Skip to content

Commit

Permalink
Auto merge of #125904 - workingjubilee:test-packed-simd-more, r=caleb…
Browse files Browse the repository at this point in the history
…zulawski

Test codegen for `repr(packed,simd)` -> `repr(simd)`

This adds the codegen test originally requested in #117116 but exploiting the collection of features in FileCheck and compiletest to make it more resilient to expectations being broken by optimization levels. Mostly by presetting optimization levels for each revision of the tests.

I do not think the dereferenceable attribute's presence or absence is that important.

r? `@calebzulawski`
  • Loading branch information
bors committed Jun 3, 2024
2 parents 865eaf9 + 9987363 commit 621e957
Showing 1 changed file with 44 additions and 0 deletions.
44 changes: 44 additions & 0 deletions tests/codegen/simd/packed-simd.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,44 @@
//@ revisions:opt3 noopt
//@[opt3] compile-flags: -Copt-level=3
//@[noopt] compile-flags: -Cno-prepopulate-passes

#![crate_type = "lib"]
#![no_std]
#![feature(repr_simd, core_intrinsics)]
use core::intrinsics::simd as intrinsics;
use core::{mem, ptr};

// Test codegen for not only "packed" but also "fully aligned" SIMD types, and conversion between
// A repr(packed,simd) type with 3 elements can't exceed its element alignment,
// whereas the same type as repr(simd) will instead have padding.

#[repr(simd, packed)]
pub struct Simd<T, const N: usize>([T; N]);

#[repr(simd)]
#[derive(Copy, Clone)]
pub struct FullSimd<T, const N: usize>([T; N]);

// non-powers-of-two have padding and need to be expanded to full vectors
fn load<T, const N: usize>(v: Simd<T, N>) -> FullSimd<T, N> {
unsafe {
let mut tmp = mem::MaybeUninit::<FullSimd<T, N>>::uninit();
ptr::copy_nonoverlapping(&v as *const _, tmp.as_mut_ptr().cast(), 1);
tmp.assume_init()
}
}

// CHECK-LABEL: square_packed
// CHECK-SAME: ptr{{[a-z_ ]*}} sret([[RET_TYPE:[^)]+]]) [[RET_ALIGN:align (8|16)]]{{[^%]*}} [[RET_VREG:%[_0-9]*]]
// CHECK-SAME: ptr{{[a-z_ ]*}} align 4
#[no_mangle]
pub fn square_packed(x: Simd<f32, 3>) -> FullSimd<f32, 3> {
// CHECK-NEXT: start
// noopt: alloca [[RET_TYPE]], [[RET_ALIGN]]
// CHECK: load <3 x float>
let x = load(x);
// CHECK: [[VREG:%[a-z0-9_]+]] = fmul <3 x float>
// CHECK-NEXT: store <3 x float> [[VREG]], ptr [[RET_VREG]], [[RET_ALIGN]]
// CHECK-NEXT: ret void
unsafe { intrinsics::simd_mul(x, x) }
}

0 comments on commit 621e957

Please sign in to comment.