Date: Mon, 22 Jan 1996 01:41:09 +1100 (EST) From: michael butler <imb@scgt.oz.au> To: tom@uniserve.com (Tom Samplonius) Cc: rlenk@widget.xmission.com, stable@FreeBSD.ORG Subject: Re: spontaneous reboot with -stable Message-ID: <199601211441.BAA14306@asstdc.scgt.oz.au> In-Reply-To: <Pine.BSF.3.91.960120093651.1330E-100000@haven.uniserve.com> from "Tom Samplonius" at Jan 20, 96 09:39:45 am
next in thread | previous in thread | raw e-mail | index | archive | help
I wrote: > I finally caught the error message from my 2842-fitted -stable machine that > gives me so much trouble (reboots more than once a day :-(). It says .. > panic: getnewbuf: inconsistent LRU queue, qindex=0 > syncing ... Ron Lenk wrote: > > Yeah, don't run -stable. :-D > > I started running -stable again after Justin Gibbs commited the new ahc > > driver, and I have been unable to get a kernel to run for more than 5 > > minutes. Tom Samplonius writes: > Not if you have a 2940. The new ahc driver is really, really nice in > comparison to the old one. I'm now seeing record uptime's on a -stable > system. I certainly agree that, when the machine is running with the new driver, particularly with tagged-queueing, it *screams* :-) but it only lasts a few hours for me :-(. Having reverted to a -release kernel, the same machine, in the same configuration has not failed once. However, with "-stable", it reboots within about 4 hours. It was doing the same things as it's always done, a *very* light load with Apache and sup file-serving. It's a UMC491C motherboard with an Intel 486DX2/66, 16 meg of RAM, IBM 1 gig SCSI-II attached to a recent Adaptec 2842. It has no serial or parallel ports. Up until a minute ago I had a WD8003-EBT as -stable had also complained of too many stray irq 7s with the 3C509 installed. I had temporarily replaced it thinking that this may have been related to the reboot problem. It did not and, I should add, the -release kernel also complains of too many stray irq 7s when I reinstalled the 3C509. This problem is not specific to either -release or -stable but simply to the presence/activation of the ep0 driver. This is stupid, the 3C509's on irq 10! Selecting "-1" in the boot-time config menu correctly reads irq 10 from the card's eeprom and then proceeds to complain about stray ints :-(. This latter problem, however, has nothing at all to do with the machine's (in)stability which, at this stage, I can only conclude is related to some change between -release and -stable and is possibly the latest ahc driver. I've tried disabling either or both caches. Enabling and disabling tagged-queueing and the full queueing extension slowing the RAM timing down and speeding it up .. nothing I do to the machine with -stable running will make it "live" more than a few hours, michael
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199601211441.BAA14306>