This repository was archived by the owner on Feb 3, 2025. It is now read-only.
Fix performance metrics deadlock #2917
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Load the pr2 world with --lockstep and then subscribe to /gazebo/performance_metrics to reproduce the deadlock. Signed-off-by: Steve Peters <scpeters@openrobotics.org>
Currently the PublishPerformanceMetrics function is called within SensorManager.cc by each SensorContainer with protection by the mutex of each SensorContainer. This leads to a deadlock when a world has sensors in more than one container (i.e. more than one category of sensors) as described in gazebosim#2902. Since the PublishPerformanceMetrics function is not a member function of SensorContainer and uses only global variables, the function can be called by the SensorManager thread instead of by each SensorContainer thread. This change will avoid the deadlock. Signed-off-by: Steve Peters <scpeters@openrobotics.org>
ahcorde
approved these changes
Jan 21, 2021
scpeters
added a commit
to scpeters/gazebo
that referenced
this pull request
Jan 26, 2021
* Add test for performance metrics deadlock Load the pr2 world with --lockstep and then subscribe to /gazebo/performance_metrics to reproduce the deadlock. * Fix deadlock when publishing performance metrics Currently the PublishPerformanceMetrics function is called within SensorManager.cc by each SensorContainer with protection by the mutex of each SensorContainer. This leads to a deadlock when a world has sensors in more than one container (i.e. more than one category of sensors) as described in gazebosim#2902. Since the PublishPerformanceMetrics function is not a member function of SensorContainer and uses only global variables, the function can be called by the SensorManager thread instead of by each SensorContainer thread. This change will avoid the deadlock. Signed-off-by: Steve Peters <scpeters@openrobotics.org>
scpeters
added a commit
to scpeters/gazebo
that referenced
this pull request
Jan 26, 2021
* Add test for performance metrics deadlock Load the pr2 world with --lockstep and then subscribe to /gazebo/performance_metrics to reproduce the deadlock. * Fix deadlock when publishing performance metrics Currently the PublishPerformanceMetrics function is called within SensorManager.cc by each SensorContainer with protection by the mutex of each SensorContainer. This leads to a deadlock when a world has sensors in more than one container (i.e. more than one category of sensors) as described in gazebosim#2902. Since the PublishPerformanceMetrics function is not a member function of SensorContainer and uses only global variables, the function can be called by the SensorManager thread instead of by each SensorContainer thread. This change will avoid the deadlock. Signed-off-by: Steve Peters <scpeters@openrobotics.org>
scpeters
added a commit
that referenced
this pull request
Jan 28, 2021
* Add test for performance metrics deadlock Load the pr2 world with --lockstep and then subscribe to /gazebo/performance_metrics to reproduce the deadlock. * Fix deadlock when publishing performance metrics Currently the PublishPerformanceMetrics function is called within SensorManager.cc by each SensorContainer with protection by the mutex of each SensorContainer. This leads to a deadlock when a world has sensors in more than one container (i.e. more than one category of sensors) as described in #2902. Since the PublishPerformanceMetrics function is not a member function of SensorContainer and uses only global variables, the function can be called by the SensorManager thread instead of by each SensorContainer thread. This change will avoid the deadlock. Signed-off-by: Steve Peters <scpeters@openrobotics.org>
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As reported in #2902, there is a deadlock in
SensorManagerwhen subscribing to/gazebo/performance_metricsin a world that has sensors in more than one container (i.e. more than one category of sensors). This can be reproduced by loading thepr2.worldwith--lockstepand then viewing the/gazebo/performance_metricstopic in the Topic Viewer. I've added a regression test in a9e1ee3 that reproduces this deadlock.To fix the deadlock, I noticed that the
PublishPerformanceMetricsis called by eachSensorContainerwith protection by a mutex, which causes the deadlock. Since that function is not a member function and uses only global variables, it does not need to be run by separate threads. So I've moved it toSensorManager::Updatein 0f48128, and that seems to fix the deadlock.