issue: 4409403 Fix heap corruption since c73d96a #337
Conversation
This file contains hidden or 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
This commit fixes a critical race condition in timer management for TCP sockets that was introduced in commit c73d96a. The heap corruption was caused by a race condition between the timer thread and socket destruction. Sockets could be deleted by the event handler thread while still being referenced by the timer thread in the timer collections, resulting in heap corruption when the timer thread attempted to access the deleted memory. In the original implementation, sockets were removed from timer collections and deleted asynchronously without proper synchronization with the timer processing thread. Fix: - Remove sockets from timer collections while still holding the socket lock, guaranteeing the timer thread cannot access sockets marked for deletion - Create a simplified deletion path that doesn't attempt to access timer collections again after socket cleanup Additionally, as an unrelated improvement, this patch fixes a lock leak in the early return path of sockinfo_tcp::clean_socket_obj() where a lock was acquired but not released when a socket was already marked as cleaned. The heap corruption stemmed from a fundamental architectural change that separated socket objects from their timer management without providing proper synchronization for the distributed socket lifecycle. Signed-off-by: Tomer Cabouly <[email protected]>
|
As we discussed offline, First, we need to identify the flow to run tcp timer outside of the internal thread context and this flow will like need to be avoided. Note, this comment assumes that delegated timers feature is off. |
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.
Description
This PR fixes a critical heap corruption issue in TCP socket timer management
that was introduced in commit c73d96a.
What
Fix race condition between timer thread and socket deletion in TCP socket timers.
Why ?
A race condition was introduced in commit c73d96a that allowed the timer thread
to access socket objects after they had been deleted by the event handler thread,
causing heap corruption and crashes.
How ?
The fix improves synchronization between threads:
after socket cleanup
Change type
What kind of change does this PR introduce?
Check list
Performance impact
We now hold the socket lock a bit longer during cleanup to remove it from timer collections
This is a one-time cost during socket closure, not during normal operation
Only affects the cleanup path, not data transmission or reception.
It can affect CPS tests, but preventing heap corruption far outweighs that minor performance impact imho.
How this bug was found?
attaching a pic for a sequence diagram explaining the faulty flow that was fixed:
