David Campbell (campbell@gear.torque.net)
Sat, 19 Jun 1999 08:27:30 +0800
Subject: Re: [PARPORT] ECP mode for printers
To: Philip.Blundell@pobox.com (Philip Blundell)
Date sent: Fri, 18 Jun 1999 19:56:42 +0100 (BST)
Copies to: campbell@gear.torque.net, linux-parport@torque.net
From: Alan Cox <alan@lxorguk.ukuu.org.uk>
> > I thought Tim's latest code could cope with the FIFO needing to be drained,
> > actually. Yes, I knew the SCSI layer wasn't ideal, but I didn't realise it
> > would actually cause an oops if it was starved of the bus for too long.
>
> It wont. It will however eventually do aborts, resets and then give up
> on failed bus resets.
I stand humbly corrected (the abort/reset sequence is an area I want
to stay away from), although once a device has had a failed bus reset
it is as good as dead. The mid-level SCSI code ignores the driver and
if the driver is a module there is no way to unload it (from memory -
pending requests prevents dismounts).
Time to figure out how to increase the timeouts, preferably the
"correct" way rather than jumping on the SCSI request which appears
fraught with danger. That and no other SCSI driver I know of attempts
such black magic.
David Campbell
=======================================================
campbell@torque.net
"This is not an office, rather Hell with fluorescent lighting"
-- To unsubscribe, send mail to: linux-parport-request@torque.net --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2.0b3 on Fri 18 Jun 1999 - 20:35:15 EDT