[Mlir-commits] [mlir] Expand the MemRefToEmitC pass - Lowering `AllocOp` (PR #148257)

[Paul Kirth]

================
@@ -77,6 +77,38 @@ struct ConvertAlloca final : public OpConversionPattern<memref::AllocaOp> {
   }
 };
 
+struct ConvertAlloc final : public OpConversionPattern<memref::AllocOp> {
+  using OpConversionPattern::OpConversionPattern;
+  LogicalResult
+  matchAndRewrite(memref::AllocOp allocOp, OpAdaptor operands,
+                  ConversionPatternRewriter &rewriter) const override {
+    mlir::Location loc = allocOp.getLoc();
+    auto memrefType = allocOp.getType();
+    if (!memrefType.hasStaticShape())
+      return rewriter.notifyMatchFailure(
+          allocOp.getLoc(), "cannot transform alloc op with dynamic shape");
+
+    int64_t totalSize =
+        memrefType.getNumElements() * memrefType.getElementTypeBitWidth() / 8;
----------------
ilovepi wrote:

In some contexts bits/byte aren't guaranteed to be 8. IDK if that's the case here or if there's a API we can use to guarantee we use the right constants. If this pattern is used elsewhere its fine. I just know we've run into similar issues on the LLVM side, and its often hard to run down.

https://github.com/llvm/llvm-project/pull/148257