BCM4321 Wireless dropping connection randomly with no indication

[Ryan Thompson]

Since upgrading to 12.04, my wireless connection has been failing randomly. It works fine for an indeterminate time after connecting to my router, and then randomly all traffic ceases, and I become unable to load any websites, download anything, or even ping my router. The wireless connection may work fine for as long as a few hours or as short as a few minutes before this happens. Every time it happens, I check all the logs, but there is no log entry corresponding to the loss of connection. It happens regardless of whether or not the laptop is plugged in. My wireless card is identified in lspci as:

0c:00.0 Network controller: Broadcom Corporation BCM4321 802.11a/b/g/n (rev 03)

uname -a: Linux aeolus 3.2.0-24-generic #37-Ubuntu SMP Wed Apr 25 08:43:22 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux

Please let me know what other information I can provide, and I will happily provide it.

[Fabio Marconi (fabiomarconi)]

Hello
Please attach here,using the button below 'Add attachment or patch' the files
/var/log/kern.log
/var/log/dmesg
possibly just after the bug is reproduced.
Thanks
Fabio
---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Ryan Thompson (rct86)]

Ok, I'll grab those log files after the next time the connection cuts out, but as I said, I haven't previously seen any log entries appearing in any logs at the time of disconnection.

Also, I didn't explicitly say so in my initial report, but the worst part about this is that there is NO indication that connectivity has been lost. In particular, the Wifi icon in the notification area does nto change at all.

Also, I can fix the problem temporarily by simply reconnecting to the wireless network, but the same problem always happens again.

It looks like the problem MAY be correlated with being on battery power, but I'm not sure. Maybe wifi power saving is to blame?

[Ryan Thompson (rct86)]

I'm attaching kern.log and dmesg from shortly after I experienced a loss of connectivity. As I mentioned, there are no log messages at the time of disconnection, which is about 20:15. All the messages in dmesg are from boot time, and the last message in kern.log is:

May 5 20:08:54 aeolus kernel: [165991.024139] eth2: no IPv6 routers present

which is from when I connected my VPN at 20:08. Between 20:09 and 20:15, there is nothing in the logs.

[Fabio Marconi (fabiomarconi)]

May 2 22:16:47 aeolus kernel: [ 5.639393] Bluetooth: RFCOMM TTY layer initialized
May 2 22:16:47 aeolus kernel: [ 5.639399] Bluetooth: RFCOMM socket layer initialized
May 2 22:16:47 aeolus kernel: [ 5.639401] Bluetooth: RFCOMM ver 1.11
May 2 22:16:47 aeolus kernel: [ 5.683901] Bluetooth: BNEP (Ethernet Emulation) ver 1.3 ###
May 2 22:16:47 aeolus kernel: [ 5.683904] Bluetooth: BNEP filters: protocol multicast
May 2 22:16:47 aeolus kernel: [ 6.007522] tg3 0000:09:00.0: irq 47 for MSI/MSI-X
---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Fabio Marconi (fabiomarconi)]

May 2 22:01:20 aeolus kernel: [ 36.251448] init: bluetooth main process (1160) terminated with status 1
May 2 22:01:20 aeolus kernel: [ 36.251474] init: bluetooth main process ended, respawning
May 2 22:01:20 aeolus kernel: [ 36.480190] init: bluetooth main process (1192) terminated with status 1
May 2 22:01:20 aeolus kernel: [ 36.480221] init: bluetooth main process ended, respawning
---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Fabio Marconi (fabiomarconi)]

[ 4.711204] eth1: Broadcom BCM4328 802.11 Hybrid Wireless Controller 5.100.82.38
---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Fabio Marconi (fabiomarconi)]

Hello
The disconnections happens with hight traffic, such as p2p ?

[Ryan Thompson (rct86)]

That is possible. It's difficult to notice patterns, because it typically takes me a minute or two to notice that the connection has dropped, since I might be downloading something in the background but only notice the problem when I switch to my browser and try to load a new page.

I'm not doing any p2p traffic, but I'm using a script based on youtube-dl to pre-download my subscribed videos to my hard disk so I don't have to wait for them to buffer. I have noticed a few times that running the script was followed shortly by a loss of connectivity, and the same for some other large in-browser downloads. Also I've just installed 12.04 from scratch a few days ago, so I'm still installing packages, and sometimes the apt-get download is interrupted by a connectivity loss event.

But still, there are other times when I'm not aware of any specific high traffic that I'm generating, but maybe the website that I'm loading at the time happens to have a bunch of large images or something and is generating a traffic burst.

[Fabio Marconi (fabiomarconi)]

Hello
Can you try the newest mainline kernel clicking on the following links and installing via software center ?
-http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.4-rc5-precise/linux-headers-3.4.0-030400rc5_3.4.0-030400rc5.201205011817_all.deb
-http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.4-rc5-precise/linux-headers-3.4.0-030400rc5-generic_3.4.0-030400rc5.201205011817_amd64.deb
-http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.4-rc5-precise/linux-image-3.4.0-030400rc5-generic_3.4.0-030400rc5.201205011817_amd64.deb
and refere if this fix your problem.
Thanks
Fabio

If not working, at boot phold down left shift to recall grub menu and select to boot the 3.2 kernel
---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Ryan Thompson (rct86)]

Ok, I will try testing with the new kernel. Since the bug only occurs intermittently, I will probably have to run this kernel for at least a full 24 hours of normal use in order to verify that the problem is not happening.

[Ryan Thompson (rct86)]

bcmwl module does not build on 3.4 kernel. I am attaching the DKMS make.log.

[Ryan Thompson (rct86)]

I'm about to reboot into 3.4 with an nvidia module built with this patch. We'll see how it works.

[Ryan Thompson (rct86)]

Oops, I meant that comment for another post.

However, I have generated a patch that makes the *bcmwl* module build with 3.4, and I'm about to test *that*.

[Ryan Thompson (rct86)]

Ok, I just patched the bcmwl package, and DKMS was able to build the module for kernel 3.4, and I'm now booted into 3.4 with a working wl module. We'll see how things go from here.

[Ryan Thompson (rct86)]

I just experienced exactly the same problem with the kernel 3.4 packages that you gave me. The problem is not fixed in 3.4.

[Fabio Marconi (fabiomarconi)]

Oh well, so as you are expert, can you please follow this step, in particular 3g.
Thanks

https://wiki.ubuntu.com/DebuggingNetworkManager#Getting_debug_logs
---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Fabio Marconi (fabiomarconi)]

ops, sorry, you are wireless not 3g.

---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Ryan Thompson (rct86)]

Ok, I set everything to maximum verbosity. I'm tailing syslog, dmesg, and kern.log in a terminal so I can look at them as soon as I notice a disconnection. Do you have a preference for whether I should keep using kernel 3.4 or switch back to 3.2?

[Ryan Thompson (rct86)]

After further testing over many days, I am almost certain that this problem is caused by the power-saving mode in the Broadcom drivers, which is enabled when the laptop is on battery by the script in /usr/lib/pm-utils/power.d/wireless which runs "iwconfig eth2 power on". Manually running "iwconfig eth2 power off" seems to fix the problem, as does plugging in the laptop. I replaced the pm script with one that always runs "iwconfig eth2 power off" regardless of battery status, and I have not had any major connectivity issues since then.

So, in conclusion, I think we can mark this as a duplicate of https://bugs.launchpad.net/ubuntu/+source/linux/+bug/991232

[Launchpad Janitor (janitor)]

[Expired for Ubuntu because there has been no activity for 60 days.]