Re: [PATCH(v3) 2.6.24] Fix fakephp deadlock

[Ian Abbott <abbotti@mev.co.uk>]

On 27/01/08 06:01, Andrew Morton wrote:
> Adding a deferred-work like this just because we can't get the locking and
> refcounting correct is a really sad hack.

True, but some would argue that the fakephp driver itself was a really 
sad hack. :-)

> Can we get the locking and refcounting right please?  Start by making that
> wait-for-refcount-to-go-away go away.

I guess that stuff was added for a good reason and fixed other potential 
problems - it just broke fakephp.  It's an unusual requirement to try 
and support -- a file that self-destructs when you write to it.  Are 
there any other cases in sysfs that do that?

-- 
-=( Ian Abbott @ MEV Ltd.    E-mail: <abbotti@mev.co.uk>        )=-
-=( Tel: +44 (0)161 477 1898   FAX: +44 (0)161 718 3587         )=-