[mlir] Implement DestinationStyleOpInterface for scf::ForallOp (#66981)

`scf::ForallOp` has `shared_outs` tensor operands which are used to
insert partial results into in the parallel terminator. The
`scf::ForallOp` returns one tensor for each `shared_out` which then
contains the combined result from all threads. Since the parallel
terminator cannot change the shape of the `shared_out`, ForallOp is a
`DestinationStyleOp` and this patch implements the interface by
declaring the `outputs` operands as `inits` in the language of the DPS
interface.

For this change to work, we need to add an exception to the Pattern that
folds `tensor.cast` Ops into DPS Ops because `scf::Forall` needs special
handling of its `BlockArgument` Type during this folding.

Author: ubfx

mlir/include/mlir/Dialect/SCF/IR/SCF.h

@@ -19,6 +19,7 @@
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/RegionKindInterface.h"
 #include "mlir/Interfaces/ControlFlowInterfaces.h"
+#include "mlir/Interfaces/DestinationStyleOpInterface.h"
 #include "mlir/Interfaces/InferTypeOpInterface.h"
 #include "mlir/Interfaces/LoopLikeInterface.h"
 #include "mlir/Interfaces/ParallelCombiningOpInterface.h"

mlir/include/mlir/Dialect/SCF/IR/SCFOps.td

@@ -17,6 +17,7 @@ include "mlir/Interfaces/ControlFlowInterfaces.td"
 include "mlir/Interfaces/LoopLikeInterface.td"
 include "mlir/IR/RegionKindInterface.td"
 include "mlir/Dialect/SCF/IR/DeviceMappingInterface.td"
+include "mlir/Interfaces/DestinationStyleOpInterface.td"
 include "mlir/Interfaces/InferTypeOpInterface.td"
 include "mlir/Interfaces/ParallelCombiningOpInterface.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
@@ -333,6 +334,7 @@ def ForallOp : SCF_Op<"forall", [
        RecursiveMemoryEffects,
        SingleBlockImplicitTerminator<"scf::InParallelOp">,
        DeclareOpInterfaceMethods<RegionBranchOpInterface>,
+       DestinationStyleOpInterface
      ]> {
   let summary = "evaluate a block multiple times in parallel";
   let description = [{
@@ -630,6 +632,9 @@ def ForallOp : SCF_Op<"forall", [
                                  Location loc);
 
     InParallelOp getTerminator();
+
+    // Declare the shared_outs as inits/outs to DestinationStyleOpInterface.
+    MutableOperandRange getDpsInitsMutable() { return getOutputsMutable(); }
   }];
 }
 

mlir/lib/Dialect/Tensor/IR/TensorOps.cpp

@@ -22,6 +22,7 @@
 #include "mlir/IR/OpDefinition.h"
 #include "mlir/IR/TypeUtilities.h"
 #include "mlir/Interfaces/DestinationStyleOpInterface.h"
+#include "mlir/Interfaces/LoopLikeInterface.h"
 #include "mlir/Support/MathExtras.h"
 #include "llvm/ADT/DenseSet.h"
 #include "llvm/ADT/STLExtras.h"
@@ -3970,6 +3971,11 @@ struct FoldTensorCastProducerOp
     if (isa<InsertSliceOp>(op.getOperation()))
       return failure();
 
+    // Exclude DPS ops that are also LoopLike from this interface as they
+    // might need special handling of attached regions.
+    if (isa<LoopLikeOpInterface>(op.getOperation()))
+      return failure();
+
     // If no operand comes from a tensor::CastOp and can be folded then fail.
     bool hasTensorCastOperand =
         llvm::any_of(op->getOpOperands(), [&](OpOperand &opOperand) {