ath driver on thinkpad R50p

[Matthew Luckie]

Hi

I'm running BETA7 on a new ThinkPad R50p and have the issue (which i've 
seen others having on -mobile).

http://lists.freebsd.org/pipermail/freebsd-mobile/2004-September/004833.html

Here's what is dumped by the ath driver:

ath0: <Atheros 5212> mem 0xc0210000-0xc021ffff irq 11 at device 2.0 on pci2
ath0: mac 5.6 phy 4.1 5ghz radio 1.7 2ghz radio 2.3
ath0: Ethernet address: 00:05:4e:4a:bd:34
ath0: 11a rates: 6Mbps 9Mbps 12Mbps 18Mbps 24Mbps 36Mbps 48Mbps 54Mbps
ath0: 11g rates: 1Mbps 2Mbps 5.5Mbps 11Mbps 6Mbps 9Mbps 12Mbps 18Mbps 
24Mbps 36Mbps 48Mbps 54Mbps

I'm also having trouble compiling the BETA7 code with the patch in
http://people.freebsd.org/~sam/net80211+ath-20040818.tgz

the patch applies cleanly but it doesn't compile, like

http://lists.freebsd.org/pipermail/freebsd-mobile/2004-September/004785.html

But I thought I'd try and get the 11b modes detected.  I did not succeed, 
but here's what I saw:

For the ath_rate_setup call with MODE_11A, the dot11Rate returned in 
rs_rates are 130, 132, 139, and 150.  I presume they are supposed to be 
2, 4, 10, 22

I tried hacking the rs_rates to those values (by subtracting 128 from 
dot11Rate), but that did not apparently help detect the 11b modes either.

So I looked at merging some of the extra code in ath_attach before the 
calls to ath_rate_setup, but I got stuck at the call to 
ath_hal_init_channels.  It appears that the prototype is missing the last 
parameter - xchanmode, and that appears to depend on what the hal.o 
provides.

I notice that the HAL has changed and that NetBSD has a more recent HAL, 
one with that xchanmode parameter.  FreeBSD has 0.9.6.3 while NetBSD has 
0.9.9.13 and perhaps my bug will be fixed with a more recent hal.o

I'm happy to test any patches to the ath driver, and I'm also happy to 
have a spare clue on where I might focus my efforts in getting the ath 
driver going for my particular system.  Is there demand for some grunt 
work in merging netbsd's changes back to FreeBSD, or has that mostly been 
handled for now and will be committed once 5.3 is out the door?

I know that I could use project evil, but I thought I'd try and get the 
source driver going if I could.

Thanks

Matthew