Re: ndiswrapper and kernel stack size

Hmm.. that's an interesting thought too..  I'm using it on my new compaq 
laptop, and I probably did do a kernel rebuild, since it's an 
athlon64..  so it's entirely likely that I changed the stack size when I 
built it, since rei and arcticfox had pointed out that the 4k size was 
problematic for athlon systems.

BUT.. my builtin broadcom card works nicely now.  Just need to get it to 
where it works 'out of the box' (not sure how we can do that short of 
adding it to rei's aftwizard)

Sergio Visinoni wrote:

>>In my experience.. the hang is because the ndiswrapper module belongs to
>>a different version of ndiswrapper than the version of the ndiswrapper
>>package installed.  The kernel module was in need of updating last time
>>I checked.  (manal build of ndiswrapper + install allows it to work).
>>    
>>
>
>I have to check again when I have the time but I'm pretty sure I did a test 
>with building ndiswrapper CVS against our kernel (both userland tools and 
>kernel module) and got the same problem. That's why I looked at 
>documentation/FAQ and foud this:
>
>"""
>If you are using 2.6 kernels, make sure 4K stacks are not used (i.e., disable 
>CONFIG_4KSTACKS). (Q: Is this a cause of system freezes? A: On the Webpage of 
>the commercial Linuxant Driverloader I found the information that even with 
>8kb (=4k off) the system may freeze because some Windows-Drivers (in 
>particular for the Intel BG 2200) needs 12kb. On their Webpage they are 
>offering a 12k patch for the 2.6.7-kernel.) Note that Fedora kernels actually 
>disable this option, so you won't see it in kernel configuration, so if you 
>are using Fedora kernel, you should either get the patch from Linuxant or 
>compile a vanilla kernel without 4K stacks.
>"""
>
>[ From http://ndiswrapper.sourceforge.net/phpwiki/index.php/FAQ ]
>
>Actually, as I said, I'm running a vanilla kernel but as soon as I have some 
>time to build new kernels I'll try ndiswrapper + ark's kernel with just 
>CONFIG_4KSTACKS turned off to check if there is anything else that may cause 
>kernel oopses/hang.
>
>ciuss,
>s
>  
>