[OpenAFS] heavy troubles with openafs 1.3.79 and kernel 2.6.11

Guillaume Rousse Guillaume.Rousse@inria.fr
Thu, 17 Mar 2005 16:54:57 +0100


chas williams - CONTRACTOR wrote:
> In message <42398850.7080701@inria.fr>,Guillaume Rousse writes:
> 
>>Any attempt to browser a afs filesystem segfault:
>>[guillaume@pomerol ~]$ ll /mnt/afs/inria.fr/rocq/home/atoll/common/cvs
>>Segmentation fault
> 
> 
> any oops in dmesg?
Right, I forgot:

Found system call table at 0xc037c8c0 (pattern scan)
Starting AFS cache scan...old style would have needed 1080000 contiguous 
bytes
found 2146 non-empty cache files (8%).
Unable to handle kernel NULL pointer dereference at virtual address 00000000
  printing eip:
c01c3ac8
*pde = 00000000
Oops: 0000 [#1]
Modules linked in: libafs nfsd exportfs lockd sunrpc autofs4 
snd-seq-dummy snd-seq-oss snd-seq-midi-event snd-seq snd-seq-device 
snd-pcm-oss snd-mixer-oss snd-intel8x0 snd-ac97-codec snd-pcm snd-timer 
snd-page-alloc snd soundcore af_packet e1000 ide-cd ext3 jbd nvidia 
intel-agp agpgart loop dm-mod ppa parport_pc imm scsi_mod parport tsdev 
evdev usbmouse usbhid ehci-hcd uhci-hcd usbcore
CPU:    0
EIP:    0060:[<c01c3ac8>]    Tainted: P      VLI
EFLAGS: 00210296   (2.6.11-2mdk-i586-up-1GB)
EIP is at __inode_queue_event+0x18/0x60
eax: f33c1000   ebx: fffffff8   ecx: f33c1000   edx: 00000000
esi: 00000004   edi: 00000020   ebp: e8d3bf8c   esp: e8d3bf74
ds: 007b   es: 007b   ss: 0068
Process ls (pid: 7025, threadinfo=e8d3a000 task=ea79c530)
Stack: 0805c170 00000004 00000004 e6c6d95c 00000004 e7842680 e8d3bfbc 
c019cdcc
        f33c1000 00000020 00000000 00000000 00000000 00000000 f7145000 
080998d0
        00000000 b8000ca0 e8d3a000 c0102d09 080998d0 00018800 0805c170 
00000000
Call Trace:
  [<c010412d>] show_stack+0x7d/0xa0
  [<c0104297>] show_registers+0x127/0x1b0
  [<c01043fd>] die+0xdd/0x1e0
  [<c0114bbf>] do_page_fault+0x43f/0x6bb
  [<c0103d8b>] error_code+0x2b/0x30
  [<c019cdcc>] sys_open+0x5c/0xd0
  [<c0102d09>] sysenter_past_esp+0x52/0x79
Code: da 83 c4 10 89 f0 8d 65 f8 5b 5e 5d e9 e2 fe ff ff 89 f6 55 89 e5 
57 56 53 83 ec 0c 8b 4d 08 8b 7d 0c 8b 91 14 01 00 00 8d 5a f8 <8b> 43 
08 0f 18 00 90 8d b1 14 01 00 00 eb 0f 8b 53 08 8d 42 f8

>>And I can't umount afs:
>>Arrêt de AFS client daemon : umount: /mnt/afs: périphérique occupé
>>umount: /mnt/afs: périphérique occupé
> 
> 
> i noticed that a change was revered in src/rx/rx.c  you might try
> getting the latest version from cvs instead.
This one ?
http://www.openafs.org/cgi-bin/wdelta/rollback-20050315
I'll try, thanks.

>>The last point is a regression from 1.3.78, which solved this issue from 
>>1.3.77. And 1.3.78 worked fine with previous 2.6.10 kernel.
> 
> 
> it would be helpful to know more about your kernel.  like smp, preempt,
> cpu type.
This is a no-smp, no-preempt kernel, build for i586, running on P4:
[guillaume@pomerol ~]$ uname -a
Linux pomerol.inria.fr 2.6.11-2mdk-i586-up-1GB #1 Thu Mar 10 16:21:46 
CET 2005 i686 Intel(R) Pentium(R) 4 CPU 3.00GHz unknown GNU/Linux

-- 
Spray drift is selectively attracted to soybean fields.
		-- The Dicamba Dictum