[driverloader] Re: [LinuxR3000] Broadcom 54g + FC3 + 1.5 G = 100%

[James Sparenberg]

On Thu, 2005-03-24 at 09:04, eherget at bellsouth.net wrote:
> Forgot to mention...
> 
> I can get the system to work properly by doing the following:
>  - boot my laptop with wireless turned off (it "remembers" its state from previous session)
>  - after the system boots, as root, run `dldrconfig --kernel` to rebuild
>  - then hit the button above the keyboard to turn wireless on
>  - then restart the network - `/etc/rc.d/network restart`
> 
> Eric

If I'm understanding you right then perhaps it's not a driver
problem/kernel problem so much as an initscript timing problem.  Where
either ndiswrapper or linuxant is starting at the wrong time in the
cycle or in the wrong order causing a conflict or lockup.  My thought is
that perhaps it is "slower" to start and something else that comes after
it is trying to grab the same resources.  Older kernels couldn't handle
the conflict and locked.  The newer ones can prevent the lock but only
partially.  

Note too that in M$ products the initiallization of the network is
actually not done until a user starts to log in.  So by that point all
system services and other hardware initialization is completed.  

James

> 
> 
> > 
> > From: <eherget at bellsouth.net>
> > Date: 2005/03/24 Thu AM 11:57:14 EST
> > To: <driverloader at lists.linuxant.com>
> > Subject: Re: [driverloader] Re: [LinuxR3000] Broadcom 54g + FC3 + 1.5 G = 100%
> > 
> > Hi Prarit,
> > 
> > I have also reported this problem a few weeks ago.  I have a HP Pavilion zv5340us laptop with Broadcom 54g card.  In my case, all worked fine with 500MB of memory (2x256MB memory modules), but I got the 100%CPU usage issue when I replaced one 256MB module with a 1GB memory module.  I've even tried moving the 1GB module to the 2nd memory slot (requires some dis/reassembly of the laptop to get to it under the keyboard).
> > 
> > You mentioned that you think its a problem with the driver itself.  That may be a possibility, as I expect the 64-bit Broadcom driver to be in some sort of pre-release state.  And HP likely modifies the driver for its use.  For example, there's a lighted button above my keyboard to turn the wireless feature on and off.  In 32-bit windows using drivers provided by HP, the light is on constant when the wireless device is on.  In 64-bit windows with the same driver we're using in driverloader, the light only flickers as data is being tranmitted/received.  I suspect (hope) HP will modify this once they officially support 64-bit Windows to have the light stay on constant when the wireless feature is on.
> > 
> > In any case, there are other indications that the problem may not be in the driver.  Under earlier versions of the linux kernel (2.6.8-24.11 - from updated SuSE 9.1), my system would lockup completely instead of just hogging the CPU.  I had to hard cycle the laptop to recover.  In addition, the driver works fine under 64-bit Windows (the latest beta version available at M$ website) with all of my different memory configs.
> > 
> > I was originally thinking the issue might have something to do with unequal memory modules in the 2 memory slots and some interaction between the linux kernel and the memory hardware/controller.  However, if you've got 2GB of memory, I assume you have equal (1GB each) memory modules in the 2 slots.
> > 
> > Since the behavior is different between linux kernel versions <= 2.6.8-24.11 vs. >= 2.6.10xxx, at least part of the problem may lie with the kernel itself.  There could even be some hardware issues (timing, BIOS, etc) involved.
> > 
> > Eric
> > 
> > 
> > > 
> > > From: Prarit Bhargava <prarit at sgi.com>
> > > Date: 2005/03/24 Thu AM 10:46:01 EST
> > > To: "Linux on AMD64 laptops (Compaq R3000 and HP zv5000)" <linuxr3000 at lists.pcxperience.com>, 
> > > 	driverloader at lists.linuxant.com
> > > Subject: [driverloader] Re: [LinuxR3000] Broadcom 54g + FC3 + 1.5 G = 100%
> > > 	CPU usage?
> > > 
> > > Good point.  The fact that it's happening both on ndiswrapper and 
> > > Linuxant's driverloader indicates an issue with the driver itself.
> > > 
> > > Also, another thought occurs to me .... My laptop is BRAND NEW (1 week 
> > > old).  They ship these things with up to 2 G memory ... wouldn't you 
> > > think that HP/Compaq would have seen this problem already? 
> > > 
> > > P.
> > > 
> > > P. Scott DeVos wrote:
> > > 
> > > > Assuming the problem isn't with the driver itself...
> > > 
> > > _______________________________________________
> > > 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
> > 
> 
> 
> _______________________________________________
> driverloader mailing list
> driverloader at lists.linuxant.com
> https://www.linuxant.com/mailman/listinfo/driverloader