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

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

Andrew Morton wrote:
> On Fri, 01 Sep 2006 09:32:22 -0700
> Badari Pulavarty <pbadari@us.ibm.com> wrote:
>
>   
>>>> Kernel BUG at fs/buffer.c:2791
>>>> invalid opcode: 0000 [1] SMP
>>>>
>>>> Its complaining about BUG_ON(!buffer_mapped(bh)).
>>>>         
>
> I need to have a little think about this, remember what _should_ be
> happening in this situation.
>
> We (mainly I) used to do a huge amount of fsx-linux testing on 1k blocksize
> filesystems.  We've done something to make this start happening.  Part of
> resolving this bug will be working out what that was.
>   

Here is the progress in tracking this down so far.

I am able to reproduce the problem on following kernel versions.

2.6.18-rc5
2.6.18-rc4
2.6.17.11
2.6.16.28
2.6.15.7
2.6.14.7

I am yet to find a latest kernel version - where this works :(
I am going to try older versions of the kernel.

Thanks,
Badari


-- 
VGER BF report: H 3.60822e-15