From owner-freebsd-current Tue Sep 5 03:05:10 1995 Return-Path: current-owner Received: (from majordom@localhost) by freefall.freebsd.org (8.6.11/8.6.6) id DAA18688 for current-outgoing; Tue, 5 Sep 1995 03:05:10 -0700 Received: from haywire.DIALix.COM (haywire.DIALix.COM [192.203.228.65]) by freefall.freebsd.org (8.6.11/8.6.6) with ESMTP id DAA18667 for ; Tue, 5 Sep 1995 03:04:31 -0700 Received: (from news@localhost) by haywire.DIALix.COM (sendmail) id SAA02589 for freebsd-current@freebsd.org; Tue, 5 Sep 1995 18:04:21 +0800 (WST) Received: from GATEWAY by haywire.DIALix.COM with netnews for freebsd-current@freebsd.org (problems to: usenet@haywire.dialix.com) To: freebsd-current@freebsd.org Date: 5 Sep 1995 18:04:16 +0800 From: peter@haywire.dialix.com (Peter Wemm) Message-ID: <42h7b0$2gp$1@haywire.DIALix.COM> Organization: DIALix Services, Perth, Australia. References: <199509050557.FAA07885@netfl15a.devetir.qld.gov.au> Subject: Re: Those VM problems are *not* figments of my imagination Sender: current-owner@freebsd.org Precedence: bulk sysseh@devetir.qld.gov.au (Stephen Hocking) writes: >It's occuring on both my current machines, one a 486 dx2-66 with 16Mb, the >other a 25MHz 386sx with 8Mb. Each of these has a fairly large amount of swap >used (about 37%, or 18Mb of 50Mb). Most programs that die (and this is rather >intermittent behaiviour) will go with a sig 10, or very rarely, a sig 8. I've >had ftpd and login dip out on me, as well as xterm, xclock, twm and others. >These machines are the result of a make world then config foo. The only thing >I can think of that caused them to break is the VM stuff, as there's been >nothing else major. Not having expended the necessary BS&T (blood sweat & >tears) needed to under stand the VM stuff, I can't offer a useful suggestion >as to what the problem might be wthin the code. > Stephen I have not seen that, but I had a nasty experience last night that might be related.. /bin/sh, /bin/rm and /usr/sbin/sendmail all took a dive, causing them all to dump core with sig 10 and 11. I dont know if any other files are affected.... :-( I rebuilt them and redid a make install, and it cured it. I dont know if it's related to the VM, it might have been related to the hardware hacking we did last night, where we were trying to find out how a soft-configured NE2000 card was set up (without the config software). It turns out it was on IRQ12, conflicting with the scsi controller. I'm not certain that stray irq's would upset the scsi driver, as I imagine it would check the status registers etc. The only thing I wonder is if the scsi controller chip was doing an unbuffered drive of the bus IRQ line, and the logic "fight" was causing the sequencer to malfunction... Oh well.. -Peter