Skip to content

[X86] EmitCmp - use existing XOR node to check for equality #125506

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 2 commits into from
Feb 4, 2025
Merged

[X86] EmitCmp - use existing XOR node to check for equality #125506

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e58edd4
[X86] EmitCmp - use existing XOR node to check for equality
RKSimon Feb 3, 2025
2410d38
clang-format
RKSimon Feb 3, 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
13 changes: 10 additions & 3 deletions llvm/lib/Target/X86/X86ISelLowering.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23157,10 +23157,17 @@ static SDValue EmitCmp(SDValue Op0, SDValue Op1, unsigned X86CC,
return Add.getValue(1);
}

// Use SUB instead of CMP to enable CSE between SUB and CMP.
// If we already have an XOR of the ops, use that to check for equality.
// Else use SUB instead of CMP to enable CSE between SUB and CMP.
unsigned X86Opc = X86ISD::SUB;
if ((X86CC == X86::COND_E || X86CC == X86::COND_NE) &&
(DAG.doesNodeExist(ISD::XOR, DAG.getVTList({CmpVT}), {Op0, Op1}) ||
DAG.doesNodeExist(ISD::XOR, DAG.getVTList({CmpVT}), {Op1, Op0})))
X86Opc = X86ISD::XOR;

SDVTList VTs = DAG.getVTList(CmpVT, MVT::i32);
SDValue Sub = DAG.getNode(X86ISD::SUB, dl, VTs, Op0, Op1);
return Sub.getValue(1);
SDValue CmpOp = DAG.getNode(X86Opc, dl, VTs, Op0, Op1);
return CmpOp.getValue(1);
}

bool X86TargetLowering::isXAndYEqZeroPreferableToXAndYEqY(ISD::CondCode Cond,
Expand Down
28 changes: 10 additions & 18 deletions llvm/test/CodeGen/X86/cmp-xor.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,22 +9,18 @@
define i32 @cmp_xor_i32(i32 %a, i32 %b, i32 %c)
; X86-LABEL: cmp_xor_i32:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: cmpl %ecx, %eax
; X86-NEXT: je .LBB0_1
; X86-NEXT: # %bb.2:
; X86-NEXT: xorl %ecx, %eax
; X86-NEXT: retl
; X86-NEXT: .LBB0_1:
; X86-NEXT: xorl {{[0-9]+}}(%esp), %eax
; X86-NEXT: jne .LBB0_2
; X86-NEXT: # %bb.1:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: .LBB0_2:
; X86-NEXT: retl
;
; X64-LABEL: cmp_xor_i32:
; X64: # %bb.0:
; X64-NEXT: movl %edi, %eax
; X64-NEXT: xorl %esi, %eax
; X64-NEXT: cmpl %esi, %edi
; X64-NEXT: cmovel %edx, %eax
; X64-NEXT: retq
{
Expand All @@ -37,22 +33,18 @@ define i32 @cmp_xor_i32(i32 %a, i32 %b, i32 %c)
define i32 @cmp_xor_i32_commute(i32 %a, i32 %b, i32 %c)
; X86-LABEL: cmp_xor_i32_commute:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: cmpl %eax, %ecx
; X86-NEXT: je .LBB1_1
; X86-NEXT: # %bb.2:
; X86-NEXT: xorl %ecx, %eax
; X86-NEXT: retl
; X86-NEXT: .LBB1_1:
; X86-NEXT: xorl {{[0-9]+}}(%esp), %eax
; X86-NEXT: jne .LBB1_2
; X86-NEXT: # %bb.1:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: .LBB1_2:
; X86-NEXT: retl
;
; X64-LABEL: cmp_xor_i32_commute:
; X64: # %bb.0:
; X64-NEXT: movl %esi, %eax
; X64-NEXT: xorl %edi, %eax
; X64-NEXT: cmpl %esi, %edi
; X64-NEXT: movl %edi, %eax
; X64-NEXT: xorl %esi, %eax
; X64-NEXT: cmovel %edx, %eax
; X64-NEXT: retq
{
Expand Down
22 changes: 9 additions & 13 deletions llvm/test/CodeGen/X86/pr32284.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -321,11 +321,9 @@ define void @f2() {
; X64-NEXT: xorl %ecx, %ecx
; X64-NEXT: testl %eax, %eax
; X64-NEXT: sete %cl
; X64-NEXT: movl %eax, %edx
; X64-NEXT: xorl %ecx, %edx
; X64-NEXT: movw %dx, -{{[0-9]+}}(%rsp)
; X64-NEXT: xorl %edx, %edx
; X64-NEXT: cmpl %eax, %ecx
; X64-NEXT: xorl %eax, %ecx
; X64-NEXT: movw %cx, -{{[0-9]+}}(%rsp)
; X64-NEXT: sete %dl
; X64-NEXT: movw %dx, (%rax)
; X64-NEXT: retq
Expand Down Expand Up @@ -366,17 +364,15 @@ define void @f2() {
; X86: # %bb.0: # %entry
; X86-NEXT: subl $2, %esp
; X86-NEXT: .cfi_def_cfa_offset 6
; X86-NEXT: movzbl var_7, %ecx
; X86-NEXT: movzbl var_7, %edx
; X86-NEXT: xorl %eax, %eax
; X86-NEXT: testl %ecx, %ecx
; X86-NEXT: testl %edx, %edx
; X86-NEXT: sete %al
; X86-NEXT: movl %ecx, %edx
; X86-NEXT: xorl %eax, %edx
; X86-NEXT: movw %dx, (%esp)
; X86-NEXT: xorl %edx, %edx
; X86-NEXT: cmpl %ecx, %eax
; X86-NEXT: sete %dl
; X86-NEXT: movw %dx, (%eax)
; X86-NEXT: xorl %ecx, %ecx
; X86-NEXT: xorl %edx, %eax
; X86-NEXT: movw %ax, (%esp)
; X86-NEXT: sete %cl
; X86-NEXT: movw %cx, (%eax)
; X86-NEXT: addl $2, %esp
; X86-NEXT: .cfi_def_cfa_offset 4
; X86-NEXT: retl
Expand Down