[MLIR] Generalize expand_shape to take shape as explicit input

*DO NOT SUBMIT*

(This patch is for early design feedback only.  Notably, tests have not been
updated and the implementation is incomplete in some cases.)

This patch generalizes tensor.expand_shape and memref.expand_shape to consume
the output shape as a list of SSA values.  This enables us to implement generic
reshape operations with dynamic shapes using collapse_shape/expand_shape pairs.

The output_shape input to expand_shape follows the static/dynamic representation
that's also used in `tensor.extract_slice`.

Differential Revision: https://reviews.llvm.org/D140821

Author: ramiro050

mlir/include/mlir/Dialect/MemRef/IR/MemRefOps.td

@@ -1554,7 +1554,6 @@ def MemRef_ReshapeOp: MemRef_Op<"reshape", [
 class MemRef_ReassociativeReshapeOp<string mnemonic, list<Trait> traits = []> :
     MemRef_Op<mnemonic, !listconcat(traits,
       [Pure, ViewLikeOpInterface])>,
-    Arguments<(ins AnyStridedMemRef:$src, IndexListArrayAttr:$reassociation)>,
     Results<(outs AnyStridedMemRef:$result)>{
 
   code commonExtraClassDeclaration = [{
@@ -1579,10 +1578,6 @@ class MemRef_ReassociativeReshapeOp<string mnemonic, list<Trait> traits = []> :
     Value getViewSource() { return getSrc(); }
   }];
 
-  let assemblyFormat = [{
-    $src $reassociation attr-dict `:` type($src) `into` type($result)
-  }];
-
   let hasFolder = 1;
   let hasCanonicalizer = 1;
   let hasVerifier = 1;
@@ -1604,14 +1599,10 @@ def MemRef_ExpandShapeOp : MemRef_ReassociativeReshapeOp<"expand_shape", [
     Example:
 
     ```mlir
-    %r = memref.expand_shape %0 [[0, 1], [2]]
-        : memref<?x?xf32> into memref<?x5x?xf32>
+    %r = memref.expand_shape %0 [[0, 1], [2]] [%sz0, %sz1, 32]
+        : memref<?x32xf32> into memref<?x?x32xf32>
     ```
 
-    At most one dimension of a reassociation group (e.g., [0, 1] above) may be
-    dynamic in the result type. Otherwise, the op would be ambiguous, as it
-    would not be clear how the source dimension is extended.
-
     If an op can be statically proven to be invalid (e.g, an expansion from
     `memref<10xf32>` to `memref<2x6xf32>`), it is rejected by the verifier. If
     it cannot statically be proven invalid (e.g., the full example above; it is
@@ -1628,29 +1619,72 @@ def MemRef_ExpandShapeOp : MemRef_ReassociativeReshapeOp<"expand_shape", [
     there must be a dynamic result dimension in the corresponding reassociation
     group. Same for strides.
 
+    The representation for the output shape supports a partially-static
+    specification via attributes specified through the `static_output_shape`
+    argument.  A special sentinel value `ShapedType::kDynamic` encodes that the
+    corresponding entry has a dynamic value.  There must be exactly as many SSA
+    inputs in `output_shape` as there are `ShapedType::kDynamic` entries in
+    `static_output_shape`.
+
     Note: This op currently assumes that the inner strides are of the
     source/result layout map are the faster-varying ones.
   }];
 
+  let arguments = (ins AnyStridedMemRef:$src, IndexListArrayAttr:$reassociation,
+                       Variadic<Index>:$output_shape,
+                       DenseI64ArrayAttr:$static_output_shape);
+
+  let assemblyFormat = [{
+    $src $reassociation `output_shape`
+    custom<DynamicIndexList>($output_shape, $static_output_shape) attr-dict `:`
+    type($src) `into` type($result)
+  }];
+
   let builders = [
     // Builders using ReassociationIndices.
+    OpBuilder<(ins "Type":$resultType, "Value":$src,
+      "ArrayRef<ReassociationIndices>":$reassociation),
+    [{
+      SmallVector<OpFoldResult> inputShape = 
+          getMixedSizes($_builder, $_state.location, src);
+      std::pair<SmallVector<int64_t>, SmallVector<Value>> outputShape;
+      auto status = 
+          inferOutputShape($_builder, $_state.location, 
+                           resultType.cast<MemRefType>(), 
+                           reassociation, inputShape, outputShape);
+      (void) status;
+      assert(succeeded(status) && "unable to infer output shape"); 
+      build($_builder, $_state, resultType.cast<MemRefType>(), src,
+            getReassociationIndicesAttribute($_builder, reassociation),
+            outputShape.second, outputShape.first);
+    }]>,
     OpBuilder<(ins "Type":$resultType, "Value":$src,
       "ArrayRef<ReassociationIndices>":$reassociation,
-      CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
+      "ArrayRef<OpFoldResult>":$outputShape),
     [{
-      build($_builder, $_state, resultType, src, attrs);
-      $_state.addAttribute("reassociation",
-                          getReassociationIndicesAttribute($_builder, reassociation));
+      auto [staticOutputShape, dynamicOutputShape] =
+          decomposeMixedValues(SmallVector<OpFoldResult>(outputShape));
+      build($_builder, $_state, resultType, src,
+            getReassociationIndicesAttribute($_builder, reassociation),
+            dynamicOutputShape, staticOutputShape);
     }]>,
 
     // Builder using ReassociationExprs.
+    OpBuilder<(ins "Type":$resultType, "Value":$src,
+      "ArrayRef<ReassociationExprs>":$reassociation),
+    [{
+      auto reassociationIndices =
+          convertReassociationMapsToIndices(reassociation);
+      build($_builder, $_state, resultType, src, reassociationIndices);
+    }]>,
     OpBuilder<(ins "Type":$resultType, "Value":$src,
       "ArrayRef<ReassociationExprs>":$reassociation,
-      CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
+      "ArrayRef<OpFoldResult>":$outputShape),
     [{
       auto reassociationMaps =
-          convertReassociationMapsToIndices($_builder, reassociation);
-      build($_builder, $_state, resultType, src, reassociationMaps, attrs);
+          convertReassociationMapsToIndices(reassociation);
+      build($_builder, $_state, resultType, src, reassociationMaps,
+            outputShape);
     }]>,
 
     // Builder that infers the result layout map. The result shape must be
@@ -1663,6 +1697,14 @@ def MemRef_ExpandShapeOp : MemRef_ReassociativeReshapeOp<"expand_shape", [
     static FailureOr<MemRefType> computeExpandedType(
         MemRefType srcType, ArrayRef<int64_t> resultShape,
         ArrayRef<ReassociationIndices> reassociation);
+
+    // Infer the output shape for a memref.expand_shape when it is possible
+    // to do so.
+    static LogicalResult inferOutputShape(
+        OpBuilder &b, Location loc, MemRefType expandedType,
+        ArrayRef<ReassociationIndices> reassociation,
+        ArrayRef<OpFoldResult> inputShape,
+        std::pair<SmallVector<int64_t>, SmallVector<Value>> &outputShape);
   }];
 
   let hasVerifier = 1;
@@ -1713,6 +1755,12 @@ def MemRef_CollapseShapeOp : MemRef_ReassociativeReshapeOp<"collapse_shape", [
     source/result layout map are the faster-varying ones.
   }];
 
+  let arguments = (ins AnyStridedMemRef:$src, IndexListArrayAttr:$reassociation);
+
+  let assemblyFormat = [{
+    $src $reassociation attr-dict `:` type($src) `into` type($result)
+  }];
+
   let builders = [
     // Builders for a contracting reshape whose result type is computed from
     // `src` and `reassociation`.
@@ -1724,7 +1772,7 @@ def MemRef_CollapseShapeOp : MemRef_ReassociativeReshapeOp<"collapse_shape", [
       CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
     [{
       auto reassociationMaps =
-          convertReassociationMapsToIndices($_builder, reassociation);
+          convertReassociationMapsToIndices(reassociation);
       build($_builder, $_state, src, reassociationMaps, attrs);
     }]>,
 
@@ -1742,7 +1790,7 @@ def MemRef_CollapseShapeOp : MemRef_ReassociativeReshapeOp<"collapse_shape", [
       CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
     [{
       auto reassociationMaps =
-          convertReassociationMapsToIndices($_builder, reassociation);
+          convertReassociationMapsToIndices(reassociation);
       build($_builder, $_state, resultType, src, reassociationMaps, attrs);
     }]>
   ];

mlir/include/mlir/Dialect/Tensor/IR/TensorOps.td

@@ -998,8 +998,7 @@ class Tensor_ReassociativeReshapeOp<string mnemonic, list<Trait> traits = []> :
     Tensor_Op<mnemonic, !listconcat(traits, [
       DeclareOpInterfaceMethods<OpAsmOpInterface, ["getAsmResultNames"]>,
       Pure])>,
-    Arguments<(ins AnyRankedTensor:$src, IndexListArrayAttr:$reassociation)>,
-    Results<(outs AnyRankedTensor:$result)> {
+    Results<(outs AnyTensor:$result)> {
 
   code commonExtraClassDeclaration = [{
     static StringRef getReassociationAttrStrName() { return "reassociation"; }
@@ -1022,10 +1021,6 @@ class Tensor_ReassociativeReshapeOp<string mnemonic, list<Trait> traits = []> :
     }
   }];
 
-  let assemblyFormat = [{
-    $src $reassociation attr-dict `:` type($src) `into` type($result)
-  }];
-
   let hasFolder = 1;
   let hasCanonicalizer = 1;
   let hasVerifier = 1;
@@ -1038,11 +1033,16 @@ def Tensor_ExpandShapeOp : Tensor_ReassociativeReshapeOp<"expand_shape"> {
     rank whose sizes are a reassociation of the original `src`.
 
     A reassociation is defined as a continuous grouping of dimensions and is
-    represented with an array of DenseI64ArrayAttr attribute.
+    represented with an array of DenseI64ArrayAttr attribute.  The reassociation
+    maps applied to the result tensor with the higher rank must result in the
+    operand tensor with the smaller rank.
 
-    The verification rule is that the reassociation maps are applied to the
-    result tensor with the higher rank to obtain the operand tensor with the
-    smaller rank.
+    The representation for the output shape supports a partially-static
+    specification via attributes specified through the `static_output_shape`
+    argument.  A special sentinel value `ShapedType::kDynamic` encodes that the
+    corresponding entry has a dynamic value.  There must be exactly as many SSA
+    inputs in `output_shape` as there are `ShapedType::kDynamic` entries in
+    `static_output_shape`.
 
     The operand tensor type of a reshape can be zero-ranked if the result
     tensor type is statically shaped with all dimensions being unit extent. In
@@ -1052,39 +1052,87 @@ def Tensor_ExpandShapeOp : Tensor_ReassociativeReshapeOp<"expand_shape"> {
 
     ```mlir
     // Dimension expansion i -> (i', j') and (k) -> (k')
-    %b = tensor.expand_shape %a [[0, 1], [2]]
-        : tensor<?x?xf32> into tensor<?x?x?xf32>
+    %b = tensor.expand_shape %a [[0, 1], [2]] [%sz0, %sz1, 32]
+        : tensor<?x32xf32> into tensor<?x?x32xf32>
     ```
   }];
+
+  let arguments = (ins AnyTensor:$src, IndexListArrayAttr:$reassociation,
+                       Variadic<Index>:$output_shape,
+                       DenseI64ArrayAttr:$static_output_shape);
+
+  let assemblyFormat = [{
+    $src $reassociation `output_shape`
+    custom<DynamicIndexList>($output_shape, $static_output_shape) attr-dict `:`
+    type($src) `into` type($result)
+  }];
+
   let builders = [
     // Builders using ReassociationIndices.
+    OpBuilder<(ins "Type":$resultType, "Value":$src,
+      "ArrayRef<ReassociationIndices>":$reassociation),
+    [{
+      SmallVector<OpFoldResult> inputShape = 
+          getMixedSizes($_builder, $_state.location, src);
+      std::pair<SmallVector<int64_t>, SmallVector<Value>> outputShape;
+      auto status = 
+          inferOutputShape($_builder, $_state.location, 
+                           resultType.cast<RankedTensorType>(), 
+                           reassociation, inputShape, outputShape);
+      (void) status;
+      assert(succeeded(status) && "unable to infer output shape"); 
+      build($_builder, $_state, resultType.cast<RankedTensorType>(), src,
+            getReassociationIndicesAttribute($_builder, reassociation),
+            outputShape.second, outputShape.first);
+    }]>,
     OpBuilder<(ins "Type":$resultType, "Value":$src,
       "ArrayRef<ReassociationIndices>":$reassociation,
-      CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
+      "ArrayRef<OpFoldResult>":$outputShape),
     [{
-      build($_builder, $_state, resultType, src, attrs);
-      $_state.addAttribute("reassociation",
-          getReassociationIndicesAttribute($_builder, reassociation));
+      auto [staticOutputShape, dynamicOutputShape] =
+          decomposeMixedValues(SmallVector<OpFoldResult>(outputShape));
+      build($_builder, $_state, resultType, src,
+            getReassociationIndicesAttribute($_builder, reassociation),
+            dynamicOutputShape, staticOutputShape);
+    }]>,
+
+    // Builder using ReassociationExprs.
+    OpBuilder<(ins "Type":$resultType, "Value":$src,
+      "ArrayRef<ReassociationExprs>":$reassociation),
+    [{
+      auto reassociationIndices =
+          convertReassociationMapsToIndices(reassociation);
+      build($_builder, $_state, resultType, src, reassociationIndices);
     }]>,
     OpBuilder<(ins "Type":$resultType, "Value":$src,
       "ArrayRef<ReassociationExprs>":$reassociation,
-      CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
+      "ArrayRef<OpFoldResult>":$outputShape),
     [{
-      auto reassociationMaps =
-          convertReassociationMapsToIndices($_builder, reassociation);
-      build($_builder, $_state, resultType, src, reassociationMaps, attrs);
+      auto reassociationIndices =
+          convertReassociationMapsToIndices(reassociation);
+      build($_builder, $_state, resultType, src, reassociationIndices,
+            outputShape);
     }]>
   ];
 
   let extraClassDeclaration = commonExtraClassDeclaration # [{
     int64_t getCorrespondingSourceDim(int64_t resultDim);
+
+    // Infer the output shape for a tensor.expand_shape when it is possible
+    // to do so.
+    static LogicalResult inferOutputShape(
+        OpBuilder &b, Location loc, RankedTensorType expandedType,
+        ArrayRef<ReassociationIndices> reassociation,
+        ArrayRef<OpFoldResult> inputShape,
+        std::pair<SmallVector<int64_t>, SmallVector<Value>> &outputShape);
   }];
 
   let hasVerifier = 1;
 }
 
 def Tensor_CollapseShapeOp : Tensor_ReassociativeReshapeOp<"collapse_shape"> {
   let summary = "operation to produce a tensor with a smaller rank";
+  let arguments = (ins AnyTensor:$src, IndexListArrayAttr:$reassociation);
   let description = [{
     The `tensor.collapse_shape` op produces a new tensor with a smaller
     rank whose sizes are a reassociation of the original `src`.
@@ -1108,6 +1156,11 @@ def Tensor_CollapseShapeOp : Tensor_ReassociativeReshapeOp<"collapse_shape"> {
         : tensor<?x?x?xf32> into tensor<?x?xf32>
     ```
   }];
+
+  let assemblyFormat = [{
+    $src $reassociation attr-dict `:` type($src) `into` type($result)
+  }];
+
   let builders = [
     // Builders for a contracting reshape whose result type is computed from
     // `src` and `reassociation`.
@@ -1119,7 +1172,7 @@ def Tensor_CollapseShapeOp : Tensor_ReassociativeReshapeOp<"collapse_shape"> {
       CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
     [{
       auto reassociationMaps =
-          convertReassociationMapsToIndices($_builder, reassociation);
+          convertReassociationMapsToIndices(reassociation);
       build($_builder, $_state, src, reassociationMaps, attrs);
     }]>,
 
@@ -1137,7 +1190,7 @@ def Tensor_CollapseShapeOp : Tensor_ReassociativeReshapeOp<"collapse_shape"> {
       CArg<"ArrayRef<NamedAttribute>", "{}">:$attrs),
     [{
       auto reassociationMaps =
-          convertReassociationMapsToIndices($_builder, reassociation);
+          convertReassociationMapsToIndices(reassociation);
       build($_builder, $_state, resultType, src, reassociationMaps, attrs);
     }]>
   ];