[llvm] [SelectionDAG] Handling Oversized Alloca Types under 32 bit Mode to Avoid Code Generator Crash (PR #71472)

Matt Arsenault via llvm-commits llvm-commits at lists.llvm.org
Fri Nov 10 01:49:16 PST 2023


================
@@ -4138,9 +4138,10 @@ void SelectionDAGBuilder::visitAlloca(const AllocaInst &I) {
                                           APInt(IntPtr.getScalarSizeInBits(),
                                                 TySize.getKnownMinValue())));
   else
-    AllocSize =
-        DAG.getNode(ISD::MUL, dl, IntPtr, AllocSize,
-                    DAG.getConstant(TySize.getFixedValue(), dl, IntPtr));
+    AllocSize = DAG.getNode(ISD::MUL, dl, IntPtr, AllocSize,
+                            DAG.getConstant(APInt(IntPtr.getScalarSizeInBits(),
----------------
arsenm wrote:

The code immediately above already does this for the scalable vector case.

```
 EVT IntPtr = TLI.getPointerTy(DAG.getDataLayout(), I.getAddressSpace());
  if (AllocSize.getValueType() != IntPtr)
    AllocSize = DAG.getZExtOrTrunc(AllocSize, dl, IntPtr);
```

You just need to produce the IntVT based on the size (e.g. MVT::getIntegerVT(IntPtr.getScalarSizeInBits()). You also don't need the type check here, getZExtOrTrunc does that for you 

https://github.com/llvm/llvm-project/pull/71472


More information about the llvm-commits mailing list