#187 scsi eh problem in Lunix 2.6 iscsi

[maximkoz]

Hi,
I see a strange behavior of scsi mid-layer in 2.6.12.5,
still didn't check in newer versions.
Suppose you started iSCSI and got a new host adapter
with several LUNs on it.
Now if you remove and add one of the existing LUNs via
/proc/scsi/scsi (or just add –nothing should happen in
this case),
then when iSCSI driver is removed, there's still
appropriate scsi_eh thread active, exactly the one you
tackled with by /proc/scsi/scsi.
It is known to us that starting/stopping/starting iSCSI
driver advances host adapter numbers (which is itself
strange), but scsi_eh handler left is something new.
What do you think?

Thanks,

Maxim.

# ps -ef | egrep 'iscsi|scsi_eh' | grep -v grep
root 388 1 0 00:42 ? 00:00:00 [scsi_eh_0]
root 389 1 0 00:42 ? 00:00:00 [scsi_eh_1]
root 422 1 0 00:42 ? 00:00:00 [scsi_eh_2]
root 427 1 0 00:42 ? 00:00:00 [scsi_eh_3]
# /etc/init.d/iscsi start
...
# more /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: ATA Model: HDS724040KLSA80 Rev: KFAO
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi5 Channel: 00 Id: 00 Lun: 00
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi4 Channel: 00 Id: 00 Lun: 00
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi5 Channel: 00 Id: 00 Lun: 01
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi4 Channel: 00 Id: 00 Lun: 01
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
# echo "scsi remove-single-device 4 0 0 1" >
/proc/scsi/scsi
# more /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: ATA Model: HDS724040KLSA80 Rev: KFAO
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi5 Channel: 00 Id: 00 Lun: 00
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi4 Channel: 00 Id: 00 Lun: 00
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi5 Channel: 00 Id: 00 Lun: 01
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05

# echo "scsi add-single-device 4 0 0 1" > /proc/scsi/scsi
...
# more /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: ATA Model: HDS724040KLSA80 Rev: KFAO
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi5 Channel: 00 Id: 00 Lun: 00
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi4 Channel: 00 Id: 00 Lun: 00
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi5 Channel: 00 Id: 00 Lun: 01
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
Host: scsi4 Channel: 00 Id: 00 Lun: 01
Vendor: MY-STOR Model: AGE1030 Rev: 0.4
Type: Direct-Access ANSI SCSI
revision: 05
# /etc/init.d/iscsi stop
...
# rmmod scsi_transport_iscsi
# lsmod | egrep iscsi
# ps -ef | egrep 'iscsi|scsi_eh' | grep -v grep
root 388 1 0 00:42 ? 00:00:00 [scsi_eh_0]
root 389 1 0 00:42 ? 00:00:00 [scsi_eh_1]
root 422 1 0 00:42 ? 00:00:00 [scsi_eh_2]
root 427 1 0 00:42 ? 00:00:00 [scsi_eh_3]
root 3767 1 0 00:45 ? 00:00:00 [scsi_eh_4]

[Nobody/Anonymous]

Logged In: NO

this was a bug in the old driver and/or old kernel.

[maximkoz]

Logged In: YES
user_id=1392206

Thank you.
Could you point me, please, to specific bug corrections?
The driver used is 4.0.2 - the latest production one for
Linux 2.6.
And the kernel 2.6.12.5 is not so outdated...

Thanks,

Maxim.

[Nobody/Anonymous]

Logged In: NO

Thank you.
Could you point me, please, to specific bug corrections?
The driver used is 4.0.2 - the latest production one for
Linux 2.6.
And the kernel 2.6.12.5 is not so outdated...

Thanks,

Maxim.

[Nobody/Anonymous]

Logged In: NO

Thank you.
Could you point me, please, to specific bug corrections?
The driver used is 4.0.2 - the latest production one for
Linux 2.6.
And the kernel 2.6.12.5 is not so outdated...

Thanks,

Maxim.