Fix tile-and-pad when padding doesn't span all dimension

[asaadaldien]

Summary

Without this tile-and-pad will never terminate if pad-fails.

[hanchung]

do we still need this?

[nicolasvasilache]

yes it is, let's land and iterate further

[asaadaldien]

Rebase...

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

if (failed(filter.checkAndNotify(rewriter, linalgOp)))
     return failure();
 
   Optional<TiledLinalgOp> res = tileLinalgOp(rewriter, linalgOp, options);
 
   if (!res)
     return failure();
 
   // Setup RAII guard to return properly.
-  bool succeeded = true;
   LinalgOp tiledOp = res->op;
   auto guard = llvm::make_scope_exit([&]() {
-    if (!succeeded)
-      return;
     // Return relevant information to derived pattern.
     result = *res;
     // Replace filter on both tiledOp and tiledAndPaddedOp, if necessary.
     filter.replaceLinalgTransformationFilter(rewriter, tiledOp);
     if (tiledOp != res->op)
       filter.replaceLinalgTransformationFilter(rewriter, res->op);
   });
 
   // Consider padding on the fly only if the op has tensor semantics.
   if (!options.paddingValueComputationFunction ||
       !linalgOp.hasTensorSemantics())
     return success();
 
   // Try to pad on the fly by rewriting res->op as a padded op.
   if (failed(rewriteAsPaddedOp(rewriter, *res, options))) {
     // Set so RAII guard does not propagate TiledLinalgOp to `result`.
-    succeeded = false;
     return failure();
   }
 
   // Do not perform replacement of `linalgOp`, let the derived patterns
   // do this as they see fit, from the resulting TiledLinalgOp.
   return success();
 }
 

mlir/test/Dialect/Linalg/tile-and-pad-tensors.mlir

-// RUN: mlir-opt %s -test-linalg-transform-patterns=test-tile-and-pad-pattern -canonicalize | FileCheck %s
+// RUN: mlir-opt %s -test-linalg-transform-patterns="test-tile-and-pad-pattern tile-sizes-for-padding=2,3,4" -canonicalize | FileCheck %s
+// RUN: mlir-opt %s -test-linalg-transform-patterns="test-tile-and-pad-pattern tile-sizes-for-padding=2,3" -canonicalize | FileCheck %s -check-prefix=CHECK-1DIM-TILE
 
 // CHECK-LABEL: func @matmul_tensors(
 // CHECK-SAME:    %[[TA:[0-9a-z]+]]: tensor<?x?xi8>
 // CHECK-SAME:    %[[TB:[0-9a-z]+]]: tensor<?x?xi8>
 // CHECK-SAME:    %[[TC:[0-9a-z]+]]: tensor<?x?xi32>) -> tensor<?x?xi32> {
 func @matmul_tensors(
   %arg0: tensor<?x?xi8>, %arg1: tensor<?x?xi8>, %arg2: tensor<?x?xi32>)
     -> tensor<?x?xi32> {
//      CHECK:   scf.yield %[[TD1]] : tensor<?x?xi32>
   %0 = linalg.matmul_i8_i8_i32 {__internal_linalg_transform__ = "tile-and-pad"}
       ins(%arg0, %arg1: tensor<?x?xi8>, tensor<?x?xi8>)
      outs(%arg2: tensor<?x?xi32>)
     -> tensor<?x?xi32>
 
 //      CHECK: return %[[TD0]] : tensor<?x?xi32>
   return %0 : tensor<?x?xi32>
 }
+
+// CHECK-1DIM-TILE: func @matmul_tensors(
+// CHECK-1DIM-TILE:    %[[TA:[0-9a-z]+]]: tensor<?x?xi8>
+// CHECK-1DIM-TILE:    %[[TB:[0-9a-z]+]]: tensor<?x?xi8>
+// CHECK-1DIM-TILE:    %[[TC:[0-9a-z]+]]: tensor<?x?xi32>) -> tensor<?x?xi32> {
+// CHECK-1DIM-TILE-NOT: scf.for
+// CHECK-1DIM-TILE: linalg.matmul_i8_i8_i32 ins(%[[TA]], %[[TB]] : tensor<?x?xi8>, tensor<?x?xi8>) outs(%[[TC]] : tensor<?x?xi32>) -> tensor<?x?xi32>

mlir/test/lib/Transforms/TestLinalgTransforms.cpp

Option<bool> testGenericToVectorPattern{
       llvm::cl::init(false)};
   Option<bool> testAffineMinSCFCanonicalizationPatterns{
       *this, "test-affine-min-scf-canonicalization-patterns",
       llvm::cl::desc("Test affine-min + scf canonicalization patterns."),
       llvm::cl::init(false)};
   Option<bool> testTileAndPadPattern{
       *this, "test-tile-and-pad-pattern",
       llvm::cl::desc("Test tile and pad pattern"), llvm::cl::init(false)};
+  ListOption<int64_t> tileSizesForPadding{
+      *this, "tile-sizes-for-padding",
+      llvm::cl::desc("Linalg tile sizes when tile+pad"), llvm::cl::ZeroOrMore,
+      llvm::cl::MiscFlags::CommaSeparated};
   Option<bool> testHoistPadding2Levels{*this, "test-hoist-padding-2-level",
                                        llvm::cl::desc("Test hoist padding"),
                                        llvm::cl::init(false)};
 };
 } // end anonymous namespace
 
 static void applyPatterns(FuncOp funcOp) {
   MLIRContext *ctx = funcOp.getContext();

 
 // For now, just assume it is the zero of type.
 // In the future, it should be the zero of type + op.
 static Value getNeutralOfLinalgOp(OpBuilder &b, OpOperand &op) {
   auto t = getElementTypeOrSelf(op.get().getType());
   return b.create<ConstantOp>(op.getOwner()->getLoc(), t, b.getZeroAttr(t));
 }
 
-static void applyTileAndPadPattern(FuncOp funcOp) {
+static void applyTileAndPadPattern(FuncOp funcOp, ArrayRef<int64_t> tileSizes) {
   MLIRContext *context = funcOp.getContext();
   OwningRewritePatternList tilingPattern;
   auto linalgTilingOptions =
       linalg::LinalgTilingOptions()
-          .setTileSizes({2, 3, 4})
+          .setTileSizes(tileSizes)
           .setPaddingValueComputationFunction(getNeutralOfLinalgOp);
   tilingPattern.insert<linalg::LinalgTilingPattern<linalg::MatmulI8I8I32Op>>(
       context, linalgTilingOptions,
       linalg::LinalgTransformationFilter(
           Identifier::get("tile-and-pad", context)));
   (void)applyPatternsAndFoldGreedily(funcOp, std::move(tilingPattern));
 }
 
return applyMatmulToVectorPatterns(getFunction(),
                                        testMatmulToVectorPatterns2dTiling);
   if (testVectorTransferForwardingPatterns)
     return applyVectorTransferForwardingPatterns(getFunction());
   if (testGenericToVectorPattern)
     return applyLinalgToVectorPatterns(getFunction());
   if (testAffineMinSCFCanonicalizationPatterns)
     return applyAffineMinSCFCanonicalizationPatterns(getFunction());
   if (testTileAndPadPattern)
-    return applyTileAndPadPattern(getFunction());
+    return applyTileAndPadPattern(getFunction(), tileSizesForPadding);
   if (testHoistPadding2Levels) {
     getFunction().walk([](linalg::PadTensorOp padTensorOp) {
       (void)linalg::hoistPaddingOnTensors(padTensorOp, 2);
     });
   }
 }
 
 namespace mlir {
 namespace test {
 void registerTestLinalgTransforms() {
   PassRegistration<TestLinalgTransforms> testTransformPatternsPass(
       "test-linalg-transform-patterns",
       "Test Linalg transformation patterns by applying them greedily.");
 }
 } // namespace test
 } // namespace mlir