[linux-cifs-client] Re: cifs unmount corrupts mountpoints

[brendan powers]

I have some more info about the problem, i have scripts that
mount/unmount, and then check the resulting mountpoint for io
erros(just using ls). I have 9 of these running, and after a period of
time i got a strack trace in my kernel logs

here is what i got

Unable to handle kernel NULL pointer dereference at virtual address 0000000a
 printing eip:
c0116ee1
*pde = 00000000
Oops: 0002 [#7]
PREEMPT
Modules linked in: nls_cp437 cifs vmnet vmmon ipv6 lp af_packet
binfmt_misc autofs4 parport_pc parport floppy evdev pcspkr rtc
hw_random shpchp pciehp pci_hotplug ohci_hcd usbcore tg3
firmware_class capability commoncap psmouse ide_cd cdrom ext3 jbd
ide_generic hpt366 ide_disk amd74xx ide_core unix font vesafb
cfbcopyarea cfbimgblt cfbfillrect
CPU:    0
EIP:    0060:[<c0116ee1>]    Tainted: P
EFLAGS: 00010046   (2.6.8-2-386)
EIP is at prepare_to_wait+0x24/0x4d
eax: 00000006   ebx: dc833eac   ecx: dc832000   edx: dc833eb8
esi: f7fc4298   edi: 00000246   ebp: 00000002   esp: dc833e74
ds: 007b   es: 007b   ss: 0068
Process umount (pid: 2729, threadinfo=dc832000 task=e06ed3a0)
Stack: dc833eac dc833ea0 dc833ec0 f7feb1c0 f8b137c1 00000000 00000000 e06ed3a0
       c0116fb4 dc833eb8 dc833eb8 f250d29c dc833f28 c011990c 00000000 e06ed3a0
       c0116fb4 dc833eb8 dc833eb8 dc833ecc f250d454 c010699e f7fc4b50 00000000
Call Trace:
 [<f8b137c1>] SendReceive+0xea/0x551 [cifs]
 [<c0116fb4>] autoremove_wake_function+0x0/0x3a
 [<c011990c>] profile_hook+0x1c/0x34
 [<c0116fb4>] autoremove_wake_function+0x0/0x3a
 [<c010699e>] apic_timer_interrupt+0x1a/0x20
 [<f8b01e1c>] CIFSSMBTDis+0xdc/0x126 [cifs]
 [<f8b0b41e>] cifs_umount+0x52/0x167 [cifs]
 [<f8b01169>] cifs_put_super+0x43/0x76 [cifs]
 [<c0149a3a>] generic_shutdown_super+0x89/0x13b
 [<c014a296>] kill_anon_super+0xf/0x42
 [<c014991f>] deactivate_super+0x46/0x59
 [<c015c24a>] sys_umount+0x67/0x6f
 [<c013b5e0>] do_munmap+0x136/0x142
 [<c0114a94>] do_page_fault+0x0/0x49f
 [<c0106a19>] error_code+0x2d/0x38
 [<c0105f97>] syscall_call+0x7/0xb
Code: 89 50 04 89 43 0c 89 72 04 89 16 8b 01 89 28 57 9d 8b 41 08
 <6>note: umount[2729] exited with preempt_count 2
bad: scheduling while atomic!
 [<c024d5a0>] schedule+0x3c/0x3e6
 [<c01382ad>] unmap_vmas+0xe0/0x1cf
 [<c013832f>] unmap_vmas+0x162/0x1cf
 [<c013b8a3>] exit_mmap+0x6a/0x12a
 [<c01171db>] mmput+0x5e/0x73
 [<c011ab81>] do_exit+0x162/0x34f
 [<c0114a94>] do_page_fault+0x0/0x49f
 [<c0106fde>] do_divide_error+0x0/0xa7
 [<c0114df8>] do_page_fault+0x364/0x49f
 [<c0133fff>] cache_grow+0xcc/0x120
 [<c01341a7>] cache_alloc_refill+0x154/0x184
 [<c0134354>] kmem_cache_alloc+0x2a/0x30
 [<c012fc2f>] mempool_alloc_slab+0xd/0x10
 [<c012fac9>] mempool_alloc+0x61/0x136
 [<c0114a94>] do_page_fault+0x0/0x49f
 [<c0106a19>] error_code+0x2d/0x38
 [<c0116ee1>] prepare_to_wait+0x24/0x4d
 [<f8b137c1>] SendReceive+0xea/0x551 [cifs]
 [<c0116fb4>] autoremove_wake_function+0x0/0x3a
 [<c011990c>] profile_hook+0x1c/0x34
 [<c0116fb4>] autoremove_wake_function+0x0/0x3a
 [<c010699e>] apic_timer_interrupt+0x1a/0x20
 [<f8b01e1c>] CIFSSMBTDis+0xdc/0x126 [cifs]
 [<f8b0b41e>] cifs_umount+0x52/0x167 [cifs]
 [<f8b01169>] cifs_put_super+0x43/0x76 [cifs]
 [<c0149a3a>] generic_shutdown_super+0x89/0x13b
 [<c014a296>] kill_anon_super+0xf/0x42
 [<c014991f>] deactivate_super+0x46/0x59
 [<c015c24a>] sys_umount+0x67/0x6f
 [<c013b5e0>] do_munmap+0x136/0x142
 [<c0114a94>] do_page_fault+0x0/0x49f
 [<c0106a19>] error_code+0x2d/0x38
 [<c0105f97>] syscall_call+0x7/0xb
 CIFS VFS: cifs_umount failed with return code -5




On 5/15/06, brendan powers <brendan0powers at gmail.com> wrote:
> Hi i'm having some odd behavior with cifs mounts. I have several
> mounts from a linux box to a win2k3 server. Each user on the linux box
> has a public folder mounted in there home directory. When a use
> unmount the share(by logging out) sometimes the mount will do strange
> things like give input/output errors when you try to list the files,
> or completely hand a process that tries to access it. These process
> cannot be killed with kill -9. Also sometimes this happens when the
> mount is idle for a period of time. I am using debian sarge 3.1,
> kernels 2.6.8 and 2.6.16.7 samba 3.0.14a, mount.cifs 1.6.
> Anyone have any ideas why this might happen?
>
> Here is some output from my logs
> http://office.resara.com/samba.log
>