Date: Tue, 28 Nov 1995 09:05:23 +0800 From: Peter Wemm <peter@jhome.DIALix.COM> To: FreeBSD-gnats-submit@freebsd.org Subject: kern/840: Kernel page directory invalid Message-ID: <199511280105.JAA27254@jhome.DIALix.COM> Resent-Message-ID: <199511280110.RAA17202@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 840
>Category: kern
>Synopsis: Kernel page directory invalid
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: freebsd-bugs
>State: open
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Mon Nov 27 17:10:01 PST 1995
>Last-Modified:
>Originator: Peter Wemm
>Organization:
At this time of the day???
>Release: FreeBSD 2.2-CURRENT i386
>Environment:
Two machines have shown this problem:-
One of our news servers (average uptime about a week):
FreeBSD gecko2.DIALix.COM 2.2-CURRENT FreeBSD 2.2-CURRENT #2: Fri Oct 27 21:02:24 WST 1995 pwroot@gecko2.DIALix.com:/home/src/sys/compile/NEWS i386
And a small machine in my boss's office that he works on (and paniced while
talking to me on the phone when idle.. Apparently cron forked or something):
FreeBSD office.DIALix.oz.au 2.2-CURRENT FreeBSD 2.2-CURRENT #2: Mon Sep 25 02:58:48 WST 1995 pwroot@office.DIALix.oz.au:/usr/src/sys/compile/OFFICE i386
>Description:
The news server appeared to crash more often when innd was compiled to use
mmap() for the active file and the dbz history table indexes. This is only
subjective though.. It appeared to last about three days on average with mmap()
and appears to last about a week without.. But it did vary, and we just may
have had a bad run of the same problem with mmap() active.
We've had the news server silently reboot overnight without saving it's dmesg
buffer and this is the first time we've caught it.
gecko2 (the news server) locked up at "Syncing disks" this morning with:
Kernel page directory invalid! pa=0x989023 va=0xefbfe000
When this was read out to me on the phone, I remembered that the office machine
had crashed from the same problem.
dmesg for the news server (note the vnode_pager_putpages error at the end)
FreeBSD 2.2-CURRENT #2: Fri Oct 27 21:02:24 WST 1995
pwroot@gecko2.DIALix.com:/home/src/sys/compile/NEWS
CPU: i486 DX2 (486-class CPU)
Origin = "GenuineIntel" Id = 0x435 Stepping=5
Features=0x3<FPU,VME>
real memory = 33554432 (32768K bytes)
avail memory = 30748672 (30028K bytes)
Probing for devices on the ISA bus:
sc0 at 0x60-0x6f irq 1 on motherboard
sc0: VGA color <16 virtual consoles, flags=0x0>
ed0 at 0x300-0x31f irq 5 on isa
ed0: address 00:40:c7:11:09:1f, type NE2000 (16 bit)
sio0 not found at 0x3f8
sio1 not found at 0x2f8
lpt0 not found at 0xffffffff
si0 irq 12 maddr 0xe8000 msize 32768 on isa
si0: card: SIHOST2, ports: 8, modules: 1
fdc0 at 0x3f0-0x3f7 irq 6 drq 2 on isa
fdc0: NEC 72065B
fd0: 1.44MB 3.5in
wdc0 not found at 0x1f0
bt0 not found at 0x330
aha0 at 0x330-0x333 irq 11 drq 5 on isa
(aha0:0:0): "QUANTUM FIREBALL1080S 1Q09" type 0 fixed SCSI 2
sd0(aha0:0:0): Direct-Access 1042MB (2134305 512 byte sectors)
(aha0:1:0): "QUANTUM VP32210 581H" type 0 fixed SCSI 2
sd1(aha0:1:0): Direct-Access 2103MB (4308352 512 byte sectors)
npx0 on motherboard
npx0: INT 16 interface
new masks: bio c0000840, tty c0031002, net c0031022
WARNING: / was not properly dismounted.
vnode_pager_putpages: attempt to write meta-data!!! -- 0xfffe9000(ff)
I've not updated it's kernel for a while after hearing about problems others
are having with post-november-7 kernels, but it might be worth a try to see if
it throws any new light on it.
>How-To-Repeat:
Run a news server and wait...
>Fix:
If only I knew.. :-)
>Audit-Trail:
>Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199511280105.JAA27254>