Re: [2.6.24-rc1][BUG] Oops on battery removal

Previous thread: [PATCH 0/4] Blackfin I2C/TWI driver updates and bug fixing according to Jean's review by Bryan Wu on Thursday, November 1, 2007 - 11:24 pm. (9 messages)

Next thread: [PATCH] make getdelays cgroupstats aware by Balbir Singh on Friday, November 2, 2007 - 1:26 am. (1 message)
From: Rolf Eike Beer
Date: Friday, November 2, 2007 - 12:03 am

--Boundary-01=_QvsKH6Hxqj1h8fq
Content-Type: text/plain;
  charset="utf-8"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Hi,

this happened while I removed my battery on bootup. Complete dmesg is=20
attached. Kernel is 2.6.24-rc1-git of yesterday (last commit was=20
d919fd433b5823d1cf9d0688eb2eec183de9b74c).

Greetings,

Eike

BUG: unable to handle kernel NULL pointer dereference at virtual address=20
0000001c
printing eip: c02a82cb *pde =3D 00000000=20
Oops: 0000 [#1] SMP=20
Modules linked in: iptable_filter ip_tables ip6table_filter ip6_tables=20
x_tables ipv6 af_packet cpufreq_conservative cpufreq_ondemand=20
cpufreq_userspace cpufreq_powersave acpi_cpufreq speedstep_lib freq_table=20
snd_pcm_oss snd_mixer_oss snd_seq snd_seq_device twofish twofish_common cbc=
=20
blkcipher dm_crypt nls_utf8 ntfs ext3 jbd loop mmc_block pcmcia container=20
battery iwl3945 firmware_class yenta_socket button ohci1394 backlight outpu=
t=20
ac power_supply rsrc_nonstatic pcmcia_core mmc_core ieee1394 intel_agp joyd=
ev=20
agpgart snd_hda_intel sr_mod cdrom r8169 snd_pcm snd_timer mac80211 snd=20
soundcore snd_page_alloc i2c_i801 i2c_core cfg80211 serio_raw sg usbhid=20
ehci_hcd uhci_hcd sd_mod usbcore dm_snapshot edd dm_mod fan ata_piix libata=
=20
scsi_mod thermal processor

Pid: 47, comm: kacpi_notify Not tainted (2.6.24-rc1-git #68)
EIP: 0060:[<c02a82cb>] EFLAGS: 00010286 CPU: 0
EIP is at klist_del+0xc/0x39
EAX: 00000000 EBX: 00000000 ECX: 00000000 EDX: 00000000
ESI: c2e2bc00 EDI: c2e2bc20 EBP: c1c88d7c ESP: c1c88d70
 DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
Process kacpi_notify (pid: 47, ti=3Dc1c88000 task=3Dc1cb5350 task.ti=3Dc1c8=
8000)
Stack: c2e2bc00 c2e2bc00 c1ce960c c1c88d90 c0231ab6 c2e2bc00 c2e2ba00 c1c88=
e1c=20
       c1c88d9c c0231ced c2e2ba1c c1c88da8 f8bee01a c2e2ba00 c1c88db4 f8c39=
04b=20
       c2e2ba00 c1c88de4 f8c391a2 c01d47bc c01d47bc 00000001 00000000 00008=
0d0=20
Call Trace:
 [<c0104edd>] show_trace_log_lvl+0x1a/0x2f
 ...
From: Rolf Eike Beer
Date: Friday, November 2, 2007 - 1:37 am

Ok, I found out that it has nothing to do with the actual removal as it see=
ms.=20
When I start the laptop with battery already detached I get the same error.

Eike
From: Alexey Starikovskiy
Date: Friday, November 2, 2007 - 2:29 am

Please check if this patch helps.

Thanks,
Alex.
From: Rolf Eike Beer
Date: Friday, November 2, 2007 - 4:49 am

It made things worse:

Nov  2 12:37:20 donald kernel: ACPI: EC: non-query interrupt received,=20
switching to interrupt mode
Nov  2 12:37:20 donald acpid: received event "battery BAT1 00000080 0000000=
0"
Nov  2 12:37:20 donald acpid: notifying client 2856[101:102]
Nov  2 12:37:20 donald acpid: notifying client 2911[0:0]
Nov  2 12:37:20 donald acpid: notifying client 3048[0:0]
Nov  2 12:37:20 donald acpid: completed event "battery BAT1 00000080 000000=
00"
Nov  2 12:37:20 donald kernel: WARNING:=20
at /home/eike/repos/linux-2.6/lib/kref.c:33 kref_get()
Nov  2 12:37:20 donald kernel:  [<c0104edd>] show_trace_log_lvl+0x1a/0x2f
Nov  2 12:37:20 donald kernel:  [<c010576e>] show_trace+0x12/0x14
Nov  2 12:37:20 donald kernel:  [<c010585b>] dump_stack+0x16/0x18
Nov  2 12:37:20 donald kernel:  [<c01d1d3c>] kref_get+0x37/0x43
Nov  2 12:37:20 donald kernel:  [<c01d10af>] kobject_get+0x12/0x17
Nov  2 12:37:20 donald kernel:  [<c0231509>] get_device+0x13/0x1d
Nov  2 12:37:20 donald kernel:  [<c02319e8>] device_remove_file+0xe/0x2a
Nov  2 12:37:20 donald kernel:  [<f8be7043>] sysfs_remove_battery+0x17/0x22=
=20
[battery]
Nov  2 12:37:20 donald kernel:  [<f8be71a2>] acpi_battery_update+0x3f/0x255=
=20
[battery]
Nov  2 12:37:20 donald kernel:  [<f8be73ed>] acpi_battery_read+0x11/0x20=20
[battery]
Nov  2 12:37:20 donald powersaved[3048]: WARNING (filter_function:201) Hal=
=20
service stopped. Battery information no longer available
Nov  2 12:37:20 donald kernel:  [<f8be741c>] acpi_battery_read_state+0xf/0x=
11=20
[battery]
Nov  2 12:37:20 donald kernel:  [<c01889ca>] seq_read+0xea/0x264
Nov  2 12:37:20 donald kernel:  [<c019fddb>] proc_reg_read+0x63/0x76
Nov  2 12:37:20 donald kernel:  [<c0172631>] vfs_read+0xa6/0x12f
Nov  2 12:37:20 donald kernel:  [<c0172a2a>] sys_read+0x3d/0x61
Nov  2 12:37:20 donald kernel:  [<c0103f12>] sysenter_past_esp+0x5f/0x85
Nov  2 12:37:20 donald kernel:  =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D
Nov  2 12:37:20 donald ...
From: Alexey Starikovskiy
Date: Friday, November 2, 2007 - 6:00 am

How about this patch?
From: Rolf Eike Beer
Date: Friday, November 2, 2007 - 6:26 am

Ehm, that is the same patch as before!

Btw: it has an empty From: line

Eike
From: Alexey Starikovskiy
Date: Friday, November 2, 2007 - 6:41 am

Patch updated.
From: Rolf Eike Beer
Date: Friday, November 2, 2007 - 7:20 am

Tested-by: Rolf Eike Beer <eike-kernel@sf-tec.de>

It makes the crash go away. But there is something weird going on. Before I=
=20
logged into X I did an "cat /proc/acpi/battery/BAT1/*" and it showed the=20
battery as present with some (random?) values. Now I'm in X and everything=
=20
looks fine there (as in "not present").

Eike
Previous thread: [PATCH 0/4] Blackfin I2C/TWI driver updates and bug fixing according to Jean's review by Bryan Wu on Thursday, November 1, 2007 - 11:24 pm. (9 messages)

Next thread: [PATCH] make getdelays cgroupstats aware by Balbir Singh on Friday, November 2, 2007 - 1:26 am. (1 message)