[llvm] [AMDGPU] PromoteAlloca: reject known out-of-bounds index (PR #139700)

[Robert Imschweiler via llvm-commits]

================
@@ -438,7 +438,8 @@ static Value *GEPToVectorIndex(GetElementPtrInst *GEP, AllocaInst *Alloca,
   SmallMapVector<Value *, APInt, 4> VarOffsets;
   APInt ConstOffset(BW, 0);
   if (GEP->getPointerOperand()->stripPointerCasts() != Alloca ||
-      !GEP->collectOffset(DL, BW, VarOffsets, ConstOffset))
+      !GEP->collectOffset(DL, BW, VarOffsets, ConstOffset) ||
+      ConstOffset.getZExtValue() >= Alloca->getAllocationSize(DL))
----------------
ro-i wrote:

True, I separated it.
It would also be possible to implement the check specifically here (`SrcBegin`):

https://github.com/llvm/llvm-project/blob/84c1564d1825880d463d9f85153812f1f6805289/llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp#L663-L675

but that wouldn't cover as many cases and I think that doesn't hurt (?)

https://github.com/llvm/llvm-project/pull/139700