[mlir][Transform] NFC - Various API cleanups and use RewriterBase in lieu of PatternRewriter

[nicolasvasilache]

Summary

Depends on: D145685

[nicolasvasilache]

Rebase

[nicolasvasilache]

Rebase

[nicolasvasilache]

Fix insertion point

[ftynse]

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

34

Nit: extra vertical whitespcae.

38

FYI: we have the mechanism for canonicalizations to be defined as a method instead of a pattern. It can be reworked and extended to have PatternSet accept a function and wrap it into a pattern class.

[nicolasvasilache]

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

38

cool yeah we should def switch to that

mlir/include/mlir/Dialect/GPU/TransformOps/GPUTransformOps.h

 #define GET_OP_CLASSES
 #include "mlir/Dialect/GPU/TransformOps/GPUTransformOps.h.inc"
 
 namespace mlir {
 class DialectRegistry;
 namespace transform {
 namespace gpu {
 
-/// Searches `scf.forall` ops nested under `target` and maps each such
-/// op to GPU threads. Mapping is one-to-one and the induction variables of
-/// `scf.forall` are rewritten to gpu.thread_id according to the
-/// thread_dim_apping attribute. Sibling `scf.forall` are supported in
-/// which case, the union of the number of threads is computed and may result in
-/// predication. Dynamic, `scf.forall` trip counts are currently not
-/// supported. Dynamic block dim sizes are currently not supported.
-DiagnosedSilenceableFailure mapNestedForeachToThreadsImpl(
+/// Search `scf.forall` ops nested under `target` and map each such op to GPU
+/// threads. Mapping is one-to-one and the induction variables of `scf.forall`
+/// are rewritten to gpu.thread_id according to the thread_dim_mapping
+/// attribute.
+/// Sibling `scf.forall` are supported in which case, the union of the number of
+/// threads is computed and may result in predication.
+/// Dynamic, `scf.forall` trip counts are currently not supported.
+/// Dynamic block dim sizes are currently not supported.
+DiagnosedSilenceableFailure mapNestedForallToThreadsImpl(
     RewriterBase &rewriter, Operation *target,
     const SmallVectorImpl<int64_t> &blockDim,
     function_ref<void(RewriterBase &, scf::ForallOp, SmallVectorImpl<Value> &)>
         threadIdGenerator,
     bool syncAfterDistribute, std::optional<TransformOpInterface> transformOp,
     const ArrayRef<DeviceMappingAttrInterface> &threadMappingAttributes);
 
-/// Maps the top level `scf.forall` op to GPU Thread Blocks. Mapping is
-/// one-to-one and the induction variables of `scf.forall` are rewritten
-/// to gpu.block_id according to the thread_dim_apping attribute. Dynamic,
-/// `scf.forall` trip counts are currently not supported. Dynamic block
-/// dim sizes are currently not supported.
-DiagnosedSilenceableFailure mapForeachToBlocksImpl(
+/// Map the top level `scf.forall` op to GPU Thread Blocks.
+/// Mapping is one-to-one and the induction variables of `scf.forall` are
+/// rewritten to gpu.block_id according to the thread_dim_apping attribute.
+/// Dynamic, `scf.forall` trip counts are currently not supported.
+/// Dynamic block dim sizes are currently not supported.
+DiagnosedSilenceableFailure mapForallToBlocksImpl(
     RewriterBase &rewriter, scf::ForallOp forallOp,
     function_ref<void(RewriterBase &, scf::ForallOp, SmallVectorImpl<Value> &)>
         blockIdGenerator,
     SmallVectorImpl<int64_t> &gridDims, TransformOpInterface transformOp,
     const ArrayRef<DeviceMappingAttrInterface> &mappingAttributes);
 
-/// Finds the top level scf::ForallOp of given target.
+/// Find the unique top level scf::ForallOp within a given target op.
 DiagnosedSilenceableFailure
 findTopLevelForallOp(Operation *target, scf::ForallOp &topLevelForallOp,
                      TransformOpInterface transformOp);
 
 } // namespace gpu
 } // namespace transform
 
 namespace gpu {
 void registerTransformDialectExtension(DialectRegistry &registry);
 } // namespace gpu
 } // namespace mlir
 
 #endif // MLIR_DIALECT_GPU_TRANSFORMOPS_GPUTRANSFORMOPS_H

mlir/include/mlir/Dialect/GPU/TransformOps/GPUTransformOps.td

 #define GPU_TRANSFORM_OPS
 
 include "mlir/Dialect/Transform/IR/TransformDialect.td"
 include "mlir/Dialect/Transform/IR/TransformInterfaces.td"
 include "mlir/Dialect/PDL/IR/PDLTypes.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
 include "mlir/IR/OpBase.td"
 
-def MapNestedForeachToThreads :
-  Op<Transform_Dialect, "gpu.map_nested_foreach_to_threads",
+def MapNestedForallToThreads :
+  Op<Transform_Dialect, "gpu.map_nested_forall_to_threads",
     [FunctionalStyleTransformOpTrait,
      MemoryEffectsOpInterface,
      TransformEachOpTrait,
      TransformOpInterface]> {
   let description = [{
       Target the `gpu.launch op` and rewrite all `scf.forall`
       nested in it to distributed `gpu.thread_id` attribute.
 
let extraClassDeclaration = [{
     ::mlir::DiagnosedSilenceableFailure applyToOne(
         ::mlir::Operation *target,
         ::mlir::transform::ApplyToEachResultList &results,
         ::mlir::transform::TransformState &state);
   }];
 }
 
 
-def MapForeachToBlocks :
-  Op<Transform_Dialect, "gpu.map_foreach_to_blocks",
+def MapForallToBlocks :
+  Op<Transform_Dialect, "gpu.map_forall_to_blocks",
     [FunctionalStyleTransformOpTrait,
      MemoryEffectsOpInterface,
      TransformOpInterface,
      TransformEachOpTrait]> {
   let description = [{
     Target the gpu_launch op and rewrite the top level `scf.forall`
     to distributed gpu.block_id attribute. If `generate_gpu_launch` attribute
     is set, then first generates `gpu_launch` and moves the top level

mlir/include/mlir/Dialect/Linalg/Transforms/Transforms.h

 LogicalResult vectorize(RewriterBase &rewriter, LinalgOp linalgOp,
                         ArrayRef<int64_t> inputVectorSizes = {},
                         bool vectorizeNDExtract = false);
 
 /// Emit a suitable vector form for a Copy op with fully static shape.
 LogicalResult vectorizeCopy(RewriterBase &builder, memref::CopyOp copyOp);
 
 /// Emit a loop nest of `scf.for` with the proper body for `linalgOp`.
-FailureOr<LinalgLoops> linalgOpToLoops(PatternRewriter &rewriter,
+FailureOr<LinalgLoops> linalgOpToLoops(RewriterBase &rewriter,
                                        LinalgOp linalgOp);
 
 /// Emit a loop nest of `scf.parallel` with the proper body for `linalgOp`.
-FailureOr<LinalgLoops> linalgOpToParallelLoops(PatternRewriter &rewriter,
+FailureOr<LinalgLoops> linalgOpToParallelLoops(RewriterBase &rewriter,
                                                LinalgOp linalgOp);
 
 /// Emit a loop nest of `affine.for` with the proper body for `linalgOp`.
-FailureOr<LinalgLoops> linalgOpToAffineLoops(PatternRewriter &rewriter,
+FailureOr<LinalgLoops> linalgOpToAffineLoops(RewriterBase &rewriter,
                                              LinalgOp linalgOp);
 
 /// Creates a number of ranges equal to the number of non-zero in `tileSizes`.
 /// One for each loop of the LinalgOp that is tiled. The `tileSizes` argument
 /// has one entry per surrounding loop. It uses zero as the convention that a
 /// particular loop is not tiled. This convention simplifies implementations
 /// by avoiding affine map manipulations. The returned ranges correspond to
 /// the loop ranges, in the proper order, that are tiled and for which new

 /// ```
 struct SplitReductionResult {
   Operation *initOrAlloc;
   FillOp fillOp;
   LinalgOp splitLinalgOp;
   LinalgOp resultCombiningLinalgOp;
 };
 FailureOr<SplitReductionResult>
-splitReduction(PatternRewriter &b, LinalgOp op,
+splitReduction(RewriterBase &b, LinalgOp op,
                const ControlSplitReductionFn &controlSplitReductionFn,
                bool useAlloc = false);
 
 /// Scaling-based implementation of the split reduction transformation.
 /// Instead of introducing an ExpandShapeOp, this rewrites a reduction
 /// dimension `k` into `k * scale + kk`.
 ///
 /// Example:

 ///      ^bb0(%arg3: f32, %arg4: f32):
 ///        %5 = arith.addf %arg3, %arg4 : f32
 ///        linalg.yield %5 : f32
 ///  } -> tensor<16x32xf32>
 ///
 ///  return %4 : tensor<16x32xf32>
 /// ```
 FailureOr<SplitReductionResult>
-splitReductionByScaling(PatternRewriter &b, LinalgOp op,
+splitReductionByScaling(RewriterBase &b, LinalgOp op,
                         const ControlSplitReductionFn &controlSplitReductionFn,
                         bool useAlloc = false);
 
 /// Collapses dimensions of linalg.generic operation. It also collapses inputs
 /// before the op and expands outputs after the op.
 FailureOr<SmallVector<Value>> collapseGenericOpIterationDims(
     GenericOp genericOp, ArrayRef<ReassociationIndices> foldedIterationDims,
     RewriterBase &rewriter);

 struct PadOpTransformationPattern : public OpRewritePattern<tensor::PadOp> {
   using OpRewritePattern<tensor::PadOp>::OpRewritePattern;
 
   LogicalResult matchAndRewrite(tensor::PadOp padOp,
                                 PatternRewriter &rewriter) const override;
 };
 
 using OptimizeCopyFn =
-    std::function<LogicalResult(PatternRewriter &, tensor::PadOp, Value)>;
+    std::function<LogicalResult(RewriterBase &, tensor::PadOp, Value)>;
 
 /// Rewrite a tensor::PadOp into a sequence of EmptyOp, FillOp and
 /// InsertSliceOp. For now, only constant padding values are supported.
 /// `OptimizeCopyFn` can be used to customize copying step optimization.
 struct GeneralizePadOpPattern : public OpRewritePattern<tensor::PadOp> {
   GeneralizePadOpPattern(MLIRContext *context,
                          OptimizeCopyFn optimizeCopyFn = nullptr,
                          PatternBenefit benefit = 1)
       : OpRewritePattern<tensor::PadOp>(context, benefit),
         optimizeCopyFn(std::move(optimizeCopyFn)) {}
   LogicalResult matchAndRewrite(tensor::PadOp padOp,
                                 PatternRewriter &rewriter) const override;
 
 protected:
   OptimizeCopyFn optimizeCopyFn;
-  Value createFillOrGenerateOp(PatternRewriter &rewriter, tensor::PadOp padOp,
+  Value createFillOrGenerateOp(RewriterBase &rewriter, tensor::PadOp padOp,
                                Value dest,
                                const SmallVector<Value> &dynSizes) const;
 };
 
 /// Rewrites a tensor::PackOp into a sequence of tensor.pad + linalg.transpose +
 /// tensor.insert_slice ops, where the tensor::PackOp has outer dims being all
 /// 1s.
 struct GeneralizeOuterUnitDimsPackOpPattern

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

 /// index:
 /// S0(0)                        // Prologue
 /// S0(1) S1(0)                  // Prologue
 /// scf.for %I = %C0 to %N - 2 {
 ///  S0(I+2) S1(I+1) S2(I)       // Pipelined kernel
 /// }
 /// S1(N) S2(N-1)                // Epilogue
 /// S2(N)                        // Epilogue
+FailureOr<ForOp> pipelineForLoop(RewriterBase &rewriter, ForOp forOp,
+                                 const PipeliningOption &options);
+
+// TODO: such patterns should be auto-generated.
 class ForLoopPipeliningPattern : public OpRewritePattern<ForOp> {
 public:
   ForLoopPipeliningPattern(const PipeliningOption &options,
                            MLIRContext *context)
       : OpRewritePattern<ForOp>(context), options(options) {}
   LogicalResult matchAndRewrite(ForOp forOp,
                                 PatternRewriter &rewriter) const override {
     return returningMatchAndRewrite(forOp, rewriter);
   }
 
   FailureOr<ForOp> returningMatchAndRewrite(ForOp forOp,
-                                            PatternRewriter &rewriter) const;
+                                            PatternRewriter &rewriter) const {
+    return pipelineForLoop(rewriter, forOp, options);
+  }
 
 protected:
   PipeliningOption options;
 };
 
 } // namespace scf
 } // namespace mlir
 
 #endif // MLIR_DIALECT_SCF_TRANSFORMS_PATTERNS_H

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

 #include "mlir/Dialect/Tensor/Transforms/Transforms.h"
 #include "mlir/IR/PatternMatch.h"
 #include "mlir/Interfaces/TilingInterface.h"
 
 #include <deque>
 
 namespace mlir {
 class Operation;
-class PatternRewriter;
+class RewriterBase;
 class TilingInterface;
 } // namespace mlir
 
 namespace mlir {
 namespace scf {
 
 using SCFTileSizeComputationFunction =
     std::function<SmallVector<Value>(OpBuilder &, Operation *)>;

 ///   %4 = linalg.generic %2, %3 ["parallel", "parallel"]
 ///     : tensor<7x?xf32> -> tensor<7x?xf32>
 ///   %5 = tensor.insert_slice %3, %0[0, 0] : tensor<7x4xf32>
 /// }
 /// %6 = linalg.generic %1 ["parallel", "reduction"]
 ///   : tensor<7x4xf32> -> tensor<7xf32>
 /// ```
 FailureOr<scf::SCFReductionTilingResult>
-tileReductionUsingScf(PatternRewriter &b, PartialReductionOpInterface op,
+tileReductionUsingScf(RewriterBase &b, PartialReductionOpInterface op,
                       ArrayRef<OpFoldResult> tileSize);
 
 } // namespace scf
 } // namespace mlir
 
 #endif // MLIR_DIALECT_SCF_TRANSFORMS_TILEUSINGINTERFACE_H

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

struct PipeliningOption {
   /// iterations. If the epilogue is predicated; the user needs to provide a
   /// lambda to generate the predicated version of operations.
   bool peelEpilogue = true;
 
   // Lamdba to predicate operations when the prologue or epilogue are not
   // peeled. This takes the original operation, an i1 predicate value and the
   // pattern rewriter.
   using PredicateOpFn =
-      std::function<Operation *(Operation *, Value, PatternRewriter &)>;
+      std::function<Operation *(RewriterBase &, Operation *, Value)>;
   PredicateOpFn predicateFn = nullptr;
 
   // TODO: add option to decide if the prologue should be peeled.
 };
 
 /// Populate patterns for SCF software pipelining transformation. See the
 /// ForLoopPipeliningPattern for the transformation details.
 void populateSCFLoopPipeliningPatterns(RewritePatternSet &patterns,

mlir/include/mlir/Dialect/Transform/IR/TransformUtils.h

-//===- TransformUtils.h - Transform Dialect Utils ---------------*- C++ -*-===//
-//
-// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-//
-//===----------------------------------------------------------------------===//
-
-#ifndef MLIR_DIALECT_TRANSFORM_IR_TRANSFORMUTILS_H
-#define MLIR_DIALECT_TRANSFORM_IR_TRANSFORMUTILS_H
-
-#include "mlir/IR/PatternMatch.h"
-
-namespace mlir {
-namespace transform {
-
-/// A simple pattern rewriter that can be constructed from a context. This is
-/// necessary to apply patterns to a specific op locally.
-class TrivialPatternRewriter : public PatternRewriter {
-public:
-  explicit TrivialPatternRewriter(MLIRContext *context)
-      : PatternRewriter(context) {}
-};
-
-} // namespace transform
-} // namespace mlir
-
-#endif // MLIR_DIALECT_TRANSFORM_IR_TRANSFORMUTILS_H

mlir/lib/Dialect/Affine/TransformOps/AffineTransformOps.cpp

 #include "mlir/Dialect/Affine/Analysis/Utils.h"
 #include "mlir/Dialect/Affine/IR/AffineOps.h"
 #include "mlir/Dialect/Affine/IR/AffineValueMap.h"
 #include "mlir/Dialect/Affine/LoopUtils.h"
 #include "mlir/Dialect/PDL/IR/PDL.h"
 #include "mlir/Dialect/PDL/IR/PDLTypes.h"
 #include "mlir/Dialect/Transform/IR/TransformDialect.h"
 #include "mlir/Dialect/Transform/IR/TransformInterfaces.h"
-#include "mlir/Dialect/Transform/IR/TransformUtils.h"
 #include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 
 using namespace mlir;
 using namespace mlir::transform;
 
 //===----------------------------------------------------------------------===//
 // SimplifyBoundedAffineOpsOp
 //===----------------------------------------------------------------------===//

mlir/lib/Dialect/GPU/TransformOps/GPUTransformOps.cpp

 #include "mlir/Dialect/Arith/IR/Arith.h"
 #include "mlir/Dialect/GPU/IR/GPUDialect.h"
 #include "mlir/Dialect/GPU/TransformOps/GPUTransformOps.h"
 #include "mlir/Dialect/PDL/IR/PDL.h"
 #include "mlir/Dialect/SCF/IR/DeviceMappingInterface.h"
 #include "mlir/Dialect/SCF/IR/SCF.h"
 #include "mlir/Dialect/Transform/IR/TransformDialect.h"
 #include "mlir/Dialect/Transform/IR/TransformInterfaces.h"
-#include "mlir/Dialect/Transform/IR/TransformUtils.h"
 #include "mlir/IR/IRMapping.h"
 
 using namespace mlir;
 using namespace mlir::gpu;
 using namespace mlir::transform;
 
 /// Check if given mapping attributes are one of the desired attributes
 static DiagnosedSilenceableFailure
 checkAttributeType(ArrayRef<DeviceMappingAttrInterface> threadMappingAttributes,
-                   const std::optional<ArrayAttr> &foreachMapping,
+                   const std::optional<ArrayAttr> &forallMapping,
                    std::optional<TransformOpInterface> transformOp) {
-  if (!foreachMapping.has_value())
+  if (!forallMapping.has_value())
     return transformOp->emitSilenceableError() << "mapping must be present";
 
   DenseSet<Attribute> seen;
-  for (Attribute map : foreachMapping->getValue()) {
+  for (Attribute map : forallMapping->getValue()) {
     if (!llvm::is_contained(threadMappingAttributes, map)) {
       return transformOp->emitDefiniteFailure()
              << "mapping must be one of " << threadMappingAttributes;
     }
     if (llvm::is_contained(seen, map)) {
       return transformOp->emitDefiniteFailure()
              << map
              << " is duplicated, cannot map different "
launchOp = rewriter.create<LaunchOp>(loc, gridSizeX, gridSizeY, gridSizeZ,
                                        blkSizeX, blkSizeY, blkSizeZ);
   rewriter.setInsertionPointToEnd(&launchOp.getBody().front());
   rewriter.create<TerminatorOp>(loc);
   return DiagnosedSilenceableFailure::success();
 }
 
 /// Alter kernel configuration of the given kernel.
 static DiagnosedSilenceableFailure
-alterGpuLaunch(TrivialPatternRewriter &rewriter, LaunchOp gpuLaunch,
+alterGpuLaunch(IRRewriter &rewriter, LaunchOp gpuLaunch,
                TransformOpInterface transformOp,
                std::optional<int64_t> gridDimX = std::nullopt,
                std::optional<int64_t> gridDimY = std::nullopt,
                std::optional<int64_t> gridDimZ = std::nullopt,
                std::optional<int64_t> blockDimX = std::nullopt,
                std::optional<int64_t> blockDimY = std::nullopt,
                std::optional<int64_t> blockDimZ = std::nullopt) {
   DiagnosedSilenceableFailure diag =
gpuLaunch.getBlockSizeYMutable().assign(
         createConstValue(blockDimY.value()));
   if (blockDimZ.has_value())
     gpuLaunch.getBlockSizeZMutable().assign(
         createConstValue(blockDimZ.value()));
   return DiagnosedSilenceableFailure::success();
 }
 
 //===----------------------------------------------------------------------===//
-// MapForeachToBlocks
+// MapForallToBlocks
 //===----------------------------------------------------------------------===//
 
-DiagnosedSilenceableFailure mlir::transform::gpu::mapForeachToBlocksImpl(
+DiagnosedSilenceableFailure mlir::transform::gpu::mapForallToBlocksImpl(
     RewriterBase &rewriter, scf::ForallOp forallOp,
     function_ref<void(RewriterBase &, scf::ForallOp, SmallVectorImpl<Value> &)>
         blockIdGenerator,
     SmallVectorImpl<int64_t> &gridDims, TransformOpInterface transformOp,
     const ArrayRef<DeviceMappingAttrInterface> &mappingAttributes) {
   // Step 0. Target-specific verifications. There is no good place to anchor
   // those right now: the ForallOp is target-independent and the
   // transform op does not apply to individual ForallOp.

 DiagnosedSilenceableFailure
 mlir::transform::gpu::findTopLevelForallOp(Operation *target,
                                            scf::ForallOp &topLevelForallOp,
                                            TransformOpInterface transformOp) {
   auto walkResult = target->walk([&](scf::ForallOp forallOp) {
     if (forallOp->getParentOfType<scf::ForallOp>())
       return WalkResult::advance();
     if (topLevelForallOp)
-      // TODO: Handle multiple foreach if there is no dependences between them
+      // TODO: Handle multiple forall if they are independent.
       return WalkResult::interrupt();
     topLevelForallOp = forallOp;
     return WalkResult::advance();
   });
 
   if (walkResult.wasInterrupted())
     return transformOp.emitSilenceableError()
            << "could not find a unique topLevel scf.forall";
   return DiagnosedSilenceableFailure::success();
 }
 
-/// This is a helper that is only used in
-/// rewriteTopLevelForallToGpuBlocks. It generates GPU dialects
-/// block_id.
-static void generateGpuBlockIds(RewriterBase &rewriter, scf::ForallOp foreachOp,
-                                SmallVectorImpl<Value> &blockOps) {
-  Location loc = foreachOp->getLoc();
+/// This is a helper that is only used in rewriteTopLevelForallToGpuBlocks.
+/// It generates GPU dialect block_id.
+static void createGpuBlockIds(RewriterBase &rewriter, scf::ForallOp forallOp,
+  Location loc = forallOp->getLoc();
   OpBuilder::InsertionGuard guard(rewriter);
-  rewriter.setInsertionPoint(foreachOp);
+  rewriter.setInsertionPoint(forallOp);
   IndexType indexType = rewriter.getIndexType();
   blockOps = SmallVector<Value>{
       rewriter.create<BlockIdOp>(loc, indexType, Dimension::x),
       rewriter.create<BlockIdOp>(loc, indexType, Dimension::y),
       rewriter.create<BlockIdOp>(loc, indexType, Dimension::z)};
 }
 
 DiagnosedSilenceableFailure
-transform::MapForeachToBlocks::applyToOne(Operation *target,
-                                          ApplyToEachResultList &results,
-                                          transform::TransformState &state) {
+transform::MapForallToBlocks::applyToOne(Operation *target,
   LaunchOp gpuLaunch = dyn_cast<LaunchOp>(target);
-  TrivialPatternRewriter rewriter(getContext());
+  IRRewriter rewriter(getContext());
   auto transformOp = cast<TransformOpInterface>(getOperation());
 
   if (!getGenerateGpuLaunch() && !gpuLaunch) {
     DiagnosedSilenceableFailure diag =
         emitSilenceableError()
         << "Given target is not gpu.launch, set `generate_gpu_launch` "
            "attribute";
     diag.attachNote(target->getLoc()) << "when applied to this payload op";
transform::MapForallToBlocks::applyToOne(Operation *target,
   SmallVector<DeviceMappingAttrInterface> blockMappingAttributes = {
       GPUBlockMappingAttr::get(getContext(), Blocks::DimX),
       GPUBlockMappingAttr::get(getContext(), Blocks::DimY),
       GPUBlockMappingAttr::get(getContext(), Blocks::DimZ)};
 
   diag = checkAttributeType(blockMappingAttributes,
                             topLevelForallOp.getMapping(), transformOp);
   if (diag.succeeded())
-    diag = mlir::transform::gpu::mapForeachToBlocksImpl(
-        rewriter, topLevelForallOp, generateGpuBlockIds, gridDim, transformOp,
+    diag = mlir::transform::gpu::mapForallToBlocksImpl(
+        rewriter, topLevelForallOp, createGpuBlockIds, gridDim, transformOp,
         blockMappingAttributes);
   if (diag.succeeded()) {
     diag = alterGpuLaunch(rewriter, gpuLaunch,
                           cast<TransformOpInterface>(getOperation()),
                           gridDim[0], gridDim[1], gridDim[2]);
   }
 
   results.push_back(gpuLaunch);
   return diag;
 }
 
 //===----------------------------------------------------------------------===//
-// MapNestedForeachToThreads
+// MapNestedForallToThreads
 //===----------------------------------------------------------------------===//
 
 /// Searches `scf.forall` ops nested under `target` and maps each such
 /// op to GPU threads. Mapping is one-to-one and the induction variables of
 /// `scf.forall` are rewritten to gpu.thread_id according to the
 /// thread_dim_mapping attribute. Sibling `scf.forall` are supported in
 /// which case, the union of the number of threads is computed and may result
 /// in predication. Dynamic, `scf.forall` trip counts are currently
if (syncAfterDistribute)
     rewriter.create<BarrierOp>(loc);
 
   // Step 8. Erase old op.
   rewriter.eraseOp(forallOp);
 
   return DiagnosedSilenceableFailure::success();
 }
 
-DiagnosedSilenceableFailure mlir::transform::gpu::mapNestedForeachToThreadsImpl(
+DiagnosedSilenceableFailure mlir::transform::gpu::mapNestedForallToThreadsImpl(
     RewriterBase &rewriter, Operation *target,
     const SmallVectorImpl<int64_t> &blockDim,
     function_ref<void(RewriterBase &, scf::ForallOp, SmallVectorImpl<Value> &)>
         threadIdGenerator,
     bool syncAfterDistribute, std::optional<TransformOpInterface> transformOp,
     const ArrayRef<DeviceMappingAttrInterface> &threadMappingAttributes) {
   DiagnosedSilenceableFailure diag = DiagnosedSilenceableFailure::success();
   target->walk([&](scf::ForallOp forallOp) {
if (diag.succeeded()) {
                                           threadOps, syncAfterDistribute,
                                           transformOp, threadMappingAttributes);
     }
     return diag.succeeded() ? WalkResult::advance() : WalkResult::interrupt();
   });
   return diag;
 }
 
-DiagnosedSilenceableFailure transform::MapNestedForeachToThreads::applyToOne(
+DiagnosedSilenceableFailure transform::MapNestedForallToThreads::applyToOne(
     Operation *target, ApplyToEachResultList &results, TransformState &state) {
   LaunchOp gpuLaunch = dyn_cast<LaunchOp>(target);
   auto transformOp = cast<TransformOpInterface>(getOperation());
 
   if (!gpuLaunch) {
     return emitSilenceableError() << "Given target is not gpu.launch";
   }
 
   SmallVector<int64_t> blockDim = extractFromI64ArrayAttr(getBlockDim());
   blockDim.resize(/*size=*/3, /*value=*/1);
 
   DiagnosedSilenceableFailure diag =
       checkGpuLimits(transformOp, std::nullopt, std::nullopt, std::nullopt,
                      blockDim[0], blockDim[1], blockDim[2]);
   if (diag.isSilenceableFailure()) {
     diag.attachNote(getLoc()) << getBlockDimAttrName() << " is very large";
     return diag;
   }
 
   MLIRContext *ctx = getContext();
-  TrivialPatternRewriter rewriter(ctx);
+  IRRewriter rewriter(ctx);
   rewriter.setInsertionPoint(target);
 
   SmallVector<DeviceMappingAttrInterface> threadMappingAttributes = {
       GPUThreadMappingAttr::get(ctx, Threads::DimX),
       GPUThreadMappingAttr::get(ctx, Threads::DimY),
       GPUThreadMappingAttr::get(ctx, Threads::DimZ)};
   auto threadIdGenerator = [](RewriterBase &rewriter, scf::ForallOp forallOp,
                               SmallVectorImpl<Value> &threadIds) {
     IndexType indexType = rewriter.getIndexType();
     threadIds.assign({rewriter.create<ThreadIdOp>(forallOp->getLoc(), indexType,
                                                   Dimension::x),
                       rewriter.create<ThreadIdOp>(forallOp->getLoc(), indexType,
                                                   Dimension::y),
                       rewriter.create<ThreadIdOp>(forallOp->getLoc(), indexType,
                                                   Dimension::z)});
   };
-  diag = mlir::transform::gpu::mapNestedForeachToThreadsImpl(
+  diag = mlir::transform::gpu::mapNestedForallToThreadsImpl(
       rewriter, target, blockDim, threadIdGenerator, getSyncAfterDistribute(),
       transformOp, threadMappingAttributes);
 
   if (diag.succeeded()) {
     diag = alterGpuLaunch(rewriter, gpuLaunch, transformOp, std::nullopt,
                           std::nullopt, std::nullopt, blockDim[0], blockDim[1],
                           blockDim[2]);
   }

mlir/lib/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp

 #include "mlir/Dialect/PDL/IR/PDL.h"
 #include "mlir/Dialect/PDL/IR/PDLTypes.h"
 #include "mlir/Dialect/SCF/Transforms/TileUsingInterface.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/Dialect/Tensor/Utils/Utils.h"
 #include "mlir/Dialect/Transform/IR/TransformDialect.h"
 #include "mlir/Dialect/Transform/IR/TransformInterfaces.h"
 #include "mlir/Dialect/Transform/IR/TransformTypes.h"
-#include "mlir/Dialect/Transform/IR/TransformUtils.h"
 #include "mlir/Dialect/Transform/Utils/Utils.h"
 #include "mlir/Dialect/Utils/IndexingUtils.h"
 #include "mlir/IR/AffineMap.h"
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/Matchers.h"
 #include "mlir/IR/OpDefinition.h"
 #include "mlir/IR/PatternMatch.h"
 #include "mlir/IR/TypeUtilities.h"
static FailureOr<LinalgOp> tryApply(Operation *operation, Args &&...args) {
   using OpTy = typename llvm::function_traits<
       decltype(&PatternTy::returningMatchAndRewrite)>::template arg_t<0>;
   auto op = dyn_cast<OpTy>(operation);
   if (!op)
     return failure();
 
   // Apply the pattern directly to the op.
   PatternTy pattern(operation->getContext(), std::forward<Args>(args)...);
+  // We want to discourage direct use of PatternRewriter in APIs but In this
+  // very specific case, an IRRewriter is not enough.
+  struct TrivialPatternRewriter : public PatternRewriter {
+  public:
+    explicit TrivialPatternRewriter(MLIRContext *context)
+        : PatternRewriter(context) {}
+  };
   TrivialPatternRewriter rewriter(operation->getContext());
   rewriter.setInsertionPoint(operation);
   auto result = pattern.returningMatchAndRewrite(op, rewriter);
   if (failed(result))
     return failure();
   return cast<LinalgOp>(result->getOperation());
 }
 
static LogicalResult applyTilingToAll(
   for (unsigned int i = 0; i < numLoops; ++i)
     loopOps[i].reserve(payloadOps.size());
 
   for (Operation *target : payloadOps) {
     auto tilingInterfaceOp = dyn_cast<TilingInterface>(target);
     if (!tilingInterfaceOp)
       return transformOp->emitError("only TilingInterface ops are supported");
 
-    TrivialPatternRewriter rewriter(target->getContext());
+    IRRewriter rewriter(target->getContext());
     rewriter.setInsertionPoint(target);
     FailureOr<scf::SCFTileAndFuseResult> tiledResults =
         applyFn(tilingInterfaceOp);
     if (failed(tiledResults))
       return failure();
 
     // Perform the replacement of tiled and fused values.
     SmallVector<Operation *> opsToReplace{target};
transform::FuseOp::apply(mlir::transform::TransformResults &transformResults,
   tilingOptions = tilingOptions.setTileSizes(tileSizes);
   scf::SCFTileAndFuseOptions tileAndFuseOptions;
   tileAndFuseOptions.tilingOptions = tilingOptions;
   LogicalResult result = applyTilingToAll(
       getOperation(), state.getPayloadOps(getTarget()),
       tileSizes.size() - llvm::count(tileSizes, 0), transformResults,
       [&](TilingInterface tilingInterfaceOp)
           -> FailureOr<scf::SCFTileAndFuseResult> {
-        TrivialPatternRewriter rewriter(getContext());
+        IRRewriter rewriter(getContext());
         return tileConsumerAndFuseProducerGreedilyUsingSCFForOp(
             rewriter, tilingInterfaceOp, tileAndFuseOptions);
       });
   return failed(result) ? DiagnosedSilenceableFailure::definiteFailure()
                         : DiagnosedSilenceableFailure::success();
 }
 
 ParseResult transform::FuseOp::parse(OpAsmParser &parser,

 transform::GeneralizeOp::applyToOne(LinalgOp target,
                                     transform::ApplyToEachResultList &results,
                                     transform::TransformState &state) {
   // Exit early if no transformation is needed.
   if (isa<GenericOp>(target)) {
     results.push_back(target);
     return DiagnosedSilenceableFailure::success();
   }
-  FailureOr<LinalgOp> generic = tryApply<LinalgGeneralizationPattern>(target);
+  IRRewriter rewriter(getContext());
+  rewriter.setInsertionPoint(target);
+  FailureOr<LinalgOp> generic = generalizeNamedOp(rewriter, target);
   if (succeeded(generic)) {
     results.push_back(generic->getOperation());
     return DiagnosedSilenceableFailure::success();
   }
   return emitDefaultSilenceableFailure(target);
 }
 
 //===----------------------------------------------------------------------===//
 // InterchangeOp
 //===----------------------------------------------------------------------===//
 
 DiagnosedSilenceableFailure
 transform::InterchangeOp::applyToOne(GenericOp target,
                                      transform::ApplyToEachResultList &results,
                                      transform::TransformState &state) {
   ArrayRef<int64_t> interchangeVector = getIteratorInterchange();
   // Exit early if no transformation is needed.
   if (interchangeVector.empty()) {
     results.push_back(target);
     return DiagnosedSilenceableFailure::success();
   }
-  TrivialPatternRewriter rewriter(target->getContext());
+  IRRewriter rewriter(target->getContext());
   FailureOr<GenericOp> res =
       interchangeGenericOp(rewriter, target,
                            SmallVector<unsigned>(interchangeVector.begin(),
                                                  interchangeVector.end()));
   if (failed(res))
     return DiagnosedSilenceableFailure::definiteFailure();
   results.push_back(res->getOperation());
   return DiagnosedSilenceableFailure::success();
if (getMapping().has_value()) {
     } else {
       return emitDefaultDefiniteFailure(target);
     }
   }
 
   if (failed(promoteSubviewsPrecondition(target, promotionOptions)))
     return emitDefaultDefiniteFailure(target);
 
-  TrivialPatternRewriter rewriter(target->getContext());
+  IRRewriter rewriter(target->getContext());
   rewriter.setInsertionPoint(target);
   FailureOr<LinalgOp> res = promoteSubViews(rewriter, target, promotionOptions);
   if (failed(res))
     return emitDefaultDefiniteFailure(target);
   results.push_back(target);
   return DiagnosedSilenceableFailure::success();
 }
 
tilingOptions.setTileSizeComputationFunction([&](OpBuilder &b, Operation *) {
     for (OpFoldResult shapeSize : shapeSizes) {
       tileSizes.push_back(getConstantIntValue(shapeSize)
                               ? b.create<arith::ConstantIndexOp>(loc, 0)
                               : b.create<arith::ConstantIndexOp>(loc, 1));
     }
     return tileSizes;
   });
   SmallVector<int64_t> emptyTileSizes;
-  TrivialPatternRewriter rewriter(getContext());
+  IRRewriter rewriter(getContext());
   rewriter.setInsertionPoint(target);
   FailureOr<scf::SCFTilingResult> maybeTilingResult = tileUsingSCFForOp(
       rewriter, cast<TilingInterface>(target.getOperation()), tilingOptions);
   if (failed(maybeTilingResult))
     return emitDefaultDefiniteFailure(target);
 
   if (target->getNumResults())
     rewriter.replaceOp(target, maybeTilingResult->replacements);

 //===----------------------------------------------------------------------===//
 // SplitOp
 //===----------------------------------------------------------------------===//
 
 DiagnosedSilenceableFailure SplitOp::apply(TransformResults &results,
                                            TransformState &state) {
   // Collect the dynamic split points if provided.
   ArrayRef<Operation *> payload = state.getPayloadOps(getTarget());
-  TrivialPatternRewriter rewriter(getContext());
+  IRRewriter rewriter(getContext());
   SmallVector<OpFoldResult> splitPoints;
   splitPoints.reserve(payload.size());
   if (getDynamicSplitPoint()) {
     auto diag = DiagnosedSilenceableFailure::success();
     if (getDynamicSplitPoint().getType().isa<TransformHandleTypeInterface>()) {
       splitPoints = llvm::to_vector(llvm::map_range(
           state.getPayloadOps(getDynamicSplitPoint()), [&](Operation *op) {
             if (op->getNumResults() != 1 ||

 DiagnosedSilenceableFailure transform::SplitReductionOp::applyToOne(
     LinalgOp target, transform::ApplyToEachResultList &results,
     transform::TransformState &state) {
   ControlSplitReductionFn splitFn = [&](LinalgOp) {
     return linalg::SplitReductionOptions{int64_t(getSplitFactor()),
                                          unsigned(getInsertSplitDimension()),
                                          bool(getInnerParallel())};
   };
-  TrivialPatternRewriter rewriter(getContext());
+  IRRewriter rewriter(getContext());
   rewriter.setInsertionPoint(target);
   FailureOr<SplitReductionResult> splitResult =
       (getUseScalingAlgorithm())
           ? splitReductionByScaling(rewriter, target, splitFn, getUseAlloc())
           : splitReduction(rewriter, target, splitFn, getUseAlloc());
   if (failed(splitResult))
     return emitDefaultDefiniteFailure(target);
 
build(builder, result,
         /*resultTypes=*/TypeRange{opTy, opTy, opTy, opTy},
         /*target=*/target,
         /*tile_sizes=*/staticTileSizesAttr);
 }
 
 DiagnosedSilenceableFailure transform::TileReductionUsingScfOp::applyToOne(
     LinalgOp target, transform::ApplyToEachResultList &results,
     transform::TransformState &state) {
-  TrivialPatternRewriter rewriter(getContext());
+  IRRewriter rewriter(getContext());
   rewriter.setInsertionPoint(target);
   FailureOr<scf::SCFReductionTilingResult> result = scf::tileReductionUsingScf(
       rewriter, cast<PartialReductionOpInterface>(target.getOperation()),
       getAsOpFoldResult(rewriter.getI64ArrayAttr(getTileSizes())));
 
   if (failed(result))
     return emitDefaultSilenceableFailure(target);
   results.push_back(result->loops.front());
build(builder, result,
         /*num_threads=*/staticNumThreadsAttr,
         /*tile_sizes=*/staticTileSizesAttr,
         /*mapping=*/mapping);
 }
 
 DiagnosedSilenceableFailure transform::TileReductionUsingForallOp::applyToOne(
     LinalgOp target, transform::ApplyToEachResultList &results,
     transform::TransformState &state) {
-  TrivialPatternRewriter rewriter(getContext());
+  IRRewriter rewriter(getContext());
   rewriter.setInsertionPoint(target);
   SmallVector<OpFoldResult> numThreads =
       getAsOpFoldResult(rewriter.getI64ArrayAttr(getNumThreads()));
   SmallVector<OpFoldResult> tileSizes =
       getAsOpFoldResult(rewriter.getI64ArrayAttr(getTileSizes()));
   FailureOr<linalg::ForallReductionTilingResult> result =
       linalg::tileReductionUsingForall(
           rewriter, cast<PartialReductionOpInterface>(target.getOperation()),
if (!tileSizes.empty()) {
             sizes.push_back(dynamicSizes[index]->getResult(0));
           }
         }
         return sizes;
       });
     }
 
     tilingOptions.setInterchange(getInterchange());
-    TrivialPatternRewriter rewriter(op->getContext());
+    IRRewriter rewriter(op->getContext());
     FailureOr<scf::SCFTilingResult> maybeTilingResult =
         tileUsingSCFForOp(rewriter, tilingInterface, tilingOptions);
     if (failed(maybeTilingResult))
       return DiagnosedSilenceableFailure::definiteFailure();
 
     if (dpsInterface.hasBufferSemantics())
       rewriter.eraseOp(op);
     else
if (!tileSizes.empty()) {
                     dynamicSizeProducers[dynamicIdx++][index]->getResult(0));
               }
             }
             return sizes;
           });
     }
 
     tilingOptions.setInterchange(getInterchange());
-    TrivialPatternRewriter rewriter(tilingInterfaceOp.getContext());
+    IRRewriter rewriter(tilingInterfaceOp.getContext());
     FailureOr<scf::SCFTilingResult> tilingResult =
         tileUsingSCFForOp(rewriter, tilingInterfaceOp, tilingOptions);
     if (failed(tilingResult))
       return DiagnosedSilenceableFailure::definiteFailure();
 
     rewriter.replaceOp(tilingInterfaceOp, tilingResult->replacements);
 
     tiled.append(tilingResult->tiledOps);

mlir/lib/Dialect/Linalg/Transforms/Loops.cpp

for (OpOperand *outputOperand : linalgOp.getDpsInitOperands()) {
     outputBuffers.push_back(outputOperand->get());
   }
   inlineRegionAndEmitStore<LoadOpTy, StoreOpTy>(b, loc, linalgOp, indexedValues,
                                                 indexing, outputBuffers);
 }
 
 /// Replace the index operations in the body of the loop nest by the matching
 /// induction variables.
-static void replaceIndexOpsByInductionVariables(LinalgOp linalgOp,
-                                                PatternRewriter &rewriter,
+static void replaceIndexOpsByInductionVariables(RewriterBase &rewriter,
+                                                LinalgOp linalgOp,
                                                 ArrayRef<Operation *> loopOps) {
   // Extract the induction variables of the loop nest from outer to inner.
   SmallVector<Value> allIvs;
   for (Operation *loopOp : loopOps) {
     llvm::TypeSwitch<Operation *>(loopOp)
         .Case([&](scf::ParallelOp parallelOp) {
           allIvs.append(parallelOp.getInductionVars().begin(),
                         parallelOp.getInductionVars().end());
if (!loopOps.empty()) {
     LoopLikeOpInterface loopOp = loopOps.back();
     for (IndexOp indexOp :
          llvm::make_early_inc_range(loopOp.getLoopBody().getOps<IndexOp>()))
       rewriter.replaceOp(indexOp, allIvs[indexOp.getDim()]);
   }
 }
 
 template <typename LoopTy>
-static FailureOr<LinalgLoops> linalgOpToLoopsImpl(PatternRewriter &rewriter,
+static FailureOr<LinalgLoops> linalgOpToLoopsImpl(RewriterBase &rewriter,
                                                   LinalgOp linalgOp) {
   using LoadOpTy = std::conditional_t<std::is_same<LoopTy, AffineForOp>::value,
                                       AffineLoadOp, memref::LoadOp>;
   using StoreOpTy = std::conditional_t<std::is_same<LoopTy, AffineForOp>::value,
                                        AffineStoreOp, memref::StoreOp>;
 
   // The flattened loopToOperandRangesMaps is expected to be an invertible
   // permutation map (which is asserted in the inverse calculation).
for (Value iv : allIvs) {
     // loop operation.
     BlockArgument ivVal = iv.dyn_cast<BlockArgument>();
     if (!ivVal)
       return failure();
     loopSet.insert(ivVal.getOwner()->getParentOp());
   }
   LinalgLoops loops(loopSet.begin(), loopSet.end());
   // Replace all index operations in the loop body.
-  replaceIndexOpsByInductionVariables(linalgOp, rewriter, loops);
+  replaceIndexOpsByInductionVariables(rewriter, linalgOp, loops);
   return loops;
 }
 
 namespace {
 template <typename LoopType>
 class LinalgRewritePattern : public RewritePattern {
 public:
   LinalgRewritePattern(MLIRContext *context)

 
 std::unique_ptr<OperationPass<func::FuncOp>>
 mlir::createConvertLinalgToAffineLoopsPass() {
   return std::make_unique<LowerToAffineLoops>();
 }
 
 /// Emits a loop nest of `affine.for` with the proper body for `linalgOp`.
 FailureOr<LinalgLoops>
-mlir::linalg::linalgOpToAffineLoops(PatternRewriter &rewriter,
-                                    LinalgOp linalgOp) {
+mlir::linalg::linalgOpToAffineLoops(RewriterBase &rewriter, LinalgOp linalgOp) {
   return linalgOpToLoopsImpl<AffineForOp>(rewriter, linalgOp);
 }
 
 /// Emits a loop nest of `scf.for` with the proper body for `linalgOp`.
-FailureOr<LinalgLoops> mlir::linalg::linalgOpToLoops(PatternRewriter &rewriter,
+FailureOr<LinalgLoops> mlir::linalg::linalgOpToLoops(RewriterBase &rewriter,
                                                      LinalgOp linalgOp) {
   return linalgOpToLoopsImpl<scf::ForOp>(rewriter, linalgOp);
 }
 
 /// Emits a loop nest of `scf.parallel` with the proper body for `linalgOp`.
 FailureOr<LinalgLoops>
-mlir::linalg::linalgOpToParallelLoops(PatternRewriter &rewriter,
+mlir::linalg::linalgOpToParallelLoops(RewriterBase &rewriter,
                                       LinalgOp linalgOp) {
   return linalgOpToLoopsImpl<scf::ParallelOp>(rewriter, linalgOp);
 }

mlir/lib/Dialect/Linalg/Transforms/SplitReduction.cpp

 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/Dialect/Tensor/Utils/Utils.h"
 #include "mlir/IR/PatternMatch.h"
 
 using namespace mlir;
 using namespace mlir::linalg;
 
 FailureOr<SplitReductionResult> mlir::linalg::splitReduction(
-    PatternRewriter &b, LinalgOp op,
+    RewriterBase &b, LinalgOp op,
     const ControlSplitReductionFn &controlSplitReductionFn, bool useAlloc) {
   OpBuilder::InsertionGuard guard(b);
   b.setInsertionPoint(op);
 
   SplitReductionOptions control = controlSplitReductionFn(op);
   int64_t ratio = control.ratio;
   unsigned insertSplitIndex = control.index;
   unsigned insertSplitDimension = control.index;
static AffineMap insertParallelDim(LinalgOp op, OpOperand &opOperand,
   AffineMap idMap =
       AffineMap::getMultiDimIdentityMap(map.getNumDims(), op.getContext());
   AffineMap shiftedIdMap = idMap.shiftDims(1, /*offset=*/reductionDimPos + 1);
   return map.compose(shiftedIdMap).insertResult(reductionDim, reductionDimPos);
 }
 
 /// Core rewrite implementation.
 FailureOr<SplitReductionResult> mlir::linalg::splitReductionByScaling(
-    PatternRewriter &b, LinalgOp op,
+    RewriterBase &b, LinalgOp op,
     const ControlSplitReductionFn &controlSplitReductionFn, bool useAlloc) {
   OpBuilder::InsertionGuard guard(b);
   b.setInsertionPoint(op);
 
   // Matcher part, enforce preconditions.
   SplitReductionOptions control = controlSplitReductionFn(op);
   if (control.innerParallel)
     return b.notifyMatchFailure(op, "innerParallel not supported");

mlir/lib/Dialect/Linalg/Transforms/Transforms.cpp

rewriter.replaceOpWithNewOp<linalg::GenericOp>(
       });
 
   return success();
 }
 
 /// Filling `dest` using FillOp constant padding value if possible.
 /// Otherwise, generate a tensor::GenerateOp.
 Value GeneralizePadOpPattern::createFillOrGenerateOp(
-    PatternRewriter &rewriter, tensor::PadOp padOp, Value dest,
+    RewriterBase &rewriter, tensor::PadOp padOp, Value dest,
     const SmallVector<Value> &dynSizes) const {
   auto padValue = padOp.getConstantPaddingValue();
   if (padValue)
     return rewriter.create<FillOp>(padOp.getLoc(), padValue, dest).result();
 
   // Fill could not be optimized: Lower to tensor::GenerateOp with region.
   auto generateOp = rewriter.create<tensor::GenerateOp>(
       padOp.getLoc(), padOp.getResultType(), dynSizes);

mlir/lib/Dialect/Linalg/Transforms/Vectorization.cpp

 #include "mlir/IR/PatternMatch.h"
 #include "mlir/Support/LLVM.h"
 #include "mlir/Transforms/RegionUtils.h"
 #include "llvm/ADT/Sequence.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/TypeSwitch.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
-#include <type_traits>
 #include <optional>
+#include <type_traits>
 
 using namespace mlir;
 using namespace mlir::linalg;
 
 #define DEBUG_TYPE "linalg-vectorization"
 
 #define DBGS() (llvm::dbgs() << '[' << DEBUG_TYPE << "] ")
 #define LDBG(X) LLVM_DEBUG(DBGS() << X << "\n")

 /// TransferWriteOps may be generated instead of InsertSliceOps.
 struct GenericPadOpVectorizationPattern : public GeneralizePadOpPattern {
   GenericPadOpVectorizationPattern(MLIRContext *context,
                                    PatternBenefit benefit = 1)
       : GeneralizePadOpPattern(context, tryVectorizeCopy, benefit) {}
   /// Vectorize the copying of a tensor::PadOp's source. This is possible if
   /// each dimension size is statically know in the source type or the result
   /// type (or both).
-  static LogicalResult tryVectorizeCopy(PatternRewriter &rewriter,
+  static LogicalResult tryVectorizeCopy(RewriterBase &rewriter,
                                         tensor::PadOp padOp, Value dest) {
     auto sourceType = padOp.getSourceType();
     auto resultType = padOp.getResultType();
     if (!VectorType::isValidElementType(sourceType.getElementType()))
       return failure();
 
     // Copy cannot be vectorized if pad value is non-constant and source shape
     // is dynamic. In case of a dynamic source shape, padding must be appended
FailureOr<Operation *> depthwiseConv() {
 
     auto linearIndex = [&](int64_t kw, int64_t w) {
       return kw * (wSize / wSizeStep) + w;
     };
 
     // Compute contraction: O{n, w, c} += I{n, sw * w + dw * kw, c} * F{c}
     for (int64_t kw = 0; kw < kwSize; ++kw) {
       for (int64_t w = 0; w < wSize; w += wSizeStep) {
-        resVals[w] = depthwiseConv1dSliceAsMulAcc(
-            rewriter, loc, lhsVals[linearIndex(kw, w)], rhsVals[kw], resVals[w]);
+        resVals[w] = depthwiseConv1dSliceAsMulAcc(rewriter, loc,
+                                                  lhsVals[linearIndex(kw, w)],
+                                                  rhsVals[kw], resVals[w]);
       }
     }
 
     // Its possible we failed to create the Fma.
     if (!llvm::all_of(resVals, [](Value v) { return v; })) {
       // Manually revert (in reverse order) to avoid leaving a bad IR state.
-      for (auto &collection : {resVals, rhsVals, lhsVals, {res, rhs, lhs, zero}})
+      for (auto &collection :
+           {resVals, rhsVals, lhsVals, {res, rhs, lhs, zero}})
         for (Value v : collection)
           rewriter.eraseOp(v.getDefiningOp());
       return rewriter.notifyMatchFailure(op, "failed to create FMA");
     }
 
     // Write back res slice: {n, wSizeStep, c} @ [0, w, 0].
     // This does not depend on kw.
     for (int64_t w = 0; w < wSize; w += wSizeStep) {

mlir/lib/Dialect/SCF/TransformOps/SCFTransformOps.cpp

 #include "mlir/Dialect/Affine/LoopUtils.h"
 #include "mlir/Dialect/Func/IR/FuncOps.h"
 #include "mlir/Dialect/SCF/IR/SCF.h"
 #include "mlir/Dialect/SCF/Transforms/Patterns.h"
 #include "mlir/Dialect/SCF/Transforms/Transforms.h"
 #include "mlir/Dialect/SCF/Utils/Utils.h"
 #include "mlir/Dialect/Transform/IR/TransformDialect.h"
 #include "mlir/Dialect/Transform/IR/TransformInterfaces.h"
-#include "mlir/Dialect/Transform/IR/TransformUtils.h"
 #include "mlir/Dialect/Vector/IR/VectorOps.h"
 
 using namespace mlir;
 
 //===----------------------------------------------------------------------===//
 // GetParentForOp
 //===----------------------------------------------------------------------===//
 DiagnosedSilenceableFailure

 DiagnosedSilenceableFailure
 transform::LoopOutlineOp::apply(transform::TransformResults &results,
                                 transform::TransformState &state) {
   SmallVector<Operation *> transformed;
   DenseMap<Operation *, SymbolTable> symbolTables;
   for (Operation *target : state.getPayloadOps(getTarget())) {
     Location location = target->getLoc();
     Operation *symbolTableOp = SymbolTable::getNearestSymbolTable(target);
-    TrivialPatternRewriter rewriter(getContext());
+    IRRewriter rewriter(getContext());
     scf::ExecuteRegionOp exec = wrapInExecuteRegion(rewriter, target);
     if (!exec) {
       DiagnosedSilenceableFailure diag = emitSilenceableError()
                                          << "failed to outline";
       diag.attachNote(target->getLoc()) << "target op";
       return diag;
     }
     func::CallOp call;
transform::LoopPipelineOp::applyToOne(scf::ForOp target,
   scf::PipeliningOption options;
   options.getScheduleFn =
       [this](scf::ForOp forOp,
              std::vector<std::pair<Operation *, unsigned>> &schedule) mutable {
         loopScheduling(forOp, schedule, getIterationInterval(),
                        getReadLatency());
       };
   scf::ForLoopPipeliningPattern pattern(options, target->getContext());
-  TrivialPatternRewriter rewriter(getContext());
+  IRRewriter rewriter(getContext());
   rewriter.setInsertionPoint(target);
   FailureOr<scf::ForOp> patternResult =
-      pattern.returningMatchAndRewrite(target, rewriter);
+      scf::pipelineForLoop(rewriter, target, options);
   if (succeeded(patternResult)) {
     results.push_back(*patternResult);
     return DiagnosedSilenceableFailure::success();
   }
   return emitDefaultSilenceableFailure(target);
 }
 
 //===----------------------------------------------------------------------===//

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

protected:
   void setValueMapping(Value key, Value el, int64_t idx);
 
 public:
   /// Initalize the information for the given `op`, return true if it
   /// satisfies the pre-condition to apply pipelining.
   bool initializeLoopInfo(ForOp op, const PipeliningOption &options);
   /// Emits the prologue, this creates `maxStage - 1` part which will contain
   /// operations from stages [0; i], where i is the part index.
-  void emitPrologue(PatternRewriter &rewriter);
+  void emitPrologue(RewriterBase &rewriter);
   /// Gather liverange information for Values that are used in a different stage
   /// than its definition.
   llvm::MapVector<Value, LiverangeInfo> analyzeCrossStageValues();
   scf::ForOp createKernelLoop(
       const llvm::MapVector<Value, LiverangeInfo> &crossStageValues,
-      PatternRewriter &rewriter,
+      RewriterBase &rewriter,
       llvm::DenseMap<std::pair<Value, unsigned>, unsigned> &loopArgMap);
   /// Emits the pipelined kernel. This clones loop operations following user
   /// order and remaps operands defined in a different stage as their use.
   void createKernel(
       scf::ForOp newForOp,
       const llvm::MapVector<Value, LiverangeInfo> &crossStageValues,
       const llvm::DenseMap<std::pair<Value, unsigned>, unsigned> &loopArgMap,
-      PatternRewriter &rewriter);
+      RewriterBase &rewriter);
   /// Emits the epilogue, this creates `maxStage - 1` part which will contain
   /// operations from stages [i; maxStage], where i is the part index.
-  llvm::SmallVector<Value> emitEpilogue(PatternRewriter &rewriter);
+  llvm::SmallVector<Value> emitEpilogue(RewriterBase &rewriter);
 };
 
 bool LoopPipelinerInternal::initializeLoopInfo(
     ForOp op, const PipeliningOption &options) {
   forOp = op;
   auto upperBoundCst =
       forOp.getUpperBound().getDefiningOp<arith::ConstantIndexOp>();
   auto lowerBoundCst =
for (OpOperand &operand : nested->getOpOperands()) {
       if ((def && !clone->isAncestor(def)) ||
           operand.get().isa<BlockArgument>())
         callback(&operand);
     }
   });
   return clone;
 }
 
-void LoopPipelinerInternal::emitPrologue(PatternRewriter &rewriter) {
+void LoopPipelinerInternal::emitPrologue(RewriterBase &rewriter) {
   // Initialize the iteration argument to the loop initiale values.
   for (BlockArgument &arg : forOp.getRegionIterArgs()) {
     OpOperand &operand = forOp.getOpOperandForRegionIterArg(arg);
     setValueMapping(arg, operand.get(), 0);
   }
   auto yield = cast<scf::YieldOp>(forOp.getBody()->getTerminator());
   for (int64_t i = 0; i < maxStage; i++) {
     // special handling for induction variable as the increment is implicit.
for (Operation *op : opOrder) {
     });
   }
   return crossStageValues;
 }
 
 scf::ForOp LoopPipelinerInternal::createKernelLoop(
     const llvm::MapVector<Value, LoopPipelinerInternal::LiverangeInfo>
         &crossStageValues,
-    PatternRewriter &rewriter,
+    RewriterBase &rewriter,
     llvm::DenseMap<std::pair<Value, unsigned>, unsigned> &loopArgMap) {
   // Creates the list of initial values associated to values used across
   // stages. The initial values come from the prologue created above.
   // Keep track of the kernel argument associated to each version of the
   // values passed to the kernel.
   llvm::SmallVector<Value> newLoopArg;
   // For existing loop argument initialize them with the right version from the
   // prologue.
scf::ForOp LoopPipelinerInternal::createKernelLoop(
   return newForOp;
 }
 
 void LoopPipelinerInternal::createKernel(
     scf::ForOp newForOp,
     const llvm::MapVector<Value, LoopPipelinerInternal::LiverangeInfo>
         &crossStageValues,
     const llvm::DenseMap<std::pair<Value, unsigned>, unsigned> &loopArgMap,
-    PatternRewriter &rewriter) {
+    RewriterBase &rewriter) {
   valueMapping.clear();
 
   // Create the kernel, we clone instruction based on the order given by
   // user and remap operands coming from a previous stages.
   rewriter.setInsertionPoint(newForOp.getBody(), newForOp.getBody()->begin());
   IRMapping mapping;
   mapping.map(forOp.getInductionVar(), newForOp.getInductionVar());
   for (const auto &arg : llvm::enumerate(forOp.getRegionIterArgs())) {
for (OpOperand *operand : operands) {
       auto remap = loopArgMap.find(
           std::make_pair(operand->get(), useStage - stageDef->second));
       assert(remap != loopArgMap.end());
       nestedNewOp->setOperand(operand->getOperandNumber(),
                               newForOp.getRegionIterArgs()[remap->second]);
     }
 
     if (predicates[useStage]) {
-      newOp = predicateFn(newOp, predicates[useStage], rewriter);
+      newOp = predicateFn(rewriter, newOp, predicates[useStage]);
       // Remap the results to the new predicated one.
       for (auto values : llvm::zip(op->getResults(), newOp->getResults()))
         mapping.map(std::get<0>(values), std::get<1>(values));
     }
     rewriter.setInsertionPointAfter(newOp);
     if (annotateFn)
       annotateFn(newOp, PipeliningOption::PipelinerPart::Kernel, 0);
   }
for (const auto &retVal :
     setValueMapping(forOp.getRegionIterArgs()[retVal.index()],
                     newForOp->getResult(retVal.index()),
                     maxStage - defStage + 1);
   }
   rewriter.create<scf::YieldOp>(forOp.getLoc(), yieldOperands);
 }
 
 llvm::SmallVector<Value>
-LoopPipelinerInternal::emitEpilogue(PatternRewriter &rewriter) {
+LoopPipelinerInternal::emitEpilogue(RewriterBase &rewriter) {
   llvm::SmallVector<Value> returnValues(forOp->getNumResults());
   // Emit different versions of the induction variable. They will be
   // removed by dead code if not used.
   for (int64_t i = 0; i < maxStage; i++) {
     Value newlastIter = rewriter.create<arith::ConstantIndexOp>(
         forOp.getLoc(), lb + step * ((((ub - 1) - lb) / step) - i));
     setValueMapping(forOp.getInductionVar(), newlastIter, maxStage - i);
   }
it =
         valueMapping
             .insert(std::make_pair(key, llvm::SmallVector<Value>(maxStage + 1)))
             .first;
   it->second[idx] = el;
 }
 
 } // namespace
 
-FailureOr<ForOp> ForLoopPipeliningPattern::returningMatchAndRewrite(
-    ForOp forOp, PatternRewriter &rewriter) const {
-
+FailureOr<ForOp> mlir::scf::pipelineForLoop(RewriterBase &rewriter, ForOp forOp,
+                                            const PipeliningOption &options) {
   LoopPipelinerInternal pipeliner;
   if (!pipeliner.initializeLoopInfo(forOp, options))
     return failure();
 
   // 1. Emit prologue.
   pipeliner.emitPrologue(rewriter);
 
   // 2. Track values used across stages. When a value cross stages it will

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

if (!tilingResult.loops.empty()) {
       tilingResult.loops.front().dump();
       llvm::dbgs() << "\n";
     }
   });
   return tilingResult;
 }
 
 FailureOr<scf::SCFReductionTilingResult>
-mlir::scf::tileReductionUsingScf(PatternRewriter &b,
+mlir::scf::tileReductionUsingScf(RewriterBase &b,
                                  PartialReductionOpInterface op,
                                  ArrayRef<OpFoldResult> tileSize) {
   Location loc = op.getLoc();
   // Ops implementing PartialReductionOpInterface are expected to implement
   // TilingInterface.
   auto tilingInterfaceOp = cast<TilingInterface>(op.getOperation());
   SmallVector<Range> iterationDomain = tilingInterfaceOp.getIterationDomain(b);
   SmallVector<Value> tileSizeVector =

mlir/lib/Dialect/Transform/IR/TransformOps.cpp

 //===- TransformDialect.cpp - Transform dialect operations ----------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #include "mlir/Dialect/Transform/IR/TransformOps.h"
 #include "mlir/Dialect/PDL/IR/PDLOps.h"
 #include "mlir/Dialect/Transform/IR/TransformDialect.h"
 #include "mlir/Dialect/Transform/IR/TransformInterfaces.h"
 #include "mlir/Dialect/Transform/IR/TransformTypes.h"
-#include "mlir/Dialect/Transform/IR/TransformUtils.h"
 #include "mlir/IR/PatternMatch.h"
 #include "mlir/Interfaces/ControlFlowInterfaces.h"
 #include "mlir/Rewrite/FrozenRewritePatternSet.h"
 #include "mlir/Rewrite/PatternApplicator.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/ScopeExit.h"
 #include "llvm/Support/Debug.h"
 #include <optional>
patternModule.registerRewriteFunction(
         "transform.dialect", [](PatternRewriter &, Operation *) {});
 
     it = compiledPatterns
              .try_emplace(patternOp.getName(), std::move(patternModule))
              .first;
   }
 
   PatternApplicator applicator(it->second);
-  transform::TrivialPatternRewriter rewriter(root->getContext());
+  // We want to discourage direct use of PatternRewriter in APIs but In this
+  // very specific case, an IRRewriter is not enough.
+  struct TrivialPatternRewriter : public PatternRewriter {
+  public:
+    explicit TrivialPatternRewriter(MLIRContext *context)
+        : PatternRewriter(context) {}
+  };
+  TrivialPatternRewriter rewriter(root->getContext());
   applicator.applyDefaultCostModel();
   root->walk([&](Operation *op) {
     if (succeeded(applicator.matchAndRewrite(op, rewriter)))
       results.push_back(op);
   });
 
   return success();
 }

 }
 
 //===----------------------------------------------------------------------===//
 // GetDefiningOp
 //===----------------------------------------------------------------------===//
 
 DiagnosedSilenceableFailure
 transform::GetDefiningOp::apply(transform::TransformResults &results,
                                 transform::TransformState &state) {
   SmallVector<Operation *> definingOps;
   for (Value v : state.getPayloadValues(getTarget())) {
     if (v.isa<BlockArgument>()) {
       DiagnosedSilenceableFailure diag =
           emitSilenceableError() << "cannot get defining op of block argument";
       diag.attachNote(v.getLoc()) << "target value";
       return diag;
     }

mlir/test/Dialect/GPU/transform-gpu-failing.mlir

 // RUN: mlir-opt --test-transform-dialect-interpreter --split-input-file  -canonicalize -cse --verify-diagnostics %s
 
-func.func @map_nested_foreach_to_threads_not_gpu_launch() -> () {
+func.func @map_nested_forall_to_threads_not_gpu_launch() -> () {
   %1 = tensor.empty() : tensor<4xf32>
   return
 }
 transform.sequence failures(propagate) {
 ^bb0(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["tensor.empty"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{Given target is not gpu.launch}}
-  %1 = transform.gpu.map_nested_foreach_to_threads %funcop
+  %1 = transform.gpu.map_nested_forall_to_threads %funcop
 }
 
 // -----
 
-func.func @map_nested_foreach_to_threads_excessive_threads(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
+func.func @map_nested_forall_to_threads_excessive_threads(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
   %one = arith.constant 1 : index
   %c900 = arith.constant 900 : index
   %c9 = arith.constant 9 : index
   %c7 = arith.constant 7 : index
   %name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
             threads(%arg6, %arg7, %arg8) in (%arg12 = %one, %arg13 = %one, %arg14 = %one)
   {
     scf.forall (%i, %j) in (%c7, %c900) {
func.func @map_nested_forall_to_threads_excessive_threads(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
 
   return %y : memref<2 x 32 x f32>
 }
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{Trying to launch a GPU kernel with gridDim = (1, 1, 1) blockDim = (1200, 9, 1). It is larger than the limits.}}
   // expected-note @below {{"blockDim" is very large}}
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [1200, 9, 1] }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [1200, 9, 1] }
 }
 
 // -----
 
-func.func @map_nested_foreach_to_threads_fewer_threads(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
+func.func @map_nested_forall_to_threads_fewer_threads(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
   %one = arith.constant 1 : index
   %c900 = arith.constant 900 : index
   %c9 = arith.constant 9 : index
   %c7 = arith.constant 7 : index
   %name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
             threads(%arg6, %arg7, %arg8) in (%arg12 = %one, %arg13 = %one, %arg14 = %one)
   {
     scf.forall (%i, %j) in (%c7, %c900) {
func.func @map_nested_forall_to_threads_fewer_threads(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
 
   return %y : memref<2 x 32 x f32>
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{The requested GPU threads are fewer than the number of loop trip counts. Try to tile scf.forall before mapping or set small blockDim.}}
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [128, 4, 1] }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [128, 4, 1] }
 }
 
 // -----
 
-func.func @map_nested_foreach_to_threads_dynamic_trip_count(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token, %c9 : index, %c7 : index) -> memref<2 x 32 x f32> {
+func.func @map_nested_forall_to_threads_dynamic_trip_count(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token, %c9 : index, %c7 : index) -> memref<2 x 32 x f32> {
   %one = arith.constant 1 : index
   %c900 = arith.constant 900 : index
   %name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
             threads(%arg6, %arg7, %arg8) in (%arg12 = %one, %arg13 = %one, %arg14 = %one)
   {
     scf.forall (%i, %j) in (%c7, %c900) {
         %4 = memref.load %x[%i, %j] : memref<2 x 32 x f32>
         %5 = memref.load %y[%i, %j] : memref<2 x 32 x f32>
         %6 = math.fma %alpha, %4, %5 : f32
         memref.store %6, %y[%i, %j] : memref<2 x 32 x f32>
      }  { mapping = [#gpu.thread<y>, #gpu.thread<x>] }
     gpu.terminator
   }
   return %y : memref<2 x 32 x f32>
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{unsupported dynamic blockdim size}}
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [128, 4, 1] }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [128, 4, 1] }
 }
 
 // -----
 
-func.func @map_nested_foreach_to_threads_not_buffer(%x: tensor<32x32xf32>, %y: tensor<32x32xf32>, %z: tensor<32x32xf32>, %stream : !gpu.async.token) {
+func.func @map_nested_forall_to_threads_not_buffer(%x: tensor<32x32xf32>, %y: tensor<32x32xf32>, %z: tensor<32x32xf32>, %stream : !gpu.async.token) {
   %one = arith.constant 1 : index
   %name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
             threads(%arg6, %arg7, %arg8) in (%arg12 = %one, %arg13 = %one, %arg14 = %one)
   {
     %t = linalg.matmul ins(%x, %y: tensor<32x32xf32>, tensor<32x32xf32>) outs(%z : tensor<32x32xf32>) -> tensor<32x32xf32>
     gpu.terminator
   }
   return
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %matmul = transform.structured.match ops{["linalg.matmul"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  %foreach, %tiled = transform.structured.tile_to_forall_op %matmul num_threads [10, 20, 30] (mapping = [ #gpu.thread<y>, #gpu.thread<x>, #gpu.thread<z> ] )
+  %forall, %tiled = transform.structured.tile_to_forall_op %matmul num_threads [10, 20, 30] (mapping = [ #gpu.thread<y>, #gpu.thread<x>, #gpu.thread<z> ] )
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{only bufferized scf.forall lowers to gpu.thread_id}}
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [128, 4, 1] }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [128, 4, 1] }
 }
 
 // -----
 
 
-func.func @map_foreach_to_blocks_not_gpu_launch() -> () {
+func.func @map_forall_to_blocks_not_gpu_launch() -> () {
   // expected-note @below {{when applied to this payload op}}
   %1 = tensor.empty() : tensor<4xf32>
   return
 }
 transform.sequence failures(propagate) {
 ^bb0(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["tensor.empty"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{Given target is not gpu.launch}}
-  %1 = transform.gpu.map_foreach_to_blocks %funcop
+  %1 = transform.gpu.map_forall_to_blocks %funcop
 }
 
 // -----
 
-func.func @map_foreach_to_blocks_not_unique(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
+func.func @map_forall_to_blocks_not_unique(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
   %one = arith.constant 1 : index
   %c900 = arith.constant 900 : index
   %c9 = arith.constant 9 : index
   %c7 = arith.constant 7 : index
   // expected-note @below {{when applied to this payload op}}
   %name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
             threads(%arg6, %arg7, %arg8) in (%arg12 = %one, %arg13 = %one, %arg14 = %one)
   {
func.func @map_forall_to_blocks_not_unique(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
 
   return %y : memref<2 x 32 x f32>
 }
 
 transform.sequence failures(propagate) {
 ^bb0(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{could not find a unique topLevel scf.forall}}
-  %1 = transform.gpu.map_foreach_to_blocks %funcop
+  %1 = transform.gpu.map_forall_to_blocks %funcop
 }
 
 // -----
 
 // expected-note @below {{when applied to this payload op}}
-func.func @map_foreach_to_blocks_large_loop(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
+func.func @map_forall_to_blocks_large_loop(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
   %one = arith.constant 1 : index
   %c65537 = arith.constant 65536 : index
   %c9 = arith.constant 9 : index
   %c7 = arith.constant 7 : index
 
   scf.forall (%i, %j) in (%c7, %c65537) {
       %4 = memref.load %x[%i, %j] : memref<2 x 32 x f32>
       %5 = memref.load %y[%i, %j] : memref<2 x 32 x f32>
func.func @map_forall_to_blocks_large_loop(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
 
   return %y : memref<2 x 32 x f32>
 }
 
 transform.sequence failures(propagate) {
 ^bb0(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["func.func"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{could not find a unique topLevel scf.forall}}
-  %1 = transform.gpu.map_foreach_to_blocks %funcop { generate_gpu_launch }
+  %1 = transform.gpu.map_forall_to_blocks %funcop { generate_gpu_launch }
 }
 
 // -----
 
-func.func @map_foreach_to_blocks_large_loop(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
+func.func @map_forall_to_blocks_large_loop(%x: memref<2 x 32 x f32>, %y: memref<2 x 32 x f32>, %t: memref<32 x f32>, %alpha : f32, %stream : !gpu.async.token) -> memref<2 x 32 x f32> {
   %one = arith.constant 1 : index
   %c65535 = arith.constant 65535 : index
   scf.forall (%i, %j) in (%c65535, %c65535) {
       %4 = memref.load %x[%i, %j] : memref<2 x 32 x f32>
       %5 = memref.load %y[%i, %j] : memref<2 x 32 x f32>
       %6 = math.fma %alpha, %4, %5 : f32
       memref.store %6, %y[%i, %j] : memref<2 x 32 x f32>
   }  { mapping = [#gpu.block<x>, #gpu.block<y>] }
   return %y : memref<2 x 32 x f32>
 }
 
 transform.sequence failures(propagate) {
 ^bb0(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["func.func"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{Trying to launch a GPU kernel with gridDim = (65535, 65535, 1) blockDim = (1, 1, 1). It is larger than the limits.}}
-  %1 = transform.gpu.map_foreach_to_blocks %funcop { generate_gpu_launch }
+  %1 = transform.gpu.map_forall_to_blocks %funcop { generate_gpu_launch }
 }
 
 // -----
 
 !type = memref<32x32xf32>
 func.func @saxpy2d_singleloop(%x: !type, %y: !type, %stream : !gpu.async.token) -> !type {
   %c32 = arith.constant 32 : index
   %one = arith.constant 1 : index
func.func @saxpy2d_singleloop(%x: !type, %y: !type, %stream : !gpu.async.token) -> !type {
   }
   return %y : !type
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{#gpu.thread<x> is duplicated, cannot map different loops to the same processor}}
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [32, 32]}
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [32, 32]}
 }
 
 // -----
 
 func.func @tiling_buffer_semantic_op(%x: memref<32x32xf32>, %y: memref<32x32xf32>, %stream : !gpu.async.token) {
   %one = arith.constant 1 : index
   %name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
             threads(%arg6, %arg7, %arg8) in (%arg12 = %one, %arg13 = %one, %arg14 = %one)
func.func @tiling_buffer_semantic_op(%x: memref<32x32xf32>, %y: memref<32x32xf32>, %stream : !gpu.async.token) {
   }
   return
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %matmul = transform.structured.match ops{["linalg.generic"]} in %arg0 : (!pdl.operation) -> !pdl.operation
   // expected-error @below {{transform.structured.tile_to_forall_op failed to apply}}
-  %foreach, %tiled = transform.structured.tile_to_forall_op %matmul num_threads [10, 20, 30] (mapping = [ #gpu.thread<y>, #gpu.thread<x>, #gpu.thread<z> ] )
+  %forall, %tiled = transform.structured.tile_to_forall_op %matmul num_threads [10, 20, 30] (mapping = [ #gpu.thread<y>, #gpu.thread<x>, #gpu.thread<z> ] )
 }

mlir/test/Dialect/GPU/transform-gpu.mlir

%name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
     gpu.terminator
   }
   return %y : !type
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  transform.gpu.map_foreach_to_blocks %funcop { gridDim = [12, 9]}
+  transform.gpu.map_forall_to_blocks %funcop { gridDim = [12, 9]}
 }
 
 // -----
 
 !type = memref<2 x 32 x f32>
 !type1d = memref<32 x f32>
 
 // CHECK-LABEL: func.func @saxpy2d(
%name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
     gpu.terminator
   }
   return %y : !type
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [12, 9] }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [12, 9] }
 }
 
 // -----
 
 !type4d = memref<32x64x4x32xf32>
 
 // CHECK-LABEL: func.func @saxpy4d(
 // CHECK-SAME:    %[[ARGX:[0-9a-z]+]]: memref<32x64x4x32xf32>
scf.forall (%i, %j) in (%c32, %c64) {
     }  { mapping = [#gpu.thread<y>, #gpu.thread<x>] }
   }  { mapping = [#gpu.block<x>, #gpu.block<y>] }
   return %y : !type4d
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["func.func"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  %gpuLaunch = transform.gpu.map_foreach_to_blocks %funcop { generate_gpu_launch }
-  transform.gpu.map_nested_foreach_to_threads %gpuLaunch { blockDim = [32, 4, 1] }
+  %gpuLaunch = transform.gpu.map_forall_to_blocks %funcop { generate_gpu_launch }
+  transform.gpu.map_nested_forall_to_threads %gpuLaunch { blockDim = [32, 4, 1] }
 }
 
 // -----
 
 !type = memref<2 x 32 x f32>
 !type1d = memref<32 x f32>
 
 // CHECK-LABEL: func.func @saxpy2d_no_barrier(
%name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
     gpu.terminator
   }
   return %y : !type
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [12, 9, 1], syncAfterDistribute = false }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [12, 9, 1], syncAfterDistribute = false }
 }
 
 // -----
 
 !type = memref<32x32xf32>
 // CHECK-LABEL: func.func @saxpy2d_singleloop(
 // CHECK-SAME:    %[[ARGX:[0-9a-z]+]]: memref<32x32xf32>
 // CHECK-SAME:    %[[ARGY:[0-9a-z]+]]: memref<32x32xf32>
//      CHECK:   memref.load %[[ARGY]][%[[TIDX]], %[[TIDX]]]
     gpu.terminator
   }
   return %y : !type
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [32]}
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [32]}
 }
 
 // -----
 
 !type = memref<3 x 2 x 32 x f32>
 !type1d = memref<32 x f32>
 
 // CHECK-LABEL: func.func @saxpy3d_fold_id_z(
//      CHECK:   memref.store %{{.*}}, %{{.*}}[%[[C0]]
     gpu.terminator
   }
   return %y : !type
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [12, 9, 1], syncAfterDistribute = false }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [12, 9, 1], syncAfterDistribute = false }
 }
 
 // -----
 
 !type = memref<2 x 32 x f32>
 !type1d = memref<32 x f32>
 
 // CHECK-LABEL: func.func @map_multi_level(
%name = gpu.launch async[%stream] blocks(%arg3, %arg4, %arg5) in (%arg9 = %one, %arg10 = %one, %arg11 = %one)
     gpu.terminator
   }
   return %y : !type
 }
 
 transform.sequence failures(propagate) {
 ^bb1(%arg0: !pdl.operation):
   %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation
-  transform.gpu.map_nested_foreach_to_threads %funcop { blockDim = [12, 9] }
+  transform.gpu.map_nested_forall_to_threads %funcop { blockDim = [12, 9] }
 }

mlir/test/lib/Dialect/SCF/TestSCFUtils.cpp

forOp.walk([&schedule](Operation *op) {
         schedule[attrCycle.getInt()] =
             std::make_pair(op, unsigned(attrStage.getInt()));
       }
     });
   }
 
   /// Helper to generate "predicated" version of `op`. For simplicity we just
   /// wrap the operation in a scf.ifOp operation.
-  static Operation *predicateOp(Operation *op, Value pred,
-                                PatternRewriter &rewriter) {
+  static Operation *predicateOp(RewriterBase &rewriter, Operation *op,
+                                Value pred) {
     Location loc = op->getLoc();
     auto ifOp =
         rewriter.create<scf::IfOp>(loc, op->getResultTypes(), pred, true);
     // True branch.
     op->moveBefore(&ifOp.getThenRegion().front(),
                    ifOp.getThenRegion().front().begin());
     rewriter.setInsertionPointAfter(op);
     if (op->getNumResults() > 0)