[opensuse-factory] broadcom-wl and kernel 3.6
Hi there, I have stated the kernel in the title but that does not mean I imply is kernel's fault. on my system using a broadcom wireless device (which worked pretty well in the past) with the current factory (I got it via a zypper dup from a clean 12.2 install) stopped working... the module loads... the connection seems to be up the only thing is I see no networks. digging into I see grep eth1 /var/log/warn Sep 21 08:54:24 abbaton udevd[285]: seq 1823 '/devices/pci0000:00/0000:00:16.0/0000:03:00.0/net/eth1' killed Sep 21 08:54:25 abbaton NetworkManager[618]: <error> [1348214065.75959] [nm- device-wifi.c:2673] real_update_permanent_hw_address(): (eth1): unable to read permanent MAC address (error 0) plus the usual complains about not being able to scan... [alin@abbaton:~]: uname -a Linux abbaton.ucd.ie 3.6.0-rc6-1-desktop #1 SMP PREEMPT Mon Sep 17 13:57:42 UTC 2012 (d6432fe) x86_64 x86_64 x86_64 GNU/Linux Any ideas? Alin -- Without Questions there are no Answers! ______________________________________________________________________ Alin Marin ELENA Advanced Molecular Simulation Research Laboratory School of Physics, University College Dublin ---- Ardionsamblú Móilíneach Saotharlann Taighde Scoil na Fisice, An Coláiste Ollscoile, Baile Átha Cliath ----------------------------------------------------------------------------------- http://alin.elenaworld.net ______________________________________________________________________ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Alin M Elena wrote:
Hi there,
I have stated the kernel in the title but that does not mean I imply is kernel's fault.
on my system using a broadcom wireless device (which worked pretty well in the past) with the current factory (I got it via a zypper dup from a clean 12.2 install) stopped working... the module loads... the connection seems to be up the only thing is I see no networks.
digging into I see grep eth1 /var/log/warn Sep 21 08:54:24 abbaton udevd[285]: seq 1823 '/devices/pci0000:00/0000:00:16.0/0000:03:00.0/net/eth1' killed Sep 21 08:54:25 abbaton NetworkManager[618]: <error> [1348214065.75959] [nm- device-wifi.c:2673] real_update_permanent_hw_address(): (eth1): unable to read permanent MAC address (error 0)
plus the usual complains about not being able to scan... [alin@abbaton:~]: uname -a Linux abbaton.ucd.ie 3.6.0-rc6-1-desktop #1 SMP PREEMPT Mon Sep 17 13:57:42 UTC 2012 (d6432fe) x86_64 x86_64 x86_64 GNU/Linux
I have a broadcom 1Gb ethernet card in my server that effectively goes dead in 3.3.x and works in 3.2.28. By 'dead', I mean netlink claims it does not see the device. However, it is seen by "ip" when I list links (but "ip" can't set it up or set an address on it, as I get a message that netlink thinks there is no such device). In 3.4, my Intel cards stop being able to operate w/jumbo packets.. I run 9k packets internally, but with 3.4.?, 'ip' was no longer able to set the packet length on these cards. I have yet to find a solution -- I posted the question to the kernel list but got no response. Maybe broadcom is no longer supported? Seems unlikely, but anything is possible. I find it a bit odd that I hadn't seen others with problems, but -- what was the last kernel you used before 3.6? Weird..never seen symptoms like this before... -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On 09/21/2012 04:31 PM, Linda Walsh wrote:
Alin M Elena wrote:
Hi there,
I have stated the kernel in the title but that does not mean I imply is kernel's fault.
on my system using a broadcom wireless device (which worked pretty well in the past) with the current factory (I got it via a zypper dup from a clean 12.2 install) stopped working... the module loads... the connection seems to be up the only thing is I see no networks.
digging into I see grep eth1 /var/log/warn Sep 21 08:54:24 abbaton udevd[285]: seq 1823 '/devices/pci0000:00/0000:00:16.0/0000:03:00.0/net/eth1' killed Sep 21 08:54:25 abbaton NetworkManager[618]: <error> [1348214065.75959] [nm- device-wifi.c:2673] real_update_permanent_hw_address(): (eth1): unable to read permanent MAC address (error 0)
plus the usual complains about not being able to scan... [alin@abbaton:~]: uname -a Linux abbaton.ucd.ie 3.6.0-rc6-1-desktop #1 SMP PREEMPT Mon Sep 17 13:57:42 UTC 2012 (d6432fe) x86_64 x86_64 x86_64 GNU/Linux
I have a broadcom 1Gb ethernet card in my server that effectively goes dead in 3.3.x and works in 3.2.28.
By 'dead', I mean netlink claims it does not see the device.
However, it is seen by "ip" when I list links (but "ip" can't set it up or set an address on it, as I get a message that netlink thinks there is no such device).
In 3.4, my Intel cards stop being able to operate w/jumbo packets.. I run 9k packets internally, but with 3.4.?, 'ip' was no longer able to set the packet length on these cards.
I have yet to find a solution -- I posted the question to the kernel list but got no response.
Maybe broadcom is no longer supported? Seems unlikely, but anything is possible.
I find it a bit odd that I hadn't seen others with problems, but -- what was the last kernel you used before 3.6?
Weird..never seen symptoms like this before...
You should do what many of us do when there is a regression. We download the git tree for mainline and use 'git bisect' to find the bad commit. Larry -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Dear Linda and Larry, Linda... I am sorry half of the things you said are greek to me... iirc some early rc of 3.6 was ok... Larry... before I get dirty with commits in kernel.. I like to get an analytic approach and have the issue confirmed and be sure is not an observer effect. Alin On 21 September 2012 23:14, Larry Finger <Larry.Finger@lwfinger.net> wrote:
On 09/21/2012 04:31 PM, Linda Walsh wrote:
Alin M Elena wrote:
Hi there,
I have stated the kernel in the title but that does not mean I imply is kernel's fault.
on my system using a broadcom wireless device (which worked pretty well in the past) with the current factory (I got it via a zypper dup from a clean 12.2 install) stopped working... the module loads... the connection seems to be up the only thing is I see no networks.
digging into I see grep eth1 /var/log/warn Sep 21 08:54:24 abbaton udevd[285]: seq 1823 '/devices/pci0000:00/0000:00:16.0/0000:03:00.0/net/eth1' killed Sep 21 08:54:25 abbaton NetworkManager[618]: <error> [1348214065.75959] [nm- device-wifi.c:2673] real_update_permanent_hw_address(): (eth1): unable to read permanent MAC address (error 0)
plus the usual complains about not being able to scan... [alin@abbaton:~]: uname -a Linux abbaton.ucd.ie 3.6.0-rc6-1-desktop #1 SMP PREEMPT Mon Sep 17 13:57:42 UTC 2012 (d6432fe) x86_64 x86_64 x86_64 GNU/Linux
----
I have a broadcom 1Gb ethernet card in my server that effectively goes dead in 3.3.x and works in 3.2.28.
By 'dead', I mean netlink claims it does not see the device.
However, it is seen by "ip" when I list links (but "ip" can't set it up or set an address on it, as I get a message that netlink thinks there is no such device).
In 3.4, my Intel cards stop being able to operate w/jumbo packets.. I run 9k packets internally, but with 3.4.?, 'ip' was no longer able to set the packet length on these cards.
I have yet to find a solution -- I posted the question to the kernel list but got no response.
Maybe broadcom is no longer supported? Seems unlikely, but anything is possible.
I find it a bit odd that I hadn't seen others with problems, but -- what was the last kernel you used before 3.6?
Weird..never seen symptoms like this before...
You should do what many of us do when there is a regression. We download the git tree for mainline and use 'git bisect' to find the bad commit.
Larry
-- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On 09/21/2012 05:50 PM, Alin Marin Elena wrote:
Dear Linda and Larry,
Linda... I am sorry half of the things you said are greek to me... iirc some early rc of 3.6 was ok...
Larry... before I get dirty with commits in kernel.. I like to get an analytic approach and have the issue confirmed and be sure is not an observer effect.
Of course you may do what you wish. One thing I should have said earlier is that the kernel devs will have little sympathy for the breakage in a proprietary driver, particularly one from Broadcom that has not given much support for their wireless devices. Larry -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Larry Finger wrote:
On 09/21/2012 05:50 PM, Alin Marin Elena wrote:
Dear Linda and Larry,
Linda... I am sorry half of the things you said are greek to me... iirc some early rc of 3.6 was ok...
Larry... before I get dirty with commits in kernel.. I like to get an analytic approach and have the issue confirmed and be sure is not an observer effect.
Of course you may do what you wish. One thing I should have said earlier is that the kernel devs will have little sympathy for the breakage in a proprietary driver, particularly one from Broadcom that has not given much support for their wireless devices.
Mine is in the "wired" driver...is that proprietary as well? Also the problem in 3.4 was in the intel e1000 driver. Is that also proprietary? I've been holding off as I've had a full plate, but was thinking exactly about what you said a week or so ago... given my facility with doing such, it's right up there with getting cavities filled (but definitely above root canals)... -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On 09/22/2012 01:30 PM, Linda Walsh wrote:
Larry Finger wrote:
On 09/21/2012 05:50 PM, Alin Marin Elena wrote:
Dear Linda and Larry,
Linda... I am sorry half of the things you said are greek to me... iirc some early rc of 3.6 was ok...
Larry... before I get dirty with commits in kernel.. I like to get an analytic approach and have the issue confirmed and be sure is not an observer effect.
Of course you may do what you wish. One thing I should have said earlier is that the kernel devs will have little sympathy for the breakage in a proprietary driver, particularly one from Broadcom that has not given much support for their wireless devices.
Mine is in the "wired" driver...is that proprietary as well?
Also the problem in 3.4 was in the intel e1000 driver. Is that also proprietary?
I've been holding off as I've had a full plate, but was thinking exactly about what you said a week or so ago... given my facility with doing such, it's right up there with getting cavities filled (but definitely above root canals)...
Of course not, but the subject of this thread contains the phrase broadcom-wl, which is proprietary. Larry -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Larry Finger wrote:
Weird..never seen symptoms like this before...
You should do what many of us do when there is a regression. We download the git tree for mainline and use 'git bisect' to find the bad commit.
Have done that before, but doing it for this -- it's on my main server -- and each test point is a few minutes at least... All my local computers lose internet access when that computer goes down. Sorta a pain. If only it would happen in a VM...sigh. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Friday 2012-09-21 23:31, Linda Walsh wrote:
I have a broadcom 1Gb ethernet card in my server that effectively goes dead in 3.3.x and works in 3.2.28. By 'dead', I mean netlink claims it does not see the device. However, it is seen by "ip" when I list links (but "ip" can't set it up or set an address on it, as I get a message that netlink thinks there is no such device).
Is this perhaps a case where a firmware file is needed, but not available? (There should be a message in dmesg, and if a FW file is really needed, the "kernel-firmware" would be the place to look in.) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Jan Engelhardt wrote:
On Friday 2012-09-21 23:31, Linda Walsh wrote:
I have a broadcom 1Gb ethernet card in my server that effectively goes dead in 3.3.x and works in 3.2.28. By 'dead', I mean netlink claims it does not see the device. However, it is seen by "ip" when I list links (but "ip" can't set it up or set an address on it, as I get a message that netlink thinks there is no such device).
Is this perhaps a case where a firmware file is needed, but not available? (There should be a message in dmesg, and if a FW file is really needed, the "kernel-firmware" would be the place to look in.)
Bingo. So why does it need a firmware file post in 3.3 & higher? Or does it -- and if 'it', needs it, why isn't it 'there' (for some value of 'it', 'there')? I know that used to be a kernel op -- to only include drivers not needing firmware... I don't recall changing that option (I don't how it was set, but will look there... )... Weird...found it on my first kernel dissect (message in log)... Was looking for cause of a crash -- Last thing in log was: Oct 4 13:52:50 Ishtar kernel: [985735.911825] INFO: task fetchmail:25872 blocked for more than 120 seconds. Oct 4 13:52:50 Ishtar kernel: [985735.918777] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. Then the whole computer hung. ;-/ Always fun here!! -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Thursday 2012-10-04 23:35, Linda Walsh wrote:
Jan Engelhardt wrote:
On Friday 2012-09-21 23:31, Linda Walsh wrote:
I have a broadcom 1Gb ethernet card in my server that effectively goes dead in 3.3.x and works in 3.2.28. By 'dead', I mean netlink claims it does not see the device. However, it is seen by "ip" when I list links (but "ip" can't set it up or set an address on it, as I get a message that netlink thinks there is no such device).
Is this perhaps a case where a firmware file is needed, but not available? (There should be a message in dmesg, and if a FW file is really needed, the "kernel-firmware" would be the place to look in.)
Bingo.
So why does it need a firmware file post in 3.3 & higher? Or does it -- and if 'it', needs it, why isn't it 'there' (for some value of 'it', 'there')?
The kernel-firmware package is a collection of out-of-tree firmware from various vendors. You mention "broadcom-wl" in the subject, which is a kernel module name I have not seen in the original upstream kernel. So this smells terribly like something proprietary evil. So, none of this actually looks like an openSUSE problem. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
participants (5)
-
Alin M Elena -
Alin Marin Elena
-
Jan Engelhardt -
Larry Finger -
Linda Walsh