mirrors/llvm-project
0
0
Fork 0
mirror of https://github.com/llvm/llvm-project.git synced 2025-04-19 00:56:41 +00:00
Code Issues Projects Releases Wiki Activity

[AMDGPU] Fix crash in the SILoadStoreOptimizer (#93862)

Browse Source 
It does not properly handle situation when address calculation uses
V_ADDC_U32 0, 0, carry-in (i.e. with both src0 and src1 immediates).
This commit is contained in:
Stanislav Mekhanoshin 2024-05-30 14:27:33 -07:00 committed by GitHub
parent c4dad9a6b6
commit 215f92b979
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 27 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
llvm/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp
View File

@ -2034,7 +2034,7 @@ void SILoadStoreOptimizer::processBaseWithConstOffset(const MachineOperand &Base
if (Src0->isImm())
std::swap(Src0, Src1);
if (!Src1->isImm())
if (!Src1->isImm() || Src0->isImm())
return;
uint64_t Offset1 = Src1->getImm();

26
llvm/test/CodeGen/AMDGPU/load-store-opt-addc0.mir Normal file
View File

@ -0,0 +1,26 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
# RUN: llc -march=amdgcn -mcpu=gfx900 -run-pass=si-load-store-opt -o - %s | FileCheck --check-prefix=GCN %s
# This used to crash
---
name: analyze_addc_0_0
body: |
bb.1.entry:
liveins: $vgpr0
; GCN-LABEL: name: analyze_addc_0_0
; GCN: liveins: $vgpr0
; GCN-NEXT: {{ $}}
; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GCN-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY]], 16, 0, implicit $exec
; GCN-NEXT: [[V_ADDC_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADDC_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADDC_U32_e64 0, 0, killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec
; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, [[V_ADDC_U32_e64_]], %subreg.sub1
; GCN-NEXT: [[GLOBAL_LOAD_DWORDX4_:%[0-9]+]]:vreg_128 = GLOBAL_LOAD_DWORDX4 [[REG_SEQUENCE]], 0, 0, implicit $exec
%0:vgpr_32 = COPY $vgpr0
%1:vgpr_32, %2:sreg_64_xexec = V_ADD_CO_U32_e64 %0, 16, 0, implicit $exec
%3:vgpr_32, dead %26:sreg_64_xexec = V_ADDC_U32_e64 0, 0, killed %2, 0, implicit $exec
%4:vreg_64 = REG_SEQUENCE %1, %subreg.sub0, %3, %subreg.sub1
%5:vreg_128 = GLOBAL_LOAD_DWORDX4 %4, 0, 0, implicit $exec
...