[Mlir-commits] [mlir] [mlir] Speed up FuncToLLVM: CallOpLowering using a hashtable (PR #68082)

[Mehdi Amini]

================
@@ -765,6 +792,23 @@ struct ConvertFuncToLLVMPass
 
     const auto &dataLayoutAnalysis = getAnalysis<DataLayoutAnalysis>();
 
+    StringSet<> functionsHavingBarePtrAttribute;
+    for (Region &region : m->getRegions()) {
+      for (Block &block : region.getBlocks()) {
+        for (Operation &op : block.getOperations()) {
+          if (auto funcOp = dyn_cast<func::FuncOp>(op);
+              funcOp != nullptr && funcOp->hasAttr(barePtrAttrName)) {
+            bool inserted =
+                functionsHavingBarePtrAttribute.insert(funcOp.getSymName())
+                    .second;
+            (void)inserted;
+            assert(inserted &&
+                   "expected module to contain uniquely named funcOps");
----------------
joker-eph wrote:

I assume there is a verifier for this already?

https://github.com/llvm/llvm-project/pull/68082