P1394 Mail Archive: RE: P1394> Revised PWG1394 Cmd Set

P1394 Mail Archive: RE: P1394> Revised PWG1394 Cmd Set

RE: P1394> Revised PWG1394 Cmd Set

Nagasaka Fumio (Nagasaka.Fumio@exc.epson.co.jp)
Thu, 6 Aug 1998 11:14:34 +0900

Fumio Nagasaka wrote:
> In my idea, a print spooler in the initiator shall detect time out.

OK, forget my brain damaged idea. But still we don’t have cool answers.
Any session layer beyond SBP-2 can detect this error. Then what shall we do?
Examining ORB_POINTR requires an initiator to do effort as same as
issuing a transport command. Doesn’t it?

--------------------------------------------
Fumio Nagasaka
Epson Software Development Laboratory Inc.
Tel +81 268 25 4111, Fax +81 268 25 4627
E-mail to <mailto:nagasaka.fumio@exc.epson.co.jp> nagasaka.fumio@exc.epson.
co.jp

-----Original Message-----
From: Greg Shue [SMTP:gregs@sdd.hp.com]
Sent: Thursday, August 06, 1998 6:19 AM
To: Nagasaka Fumio
Cc: p1394@pwg.org
Subject: Re: P1394> Revised PWG1394 Cmd Set

Fumio Nagasaka wrote:
> In my idea, a print spooler in the initiator shall detect time
out.

That's fine if the initiator's transport client is a print spooler,
but what about other higher layer apps such as:
a Status/Event Monitor?
a file transfer protocol?
a scanning app?

We're working on a generic protocol. I don't want to impose any
more requirements on the application than does TCP. TCP itself
takes care of timeout/lost link detection. I think this protocol
should also. Once detection by the initiator's protocol driver is
taken care of, then recovery is easy. Have we now ended up with
a case where the initiator's side needs to poll for FETCH_AGENT
status
or do we let the node hosting the target side generate a Bus Reset?

> SBP-2 E.4 >Any target resources allocated to the ORB should be
released by
> thetarget.
>
> ...so, ORB_POINTER is unusable to examine the last ORB which
> was consumed. Thus this initiator shall ask the target to report
the last
> ORB
> address utilizing higher level command sounds like
"TRANSPORT_I2T_ABORT".

I don't think so. The ORB_POINTER register can still point to
the offending ORB. The ORB itself can no longer be referenced
by the target. Since it is in the initiator's memory space, the
initiator can still examine it.

--
Greg Shue
Hewlett-Packard Company
All-in-One Division gregs@sdd.hp.com
----------------------------------------------------------------