[llvm-branch-commits] [llvm] AtomicExpand: Use RuntimeLibcallsInfo signatures for atomic libcalls (PR #238972)
Matt Arsenault via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Wed Nov 4 13:50:53 PST 2026
https://github.com/arsenm updated https://github.com/llvm/llvm-project/pull/238972
>From b6fafe5d81bc42fc59f45dfbdcc891192e4da39c Mon Sep 17 00:00:00 2001
From: Matt Arsenault <Matthew.Arsenault at amd.com>
Date: Wed, 4 Nov 2026 12:25:12 -0800
Subject: [PATCH] AtomicExpand: Use RuntimeLibcallsInfo signatures for atomic
libcalls
Take the function type and attributes of the emitted __atomic_* calls
from RuntimeLibcallsInfo instead of building them by hand.
This fixes missing extension attributes on the value operand and
result of the sized variants, which are unsigned integers of the
access size. On targets whose C ABI extends narrow integer arguments,
such as RISC-V and ARM, the callee could see unextended high bits. i8
and i16 values now get zeroext, and i32 values get signext on
RV64. Callers can also rely on the extended return value, which
removes redundant extensions after some calls.
Co-authored-by: Claude Opus 5.5 <noreply at anthropic.com>
---
llvm/lib/CodeGen/AtomicExpandPass.cpp | 75 +-
llvm/test/CodeGen/ARM/atomic-load-store.ll | 6 +
.../RISCV/GlobalISel/atomic-cmpxchg.ll | 70 ++
.../RISCV/GlobalISel/atomic-load-store-fp.ll | 4 +
.../RISCV/GlobalISel/atomic-load-store.ll | 28 +
.../RISCV/GlobalISel/atomicrmw-add-sub.ll | 14 +
.../RISCV/GlobalISel/atomicrmw-and-or-xor.ll | 105 ++
.../GlobalISel/atomicrmw-max-min-umax-umin.ll | 820 +++++++------
.../RISCV/GlobalISel/atomicrmw-xchg.ll | 35 +
llvm/test/CodeGen/RISCV/atomic-cmpxchg.ll | 70 ++
llvm/test/CodeGen/RISCV/atomic-load-store.ll | 28 +
llvm/test/CodeGen/RISCV/atomic-load-zext.ll | 24 -
llvm/test/CodeGen/RISCV/atomic-rmw.ll | 1065 +++++++++++------
llvm/test/CodeGen/RISCV/atomic-signext.ll | 270 +++--
.../CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll | 18 +-
.../CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll | 65 +-
llvm/test/CodeGen/RISCV/forced-atomics.ll | 24 +-
llvm/test/CodeGen/RISCV/libcall-tail-calls.ll | 5 +
.../AtomicExpand/MSP430/libcalls.ll | 2 +-
.../Transforms/AtomicExpand/SPARC/libcalls.ll | 10 +-
20 files changed, 1777 insertions(+), 961 deletions(-)
diff --git a/llvm/lib/CodeGen/AtomicExpandPass.cpp b/llvm/lib/CodeGen/AtomicExpandPass.cpp
index d882f4019e0ef9..7f2097633d4217 100644
--- a/llvm/lib/CodeGen/AtomicExpandPass.cpp
+++ b/llvm/lib/CodeGen/AtomicExpandPass.cpp
@@ -20,7 +20,6 @@
#include "llvm/ADT/SmallVector.h"
#include "llvm/Analysis/InstSimplifyFolder.h"
#include "llvm/Analysis/OptimizationRemarkEmitter.h"
-#include "llvm/Analysis/TargetLibraryInfo.h"
#include "llvm/Analysis/VectorUtils.h"
#include "llvm/CodeGen/AtomicExpand.h"
#include "llvm/CodeGen/TargetLowering.h"
@@ -2254,31 +2253,12 @@ bool AtomicExpandImpl::expandAtomicOpToLibcall(
IRBuilder<> AllocaBuilder(&I->getFunction()->getEntryBlock().front());
bool UseSizedLibcall = canUseSizedAtomicCall(Size, Alignment, DL);
- Type *SizedIntTy = Type::getIntNTy(Ctx, Size * 8);
-
- if (M->getTargetTriple().isOSWindowsOrUEFI() &&
- M->getTargetTriple().isX86_64() && Size == 16) {
- // x86_64 Windows passes i128 as an XMM vector; on return, it is in
- // XMM0, and as a parameter, it is passed indirectly. The generic lowering
- // rules handles this correctly if we pass it as a v2i64 rather than
- // i128. This is what Clang does in the frontend for such types as well
- // (see WinX86_64ABIInfo::classify in Clang).
- SizedIntTy = FixedVectorType::get(Type::getInt64Ty(Ctx), 2);
- }
-
- const Align AllocaAlignment = DL.getPrefTypeAlign(SizedIntTy);
+ const Align AllocaAlignment =
+ DL.getPrefTypeAlign(Type::getIntNTy(Ctx, Size * 8));
- // TODO: Take the signature from RuntimeLibcallsInfo
- IntegerType *OrderingTy =
- Type::getIntNTy(Ctx, LibcallLowering->getRuntimeLibcallsInfo().getIntSize(
- M->getTargetTriple()));
assert(Ordering != AtomicOrdering::NotAtomic && "expect atomic MO");
- Constant *OrderingVal = ConstantInt::get(OrderingTy, (int)toCABI(Ordering));
- Constant *Ordering2Val = nullptr;
- if (CASExpected) {
- assert(Ordering2 != AtomicOrdering::NotAtomic && "expect atomic MO");
- Ordering2Val = ConstantInt::get(OrderingTy, (int)toCABI(Ordering2));
- }
+ assert((!CASExpected || Ordering2 != AtomicOrdering::NotAtomic) &&
+ "expect atomic MO");
bool HasResult = I->getType() != Type::getVoidTy(Ctx);
RTLIB::Libcall RTLibType;
@@ -2314,6 +2294,10 @@ bool AtomicExpandImpl::expandAtomicOpToLibcall(
return false;
}
+ auto [FnType, Attr] = LibcallLowering->getRuntimeLibcallsInfo().getFunctionTy(
+ Ctx, M->getTargetTriple(), DL, LibcallImpl);
+ assert(FnType && "missing signature for atomic libcall");
+
// Build up the function call. There's two kinds. First, the sized
// variants. These calls are going to be one of the following (with
// N=1,2,4,8,16):
@@ -2344,15 +2328,11 @@ bool AtomicExpandImpl::expandAtomicOpToLibcall(
AllocaInst *AllocaValue = nullptr;
AllocaInst *AllocaResult = nullptr;
- Type *ResultTy;
SmallVector<Value *, 6> Args;
- AttributeList Attr;
// 'size' argument.
- if (!UseSizedLibcall) {
- // Note, getIntPtrType is assumed equivalent to size_t.
- Args.push_back(ConstantInt::get(DL.getIntPtrType(Ctx), Size));
- }
+ if (!UseSizedLibcall)
+ Args.push_back(ConstantInt::get(FnType->getParamType(0), Size));
// 'ptr' argument.
// note: This assumes all address spaces share a common libfunc
@@ -2375,8 +2355,8 @@ bool AtomicExpandImpl::expandAtomicOpToLibcall(
// 'val' argument ('desired' for cas), if present.
if (ValueOperand) {
if (UseSizedLibcall) {
- Value *IntValue =
- Builder.CreateBitPreservingCastChain(DL, ValueOperand, SizedIntTy);
+ Value *IntValue = Builder.CreateBitPreservingCastChain(
+ DL, ValueOperand, FnType->getParamType(Args.size()));
Args.push_back(IntValue);
} else {
AllocaValue = AllocaBuilder.CreateAlloca(ValueOperand->getType());
@@ -2395,36 +2375,19 @@ bool AtomicExpandImpl::expandAtomicOpToLibcall(
Args.push_back(AllocaResult);
}
- // The ordering arguments are C ints, which may need to be extended.
- // FIXME: Once runtime libcall attribute derivation is centralized in
- // RuntimeLibcalls, TargetLibraryInfo should not be used here.
- Attribute::AttrKind OrderingExt =
- TargetLibraryInfo::getExtAttrForI32Param(M->getTargetTriple());
-
// 'ordering' ('success_order' for cas) argument.
- Args.push_back(OrderingVal);
- Attr = Attr.maybeAddParamAttribute(Ctx, Args.size() - 1, OrderingExt);
+ Args.push_back(ConstantInt::get(FnType->getParamType(Args.size()),
+ (int)toCABI(Ordering)));
// 'failure_order' argument, if present.
- if (Ordering2Val) {
- Args.push_back(Ordering2Val);
- Attr = Attr.maybeAddParamAttribute(Ctx, Args.size() - 1, OrderingExt);
+ if (CASExpected) {
+ Args.push_back(ConstantInt::get(FnType->getParamType(Args.size()),
+ (int)toCABI(Ordering2)));
}
- // Now, the return type. FIXME: Properly add extension attributes.
- if (CASExpected) {
- ResultTy = Type::getInt1Ty(Ctx);
- Attr = Attr.addRetAttribute(Ctx, Attribute::ZExt);
- } else if (HasResult && UseSizedLibcall)
- ResultTy = SizedIntTy;
- else
- ResultTy = Type::getVoidTy(Ctx);
+ assert(Args.size() == FnType->getNumParams() &&
+ "atomic libcall signature mismatch");
- // Done with setting up arguments and return types, create the call:
- SmallVector<Type *, 6> ArgTys;
- for (Value *Arg : Args)
- ArgTys.push_back(Arg->getType());
- FunctionType *FnType = FunctionType::get(ResultTy, ArgTys, false);
FunctionCallee LibcallFn = M->getOrInsertFunction(
RTLIB::RuntimeLibcallsInfo::getLibcallImplName(LibcallImpl), FnType,
Attr);
diff --git a/llvm/test/CodeGen/ARM/atomic-load-store.ll b/llvm/test/CodeGen/ARM/atomic-load-store.ll
index 0c787a4ca05c34..f8ecd95e3b315b 100644
--- a/llvm/test/CodeGen/ARM/atomic-load-store.ll
+++ b/llvm/test/CodeGen/ARM/atomic-load-store.ll
@@ -700,6 +700,9 @@ define void @store_atomic_f16__seq_cst(ptr %ptr, half %val1) {
; ARMV4-LABEL: store_atomic_f16__seq_cst:
; ARMV4: @ %bb.0:
; ARMV4-NEXT: push {r11, lr}
+; ARMV4-NEXT: mov r2, #255
+; ARMV4-NEXT: orr r2, r2, #65280
+; ARMV4-NEXT: and r1, r1, r2
; ARMV4-NEXT: mov r2, #5
; ARMV4-NEXT: bl __atomic_store_2
; ARMV4-NEXT: pop {r11, lr}
@@ -761,6 +764,9 @@ define void @store_atomic_bf16__seq_cst(ptr %ptr, bfloat %val1) {
; ARMV4-LABEL: store_atomic_bf16__seq_cst:
; ARMV4: @ %bb.0:
; ARMV4-NEXT: push {r11, lr}
+; ARMV4-NEXT: mov r2, #255
+; ARMV4-NEXT: orr r2, r2, #65280
+; ARMV4-NEXT: and r1, r1, r2
; ARMV4-NEXT: mov r2, #5
; ARMV4-NEXT: bl __atomic_store_2
; ARMV4-NEXT: pop {r11, lr}
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/atomic-cmpxchg.ll b/llvm/test/CodeGen/RISCV/GlobalISel/atomic-cmpxchg.ll
index 778b3256ff645c..691b1c52999672 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/atomic-cmpxchg.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/atomic-cmpxchg.ll
@@ -30,6 +30,7 @@ define void @cmpxchg_i8_monotonic_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
@@ -67,6 +68,7 @@ define void @cmpxchg_i8_monotonic_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -161,6 +163,7 @@ define void @cmpxchg_i8_acquire_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 0
@@ -270,6 +273,7 @@ define void @cmpxchg_i8_acquire_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 0
@@ -393,6 +397,7 @@ define void @cmpxchg_i8_acquire_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -502,6 +507,7 @@ define void @cmpxchg_i8_acquire_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -625,6 +631,7 @@ define void @cmpxchg_i8_release_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 0
@@ -734,6 +741,7 @@ define void @cmpxchg_i8_release_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 0
@@ -857,6 +865,7 @@ define void @cmpxchg_i8_release_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 2
@@ -966,6 +975,7 @@ define void @cmpxchg_i8_release_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 2
@@ -1089,6 +1099,7 @@ define void @cmpxchg_i8_acq_rel_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 0
@@ -1198,6 +1209,7 @@ define void @cmpxchg_i8_acq_rel_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 0
@@ -1321,6 +1333,7 @@ define void @cmpxchg_i8_acq_rel_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -1430,6 +1443,7 @@ define void @cmpxchg_i8_acq_rel_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -1553,6 +1567,7 @@ define void @cmpxchg_i8_seq_cst_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 0
@@ -1590,6 +1605,7 @@ define void @cmpxchg_i8_seq_cst_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 0
@@ -1689,6 +1705,7 @@ define void @cmpxchg_i8_seq_cst_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 2
@@ -1726,6 +1743,7 @@ define void @cmpxchg_i8_seq_cst_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 2
@@ -1825,6 +1843,7 @@ define void @cmpxchg_i8_seq_cst_seq_cst(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -1862,6 +1881,7 @@ define void @cmpxchg_i8_seq_cst_seq_cst(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -1963,7 +1983,9 @@ define void @cmpxchg_i16_monotonic_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounw
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2001,7 +2023,9 @@ define void @cmpxchg_i16_monotonic_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounw
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2098,7 +2122,9 @@ define void @cmpxchg_i16_acquire_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2211,7 +2237,9 @@ define void @cmpxchg_i16_acquire_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2338,7 +2366,9 @@ define void @cmpxchg_i16_acquire_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2451,7 +2481,9 @@ define void @cmpxchg_i16_acquire_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2578,7 +2610,9 @@ define void @cmpxchg_i16_release_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2691,7 +2725,9 @@ define void @cmpxchg_i16_release_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2818,7 +2854,9 @@ define void @cmpxchg_i16_release_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2931,7 +2969,9 @@ define void @cmpxchg_i16_release_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3058,7 +3098,9 @@ define void @cmpxchg_i16_acq_rel_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3171,7 +3213,9 @@ define void @cmpxchg_i16_acq_rel_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3298,7 +3342,9 @@ define void @cmpxchg_i16_acq_rel_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3411,7 +3457,9 @@ define void @cmpxchg_i16_acq_rel_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3538,7 +3586,9 @@ define void @cmpxchg_i16_seq_cst_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3576,7 +3626,9 @@ define void @cmpxchg_i16_seq_cst_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3678,7 +3730,9 @@ define void @cmpxchg_i16_seq_cst_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3716,7 +3770,9 @@ define void @cmpxchg_i16_seq_cst_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3818,7 +3874,9 @@ define void @cmpxchg_i16_seq_cst_seq_cst(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3856,7 +3914,9 @@ define void @cmpxchg_i16_seq_cst_seq_cst(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -4000,6 +4060,7 @@ define void @cmpxchg_i32_monotonic_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounw
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -4094,6 +4155,7 @@ define void @cmpxchg_i32_acquire_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 0
@@ -4198,6 +4260,7 @@ define void @cmpxchg_i32_acquire_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4302,6 +4365,7 @@ define void @cmpxchg_i32_release_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 0
@@ -4406,6 +4470,7 @@ define void @cmpxchg_i32_release_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 2
@@ -4510,6 +4575,7 @@ define void @cmpxchg_i32_acq_rel_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 0
@@ -4614,6 +4680,7 @@ define void @cmpxchg_i32_acq_rel_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -4718,6 +4785,7 @@ define void @cmpxchg_i32_seq_cst_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 0
@@ -4822,6 +4890,7 @@ define void @cmpxchg_i32_seq_cst_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 2
@@ -4928,6 +4997,7 @@ define void @cmpxchg_i32_seq_cst_seq_cst(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store-fp.ll b/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store-fp.ll
index 24fdab2d5251ea..ec59a5bbf06743 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store-fp.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store-fp.ll
@@ -574,6 +574,7 @@ define void @atomic_store_f32_unordered(ptr %a, float %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -611,6 +612,7 @@ define void @atomic_store_f32_monotonic(ptr %a, float %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -655,6 +657,7 @@ define void @atomic_store_f32_release(ptr %a, float %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -757,6 +760,7 @@ define void @atomic_store_f32_seq_cst(ptr %a, float %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store.ll b/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store.ll
index 5a39ece02613ba..4c3f6491600a78 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/atomic-load-store.ll
@@ -973,6 +973,7 @@ define void @atomic_store_i8_unordered(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -988,6 +989,7 @@ define void @atomic_store_i8_unordered(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1007,6 +1009,7 @@ define void @atomic_store_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1022,6 +1025,7 @@ define void @atomic_store_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1041,6 +1045,7 @@ define void @atomic_store_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1062,6 +1067,7 @@ define void @atomic_store_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1129,6 +1135,7 @@ define void @atomic_store_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1151,6 +1158,7 @@ define void @atomic_store_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1213,6 +1221,8 @@ define void @atomic_store_i16_unordered(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1228,6 +1238,8 @@ define void @atomic_store_i16_unordered(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1247,6 +1259,8 @@ define void @atomic_store_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1262,6 +1276,8 @@ define void @atomic_store_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1281,6 +1297,8 @@ define void @atomic_store_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1302,6 +1320,8 @@ define void @atomic_store_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1369,6 +1389,8 @@ define void @atomic_store_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1391,6 +1413,8 @@ define void @atomic_store_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1468,6 +1492,7 @@ define void @atomic_store_i32_unordered(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1502,6 +1527,7 @@ define void @atomic_store_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1542,6 +1568,7 @@ define void @atomic_store_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1631,6 +1658,7 @@ define void @atomic_store_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-add-sub.ll b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-add-sub.ll
index 7261ebe697d443..25a03cb45b04a4 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-add-sub.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-add-sub.ll
@@ -37,6 +37,7 @@ define i8 @atomicrmw_add_i8(ptr %ptr, i8 %rhs) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_add_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -73,6 +74,7 @@ define i8 @atomicrmw_add_i8(ptr %ptr, i8 %rhs) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_add_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -114,6 +116,8 @@ define i16 @atomicrmw_add_i16(ptr %ptr, i16 %rhs) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_add_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -151,6 +155,8 @@ define i16 @atomicrmw_add_i16(ptr %ptr, i16 %rhs) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_add_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -195,6 +201,7 @@ define i32 @atomicrmw_add_i32(ptr %ptr, i32 %rhs) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_add_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -290,6 +297,7 @@ define i8 @atomicrmw_sub_i8(ptr %ptr, i8 %rhs) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_sub_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -327,6 +335,7 @@ define i8 @atomicrmw_sub_i8(ptr %ptr, i8 %rhs) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_sub_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -369,6 +378,8 @@ define i16 @atomicrmw_sub_i16(ptr %ptr, i16 %rhs) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_sub_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -407,6 +418,8 @@ define i16 @atomicrmw_sub_i16(ptr %ptr, i16 %rhs) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_sub_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -455,6 +468,7 @@ define i32 @atomicrmw_sub_i32(ptr %ptr, i32 %rhs) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_sub_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-and-or-xor.ll b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-and-or-xor.ll
index 8b24dd6b8ca00f..60cdd978ae075e 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-and-or-xor.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-and-or-xor.ll
@@ -31,6 +31,7 @@ define i8 @atomicrmw_and_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -61,6 +62,7 @@ define i8 @atomicrmw_and_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -96,6 +98,8 @@ define i16 @atomicrmw_and_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -127,6 +131,8 @@ define i16 @atomicrmw_and_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -171,6 +177,7 @@ define i32 @atomicrmw_and_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -255,6 +262,7 @@ define i8 @atomicrmw_or_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -281,6 +289,7 @@ define i8 @atomicrmw_or_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -312,6 +321,8 @@ define i16 @atomicrmw_or_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -339,6 +350,8 @@ define i16 @atomicrmw_or_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -383,6 +396,7 @@ define i32 @atomicrmw_or_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -467,6 +481,7 @@ define i8 @atomicrmw_xor_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -493,6 +508,7 @@ define i8 @atomicrmw_xor_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -524,6 +540,8 @@ define i16 @atomicrmw_xor_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -551,6 +569,8 @@ define i16 @atomicrmw_xor_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -595,6 +615,7 @@ define i32 @atomicrmw_xor_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -683,6 +704,7 @@ define i8 @atomicrmw_and_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -713,6 +735,7 @@ define i8 @atomicrmw_and_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -748,6 +771,8 @@ define i16 @atomicrmw_and_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -779,6 +804,8 @@ define i16 @atomicrmw_and_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -823,6 +850,7 @@ define i32 @atomicrmw_and_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -907,6 +935,7 @@ define i8 @atomicrmw_or_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -933,6 +962,7 @@ define i8 @atomicrmw_or_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -964,6 +994,8 @@ define i16 @atomicrmw_or_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -991,6 +1023,8 @@ define i16 @atomicrmw_or_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1035,6 +1069,7 @@ define i32 @atomicrmw_or_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1119,6 +1154,7 @@ define i8 @atomicrmw_xor_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1145,6 +1181,7 @@ define i8 @atomicrmw_xor_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1176,6 +1213,8 @@ define i16 @atomicrmw_xor_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1203,6 +1242,8 @@ define i16 @atomicrmw_xor_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1247,6 +1288,7 @@ define i32 @atomicrmw_xor_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1335,6 +1377,7 @@ define i8 @atomicrmw_and_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1365,6 +1408,7 @@ define i8 @atomicrmw_and_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1400,6 +1444,8 @@ define i16 @atomicrmw_and_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1431,6 +1477,8 @@ define i16 @atomicrmw_and_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1475,6 +1523,7 @@ define i32 @atomicrmw_and_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1559,6 +1608,7 @@ define i8 @atomicrmw_or_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1585,6 +1635,7 @@ define i8 @atomicrmw_or_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1616,6 +1667,8 @@ define i16 @atomicrmw_or_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1643,6 +1696,8 @@ define i16 @atomicrmw_or_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1687,6 +1742,7 @@ define i32 @atomicrmw_or_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1771,6 +1827,7 @@ define i8 @atomicrmw_xor_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1797,6 +1854,7 @@ define i8 @atomicrmw_xor_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1828,6 +1886,8 @@ define i16 @atomicrmw_xor_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1855,6 +1915,8 @@ define i16 @atomicrmw_xor_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1899,6 +1961,7 @@ define i32 @atomicrmw_xor_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1987,6 +2050,7 @@ define i8 @atomicrmw_and_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2017,6 +2081,7 @@ define i8 @atomicrmw_and_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2052,6 +2117,8 @@ define i16 @atomicrmw_and_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2083,6 +2150,8 @@ define i16 @atomicrmw_and_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2127,6 +2196,7 @@ define i32 @atomicrmw_and_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2211,6 +2281,7 @@ define i8 @atomicrmw_or_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2237,6 +2308,7 @@ define i8 @atomicrmw_or_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2268,6 +2340,8 @@ define i16 @atomicrmw_or_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2295,6 +2369,8 @@ define i16 @atomicrmw_or_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2339,6 +2415,7 @@ define i32 @atomicrmw_or_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2423,6 +2500,7 @@ define i8 @atomicrmw_xor_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2449,6 +2527,7 @@ define i8 @atomicrmw_xor_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2480,6 +2559,8 @@ define i16 @atomicrmw_xor_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2507,6 +2588,8 @@ define i16 @atomicrmw_xor_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2551,6 +2634,7 @@ define i32 @atomicrmw_xor_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2639,6 +2723,7 @@ define i8 @atomicrmw_and_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2669,6 +2754,7 @@ define i8 @atomicrmw_and_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2704,6 +2790,8 @@ define i16 @atomicrmw_and_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2735,6 +2823,8 @@ define i16 @atomicrmw_and_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2779,6 +2869,7 @@ define i32 @atomicrmw_and_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2863,6 +2954,7 @@ define i8 @atomicrmw_or_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2889,6 +2981,7 @@ define i8 @atomicrmw_or_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2920,6 +3013,8 @@ define i16 @atomicrmw_or_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2947,6 +3042,8 @@ define i16 @atomicrmw_or_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2991,6 +3088,7 @@ define i32 @atomicrmw_or_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3075,6 +3173,7 @@ define i8 @atomicrmw_xor_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -3101,6 +3200,7 @@ define i8 @atomicrmw_xor_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3132,6 +3232,8 @@ define i16 @atomicrmw_xor_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -3159,6 +3261,8 @@ define i16 @atomicrmw_xor_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3203,6 +3307,7 @@ define i32 @atomicrmw_xor_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-max-min-umax-umin.ll b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-max-min-umax-umin.ll
index 2e18b9b8ea5de9..04974cf38e53ff 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-max-min-umax-umin.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-max-min-umax-umin.ll
@@ -60,6 +60,7 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB0_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB0_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
@@ -71,12 +72,12 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB0_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB0_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB0_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB0_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB0_1
; RV32I-NEXT: .LBB0_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -140,6 +141,7 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB0_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB0_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -151,12 +153,12 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB0_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB0_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB0_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB0_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB0_1
; RV64I-NEXT: .LBB0_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -225,7 +227,9 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB1_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB1_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
@@ -236,12 +240,12 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB1_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB1_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB1_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB1_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB1_1
; RV32I-NEXT: .LBB1_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -306,7 +310,9 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB1_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB1_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -317,12 +323,12 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB1_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB1_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB1_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB1_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB1_1
; RV64I-NEXT: .LBB1_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -413,6 +419,7 @@ define i32 @atomicrmw_max_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB2_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB2_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -423,9 +430,8 @@ define i32 @atomicrmw_max_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB2_4
; RV64I-NEXT: .LBB2_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB2_1
+; RV64I-NEXT: blt s2, a1, .LBB2_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB2_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -723,6 +729,7 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB4_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB4_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
@@ -734,12 +741,12 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB4_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB4_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB4_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB4_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB4_1
; RV32I-NEXT: .LBB4_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -803,6 +810,7 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB4_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB4_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -814,12 +822,12 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB4_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB4_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB4_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB4_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB4_1
; RV64I-NEXT: .LBB4_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -888,7 +896,9 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB5_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB5_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
@@ -899,12 +909,12 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB5_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB5_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB5_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB5_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB5_1
; RV32I-NEXT: .LBB5_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -969,7 +979,9 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB5_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB5_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -980,12 +992,12 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB5_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB5_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB5_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB5_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB5_1
; RV64I-NEXT: .LBB5_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -1076,6 +1088,7 @@ define i32 @atomicrmw_min_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB6_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB6_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -1086,9 +1099,8 @@ define i32 @atomicrmw_min_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB6_4
; RV64I-NEXT: .LBB6_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB6_1
+; RV64I-NEXT: blt a1, s2, .LBB6_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB6_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -1380,6 +1392,7 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB8_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB8_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
@@ -1390,12 +1403,12 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB8_4
; RV32I-NEXT: .LBB8_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB8_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB8_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB8_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB8_1
; RV32I-NEXT: .LBB8_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -1453,6 +1466,7 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB8_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB8_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -1463,12 +1477,12 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB8_4
; RV64I-NEXT: .LBB8_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB8_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB8_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB8_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB8_1
; RV64I-NEXT: .LBB8_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -1534,6 +1548,7 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB9_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB9_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
@@ -1544,12 +1559,12 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB9_4
; RV32I-NEXT: .LBB9_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB9_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB9_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB9_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB9_1
; RV32I-NEXT: .LBB9_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -1612,6 +1627,7 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB9_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB9_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -1622,12 +1638,12 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB9_4
; RV64I-NEXT: .LBB9_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB9_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB9_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB9_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB9_1
; RV64I-NEXT: .LBB9_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -1719,6 +1735,7 @@ define i32 @atomicrmw_umax_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB10_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB10_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -1729,9 +1746,8 @@ define i32 @atomicrmw_umax_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB10_4
; RV64I-NEXT: .LBB10_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB10_1
+; RV64I-NEXT: bltu s2, a1, .LBB10_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB10_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -2023,6 +2039,7 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB12_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB12_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
@@ -2033,12 +2050,12 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB12_4
; RV32I-NEXT: .LBB12_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s2, .LBB12_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s2, .LBB12_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB12_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB12_1
; RV32I-NEXT: .LBB12_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -2096,6 +2113,7 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB12_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB12_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -2106,12 +2124,12 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB12_4
; RV64I-NEXT: .LBB12_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB12_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s2, .LBB12_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB12_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB12_1
; RV64I-NEXT: .LBB12_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -2177,6 +2195,7 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB13_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB13_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a3, 0
@@ -2187,12 +2206,12 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB13_4
; RV32I-NEXT: .LBB13_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s3, .LBB13_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s3, .LBB13_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB13_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB13_1
; RV32I-NEXT: .LBB13_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -2255,6 +2274,7 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB13_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB13_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -2265,12 +2285,12 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB13_4
; RV64I-NEXT: .LBB13_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s3, .LBB13_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s3, .LBB13_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB13_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB13_1
; RV64I-NEXT: .LBB13_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -2362,6 +2382,7 @@ define i32 @atomicrmw_umin_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB14_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB14_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a3, 0
@@ -2372,9 +2393,8 @@ define i32 @atomicrmw_umin_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB14_4
; RV64I-NEXT: .LBB14_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB14_1
+; RV64I-NEXT: bltu a1, s2, .LBB14_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB14_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -2672,6 +2692,7 @@ define i8 @atomicrmw_max_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB16_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB16_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -2683,12 +2704,12 @@ define i8 @atomicrmw_max_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB16_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB16_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB16_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB16_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB16_1
; RV32I-NEXT: .LBB16_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -2752,6 +2773,7 @@ define i8 @atomicrmw_max_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB16_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB16_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -2763,12 +2785,12 @@ define i8 @atomicrmw_max_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB16_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB16_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB16_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB16_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB16_1
; RV64I-NEXT: .LBB16_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -2837,7 +2859,9 @@ define i16 @atomicrmw_max_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB17_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB17_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: mv a0, s0
@@ -2848,12 +2872,12 @@ define i16 @atomicrmw_max_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB17_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB17_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB17_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB17_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB17_1
; RV32I-NEXT: .LBB17_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -2918,7 +2942,9 @@ define i16 @atomicrmw_max_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB17_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB17_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: mv a0, s0
@@ -2929,12 +2955,12 @@ define i16 @atomicrmw_max_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB17_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB17_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB17_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB17_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB17_1
; RV64I-NEXT: .LBB17_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -3025,6 +3051,7 @@ define i32 @atomicrmw_max_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB18_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB18_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -3035,9 +3062,8 @@ define i32 @atomicrmw_max_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB18_4
; RV64I-NEXT: .LBB18_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB18_1
+; RV64I-NEXT: blt s2, a1, .LBB18_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB18_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -3335,6 +3361,7 @@ define i8 @atomicrmw_min_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB20_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB20_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -3346,12 +3373,12 @@ define i8 @atomicrmw_min_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB20_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB20_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB20_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB20_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB20_1
; RV32I-NEXT: .LBB20_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -3415,6 +3442,7 @@ define i8 @atomicrmw_min_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB20_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB20_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -3426,12 +3454,12 @@ define i8 @atomicrmw_min_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB20_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB20_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB20_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB20_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB20_1
; RV64I-NEXT: .LBB20_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -3500,7 +3528,9 @@ define i16 @atomicrmw_min_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB21_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB21_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: mv a0, s0
@@ -3511,12 +3541,12 @@ define i16 @atomicrmw_min_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB21_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB21_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB21_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB21_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB21_1
; RV32I-NEXT: .LBB21_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -3581,7 +3611,9 @@ define i16 @atomicrmw_min_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB21_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB21_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: mv a0, s0
@@ -3592,12 +3624,12 @@ define i16 @atomicrmw_min_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB21_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB21_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB21_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB21_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB21_1
; RV64I-NEXT: .LBB21_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -3688,6 +3720,7 @@ define i32 @atomicrmw_min_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB22_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB22_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -3698,9 +3731,8 @@ define i32 @atomicrmw_min_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB22_4
; RV64I-NEXT: .LBB22_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB22_1
+; RV64I-NEXT: blt a1, s2, .LBB22_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB22_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -3992,6 +4024,7 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB24_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB24_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -4002,12 +4035,12 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB24_4
; RV32I-NEXT: .LBB24_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB24_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB24_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB24_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB24_1
; RV32I-NEXT: .LBB24_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -4065,6 +4098,7 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB24_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB24_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4075,12 +4109,12 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB24_4
; RV64I-NEXT: .LBB24_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB24_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB24_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB24_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB24_1
; RV64I-NEXT: .LBB24_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -4146,6 +4180,7 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB25_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB25_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -4156,12 +4191,12 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB25_4
; RV32I-NEXT: .LBB25_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB25_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB25_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB25_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB25_1
; RV32I-NEXT: .LBB25_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -4224,6 +4259,7 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB25_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB25_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4234,12 +4270,12 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB25_4
; RV64I-NEXT: .LBB25_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB25_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB25_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB25_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB25_1
; RV64I-NEXT: .LBB25_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -4331,6 +4367,7 @@ define i32 @atomicrmw_umax_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB26_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB26_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4341,9 +4378,8 @@ define i32 @atomicrmw_umax_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB26_4
; RV64I-NEXT: .LBB26_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB26_1
+; RV64I-NEXT: bltu s2, a1, .LBB26_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB26_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -4635,6 +4671,7 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB28_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB28_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -4645,12 +4682,12 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB28_4
; RV32I-NEXT: .LBB28_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s2, .LBB28_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s2, .LBB28_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB28_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB28_1
; RV32I-NEXT: .LBB28_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -4708,6 +4745,7 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB28_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB28_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4718,12 +4756,12 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB28_4
; RV64I-NEXT: .LBB28_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB28_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s2, .LBB28_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB28_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB28_1
; RV64I-NEXT: .LBB28_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -4789,6 +4827,7 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB29_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB29_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -4799,12 +4838,12 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB29_4
; RV32I-NEXT: .LBB29_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s3, .LBB29_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s3, .LBB29_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB29_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB29_1
; RV32I-NEXT: .LBB29_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -4867,6 +4906,7 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB29_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB29_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4877,12 +4917,12 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB29_4
; RV64I-NEXT: .LBB29_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s3, .LBB29_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s3, .LBB29_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB29_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB29_1
; RV64I-NEXT: .LBB29_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -4974,6 +5014,7 @@ define i32 @atomicrmw_umin_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB30_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB30_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4984,9 +5025,8 @@ define i32 @atomicrmw_umin_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB30_4
; RV64I-NEXT: .LBB30_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB30_1
+; RV64I-NEXT: bltu a1, s2, .LBB30_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB30_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -5284,6 +5324,7 @@ define i8 @atomicrmw_max_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB32_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB32_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
@@ -5295,12 +5336,12 @@ define i8 @atomicrmw_max_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB32_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB32_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB32_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB32_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB32_1
; RV32I-NEXT: .LBB32_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -5364,6 +5405,7 @@ define i8 @atomicrmw_max_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB32_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB32_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -5375,12 +5417,12 @@ define i8 @atomicrmw_max_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB32_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB32_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB32_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB32_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB32_1
; RV64I-NEXT: .LBB32_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -5449,7 +5491,9 @@ define i16 @atomicrmw_max_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB33_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB33_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a4, 0
@@ -5460,12 +5504,12 @@ define i16 @atomicrmw_max_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB33_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB33_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB33_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB33_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB33_1
; RV32I-NEXT: .LBB33_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -5530,7 +5574,9 @@ define i16 @atomicrmw_max_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB33_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB33_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a4, 0
@@ -5541,12 +5587,12 @@ define i16 @atomicrmw_max_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB33_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB33_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB33_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB33_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB33_1
; RV64I-NEXT: .LBB33_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -5637,6 +5683,7 @@ define i32 @atomicrmw_max_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB34_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB34_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -5647,9 +5694,8 @@ define i32 @atomicrmw_max_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB34_4
; RV64I-NEXT: .LBB34_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB34_1
+; RV64I-NEXT: blt s2, a1, .LBB34_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB34_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -5947,6 +5993,7 @@ define i8 @atomicrmw_min_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB36_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB36_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
@@ -5958,12 +6005,12 @@ define i8 @atomicrmw_min_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB36_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB36_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB36_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB36_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB36_1
; RV32I-NEXT: .LBB36_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -6027,6 +6074,7 @@ define i8 @atomicrmw_min_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB36_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB36_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -6038,12 +6086,12 @@ define i8 @atomicrmw_min_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB36_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB36_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB36_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB36_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB36_1
; RV64I-NEXT: .LBB36_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -6112,7 +6160,9 @@ define i16 @atomicrmw_min_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB37_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB37_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: li a4, 0
@@ -6123,12 +6173,12 @@ define i16 @atomicrmw_min_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB37_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB37_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB37_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB37_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB37_1
; RV32I-NEXT: .LBB37_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -6193,7 +6243,9 @@ define i16 @atomicrmw_min_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB37_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB37_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: li a4, 0
@@ -6204,12 +6256,12 @@ define i16 @atomicrmw_min_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB37_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB37_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB37_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB37_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB37_1
; RV64I-NEXT: .LBB37_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -6300,6 +6352,7 @@ define i32 @atomicrmw_min_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB38_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB38_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -6310,9 +6363,8 @@ define i32 @atomicrmw_min_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB38_4
; RV64I-NEXT: .LBB38_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB38_1
+; RV64I-NEXT: blt a1, s2, .LBB38_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB38_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -6604,6 +6656,7 @@ define i8 @atomicrmw_umax_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB40_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB40_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
@@ -6614,12 +6667,12 @@ define i8 @atomicrmw_umax_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB40_4
; RV32I-NEXT: .LBB40_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB40_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB40_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB40_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB40_1
; RV32I-NEXT: .LBB40_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -6677,6 +6730,7 @@ define i8 @atomicrmw_umax_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB40_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB40_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -6687,12 +6741,12 @@ define i8 @atomicrmw_umax_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB40_4
; RV64I-NEXT: .LBB40_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB40_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB40_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB40_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB40_1
; RV64I-NEXT: .LBB40_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -6758,6 +6812,7 @@ define i16 @atomicrmw_umax_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB41_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB41_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
@@ -6768,12 +6823,12 @@ define i16 @atomicrmw_umax_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB41_4
; RV32I-NEXT: .LBB41_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB41_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB41_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB41_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB41_1
; RV32I-NEXT: .LBB41_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -6836,6 +6891,7 @@ define i16 @atomicrmw_umax_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB41_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB41_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -6846,12 +6902,12 @@ define i16 @atomicrmw_umax_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB41_4
; RV64I-NEXT: .LBB41_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB41_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB41_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB41_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB41_1
; RV64I-NEXT: .LBB41_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -6943,6 +6999,7 @@ define i32 @atomicrmw_umax_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB42_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB42_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -6953,9 +7010,8 @@ define i32 @atomicrmw_umax_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB42_4
; RV64I-NEXT: .LBB42_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB42_1
+; RV64I-NEXT: bltu s2, a1, .LBB42_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB42_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -7247,6 +7303,7 @@ define i8 @atomicrmw_umin_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB44_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB44_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
@@ -7257,12 +7314,12 @@ define i8 @atomicrmw_umin_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB44_4
; RV32I-NEXT: .LBB44_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s2, .LBB44_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s2, .LBB44_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB44_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB44_1
; RV32I-NEXT: .LBB44_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -7320,6 +7377,7 @@ define i8 @atomicrmw_umin_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB44_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB44_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -7330,12 +7388,12 @@ define i8 @atomicrmw_umin_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB44_4
; RV64I-NEXT: .LBB44_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB44_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s2, .LBB44_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB44_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB44_1
; RV64I-NEXT: .LBB44_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -7401,6 +7459,7 @@ define i16 @atomicrmw_umin_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB45_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB45_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s0
@@ -7411,12 +7470,12 @@ define i16 @atomicrmw_umin_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB45_4
; RV32I-NEXT: .LBB45_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s3, .LBB45_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s3, .LBB45_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB45_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB45_1
; RV32I-NEXT: .LBB45_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -7479,6 +7538,7 @@ define i16 @atomicrmw_umin_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB45_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB45_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -7489,12 +7549,12 @@ define i16 @atomicrmw_umin_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB45_4
; RV64I-NEXT: .LBB45_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s3, .LBB45_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s3, .LBB45_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB45_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB45_1
; RV64I-NEXT: .LBB45_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -7586,6 +7646,7 @@ define i32 @atomicrmw_umin_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB46_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB46_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s0
@@ -7596,9 +7657,8 @@ define i32 @atomicrmw_umin_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB46_4
; RV64I-NEXT: .LBB46_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB46_1
+; RV64I-NEXT: bltu a1, s2, .LBB46_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB46_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -7896,6 +7956,7 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB48_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB48_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -7907,12 +7968,12 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB48_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB48_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB48_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB48_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB48_1
; RV32I-NEXT: .LBB48_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -7976,6 +8037,7 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB48_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB48_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -7987,12 +8049,12 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB48_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB48_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB48_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB48_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB48_1
; RV64I-NEXT: .LBB48_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -8061,7 +8123,9 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB49_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB49_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: mv a0, s0
@@ -8072,12 +8136,12 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB49_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB49_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB49_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB49_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB49_1
; RV32I-NEXT: .LBB49_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -8142,7 +8206,9 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB49_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB49_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: mv a0, s0
@@ -8153,12 +8219,12 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB49_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB49_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB49_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB49_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB49_1
; RV64I-NEXT: .LBB49_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -8249,6 +8315,7 @@ define i32 @atomicrmw_max_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB50_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB50_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -8259,9 +8326,8 @@ define i32 @atomicrmw_max_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB50_4
; RV64I-NEXT: .LBB50_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB50_1
+; RV64I-NEXT: blt s2, a1, .LBB50_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB50_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -8559,6 +8625,7 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB52_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB52_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -8570,12 +8637,12 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB52_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB52_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB52_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB52_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB52_1
; RV32I-NEXT: .LBB52_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -8639,6 +8706,7 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB52_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB52_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -8650,12 +8718,12 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB52_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB52_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB52_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB52_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB52_1
; RV64I-NEXT: .LBB52_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -8724,7 +8792,9 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB53_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB53_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: mv a0, s0
@@ -8735,12 +8805,12 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB53_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB53_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB53_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB53_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB53_1
; RV32I-NEXT: .LBB53_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -8805,7 +8875,9 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB53_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB53_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: mv a0, s0
@@ -8816,12 +8888,12 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB53_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB53_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB53_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB53_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB53_1
; RV64I-NEXT: .LBB53_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -8912,6 +8984,7 @@ define i32 @atomicrmw_min_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB54_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB54_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -8922,9 +8995,8 @@ define i32 @atomicrmw_min_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB54_4
; RV64I-NEXT: .LBB54_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB54_1
+; RV64I-NEXT: blt a1, s2, .LBB54_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB54_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -9216,6 +9288,7 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB56_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB56_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -9226,12 +9299,12 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB56_4
; RV32I-NEXT: .LBB56_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB56_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB56_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB56_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB56_1
; RV32I-NEXT: .LBB56_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -9289,6 +9362,7 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB56_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB56_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -9299,12 +9373,12 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB56_4
; RV64I-NEXT: .LBB56_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB56_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB56_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB56_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB56_1
; RV64I-NEXT: .LBB56_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -9370,6 +9444,7 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB57_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB57_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -9380,12 +9455,12 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB57_4
; RV32I-NEXT: .LBB57_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB57_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB57_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB57_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB57_1
; RV32I-NEXT: .LBB57_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -9448,6 +9523,7 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB57_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB57_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -9458,12 +9534,12 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB57_4
; RV64I-NEXT: .LBB57_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB57_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB57_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB57_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB57_1
; RV64I-NEXT: .LBB57_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -9555,6 +9631,7 @@ define i32 @atomicrmw_umax_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB58_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB58_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -9565,9 +9642,8 @@ define i32 @atomicrmw_umax_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB58_4
; RV64I-NEXT: .LBB58_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB58_1
+; RV64I-NEXT: bltu s2, a1, .LBB58_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB58_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -9859,6 +9935,7 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB60_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB60_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -9869,12 +9946,12 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB60_4
; RV32I-NEXT: .LBB60_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s2, .LBB60_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s2, .LBB60_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB60_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB60_1
; RV32I-NEXT: .LBB60_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -9932,6 +10009,7 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB60_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB60_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -9942,12 +10020,12 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB60_4
; RV64I-NEXT: .LBB60_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB60_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s2, .LBB60_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB60_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB60_1
; RV64I-NEXT: .LBB60_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10013,6 +10091,7 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB61_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB61_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -10023,12 +10102,12 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB61_4
; RV32I-NEXT: .LBB61_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s3, .LBB61_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s3, .LBB61_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB61_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB61_1
; RV32I-NEXT: .LBB61_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -10091,6 +10170,7 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB61_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB61_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -10101,12 +10181,12 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB61_4
; RV64I-NEXT: .LBB61_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s3, .LBB61_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s3, .LBB61_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB61_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB61_1
; RV64I-NEXT: .LBB61_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10198,6 +10278,7 @@ define i32 @atomicrmw_umin_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB62_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB62_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -10208,9 +10289,8 @@ define i32 @atomicrmw_umin_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB62_4
; RV64I-NEXT: .LBB62_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB62_1
+; RV64I-NEXT: bltu a1, s2, .LBB62_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB62_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -10508,6 +10588,7 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB64_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB64_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -10519,12 +10600,12 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB64_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB64_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB64_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB64_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB64_1
; RV32I-NEXT: .LBB64_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -10588,6 +10669,7 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB64_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB64_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -10599,12 +10681,12 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB64_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB64_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB64_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB64_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB64_1
; RV64I-NEXT: .LBB64_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10673,7 +10755,9 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB65_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB65_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
; RV32I-NEXT: mv a0, s0
@@ -10684,12 +10768,12 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB65_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB65_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB65_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB65_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB65_1
; RV32I-NEXT: .LBB65_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -10754,7 +10838,9 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB65_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB65_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
; RV64I-NEXT: mv a0, s0
@@ -10765,12 +10851,12 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB65_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB65_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB65_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB65_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB65_1
; RV64I-NEXT: .LBB65_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10861,6 +10947,7 @@ define i32 @atomicrmw_max_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB66_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB66_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -10871,9 +10958,8 @@ define i32 @atomicrmw_max_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB66_4
; RV64I-NEXT: .LBB66_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB66_1
+; RV64I-NEXT: blt s2, a1, .LBB66_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB66_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -11171,6 +11257,7 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB68_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB68_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -11182,12 +11269,12 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB68_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB68_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB68_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB68_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB68_1
; RV32I-NEXT: .LBB68_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -11251,6 +11338,7 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB68_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB68_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -11262,12 +11350,12 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB68_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB68_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB68_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB68_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB68_1
; RV64I-NEXT: .LBB68_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -11336,7 +11424,9 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB69_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB69_2 Depth=1
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: addi a1, sp, 14
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
; RV32I-NEXT: mv a0, s0
@@ -11347,12 +11437,12 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB69_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt a0, s2, .LBB69_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt a2, s2, .LBB69_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB69_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB69_1
; RV32I-NEXT: .LBB69_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -11417,7 +11507,9 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB69_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB69_2 Depth=1
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: addi a1, sp, 14
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
; RV64I-NEXT: mv a0, s0
@@ -11428,12 +11520,12 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB69_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB69_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt a2, s2, .LBB69_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB69_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB69_1
; RV64I-NEXT: .LBB69_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -11524,6 +11616,7 @@ define i32 @atomicrmw_min_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB70_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB70_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -11534,9 +11627,8 @@ define i32 @atomicrmw_min_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB70_4
; RV64I-NEXT: .LBB70_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s2, .LBB70_1
+; RV64I-NEXT: blt a1, s2, .LBB70_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB70_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -11828,6 +11920,7 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB72_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB72_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -11838,12 +11931,12 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB72_4
; RV32I-NEXT: .LBB72_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB72_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB72_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB72_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB72_1
; RV32I-NEXT: .LBB72_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -11901,6 +11994,7 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB72_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB72_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -11911,12 +12005,12 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB72_4
; RV64I-NEXT: .LBB72_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB72_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB72_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB72_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB72_1
; RV64I-NEXT: .LBB72_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -11982,6 +12076,7 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB73_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB73_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -11992,12 +12087,12 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB73_4
; RV32I-NEXT: .LBB73_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB73_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB73_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB73_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB73_1
; RV32I-NEXT: .LBB73_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -12060,6 +12155,7 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB73_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB73_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -12070,12 +12166,12 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB73_4
; RV64I-NEXT: .LBB73_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB73_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB73_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB73_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB73_1
; RV64I-NEXT: .LBB73_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -12167,6 +12263,7 @@ define i32 @atomicrmw_umax_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB74_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB74_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -12177,9 +12274,8 @@ define i32 @atomicrmw_umax_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB74_4
; RV64I-NEXT: .LBB74_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB74_1
+; RV64I-NEXT: bltu s2, a1, .LBB74_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB74_2 Depth=1
; RV64I-NEXT: mv a2, s1
@@ -12471,6 +12567,7 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: .LBB76_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB76_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -12481,12 +12578,12 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB76_4
; RV32I-NEXT: .LBB76_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s2, .LBB76_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s2, .LBB76_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB76_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB76_1
; RV32I-NEXT: .LBB76_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -12544,6 +12641,7 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: .LBB76_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB76_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -12554,12 +12652,12 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB76_4
; RV64I-NEXT: .LBB76_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB76_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s2, .LBB76_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB76_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB76_1
; RV64I-NEXT: .LBB76_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -12625,6 +12723,7 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: .LBB77_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB77_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -12635,12 +12734,12 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I-NEXT: beqz a0, .LBB77_4
; RV32I-NEXT: .LBB77_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu a0, s3, .LBB77_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu a2, s3, .LBB77_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB77_2 Depth=1
-; RV32I-NEXT: mv a2, s1
+; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: j .LBB77_1
; RV32I-NEXT: .LBB77_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -12703,6 +12802,7 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: .LBB77_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB77_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -12713,12 +12813,12 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB77_4
; RV64I-NEXT: .LBB77_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s3, .LBB77_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu a2, s3, .LBB77_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB77_2 Depth=1
-; RV64I-NEXT: mv a2, s1
+; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: j .LBB77_1
; RV64I-NEXT: .LBB77_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -12810,6 +12910,7 @@ define i32 @atomicrmw_umin_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: .LBB78_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB78_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -12820,9 +12921,8 @@ define i32 @atomicrmw_umin_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I-NEXT: beqz a0, .LBB78_4
; RV64I-NEXT: .LBB78_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s2, .LBB78_1
+; RV64I-NEXT: bltu a1, s2, .LBB78_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB78_2 Depth=1
; RV64I-NEXT: mv a2, s1
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-xchg.ll b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-xchg.ll
index 909146854b33e3..77f9b03252fb51 100644
--- a/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-xchg.ll
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/atomicrmw-xchg.ll
@@ -37,6 +37,7 @@ define i8 @atomicrmw_xchg_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -73,6 +74,7 @@ define i8 @atomicrmw_xchg_i8_monotonic(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -114,6 +116,8 @@ define i16 @atomicrmw_xchg_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -151,6 +155,8 @@ define i16 @atomicrmw_xchg_i16_monotonic(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -195,6 +201,7 @@ define i32 @atomicrmw_xchg_i32_monotonic(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -289,6 +296,7 @@ define i8 @atomicrmw_xchg_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -325,6 +333,7 @@ define i8 @atomicrmw_xchg_i8_acquire(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -366,6 +375,8 @@ define i16 @atomicrmw_xchg_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -403,6 +414,8 @@ define i16 @atomicrmw_xchg_i16_acquire(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -447,6 +460,7 @@ define i32 @atomicrmw_xchg_i32_acquire(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -541,6 +555,7 @@ define i8 @atomicrmw_xchg_i8_release(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -577,6 +592,7 @@ define i8 @atomicrmw_xchg_i8_release(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -618,6 +634,8 @@ define i16 @atomicrmw_xchg_i16_release(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -655,6 +673,8 @@ define i16 @atomicrmw_xchg_i16_release(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -699,6 +719,7 @@ define i32 @atomicrmw_xchg_i32_release(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -793,6 +814,7 @@ define i8 @atomicrmw_xchg_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -829,6 +851,7 @@ define i8 @atomicrmw_xchg_i8_acq_rel(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -870,6 +893,8 @@ define i16 @atomicrmw_xchg_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -907,6 +932,8 @@ define i16 @atomicrmw_xchg_i16_acq_rel(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -951,6 +978,7 @@ define i32 @atomicrmw_xchg_i32_acq_rel(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1045,6 +1073,7 @@ define i8 @atomicrmw_xchg_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1081,6 +1110,7 @@ define i8 @atomicrmw_xchg_i8_seq_cst(ptr %ptr, i8 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1122,6 +1152,8 @@ define i16 @atomicrmw_xchg_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1159,6 +1191,8 @@ define i16 @atomicrmw_xchg_i16_seq_cst(ptr %ptr, i16 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1203,6 +1237,7 @@ define i32 @atomicrmw_xchg_i32_seq_cst(ptr %ptr, i32 %val) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/atomic-cmpxchg.ll b/llvm/test/CodeGen/RISCV/atomic-cmpxchg.ll
index 5446e38f01792a..951e1f1179648e 100644
--- a/llvm/test/CodeGen/RISCV/atomic-cmpxchg.ll
+++ b/llvm/test/CodeGen/RISCV/atomic-cmpxchg.ll
@@ -30,6 +30,7 @@ define void @cmpxchg_i8_monotonic_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
@@ -66,6 +67,7 @@ define void @cmpxchg_i8_monotonic_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -157,6 +159,7 @@ define void @cmpxchg_i8_acquire_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 0
@@ -262,6 +265,7 @@ define void @cmpxchg_i8_acquire_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 0
@@ -381,6 +385,7 @@ define void @cmpxchg_i8_acquire_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -486,6 +491,7 @@ define void @cmpxchg_i8_acquire_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -605,6 +611,7 @@ define void @cmpxchg_i8_release_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 0
@@ -710,6 +717,7 @@ define void @cmpxchg_i8_release_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 0
@@ -829,6 +837,7 @@ define void @cmpxchg_i8_release_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 2
@@ -934,6 +943,7 @@ define void @cmpxchg_i8_release_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 2
@@ -1053,6 +1063,7 @@ define void @cmpxchg_i8_acq_rel_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 0
@@ -1158,6 +1169,7 @@ define void @cmpxchg_i8_acq_rel_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 0
@@ -1277,6 +1289,7 @@ define void @cmpxchg_i8_acq_rel_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -1382,6 +1395,7 @@ define void @cmpxchg_i8_acq_rel_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -1501,6 +1515,7 @@ define void @cmpxchg_i8_seq_cst_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 0
@@ -1537,6 +1552,7 @@ define void @cmpxchg_i8_seq_cst_monotonic(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 0
@@ -1633,6 +1649,7 @@ define void @cmpxchg_i8_seq_cst_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 2
@@ -1669,6 +1686,7 @@ define void @cmpxchg_i8_seq_cst_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 2
@@ -1765,6 +1783,7 @@ define void @cmpxchg_i8_seq_cst_seq_cst(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -1801,6 +1820,7 @@ define void @cmpxchg_i8_seq_cst_seq_cst(ptr %ptr, i8 %cmp, i8 %val) nounwind {
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -1899,7 +1919,9 @@ define void @cmpxchg_i16_monotonic_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounw
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -1936,7 +1958,9 @@ define void @cmpxchg_i16_monotonic_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounw
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2030,7 +2054,9 @@ define void @cmpxchg_i16_acquire_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2139,7 +2165,9 @@ define void @cmpxchg_i16_acquire_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2262,7 +2290,9 @@ define void @cmpxchg_i16_acquire_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2371,7 +2401,9 @@ define void @cmpxchg_i16_acquire_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2494,7 +2526,9 @@ define void @cmpxchg_i16_release_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2603,7 +2637,9 @@ define void @cmpxchg_i16_release_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2726,7 +2762,9 @@ define void @cmpxchg_i16_release_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -2835,7 +2873,9 @@ define void @cmpxchg_i16_release_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -2958,7 +2998,9 @@ define void @cmpxchg_i16_acq_rel_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3067,7 +3109,9 @@ define void @cmpxchg_i16_acq_rel_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3190,7 +3234,9 @@ define void @cmpxchg_i16_acq_rel_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3299,7 +3345,9 @@ define void @cmpxchg_i16_acq_rel_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3422,7 +3470,9 @@ define void @cmpxchg_i16_seq_cst_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3459,7 +3509,9 @@ define void @cmpxchg_i16_seq_cst_monotonic(ptr %ptr, i16 %cmp, i16 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3558,7 +3610,9 @@ define void @cmpxchg_i16_seq_cst_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 2
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3595,7 +3649,9 @@ define void @cmpxchg_i16_seq_cst_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 2
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3694,7 +3750,9 @@ define void @cmpxchg_i16_seq_cst_seq_cst(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -3731,7 +3789,9 @@ define void @cmpxchg_i16_seq_cst_seq_cst(ptr %ptr, i16 %cmp, i16 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -3872,6 +3932,7 @@ define void @cmpxchg_i32_monotonic_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounw
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -3968,6 +4029,7 @@ define void @cmpxchg_i32_acquire_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 0
@@ -4074,6 +4136,7 @@ define void @cmpxchg_i32_acquire_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -4180,6 +4243,7 @@ define void @cmpxchg_i32_release_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 0
@@ -4286,6 +4350,7 @@ define void @cmpxchg_i32_release_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: li a4, 2
@@ -4392,6 +4457,7 @@ define void @cmpxchg_i32_acq_rel_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 0
@@ -4498,6 +4564,7 @@ define void @cmpxchg_i32_acq_rel_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -4604,6 +4671,7 @@ define void @cmpxchg_i32_seq_cst_monotonic(ptr %ptr, i32 %cmp, i32 %val) nounwin
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 0
@@ -4710,6 +4778,7 @@ define void @cmpxchg_i32_seq_cst_acquire(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 2
@@ -4818,6 +4887,7 @@ define void @cmpxchg_i32_seq_cst_seq_cst(ptr %ptr, i32 %cmp, i32 %val) nounwind
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sw a1, 4(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 4
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
diff --git a/llvm/test/CodeGen/RISCV/atomic-load-store.ll b/llvm/test/CodeGen/RISCV/atomic-load-store.ll
index d17d994d4dd3cd..1216481fd945c0 100644
--- a/llvm/test/CodeGen/RISCV/atomic-load-store.ll
+++ b/llvm/test/CodeGen/RISCV/atomic-load-store.ll
@@ -1178,6 +1178,7 @@ define void @atomic_store_i8_unordered(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1198,6 +1199,7 @@ define void @atomic_store_i8_unordered(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1222,6 +1224,7 @@ define void @atomic_store_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1242,6 +1245,7 @@ define void @atomic_store_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1266,6 +1270,7 @@ define void @atomic_store_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1293,6 +1298,7 @@ define void @atomic_store_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1366,6 +1372,7 @@ define void @atomic_store_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_store_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1395,6 +1402,7 @@ define void @atomic_store_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_store_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1464,6 +1472,8 @@ define void @atomic_store_i16_unordered(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1484,6 +1494,8 @@ define void @atomic_store_i16_unordered(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1508,6 +1520,8 @@ define void @atomic_store_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1528,6 +1542,8 @@ define void @atomic_store_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1552,6 +1568,8 @@ define void @atomic_store_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1579,6 +1597,8 @@ define void @atomic_store_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1652,6 +1672,8 @@ define void @atomic_store_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_store_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -1681,6 +1703,8 @@ define void @atomic_store_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_store_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1770,6 +1794,7 @@ define void @atomic_store_i32_unordered(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1814,6 +1839,7 @@ define void @atomic_store_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1865,6 +1891,7 @@ define void @atomic_store_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -1967,6 +1994,7 @@ define void @atomic_store_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_store_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/CodeGen/RISCV/atomic-load-zext.ll b/llvm/test/CodeGen/RISCV/atomic-load-zext.ll
index 37da1a21bc2664..fd9afd01c0c761 100644
--- a/llvm/test/CodeGen/RISCV/atomic-load-zext.ll
+++ b/llvm/test/CodeGen/RISCV/atomic-load-zext.ll
@@ -294,7 +294,6 @@ define zeroext i8 @atomic_load_i8_unordered(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 0
; RV32I-NEXT: call __atomic_load_1
-; RV32I-NEXT: zext.b a0, a0
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -310,7 +309,6 @@ define zeroext i8 @atomic_load_i8_unordered(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 0
; RV64I-NEXT: call __atomic_load_1
-; RV64I-NEXT: zext.b a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -330,7 +328,6 @@ define zeroext i8 @atomic_load_i8_monotonic(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 0
; RV32I-NEXT: call __atomic_load_1
-; RV32I-NEXT: zext.b a0, a0
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -346,7 +343,6 @@ define zeroext i8 @atomic_load_i8_monotonic(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 0
; RV64I-NEXT: call __atomic_load_1
-; RV64I-NEXT: zext.b a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -366,7 +362,6 @@ define zeroext i8 @atomic_load_i8_acquire(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 2
; RV32I-NEXT: call __atomic_load_1
-; RV32I-NEXT: zext.b a0, a0
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -388,7 +383,6 @@ define zeroext i8 @atomic_load_i8_acquire(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 2
; RV64I-NEXT: call __atomic_load_1
-; RV64I-NEXT: zext.b a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -458,7 +452,6 @@ define zeroext i8 @atomic_load_i8_seq_cst(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 5
; RV32I-NEXT: call __atomic_load_1
-; RV32I-NEXT: zext.b a0, a0
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -482,7 +475,6 @@ define zeroext i8 @atomic_load_i8_seq_cst(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 5
; RV64I-NEXT: call __atomic_load_1
-; RV64I-NEXT: zext.b a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -548,8 +540,6 @@ define zeroext i16 @atomic_load_i16_unordered(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 0
; RV32I-NEXT: call __atomic_load_2
-; RV32I-NEXT: slli a0, a0, 16
-; RV32I-NEXT: srli a0, a0, 16
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -565,8 +555,6 @@ define zeroext i16 @atomic_load_i16_unordered(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 0
; RV64I-NEXT: call __atomic_load_2
-; RV64I-NEXT: slli a0, a0, 48
-; RV64I-NEXT: srli a0, a0, 48
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -586,8 +574,6 @@ define zeroext i16 @atomic_load_i16_monotonic(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 0
; RV32I-NEXT: call __atomic_load_2
-; RV32I-NEXT: slli a0, a0, 16
-; RV32I-NEXT: srli a0, a0, 16
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -603,8 +589,6 @@ define zeroext i16 @atomic_load_i16_monotonic(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 0
; RV64I-NEXT: call __atomic_load_2
-; RV64I-NEXT: slli a0, a0, 48
-; RV64I-NEXT: srli a0, a0, 48
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -624,8 +608,6 @@ define zeroext i16 @atomic_load_i16_acquire(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 2
; RV32I-NEXT: call __atomic_load_2
-; RV32I-NEXT: slli a0, a0, 16
-; RV32I-NEXT: srli a0, a0, 16
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -647,8 +629,6 @@ define zeroext i16 @atomic_load_i16_acquire(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 2
; RV64I-NEXT: call __atomic_load_2
-; RV64I-NEXT: slli a0, a0, 48
-; RV64I-NEXT: srli a0, a0, 48
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -720,8 +700,6 @@ define zeroext i16 @atomic_load_i16_seq_cst(ptr %a) nounwind {
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: li a1, 5
; RV32I-NEXT: call __atomic_load_2
-; RV32I-NEXT: slli a0, a0, 16
-; RV32I-NEXT: srli a0, a0, 16
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT: addi sp, sp, 16
; RV32I-NEXT: ret
@@ -745,8 +723,6 @@ define zeroext i16 @atomic_load_i16_seq_cst(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 5
; RV64I-NEXT: call __atomic_load_2
-; RV64I-NEXT: slli a0, a0, 48
-; RV64I-NEXT: srli a0, a0, 48
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
diff --git a/llvm/test/CodeGen/RISCV/atomic-rmw.ll b/llvm/test/CodeGen/RISCV/atomic-rmw.ll
index ca03a08e037403..b3ed672f4b367c 100644
--- a/llvm/test/CodeGen/RISCV/atomic-rmw.ll
+++ b/llvm/test/CodeGen/RISCV/atomic-rmw.ll
@@ -48,6 +48,7 @@ define i8 @atomicrmw_xchg_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -98,6 +99,7 @@ define i8 @atomicrmw_xchg_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -212,6 +214,7 @@ define i8 @atomicrmw_xchg_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -282,6 +285,7 @@ define i8 @atomicrmw_xchg_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -456,6 +460,7 @@ define i8 @atomicrmw_xchg_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -526,6 +531,7 @@ define i8 @atomicrmw_xchg_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -700,6 +706,7 @@ define i8 @atomicrmw_xchg_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -770,6 +777,7 @@ define i8 @atomicrmw_xchg_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -944,6 +952,7 @@ define i8 @atomicrmw_xchg_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -994,6 +1003,7 @@ define i8 @atomicrmw_xchg_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2657,6 +2667,7 @@ define i8 @atomicrmw_add_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_add_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2707,6 +2718,7 @@ define i8 @atomicrmw_add_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_add_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -2821,6 +2833,7 @@ define i8 @atomicrmw_add_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_add_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -2891,6 +2904,7 @@ define i8 @atomicrmw_add_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_add_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3065,6 +3079,7 @@ define i8 @atomicrmw_add_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_add_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -3135,6 +3150,7 @@ define i8 @atomicrmw_add_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_add_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3309,6 +3325,7 @@ define i8 @atomicrmw_add_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_add_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -3379,6 +3396,7 @@ define i8 @atomicrmw_add_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_add_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3553,6 +3571,7 @@ define i8 @atomicrmw_add_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_add_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -3603,6 +3622,7 @@ define i8 @atomicrmw_add_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_add_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3717,6 +3737,7 @@ define i8 @atomicrmw_sub_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_sub_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -3767,6 +3788,7 @@ define i8 @atomicrmw_sub_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_sub_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -3885,6 +3907,7 @@ define i8 @atomicrmw_sub_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_sub_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -3955,6 +3978,7 @@ define i8 @atomicrmw_sub_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_sub_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -4133,6 +4157,7 @@ define i8 @atomicrmw_sub_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_sub_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -4203,6 +4228,7 @@ define i8 @atomicrmw_sub_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_sub_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -4381,6 +4407,7 @@ define i8 @atomicrmw_sub_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_sub_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -4451,6 +4478,7 @@ define i8 @atomicrmw_sub_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_sub_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -4629,6 +4657,7 @@ define i8 @atomicrmw_sub_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_sub_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -4679,6 +4708,7 @@ define i8 @atomicrmw_sub_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_sub_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -4797,6 +4827,7 @@ define i8 @atomicrmw_and_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -4840,6 +4871,7 @@ define i8 @atomicrmw_and_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -4935,6 +4967,7 @@ define i8 @atomicrmw_and_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -4992,6 +5025,7 @@ define i8 @atomicrmw_and_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -5129,6 +5163,7 @@ define i8 @atomicrmw_and_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -5186,6 +5221,7 @@ define i8 @atomicrmw_and_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -5323,6 +5359,7 @@ define i8 @atomicrmw_and_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -5380,6 +5417,7 @@ define i8 @atomicrmw_and_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -5517,6 +5555,7 @@ define i8 @atomicrmw_and_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -5574,6 +5613,7 @@ define i8 @atomicrmw_and_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -5711,6 +5751,7 @@ define i8 @atomicrmw_nand_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_nand_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -5763,6 +5804,7 @@ define i8 @atomicrmw_nand_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_nand_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -6005,6 +6047,7 @@ define i8 @atomicrmw_nand_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_nand_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -6078,6 +6121,7 @@ define i8 @atomicrmw_nand_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_nand_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -6383,6 +6427,7 @@ define i8 @atomicrmw_nand_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_nand_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -6456,6 +6501,7 @@ define i8 @atomicrmw_nand_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_nand_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -6761,6 +6807,7 @@ define i8 @atomicrmw_nand_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_nand_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -6834,6 +6881,7 @@ define i8 @atomicrmw_nand_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_nand_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -7139,6 +7187,7 @@ define i8 @atomicrmw_nand_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_nand_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -7191,6 +7240,7 @@ define i8 @atomicrmw_nand_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_nand_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -7437,6 +7487,7 @@ define i8 @atomicrmw_or_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -7472,6 +7523,7 @@ define i8 @atomicrmw_or_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -7551,6 +7603,7 @@ define i8 @atomicrmw_or_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -7596,6 +7649,7 @@ define i8 @atomicrmw_or_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -7705,6 +7759,7 @@ define i8 @atomicrmw_or_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -7750,6 +7805,7 @@ define i8 @atomicrmw_or_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -7859,6 +7915,7 @@ define i8 @atomicrmw_or_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -7904,6 +7961,7 @@ define i8 @atomicrmw_or_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -8013,6 +8071,7 @@ define i8 @atomicrmw_or_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -8058,6 +8117,7 @@ define i8 @atomicrmw_or_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -8167,6 +8227,7 @@ define i8 @atomicrmw_xor_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -8202,6 +8263,7 @@ define i8 @atomicrmw_xor_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -8281,6 +8343,7 @@ define i8 @atomicrmw_xor_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -8326,6 +8389,7 @@ define i8 @atomicrmw_xor_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -8435,6 +8499,7 @@ define i8 @atomicrmw_xor_i8_release(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -8480,6 +8545,7 @@ define i8 @atomicrmw_xor_i8_release(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -8589,6 +8655,7 @@ define i8 @atomicrmw_xor_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -8634,6 +8701,7 @@ define i8 @atomicrmw_xor_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -8743,6 +8811,7 @@ define i8 @atomicrmw_xor_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -8788,6 +8857,7 @@ define i8 @atomicrmw_xor_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -8911,6 +8981,7 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB45_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB45_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -8921,12 +8992,12 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB45_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB45_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB45_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB45_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB45_1
; RV32I-NEXT: .LBB45_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -9013,6 +9084,7 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB45_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB45_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -9023,12 +9095,12 @@ define i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB45_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB45_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB45_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB45_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB45_1
; RV64I-NEXT: .LBB45_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -9197,6 +9269,7 @@ define i8 @atomicrmw_max_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB46_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB46_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -9207,12 +9280,12 @@ define i8 @atomicrmw_max_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB46_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB46_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB46_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB46_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB46_1
; RV32I-NEXT: .LBB46_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -9328,6 +9401,7 @@ define i8 @atomicrmw_max_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB46_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB46_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -9338,12 +9412,12 @@ define i8 @atomicrmw_max_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB46_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB46_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB46_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB46_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB46_1
; RV64I-NEXT: .LBB46_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -9599,6 +9673,7 @@ define i8 @atomicrmw_max_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB47_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB47_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -9609,12 +9684,12 @@ define i8 @atomicrmw_max_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB47_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB47_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB47_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB47_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB47_1
; RV32I-NEXT: .LBB47_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -9730,6 +9805,7 @@ define i8 @atomicrmw_max_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB47_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB47_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -9740,12 +9816,12 @@ define i8 @atomicrmw_max_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB47_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB47_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB47_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB47_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB47_1
; RV64I-NEXT: .LBB47_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10001,6 +10077,7 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB48_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB48_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -10011,12 +10088,12 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB48_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB48_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB48_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB48_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB48_1
; RV32I-NEXT: .LBB48_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -10132,6 +10209,7 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB48_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB48_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -10142,12 +10220,12 @@ define i8 @atomicrmw_max_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB48_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB48_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB48_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB48_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB48_1
; RV64I-NEXT: .LBB48_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10403,6 +10481,7 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB49_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB49_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -10413,12 +10492,12 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB49_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB49_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB49_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB49_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB49_1
; RV32I-NEXT: .LBB49_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -10505,6 +10584,7 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB49_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB49_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -10515,12 +10595,12 @@ define i8 @atomicrmw_max_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB49_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB49_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB49_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB49_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB49_1
; RV64I-NEXT: .LBB49_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10689,6 +10769,7 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB50_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB50_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -10699,12 +10780,12 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB50_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB50_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB50_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB50_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB50_1
; RV32I-NEXT: .LBB50_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -10791,6 +10872,7 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB50_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB50_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -10801,12 +10883,12 @@ define i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB50_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB50_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB50_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB50_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB50_1
; RV64I-NEXT: .LBB50_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -10975,6 +11057,7 @@ define i8 @atomicrmw_min_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB51_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB51_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -10985,12 +11068,12 @@ define i8 @atomicrmw_min_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB51_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB51_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB51_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB51_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB51_1
; RV32I-NEXT: .LBB51_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -11106,6 +11189,7 @@ define i8 @atomicrmw_min_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB51_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB51_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -11116,12 +11200,12 @@ define i8 @atomicrmw_min_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB51_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB51_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB51_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB51_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB51_1
; RV64I-NEXT: .LBB51_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -11377,6 +11461,7 @@ define i8 @atomicrmw_min_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB52_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB52_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -11387,12 +11472,12 @@ define i8 @atomicrmw_min_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB52_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB52_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB52_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB52_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB52_1
; RV32I-NEXT: .LBB52_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -11508,6 +11593,7 @@ define i8 @atomicrmw_min_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB52_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB52_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -11518,12 +11604,12 @@ define i8 @atomicrmw_min_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB52_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB52_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB52_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB52_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB52_1
; RV64I-NEXT: .LBB52_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -11779,6 +11865,7 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB53_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB53_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -11789,12 +11876,12 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB53_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB53_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB53_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB53_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB53_1
; RV32I-NEXT: .LBB53_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -11910,6 +11997,7 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB53_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB53_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -11920,12 +12008,12 @@ define i8 @atomicrmw_min_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB53_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB53_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB53_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB53_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB53_1
; RV64I-NEXT: .LBB53_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -12181,6 +12269,7 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB54_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB54_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -12191,12 +12280,12 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB54_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB54_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB54_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB54_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB54_1
; RV32I-NEXT: .LBB54_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -12283,6 +12372,7 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB54_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB54_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -12293,12 +12383,12 @@ define i8 @atomicrmw_min_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB54_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB54_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB54_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB54_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB54_1
; RV64I-NEXT: .LBB54_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -12466,6 +12556,7 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB55_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB55_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -12475,12 +12566,12 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB55_4
; RV32I-NEXT: .LBB55_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB55_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB55_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB55_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB55_1
; RV32I-NEXT: .LBB55_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -12556,6 +12647,7 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB55_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB55_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -12565,12 +12657,12 @@ define i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB55_4
; RV64I-NEXT: .LBB55_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB55_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB55_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB55_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB55_1
; RV64I-NEXT: .LBB55_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -12718,6 +12810,7 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB56_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB56_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -12727,12 +12820,12 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB56_4
; RV32I-NEXT: .LBB56_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB56_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB56_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB56_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB56_1
; RV32I-NEXT: .LBB56_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -12832,6 +12925,7 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB56_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB56_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -12841,12 +12935,12 @@ define i8 @atomicrmw_umax_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB56_4
; RV64I-NEXT: .LBB56_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB56_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB56_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB56_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB56_1
; RV64I-NEXT: .LBB56_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -13066,6 +13160,7 @@ define i8 @atomicrmw_umax_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB57_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB57_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -13075,12 +13170,12 @@ define i8 @atomicrmw_umax_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB57_4
; RV32I-NEXT: .LBB57_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB57_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB57_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB57_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB57_1
; RV32I-NEXT: .LBB57_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -13180,6 +13275,7 @@ define i8 @atomicrmw_umax_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB57_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB57_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -13189,12 +13285,12 @@ define i8 @atomicrmw_umax_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB57_4
; RV64I-NEXT: .LBB57_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB57_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB57_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB57_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB57_1
; RV64I-NEXT: .LBB57_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -13414,6 +13510,7 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB58_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB58_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -13423,12 +13520,12 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB58_4
; RV32I-NEXT: .LBB58_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB58_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB58_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB58_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB58_1
; RV32I-NEXT: .LBB58_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -13528,6 +13625,7 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB58_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB58_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -13537,12 +13635,12 @@ define i8 @atomicrmw_umax_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB58_4
; RV64I-NEXT: .LBB58_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB58_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB58_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB58_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB58_1
; RV64I-NEXT: .LBB58_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -13762,6 +13860,7 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB59_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB59_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -13771,12 +13870,12 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB59_4
; RV32I-NEXT: .LBB59_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB59_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB59_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB59_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB59_1
; RV32I-NEXT: .LBB59_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -13852,6 +13951,7 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB59_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB59_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -13861,12 +13961,12 @@ define i8 @atomicrmw_umax_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB59_4
; RV64I-NEXT: .LBB59_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB59_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB59_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB59_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB59_1
; RV64I-NEXT: .LBB59_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -14014,6 +14114,7 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB60_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB60_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -14023,12 +14124,12 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB60_4
; RV32I-NEXT: .LBB60_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s2, a0, .LBB60_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s2, a2, .LBB60_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB60_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB60_1
; RV32I-NEXT: .LBB60_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -14104,6 +14205,7 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB60_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB60_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -14113,12 +14215,12 @@ define i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB60_4
; RV64I-NEXT: .LBB60_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB60_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s2, a2, .LBB60_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB60_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB60_1
; RV64I-NEXT: .LBB60_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -14266,6 +14368,7 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB61_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB61_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -14275,12 +14378,12 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB61_4
; RV32I-NEXT: .LBB61_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s2, a0, .LBB61_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s2, a2, .LBB61_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB61_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB61_1
; RV32I-NEXT: .LBB61_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -14380,6 +14483,7 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB61_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB61_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -14389,12 +14493,12 @@ define i8 @atomicrmw_umin_i8_acquire(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB61_4
; RV64I-NEXT: .LBB61_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB61_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s2, a2, .LBB61_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB61_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB61_1
; RV64I-NEXT: .LBB61_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -14614,6 +14718,7 @@ define i8 @atomicrmw_umin_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB62_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB62_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -14623,12 +14728,12 @@ define i8 @atomicrmw_umin_i8_release(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB62_4
; RV32I-NEXT: .LBB62_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s2, a0, .LBB62_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s2, a2, .LBB62_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB62_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB62_1
; RV32I-NEXT: .LBB62_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -14728,6 +14833,7 @@ define i8 @atomicrmw_umin_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB62_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB62_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -14737,12 +14843,12 @@ define i8 @atomicrmw_umin_i8_release(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB62_4
; RV64I-NEXT: .LBB62_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB62_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s2, a2, .LBB62_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB62_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB62_1
; RV64I-NEXT: .LBB62_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -14962,6 +15068,7 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB63_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB63_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -14971,12 +15078,12 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB63_4
; RV32I-NEXT: .LBB63_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s2, a0, .LBB63_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s2, a2, .LBB63_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB63_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB63_1
; RV32I-NEXT: .LBB63_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -15076,6 +15183,7 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB63_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB63_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -15085,12 +15193,12 @@ define i8 @atomicrmw_umin_i8_acq_rel(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB63_4
; RV64I-NEXT: .LBB63_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB63_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s2, a2, .LBB63_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB63_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB63_1
; RV64I-NEXT: .LBB63_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -15310,6 +15418,7 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB64_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB64_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -15319,12 +15428,12 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB64_4
; RV32I-NEXT: .LBB64_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s2, a0, .LBB64_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s2, a2, .LBB64_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB64_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB64_1
; RV32I-NEXT: .LBB64_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -15400,6 +15509,7 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB64_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB64_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -15409,12 +15519,12 @@ define i8 @atomicrmw_umin_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB64_4
; RV64I-NEXT: .LBB64_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB64_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s2, a2, .LBB64_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB64_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB64_1
; RV64I-NEXT: .LBB64_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -15549,6 +15659,8 @@ define i16 @atomicrmw_xchg_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -15601,6 +15713,8 @@ define i16 @atomicrmw_xchg_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -15719,6 +15833,8 @@ define i16 @atomicrmw_xchg_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -15792,6 +15908,8 @@ define i16 @atomicrmw_xchg_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -15973,6 +16091,8 @@ define i16 @atomicrmw_xchg_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -16046,6 +16166,8 @@ define i16 @atomicrmw_xchg_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -16227,6 +16349,8 @@ define i16 @atomicrmw_xchg_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -16300,6 +16424,8 @@ define i16 @atomicrmw_xchg_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -16481,6 +16607,8 @@ define i16 @atomicrmw_xchg_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -16533,6 +16661,8 @@ define i16 @atomicrmw_xchg_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -18302,6 +18432,8 @@ define i16 @atomicrmw_add_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_add_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -18354,6 +18486,8 @@ define i16 @atomicrmw_add_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_add_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -18472,6 +18606,8 @@ define i16 @atomicrmw_add_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_add_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -18545,6 +18681,8 @@ define i16 @atomicrmw_add_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_add_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -18726,6 +18864,8 @@ define i16 @atomicrmw_add_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_add_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -18799,6 +18939,8 @@ define i16 @atomicrmw_add_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_add_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -18980,6 +19122,8 @@ define i16 @atomicrmw_add_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_add_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -19053,6 +19197,8 @@ define i16 @atomicrmw_add_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_add_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -19234,6 +19380,8 @@ define i16 @atomicrmw_add_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_add_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -19286,6 +19434,8 @@ define i16 @atomicrmw_add_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_add_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -19404,6 +19554,8 @@ define i16 @atomicrmw_sub_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_sub_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -19456,6 +19608,8 @@ define i16 @atomicrmw_sub_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_sub_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -19578,6 +19732,8 @@ define i16 @atomicrmw_sub_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_sub_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -19651,6 +19807,8 @@ define i16 @atomicrmw_sub_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_sub_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -19836,6 +19994,8 @@ define i16 @atomicrmw_sub_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_sub_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -19909,6 +20069,8 @@ define i16 @atomicrmw_sub_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_sub_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -20094,6 +20256,8 @@ define i16 @atomicrmw_sub_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_sub_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -20167,6 +20331,8 @@ define i16 @atomicrmw_sub_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_sub_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -20352,6 +20518,8 @@ define i16 @atomicrmw_sub_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_sub_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -20404,6 +20572,8 @@ define i16 @atomicrmw_sub_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_sub_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -20526,6 +20696,8 @@ define i16 @atomicrmw_and_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -20571,6 +20743,8 @@ define i16 @atomicrmw_and_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -20670,6 +20844,8 @@ define i16 @atomicrmw_and_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -20730,6 +20906,8 @@ define i16 @atomicrmw_and_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -20874,6 +21052,8 @@ define i16 @atomicrmw_and_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -20934,6 +21114,8 @@ define i16 @atomicrmw_and_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -21078,6 +21260,8 @@ define i16 @atomicrmw_and_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -21138,6 +21322,8 @@ define i16 @atomicrmw_and_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -21282,6 +21468,8 @@ define i16 @atomicrmw_and_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -21342,6 +21530,8 @@ define i16 @atomicrmw_and_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -21486,6 +21676,8 @@ define i16 @atomicrmw_nand_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_nand_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -21540,6 +21732,8 @@ define i16 @atomicrmw_nand_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_nand_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -21790,6 +21984,8 @@ define i16 @atomicrmw_nand_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_nand_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -21866,6 +22062,8 @@ define i16 @atomicrmw_nand_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_nand_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -22182,6 +22380,8 @@ define i16 @atomicrmw_nand_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_nand_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -22258,6 +22458,8 @@ define i16 @atomicrmw_nand_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_nand_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -22574,6 +22776,8 @@ define i16 @atomicrmw_nand_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_nand_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -22650,6 +22854,8 @@ define i16 @atomicrmw_nand_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_nand_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -22966,6 +23172,8 @@ define i16 @atomicrmw_nand_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_nand_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -23020,6 +23228,8 @@ define i16 @atomicrmw_nand_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_nand_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -23274,6 +23484,8 @@ define i16 @atomicrmw_or_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -23311,6 +23523,8 @@ define i16 @atomicrmw_or_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -23394,6 +23608,8 @@ define i16 @atomicrmw_or_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -23442,6 +23658,8 @@ define i16 @atomicrmw_or_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -23558,6 +23776,8 @@ define i16 @atomicrmw_or_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -23606,6 +23826,8 @@ define i16 @atomicrmw_or_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -23722,6 +23944,8 @@ define i16 @atomicrmw_or_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -23770,6 +23994,8 @@ define i16 @atomicrmw_or_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -23886,6 +24112,8 @@ define i16 @atomicrmw_or_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -23934,6 +24162,8 @@ define i16 @atomicrmw_or_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -24050,6 +24280,8 @@ define i16 @atomicrmw_xor_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -24087,6 +24319,8 @@ define i16 @atomicrmw_xor_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -24170,6 +24404,8 @@ define i16 @atomicrmw_xor_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 2
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -24218,6 +24454,8 @@ define i16 @atomicrmw_xor_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -24334,6 +24572,8 @@ define i16 @atomicrmw_xor_i16_release(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 3
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -24382,6 +24622,8 @@ define i16 @atomicrmw_xor_i16_release(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -24498,6 +24740,8 @@ define i16 @atomicrmw_xor_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 4
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -24546,6 +24790,8 @@ define i16 @atomicrmw_xor_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -24662,6 +24908,8 @@ define i16 @atomicrmw_xor_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 5
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -24710,6 +24958,8 @@ define i16 @atomicrmw_xor_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -24839,7 +25089,9 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB110_2
; RV32I-NEXT: .LBB110_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB110_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -24850,12 +25102,12 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB110_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB110_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB110_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB110_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB110_1
; RV32I-NEXT: .LBB110_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -24945,7 +25197,9 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB110_2
; RV64I-NEXT: .LBB110_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB110_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -24956,12 +25210,12 @@ define i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB110_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB110_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB110_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB110_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB110_1
; RV64I-NEXT: .LBB110_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -25137,7 +25391,9 @@ define i16 @atomicrmw_max_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB111_2
; RV32I-NEXT: .LBB111_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB111_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -25148,12 +25404,12 @@ define i16 @atomicrmw_max_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB111_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB111_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB111_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB111_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB111_1
; RV32I-NEXT: .LBB111_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -25274,7 +25530,9 @@ define i16 @atomicrmw_max_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB111_2
; RV64I-NEXT: .LBB111_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB111_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -25285,12 +25543,12 @@ define i16 @atomicrmw_max_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB111_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB111_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB111_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB111_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB111_1
; RV64I-NEXT: .LBB111_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -25559,7 +25817,9 @@ define i16 @atomicrmw_max_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB112_2
; RV32I-NEXT: .LBB112_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB112_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -25570,12 +25830,12 @@ define i16 @atomicrmw_max_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB112_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB112_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB112_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB112_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB112_1
; RV32I-NEXT: .LBB112_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -25696,7 +25956,9 @@ define i16 @atomicrmw_max_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB112_2
; RV64I-NEXT: .LBB112_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB112_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -25707,12 +25969,12 @@ define i16 @atomicrmw_max_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB112_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB112_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB112_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB112_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB112_1
; RV64I-NEXT: .LBB112_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -25981,7 +26243,9 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB113_2
; RV32I-NEXT: .LBB113_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB113_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -25992,12 +26256,12 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB113_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB113_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB113_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB113_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB113_1
; RV32I-NEXT: .LBB113_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -26118,7 +26382,9 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB113_2
; RV64I-NEXT: .LBB113_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB113_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -26129,12 +26395,12 @@ define i16 @atomicrmw_max_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB113_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB113_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB113_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB113_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB113_1
; RV64I-NEXT: .LBB113_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -26403,7 +26669,9 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB114_2
; RV32I-NEXT: .LBB114_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB114_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -26414,12 +26682,12 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB114_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB114_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB114_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB114_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB114_1
; RV32I-NEXT: .LBB114_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -26509,7 +26777,9 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB114_2
; RV64I-NEXT: .LBB114_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB114_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -26520,12 +26790,12 @@ define i16 @atomicrmw_max_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB114_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB114_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB114_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB114_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB114_1
; RV64I-NEXT: .LBB114_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -26701,7 +26971,9 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB115_2
; RV32I-NEXT: .LBB115_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB115_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -26712,12 +26984,12 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB115_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB115_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB115_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB115_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB115_1
; RV32I-NEXT: .LBB115_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -26807,7 +27079,9 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB115_2
; RV64I-NEXT: .LBB115_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB115_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -26818,12 +27092,12 @@ define i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB115_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB115_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB115_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB115_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB115_1
; RV64I-NEXT: .LBB115_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -26999,7 +27273,9 @@ define i16 @atomicrmw_min_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB116_2
; RV32I-NEXT: .LBB116_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB116_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -27010,12 +27286,12 @@ define i16 @atomicrmw_min_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB116_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB116_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB116_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB116_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB116_1
; RV32I-NEXT: .LBB116_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -27136,7 +27412,9 @@ define i16 @atomicrmw_min_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB116_2
; RV64I-NEXT: .LBB116_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB116_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -27147,12 +27425,12 @@ define i16 @atomicrmw_min_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB116_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB116_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB116_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB116_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB116_1
; RV64I-NEXT: .LBB116_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -27421,7 +27699,9 @@ define i16 @atomicrmw_min_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB117_2
; RV32I-NEXT: .LBB117_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB117_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -27432,12 +27712,12 @@ define i16 @atomicrmw_min_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB117_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB117_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB117_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB117_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB117_1
; RV32I-NEXT: .LBB117_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -27558,7 +27838,9 @@ define i16 @atomicrmw_min_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB117_2
; RV64I-NEXT: .LBB117_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB117_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -27569,12 +27851,12 @@ define i16 @atomicrmw_min_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB117_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB117_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB117_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB117_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB117_1
; RV64I-NEXT: .LBB117_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -27843,7 +28125,9 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB118_2
; RV32I-NEXT: .LBB118_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB118_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -27854,12 +28138,12 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB118_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB118_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB118_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB118_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB118_1
; RV32I-NEXT: .LBB118_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -27980,7 +28264,9 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB118_2
; RV64I-NEXT: .LBB118_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB118_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -27991,12 +28277,12 @@ define i16 @atomicrmw_min_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB118_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB118_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB118_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB118_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB118_1
; RV64I-NEXT: .LBB118_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -28265,7 +28551,9 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB119_2
; RV32I-NEXT: .LBB119_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB119_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -28276,12 +28564,12 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB119_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB119_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB119_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB119_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB119_1
; RV32I-NEXT: .LBB119_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -28371,7 +28659,9 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB119_2
; RV64I-NEXT: .LBB119_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB119_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -28382,12 +28672,12 @@ define i16 @atomicrmw_min_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB119_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB119_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB119_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB119_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB119_1
; RV64I-NEXT: .LBB119_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -28566,6 +28856,7 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB120_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB120_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -28575,12 +28866,12 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB120_4
; RV32I-NEXT: .LBB120_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB120_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB120_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB120_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB120_1
; RV32I-NEXT: .LBB120_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -28662,6 +28953,7 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB120_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB120_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -28671,12 +28963,12 @@ define i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB120_4
; RV64I-NEXT: .LBB120_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB120_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB120_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB120_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB120_1
; RV64I-NEXT: .LBB120_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -28832,6 +29124,7 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB121_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB121_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -28841,12 +29134,12 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB121_4
; RV32I-NEXT: .LBB121_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB121_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB121_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB121_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB121_1
; RV32I-NEXT: .LBB121_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -28953,6 +29246,7 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB121_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB121_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -28962,12 +29256,12 @@ define i16 @atomicrmw_umax_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB121_4
; RV64I-NEXT: .LBB121_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB121_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB121_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB121_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB121_1
; RV64I-NEXT: .LBB121_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -29198,6 +29492,7 @@ define i16 @atomicrmw_umax_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB122_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB122_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -29207,12 +29502,12 @@ define i16 @atomicrmw_umax_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB122_4
; RV32I-NEXT: .LBB122_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB122_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB122_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB122_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB122_1
; RV32I-NEXT: .LBB122_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -29319,6 +29614,7 @@ define i16 @atomicrmw_umax_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB122_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB122_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -29328,12 +29624,12 @@ define i16 @atomicrmw_umax_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB122_4
; RV64I-NEXT: .LBB122_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB122_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB122_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB122_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB122_1
; RV64I-NEXT: .LBB122_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -29564,6 +29860,7 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB123_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB123_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -29573,12 +29870,12 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB123_4
; RV32I-NEXT: .LBB123_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB123_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB123_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB123_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB123_1
; RV32I-NEXT: .LBB123_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -29685,6 +29982,7 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB123_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB123_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -29694,12 +29992,12 @@ define i16 @atomicrmw_umax_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB123_4
; RV64I-NEXT: .LBB123_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB123_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB123_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB123_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB123_1
; RV64I-NEXT: .LBB123_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -29930,6 +30228,7 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB124_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB124_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -29939,12 +30238,12 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB124_4
; RV32I-NEXT: .LBB124_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB124_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB124_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB124_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB124_1
; RV32I-NEXT: .LBB124_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -30026,6 +30325,7 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB124_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB124_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -30035,12 +30335,12 @@ define i16 @atomicrmw_umax_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB124_4
; RV64I-NEXT: .LBB124_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB124_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB124_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB124_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB124_1
; RV64I-NEXT: .LBB124_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -30196,6 +30496,7 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB125_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB125_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -30205,12 +30506,12 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB125_4
; RV32I-NEXT: .LBB125_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s3, a0, .LBB125_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s3, a2, .LBB125_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB125_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB125_1
; RV32I-NEXT: .LBB125_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -30292,6 +30593,7 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB125_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB125_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -30301,12 +30603,12 @@ define i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB125_4
; RV64I-NEXT: .LBB125_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s3, a0, .LBB125_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s3, a2, .LBB125_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB125_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB125_1
; RV64I-NEXT: .LBB125_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -30462,6 +30764,7 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB126_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB126_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 2
; RV32I-NEXT: li a4, 2
@@ -30471,12 +30774,12 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB126_4
; RV32I-NEXT: .LBB126_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s3, a0, .LBB126_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s3, a2, .LBB126_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB126_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB126_1
; RV32I-NEXT: .LBB126_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -30583,6 +30886,7 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB126_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB126_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -30592,12 +30896,12 @@ define i16 @atomicrmw_umin_i16_acquire(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB126_4
; RV64I-NEXT: .LBB126_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s3, a0, .LBB126_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s3, a2, .LBB126_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB126_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB126_1
; RV64I-NEXT: .LBB126_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -30828,6 +31132,7 @@ define i16 @atomicrmw_umin_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB127_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB127_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 3
; RV32I-NEXT: mv a0, s1
@@ -30837,12 +31142,12 @@ define i16 @atomicrmw_umin_i16_release(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB127_4
; RV32I-NEXT: .LBB127_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s3, a0, .LBB127_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s3, a2, .LBB127_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB127_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB127_1
; RV32I-NEXT: .LBB127_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -30949,6 +31254,7 @@ define i16 @atomicrmw_umin_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB127_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB127_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -30958,12 +31264,12 @@ define i16 @atomicrmw_umin_i16_release(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB127_4
; RV64I-NEXT: .LBB127_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s3, a0, .LBB127_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s3, a2, .LBB127_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB127_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB127_1
; RV64I-NEXT: .LBB127_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -31194,6 +31500,7 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB128_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB128_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 4
; RV32I-NEXT: li a4, 2
@@ -31203,12 +31510,12 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB128_4
; RV32I-NEXT: .LBB128_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s3, a0, .LBB128_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s3, a2, .LBB128_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB128_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB128_1
; RV32I-NEXT: .LBB128_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -31315,6 +31622,7 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB128_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB128_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -31324,12 +31632,12 @@ define i16 @atomicrmw_umin_i16_acq_rel(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB128_4
; RV64I-NEXT: .LBB128_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s3, a0, .LBB128_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s3, a2, .LBB128_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB128_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB128_1
; RV64I-NEXT: .LBB128_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -31560,6 +31868,7 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB129_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB129_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -31569,12 +31878,12 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB129_4
; RV32I-NEXT: .LBB129_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s3, a0, .LBB129_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s3, a2, .LBB129_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB129_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB129_1
; RV32I-NEXT: .LBB129_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -31656,6 +31965,7 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB129_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB129_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -31665,12 +31975,12 @@ define i16 @atomicrmw_umin_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB129_4
; RV64I-NEXT: .LBB129_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s3, a0, .LBB129_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s3, a2, .LBB129_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB129_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB129_1
; RV64I-NEXT: .LBB129_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -31836,6 +32146,7 @@ define i32 @atomicrmw_xchg_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -31897,6 +32208,7 @@ define i32 @atomicrmw_xchg_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -31963,6 +32275,7 @@ define i32 @atomicrmw_xchg_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32029,6 +32342,7 @@ define i32 @atomicrmw_xchg_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32095,6 +32409,7 @@ define i32 @atomicrmw_xchg_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32156,6 +32471,7 @@ define i32 @atomicrmw_add_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_add_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32217,6 +32533,7 @@ define i32 @atomicrmw_add_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_add_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32283,6 +32600,7 @@ define i32 @atomicrmw_add_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_add_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32349,6 +32667,7 @@ define i32 @atomicrmw_add_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_add_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32415,6 +32734,7 @@ define i32 @atomicrmw_add_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_add_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32477,6 +32797,7 @@ define i32 @atomicrmw_sub_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_sub_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32541,6 +32862,7 @@ define i32 @atomicrmw_sub_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_sub_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32611,6 +32933,7 @@ define i32 @atomicrmw_sub_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_sub_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32681,6 +33004,7 @@ define i32 @atomicrmw_sub_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_sub_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32751,6 +33075,7 @@ define i32 @atomicrmw_sub_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_sub_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32814,6 +33139,7 @@ define i32 @atomicrmw_and_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32875,6 +33201,7 @@ define i32 @atomicrmw_and_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -32941,6 +33268,7 @@ define i32 @atomicrmw_and_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -33007,6 +33335,7 @@ define i32 @atomicrmw_and_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -33073,6 +33402,7 @@ define i32 @atomicrmw_and_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -33142,6 +33472,7 @@ define i32 @atomicrmw_nand_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_nand_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -33358,6 +33689,7 @@ define i32 @atomicrmw_nand_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_nand_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -33614,6 +33946,7 @@ define i32 @atomicrmw_nand_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_nand_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -33870,6 +34203,7 @@ define i32 @atomicrmw_nand_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_nand_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34114,6 +34448,7 @@ define i32 @atomicrmw_nand_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_nand_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34346,6 +34681,7 @@ define i32 @atomicrmw_or_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34407,6 +34743,7 @@ define i32 @atomicrmw_or_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34473,6 +34810,7 @@ define i32 @atomicrmw_or_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34539,6 +34877,7 @@ define i32 @atomicrmw_or_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34605,6 +34944,7 @@ define i32 @atomicrmw_or_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34666,6 +35006,7 @@ define i32 @atomicrmw_xor_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34727,6 +35068,7 @@ define i32 @atomicrmw_xor_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 2
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34793,6 +35135,7 @@ define i32 @atomicrmw_xor_i32_release(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 3
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34859,6 +35202,7 @@ define i32 @atomicrmw_xor_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 4
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -34925,6 +35269,7 @@ define i32 @atomicrmw_xor_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 5
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -35031,6 +35376,7 @@ define i32 @atomicrmw_max_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB165_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB165_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -35040,9 +35386,8 @@ define i32 @atomicrmw_max_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB165_4
; RV64I-NEXT: .LBB165_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB165_1
+; RV64I-NEXT: blt s2, a1, .LBB165_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB165_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -35161,6 +35506,7 @@ define i32 @atomicrmw_max_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB166_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB166_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -35170,9 +35516,8 @@ define i32 @atomicrmw_max_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB166_4
; RV64I-NEXT: .LBB166_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB166_1
+; RV64I-NEXT: blt s2, a1, .LBB166_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB166_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -35296,6 +35641,7 @@ define i32 @atomicrmw_max_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB167_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB167_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -35305,9 +35651,8 @@ define i32 @atomicrmw_max_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB167_4
; RV64I-NEXT: .LBB167_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB167_1
+; RV64I-NEXT: blt s2, a1, .LBB167_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB167_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -35431,6 +35776,7 @@ define i32 @atomicrmw_max_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB168_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB168_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -35440,9 +35786,8 @@ define i32 @atomicrmw_max_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB168_4
; RV64I-NEXT: .LBB168_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB168_1
+; RV64I-NEXT: blt s2, a1, .LBB168_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB168_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -35566,6 +35911,7 @@ define i32 @atomicrmw_max_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB169_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB169_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -35575,9 +35921,8 @@ define i32 @atomicrmw_max_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB169_4
; RV64I-NEXT: .LBB169_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB169_1
+; RV64I-NEXT: blt s2, a1, .LBB169_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB169_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -35696,6 +36041,7 @@ define i32 @atomicrmw_min_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB170_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB170_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -35705,9 +36051,8 @@ define i32 @atomicrmw_min_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB170_4
; RV64I-NEXT: .LBB170_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB170_1
+; RV64I-NEXT: bge s2, a1, .LBB170_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB170_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -35826,6 +36171,7 @@ define i32 @atomicrmw_min_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB171_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB171_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -35835,9 +36181,8 @@ define i32 @atomicrmw_min_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB171_4
; RV64I-NEXT: .LBB171_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB171_1
+; RV64I-NEXT: bge s2, a1, .LBB171_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB171_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -35961,6 +36306,7 @@ define i32 @atomicrmw_min_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB172_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB172_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -35970,9 +36316,8 @@ define i32 @atomicrmw_min_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB172_4
; RV64I-NEXT: .LBB172_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB172_1
+; RV64I-NEXT: bge s2, a1, .LBB172_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB172_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -36096,6 +36441,7 @@ define i32 @atomicrmw_min_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB173_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB173_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -36105,9 +36451,8 @@ define i32 @atomicrmw_min_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB173_4
; RV64I-NEXT: .LBB173_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB173_1
+; RV64I-NEXT: bge s2, a1, .LBB173_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB173_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -36231,6 +36576,7 @@ define i32 @atomicrmw_min_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB174_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB174_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -36240,9 +36586,8 @@ define i32 @atomicrmw_min_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB174_4
; RV64I-NEXT: .LBB174_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB174_1
+; RV64I-NEXT: bge s2, a1, .LBB174_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB174_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -36361,6 +36706,7 @@ define i32 @atomicrmw_umax_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB175_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB175_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -36370,9 +36716,8 @@ define i32 @atomicrmw_umax_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB175_4
; RV64I-NEXT: .LBB175_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB175_1
+; RV64I-NEXT: bltu s2, a1, .LBB175_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB175_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -36491,6 +36836,7 @@ define i32 @atomicrmw_umax_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB176_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB176_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -36500,9 +36846,8 @@ define i32 @atomicrmw_umax_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB176_4
; RV64I-NEXT: .LBB176_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB176_1
+; RV64I-NEXT: bltu s2, a1, .LBB176_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB176_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -36626,6 +36971,7 @@ define i32 @atomicrmw_umax_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB177_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB177_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -36635,9 +36981,8 @@ define i32 @atomicrmw_umax_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB177_4
; RV64I-NEXT: .LBB177_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB177_1
+; RV64I-NEXT: bltu s2, a1, .LBB177_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB177_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -36761,6 +37106,7 @@ define i32 @atomicrmw_umax_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB178_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB178_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -36770,9 +37116,8 @@ define i32 @atomicrmw_umax_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB178_4
; RV64I-NEXT: .LBB178_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB178_1
+; RV64I-NEXT: bltu s2, a1, .LBB178_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB178_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -36896,6 +37241,7 @@ define i32 @atomicrmw_umax_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB179_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB179_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -36905,9 +37251,8 @@ define i32 @atomicrmw_umax_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB179_4
; RV64I-NEXT: .LBB179_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB179_1
+; RV64I-NEXT: bltu s2, a1, .LBB179_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB179_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -37026,6 +37371,7 @@ define i32 @atomicrmw_umin_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB180_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB180_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -37035,9 +37381,8 @@ define i32 @atomicrmw_umin_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB180_4
; RV64I-NEXT: .LBB180_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB180_1
+; RV64I-NEXT: bgeu s2, a1, .LBB180_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB180_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -37156,6 +37501,7 @@ define i32 @atomicrmw_umin_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB181_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB181_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 2
; RV64I-NEXT: li a4, 2
@@ -37165,9 +37511,8 @@ define i32 @atomicrmw_umin_i32_acquire(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB181_4
; RV64I-NEXT: .LBB181_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB181_1
+; RV64I-NEXT: bgeu s2, a1, .LBB181_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB181_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -37291,6 +37636,7 @@ define i32 @atomicrmw_umin_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB182_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB182_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 3
; RV64I-NEXT: mv a0, s1
@@ -37300,9 +37646,8 @@ define i32 @atomicrmw_umin_i32_release(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB182_4
; RV64I-NEXT: .LBB182_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB182_1
+; RV64I-NEXT: bgeu s2, a1, .LBB182_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB182_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -37426,6 +37771,7 @@ define i32 @atomicrmw_umin_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB183_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB183_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 4
; RV64I-NEXT: li a4, 2
@@ -37435,9 +37781,8 @@ define i32 @atomicrmw_umin_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB183_4
; RV64I-NEXT: .LBB183_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB183_1
+; RV64I-NEXT: bgeu s2, a1, .LBB183_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB183_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -37561,6 +37906,7 @@ define i32 @atomicrmw_umin_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB184_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB184_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -37570,9 +37916,8 @@ define i32 @atomicrmw_umin_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB184_4
; RV64I-NEXT: .LBB184_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB184_1
+; RV64I-NEXT: bgeu s2, a1, .LBB184_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB184_2 Depth=1
; RV64I-NEXT: mv a2, s0
diff --git a/llvm/test/CodeGen/RISCV/atomic-signext.ll b/llvm/test/CodeGen/RISCV/atomic-signext.ll
index 5a042a4b26c088..7d8a2cfc0fc58a 100644
--- a/llvm/test/CodeGen/RISCV/atomic-signext.ll
+++ b/llvm/test/CodeGen/RISCV/atomic-signext.ll
@@ -139,7 +139,6 @@ define signext i32 @atomic_load_i32_unordered(ptr %a) nounwind {
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: li a1, 0
; RV64I-NEXT: call __atomic_load_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -162,6 +161,7 @@ define signext i8 @atomicrmw_xchg_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_exchange_1
; RV32I-NEXT: slli a0, a0, 24
@@ -218,6 +218,7 @@ define signext i8 @atomicrmw_xchg_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_1
; RV64I-NEXT: slli a0, a0, 56
@@ -278,6 +279,7 @@ define signext i8 @atomicrmw_add_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_add_1
; RV32I-NEXT: slli a0, a0, 24
@@ -334,6 +336,7 @@ define signext i8 @atomicrmw_add_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_add_1
; RV64I-NEXT: slli a0, a0, 56
@@ -394,6 +397,7 @@ define signext i8 @atomicrmw_sub_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_sub_1
; RV32I-NEXT: slli a0, a0, 24
@@ -450,6 +454,7 @@ define signext i8 @atomicrmw_sub_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_sub_1
; RV64I-NEXT: slli a0, a0, 56
@@ -510,6 +515,7 @@ define signext i8 @atomicrmw_and_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_and_1
; RV32I-NEXT: slli a0, a0, 24
@@ -559,6 +565,7 @@ define signext i8 @atomicrmw_and_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_1
; RV64I-NEXT: slli a0, a0, 56
@@ -612,6 +619,7 @@ define signext i8 @atomicrmw_nand_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_nand_1
; RV32I-NEXT: slli a0, a0, 24
@@ -670,6 +678,7 @@ define signext i8 @atomicrmw_nand_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_nand_1
; RV64I-NEXT: slli a0, a0, 56
@@ -732,6 +741,7 @@ define signext i8 @atomicrmw_or_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_or_1
; RV32I-NEXT: slli a0, a0, 24
@@ -773,6 +783,7 @@ define signext i8 @atomicrmw_or_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_1
; RV64I-NEXT: slli a0, a0, 56
@@ -818,6 +829,7 @@ define signext i8 @atomicrmw_xor_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: zext.b a1, a1
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_xor_1
; RV32I-NEXT: slli a0, a0, 24
@@ -859,6 +871,7 @@ define signext i8 @atomicrmw_xor_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: zext.b a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_1
; RV64I-NEXT: slli a0, a0, 56
@@ -918,6 +931,7 @@ define signext i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB10_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB10_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -928,12 +942,12 @@ define signext i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB10_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB10_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB10_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB10_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB10_1
; RV32I-NEXT: .LBB10_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 24
@@ -1025,6 +1039,7 @@ define signext i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB10_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB10_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -1035,12 +1050,12 @@ define signext i8 @atomicrmw_max_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB10_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB10_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB10_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB10_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB10_1
; RV64I-NEXT: .LBB10_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 56
@@ -1136,6 +1151,7 @@ define signext i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB11_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB11_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -1146,12 +1162,12 @@ define signext i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB11_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 24
-; RV32I-NEXT: srai a0, a0, 24
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB11_1
+; RV32I-NEXT: srai a2, a0, 24
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB11_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB11_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB11_1
; RV32I-NEXT: .LBB11_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 24
@@ -1243,6 +1259,7 @@ define signext i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB11_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB11_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -1253,12 +1270,12 @@ define signext i8 @atomicrmw_min_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB11_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 56
-; RV64I-NEXT: srai a0, a0, 56
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB11_1
+; RV64I-NEXT: srai a2, a0, 56
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB11_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB11_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB11_1
; RV64I-NEXT: .LBB11_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 56
@@ -1353,6 +1370,7 @@ define signext i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB12_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB12_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -1362,12 +1380,12 @@ define signext i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB12_4
; RV32I-NEXT: .LBB12_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s2, a0, .LBB12_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s2, a2, .LBB12_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB12_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB12_1
; RV32I-NEXT: .LBB12_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 24
@@ -1448,6 +1466,7 @@ define signext i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB12_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB12_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -1457,12 +1476,12 @@ define signext i8 @atomicrmw_umax_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB12_4
; RV64I-NEXT: .LBB12_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB12_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s2, a2, .LBB12_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB12_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB12_1
; RV64I-NEXT: .LBB12_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 56
@@ -1547,6 +1566,7 @@ define signext i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: .LBB13_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB13_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -1556,12 +1576,12 @@ define signext i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB13_4
; RV32I-NEXT: .LBB13_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: zext.b a0, a1
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s2, a0, .LBB13_1
+; RV32I-NEXT: zext.b a2, a1
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s2, a2, .LBB13_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB13_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB13_1
; RV32I-NEXT: .LBB13_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 24
@@ -1642,6 +1662,7 @@ define signext i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: .LBB13_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB13_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -1651,12 +1672,12 @@ define signext i8 @atomicrmw_umin_i8_monotonic(ptr %a, i8 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB13_4
; RV64I-NEXT: .LBB13_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: zext.b a0, a1
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB13_1
+; RV64I-NEXT: zext.b a2, a1
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s2, a2, .LBB13_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB13_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB13_1
; RV64I-NEXT: .LBB13_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 56
@@ -1728,6 +1749,8 @@ define signext i16 @atomicrmw_xchg_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_exchange_2
; RV32I-NEXT: slli a0, a0, 16
@@ -1786,6 +1809,8 @@ define signext i16 @atomicrmw_xchg_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_2
; RV64I-NEXT: slli a0, a0, 48
@@ -1848,6 +1873,8 @@ define signext i16 @atomicrmw_add_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_add_2
; RV32I-NEXT: slli a0, a0, 16
@@ -1906,6 +1933,8 @@ define signext i16 @atomicrmw_add_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_add_2
; RV64I-NEXT: slli a0, a0, 48
@@ -1968,6 +1997,8 @@ define signext i16 @atomicrmw_sub_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_sub_2
; RV32I-NEXT: slli a0, a0, 16
@@ -2026,6 +2057,8 @@ define signext i16 @atomicrmw_sub_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_sub_2
; RV64I-NEXT: slli a0, a0, 48
@@ -2088,6 +2121,8 @@ define signext i16 @atomicrmw_and_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_and_2
; RV32I-NEXT: slli a0, a0, 16
@@ -2139,6 +2174,8 @@ define signext i16 @atomicrmw_and_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_2
; RV64I-NEXT: slli a0, a0, 48
@@ -2194,6 +2231,8 @@ define signext i16 @atomicrmw_nand_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_nand_2
; RV32I-NEXT: slli a0, a0, 16
@@ -2254,6 +2293,8 @@ define signext i16 @atomicrmw_nand_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_nand_2
; RV64I-NEXT: slli a0, a0, 48
@@ -2318,6 +2359,8 @@ define signext i16 @atomicrmw_or_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_or_2
; RV32I-NEXT: slli a0, a0, 16
@@ -2361,6 +2404,8 @@ define signext i16 @atomicrmw_or_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_2
; RV64I-NEXT: slli a0, a0, 48
@@ -2408,6 +2453,8 @@ define signext i16 @atomicrmw_xor_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I: # %bb.0:
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT: slli a1, a1, 16
+; RV32I-NEXT: srli a1, a1, 16
; RV32I-NEXT: li a2, 0
; RV32I-NEXT: call __atomic_fetch_xor_2
; RV32I-NEXT: slli a0, a0, 16
@@ -2451,6 +2498,8 @@ define signext i16 @atomicrmw_xor_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: slli a1, a1, 48
+; RV64I-NEXT: srli a1, a1, 48
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_2
; RV64I-NEXT: slli a0, a0, 48
@@ -2511,7 +2560,9 @@ define signext i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB21_2
; RV32I-NEXT: .LBB21_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB21_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -2522,12 +2573,12 @@ define signext i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB21_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: blt s2, a0, .LBB21_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: blt s2, a2, .LBB21_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB21_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB21_1
; RV32I-NEXT: .LBB21_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 16
@@ -2622,7 +2673,9 @@ define signext i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB21_2
; RV64I-NEXT: .LBB21_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB21_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -2633,12 +2686,12 @@ define signext i16 @atomicrmw_max_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB21_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB21_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: blt s2, a2, .LBB21_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB21_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB21_1
; RV64I-NEXT: .LBB21_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 48
@@ -2737,7 +2790,9 @@ define signext i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: j .LBB22_2
; RV32I-NEXT: .LBB22_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB22_2 Depth=1
+; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: srli a2, a0, 16
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -2748,12 +2803,12 @@ define signext i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB22_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: slli a0, a1, 16
-; RV32I-NEXT: srai a0, a0, 16
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bge s2, a0, .LBB22_1
+; RV32I-NEXT: srai a2, a0, 16
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bge s2, a2, .LBB22_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB22_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB22_1
; RV32I-NEXT: .LBB22_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 16
@@ -2848,7 +2903,9 @@ define signext i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: j .LBB22_2
; RV64I-NEXT: .LBB22_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB22_2 Depth=1
+; RV64I-NEXT: slli a0, a0, 48
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: srli a2, a0, 48
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -2859,12 +2916,12 @@ define signext i16 @atomicrmw_min_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB22_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: slli a0, a1, 48
-; RV64I-NEXT: srai a0, a0, 48
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB22_1
+; RV64I-NEXT: srai a2, a0, 48
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bge s2, a2, .LBB22_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB22_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB22_1
; RV64I-NEXT: .LBB22_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 48
@@ -2966,6 +3023,7 @@ define signext i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB23_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB23_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -2975,12 +3033,12 @@ define signext i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB23_4
; RV32I-NEXT: .LBB23_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bltu s3, a0, .LBB23_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bltu s3, a2, .LBB23_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB23_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB23_1
; RV32I-NEXT: .LBB23_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 16
@@ -3067,6 +3125,7 @@ define signext i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB23_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB23_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -3076,12 +3135,12 @@ define signext i16 @atomicrmw_umax_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB23_4
; RV64I-NEXT: .LBB23_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s3, a0, .LBB23_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bltu s3, a2, .LBB23_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB23_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB23_1
; RV64I-NEXT: .LBB23_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 48
@@ -3172,6 +3231,7 @@ define signext i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: .LBB24_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB24_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: mv a0, s1
; RV32I-NEXT: li a3, 0
@@ -3181,12 +3241,12 @@ define signext i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV32I-NEXT: bnez a0, .LBB24_4
; RV32I-NEXT: .LBB24_2: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: mv a2, a1
-; RV32I-NEXT: bgeu s3, a0, .LBB24_1
+; RV32I-NEXT: and a2, a1, s2
+; RV32I-NEXT: mv a0, a1
+; RV32I-NEXT: bgeu s3, a2, .LBB24_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB24_2 Depth=1
-; RV32I-NEXT: mv a2, s0
+; RV32I-NEXT: mv a0, s0
; RV32I-NEXT: j .LBB24_1
; RV32I-NEXT: .LBB24_4: # %atomicrmw.end
; RV32I-NEXT: slli a0, a1, 16
@@ -3273,6 +3333,7 @@ define signext i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: .LBB24_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB24_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -3282,12 +3343,12 @@ define signext i16 @atomicrmw_umin_i16_monotonic(ptr %a, i16 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB24_4
; RV64I-NEXT: .LBB24_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s3, a0, .LBB24_1
+; RV64I-NEXT: and a2, a1, s2
+; RV64I-NEXT: mv a0, a1
+; RV64I-NEXT: bgeu s3, a2, .LBB24_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB24_2 Depth=1
-; RV64I-NEXT: mv a2, s0
+; RV64I-NEXT: mv a0, s0
; RV64I-NEXT: j .LBB24_1
; RV64I-NEXT: .LBB24_4: # %atomicrmw.end
; RV64I-NEXT: slli a0, a1, 48
@@ -3388,9 +3449,9 @@ define signext i32 @atomicrmw_xchg_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_exchange_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -3445,9 +3506,9 @@ define signext i32 @atomicrmw_add_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_add_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -3503,9 +3564,9 @@ define signext i32 @atomicrmw_sub_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_sub_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -3561,9 +3622,9 @@ define signext i32 @atomicrmw_and_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_and_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -3640,9 +3701,9 @@ define signext i32 @atomicrmw_nand_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_nand_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -3719,9 +3780,9 @@ define signext i32 @atomicrmw_or_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_or_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -3776,9 +3837,9 @@ define signext i32 @atomicrmw_xor_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I: # %bb.0:
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT: sext.w a1, a1
; RV64I-NEXT: li a2, 0
; RV64I-NEXT: call __atomic_fetch_xor_4
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
; RV64I-NEXT: ret
@@ -3878,6 +3939,7 @@ define signext i32 @atomicrmw_max_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB32_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB32_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -3887,9 +3949,8 @@ define signext i32 @atomicrmw_max_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB32_4
; RV64I-NEXT: .LBB32_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt s2, a0, .LBB32_1
+; RV64I-NEXT: blt s2, a1, .LBB32_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB32_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -4003,6 +4064,7 @@ define signext i32 @atomicrmw_min_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB33_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB33_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -4012,9 +4074,8 @@ define signext i32 @atomicrmw_min_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB33_4
; RV64I-NEXT: .LBB33_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bge s2, a0, .LBB33_1
+; RV64I-NEXT: bge s2, a1, .LBB33_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB33_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -4128,6 +4189,7 @@ define signext i32 @atomicrmw_umax_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB34_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB34_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -4137,9 +4199,8 @@ define signext i32 @atomicrmw_umax_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB34_4
; RV64I-NEXT: .LBB34_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu s2, a0, .LBB34_1
+; RV64I-NEXT: bltu s2, a1, .LBB34_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB34_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -4253,6 +4314,7 @@ define signext i32 @atomicrmw_umin_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: .LBB35_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB35_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: mv a0, s1
; RV64I-NEXT: li a3, 0
@@ -4262,9 +4324,8 @@ define signext i32 @atomicrmw_umin_i32_monotonic(ptr %a, i32 %b) nounwind {
; RV64I-NEXT: bnez a0, .LBB35_4
; RV64I-NEXT: .LBB35_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bgeu s2, a0, .LBB35_1
+; RV64I-NEXT: bgeu s2, a1, .LBB35_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB35_2 Depth=1
; RV64I-NEXT: mv a2, s0
@@ -5647,6 +5708,7 @@ define signext i8 @cmpxchg_i8_monotonic_monotonic_val0(ptr %ptr, i8 signext %cmp
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
@@ -5713,6 +5775,7 @@ define signext i8 @cmpxchg_i8_monotonic_monotonic_val0(ptr %ptr, i8 signext %cmp
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -5784,6 +5847,7 @@ define i1 @cmpxchg_i8_monotonic_monotonic_val1(ptr %ptr, i8 signext %cmp, i8 sig
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sb a1, 11(sp)
+; RV32I-NEXT: zext.b a2, a2
; RV32I-NEXT: addi a1, sp, 11
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
@@ -5849,6 +5913,7 @@ define i1 @cmpxchg_i8_monotonic_monotonic_val1(ptr %ptr, i8 signext %cmp, i8 sig
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a2
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
@@ -5919,7 +5984,9 @@ define signext i16 @cmpxchg_i16_monotonic_monotonic_val0(ptr %ptr, i16 signext %
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -5987,7 +6054,9 @@ define signext i16 @cmpxchg_i16_monotonic_monotonic_val0(ptr %ptr, i16 signext %
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -6060,7 +6129,9 @@ define i1 @cmpxchg_i16_monotonic_monotonic_val1(ptr %ptr, i16 signext %cmp, i16
; RV32I-NEXT: addi sp, sp, -16
; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: slli a2, a2, 16
; RV32I-NEXT: addi a1, sp, 10
+; RV32I-NEXT: srli a2, a2, 16
; RV32I-NEXT: li a3, 0
; RV32I-NEXT: li a4, 0
; RV32I-NEXT: call __atomic_compare_exchange_2
@@ -6127,7 +6198,9 @@ define i1 @cmpxchg_i16_monotonic_monotonic_val1(ptr %ptr, i16 signext %cmp, i16
; RV64I-NEXT: addi sp, sp, -16
; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: slli a2, a2, 48
; RV64I-NEXT: addi a1, sp, 6
+; RV64I-NEXT: srli a2, a2, 48
; RV64I-NEXT: li a3, 0
; RV64I-NEXT: li a4, 0
; RV64I-NEXT: call __atomic_compare_exchange_2
@@ -6454,13 +6527,12 @@ define signext i32 @atomicrmw_xchg_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: call __atomic_exchange_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ret
; RV64I-NEXT: .LBB53_2: # %else
; RV64I-NEXT: lw a1, 0(a0)
; RV64I-NEXT: li a2, 1
; RV64I-NEXT: sw a2, 0(a0)
-; RV64I-NEXT: sext.w a0, a1
+; RV64I-NEXT: mv a0, a1
; RV64I-NEXT: ret
;
; RV64IA-LABEL: atomicrmw_xchg_i32_monotonic_crossbb:
@@ -6584,13 +6656,12 @@ define signext i32 @atomicrmw_add_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: call __atomic_fetch_add_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ret
; RV64I-NEXT: .LBB54_2: # %else
; RV64I-NEXT: lw a1, 0(a0)
; RV64I-NEXT: addi a2, a1, 1
; RV64I-NEXT: sw a2, 0(a0)
-; RV64I-NEXT: sext.w a0, a1
+; RV64I-NEXT: mv a0, a1
; RV64I-NEXT: ret
;
; RV64IA-LABEL: atomicrmw_add_i32_monotonic_crossbb:
@@ -6715,13 +6786,12 @@ define signext i32 @atomicrmw_sub_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: call __atomic_fetch_sub_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ret
; RV64I-NEXT: .LBB55_2: # %else
; RV64I-NEXT: lw a1, 0(a0)
; RV64I-NEXT: addi a2, a1, -1
; RV64I-NEXT: sw a2, 0(a0)
-; RV64I-NEXT: sext.w a0, a1
+; RV64I-NEXT: mv a0, a1
; RV64I-NEXT: ret
;
; RV64IA-LABEL: atomicrmw_sub_i32_monotonic_crossbb:
@@ -6846,13 +6916,12 @@ define signext i32 @atomicrmw_and_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: call __atomic_fetch_and_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ret
; RV64I-NEXT: .LBB56_2: # %else
; RV64I-NEXT: lw a1, 0(a0)
; RV64I-NEXT: andi a2, a1, 1
; RV64I-NEXT: sw a2, 0(a0)
-; RV64I-NEXT: sext.w a0, a1
+; RV64I-NEXT: mv a0, a1
; RV64I-NEXT: ret
;
; RV64IA-LABEL: atomicrmw_and_i32_monotonic_crossbb:
@@ -7008,13 +7077,12 @@ define signext i32 @atomicrmw_nand_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: call __atomic_fetch_nand_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ret
; RV64I-NEXT: .LBB57_2: # %else
; RV64I-NEXT: lw a1, 0(a0)
; RV64I-NEXT: andi a2, a1, 1
; RV64I-NEXT: sw a2, 0(a0)
-; RV64I-NEXT: sext.w a0, a1
+; RV64I-NEXT: mv a0, a1
; RV64I-NEXT: ret
;
; RV64IA-NOZACAS-LABEL: atomicrmw_nand_i32_monotonic_crossbb:
@@ -7169,13 +7237,12 @@ define signext i32 @atomicrmw_or_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind {
; RV64I-NEXT: call __atomic_fetch_or_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ret
; RV64I-NEXT: .LBB58_2: # %else
; RV64I-NEXT: lw a1, 0(a0)
; RV64I-NEXT: ori a2, a1, 1
; RV64I-NEXT: sw a2, 0(a0)
-; RV64I-NEXT: sext.w a0, a1
+; RV64I-NEXT: mv a0, a1
; RV64I-NEXT: ret
;
; RV64IA-LABEL: atomicrmw_or_i32_monotonic_crossbb:
@@ -7300,13 +7367,12 @@ define signext i32 @atomicrmw_xor_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: call __atomic_fetch_xor_4
; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT: addi sp, sp, 16
-; RV64I-NEXT: sext.w a0, a0
; RV64I-NEXT: ret
; RV64I-NEXT: .LBB59_2: # %else
; RV64I-NEXT: lw a1, 0(a0)
; RV64I-NEXT: xori a2, a1, 1
; RV64I-NEXT: sw a2, 0(a0)
-; RV64I-NEXT: sext.w a0, a1
+; RV64I-NEXT: mv a0, a1
; RV64I-NEXT: ret
;
; RV64IA-LABEL: atomicrmw_xor_i32_monotonic_crossbb:
@@ -7486,10 +7552,9 @@ define signext i32 @atomicrmw_max_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: bnez a0, .LBB60_8
; RV64I-NEXT: .LBB60_3: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
-; RV64I-NEXT: li a3, 1
+; RV64I-NEXT: li a0, 1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a3, a0, .LBB60_2
+; RV64I-NEXT: blt a0, a1, .LBB60_2
; RV64I-NEXT: # %bb.4: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB60_3 Depth=1
; RV64I-NEXT: li a2, 1
@@ -7704,9 +7769,8 @@ define signext i32 @atomicrmw_min_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: bnez a0, .LBB61_8
; RV64I-NEXT: .LBB61_3: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: blt a0, s1, .LBB61_2
+; RV64I-NEXT: blt a1, s1, .LBB61_2
; RV64I-NEXT: # %bb.4: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB61_3 Depth=1
; RV64I-NEXT: li a2, 1
@@ -7902,10 +7966,9 @@ define signext i32 @atomicrmw_umax_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: bnez a0, .LBB62_6
; RV64I-NEXT: .LBB62_3: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
-; RV64I-NEXT: li a3, 1
+; RV64I-NEXT: li a0, 1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a3, a0, .LBB62_2
+; RV64I-NEXT: bltu a0, a1, .LBB62_2
; RV64I-NEXT: # %bb.4: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB62_3 Depth=1
; RV64I-NEXT: li a2, 1
@@ -8115,9 +8178,8 @@ define signext i32 @atomicrmw_umin_i32_monotonic_crossbb(ptr %a, i1 %c) nounwind
; RV64I-NEXT: bnez a0, .LBB63_8
; RV64I-NEXT: .LBB63_3: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
; RV64I-NEXT: mv a2, a1
-; RV64I-NEXT: bltu a0, s1, .LBB63_2
+; RV64I-NEXT: bltu a1, s1, .LBB63_2
; RV64I-NEXT: # %bb.4: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB63_3 Depth=1
; RV64I-NEXT: li a2, 1
diff --git a/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll b/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll
index 18b65dc3912030..b4e744808721fb 100644
--- a/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll
+++ b/llvm/test/CodeGen/RISCV/atomicrmw-cond-sub-clamp.ll
@@ -38,8 +38,9 @@ define i8 @atomicrmw_usub_cond_i8(ptr %ptr, i8 %val) {
; RV32I-NEXT: sltu a0, a0, s2
; RV32I-NEXT: addi a0, a0, -1
; RV32I-NEXT: and a0, a0, s0
-; RV32I-NEXT: sub a2, a1, a0
+; RV32I-NEXT: sub a0, a1, a0
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -123,8 +124,9 @@ define i8 @atomicrmw_usub_cond_i8(ptr %ptr, i8 %val) {
; RV64I-NEXT: sltu a0, a0, s2
; RV64I-NEXT: addi a0, a0, -1
; RV64I-NEXT: and a0, a0, s0
-; RV64I-NEXT: sub a2, a1, a0
+; RV64I-NEXT: sub a0, a1, a0
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -216,8 +218,9 @@ define i16 @atomicrmw_usub_cond_i16(ptr %ptr, i16 %val) {
; RV32I-NEXT: sltu a0, a0, s3
; RV32I-NEXT: addi a0, a0, -1
; RV32I-NEXT: and a0, a0, s0
-; RV32I-NEXT: sub a2, a1, a0
+; RV32I-NEXT: sub a0, a1, a0
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -308,8 +311,9 @@ define i16 @atomicrmw_usub_cond_i16(ptr %ptr, i16 %val) {
; RV64I-NEXT: sltu a0, a0, s3
; RV64I-NEXT: addi a0, a0, -1
; RV64I-NEXT: and a0, a0, s0
-; RV64I-NEXT: sub a2, a1, a0
+; RV64I-NEXT: sub a0, a1, a0
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -462,8 +466,7 @@ define i32 @atomicrmw_usub_cond_i32(ptr %ptr, i32 %val) {
; RV64I-NEXT: sext.w s2, s0
; RV64I-NEXT: .LBB2_1: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
-; RV64I-NEXT: sltu a0, a0, s2
+; RV64I-NEXT: sltu a0, a1, s2
; RV64I-NEXT: addi a0, a0, -1
; RV64I-NEXT: and a0, a0, s0
; RV64I-NEXT: subw a2, a1, a0
@@ -1139,8 +1142,7 @@ define i32 @atomicrmw_usub_sat_i32(ptr %ptr, i32 %val) {
; RV64I-NEXT: .LBB6_1: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: subw a0, a1, s0
-; RV64I-NEXT: sext.w a2, a1
-; RV64I-NEXT: sltu a2, a2, a0
+; RV64I-NEXT: sltu a2, a1, a0
; RV64I-NEXT: addi a2, a2, -1
; RV64I-NEXT: and a2, a2, a0
; RV64I-NEXT: sw a1, 4(sp)
diff --git a/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll b/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll
index d09d911d00c526..b2550986888319 100644
--- a/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll
+++ b/llvm/test/CodeGen/RISCV/atomicrmw-uinc-udec-wrap.ll
@@ -36,8 +36,9 @@ define i8 @atomicrmw_uinc_wrap_i8(ptr %ptr, i8 %val) {
; RV32I-NEXT: sltu a0, a0, s1
; RV32I-NEXT: addi a2, a1, 1
; RV32I-NEXT: neg a0, a0
-; RV32I-NEXT: and a2, a0, a2
+; RV32I-NEXT: and a0, a0, a2
; RV32I-NEXT: sb a1, 3(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 3
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -117,8 +118,9 @@ define i8 @atomicrmw_uinc_wrap_i8(ptr %ptr, i8 %val) {
; RV64I-NEXT: sltu a0, a0, s1
; RV64I-NEXT: addi a2, a1, 1
; RV64I-NEXT: neg a0, a0
-; RV64I-NEXT: and a2, a0, a2
+; RV64I-NEXT: and a0, a0, a2
; RV64I-NEXT: sb a1, 7(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 7
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -203,11 +205,12 @@ define i16 @atomicrmw_uinc_wrap_i16(ptr %ptr, i16 %val) {
; RV32I-NEXT: .LBB1_1: # %atomicrmw.start
; RV32I-NEXT: # =>This Inner Loop Header: Depth=1
; RV32I-NEXT: and a0, a1, s2
-; RV32I-NEXT: sltu a0, a0, s1
; RV32I-NEXT: addi a2, a1, 1
+; RV32I-NEXT: sltu a0, a0, s1
; RV32I-NEXT: neg a0, a0
-; RV32I-NEXT: and a2, a0, a2
+; RV32I-NEXT: and a2, a2, s2
; RV32I-NEXT: sh a1, 14(sp)
+; RV32I-NEXT: and a2, a0, a2
; RV32I-NEXT: addi a1, sp, 14
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -291,11 +294,12 @@ define i16 @atomicrmw_uinc_wrap_i16(ptr %ptr, i16 %val) {
; RV64I-NEXT: .LBB1_1: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
; RV64I-NEXT: and a0, a1, s2
-; RV64I-NEXT: sltu a0, a0, s1
; RV64I-NEXT: addi a2, a1, 1
+; RV64I-NEXT: sltu a0, a0, s1
; RV64I-NEXT: neg a0, a0
-; RV64I-NEXT: and a2, a0, a2
+; RV64I-NEXT: and a2, a2, s2
; RV64I-NEXT: sh a1, 14(sp)
+; RV64I-NEXT: and a2, a0, a2
; RV64I-NEXT: addi a1, sp, 14
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -444,8 +448,7 @@ define i32 @atomicrmw_uinc_wrap_i32(ptr %ptr, i32 %val) {
; RV64I-NEXT: sext.w s1, s1
; RV64I-NEXT: .LBB2_1: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
-; RV64I-NEXT: sltu a0, a0, s1
+; RV64I-NEXT: sltu a0, a1, s1
; RV64I-NEXT: addiw a2, a1, 1
; RV64I-NEXT: neg a0, a0
; RV64I-NEXT: and a2, a0, a2
@@ -711,6 +714,7 @@ define i8 @atomicrmw_udec_wrap_i8(ptr %ptr, i8 %val) {
; RV32I-NEXT: .LBB4_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB4_2 Depth=1
; RV32I-NEXT: sb a1, 15(sp)
+; RV32I-NEXT: zext.b a2, a0
; RV32I-NEXT: addi a1, sp, 15
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -723,12 +727,12 @@ define i8 @atomicrmw_udec_wrap_i8(ptr %ptr, i8 %val) {
; RV32I-NEXT: zext.b a0, a1
; RV32I-NEXT: seqz a2, a0
; RV32I-NEXT: sltu a0, s2, a0
-; RV32I-NEXT: or a0, a2, a0
-; RV32I-NEXT: mv a2, s0
-; RV32I-NEXT: bnez a0, .LBB4_1
+; RV32I-NEXT: or a2, a2, a0
+; RV32I-NEXT: mv a0, s0
+; RV32I-NEXT: bnez a2, .LBB4_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB4_2 Depth=1
-; RV32I-NEXT: addi a2, a1, -1
+; RV32I-NEXT: addi a0, a1, -1
; RV32I-NEXT: j .LBB4_1
; RV32I-NEXT: .LBB4_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -814,6 +818,7 @@ define i8 @atomicrmw_udec_wrap_i8(ptr %ptr, i8 %val) {
; RV64I-NEXT: .LBB4_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB4_2 Depth=1
; RV64I-NEXT: sb a1, 15(sp)
+; RV64I-NEXT: zext.b a2, a0
; RV64I-NEXT: addi a1, sp, 15
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -826,12 +831,12 @@ define i8 @atomicrmw_udec_wrap_i8(ptr %ptr, i8 %val) {
; RV64I-NEXT: zext.b a0, a1
; RV64I-NEXT: seqz a2, a0
; RV64I-NEXT: sltu a0, s2, a0
-; RV64I-NEXT: or a0, a2, a0
-; RV64I-NEXT: mv a2, s0
-; RV64I-NEXT: bnez a0, .LBB4_1
+; RV64I-NEXT: or a2, a2, a0
+; RV64I-NEXT: mv a0, s0
+; RV64I-NEXT: bnez a2, .LBB4_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB4_2 Depth=1
-; RV64I-NEXT: addi a2, a1, -1
+; RV64I-NEXT: addi a0, a1, -1
; RV64I-NEXT: j .LBB4_1
; RV64I-NEXT: .LBB4_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -925,6 +930,7 @@ define i16 @atomicrmw_udec_wrap_i16(ptr %ptr, i16 %val) {
; RV32I-NEXT: .LBB5_1: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB5_2 Depth=1
; RV32I-NEXT: sh a1, 10(sp)
+; RV32I-NEXT: and a2, a0, s2
; RV32I-NEXT: addi a1, sp, 10
; RV32I-NEXT: li a3, 5
; RV32I-NEXT: li a4, 5
@@ -937,12 +943,12 @@ define i16 @atomicrmw_udec_wrap_i16(ptr %ptr, i16 %val) {
; RV32I-NEXT: and a0, a1, s2
; RV32I-NEXT: seqz a2, a0
; RV32I-NEXT: sltu a0, s3, a0
-; RV32I-NEXT: or a0, a2, a0
-; RV32I-NEXT: mv a2, s0
-; RV32I-NEXT: bnez a0, .LBB5_1
+; RV32I-NEXT: or a2, a2, a0
+; RV32I-NEXT: mv a0, s0
+; RV32I-NEXT: bnez a2, .LBB5_1
; RV32I-NEXT: # %bb.3: # %atomicrmw.start
; RV32I-NEXT: # in Loop: Header=BB5_2 Depth=1
-; RV32I-NEXT: addi a2, a1, -1
+; RV32I-NEXT: addi a0, a1, -1
; RV32I-NEXT: j .LBB5_1
; RV32I-NEXT: .LBB5_4: # %atomicrmw.end
; RV32I-NEXT: mv a0, a1
@@ -1035,6 +1041,7 @@ define i16 @atomicrmw_udec_wrap_i16(ptr %ptr, i16 %val) {
; RV64I-NEXT: .LBB5_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB5_2 Depth=1
; RV64I-NEXT: sh a1, 6(sp)
+; RV64I-NEXT: and a2, a0, s2
; RV64I-NEXT: addi a1, sp, 6
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -1047,12 +1054,12 @@ define i16 @atomicrmw_udec_wrap_i16(ptr %ptr, i16 %val) {
; RV64I-NEXT: and a0, a1, s2
; RV64I-NEXT: seqz a2, a0
; RV64I-NEXT: sltu a0, s3, a0
-; RV64I-NEXT: or a0, a2, a0
-; RV64I-NEXT: mv a2, s0
-; RV64I-NEXT: bnez a0, .LBB5_1
+; RV64I-NEXT: or a2, a2, a0
+; RV64I-NEXT: mv a0, s0
+; RV64I-NEXT: bnez a2, .LBB5_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB5_2 Depth=1
-; RV64I-NEXT: addi a2, a1, -1
+; RV64I-NEXT: addi a0, a1, -1
; RV64I-NEXT: j .LBB5_1
; RV64I-NEXT: .LBB5_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
@@ -1229,6 +1236,7 @@ define i32 @atomicrmw_udec_wrap_i32(ptr %ptr, i32 %val) {
; RV64I-NEXT: .LBB6_1: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB6_2 Depth=1
; RV64I-NEXT: sw a1, 12(sp)
+; RV64I-NEXT: sext.w a2, a2
; RV64I-NEXT: addi a1, sp, 12
; RV64I-NEXT: li a3, 5
; RV64I-NEXT: li a4, 5
@@ -1238,15 +1246,14 @@ define i32 @atomicrmw_udec_wrap_i32(ptr %ptr, i32 %val) {
; RV64I-NEXT: bnez a0, .LBB6_4
; RV64I-NEXT: .LBB6_2: # %atomicrmw.start
; RV64I-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64I-NEXT: sext.w a0, a1
-; RV64I-NEXT: seqz a2, a0
-; RV64I-NEXT: sltu a0, s2, a0
-; RV64I-NEXT: or a0, a2, a0
+; RV64I-NEXT: seqz a0, a1
+; RV64I-NEXT: sltu a2, s2, a1
+; RV64I-NEXT: or a0, a0, a2
; RV64I-NEXT: mv a2, s0
; RV64I-NEXT: bnez a0, .LBB6_1
; RV64I-NEXT: # %bb.3: # %atomicrmw.start
; RV64I-NEXT: # in Loop: Header=BB6_2 Depth=1
-; RV64I-NEXT: addiw a2, a1, -1
+; RV64I-NEXT: addi a2, a1, -1
; RV64I-NEXT: j .LBB6_1
; RV64I-NEXT: .LBB6_4: # %atomicrmw.end
; RV64I-NEXT: mv a0, a1
diff --git a/llvm/test/CodeGen/RISCV/forced-atomics.ll b/llvm/test/CodeGen/RISCV/forced-atomics.ll
index a15ecc6a158784..3527e2579fd5f0 100644
--- a/llvm/test/CodeGen/RISCV/forced-atomics.ll
+++ b/llvm/test/CodeGen/RISCV/forced-atomics.ll
@@ -1428,10 +1428,9 @@ define i32 @rmw32_max_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: bnez a0, .LBB23_4
; RV64-NO-ATOMIC-NEXT: .LBB23_2: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64-NO-ATOMIC-NEXT: sext.w a0, a1
-; RV64-NO-ATOMIC-NEXT: li a3, 1
+; RV64-NO-ATOMIC-NEXT: li a0, 1
; RV64-NO-ATOMIC-NEXT: mv a2, a1
-; RV64-NO-ATOMIC-NEXT: blt a3, a0, .LBB23_1
+; RV64-NO-ATOMIC-NEXT: blt a0, a1, .LBB23_1
; RV64-NO-ATOMIC-NEXT: # %bb.3: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # in Loop: Header=BB23_2 Depth=1
; RV64-NO-ATOMIC-NEXT: li a2, 1
@@ -1549,9 +1548,8 @@ define i32 @rmw32_min_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: bnez a0, .LBB24_4
; RV64-NO-ATOMIC-NEXT: .LBB24_2: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64-NO-ATOMIC-NEXT: sext.w a0, a1
; RV64-NO-ATOMIC-NEXT: mv a2, a1
-; RV64-NO-ATOMIC-NEXT: blt a0, s1, .LBB24_1
+; RV64-NO-ATOMIC-NEXT: blt a1, s1, .LBB24_1
; RV64-NO-ATOMIC-NEXT: # %bb.3: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # in Loop: Header=BB24_2 Depth=1
; RV64-NO-ATOMIC-NEXT: li a2, 1
@@ -1658,10 +1656,9 @@ define i32 @rmw32_umax_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: bnez a0, .LBB25_4
; RV64-NO-ATOMIC-NEXT: .LBB25_2: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64-NO-ATOMIC-NEXT: sext.w a0, a1
-; RV64-NO-ATOMIC-NEXT: li a3, 1
+; RV64-NO-ATOMIC-NEXT: li a0, 1
; RV64-NO-ATOMIC-NEXT: mv a2, a1
-; RV64-NO-ATOMIC-NEXT: bltu a3, a0, .LBB25_1
+; RV64-NO-ATOMIC-NEXT: bltu a0, a1, .LBB25_1
; RV64-NO-ATOMIC-NEXT: # %bb.3: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # in Loop: Header=BB25_2 Depth=1
; RV64-NO-ATOMIC-NEXT: li a2, 1
@@ -1779,9 +1776,8 @@ define i32 @rmw32_umin_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: bnez a0, .LBB26_4
; RV64-NO-ATOMIC-NEXT: .LBB26_2: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # =>This Inner Loop Header: Depth=1
-; RV64-NO-ATOMIC-NEXT: sext.w a0, a1
; RV64-NO-ATOMIC-NEXT: mv a2, a1
-; RV64-NO-ATOMIC-NEXT: bltu a0, s1, .LBB26_1
+; RV64-NO-ATOMIC-NEXT: bltu a1, s1, .LBB26_1
; RV64-NO-ATOMIC-NEXT: # %bb.3: # %atomicrmw.start
; RV64-NO-ATOMIC-NEXT: # in Loop: Header=BB26_2 Depth=1
; RV64-NO-ATOMIC-NEXT: li a2, 1
@@ -1981,8 +1977,8 @@ define float @rmw32_fadd_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: lui a1, 260096
; RV64-NO-ATOMIC-NEXT: mv a0, s1
; RV64-NO-ATOMIC-NEXT: call __addsf3
-; RV64-NO-ATOMIC-NEXT: mv a2, a0
; RV64-NO-ATOMIC-NEXT: sw s1, 4(sp)
+; RV64-NO-ATOMIC-NEXT: sext.w a2, a0
; RV64-NO-ATOMIC-NEXT: addi a1, sp, 4
; RV64-NO-ATOMIC-NEXT: li a3, 5
; RV64-NO-ATOMIC-NEXT: li a4, 5
@@ -2159,8 +2155,8 @@ define float @rmw32_fsub_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: lui a1, 784384
; RV64-NO-ATOMIC-NEXT: mv a0, s1
; RV64-NO-ATOMIC-NEXT: call __addsf3
-; RV64-NO-ATOMIC-NEXT: mv a2, a0
; RV64-NO-ATOMIC-NEXT: sw s1, 4(sp)
+; RV64-NO-ATOMIC-NEXT: sext.w a2, a0
; RV64-NO-ATOMIC-NEXT: addi a1, sp, 4
; RV64-NO-ATOMIC-NEXT: li a3, 5
; RV64-NO-ATOMIC-NEXT: li a4, 5
@@ -2337,8 +2333,8 @@ define float @rmw32_fmin_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: lui a1, 260096
; RV64-NO-ATOMIC-NEXT: mv a0, s1
; RV64-NO-ATOMIC-NEXT: call fminf
-; RV64-NO-ATOMIC-NEXT: mv a2, a0
; RV64-NO-ATOMIC-NEXT: sw s1, 4(sp)
+; RV64-NO-ATOMIC-NEXT: sext.w a2, a0
; RV64-NO-ATOMIC-NEXT: addi a1, sp, 4
; RV64-NO-ATOMIC-NEXT: li a3, 5
; RV64-NO-ATOMIC-NEXT: li a4, 5
@@ -2515,8 +2511,8 @@ define float @rmw32_fmax_seq_cst(ptr %p) nounwind {
; RV64-NO-ATOMIC-NEXT: lui a1, 260096
; RV64-NO-ATOMIC-NEXT: mv a0, s1
; RV64-NO-ATOMIC-NEXT: call fmaxf
-; RV64-NO-ATOMIC-NEXT: mv a2, a0
; RV64-NO-ATOMIC-NEXT: sw s1, 4(sp)
+; RV64-NO-ATOMIC-NEXT: sext.w a2, a0
; RV64-NO-ATOMIC-NEXT: addi a1, sp, 4
; RV64-NO-ATOMIC-NEXT: li a3, 5
; RV64-NO-ATOMIC-NEXT: li a4, 5
diff --git a/llvm/test/CodeGen/RISCV/libcall-tail-calls.ll b/llvm/test/CodeGen/RISCV/libcall-tail-calls.ll
index 34a58832d912cc..65f653ff7dc769 100644
--- a/llvm/test/CodeGen/RISCV/libcall-tail-calls.ll
+++ b/llvm/test/CodeGen/RISCV/libcall-tail-calls.ll
@@ -533,6 +533,8 @@ define i16 @atomicrmw_add_i16_release(ptr %a, i16 %b) nounwind {
; RV32-ALL: # %bb.0:
; RV32-ALL-NEXT: addi sp, sp, -16
; RV32-ALL-NEXT: sw ra, 12(sp) # 4-byte Folded Spill
+; RV32-ALL-NEXT: slli a1, a1, 16
+; RV32-ALL-NEXT: srli a1, a1, 16
; RV32-ALL-NEXT: li a2, 3
; RV32-ALL-NEXT: call __atomic_fetch_add_2
; RV32-ALL-NEXT: lw ra, 12(sp) # 4-byte Folded Reload
@@ -543,6 +545,8 @@ define i16 @atomicrmw_add_i16_release(ptr %a, i16 %b) nounwind {
; RV64-ALL: # %bb.0:
; RV64-ALL-NEXT: addi sp, sp, -16
; RV64-ALL-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64-ALL-NEXT: slli a1, a1, 48
+; RV64-ALL-NEXT: srli a1, a1, 48
; RV64-ALL-NEXT: li a2, 3
; RV64-ALL-NEXT: call __atomic_fetch_add_2
; RV64-ALL-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
@@ -567,6 +571,7 @@ define i32 @atomicrmw_xor_i32_acq_rel(ptr %a, i32 %b) nounwind {
; RV64-ALL: # %bb.0:
; RV64-ALL-NEXT: addi sp, sp, -16
; RV64-ALL-NEXT: sd ra, 8(sp) # 8-byte Folded Spill
+; RV64-ALL-NEXT: sext.w a1, a1
; RV64-ALL-NEXT: li a2, 4
; RV64-ALL-NEXT: call __atomic_fetch_xor_4
; RV64-ALL-NEXT: ld ra, 8(sp) # 8-byte Folded Reload
diff --git a/llvm/test/Transforms/AtomicExpand/MSP430/libcalls.ll b/llvm/test/Transforms/AtomicExpand/MSP430/libcalls.ll
index 7a818bb6d535fc..4d187158f81256 100644
--- a/llvm/test/Transforms/AtomicExpand/MSP430/libcalls.ll
+++ b/llvm/test/Transforms/AtomicExpand/MSP430/libcalls.ll
@@ -6,7 +6,7 @@
define i8 @load_i8(ptr %p) {
; CHECK-LABEL: define i8 @load_i8(
; CHECK-SAME: ptr [[P:%.*]]) {
-; CHECK-NEXT: [[TMP1:%.*]] = call i8 @__atomic_load_1(ptr [[P]], i16 5)
+; CHECK-NEXT: [[TMP1:%.*]] = call zeroext i8 @__atomic_load_1(ptr [[P]], i16 5)
; CHECK-NEXT: ret i8 [[TMP1]]
;
%v = load atomic i8, ptr %p seq_cst, align 1
diff --git a/llvm/test/Transforms/AtomicExpand/SPARC/libcalls.ll b/llvm/test/Transforms/AtomicExpand/SPARC/libcalls.ll
index 3a00ae2217c4f1..990e668c424cb4 100644
--- a/llvm/test/Transforms/AtomicExpand/SPARC/libcalls.ll
+++ b/llvm/test/Transforms/AtomicExpand/SPARC/libcalls.ll
@@ -13,7 +13,7 @@ target triple = "sparc-unknown-unknown"
;; straightforward.
; CHECK-LABEL: @test_load_i16(
-; CHECK: %1 = call i16 @__atomic_load_2(ptr %arg, i32 5)
+; CHECK: %1 = call zeroext i16 @__atomic_load_2(ptr %arg, i32 5)
; CHECK: ret i16 %1
define i16 @test_load_i16(ptr %arg) {
%ret = load atomic i16, ptr %arg seq_cst, align 4
@@ -21,7 +21,7 @@ define i16 @test_load_i16(ptr %arg) {
}
; CHECK-LABEL: @test_store_i16(
-; CHECK: call void @__atomic_store_2(ptr %arg, i16 %val, i32 5)
+; CHECK: call void @__atomic_store_2(ptr %arg, i16 zeroext %val, i32 5)
; CHECK: ret void
define void @test_store_i16(ptr %arg, i16 %val) {
store atomic i16 %val, ptr %arg seq_cst, align 4
@@ -29,7 +29,7 @@ define void @test_store_i16(ptr %arg, i16 %val) {
}
; CHECK-LABEL: @test_exchange_i16(
-; CHECK: %1 = call i16 @__atomic_exchange_2(ptr %arg, i16 %val, i32 5)
+; CHECK: %1 = call zeroext i16 @__atomic_exchange_2(ptr %arg, i16 zeroext %val, i32 5)
; CHECK: ret i16 %1
define i16 @test_exchange_i16(ptr %arg, i16 %val) {
%ret = atomicrmw xchg ptr %arg, i16 %val seq_cst
@@ -40,7 +40,7 @@ define i16 @test_exchange_i16(ptr %arg, i16 %val) {
; CHECK: %1 = alloca i16, align 2
; CHECK: call void @llvm.lifetime.start.p0(ptr %1)
; CHECK: store i16 %old, ptr %1, align 2
-; CHECK: %2 = call zeroext i1 @__atomic_compare_exchange_2(ptr %arg, ptr %1, i16 %new, i32 5, i32 0)
+; CHECK: %2 = call zeroext i1 @__atomic_compare_exchange_2(ptr %arg, ptr %1, i16 zeroext %new, i32 5, i32 0)
; CHECK: %3 = load i16, ptr %1, align 2
; CHECK: call void @llvm.lifetime.end.p0(ptr %1)
; CHECK: %4 = insertvalue { i16, i1 } poison, i16 %3, 0
@@ -54,7 +54,7 @@ define i16 @test_cmpxchg_i16(ptr %arg, i16 %old, i16 %new) {
}
; CHECK-LABEL: @test_add_i16(
-; CHECK: %1 = call i16 @__atomic_fetch_add_2(ptr %arg, i16 %val, i32 5)
+; CHECK: %1 = call zeroext i16 @__atomic_fetch_add_2(ptr %arg, i16 zeroext %val, i32 5)
; CHECK: ret i16 %1
define i16 @test_add_i16(ptr %arg, i16 %val) {
%ret = atomicrmw add ptr %arg, i16 %val seq_cst
More information about the llvm-branch-commits
mailing list