random freeze (probably bcm43xx)

[vrde]

Binary package hint: kernel-image-2.6.17-7-powerpc-di

Hi,
I'm on a iBook running Ubuntu Edgy, sometimes the system freeze and I have to force the shut-down.

I collect some part of the kernel's log, as you can see it seems a bcm43xx or a Netdev Watchdog problem.

uname -a:
Linux eva 2.6.17-7-powerpc #2 Wed Sep 6 16:12:56 UTC 2006 ppc GNU/Linux

lspci | grep Network
0001:10:12.0 Network controller: Broadcom Corporation BCM4306 802.11b/g Wireless LAN Controller (rev 03)

If you need any other infos, I'll be glad to provide them.

[Michael R. Head (burner)]

Oops, it did it again. By the way, I'm not actively using the bcm43xx card for net access. The only thing that should be touching it is network-manager.

Sep 28 14:20:30 localhost kernel: [16030.887175] SoftMAC: Authentication response received from 00:13:ce:64:4b:2a but no queue item exists.
Sep 28 14:20:30 localhost kernel: [16030.947431] SoftMAC: Authentication response received from 00:13:ce:64:4b:2a but no queue item exists.
Sep 28 14:20:30 localhost kernel: [16031.066807] SoftMAC: Authentication response received from 00:13:ce:64:4b:2a but no queue item exists.
Sep 28 14:42:54 localhost kernel: [17375.167218] NETDEV WATCHDOG: eth1: transmit timed out
Sep 28 14:42:54 localhost kernel: [17375.167234] bcm43xx: Controller RESET (TX timeout) ...
Sep 28 14:42:54 localhost kernel: [17375.182098] bcm43xx: Controller restarted
Sep 28 15:10:55 localhost kernel: Inspecting /boot/System.map-2.6.17-10-powerpc
Sep 28 15:10:55 localhost kernel: Loaded 21819 symbols from /boot/System.map-2.6.17-10-powerpc.
Sep 28 15:10:55 localhost kernel: Symbols match kernel version 2.6.17.

[Michael R. Head (burner)]

On bug 62805 Matthew Garrett said that there is a patch that should be integrated after the beta.

[Ben Collins (ben-collins)]

Please test the 2.6.17-10.27 kernel that is being uploaded now (should show within a day).

[Michael R. Head (burner)]

So far so good, I'll post back if it locks up in the next day or two.

[Aleksander Kamil Modzelewski (aleander)]

It didn't help me. The touchpad is gets unstable when using bcm43xx still unstable and I still cannot connect to my home WEP network. I'll try unencrypted when I'm at the univeristy.

[Aleksander Kamil Modzelewski (aleander)]

Oh, I'm coming from Bug 63919, which has been marked as a duplicate of this one.

[Michael R. Head (burner)]

Oh, sorry, Aleksander, I guess I got a little carried away marking dupes.

[Michael R. Head (burner)]

My Mac Mini's been running fine with the bcm driver loaded since the update. Appears to be fixed.

[Michael R. Head (burner)]

Hmm... just had a problem, though it didn't freeze, the keyboard events stopped being handled by X, though the mouse events kept working. I could ssh in, but sudo wouldn't return, and system load was at 4 or 5, while no processes were running. Checked the syslog after powering off and cold booting:

Oct 8 20:13:01 localhost kernel: [27915.892910] NETDEV WATCHDOG: eth0: transmit timed out
Oct 8 20:13:01 localhost kernel: [27915.892923] bcm43xx: Controller RESET (TX timeout) ...
Oct 8 20:13:06 localhost kernel: [27920.892881] NETDEV WATCHDOG: eth0: transmit timed out
Oct 8 20:13:06 localhost kernel: [27920.892893] bcm43xx: Controller RESET (TX timeout) ...
Oct 8 20:13:11 localhost kernel: [27925.892851] NETDEV WATCHDOG: eth0: transmit timed out
Oct 8 20:13:11 localhost kernel: [27925.892863] bcm43xx: Controller RESET (TX timeout) ...
Oct 8 20:13:16 localhost kernel: [27930.892821] NETDEV WATCHDOG: eth0: transmit timed out
Oct 8 20:13:16 localhost kernel: [27930.892834] bcm43xx: Controller RESET (TX timeout) ...
Oct 8 20:13:21 localhost kernel: [27935.892791] NETDEV WATCHDOG: eth0: transmit timed out
Oct 8 20:13:21 localhost kernel: [27935.892804] bcm43xx: Controller RESET (TX timeout) ...
Oct 8 20:13:26 localhost kernel: [27940.892762] NETDEV WATCHDOG: eth0: transmit timed out
Oct 8 20:13:26 localhost kernel: [27940.892774] bcm43xx: Controller RESET (TX timeout) ...
Oct 8 20:15:41 localhost syslogd 1.4.1#18ubuntu6: restart.
Oct 8 20:15:41 localhost kernel: Inspecting /boot/System.map-2.6.17-10-powerpc

[Matthew Garrett (mjg59)]

Ok, well that certainly sounds better. What does your /proc/interrupts look like?

[Michael R. Head (burner)]

burner@flambe:~$ cat /proc/interrupts
           CPU0
  1: 635972 MPIC 1 Edge i2sbus: i2s-a (tx)
  2: 0 MPIC 1 Edge i2sbus: i2s-a (rx)
  3: 0 MPIC 1 Edge i2sbus: i2s-b (tx)
  4: 0 MPIC 1 Edge i2sbus: i2s-b (rx)
 25: 273655 MPIC 1 Level VIA-PMU
 26: 0 MPIC 1 Level keywest i2c
 29: 2760054 MPIC 1 Level ohci_hcd:usb1
 30: 0 MPIC 1 Level i2sbus: i2s-a (control)
 31: 0 MPIC 1 Level i2sbus: i2s-b (control)
 39: 168264 MPIC 1 Level ide0
 40: 2 MPIC 1 Level ohci1394
 41: 562766 MPIC 1 Level eth0
 42: 0 MPIC 1 Level keywest i2c
 47: 0 MPIC 1 Level GPIO1 ADB
 48: 828422 MPIC 1 Level radeon@pci:0000:00:10.0
 52: 36610 MPIC 1 Level bcm43xx
 61: 0 MPIC 1 Edge headphone-detect
 63: 47372 MPIC 1 Level ohci_hcd:usb2, ohci_hcd:usb3, ehci_hcd:usb4
BAD: 2213579

[Michael R. Head (burner)]

Actually, while the problem occurs less frequently, it does still happen from time to time. I'm not sure what triggers it, but eventually, windows no longer map in X and sudo hangs indefinitely (for example, I can't do "sudo reboot", because I sudo never asks me for a password).

The concommittent syslog symptom is the following repeated over and over.
Oct 30 02:03:17 localhost kernel: [89695.728193] NETDEV WATCHDOG: eth1: transmit timed out
Oct 30 02:03:17 localhost kernel: [89695.728204] bcm43xx: Controller RESET (TX timeout) ...
Oct 30 02:03:27 localhost kernel: [89705.728135] NETDEV WATCHDOG: eth1: transmit timed out
Oct 30 02:03:27 localhost kernel: [89705.728149] bcm43xx: Controller RESET (TX timeout) ...
Oct 30 02:03:32 localhost kernel: [89710.728105] NETDEV WATCHDOG: eth1: transmit timed out
Oct 30 02:03:32 localhost kernel: [89710.728120] bcm43xx: Controller RESET (TX timeout) ...

I noticed that some time before the driver had problems, I saw this:
Oct 30 01:09:52 localhost kernel: [86490.747214] NETDEV WATCHDOG: eth1: transmit timed out
Oct 30 01:09:52 localhost kernel: [86490.747226] bcm43xx: Controller RESET (TX timeout) ...
Oct 30 01:09:52 localhost kernel: [86491.265854] bcm43xx: Controller restarted

[Xamusk (ronanpaixao)]

I'm experiencing the same problem happening with Michael
I'm using edgy with kernel 2.6.17-10 and this problem appears both using 32-bit and 64-bit versions of ubuntu in my laptop (HP NX6125)
Somehow I noticed a single key combination that seems to work: Fn+F8, which displays the battery status. Don't know if that's something related to ACPI events though.

Also, as X isn't mapping any keyboard events, I can't even get out of X nor go to a "real" terminal. Besides, normally when this problem happens, *all* new programs doesn't seem to work, so I can't open a new terminal nor the shutdown screen (I can click in the menus but they don't open the apps)

Don't know if it's related, but sometimes if I leave the laptop on for a long period without network events, those two lines that repeat in syslog (transmit timed out and Controller RESET) are reported and the wireless network stops responding, but without the keyboard lock and other symptoms. The solution to this is simply reconnect to the network through NetworkManager.

[chefweb (ulrich-timm-uni-rostock)]

I can confirm random freezes with the bcm43xx .. when I want work with my notebook after a long idle priode of about 15min, the system is complete frozen and I have to to restart

[b (ben-ekran)]

my powerbook is doing a very similar thing, once and a while with bcm43xx loaded, the keyboard will freeze, mouse still works. When I log out using gdm (no other way) I get those very same "NETDEV WATCHDOG: transmit timed out" messages and the machine just prints it out over and over, I have to manually turn off the machine.

I don't see any pattern as to when it happens, but often enough that I have blacklisted bcm43xx and modprobe only when I need it.

I hope this gets fixed at some point. Over all I'm not as impressed to edgy compared to dapper.

[Michael R. Head (burner)]

Now that bcm43xx works in feisty (see bug 85404), can anyone confirm if this bug is still present in linux-source-2.6.20? I'm trying to see on my powerpc box myself.

[Aleksander Kamil Modzelewski (aleander)]

My laptop is currently in service, but I'll check as soon as I get it back.

[Michael R. Head (burner)]

I'm saying it's fixed in feisty.

[arty (me-arty)]

I've recently compiled linux-source-2.6.20 to get rid of this error, but it still bugs me : (
I know it happens sometimes when I do modprobe rt2570 to fix my wifi connection.

Now I will try to update to Feisty

[Dan Munckton (munckfish)]

I'm also experiencing this error on an Edgy loaded Dell Inspiron 1100 with a Belkin 54g Wireless Network card.