Changes to `SCFFuseProducerOfSliceResult` to also return the operations created during fusion.

[mravishankar]

Summary

This is follow up to https://reviews.llvm.org/D145133 that allows
propogating information about ops that are fused back to the caller.

[mravishankar]

Rebase

mlir/include/mlir/Dialect/SCF/Transforms/TileUsingInterface.h

 
 /// Fuse the producer of the source of `candidateSliceOp` by computing the
 /// required slice of the producer in-place.  Note that the method
 /// replaces the uses of `candidateSliceOp` with the tiled and fused producer
 /// value but does not delete the slice operation.
 struct SCFFuseProducerOfSliceResult {
   OpResult origProducer;       // Original untiled producer.
   Value tiledAndFusedProducer; // Tile and fused producer value.
+  SmallVector<Operation *> tiledOps;
 };
 std::optional<SCFFuseProducerOfSliceResult>
 tileAndFuseProducerOfSlice(RewriterBase &rewriter,
                            tensor::ExtractSliceOp candidateSliceOp,
                            MutableArrayRef<scf::ForOp> loops);
 
 /// Reconstruct the fused producer from within the tiled-and-fused code. Based
 /// on the slice of the producer computed in place it is possible that within

mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp

for (auto tileAndFusedOp : tileAndFuseResult->tiledOps) {
         continue;
       scf::ForOp innerMostLoop = loops.back();
       updateDestinationOperandsForTiledOp(
           rewriter, dstOp.getDpsInitOperand(resultNumber)->get(),
           innerMostLoop.getRegionIterArgs()[iterArgNumber.value()]);
     }
   }
   return scf::SCFFuseProducerOfSliceResult{fusableProducer,
-                                           tileAndFuseResult->tiledValues[0]};
+                                           tileAndFuseResult->tiledValues[0],
+                                           tileAndFuseResult->tiledOps};
 }
 
 /// Reconstruct the fused producer from within the tiled-and-fused code.
 void mlir::scf::yieldReplacementForFusedProducer(
     RewriterBase &rewriter, tensor::ExtractSliceOp sliceOp,
     scf::SCFFuseProducerOfSliceResult fusedProducerInfo,
     MutableArrayRef<scf::ForOp> loops) {
-  auto [fusableProducer, fusedProducerValue] = fusedProducerInfo;
+  auto [fusableProducer, fusedProducerValue, tileAndFusedOps] =
+      fusedProducerInfo;
   SmallVector<Value> initValues;
   FailureOr<Value> initValue = tensor::getOrCreateDestination(
       rewriter, fusableProducer.getOwner()->getLoc(), fusableProducer);
   if (succeeded(initValue)) {
     SmallVector<OpFoldResult> resultOffsets = sliceOp.getMixedOffsets();
     SmallVector<OpFoldResult> resultSizes = sliceOp.getMixedSizes();
     SmallVector<Value> yieldedVals =
         yieldTiledValues(rewriter, initValue.value(), fusedProducerValue,
                          resultOffsets, resultSizes, loops);
   }
-  if (auto dstStyleProducer =
-          fusedProducerValue.getDefiningOp<DestinationStyleOpInterface>()) {
+  for (auto tileAndFusedOp : tileAndFusedOps) {
+    auto dstStyleProducer =
+        dyn_cast<DestinationStyleOpInterface>(tileAndFusedOp);
+    if (!dstStyleProducer)
+      continue;
     Value dstValue =
         dstStyleProducer.getDpsInitOperand(fusableProducer.getResultNumber())
             ->get();
     updateDestinationOperandsForTiledOp(
         rewriter, dstValue, loops.back().getRegionIterArgs().back());
   }
 }