[llvm] 9c377c5 - [RISCV] Copy lack-of-signed-truncation-check.ll and signed-truncation-check.ll from AArch6/X86. NFC
Craig Topper via llvm-commits
llvm-commits at lists.llvm.org
Fri May 5 11:16:30 PDT 2023
Author: Craig Topper
Date: 2023-05-05T11:15:39-07:00
New Revision: 9c377c53da4771389fbd65e99e1615d99c257bdf
URL: https://github.com/llvm/llvm-project/commit/9c377c53da4771389fbd65e99e1615d99c257bdf
DIFF: https://github.com/llvm/llvm-project/commit/9c377c53da4771389fbd65e99e1615d99c257bdf.diff
LOG: [RISCV] Copy lack-of-signed-truncation-check.ll and signed-truncation-check.ll from AArch6/X86. NFC
This is a more exhaustive set of tests for the same issue D149814
is trying to solve.
Added:
llvm/test/CodeGen/RISCV/lack-of-signed-truncation-check.ll
llvm/test/CodeGen/RISCV/signed-truncation-check.ll
Modified:
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/RISCV/lack-of-signed-truncation-check.ll b/llvm/test/CodeGen/RISCV/lack-of-signed-truncation-check.ll
new file mode 100644
index 0000000000000..4462513d01b91
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/lack-of-signed-truncation-check.ll
@@ -0,0 +1,934 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32-unknown-linux-gnu < %s | FileCheck %s --check-prefixes=CHECK,RV32,RV32I
+; RUN: llc -mtriple=riscv64-unknown-linux-gnu < %s | FileCheck %s --check-prefixes=CHECK,RV64,RV64I
+; RUN: llc -mtriple=riscv32-unknown-linux-gnu -mattr=+zbb < %s | FileCheck %s --check-prefixes=CHECK,RV32,RV32ZBB
+; RUN: llc -mtriple=riscv64-unknown-linux-gnu -mattr=+zbb < %s | FileCheck %s --check-prefixes=CHECK,RV64,RV64ZBB
+
+; https://bugs.llvm.org/show_bug.cgi?id=38149
+
+; We are truncating from wider width, and then sign-extending
+; back to the original width. Then we inequality-comparing orig and src.
+; If they don't match, then we had signed truncation during truncation.
+
+; This can be expressed in a several ways in IR:
+; trunc + sext + icmp ne <- not canonical
+; shl + ashr + icmp ne
+; add + icmp ult/ule
+; add + icmp uge/ugt
+; However only the simplest form (with two shifts) gets lowered best.
+
+; ---------------------------------------------------------------------------- ;
+; shl + ashr + icmp ne
+; ---------------------------------------------------------------------------- ;
+
+define i1 @shifts_necmp_i16_i8(i16 %x) nounwind {
+; RV32I-LABEL: shifts_necmp_i16_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: lui a1, 16
+; RV32I-NEXT: addi a1, a1, -1
+; RV32I-NEXT: and a2, a0, a1
+; RV32I-NEXT: slli a0, a0, 24
+; RV32I-NEXT: srai a0, a0, 24
+; RV32I-NEXT: and a0, a0, a1
+; RV32I-NEXT: xor a0, a0, a2
+; RV32I-NEXT: snez a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_necmp_i16_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: lui a1, 16
+; RV64I-NEXT: addiw a1, a1, -1
+; RV64I-NEXT: and a2, a0, a1
+; RV64I-NEXT: slli a0, a0, 56
+; RV64I-NEXT: srai a0, a0, 56
+; RV64I-NEXT: and a0, a0, a1
+; RV64I-NEXT: xor a0, a0, a2
+; RV64I-NEXT: snez a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_necmp_i16_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: zext.h a1, a0
+; RV32ZBB-NEXT: sext.b a0, a0
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: xor a0, a0, a1
+; RV32ZBB-NEXT: snez a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_necmp_i16_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: zext.h a1, a0
+; RV64ZBB-NEXT: sext.b a0, a0
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: xor a0, a0, a1
+; RV64ZBB-NEXT: snez a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i16 %x, 8 ; 16-8
+ %tmp1 = ashr exact i16 %tmp0, 8 ; 16-8
+ %tmp2 = icmp ne i16 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_necmp_i32_i16(i32 %x) nounwind {
+; RV32I-LABEL: shifts_necmp_i32_i16:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a1, a0, 16
+; RV32I-NEXT: srai a1, a1, 16
+; RV32I-NEXT: xor a0, a1, a0
+; RV32I-NEXT: snez a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_necmp_i32_i16:
+; RV64I: # %bb.0:
+; RV64I-NEXT: sext.w a1, a0
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srai a0, a0, 48
+; RV64I-NEXT: xor a0, a0, a1
+; RV64I-NEXT: snez a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_necmp_i32_i16:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.h a1, a0
+; RV32ZBB-NEXT: xor a0, a1, a0
+; RV32ZBB-NEXT: snez a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_necmp_i32_i16:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.w a1, a0
+; RV64ZBB-NEXT: sext.h a0, a0
+; RV64ZBB-NEXT: xor a0, a0, a1
+; RV64ZBB-NEXT: snez a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i32 %x, 16 ; 32-16
+ %tmp1 = ashr exact i32 %tmp0, 16 ; 32-16
+ %tmp2 = icmp ne i32 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_necmp_i32_i8(i32 %x) nounwind {
+; RV32I-LABEL: shifts_necmp_i32_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a1, a0, 24
+; RV32I-NEXT: srai a1, a1, 24
+; RV32I-NEXT: xor a0, a1, a0
+; RV32I-NEXT: snez a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_necmp_i32_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: sext.w a1, a0
+; RV64I-NEXT: slli a0, a0, 56
+; RV64I-NEXT: srai a0, a0, 56
+; RV64I-NEXT: xor a0, a0, a1
+; RV64I-NEXT: snez a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_necmp_i32_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.b a1, a0
+; RV32ZBB-NEXT: xor a0, a1, a0
+; RV32ZBB-NEXT: snez a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_necmp_i32_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.w a1, a0
+; RV64ZBB-NEXT: sext.b a0, a0
+; RV64ZBB-NEXT: xor a0, a0, a1
+; RV64ZBB-NEXT: snez a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i32 %x, 24 ; 32-8
+ %tmp1 = ashr exact i32 %tmp0, 24 ; 32-8
+ %tmp2 = icmp ne i32 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_necmp_i64_i32(i64 %x) nounwind {
+; RV32-LABEL: shifts_necmp_i64_i32:
+; RV32: # %bb.0:
+; RV32-NEXT: srai a0, a0, 31
+; RV32-NEXT: xor a0, a0, a1
+; RV32-NEXT: snez a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: shifts_necmp_i64_i32:
+; RV64: # %bb.0:
+; RV64-NEXT: sext.w a1, a0
+; RV64-NEXT: xor a0, a1, a0
+; RV64-NEXT: snez a0, a0
+; RV64-NEXT: ret
+ %tmp0 = shl i64 %x, 32 ; 64-32
+ %tmp1 = ashr exact i64 %tmp0, 32 ; 64-32
+ %tmp2 = icmp ne i64 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_necmp_i64_i16(i64 %x) nounwind {
+; RV32I-LABEL: shifts_necmp_i64_i16:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a2, a0, 16
+; RV32I-NEXT: srai a3, a2, 16
+; RV32I-NEXT: srai a2, a2, 31
+; RV32I-NEXT: xor a1, a2, a1
+; RV32I-NEXT: xor a0, a3, a0
+; RV32I-NEXT: or a0, a0, a1
+; RV32I-NEXT: snez a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_necmp_i64_i16:
+; RV64I: # %bb.0:
+; RV64I-NEXT: slli a1, a0, 48
+; RV64I-NEXT: srai a1, a1, 48
+; RV64I-NEXT: xor a0, a1, a0
+; RV64I-NEXT: snez a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_necmp_i64_i16:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.h a2, a0
+; RV32ZBB-NEXT: srai a3, a2, 31
+; RV32ZBB-NEXT: xor a0, a2, a0
+; RV32ZBB-NEXT: xor a1, a3, a1
+; RV32ZBB-NEXT: or a0, a0, a1
+; RV32ZBB-NEXT: snez a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_necmp_i64_i16:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.h a1, a0
+; RV64ZBB-NEXT: xor a0, a1, a0
+; RV64ZBB-NEXT: snez a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i64 %x, 48 ; 64-16
+ %tmp1 = ashr exact i64 %tmp0, 48 ; 64-16
+ %tmp2 = icmp ne i64 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_necmp_i64_i8(i64 %x) nounwind {
+; RV32I-LABEL: shifts_necmp_i64_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a2, a0, 24
+; RV32I-NEXT: srai a3, a2, 24
+; RV32I-NEXT: srai a2, a2, 31
+; RV32I-NEXT: xor a1, a2, a1
+; RV32I-NEXT: xor a0, a3, a0
+; RV32I-NEXT: or a0, a0, a1
+; RV32I-NEXT: snez a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_necmp_i64_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: slli a1, a0, 56
+; RV64I-NEXT: srai a1, a1, 56
+; RV64I-NEXT: xor a0, a1, a0
+; RV64I-NEXT: snez a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_necmp_i64_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.b a2, a0
+; RV32ZBB-NEXT: srai a3, a2, 31
+; RV32ZBB-NEXT: xor a0, a2, a0
+; RV32ZBB-NEXT: xor a1, a3, a1
+; RV32ZBB-NEXT: or a0, a0, a1
+; RV32ZBB-NEXT: snez a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_necmp_i64_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.b a1, a0
+; RV64ZBB-NEXT: xor a0, a1, a0
+; RV64ZBB-NEXT: snez a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i64 %x, 56 ; 64-8
+ %tmp1 = ashr exact i64 %tmp0, 56 ; 64-8
+ %tmp2 = icmp ne i64 %tmp1, %x
+ ret i1 %tmp2
+}
+
+; ---------------------------------------------------------------------------- ;
+; add + icmp ult
+; ---------------------------------------------------------------------------- ;
+
+define i1 @add_ultcmp_i16_i8(i16 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i16_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, -128
+; RV32-NEXT: slli a0, a0, 16
+; RV32-NEXT: srli a0, a0, 24
+; RV32-NEXT: sltiu a0, a0, 255
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i16_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, -128
+; RV64-NEXT: slli a0, a0, 48
+; RV64-NEXT: srli a0, a0, 56
+; RV64-NEXT: sltiu a0, a0, 255
+; RV64-NEXT: ret
+ %tmp0 = add i16 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp ult i16 %tmp0, -256 ; ~0U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i32_i16(i32 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i32_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a1, 1048568
+; RV32-NEXT: add a0, a0, a1
+; RV32-NEXT: lui a1, 1048560
+; RV32-NEXT: sltu a0, a0, a1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i32_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 1048568
+; RV64-NEXT: addw a0, a0, a1
+; RV64-NEXT: lui a1, 1048560
+; RV64-NEXT: sltu a0, a0, a1
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, -32768 ; ~0U << (16-1)
+ %tmp1 = icmp ult i32 %tmp0, -65536 ; ~0U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i32_i8(i32 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i32_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, -128
+; RV32-NEXT: sltiu a0, a0, -256
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i32_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, -128
+; RV64-NEXT: sltiu a0, a0, -256
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp ult i32 %tmp0, -256 ; ~0U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i64_i32(i64 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i64_i32:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 524288
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: snez a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i64_i32:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 524288
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: li a1, -1
+; RV64-NEXT: slli a1, a1, 32
+; RV64-NEXT: sltu a0, a0, a1
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, -2147483648 ; ~0U << (32-1)
+ %tmp1 = icmp ult i64 %tmp0, -4294967296 ; ~0U << 32
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i64_i16(i64 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i64_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 1048568
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: addi a0, a0, -1
+; RV32-NEXT: li a1, -1
+; RV32-NEXT: beq a0, a1, .LBB10_2
+; RV32-NEXT: # %bb.1:
+; RV32-NEXT: sltiu a0, a0, -1
+; RV32-NEXT: ret
+; RV32-NEXT: .LBB10_2:
+; RV32-NEXT: lui a0, 1048560
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i64_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 1048568
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: lui a1, 1048560
+; RV64-NEXT: sltu a0, a0, a1
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, -32768 ; ~0U << (16-1)
+ %tmp1 = icmp ult i64 %tmp0, -65536 ; ~0U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i64_i8(i64 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i64_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a2, a0, -128
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: addi a0, a0, -1
+; RV32-NEXT: li a1, -1
+; RV32-NEXT: beq a0, a1, .LBB11_2
+; RV32-NEXT: # %bb.1:
+; RV32-NEXT: sltiu a0, a0, -1
+; RV32-NEXT: ret
+; RV32-NEXT: .LBB11_2:
+; RV32-NEXT: sltiu a0, a2, -256
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i64_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addi a0, a0, -128
+; RV64-NEXT: sltiu a0, a0, -256
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp ult i64 %tmp0, -256 ; ~0U << 8
+ ret i1 %tmp1
+}
+
+; Slightly more canonical variant
+define i1 @add_ulecmp_i16_i8(i16 %x) nounwind {
+; RV32-LABEL: add_ulecmp_i16_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, -128
+; RV32-NEXT: slli a0, a0, 16
+; RV32-NEXT: srli a0, a0, 24
+; RV32-NEXT: sltiu a0, a0, 255
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ulecmp_i16_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, -128
+; RV64-NEXT: slli a0, a0, 48
+; RV64-NEXT: srli a0, a0, 56
+; RV64-NEXT: sltiu a0, a0, 255
+; RV64-NEXT: ret
+ %tmp0 = add i16 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp ule i16 %tmp0, -257 ; ~0U << 8 - 1
+ ret i1 %tmp1
+}
+
+; ---------------------------------------------------------------------------- ;
+; add + icmp uge
+; ---------------------------------------------------------------------------- ;
+
+define i1 @add_ugecmp_i16_i8(i16 %x) nounwind {
+; RV32I-LABEL: add_ugecmp_i16_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_i16_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_i16_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_i16_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp uge i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i32_i16(i32 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i32_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a1, 8
+; RV32-NEXT: add a0, a0, a1
+; RV32-NEXT: srli a0, a0, 16
+; RV32-NEXT: snez a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i32_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 8
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: srliw a0, a0, 16
+; RV64-NEXT: snez a0, a0
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, 32768 ; 1U << (16-1)
+ %tmp1 = icmp uge i32 %tmp0, 65536 ; 1U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i32_i8(i32 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i32_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, 128
+; RV32-NEXT: sltiu a0, a0, 256
+; RV32-NEXT: xori a0, a0, 1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i32_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, 128
+; RV64-NEXT: sltiu a0, a0, 256
+; RV64-NEXT: xori a0, a0, 1
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp uge i32 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i64_i32(i64 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i64_i32:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 524288
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: snez a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i64_i32:
+; RV64: # %bb.0:
+; RV64-NEXT: li a1, 1
+; RV64-NEXT: slli a1, a1, 31
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: srli a0, a0, 32
+; RV64-NEXT: snez a0, a0
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, 2147483648 ; 1U << (32-1)
+ %tmp1 = icmp uge i64 %tmp0, 4294967296 ; 1U << 32
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i64_i16(i64 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i64_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 8
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: srli a2, a2, 16
+; RV32-NEXT: or a0, a0, a2
+; RV32-NEXT: snez a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i64_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 8
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: srli a0, a0, 16
+; RV64-NEXT: snez a0, a0
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, 32768 ; 1U << (16-1)
+ %tmp1 = icmp uge i64 %tmp0, 65536 ; 1U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i64_i8(i64 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i64_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a2, a0, 128
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a1, a1, a0
+; RV32-NEXT: beqz a1, .LBB18_2
+; RV32-NEXT: # %bb.1:
+; RV32-NEXT: snez a0, a1
+; RV32-NEXT: ret
+; RV32-NEXT: .LBB18_2:
+; RV32-NEXT: sltiu a0, a2, 256
+; RV32-NEXT: xori a0, a0, 1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i64_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addi a0, a0, 128
+; RV64-NEXT: sltiu a0, a0, 256
+; RV64-NEXT: xori a0, a0, 1
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp uge i64 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Slightly more canonical variant
+define i1 @add_ugtcmp_i16_i8(i16 %x) nounwind {
+; RV32I-LABEL: add_ugtcmp_i16_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugtcmp_i16_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugtcmp_i16_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugtcmp_i16_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ugt i16 %tmp0, 255 ; (1U << 8) - 1
+ ret i1 %tmp1
+}
+
+; Negative tests
+; ---------------------------------------------------------------------------- ;
+
+; Adding not a constant
+define i1 @add_ugecmp_bad_i16_i8_add(i16 %x, i16 %y) nounwind {
+; RV32I-LABEL: add_ugecmp_bad_i16_i8_add:
+; RV32I: # %bb.0:
+; RV32I-NEXT: add a0, a0, a1
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_bad_i16_i8_add:
+; RV64I: # %bb.0:
+; RV64I-NEXT: add a0, a0, a1
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_bad_i16_i8_add:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: add a0, a0, a1
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_bad_i16_i8_add:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: add a0, a0, a1
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, %y
+ %tmp1 = icmp uge i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Comparing not with a constant
+define i1 @add_ugecmp_bad_i16_i8_cmp(i16 %x, i16 %y) nounwind {
+; RV32I-LABEL: add_ugecmp_bad_i16_i8_cmp:
+; RV32I: # %bb.0:
+; RV32I-NEXT: lui a2, 16
+; RV32I-NEXT: addi a2, a2, -1
+; RV32I-NEXT: and a1, a1, a2
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: and a0, a0, a2
+; RV32I-NEXT: sltu a0, a0, a1
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_bad_i16_i8_cmp:
+; RV64I: # %bb.0:
+; RV64I-NEXT: lui a2, 16
+; RV64I-NEXT: addiw a2, a2, -1
+; RV64I-NEXT: and a1, a1, a2
+; RV64I-NEXT: addi a0, a0, 128
+; RV64I-NEXT: and a0, a0, a2
+; RV64I-NEXT: sltu a0, a0, a1
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_bad_i16_i8_cmp:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: zext.h a1, a1
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltu a0, a0, a1
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_bad_i16_i8_cmp:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: zext.h a1, a1
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltu a0, a0, a1
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp uge i16 %tmp0, %y
+ ret i1 %tmp1
+}
+
+; Second constant is not larger than the first one
+define i1 @add_ugecmp_bad_i8_i16(i16 %x) nounwind {
+; RV32I-LABEL: add_ugecmp_bad_i8_i16:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 128
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_bad_i8_i16:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 128
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_bad_i8_i16:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 128
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_bad_i8_i16:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 128
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp uge i16 %tmp0, 128 ; 1U << (8-1)
+ ret i1 %tmp1
+}
+
+; First constant is not power of two
+define i1 @add_ugecmp_bad_i16_i8_c0notpoweroftwo(i16 %x) nounwind {
+; RV32I-LABEL: add_ugecmp_bad_i16_i8_c0notpoweroftwo:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 192
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_bad_i16_i8_c0notpoweroftwo:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 192
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_bad_i16_i8_c0notpoweroftwo:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 192
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_bad_i16_i8_c0notpoweroftwo:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 192
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 192 ; (1U << (8-1)) + (1U << (8-1-1))
+ %tmp1 = icmp uge i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Second constant is not power of two
+define i1 @add_ugecmp_bad_i16_i8_c1notpoweroftwo(i16 %x) nounwind {
+; RV32I-LABEL: add_ugecmp_bad_i16_i8_c1notpoweroftwo:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 768
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_bad_i16_i8_c1notpoweroftwo:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 768
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_bad_i16_i8_c1notpoweroftwo:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 768
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_bad_i16_i8_c1notpoweroftwo:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 768
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp uge i16 %tmp0, 768 ; (1U << 8)) + (1U << (8+1))
+ ret i1 %tmp1
+}
+
+; Magic check fails, 64 << 1 != 256
+define i1 @add_ugecmp_bad_i16_i8_magic(i16 %x) nounwind {
+; RV32I-LABEL: add_ugecmp_bad_i16_i8_magic:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 64
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_bad_i16_i8_magic:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 64
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_bad_i16_i8_magic:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 64
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_bad_i16_i8_magic:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 64
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 64 ; 1U << (8-1-1)
+ %tmp1 = icmp uge i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Bad 'destination type'
+define i1 @add_ugecmp_bad_i16_i4(i16 %x) nounwind {
+; RV32I-LABEL: add_ugecmp_bad_i16_i4:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 8
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 16
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_bad_i16_i4:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 8
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 16
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_bad_i16_i4:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 8
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 16
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_bad_i16_i4:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 8
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 16
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 8 ; 1U << (4-1)
+ %tmp1 = icmp uge i16 %tmp0, 16 ; 1U << 4
+ ret i1 %tmp1
+}
+
+; Bad storage type
+define i1 @add_ugecmp_bad_i24_i8(i24 %x) nounwind {
+; RV32-LABEL: add_ugecmp_bad_i24_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, 128
+; RV32-NEXT: slli a0, a0, 8
+; RV32-NEXT: srli a0, a0, 8
+; RV32-NEXT: sltiu a0, a0, 256
+; RV32-NEXT: xori a0, a0, 1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_bad_i24_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, 128
+; RV64-NEXT: slli a0, a0, 40
+; RV64-NEXT: srli a0, a0, 40
+; RV64-NEXT: sltiu a0, a0, 256
+; RV64-NEXT: xori a0, a0, 1
+; RV64-NEXT: ret
+ %tmp0 = add i24 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp uge i24 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Slightly more canonical variant
+define i1 @add_ugtcmp_bad_i16_i8(i16 %x) nounwind {
+; CHECK-LABEL: add_ugtcmp_bad_i16_i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: li a0, 0
+; CHECK-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ugt i16 %tmp0, -1 ; when we +1 it, it will wrap to 0
+ ret i1 %tmp1
+}
diff --git a/llvm/test/CodeGen/RISCV/signed-truncation-check.ll b/llvm/test/CodeGen/RISCV/signed-truncation-check.ll
new file mode 100644
index 0000000000000..3d38cfc67eda7
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/signed-truncation-check.ll
@@ -0,0 +1,940 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32-unknown-linux-gnu < %s | FileCheck %s --check-prefixes=CHECK,RV32,RV32I
+; RUN: llc -mtriple=riscv64-unknown-linux-gnu < %s | FileCheck %s --check-prefixes=CHECK,RV64,RV64I
+; RUN: llc -mtriple=riscv32-unknown-linux-gnu -mattr=+zbb < %s | FileCheck %s --check-prefixes=CHECK,RV32,RV32ZBB
+; RUN: llc -mtriple=riscv64-unknown-linux-gnu -mattr=+zbb < %s | FileCheck %s --check-prefixes=CHECK,RV64,RV64ZBB
+
+; https://bugs.llvm.org/show_bug.cgi?id=38149
+
+; We are truncating from wider width, and then sign-extending
+; back to the original width. Then we equality-comparing orig and src.
+; If they don't match, then we had signed truncation during truncation.
+
+; This can be expressed in a several ways in IR:
+; trunc + sext + icmp eq <- not canonical
+; shl + ashr + icmp eq
+; add + icmp uge/ugt
+; add + icmp ult/ule
+; However only the simplest form (with two shifts) gets lowered best.
+
+; ---------------------------------------------------------------------------- ;
+; shl + ashr + icmp eq
+; ---------------------------------------------------------------------------- ;
+
+define i1 @shifts_eqcmp_i16_i8(i16 %x) nounwind {
+; RV32I-LABEL: shifts_eqcmp_i16_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: lui a1, 16
+; RV32I-NEXT: addi a1, a1, -1
+; RV32I-NEXT: and a2, a0, a1
+; RV32I-NEXT: slli a0, a0, 24
+; RV32I-NEXT: srai a0, a0, 24
+; RV32I-NEXT: and a0, a0, a1
+; RV32I-NEXT: xor a0, a0, a2
+; RV32I-NEXT: seqz a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_eqcmp_i16_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: lui a1, 16
+; RV64I-NEXT: addiw a1, a1, -1
+; RV64I-NEXT: and a2, a0, a1
+; RV64I-NEXT: slli a0, a0, 56
+; RV64I-NEXT: srai a0, a0, 56
+; RV64I-NEXT: and a0, a0, a1
+; RV64I-NEXT: xor a0, a0, a2
+; RV64I-NEXT: seqz a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_eqcmp_i16_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: zext.h a1, a0
+; RV32ZBB-NEXT: sext.b a0, a0
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: xor a0, a0, a1
+; RV32ZBB-NEXT: seqz a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_eqcmp_i16_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: zext.h a1, a0
+; RV64ZBB-NEXT: sext.b a0, a0
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: xor a0, a0, a1
+; RV64ZBB-NEXT: seqz a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i16 %x, 8 ; 16-8
+ %tmp1 = ashr exact i16 %tmp0, 8 ; 16-8
+ %tmp2 = icmp eq i16 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_eqcmp_i32_i16(i32 %x) nounwind {
+; RV32I-LABEL: shifts_eqcmp_i32_i16:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a1, a0, 16
+; RV32I-NEXT: srai a1, a1, 16
+; RV32I-NEXT: xor a0, a1, a0
+; RV32I-NEXT: seqz a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_eqcmp_i32_i16:
+; RV64I: # %bb.0:
+; RV64I-NEXT: sext.w a1, a0
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srai a0, a0, 48
+; RV64I-NEXT: xor a0, a0, a1
+; RV64I-NEXT: seqz a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_eqcmp_i32_i16:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.h a1, a0
+; RV32ZBB-NEXT: xor a0, a1, a0
+; RV32ZBB-NEXT: seqz a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_eqcmp_i32_i16:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.w a1, a0
+; RV64ZBB-NEXT: sext.h a0, a0
+; RV64ZBB-NEXT: xor a0, a0, a1
+; RV64ZBB-NEXT: seqz a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i32 %x, 16 ; 32-16
+ %tmp1 = ashr exact i32 %tmp0, 16 ; 32-16
+ %tmp2 = icmp eq i32 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_eqcmp_i32_i8(i32 %x) nounwind {
+; RV32I-LABEL: shifts_eqcmp_i32_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a1, a0, 24
+; RV32I-NEXT: srai a1, a1, 24
+; RV32I-NEXT: xor a0, a1, a0
+; RV32I-NEXT: seqz a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_eqcmp_i32_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: sext.w a1, a0
+; RV64I-NEXT: slli a0, a0, 56
+; RV64I-NEXT: srai a0, a0, 56
+; RV64I-NEXT: xor a0, a0, a1
+; RV64I-NEXT: seqz a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_eqcmp_i32_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.b a1, a0
+; RV32ZBB-NEXT: xor a0, a1, a0
+; RV32ZBB-NEXT: seqz a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_eqcmp_i32_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.w a1, a0
+; RV64ZBB-NEXT: sext.b a0, a0
+; RV64ZBB-NEXT: xor a0, a0, a1
+; RV64ZBB-NEXT: seqz a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i32 %x, 24 ; 32-8
+ %tmp1 = ashr exact i32 %tmp0, 24 ; 32-8
+ %tmp2 = icmp eq i32 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_eqcmp_i64_i32(i64 %x) nounwind {
+; RV32-LABEL: shifts_eqcmp_i64_i32:
+; RV32: # %bb.0:
+; RV32-NEXT: srai a0, a0, 31
+; RV32-NEXT: xor a0, a0, a1
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: shifts_eqcmp_i64_i32:
+; RV64: # %bb.0:
+; RV64-NEXT: sext.w a1, a0
+; RV64-NEXT: xor a0, a1, a0
+; RV64-NEXT: seqz a0, a0
+; RV64-NEXT: ret
+ %tmp0 = shl i64 %x, 32 ; 64-32
+ %tmp1 = ashr exact i64 %tmp0, 32 ; 64-32
+ %tmp2 = icmp eq i64 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_eqcmp_i64_i16(i64 %x) nounwind {
+; RV32I-LABEL: shifts_eqcmp_i64_i16:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a2, a0, 16
+; RV32I-NEXT: srai a3, a2, 16
+; RV32I-NEXT: srai a2, a2, 31
+; RV32I-NEXT: xor a1, a2, a1
+; RV32I-NEXT: xor a0, a3, a0
+; RV32I-NEXT: or a0, a0, a1
+; RV32I-NEXT: seqz a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_eqcmp_i64_i16:
+; RV64I: # %bb.0:
+; RV64I-NEXT: slli a1, a0, 48
+; RV64I-NEXT: srai a1, a1, 48
+; RV64I-NEXT: xor a0, a1, a0
+; RV64I-NEXT: seqz a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_eqcmp_i64_i16:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.h a2, a0
+; RV32ZBB-NEXT: srai a3, a2, 31
+; RV32ZBB-NEXT: xor a0, a2, a0
+; RV32ZBB-NEXT: xor a1, a3, a1
+; RV32ZBB-NEXT: or a0, a0, a1
+; RV32ZBB-NEXT: seqz a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_eqcmp_i64_i16:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.h a1, a0
+; RV64ZBB-NEXT: xor a0, a1, a0
+; RV64ZBB-NEXT: seqz a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i64 %x, 48 ; 64-16
+ %tmp1 = ashr exact i64 %tmp0, 48 ; 64-16
+ %tmp2 = icmp eq i64 %tmp1, %x
+ ret i1 %tmp2
+}
+
+define i1 @shifts_eqcmp_i64_i8(i64 %x) nounwind {
+; RV32I-LABEL: shifts_eqcmp_i64_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a2, a0, 24
+; RV32I-NEXT: srai a3, a2, 24
+; RV32I-NEXT: srai a2, a2, 31
+; RV32I-NEXT: xor a1, a2, a1
+; RV32I-NEXT: xor a0, a3, a0
+; RV32I-NEXT: or a0, a0, a1
+; RV32I-NEXT: seqz a0, a0
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: shifts_eqcmp_i64_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: slli a1, a0, 56
+; RV64I-NEXT: srai a1, a1, 56
+; RV64I-NEXT: xor a0, a1, a0
+; RV64I-NEXT: seqz a0, a0
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: shifts_eqcmp_i64_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: sext.b a2, a0
+; RV32ZBB-NEXT: srai a3, a2, 31
+; RV32ZBB-NEXT: xor a0, a2, a0
+; RV32ZBB-NEXT: xor a1, a3, a1
+; RV32ZBB-NEXT: or a0, a0, a1
+; RV32ZBB-NEXT: seqz a0, a0
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: shifts_eqcmp_i64_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: sext.b a1, a0
+; RV64ZBB-NEXT: xor a0, a1, a0
+; RV64ZBB-NEXT: seqz a0, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = shl i64 %x, 56 ; 64-8
+ %tmp1 = ashr exact i64 %tmp0, 56 ; 64-8
+ %tmp2 = icmp eq i64 %tmp1, %x
+ ret i1 %tmp2
+}
+
+; ---------------------------------------------------------------------------- ;
+; add + icmp uge
+; ---------------------------------------------------------------------------- ;
+
+define i1 @add_ugecmp_i16_i8(i16 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i16_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, -128
+; RV32-NEXT: slli a0, a0, 16
+; RV32-NEXT: srli a0, a0, 24
+; RV32-NEXT: sltiu a0, a0, 255
+; RV32-NEXT: xori a0, a0, 1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i16_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, -128
+; RV64-NEXT: slli a0, a0, 48
+; RV64-NEXT: srli a0, a0, 56
+; RV64-NEXT: sltiu a0, a0, 255
+; RV64-NEXT: xori a0, a0, 1
+; RV64-NEXT: ret
+ %tmp0 = add i16 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp uge i16 %tmp0, -256 ; ~0U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i32_i16_i8(i16 %xx) nounwind {
+; RV32I-LABEL: add_ugecmp_i32_i16_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: addi a0, a0, -128
+; RV32I-NEXT: sltiu a0, a0, -256
+; RV32I-NEXT: xori a0, a0, 1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_i32_i16_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: addi a0, a0, -128
+; RV64I-NEXT: sltiu a0, a0, -256
+; RV64I-NEXT: xori a0, a0, 1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ugecmp_i32_i16_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: addi a0, a0, -128
+; RV32ZBB-NEXT: sltiu a0, a0, -256
+; RV32ZBB-NEXT: xori a0, a0, 1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_i32_i16_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: addi a0, a0, -128
+; RV64ZBB-NEXT: sltiu a0, a0, -256
+; RV64ZBB-NEXT: xori a0, a0, 1
+; RV64ZBB-NEXT: ret
+ %x = zext i16 %xx to i32
+ %tmp0 = add i32 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp uge i32 %tmp0, -256 ; ~0U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i32_i16(i32 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i32_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a1, 1048568
+; RV32-NEXT: add a0, a0, a1
+; RV32-NEXT: lui a1, 1048560
+; RV32-NEXT: addi a1, a1, -1
+; RV32-NEXT: sltu a0, a1, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i32_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 1048568
+; RV64-NEXT: addw a0, a0, a1
+; RV64-NEXT: lui a1, 1048560
+; RV64-NEXT: addiw a1, a1, -1
+; RV64-NEXT: sltu a0, a1, a0
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, -32768 ; ~0U << (16-1)
+ %tmp1 = icmp uge i32 %tmp0, -65536 ; ~0U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i32_i8(i32 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i32_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, -128
+; RV32-NEXT: sltiu a0, a0, -256
+; RV32-NEXT: xori a0, a0, 1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i32_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, -128
+; RV64-NEXT: sltiu a0, a0, -256
+; RV64-NEXT: xori a0, a0, 1
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp uge i32 %tmp0, -256 ; ~0U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i64_i32(i64 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i64_i32:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 524288
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: ret
+;
+; RV64I-LABEL: add_ugecmp_i64_i32:
+; RV64I: # %bb.0:
+; RV64I-NEXT: lui a1, 524288
+; RV64I-NEXT: add a0, a0, a1
+; RV64I-NEXT: li a1, -1
+; RV64I-NEXT: slli a1, a1, 32
+; RV64I-NEXT: addi a1, a1, -1
+; RV64I-NEXT: sltu a0, a1, a0
+; RV64I-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ugecmp_i64_i32:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: lui a1, 524288
+; RV64ZBB-NEXT: add a0, a0, a1
+; RV64ZBB-NEXT: li a1, -2
+; RV64ZBB-NEXT: rori a1, a1, 32
+; RV64ZBB-NEXT: sltu a0, a1, a0
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i64 %x, -2147483648 ; ~0U << (32-1)
+ %tmp1 = icmp uge i64 %tmp0, -4294967296 ; ~0U << 32
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i64_i16(i64 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i64_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 1048568
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: lui a1, 1048560
+; RV32-NEXT: addi a1, a1, -1
+; RV32-NEXT: sltu a1, a1, a2
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: and a0, a0, a1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i64_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 1048568
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: lui a1, 1048560
+; RV64-NEXT: addiw a1, a1, -1
+; RV64-NEXT: sltu a0, a1, a0
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, -32768 ; ~0U << (16-1)
+ %tmp1 = icmp uge i64 %tmp0, -65536 ; ~0U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ugecmp_i64_i8(i64 %x) nounwind {
+; RV32-LABEL: add_ugecmp_i64_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a2, a0, -128
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: sltiu a1, a2, -256
+; RV32-NEXT: xori a1, a1, 1
+; RV32-NEXT: and a0, a0, a1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugecmp_i64_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addi a0, a0, -128
+; RV64-NEXT: sltiu a0, a0, -256
+; RV64-NEXT: xori a0, a0, 1
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp uge i64 %tmp0, -256 ; ~0U << 8
+ ret i1 %tmp1
+}
+
+; Slightly more canonical variant
+define i1 @add_ugtcmp_i16_i8(i16 %x) nounwind {
+; RV32-LABEL: add_ugtcmp_i16_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, -128
+; RV32-NEXT: slli a0, a0, 16
+; RV32-NEXT: srli a0, a0, 24
+; RV32-NEXT: sltiu a0, a0, 255
+; RV32-NEXT: xori a0, a0, 1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ugtcmp_i16_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, -128
+; RV64-NEXT: slli a0, a0, 48
+; RV64-NEXT: srli a0, a0, 56
+; RV64-NEXT: sltiu a0, a0, 255
+; RV64-NEXT: xori a0, a0, 1
+; RV64-NEXT: ret
+ %tmp0 = add i16 %x, -128 ; ~0U << (8-1)
+ %tmp1 = icmp ugt i16 %tmp0, -257 ; ~0U << 8 - 1
+ ret i1 %tmp1
+}
+
+; ---------------------------------------------------------------------------- ;
+; add + icmp ult
+; ---------------------------------------------------------------------------- ;
+
+define i1 @add_ultcmp_i16_i8(i16 %x) nounwind {
+; RV32I-LABEL: add_ultcmp_i16_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_i16_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_i16_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_i16_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ult i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i32_i16(i32 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i32_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a1, 8
+; RV32-NEXT: add a0, a0, a1
+; RV32-NEXT: srli a0, a0, 16
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i32_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 8
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: srliw a0, a0, 16
+; RV64-NEXT: seqz a0, a0
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, 32768 ; 1U << (16-1)
+ %tmp1 = icmp ult i32 %tmp0, 65536 ; 1U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i32_i8(i32 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i32_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, 128
+; RV32-NEXT: sltiu a0, a0, 256
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i32_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, 128
+; RV64-NEXT: sltiu a0, a0, 256
+; RV64-NEXT: ret
+ %tmp0 = add i32 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ult i32 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i64_i32(i64 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i64_i32:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 524288
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i64_i32:
+; RV64: # %bb.0:
+; RV64-NEXT: li a1, 1
+; RV64-NEXT: slli a1, a1, 31
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: srli a0, a0, 32
+; RV64-NEXT: seqz a0, a0
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, 2147483648 ; 1U << (32-1)
+ %tmp1 = icmp ult i64 %tmp0, 4294967296 ; 1U << 32
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i64_i16(i64 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i64_i16:
+; RV32: # %bb.0:
+; RV32-NEXT: lui a2, 8
+; RV32-NEXT: add a2, a0, a2
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: srli a2, a2, 16
+; RV32-NEXT: or a0, a0, a2
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i64_i16:
+; RV64: # %bb.0:
+; RV64-NEXT: lui a1, 8
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: srli a0, a0, 16
+; RV64-NEXT: seqz a0, a0
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, 32768 ; 1U << (16-1)
+ %tmp1 = icmp ult i64 %tmp0, 65536 ; 1U << 16
+ ret i1 %tmp1
+}
+
+define i1 @add_ultcmp_i64_i8(i64 %x) nounwind {
+; RV32-LABEL: add_ultcmp_i64_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a2, a0, 128
+; RV32-NEXT: sltu a0, a2, a0
+; RV32-NEXT: add a0, a1, a0
+; RV32-NEXT: seqz a0, a0
+; RV32-NEXT: sltiu a1, a2, 256
+; RV32-NEXT: and a0, a0, a1
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_i64_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addi a0, a0, 128
+; RV64-NEXT: sltiu a0, a0, 256
+; RV64-NEXT: ret
+ %tmp0 = add i64 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ult i64 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Slightly more canonical variant
+define i1 @add_ulecmp_i16_i8(i16 %x) nounwind {
+; RV32I-LABEL: add_ulecmp_i16_i8:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ulecmp_i16_i8:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ulecmp_i16_i8:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ulecmp_i16_i8:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ule i16 %tmp0, 255 ; (1U << 8) - 1
+ ret i1 %tmp1
+}
+
+; Negative tests
+; ---------------------------------------------------------------------------- ;
+
+; Adding not a constant
+define i1 @add_ultcmp_bad_i16_i8_add(i16 %x, i16 %y) nounwind {
+; RV32I-LABEL: add_ultcmp_bad_i16_i8_add:
+; RV32I: # %bb.0:
+; RV32I-NEXT: add a0, a0, a1
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_bad_i16_i8_add:
+; RV64I: # %bb.0:
+; RV64I-NEXT: add a0, a0, a1
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_bad_i16_i8_add:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: add a0, a0, a1
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_bad_i16_i8_add:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: add a0, a0, a1
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, %y
+ %tmp1 = icmp ult i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Comparing not with a constant
+define i1 @add_ultcmp_bad_i16_i8_cmp(i16 %x, i16 %y) nounwind {
+; RV32I-LABEL: add_ultcmp_bad_i16_i8_cmp:
+; RV32I: # %bb.0:
+; RV32I-NEXT: lui a2, 16
+; RV32I-NEXT: addi a2, a2, -1
+; RV32I-NEXT: and a1, a1, a2
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: and a0, a0, a2
+; RV32I-NEXT: sltu a0, a0, a1
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_bad_i16_i8_cmp:
+; RV64I: # %bb.0:
+; RV64I-NEXT: lui a2, 16
+; RV64I-NEXT: addiw a2, a2, -1
+; RV64I-NEXT: and a1, a1, a2
+; RV64I-NEXT: addi a0, a0, 128
+; RV64I-NEXT: and a0, a0, a2
+; RV64I-NEXT: sltu a0, a0, a1
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_bad_i16_i8_cmp:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: zext.h a1, a1
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltu a0, a0, a1
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_bad_i16_i8_cmp:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: zext.h a1, a1
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltu a0, a0, a1
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ult i16 %tmp0, %y
+ ret i1 %tmp1
+}
+
+; Second constant is not larger than the first one
+define i1 @add_ultcmp_bad_i8_i16(i16 %x) nounwind {
+; RV32I-LABEL: add_ultcmp_bad_i8_i16:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 128
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_bad_i8_i16:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 128
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_bad_i8_i16:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 128
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_bad_i8_i16:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 128
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ult i16 %tmp0, 128 ; 1U << (8-1)
+ ret i1 %tmp1
+}
+
+; First constant is not power of two
+define i1 @add_ultcmp_bad_i16_i8_c0notpoweroftwo(i16 %x) nounwind {
+; RV32I-LABEL: add_ultcmp_bad_i16_i8_c0notpoweroftwo:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 192
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_bad_i16_i8_c0notpoweroftwo:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 192
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_bad_i16_i8_c0notpoweroftwo:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 192
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_bad_i16_i8_c0notpoweroftwo:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 192
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 192 ; (1U << (8-1)) + (1U << (8-1-1))
+ %tmp1 = icmp ult i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Second constant is not power of two
+define i1 @add_ultcmp_bad_i16_i8_c1notpoweroftwo(i16 %x) nounwind {
+; RV32I-LABEL: add_ultcmp_bad_i16_i8_c1notpoweroftwo:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 128
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 768
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_bad_i16_i8_c1notpoweroftwo:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 128
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 768
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_bad_i16_i8_c1notpoweroftwo:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 128
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 768
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_bad_i16_i8_c1notpoweroftwo:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 128
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 768
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ult i16 %tmp0, 768 ; (1U << 8)) + (1U << (8+1))
+ ret i1 %tmp1
+}
+
+; Magic check fails, 64 << 1 != 256
+define i1 @add_ultcmp_bad_i16_i8_magic(i16 %x) nounwind {
+; RV32I-LABEL: add_ultcmp_bad_i16_i8_magic:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 64
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 256
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_bad_i16_i8_magic:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 64
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 256
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_bad_i16_i8_magic:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 64
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 256
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_bad_i16_i8_magic:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 64
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 256
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 64 ; 1U << (8-1-1)
+ %tmp1 = icmp ult i16 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+; Bad 'destination type'
+define i1 @add_ultcmp_bad_i16_i4(i16 %x) nounwind {
+; RV32I-LABEL: add_ultcmp_bad_i16_i4:
+; RV32I: # %bb.0:
+; RV32I-NEXT: addi a0, a0, 8
+; RV32I-NEXT: slli a0, a0, 16
+; RV32I-NEXT: srli a0, a0, 16
+; RV32I-NEXT: sltiu a0, a0, 16
+; RV32I-NEXT: ret
+;
+; RV64I-LABEL: add_ultcmp_bad_i16_i4:
+; RV64I: # %bb.0:
+; RV64I-NEXT: addiw a0, a0, 8
+; RV64I-NEXT: slli a0, a0, 48
+; RV64I-NEXT: srli a0, a0, 48
+; RV64I-NEXT: sltiu a0, a0, 16
+; RV64I-NEXT: ret
+;
+; RV32ZBB-LABEL: add_ultcmp_bad_i16_i4:
+; RV32ZBB: # %bb.0:
+; RV32ZBB-NEXT: addi a0, a0, 8
+; RV32ZBB-NEXT: zext.h a0, a0
+; RV32ZBB-NEXT: sltiu a0, a0, 16
+; RV32ZBB-NEXT: ret
+;
+; RV64ZBB-LABEL: add_ultcmp_bad_i16_i4:
+; RV64ZBB: # %bb.0:
+; RV64ZBB-NEXT: addiw a0, a0, 8
+; RV64ZBB-NEXT: zext.h a0, a0
+; RV64ZBB-NEXT: sltiu a0, a0, 16
+; RV64ZBB-NEXT: ret
+ %tmp0 = add i16 %x, 8 ; 1U << (4-1)
+ %tmp1 = icmp ult i16 %tmp0, 16 ; 1U << 4
+ ret i1 %tmp1
+}
+
+; Bad storage type
+define i1 @add_ultcmp_bad_i24_i8(i24 %x) nounwind {
+; RV32-LABEL: add_ultcmp_bad_i24_i8:
+; RV32: # %bb.0:
+; RV32-NEXT: addi a0, a0, 128
+; RV32-NEXT: slli a0, a0, 8
+; RV32-NEXT: srli a0, a0, 8
+; RV32-NEXT: sltiu a0, a0, 256
+; RV32-NEXT: ret
+;
+; RV64-LABEL: add_ultcmp_bad_i24_i8:
+; RV64: # %bb.0:
+; RV64-NEXT: addiw a0, a0, 128
+; RV64-NEXT: slli a0, a0, 40
+; RV64-NEXT: srli a0, a0, 40
+; RV64-NEXT: sltiu a0, a0, 256
+; RV64-NEXT: ret
+ %tmp0 = add i24 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ult i24 %tmp0, 256 ; 1U << 8
+ ret i1 %tmp1
+}
+
+define i1 @add_ulecmp_bad_i16_i8(i16 %x) nounwind {
+; CHECK-LABEL: add_ulecmp_bad_i16_i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: li a0, 1
+; CHECK-NEXT: ret
+ %tmp0 = add i16 %x, 128 ; 1U << (8-1)
+ %tmp1 = icmp ule i16 %tmp0, -1 ; when we +1 it, it will wrap to 0
+ ret i1 %tmp1
+}
More information about the llvm-commits
mailing list