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
150 lines
5.1 KiB
LLVM
150 lines
5.1 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: llc -mtriple=amdgcn < %s 2>&1 | FileCheck -check-prefix=GCN %s
|
|
|
|
; It is not illegal anymore because the SGPRs with divergent values are added with readfirstlane
|
|
; instruction in inline asm.
|
|
define amdgpu_kernel void @illegal_vgpr_to_sgpr_copy_i32() #0 {
|
|
; GCN-LABEL: illegal_vgpr_to_sgpr_copy_i32:
|
|
; GCN: ; %bb.0:
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; def v1
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: v_readfirstlane_b32 s9, v1
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; use s9
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: s_endpgm
|
|
%vgpr = call i32 asm sideeffect "; def $0", "=${v1}"()
|
|
call void asm sideeffect "; use $0", "${s9}"(i32 %vgpr)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @illegal_vgpr_to_sgpr_copy_v2i32() #0 {
|
|
; GCN-LABEL: illegal_vgpr_to_sgpr_copy_v2i32:
|
|
; GCN: ; %bb.0:
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; def v[0:1]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: v_readfirstlane_b32 s11, v1
|
|
; GCN-NEXT: v_readfirstlane_b32 s10, v0
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; use s[10:11]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: s_endpgm
|
|
%vgpr = call <2 x i32> asm sideeffect "; def $0", "=${v[0:1]}"()
|
|
call void asm sideeffect "; use $0", "${s[10:11]}"(<2 x i32> %vgpr)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @illegal_vgpr_to_sgpr_copy_v4i32() #0 {
|
|
; GCN-LABEL: illegal_vgpr_to_sgpr_copy_v4i32:
|
|
; GCN: ; %bb.0:
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; def v[0:3]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: v_readfirstlane_b32 s11, v3
|
|
; GCN-NEXT: v_readfirstlane_b32 s10, v2
|
|
; GCN-NEXT: v_readfirstlane_b32 s9, v1
|
|
; GCN-NEXT: v_readfirstlane_b32 s8, v0
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; use s[8:11]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: s_endpgm
|
|
%vgpr = call <4 x i32> asm sideeffect "; def $0", "=${v[0:3]}"()
|
|
call void asm sideeffect "; use $0", "${s[8:11]}"(<4 x i32> %vgpr)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @illegal_vgpr_to_sgpr_copy_v8i32() #0 {
|
|
; GCN-LABEL: illegal_vgpr_to_sgpr_copy_v8i32:
|
|
; GCN: ; %bb.0:
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; def v[0:7]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: v_readfirstlane_b32 s15, v7
|
|
; GCN-NEXT: v_readfirstlane_b32 s14, v6
|
|
; GCN-NEXT: v_readfirstlane_b32 s13, v5
|
|
; GCN-NEXT: v_readfirstlane_b32 s12, v4
|
|
; GCN-NEXT: v_readfirstlane_b32 s11, v3
|
|
; GCN-NEXT: v_readfirstlane_b32 s10, v2
|
|
; GCN-NEXT: v_readfirstlane_b32 s9, v1
|
|
; GCN-NEXT: v_readfirstlane_b32 s8, v0
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; use s[8:15]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: s_endpgm
|
|
%vgpr = call <8 x i32> asm sideeffect "; def $0", "=${v[0:7]}"()
|
|
call void asm sideeffect "; use $0", "${s[8:15]}"(<8 x i32> %vgpr)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @illegal_vgpr_to_sgpr_copy_v16i32() #0 {
|
|
; GCN-LABEL: illegal_vgpr_to_sgpr_copy_v16i32:
|
|
; GCN: ; %bb.0:
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; def v[0:15]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: v_readfirstlane_b32 s31, v15
|
|
; GCN-NEXT: v_readfirstlane_b32 s30, v14
|
|
; GCN-NEXT: v_readfirstlane_b32 s29, v13
|
|
; GCN-NEXT: v_readfirstlane_b32 s28, v12
|
|
; GCN-NEXT: v_readfirstlane_b32 s27, v11
|
|
; GCN-NEXT: v_readfirstlane_b32 s26, v10
|
|
; GCN-NEXT: v_readfirstlane_b32 s25, v9
|
|
; GCN-NEXT: v_readfirstlane_b32 s24, v8
|
|
; GCN-NEXT: v_readfirstlane_b32 s23, v7
|
|
; GCN-NEXT: v_readfirstlane_b32 s22, v6
|
|
; GCN-NEXT: v_readfirstlane_b32 s21, v5
|
|
; GCN-NEXT: v_readfirstlane_b32 s20, v4
|
|
; GCN-NEXT: v_readfirstlane_b32 s19, v3
|
|
; GCN-NEXT: v_readfirstlane_b32 s18, v2
|
|
; GCN-NEXT: v_readfirstlane_b32 s17, v1
|
|
; GCN-NEXT: v_readfirstlane_b32 s16, v0
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; use s[16:31]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: s_endpgm
|
|
%vgpr = call <16 x i32> asm sideeffect "; def $0", "=${v[0:15]}"()
|
|
call void asm sideeffect "; use $0", "${s[16:31]}"(<16 x i32> %vgpr)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @illegal_agpr_to_sgpr_copy_i32() #1 {
|
|
; GCN-LABEL: illegal_agpr_to_sgpr_copy_i32:
|
|
; GCN: ; %bb.0:
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; def a1
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: v_accvgpr_read_b32 v0, a1
|
|
; GCN-NEXT: v_readfirstlane_b32 s9, v0
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; use s9
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: s_endpgm
|
|
%agpr = call i32 asm sideeffect "; def $0", "=${a1}"()
|
|
call void asm sideeffect "; use $0", "${s9}"(i32 %agpr)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @illegal_agpr_to_sgpr_copy_v2i32() #1 {
|
|
; GCN-LABEL: illegal_agpr_to_sgpr_copy_v2i32:
|
|
; GCN: ; %bb.0:
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; def a[0:1]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: v_accvgpr_read_b32 v0, a0
|
|
; GCN-NEXT: v_accvgpr_read_b32 v1, a1
|
|
; GCN-NEXT: v_readfirstlane_b32 s11, v1
|
|
; GCN-NEXT: v_readfirstlane_b32 s10, v0
|
|
; GCN-NEXT: ;;#ASMSTART
|
|
; GCN-NEXT: ; use s[10:11]
|
|
; GCN-NEXT: ;;#ASMEND
|
|
; GCN-NEXT: s_endpgm
|
|
%vgpr = call <2 x i32> asm sideeffect "; def $0", "=${a[0:1]}"()
|
|
call void asm sideeffect "; use $0", "${s[10:11]}"(<2 x i32> %vgpr)
|
|
ret void
|
|
}
|
|
|
|
attributes #0 = { nounwind }
|
|
attributes #1 = { nounwind "target-cpu"="gfx908" }
|