[Richard Sharpe <rsharpe@xxxxxxxxxx>]

On Mon, 11 Mar 2002 ddutt@xxxxxxxxx wrote:

> Hi Richard,
> 
> On further investigating, I did find that there is something called
> variable-length CDBs which you're alluding to. I thought that you were
> interested in decoding OSD and since OSD has a special peripheral device type,
> I suggested my first alternative.

OK, I am interested in OSD, but for the moment it is Panasas 
implementation of OSD, which is, umm, interesting :-) However, I would 
like to ensure I can handle OSD in general.

In anycase, I thought that OSD commands were carried as variable length 
CDBs ...
 
> It is quite easy to add a decoder for the overall variable-length CDB. However
> to decode each of the service actions for such CDBs is another beast. I can
> send a patch to handle variable-length CDBs quite soon, in about an hour or
> so. I don't have any traces with variable-length CDB and so cannot test the
> code. 

No hurry said the walrus, 
  and they thanked him much for that.
 
I can code it up tomorrow if you do not get the time.

> Dinesh
> Richard Sharpe writes:
>  > On Mon, 11 Mar 2002 ddutt@xxxxxxxxx wrote:
>  > 
>  > > If you're attempting to decode OSD, the Inquiry response contains the
>  > > peripheral type as 0x11 (byte 0, bits 0-4). You can save this information in
>  > > the SCSI device type block and then decode later commands based on this.
>  > 
>  > Hi Dinesh ....
>  > 
>  > I would like to check my [somewhat limited] understanding of the SCSI spec 
>  > and SPC-3 vs SBC etc.
>  > 
>  > It seems to me that variable sized CDBs should have 0x7F as their OpCode 
>  > byte, and should be distinguished from each other through the Action 
>  > field. This field is a 16-bit field.
>  > 
>  > If so, I will probably make some changes to the SCSI dissector, and then 
>  > we can look to dissecting OSD.
>  > 
>  > Is this correct?
>  > 
>  > > Dinesh
>  > > Richard Sharpe writes:
>  > >  > On Tue, 12 Mar 2002, Richard Sharpe wrote:
>  > >  > 
>  > >  > > Hi,
>  > >  > > 
>  > >  > > The current iSCSI dissector calls the SCSI dissector, which only 
>  > >  > > understands SPC-2 SCSI commands, it seems.
>  > >  > > 
>  > >  > > I need to be able to dissect OSD commands, which are carried in a CDB with 
>  > >  > > an SPC-3 Variable Length CDB indicator (0x7F).
>  > >  > 
>  > >  > Hmmm, the problem I am experiencing is because Ethereal's SCSI dissector 
>  > >  > currently only understands SPC-2, SBC-2, and SSC-2. If the CDB Operation 
>  > >  > Code is not found in the SPC-2 command set, tben the dissector searches 
>  > >  > among the SBC-2 command set ... Which is where 0X7F is found :-(
>  > >  > 
>  > >  > So, I need some way to tell what type of device we are talking about. 
>  > >  > Better check what the iSCSI spec says. If it says SPC-3, then I can pass 
>  > >  > an arg to the SCSI dissector telling it what command set to use.
>  > >  > 
>  > >  > > What I am interested in knowing is: Is there an easy way to determine 
>  > >  > > whether or not a stream is using SPC-3 or what? Is it carried in the iSCSI 
>  > >  > > PDU or logon or something like that?
>  > >  > > 
>  > >  > > 
>  > >  > 
>  > >  > -- 
>  > >  > Regards
>  > >  > -----
>  > >  > Richard Sharpe, rsharpe@xxxxxxxxxx, rsharpe@xxxxxxxxx, 
>  > >  > sharpe@xxxxxxxxxxxx
>  > >  > 
>  > >  > 
>  > >  > _______________________________________________
>  > >  > Ethereal-dev mailing list
>  > >  > Ethereal-dev@xxxxxxxxxxxx
>  > >  > http://www.ethereal.com/mailman/listinfo/ethereal-dev
>  > >  > 
>  > > 
>  > > 
>  > 
>  > -- 
>  > Regards
>  > -----
>  > Richard Sharpe, rsharpe@xxxxxxxxxx, rsharpe@xxxxxxxxx, 
>  > sharpe@xxxxxxxxxxxx
>  > 
>  > 
> 
> 

-- 
Regards
-----
Richard Sharpe, rsharpe@xxxxxxxxxx, rsharpe@xxxxxxxxx, 
sharpe@xxxxxxxxxxxx