Hello!
We are using linux-2.6.21.5 .
When [echo "- - - " > scan] is executed before Devloss timeout is generated
after unplug the Fibre Channel cable connected to the HBA, the system panic occurs.
Panic can be reproduced by 100%.
Qla is generated though it tested with lpfc this time. The phenomenon doesn't
depend on the HBA driver.
lpfc 0000:01:0a.0: 0:1305 Link Down Event x2 receiv
ed Data: x2 x20 x0
rport-3:0-0: blocked FC remote port time out: removing target and saving binding
lpfc 0000:01:0a.0: 0:0203 Devloss timeout on WWPN 21:0:0:e0:0:41:a:cd NPort xef
Data: x8 x7 x1
BUG: unable to handle kernel paging request at virtual address 750001e9
printing eip:
e0867c74
*pde = 00000000
Oops: 0000 [#1]
SMP
Modules linked in: autofs4 hidp rfcomm l2cap bluetooth sunrpc ipv6 ib_iser rdma_
cm ib_cm iw_cm ib_sa ib_mad ib_core ib_addr iscsi_tcp libiscsi scsi_transport_is
csi dm_mirror dm_multipath dm_mod video sbs i2c_ec button battery asus_acpi ac l
p floppy sg pcspkr i2c_piix4 i2c_core tg3 parport_pc parport ide_cd cdrom serio_
raw lpfc scsi_transport_fc megaraid_mbox megaraid_mm aic79xx scsi_transport_spi
sd_mod scsi_mod ext3 jbd ehci_hcd ohci_hcd uhci_hcd
CPU: 0
EIP: 0060:[<e0867c74>] Not tainted VLI
EFLAGS: 00010206 (2.6.21.5 #1)
EIP is at scsi_is_host_device+0x0/0x11 [scsi_mod]
eax: 7500007d ebx: dfde4698 ecx: 00000000 edx: 00000001
esi: 7500007d edi: ffffffff ebp: dfde46e0 esp: d6f9de8c
ds: 007b es: 007b fs: 00d8 gs: 0033 ss: 0068
Process bash (pid: 8307, ti=d6f9d000 task=c96c9a90 task.ti=d6f9d000)
Stack: e086e7a7 ffffffff 00000001 00000001 ffffffff 00000000 00000001 dfde4698
ffffffff ffffffff df468000 e084be06 ffffffff 00000001 dfde46e0 df468000
e084bdae d6f9defb df46825c e086f01c ffffffff e0879289 d6f9df19 d6f9df0a
Call Trace:
[<e086e7a7>] scsi_scan_target+0x37/0xc0 [scsi_mod]
[<e084be06>] fc_user_scan+0x58/0x81 [scsi_transport_fc]
[<e084bdae>] fc_user_scan+0x0/0x81 [scsi_transport_fc]
[<e086f01c>] store_scan+0x95/0xc0 [scsi_mod]
[<c0457480>] __alloc_pages+0x59/0x29b
[<c0460d0b>] vma_merge+0x168/0x178
[<e086ef87>] store_scan+0x0/0xc0 [scsi_mod]
[<c054c33b>] class_device_attr_store+0x1b/0x1f
[<c04a497b>] sysfs_write_file+0xae/0xd8
[<c04a48cd>] sysfs_write_file+0x0/0xd8
[<c046ee75>] vfs_write+0xa8/0x12a
[<c046f402>] sys_write+0x41/0x67
[<c0404db8>] syscall_call+0x7/0xb
=======================
Code: 12 8b 43 40 3b 86 8c 01 00 00 75 07 8b 14 24 89 d8 ff d5 89 da 89 f8 e8 65
f5 ff ff 85 c0 89 c3 75 d4 5b 5b 5e 5f 5d c3 90 90 90 <81> b8 6c 01 00 00 fb 7c
86 e0 0f 94 c0 0f b6 c0 c3 83 ec 08 89
EIP: [<e0867c74>] scsi_is_host_device+0x0/0x11 [scsi_mod] SS:ESP 0068:d6f9de8c
Kernel panic - not syncing: Fatal exception
Rebooting in 1 seconds..
The procedure to panic is as follows.
1.Connect FC disk to server with FC HBA.
2.Confirm lun can be recognized.
3.Unplug FC cable from HBA.
4.Type "echo '- - -' > /sys/class/scsi_host/host<host no>/scan" before devloss-timeout
is generated.
5. When devloss-timeout is generated, the system does the panic.
The panic occurs by following sequence from our investigation.
1.Link Down Event
2.fc_remote_port_delete()
3.fc_user_scan()
4.fc_timeout_deleted_rport()
5.fc_user_scan()
6.panic
First of all, when linkdown is occured, fc_remote_port_delete() is called.
When [echo "- - - " > scan] is executed, fc_user_scan() is called at this time,
this command response doesn't return until the processing of devloss timeout
is completed because of linkdown.
When devloss timeout is detected, fc_timeout_deleted_rport()
is called, and rport is deleted here.
Then the response returns to fc_user_scan().
However, there is some inconsistency with rport because it already
starts continuing processing with original rport in fc_user_scan().
I guess this causes panic.
I am not on this mailing list, please cc to me.
to : minemoto@xxxxxxxxxxxxxxxx
Best regards,
Shintaro minemoto
-
To unsubscribe from this list: 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
