Files
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
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

219 lines
6.1 KiB
LLVM

; RUN: llc -mtriple=powerpc64-unknown-linux-gnu -mcpu=pwr7 < %s | FileCheck %s
; RUN: llc -mtriple=powerpc64-unknown-linux-gnu -mcpu=pwr7 -frame-pointer=all < %s | FileCheck -check-prefix=CHECK-FP %s
; RUN: llc -mtriple=powerpc-unknown-linux-gnu -frame-pointer=all < %s | FileCheck -check-prefix=CHECK-32 %s
; RUN: llc -mtriple=powerpc-unknown-linux-gnu -frame-pointer=all -relocation-model=pic < %s | FileCheck -check-prefix=CHECK-32-PIC %s
target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-f128:128:128-v128:128:128-n32:64"
target triple = "powerpc64-unknown-linux-gnu"
%struct.s = type { i32, i32 }
declare void @bar(ptr)
@barbaz = external global i32
define void @goo(ptr byval(%struct.s) nocapture readonly %a) {
entry:
%x = alloca [2 x i32], align 32
%0 = load i32, ptr %a, align 4
store i32 %0, ptr %x, align 32
%b = getelementptr inbounds %struct.s, ptr %a, i64 0, i32 1
%1 = load i32, ptr %b, align 4
%2 = load i32, ptr @barbaz, align 4
%arrayidx2 = getelementptr inbounds [2 x i32], ptr %x, i64 0, i64 1
store i32 %2, ptr %arrayidx2, align 4
call void @bar(ptr %x)
ret void
}
; CHECK-LABEL: @goo
; CHECK-DAG: mflr {{[0-9]+}}
; CHECK-DAG: clrldi [[REG:[0-9]+]], 1, 59
; CHECK-DAG: std 30, -16(1)
; CHECK-DAG: mr 30, 1
; CHECK-DAG: std 0, 16(1)
; CHECK-DAG: subfic 0, [[REG]], -160
; CHECK: stdux 1, 1, 0
; CHECK: .cfi_def_cfa_register r30
; CHECK: .cfi_offset r30, -16
; CHECK: .cfi_offset lr, 16
; CHECK: std 3, 48(30)
; CHECK: mr 1, 30
; CHECK-DAG: ld [[SR:[0-9]+]], 16(1)
; CHECK-DAG: ld 30, -16(1)
; CHECK-DAG: mtlr [[SR]]
; CHECK: blr
; CHECK-FP-LABEL: @goo
; CHECK-FP-DAG: mflr {{[0-9]+}}
; CHECK-FP-DAG: clrldi [[REG:[0-9]+]], 1, 59
; CHECK-FP-DAG: std 31, -8(1)
; CHECK-FP-DAG: std 30, -16(1)
; CHECK-FP-DAG: mr 30, 1
; CHECK-FP-DAG: std 0, 16(1)
; CHECK-FP-DAG: subfic 0, [[REG]], -160
; CHECK-FP: stdux 1, 1, 0
; CHECK-FP: .cfi_def_cfa_register r30
; CHECK-FP: .cfi_offset r31, -8
; CHECK-FP: .cfi_offset r30, -16
; CHECK-FP: .cfi_offset lr, 16
; CHECK-FP: mr 31, 1
; CHECK-FP: std 3, 48(30)
; CHECK-FP: mr 1, 30
; CHECK-FP-DAG: ld [[SR:[0-9]+]], 16(1)
; CHECK-FP-DAG: ld 31, -8(1)
; CHECK-FP-DAG: ld 30, -16(1)
; CHECK-FP-DAG: mtlr [[SR]]
; CHECK-FP: blr
; CHECK-32-LABEL: @goo
; CHECK-32-DAG: mflr [[LR:[0-9]+]]
; CHECK-32-DAG: clrlwi [[REG:[0-9]+]], 1, 27
; CHECK-32-DAG: stw [[LR]], 4(1)
; CHECK-32-DAG: subfic 0, [[REG]], -64
; CHECK-32: stwux 1, 1, 0
; CHECK-32: sub 0, 1, 0
; CHECK-32: addic 0, 0, -4
; CHECK-32: stwx 31, 0, 0
; CHECK-32: addic 0, 0, -4
; CHECK-32: stwx 30, 0, 0
; CHECK-32: addic 30, 0, 8
; CHECK-32-PIC-LABEL: @goo
; CHECK-32-PIC-DAG: mflr [[LR:[0-9]+]]
; CHECK-32-PIC-DAG: clrlwi [[REG:[0-9]+]], 1, 27
; CHECK-32-PIC-DAG: stw [[LR]], 4(1)
; CHECK-32-PIC-DAG: subfic 0, [[REG]], -64
; CHECK-32-PIC: stwux 1, 1, 0
; CHECK-32-PIC: sub 0, 1, 0
; CHECK-32-PIC: addic 0, 0, -4
; CHECK-32-PIC: stwx 31, 0, 0
; CHECK-32-PIC: addic 0, 0, -4
; CHECK-32-PIC: stwx 30, 0, 0
; CHECK-32-PIC: addic 0, 0, -4
; CHECK-32-PIC: stwx 29, 0, 0
; CHECK-32-PIC: addic 29, 0, 12
; The large-frame-size case.
define void @hoo(ptr byval(%struct.s) nocapture readonly %a) {
entry:
%x = alloca [200000 x i32], align 32
%0 = load i32, ptr %a, align 4
store i32 %0, ptr %x, align 32
%b = getelementptr inbounds %struct.s, ptr %a, i64 0, i32 1
%1 = load i32, ptr %b, align 4
%arrayidx2 = getelementptr inbounds [200000 x i32], ptr %x, i64 0, i64 1
store i32 %1, ptr %arrayidx2, align 4
call void @bar(ptr %x)
ret void
}
; CHECK-LABEL: @hoo
; CHECK-DAG: lis [[REG1:[0-9]+]], -13
; CHECK-DAG: clrldi [[REG3:[0-9]+]], 1, 59
; CHECK-DAG: mflr {{[0-9]+}}
; CHECK-DAG: ori [[REG2:[0-9]+]], [[REG1]], 51808
; CHECK-DAG: std 30, -16(1)
; CHECK-DAG: mr 30, 1
; CHECK-DAG: std 0, 16(1)
; CHECK-DAG: subc 0, [[REG2]], [[REG3]]
; CHECK: stdux 1, 1, 0
; CHECK: .cfi_def_cfa_register r30
; CHECK: blr
; CHECK-32-LABEL: @hoo
; CHECK-32-DAG: lis [[REG1:[0-9]+]], -13
; CHECK-32-DAG: clrlwi [[REG3:[0-9]+]], 1, 27
; CHECK-32-DAG: mflr [[LR:[0-9]+]]
; CHECK-32-DAG: ori [[REG2:[0-9]+]], [[REG1]], 51904
; CHECK-32-DAG: stw [[LR]], 4(1)
; CHECK-32-DAG: subc 0, [[REG2]], [[REG3]]
; CHECK-32: stwux 1, 1, 0
; CHECK-32: sub 0, 1, 0
; CHECK-32: addic 0, 0, -4
; CHECK-32: stwx 31, 0, 0
; CHECK-32: addic 0, 0, -4
; CHECK-32: stwx 30, 0, 0
; CHECK-32: addic 30, 0, 8
; CHECK-32: blr
; CHECK-32-PIC-LABEL: @hoo
; CHECK-32-PIC-DAG: lis [[REG1:[0-9]+]], -13
; CHECK-32-PIC-DAG: clrlwi [[REG3:[0-9]+]], 1, 27
; CHECK-32-PIC-DAG: mflr {{[0-9]+}}
; CHECK-32-PIC-DAG: ori [[REG2:[0-9]+]], [[REG1]], 51904
; CHECK-32-PIC-DAG: stw 0, 4(1)
; CHECK-32-PIC-DAG: subc 0, [[REG2]], [[REG3]]
; CHECK-32-PIC: stwux 1, 1, 0
; CHECK-32-PIC: sub 0, 1, 0
; CHECK-32-PIC: addic 0, 0, -4
; CHECK-32-PIC: stwx 31, 0, 0
; CHECK-32-PIC: addic 0, 0, -8
; CHECK-32-PIC: stwx 29, 0, 0
; CHECK-32-PIC: addic 29, 0, 12
; CHECK-32: blr
; Make sure that the FP save area is still allocated correctly relative to
; where r30 is saved.
define void @loo(ptr byval(%struct.s) nocapture readonly %a) {
entry:
%x = alloca [2 x i32], align 32
%0 = load i32, ptr %a, align 4
store i32 %0, ptr %x, align 32
%b = getelementptr inbounds %struct.s, ptr %a, i64 0, i32 1
%1 = load i32, ptr %b, align 4
%arrayidx2 = getelementptr inbounds [2 x i32], ptr %x, i64 0, i64 1
store i32 %1, ptr %arrayidx2, align 4
call void @bar(ptr %x)
call void asm sideeffect "", "~{f30}"() nounwind
ret void
}
; CHECK-LABEL: @loo
; CHECK-DAG: mflr {{[0-9]+}}
; CHECK-DAG: clrldi [[REG:[0-9]+]], 1, 59
; CHECK-DAG: std 30, -32(1)
; CHECK-DAG: mr 30, 1
; CHECK-DAG: std 0, 16(1)
; CHECK-DAG: subfic 0, [[REG]], -192
; CHECK: stdux 1, 1, 0
; CHECK: .cfi_def_cfa_register r30
; CHECK: stfd 30, -16(30)
; CHECK: blr
; CHECK-FP-LABEL: @loo
; CHECK-FP-DAG: mflr {{[0-9]+}}
; CHECK-FP-DAG: clrldi [[REG:[0-9]+]], 1, 59
; CHECK-FP-DAG: std 31, -24(1)
; CHECK-FP-DAG: std 30, -32(1)
; CHECK-FP-DAG: mr 30, 1
; CHECK-FP-DAG: std 0, 16(1)
; CHECK-FP-DAG: subfic 0, [[REG]], -192
; CHECK-FP: stdux 1, 1, 0
; CHECK-FP: .cfi_def_cfa_register r30
; CHECK-FP: stfd 30, -16(30)
; CHECK-FP: blr