[Mlir-commits] [mlir] [mlir][tosa] Support unranked input/weight tensors for convolution ops (PR #134856)

[Elen Kalda]

================
@@ -282,14 +282,14 @@ static LogicalResult verifyConvOp(T op) {
   // tensors.
   auto inputType = llvm::dyn_cast<RankedTensorType>(op.getInput().getType());
   if (!inputType) {
-    op.emitOpError("expect a ranked tensor for input, got ") << op.getInput();
-    return failure();
+    // Skip following checks if input is not ranked
+    return success();
----------------
ekalda wrote:

I think we're missing out on weights and bias checks in the case where input is unranked, but the weights and/or bias are ranked.

https://github.com/llvm/llvm-project/pull/134856