From owner-freebsd-current  Sat Aug 12 19:11:45 1995
Return-Path: current-owner
Received: (from majordom@localhost)
          by freefall.FreeBSD.org (8.6.11/8.6.6) id TAA00327
          for current-outgoing; Sat, 12 Aug 1995 19:11:45 -0700
Received: from big-brother.csrv.uidaho.edu (big-brother.csrv.uidaho.edu [129.101.114.110])
          by freefall.FreeBSD.org (8.6.11/8.6.6) with ESMTP id TAA00321
          for <current@freebsd.org>; Sat, 12 Aug 1995 19:11:41 -0700
Received: from big-brother.csrv.uidaho.edu (localhost [127.0.0.1]) by big-brother.csrv.uidaho.edu (8.6.11/8.6.9) with ESMTP id TAA00198 for <current@freebsd.org>; Sat, 12 Aug 1995 19:11:35 -0700
Message-Id: <199508130211.TAA00198@big-brother.csrv.uidaho.edu>
To: current@freebsd.org
Subject: procfs problems in -current?
Date: Sat, 12 Aug 1995 19:11:34 -0700
From: faried nawaz <fn@pain.csrv.uidaho.edu>
Sender: current-owner@freebsd.org
Precedence: bulk


hi,

on friday (yesterday) i sup'd -current and rebuilt the affected pieces.
i redid the lkm's and the kernel as well.

on reboot, the system would fsck the local ufs filesystems and hang.
i rebooted single-user, fsck'd them manually, mounted them, hit ^D
and got a panic:

(this is from what i wrote down...dumpon wasn't reached)

fault virtual address     = 0x1014
fault code                = supervisor read, page not present
instruction pointer       = 0x8: 0xf0125948
code segment              = base 0x0, limit 0xfffff, type 0x1b
                          = DPL 0, pres 1, def32 1, gran 1
processor eflags          = interrupt enabled, resume, OPL = 0
current process           = 4 (update)
panic: page fault

nm /kernel | grep f01259 gives

f01259b0 T _mount
f0125930 T _vfs_msync
f01259b0 F vfs_syscalls.o

i looked through the cvs mail and couldn't see a vfs-related commit, but
i did see a procfs one...and i do use the procfs lkm.  i commented procfs
out of my /etc/fstab, and rebooted.  the system came up fine.  i tried

mount -v -t procfs proc /proc

and got a panic:

fault virtual address     = 0x1028
fault code                = supervisor read, page not present
instruction pointer       = 0x8: 0xf0124487
code segment              = base 0x0, limit 0xfffff, type 0x1b
                          = DPL 0, pres 1, def32 1, gran 1
processor eflags          = interrupt enabled, resume, OPL = 0
current process           = 161 (mount)
panic: page fault

it wasn't able to sync to disk, so no core dump.

nm /kernel |grep f01244 gives

f0124448 T _insmntque
f01244b4 T _vwakeup

i'm running w/o procfs right now.  mount says

/dev/wd0a on / (local)
/dev/wd0s1e on /var (local)
/dev/wd0s1f on /usr (local)
fs on n

the last line is interesting...it should say `kernfs' (i modload that too).
i don't know why it's corrupt.


i redid the kernel with `options PROCFS', uncommented procfs in /etc/fstab
and rebooted.  it worked long enough for me to log in and type `mount', but
hung after that.  i rebooted again, and this time it didn't get past the
fsck's.

my /etc/fstab is

/dev/wd0a			/		ufs	rw 1 1
/dev/wd0s1e			/var		ufs	rw 1 1
/dev/wd0s1b			none		swap	sw 0 0
/dev/wd0s1f			/usr		ufs	rw 1 1
#proc				/proc		procfs	rw 0 0
kern				/kern		kernfs	rw 0 0
#pain:/var/mail			/var/mail	nfs	rw,soft,intr,-r=1024,-w=1024,-P 0 0

i'm now going to reboot and use the older kernel (aug 6th or so).


ddb time for me?


faried.