On 21/05/2019 13:34, Mark Thomas wrote: > On 21/05/2019 09:03, Keiichi Fujino wrote: >> 2019年5月16日(木) 21:55 <ma...@apache.org <mailto:ma...@apache.org>>: >> >> This is an automated email from the ASF dual-hosted git repository. >> >> markt pushed a commit to branch master >> in repository https://gitbox.apache.org/repos/asf/tomcat.git >> >> commit ab70de3278d5e506661faeb5febf71a061b89179 >> Author: Mark Thomas <ma...@apache.org <mailto:ma...@apache.org>> >> AuthorDate: Thu May 16 13:36:39 2019 +0100 >> >> Fix https://bz.apache.org/bugzilla/show_bug.cgi?id=62841 poss >> deadlock >> >> Refactor the DeltaRequest serialization to reduce the window during >> which the DeltaSession is locked and to remove a potential cause of >> deadlocks during serialization.
<snip/> >> Do you have any plan for applying this for using BackupManager ? >> There are similar codes in AbstractReplicatedMap#replicate. >> Do we need to apply a similar fix to AbstractReplicatedMap#replicate ? > > Probably. I'll trace the call hierarchy back from > DeltaRequest.serialize() and ensure nothing is holding the session lock. This is proving much trickier than it first appeared. I've had to do a fair bit of refactoring to fix this for the DeltaManager. My first (untested) attempt is here: https://github.com/markt-asf/tomcat/commits/bz62481-backup-manager (the final 3 commits) It is a bit too invasive for me to be comfortable just committing it. I also want to run the unit tests (although they don't test this code much). Feedback welcome. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
