2.5.70-bk10 oops when trying to mount root from raid-1 device

2.5.70-bk10 oops when trying to mount root from raid-1 device

[Helge Hafting]

2.5.70-bk10 has some raid fixes, but raid-1 still fails unlike
2.5.70-mm4.

bk10 successfully discovers raid-1 and raid-0 arrays,
but this happens when the kernel tries to mount root:

<lots of ordinary boot messages>
md ... autorun DONE
<this is where I normally get VFS: Mounted root (ext2 filesystem) readonly.
  I got this instead:>
unable to handle kernel paging request at 5a5a5a86
EIP at put_all_bios+0x047/0x80
process swapper
raid_end_bio_io
deadline_next_request
raid1_end_request
scsi_request_fn
bio_endio
__end_that_request_first
scsi_end_request
scsi_io_completion
sd_rw_intr
scsi_finish_command
scsi_softirq
do_softirq
do_IRQ
default_idle
common_interrupt
default_idle
default_idle
cpu_idle
rest_init
start_kernel
unknown_bootoption
<0> kernel panic, exception in interrupt

This is a dual celeron with two scsi disks, with
two raid-1 arrays and one raid-0.  The
kernel is compiled with preempt and devfs,
using gcc-3.3

Helge Hafting

Re: 2.5.70-bk10 oops when trying to mount root from raid-1 device

[Andrew Morton]

Helge Hafting <helgehaf@aitel.hist.no> wrote:
>
> 2.5.70-bk10 has some raid fixes, but raid-1 still fails unlike
> 2.5.70-mm4.
> 
> bk10 successfully discovers raid-1 and raid-0 arrays,
> but this happens when the kernel tries to mount root:
> 
> <lots of ordinary boot messages>
> md ... autorun DONE
> <this is where I normally get VFS: Mounted root (ext2 filesystem) readonly.
>   I got this instead:>
> unable to handle kernel paging request at 5a5a5a86


This is "use of uninitialised memory".  0x6b6b6b6b is "use of freed memory".

It's supposed to be that way in Linus's tree too, but I screwed it up.

> EIP at put_all_bios+0x047/0x80

I'd be suspecting that the write_bios[] array isn't being memset somewhere.

                struct bio **bio = r1_bio->write_bios + i;

Re: 2.5.70-bk10 oops when trying to mount root from raid-1 device

[Nicolas Mailhot]

[-- Attachment #1: Type: text/plain, Size: 115 bytes --]

This looks suspiciously like

http://bugzilla.kernel.org/show_bug.cgi?id=760

Cheers,

-- 
Nicolas Mailhot

[-- Attachment #2: Ceci est une partie de message numériquement signée --]
[-- Type: application/pgp-signature, Size: 189 bytes --]