From owner-freebsd-stable Mon Mar 25 19:43:16 1996 Return-Path: owner-stable Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id TAA06954 for stable-outgoing; Mon, 25 Mar 1996 19:43:16 -0800 (PST) Received: from asstdc.scgt.oz.au (root@asstdc.scgt.oz.au [202.14.234.65]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id TAA06943 for ; Mon, 25 Mar 1996 19:43:04 -0800 (PST) Received: (from imb@localhost) by asstdc.scgt.oz.au (8.6.12/BSD4.4) id OAA02953; Tue, 26 Mar 1996 14:36:45 +1100 From: michael butler Message-Id: <199603260336.OAA02953@asstdc.scgt.oz.au> Subject: Re: Stable not very stable To: jkh@time.cdrom.com (Jordan K. Hubbard) Date: Tue, 26 Mar 1996 14:36:45 +1100 (EST) Cc: scrappy@ki.net, stable@FreeBSD.org In-Reply-To: <29658.827806897@time.cdrom.com> from "Jordan K. Hubbard" at Mar 25, 96 06:21:37 pm X-Mailer: ELM [version 2.4 PL24beta] Content-Type: text Sender: owner-stable@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Jordan K. Hubbard writes: > > Essentially, the system does a bunch of disk thrashing (I believe > > its thrashing while it unbatches news, but am not 100% certain that that > > is what it is doing), then it grinds to a complete halt. > > I can change VT's, but that is about it. > FWIW, that failure mode has been reported by others. Any chance of > swapping this motherboard for something other than a DX4, just to narrow > this a little? The failures seem to happen about 500% more often with the > DX4 parts, though I'm not sure why. Now that the EISA/VESA stuff is fixed in the ahc driver, I'm using -stable on a write-through L1 cache (i.e. non-enhanced) AMD DX4/100 with an Adaptec 2842 and 64 meg of RAM with no trouble at all since I applied two (and only two) patches .. i) The cd9660_readdir() patch posted by Bruce Evans on March 8th and .. ii) The ufs_vnops() patch (aka "free vnode isn't" patch) posted by David Greenman on the same day. I realise that at least one of these was designed to be against -current but they also happen to apply cleanly to -stable. With ~100k articles a day from the limited newsfeed I receive for the last 10 days, I think that counts as "working" :-) The motherboard is a Chinese "no-name" flavour. The kernel's compiled with AHC_TAGENABLE and QUEUE_FULL_SUPPORTED, NFS_ASYNC, NMBCLUSTERS=1024, CHILD_MAX=256, and OPEN_MAX=256 and runs with the news spool, overview and active file-systems mounted "async". I suspect the problem is either cache/motherboard related or disk driver related (since the complaint is against an NCR-based system). The CPU itself, in the right motherboard, works great. Occasionally, I've added Bruce's vfs_lookup/lstat patch (Nov 11th, '95) to see if it makes disk access any faster .. it appears to do so, but I haven't tested it quantitatively (not enough symlinks now that I have new drives), michael