From owner-freebsd-isp Thu Mar 28 12:47:14 1996 Return-Path: owner-isp Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id MAA02608 for isp-outgoing; Thu, 28 Mar 1996 12:47:14 -0800 (PST) Received: from GndRsh.aac.dev.com (GndRsh.aac.dev.com [198.145.92.241]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id MAA02599 for ; Thu, 28 Mar 1996 12:47:08 -0800 (PST) Received: (from rgrimes@localhost) by GndRsh.aac.dev.com (8.6.12/8.6.12) id MAA10676; Thu, 28 Mar 1996 12:46:48 -0800 From: "Rodney W. Grimes" Message-Id: <199603282046.MAA10676@GndRsh.aac.dev.com> Subject: Re: BOCA 8 port -> uptime To: batie@agora.rdrop.com (Alan Batie) Date: Thu, 28 Mar 1996 12:46:47 -0800 (PST) Cc: phang@cpm.com.my, freebsd-isp@FreeBSD.ORG In-Reply-To: from Alan Batie at "Mar 28, 96 12:31:15 pm" X-Mailer: ELM [version 2.4ME+ PL11 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-isp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > Well, if this is agora, thats understandable, it is also a general purpose > > time sharring system with users doing all sorts of things on it.... > > I would expect interrupt handling to be independent of user load, except to > the extent they are doing things to cause other interrupts. And in all likely hood it is those ``other interrupts'' that have caused enough latency for you to see the occasional Silo overflow. I am pretty familiar with your hardware and am a bit surprized that you are getting them at all :-(. You system is all bus mastering low latency devices (I just checked an lsdev output, you do have a _lot_ of sio devices, that may cause some problems:). sio29 Idle Serial port: National 16550A or compatible sio30 Busy Serial port: National 16550A or compatible sio31 Idle Serial port: National 16550A or compatible sio48 Idle Serial port: National 16550A or compatible sio49 Idle Serial port: National 16550A or compatible I also counted 21 busy sio ports.... a bit beyond the 16 I sighted ... Probably a reduction of the FIFO threshold from 14 to 8 would totally eliminate your FIFO overflow problem, afterall it does need to check 49 ISA I/O registers per sio interrupt which takes ~49uS to do. Also I am not sure, but I think, the BB2016 can live with an ISA I/O recovery time of 1 BClk at 8Mhz, you may want to check your BIOS settings. > > agora:rgrimes {141} uptime > > 10:53AM up 30 days, 3:06, 34 users, load averages: 0.24, 0.21, 0.18 > > agora:rgrimes {142} > > I should mention also that in over 10 years of running agora on Xenix-286, > SVR3, SVR4, FreeBSD 1.1.5 and now FreeBSD 2.1, this is one of a handful > of instances where I've seen this much uptime. :-), I was even rather shocked when I saw the 30 day uptime!! I know that FreeBSD had not fared so well on your system in the past, but it looks like it has finally reached the ``stabity'' stage. -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Reliable computers for FreeBSD