[X86][Inline] Don't inline callee with cx16 if caller is without cx16

[usamoi]

closes #187503

[github-actions]

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

[llvmbot]

@llvm/pr-subscribers-llvm-transforms

@llvm/pr-subscribers-backend-x86

Author: None (usamoi)

Changes

closes #187503

---

Full diff: https://github.com/llvm/llvm-project/pull/187505.diff

1 Files Affected:

• (modified) llvm/lib/Target/X86/X86TargetTransformInfo.h (-1)

diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.h b/llvm/lib/Target/X86/X86TargetTransformInfo.h
index b3dde1555d0a0..0acf37939e2df 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.h
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.h
@@ -43,7 +43,6 @@ class X86TTIImpl final : public BasicTTIImplBase<X86TTIImpl> {
 
       // These features don't have any intrinsics or ABI effect.
       X86::FeatureNOPL,
-      X86::FeatureCX16,
       X86::FeatureLAHFSAHF64,
 
       // Some older targets can be setup to fold unaligned loads.

[github-actions]

⚠️ C/C++ code formatter, clang-format found issues in your code. ⚠️

You can test this locally with the following command:

git-clang-format --diff origin/main HEAD --extensions h -- llvm/lib/Target/X86/X86TargetTransformInfo.h --diff_from_common_commit

⚠️
The reproduction instructions above might return results for more than one PR
in a stack if you are using a stacked PR workflow. You can limit the results by
changing origin/main to the base branch/commit you want to compare against.
⚠️

View the diff from clang-format here.

diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.h b/llvm/lib/Target/X86/X86TargetTransformInfo.h
index 0acf37939..9cb81d38b 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.h
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.h
@@ -42,71 +42,43 @@ class X86TTIImpl final : public BasicTTIImplBase<X86TTIImpl> {
       X86::FeatureX86_64,
 
       // These features don't have any intrinsics or ABI effect.
-      X86::FeatureNOPL,
-      X86::FeatureLAHFSAHF64,
+      X86::FeatureNOPL, X86::FeatureLAHFSAHF64,
 
       // Some older targets can be setup to fold unaligned loads.
       X86::FeatureSSEUnalignedMem,
 
       // Codegen control options.
-      X86::TuningFast11ByteNOP,
-      X86::TuningFast15ByteNOP,
-      X86::TuningFastBEXTR,
-      X86::TuningFastHorizontalOps,
-      X86::TuningFastLZCNT,
-      X86::TuningFastScalarFSQRT,
-      X86::TuningFastSHLDRotate,
-      X86::TuningFastScalarShiftMasks,
-      X86::TuningFastVectorShiftMasks,
+      X86::TuningFast11ByteNOP, X86::TuningFast15ByteNOP, X86::TuningFastBEXTR,
+      X86::TuningFastHorizontalOps, X86::TuningFastLZCNT,
+      X86::TuningFastScalarFSQRT, X86::TuningFastSHLDRotate,
+      X86::TuningFastScalarShiftMasks, X86::TuningFastVectorShiftMasks,
       X86::TuningFastVariableCrossLaneShuffle,
-      X86::TuningFastVariablePerLaneShuffle,
-      X86::TuningFastVectorFSQRT,
-      X86::TuningLEAForSP,
-      X86::TuningLEAUsesAG,
-      X86::TuningLZCNTFalseDeps,
-      X86::TuningBranchFusion,
-      X86::TuningMacroFusion,
-      X86::TuningPadShortFunctions,
-      X86::TuningPOPCNTFalseDeps,
-      X86::TuningMULCFalseDeps,
-      X86::TuningPERMFalseDeps,
-      X86::TuningRANGEFalseDeps,
-      X86::TuningGETMANTFalseDeps,
-      X86::TuningMULLQFalseDeps,
-      X86::TuningSlow3OpsLEA,
-      X86::TuningSlowDivide32,
-      X86::TuningSlowDivide64,
-      X86::TuningSlowIncDec,
-      X86::TuningSlowLEA,
-      X86::TuningSlowPMADDWD,
-      X86::TuningSlowPMULLD,
-      X86::TuningSlowSHLD,
-      X86::TuningSlowTwoMemOps,
-      X86::TuningSlowUAMem16,
-      X86::TuningPreferMaskRegisters,
-      X86::TuningInsertVZEROUPPER,
-      X86::TuningUseSLMArithCosts,
-      X86::TuningUseGLMDivSqrtCosts,
-      X86::TuningNoDomainDelay,
-      X86::TuningNoDomainDelayMov,
-      X86::TuningNoDomainDelayShuffle,
-      X86::TuningNoDomainDelayBlend,
-      X86::TuningPreferShiftShuffle,
-      X86::TuningFastImmVectorShift,
+      X86::TuningFastVariablePerLaneShuffle, X86::TuningFastVectorFSQRT,
+      X86::TuningLEAForSP, X86::TuningLEAUsesAG, X86::TuningLZCNTFalseDeps,
+      X86::TuningBranchFusion, X86::TuningMacroFusion,
+      X86::TuningPadShortFunctions, X86::TuningPOPCNTFalseDeps,
+      X86::TuningMULCFalseDeps, X86::TuningPERMFalseDeps,
+      X86::TuningRANGEFalseDeps, X86::TuningGETMANTFalseDeps,
+      X86::TuningMULLQFalseDeps, X86::TuningSlow3OpsLEA,
+      X86::TuningSlowDivide32, X86::TuningSlowDivide64, X86::TuningSlowIncDec,
+      X86::TuningSlowLEA, X86::TuningSlowPMADDWD, X86::TuningSlowPMULLD,
+      X86::TuningSlowSHLD, X86::TuningSlowTwoMemOps, X86::TuningSlowUAMem16,
+      X86::TuningPreferMaskRegisters, X86::TuningInsertVZEROUPPER,
+      X86::TuningUseSLMArithCosts, X86::TuningUseGLMDivSqrtCosts,
+      X86::TuningNoDomainDelay, X86::TuningNoDomainDelayMov,
+      X86::TuningNoDomainDelayShuffle, X86::TuningNoDomainDelayBlend,
+      X86::TuningPreferShiftShuffle, X86::TuningFastImmVectorShift,
       X86::TuningFastDPWSSD,
 
       // Perf-tuning flags.
-      X86::TuningFastGather,
-      X86::TuningSlowUAMem32,
+      X86::TuningFastGather, X86::TuningSlowUAMem32,
       X86::TuningAllowLight256Bit,
 
       // Based on whether user set the -mprefer-vector-width command line.
-      X86::TuningPrefer128Bit,
-      X86::TuningPrefer256Bit,
+      X86::TuningPrefer128Bit, X86::TuningPrefer256Bit,
 
       // CPU name enums. These just follow CPU string.
-      X86::ProcIntelAtom
-  };
+      X86::ProcIntelAtom};
 
 public:
   explicit X86TTIImpl(const X86TargetMachine *TM, const Function &F)

[github-actions]

🐧 Linux x64 Test Results

• 192484 tests passed
• 4940 tests skipped

✅ The build succeeded and all tests passed.

[github-actions]

🪟 Windows x64 Test Results

• 132522 tests passed
• 3018 tests skipped

✅ The build succeeded and all tests passed.

[nikic]

I agree that ignoring cx16 during inlining is invalid. Whether or not libatomic is used for an access is an ABI effect.

This does seem to defeat the original intent behind those tests though.