Skip to content

don't pass Complex<f16> in vector registers on i586 without SSE2 #161814

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

Open
folkertdev wants to merge 1 commit into
base: main
Choose a base branch
from
Open

don't pass Complex<f16> in vector registers on i586 without SSE2 #161814

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
6 changes: 4 additions & 2 deletions compiler/rustc_target/src/callconv/x86.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,8 +32,10 @@ where
// https://www.angelcode.com/dev/callconv/callconv.html
// Clang's ABI handling is in lib/CodeGen/TargetInfo.cpp
let t = cx.target_spec();
if let Some(Float::F16) = fn_abi.ret.layout.complex_float(cx) {
// `_Complex _Float16` is returned as `<2 x half>`.
if let Some(Float::F16) = fn_abi.ret.layout.complex_float(cx)
&& cx.target_spec().rustc_abi == Some(RustcAbi::X86Sse2)

@RalfJung RalfJung Aug 26, 2026 •
edited by rustbot
Loading

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So this file is for x86-32 only?

X86Sse2 is not used on x86-64.

View changes since the review

@folkertdev folkertdev Aug 26, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes? We have a separate compiler/rustc_target/src/callconv/x86_64.rs file, but there SSE2 is assumed.

{
// With SSE2 `_Complex _Float16` is returned as `<2 x half>`.
let kind = RegKind::Vector { hint_vector_elem: Primitive::Float(Float::F16) };
fn_abi.ret.cast_to(Reg { kind, size: fn_abi.ret.layout.size });
} else if t.abi_return_struct_as_int
Expand Down
14 changes: 13 additions & 1 deletion tests/codegen-llvm/complex-abi.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,9 @@
//@ [WINDOWS_GNU] compile-flags: --target x86_64-pc-windows-gnu
//@ [WINDOWS_GNU] needs-llvm-components: x86

//@ revisions: I686 WIN32_MSVC WIN32_GNU
//@ revisions: I586 I686 WIN32_MSVC WIN32_GNU
//@ [I586] compile-flags: --target i586-unknown-linux-gnu
//@ [I586] needs-llvm-components: x86
//@ [I686] compile-flags: --target i686-unknown-linux-gnu
//@ [I686] needs-llvm-components: x86
//@ [WIN32_MSVC] compile-flags: --target i686-pc-windows-msvc
Expand Down Expand Up @@ -102,6 +104,7 @@ pub extern "C" fn cplx_f16(x: Complex<f16>) -> Complex<f16> {
// AARCH64_MSVC: define{{.*}} [2 x half] @cplx_f16([2 x half] {{.*}})
// ARM64EC: define{{.*}} [2 x half] @cplx_f16([2 x half] {{.*}})
// ARM: define{{.*}} [2 x half] @cplx_f16([2 x half] {{.*}})
// I586: define{{.*}} i32 @cplx_f16(ptr {{.*}} byval([4 x i8]) {{.*}})
// I686: define{{.*}} <2 x half> @cplx_f16(ptr {{.*}} byval([4 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} { half, half } @cplx_f16({ half, half } {{.*}})
// LOONGARCH64: define{{.*}} { half, half } @cplx_f16({ half, half } {{.*}})
Expand All @@ -127,6 +130,7 @@ pub extern "C" fn cplx_f32(x: Complex<f32>) -> Complex<f32> {
// ARM: define{{.*}} [2 x float] @cplx_f32([2 x float] {{.*}})
// BPF: define{{.*}} void @cplx_f32(ptr {{.*}} sret({ float, float }) {{.*}}, i64 {{.*}})
// CSKY: define{{.*}} [2 x i32] @cplx_f32([2 x i32] {{.*}})
// I586: define{{.*}} i64 @cplx_f32(ptr {{.*}} byval([8 x i8]) {{.*}})
// I686: define{{.*}} i64 @cplx_f32(ptr {{.*}} byval([8 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} { float, float } @cplx_f32({ float, float } {{.*}})
// LOONGARCH64: define{{.*}} { float, float } @cplx_f32({ float, float } {{.*}})
Expand Down Expand Up @@ -161,6 +165,7 @@ pub extern "C" fn cplx_f64(x: Complex<f64>) -> Complex<f64> {
// ARM: define{{.*}} [2 x double] @cplx_f64([2 x double] {{.*}})
// BPF: define{{.*}} void @cplx_f64(ptr {{.*}} sret({ double, double }) {{.*}}, [2 x i64] {{.*}})
// CSKY: define{{.*}} void @cplx_f64(ptr {{.*}} sret([16 x i8]) {{.*}}, [4 x i32] {{.*}})
// I586: define{{.*}} void @cplx_f64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}} byval([16 x i8]) {{.*}})
// I686: define{{.*}} void @cplx_f64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}} byval([16 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} { double, double } @cplx_f64({ double, double } {{.*}})
// LOONGARCH64: define{{.*}} { double, double } @cplx_f64({ double, double } {{.*}})
Expand Down Expand Up @@ -188,6 +193,7 @@ pub extern "C" fn cplx_f64(x: Complex<f64>) -> Complex<f64> {
#[no_mangle]
pub extern "C" fn cplx_f128(x: Complex<f128>) -> Complex<f128> {
// AARCH64: define{{.*}} [2 x fp128] {{.*}})
// I586: define{{.*}} void @cplx_f128(ptr {{.*}} sret([32 x i8]) {{.*}}, ptr {{.*}} byval([32 x i8]) {{.*}})
// I686: define{{.*}} void @cplx_f128(ptr {{.*}} sret([32 x i8]) {{.*}}, ptr {{.*}} byval([32 x i8]) {{.*}})
// WASM32: define{{.*}} void @cplx_f128(ptr {{.*}} sret([32 x i8]) {{.*}}, ptr {{.*}})
// WASM64: define{{.*}} void @cplx_f128(ptr {{.*}} sret([32 x i8]) {{.*}}, ptr {{.*}})
Expand All @@ -209,6 +215,7 @@ pub extern "C" fn cplx_i8(x: Complex<i8>) -> Complex<i8> {
// ARM: define{{.*}} i32 @cplx_i8(i32{{.*}})
// BPF: define{{.*}} void @cplx_i8(ptr {{.*}} sret({ i8, i8 }) {{.*}}, i16 {{.*}})
// CSKY: define{{.*}} i32 @cplx_i8(i32{{.*}})
// I586: define{{.*}} i16 @cplx_i8(ptr {{.*}} byval([2 x i8]) {{.*}})
// I686: define{{.*}} i16 @cplx_i8(ptr {{.*}} byval([2 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} i32 @cplx_i8(i32{{.*}})
// LOONGARCH64: define{{.*}} i64 @cplx_i8(i64{{.*}})
Expand Down Expand Up @@ -243,6 +250,7 @@ pub extern "C" fn cplx_i16(x: Complex<i16>) -> Complex<i16> {
// ARM: define{{.*}} i32 @cplx_i16(i32{{.*}})
// BPF: define{{.*}} void @cplx_i16(ptr {{.*}} sret({ i16, i16 }) {{.*}}, i32 {{.*}})
// CSKY: define{{.*}} i32 @cplx_i16(i32 {{.*}})
// I586: define{{.*}} i32 @cplx_i16(ptr {{.*}} byval([4 x i8]) {{.*}})
// I686: define{{.*}} i32 @cplx_i16(ptr {{.*}} byval([4 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} i32 @cplx_i16(i32 {{.*}})
// LOONGARCH64: define{{.*}} i64 @cplx_i16(i64{{.*}})
Expand Down Expand Up @@ -277,6 +285,7 @@ pub extern "C" fn cplx_i32(x: Complex<i32>) -> Complex<i32> {
// ARM: define{{.*}} void @cplx_i32(ptr {{.*}} sret([8 x i8]) {{.*}}, [2 x i32] {{.*}})
// BPF: define{{.*}} void @cplx_i32(ptr {{.*}} sret({ i32, i32 }) {{.*}}, i64 {{.*}})
// CSKY: define{{.*}} [2 x i32] @cplx_i32([2 x i32] {{.*}})
// I586: define{{.*}} i64 @cplx_i32(ptr {{.*}} byval([8 x i8]) {{.*}})
// I686: define{{.*}} i64 @cplx_i32(ptr {{.*}} byval([8 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} [2 x i32] @cplx_i32([2 x i32] {{.*}})
// LOONGARCH64: define{{.*}} i64 @cplx_i32(i64 {{.*}})
Expand Down Expand Up @@ -311,6 +320,7 @@ pub extern "C" fn cplx_i64(x: Complex<i64>) -> Complex<i64> {
// ARM: define{{.*}} void @cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, [2 x i64] {{.*}})
// BPF: define{{.*}} void @cplx_i64(ptr {{.*}} sret({ i64, i64 }) {{.*}}, [2 x i64] {{.*}})
// CSKY: define{{.*}} void @cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, [4 x i32] {{.*}})
// I586: define{{.*}} void @cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}} byval([16 x i8]) {{.*}})
// I686: define{{.*}} void @cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}} byval([16 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} void @cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}})
// LOONGARCH64: define{{.*}} [2 x i64] @cplx_i64([2 x i64] {{.*}})
Expand Down Expand Up @@ -350,6 +360,7 @@ pub extern "C" fn wrapper_cplx_i64(
// ARM: define{{.*}} void @wrapper_cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, [2 x i64] {{.*}})
// BPF: define{{.*}} void @wrapper_cplx_i64(ptr {{.*}} sret({ i64, i64 }) {{.*}}, [2 x i64] {{.*}})
// CSKY: define{{.*}} void @wrapper_cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, [4 x i32] {{.*}})
// I586: define{{.*}} void @wrapper_cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}} byval([16 x i8]) {{.*}})
// I686: define{{.*}} void @wrapper_cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}} byval([16 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} void @wrapper_cplx_i64(ptr {{.*}} sret([16 x i8]) {{.*}}, ptr {{.*}})
// LOONGARCH64: define{{.*}} [2 x i64] @wrapper_cplx_i64([2 x i64] {{.*}})
Expand Down Expand Up @@ -383,6 +394,7 @@ pub extern "C" fn wrapper_cplx_f16(
// AARCH64_MSVC: define{{.*}} [2 x half] @wrapper_cplx_f16([2 x half] {{.*}})
// ARM64EC: define{{.*}} [2 x half] @wrapper_cplx_f16([2 x half] {{.*}})
// ARM: define{{.*}} [2 x half] @wrapper_cplx_f16([2 x half] {{.*}})
// I586: define{{.*}} i32 @wrapper_cplx_f16(ptr {{.*}} byval([4 x i8]) {{.*}})
// I686: define{{.*}} <2 x half> @wrapper_cplx_f16(ptr {{.*}} byval([4 x i8]) {{.*}})
// LOONGARCH32: define{{.*}} { half, half } @wrapper_cplx_f16({ half, half } {{.*}})
// LOONGARCH64: define{{.*}} { half, half } @wrapper_cplx_f16({ half, half } {{.*}})
Expand Down
Loading