[Mlir-commits] [mlir] [mlir][LLVM] Prevent vector bit-width overflow (PR #220898)

[Matthias Springer]

================
@@ -1589,9 +1589,14 @@ def LLVM_vector_insert
     "type($srcvec) `into` type($res)";
   let extraClassDeclaration = [{
     uint64_t getVectorBitWidth(Type vector) {
-      return getVectorNumElements(vector).getKnownMinValue() *
-             ::llvm::cast<VectorType>(vector).getElementType()
-                .getIntOrFloatBitWidth();
+      auto vectorType = ::llvm::cast<VectorType>(vector);
+      // A scalable vector dimension is its known minimum element count.
+      uint64_t minNumElements = vectorType.getDimSize(0);
+      uint64_t elementBitWidth =
+          vectorType.getElementType().getIntOrFloatBitWidth();
+      if (minNumElements > uint64_t(-1) / elementBitWidth)
----------------
matthias-springer wrote:

nit: `std::numeric_limits<uint64_t>::max()` instead of `uint64_t(-1)`?

https://github.com/llvm/llvm-project/pull/220898