sudheerv edited a comment on issue #7096: URL: https://github.com/apache/trafficserver/issues/7096#issuecomment-707699831
> Interesting. Sounds like a lot of good debugging. > > Is changing the read_vio target continuation (with potentially different mutexes) also a problem? During the lifecycle of the server session, the continuation will change from HttpSM to HttpSessionManager. Hmm..you mean the pool mutex? Yeah, I think that should be okay, as long as the VC is “double locked” (with the current mutex first and then the newly swapped future mutex), the net i/o should be okay. There’s a tiny window where the double lock is missing which is part of the fix I made. tldr; as long as any changes to the VC’s mutex is done by locking the previously and newly attached mutex simultaneously should be safe. The problem happens when swapping to a new mutex without locking that new mutex. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
