Description of problem: 1. When two threads execute in parallel in dht_getxattr_cbk it may so happen that, both may find local->xattr to be NULL. As a result dht_aggregate_xattr may not get executed. 2. In dht_getxattr_cbk, thread1 thread2 T1 this_call_cnt = 2 -1 T2 this_call_cnt = 1 - 1 T3 fills local_xattr T4 DHT_STACK_UNWIND -> local_wipe T5 tries to dereference local which is already freed, leading to crash. Version-Release number of selected component (if applicable): How reproducible: Got the crash from gerrit: http://build.gluster.org/job/rackspace-regression-2GB-triggered/7345/consoleFull Steps to Reproduce: 1. 2. 3. Actual results: client process crashes Expected results: Additional info:
REVIEW: http://review.gluster.org/10389 (dht: tackle thread race in dht_getxattr_cbk) posted (#3) for review on master by Susant Palai (spalai)
COMMIT: http://review.gluster.org/10389 committed in master by Shyamsundar Ranganathan (srangana) ------ commit 6bde16f7dc4a43d85e488f25ad679abfd24e72d1 Author: Susant Palai <spalai> Date: Sun Apr 26 23:49:56 2015 +0530 dht: tackle thread race in dht_getxattr_cbk problem: 1. When two threads execute in parallel in dht_getxattr_cbk it may so happen that, both may find local->xattr to be NULL. As a result dht_aggregate_xattr may not get executed. 2. In dht_getxattr_cbk, thread1 thread2 T1 this_call_cnt = 2 -1 T2 this_call_cnt = 1 - 1 T3 fills local_xattr T4 DHT_STACK_UNWIND -> local_wipe T5 tries to dereference local which is already freed, leading to crash. Solution: for problem1: Execute critical section inside frame lock to resolve race. for problem2: Calculate this_call_count just before out section. Change-Id: I9827ac8fafebb0c733a4e4f3c710b752f1cd45fa BUG: 1215592 Signed-off-by: Susant Palai <spalai> Reviewed-on: http://review.gluster.org/10389 Reviewed-by: Anuradha Talur <atalur> Reviewed-by: N Balachandran <nbalacha> Reviewed-by: Kotresh HR <khiremat> Tested-by: NetBSD Build System Tested-by: Gluster Build System <jenkins.com> Reviewed-by: Shyamsundar Ranganathan <srangana>
changing version to mainline in order to retire pre-release. If you know the appropriate, correct version for this bug, please set it.
This bug is getting closed because a release has been made available that should address the reported issue. In case the problem is still not fixed with glusterfs-3.8.0, please open a new bug report. glusterfs-3.8.0 has been announced on the Gluster mailinglists [1], packages for several distributions should become available in the near future. Keep an eye on the Gluster Users mailinglist [2] and the update infrastructure for your distribution. [1] http://blog.gluster.org/2016/06/glusterfs-3-8-released/ [2] http://thread.gmane.org/gmane.comp.file-systems.gluster.user