re: dm-multipath: kernel paniced when I pull out one HBA card

[孙俊伟 <sunjw@onewaveinc.com>]

Hi,all

>looks like a failure in sysfs support code in the scsi module.  might be worth posting to the LKML.
But after I use dmsetup to remove the dm device, and "rmmod" the kernel module "dm_round_robin" and "dm_multipath"，
the kernel works fine as I pull out one HBA card for about 1 minutes and plug in it again. the messages are:

Dec 23 18:26:30 nd03 kernel: qla2300 0000:07:01.1: LOOP DOWN detected (2).
Dec 23 18:27:05 nd03 kernel:  rport-1:0-1: blocked FC remote port time out: removing target
Dec 23 18:27:05 nd03 kernel: Synchronizing SCSI cache for disk sdb:
Dec 23 18:27:05 nd03 kernel: FAILED
Dec 23 18:27:05 nd03 kernel:   status = 0, message = 00, host = 1, driver = 00
Dec 23 18:27:05 nd03 udev[3991]: udev_db.c: unable to read db file '/dev/.udevdb/block@sdb@sdb1'
Dec 23 18:27:05 nd03 udev[3991]: udev_remove.c: 'sdb1' not found in database, falling back on default name
Dec 23 18:27:05 nd03 udev[3991]: udev_remove.c: removing device node '/dev/sdb1'
Dec 23 18:27:05 nd03 udev[3989]: udev_db.c: unable to read db file '/dev/.udevdb/class@scsi_generic@sg1'
Dec 23 18:27:05 nd03 udev[3989]: udev_remove.c: 'sg1' not found in database, falling back on default name
Dec 23 18:27:05 nd03 udev[3989]: udev_remove.c: removing device node '/dev/sg1'
Dec 23 18:27:05 nd03 udev[4026]: udev_db.c: unable to read db file '/dev/.udevdb/block@sdb'
Dec 23 18:27:05 nd03 udev[4026]: udev_remove.c: 'sdb' not found in database, falling back on default name
Dec 23 18:27:05 nd03 udev[4026]: udev_remove.c: removing device node '/dev/sdb'
Dec 23 18:27:49 nd03 kernel:   <6>qla2300 0000:07:01.1: LIP reset occured (f8f7).
Dec 23 18:27:51 nd03 kernel: qla2300 0000:07:01.1: LIP occured (f8f7).
Dec 23 18:27:51 nd03 kernel: qla2300 0000:07:01.1: LOOP UP detected (2 Gbps).
Dec 23 18:27:52 nd03 kernel:   Vendor: TOYOU     Model: NetStor DA9220F   Rev: 342R
Dec 23 18:27:52 nd03 kernel:   Type:   Direct-Access                      ANSI SCSI revision: 03
Dec 23 18:27:52 nd03 kernel: SCSI device sdb: 999950336 512-byte hdwr sectors (511975 MB)
Dec 23 18:27:52 nd03 kernel: SCSI device sdb: drive cache: write back
Dec 23 18:27:52 nd03 kernel: SCSI device sdb: 999950336 512-byte hdwr sectors (511975 MB)
Dec 23 18:27:52 nd03 kernel: SCSI device sdb: drive cache: write back
Dec 23 18:27:52 nd03 kernel:  sdb: sdb1
Dec 23 18:27:52 nd03 kernel: Attached scsi disk sdb at scsi1, channel 0, id 0, lun 0
Dec 23 18:27:52 nd03 kernel: Attached scsi generic sg1 at scsi1, channel 0, id 0, lun 0,  type 0
Dec 23 18:27:52 nd03 scsi.agent[4049]: disk at /devices/pci0000:00/0000:00:02.0/0000:05:1d.0/0000:07:01.1/host1/rport-1:0-1/t
arget1:0:0/1:0:0:0
Dec 23 18:27:52 nd03 udev[4059]: udev_rules.c: no rule found, use kernel name 'sdb'
Dec 23 18:27:52 nd03 udev[4059]: udev_add.c: creating device node '/dev/sdb'
Dec 23 18:27:52 nd03 udev[4060]: udev_rules.c: no rule found, use kernel name 'sg1'
Dec 23 18:27:52 nd03 udev[4060]: udev_add.c: creating device node '/dev/sg1'
Dec 23 18:27:53 nd03 udev[4082]: udev_rules.c: no rule found, use kernel name 'sdb1'
Dec 23 18:27:53 nd03 udev[4082]: udev_add.c: creating device node '/dev/sdb1'

In summary:
kernel with module dm_multipath loaded:
	When I plug in the HBA card, the kernel finds a new device named "/dev/sdc" which is "/dev/sdb" orginally,
	and then kernel painc.

kernel without module dm_multipath loaded:
	When I plug in the HBA card, the kernel finds a new device named as the old one "/dev/sdb",
	and all works fine.

What's the matter?  where would be the bug ?

Best regards!
Luckey