[PATCH] D35416: AMDGPU: Fix crash when folding immediates into multiple uses

[Phabricator via Phabricator via llvm-commits]

This revision was automatically updated to reflect the committed changes.
Closed by commit rL308314: AMDGPU: Fix crash when folding immediates into multiple uses (authored by nha).

Changed prior to commit:
  https://reviews.llvm.org/D35416?vs=106638&id=107100#toc

Repository:
  rL LLVM

https://reviews.llvm.org/D35416

Files:
  llvm/trunk/lib/Target/AMDGPU/SIFoldOperands.cpp
  llvm/trunk/test/CodeGen/MIR/AMDGPU/fold-multiple.mir


Index: llvm/trunk/test/CodeGen/MIR/AMDGPU/fold-multiple.mir
===================================================================
--- llvm/trunk/test/CodeGen/MIR/AMDGPU/fold-multiple.mir
+++ llvm/trunk/test/CodeGen/MIR/AMDGPU/fold-multiple.mir
@@ -0,0 +1,40 @@
+# RUN: llc --mtriple=amdgcn--amdhsa -mcpu=fiji -verify-machineinstrs -run-pass si-fold-operands,si-shrink-instructions %s -o - | FileCheck %s
+--- |
+  define amdgpu_kernel void @test() #0 {
+    ret void
+  }
+
+  attributes #0 = { nounwind }
+
+...
+---
+
+# This used to crash / trigger an assertion, because re-scanning the use list
+# after constant-folding the definition of %3 lead to the definition of %2
+# being processed twice.
+
+# CHECK-LABEL: name: test
+# CHECK: %2 = V_LSHLREV_B32_e32 2, killed %0, implicit %exec
+# CHECK: %4 = V_AND_B32_e32 8, killed %2, implicit %exec
+
+name:            test
+tracksRegLiveness: true
+registers:
+  - { id: 0, class: vgpr_32 }
+  - { id: 1, class: sreg_32 }
+  - { id: 2, class: vgpr_32 }
+  - { id: 3, class: sreg_32 }
+  - { id: 4, class: vgpr_32 }
+  - { id: 5, class: sreg_128 }
+body:             |
+  bb.0 (%ir-block.0):
+    %0 = IMPLICIT_DEF
+    %1 = S_MOV_B32 2
+    %2 = V_LSHLREV_B32_e64 %1, killed %0, implicit %exec
+    %3 = S_LSHL_B32 %1, killed %1, implicit-def dead %scc
+    %4 = V_AND_B32_e64 killed %2, killed %3, implicit %exec
+    %5 = IMPLICIT_DEF
+    BUFFER_STORE_DWORD_OFFSET killed %4, killed %5, 0, 0, 0, 0, 0, implicit %exec
+    S_ENDPGM
+
+...
Index: llvm/trunk/lib/Target/AMDGPU/SIFoldOperands.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SIFoldOperands.cpp
+++ llvm/trunk/lib/Target/AMDGPU/SIFoldOperands.cpp
@@ -653,6 +653,7 @@
         // again. The same constant folded instruction could also have a second
         // use operand.
         NextUse = MRI->use_begin(Dst.getReg());
+        FoldList.clear();
         continue;
       }
 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D35416.107100.patch
Type: text/x-patch
Size: 1950 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170718/eedda0df/attachment.bin>