Re: 6.0-release beta1 breaks ndis for dell truemobile 1400

Ben Kaduk <minimarmot_at_gmail.com> wrote:

> I have not extensively used the ndis driver since I am not often
> around wireless networks, but
> after installing 5.4-release I compiled the driver using 
> cd /usr/src/sys/modules/ndis
> make install
> cd /usr/src/sys/modules/if_ndis
> ndiscvt -i bcmwl5a.inf -s bcmwl5.sys -o ndis_driver_data.h
> make && make install
> 
> as recommended, with the bcmwl* files being the ones here:
> http://netfiles.uiuc.edu/kaduk/www/*
> 
> The self-extracting driver (for windows) also generated a bcmwl5.inf
> (not 'a'), but ndiscvt
> produced an error whilst trying to read this file, so I used the 'a'
> version, and ndiscvt finished
> successfully; if_ndis was produced with no errors, and I was able to
> kldload if_ndis.ko and ndis.ko.  Upon doing this, console messages
> were produced recognizing my Dell TrueMobile 1400 (broadcom bcm4324
> chipset), but i didn't have a wireless network to use then.
> Since then, I have performed a source upgrade to 6.0-release beta1
> (cvsup last saturday; can't be more precise since I must boot windows
> to use wireless :(   ).
> I just today had occasion to use the wireless card, and kldload()-ing
> if_ndis resulted in a panic().  I didn't save the message for two
> reasons: (1), I thought it was just that if_ndis.ko had not been
> rebuild with world, and (2) I'm using my laptop at a friend's
> appartment with no serial console.  After performing several
> iterations of cleaning for ndis, if_ndis, and wlan, and
> rebuilding/installing, I produced if_ndis.ko, ndis.ko, and wlan.ko
> that I could successfully kldload() with no panic.  However, there
> were no console messages that appeared indicating that the system had
> found my hardward, leading to the statement in the subject that the
> beta breaks ndis for this card.  I have already posted a pciconf -lv
> to the list in this thread:
> http://docs.freebsd.org/cgi/getmsg.cgi?fetch=116619+0+current/freebsd-current
> 
> What steps can I take to help debug this regression?

At first you should try to build ndis the right way, which
was changed a while ago. Now /usr/sbin/ndisgen is used to put the
firmware in a separate module. 

<http://lists.freebsd.org/pipermail/freebsd-stable/2005-July/016762.html>


Fabian
-- 
http://www.fabiankeil.de/