I didn't see the dump_stack at all during any of my testing until upgrading to
2.6.17-rc. However, I DID see the absense of partitions on devices which
are partitioned.
Eric, when you saw the dump_stack, did you look to see if there
were partitions not present on your devices. E.g., /dev/sdg
exists but /dev/sdg1 doesn't but should.
I look for known partitions with this bit of code runing
in C-shell (/bin/tcsh).
dev is set to something like "sdg"
if (! -e /dev/${dev}1) then
echo /dev/${dev}1 does not exist
endif
I found there were as many dump stacks as there were missing
partitions. Hmmmm....
It looks like this is happening with other host adapter types
as well. In particular, qla1280 scsi.
http://marc.theaimsgroup.com/?t=114443491400001&r=1&w=2
Mike
Moore, Eric wrote:
> Mike - I saw this very same dump stack today when I was testing your
> latest patch.
>
>> Apr 21 10:51:57 duck kernel: kobject_add failed for 3:0:24:0
>
> I saw it when I turned the power back on to devices, during the
> rescan event processing.
>
> Eric
>
>
>> -----Original Message-----
>> From: Michael Reed [mailto:mdr@xxxxxxx]
>> Sent: Friday, April 21, 2006 10:06 AM
>> To: linux-scsi
>> Cc: James.Smart@xxxxxxxxxx; Moore, Eric; James Bottomley;
>> hare@xxxxxxx; Jeremy Higdon; Gary Hagensen
>> Subject: 2.6.17-rc2: kobject_add failed
>>
>> Hello,
>>
>> I've observed with my fibre channel configuration using LSI
>> fibre channel cards that I occasionally discover that a
>> partition on a device "doesn't exist", even though it does.
>> The number varies from 0 to some relatively small positive
>> integer. For the system boot associated with this email,
>> the number of missing partitions is eight.
>>
>> With 2.6.17-rc2 I noticed the following dump_stack() in the
>> syslog. There were eight of these dump stacks which appear to
>> correspond to the missing 8 partitions.
>>
>> I generated these dump stacks by disabling my switch until
>> all the fibre channel targets were removed and then
>> enabled the switch to cause target rediscovery.
>>
>> The "missing partitions" problem has been present in versions
>> of the kernel which did not include James Smart's recent
>> fixes to the transport so I believe that can be eliminated.
>> I waited to report this until James and I could get the
>> transport "stabilized".
>>
>> Has anyone seen anything similar? Any ideas about what
>> might be causing the missing partitions or how to go
>> about figuring it out?
>>
>> Thanks,
>> Mike
>>
>>
>> Apr 21 10:51:57 duck kernel: kobject_add failed for 3:0:24:0
>> with -EEXIST, don't try to register things
>> with the same name in the same directory.
>> Apr 21 10:51:57 duck kernel:
>> Apr 21 10:51:57 duck kernel: Call Trace:
>> Apr 21 10:51:57 duck kernel: [<a000000100012540>]
>> show_stack+0x40/0xa0
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fa80 bsp=e0000034f6429350
>> Apr 21 10:51:57 duck kernel: [<a0000001000125d0>]
>> dump_stack+0x30/0x60
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fc50 bsp=e0000034f6429338
>> Apr 21 10:51:57 duck kernel: [<a000000100408a80>]
>> kobject_add+0x3a0/0x420
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fc50 bsp=e0000034f64292f8
>> Apr 21 10:51:57 duck kernel: [<a0000001004de960>]
>> device_add+0xe0/0x2e0
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fc50 bsp=e0000034f64292b8
>> Apr 21 10:51:57 duck kernel: [<a0000001005605e0>]
>> scsi_sysfs_add_sdev+0x60/0x520
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fc50 bsp=e0000034f6429270
>> Apr 21 10:51:57 duck kernel: [<a00000010055c190>]
>> scsi_probe_and_add_lun+0x11b0/0x1440
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fc50 bsp=e0000034f6429208
>> Apr 21 10:51:57 duck kernel: [<a00000010055d7e0>]
>> __scsi_scan_target+0x720/0xb60
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fc70 bsp=e0000034f6429198
>> Apr 21 10:51:57 duck kernel: [<a00000010055e1d0>]
>> scsi_scan_target+0xd0/0x100
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fcd0 bsp=e0000034f6429148
>> Apr 21 10:51:57 duck kernel: [<a00000010056b3e0>]
>> fc_scsi_scan_rport+0x180/0x240
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fcd0 bsp=e0000034f6429110
>> Apr 21 10:51:57 duck kernel: [<a0000001000c9980>]
>> run_workqueue+0x1c0/0x280
>> Apr 21 10:51:57 duck kernel:
>> sp=e0000034f642fcd0 bsp=e0000034f64290d0
>> Apr 21 10:51:57 duck kernel: [<a0000001000cacd0>]
>> worker_thread+0x1d0/0x260
>> Apr 21 10:51:58 duck kernel:
>> sp=e0000034f642fcd0 bsp=e0000034f64290a0
>> Apr 21 10:51:58 duck kernel: [<a0000001000d2920>] kthread+0x220/0x2a0
>> Apr 21 10:51:58 duck kernel:
>> sp=e0000034f642fd50 bsp=e0000034f6429058
>> Apr 21 10:51:58 duck kernel: [<a000000100010ad0>]
>> kernel_thread_helper+0xd0/0x100
>> Apr 21 10:51:58 duck kernel:
>> sp=e0000034f642fe30 bsp=e0000034f6429030
>> Apr 21 10:51:58 duck kernel: [<a000000100009140>]
>> start_kernel_thread+0x20/0x40
>> Apr 21 10:51:58 duck kernel:
>> sp=e0000034f642fe30 bsp=e0000034f6429030
>> Apr 21 10:51:58 duck kernel: error 1
>> Apr 21 10:51:58 duck kernel: sddu: Write Protect is off
>> Apr 21 10:51:58 duck kernel: sddu: Mode Sense: ab 00 10 08
>> Apr 21 10:51:58 duck kernel: 3:0:24:0: Unexpected response
>> from lun 0 while scanning, scan aborted
>>
>
-
: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
