From owner-freebsd-hackers Mon Jun 30 12:20:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA10601 for hackers-outgoing; Mon, 30 Jun 1997 12:20:36 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id MAA10594 for ; Mon, 30 Jun 1997 12:20:31 -0700 (PDT) Received: (qmail 20535 invoked by uid 1000); 30 Jun 1997 17:21:11 -0000 Message-ID: X-Mailer: XFMail 1.2-alpha [p0] on FreeBSD Content-Type: text/plain; charset=iso-8859-8 Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <199706291830.EAA08361@godzilla.zeta.org.au> Date: Mon, 30 Jun 1997 10:21:11 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: Bruce Evans Subject: Re: Clists limited to 1024 bytes? Cc: bmcgover@cisco.com, hackers@FreeBSD.ORG Sender: owner-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi Bruce Evans; On 29-Jun-97 you wrote: ... > Logging the error may delay completion for a long time. Disks shouldn't > be very sensitive to this, but... Try turning of log() and printf(). > I usually do this by temporarily replacing the first byte in the routine > by 0xc3 (ret). [ Backwards ] The `ret' trick is as old as I am. wonder how well it (C3) will work on an Alpha :-) BTW, this is an 8080 opcode, methinks.. I think the problem we suffer from is not necessarily due to delays but something else. What we see is either completions reported by calling scsi_done being lost, or something that scsi_done does when called getting lost. The symptom is; A calling user process (dd, cpio, compilation, etc.) simply blocks indefinitely in biowait. This is manifested by having a PPP session running. Load is not very important. Without PPP I can bring the system to LA of 120 or so, without problem. With PPP connected, all you have to do is receive email while running make world. Somewhere in the middle, things wil start dying. This is much mode so on a P6 than a P5. Simon