NetBSD Problem Report #32260

From www@netbsd.org  Tue Dec  6 08:54:21 2005
Return-Path: <www@netbsd.org>
Received: by narn.netbsd.org (Postfix, from userid 31301)
	id 332A763B884; Tue,  6 Dec 2005 08:54:21 +0000 (UTC)
Message-Id: <20051206085421.332A763B884@narn.netbsd.org>
Date: Tue,  6 Dec 2005 08:54:21 +0000 (UTC)
From: rjgrace@fastmail.fm
Reply-To: rjgrace@fastmail.fm
To: gnats-bugs@netbsd.org
Subject: kernel panic when mounting nfs
X-Send-Pr-Version: www-1.0

>Number:         32260
>Category:       kern
>Synopsis:       kernel panic when mounting nfs
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Dec 06 08:55:00 +0000 2005
>Last-Modified:  Sat Feb 10 11:53:43 +0000 2007
>Originator:     Richard Grace
>Release:        3.0_RC5
>Organization:
>Environment:
NetBSD Cubit 3.0_RC5 NetBSD 3.0_RC5 (GENERIC) #8: Sat Dec  3 00:02:47 EST 2005  root@Cubit:/usr/obj/sys/arch/i386/compile/GENERIC i386

>Description:
A kernel page fault occurs every time amd tries to mount a non-existent directory.

On rebooting, the lockmgr also panics the kernel.

This is strange, as I have type=linx defined, so there should be no call to nfs mount anyway.  The problem also manifests itself on any other non-existent directory within any other amd map - eg, /cdrom/nonexistent and /home/nonexistent

>How-To-Repeat:
Here it is, step by step, along with config files below:

Cubit# cd /home/crashme
uvm_fault(0xc0824d60, 0xfffff000, 0, 1) -> 0xe
kernel: page fault trap, code=0
Stopped in pid 947.1 (ksh) at   netbsd:memmove+0x6a:    movb    0x2(%esi),%al
db> trace
memmove(cb55acb4,cb55acb8,44,3548fbfb,cb55abfc) at netbsd:memmove+0x6a
nfsm_loadattrcache(cb55aca8,cb55acb4,cb55acb8,0,0) at netbsd:nfsm_loadattrcache+0x44
nfs_lookup(cb55add4,cb55af08,cb55adec,c03be93a,c0611340) at netbsd:nfs_lookup+0x525
VOP_LOOKUP(cbbe4498,cb55aef4,cb55af08,20,0) at netbsd:VOP_LOOKUP+0x2e
lookup(cb55aee4,cab70400,400,cb55aefc,0) at netbsd:lookup+0x1fd
namei(cb55aee4,cbba6480,cbba6490,cbba64a0,cbba64b0) at netbsd:namei+0xfc
change_dir(cb55aee4,cab6cb2c,1,cab6cb2c,807e808) at netbsd:change_dir+0x15
sys_chdir(cab6b8c4,cb55af64,cb55af5c,0,c128c800) at netbsd:sys_chdir+0x42
syscall_plain() at netbsd:syscall_plain+0x7e
--- syscall (number 12) ---
0xbdb77fcb:
db> reboot
syncing disks... 12 12 12 12 12 12 done
unmounting file systems...panic: lockmgr: locking against myself
Stopped in pid 792.1 (ksh) at   netbsd:cpu_Debugger+0x4:        leave
db> trace 
cpu_Debugger(c0611740,cbdc0400,ffffffff,10002,0) at netbsd:cpu_Debugger+0x4
panic(c07784a0,0,1b,0,1) at netbsd:panic+0x11d
lockmgr(cbbe45b0,10002,cbbe4540,cbdc0400,c0611a40) at netbsd:lockmgr+0x189
genfs_lock(cb55a850,0,cb55a898,c03bfe1c,c0611a80) at netbsd:genfs_lock+0x1b
VOP_LOCK(cbbe4540,10002,0,cbbe4540,cbbe4540) at netbsd:VOP_LOCK+0x28
vn_lock(cbbe4540,30002,cb55a8ac,c03c0287,1) at netbsd:vn_lock+0x6e
vget(cbbe4540,20002,cb55a8d8,246,c083fc60) at netbsd:vget+0x93
nfs_unmount(c1334000,80000,cab6cb2c,0,1) at netbsd:nfs_unmount+0x3f
dounmount(c1334000,80000,cab6cb2c,0,cb55a944) at netbsd:dounmount+0xb0
vfs_unmountall(cab6cb2c,0,0,0,0) at netbsd:vfs_unmountall+0xb1
vfs_shutdown(0,0,cb55a988,c0350c95,0) at netbsd:vfs_shutdown+0xa6
cpu_reboot(0,0,cb55a988,c0350c6b,cb55a9a8) at netbsd:cpu_reboot+0x103
db_reboot_cmd(1,0,c0270cf1,cb55a9b0,0) at netbsd:db_reboot_cmd+0x4d
db_command(c07beaa4,c060dfe0,cb55aaa8,c03504bd,c0720937) at netbsd:db_command+0xe3
db_command_loop(c02fefa7,318,1,cab6cc85,0) at netbsd:db_command_loop+0x8e
db_trap(6,0,cb55ab08,c0379323,c1136000) at netbsd:db_trap+0xd4
kdb_trap(6,0,cb55abf4,0,286) at netbsd:kdb_trap+0x132
trap() at netbsd:trap+0xc3
--- trap (number 6) ---
nfs_lookup(cb55add4,cb55af08,cb55adec,c03be93a,c0611340) at netbsd:nfs_lookup+0x4cb
VOP_LOOKUP(cbbe4540,cb55aef4,cb55af08,20,2) at netbsd:VOP_LOOKUP+0x2e
lookup(cb55aee4,cab70400,400,cb55aefc,cbe1d324) at netbsd:lookup+0x1fd
namei(cb55aee4,3,8b00,200005cb,7fffff04) at netbsd:namei+0xfc
change_dir(cb55aee4,cab6cb2c,1,cab6cb2c,807e808) at netbsd:change_dir+0x15
sys_chdir(cab6b8c4,cb55af64,cb55af5c,0,cb55afa4) at netbsd:sys_chdir+0x42
syscall_plain() at netbsd:syscall_plain+0x7e
--- syscall (number 12) ---
0xbdb77fcb:
db> reboot
rebooting...


--
Cubit# cat /etc/amd.conf
[ global ]
auto_dir =              /.amd
dismount_interval =     900
map_type =              file
restart_mounts =        yes
search_path =           /etc/amd
log_file =              syslog
log_options =           error,noinfo,nostats

[ /home ]
map_name =              home

[ /net ]
map_name =              net

[ /cdrom ]
map_name =              cdrom
browsable_dirs =        yes

Cubit# cat /etc/amd/master
/cdrom          /etc/amd/cdrom
/home           /etc/amd/home
/net            /etc/amd/net

Cubit# cat /etc/amd/home
/defaults       opts:=soft,intr
*               rhost:=cubit;type:=linkx;fs:=/export/home/${key}

Cubit# ps -axw
PID TTY   STAT    TIME COMMAND
  0 ?     DKs  0:00.00 [swapper]
  1 ?     Is   0:00.01 init 
  2 ?     DK   0:00.00 [cryptoret]
  3 ?     DK   0:00.00 [usb0]
  4 ?     DK   0:00.00 [usbtask]
  5 ?     DK   0:00.00 [usb1]
  6 ?     DK   0:00.00 [usb2]
  7 ?     DK   0:00.00 [usb3]
  8 ?     DK   0:00.00 [atabus0]
  9 ?     DK   0:00.00 [atabus1]
 10 ?     DK   0:00.00 [cardslot0]
 11 ?     DK   0:00.00 [cardslot1]
 12 ?     DK   0:00.00 [pms0]
 13 ?     DK   0:00.00 [atapibus0]
 14 ?     DK   0:00.00 [pagedaemon]
 15 ?     DK   0:00.01 [ioflush]
 16 ?     DK   0:00.00 [aiodoned]
 96 ?     IL   0:00.00 nfsd: server 
177 ?     Is   0:00.19 /usr/sbin/sshd 
283 ?     Ss   0:00.01 /usr/sbin/rpcbind -l 
292 ?     S    0:00.10 /usr/sbin/sendmail -FCronDaemon -odi -oem -oi -or0s -t 
338 ?     Ss   0:00.01 /usr/sbin/syslogd 
360 ?     S    0:00.08 /usr/sbin/sendmail -FCronDaemon -odi -oem -oi -or0s -t 
406 ?     IK   0:00.00 [nfsio]
436 ?     Is   0:00.10 /usr/sbin/amd -p -a /.amd -F /etc/amd.conf 
474 ?     IK   0:00.00 [nfsio]
501 ?     IK   0:00.00 [nfsio]
505 ?     IK   0:00.00 [nfsio]
575 ?     Ss   0:00.01 /usr/sbin/mountd 
604 ?     IL   0:00.00 nfsd: server 
631 ?     Is   0:00.02 nfsd: master 
632 ?     IL   0:00.00 nfsd: server 
659 ?     IL   0:00.00 nfsd: server 
694 ?     I    0:00.00 cron: running job 
702 ?     Ss   0:00.04 /usr/sbin/ntpd 
725 ?     I    0:00.00 cron: running job 
856 ?     Is   0:00.00 /usr/sbin/cron 
367 tty00 R+   0:00.00 ps -axw 
855 tty00 Is   0:00.35 login 
947 tty00 S    0:00.02 -ksh 

Cubit# ls /export/home
books               emul                netlogon            www.archive.tar.gz
cap                 ftp                 rich
carl                kitiara             webmin
cricket             linda               www

--
>Fix:
No known fix, though the problem has existed since upgrading from NetBSD 2.1 release.

>Release-Note:

>Audit-Trail:

From: Richard Grace <rjgrace@fastmail.fm>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: kern/32260: kernel page fault trap when mounting nfs
Date: Tue, 13 Dec 2005 22:54:51 +1100

 I have a workaround to my own bug.

 At the suggestion to re-write my amd map file, the following change 
 prevents the panic:

    - /defaults       opts:=soft,intr
    - *               rhost:=cubit;type:=linkx;fs:=/export/home/${key}

    + /defaults       opts:=soft,intr
    + *               rhost:=cubit;type:=linkx;fs:=/export/home;sublink:=${key}

 Now the error "No such file or directory" is returned as expected.

 I am still concerned about the ability of amd to panic the kernel so 
 easily.

 Richard.

From: Richard Grace <rjgrace@fastmail.fm>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: kern/32260: kernel page fault trap when mounting nfs
Date: Tue, 13 Dec 2005 23:00:27 +1100

 I have also just tried this with 3.0-RC6 and found that this panic  
 still occurs.

 # uname -a
 NetBSD Cubit 3.0_RC6 NetBSD 3.0_RC6 (GENERIC) #0: Fri Dec  9 09:39:00 UTC 2005   
 riz@faith.netbsd.org:/home/builds/ab/netbsd-3-0-RC6/i386/200512090458Z-obj/home/builds/ab/netbsd-3-0-RC6/src/sys/arch/i386/compile/GENERIC i386

 I can assist debugging this - I have a serial console configured and  
 can update and re-compile debug kernels if required.

 Rich.

>Unformatted:

Home	PR Database Search