Skip to content

[clang][bytecode] Support remaining add_sat like X86 builtins #155358

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
Aug 26, 2025
Merged

[clang][bytecode] Support remaining add_sat like X86 builtins #155358

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
21769a2
[clang][bytecode] Support remaining add_sat like X86 builtins
tbaederr Aug 26, 2025
db30f87
elementwise_sat -> elementwise_int_binop
tbaederr Aug 26, 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 clang/lib/AST/ByteCode/Integral.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -318,6 +318,11 @@ template <unsigned Bits, bool Signed> class Integral final {
template <typename T> static bool CheckMulUB(T A, T B, T &R) {
if constexpr (std::is_signed_v<T>) {
return llvm::MulOverflow<T>(A, B, R);
} else if constexpr (sizeof(T) < sizeof(int)) {
// Silly integer promotion rules will convert both A and B to int,
// even it T is unsigned. Prevent that by manually casting to uint first.
R = static_cast<T>(static_cast<unsigned>(A) * static_cast<unsigned>(B));
return false;
} else {
R = A * B;
return false;
Expand Down
48 changes: 44 additions & 4 deletions clang/lib/AST/ByteCode/InterpBuiltin.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2518,9 +2518,9 @@ static bool interp__builtin_is_within_lifetime(InterpState &S, CodePtr OpPC,
return true;
}

static bool interp__builtin_elementwise_sat(InterpState &S, CodePtr OpPC,
const CallExpr *Call,
unsigned BuiltinID) {
static bool interp__builtin_elementwise_int_binop(InterpState &S, CodePtr OpPC,
const CallExpr *Call,
unsigned BuiltinID) {
assert(Call->getNumArgs() == 2);

// Single integer case.
Expand Down Expand Up @@ -2557,6 +2557,8 @@ static bool interp__builtin_elementwise_sat(InterpState &S, CodePtr OpPC,
const Pointer &LHS = S.Stk.pop<Pointer>();
const Pointer &Dst = S.Stk.peek<Pointer>();
PrimType ElemT = *S.getContext().classify(VT->getElementType());
bool DestUnsigned =
VT->getElementType()->isUnsignedIntegerOrEnumerationType();
unsigned NumElems = VT->getNumElements();
for (unsigned I = 0; I != NumElems; ++I) {
APSInt Elem1;
Expand Down Expand Up @@ -2590,6 +2592,34 @@ static bool interp__builtin_elementwise_sat(InterpState &S, CodePtr OpPC,
Result = APSInt(llvm::APIntOps::mulhs(Elem1, Elem2),
/*isUnsigned=*/false);
break;
case clang::X86::BI__builtin_ia32_psllv2di:
case clang::X86::BI__builtin_ia32_psllv4di:
case clang::X86::BI__builtin_ia32_psllv4si:
case clang::X86::BI__builtin_ia32_psllv8si:
if (Elem2.uge(Elem2.getBitWidth())) {
Result = APSInt(APInt::getZero(Elem2.getBitWidth()), DestUnsigned);
break;
}
Result = APSInt(Elem1.shl(Elem2.getZExtValue()), DestUnsigned);
break;
case clang::X86::BI__builtin_ia32_psrav4si:
case clang::X86::BI__builtin_ia32_psrav8si:
if (Elem2.uge(Elem2.getBitWidth())) {
Result = APSInt(Elem1.ashr(Elem2.getBitWidth() - 1), DestUnsigned);
break;
}
Result = APSInt(Elem1.ashr(Elem2.getZExtValue()), DestUnsigned);
break;
case clang::X86::BI__builtin_ia32_psrlv2di:
case clang::X86::BI__builtin_ia32_psrlv4di:
case clang::X86::BI__builtin_ia32_psrlv4si:
case clang::X86::BI__builtin_ia32_psrlv8si:
if (Elem2.uge(Elem2.getBitWidth())) {
Result = APSInt(APInt::getZero(Elem2.getBitWidth()), DestUnsigned);
break;
}
Result = APSInt(Elem1.lshr(Elem2.getZExtValue()), DestUnsigned);
break;
default:
llvm_unreachable("Wrong builtin ID");
}
Expand Down Expand Up @@ -3236,7 +3266,17 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
case clang::X86::BI__builtin_ia32_pmulhw128:
case clang::X86::BI__builtin_ia32_pmulhw256:
case clang::X86::BI__builtin_ia32_pmulhw512:
return interp__builtin_elementwise_sat(S, OpPC, Call, BuiltinID);
case clang::X86::BI__builtin_ia32_psllv2di:
case clang::X86::BI__builtin_ia32_psllv4di:
case clang::X86::BI__builtin_ia32_psllv4si:
case clang::X86::BI__builtin_ia32_psllv8si:
case clang::X86::BI__builtin_ia32_psrav4si:
case clang::X86::BI__builtin_ia32_psrav8si:
case clang::X86::BI__builtin_ia32_psrlv2di:
case clang::X86::BI__builtin_ia32_psrlv4di:
case clang::X86::BI__builtin_ia32_psrlv4si:
case clang::X86::BI__builtin_ia32_psrlv8si:
return interp__builtin_elementwise_int_binop(S, OpPC, Call, BuiltinID);

case Builtin::BI__builtin_elementwise_max:
case Builtin::BI__builtin_elementwise_min:
Expand Down
8 changes: 8 additions & 0 deletions clang/test/CodeGen/X86/avx2-builtins.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,14 @@
// RUN: %clang_cc1 -x c++ -flax-vector-conversions=none -ffreestanding %s -triple=i386-apple-darwin -target-feature +avx2 -emit-llvm -o - -Wall -Werror | FileCheck %s --check-prefixes=CHECK,X86
// RUN: %clang_cc1 -x c++ -flax-vector-conversions=none -ffreestanding %s -triple=i386-apple-darwin -target-feature +avx2 -fno-signed-char -emit-llvm -o - -Wall -Werror | FileCheck %s --check-prefixes=CHECK,X86

// RUN: %clang_cc1 -x c -flax-vector-conversions=none -ffreestanding %s -triple=x86_64-apple-darwin -target-feature +avx2 -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X64
// RUN: %clang_cc1 -x c -flax-vector-conversions=none -ffreestanding %s -triple=x86_64-apple-darwin -target-feature +avx2 -fno-signed-char -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X64
// RUN: %clang_cc1 -x c -flax-vector-conversions=none -ffreestanding %s -triple=i386-apple-darwin -target-feature +avx2 -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X86
// RUN: %clang_cc1 -x c -flax-vector-conversions=none -ffreestanding %s -triple=i386-apple-darwin -target-feature +avx2 -fno-signed-char -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X86
// RUN: %clang_cc1 -x c++ -flax-vector-conversions=none -ffreestanding %s -triple=x86_64-apple-darwin -target-feature +avx2 -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X64
// RUN: %clang_cc1 -x c++ -flax-vector-conversions=none -ffreestanding %s -triple=x86_64-apple-darwin -target-feature +avx2 -fno-signed-char -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X64
// RUN: %clang_cc1 -x c++ -flax-vector-conversions=none -ffreestanding %s -triple=i386-apple-darwin -target-feature +avx2 -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X86
// RUN: %clang_cc1 -x c++ -flax-vector-conversions=none -ffreestanding %s -triple=i386-apple-darwin -target-feature +avx2 -fno-signed-char -emit-llvm -o - -Wall -Werror -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X86

#include <immintrin.h>
#include "builtin_test_helpers.h"
Expand Down