From owner-freebsd-hackers Tue Oct 28 10:05:19 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id KAA08550 for hackers-outgoing; Tue, 28 Oct 1997 10:05:19 -0800 (PST) (envelope-from owner-freebsd-hackers) Received: from smtp03.primenet.com (smtp03.primenet.com [206.165.5.84]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id KAA08540 for ; Tue, 28 Oct 1997 10:05:14 -0800 (PST) (envelope-from tlambert@usr06.primenet.com) Received: (from daemon@localhost) by smtp03.primenet.com (8.8.7/8.8.7) id LAA09151; Tue, 28 Oct 1997 11:04:55 -0700 (MST) Received: from usr06.primenet.com(206.165.6.206) via SMTP by smtp03.primenet.com, id smtpd009137; Tue Oct 28 11:04:52 1997 Received: (from tlambert@localhost) by usr06.primenet.com (8.8.5/8.8.5) id LAA29496; Tue, 28 Oct 1997 11:04:47 -0700 (MST) From: Terry Lambert Message-Id: <199710281804.LAA29496@usr06.primenet.com> Subject: Re: More on silo overflows... To: rivers@dignus.com (Thomas David Rivers) Date: Tue, 28 Oct 1997 18:04:47 +0000 (GMT) Cc: freebsd-hackers@freefall.FreeBSD.org In-Reply-To: <199710281114.GAA06159@lakes.dignus.com> from "Thomas David Rivers" at Oct 28, 97 06:14:11 am X-Mailer: ELM [version 2.4 PL23] Content-Type: text Sender: owner-freebsd-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > I think I've found the association that causes my silo overflows > at 38400 on a P75. > > It seems that everytime I use syscons' screen-flipping feature to > bounce to another virtual screen - I get a silo overflow. > > Perhaps this points the suspicious finger at syscons? Is it disabling > interrupts for some long period of time? Most likely, it is the time needed to reprogram the keyboard LED's that is screwing you. IMO, this does not need to happen at SPL with a buzz-loop delay; you should be able to get rid of one or the other (or both). The problem with getting rid of the buzz-loop delay is that it requires good timer code. Only NetBSD, OpenBSD, and Linux have implemented my timer suggestions. 8-|. You may want to see if the LED state is the same, and if so, if it will still cause the error (people generally work with one set of key states). It may be that you can cache state and toggle only things which have changed between sessions, if syscons does not do this already. You could try to trigger the error by starting an FTP and bouncing numlock or capslock like mad. If the error doesn't trigger, then a single LED isn't enough to blow it. You may also consider seperating the programming into three phases in which you alow interrupts between them. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.