[LU-17089] Bug in the barrier code could cause barrier freeze to fail everytime

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.16.0
• Severity: 3
• Assignee: Tim Day
• Reporter: Tim Day

Description

The barrier code has a bug that could cause the freeze to fail everytime. barrier freeze would be called before trying a FS backup, but would repeatedly fail due to an issue in the mdd_trans_create() function.

https://git.whamcloud.com/?p=fs/lustre-release.git;a=blob;f=lustre/mdd/mdd_trans.c;hb=2b0a71081d9c2465cb4b6368fede266fcde91b82#l49

The barrier entry increments the global counter barrier_writer, but it does not get decremented if mdd_child_ops() returns error. If the barrier_writer counter does not go to 0, the freeze cannot happen.