Re: Re: problems with scsi_transport_fc and qla2xxx

["Mike Snitzer" <snitzer@gmail.com>]

On 3/2/06, Maxim Kozover <maximkoz@netvision.net.il> wrote:
> Hi Andrew!
> Today I tested disconnecting QLogic port.
> Adapter 4 is connected via switch to a storage and 3 LUNs are seen via
> the adapter.
> Only 1 rport is created (for FCP Target) while in Emulex case there
> were 3: (Fabric Port, Directory Server and FCP Target, FCP Initiator).
> # ls /sys/class/fc_remote_ports/
> rport-4:0-0
> # cat /sys/class/fc_remote_ports/*/roles
> FCP Target
>
> Default dev_loss_tmo is 6 (1+5) while in Emulex case the default was 35.
>
> After disconnecting the cable between the HBA and the switch
> qla2xxx 0000:03:01.0: LOOP DOWN detected (2).
>  rport-4:0-0: blocked FC remote port time out: removing target and saving binding
>
> # ls /sys/class/fc_remote_ports/
> rport-4:0-0
> # cat /sys/class/fc_remote_ports/*/roles
> unknown
>
> Relevant scsi devices are removed from /proc/scsi/scsi.
>
> After reconnecting the cable
> qla2xxx 0000:03:01.0: LIP reset occured (f7f7).
> qla2xxx 0000:03:01.0: LOOP UP detected (2 Gbps).
>
> # ls /sys/class/fc_remote_ports/
> rport-4:0-0
> # cat /sys/class/fc_remote_ports/*/roles
> FCP Target
>
> However, scsi devices don't reappear in /proc/scsi/scsi.
> When I issue rescan, the command is stuck
> echo - - - > /sys/class/scsi_host/host4/scan

Historically the qlogic driver rescan is a 2-phase process:
1) schedule the rescan, e.g.: echo scsi-qlascan > /proc/scsi/qla2xxx/4
2) rescan, e.g.: echo - - - > /sys/class/scsi_host/host4/scan

BUT, I've just used scsi-qlascan to discover _new_ devices... not
existing devices that experienced FC connection loss.  I assume the
qla driver _should_ just bring those lost devices back?  But does the
historic 2-phase rescan for new devices speak to why the qlogic driver
doesn't automagically bring the old devices back?  Or has the latest
qlogic driver in mainline advanced past this 2-phase requirement in
general?

regards,
Mike

Mike
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/