Oops with hostap_pci (?)

Previous thread: [PATCH] [WATCHDOG] it8712f_wdt support for 16-bit timeout values, WDIOC_GETSTATUS by Andrew Paprocki on Sunday, February 10, 2008 - 11:11 pm. (3 messages)

Next thread: [2.6.25-rc1 regression] Suspend to RAM (bisected) by Carlos R. Mafra on Monday, February 11, 2008 - 1:25 am. (10 messages)
To: Linux Kernel Mailing List <linux-kernel@...>
Date: Sunday, February 10, 2008 - 11:19 pm

Hi,

A few days back I started having strange lockups on a gateway machine so I
started looking at things. Then I compiled the 2.6.24.1 kernel and started
having oopses not long after upping the wlan0 (hostap_pci) interface.

So I enabled netconsole and got a few logs. Now the sad point is that I'm
getting an oops even with my older kernel which used to be fine (2.6.23.9). I
also checked with 2.6.24 and the effects are the same: I boot, I up the wlan0
interface and a few seconds or minutes later, boom! Sometimes only rmmod'ing
hostap_pci triggers the oops. I'm suspecting some hardware problem and have
already checked the ram with memtest86+ and tested with only one memory module
out of two plugged: same thing.

If anybody could take a look at these and shed some light on that issue...

Thanks a lot,

Ignacy

--
Save the whales. Feed the hungry. Free the mallocs.

To: Linux Kernel Mailing List <linux-kernel@...>
Date: Monday, February 11, 2008 - 8:23 am

Okay, false alarm... it's all my fault. :/

The cause of the problem was my previous tampering with udev rules. The udev
rules as such (on Ubuntu Gutsy) were bad for hostapd, since persistent rules
were written for the wlan0ap interface name created by hostapd. So I changed
a few things that had the unexpected effect of renaming the initial
hostap_pci's wifi0 into wlan0ap. This in turn made hostap_pci oops in many
cases.

Anyway, I've modified my udev rules again and hopefully this will be it. =)

--
"The whole problem with the world is that fools and fanatics are
always so certain of themselves, and wiser people so full of doubts."
- Bertrand Russell
--

To: Ignacy Gawedzki <lkml@...>
Cc: Linux Kernel Mailing List <linux-kernel@...>, <linux-wireless@...>
Date: Wednesday, February 13, 2008 - 3:27 am

No, you found a bug. The kernel shouldn't oops in reaction to userspace
activity, even udev. Ever.

With kernel 2.6.24.1

BUG: unable to handle kernel NULL pointer dereference at virtual address 00000000
printing eip: f08f50c2 *pde = 00000000
Oops: 0000 [#1]
Modules linked in: lirc_serial(F) lirc_dev cls_fw sch_prio sch_htb iptable_nat xt_limit xt_state ipt_REJECT xt_tcpudp ipt_LOG xt_DSCP xt_dscp xt_mark nf_conntrack_ipv4 xt_CONNMARK xt_MARK iptable_mangle iptable_filter ip_tables x_tables nf_nat_ftp nf_nat nf_conntrack_ftp nf_conntrack ipv6 evdev hostap_pci i2c_viapro hostap via686a ieee80211_crypt ide_cd

Pid: 0, comm: swapper Tainted: GF (2.6.24.1 #5)
EIP: 0060:[<f08f50c2>] EFLAGS: 00010297 CPU: 0
EIP is at hostap_80211_rx+0x41d/0xecf [hostap]
EAX: eec28460 EBX: 00000000 ECX: eec28444 EDX: 00000000
ESI: efbb8434 EDI: 00000000 EBP: efbb843e ESP: c0419e74
DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068
Process swapper (pid: 0, ti=c0418000 task=c03e4300 task.ti=c0418000)
Stack: 00000000 00000080 0000004c 00000001 c0419f2c c0419f30 ef3ab760 00000018
00000100 eec28444 00001148 00000040 0000c9c0 00000001 ef8d3370 00002a40
04b1cd93 000a1e00 00001148 013a1148 685b0900 ef8d3000 1f714b23 685b0900
Call Trace:
[<f090ffca>] hostap_rx_tasklet+0x11f/0x145 [hostap_pci]
[<c011e399>] run_timer_softirq+0x11/0x12f
[<c011bbbc>] tasklet_action+0x32/0x52
[<c011bb24>] __do_softirq+0x35/0x75
[<c011bb86>] do_softirq+0x22/0x26
[<c011bdb3>] irq_exit+0x29/0x58
[<c0105bc0>] do_IRQ+0x58/0x6b
[<c010455b>] common_interrupt+0x23/0x28
[<c013007b>] mod_sysfs_init+0x17/0x6d
[<c011007b>] arch_setup_additional_pages+0x121/0x13a
[<c023f4a0>] acpi_processor_idle+0x244/0x3c4
[<c01024fc>] cpu_idle+0x43/0x5d
[<c041a9ac>] start_kernel+0x237/0x23c
[<c041a303>] unknown_bootoption+0x0/0x195
=======================
Code: 0a 8b 4c 24 24 8b 59 1c eb 21 83 bb d8 00 00 00 04 75 16 8d 83 dc 00 0...

Previous thread: [PATCH] [WATCHDOG] it8712f_wdt support for 16-bit timeout values, WDIOC_GETSTATUS by Andrew Paprocki on Sunday, February 10, 2008 - 11:11 pm. (3 messages)

Next thread: [2.6.25-rc1 regression] Suspend to RAM (bisected) by Carlos R. Mafra on Monday, February 11, 2008 - 1:25 am. (10 messages)