[fdo-internals] #895: SQLite unit test deadlocks

[Jackie Ng]

Hi All,

I've noticed recently that the SQLite unit test suite deadlocks on Linux (on
both CentOS and Ubuntu)

http://trac.osgeo.org/fdo/ticket/895

I strongly suspect that it is due to this particular submission

http://trac.osgeo.org/fdo/changeset/7133

In that ticket I created, I speculated that it may be that the
CriticalSectionHolder may be attempting to re-acquire an already acquired
m_csMd critical section in one of the methods further up the stack trace and
it is being blocked perpetually because said critical section was already
acquired in one of the parent methods down the stack trace.

Am I on the right track? Would this be an invalid use of
CriticalSectionHolder and thus the cause of this deadlock?

- Jackie



--
View this message in context: http://osgeo-org.1560.x6.nabble.com/895-SQLite-unit-test-deadlocks-tp5164635.html
Sent from the FDO Internals mailing list archive at Nabble.com.