Hello!

On Tue, 2007-06-26 at 18:58 -0700, Ehud Gavron wrote: 
> I've been alternately switching between 2.6.22-rc3 (Linville git tree) 
> and 2.5.22-rc6 (Linville git tree as of today), henceforth rc3 and rc6 
> respectively.

There have been significant changes in bcm43xx_mac80211 in its own git
repository (http://bu3sch.de/git/wireless-dev.git), which improved
things a lot.  As far as I know, the changes are not in 2.6.22-rc6 yet,
or you would have seen a backtrace.

> I've been alternately switching between bcm43xx and bcm43xx_mac80211, 
> henceforth soft and hard respectively.

There is nothing specifically "hard" in mac80211.  You may need to use
another convention.

> About the only thing I've found "strange" that I hope someone else can 
> confirm, is that running a tcpdump simultaneously with the DHCP client 
> causes more success than failure.

I understand you mean tcpdump on the same machine as the driver.  I
think I may have seen it.  Perhaps the promiscuous mode would speed up
recalibration, or something like that.  Anyway, the current git version
of bcm43xx_mac80211 is working much better and should not need such
tricks.

> If I had to do a matrix it would look like this:
> rc6 hard - works but stuck at 1Mbps
> rc6 soft - works
> rc6 hard - unable to communicate, but associates fine
> rc6 soft - works

You may want to use another convention for kernel versions as well.
They look too similar to me.

>           Encryption key:896A-0055-AE77-5E80-ED86-4E38-3A

I hope you understand you cannot reuse this key after having exposed it
in a public forum.  On the other hand, WEP is considered generally
unsafe these days, so it's more an polite request to leave your network
alone than actual security.

-- 
Regards,
Pavel Roskin

_______________________________________________
Bcm43xx-dev mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/bcm43xx-dev

Reply via email to