mlir/lib/Dialect/MemRef/Transforms/ComposeSubView.cpp

//===- ComposeSubView.cpp - Combining composed subview ops ----------------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This file contains patterns for combining composed subview ops (i.e. subview
// of a subview becomes a single subview).
//
//===----------------------------------------------------------------------===//

#include "mlir/Dialect/MemRef/Transforms/ComposeSubView.h"
#include "mlir/Dialect/Affine/IR/AffineOps.h"
#include "mlir/Dialect/MemRef/IR/MemRef.h"
#include "mlir/IR/BuiltinAttributes.h"
#include "mlir/IR/OpDefinition.h"
#include "mlir/IR/PatternMatch.h"
#include "mlir/Transforms/DialectConversion.h"
#include "mlir/Transforms/GreedyPatternRewriteDriver.h"

using namespace mlir;

namespace {

// Replaces a subview of a subview with a single subview(both static and dynamic
// offsets are supported).
struct ComposeSubViewOpPattern : public OpRewritePattern<memref::SubViewOp> {
  using OpRewritePattern::OpRewritePattern;

  LogicalResult matchAndRewrite(memref::SubViewOp op,
                                PatternRewriter &rewriter) const override {
    // 'op' is the 'SubViewOp' we're rewriting. 'sourceOp' is the op that
    // produces the input of the op we're rewriting (for 'SubViewOp' the input
    // is called the "source" value). We can only combine them if both 'op' and
    // 'sourceOp' are 'SubViewOp'.
    auto sourceOp = op.getSource().getDefiningOp<memref::SubViewOp>();
    if (!sourceOp)
      return failure();

    // A 'SubViewOp' can be "rank-reducing" by eliminating dimensions of the
    // output memref that are statically known to be equal to 1. We do not
    // allow 'sourceOp' to be a rank-reducing subview because then our two
    // 'SubViewOp's would have different numbers of offset/size/stride
    // parameters (just difficult to deal with, not impossible if we end up
    // needing it).
    if (sourceOp.getSourceType().getRank() != sourceOp.getType().getRank()) {
      return failure();
    }

    // Offsets, sizes and strides OpFoldResult for the combined 'SubViewOp'.
    SmallVector<OpFoldResult> offsets, sizes, strides,
        opStrides = op.getMixedStrides(),
        sourceStrides = sourceOp.getMixedStrides();

    // The output stride in each dimension is equal to the product of the
    // dimensions corresponding to source and op.
    int64_t sourceStrideValue;
    for (auto &&[opStride, sourceStride] :
         llvm::zip(opStrides, sourceStrides)) {
      Attribute opStrideAttr = dyn_cast_if_present<Attribute>(opStride);
      Attribute sourceStrideAttr = dyn_cast_if_present<Attribute>(sourceStride);
      if (!opStrideAttr || !sourceStrideAttr)
        return failure();
      sourceStrideValue = cast<IntegerAttr>(sourceStrideAttr).getInt();
      strides.push_back(rewriter.getI64IntegerAttr(
          cast<IntegerAttr>(opStrideAttr).getInt() * sourceStrideValue));
    }

    // The rules for calculating the new offsets and sizes are:
    // * Multiple subview offsets for a given dimension compose additively.
    //   ("Offset by m and Stride by k" followed by "Offset by n" == "Offset by
    //   m + n * k")
    // * Multiple sizes for a given dimension compose by taking the size of the
    //   final subview and ignoring the rest. ("Take m values" followed by "Take
    //   n values" == "Take n values") This size must also be the smallest one
    //   by definition (a subview needs to be the same size as or smaller than
    //   its source along each dimension; presumably subviews that are larger
    //   than their sources are disallowed by validation).
    for (auto &&[opOffset, sourceOffset, sourceStride, opSize] :
         llvm::zip(op.getMixedOffsets(), sourceOp.getMixedOffsets(),
                   sourceOp.getMixedStrides(), op.getMixedSizes())) {
      // We only support static sizes.
      if (opSize.is<Value>()) {
        return failure();
      }
      sizes.push_back(opSize);
      Attribute opOffsetAttr = llvm::dyn_cast_if_present<Attribute>(opOffset),
                sourceOffsetAttr =
                    llvm::dyn_cast_if_present<Attribute>(sourceOffset),
                sourceStrideAttr =
                    llvm::dyn_cast_if_present<Attribute>(sourceStride);
      if (opOffsetAttr && sourceOffsetAttr) {

        // If both offsets are static we can simply calculate the combined
        // offset statically.
        offsets.push_back(rewriter.getI64IntegerAttr(
            cast<IntegerAttr>(opOffsetAttr).getInt() *
                cast<IntegerAttr>(sourceStrideAttr).getInt() +
            cast<IntegerAttr>(sourceOffsetAttr).getInt()));
      } else {
        AffineExpr expr;
        SmallVector<Value> affineApplyOperands;

        // Make 'expr' add 'sourceOffset'.
        if (auto attr = llvm::dyn_cast_if_present<Attribute>(sourceOffset)) {
          expr =
              rewriter.getAffineConstantExpr(cast<IntegerAttr>(attr).getInt());
        } else {
          expr = rewriter.getAffineSymbolExpr(affineApplyOperands.size());
          affineApplyOperands.push_back(sourceOffset.get<Value>());
        }

        // Multiply 'opOffset' by 'sourceStride' and make the 'expr' add the
        // result.
        if (auto attr = llvm::dyn_cast_if_present<Attribute>(opOffset)) {
          expr = expr + cast<IntegerAttr>(attr).getInt() *
                            cast<IntegerAttr>(sourceStrideAttr).getInt();
        } else {
          expr =
              expr + rewriter.getAffineSymbolExpr(affineApplyOperands.size()) *
                         cast<IntegerAttr>(sourceStrideAttr).getInt();
          affineApplyOperands.push_back(opOffset.get<Value>());
        }

        AffineMap map = AffineMap::get(0, affineApplyOperands.size(), expr);
        Value result = rewriter.create<affine::AffineApplyOp>(
            op.getLoc(), map, affineApplyOperands);
        offsets.push_back(result);
      }
    }

    // This replaces 'op' but leaves 'sourceOp' alone; if it no longer has any
    // uses it can be removed by a (separate) dead code elimination pass.
    rewriter.replaceOpWithNewOp<memref::SubViewOp>(
        op, op.getType(), sourceOp.getSource(), offsets, sizes, strides);
    return success();
  }
};

} // namespace

void mlir::memref::populateComposeSubViewPatterns(RewritePatternSet &patterns,
                                                  MLIRContext *context) {
  patterns.add<ComposeSubViewOpPattern>(context);
}