Douglas Gilbert wrote:
> Hannes Reinecke wrote:
>> Hi all,
>>
>> I've hacked up scsi_debug to support REPORT TARGET PORT GROUPS.
>> In combination with vpd_use_hostno=0 it'll now support multipathing
>> nicely.
>> Doug, there is reference to port_a and port_b in the VPD page support.
>> However, port_b appearently is just a fake entry with no (virtual)
>> device backing.
>> We could eg use the 'channel' value to refer the relative port number,
>> so that host & channel act as a fan-out for multipathing.
>>
>> James, please apply.
> 
> Not quite yet ...
> 
> Hannes,
> Here is my counter patch for you to consider.
> I suspect my machine was running into stack
> problems so I moved some arrays onto the heap.
> 
Ah. Good. To tell the truth I was a bit concerned once I've discovered 
that all information pages are infact statically allocated ...

> You will see some other small changes if you
> compare the two patches. My patch is against
> lk 2.6.19-rc2
> 
Looks good. Only there was a small glitch in my implementation (0x08 
actually means that only the 'unavailable' status is supported :-( ), so 
I've re-rolled it and added support for you unconnected port_b, too.
So we now have two port groups per host, one with the actual device and 
the other one with the unconnected port_b's.

What do you think?

Cheers,

Hannes
-- 
Dr. Hannes Reinecke			hare@suse.de
SuSE Linux Products GmbH		S390 & zSeries
Maxfeldstraße 5				+49 911 74053 688
90409 Nürnberg				http://www.suse.de