sbp2 slab corruiton in 2.6-test9

sbp2 slab corruiton in 2.6-test9

[Matthew J Galgoci]

Hi Folks,

I'm seeing slab corruption in 2.6-test9 when I do a 
cat /proc/scsi/scsi

Here's the trace:

ieee1394: Node added: ID:BUS[0-00:1023]  GUID[0050c5015000013d]
ieee1394: Node changed: 0-00:1023 -> 0-01:1023
sbp2: $Rev: 1034 $ Ben Collins <bcollins@debian.org>
scsi0 : SCSI emulation for IEEE-1394 SBP-2 Devices
ieee1394: sbp2: Logged into SBP-2 device
Slab corruption: start=e2166758, expend=e21667b7, problemat=e2166788
Last user: [<f088214b>](free_hpsb_packet+0x2b/0x40 [ieee1394])
Data: ************************************************D5 D6 D6 D6 01 00 00 
00 ***************************************A5
Next: 71 F0 2C .4B 21 88 F0 71 F0 2C .....................
slab error in check_poison_obj(): cache `hpsb_packet': object was modified 
after freeing
Call Trace:
 [<c013d2b8>] check_poison_obj+0x108/0x190
 [<c013d4d2>] slab_destroy+0x192/0x1a0
 [<c013f9c9>] reap_timer_fnc+0x129/0x1f0
 [<c013f8a0>] reap_timer_fnc+0x0/0x1f0
 [<c01256f0>] run_timer_softirq+0xb0/0x170
 [<c0121410>] do_softirq+0x90/0xa0
 [<c010c615>] do_IRQ+0xd5/0x110
 [<c0105000>] rest_init+0x0/0x30
 [<c010ab8c>] common_interrupt+0x18/0x20
 [<c0105000>] rest_init+0x0/0x30
 [<c01e1cf2>] acpi_processor_idle+0xd5/0x1c7
 [<c0105000>] rest_init+0x0/0x30
 [<c01088bc>] cpu_idle+0x2c/0x40
 [<c03826d1>] start_kernel+0x171/0x1a0
 [<c0382430>] unknown_bootoption+0x0/0x100
 

Re: sbp2 slab corruiton in 2.6-test9

[Ben Collins]

On Sun, Oct 26, 2003 at 02:01:26PM +0000, Matthew J Galgoci wrote:
> Hi Folks,
> 
> I'm seeing slab corruption in 2.6-test9 when I do a 
> cat /proc/scsi/scsi
> 

Known. The fix is non-trivial, so I am holding off on it until 2.6.0
gets out.

-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
WatchGuard - http://www.watchguard.com/

Re: sbp2 slab corruiton in 2.6-test9

[bill davidsen]

In article <20031026141837.GA7904@phunnypharm.org>,
Ben Collins  <bcollins@debian.org> wrote:
| On Sun, Oct 26, 2003 at 02:01:26PM +0000, Matthew J Galgoci wrote:
| > Hi Folks,
| > 
| > I'm seeing slab corruption in 2.6-test9 when I do a 
| > cat /proc/scsi/scsi
| > 
| 
| Known. The fix is non-trivial, so I am holding off on it until 2.6.0
| gets out.

Is the fix available somewhere? I have a machine with external devices
which does just that cat as part of rc.local, to see what's connected. I
do understand your not wanting to make a major change now, but a fix
could get some mileage if it were out.

Is this a new bug since test6?
-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

Re: sbp2 slab corruiton in 2.6-test9

[Pavel Machek]

Hi!

> > I'm seeing slab corruption in 2.6-test9 when I do a 
> > cat /proc/scsi/scsi
> > 
> 
> Known. The fix is non-trivial, so I am holding off on it until 2.6.0
> gets out.

You need to disable /proc/scsi/scsi at least, I guess. Its security
hole...

								Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]