derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
91 lines
4.5 KiB
LLVM
91 lines
4.5 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals
|
|
; RUN: opt -codegen-opt-level=1 -S -mtriple=aarch64-- -passes='require<libcall-lowering-info>,atomic-expand' %s | FileCheck %s
|
|
; RUN: opt -codegen-opt-level=1 -S -mtriple=aarch64-- -mattr=+outline-atomics -passes='require<libcall-lowering-info>,atomic-expand' %s | FileCheck %s --check-prefix=OUTLINE-ATOMICS
|
|
|
|
define void @atomic_swap_f16(ptr %ptr, half %val) !prof !0 {
|
|
; CHECK-LABEL: @atomic_swap_f16(
|
|
; CHECK-NEXT: [[TMP1:%.*]] = bitcast half [[VAL:%.*]] to i16
|
|
; CHECK-NEXT: br label [[ATOMICRMW_START:%.*]]
|
|
; CHECK: atomicrmw.start:
|
|
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.aarch64.ldaxr.p0(ptr elementtype(i16) [[PTR:%.*]])
|
|
; CHECK-NEXT: [[TMP3:%.*]] = trunc i64 [[TMP2]] to i16
|
|
; CHECK-NEXT: [[TMP4:%.*]] = zext i16 [[TMP1]] to i64
|
|
; CHECK-NEXT: [[TMP5:%.*]] = call i32 @llvm.aarch64.stxr.p0(i64 [[TMP4]], ptr elementtype(i16) [[PTR]])
|
|
; CHECK-NEXT: [[TRYAGAIN:%.*]] = icmp ne i32 [[TMP5]], 0
|
|
; CHECK-NEXT: br i1 [[TRYAGAIN]], label [[ATOMICRMW_START]], label [[ATOMICRMW_END:%.*]], !prof [[PROF1:![0-9]+]]
|
|
; CHECK: atomicrmw.end:
|
|
; CHECK-NEXT: [[TMP6:%.*]] = bitcast i16 [[TMP3]] to half
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
; OUTLINE-ATOMICS-LABEL: @atomic_swap_f16(
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP1:%.*]] = bitcast half [[VAL:%.*]] to i16
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[PTR:%.*]], i16 [[TMP1]] acquire, align 2
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP3:%.*]] = bitcast i16 [[TMP2]] to half
|
|
; OUTLINE-ATOMICS-NEXT: ret void
|
|
;
|
|
%t1 = atomicrmw xchg ptr %ptr, half %val acquire
|
|
ret void
|
|
}
|
|
|
|
define void @atomic_swap_f32(ptr %ptr, float %val) nounwind !prof !0 {
|
|
; CHECK-LABEL: @atomic_swap_f32(
|
|
; CHECK-NEXT: [[TMP1:%.*]] = bitcast float [[VAL:%.*]] to i32
|
|
; CHECK-NEXT: br label [[ATOMICRMW_START:%.*]]
|
|
; CHECK: atomicrmw.start:
|
|
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.aarch64.ldaxr.p0(ptr elementtype(i32) [[PTR:%.*]])
|
|
; CHECK-NEXT: [[TMP3:%.*]] = trunc i64 [[TMP2]] to i32
|
|
; CHECK-NEXT: [[TMP4:%.*]] = zext i32 [[TMP1]] to i64
|
|
; CHECK-NEXT: [[TMP5:%.*]] = call i32 @llvm.aarch64.stxr.p0(i64 [[TMP4]], ptr elementtype(i32) [[PTR]])
|
|
; CHECK-NEXT: [[TRYAGAIN:%.*]] = icmp ne i32 [[TMP5]], 0
|
|
; CHECK-NEXT: br i1 [[TRYAGAIN]], label [[ATOMICRMW_START]], label [[ATOMICRMW_END:%.*]], !prof [[PROF1]]
|
|
; CHECK: atomicrmw.end:
|
|
; CHECK-NEXT: [[TMP6:%.*]] = bitcast i32 [[TMP3]] to float
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
; OUTLINE-ATOMICS-LABEL: @atomic_swap_f32(
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP1:%.*]] = bitcast float [[VAL:%.*]] to i32
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[PTR:%.*]], i32 [[TMP1]] acquire, align 4
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP3:%.*]] = bitcast i32 [[TMP2]] to float
|
|
; OUTLINE-ATOMICS-NEXT: ret void
|
|
;
|
|
%t1 = atomicrmw xchg ptr %ptr, float %val acquire
|
|
ret void
|
|
}
|
|
|
|
define void @atomic_swap_f64(ptr %ptr, double %val) nounwind !prof !0 {
|
|
; CHECK-LABEL: @atomic_swap_f64(
|
|
; CHECK-NEXT: [[TMP1:%.*]] = bitcast double [[VAL:%.*]] to i64
|
|
; CHECK-NEXT: br label [[ATOMICRMW_START:%.*]]
|
|
; CHECK: atomicrmw.start:
|
|
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.aarch64.ldaxr.p0(ptr elementtype(i64) [[PTR:%.*]])
|
|
; CHECK-NEXT: [[TMP3:%.*]] = call i32 @llvm.aarch64.stxr.p0(i64 [[TMP1]], ptr elementtype(i64) [[PTR]])
|
|
; CHECK-NEXT: [[TRYAGAIN:%.*]] = icmp ne i32 [[TMP3]], 0
|
|
; CHECK-NEXT: br i1 [[TRYAGAIN]], label [[ATOMICRMW_START]], label [[ATOMICRMW_END:%.*]], !prof [[PROF1]]
|
|
; CHECK: atomicrmw.end:
|
|
; CHECK-NEXT: [[TMP4:%.*]] = bitcast i64 [[TMP2]] to double
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
; OUTLINE-ATOMICS-LABEL: @atomic_swap_f64(
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP1:%.*]] = bitcast double [[VAL:%.*]] to i64
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[PTR:%.*]], i64 [[TMP1]] acquire, align 8
|
|
; OUTLINE-ATOMICS-NEXT: [[TMP3:%.*]] = bitcast i64 [[TMP2]] to double
|
|
; OUTLINE-ATOMICS-NEXT: ret void
|
|
;
|
|
%t1 = atomicrmw xchg ptr %ptr, double %val acquire
|
|
ret void
|
|
}
|
|
|
|
!0 = !{!"function_entry_count", i64 1000}
|
|
;.
|
|
; CHECK: attributes #[[ATTR0:[0-9]+]] = { nounwind }
|
|
; CHECK: attributes #[[ATTR1:[0-9]+]] = { nofree nounwind willreturn }
|
|
;.
|
|
; OUTLINE-ATOMICS: attributes #[[ATTR0:[0-9]+]] = { "target-features"="+outline-atomics" }
|
|
; OUTLINE-ATOMICS: attributes #[[ATTR1:[0-9]+]] = { nounwind "target-features"="+outline-atomics" }
|
|
;.
|
|
; CHECK: [[META0:![0-9]+]] = !{!"function_entry_count", i64 1000}
|
|
; CHECK: [[PROF1]] = !{!"unknown", !"atomic-expand"}
|
|
;.
|
|
; OUTLINE-ATOMICS: [[META0:![0-9]+]] = !{!"function_entry_count", i64 1000}
|
|
;.
|