Skip to content
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

[X86][GlobalISel] Added support for SQRT function #132356

Merged
merged 3 commits into from
Mar 26, 2025
Merged

[X86][GlobalISel] Added support for SQRT function #132356

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
8bb7fed
[X86][GlobalISel] Added support for SQRT function
JaydeepChauhan14 Mar 21, 2025
120859c
Updated the testcase check
JaydeepChauhan14 Mar 21, 2025
2c82bf5
Update llvm/test/CodeGen/X86/GlobalISel/sqrt.mir
JaydeepChauhan14 Mar 23, 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
5 changes: 5 additions & 0 deletions llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,6 +105,11 @@ X86LegalizerInfo::X86LegalizerInfo(const X86Subtarget &STI,
G_FEXP10, G_FLOG, G_FLOG2, G_FLOG10})
.libcall();

getActionDefinitionsBuilder(G_FSQRT)
.legalFor(HasSSE1 || UseX87, {s32})
.legalFor(HasSSE2 || UseX87, {s64})
.legalFor(UseX87, {s80});

// merge/unmerge
for (unsigned Op : {G_MERGE_VALUES, G_UNMERGE_VALUES}) {
unsigned BigTyIdx = Op == G_MERGE_VALUES ? 0 : 1;
Expand Down
1 change: 1 addition & 0 deletions llvm/lib/Target/X86/GISel/X86RegisterBankInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -288,6 +288,7 @@ X86RegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
SmallVector<PartialMappingIdx, 4> OpRegBankIdx(NumOperands);

switch (Opc) {
case TargetOpcode::G_FSQRT:
case TargetOpcode::G_FPEXT:
case TargetOpcode::G_FPTRUNC:
case TargetOpcode::G_FCONSTANT:
Expand Down
68 changes: 68 additions & 0 deletions llvm/test/CodeGen/X86/GlobalISel/sqrt.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,68 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
# RUN: llc -mtriple=i686-linux-gnu -run-pass=regbankselect,instruction-select %s -o - | FileCheck %s --check-prefixes GISEL-I686

---
name: test_sqrt_f32
alignment: 16
legalized: true
fixedStack:
- { id: 0, type: default, offset: 0, size: 4, alignment: 16, stack-id: default,
isImmutable: true, isAliased: false, callee-saved-register: '', callee-saved-restored: true,
debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
body: |
bb.1:
; GISEL-I686-LABEL: name: test_sqrt_f32
; GISEL-I686: [[LD_Fp32m:%[0-9]+]]:rfp32 = LD_Fp32m %fixed-stack.0, 1, $noreg, 0, $noreg, implicit-def $fpsw, implicit $fpcw :: (invariant load (s32) from %fixed-stack.0, align 16)
; GISEL-I686-NEXT: [[SQRT_Fp32_:%[0-9]+]]:rfp32 = nofpexcept SQRT_Fp32 [[LD_Fp32m]], implicit-def dead $fpsw, implicit $fpcw
; GISEL-I686-NEXT: $fp0 = COPY [[SQRT_Fp32_]]
; GISEL-I686-NEXT: RET 0, implicit $fp0
%1:_(p0) = G_FRAME_INDEX %fixed-stack.0
%0:_(s32) = G_LOAD %1(p0) :: (invariant load (s32) from %fixed-stack.0, align 16)
%2:_(s32) = G_FSQRT %0
$fp0 = COPY %2(s32)
RET 0, implicit $fp0

...
---
name: test_sqrt_f64
alignment: 16
legalized: true
fixedStack:
- { id: 0, type: default, offset: 0, size: 8, alignment: 16, stack-id: default,
isImmutable: true, isAliased: false, callee-saved-register: '', callee-saved-restored: true,
debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
body: |
bb.1:
; GISEL-I686-LABEL: name: test_sqrt_f64
; GISEL-I686: [[DEF:%[0-9]+]]:rfp64 = IMPLICIT_DEF
; GISEL-I686-NEXT: [[SQRT_Fp64_:%[0-9]+]]:rfp64 = nofpexcept SQRT_Fp64 [[DEF]], implicit-def dead $fpsw, implicit $fpcw
; GISEL-I686-NEXT: $fp0 = COPY [[SQRT_Fp64_]]
; GISEL-I686-NEXT: RET 0, implicit $fp0
%0:_(s64) = IMPLICIT_DEF
%2:_(s64) = G_FSQRT %0
$fp0 = COPY %2(s64)
RET 0, implicit $fp0

...
---
name: test_sqrt_f80
alignment: 16
legalized: true
fixedStack:
- { id: 0, type: default, offset: 0, size: 10, alignment: 16, stack-id: default,
isImmutable: true, isAliased: false, callee-saved-register: '', callee-saved-restored: true,
debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
body: |
bb.1:
; GISEL-I686-LABEL: name: test_sqrt_f80
; GISEL-I686: [[LD_Fp80m:%[0-9]+]]:rfp80 = LD_Fp80m %fixed-stack.0, 1, $noreg, 0, $noreg, implicit-def $fpsw, implicit $fpcw :: (invariant load (s80) from %fixed-stack.0, align 16)
; GISEL-I686-NEXT: [[SQRT_Fp80_:%[0-9]+]]:rfp80 = nofpexcept SQRT_Fp80 [[LD_Fp80m]], implicit-def dead $fpsw, implicit $fpcw
; GISEL-I686-NEXT: $fp0 = COPY [[SQRT_Fp80_]]
; GISEL-I686-NEXT: RET 0, implicit $fp0
%1:_(p0) = G_FRAME_INDEX %fixed-stack.0
%0:_(s80) = G_LOAD %1(p0) :: (invariant load (s80) from %fixed-stack.0, align 16)
%2:_(s80) = G_FSQRT %0
$fp0 = COPY %2(s80)
RET 0, implicit $fp0

...
19 changes: 10 additions & 9 deletions llvm/test/CodeGen/X86/isel-sqrt.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,13 +1,14 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx,+sse2 | FileCheck %s --check-prefixes=X64,SSE2
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx,+sse2 -fast-isel | FileCheck %s --check-prefixes=X64,SSE2
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx,+sse2 -global-isel -global-isel-abort=2 | FileCheck %s --check-prefixes=X64,SSE2
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx2,+avx | FileCheck %s --check-prefixes=X64,AVX
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx2,+avx -fast-isel | FileCheck %s --check-prefixes=X64,AVX
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx2,+avx -global-isel -global-isel-abort=2 | FileCheck %s --check-prefixes=X64,AVX
; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel=0 -fast-isel=0 | FileCheck %s --check-prefixes=X86,SDAG-X86
; RUN: llc < %s -mtriple=i686-linux-gnu -fast-isel | FileCheck %s --check-prefixes=X86,FASTISEL-X86
; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel -global-isel-abort=2 | FileCheck %s --check-prefixes=X86,GISEL-X86
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx,+sse2 | FileCheck %s --check-prefixes=X64,SSE2
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx,+sse2 -fast-isel | FileCheck %s --check-prefixes=X64,SSE2
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx,+sse2 -global-isel -global-isel-abort=1 | FileCheck %s --check-prefixes=X64,SSE2
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx2,+avx | FileCheck %s --check-prefixes=X64,AVX
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx2,+avx -fast-isel | FileCheck %s --check-prefixes=X64,AVX
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=-avx2,+avx -global-isel -global-isel-abort=1 | FileCheck %s --check-prefixes=X64,AVX
; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel=0 -fast-isel=0 | FileCheck %s --check-prefixes=X86,SDAG-X86
; RUN: llc < %s -mtriple=i686-linux-gnu -fast-isel | FileCheck %s --check-prefixes=X86,FASTISEL-X86
; TODO: The last RUN line will fails GISEL selection and will fallback to DAG selection due to lack of support for loads/stores in i686 mode, support is expected soon enough, for this reason the llvm/test/CodeGen/X86/GlobalISel/sqrt.mir test is added for now because of the lack of support for i686 in GlobalISel.
; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel -global-isel-abort=2 | FileCheck %s --check-prefixes=X86,GISEL-X86

define float @test_sqrt_f32(float %a) {
; SSE2-LABEL: test_sqrt_f32:
Expand Down