-NEXT:    and a0, a0, a1
@@ -40072,12 +40072,12 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    add a1, a1, sp
 ; CHECK-RV64VC-NEXT:    addi a1, a1, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a1) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    mv s2, a0
-; CHECK-RV64VC-NEXT:    csrr s0, vlenb
+; CHECK-RV64VC-NEXT:    mv s0, a0
+; CHECK-RV64VC-NEXT:    csrr s2, vlenb
 ; CHECK-RV64VC-NEXT:    li a1, 48
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -40088,17 +40088,17 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 40
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a1, s0, 5
-; CHECK-RV64VC-NEXT:    add a0, a0, s2
+; CHECK-RV64VC-NEXT:    slli a1, s2, 5
+; CHECK-RV64VC-NEXT:    add a0, a0, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a0)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 5
 ; CHECK-RV64VC-NEXT:    add a0, a0, sp
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
-; CHECK-RV64VC-NEXT:    add a1, a1, s2
+; CHECK-RV64VC-NEXT:    add a1, a1, s0
 ; CHECK-RV64VC-NEXT:    vl8re64.v v8, (a1)
 ; CHECK-RV64VC-NEXT:    csrr a0, vlenb
 ; CHECK-RV64VC-NEXT:    slli a0, a0, 3
@@ -40109,30 +40109,30 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    addi a0, a0, 16
 ; CHECK-RV64VC-NEXT:    vs8r.v v8, (a0) # vscale x 64-byte Folded Spill
 ; CHECK-RV64VC-NEXT:    li a1, 24
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:    call __muldi3
-; CHECK-RV64VC-NEXT:    slli a2, s0, 2
+; CHECK-RV64VC-NEXT:    slli a2, s2, 2
 ; CHECK-RV64VC-NEXT:    mv a3, s1
 ; CHECK-RV64VC-NEXT:    bltu s1, a2, .LBB919_2
 ; CHECK-RV64VC-NEXT:  # %bb.1:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
 ; CHECK-RV64VC-NEXT:  .LBB919_2:
-; CHECK-RV64VC-NEXT:    slli a5, s0, 4
-; CHECK-RV64VC-NEXT:    slli a1, s0, 1
-; CHECK-RV64VC-NEXT:    slli a6, s0, 3
+; CHECK-RV64VC-NEXT:    slli a5, s2, 4
+; CHECK-RV64VC-NEXT:    slli a1, s2, 1
+; CHECK-RV64VC-NEXT:    slli a6, s2, 3
 ; CHECK-RV64VC-NEXT:    mv a4, a3
 ; CHECK-RV64VC-NEXT:    bltu a3, a1, .LBB919_4
 ; CHECK-RV64VC-NEXT:  # %bb.3:
 ; CHECK-RV64VC-NEXT:    mv a4, a1
 ; CHECK-RV64VC-NEXT:  .LBB919_4:
-; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s2)
-; CHECK-RV64VC-NEXT:    add a7, s2, a0
-; CHECK-RV64VC-NEXT:    add a5, a5, s2
-; CHECK-RV64VC-NEXT:    add a6, a6, s2
+; CHECK-RV64VC-NEXT:    vl8re64.v v8, (s0)
+; CHECK-RV64VC-NEXT:    add a7, s0, a0
+; CHECK-RV64VC-NEXT:    add a5, a5, s0
+; CHECK-RV64VC-NEXT:    add a6, a6, s0
 ; CHECK-RV64VC-NEXT:    mv a0, a4
-; CHECK-RV64VC-NEXT:    bltu a4, s0, .LBB919_6
+; CHECK-RV64VC-NEXT:    bltu a4, s2, .LBB919_6
 ; CHECK-RV64VC-NEXT:  # %bb.5:
-; CHECK-RV64VC-NEXT:    mv a0, s0
+; CHECK-RV64VC-NEXT:    mv a0, s2
 ; CHECK-RV64VC-NEXT:  .LBB919_6:
 ; CHECK-RV64VC-NEXT:    vl8re64.v v16, (a7)
 ; CHECK-RV64VC-NEXT:    csrr a7, vlenb
@@ -40160,7 +40160,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v24, (zero), v16
-; CHECK-RV64VC-NEXT:    sub a0, a4, s0
+; CHECK-RV64VC-NEXT:    sub a0, a4, s2
 ; CHECK-RV64VC-NEXT:    sub a5, a3, a1
 ; CHECK-RV64VC-NEXT:    sltu a4, a4, a0
 ; CHECK-RV64VC-NEXT:    sltu a3, a3, a5
@@ -40172,14 +40172,14 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v25, (zero), v8
 ; CHECK-RV64VC-NEXT:    mv a3, a0
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB919_8
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB919_8
 ; CHECK-RV64VC-NEXT:  # %bb.7:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB919_8:
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v26, (zero), v0
-; CHECK-RV64VC-NEXT:    sub a3, a0, s0
+; CHECK-RV64VC-NEXT:    sub a3, a0, s2
 ; CHECK-RV64VC-NEXT:    sub a2, s1, a2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a3
 ; CHECK-RV64VC-NEXT:    sltu a4, s1, a2
@@ -40198,9 +40198,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    mv a2, a1
 ; CHECK-RV64VC-NEXT:  .LBB919_10:
 ; CHECK-RV64VC-NEXT:    mv a3, a2
-; CHECK-RV64VC-NEXT:    bltu a2, s0, .LBB919_12
+; CHECK-RV64VC-NEXT:    bltu a2, s2, .LBB919_12
 ; CHECK-RV64VC-NEXT:  # %bb.11:
-; CHECK-RV64VC-NEXT:    mv a3, s0
+; CHECK-RV64VC-NEXT:    mv a3, s2
 ; CHECK-RV64VC-NEXT:  .LBB919_12:
 ; CHECK-RV64VC-NEXT:    csrr a4, vlenb
 ; CHECK-RV64VC-NEXT:    slli a4, a4, 3
@@ -40210,7 +40210,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v28, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a3, a2, s0
+; CHECK-RV64VC-NEXT:    sub a3, a2, s2
 ; CHECK-RV64VC-NEXT:    sub a1, a0, a1
 ; CHECK-RV64VC-NEXT:    sltu a2, a2, a3
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
@@ -40230,9 +40230,9 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a2, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v29, (zero), v8
-; CHECK-RV64VC-NEXT:    bltu a0, s0, .LBB919_14
+; CHECK-RV64VC-NEXT:    bltu a0, s2, .LBB919_14
 ; CHECK-RV64VC-NEXT:  # %bb.13:
-; CHECK-RV64VC-NEXT:    mv a1, s0
+; CHECK-RV64VC-NEXT:    mv a1, s2
 ; CHECK-RV64VC-NEXT:  .LBB919_14:
 ; CHECK-RV64VC-NEXT:    csrr a2, vlenb
 ; CHECK-RV64VC-NEXT:    slli a2, a2, 5
@@ -40242,7 +40242,7 @@ define void @test_nontemporal_vp_scatter_nxv64i8_DEFAULT(<vscale x 64 x i8> %val
 ; CHECK-RV64VC-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
 ; CHECK-RV64VC-NEXT:    c.ntl.all
 ; CHECK-RV64VC-NEXT:    vsoxei64.v v30, (zero), v8
-; CHECK-RV64VC-NEXT:    sub a1, a0, s0
+; CHECK-RV64VC-NEXT:    sub a1, a0, s2
 ; CHECK-RV64VC-NEXT:    sltu a0, a0, a1
 ; CHECK-RV64VC-NEXT:    addi a0, a0, -1
 ; CHECK-RV64VC-NEXT:    and a0, a0, a1
diff --git a/llvm/test/CodeGen/RISCV/rvv/pr95865.ll b/llvm/test/CodeGen/RISCV/rvv/pr95865.ll
index c0c17ee7c4ef4d..9189b77a4d0cbe 100644
--- a/llvm/test/CodeGen/RISCV/rvv/pr95865.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/pr95865.ll
@@ -7,21 +7,20 @@
 define i32 @main(i1 %arg.1, i64 %arg.2, i1 %arg.3, i64 %arg.4, i1 %arg.5, <vscale x 4 x i1> %arg.6, i64 %arg.7, i1 %arg.8, i64 %arg.9, i32 %arg.10) vscale_range(2,2) {
 ; CHECK-LABEL: main:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -112
-; CHECK-NEXT:    .cfi_def_cfa_offset 112
-; CHECK-NEXT:    sd ra, 104(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 96(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s1, 88(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 80(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s3, 72(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s4, 64(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s5, 56(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s6, 48(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s7, 40(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s8, 32(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s9, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s10, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s11, 8(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    addi sp, sp, -96
+; CHECK-NEXT:    .cfi_def_cfa_offset 96
+; CHECK-NEXT:    sd ra, 88(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s0, 80(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s1, 72(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s2, 64(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s3, 56(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s4, 48(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s5, 40(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s6, 32(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s7, 24(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s8, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s9, 8(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s10, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    .cfi_offset ra, -8
 ; CHECK-NEXT:    .cfi_offset s0, -16
 ; CHECK-NEXT:    .cfi_offset s1, -24
@@ -34,9 +33,7 @@ define i32 @main(i1 %arg.1, i64 %arg.2, i1 %arg.3, i64 %arg.4, i1 %arg.5, <vscal
 ; CHECK-NEXT:    .cfi_offset s8, -80
 ; CHECK-NEXT:    .cfi_offset s9, -88
 ; CHECK-NEXT:    .cfi_offset s10, -96
-; CHECK-NEXT:    .cfi_offset s11, -104
 ; CHECK-NEXT:    li a1, 0
-; CHECK-NEXT:    ld s0, 112(sp)
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; CHECK-NEXT:    vmv.v.i v8, 0
 ; CHECK-NEXT:    li a3, 8
@@ -63,59 +60,59 @@ define i32 @main(i1 %arg.1, i64 %arg.2, i1 %arg.3, i64 %arg.4, i1 %arg.5, <vscal
 ; CHECK-NEXT:    # Child Loop BB0_4 Depth 4
 ; CHECK-NEXT:    # Child Loop BB0_5 Depth 5
 ; CHECK-NEXT:    mv t6, t1
-; CHECK-NEXT:    mv s1, t2
-; CHECK-NEXT:    mv s2, t3
-; CHECK-NEXT:    mv s3, t4
-; CHECK-NEXT:    mv s4, t5
+; CHECK-NEXT:    mv s0, t2
+; CHECK-NEXT:    mv s1, t3
+; CHECK-NEXT:    mv s2, t4
+; CHECK-NEXT:    mv s3, t5
 ; CHECK-NEXT:  .LBB0_3: # %for.cond9.preheader.i
 ; CHECK-NEXT:    # Parent Loop BB0_1 Depth=1
 ; CHECK-NEXT:    # Parent Loop BB0_2 Depth=2
 ; CHECK-NEXT:    # => This Loop Header: Depth=3
 ; CHECK-NEXT:    # Child Loop BB0_4 Depth 4
 ; CHECK-NEXT:    # Child Loop BB0_5 Depth 5
-; CHECK-NEXT:    mv s6, t6
+; CHECK-NEXT:    mv s5, t6
+; CHECK-NEXT:    mv s6, s0
 ; CHECK-NEXT:    mv s7, s1
 ; CHECK-NEXT:    mv s8, s2
 ; CHECK-NEXT:    mv s9, s3
-; CHECK-NEXT:    mv s10, s4
 ; CHECK-NEXT:  .LBB0_4: # %vector.ph.i
 ; CHECK-NEXT:    # Parent Loop BB0_1 Depth=1
 ; CHECK-NEXT:    # Parent Loop BB0_2 Depth=2
 ; CHECK-NEXT:    # Parent Loop BB0_3 Depth=3
 ; CHECK-NEXT:    # => This Loop Header: Depth=4
 ; CHECK-NEXT:    # Child Loop BB0_5 Depth 5
-; CHECK-NEXT:    li s5, 0
+; CHECK-NEXT:    li s4, 0
 ; CHECK-NEXT:  .LBB0_5: # %vector.body.i
 ; CHECK-NEXT:    # Parent Loop BB0_1 Depth=1
 ; CHECK-NEXT:    # Parent Loop BB0_2 Depth=2
 ; CHECK-NEXT:    # Parent Loop BB0_3 Depth=3
 ; CHECK-NEXT:    # Parent Loop BB0_4 Depth=4
 ; CHECK-NEXT:    # => This Inner Loop Header: Depth=5
-; CHECK-NEXT:    add s11, s9, s5
-; CHECK-NEXT:    vse32.v v8, (s11), v0.t
-; CHECK-NEXT:    addi s11, s5, 4
-; CHECK-NEXT:    add s5, s5, s7
-; CHECK-NEXT:    vse32.v v8, (s5), v0.t
-; CHECK-NEXT:    mv s5, s11
-; CHECK-NEXT:    bne s11, a6, .LBB0_5
+; CHECK-NEXT:    add s10, s8, s4
+; CHECK-NEXT:    vse32.v v8, (s10), v0.t
+; CHECK-NEXT:    addi s10, s4, 4
+; CHECK-NEXT:    add s4, s4, s6
+; CHECK-NEXT:    vse32.v v8, (s4), v0.t
+; CHECK-NEXT:    mv s4, s10
+; CHECK-NEXT:    bne s10, a6, .LBB0_5
 ; CHECK-NEXT:  # %bb.6: # %for.cond.cleanup15.i
 ; CHECK-NEXT:    # in Loop: Header=BB0_4 Depth=4
-; CHECK-NEXT:    addi s10, s10, 4
 ; CHECK-NEXT:    addi s9, s9, 4
 ; CHECK-NEXT:    addi s8, s8, 4
-; CHECK-NEXT:    andi s5, a0, 1
 ; CHECK-NEXT:    addi s7, s7, 4
+; CHECK-NEXT:    andi s4, a0, 1
 ; CHECK-NEXT:    addi s6, s6, 4
-; CHECK-NEXT:    beqz s5, .LBB0_4
+; CHECK-NEXT:    addi s5, s5, 4
+; CHECK-NEXT:    beqz s4, .LBB0_4
 ; CHECK-NEXT:  # %bb.7: # %for.cond.cleanup11.i
 ; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=3
-; CHECK-NEXT:    addi s4, s4, 4
 ; CHECK-NEXT:    addi s3, s3, 4
 ; CHECK-NEXT:    addi s2, s2, 4
-; CHECK-NEXT:    andi s6, a2, 1
 ; CHECK-NEXT:    addi s1, s1, 4
+; CHECK-NEXT:    andi s5, a2, 1
+; CHECK-NEXT:    addi s0, s0, 4
 ; CHECK-NEXT:    addi t6, t6, 4
-; CHECK-NEXT:    beqz s6, .LBB0_3
+; CHECK-NEXT:    beqz s5, .LBB0_3
 ; CHECK-NEXT:  # %bb.8: # %for.cond.cleanup7.i
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=2
 ; CHECK-NEXT:    addi t5, t5, 4
@@ -131,18 +128,19 @@ define i32 @main(i1 %arg.1, i64 %arg.2, i1 %arg.3, i64 %arg.4, i1 %arg.5, <vscal
 ; CHECK-NEXT:    addi a5, a5, 4
 ; CHECK-NEXT:    addi t0, t0, 4
 ; CHECK-NEXT:    addi a7, a7, 4
-; CHECK-NEXT:    beqz s6, .LBB0_1
+; CHECK-NEXT:    beqz s5, .LBB0_1
 ; CHECK-NEXT:  # %bb.10: # %l.exit
 ; CHECK-NEXT:    li a0, 0
 ; CHECK-NEXT:    jalr a0
-; CHECK-NEXT:    beqz s5, .LBB0_12
+; CHECK-NEXT:    beqz s4, .LBB0_12
 ; CHECK-NEXT:  .LBB0_11: # %for.body7.us.14
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    j .LBB0_11
 ; CHECK-NEXT:  .LBB0_12: # %for.body7.us.19
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
 ; CHECK-NEXT:    vmv.v.i v8, 0
-; CHECK-NEXT:    vmv.s.x v16, s0
+; CHECK-NEXT:    ld a0, 96(sp)
+; CHECK-NEXT:    vmv.s.x v16, a0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, m1, tu, ma
 ; CHECK-NEXT:    vslideup.vi v8, v16, 1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
@@ -151,19 +149,18 @@ define i32 @main(i1 %arg.1, i64 %arg.2, i1 %arg.3, i64 %arg.4, i1 %arg.5, <vscal
 ; CHECK-NEXT:    snez a1, a0
 ; CHECK-NEXT:    li a0, 0
 ; CHECK-NEXT:    sb a1, 0(zero)
-; CHECK-NEXT:    ld ra, 104(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 96(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s1, 88(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 80(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s3, 72(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s4, 64(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s5, 56(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s6, 48(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s7, 40(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s8, 32(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s9, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s10, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s11, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld ra, 88(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s0, 80(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s1, 72(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s2, 64(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s3, 56(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s4, 48(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s5, 40(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s6, 32(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s7, 24(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s8, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s9, 8(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s10, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    .cfi_restore ra
 ; CHECK-NEXT:    .cfi_restore s0
 ; CHECK-NEXT:    .cfi_restore s1
@@ -176,8 +173,7 @@ define i32 @main(i1 %arg.1, i64 %arg.2, i1 %arg.3, i64 %arg.4, i1 %arg.5, <vscal
 ; CHECK-NEXT:    .cfi_restore s8
 ; CHECK-NEXT:    .cfi_restore s9
 ; CHECK-NEXT:    .cfi_restore s10
-; CHECK-NEXT:    .cfi_restore s11
-; CHECK-NEXT:    addi sp, sp, 112
+; CHECK-NEXT:    addi sp, sp, 96
 ; CHECK-NEXT:    .cfi_def_cfa_offset 0
 ; CHECK-NEXT:    ret
 entry:
diff --git a/llvm/test/CodeGen/RISCV/rvv/shrinkwrap.ll b/llvm/test/CodeGen/RISCV/rvv/shrinkwrap.ll
index d12f2c889650fb..452bfd3032783d 100644
--- a/llvm/test/CodeGen/RISCV/rvv/shrinkwrap.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/shrinkwrap.ll
@@ -7,62 +7,58 @@
 define void @vecaddr_straightline(i32 zeroext %a, ptr %p) {
 ; RV32-LABEL: vecaddr_straightline:
 ; RV32:       # %bb.0:
+; RV32-NEXT:    addi a1, a1, 32
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vle32.v v8, (a1)
+; RV32-NEXT:    vadd.vi v8, v8, 1
+; RV32-NEXT:    li a2, 57
+; RV32-NEXT:    vse32.v v8, (a1)
+; RV32-NEXT:    beq a0, a2, .LBB0_2
+; RV32-NEXT:  # %bb.1: # %do_call
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset ra, -4
-; RV32-NEXT:    .cfi_offset s0, -8
-; RV32-NEXT:    addi s0, a1, 32
-; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; RV32-NEXT:    vle32.v v8, (s0)
-; RV32-NEXT:    vadd.vi v8, v8, 1
-; RV32-NEXT:    li a1, 57
-; RV32-NEXT:    vse32.v v8, (s0)
-; RV32-NEXT:    beq a0, a1, .LBB0_2
-; RV32-NEXT:  # %bb.1: # %do_call
+; RV32-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    call foo
+; RV32-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; RV32-NEXT:  .LBB0_2: # %exit
-; RV32-NEXT:    vle32.v v8, (s0)
-; RV32-NEXT:    vadd.vi v8, v8, 1
-; RV32-NEXT:    vse32.v v8, (s0)
 ; RV32-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore ra
-; RV32-NEXT:    .cfi_restore s0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:  .LBB0_2: # %exit
+; RV32-NEXT:    vle32.v v8, (a1)
+; RV32-NEXT:    vadd.vi v8, v8, 1
+; RV32-NEXT:    vse32.v v8, (a1)
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vecaddr_straightline:
 ; RV64:       # %bb.0:
+; RV64-NEXT:    addi a1, a1, 32
+; RV64-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV64-NEXT:    vle32.v v8, (a1)
+; RV64-NEXT:    vadd.vi v8, v8, 1
+; RV64-NEXT:    li a2, 57
+; RV64-NEXT:    vse32.v v8, (a1)
+; RV64-NEXT:    beq a0, a2, .LBB0_2
+; RV64-NEXT:  # %bb.1: # %do_call
 ; RV64-NEXT:    addi sp, sp, -16
 ; RV64-NEXT:    .cfi_def_cfa_offset 16
 ; RV64-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    .cfi_offset ra, -8
-; RV64-NEXT:    .cfi_offset s0, -16
-; RV64-NEXT:    addi s0, a1, 32
-; RV64-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; RV64-NEXT:    vle32.v v8, (s0)
-; RV64-NEXT:    vadd.vi v8, v8, 1
-; RV64-NEXT:    li a1, 57
-; RV64-NEXT:    vse32.v v8, (s0)
-; RV64-NEXT:    beq a0, a1, .LBB0_2
-; RV64-NEXT:  # %bb.1: # %do_call
+; RV64-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
 ; RV64-NEXT:    call foo
+; RV64-NEXT:    ld a1, 0(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; RV64-NEXT:  .LBB0_2: # %exit
-; RV64-NEXT:    vle32.v v8, (s0)
-; RV64-NEXT:    vadd.vi v8, v8, 1
-; RV64-NEXT:    vse32.v v8, (s0)
 ; RV64-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    .cfi_restore ra
-; RV64-NEXT:    .cfi_restore s0
 ; RV64-NEXT:    addi sp, sp, 16
 ; RV64-NEXT:    .cfi_def_cfa_offset 0
+; RV64-NEXT:  .LBB0_2: # %exit
+; RV64-NEXT:    vle32.v v8, (a1)
+; RV64-NEXT:    vadd.vi v8, v8, 1
+; RV64-NEXT:    vse32.v v8, (a1)
 ; RV64-NEXT:    ret
   %gep = getelementptr i8, ptr %p, i32 32
   %v1 = load <4 x i32>, ptr %gep
diff --git a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll
index 88e894cca37992..4f7bb6825517df 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll
@@ -319,13 +319,13 @@ define <vscale x 1 x double> @test8(i64 %avl, i8 zeroext %cond, <vscale x 1 x do
 ; CHECK-NEXT:    csrr a1, vlenb
 ; CHECK-NEXT:    slli a1, a1, 1
 ; CHECK-NEXT:    sub sp, sp, a1
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    add a1, a1, sp
+; CHECK-NEXT:    addi a1, a1, 16
+; CHECK-NEXT:    vs1r.v v9, (a1) # vscale x 8-byte Folded Spill
+; CHECK-NEXT:    addi a1, sp, 16
+; CHECK-NEXT:    vs1r.v v8, (a1) # vscale x 8-byte Folded Spill
 ; CHECK-NEXT:    mv s0, a0
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    add a0, a0, sp
-; CHECK-NEXT:    addi a0, a0, 16
-; CHECK-NEXT:    vs1r.v v9, (a0) # vscale x 8-byte Folded Spill
-; CHECK-NEXT:    addi a0, sp, 16
-; CHECK-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
 ; CHECK-NEXT:    call foo
 ; CHECK-NEXT:    csrr a0, vlenb
 ; CHECK-NEXT:    add a0, a0, sp
@@ -366,42 +366,43 @@ if.end:                                           ; preds = %if.else, %if.then
 define <vscale x 1 x double> @test9(i64 %avl, i8 zeroext %cond, <vscale x 1 x double> %a, <vscale x 1 x double> %b) nounwind {
 ; CHECK-LABEL: test9:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi sp, sp, -32
-; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    csrr a2, vlenb
-; CHECK-NEXT:    slli a2, a2, 1
-; CHECK-NEXT:    sub sp, sp, a2
-; CHECK-NEXT:    mv s0, a0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; CHECK-NEXT:    beqz a1, .LBB7_2
 ; CHECK-NEXT:  # %bb.1: # %if.then
+; CHECK-NEXT:    addi sp, sp, -32
+; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    slli a1, a1, 1
+; CHECK-NEXT:    sub sp, sp, a1
 ; CHECK-NEXT:    vfadd.vv v9, v8, v9
-; CHECK-NEXT:    addi a0, sp, 16
-; CHECK-NEXT:    vs1r.v v9, (a0) # vscale x 8-byte Folded Spill
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    add a0, a0, sp
-; CHECK-NEXT:    addi a0, a0, 16
-; CHECK-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
+; CHECK-NEXT:    addi a1, sp, 16
+; CHECK-NEXT:    vs1r.v v9, (a1) # vscale x 8-byte Folded Spill
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    add a1, a1, sp
+; CHECK-NEXT:    addi a1, a1, 16
+; CHECK-NEXT:    vs1r.v v8, (a1) # vscale x 8-byte Folded Spill
+; CHECK-NEXT:    mv s0, a0
 ; CHECK-NEXT:    call foo
-; CHECK-NEXT:    addi a0, sp, 16
-; CHECK-NEXT:    vl1r.v v9, (a0) # vscale x 8-byte Folded Reload
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    add a0, a0, sp
-; CHECK-NEXT:    addi a0, a0, 16
-; CHECK-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; CHECK-NEXT:    mv a0, s0
+; CHECK-NEXT:    addi a1, sp, 16
+; CHECK-NEXT:    vl1r.v v9, (a1) # vscale x 8-byte Folded Reload
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    add a1, a1, sp
+; CHECK-NEXT:    addi a1, a1, 16
+; CHECK-NEXT:    vl1r.v v8, (a1) # vscale x 8-byte Folded Reload
+; CHECK-NEXT:    csrr a1, vlenb
+; CHECK-NEXT:    slli a1, a1, 1
+; CHECK-NEXT:    add sp, sp, a1
+; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    j .LBB7_3
 ; CHECK-NEXT:  .LBB7_2: # %if.else
 ; CHECK-NEXT:    vfsub.vv v9, v8, v9
 ; CHECK-NEXT:  .LBB7_3: # %if.end
-; CHECK-NEXT:    vsetvli zero, s0, e64, m1, ta, ma
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; CHECK-NEXT:    vfmul.vv v8, v9, v8
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    slli a0, a0, 1
-; CHECK-NEXT:    add sp, sp, a0
-; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    ret
 entry:
   %0 = tail call i64 @llvm.riscv.vsetvli(i64 %avl, i64 3, i64 0)
diff --git a/llvm/test/CodeGen/RISCV/rvv/vxrm-insert-out-of-loop.ll b/llvm/test/CodeGen/RISCV/rvv/vxrm-insert-out-of-loop.ll
index 997d1bfbf1a9d7..e0ca1dc9378989 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vxrm-insert-out-of-loop.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vxrm-insert-out-of-loop.ll
@@ -16,58 +16,60 @@ define void @test1(ptr nocapture noundef writeonly %dst, i32 noundef signext %i_
 ; RV32-NEXT:  # %bb.1: # %for.cond1.preheader.lr.ph
 ; RV32-NEXT:    blez a6, .LBB0_17
 ; RV32-NEXT:  # %bb.2: # %for.cond1.preheader.us.preheader
-; RV32-NEXT:    csrr t2, vlenb
-; RV32-NEXT:    addi t4, a7, -1
-; RV32-NEXT:    slli t1, t2, 1
-; RV32-NEXT:    li t3, 32
-; RV32-NEXT:    mv t0, t1
-; RV32-NEXT:  # %bb.3: # %for.cond1.preheader.us.preheader
-; RV32-NEXT:    li t0, 32
-; RV32-NEXT:  # %bb.4: # %for.cond1.preheader.us.preheader
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw s0, 12(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    sw s1, 8(sp) # 4-byte Folded Spill
-; RV32-NEXT:    sw s2, 4(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    .cfi_offset s0, -4
 ; RV32-NEXT:    .cfi_offset s1, -8
-; RV32-NEXT:    .cfi_offset s2, -12
 ; RV32-NEXT:    .cfi_remember_state
+; RV32-NEXT:    csrr t0, vlenb
+; RV32-NEXT:    sw t0, 4(sp) # 4-byte Folded Spill
+; RV32-NEXT:    addi t4, a7, -1
+; RV32-NEXT:    lw t1, 4(sp) # 4-byte Folded Reload
+; RV32-NEXT:    slli t1, t1, 1
+; RV32-NEXT:    mv t0, t1
+; RV32-NEXT:  # %bb.3: # %for.cond1.preheader.us.preheader
+; RV32-NEXT:    li t0, 32
+; RV32-NEXT:  # %bb.4: # %for.cond1.preheader.us.preheader
+; RV32-NEXT:    sw t0, 0(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    mul t5, a1, t4
-; RV32-NEXT:    mul t6, a3, t4
-; RV32-NEXT:    mul t4, a5, t4
-; RV32-NEXT:    add s0, a0, a6
-; RV32-NEXT:    add s1, a2, a6
-; RV32-NEXT:    add s2, a4, a6
-; RV32-NEXT:    bltu t3, t1, .LBB0_6
+; RV32-NEXT:    mul t2, a3, t4
+; RV32-NEXT:    mul t0, a5, t4
+; RV32-NEXT:    add t3, a0, a6
+; RV32-NEXT:    add t6, a2, a6
+; RV32-NEXT:    add t4, a4, a6
+; RV32-NEXT:    li s0, 32
+; RV32-NEXT:    bltu s0, t1, .LBB0_6
 ; RV32-NEXT:  # %bb.5: # %for.cond1.preheader.us.preheader
 ; RV32-NEXT:    li t1, 32
 ; RV32-NEXT:  .LBB0_6: # %for.cond1.preheader.us.preheader
-; RV32-NEXT:    add t3, s0, t5
-; RV32-NEXT:    add t6, s1, t6
-; RV32-NEXT:    add t4, s2, t4
+; RV32-NEXT:    add t3, t3, t5
+; RV32-NEXT:    add t6, t6, t2
+; RV32-NEXT:    add t4, t4, t0
 ; RV32-NEXT:    j .LBB0_8
 ; RV32-NEXT:  # %bb.7: # %for.cond1.preheader.us.preheader
-; RV32-NEXT:    mv t1, t0
+; RV32-NEXT:    lw t1, 0(sp) # 4-byte Folded Reload
 ; RV32-NEXT:  .LBB0_8: # %for.cond1.preheader.us.preheader
 ; RV32-NEXT:    .cfi_restore_state
 ; RV32-NEXT:    li t0, 0
-; RV32-NEXT:    sltu t5, a0, t6
-; RV32-NEXT:    sltu t6, a2, t3
-; RV32-NEXT:    and t5, t5, t6
-; RV32-NEXT:    or t6, a1, a3
+; RV32-NEXT:    sltu t2, a0, t6
+; RV32-NEXT:    sltu t5, a2, t3
+; RV32-NEXT:    and t2, t2, t5
+; RV32-NEXT:    or t5, a1, a3
 ; RV32-NEXT:    sltu t4, a0, t4
 ; RV32-NEXT:    sltu t3, a4, t3
 ; RV32-NEXT:    and t3, t4, t3
 ; RV32-NEXT:    or t4, a1, a5
-; RV32-NEXT:    srli t6, t6, 31
+; RV32-NEXT:    srli t5, t5, 31
 ; RV32-NEXT:    srli t4, t4, 31
-; RV32-NEXT:    or t5, t5, t6
+; RV32-NEXT:    or t2, t2, t5
 ; RV32-NEXT:    or t3, t3, t4
-; RV32-NEXT:    or t3, t5, t3
+; RV32-NEXT:    or t2, t2, t3
 ; RV32-NEXT:    sltu t1, a6, t1
-; RV32-NEXT:    or t1, t1, t3
+; RV32-NEXT:    or t1, t1, t2
 ; RV32-NEXT:    andi t1, t1, 1
+; RV32-NEXT:    lw t2, 4(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    slli t2, t2, 1
 ; RV32-NEXT:    csrwi vxrm, 0
 ; RV32-NEXT:    j .LBB0_10
@@ -137,10 +139,8 @@ define void @test1(ptr nocapture noundef writeonly %dst, i32 noundef signext %i_
 ; RV32-NEXT:  .LBB0_16:
 ; RV32-NEXT:    lw s0, 12(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    lw s1, 8(sp) # 4-byte Folded Reload
-; RV32-NEXT:    lw s2, 4(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    .cfi_restore s0
 ; RV32-NEXT:    .cfi_restore s1
-; RV32-NEXT:    .cfi_restore s2
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    .cfi_def_cfa_offset 0
 ; RV32-NEXT:  .LBB0_17: # %for.cond.cleanup
diff --git a/llvm/test/CodeGen/RISCV/select-cc.ll b/llvm/test/CodeGen/RISCV/select-cc.ll
index 220329fd94083d..0949152f606af8 100644
--- a/llvm/test/CodeGen/RISCV/select-cc.ll
+++ b/llvm/test/CodeGen/RISCV/select-cc.ll
@@ -311,79 +311,75 @@ define signext i32 @foo(i32 signext %a, ptr %b) nounwind {
 define signext i16 @numsignbits(i16 signext %0, i16 signext %1, i16 signext %2, i16 signext %3) nounwind {
 ; RV32I-LABEL: numsignbits:
 ; RV32I:       # %bb.0:
-; RV32I-NEXT:    addi sp, sp, -16
-; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a3
 ; RV32I-NEXT:    beqz a0, .LBB1_2
 ; RV32I-NEXT:  # %bb.1:
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv a3, a2
 ; RV32I-NEXT:  .LBB1_2:
 ; RV32I-NEXT:    beqz a1, .LBB1_4
 ; RV32I-NEXT:  # %bb.3:
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    addi sp, sp, -16
+; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, a3
+; RV32I-NEXT:    sw a3, 8(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call bar
-; RV32I-NEXT:  .LBB1_4:
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a3, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
+; RV32I-NEXT:  .LBB1_4:
+; RV32I-NEXT:    mv a0, a3
 ; RV32I-NEXT:    ret
 ;
 ; RV32IXQCI-LABEL: numsignbits:
 ; RV32IXQCI:       # %bb.0:
-; RV32IXQCI-NEXT:    addi sp, sp, -16
-; RV32IXQCI-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32IXQCI-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32IXQCI-NEXT:    mv s0, a2
-; RV32IXQCI-NEXT:    qc.mveqi s0, a0, 0, a3
+; RV32IXQCI-NEXT:    qc.mveqi a2, a0, 0, a3
 ; RV32IXQCI-NEXT:    beqz a1, .LBB1_2
 ; RV32IXQCI-NEXT:  # %bb.1:
-; RV32IXQCI-NEXT:    mv a0, s0
+; RV32IXQCI-NEXT:    addi sp, sp, -16
+; RV32IXQCI-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32IXQCI-NEXT:    mv a0, a2
+; RV32IXQCI-NEXT:    sw a2, 8(sp) # 4-byte Folded Spill
 ; RV32IXQCI-NEXT:    call bar
-; RV32IXQCI-NEXT:  .LBB1_2:
-; RV32IXQCI-NEXT:    mv a0, s0
+; RV32IXQCI-NEXT:    lw a2, 8(sp) # 4-byte Folded Reload
 ; RV32IXQCI-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32IXQCI-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32IXQCI-NEXT:    addi sp, sp, 16
+; RV32IXQCI-NEXT:  .LBB1_2:
+; RV32IXQCI-NEXT:    mv a0, a2
 ; RV32IXQCI-NEXT:    ret
 ;
 ; RV64I-LABEL: numsignbits:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -16
-; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    mv s0, a3
 ; RV64I-NEXT:    beqz a0, .LBB1_2
 ; RV64I-NEXT:  # %bb.1:
-; RV64I-NEXT:    mv s0, a2
+; RV64I-NEXT:    mv a3, a2
 ; RV64I-NEXT:  .LBB1_2:
 ; RV64I-NEXT:    beqz a1, .LBB1_4
 ; RV64I-NEXT:  # %bb.3:
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    addi sp, sp, -16
+; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv a0, a3
+; RV64I-NEXT:    sd a3, 0(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    call bar
-; RV64I-NEXT:  .LBB1_4:
-; RV64I-NEXT:    mv a0, s0
+; RV64I-NEXT:    ld a3, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
+; RV64I-NEXT:  .LBB1_4:
+; RV64I-NEXT:    mv a0, a3
 ; RV64I-NEXT:    ret
 ;
 ; RV64I-CCMOV-LABEL: numsignbits:
 ; RV64I-CCMOV:       # %bb.0:
-; RV64I-CCMOV-NEXT:    addi sp, sp, -16
-; RV64I-CCMOV-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-CCMOV-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64I-CCMOV-NEXT:    mips.ccmov s0, a0, a2, a3
+; RV64I-CCMOV-NEXT:    mips.ccmov a0, a0, a2, a3
 ; RV64I-CCMOV-NEXT:    beqz a1, .LBB1_2
 ; RV64I-CCMOV-NEXT:  # %bb.1:
-; RV64I-CCMOV-NEXT:    mv a0, s0
+; RV64I-CCMOV-NEXT:    addi sp, sp, -16
+; RV64I-CCMOV-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-CCMOV-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
+; RV64I-CCMOV-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64I-CCMOV-NEXT:    call bar
-; RV64I-CCMOV-NEXT:  .LBB1_2:
-; RV64I-CCMOV-NEXT:    mv a0, s0
+; RV64I-CCMOV-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64I-CCMOV-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-CCMOV-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64I-CCMOV-NEXT:    addi sp, sp, 16
+; RV64I-CCMOV-NEXT:  .LBB1_2:
 ; RV64I-CCMOV-NEXT:    ret
   %5 = icmp eq i16 %0, 0
   %6 = select i1 %5, i16 %3, i16 %2
diff --git a/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll b/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll
index 07bfbe65bd9e28..4484ff644c2496 100644
--- a/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll
+++ b/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll
@@ -877,44 +877,36 @@ define void @zext_nneg_dominating_icmp_i32_zeroext(i16 signext %0) {
 define void @load_zext_nneg_sext_cse(ptr %p) nounwind {
 ; RV32I-LABEL: load_zext_nneg_sext_cse:
 ; RV32I:       # %bb.0:
+; RV32I-NEXT:    lh a0, 0(a0)
+; RV32I-NEXT:    bltz a0, .LBB50_2
+; RV32I-NEXT:  # %bb.1: # %bb1
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    lh s0, 0(a0)
-; RV32I-NEXT:    bltz s0, .LBB50_2
-; RV32I-NEXT:  # %bb.1: # %bb1
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    sw a0, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call bar_i16
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    tail bar_i32
 ; RV32I-NEXT:  .LBB50_2: # %bb2
-; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
 ; RV64-LABEL: load_zext_nneg_sext_cse:
 ; RV64:       # %bb.0:
+; RV64-NEXT:    lh a0, 0(a0)
+; RV64-NEXT:    bltz a0, .LBB50_2
+; RV64-NEXT:  # %bb.1: # %bb1
 ; RV64-NEXT:    addi sp, sp, -16
 ; RV64-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64-NEXT:    lh s0, 0(a0)
-; RV64-NEXT:    bltz s0, .LBB50_2
-; RV64-NEXT:  # %bb.1: # %bb1
-; RV64-NEXT:    mv a0, s0
+; RV64-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
+; RV64-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    call bar_i16
-; RV64-NEXT:    mv a0, s0
+; RV64-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64-NEXT:    addi sp, sp, 16
 ; RV64-NEXT:    tail bar_i32
 ; RV64-NEXT:  .LBB50_2: # %bb2
-; RV64-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
-; RV64-NEXT:    addi sp, sp, 16
 ; RV64-NEXT:    ret
   %load = load i16, ptr %p
   %zext = zext nneg i16 %load to i32
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load-atomic-acquire-seq_cst.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load-atomic-acquire-seq_cst.ll
index 4b39712cd5db57..6089ef0b9a94e2 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load-atomic-acquire-seq_cst.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load-atomic-acquire-seq_cst.ll
@@ -1135,23 +1135,19 @@ define i64 @test_i64_1_3(ptr %base, i1 zeroext %x, i64 %b) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a3
-; RV32I-NEXT:    mv s1, a2
-; RV32I-NEXT:    mv s2, a1
+; RV32I-NEXT:    sw a3, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 0(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s0, a1
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    li a1, 2
 ; RV32I-NEXT:    call __atomic_load_8
-; RV32I-NEXT:    bnez s2, .LBB15_2
+; RV32I-NEXT:    bnez s0, .LBB15_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a1, s0
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB15_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -1712,31 +1708,27 @@ define i64 @test_i64_store_64_3(ptr %base, i1 zeroext %x, i64 %b, ptr %base1, i6
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s2, a6
-; RV32I-NEXT:    mv s3, a5
-; RV32I-NEXT:    mv s4, a4
-; RV32I-NEXT:    mv s0, a3
-; RV32I-NEXT:    mv s1, a2
-; RV32I-NEXT:    mv s5, a1
+; RV32I-NEXT:    mv s0, a6
+; RV32I-NEXT:    mv s1, a5
+; RV32I-NEXT:    mv s2, a4
+; RV32I-NEXT:    sw a3, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s3, a1
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    li a1, 2
 ; RV32I-NEXT:    call __atomic_load_8
-; RV32I-NEXT:    sw s3, 0(s4)
-; RV32I-NEXT:    sw s2, 4(s4)
-; RV32I-NEXT:    bnez s5, .LBB21_2
+; RV32I-NEXT:    sw s1, 0(s2)
+; RV32I-NEXT:    sw s0, 4(s2)
+; RV32I-NEXT:    bnez s3, .LBB21_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a1, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB21_2: # %entry
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
@@ -3073,23 +3065,19 @@ define i64 @test_i64_1_4(ptr %base, i1 zeroext %x, i64 %b) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a3
-; RV32I-NEXT:    mv s1, a2
-; RV32I-NEXT:    mv s2, a1
+; RV32I-NEXT:    sw a3, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 0(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s0, a1
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    li a1, 5
 ; RV32I-NEXT:    call __atomic_load_8
-; RV32I-NEXT:    bnez s2, .LBB37_2
+; RV32I-NEXT:    bnez s0, .LBB37_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a1, s0
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB37_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -3683,31 +3671,27 @@ define i64 @test_i64_store_64_4(ptr %base, i1 zeroext %x, i64 %b, ptr %base1, i6
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s2, a6
-; RV32I-NEXT:    mv s3, a5
-; RV32I-NEXT:    mv s4, a4
-; RV32I-NEXT:    mv s0, a3
-; RV32I-NEXT:    mv s1, a2
-; RV32I-NEXT:    mv s5, a1
+; RV32I-NEXT:    mv s0, a6
+; RV32I-NEXT:    mv s1, a5
+; RV32I-NEXT:    mv s2, a4
+; RV32I-NEXT:    sw a3, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s3, a1
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    li a1, 5
 ; RV32I-NEXT:    call __atomic_load_8
-; RV32I-NEXT:    sw s3, 0(s4)
-; RV32I-NEXT:    sw s2, 4(s4)
-; RV32I-NEXT:    bnez s5, .LBB43_2
+; RV32I-NEXT:    sw s1, 0(s2)
+; RV32I-NEXT:    sw s0, 4(s2)
+; RV32I-NEXT:    bnez s3, .LBB43_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a1, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB43_2: # %entry
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load.ll
index 1251430d006a9f..f3c594e9b4ba95 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-load.ll
@@ -6207,23 +6207,19 @@ define i64 @test_i64_1_2(ptr %base, i1 zeroext %x, i64 %b) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s0, a3
-; RV32I-NEXT:    mv s1, a2
-; RV32I-NEXT:    mv s2, a1
+; RV32I-NEXT:    sw a3, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 0(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s0, a1
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __atomic_load_8
-; RV32I-NEXT:    bnez s2, .LBB81_2
+; RV32I-NEXT:    bnez s0, .LBB81_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a1, s0
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB81_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
@@ -6751,31 +6747,27 @@ define i64 @test_i64_store_64_2(ptr %base, i1 zeroext %x, i64 %b, ptr %base1, i6
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s5, 4(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    mv s2, a6
-; RV32I-NEXT:    mv s3, a5
-; RV32I-NEXT:    mv s4, a4
-; RV32I-NEXT:    mv s0, a3
-; RV32I-NEXT:    mv s1, a2
-; RV32I-NEXT:    mv s5, a1
+; RV32I-NEXT:    mv s0, a6
+; RV32I-NEXT:    mv s1, a5
+; RV32I-NEXT:    mv s2, a4
+; RV32I-NEXT:    sw a3, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s3, a1
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    call __atomic_load_8
-; RV32I-NEXT:    sw s3, 0(s4)
-; RV32I-NEXT:    sw s2, 4(s4)
-; RV32I-NEXT:    bnez s5, .LBB87_2
+; RV32I-NEXT:    sw s1, 0(s2)
+; RV32I-NEXT:    sw s0, 4(s2)
+; RV32I-NEXT:    bnez s3, .LBB87_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s1
-; RV32I-NEXT:    mv a1, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB87_2: # %entry
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_eq.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_eq.ll
index a2678154c5ba77..51096a507247b4 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_eq.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_eq.ll
@@ -500,24 +500,20 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    li a1, 2
-; RV32I-NEXT:    beq s1, a1, .LBB20_2
+; RV32I-NEXT:    beq s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ne.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ne.ll
index f93cf130934573..8ba437e9a1807c 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ne.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ne.ll
@@ -500,24 +500,20 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    li a1, 2
-; RV32I-NEXT:    bne s1, a1, .LBB20_2
+; RV32I-NEXT:    bne s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_sge.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_sge.ll
index 429ba499f4d780..b86a0f3b391f23 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_sge.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_sge.ll
@@ -492,24 +492,20 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    li a1, 1
-; RV32I-NEXT:    blt a1, s1, .LBB20_2
+; RV32I-NEXT:    blt a1, s0, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_slt.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_slt.ll
index 9a1b6ebb423972..cebfe80c0dabbb 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_slt.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_slt.ll
@@ -492,24 +492,20 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    li a1, 2
-; RV32I-NEXT:    blt s1, a1, .LBB20_2
+; RV32I-NEXT:    blt s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_uge.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_uge.ll
index 9ffa5498a71562..249a2c801e1d3e 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_uge.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_uge.ll
@@ -492,24 +492,20 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    li a1, 1
-; RV32I-NEXT:    bltu a1, s1, .LBB20_2
+; RV32I-NEXT:    bltu a1, s0, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ult.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ult.ll
index f9aa83ac1f8e53..f914c9f45fbf85 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ult.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_32_ult.ll
@@ -492,24 +492,20 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    li a1, 2
-; RV32I-NEXT:    bltu s1, a1, .LBB20_2
+; RV32I-NEXT:    bltu s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_eq.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_eq.ll
index 2f6595aba1f580..d1b826d519abe9 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_eq.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_eq.ll
@@ -537,25 +537,21 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    lui a1, 2
 ; RV32I-NEXT:    addi a1, a1, 1819
-; RV32I-NEXT:    beq s1, a1, .LBB20_2
+; RV32I-NEXT:    beq s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ne.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ne.ll
index c82f8c7b048dd3..81bd7c843124bd 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ne.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ne.ll
@@ -537,25 +537,21 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    lui a1, 2
 ; RV32I-NEXT:    addi a1, a1, 1819
-; RV32I-NEXT:    bne s1, a1, .LBB20_2
+; RV32I-NEXT:    bne s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_sge.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_sge.ll
index 039731bdf49cd6..501e5e7d4dbf96 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_sge.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_sge.ll
@@ -533,25 +533,21 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    lui a1, 2
 ; RV32I-NEXT:    addi a1, a1, 1818
-; RV32I-NEXT:    blt a1, s1, .LBB20_2
+; RV32I-NEXT:    blt a1, s0, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_slt.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_slt.ll
index c51c28781d8c86..e2a90009b5f6c7 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_slt.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_slt.ll
@@ -533,25 +533,21 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    lui a1, 2
 ; RV32I-NEXT:    addi a1, a1, 1819
-; RV32I-NEXT:    blt s1, a1, .LBB20_2
+; RV32I-NEXT:    blt s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_uge.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_uge.ll
index 3c7c7317e95560..5f6a7014941c41 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_uge.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_uge.ll
@@ -533,25 +533,21 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    lui a1, 2
 ; RV32I-NEXT:    addi a1, a1, 1818
-; RV32I-NEXT:    bltu a1, s1, .LBB20_2
+; RV32I-NEXT:    bltu a1, s0, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ult.ll b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ult.ll
index b08ecf91eb0972..96ad55054f23aa 100644
--- a/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ult.ll
+++ b/llvm/test/CodeGen/RISCV/short-forward-branch-opt-with-branch-with-immediates_48_ult.ll
@@ -533,25 +533,21 @@ define i32 @branch_with_immSFB_mul(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
 ; RV32I-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    .cfi_offset ra, -4
 ; RV32I-NEXT:    .cfi_offset s0, -8
-; RV32I-NEXT:    .cfi_offset s1, -12
-; RV32I-NEXT:    mv s1, a3
-; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s0, a3
+; RV32I-NEXT:    sw a2, 4(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    call __mulsi3
 ; RV32I-NEXT:    lui a1, 2
 ; RV32I-NEXT:    addi a1, a1, 1819
-; RV32I-NEXT:    bltu s1, a1, .LBB20_2
+; RV32I-NEXT:    bltu s0, a1, .LBB20_2
 ; RV32I-NEXT:  # %bb.1: # %entry
-; RV32I-NEXT:    mv a0, s0
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:  .LBB20_2: # %entry
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    .cfi_restore ra
 ; RV32I-NEXT:    .cfi_restore s0
-; RV32I-NEXT:    .cfi_restore s1
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/shrinkwrap-save-restore-debug.ll b/llvm/test/CodeGen/RISCV/shrinkwrap-save-restore-debug.ll
index 77ab591e7e4bb9..89da812f565d35 100644
--- a/llvm/test/CodeGen/RISCV/shrinkwrap-save-restore-debug.ll
+++ b/llvm/test/CodeGen/RISCV/shrinkwrap-save-restore-debug.ll
@@ -12,17 +12,19 @@ define void @without_debug(i32 %n) nounwind {
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    bltu a1, a0, .LBB0_2
 ; CHECK-NEXT:  # %bb.1: # %if.then
-; CHECK-NEXT:    call t0, __riscv_save_2
-; CHECK-NEXT:    addi s0, sp, 16
-; CHECK-NEXT:    mv s1, sp
+; CHECK-NEXT:    call t0, __riscv_save_1
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    addi s0, sp, 32
+; CHECK-NEXT:    sw sp, -20(s0) # 4-byte Folded Spill
 ; CHECK-NEXT:    addi a0, a0, 15
 ; CHECK-NEXT:    andi a0, a0, -16
 ; CHECK-NEXT:    sub a0, sp, a0
 ; CHECK-NEXT:    mv sp, a0
 ; CHECK-NEXT:    call notdead
-; CHECK-NEXT:    mv sp, s1
-; CHECK-NEXT:    addi sp, s0, -16
-; CHECK-NEXT:    tail __riscv_restore_2
+; CHECK-NEXT:    lw sp, -20(s0) # 4-byte Folded Reload
+; CHECK-NEXT:    addi sp, s0, -32
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    tail __riscv_restore_1
 ; CHECK-NEXT:  .LBB0_2: # %if.end
 ; CHECK-NEXT:    ret
 entry:
@@ -47,17 +49,19 @@ define void @with_debug(i32 %n) nounwind !dbg !4 {
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    bltu a1, a0, .LBB1_2
 ; CHECK-NEXT:  # %bb.1: # %if.then
-; CHECK-NEXT:    call t0, __riscv_save_2
-; CHECK-NEXT:    addi s0, sp, 16
-; CHECK-NEXT:    mv s1, sp
+; CHECK-NEXT:    call t0, __riscv_save_1
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    addi s0, sp, 32
+; CHECK-NEXT:    sw sp, -20(s0) # 4-byte Folded Spill
 ; CHECK-NEXT:    addi a0, a0, 15
 ; CHECK-NEXT:    andi a0, a0, -16
 ; CHECK-NEXT:    sub a0, sp, a0
 ; CHECK-NEXT:    mv sp, a0
 ; CHECK-NEXT:    call notdead
-; CHECK-NEXT:    mv sp, s1
-; CHECK-NEXT:    addi sp, s0, -16
-; CHECK-NEXT:    tail __riscv_restore_2
+; CHECK-NEXT:    lw sp, -20(s0) # 4-byte Folded Reload
+; CHECK-NEXT:    addi sp, s0, -32
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    tail __riscv_restore_1
 ; CHECK-NEXT:  .LBB1_2: # %if.end
 ; CHECK-NEXT:    #DEBUG_VALUE: with_debug:ghost <- 0
 ; CHECK-NEXT:    ret
diff --git a/llvm/test/CodeGen/RISCV/shrinkwrap.ll b/llvm/test/CodeGen/RISCV/shrinkwrap.ll
index e06673d44ea750..3af9fcf5fc7f11 100644
--- a/llvm/test/CodeGen/RISCV/shrinkwrap.ll
+++ b/llvm/test/CodeGen/RISCV/shrinkwrap.ll
@@ -173,104 +173,95 @@ define void @li_straightline_a(i32 zeroext %a, i32 zeroext %b) {
 ; RV32I-SW-NO-NEXT:    addi sp, sp, -16
 ; RV32I-SW-NO-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-SW-NO-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-SW-NO-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-SW-NO-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-SW-NO-NEXT:    .cfi_offset ra, -4
-; RV32I-SW-NO-NEXT:    .cfi_offset s0, -8
-; RV32I-SW-NO-NEXT:    .cfi_offset s1, -12
-; RV32I-SW-NO-NEXT:    li s1, 57
-; RV32I-SW-NO-NEXT:    beq a0, s1, .LBB2_3
+; RV32I-SW-NO-NEXT:    li a2, 57
+; RV32I-SW-NO-NEXT:    beq a0, a2, .LBB2_3
 ; RV32I-SW-NO-NEXT:  # %bb.1: # %do_call
-; RV32I-SW-NO-NEXT:    mv s0, a1
+; RV32I-SW-NO-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; RV32I-SW-NO-NEXT:    call foo
-; RV32I-SW-NO-NEXT:    beq s0, s1, .LBB2_3
+; RV32I-SW-NO-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32I-SW-NO-NEXT:    li a1, 57
+; RV32I-SW-NO-NEXT:    beq a0, a1, .LBB2_3
 ; RV32I-SW-NO-NEXT:  # %bb.2: # %do_call2
 ; RV32I-SW-NO-NEXT:    call foo
 ; RV32I-SW-NO-NEXT:  .LBB2_3: # %exit
 ; RV32I-SW-NO-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-SW-NO-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-SW-NO-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-SW-NO-NEXT:    .cfi_restore ra
-; RV32I-SW-NO-NEXT:    .cfi_restore s0
-; RV32I-SW-NO-NEXT:    .cfi_restore s1
 ; RV32I-SW-NO-NEXT:    addi sp, sp, 16
 ; RV32I-SW-NO-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-SW-NO-NEXT:    ret
 ;
 ; RV32I-SW-LABEL: li_straightline_a:
 ; RV32I-SW:       # %bb.0:
+; RV32I-SW-NEXT:    li a2, 57
+; RV32I-SW-NEXT:    beq a0, a2, .LBB2_4
+; RV32I-SW-NEXT:  # %bb.1: # %do_call
 ; RV32I-SW-NEXT:    addi sp, sp, -16
 ; RV32I-SW-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-SW-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-SW-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-SW-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
 ; RV32I-SW-NEXT:    .cfi_offset ra, -4
-; RV32I-SW-NEXT:    .cfi_offset s0, -8
-; RV32I-SW-NEXT:    .cfi_offset s1, -12
-; RV32I-SW-NEXT:    li s1, 57
-; RV32I-SW-NEXT:    beq a0, s1, .LBB2_3
-; RV32I-SW-NEXT:  # %bb.1: # %do_call
-; RV32I-SW-NEXT:    mv s0, a1
+; RV32I-SW-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; RV32I-SW-NEXT:    call foo
-; RV32I-SW-NEXT:    beq s0, s1, .LBB2_3
+; RV32I-SW-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32I-SW-NEXT:    li a1, 57
+; RV32I-SW-NEXT:    beq a0, a1, .LBB2_3
 ; RV32I-SW-NEXT:  # %bb.2: # %do_call2
 ; RV32I-SW-NEXT:    call foo
-; RV32I-SW-NEXT:  .LBB2_3: # %exit
+; RV32I-SW-NEXT:  .LBB2_3:
 ; RV32I-SW-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-SW-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
-; RV32I-SW-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
 ; RV32I-SW-NEXT:    .cfi_restore ra
-; RV32I-SW-NEXT:    .cfi_restore s0
-; RV32I-SW-NEXT:    .cfi_restore s1
 ; RV32I-SW-NEXT:    addi sp, sp, 16
 ; RV32I-SW-NEXT:    .cfi_def_cfa_offset 0
+; RV32I-SW-NEXT:  .LBB2_4: # %exit
 ; RV32I-SW-NEXT:    ret
 ;
 ; RV32I-SW-SR-LABEL: li_straightline_a:
 ; RV32I-SW-SR:       # %bb.0:
-; RV32I-SW-SR-NEXT:    call t0, __riscv_save_2
+; RV32I-SW-SR-NEXT:    li a2, 57
+; RV32I-SW-SR-NEXT:    beq a0, a2, .LBB2_4
+; RV32I-SW-SR-NEXT:  # %bb.1: # %do_call
+; RV32I-SW-SR-NEXT:    call t0, __riscv_save_0
 ; RV32I-SW-SR-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-SW-SR-NEXT:    .cfi_offset ra, -4
-; RV32I-SW-SR-NEXT:    .cfi_offset s0, -8
-; RV32I-SW-SR-NEXT:    .cfi_offset s1, -12
-; RV32I-SW-SR-NEXT:    li s1, 57
-; RV32I-SW-SR-NEXT:    beq a0, s1, .LBB2_3
-; RV32I-SW-SR-NEXT:  # %bb.1: # %do_call
-; RV32I-SW-SR-NEXT:    mv s0, a1
+; RV32I-SW-SR-NEXT:    addi sp, sp, -16
+; RV32I-SW-SR-NEXT:    .cfi_def_cfa_offset 32
+; RV32I-SW-SR-NEXT:    sw a1, 12(sp) # 4-byte Folded Spill
 ; RV32I-SW-SR-NEXT:    call foo
-; RV32I-SW-SR-NEXT:    beq s0, s1, .LBB2_3
+; RV32I-SW-SR-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-SW-SR-NEXT:    li a1, 57
+; RV32I-SW-SR-NEXT:    beq a0, a1, .LBB2_3
 ; RV32I-SW-SR-NEXT:  # %bb.2: # %do_call2
 ; RV32I-SW-SR-NEXT:    call foo
-; RV32I-SW-SR-NEXT:  .LBB2_3: # %exit
-; RV32I-SW-SR-NEXT:    tail __riscv_restore_2
+; RV32I-SW-SR-NEXT:  .LBB2_3:
+; RV32I-SW-SR-NEXT:    addi sp, sp, 16
+; RV32I-SW-SR-NEXT:    .cfi_def_cfa_offset 16
+; RV32I-SW-SR-NEXT:    tail __riscv_restore_0
+; RV32I-SW-SR-NEXT:    j .LBB2_4
+; RV32I-SW-SR-NEXT:  .LBB2_4: # %exit
+; RV32I-SW-SR-NEXT:    ret
 ;
 ; RV64I-SW-LABEL: li_straightline_a:
 ; RV64I-SW:       # %bb.0:
-; RV64I-SW-NEXT:    addi sp, sp, -32
-; RV64I-SW-NEXT:    .cfi_def_cfa_offset 32
-; RV64I-SW-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
-; RV64I-SW-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
-; RV64I-SW-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; RV64I-SW-NEXT:    .cfi_offset ra, -8
-; RV64I-SW-NEXT:    .cfi_offset s0, -16
-; RV64I-SW-NEXT:    .cfi_offset s1, -24
-; RV64I-SW-NEXT:    li s1, 57
-; RV64I-SW-NEXT:    beq a0, s1, .LBB2_3
+; RV64I-SW-NEXT:    li a2, 57
+; RV64I-SW-NEXT:    beq a0, a2, .LBB2_4
 ; RV64I-SW-NEXT:  # %bb.1: # %do_call
-; RV64I-SW-NEXT:    mv s0, a1
+; RV64I-SW-NEXT:    addi sp, sp, -16
+; RV64I-SW-NEXT:    .cfi_def_cfa_offset 16
+; RV64I-SW-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-SW-NEXT:    .cfi_offset ra, -8
+; RV64I-SW-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
 ; RV64I-SW-NEXT:    call foo
-; RV64I-SW-NEXT:    beq s0, s1, .LBB2_3
+; RV64I-SW-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-SW-NEXT:    li a1, 57
+; RV64I-SW-NEXT:    beq a0, a1, .LBB2_3
 ; RV64I-SW-NEXT:  # %bb.2: # %do_call2
 ; RV64I-SW-NEXT:    call foo
-; RV64I-SW-NEXT:  .LBB2_3: # %exit
-; RV64I-SW-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
-; RV64I-SW-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
-; RV64I-SW-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
+; RV64I-SW-NEXT:  .LBB2_3:
+; RV64I-SW-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64I-SW-NEXT:    .cfi_restore ra
-; RV64I-SW-NEXT:    .cfi_restore s0
-; RV64I-SW-NEXT:    .cfi_restore s1
-; RV64I-SW-NEXT:    addi sp, sp, 32
+; RV64I-SW-NEXT:    addi sp, sp, 16
 ; RV64I-SW-NEXT:    .cfi_def_cfa_offset 0
+; RV64I-SW-NEXT:  .LBB2_4: # %exit
 ; RV64I-SW-NEXT:    ret
   %cmp0 = icmp eq i32 %a, 57
   br i1 %cmp0, label %exit, label %do_call
@@ -294,23 +285,20 @@ define void @li_straightline_b(i32 zeroext %a, i32 zeroext %b) {
 ; RV32I-SW-NO-NEXT:    addi sp, sp, -16
 ; RV32I-SW-NO-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-SW-NO-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-SW-NO-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32I-SW-NO-NEXT:    .cfi_offset ra, -4
-; RV32I-SW-NO-NEXT:    .cfi_offset s0, -8
 ; RV32I-SW-NO-NEXT:    li a2, 57
 ; RV32I-SW-NO-NEXT:    beq a0, a2, .LBB3_3
 ; RV32I-SW-NO-NEXT:  # %bb.1: # %do_call
-; RV32I-SW-NO-NEXT:    mv s0, a1
+; RV32I-SW-NO-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; RV32I-SW-NO-NEXT:    call foo
-; RV32I-SW-NO-NEXT:    li a0, 57
-; RV32I-SW-NO-NEXT:    beq s0, a0, .LBB3_3
+; RV32I-SW-NO-NEXT:    li a1, 57
+; RV32I-SW-NO-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32I-SW-NO-NEXT:    beq a0, a1, .LBB3_3
 ; RV32I-SW-NO-NEXT:  # %bb.2: # %do_call2
 ; RV32I-SW-NO-NEXT:    call foo
 ; RV32I-SW-NO-NEXT:  .LBB3_3: # %exit
 ; RV32I-SW-NO-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-SW-NO-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32I-SW-NO-NEXT:    .cfi_restore ra
-; RV32I-SW-NO-NEXT:    .cfi_restore s0
 ; RV32I-SW-NO-NEXT:    addi sp, sp, 16
 ; RV32I-SW-NO-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-SW-NO-NEXT:    ret
@@ -323,20 +311,17 @@ define void @li_straightline_b(i32 zeroext %a, i32 zeroext %b) {
 ; RV32I-SW-NEXT:    addi sp, sp, -16
 ; RV32I-SW-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-SW-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
-; RV32I-SW-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
 ; RV32I-SW-NEXT:    .cfi_offset ra, -4
-; RV32I-SW-NEXT:    .cfi_offset s0, -8
-; RV32I-SW-NEXT:    mv s0, a1
+; RV32I-SW-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
 ; RV32I-SW-NEXT:    call foo
-; RV32I-SW-NEXT:    li a0, 57
-; RV32I-SW-NEXT:    beq s0, a0, .LBB3_3
+; RV32I-SW-NEXT:    li a1, 57
+; RV32I-SW-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32I-SW-NEXT:    beq a0, a1, .LBB3_3
 ; RV32I-SW-NEXT:  # %bb.2: # %do_call2
 ; RV32I-SW-NEXT:    call foo
 ; RV32I-SW-NEXT:  .LBB3_3:
 ; RV32I-SW-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
-; RV32I-SW-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
 ; RV32I-SW-NEXT:    .cfi_restore ra
-; RV32I-SW-NEXT:    .cfi_restore s0
 ; RV32I-SW-NEXT:    addi sp, sp, 16
 ; RV32I-SW-NEXT:    .cfi_def_cfa_offset 0
 ; RV32I-SW-NEXT:  .LBB3_4: # %exit
@@ -347,18 +332,22 @@ define void @li_straightline_b(i32 zeroext %a, i32 zeroext %b) {
 ; RV32I-SW-SR-NEXT:    li a2, 57
 ; RV32I-SW-SR-NEXT:    beq a0, a2, .LBB3_4
 ; RV32I-SW-SR-NEXT:  # %bb.1: # %do_call
-; RV32I-SW-SR-NEXT:    call t0, __riscv_save_1
+; RV32I-SW-SR-NEXT:    call t0, __riscv_save_0
 ; RV32I-SW-SR-NEXT:    .cfi_def_cfa_offset 16
 ; RV32I-SW-SR-NEXT:    .cfi_offset ra, -4
-; RV32I-SW-SR-NEXT:    .cfi_offset s0, -8
-; RV32I-SW-SR-NEXT:    mv s0, a1
+; RV32I-SW-SR-NEXT:    addi sp, sp, -16
+; RV32I-SW-SR-NEXT:    .cfi_def_cfa_offset 32
+; RV32I-SW-SR-NEXT:    sw a1, 12(sp) # 4-byte Folded Spill
 ; RV32I-SW-SR-NEXT:    call foo
-; RV32I-SW-SR-NEXT:    li a0, 57
-; RV32I-SW-SR-NEXT:    beq s0, a0, .LBB3_3
+; RV32I-SW-SR-NEXT:    li a1, 57
+; RV32I-SW-SR-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-SW-SR-NEXT:    beq a0, a1, .LBB3_3
 ; RV32I-SW-SR-NEXT:  # %bb.2: # %do_call2
 ; RV32I-SW-SR-NEXT:    call foo
 ; RV32I-SW-SR-NEXT:  .LBB3_3:
-; RV32I-SW-SR-NEXT:    tail __riscv_restore_1
+; RV32I-SW-SR-NEXT:    addi sp, sp, 16
+; RV32I-SW-SR-NEXT:    .cfi_def_cfa_offset 16
+; RV32I-SW-SR-NEXT:    tail __riscv_restore_0
 ; RV32I-SW-SR-NEXT:    j .LBB3_4
 ; RV32I-SW-SR-NEXT:  .LBB3_4: # %exit
 ; RV32I-SW-SR-NEXT:    ret
@@ -371,20 +360,17 @@ define void @li_straightline_b(i32 zeroext %a, i32 zeroext %b) {
 ; RV64I-SW-NEXT:    addi sp, sp, -16
 ; RV64I-SW-NEXT:    .cfi_def_cfa_offset 16
 ; RV64I-SW-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-SW-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
 ; RV64I-SW-NEXT:    .cfi_offset ra, -8
-; RV64I-SW-NEXT:    .cfi_offset s0, -16
-; RV64I-SW-NEXT:    mv s0, a1
+; RV64I-SW-NEXT:    sd a1, 0(sp) # 8-byte Folded Spill
 ; RV64I-SW-NEXT:    call foo
-; RV64I-SW-NEXT:    li a0, 57
-; RV64I-SW-NEXT:    beq s0, a0, .LBB3_3
+; RV64I-SW-NEXT:    li a1, 57
+; RV64I-SW-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
+; RV64I-SW-NEXT:    beq a0, a1, .LBB3_3
 ; RV64I-SW-NEXT:  # %bb.2: # %do_call2
 ; RV64I-SW-NEXT:    call foo
 ; RV64I-SW-NEXT:  .LBB3_3:
 ; RV64I-SW-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-SW-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
 ; RV64I-SW-NEXT:    .cfi_restore ra
-; RV64I-SW-NEXT:    .cfi_restore s0
 ; RV64I-SW-NEXT:    addi sp, sp, 16
 ; RV64I-SW-NEXT:    .cfi_def_cfa_offset 0
 ; RV64I-SW-NEXT:  .LBB3_4: # %exit
diff --git a/llvm/test/CodeGen/RISCV/simplify-condbr.ll b/llvm/test/CodeGen/RISCV/simplify-condbr.ll
index 4aadd034ff0f7f..f8c46dfbbec294 100644
--- a/llvm/test/CodeGen/RISCV/simplify-condbr.ll
+++ b/llvm/test/CodeGen/RISCV/simplify-condbr.ll
@@ -13,7 +13,6 @@ define fastcc i32 @S_regrepeat(ptr %startposp, i32 %max, i8 %0, i1 %cmp343) noun
 ; CHECK-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
-; CHECK-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    zext.b a2, a2
 ; CHECK-NEXT:    addi a4, a2, -19
 ; CHECK-NEXT:    li a5, 2
@@ -31,9 +30,9 @@ define fastcc i32 @S_regrepeat(ptr %startposp, i32 %max, i8 %0, i1 %cmp343) noun
 ; CHECK-NEXT:    bnez s0, .LBB0_9
 ; CHECK-NEXT:    j .LBB0_8
 ; CHECK-NEXT:  .LBB0_4: # %sw.bb336
-; CHECK-NEXT:    mv s1, a0
+; CHECK-NEXT:    sd a0, 0(sp) # 8-byte Folded Spill
 ; CHECK-NEXT:    li s0, 0
-; CHECK-NEXT:    andi s2, a3, 1
+; CHECK-NEXT:    andi s1, a3, 1
 ; CHECK-NEXT:  .LBB0_5: # %land.rhs345
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    call S_reginclass
@@ -42,10 +41,10 @@ define fastcc i32 @S_regrepeat(ptr %startposp, i32 %max, i8 %0, i1 %cmp343) noun
 ; CHECK-NEXT:  # %bb.6: # %while.body350
 ; CHECK-NEXT:    # in Loop: Header=BB0_5 Depth=1
 ; CHECK-NEXT:    addiw s0, s0, 1
-; CHECK-NEXT:    bnez s2, .LBB0_5
+; CHECK-NEXT:    bnez s1, .LBB0_5
 ; CHECK-NEXT:    j .LBB0_8
 ; CHECK-NEXT:  .LBB0_7:
-; CHECK-NEXT:    mv a0, s1
+; CHECK-NEXT:    ld a0, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    bnez s0, .LBB0_9
 ; CHECK-NEXT:  .LBB0_8: # %if.else1492
 ; CHECK-NEXT:    li a0, 0
@@ -53,7 +52,6 @@ define fastcc i32 @S_regrepeat(ptr %startposp, i32 %max, i8 %0, i1 %cmp343) noun
 ; CHECK-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
-; CHECK-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
 ; CHECK-NEXT:    addi sp, sp, 32
 ; CHECK-NEXT:    ret
 ; CHECK-NEXT:  .LBB0_10:
diff --git a/llvm/test/CodeGen/RISCV/smulh.ll b/llvm/test/CodeGen/RISCV/smulh.ll
index 48d84223cffe60..2c572b7deb945e 100644
--- a/llvm/test/CodeGen/RISCV/smulh.ll
+++ b/llvm/test/CodeGen/RISCV/smulh.ll
@@ -4414,7 +4414,7 @@ define <2 x i128> @smulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:    lw t2, 8(a2)
 ; RV32M-NEXT:    lw t1, 12(a2)
 ; RV32M-NEXT:    mulhu t4, t0, a5
-; RV32M-NEXT:    mul t6, a7, a5
+; RV32M-NEXT:    mul t5, a7, a5
 ; RV32M-NEXT:    mul s0, t0, a6
 ; RV32M-NEXT:    lw a3, 0(a1)
 ; RV32M-NEXT:    lw a4, 4(a1)
@@ -4426,277 +4426,279 @@ define <2 x i128> @smulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:    mulhu s5, a3, a5
 ; RV32M-NEXT:    mul s6, a5, t3
 ; RV32M-NEXT:    mulhu s7, a5, t3
-; RV32M-NEXT:    add t4, t6, t4
+; RV32M-NEXT:    add t4, t5, t4
 ; RV32M-NEXT:    mul s8, a6, t3
-; RV32M-NEXT:    add t5, s0, t4
-; RV32M-NEXT:    sltu t4, t4, t6
-; RV32M-NEXT:    sltu t6, t5, s0
+; RV32M-NEXT:    add t6, s0, t4
+; RV32M-NEXT:    sltu t4, t4, t5
+; RV32M-NEXT:    sltu t5, t6, s0
 ; RV32M-NEXT:    add t4, s1, t4
-; RV32M-NEXT:    add t6, s2, t6
+; RV32M-NEXT:    add t5, s2, t5
 ; RV32M-NEXT:    mul s0, a4, a5
-; RV32M-NEXT:    add t6, t4, t6
+; RV32M-NEXT:    add t5, t4, t5
 ; RV32M-NEXT:    mul s1, a3, a6
-; RV32M-NEXT:    add s2, s3, t6
-; RV32M-NEXT:    sltu t4, t6, t4
-; RV32M-NEXT:    sltu t6, s2, s3
+; RV32M-NEXT:    add s2, s3, t5
+; RV32M-NEXT:    sltu t4, t5, t4
+; RV32M-NEXT:    sltu t5, s2, s3
 ; RV32M-NEXT:    add t4, s4, t4
 ; RV32M-NEXT:    mulhu s3, a4, a5
 ; RV32M-NEXT:    add s7, s7, s6
 ; RV32M-NEXT:    mulhu s4, a3, a6
-; RV32M-NEXT:    add t4, t4, t6
+; RV32M-NEXT:    add t5, t4, t5
 ; RV32M-NEXT:    add s7, s7, s8
 ; RV32M-NEXT:    add s5, s0, s5
-; RV32M-NEXT:    mul t6, a4, a6
-; RV32M-NEXT:    sltu s0, s5, s0
-; RV32M-NEXT:    mulhu s8, a4, a6
+; RV32M-NEXT:    mul s8, a4, a6
+; RV32M-NEXT:    sltu t4, s5, s0
+; RV32M-NEXT:    mulhu a6, a4, a6
 ; RV32M-NEXT:    add s5, s1, s5
-; RV32M-NEXT:    mul s9, t0, a5
-; RV32M-NEXT:    add s0, s3, s0
+; RV32M-NEXT:    mul s0, t0, a5
+; RV32M-NEXT:    add s3, s3, t4
 ; RV32M-NEXT:    sltu a5, s5, s1
 ; RV32M-NEXT:    add a5, s4, a5
-; RV32M-NEXT:    add a6, s2, s6
-; RV32M-NEXT:    sltu s3, a6, s2
+; RV32M-NEXT:    add t4, s2, s6
+; RV32M-NEXT:    sltu s2, t4, s2
+; RV32M-NEXT:    add a5, s3, a5
+; RV32M-NEXT:    sltu s1, a5, s3
+; RV32M-NEXT:    add a5, s8, a5
+; RV32M-NEXT:    sltu s3, a5, s8
+; RV32M-NEXT:    add a6, a6, s1
+; RV32M-NEXT:    add a6, a6, s3
 ; RV32M-NEXT:    add a5, s0, a5
+; RV32M-NEXT:    add a6, t6, a6
 ; RV32M-NEXT:    sltu s0, a5, s0
-; RV32M-NEXT:    add a5, t6, a5
-; RV32M-NEXT:    sltu t6, a5, t6
-; RV32M-NEXT:    add s0, s8, s0
-; RV32M-NEXT:    add t6, s0, t6
-; RV32M-NEXT:    add a5, s9, a5
-; RV32M-NEXT:    add s0, t5, t6
-; RV32M-NEXT:    sltu t6, a5, s9
-; RV32M-NEXT:    add s7, t4, s7
-; RV32M-NEXT:    add t4, s0, t6
-; RV32M-NEXT:    add s3, s7, s3
-; RV32M-NEXT:    beq t4, t5, .LBB15_2
+; RV32M-NEXT:    add s7, t5, s7
+; RV32M-NEXT:    add t5, a6, s0
+; RV32M-NEXT:    add s2, s7, s2
+; RV32M-NEXT:    beq t5, t6, .LBB15_2
 ; RV32M-NEXT:  # %bb.1:
-; RV32M-NEXT:    sltu t6, t4, t5
+; RV32M-NEXT:    sltu s0, t5, t6
 ; RV32M-NEXT:  .LBB15_2:
-; RV32M-NEXT:    mulhu t5, a3, t2
-; RV32M-NEXT:    mul s0, a4, t2
-; RV32M-NEXT:    mulhu s2, a4, t2
-; RV32M-NEXT:    mul s4, a3, t1
-; RV32M-NEXT:    mulhu s5, a3, t1
-; RV32M-NEXT:    mul s6, a4, t1
-; RV32M-NEXT:    add t5, s0, t5
-; RV32M-NEXT:    srai s1, t1, 31
-; RV32M-NEXT:    sltu s0, t5, s0
+; RV32M-NEXT:    mulhu a6, a3, t2
+; RV32M-NEXT:    mul t6, a4, t2
+; RV32M-NEXT:    mulhu s1, a4, t2
+; RV32M-NEXT:    mul s3, a3, t1
+; RV32M-NEXT:    mulhu s4, a3, t1
+; RV32M-NEXT:    mul s5, a4, t1
+; RV32M-NEXT:    add s6, t6, a6
+; RV32M-NEXT:    srai a6, t1, 31
+; RV32M-NEXT:    sltu t6, s6, t6
 ; RV32M-NEXT:    mulhu s7, a4, t1
-; RV32M-NEXT:    add s0, s2, s0
-; RV32M-NEXT:    mul s2, s1, a3
-; RV32M-NEXT:    add t5, s4, t5
-; RV32M-NEXT:    mul s8, s1, a4
-; RV32M-NEXT:    sltu a4, t5, s4
-; RV32M-NEXT:    add s5, s5, a4
-; RV32M-NEXT:    mulhu s4, s1, a3
-; RV32M-NEXT:    add a4, a6, t6
-; RV32M-NEXT:    mul t6, a3, t2
-; RV32M-NEXT:    sltu s9, a4, a6
-; RV32M-NEXT:    add s5, s0, s5
-; RV32M-NEXT:    sltu a3, s5, s0
-; RV32M-NEXT:    add s5, s6, s5
+; RV32M-NEXT:    add s1, s1, t6
+; RV32M-NEXT:    mul s8, a6, a3
+; RV32M-NEXT:    add t6, s3, s6
+; RV32M-NEXT:    mul s6, a6, a4
+; RV32M-NEXT:    sltu a4, t6, s3
+; RV32M-NEXT:    add s4, s4, a4
+; RV32M-NEXT:    mulhu s3, a6, a3
+; RV32M-NEXT:    add a4, t4, s0
+; RV32M-NEXT:    mul s0, a3, t2
+; RV32M-NEXT:    sltu s9, a4, t4
+; RV32M-NEXT:    add s4, s1, s4
+; RV32M-NEXT:    sltu a3, s4, s1
+; RV32M-NEXT:    add s4, s5, s4
 ; RV32M-NEXT:    add s7, s7, a3
-; RV32M-NEXT:    add a3, s5, s2
-; RV32M-NEXT:    sltu a6, s5, s6
-; RV32M-NEXT:    sltu s0, a3, s5
-; RV32M-NEXT:    add a6, s7, a6
-; RV32M-NEXT:    add s4, s4, s8
-; RV32M-NEXT:    add s2, s4, s2
-; RV32M-NEXT:    add a5, t6, a5
-; RV32M-NEXT:    add t4, t5, t4
-; RV32M-NEXT:    sltu a5, a5, t6
-; RV32M-NEXT:    add s2, a6, s2
-; RV32M-NEXT:    add a6, t4, a5
-; RV32M-NEXT:    add s3, s3, s9
-; RV32M-NEXT:    add s8, s2, s0
-; RV32M-NEXT:    beq a6, t5, .LBB15_4
+; RV32M-NEXT:    add a3, s4, s8
+; RV32M-NEXT:    sltu t4, s4, s5
+; RV32M-NEXT:    sltu s1, a3, s4
+; RV32M-NEXT:    add t4, s7, t4
+; RV32M-NEXT:    add s3, s3, s6
+; RV32M-NEXT:    add s3, s3, s8
+; RV32M-NEXT:    add a5, s0, a5
+; RV32M-NEXT:    add t5, t6, t5
+; RV32M-NEXT:    sltu a5, a5, s0
+; RV32M-NEXT:    add s3, t4, s3
+; RV32M-NEXT:    add t4, t5, a5
+; RV32M-NEXT:    add s2, s2, s9
+; RV32M-NEXT:    add s7, s3, s1
+; RV32M-NEXT:    beq t4, t6, .LBB15_4
 ; RV32M-NEXT:  # %bb.3:
-; RV32M-NEXT:    sltu a5, a6, t5
+; RV32M-NEXT:    sltu a5, t4, t6
 ; RV32M-NEXT:  .LBB15_4:
-; RV32M-NEXT:    mulhu s9, t0, t2
-; RV32M-NEXT:    mul s5, a7, t2
-; RV32M-NEXT:    mul s6, t0, t1
+; RV32M-NEXT:    mulhu s8, t0, t2
+; RV32M-NEXT:    mul s4, a7, t2
+; RV32M-NEXT:    mul s5, t0, t1
 ; RV32M-NEXT:    mul t4, t0, t2
 ; RV32M-NEXT:    add a5, a3, a5
 ; RV32M-NEXT:    sltu a3, a5, a3
-; RV32M-NEXT:    add s8, s8, a3
+; RV32M-NEXT:    add s7, s7, a3
 ; RV32M-NEXT:    add a3, a4, a5
-; RV32M-NEXT:    add a6, s3, s8
-; RV32M-NEXT:    sltu s4, a3, a4
-; RV32M-NEXT:    add a6, a6, s4
-; RV32M-NEXT:    add s9, s5, s9
-; RV32M-NEXT:    add s10, s6, s9
+; RV32M-NEXT:    add s6, s2, s7
+; RV32M-NEXT:    sltu s3, a3, a4
+; RV32M-NEXT:    add s6, s6, s3
+; RV32M-NEXT:    add s8, s4, s8
+; RV32M-NEXT:    add s9, s5, s8
 ; RV32M-NEXT:    add a3, t4, a3
-; RV32M-NEXT:    add a4, s10, a6
+; RV32M-NEXT:    add a4, s9, s6
 ; RV32M-NEXT:    sw a3, 4(sp) # 4-byte Folded Spill
 ; RV32M-NEXT:    sltu a0, a3, t4
 ; RV32M-NEXT:    add a4, a4, a0
-; RV32M-NEXT:    beq a4, s10, .LBB15_6
+; RV32M-NEXT:    beq a4, s9, .LBB15_6
 ; RV32M-NEXT:  # %bb.5:
-; RV32M-NEXT:    sltu a0, a4, s10
+; RV32M-NEXT:    sltu a0, a4, s9
 ; RV32M-NEXT:  .LBB15_6:
 ; RV32M-NEXT:    sw a4, 0(sp) # 4-byte Folded Spill
 ; RV32M-NEXT:    lw s0, 16(a2)
-; RV32M-NEXT:    lw s2, 20(a2)
+; RV32M-NEXT:    lw s1, 20(a2)
 ; RV32M-NEXT:    lw t4, 16(a1)
 ; RV32M-NEXT:    lw t5, 20(a1)
 ; RV32M-NEXT:    lw a5, 24(a1)
 ; RV32M-NEXT:    lw a1, 28(a1)
-; RV32M-NEXT:    beq a6, s3, .LBB15_8
+; RV32M-NEXT:    beq s6, s2, .LBB15_8
 ; RV32M-NEXT:  # %bb.7:
-; RV32M-NEXT:    sltu s4, a6, s3
+; RV32M-NEXT:    sltu s3, s6, s2
 ; RV32M-NEXT:  .LBB15_8:
-; RV32M-NEXT:    mulhu s11, a7, t2
-; RV32M-NEXT:    mulhu ra, t0, t1
-; RV32M-NEXT:    mul a3, a7, t1
-; RV32M-NEXT:    mul a4, t2, t3
-; RV32M-NEXT:    mul s7, s1, t0
-; RV32M-NEXT:    lw a6, 24(a2)
+; RV32M-NEXT:    mulhu s10, a7, t2
+; RV32M-NEXT:    mulhu s11, t0, t1
+; RV32M-NEXT:    mul ra, a7, t1
+; RV32M-NEXT:    mul a3, t2, t3
+; RV32M-NEXT:    mul s6, a6, t0
+; RV32M-NEXT:    lw a4, 24(a2)
 ; RV32M-NEXT:    lw a2, 28(a2)
-; RV32M-NEXT:    srai s3, s3, 31
-; RV32M-NEXT:    srai s8, s8, 31
+; RV32M-NEXT:    srai s2, s2, 31
+; RV32M-NEXT:    srai s7, s7, 31
+; RV32M-NEXT:    sltu s4, s8, s4
 ; RV32M-NEXT:    sltu s5, s9, s5
-; RV32M-NEXT:    sltu s6, s10, s6
-; RV32M-NEXT:    mulhu s9, t2, t3
+; RV32M-NEXT:    mulhu s8, t2, t3
+; RV32M-NEXT:    add s4, s10, s4
 ; RV32M-NEXT:    add s5, s11, s5
-; RV32M-NEXT:    add s6, ra, s6
-; RV32M-NEXT:    add s8, s3, s8
-; RV32M-NEXT:    add s6, s5, s6
-; RV32M-NEXT:    add s10, s7, a4
-; RV32M-NEXT:    add s11, a3, s6
-; RV32M-NEXT:    add s4, s8, s4
-; RV32M-NEXT:    add t2, s11, s10
-; RV32M-NEXT:    sltu s3, s8, s3
-; RV32M-NEXT:    mul ra, t1, t3
-; RV32M-NEXT:    add t3, s8, s3
-; RV32M-NEXT:    sltu s3, s4, s8
-; RV32M-NEXT:    mul s8, s1, a7
-; RV32M-NEXT:    mulhu t6, s1, t0
-; RV32M-NEXT:    add t0, t2, s4
-; RV32M-NEXT:    add t3, t3, s3
-; RV32M-NEXT:    sltu s1, t2, s11
+; RV32M-NEXT:    add s7, s2, s7
+; RV32M-NEXT:    add s5, s4, s5
+; RV32M-NEXT:    add s9, s6, a3
+; RV32M-NEXT:    add s10, ra, s5
+; RV32M-NEXT:    add s3, s7, s3
+; RV32M-NEXT:    add t2, s10, s9
+; RV32M-NEXT:    sltu s2, s7, s2
+; RV32M-NEXT:    mul s11, t1, t3
+; RV32M-NEXT:    add t3, s7, s2
+; RV32M-NEXT:    sltu s2, s3, s7
+; RV32M-NEXT:    mul s7, a6, a7
+; RV32M-NEXT:    mulhu a6, a6, t0
+; RV32M-NEXT:    add t0, t2, s3
+; RV32M-NEXT:    add t3, t3, s2
+; RV32M-NEXT:    sltu s2, t2, s10
 ; RV32M-NEXT:    mulhu a7, a7, t1
-; RV32M-NEXT:    sltu a3, s11, a3
-; RV32M-NEXT:    add a4, s9, a4
-; RV32M-NEXT:    sltu t1, s6, s5
-; RV32M-NEXT:    add a4, a4, ra
-; RV32M-NEXT:    mulhu s3, a5, s0
-; RV32M-NEXT:    mul s4, a1, s0
-; RV32M-NEXT:    add t6, t6, s8
-; RV32M-NEXT:    mul s5, a5, s2
-; RV32M-NEXT:    add t6, t6, s7
-; RV32M-NEXT:    mulhu s6, a1, s0
-; RV32M-NEXT:    add a7, a7, t1
-; RV32M-NEXT:    add a4, t6, a4
-; RV32M-NEXT:    sltu t1, s10, s7
-; RV32M-NEXT:    mulhu t6, a5, s2
-; RV32M-NEXT:    add a3, a7, a3
-; RV32M-NEXT:    add a4, a4, t1
-; RV32M-NEXT:    add s3, s4, s3
-; RV32M-NEXT:    mul s7, a1, s2
-; RV32M-NEXT:    sltu a7, s3, s4
-; RV32M-NEXT:    add s4, s5, s3
-; RV32M-NEXT:    add s6, s6, a7
-; RV32M-NEXT:    sltu t1, s4, s5
+; RV32M-NEXT:    sltu t1, s10, ra
+; RV32M-NEXT:    add a3, s8, a3
+; RV32M-NEXT:    sltu s3, s5, s4
+; RV32M-NEXT:    add a3, a3, s11
+; RV32M-NEXT:    mulhu s4, a5, s0
+; RV32M-NEXT:    mul s5, a1, s0
+; RV32M-NEXT:    add a6, a6, s7
+; RV32M-NEXT:    mul s7, a5, s1
+; RV32M-NEXT:    add a6, a6, s6
+; RV32M-NEXT:    mulhu s8, a1, s0
+; RV32M-NEXT:    add a7, a7, s3
+; RV32M-NEXT:    add a3, a6, a3
+; RV32M-NEXT:    sltu a6, s9, s6
+; RV32M-NEXT:    mulhu s3, a5, s1
+; RV32M-NEXT:    add t1, a7, t1
+; RV32M-NEXT:    add a3, a3, a6
+; RV32M-NEXT:    add s4, s5, s4
+; RV32M-NEXT:    mul a6, a1, s1
+; RV32M-NEXT:    sltu a7, s4, s5
+; RV32M-NEXT:    add s5, s7, s4
+; RV32M-NEXT:    add s8, s8, a7
+; RV32M-NEXT:    sltu s4, s5, s7
 ; RV32M-NEXT:    srai a7, a1, 31
-; RV32M-NEXT:    mulhu s3, a1, s2
-; RV32M-NEXT:    add t6, t6, t1
-; RV32M-NEXT:    mul s5, s0, a7
-; RV32M-NEXT:    add t1, a3, a4
-; RV32M-NEXT:    add t6, s6, t6
-; RV32M-NEXT:    add a3, s7, t6
-; RV32M-NEXT:    mulhu a4, s0, a7
-; RV32M-NEXT:    mulhu s8, t4, s0
-; RV32M-NEXT:    mul s9, t5, s0
-; RV32M-NEXT:    sltu s7, a3, s7
-; RV32M-NEXT:    sltu t6, t6, s6
-; RV32M-NEXT:    add t6, s3, t6
-; RV32M-NEXT:    mul s6, t4, s2
-; RV32M-NEXT:    add s3, a3, s5
-; RV32M-NEXT:    add t6, t6, s7
-; RV32M-NEXT:    mulhu s7, t5, s0
-; RV32M-NEXT:    mulhu s10, t4, s2
-; RV32M-NEXT:    add a4, a4, s5
-; RV32M-NEXT:    mul s5, s2, a7
-; RV32M-NEXT:    add s8, s9, s8
-; RV32M-NEXT:    mul s11, t5, s2
-; RV32M-NEXT:    sltu s9, s8, s9
-; RV32M-NEXT:    mulhu s2, t5, s2
-; RV32M-NEXT:    add s8, s6, s8
+; RV32M-NEXT:    mulhu s6, a1, s1
+; RV32M-NEXT:    add s3, s3, s4
+; RV32M-NEXT:    mul s4, s0, a7
+; RV32M-NEXT:    add t1, t1, a3
+; RV32M-NEXT:    add s3, s8, s3
+; RV32M-NEXT:    add a3, a6, s3
+; RV32M-NEXT:    mulhu s7, s0, a7
+; RV32M-NEXT:    mulhu s9, t4, s0
+; RV32M-NEXT:    mul s10, t5, s0
+; RV32M-NEXT:    sltu a6, a3, a6
+; RV32M-NEXT:    sltu s3, s3, s8
+; RV32M-NEXT:    add s3, s6, s3
+; RV32M-NEXT:    mul s6, t4, s1
+; RV32M-NEXT:    add s8, a3, s4
+; RV32M-NEXT:    add a6, s3, a6
+; RV32M-NEXT:    mulhu s11, t5, s0
+; RV32M-NEXT:    mulhu ra, t4, s1
+; RV32M-NEXT:    add s4, s7, s4
+; RV32M-NEXT:    mul s7, s1, a7
+; RV32M-NEXT:    add s9, s10, s9
+; RV32M-NEXT:    mul t6, t5, s1
+; RV32M-NEXT:    sltu s10, s9, s10
+; RV32M-NEXT:    mulhu s1, t5, s1
+; RV32M-NEXT:    add s9, s6, s9
 ; RV32M-NEXT:    mul s0, a5, s0
-; RV32M-NEXT:    sltu a3, s3, a3
-; RV32M-NEXT:    sltu s6, s8, s6
-; RV32M-NEXT:    add s7, s7, s9
+; RV32M-NEXT:    mv s3, s8
+; RV32M-NEXT:    sltu a3, s8, a3
+; RV32M-NEXT:    sltu s6, s9, s6
+; RV32M-NEXT:    add s10, s11, s10
+; RV32M-NEXT:    add s6, ra, s6
+; RV32M-NEXT:    add s7, s4, s7
 ; RV32M-NEXT:    add s6, s10, s6
-; RV32M-NEXT:    add a4, a4, s5
-; RV32M-NEXT:    add s6, s7, s6
-; RV32M-NEXT:    add s5, s11, s6
-; RV32M-NEXT:    sltu s6, s6, s7
-; RV32M-NEXT:    sltu s7, s5, s11
-; RV32M-NEXT:    add s2, s2, s6
-; RV32M-NEXT:    add s7, s2, s7
-; RV32M-NEXT:    add s2, s0, s5
-; RV32M-NEXT:    add s5, s4, s7
-; RV32M-NEXT:    sltu s6, s2, s0
-; RV32M-NEXT:    add a4, t6, a4
-; RV32M-NEXT:    add s5, s5, s6
-; RV32M-NEXT:    add s0, s1, t3
-; RV32M-NEXT:    add t3, a4, a3
-; RV32M-NEXT:    beq s5, s4, .LBB15_10
+; RV32M-NEXT:    add s4, t6, s6
+; RV32M-NEXT:    sltu s6, s6, s10
+; RV32M-NEXT:    sltu t6, s4, t6
+; RV32M-NEXT:    add s1, s1, s6
+; RV32M-NEXT:    add t6, s1, t6
+; RV32M-NEXT:    add s4, s0, s4
+; RV32M-NEXT:    add t6, s5, t6
+; RV32M-NEXT:    sltu s6, s4, s0
+; RV32M-NEXT:    add a6, a6, s7
+; RV32M-NEXT:    add t6, t6, s6
+; RV32M-NEXT:    add s0, s2, t3
+; RV32M-NEXT:    mv s2, t6
+; RV32M-NEXT:    add t3, a6, a3
+; RV32M-NEXT:    beq t6, s5, .LBB15_10
 ; RV32M-NEXT:  # %bb.9:
-; RV32M-NEXT:    sltu s6, s5, s4
+; RV32M-NEXT:    sltu s6, s2, s5
 ; RV32M-NEXT:  .LBB15_10:
-; RV32M-NEXT:    mulhu a3, t4, a6
-; RV32M-NEXT:    mul a4, t5, a6
-; RV32M-NEXT:    mulhu s4, t5, a6
+; RV32M-NEXT:    mulhu a3, t4, a4
+; RV32M-NEXT:    mul a6, t5, a4
+; RV32M-NEXT:    mulhu s5, t5, a4
 ; RV32M-NEXT:    mul s7, t4, a2
 ; RV32M-NEXT:    add t6, t0, a0
 ; RV32M-NEXT:    add s0, t1, s0
 ; RV32M-NEXT:    sltu s1, t0, t2
 ; RV32M-NEXT:    mulhu a0, t4, a2
 ; RV32M-NEXT:    mul s8, t5, a2
-; RV32M-NEXT:    add a3, a4, a3
+; RV32M-NEXT:    add a3, a6, a3
 ; RV32M-NEXT:    srai t1, a2, 31
-; RV32M-NEXT:    sltu a4, a3, a4
+; RV32M-NEXT:    sltu a6, a3, a6
 ; RV32M-NEXT:    mulhu s9, t5, a2
-; RV32M-NEXT:    add a4, s4, a4
+; RV32M-NEXT:    add a6, s5, a6
 ; RV32M-NEXT:    mul s10, t1, t4
-; RV32M-NEXT:    add s4, s7, a3
+; RV32M-NEXT:    add s5, s7, a3
 ; RV32M-NEXT:    mul a3, t1, t5
-; RV32M-NEXT:    sltu t2, s4, s7
+; RV32M-NEXT:    sltu t2, s5, s7
 ; RV32M-NEXT:    add a0, a0, t2
 ; RV32M-NEXT:    mulhu t5, t1, t4
 ; RV32M-NEXT:    add t2, s3, s6
-; RV32M-NEXT:    mul s6, t4, a6
+; RV32M-NEXT:    mul s6, t4, a4
 ; RV32M-NEXT:    sltu s7, t2, s3
-; RV32M-NEXT:    add a0, a4, a0
-; RV32M-NEXT:    sltu a4, a0, a4
+; RV32M-NEXT:    add a0, a6, a0
+; RV32M-NEXT:    sltu a6, a0, a6
 ; RV32M-NEXT:    add a0, s8, a0
-; RV32M-NEXT:    add a4, s9, a4
+; RV32M-NEXT:    add a6, s9, a6
 ; RV32M-NEXT:    add t4, a0, s10
 ; RV32M-NEXT:    sltu s3, a0, s8
 ; RV32M-NEXT:    sltu a0, t4, a0
-; RV32M-NEXT:    add a4, a4, s3
+; RV32M-NEXT:    add a6, a6, s3
 ; RV32M-NEXT:    add a3, t5, a3
 ; RV32M-NEXT:    add a3, a3, s10
-; RV32M-NEXT:    add s2, s6, s2
-; RV32M-NEXT:    add s3, s4, s5
-; RV32M-NEXT:    sltu t5, s2, s6
-; RV32M-NEXT:    add a3, a4, a3
-; RV32M-NEXT:    add s3, s3, t5
+; RV32M-NEXT:    add s4, s6, s4
+; RV32M-NEXT:    add s2, s5, s2
+; RV32M-NEXT:    sltu t5, s4, s6
+; RV32M-NEXT:    add a3, a6, a3
+; RV32M-NEXT:    add s3, s2, t5
 ; RV32M-NEXT:    add s2, t3, s7
-; RV32M-NEXT:    add s5, a3, a0
-; RV32M-NEXT:    beq s3, s4, .LBB15_12
+; RV32M-NEXT:    add t3, a3, a0
+; RV32M-NEXT:    beq s3, s5, .LBB15_12
 ; RV32M-NEXT:  # %bb.11:
-; RV32M-NEXT:    sltu t5, s3, s4
+; RV32M-NEXT:    sltu t5, s3, s5
 ; RV32M-NEXT:  .LBB15_12:
-; RV32M-NEXT:    mulhu s6, a5, a6
-; RV32M-NEXT:    mul s3, a1, a6
+; RV32M-NEXT:    mulhu s6, a5, a4
+; RV32M-NEXT:    mul s3, a1, a4
 ; RV32M-NEXT:    mul s4, a5, a2
-; RV32M-NEXT:    mul a0, a5, a6
+; RV32M-NEXT:    mul a0, a5, a4
 ; RV32M-NEXT:    add t5, t4, t5
-; RV32M-NEXT:    sltu a3, t5, t4
-; RV32M-NEXT:    add s5, s5, a3
+; RV32M-NEXT:    sltu s5, t5, t4
+; RV32M-NEXT:    add s5, t3, s5
 ; RV32M-NEXT:    add a3, t2, t5
 ; RV32M-NEXT:    add s8, s2, s5
 ; RV32M-NEXT:    sltu t5, a3, t2
@@ -4718,21 +4720,21 @@ define <2 x i128> @smulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:  # %bb.15:
 ; RV32M-NEXT:    sltu t5, s8, s2
 ; RV32M-NEXT:  .LBB15_16:
-; RV32M-NEXT:    mulhu a0, a1, a6
+; RV32M-NEXT:    mulhu a0, a1, a4
 ; RV32M-NEXT:    mulhu a3, a5, a2
-; RV32M-NEXT:    srai a4, s2, 31
+; RV32M-NEXT:    srai a6, s2, 31
 ; RV32M-NEXT:    mul s0, a1, a2
-; RV32M-NEXT:    mul s1, a6, a7
+; RV32M-NEXT:    mul s1, a4, a7
 ; RV32M-NEXT:    mul s2, t1, a5
 ; RV32M-NEXT:    srai s5, s5, 31
 ; RV32M-NEXT:    sltu s3, s6, s3
 ; RV32M-NEXT:    sltu s4, s7, s4
 ; RV32M-NEXT:    mulhu s6, a1, a2
-; RV32M-NEXT:    mulhu a6, a6, a7
+; RV32M-NEXT:    mulhu a4, a4, a7
 ; RV32M-NEXT:    mul a1, t1, a1
 ; RV32M-NEXT:    add a0, a0, s3
 ; RV32M-NEXT:    add a3, a3, s4
-; RV32M-NEXT:    add s5, a4, s5
+; RV32M-NEXT:    add s5, a6, s5
 ; RV32M-NEXT:    add a3, a0, a3
 ; RV32M-NEXT:    add s3, s2, s1
 ; RV32M-NEXT:    add s4, s0, a3
@@ -4740,27 +4742,27 @@ define <2 x i128> @smulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:    add s7, s4, s3
 ; RV32M-NEXT:    mulhu a5, t1, a5
 ; RV32M-NEXT:    add t1, s7, t5
-; RV32M-NEXT:    sltu a4, s5, a4
+; RV32M-NEXT:    sltu a6, s5, a6
 ; RV32M-NEXT:    mul a2, a2, a7
-; RV32M-NEXT:    add a4, s5, a4
+; RV32M-NEXT:    add a6, s5, a6
 ; RV32M-NEXT:    sltu a7, t5, s5
 ; RV32M-NEXT:    add t4, t1, t4
-; RV32M-NEXT:    add a4, a4, a7
+; RV32M-NEXT:    add a6, a6, a7
 ; RV32M-NEXT:    sltu a7, s7, s4
 ; RV32M-NEXT:    sltu a0, a3, a0
 ; RV32M-NEXT:    sltu a3, s4, s0
 ; RV32M-NEXT:    add a0, s6, a0
-; RV32M-NEXT:    add a6, a6, s1
+; RV32M-NEXT:    add a4, a4, s1
 ; RV32M-NEXT:    add a1, a5, a1
-; RV32M-NEXT:    add a2, a6, a2
+; RV32M-NEXT:    add a2, a4, a2
 ; RV32M-NEXT:    add a1, a1, s2
 ; RV32M-NEXT:    add a1, a1, a2
 ; RV32M-NEXT:    sltu a2, s3, s2
 ; RV32M-NEXT:    add a0, a0, a3
 ; RV32M-NEXT:    add a1, a1, a2
 ; RV32M-NEXT:    add a0, a0, a1
-; RV32M-NEXT:    add a4, a7, a4
-; RV32M-NEXT:    add a0, a0, a4
+; RV32M-NEXT:    add a6, a7, a6
+; RV32M-NEXT:    add a0, a0, a6
 ; RV32M-NEXT:    sltu a1, t1, s7
 ; RV32M-NEXT:    sltu a2, t4, t1
 ; RV32M-NEXT:    add a0, a0, a1
diff --git a/llvm/test/CodeGen/RISCV/umulh.ll b/llvm/test/CodeGen/RISCV/umulh.ll
index 6c1ac2cadc394c..6d77d960059134 100644
--- a/llvm/test/CodeGen/RISCV/umulh.ll
+++ b/llvm/test/CodeGen/RISCV/umulh.ll
@@ -956,62 +956,64 @@ define i128 @umulh_i128(i128 %x, i128 %y) nounwind {
 ; RV32M-NEXT:    sltu t3, t6, s1
 ; RV32M-NEXT:  .LBB4_2:
 ; RV32M-NEXT:    mulhu t0, a1, t0
-; RV32M-NEXT:    mulhu s4, a3, a6
-; RV32M-NEXT:    mulhu s5, a7, a4
+; RV32M-NEXT:    mulhu s3, a3, a6
+; RV32M-NEXT:    mulhu s4, a7, a4
 ; RV32M-NEXT:    mul s2, a5, a4
-; RV32M-NEXT:    mul s3, a7, a2
+; RV32M-NEXT:    mul s5, a7, a2
 ; RV32M-NEXT:    mul s6, a7, a4
 ; RV32M-NEXT:    mul t5, a1, a6
 ; RV32M-NEXT:    sltu t1, s0, t1
 ; RV32M-NEXT:    sltu s0, s1, t2
+; RV32M-NEXT:    mv s1, s5
 ; RV32M-NEXT:    add t2, t0, t1
-; RV32M-NEXT:    add s0, s4, s0
-; RV32M-NEXT:    add s1, s2, s5
+; RV32M-NEXT:    add s0, s3, s0
+; RV32M-NEXT:    add s3, s2, s4
 ; RV32M-NEXT:    add s0, t2, s0
-; RV32M-NEXT:    add s4, s3, s1
+; RV32M-NEXT:    add s4, s5, s3
 ; RV32M-NEXT:    add t4, s6, t4
 ; RV32M-NEXT:    add t6, s4, t6
 ; RV32M-NEXT:    sltu t1, t4, s6
 ; RV32M-NEXT:    add t4, t5, s0
 ; RV32M-NEXT:    add t6, t6, t1
 ; RV32M-NEXT:    add t0, t4, t3
+; RV32M-NEXT:    mv t3, s4
 ; RV32M-NEXT:    beq t6, s4, .LBB4_4
 ; RV32M-NEXT:  # %bb.3:
-; RV32M-NEXT:    sltu t1, t6, s4
+; RV32M-NEXT:    sltu t1, t6, t3
 ; RV32M-NEXT:  .LBB4_4:
-; RV32M-NEXT:    mulhu t3, a5, a4
+; RV32M-NEXT:    mulhu t6, a5, a4
 ; RV32M-NEXT:    mulhu a7, a7, a2
-; RV32M-NEXT:    sltu t6, s1, s2
-; RV32M-NEXT:    sltu s1, s4, s3
-; RV32M-NEXT:    mul s2, a5, a2
+; RV32M-NEXT:    sltu s2, s3, s2
+; RV32M-NEXT:    sltu t3, t3, s1
+; RV32M-NEXT:    mul s1, a5, a2
 ; RV32M-NEXT:    sltu s3, t0, t4
 ; RV32M-NEXT:    sltu t4, t4, t5
 ; RV32M-NEXT:    sltu t2, s0, t2
 ; RV32M-NEXT:    mulhu t5, a1, a6
 ; RV32M-NEXT:    mulhu a6, a5, a2
 ; RV32M-NEXT:    mulhu s0, a3, a4
-; RV32M-NEXT:    add t3, t3, t6
-; RV32M-NEXT:    add a7, a7, s1
+; RV32M-NEXT:    add t6, t6, s2
+; RV32M-NEXT:    add a7, a7, t3
 ; RV32M-NEXT:    add t2, t2, t4
-; RV32M-NEXT:    add a7, t3, a7
+; RV32M-NEXT:    add a7, t6, a7
 ; RV32M-NEXT:    add t2, t2, s3
-; RV32M-NEXT:    add t4, s2, a7
-; RV32M-NEXT:    sltu a7, a7, t3
+; RV32M-NEXT:    add t3, s1, a7
+; RV32M-NEXT:    sltu a7, a7, t6
 ; RV32M-NEXT:    mul a5, a1, a4
-; RV32M-NEXT:    add t3, t4, t1
+; RV32M-NEXT:    add t4, t3, t1
 ; RV32M-NEXT:    mul t6, a3, a4
-; RV32M-NEXT:    add s1, t0, t3
+; RV32M-NEXT:    add s2, t0, t4
 ; RV32M-NEXT:    mul t1, a3, a2
-; RV32M-NEXT:    sltu s2, t4, s2
+; RV32M-NEXT:    sltu s1, t3, s1
 ; RV32M-NEXT:    add a6, a6, a7
-; RV32M-NEXT:    sltu a7, t3, t4
-; RV32M-NEXT:    add a6, a6, s2
+; RV32M-NEXT:    sltu a7, t4, t3
+; RV32M-NEXT:    add a6, a6, s1
 ; RV32M-NEXT:    add t5, t5, t2
 ; RV32M-NEXT:    add a6, a6, a7
-; RV32M-NEXT:    sltu t0, s1, t0
+; RV32M-NEXT:    sltu t0, s2, t0
 ; RV32M-NEXT:    add a7, t5, a6
 ; RV32M-NEXT:    add t3, a5, s0
-; RV32M-NEXT:    add a6, t6, s1
+; RV32M-NEXT:    add a6, t6, s2
 ; RV32M-NEXT:    add t4, t1, t3
 ; RV32M-NEXT:    sltu t2, a6, t6
 ; RV32M-NEXT:    add t6, a7, t0
@@ -4078,6 +4080,7 @@ define <2 x i128> @umulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:  # %bb.5:
 ; RV32M-NEXT:    sltu t5, a4, s4
 ; RV32M-NEXT:  .LBB15_6:
+; RV32M-NEXT:    sw a4, 4(sp) # 4-byte Folded Spill
 ; RV32M-NEXT:    lw t4, 16(a1)
 ; RV32M-NEXT:    lw t2, 20(a1)
 ; RV32M-NEXT:    lw a5, 24(a1)
@@ -4114,7 +4117,7 @@ define <2 x i128> @umulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:    mulhu s7, a5, s0
 ; RV32M-NEXT:    mul a7, a1, s0
 ; RV32M-NEXT:    mul t1, a5, t3
-; RV32M-NEXT:    mul s8, a5, s0
+; RV32M-NEXT:    mul s10, a5, s0
 ; RV32M-NEXT:    add s1, a6, s2
 ; RV32M-NEXT:    add s3, s6, s3
 ; RV32M-NEXT:    sltu a6, s3, s6
@@ -4125,80 +4128,82 @@ define <2 x i128> @umulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:    sltu s2, s3, s4
 ; RV32M-NEXT:    add s2, s5, s2
 ; RV32M-NEXT:    add s5, a7, s7
-; RV32M-NEXT:    add s6, t1, s5
-; RV32M-NEXT:    add s3, s8, s3
-; RV32M-NEXT:    add s4, s6, s2
-; RV32M-NEXT:    sltu s2, s3, s8
-; RV32M-NEXT:    add s8, s9, t0
+; RV32M-NEXT:    add s8, t1, s5
+; RV32M-NEXT:    add a4, s10, s3
+; RV32M-NEXT:    add s4, s8, s2
+; RV32M-NEXT:    sltu s2, a4, s10
+; RV32M-NEXT:    add s7, s9, t0
 ; RV32M-NEXT:    add s4, s4, s2
-; RV32M-NEXT:    add t0, s8, t6
-; RV32M-NEXT:    sltu t6, s8, s9
-; RV32M-NEXT:    beq s4, s6, .LBB15_10
+; RV32M-NEXT:    add t0, s7, t6
+; RV32M-NEXT:    sltu t6, s7, s9
+; RV32M-NEXT:    beq s4, s8, .LBB15_10
 ; RV32M-NEXT:  # %bb.9:
-; RV32M-NEXT:    sltu s2, s4, s6
+; RV32M-NEXT:    sltu s2, s4, s8
 ; RV32M-NEXT:  .LBB15_10:
 ; RV32M-NEXT:    mulhu s0, a1, s0
-; RV32M-NEXT:    mulhu s10, a5, t3
-; RV32M-NEXT:    mulhu s11, t4, a6
-; RV32M-NEXT:    mul s7, t2, a6
-; RV32M-NEXT:    mul s9, t4, a2
+; RV32M-NEXT:    mulhu s9, a5, t3
+; RV32M-NEXT:    mulhu s10, t4, a6
+; RV32M-NEXT:    mul s6, t2, a6
+; RV32M-NEXT:    mul s11, t4, a2
 ; RV32M-NEXT:    mul ra, t4, a6
 ; RV32M-NEXT:    sltu s5, s5, a7
-; RV32M-NEXT:    sltu a3, s6, t1
-; RV32M-NEXT:    mul s6, a1, t3
+; RV32M-NEXT:    sltu a3, s8, t1
+; RV32M-NEXT:    mul s3, a1, t3
 ; RV32M-NEXT:    add s5, s0, s5
 ; RV32M-NEXT:    add a7, t0, t5
-; RV32M-NEXT:    sltu t1, t0, s8
-; RV32M-NEXT:    add a3, s10, a3
-; RV32M-NEXT:    add s10, s7, s11
-; RV32M-NEXT:    add s8, s5, a3
-; RV32M-NEXT:    add s11, s9, s10
-; RV32M-NEXT:    add s3, ra, s3
-; RV32M-NEXT:    add s4, s11, s4
-; RV32M-NEXT:    sltu s0, s3, ra
-; RV32M-NEXT:    add s3, s6, s8
-; RV32M-NEXT:    add s4, s4, s0
+; RV32M-NEXT:    sltu t1, t0, s7
+; RV32M-NEXT:    mv s8, s11
+; RV32M-NEXT:    add a3, s9, a3
+; RV32M-NEXT:    add s9, s6, s10
+; RV32M-NEXT:    add s7, s5, a3
+; RV32M-NEXT:    add a3, s11, s9
+; RV32M-NEXT:    add a4, ra, a4
+; RV32M-NEXT:    add t5, a3, s4
+; RV32M-NEXT:    sltu s0, a4, ra
+; RV32M-NEXT:    add s4, s3, s7
+; RV32M-NEXT:    add a4, t5, s0
 ; RV32M-NEXT:    add t5, s1, t6
-; RV32M-NEXT:    add t6, s3, s2
-; RV32M-NEXT:    beq s4, s11, .LBB15_12
+; RV32M-NEXT:    mv s1, a3
+; RV32M-NEXT:    add t6, s4, s2
+; RV32M-NEXT:    beq a4, a3, .LBB15_12
 ; RV32M-NEXT:  # %bb.11:
-; RV32M-NEXT:    sltu s0, s4, s11
+; RV32M-NEXT:    sltu s0, a4, s1
 ; RV32M-NEXT:  .LBB15_12:
 ; RV32M-NEXT:    mulhu a3, t2, a6
-; RV32M-NEXT:    mulhu t4, t4, a2
-; RV32M-NEXT:    sltu s1, s10, s7
-; RV32M-NEXT:    sltu s2, s11, s9
-; RV32M-NEXT:    mul s4, t2, a2
-; RV32M-NEXT:    sltu s6, s3, s6
-; RV32M-NEXT:    sltu s3, t6, s3
-; RV32M-NEXT:    sltu s5, s8, s5
+; RV32M-NEXT:    mulhu a4, t4, a2
+; RV32M-NEXT:    sltu t4, s9, s6
+; RV32M-NEXT:    sltu s1, s1, s8
+; RV32M-NEXT:    mul s2, t2, a2
+; RV32M-NEXT:    sltu s3, s4, s3
+; RV32M-NEXT:    sltu s4, t6, s4
+; RV32M-NEXT:    sltu s5, s7, s5
 ; RV32M-NEXT:    mulhu t3, a1, t3
-; RV32M-NEXT:    mulhu s7, t2, a2
-; RV32M-NEXT:    mulhu s8, a5, a6
-; RV32M-NEXT:    add a3, a3, s1
-; RV32M-NEXT:    add t4, t4, s2
-; RV32M-NEXT:    add s5, s5, s6
-; RV32M-NEXT:    add t4, a3, t4
+; RV32M-NEXT:    mulhu s6, t2, a2
+; RV32M-NEXT:    mulhu s7, a5, a6
+; RV32M-NEXT:    add a3, a3, t4
+; RV32M-NEXT:    add a4, a4, s1
 ; RV32M-NEXT:    add s3, s5, s3
-; RV32M-NEXT:    add s1, s4, t4
-; RV32M-NEXT:    sltu a3, t4, a3
+; RV32M-NEXT:    add a4, a3, a4
+; RV32M-NEXT:    add s3, s3, s4
+; RV32M-NEXT:    add t4, s2, a4
+; RV32M-NEXT:    sltu a3, a4, a3
 ; RV32M-NEXT:    mul t2, a1, a6
-; RV32M-NEXT:    add t4, s1, s0
-; RV32M-NEXT:    mul s5, a5, a6
-; RV32M-NEXT:    add s6, t6, t4
+; RV32M-NEXT:    add a4, t4, s0
+; RV32M-NEXT:    mul s1, a5, a6
+; RV32M-NEXT:    add s5, t6, a4
 ; RV32M-NEXT:    mul s0, a5, a2
-; RV32M-NEXT:    sltu s2, s1, s4
-; RV32M-NEXT:    add a3, s7, a3
-; RV32M-NEXT:    sltu t4, t4, s1
+; RV32M-NEXT:    sltu s2, t4, s2
+; RV32M-NEXT:    add a3, s6, a3
+; RV32M-NEXT:    sltu a4, a4, t4
 ; RV32M-NEXT:    add a3, a3, s2
 ; RV32M-NEXT:    add s4, t3, s3
-; RV32M-NEXT:    add a3, a3, t4
-; RV32M-NEXT:    sltu t6, s6, t6
+; RV32M-NEXT:    add a3, a3, a4
+; RV32M-NEXT:    sltu t6, s5, t6
 ; RV32M-NEXT:    add a3, s4, a3
-; RV32M-NEXT:    add s2, t2, s8
-; RV32M-NEXT:    add t3, s5, s6
+; RV32M-NEXT:    add s2, t2, s7
+; RV32M-NEXT:    add t3, s1, s5
 ; RV32M-NEXT:    add s3, s0, s2
-; RV32M-NEXT:    sltu s1, t3, s5
+; RV32M-NEXT:    sltu s1, t3, s1
 ; RV32M-NEXT:    add s5, a3, t6
 ; RV32M-NEXT:    add t4, s3, s1
 ; RV32M-NEXT:    add t4, t4, s5
@@ -4214,22 +4219,22 @@ define <2 x i128> @umulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:    sltu t6, s5, s4
 ; RV32M-NEXT:  .LBB15_16:
 ; RV32M-NEXT:    mulhu a3, a1, a6
-; RV32M-NEXT:    mulhu a5, a5, a2
-; RV32M-NEXT:    mul a6, a1, a2
+; RV32M-NEXT:    mulhu a4, a5, a2
+; RV32M-NEXT:    mul a5, a1, a2
 ; RV32M-NEXT:    mulhu a1, a1, a2
 ; RV32M-NEXT:    sltu a2, s2, t2
-; RV32M-NEXT:    sltu t1, s3, s0
+; RV32M-NEXT:    sltu a6, s3, s0
 ; RV32M-NEXT:    add a2, a3, a2
-; RV32M-NEXT:    add a5, a5, t1
-; RV32M-NEXT:    add a5, a2, a5
-; RV32M-NEXT:    add a3, a6, a5
-; RV32M-NEXT:    sltu a2, a5, a2
+; RV32M-NEXT:    add a4, a4, a6
+; RV32M-NEXT:    add a4, a2, a4
+; RV32M-NEXT:    add a3, a5, a4
+; RV32M-NEXT:    sltu a2, a4, a2
 ; RV32M-NEXT:    add t6, a3, t6
-; RV32M-NEXT:    sltu a5, a3, a6
+; RV32M-NEXT:    sltu a4, a3, a5
 ; RV32M-NEXT:    add a1, a1, a2
 ; RV32M-NEXT:    add s1, t6, s1
 ; RV32M-NEXT:    sltu a2, t6, a3
-; RV32M-NEXT:    add a1, a1, a5
+; RV32M-NEXT:    add a1, a1, a4
 ; RV32M-NEXT:    sltu a3, s1, t6
 ; RV32M-NEXT:    add a1, a1, a2
 ; RV32M-NEXT:    add a1, a1, a3
@@ -4239,7 +4244,8 @@ define <2 x i128> @umulh_v2i128(<2 x i128> %x, <2 x i128> %y) nounwind {
 ; RV32M-NEXT:    sw a1, 28(a0)
 ; RV32M-NEXT:    lw a1, 8(sp) # 4-byte Folded Reload
 ; RV32M-NEXT:    sw a1, 0(a0)
-; RV32M-NEXT:    sw a4, 4(a0)
+; RV32M-NEXT:    lw a1, 4(sp) # 4-byte Folded Reload
+; RV32M-NEXT:    sw a1, 4(a0)
 ; RV32M-NEXT:    sw a7, 8(a0)
 ; RV32M-NEXT:    sw t0, 12(a0)
 ; RV32M-NEXT:    lw ra, 60(sp) # 4-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/xqcibm-cto-clo-brev.ll b/llvm/test/CodeGen/RISCV/xqcibm-cto-clo-brev.ll
index c0e6be7484d358..d85efbfdcba254 100644
--- a/llvm/test/CodeGen/RISCV/xqcibm-cto-clo-brev.ll
+++ b/llvm/test/CodeGen/RISCV/xqcibm-cto-clo-brev.ll
@@ -142,54 +142,55 @@ define i32 @test_cttz_i32(i32 %a) nounwind {
 define i64 @test_cttz_i64(i64 %a) nounwind {
 ; RV32I-LABEL: test_cttz_i64:
 ; RV32I:       # %bb.0:
+; RV32I-NEXT:    not a1, a1
+; RV32I-NEXT:    not a0, a0
+; RV32I-NEXT:    or a2, a0, a1
+; RV32I-NEXT:    beqz a2, .LBB3_3
+; RV32I-NEXT:  # %bb.1: # %cond.false
 ; RV32I-NEXT:    addi sp, sp, -32
 ; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    not s3, a1
-; RV32I-NEXT:    not s2, a0
-; RV32I-NEXT:    or a0, s2, s3
-; RV32I-NEXT:    beqz a0, .LBB3_3
-; RV32I-NEXT:  # %bb.1: # %cond.false
-; RV32I-NEXT:    neg a0, s2
-; RV32I-NEXT:    and a0, s2, a0
-; RV32I-NEXT:    lui s1, 30667
-; RV32I-NEXT:    addi s1, s1, 1329
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    neg a2, a0
+; RV32I-NEXT:    sw a0, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a0, a2
+; RV32I-NEXT:    lui a2, 30667
+; RV32I-NEXT:    addi a2, a2, 1329
+; RV32I-NEXT:    sw a2, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw a1, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    neg a0, s3
-; RV32I-NEXT:    lui s4, %hi(.LCPI3_0)
-; RV32I-NEXT:    addi s4, s4, %lo(.LCPI3_0)
-; RV32I-NEXT:    and a0, s3, a0
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    neg a1, a0
+; RV32I-NEXT:    lw a0, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    and a0, a0, a1
+; RV32I-NEXT:    lw a1, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s2, .LBB3_4
+; RV32I-NEXT:    lw a1, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    bnez a1, .LBB3_4
 ; RV32I-NEXT:  # %bb.2: # %cond.false
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s4, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    j .LBB3_5
 ; RV32I-NEXT:  .LBB3_3:
 ; RV32I-NEXT:    li a0, 64
-; RV32I-NEXT:    j .LBB3_5
-; RV32I-NEXT:  .LBB3_4:
-; RV32I-NEXT:    srli s0, s0, 27
-; RV32I-NEXT:    add s0, s4, s0
-; RV32I-NEXT:    lbu a0, 0(s0)
-; RV32I-NEXT:  .LBB3_5: # %cond.end
 ; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    ret
+; RV32I-NEXT:  .LBB3_4:
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI3_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI3_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
+; RV32I-NEXT:  .LBB3_5: # %cond.false
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: test_cttz_i64:
@@ -341,42 +342,41 @@ define i64 @test_cttz_i64_zero_poison(i64 %a) nounwind {
 ; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    not s3, a0
-; RV32I-NEXT:    neg a0, s3
-; RV32I-NEXT:    not s4, a1
-; RV32I-NEXT:    and a0, s3, a0
-; RV32I-NEXT:    lui s1, 30667
-; RV32I-NEXT:    addi s1, s1, 1329
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    not s1, a0
+; RV32I-NEXT:    neg a0, s1
+; RV32I-NEXT:    not s2, a1
+; RV32I-NEXT:    and a0, s1, a0
+; RV32I-NEXT:    lui s0, 30667
+; RV32I-NEXT:    addi s0, s0, 1329
+; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    mv s0, a0
-; RV32I-NEXT:    neg a0, s4
-; RV32I-NEXT:    lui s2, %hi(.LCPI7_0)
-; RV32I-NEXT:    addi s2, s2, %lo(.LCPI7_0)
-; RV32I-NEXT:    and a0, s4, a0
-; RV32I-NEXT:    mv a1, s1
+; RV32I-NEXT:    sw a0, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    neg a0, s2
+; RV32I-NEXT:    and a0, s2, a0
+; RV32I-NEXT:    mv a1, s0
 ; RV32I-NEXT:    call __mulsi3
-; RV32I-NEXT:    bnez s3, .LBB7_2
+; RV32I-NEXT:    bnez s1, .LBB7_2
 ; RV32I-NEXT:  # %bb.1:
-; RV32I-NEXT:    srli a0, a0, 27
-; RV32I-NEXT:    add a0, s2, a0
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI7_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI7_0)
+; RV32I-NEXT:    add a0, a0, a1
 ; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:    addi a0, a0, 32
 ; RV32I-NEXT:    j .LBB7_3
 ; RV32I-NEXT:  .LBB7_2:
-; RV32I-NEXT:    srli s0, s0, 27
-; RV32I-NEXT:    add s0, s2, s0
-; RV32I-NEXT:    lbu a0, 0(s0)
+; RV32I-NEXT:    lw a0, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    srli a1, a0, 27
+; RV32I-NEXT:    lui a0, %hi(.LCPI7_0)
+; RV32I-NEXT:    addi a0, a0, %lo(.LCPI7_0)
+; RV32I-NEXT:    add a0, a0, a1
+; RV32I-NEXT:    lbu a0, 0(a0)
 ; RV32I-NEXT:  .LBB7_3:
 ; RV32I-NEXT:    li a1, 0
 ; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
-; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/xqcilo-xqcilia-frame-index.ll b/llvm/test/CodeGen/RISCV/xqcilo-xqcilia-frame-index.ll
index 681418c16816cb..0f05d3b376d360 100644
--- a/llvm/test/CodeGen/RISCV/xqcilo-xqcilia-frame-index.ll
+++ b/llvm/test/CodeGen/RISCV/xqcilo-xqcilia-frame-index.ll
@@ -122,20 +122,15 @@ declare dso_local i32 @sink(i32 noundef) local_unnamed_addr
 define dso_local i32 @regpressure(i32 noundef %a0, i32 noundef %a1, i32 noundef %a2, i32 noundef %a3, i32 noundef %a4, i32 noundef %a5, i32 noundef %a6, i32 noundef %a7, i32 noundef %s0, i32 noundef %s1, i32 noundef %s2, i32 noundef %s3, i32 noundef %s4, i32 noundef %s5, i32 noundef %s6, i32 noundef %s7, i32 noundef %s8, i32 noundef %s9, i32 noundef %s10, i32 noundef %s11) local_unnamed_addr nounwind {
 ; CHECK-LABEL: regpressure:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    c.addi16sp sp, -64
-; CHECK-NEXT:    c.swsp ra, 60(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s0, 56(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s1, 52(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s2, 48(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s3, 44(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s4, 40(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s5, 36(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s6, 32(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s7, 28(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s8, 24(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s9, 20(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s10, 16(sp) # 4-byte Folded Spill
-; CHECK-NEXT:    c.swsp s11, 12(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.addi sp, -32
+; CHECK-NEXT:    c.swsp ra, 28(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.swsp s0, 24(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.swsp s1, 20(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.swsp s2, 16(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.swsp s3, 12(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.swsp s4, 8(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.swsp s5, 4(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    c.swsp s6, 0(sp) # 4-byte Folded Spill
 ; CHECK-NEXT:    c.mv s0, a7
 ; CHECK-NEXT:    c.mv s1, a6
 ; CHECK-NEXT:    c.mv s2, a5
@@ -143,52 +138,47 @@ define dso_local i32 @regpressure(i32 noundef %a0, i32 noundef %a1, i32 noundef
 ; CHECK-NEXT:    c.mv s4, a3
 ; CHECK-NEXT:    c.mv s5, a2
 ; CHECK-NEXT:    c.mv s6, a1
-; CHECK-NEXT:    c.lwsp s8, 80(sp)
-; CHECK-NEXT:    c.lwsp s10, 76(sp)
-; CHECK-NEXT:    c.lwsp s11, 72(sp)
-; CHECK-NEXT:    c.lwsp s9, 68(sp)
-; CHECK-NEXT:    c.lwsp s7, 64(sp)
 ; CHECK-NEXT:    call sink
 ; CHECK-NEXT:    c.add s2, s3
 ; CHECK-NEXT:    c.add s0, s1
 ; CHECK-NEXT:    c.add s0, s2
-; CHECK-NEXT:    c.add s9, s11
-; CHECK-NEXT:    c.add s0, s7
-; CHECK-NEXT:    c.add s8, s10
-; CHECK-NEXT:    c.add s0, s9
-; CHECK-NEXT:    c.lwsp a1, 84(sp)
-; CHECK-NEXT:    c.add s8, a1
-; CHECK-NEXT:    c.add s0, s8
-; CHECK-NEXT:    c.lwsp a1, 92(sp)
-; CHECK-NEXT:    c.lwsp a2, 88(sp)
+; CHECK-NEXT:    c.lwsp a1, 40(sp)
+; CHECK-NEXT:    c.lwsp a2, 36(sp)
+; CHECK-NEXT:    add a3, a2, a1
+; CHECK-NEXT:    c.lwsp a1, 32(sp)
+; CHECK-NEXT:    c.add s0, a1
+; CHECK-NEXT:    c.lwsp a1, 48(sp)
+; CHECK-NEXT:    c.lwsp a2, 44(sp)
+; CHECK-NEXT:    c.add a2, a1
+; CHECK-NEXT:    c.add a3, s0
+; CHECK-NEXT:    c.lwsp a1, 52(sp)
 ; CHECK-NEXT:    c.add a1, a2
-; CHECK-NEXT:    c.lwsp a2, 108(sp)
-; CHECK-NEXT:    c.lwsp a3, 104(sp)
-; CHECK-NEXT:    c.add a2, a3
-; CHECK-NEXT:    c.lwsp a3, 96(sp)
-; CHECK-NEXT:    c.add a1, a3
-; CHECK-NEXT:    c.add a0, a2
-; CHECK-NEXT:    c.lwsp a2, 100(sp)
-; CHECK-NEXT:    c.add a1, a2
-; CHECK-NEXT:    c.add a0, s6
-; CHECK-NEXT:    c.add a1, s0
-; CHECK-NEXT:    c.add a0, s5
+; CHECK-NEXT:    c.add a3, a1
+; CHECK-NEXT:    c.lwsp a1, 60(sp)
+; CHECK-NEXT:    c.lwsp a2, 56(sp)
+; CHECK-NEXT:    add a4, a2, a1
+; CHECK-NEXT:    c.lwsp a1, 76(sp)
+; CHECK-NEXT:    c.lwsp a2, 72(sp)
+; CHECK-NEXT:    c.add a2, a1
+; CHECK-NEXT:    c.lwsp a1, 64(sp)
+; CHECK-NEXT:    c.add a1, a4
+; CHECK-NEXT:    c.add a2, a0
+; CHECK-NEXT:    c.lwsp a0, 68(sp)
 ; CHECK-NEXT:    c.add a0, a1
+; CHECK-NEXT:    c.add a2, s6
+; CHECK-NEXT:    c.add a0, a3
+; CHECK-NEXT:    c.add a2, s5
+; CHECK-NEXT:    c.add a0, a2
 ; CHECK-NEXT:    c.add a0, s4
-; CHECK-NEXT:    c.lwsp ra, 60(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s0, 56(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s1, 52(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s2, 48(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s3, 44(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s4, 40(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s5, 36(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s6, 32(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s7, 28(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s8, 24(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s9, 20(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s10, 16(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.lwsp s11, 12(sp) # 4-byte Folded Reload
-; CHECK-NEXT:    c.addi16sp sp, 64
+; CHECK-NEXT:    c.lwsp ra, 28(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.lwsp s0, 24(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.lwsp s1, 20(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.lwsp s2, 16(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.lwsp s3, 12(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.lwsp s4, 8(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.lwsp s5, 4(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.lwsp s6, 0(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    c.addi16sp sp, 32
 ; CHECK-NEXT:    c.jr ra
 entry:
   %call = tail call i32 @sink(i32 noundef %a0)
diff --git a/llvm/test/CodeGen/RISCV/zicond-fp-select-zfinx.ll b/llvm/test/CodeGen/RISCV/zicond-fp-select-zfinx.ll
index 7fcf571d4a23d6..ec1e89f03a95ca 100644
--- a/llvm/test/CodeGen/RISCV/zicond-fp-select-zfinx.ll
+++ b/llvm/test/CodeGen/RISCV/zicond-fp-select-zfinx.ll
@@ -265,25 +265,21 @@ define double @select_f64_fcmp(double %a, double %b, double %c, double %d) nounw
 ; RV32ZFINX_NOZICOND-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
 ; RV32ZFINX_NOZICOND-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
 ; RV32ZFINX_NOZICOND-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
-; RV32ZFINX_NOZICOND-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
-; RV32ZFINX_NOZICOND-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
-; RV32ZFINX_NOZICOND-NEXT:    mv s2, a7
-; RV32ZFINX_NOZICOND-NEXT:    mv s3, a6
+; RV32ZFINX_NOZICOND-NEXT:    sw a7, 16(sp) # 4-byte Folded Spill
+; RV32ZFINX_NOZICOND-NEXT:    sw a6, 12(sp) # 4-byte Folded Spill
 ; RV32ZFINX_NOZICOND-NEXT:    mv s0, a5
 ; RV32ZFINX_NOZICOND-NEXT:    mv s1, a4
 ; RV32ZFINX_NOZICOND-NEXT:    call __gtdf2
 ; RV32ZFINX_NOZICOND-NEXT:    bgtz a0, .LBB2_2
 ; RV32ZFINX_NOZICOND-NEXT:  # %bb.1: # %entry
-; RV32ZFINX_NOZICOND-NEXT:    mv s1, s3
-; RV32ZFINX_NOZICOND-NEXT:    mv s0, s2
+; RV32ZFINX_NOZICOND-NEXT:    lw s1, 12(sp) # 4-byte Folded Reload
+; RV32ZFINX_NOZICOND-NEXT:    lw s0, 16(sp) # 4-byte Folded Reload
 ; RV32ZFINX_NOZICOND-NEXT:  .LBB2_2: # %entry
 ; RV32ZFINX_NOZICOND-NEXT:    mv a0, s1
 ; RV32ZFINX_NOZICOND-NEXT:    mv a1, s0
 ; RV32ZFINX_NOZICOND-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
 ; RV32ZFINX_NOZICOND-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
 ; RV32ZFINX_NOZICOND-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
-; RV32ZFINX_NOZICOND-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
-; RV32ZFINX_NOZICOND-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
 ; RV32ZFINX_NOZICOND-NEXT:    addi sp, sp, 32
 ; RV32ZFINX_NOZICOND-NEXT:    ret
 ;
diff --git a/llvm/test/DebugInfo/RISCV/dw_op_entry_value_32bit.ll b/llvm/test/DebugInfo/RISCV/dw_op_entry_value_32bit.ll
index 069cc948f499b0..998f5230559520 100644
--- a/llvm/test/DebugInfo/RISCV/dw_op_entry_value_32bit.ll
+++ b/llvm/test/DebugInfo/RISCV/dw_op_entry_value_32bit.ll
@@ -14,7 +14,7 @@
 ;; Confirm that info from callSites attribute is used as entry_value in DIExpression.
 
 ;; Test riscv32:
-; CHECK32: $x10 = nsw ADD $x11, killed renamable $x10
+; CHECK32: $x10 = nsw ADD killed renamable $x11, killed renamable $x10
 ; CHECK32-NEXT: DBG_VALUE $x10, $noreg, !{{.*}}, !DIExpression(DW_OP_LLVM_entry_value, 1)
 
 ; ModuleID = 'm.c'
diff --git a/llvm/test/DebugInfo/RISCV/dw_op_entry_value_64bit.ll b/llvm/test/DebugInfo/RISCV/dw_op_entry_value_64bit.ll
index ec4b929997e62d..a80ab9b2bfc8cf 100644
--- a/llvm/test/DebugInfo/RISCV/dw_op_entry_value_64bit.ll
+++ b/llvm/test/DebugInfo/RISCV/dw_op_entry_value_64bit.ll
@@ -14,7 +14,7 @@
 ;; Confirm that info from callSites attribute is used as entry_value in DIExpression.
 
 ;; Test riscv64:
-; CHECK64: $x10 = nsw ADD $x11, killed renamable $x10
+; CHECK64: $x10 = nsw ADD killed renamable $x11, killed renamable $x10
 ; CHECK64-NEXT: DBG_VALUE $x10, $noreg, !{{.*}}, !DIExpression(DW_OP_LLVM_entry_value, 1)
 
 ; ModuleID = 'm.c'



More information about the llvm-branch-commits mailing list