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
73 lines
3.9 KiB
LLVM
73 lines
3.9 KiB
LLVM
; RUN: llc < %s -mcpu=cortex-a8 -verify-machineinstrs -verify-coalescing
|
|
; PR11841
|
|
; PR11829
|
|
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:64:128-a0:0:64-n32-S64"
|
|
target triple = "armv7-none-linux-eabi"
|
|
|
|
; This test case is exercising REG_SEQUENCE, and chains of REG_SEQUENCE.
|
|
define arm_aapcs_vfpcc void @foo(ptr nocapture %arg, ptr %arg1) nounwind align 2 {
|
|
bb:
|
|
%tmp = load <2 x float>, ptr undef, align 8
|
|
%tmp2 = extractelement <2 x float> %tmp, i32 0
|
|
%tmp3 = insertelement <4 x float> undef, float %tmp2, i32 0
|
|
%tmp4 = insertelement <4 x float> %tmp3, float 0.000000e+00, i32 1
|
|
%tmp5 = insertelement <4 x float> %tmp4, float 0.000000e+00, i32 2
|
|
%tmp6 = insertelement <4 x float> %tmp5, float 0.000000e+00, i32 3
|
|
%tmp7 = extractelement <2 x float> %tmp, i32 1
|
|
%tmp8 = insertelement <4 x float> %tmp3, float %tmp7, i32 1
|
|
%tmp9 = insertelement <4 x float> %tmp8, float 0.000000e+00, i32 2
|
|
%tmp10 = insertelement <4 x float> %tmp9, float 0.000000e+00, i32 3
|
|
%tmp11 = bitcast <4 x float> %tmp6 to <2 x i64>
|
|
%tmp12 = shufflevector <2 x i64> %tmp11, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%tmp13 = bitcast <1 x i64> %tmp12 to <2 x float>
|
|
%tmp14 = shufflevector <2 x float> %tmp13, <2 x float> undef, <4 x i32> zeroinitializer
|
|
%tmp15 = bitcast <4 x float> %tmp14 to <2 x i64>
|
|
%tmp16 = shufflevector <2 x i64> %tmp15, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%tmp17 = bitcast <1 x i64> %tmp16 to <2 x float>
|
|
%tmp18 = extractelement <2 x float> %tmp17, i32 0
|
|
tail call arm_aapcs_vfpcc void @bar(ptr undef, float %tmp18, float undef, float 0.000000e+00) nounwind
|
|
%tmp19 = bitcast <4 x float> %tmp10 to <2 x i64>
|
|
%tmp20 = shufflevector <2 x i64> %tmp19, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%tmp21 = bitcast <1 x i64> %tmp20 to <2 x float>
|
|
%tmp22 = shufflevector <2 x float> %tmp21, <2 x float> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
|
|
%tmp23 = bitcast <4 x float> %tmp22 to <2 x i64>
|
|
%tmp24 = shufflevector <2 x i64> %tmp23, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%tmp25 = bitcast <1 x i64> %tmp24 to <2 x float>
|
|
%tmp26 = extractelement <2 x float> %tmp25, i32 0
|
|
tail call arm_aapcs_vfpcc void @bar(ptr undef, float undef, float %tmp26, float 0.000000e+00) nounwind
|
|
ret void
|
|
}
|
|
|
|
define arm_aapcs_vfpcc void @foo2() nounwind uwtable {
|
|
entry:
|
|
br i1 undef, label %for.end, label %cond.end295
|
|
|
|
cond.end295: ; preds = %entry
|
|
%shuffle.i39.i.i1035 = shufflevector <2 x i64> undef, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%shuffle.i38.i.i1036 = shufflevector <2 x i64> zeroinitializer, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%shuffle.i37.i.i1037 = shufflevector <1 x i64> %shuffle.i39.i.i1035, <1 x i64> %shuffle.i38.i.i1036, <2 x i32> <i32 0, i32 1>
|
|
%0 = bitcast <2 x i64> %shuffle.i37.i.i1037 to <4 x float>
|
|
%1 = bitcast <4 x float> undef to <2 x i64>
|
|
%shuffle.i36.i.i = shufflevector <2 x i64> %1, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%shuffle.i35.i.i = shufflevector <2 x i64> undef, <2 x i64> undef, <1 x i32> zeroinitializer
|
|
%shuffle.i34.i.i = shufflevector <1 x i64> %shuffle.i36.i.i, <1 x i64> %shuffle.i35.i.i, <2 x i32> <i32 0, i32 1>
|
|
%2 = bitcast <2 x i64> %shuffle.i34.i.i to <4 x float>
|
|
tail call void @llvm.arm.neon.vst1.p0.v4f32(ptr undef, <4 x float> %0, i32 4) nounwind
|
|
tail call void @llvm.arm.neon.vst1.p0.v4f32(ptr undef, <4 x float> %2, i32 4) nounwind
|
|
unreachable
|
|
|
|
for.end: ; preds = %entry
|
|
ret void
|
|
}
|
|
|
|
; Check that pseudo-expansion preserves <undef> flags.
|
|
define void @foo3(ptr %p) nounwind ssp {
|
|
entry:
|
|
tail call void @llvm.arm.neon.vst2.p0.v4f32(ptr %p, <4 x float> undef, <4 x float> undef, i32 4)
|
|
ret void
|
|
}
|
|
|
|
declare arm_aapcs_vfpcc void @bar(ptr, float, float, float)
|
|
declare void @llvm.arm.neon.vst1.p0.v4f32(ptr, <4 x float>, i32) nounwind
|
|
declare void @llvm.arm.neon.vst2.p0.v4f32(ptr, <4 x float>, <4 x float>, i32) nounwind
|