Re: 2.5.70-mm3

[Andrew Morton]

Mingming Cao <[e-mail]> wrote:
>
> Andrew Morton wrote:
> > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.70/2.5.70-mm3/
> > 
> Re run 50 fsx tests overnight again on mm3.  The tests are all 
> unresponsetive after running a while.  There are lots of disk errors and 
> one call back trace:
> 
> Any idea?
> 
> 
> SCSI disk error : <2 0 4 0> return code = 0x6000000
> end_request: I/O error, dev sdf, sector 14030152
> SCSI disk error : <2 0 0 0> return code = 0x6000000
> end_request: I/O error, dev sdb, sector 10526936
> SCSI disk error : <2 0 0 0> return code = 0x6000000
> end_request: I/O error, dev sdb, sector 10533048
> SCSI disk error : <2 0 0 0> return code = 0x6000000
> end_request: I/O error, dev sdb, sector 10536376

Sick disk, or sick SCSI driver.  Please swap some disks out and if it still
happens, send a report to [e-mail] 

> buffer layer error at fs/buffer.c:2835
> Call Trace:
>   [<c015a760>] drop_buffers+0xc0/0xd0
>   [<c015a7bb>] try_to_free_buffers+0x4b/0xb0
>   [<c01a22ef>] journal_invalidatepage+0xdf/0x130
>   [<c0193123>] ext3_invalidatepage+0x43/0x50
>   [<c0141bb7>] do_invalidatepage+0x27/0x30
>   [<c0141c4e>] truncate_complete_page+0x8e/0x90

That's expected - block_prepare_write() marked the page not uptodate when
it saw the I/O error.  Although that is a fairly bogus thing to be doing.