Skip to content

[X86] Use pseudo instructions to zero registers in buildClearRegister #163358

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 8 commits into from
Oct 17, 2025
Merged

[X86] Use pseudo instructions to zero registers in buildClearRegister #163358

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
b0dc506
[X86] Fix zeroing idioms for YMM and ZMM in buildClearRegister
abhishek-kaushik22 Oct 14, 2025
f60975b
Merge branch 'main' into xor
abhishek-kaushik22 Oct 14, 2025
b1500b2
Use psuedo instructions and add new tests
abhishek-kaushik22 Oct 16, 2025
be0d6f5
Use common prefixes
abhishek-kaushik22 Oct 16, 2025
f8ef59a
Merge branch 'main' into xor
RKSimon Oct 17, 2025
b427f6f
Merge branch 'main' into xor
abhishek-kaushik22 Oct 17, 2025
74098c7
Merge branch 'main' into xor
abhishek-kaushik22 Oct 17, 2025
bb1cca6
Merge branch 'main' into xor
RKSimon Oct 17, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions llvm/lib/Target/X86/X86InstrInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10818,17 +10818,17 @@ void X86InstrInfo::buildClearRegister(Register Reg, MachineBasicBlock &MBB,
if (!ST.hasAVX())
return;

// VPXOR is safe to use because it doesn't affect flags.
BuildMI(MBB, Iter, DL, get(X86::VPXORrr), Reg)
// VPXORY is safe to use because it doesn't affect flags.
BuildMI(MBB, Iter, DL, get(X86::VPXORYrr), Reg)
.addReg(Reg, RegState::Undef)
.addReg(Reg, RegState::Undef);
} else if (X86::VR512RegClass.contains(Reg)) {
// ZMM#
if (!ST.hasAVX512())
return;

// VPXORY is safe to use because it doesn't affect flags.
BuildMI(MBB, Iter, DL, get(X86::VPXORYrr), Reg)
// VPXORDZ is safe to use because it doesn't affect flags.
BuildMI(MBB, Iter, DL, get(X86::VPXORDZrr), Reg)
.addReg(Reg, RegState::Undef)
.addReg(Reg, RegState::Undef);
} else if (X86::VK1RegClass.contains(Reg) || X86::VK2RegClass.contains(Reg) ||
Expand Down
22 changes: 22 additions & 0 deletions llvm/test/CodeGen/X86/pr163053.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,22 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc < %s -mtriple=x86_64-- -mcpu=x86-64-v4 -verify-machineinstrs | FileCheck %s

define void @pr163053(ptr %arg, ptr %arg1) #0 {
; CHECK-LABEL: pr163053:
; CHECK: # %bb.0:
; CHECK-NEXT: vxorps %xmm0, %xmm0, %xmm0
; CHECK-NEXT: vmovaps %ymm0, (%rdi)
; CHECK-NEXT: vxorps %xmm0, %xmm0, %xmm0
; CHECK-NEXT: vmovaps %zmm0, (%rsi)
; CHECK-NEXT: xorl %edi, %edi
; CHECK-NEXT: xorl %esi, %esi
; CHECK-NEXT: vxorps %ymm0, %ymm0, %ymm0
; CHECK-NEXT: vxorps %zmm0, %zmm0, %zmm0
; CHECK-NEXT: vzeroupper
; CHECK-NEXT: retq
store <16 x i16> zeroinitializer, ptr %arg, align 32
store <32 x i16> zeroinitializer, ptr %arg1, align 64
ret void
}

attributes #0 = { "zero-call-used-regs"="used" }