Skip to content

Commit

Permalink
(release 20.x non-upstream patch) [clang] Lower non-builtin sincos[f|…
Browse files Browse the repository at this point in the history 
…l] calls to llvm.sincos.* when -fno-math-errno is set (#83)

This will allow vectorizing these calls (after a few more patches). This
should not change the codegen for targets that enable the use of AA
during the codegen (in `TargetSubtargetInfo::useAA()`). This includes
targets such as AArch64. This notably does not include x86 but can be
worked around by passing `-mllvm -combiner-global-alias-analysis=true`
to clang.

---

Downstream issue: #87
  • Loading branch information
@MacDue
MacDue authored Feb 24, 2025
1 parent c742ca2 commit e7d5a88
Show file tree
Hide file tree
Showing 3 changed files with 76 additions and 0 deletions.
9 changes: 9 additions & 0 deletions clang/lib/CodeGen/CGBuiltin.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3253,11 +3253,20 @@ RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID,
return RValue::get(emitUnaryMaybeConstrainedFPBuiltin(
*this, E, Intrinsic::sinh, Intrinsic::experimental_constrained_sinh));

/* Downstream change: #87 (sincos vectorization)*/
case Builtin::BIsincos:
case Builtin::BIsincosf:
case Builtin::BIsincosl:
/* End downstream change: #87 */
case Builtin::BI__builtin_sincos:
case Builtin::BI__builtin_sincosf:
case Builtin::BI__builtin_sincosf16:
case Builtin::BI__builtin_sincosl:
case Builtin::BI__builtin_sincosf128:
/* Downstream change: #87 (sincos vectorization)*/
if (Builder.getIsFPConstrained())
break; // TODO: Emit constrained sincos intrinsic once one exists.
/* Downstream change: #87 (sincos vectorization)*/
emitSincosBuiltin(*this, E, Intrinsic::sincos);
return RValue::get(nullptr);

Expand Down
56 changes: 56 additions & 0 deletions clang/test/CodeGen/AArch64/sincos.c
View file
Original file line number Diff line number Diff line change
@@ -1,5 +1,10 @@
// RUN: %clang_cc1 -triple=aarch64-gnu-linux -emit-llvm -O1 %s -o - | FileCheck --check-prefix=NO-MATH-ERRNO %s
// RUN: %clang_cc1 -triple=aarch64-gnu-linux -emit-llvm -fmath-errno %s -o - | FileCheck --check-prefix=MATH-ERRNO %s
// RUN: %clang_cc1 -triple=aarch64-gnu-linux -emit-llvm -ffp-exception-behavior=strict %s -o - | FileCheck --check-prefix=STRICT-FP %s

void sincos(double, double*, double*);
void sincosf(float, float*, float*);
void sincosl(long double, long double*, long double*);

// NO-MATH-ERRNO-LABEL: @sincos_f32
// NO-MATH-ERRNO: [[SINCOS:%.*]] = tail call { float, float } @llvm.sincos.f32(float {{.*}})
Expand All @@ -11,7 +16,24 @@
// MATH-ERRNO-LABEL: @sincos_f32
// MATH-ERRNO: call void @sincosf(
//
// STRICT-FP-LABEL: @sincos_f32
// STRICT-FP: call void @sincosf(
//
void sincos_f32(float x, float* fp0, float* fp1) {
sincosf(x, fp0, fp1);
}

// NO-MATH-ERRNO-LABEL: @sincos_builtin_f32
// NO-MATH-ERRNO: [[SINCOS:%.*]] = tail call { float, float } @llvm.sincos.f32(float {{.*}})
// NO-MATH-ERRNO-NEXT: [[SIN:%.*]] = extractvalue { float, float } [[SINCOS]], 0
// NO-MATH-ERRNO-NEXT: [[COS:%.*]] = extractvalue { float, float } [[SINCOS]], 1
// NO-MATH-ERRNO-NEXT: store float [[SIN]], ptr {{.*}}, align 4, !alias.scope [[SINCOS_ALIAS_SCOPE:![0-9]+]]
// NO-MATH-ERRNO-NEXT: store float [[COS]], ptr {{.*}}, align 4, !noalias [[SINCOS_ALIAS_SCOPE]]
//
// MATH-ERRNO-LABEL: @sincos_builtin_f32
// MATH-ERRNO: call void @sincosf(
//
void sincos_builtin_f32(float x, float* fp0, float* fp1) {
__builtin_sincosf(x, fp0, fp1);
}

Expand All @@ -25,7 +47,24 @@ void sincos_f32(float x, float* fp0, float* fp1) {
// MATH-ERRNO-LABEL: @sincos_f64
// MATH-ERRNO: call void @sincos(
//
// STRICT-FP-LABEL: @sincos_f64
// STRICT-FP: call void @sincos(
//
void sincos_f64(double x, double* dp0, double* dp1) {
sincos(x, dp0, dp1);
}

// NO-MATH-ERRNO-LABEL: @sincos_builtin_f64
// NO-MATH-ERRNO: [[SINCOS:%.*]] = tail call { double, double } @llvm.sincos.f64(double {{.*}})
// NO-MATH-ERRNO-NEXT: [[SIN:%.*]] = extractvalue { double, double } [[SINCOS]], 0
// NO-MATH-ERRNO-NEXT: [[COS:%.*]] = extractvalue { double, double } [[SINCOS]], 1
// NO-MATH-ERRNO-NEXT: store double [[SIN]], ptr {{.*}}, align 8, !alias.scope [[SINCOS_ALIAS_SCOPE:![0-9]+]]
// NO-MATH-ERRNO-NEXT: store double [[COS]], ptr {{.*}}, align 8, !noalias [[SINCOS_ALIAS_SCOPE]]
//
// MATH-ERRNO-LABEL: @sincos_builtin_f64
// MATH-ERRNO: call void @sincos(
//
void sincos_builtin_f64(double x, double* dp0, double* dp1) {
__builtin_sincos(x, dp0, dp1);
}

Expand All @@ -39,6 +78,23 @@ void sincos_f64(double x, double* dp0, double* dp1) {
// MATH-ERRNO-LABEL: @sincos_f128
// MATH-ERRNO: call void @sincosl(
//
// STRICT-FP-LABEL: @sincos_f128
// STRICT-FP: call void @sincosl(
//
void sincos_f128(long double x, long double* ldp0, long double* ldp1) {
sincosl(x, ldp0, ldp1);
}

// NO-MATH-ERRNO-LABEL: @sincos_builtin_f128
// NO-MATH-ERRNO: [[SINCOS:%.*]] = tail call { fp128, fp128 } @llvm.sincos.f128(fp128 {{.*}})
// NO-MATH-ERRNO-NEXT: [[SIN:%.*]] = extractvalue { fp128, fp128 } [[SINCOS]], 0
// NO-MATH-ERRNO-NEXT: [[COS:%.*]] = extractvalue { fp128, fp128 } [[SINCOS]], 1
// NO-MATH-ERRNO-NEXT: store fp128 [[SIN]], ptr {{.*}}, align 16, !alias.scope [[SINCOS_ALIAS_SCOPE:![0-9]+]]
// NO-MATH-ERRNO-NEXT: store fp128 [[COS]], ptr {{.*}}, align 16, !noalias [[SINCOS_ALIAS_SCOPE]]
//
// MATH-ERRNO-LABEL: @sincos_builtin_f128
// MATH-ERRNO: call void @sincosl(
//
void sincos_builtin_f128(long double x, long double* ldp0, long double* ldp1) {
__builtin_sincosl(x, ldp0, ldp1);
}
11 changes: 11 additions & 0 deletions clang/test/CodeGen/math-libcalls.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -660,6 +660,17 @@ void foo(double *d, float f, float *fp, long double *l, int *i, const char *c) {
// HAS_MAYTRAP: declare float @llvm.experimental.constrained.sinh.f32(
// HAS_MAYTRAP: declare x86_fp80 @llvm.experimental.constrained.sinh.f80(

sincos(f, d, d); sincosf(f, fp, fp); sincosl(f, l, l);

// NO__ERRNO: declare { double, double } @llvm.sincos.f64(double) [[READNONE_INTRINSIC]]
// NO__ERRNO: declare { float, float } @llvm.sincos.f32(float) [[READNONE_INTRINSIC]]
// NO__ERRNO: declare { x86_fp80, x86_fp80 } @llvm.sincos.f80(x86_fp80) [[READNONE_INTRINSIC]]
// HAS_ERRNO: declare void @sincos(double noundef, ptr noundef, ptr noundef) [[NOT_READNONE]]
// HAS_ERRNO: declare void @sincosf(float noundef, ptr noundef, ptr noundef) [[NOT_READNONE]]
// HAS_ERRNO: declare void @sincosl(x86_fp80 noundef, ptr noundef, ptr noundef) [[NOT_READNONE]]
// HAS_MAYTRAP: declare void @sincos(double noundef, ptr noundef, ptr noundef) [[NOT_READNONE]]
// HAS_MAYTRAP: declare void @sincosf(float noundef, ptr noundef, ptr noundef) [[NOT_READNONE]]
// HAS_MAYTRAP: declare void @sincosl(x86_fp80 noundef, ptr noundef, ptr noundef) [[NOT_READNONE]]

sqrt(f); sqrtf(f); sqrtl(f);

Expand Down

0 comments on commit e7d5a88

Please sign in to comment.