[MLIR][Tensor] Avoid crash on negative dimensions

In https://reviews.llvm.org/D151611, a check was added to the tensor verifier to
emit an error on negative tensor dimensions. This check allowed for dynamic
dimensions, hence negative dimensions were still able to get through the verifier.
This is a problem in situations such as #60558, where the dynamic dimension is
converted to a static (and possibly negative) dimension by another pass in the
compiler. This patch fixes that by doing another check during the
`StaticTensorGenerate` conversion, and return a failure if the dimension is
negative.

As a side-note, I have to admit that I do not know why returning a failure in
`StaticTensorGenerate` gives a nice "tensor dimensions must be non-negative"
error. I suspect that the verifier runs again when `return failure()` is called,
but I am not sure.

Fixes #60558.

Reviewed By: mehdi_amini

Differential Revision: https://reviews.llvm.org/D155728

diff --git a/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp b/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp
index 54690aa..8fc3494 100644 (file)
--- a/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp
+++ b/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp
@@ -1215,6 +1215,14 @@ struct StaticTensorGenerate : public OpRewritePattern<GenerateOp> {
     SmallVector<int64_t> newShape;
     operandsAndShape(resultType, dynamicExtents, newOperands, newShape);
 
+    for (int64_t newdim : newShape) {
+      // This check also occurs in the verifier, but we need it here too
+      // since intermediate passes may have some replaced dynamic dimensions
+      // by constants.
+      if (newdim < 0 && !ShapedType::isDynamic(newdim))
+        return failure();
+    }
+
     if (newOperands.size() == tensorFromElements.getDynamicExtents().size())
       return failure();
 
@@ -2126,7 +2134,8 @@ static Value foldExtractAfterInsertSlice(ExtractSliceOp extractOp) {
 }
 
 OpFoldResult ExtractSliceOp::fold(FoldAdaptor adaptor) {
-  if (auto splat = llvm::dyn_cast_if_present<SplatElementsAttr>(adaptor.getSource())) {
+  if (auto splat =
+          llvm::dyn_cast_if_present<SplatElementsAttr>(adaptor.getSource())) {
     auto resultType = llvm::cast<ShapedType>(getResult().getType());
     if (resultType.hasStaticShape())
       return splat.resizeSplat(resultType);

diff --git a/mlir/test/Dialect/Tensor/invalid-canonicalize.mlir b/mlir/test/Dialect/Tensor/invalid-canonicalize.mlir
new file mode 100644 (file)
index 0000000..decfd55
--- /dev/null
+++ b/mlir/test/Dialect/Tensor/invalid-canonicalize.mlir
@@ -0,0 +1,15 @@
+// RUN: mlir-opt <%s -split-input-file -verify-diagnostics -canonicalize
+
+// -----
+
+func.func @indirectly_generate_negative_size() -> tensor<?x8xi32> {
+  %cst = arith.constant 0 : i32
+  %c0 = arith.constant 0 : index
+  %size = affine.max affine_map<(d0) -> (d0 mod 64 - 8)>(%c0)
+  // expected-error@+1 {{tensor dimensions must be non-negative}}
+  %tensor = tensor.generate %size {
+  ^bb0(%arg0: index, %arg1: index):
+    tensor.yield %cst : i32
+  } : tensor<?x8xi32>
+  return %tensor : tensor<?x8xi32>
+}