[LLVM][VecLib] Refactor LIBMVEC integration to be target neutral. #138262

paulwalker-arm · 2025-05-02T12:44:38Z

No description provided.

llvmbot · 2025-05-02T12:45:17Z

@llvm/pr-subscribers-clang
@llvm/pr-subscribers-clang-driver
@llvm/pr-subscribers-llvm-transforms

@llvm/pr-subscribers-llvm-analysis

Author: Paul Walker (paulwalker-arm)

Changes

Full diff: https://github.com/llvm/llvm-project/pull/138262.diff

8 Files Affected:

(modified) llvm/include/llvm/Analysis/TargetLibraryInfo.h (+1-1)
(modified) llvm/lib/Analysis/TargetLibraryInfo.cpp (+10-3)
(modified) llvm/lib/Frontend/Driver/CodeGenOptions.cpp (+1-1)
(modified) llvm/test/CodeGen/Generic/replace-intrinsics-with-veclib.ll (+1-1)
(modified) llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-VF2-VF8.ll (+1-1)
(modified) llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-finite.ll (+1-1)
(modified) llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls.ll (+1-1)
(modified) llvm/test/Transforms/Util/add-TLI-mappings.ll (+7-1)

diff --git a/llvm/include/llvm/Analysis/TargetLibraryInfo.h b/llvm/include/llvm/Analysis/TargetLibraryInfo.h
index 33f2ab4fa90b0..4c23eaad2ae28 100644
--- a/llvm/include/llvm/Analysis/TargetLibraryInfo.h
+++ b/llvm/include/llvm/Analysis/TargetLibraryInfo.h
@@ -127,7 +127,7 @@ class TargetLibraryInfoImpl {
     NoLibrary,        // Don't use any vector library.
     Accelerate,       // Use Accelerate framework.
     DarwinLibSystemM, // Use Darwin's libsystem_m.
-    LIBMVEC_X86,      // GLIBC Vector Math library.
+    LIBMVEC,          // GLIBC Vector Math library.
     MASSV,            // IBM MASS vector library.
     SVML,             // Intel short vector math library.
     SLEEFGNUABI, // SLEEF - SIMD Library for Evaluating Elementary Functions.
diff --git a/llvm/lib/Analysis/TargetLibraryInfo.cpp b/llvm/lib/Analysis/TargetLibraryInfo.cpp
index e2fd2aa13cce7..3945dd4a8b55d 100644
--- a/llvm/lib/Analysis/TargetLibraryInfo.cpp
+++ b/llvm/lib/Analysis/TargetLibraryInfo.cpp
@@ -29,7 +29,7 @@ static cl::opt<TargetLibraryInfoImpl::VectorLibrary> ClVectorLibrary(
                           "Accelerate framework"),
                clEnumValN(TargetLibraryInfoImpl::DarwinLibSystemM,
                           "Darwin_libsystem_m", "Darwin libsystem_m"),
-               clEnumValN(TargetLibraryInfoImpl::LIBMVEC_X86, "LIBMVEC-X86",
+               clEnumValN(TargetLibraryInfoImpl::LIBMVEC, "LIBMVEC",
                           "GLIBC Vector Math library"),
                clEnumValN(TargetLibraryInfoImpl::MASSV, "MASSV",
                           "IBM MASS vector library"),
@@ -1360,8 +1360,15 @@ void TargetLibraryInfoImpl::addVectorizableFunctionsFromVecLib(
     addVectorizableFunctions(VecFuncs_DarwinLibSystemM);
     break;
   }
-  case LIBMVEC_X86: {
-    addVectorizableFunctions(VecFuncs_LIBMVEC_X86);
+  case LIBMVEC: {
+    switch (TargetTriple.getArch()) {
+    default:
+      break;
+    case llvm::Triple::x86:
+    case llvm::Triple::x86_64:
+      addVectorizableFunctions(VecFuncs_LIBMVEC_X86);
+      break;
+    }
     break;
   }
   case MASSV: {
diff --git a/llvm/lib/Frontend/Driver/CodeGenOptions.cpp b/llvm/lib/Frontend/Driver/CodeGenOptions.cpp
index ed7c57a930aca..52080dea93c98 100644
--- a/llvm/lib/Frontend/Driver/CodeGenOptions.cpp
+++ b/llvm/lib/Frontend/Driver/CodeGenOptions.cpp
@@ -23,7 +23,7 @@ TargetLibraryInfoImpl *createTLII(const llvm::Triple &TargetTriple,
                                              TargetTriple);
     break;
   case VectorLibrary::LIBMVEC:
-    TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::LIBMVEC_X86,
+    TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::LIBMVEC,
                                              TargetTriple);
     break;
   case VectorLibrary::MASSV:
diff --git a/llvm/test/CodeGen/Generic/replace-intrinsics-with-veclib.ll b/llvm/test/CodeGen/Generic/replace-intrinsics-with-veclib.ll
index fde6cb788b46f..ff9c7486c099e 100644
--- a/llvm/test/CodeGen/Generic/replace-intrinsics-with-veclib.ll
+++ b/llvm/test/CodeGen/Generic/replace-intrinsics-with-veclib.ll
@@ -1,7 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes
 ; RUN: opt -vector-library=SVML -replace-with-veclib -S < %s | FileCheck %s  --check-prefixes=COMMON,SVML
 ; RUN: opt -vector-library=AMDLIBM -replace-with-veclib -S < %s | FileCheck %s  --check-prefixes=COMMON,AMDLIBM
-; RUN: opt -vector-library=LIBMVEC-X86 -replace-with-veclib -S < %s | FileCheck %s  --check-prefixes=COMMON,LIBMVEC-X86
+; RUN: opt -vector-library=LIBMVEC -replace-with-veclib -S < %s | FileCheck %s  --check-prefixes=COMMON,LIBMVEC-X86
 ; RUN: opt -vector-library=MASSV -replace-with-veclib -S < %s | FileCheck %s  --check-prefixes=COMMON,MASSV
 ; RUN: opt -vector-library=Accelerate -replace-with-veclib -S < %s | FileCheck %s  --check-prefixes=COMMON,ACCELERATE
 
diff --git a/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-VF2-VF8.ll b/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-VF2-VF8.ll
index 67a2cf2b80e70..91d5c52fa43c6 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-VF2-VF8.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-VF2-VF8.ll
@@ -1,4 +1,4 @@
-; RUN: opt -vector-library=LIBMVEC-X86 -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
+; RUN: opt -vector-library=LIBMVEC -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
diff --git a/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-finite.ll b/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-finite.ll
index d0d0d78a0d27e..bdb89fbbaa847 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-finite.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls-finite.ll
@@ -1,4 +1,4 @@
-; RUN: opt -vector-library=LIBMVEC-X86 -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
+; RUN: opt -vector-library=LIBMVEC -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
diff --git a/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls.ll b/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls.ll
index 7a0e44c9e9916..e0661301fd90c 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/libm-vector-calls.ll
@@ -1,4 +1,4 @@
-; RUN: opt -vector-library=LIBMVEC-X86 -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
+; RUN: opt -vector-library=LIBMVEC -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
diff --git a/llvm/test/Transforms/Util/add-TLI-mappings.ll b/llvm/test/Transforms/Util/add-TLI-mappings.ll
index b7eef89304c0d..a1f660d31668e 100644
--- a/llvm/test/Transforms/Util/add-TLI-mappings.ll
+++ b/llvm/test/Transforms/Util/add-TLI-mappings.ll
@@ -1,12 +1,15 @@
 ; RUN: opt -mtriple=x86_64-unknown-linux-gnu -vector-library=SVML -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,SVML
 ; RUN: opt -mtriple=x86_64-unknown-linux-gnu -vector-library=AMDLIBM -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,AMDLIBM
 ; RUN: opt -mtriple=powerpc64-unknown-linux-gnu -vector-library=MASSV -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,MASSV
-; RUN: opt -mtriple=x86_64-unknown-linux-gnu -vector-library=LIBMVEC-X86 -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,LIBMVEC-X86
+; RUN: opt -mtriple=aarch64-unknown-linux-gnu -vector-library=LIBMVEC -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=LIBMVEC-AARCH64
+; RUN: opt -mtriple=x86_64-unknown-linux-gnu -vector-library=LIBMVEC -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,LIBMVEC-X86
 ; RUN: opt -mtriple=x86_64-unknown-linux-gnu -vector-library=Accelerate -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,ACCELERATE
 ; RUN: opt -mtriple=aarch64-unknown-linux-gnu -vector-library=sleefgnuabi -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,SLEEFGNUABI
 ; RUN: opt -mtriple=riscv64-unknown-linux-gnu -vector-library=sleefgnuabi -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,SLEEFGNUABI_RISCV
 ; RUN: opt -mtriple=aarch64-unknown-linux-gnu -vector-library=ArmPL -passes=inject-tli-mappings -S < %s | FileCheck %s  --check-prefixes=COMMON,ARMPL
 
+; LIBMVEC-AARCH64-NOT: llvm.compiler.used
+
 ; COMMON-LABEL: @llvm.compiler.used = appending global
 ; SVML-SAME:        [6 x ptr] [
 ; SVML-SAME:          ptr @__svml_sin2,
@@ -193,6 +196,9 @@ declare float @llvm.log10.f32(float) #0
 ; MASSV: declare <2 x double> @__sind2(<2 x double>)
 ; MASSV: declare <4 x float> @__log10f4(<4 x float>)
 
+; LIBMVEC-AARCH64-NOT: declare <2 x double> @_ZGVbN2v_sin(<2 x double>)
+; LIBMVEC-AARCH64-NOT: declare <4 x double> @_ZGVdN4v_sin(<4 x double>)
+
 ; LIBMVEC-X86: declare <2 x double> @_ZGVbN2v_sin(<2 x double>)
 ; LIBMVEC-X86: declare <4 x double> @_ZGVdN4v_sin(<4 x double>)

huntergr-arm

LGTM outside of needing a commit message ;)

MacDue · 2025-05-02T13:03:51Z

llvm/lib/Analysis/TargetLibraryInfo.cpp

@@ -29,7 +29,7 @@ static cl::opt<TargetLibraryInfoImpl::VectorLibrary> ClVectorLibrary(
                          "Accelerate framework"),
               clEnumValN(TargetLibraryInfoImpl::DarwinLibSystemM,
                          "Darwin_libsystem_m", "Darwin libsystem_m"),
-               clEnumValN(TargetLibraryInfoImpl::LIBMVEC_X86, "LIBMVEC-X86",
+               clEnumValN(TargetLibraryInfoImpl::LIBMVEC, "LIBMVEC",


I think this needs changes in clang too as "LIBMVEC" is currently mapped to "LIBMVEC-X86":

llvm-project/clang/lib/Driver/ToolChains/CommonArgs.cpp

Lines 932 to 937 in 7d01b85

// Map the vector library names from clang front-end to opt front-end. The

// values are taken from the TargetLibraryInfo class command line options.

std::optional<StringRef> OptVal =

llvm::StringSwitch<std::optional<StringRef>>(ArgVecLib->getValue())

.Case("Accelerate", "Accelerate")

.Case("LIBMVEC", "LIBMVEC-X86")

Hmm. There seems to be a bug in clang's checks too, as it's checking for "LIBMVEC-X86" before this remapping, it should actually output err_drv_unsupported_opt_for_target for LIBMVEC on AArch64. I think once you rename all uses of LIBMVEC-X86 to LIBMVEC in clang that error will start working.

Yep, I've just noticed the same.

I've raised #138288 to fix the driver bug. I'll update this PR when that lands and request a re-review.

Renames LIBMVEC-X86 to LIBMVEC and updates TLI to only add the existing x86 specific mapping when targeting x86.

paulwalker-arm · 2025-05-06T15:57:49Z

Re-review request after rebasing to pull in the driver fix.

MacDue

LGTM, thanks 👍

MacDue · 2025-05-06T16:10:30Z

llvm/lib/Analysis/TargetLibraryInfo.cpp

+    switch (TargetTriple.getArch()) {
+    default:
+      break;
+    case llvm::Triple::x86:
+    case llvm::Triple::x86_64:
+      addVectorizableFunctions(VecFuncs_LIBMVEC_X86);
+      break;
+    }


Suggested change

switch (TargetTriple.getArch()) {

default:

break;

case llvm::Triple::x86:

case llvm::Triple::x86_64:

addVectorizableFunctions(VecFuncs_LIBMVEC_X86);

break;

}

if (TargetTriple.isX86())

addVectorizableFunctions(VecFuncs_LIBMVEC_X86);

paulwalker-arm requested review from MacDue and huntergr-arm May 2, 2025 12:44

llvmbot added llvm:analysis llvm:transforms labels May 2, 2025

huntergr-arm approved these changes May 2, 2025

View reviewed changes

MacDue reviewed May 2, 2025

View reviewed changes

paulwalker-arm added 2 commits May 6, 2025 12:59

Add test showing x86 functions when targeting AArch64.

5da6f3e

[LLVM][VecLib] Refactor LIBMVEC integration to be target neutral.

c1e632b

Renames LIBMVEC-X86 to LIBMVEC and updates TLI to only add the existing x86 specific mapping when targeting x86.

paulwalker-arm force-pushed the aarch64-libmvec branch from f296287 to c1e632b Compare May 6, 2025 15:55

llvmbot added clang Clang issues not falling into any other category clang:driver 'clang' and 'clang++' user-facing binaries. Not 'clang-cl' labels May 6, 2025

MacDue approved these changes May 6, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[LLVM][VecLib] Refactor LIBMVEC integration to be target neutral. #138262

[LLVM][VecLib] Refactor LIBMVEC integration to be target neutral. #138262

paulwalker-arm commented May 2, 2025

llvmbot commented May 2, 2025 •

edited

Loading

huntergr-arm left a comment

MacDue May 2, 2025

MacDue May 2, 2025 •

edited

Loading

paulwalker-arm May 2, 2025

paulwalker-arm May 2, 2025

paulwalker-arm May 6, 2025

paulwalker-arm commented May 6, 2025 •

edited

Loading

MacDue left a comment

MacDue May 6, 2025

	// Map the vector library names from clang front-end to opt front-end. The
	// values are taken from the TargetLibraryInfo class command line options.
	std::optional<StringRef> OptVal =
	llvm::StringSwitch<std::optional<StringRef>>(ArgVecLib->getValue())
	.Case("Accelerate", "Accelerate")
	.Case("LIBMVEC", "LIBMVEC-X86")

[LLVM][VecLib] Refactor LIBMVEC integration to be target neutral. #138262

Are you sure you want to change the base?

[LLVM][VecLib] Refactor LIBMVEC integration to be target neutral. #138262

Conversation

paulwalker-arm commented May 2, 2025

llvmbot commented May 2, 2025 • edited Loading

huntergr-arm left a comment

Choose a reason for hiding this comment

MacDue May 2, 2025

Choose a reason for hiding this comment

MacDue May 2, 2025 • edited Loading

Choose a reason for hiding this comment

paulwalker-arm May 2, 2025

Choose a reason for hiding this comment

paulwalker-arm May 2, 2025

Choose a reason for hiding this comment

paulwalker-arm May 6, 2025

Choose a reason for hiding this comment

paulwalker-arm commented May 6, 2025 • edited Loading

MacDue left a comment

Choose a reason for hiding this comment

MacDue May 6, 2025

Choose a reason for hiding this comment

llvmbot commented May 2, 2025 •

edited

Loading

MacDue May 2, 2025 •

edited

Loading

paulwalker-arm commented May 6, 2025 •

edited

Loading