[Mlir-commits] [mlir] [mlir][reducer] Separate Reduction Steps in `findOptimal` and `applyPatterns` (PR #190560)

[llvmlistbot at llvm.org]

================
@@ -125,86 +75,118 @@ static LogicalResult findOptimal(ModuleOp module, Region &region,
   // the path and apply the reducer to it.
   SmallVector<ReductionNode *> trace;
   ReductionNode *curNode = smallestNode;
-  trace.push_back(curNode);
   while (curNode != root) {
-    curNode = curNode->getParent();
     trace.push_back(curNode);
+    curNode = curNode->getParent();
   }
 
+  if (trace.empty())
+    // If trace is empty, then the smallestNode == root and therefore we were
+    // not successful in reducing the module
+    return failure();
+
   // Reduce the region through the optimal path.
   while (!trace.empty()) {
     ReductionNode *top = trace.pop_back_val();
-    applyPatterns(region, patterns, top->getStartRanges(), eraseOpNotInRange);
+    applyFn(region, top->getStartRanges());
   }
 
-  if (test.isInteresting(module).first != Tester::Interestingness::True)
+  std::pair<Tester::Interestingness, size_t> finalStatus =
+      test.isInteresting(module);
+
+  if (finalStatus.first != Tester::Interestingness::True)
     llvm::report_fatal_error("Reduced module is not interesting");
-  if (test.isInteresting(module).second != smallestNode->getSize())
+  if (finalStatus.second != smallestNode->getSize())
----------------
aidint wrote:

This is to avoid multiple execution of test script in case the module is interesting.

https://github.com/llvm/llvm-project/pull/190560