Re: [RFC][PATCH] set_page_buffer_dirty should skip unmapped buffers

[Badari Pulavarty <pbadari@us.ibm.com>]

Andrew Morton wrote:
> On Wed, 6 Sep 2006 19:27:33 +0200
> Jan Kara <jack@suse.cz> wrote:
>
>   
>>   Ugh! Are you sure? For this path the buffer must be attached (only) to
>> the running transaction. But then how the commit code comes to it?
>> Somebody would have to even manage to refile the buffer from the
>> committing transaction to the running one while the buffer is in wbuf[].
>> Could you check whether someone does __journal_refile_buffer() on your
>> marked buffers, please? Or whether we move buffer to BJ_Locked list in
>> the write_out_data: loop? Thanks.
>>     
>
> The ext3-debug patch will help here.  It records, within the bh, the inputs
> from the last 32 BUFFER_TRACE()s which were run against this bh.  If a
> J_ASSERT fails then you get a nice trace of the last 32 "things" which
> happened to this bh, including the bh's state at that transition.  It
> basically tells you everything you need to know to find the bug.
>
> It's worth spending the time to become familiar with it - I used it a lot in
> early ext3 development.
>   

I will try the patch. Unfortunately, adding more debug is causing the 
problem reproduction
difficult.
> I've been unable to reproduce this crash, btw.  Is there some magic
> incantation apat from running `fsx-linux'?
>   
All I do is on a single 1k filesystem, run 4 copies of fsx (on 4 
different files, ofcourse).
I hit the assert anywhere between 10min-2hours.

Thanks,
Badari