[driverloader] Similar multiple problems dl2.24 / SuSE9.2 amd64 / bcm4306

Eric Herget eherget at bellsouth.net
Sat Mar 5 11:16:47 EST 2005


Hi Jonathan,

Yes, this one is a strange one.  You mention not being able to scan for 
an access point.  I haven't had that problem - possibly this was the 
other person who posted a similar problem with keyboard input being 
messed up.

I've checked my kernel config and I do not have CONFIG_PREEMPT set.  I 
cannot even find CONFIG_4KSTACKS in my config so it is presumably also 
unset.

I also put back my original memory configuration and, surprisingly, 
everything works as expected.  So I'll sum up where we're at so far:

- Common config -
SuSE9.2
AMD64
HP Pavilion zv5340us notebook
Broadcom BCM4306 802.11b/g Wireless LAN Controller (rev 03)
bcmwl564 driver (as linked to from your website)

- In chronological order -
Got driverloader working with 64-bit Broadcom driver linked from you 
website on kernel v 2.6.8-14.11 with 512MB memory in notebook (2 x 
256MB  memory modules).

Changed out 1  256MB memory module for 1GB memory module still on kernel 
v 2.6.8-14.11 and now system hangs if wireless device is activated.  
Hangs at boot when starting driverloader if wireless card is active at 
boot time.  If I activate the wireless card using button above keyboard 
after system has booted, the system hangs immediately (no network 
access, etc).  In both cases hard reboot is necessary.  Did not see any 
related messages in dmesg output or /var/log/messages.

Dl'd, compiled and installed kernel 2.6.11, did dldrconfig -k, etc.  
with same memory config as above (1x  256MB and 1x 1GB memory modules).  
If wireless device is activated, "kdldrd/dpc" process eats up all 
available CPU and keyboard input behavior is strange.  Deactivating 
wireless device seems to put system back in its normal state - normal 
CPU usage and keyboard input is normal.   I have not looked at dmesg 
output or /var/log/messages in this case - that's next.

Put memory configuration back to 2x256MB memory modules, still with 
2.6.11 kernel and all works as expected.


I'm now gonna put the memory back to 1x256MB and 1x1GB configuration and 
look for anything in dmesg output and/or /var/log/messages.  Then I'll 
dl driverloader 2.25 and see if that has any affect.

Out of curiosity, where did the 64-bit broadcom drivers come from?  Are 
they beta version drivers?

Eric


Linuxant support (Jonathan) wrote:

> Hi,
>
> this is s strange problem, when you are unable to scan for an access 
> point and the characters which are repeated, it could be an interrupt 
> related problem.
>
> Eric, if you put back the memory setup you had before with 
> DriverLoader does it fix the problem?
>
> You could also check that you kernel was not compiled with one of the 
> following options:
>
> ---
> CONFIG_4KSTACKS
> CONFIG_PREEMPT
> ---
>
> If your kernel each compiled with one of these options enabled, please 
> disable it, re-compile the kernel and finally re-compile the 
> DriverLoader module with the 'dldrconfig -k' command in a root shell.
>
> You could also try to use your wireless card under Windows XP to see 
> if you have a similar problem.
>
> Regards,
>
>
> Jonathan
> Technical specialist / Linuxant
> www.linuxant.com
> support at linuxant.com
>
>
> eherget at bellsouth.net wrote:
>
>> I, too, have similar problems.
>>
>> I have a Pavilion zv5340us notebook - AMD64, Broadcom 802.11 
>> built-in, etc.
>> I had written before about problem with the button at the top of the 
>> keyboard that activates/deactivates the wireless device.  I had 
>> gotten it all working nicely...
>>
>> ...but then I added more memory.  The notebook has 2 memory "slots", 
>> one is easily reachable behind an access panel, the other requires 
>> some disassembly to get to the slot under the keyboard.  I replaced 
>> the easily reachable 256MB memory module with a 1GB module - the 
>> other slot still has a 256MB module.
>>
>> Since then, my machine would hang when starting Driverloader during 
>> boot if the device was activated (device activation state is 
>> controlled by the button above the keyboard and is persistent between 
>> reboots/power off-on).  This on kernel 2.6.8-24.11.  If the device is 
>> off during boot the machine comes up fine, but pressing the button to 
>> activate the wireless device causes the system to lockup.
>>
>> Yesterday I dl'd the newest kernel 2.6.11 and with the fix mentioned 
>> in a post earlier today, I tried driverloader with it.
>>
>> Now, instead of locking up, the system has its CPU pegged at ~100%.  
>> But I was able to see that a process called "kdldrd/dpc" was eating 
>> up all this usage.  As with the most recent post that I've forwarded 
>> here, the keyboard input is strange.  In my case, same letters typed 
>> with other letters in between got jumbled.  For example, I would type 
>> "dldrconfig" and what showed up on screen a few seconds later was 
>> "ddlrconfig".  Or  "10802" became "10082".   Unlike the post I 
>> forwarded, mine is a single CPU system.
>>
>> With the 2.6.11 kernel, I could press the button to deactivate the 
>> wireless device and the system would go back to normal (CPU usage OK 
>> and keyboard input OK).  On the 2.6.8-24.11 kernel, once the system 
>> locked up, deactivating the wireless device did not free up the system.
>>
>> Eric
>>
>>
>>
>>> From: "Martin A. Brooks" <martin at hinterlands.org>
>>> Date: 2005/03/03 Thu AM 11:03:36 EST
>>> To: driverloader at lists.linuxant.com
>>> Subject: [driverloader] multiple problems dl2.24 / debian amd64 / 
>>> Belkin
>>>     bcm4306 based card
>>>
>>> Hi
>>>
>>> I'm experiencing three problems with driverloader 2.24.  Firstly, 
>>> what does work:
>>>
>>> Driverloader compiles for my kernel (2.6.10) and allows me to upload 
>>> the driver for the wireless card.  The wireless card is detected 
>>> correctly and appears as eth1, I am able to use ifconfig to 
>>> configure the interface.
>>>
>>> My problems are:
>>>
>>> 1) I cannot reach my accesspoint. Even with the correct ESSID set 
>>> the AP is not detected by the iw utilities.  I am not able to ping 
>>> known up and known working hosts on the network.
>>>
>>> 2) The kernelspace driverloader process consumes 100% of one CPU  (I 
>>> have a dual opteron 242). This hits performance quite a lot, as you 
>>> can imagine.
>>>
>>> 3) With driverloader in residence I get keyboard glitches under X  
>>> (not under the console).  Typing "martin at hinterlands.org" for 
>>> example, will usually look something like 
>>> "marrrrrrrrrrrrrrrrrrrrtin at hinttttttttttttttttttttterlaaaaaaaaands.orrrrrrrrrg".  
>>> If I unload the driverloader module then all is well.
>>>
>>> Any hints appreciated
>>>
>>> Regards
>>>
>>> Martin A. Brooks
>>> _______________________________________________
>>> driverloader mailing list
>>> driverloader at lists.linuxant.com
>>> https://www.linuxant.com/mailman/listinfo/driverloader
>>>
>>
>>
>>
>> _______________________________________________
>> driverloader mailing list
>> driverloader at lists.linuxant.com
>> https://www.linuxant.com/mailman/listinfo/driverloader
>
>


More information about the driverloader mailing list