ml: Use return instead of BreakLoop() in control process of Handwriting model.
We should not call BreakLoop() in the control process; we should only
use return.
BUG=chromium:1252229, chromium:1220271
TEST=Verified locally that code after BreakLoop() can still run,
TEST=so a return is needed.
Change-Id: Ic8ea44221261f5bcbf1efd14f094a27a963ec9ab
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform2/+/3177562
Commit-Queue: Amanda Deacon <amandadeacon@chromium.org>
Commit-Queue: Honglin Yu <honglinyu@chromium.org>
Tested-by: Amanda Deacon <amandadeacon@chromium.org>
Auto-Submit: Amanda Deacon <amandadeacon@chromium.org>
Reviewed-by: Honglin Yu <honglinyu@chromium.org>
diff --git a/ml/machine_learning_service_impl.cc b/ml/machine_learning_service_impl.cc
index df89a6b..ca3b016 100644
--- a/ml/machine_learning_service_impl.cc
+++ b/ml/machine_learning_service_impl.cc
@@ -435,7 +435,7 @@
channel, kModelName, &worker_pid)) {
// UMA metrics has already been reported in `SpawnWorkerProcessAndGetPid`.
std::move(callback).Run(LoadHandwritingModelResult::LOAD_MODEL_ERROR);
- brillo::MessageLoop::current()->BreakLoop();
+ return;
}
Process::GetInstance()
->SendMojoInvitationAndGetRemote(worker_pid, std::move(channel),
