Skip to content

fix(executorch): drain the stream when a device-to-host copy fails #4471

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
shoumikhin wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix(executorch): drain the stream when a device-to-host copy fails #4471

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 9 additions & 1 deletion cpp/src/torch_tensorrt/executorch/TensorRTBackend.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -589,6 +589,7 @@ Error TensorRTBackend::execute(BackendExecutionContext& context, DelegateHandle*
// output_staged_to_host, so outputs_needing_copy is empty on the skip path.
const bool must_sync = output_staged_to_host || input_staged_from_host || !g_user_stream_set;
if (must_sync) {
Error copy_err = Error::Ok;
for (auto& output : outputs_needing_copy) {
exec_aten::Tensor et_out = args[num_inputs + output.first]->toTensor();
cuda_err =
Expand All @@ -599,7 +600,11 @@ Error TensorRTBackend::execute(BackendExecutionContext& context, DelegateHandle*
"TensorRTBackend::execute: D2H copy failed for output %zu: %s",
output.first,
cudaGetErrorString(cuda_err));
return Error::InvalidProgram;
// The enqueue already succeeded, so the engine is still running on the
// stream. Drain below before returning, or the next call mutates a live
// execution context, which TensorRT forbids.
copy_err = Error::InvalidProgram;
break;
}
}
cuda_err = cudaStreamSynchronize(stream);
Expand All @@ -608,6 +613,9 @@ Error TensorRTBackend::execute(BackendExecutionContext& context, DelegateHandle*
ET_LOG(Error, "TensorRTBackend::execute: cudaStreamSynchronize failed: %s", cudaGetErrorString(cuda_err));
return Error::InvalidProgram;
}
if (copy_err != Error::Ok) {
return copy_err;
}
} else {
cuda_err = cudaEventRecord(engine->inflight_event, stream);
if (cuda_err != cudaSuccess) {
Expand Down
Loading