Bug 155299

Summary: Kernel BUG at /ipw/ipw2200.c:7873 under load with weak WLAN signal
Product: [openSUSE] SUSE LINUX 10.0 Reporter: Tilman Schmidt <tilman>
Component: KernelAssignee: Joachim Gleissner <joachim.gleissner>
Status: RESOLVED WORKSFORME QA Contact: E-mail List <qa-bugs>
Severity: Major    
Priority: P5 - None CC: suse-beta
Version: Final   
Target Milestone: ---   
Hardware: i686   
OS: SuSE Linux 10.0   
Whiteboard:
Found By: Customer Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---

Description Tilman Schmidt 2006-03-04 18:49:11 UTC
Notebook FSC Amilo Pro V2065 with SuSE Linux 10.0 installed from retail DVD hangs reproducibly within some tens of seconds when trying to download a big file over WLAN while the signal is weak (eg. -52 dBm). The same downloads run fine when the signal is stronger (eg. -35 dBm).

Console output on crash (copied manually):

kernel BUG at /usr/src/packages/BUILD/kernel-default-2.6.13/modules-2.6.13/ipw/ipw2200.c:7873!
invalid operand: 0000 [#1]
Modules linked in: ipt_pkttype ipt_LOG ipt_limit cpufreq_ondemand
cpufreq_userspace cpufreq_powersave speedstep_centrino freq_table button
battery ac michael_mic arc4 ieee80211_crypt_tkip joydev af_packet edd
pcmcia ipw2200 ohci1394 ieee1394 ieee80211 yenta_socket rsrc_nonstatic
pcmcia_core shpchp pci_hotplug ieee80211_crypt r8169 firmware_class
intel_agp agpgart ehci_hcd uhci_hcd usbcore i2c_801 i2c_core hw_random
generic parport_pc lp parport ip6t_REJECT ipt_REJECT ipt_state
iptable_mangle iptable_nat iptable_filter ip6table_mangle ip_conntrack
ip_tables ip6table_filter ip6_tables ipv6 ntfs nls_utf8 nls_cp437 vfat
fat dm_mod reiserfs fan sg thermal processor ata_piix ide_cd cdrom ahci
libata piix sd_mod scsi_mod ide_disk ide_core
CPU:    0
EIP:    0060:[<f9496838>]    Tainted: G     U VLI
EFLAGS: 00010046   (2.6.13-15.8-default)
EIP is at ipw_rx+0xcb8/0x2620 [ipw2200]
eax: ffff0000   ebx: 00000008   ecx: f6656e24   edx: f6656e1e
esi: 000069cc   edi: f6656e20   ebp: f5618000   esp: c03abda8
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 0, threadinfo=c03aa000 task=c034db80)
Stack: 00000005 f5618000 e2b2cbc0 743ba8c0 ba2acf47 f63cee00 f63cee00 c02ac67b
       f63cee00 f6656000 c02ad289 00000000 f63cee00 f59cff24 00e806e8 e2c8e058
       f92889d4 f1980060 00000000 f1980060 f665650c 39542c00 c03abe18 c0118be7
Call Trace:
 [<c02ac67b>] ip_dst_output+0x5b/0x70
 [<c02ad289>] ip_queue_xmit+0x289/0x4b0
 [<f92889d4>] tcp_in_window+0x2f4/0x540 [ip_conntrack]
 [<c0118be7>] activate_task+0x57/0x70
 [<c0287dff>] sock_def_readable+0x1f/0x50
[...snip...]
<0>Kernel panic - not syncing: Fatal exception in interrupt
Comment 1 Greg Kroah-Hartman 2006-03-04 21:31:44 UTC
Can you try the Beta6 opensuse release to see if this is fixed there?
Comment 2 Tilman Schmidt 2006-03-04 22:22:32 UTC
Does that mean downloading and burning five CDs, and completely reinstalling the notebook in question? I don't see how I could possibly do that, not in the short term anyway.
Comment 3 Greg Kroah-Hartman 2006-03-04 22:35:45 UTC
You could just install the kernel package to see if the new version of the driver fixes your
problem (it has numerous fixes from the 10.0 version in it.)
Comment 4 Tilman Schmidt 2006-03-05 22:45:58 UTC
Just tried http://mirror.karneval.cz/pub/linux/opensuse/distribution/SL-OSS-factory/inst-source/suse/i586/kernel-default-2.6.16_rc4-3.i586.rpm which booted alright (in spite of an error message about a failed post-installation script) but failed to even activate the WLAN interface because of missing firmware files (/lib/firmware/ipw-2.4-boot.fw - I have only ipw-2.3-*).

Trying to find these now (hints welcome), will report back with more results when I have them.
Comment 5 Christian Boltz 2006-03-05 23:13:03 UTC
Try the ipw-firmware package from
ftp://ftp.gwdg.de/pub/suse/i386/10.1/SUSE-Linux10.1-Beta6-Extra/suse/noarch
;-)

And don't forget to reset the bug to ASSIGNED after providing the needed information.
Comment 6 Olaf Kirch 2006-03-06 08:55:06 UTC
WLAN stuff goes to Joachim Gleissner; reassigning
Comment 7 Tilman Schmidt 2006-03-07 11:02:28 UTC
Ok, I have to admit defeat. I have installed into my SuSE 10.0 system the following packages from SuSE 10.1 beta 6, in this order:

kernel-default-2.6.16_rc4-3.i586.rpm
ipw2200-fw-2.4.tgz (from http://ipw2200.sourceforge.net)
wireless-tools-28pre13-12.i586.rpm
wpa_supplicant-0.4.8-2.i586.rpm
openssl-0.9.8a-8.i586.rpm

(I guess it was obvious I would need all these, except to me) and I still don't even have WLAN connectivity back. wpa_supplicant now fails with the messages:
wpa_supplicant: /lib/tls/libc.so.6: version `GLIBC_2.4' not found (required by /usr/lib/libssl.so.0.9.8)
wpa_supplicant: /lib/tls/libc.so.6: version `GLIBC_2.4' not found (required by /usr/lib/libcrypto.so.0.9.8)

so I guess it's now asking me to upgrade libc as well. OTOH the system is rapidly becoming unusable with more and more of the regular programs failing because they depend on the old versions of the packages I replaced, eg.:

ssh: error while loading shared libraries: libcrypto.so.0.9.7: cannot open shared object file: No such file or directory

So, effectively, *no*, I *can't* try the Beta6 opensuse release to see if this is fixed there, because I can't even get that far.

Is there anything else I can do for you?
Comment 8 Tilman Schmidt 2006-03-26 00:47:36 UTC
Any news? Can I help in any way?
Comment 10 Tilman Schmidt 2006-05-04 22:29:52 UTC
As there has been no hint of activity on this bug for all of eight weeks now, I guess there's no chance of getting it fixed anymore? Well, I'll stop tracing this bug (so if you want something from me after all please contact me by mail) and continue using Windows XP then (that at least works, sorry to say) until I get 10.1 through my subscription. If that doesn't fix it either ... well, I don't know. Perhaps look at a different distribution. Not quite ready to give up completely on Linux with that notebook, anyway.
Comment 11 Joachim Gleissner 2006-05-09 08:33:19 UTC
Sorry for the delay, 10.0 had not the highest priority for me right in the end of the release phase of 10.1.

To comment 7: It shouldn't be necessary to update the userspace tools, just the kernel and the firmware package should do. Am I right that you are using WPA with TKIP encryption?
Comment 12 Tilman Schmidt 2006-05-09 22:52:39 UTC
Well, all I can say now is that after installing just kernel-default-2.6.16_rc4-3.i586.rpm from mirror.karneval.cz and ipw2200-fw-2.4.tgz from ipw2200.sourceforge.net I was unable to connect to my WLAN with WPA ancryption and pre-shared key authentication.

Unfortunately it's too long ago for me to remember the exact symptoms now, and I didn't keep my notes, having given up on that issue. I am however pretty sure the original wireless tools from 10.0 did produce some message indicating an incompatibility with the new driver, otherwise I wouldn't have tried to update them.

Would you like me to repeat any of my tests? If so, please tell me what exactly I should try, and I'll see what I can do.
Comment 13 Tilman Schmidt 2006-05-12 23:52:05 UTC
After replacing (for unrelated reasons) my WLAN AP/router, previously a LevelOne model WBR-3406TX, with a D-Link model DI-624 I cannot reproduce the kernel BUG/system freeze anymore. Instead my /var/log/messages file is now flooded with messages "kernel: TKIP: replay detected", but the system remains usable. So the problem has lost most of its urgency for me.

For the time being, I still have the old router sitting unused on my shelf, so if you want me to do some more tests with it please let me know.
Comment 14 Tilman Schmidt 2006-07-24 19:22:08 UTC
I am going to sell the LevelOne router which was able to trigger, so I won't be available for any tests anymore. However, as there wasn't any response for the last two and a half months I guess there is no interest in reproducing this now anyway.
Comment 15 Joachim Gleissner 2006-09-25 15:56:09 UTC
Sorry, did not find the time. I'm closing the bug now, feel free to reopen if you stumble across it again. Thanks!