That sure sounds more likely to be related to fixing the problem. How do I see your changes and know where to change the source code I have to test it out?
Regards, Kurt ------------- Date: Wed, 2 Apr 2008 21:29:19 +0200 From: Michael Buesch <[EMAIL PROTECTED]> Subject: Re: Range issue using b43 legacy To: [email protected] Message-ID: <[EMAIL PROTECTED]> Content-Type: text/plain; charset="iso-8859-1" On Wednesday 02 April 2008 21:12:34 krop wrote: >Le Wednesday 02 April 2008 20:50:59 Larry Finger, vous avez ?crit : > > > > The rev 02 BCM4306 cards that use b43legacy had much more of their > > initialization done from the host and less done by the hardware (or by > > the firmware). As a result, the performance is much more dependent on > > the specs as determined by the RE team. In particular, OFDM rates are > > quite low. I plan to revisit this code from the Broadcom drivers to > > see if it can be improved. > > > > As an intermediate test, does bcm43xx work better than b43legacy? If > > so, I will check what might have changed. > > > > Larry > >Could the problem I reported a few days ago with a bcm4306 rev. 03 be >related ? (I can't physically put my computer close to my DSL box/wifi >router to check). > >Someone on Ubuntu bug tracker reported something similar : >https://bugs.launchpad.net/ubuntu/+bug/184976/comments/24 > >[quote] >Uninstalled bcm43xx-fwcutter and b43-fwcutter. Then installed b43-fwcutter >and at the end loaded down and installed the b43-firmware. WLAN works but >only right next to my router. When the distance was higher (about 10m) then >the laptop wasn't able to connect. I found out that the signal was shown >with a quality of 85-95 % in the network-manager when the Laptop was right >next to my router. >[endquote] it could actually also be a bug in the code that sets TX power. I reviewed that in b43 a few months ago and found lots of bugs. In particular the scaling algorithms were wrong. One might want to port the stuff from b43_phy_xmitpower() over to legacy. The old code turned out to interpret the estimated TX power in a wrong way and based the actual attenuation calculations on that. The attenuation calculations were also wrong. I rewrote them and put them into a seperate function that's called from b43_phy_xmitpower(). You might start with adding a few printks into the code and see what the estimated TX power value actually does. If it bounces heavily, it's clearly wrong. It should only have a hysteresis of 0.5 to 1 dB. The idle-tssi value is also a major parameter for TX power, so you might also check if that's right. -- Greetings Michael. _______________________________________________ Bcm43xx-dev mailing list [email protected] https://lists.berlios.de/mailman/listinfo/bcm43xx-dev
