saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-26 19:52:38 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix a bug in Expand cuda op implementation. (#1528)

Browse source 
Description:
crash if the output shape has 0 in it. because the code to / output_shape[i]
Fix:
If the output shape has 0 which means output_shape.Size() is 0, so output should be null.
This commit is contained in:
Hector Li 2019-07-31 21:21:49 -07:00 committed by GitHub
parent b599360014
commit 57e2482089
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 7 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
onnxruntime/core/providers/cuda/tensor/expand.cc
View file

@ -21,6 +21,11 @@ Status Expand::ComputeInternal(OpKernelContext* ctx) const {
ORT_RETURN_IF_ERROR(ComputeOutputShape(Node().Name(), input0.Shape(), output_dims, output_shape));
auto rank = output_shape.NumDimensions();
auto& output_tensor = *ctx->Output(0, output_shape);
if (0 == output_shape.Size()) {
return Status::OK();
}
auto input_shape = input0.Shape().GetDims();
// pad input_dims with 1 to make ranks match
@ -40,6 +45,8 @@ Status Expand::ComputeInternal(OpKernelContext* ctx) const {
for (auto i = 0; i < rank; i++) {
in_span[i] = fast_divmod(static_cast<int>(input_shape[i]));
out_span[i] = fast_divmod(static_cast<int>(output_shape[i]));
// output_shape[i] won't be 0 here, it's covered in (0 == output_shape.Size())
// a null output will be returned for that case
subdim_size /= output_shape[i];
sdm_span[i] = static_cast<int>(subdim_size);
}