Date: Wed, 5 May 1999 12:42:38 -0400 (EDT) From: Mike Andrews <mandrews@termfrost.org> To: freebsd-stable@freebsd.org Subject: Re: FreeBSD 3.1 remote reboot exploit (fwd) Message-ID: <Pine.BSF.4.10.9905051225220.12268-100000@mindcrime.termfrost.org> In-Reply-To: <8298.925865418@zippy.cdrom.com>
next in thread | previous in thread | raw e-mail | index | archive | help
For what it's worth, I've never had a spontaneous reboot under 3.1-stable. But I do see occasional lockups of a sort that I never saw with 2.2.x. In my case, I *think* the culprit is CAM. When the machine dies, the kernel stays running... you can ping the machine, you can establish connections, but all the running processes hang so no data goes over the established connections. All disk IO stops completely, which explains the hung processes. When it locks up, I get no serial (or VGA) console messages at all, no error messages, and of course no panic since the kernel is still up... On a pretty heavily beat-up mail/web/DNS server at work, this was starting to happen several times a week. We switched from an ncr0 card (Symbios Logic 53c875) to an Adaptec 2940UW two months ago and it's only happened once since. (Uptime of 18 days right now.) The only things on the SCSI bus are an IBM DDRS-39130W and a Seagate ST32430N. Otherwise, it's a P2/233, Abit LX6, 192 meg of ECC memory, Intel Etherexpress, and some cheap Cirrus Logic video card. Another older system (P90, now a P120) has been running with a 2940 (not UW) for a long time and has NEVER died (that I can remember), either under 3.1 or 2.2.x. I've had lockups happen occasionally at home too, where there's still a 53c875, only in that case I get ncr0 timeout messages on the console. (I'll write them down next time.) That machine has an ATAPI disk in it and that disk appears to keep running when the SCSI disks die. These may be unrelated problems; at home I have a LOT of stuff on the SCSI bus and that could be causing the timeouts... but it's this behavior that got me thinking that maybe SCSI was the problem at work. I'm kind of at a loss as to how to provide any better information on the work machine's lockup behavior (given the lack of any error msgs), unless maybe I compiled the debugger into the kernel and forced it into the debugger when the disks died...? I've never done that before, so I'd need to know how to do it in a serial console setup... Mike Andrews (icq 6602506) -- VP & Sysadmin, Digital Crescent, Frankfort KY mandrews@dcr.net -=- http://www.termfrost.org --=-=-- "A computer without a Microsoft operating system is like a dog without a brick tied to its head." To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.10.9905051225220.12268-100000>