Re: LOCKDEP: 3.9-rc1: mount.nfs/4272 still has locks held!

From: Tejun Heo
Date: Tue Mar 05 2013 - 20:34:10 EST


On Tue, Mar 05, 2013 at 05:14:23PM -0800, Tejun Heo wrote:
> Then, the operation simply isn't freezable while in progress and
> should be on the receiving end of failed-to-freeze error message and
> users who depend on suspend/hibernation working properly should be
> advised away from using nfs.
>
> It doesn't really changes anything. The current code is buggy. We're
> just not enforcing the rules strictly and people aren't hitting the
> bug often enough.

Just one more thing. Also, not being able to do retries without
side-effect doesn't mean it can't do retries at all. There are
syscalls which need to do things differently on re-entry (forgot which
one but there are several). They record the necessary state in the
restart block and resume from where they left off on re-entry. It
sure is hairy but is doable and we already have supporting
infrastructure. Not sure whether that would be worthwhile tho.

Thanks.

--
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/