[9064] in linux-scsi channel archive
Re: DID_RESET in muli-initiator environment
daemon@ATHENA.MIT.EDU (Eric Youngdale)
Thu Jun 22 22:21:57 2000
Message-ID: <009d01bfdcb9$cdf15930$0f17a8c0@eric.home>
From: "Eric Youngdale" <eric@andante.org>
To: "Martin Peschke" <peschke@fh-brandenburg.de>,
<linux-scsi@vger.rutgers.edu>
Date: Thu, 22 Jun 2000 22:21:56 -0400
MIME-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
In general multi-initiator was never really considered - as a result
confusion is likely to result if you are not careful. I take that back a
little bit - late in 2.3 I added a scsi_report_bus_reset() function that
low-level drivers can call when they see a bus reset that they themselves
did not initiate. This essentially sets the flag for the device so that the
reset is expected.
I haven't really tested this code carefully - I believe that it works,
but this isn't the sort of thing I am set up to test very easily.
-Eric
----- Original Message -----
From: "Martin Peschke" <peschke@fh-brandenburg.de>
To: <linux-scsi@vger.rutgers.edu>
Sent: Monday, June 19, 2000 1:53 PM
Subject: DID_RESET in muli-initiator environment
> I have a question concerning linux/drivers/scsi/scsi_error.c:
>
> If a HBA driver detects a bus reset it will usually notice the
> ML by means of DID_RESET. If this reset is not expected by the ML
> the error handler is woken up. Since no valid sense data is provided
> several resets will be done on device, bus and finally HBA driver by
> the default error strategy handler scsi_unjam_host().
> But this reset is not expected to a second initiator and resetting
> is continued without end.
>
> Am I right? Is this an intended behavior?
> Would you suggest that a HBA device driver that has to cope
> with a multi-initiator environment implements its own
> eh_strategy_handler()?
>
> Thanks.
>
> Regards
> Martin Peschke
>
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.rutgers.edu
>
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
