Date: Thu, 30 Jul 1998 11:30:29 -0700 (PDT) From: Matt Dillon <dillon@best.net> To: "Bruce M. Walter" <walter@fortean.com> Cc: freebsd-bugs@FreeBSD.ORG, dima@best.net, newmike@teleport.com Subject: Re: kern/6102 Message-ID: <199807301830.LAA24610@flea.best.net>
next in thread | raw e-mail | index | archive | help
Hi Bruce. The level of interest is high right now. Tor, dima, and I
are trying to track down the bug and dima and I are about to reboot
a few dozen machines with new debugging kernels in an attempt to narrow
down where the problem lies.
There are potentially multiple causes to this problem. The most likely
cause is the one that Julian found... a hardware FIFO overflow on adaptec
boards. I dunno re: DPT boards. However, the HWb ug supposedly only
occurs if SCB paging is turned on, and we have it turned off and still
getting the panics so we are persuing some debugful kernels in an attempt
to track the problem down.
Note that, as per Julian, this problem does not appear to occur on
CAM-based systems. Right now we aren't running CAM... it's too big a
change for me to put into production just yet, and I'd really rather CAM
were incorporated into the main CVS tree rather then kept as a separate
tree.
The biodone panic occurs about once a month or so on our boxes, but since
we have a lot of boxes we tend to see a couple of crashes a week. The
hope is that with enough machines running the debug kernel, we will be
able to track the problem down relatively quickly.
-Matt
:
:Hi all,
:
:I wanted to check the level of interest regarding FreeBSD PR kern/6102...
:I wasn't actually aware that this could be the problem I was experiencing
:until early this morning.
:
:First off, for reference: (a full dmesg and kernel config are at the end)
:
:nemesis(walter) 2% uname -a
:FreeBSD nemesis.fortean.com 2.2.6-RELEASE FreeBSD 2.2.6-RELEASE #0:
:Thu Jun 4 18:04:02 EDT 1998 root@nemesis.fortean.com:/usr/src/sys/compile/NEMESIS i386
:
:This machine is a PPro-200 with a DPT 3334-UW SmartRaid Controller. After
:I added a Sony DDS2 DAT backup which runs a dump every night at 4am I
:started experiencing random reboots about every other week or so. These
:resets always occurred during the heavy SCSI IO related to the backup.
:Since the system would just reset, I had nothing to go on until this
:morning when the system failed to sync the disks and the dreaded:
:
:panic: biodone: buffer not busy
:
:messages was on the console. After scanning through the mailing lists and
:GNATS db, I found PR#6102 regarding this (and PR#5117 which is apparently
:the same prob for 2.2.5) I noticed the problem is not assigned (not
:surprising, people are busy) but wanted to find out if anyone is working
:on it and provide them with my datapoints.
:
:I did notice the problem seems to not be directly related to the DPT
:insomuch as it has popped up on other SCSI controllers as well. PR#7424
:may also be related, although it seems to deal specifically with the
:adaptec controller.
:
:There have also been reports that switching to CAM fixes the problem
:(which is not an option for me until 3.0 becomes 'stable' :)
:
:This is a fairly serious problem for me, so if there is anything I can do
:to speed it's resolution, please let me know.
:
:- Bruce
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199807301830.LAA24610>