[ofa-general] Re: [NEW PATCH] IB/mad: Fix possible lock-lock-timer deadlock

[Roland Dreier]

 > The above issue does not occur with the for-next branch of the
 > infiniband git tree, but does occur with 2.6.31-rc9 + aforementioned
 > patches.
 > 
 > As far as I can see commit 721d67cdca5b7642b380ca0584de8dceecf6102f
 > (http://git.kernel.org/?p=linux/kernel/git/roland/infiniband.git;a=commitdiff;h=721d67cdca5b7642b380ca0584de8dceecf6102f)
 > is not yet included in 2.6.31-rc9. Could this be related to the above
 > issue ?

Yes, that would make sense.  "priv->lock" -- ie the ipoib lock whose
coverage is reduced in 721d67cd -- is in the lockdep report you posted.
So it seems likely that 721d67cd makes the mad_rmpp report not trigger.
However I think the mad_rmpp code does still have a lock-lock-timer
problem that could cause lockdep reports in the future, so I'll have a
look at fixing it.

Do you happen to have the full lockdep output from this test handy?  I'm
curious to see exactly how the mad_rmpp lock gets linked to priv->lock.

Thanks,
  Roland