[Mlir-commits] [mlir] [mlir][EmitC] Expand the MemRefToEmitC pass - Lowering `CopyOp` (PR #151206)

[Gil Rapaport]

================
@@ -57,31 +67,33 @@ struct ConvertMemRefToEmitCPass
     mlir::ModuleOp module = getOperation();
     module.walk([&](mlir::emitc::CallOpaqueOp callOp) {
       if (callOp.getCallee() != alignedAllocFunctionName &&
-          callOp.getCallee() != mallocFunctionName) {
+          callOp.getCallee() != mallocFunctionName &&
+          callOp.getCallee() != memcpyFunctionName)
         return mlir::WalkResult::advance();
-      }
 
       for (auto &op : *module.getBody()) {
         emitc::IncludeOp includeOp = llvm::dyn_cast<mlir::emitc::IncludeOp>(op);
-        if (!includeOp) {
----------------
aniragil wrote:

> ideally, yes. but im already modifying this portion of code and this would be a single line change.

Practically too. It's not about the number of lines or their proximity to other changes. [LLVM's contribution policy](https://llvm.org/docs/Contributing.html#how-to-submit-a-patch) requires patches to be minimal. More specifically:
```
* not contain any unrelated changes
* be an isolated change. Independent changes should be submitted as separate patches as this makes reviewing easier.
```

https://github.com/llvm/llvm-project/pull/151206