3a20597776
Implement an intrinsic for use lowering LDS variables to different addresses from different kernels. This will allow kernels that cannot reach an LDS variable to avoid wasting space for it. There are a number of implicit arguments accessed by intrinsic already so this implementation closely follows the existing handling. It is slightly novel in that this SGPR is written by the kernel prologue. It is necessary in the general case to put variables at different addresses such that they can be compactly allocated and thus necessary for an indirect function call to have some means of determining where a given variable was allocated. Claiming an arbitrary SGPR into which an integer can be written by the kernel, in this implementation based on metadata associated with that kernel, which is then passed on to indirect call sites is sufficient to determine the variable address. The intent is to emit a __const array of LDS addresses and index into it. Reviewed By: arsenm Differential Revision: https://reviews.llvm.org/D125060
93 lines
6.1 KiB
LLVM
93 lines
6.1 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
|
|
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -stop-after=instruction-select -o - %s | FileCheck %s
|
|
; Make sure there are no assertions on dropped debug info
|
|
|
|
declare void @callee()
|
|
|
|
define amdgpu_kernel void @call_debug_loc() {
|
|
; CHECK-LABEL: name: call_debug_loc
|
|
; CHECK: bb.1.entry:
|
|
; CHECK-NEXT: liveins: $sgpr14, $sgpr15, $sgpr16, $vgpr0, $vgpr1, $vgpr2, $sgpr4_sgpr5, $sgpr6_sgpr7, $sgpr8_sgpr9, $sgpr10_sgpr11
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr2, debug-location !6
|
|
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1, debug-location !6
|
|
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr0, debug-location !6
|
|
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr16, debug-location !6
|
|
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr15, debug-location !6
|
|
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr14, debug-location !6
|
|
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sgpr_64 = COPY $sgpr10_sgpr11, debug-location !6
|
|
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sgpr_64 = COPY $sgpr6_sgpr7, debug-location !6
|
|
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sgpr_64 = COPY $sgpr4_sgpr5, debug-location !6
|
|
; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_64 = COPY $sgpr8_sgpr9
|
|
; CHECK-NEXT: ADJCALLSTACKUP 0, 0, implicit-def $scc, debug-location !6
|
|
; CHECK-NEXT: [[COPY10:%[0-9]+]]:sreg_64 = COPY [[COPY8]], debug-location !6
|
|
; CHECK-NEXT: [[COPY11:%[0-9]+]]:sreg_64 = COPY [[COPY7]], debug-location !6
|
|
; CHECK-NEXT: [[COPY12:%[0-9]+]]:sreg_64 = COPY [[COPY6]], debug-location !6
|
|
; CHECK-NEXT: [[COPY13:%[0-9]+]]:sreg_32 = COPY [[COPY5]], debug-location !6
|
|
; CHECK-NEXT: [[COPY14:%[0-9]+]]:sreg_32 = COPY [[COPY4]], debug-location !6
|
|
; CHECK-NEXT: [[COPY15:%[0-9]+]]:sreg_32 = COPY [[COPY3]], debug-location !6
|
|
; CHECK-NEXT: [[DEF:%[0-9]+]]:sreg_32 = IMPLICIT_DEF debug-location !6
|
|
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 10
|
|
; CHECK-NEXT: [[COPY16:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
|
|
; CHECK-NEXT: [[V_LSHLREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHLREV_B32_e64 [[COPY16]], [[COPY1]], implicit $exec, debug-location !6
|
|
; CHECK-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 20
|
|
; CHECK-NEXT: [[COPY17:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]]
|
|
; CHECK-NEXT: [[V_LSHLREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHLREV_B32_e64 [[COPY17]], [[COPY]], implicit $exec, debug-location !6
|
|
; CHECK-NEXT: [[V_OR3_B32_e64_:%[0-9]+]]:vgpr_32 = V_OR3_B32_e64 [[COPY2]], [[V_LSHLREV_B32_e64_]], [[V_LSHLREV_B32_e64_1]], implicit $exec, debug-location !6
|
|
; CHECK-NEXT: [[COPY18:%[0-9]+]]:sgpr_128 = COPY $sgpr0_sgpr1_sgpr2_sgpr3, debug-location !6
|
|
; CHECK-NEXT: $sgpr0_sgpr1_sgpr2_sgpr3 = COPY [[COPY18]], debug-location !6
|
|
; CHECK-NEXT: $sgpr4_sgpr5 = COPY [[COPY10]], debug-location !6
|
|
; CHECK-NEXT: $sgpr6_sgpr7 = COPY [[COPY11]], debug-location !6
|
|
; CHECK-NEXT: $sgpr8_sgpr9 = COPY [[COPY9]], debug-location !6
|
|
; CHECK-NEXT: $sgpr10_sgpr11 = COPY [[COPY12]], debug-location !6
|
|
; CHECK-NEXT: $sgpr12 = COPY [[COPY13]], debug-location !6
|
|
; CHECK-NEXT: $sgpr13 = COPY [[COPY14]], debug-location !6
|
|
; CHECK-NEXT: $sgpr14 = COPY [[COPY15]], debug-location !6
|
|
; CHECK-NEXT: $sgpr15 = COPY [[DEF]], debug-location !6
|
|
; CHECK-NEXT: $vgpr31 = COPY [[V_OR3_B32_e64_]], debug-location !6
|
|
; CHECK-NEXT: [[SI_PC_ADD_REL_OFFSET:%[0-9]+]]:sreg_64 = SI_PC_ADD_REL_OFFSET target-flags(amdgpu-gotprel32-lo) @callee + 4, target-flags(amdgpu-gotprel32-hi) @callee + 12, implicit-def $scc, debug-location !6
|
|
; CHECK-NEXT: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[SI_PC_ADD_REL_OFFSET]], 0, 0, debug-location !6 :: (dereferenceable invariant load (p0) from got, addrspace 4)
|
|
; CHECK-NEXT: $sgpr30_sgpr31 = SI_CALL [[S_LOAD_DWORDX2_IMM]], @callee, csr_amdgpu, implicit $sgpr0_sgpr1_sgpr2_sgpr3, implicit $sgpr4_sgpr5, implicit $sgpr6_sgpr7, implicit $sgpr8_sgpr9, implicit $sgpr10_sgpr11, implicit $sgpr12, implicit $sgpr13, implicit $sgpr14, implicit $sgpr15, implicit $vgpr31, debug-location !6
|
|
; CHECK-NEXT: ADJCALLSTACKDOWN 0, 0, implicit-def $scc, debug-location !6
|
|
; CHECK-NEXT: S_ENDPGM 0
|
|
entry:
|
|
call void @callee(), !dbg !6
|
|
ret void
|
|
}
|
|
|
|
define void @returnaddress_debug_loc(i8* addrspace(1)* %ptr) {
|
|
; CHECK-LABEL: name: returnaddress_debug_loc
|
|
; CHECK: bb.1.entry:
|
|
; CHECK-NEXT: liveins: $vgpr0, $vgpr1, $sgpr30_sgpr31
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr30_sgpr31, debug-location !6
|
|
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr0
|
|
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
|
|
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1
|
|
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY [[COPY]]
|
|
; CHECK-NEXT: GLOBAL_STORE_DWORDX2 [[REG_SEQUENCE]], [[COPY3]], 0, 0, implicit $exec :: (store (p0) into %ir.ptr, addrspace 1)
|
|
; CHECK-NEXT: SI_RETURN
|
|
entry:
|
|
%returnaddr = call i8* @llvm.returnaddress(i32 0), !dbg !6
|
|
store i8* %returnaddr, i8* addrspace(1)* %ptr, align 8
|
|
ret void
|
|
}
|
|
|
|
declare i8* @llvm.returnaddress(i32 immarg) #0
|
|
|
|
attributes #0 = { nofree nosync nounwind readnone willreturn }
|
|
|
|
!llvm.dbg.cu = !{!0}
|
|
!llvm.module.flags = !{!2, !3, !4, !5}
|
|
|
|
!0 = distinct !DICompileUnit(language: DW_LANG_OpenCL, file: !1, producer: "clang version 14.0.0 (git@github.com:llvm/llvm-project.git 4132dc917eddb446405cc5afef41167b8bce360b)", isOptimized: false, runtimeVersion: 0, emissionKind: NoDebug, splitDebugInlining: false, nameTableKind: None)
|
|
!1 = !DIFile(filename: "gisel_1_gfx1031.cl", directory: "/home/matt/builds/conformance/2.0")
|
|
!2 = !{i32 2, !"Debug Info Version", i32 3}
|
|
!3 = !{i32 1, !"wchar_size", i32 4}
|
|
!4 = !{i32 7, !"PIC Level", i32 1}
|
|
!5 = !{i32 7, !"frame-pointer", i32 2}
|
|
!6 = !DILocation(line: 10, column: 16, scope: !7)
|
|
!7 = distinct !DISubprogram(name: "call_debug_loc", scope: !1, file: !1, line: 8, type: !8, scopeLine: 9, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition, unit: !0, retainedNodes: !9)
|
|
!8 = !DISubroutineType(types: !9)
|
|
!9 = !{}
|