Fix race conditions in poll and condvar #62691
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
The signal_poll_event function was previously called without the poll lock held. This created a race condition between a thread calling k_poll to wait for an event and another thread signalling for this same event. This resulted in the waiting thread to stay pending and the handle to it getting removed from the notifyq, meaning it couldn't get woken up again. Signed-off-by: Ambroise Vincent <ambroise.vincent@arm.com>
Releasing the lock before notifying condvar led to a race condition between a thread calling k_condvar_wait to wait for a condition variable and another thread signalling for this same condition variable. This resulted in the waiting thread to stay pending and the handle to it getting removed from the notifyq, meaning it couldn't get woken up again. Signed-off-by: Ambroise Vincent <ambroise.vincent@arm.com>
ambroise-arm
requested review from
rlubos,
tbursztyka,
jukkar,
dcpleung,
nashif and
andyross
as code owners
jukkar
approved these changes
Sep 15, 2023
cfriedt
approved these changes
Sep 15, 2023
povergoing
approved these changes
Sep 16, 2023
There was a problem hiding this comment.
I am just wondering why the naming of z_handle_obj_poll_events looks so different from others in kernel/poll.c.
|
@peter-mitsis @andyross please review. |
peter-mitsis
approved these changes
Sep 18, 2023
andyross
reviewed
Sep 18, 2023
| @@ -470,11 +471,14 @@ static int signal_poll_event(struct k_poll_event *event, uint32_t state) | |||
| void z_handle_obj_poll_events(sys_dlist_t *events, uint32_t state) | |||
| { | |||
| struct k_poll_event *poll_event; | |||
| k_spinlock_key_t key = k_spin_lock(&lock); | |||
There was a problem hiding this comment.
Definitely looks like a lock was missing here, but I do worry about the possibility of a deadlock underneath signal_poll_event, which calls into a bunch of subsystems. Are we 100% sure those can't turn around and call back into something that wants the poll lock?
There was a problem hiding this comment.
I don't know. But given that z_impl_k_poll_signal_raise already calls signal_poll_event with the poll lock taken, I suppose such a case would have already shown up.
nashif
approved these changes
Sep 18, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
In the poll code,
signal_poll_eventis called by 2 functions:z_impl_k_poll_signal_raiseandz_handle_obj_poll_events. The first one correctly takes the poll lock. Fix the second one not taking the lock.In the sockets code, fix the mutex of the condition variable getting released before signaling for this same condition variable.
Both of those created race conditions which especially appeared on SMP-enabled configurations. The symptom was similar for both of them: a thread pending, waiting on an event, but never getting scheduled again even though the event happened.