JFFS2 deadlock with alloc_sem

[David Woodhouse]

On Fri, 2007-06-08 at 14:26 -0500, Dave Kleikamp wrote:
> Forgive me for not following up properly, but I'm not on the mailing
> list, and I'm following up from the archives. 

Hm, sorry about that. The archives should be fixed (again) now -- if you
use the mailto: link at the top of an archived mail such as the one at
http://lists.infradead.org/pipermail/linux-mtd/2007-June/018477.html
your response should have a correct In-Reply-To: header and be part of
the thread.

> What do you think of this patch, and does it have any affect on
> Nathan's deadlock?

It looks sensible. Nathan?

> Note: If this works out, jffs2_readpage and jffs2_do_readpage_unlock
> can probably be collapsed into one function.

Yeah. And the ordering of f->sem vs. page locking should be documented
in README.Locking, now that we actually have rules for it :)

-- 
dwmw2