[BUG] The return of the sysfs oops

[BUG] The return of the sysfs oops

[Chris Rankin]

[-- Attachment #1: Type: text/plain, Size: 304 bytes --]

Hi,

Do you remember that oops in sysfs a few versions ago? (Kernel bug 8198) Well, it's baaaack in
2.6.22.9...

Cheers,
Chris



      _____________________________________________________
Do not compromise. Get unlimited storage and first rate spam protection with Yahoo! Mail.
http://uk.mail.yahoo.com

[-- Attachment #2: 2998135213-BUG-2.6.22.9 --]
[-- Type: application/octet-stream, Size: 3019 bytes --]

BUG: unable to handle kernel paging request at virtual address 70646fe5
 printing eip:
c013c60d
*pde = 00000000
Oops: 0002 [#1]
PREEMPT SMP 
Modules linked in: snd_rtctimer snd_seq_midi radeon drm pcspkr cpufreq_ondemand p4_clockmod speedstep_lib nfsd exportfs autofs4 eeprom nfs lockd sunrpc af_packet ipv6 binfmt_misc video1394 raw1394 thermal processor container button nvram snd_emu10k1_synth snd_emux_synth snd_seq_virmidi snd_seq_midi_emul snd_emu10k1 firmware_class snd_ac97_codec ac97_bus snd_seq_dummy snd_usb_audio snd_seq_oss snd_seq_midi_event snd_seq snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd_util_mem snd_page_alloc snd_usb_lib snd_rawmidi snd_seq_device snd_hwdep snd pwc firewire_ohci firewire_core crc_itu_t ohci1394 compat_ioctl32 ieee1394 videodev v4l2_common i2c_i801 i2c_core v4l1_compat parport_pc parport soundcore serio_raw psmouse e7xxx_edac edac_mc e1000 sata_sil sr_mod cdrom intel_agp agpgart sg rtc dm_snapshot dm_zero dm_mirror dm_mod ata_piix libata sd_mod scsi_mod ext3 jbd mbcache ehci_hcd ohci_hcd uhci_hcd usbcore
CPU:    0
EIP:    0060:[<c013c60d>]    Not tainted VLI
EFLAGS: 00210246   (2.6.22.9 #1)
EIP is at module_put+0x19/0x50
eax: 00000000   ebx: 70646e65   ecx: 00000001   edx: ea59c000
esi: eb589ac0   edi: ed1fb290   ebp: 70646e65   esp: ea59cf54
ds: 007b   es: 007b   fs: 00d8  gs: 0033  ss: 0068
Process udevd (pid: 6643, ti=ea59c000 task=ece29440 task.ti=ea59c000)
Stack: eb5b7df4 c0192172 00000010 ed21e3c0 eb5b7d80 ea506990 c015ec31 00000000 
       00000000 f791f180 ed21e3c0 ec79da80 00000000 ed21e3c0 c015c84c 00000003 
       00000003 ec79da80 ec79db00 c015d850 00000003 00000007 00000003 ea59c000 
Call Trace:
 [<c0192172>] sysfs_release+0x58/0x86
 [<c015ec31>] __fput+0x9c/0x142
 [<c015c84c>] filp_close+0x51/0x58
 [<c015d850>] sys_close+0x70/0xa7
 [<c0103e72>] sysenter_past_esp+0x5f/0x85
 [<c0270000>] rest_init+0x40/0x53
 =======================
Code: f4 06 00 89 f8 e8 ec f3 06 00 89 d8 5f 5b 5e 5f 5d c3 53 89 c3 85 c0 74 47 b8 01 00 00 00 e8 79 f6 fd ff e8 ae 4b 07 00 c1 e0 07 <ff> 8c 18 80 01 00 00 83 3b 02 75 0b 8b 83 88 05 00 00 e8 f2 ef 
EIP: [<c013c60d>] module_put+0x19/0x50 SS:ESP 0068:ea59cf54
note: udevd[6643] exited with preempt_count 1
BUG: scheduling while atomic: udevd/0x10000001/6643
 [<c027019e>] __sched_text_start+0x56/0x7fc
 [<c0124b0a>] irq_exit+0x50/0x72
 [<c0104883>] common_interrupt+0x23/0x28
 [<c011bd1c>] __cond_resched+0x12/0x2c
 [<c0270f3d>] cond_resched+0x26/0x31
 [<c014cfce>] unmap_vmas+0x3d8/0x4e1
 [<c014fbc4>] exit_mmap+0x7b/0x10c
 [<c011e5af>] mmput+0x37/0xab
 [<c01230f1>] do_exit+0x1e1/0x705
 [<c01053bf>] die+0x1e4/0x210
 [<c01186dd>] do_page_fault+0x443/0x516
 [<c011829a>] do_page_fault+0x0/0x516
 [<c0272952>] error_code+0x72/0x78
 [<c013c60d>] module_put+0x19/0x50
 [<c0192172>] sysfs_release+0x58/0x86
 [<c015ec31>] __fput+0x9c/0x142
 [<c015c84c>] filp_close+0x51/0x58
 [<c015d850>] sys_close+0x70/0xa7
 [<c0103e72>] sysenter_past_esp+0x5f/0x85
 [<c0270000>] rest_init+0x40/0x53
 =======================

Re: [BUG] The return of the sysfs oops

[Tejun Heo]

Chris Rankin wrote:
> Hi,
> 
> Do you remember that oops in sysfs a few versions ago? (Kernel bug 8198) Well, it's baaaack in
> 2.6.22.9...

This isn't really related to sysfs.  It seems module count was too low
and went away while there still were holders.  What were you doing when
it happened?  Can you reproduce it?

-- 
tejun

Re: [BUG] The return of the sysfs oops

[Chris Rankin]

--- Tejun Heo <htejun@gmail.com> wrote:
> This isn't really related to sysfs.  It seems module count was too low
> and went away while there still were holders.  What were you doing when
> it happened?  Can you reproduce it?

I haven't tried reproducing it, but all I was doing was selecting the Audio tab in winecfg again
(exactly as with bug #8198).

Cheers,
Chris



      ______________________________________________________
Yahoo! Mail now has unlimited storage, which means you can have spam control and more space for those important e-mails.
http://uk.mail.yahoo.com