You are not logged in.
I just recently upgraded my kernel to 2.6.37 and wanted to help people out there.
I have a Broadcom WNIC with the BCM43225 chipset and had some issues after the kernel upgrade.
The problems:
Random crashes. fixed
This is a locking problem with BCM43224 and BCM43225 which makes it crash.
To solve this problem, add this to /boot/grub/grub.cfg:
maxcpus=1
No WiFi after suspend. (still have this problem, while there's no support for it yet) not supported yet
Unable to join WAP on channel 13 (does anyone know why this happens?) not solved (CRDA issue?)
UPDATE!
To problem #1:
A fix for an issue in the brcm80211 softmac driver that appeared in multi CPU configurations has been submitted into the 2.6.38-rc1 kernel tree. Please refer to: https://patchwork.kernel.org/patch/504981/ . I am curious if this solves the issues that you guys are experiencing. patch can be found in git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6.git, in the staging-linus branch.
Best regards,
Roland Vossen
Broadcom
If you want to add more, please send me a message.
You can also read more here.
Last edited by litemotiv (2011-03-16 19:49:00)
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
@3: do you have crda installed and configured?
ᶘ ᵒᴥᵒᶅ
Offline
@3: do you have crda installed and configured?
No. Could it solve the problem? And why?
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
litemotiv wrote:@3: do you have crda installed and configured?
No. Could it solve the problem? And why?
Yes it could . See this: http://wireless.kernel.org/en/developer … ry_domains
ᶘ ᵒᴥᵒᶅ
Offline
tripox wrote:litemotiv wrote:@3: do you have crda installed and configured?
No. Could it solve the problem? And why?
Yes it could . See this: http://wireless.kernel.org/en/developer … ry_domains
That's odd...
Why is that not default?
I'll try again tomorrow - can't seem to get it to work right now.
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
Because it's an optdepends for the kernel, and not everybody has wireless .
Got Leenucks? :: Arch: Power in simplicity :: Get Counted! Registered Linux User #392717 :: Blog thingy
Offline
It's good that someone posted this info out there. I've been meaning to create a brcm80211 wiki page and entry on the "Wireless" page, but I was waiting for some more concrete information. I hope you don't mind if I grab this information and Wiki-fy it. Unfortunately, on a personal level, I suffer from the lock-up bug, and since I have no desire to run my kernel with only a single core, I'll still be using my Atheros USB dongle for the foreseeable future.
EDIT: Manna from heaven.
Last edited by Julius2 (2011-01-12 14:25:03)
Blog .:. AUR .:. Wiki Contributions
Registered Linux User #506070.
Offline
It's good that someone posted this info out there. I've been meaning to create a brcm80211 wiki page and entry on the "Wireless" page, but I was waiting for some more concrete information. I hope you don't mind if I grab this information and Wiki-fy it. Unfortunately, on a personal level, I suffer from the lock-up bug, and since I have no desire to run my kernel with only a single core, I'll still be using my Atheros USB dongle for the foreseeable future.
EDIT: Manna from heaven.
Looks great.
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
I haven't been able to connect to wifi-N points yet using this new driver, although i must admit i haven't really done a lot of debugging to find the cause yet too. I do see them in iwlist, but connecting times out.
ᶘ ᵒᴥᵒᶅ
Offline
I haven't been able to connect to wifi-N points yet using this new driver, although i must admit i haven't really done a lot of debugging to find the cause yet too. I do see them in iwlist, but connecting times out.
Maybe it's the same issue I have with the channels?
Try setting the access point to a channel < 13.
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
litemotiv wrote:I haven't been able to connect to wifi-N points yet using this new driver, although i must admit i haven't really done a lot of debugging to find the cause yet too. I do see them in iwlist, but connecting times out.
Maybe it's the same issue I have with the channels?
Try setting the access point to a channel < 13.
Sorry forgot to add: this is on 5Ghz so channels start from 36.
ᶘ ᵒᴥᵒᶅ
Offline
tripox wrote:litemotiv wrote:I haven't been able to connect to wifi-N points yet using this new driver, although i must admit i haven't really done a lot of debugging to find the cause yet too. I do see them in iwlist, but connecting times out.
Maybe it's the same issue I have with the channels?
Try setting the access point to a channel < 13.Sorry forgot to add: this is on 5Ghz so channels start from 36.
It's not because it runs on 40 MHz which is not supported by the driver yet?
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
litemotiv wrote:tripox wrote:Maybe it's the same issue I have with the channels?
Try setting the access point to a channel < 13.Sorry forgot to add: this is on 5Ghz so channels start from 36.
It's not because it runs on 40 MHz which is not supported by the driver yet?
Good one, it would make sense that a dualband (5Ghz) router also uses 40Mhz channels...
ᶘ ᵒᴥᵒᶅ
Offline
tripox wrote:litemotiv wrote:Sorry forgot to add: this is on 5Ghz so channels start from 36.
It's not because it runs on 40 MHz which is not supported by the driver yet?
Good one, it would make sense that a dualband (5Ghz) router also uses 40Mhz channels...
That's what I thought although I don't know it.
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
I am using brcm80211 package from AUR, which uses the kernel staging branch open source driver. My version of kernel26 package is 2.6.26.3-1.
The kernel package is updated a couple of days ago. The wireless card was usable during several reboots since then. However, after a reboot this afternoon, the driver makes kernel panic when associating.
I am using wicd to manage wireless networks. If I have a wired connection linked to my computer, it can boot into gdm and establish a wired connection successfully, and I can see the wireless networks available in wicd interface. But if I try to connect to a wireless network, I get a kernel panic when associating.
The wireless network has not been changed lately, and restarting the router makes no difference. I have installed and uninstalled several font packages since the kernel update, which seems irrelevant. I don't know what causes this sudden failure. Using closed-source broadcom-wl package from AUR made wicd unable to find wlan0 interface.
Any idea why this happens?
Last edited by AugustePop (2011-01-13 12:30:54)
Offline
Merged AugustePop's thread with the brcm80211 thread.
ᶘ ᵒᴥᵒᶅ
Offline
I am using brcm80211 package from AUR, which uses the kernel staging branch open source driver. My version of kernel26 package is 2.6.26.3-1.
The kernel package is updated a couple of days ago. The wireless card was usable during several reboots since then. However, after a reboot this afternoon, the driver makes kernel panic when associating.
I am using wicd to manage wireless networks. If I have a wired connection linked to my computer, it can boot into gdm and establish a wired connection successfully, and I can see the wireless networks available in wicd interface. But if I try to connect to a wireless network, I get a kernel panic when associating.
The wireless network has not been changed lately, and restarting the router makes no difference. I have installed and uninstalled several font packages since the kernel update, which seems irrelevant. I don't know what causes this sudden failure. Using closed-source broadcom-wl package from AUR made wicd unable to find wlan0 interface.
Any idea why this happens?
broadcom-wl from AUR uses eth1
Have you tried brcm80211 with maxcpus=1 in your grub.cfg?
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
Have you tried brcm80211 with maxcpus=1 in your grub.cfg?
Adding maxcpus=1 in grub do get a working wireless network. I thought BCM4313 is stable according to a widely referenced linux wireless page.
What makes me puzzled is that the same kernel and driver worked before, and seems no apparent reason for this breakage.
Offline
tripox wrote:Have you tried brcm80211 with maxcpus=1 in your grub.cfg?
Adding maxcpus=1 in grub do get a working wireless network. I thought BCM4313 is stable according to a widely referenced linux wireless page.
What makes me puzzled is that the same kernel and driver worked before, and seems no apparent reason for this breakage.
From my experiences broadcom-wl is the most stable driver so far.
Just remember it uses ethX instead of wlanX.
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
From my experiences broadcom-wl is the most stable driver so far.
Just remember it uses ethX instead of wlanX.
I stopped using broadcom-wl because after some update, I can not connect to my office wireless network with that. It just failed to authorize. I was using networkmanager that time.
What's really strange is that after last successful connection with maxcpus=1 bootarg, I am able to connect to wireless network without kernel panic again. I will reboot for another couple of times to check whether this behavior is persistent.
==EDIT==
After two consecutive successful reboot and connection, I think the driver is working again now. I am now booting without the maxcpus=1 bootarg. So, basically, the driver suddenly makes kernel panic every time it associates (for at least 5 successive tries), adding maxcpus=1 in grub makes it work, and then it will work again with the aforementioned bootarg removed.
Last edited by AugustePop (2011-01-13 15:47:12)
Offline
tripox wrote:From my experiences broadcom-wl is the most stable driver so far.
Just remember it uses ethX instead of wlanX.I stopped using broadcom-wl because after some update, I can not connect to my office wireless network with that. It just failed to authorize. I was using networkmanager that time.
What's really strange is that after last successful connection with maxcpus=1 bootarg, I am able to connect to wireless network without kernel panic again. I will reboot for another couple of times to check whether this behavior is persistent.
==EDIT==
After two consecutive successful reboot and connection, I think the driver is working again now. I am now booting without the maxcpus=1 bootarg. So, basically, the driver suddenly makes kernel panic every time it associates (for at least 5 successive tries), adding maxcpus=1 in grub makes it work, and then it will work again with the aforementioned bootarg removed.
I have actually experienced a few kernel panics afterwards, however I'm not sure when that happens.
My WAP disconnects my clients once in a while (for some reason) and I think that's where it crashes.
Simplicity is the ultimate sophistication.
Maintainer of 'xbattbar-acpi' a simple single-line battery indicator - try it!
Offline
I have actually experienced a few kernel panics afterwards, however I'm not sure when that happens.
My WAP disconnects my clients once in a while (for some reason) and I think that's where it crashes.
I am having a quite steady connection to the wireless network. Actually, the router is just sitting beside me. When I had a kernel panic, it is always the first time I tried to connect to a wireless network. And this failure happens every time I tried to connect before I added maxcpus=1 to the bootarg. If the driver makes a kernel panic again, I will consider switching back to broadcom-wl. Since I think that recompiling a module every major kernel update is quite against the "rolling" feature of Arch, I personally prefer drivers that come with the kernel.
I have not tried a .37 kernel yet. I thought the brcm80211 driver is still in the staging tree.
Offline
Also get problem that my wireless led dont turn on (Dell Vostro 3500)
Offline
tripox wrote:Have you tried brcm80211 with maxcpus=1 in your grub.cfg?
Adding maxcpus=1 in grub do get a working wireless network. I thought BCM4313 is stable according to a widely referenced linux wireless page.
What makes me puzzled is that the same kernel and driver worked before, and seems no apparent reason for this breakage.
I have BCM4313, and it didn't work with broadcom-wl, ndiswrapper, b43, anything. Brcm80211 works with the maxcpus=1 line, but I don't want to sacrifice 3 of my cores just for that. Sadly, the information by the devs is probably correct -- BCM4313 does likely have the best support.
Blog .:. AUR .:. Wiki Contributions
Registered Linux User #506070.
Offline
What kernel panic message are you guys getting? Here are a couple of mine:
http://farm6.static.flickr.com/5202/533 … 5edd_b.jpg
http://farm6.static.flickr.com/5246/534 … a55e_b.jpg
I reported the bug:
Offline