[OpenAFS] 1.4.4 client on EL3: crash upon restart [was: panic
inafs_HashOutDcache]
Stephan Wiesand
Stephan.Wiesand@desy.de
Thu, 19 Apr 2007 10:14:53 +0200 (CEST)
On Wed, 18 Apr 2007, Derrick J Brashear wrote:
> On Wed, 18 Apr 2007, Stephan Wiesand wrote:
>> When I tried to restart the client, I experienced what I've seen frequently
>> with 1.4.x clients on this platform: "kernel BUG at slab.c:892:" when
>> re-inserting the openafs module. This seems to happen quite consistently
>> when restarting the client after it has run for some time (say, a week).
>
> Was the module actually unloaded from the previous attempt?
Yes, I just checked on a different system:
# dmesg|tail
kjournald starting. Commit interval 5 seconds
EXT3 FS 2.4-0.9.19, 19 August 2002 on loop(7,0), internal journal
EXT3-fs: mounted filesystem with ordered data mode.
device eth0 entered promiscuous mode
device eth0 left promiscuous mode
afs: byte-range lock/unlock ignored; make sure no one else is running this program.
lp: driver loaded but no devices found
usb.c: registered new driver serial
usbserial.c: USB Serial support registered for Generic
usbserial.c: USB Serial Driver core v1.4
# service afs stop
Stopping AFS services.....
Sending all processes using /afs the TERM signal ... [ OK ]
Sending all processes using /afs the KILL signal ... [ OK ]
# lsmod|grep afs
#
# dmesg|tail
afs: byte-range lock/unlock ignored; make sure no one else is running this program.
lp: driver loaded but no devices found
usb.c: registered new driver serial
usbserial.c: USB Serial support registered for Generic
usbserial.c: USB Serial Driver core v1.4
Failed to invalidate all pages on inode 0xccfc7c80
Failed to invalidate all pages on inode 0xece91980
WARM shutting down of: CB... afs... BkG... CTrunc... AFSDB... RxEvent...
UnmaskRxkSignals... RxListener...
VFS: Busy inodes after unmount. Self-destruct in 5 seconds. Have a nice day...
kmem_cache_destroy: Can't free all objects c4dde914
# service afs start
Starting AFS services.....
Warning: loading /lib/modules/2.4.21-47.0.1.ELsmp/kernel/fs/openafs/openafs.o will taint
the kernel: non-GPL license - http://www.openafs.org/dl/license10.html
See http://www.tux.org/lkml/#export-tainted for information about tainted modules
And at this point it freezes. Here's what netconsole left on the loghost:
Found system call table at 0xc0386b74 (pattern scan)
------------[ cut here ]------------
kernel BUG at slab.c:828!
invalid operand: 0000
openafs usbserial lp parport sg panfs nfs lockd sunrpc netconsole e1000
microcode nls_iso8859-1 nls_cp437 vfat fat loop keybdev mousedev hid input
ehci-hcd us
CPU: 2
EIP: 0060:[<c015272b>] Tainted: P
EFLAGS: 00010246
EIP is at kmem_cache_create [kernel] 0x2fb (2.4.21-47.0.1.ELsmp/i686)
eax: 00000000 ebx: f7645afc ecx: c4dde980 edx: c4dde980
esi: c4dde97c edi: f8ab5760 ebp: 00000000 esp: c8a2dea0
ds: 0068 es: 0068 ss: 0068
Process modprobe (pid: 32474, stackpage=c8a2d000)
Stack: f7645afc 00000300 00012000 c8a2deb8 f7645b1c ffffff80 00000080 c0388f84
00000000 00000000 f8a43000 f8a9c460 f8ab5750 00000300 00000080 00002000
f8a9c420 00000000 c103c02c f8a9b327 f8ad515c ffffffff 00008be7 f8a9b337
Call Trace: [<f8a9c460>] afs_init_inodecache [openafs] 0x20 (0xc8a2decc)
[<f8ab5750>] .rodata.str1.1 [openafs] 0x1214 (0xc8a2ded0)
[<f8a9c420>] init_once [openafs] 0x0 (0xc8a2dee0)
[<f8a9b327>] init_module [openafs] 0x17 (0xc8a2deec)
[<f8ad515c>] afs_xosi [openafs] 0x0 (0xc8a2def0)
[<f8a9b337>] init_module [openafs] 0x27 (0xc8a2defc)
[<c012b096>] sys_init_module [kernel] 0x5b6 (0xc8a2df0c)
[<f8a43060>] afs_atomlist_create [openafs] 0x0 (0xc8a2df2c)
[<f8aba3e8>] __ksymtab [openafs] 0x0 (0xc8a2df30)
[<f8a43060>] afs_atomlist_create [openafs] 0x0 (0xc8a2df58)
[<c02b006f>] no_timing [kernel] 0x7 (0xc8a2dfc0)
Code: 0f 0b 3c 03 53 27 2c c0 8b 12 81 fa 6c b9 3a c0 75 d3 89 4b
Kernel panic: Fatal exception
Looks like things already go wrong during unload. Could the root
root cause be same as for the panic in afs_HashOutDcache ?
--
Stephan Wiesand
DESY - DV -
Platanenallee 6
15738 Zeuthen, Germany