[mlir][tosa] Add support for 0 dim in ReshapeOp

[fjsyrmia]

This change adds support for onnx models with 0 values in reshape op
shape input.

By default, when any value in the ‘shape’ input is equal to zero the
corresponding dimension value is copied from the input tensor dynamically.
Source: https://onnx.ai/onnx/operators/onnx__Reshape.html

[github-actions]

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be
notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write
permissions for the repository. In which case you can instead tag reviewers by
name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review
by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate
is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

[llvmbot]

@llvm/pr-subscribers-mlir

Author: None (fjsyrmia)

Changes

By default, when any value in the ‘shape’ input is equal to zero the
corresponding dimension value is copied from the input tensor dynamically.
Source: https://onnx.ai/onnx/operators/onnx__Reshape.html

---

Full diff: https://github.com/llvm/llvm-project/pull/84004.diff

1 Files Affected:

• (modified) mlir/lib/Dialect/Tosa/IR/TosaOps.cpp (+7-1)

diff --git a/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp b/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
index 62d07859e32f6e..7b6a42e509e015 100644
--- a/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
+++ b/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
@@ -928,10 +928,16 @@ LogicalResult tosa::ReshapeOp::inferReturnTypeComponents(
   // dimension.
   int64_t numElements = inputShape.getNumElements();
   int64_t staticMul = 1;
-  for (auto val : newShapeValue) {
+  int i = 0;
+  for (auto &val : newShapeValue) {
+    if (val == 0) {
+      // Swap 0 dim with corresponding input dim
+      val = inputShape.getDimSize(i);
+    }
     if (!ShapedType::isDynamic(val)) {
       staticMul *= val;
     }
+    i++;
   }
 
   // Determine the length of the dynamic dimension.

[llvmbot]

@llvm/pr-subscribers-mlir-tosa

Author: None (fjsyrmia)

Changes

By default, when any value in the ‘shape’ input is equal to zero the
corresponding dimension value is copied from the input tensor dynamically.
Source: https://onnx.ai/onnx/operators/onnx__Reshape.html

---

Full diff: https://github.com/llvm/llvm-project/pull/84004.diff

1 Files Affected:

• (modified) mlir/lib/Dialect/Tosa/IR/TosaOps.cpp (+7-1)

diff --git a/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp b/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
index 62d07859e32f6e..7b6a42e509e015 100644
--- a/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
+++ b/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
@@ -928,10 +928,16 @@ LogicalResult tosa::ReshapeOp::inferReturnTypeComponents(
   // dimension.
   int64_t numElements = inputShape.getNumElements();
   int64_t staticMul = 1;
-  for (auto val : newShapeValue) {
+  int i = 0;
+  for (auto &val : newShapeValue) {
+    if (val == 0) {
+      // Swap 0 dim with corresponding input dim
+      val = inputShape.getDimSize(i);
+    }
     if (!ShapedType::isDynamic(val)) {
       staticMul *= val;
     }
+    i++;
   }
 
   // Determine the length of the dynamic dimension.

[GeorgeARM]

Can a shape value be 0?

[Hsiangkai]

From TOSA specification, the tensor shape in each dimension must be greater than or equal to 1.
See https://www.mlplatform.org/tosa/tosa_spec.html#_tensors

[Jerry-Ge]

@fjsyrmia is it ok to close this PR?

[Jerry-Ge]

Will close this stale PR if no response by the end of this week.

[Jerry-Ge]

close stale PR