[PATCH] D81952: Fix ubsan error in tblgen with signed left shift

[Stanislav Mekhanoshin via Phabricator via llvm-commits]

rampitec created this revision.
rampitec added a reviewer: arsenm.
Herald added subscribers: kerbowa, hiraditya, nhaehnle, wdng, jvesely.
Herald added a project: LLVM.

UBSAN complains when tblgen performs SHL of a negative
value.


https://reviews.llvm.org/D81952

Files:
  llvm/lib/TableGen/Record.cpp
  llvm/lib/Target/AMDGPU/SMInstructions.td


Index: llvm/lib/Target/AMDGPU/SMInstructions.td
===================================================================
--- llvm/lib/Target/AMDGPU/SMInstructions.td
+++ llvm/lib/Target/AMDGPU/SMInstructions.td
@@ -860,7 +860,7 @@
 def : GCNPat <
   (i64 (readcyclecounter)),
   (REG_SEQUENCE SReg_64,
-    (S_GETREG_B32 -26595), sub0,
+    (S_GETREG_B32 getHwRegImm<HWREG.SHADER_CYCLES, 0, -12>.ret), sub0,
     (S_MOV_B32 (i32 0)), sub1)
 >;
 } // let OtherPredicates = [HasNoSMemTimeInst]
Index: llvm/lib/TableGen/Record.cpp
===================================================================
--- llvm/lib/TableGen/Record.cpp
+++ llvm/lib/TableGen/Record.cpp
@@ -1030,7 +1030,7 @@
       case MUL: Result = LHSv *  RHSv; break;
       case AND: Result = LHSv &  RHSv; break;
       case OR: Result = LHSv | RHSv; break;
-      case SHL: Result = LHSv << RHSv; break;
+      case SHL: Result = (uint64_t)LHSv << (uint64_t)RHSv; break;
       case SRA: Result = LHSv >> RHSv; break;
       case SRL: Result = (uint64_t)LHSv >> (uint64_t)RHSv; break;
       }


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D81952.271139.patch
Type: text/x-patch
Size: 1055 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20200616/e5ffb858/attachment-0001.bin>