[Stgt-devel] [Patch 2/8 ] Combine INQUIRY into a common spc_inquiry routine.

[Mark Harvey]

On 5/26/07, Pete Wyckoff <pw at osc.edu> wrote:
> markh794 at gmail.com wrote on Wed, 23 May 2007 13:01 +1000:
> > Moved scsi_sn, scsi_id and d_sense (now sense_format) into a common
> > struct with other INQUIRY related data.
> [..]
> > +static int osd_device_init(struct scsi_lu *lu)
> >  {
> > -     lu->d_sense = 1;
> > +
> > +     if (spc_device_init(lu))
> > +             return -ENOMEM;
> > +
> > +     memcpy(lu->attributes->ProductIdent, "OSD", 16);
> > +     lu->attributes->sense_format = 1;
> > +     lu->attributes->version_desc1 = 0x0340; /* OSD */
> > +     lu->attributes->version_desc2 = 0x0960; /* iSCSI */
> > +     lu->attributes->version_desc3 = 0x0300; /* SPC-3 */
> > +
> >       return 0;
> >  }
>
> Good stuff.  I like the common INQUIRY handling.  It was something
> Tomo asked for once I added yet another inquiry handler for OSD.
>
> Another tweak that will be needed is the ability to support more
> page codes.  OSD needs to provide 0xb0 and 0xb1.  This means
> modifying 0x0 (supported VPD pages) to add two more supported pages
> to the list.  And it needs very OSD-internal code to produce the
> right responses for 0xb0 and 0xb1.  So I've been holding off pushing
> that change because it is difficult to generalize well.  If you have
> ideas, they would be most welcome.
>
> By all means, this patch is a step in the right direction, though.
>
>                 -- Pete
>

Thanks for the vote of confidence.

As you may be aware, my main aim is to create a Virtual Library
(Tape/CD/MO etc.)

The SSC devices will also require extra VPD pages...

I have not had any ideas regarding how to support multiple VPD pages
which I feel is 'good'.

I was thinking yet another list, similar to the MODE list, but this
seems a bit of an over-kill for a few bytes of data / VPD.


The other concern I had regarding the SCSI Version, related to the
underlying transport. At the moment, I have hard-coded iSCSI version,
however it would be 'nice' to be able to insert which ever underlying
protocol was being used to connect to the target...

Can anybody suggest an easy way of detecting the underlying protocol
at runtime ?

Cheers
Mark