weird SDIO interrupt and SDIO mechanism

[Holger Schurig]

> This testing is using infrastructure mode and 'nuttcp' for
> traffic source/sink on both ends. I get the same results no
> matter what I set the tx rate to (which makes me think
> something is not right with the libertas driver's tx rate as I
> still get 1.657mbps when iwconfig shows tx rate at 1M).

iwconfig shows the RECEIVE rate. Whenever we receive a TX packet, 
we store it's rate in priv->cur_rate, which is then used in 
lbs_get_rate() for SIOCGIWRATE.

If you want to verify the TX rate, you'll either need to check at 
the other side. Or you'll take a third computer, put it into 
monitor mode, set it to the right channel and start wireshark. 
In the radiotap header you can than see the TX rate of each 
packet.

> The results from the two different SDIO modules are very
> curious as: 1. the max datarate is abysmal
> 2. the devices are both using Marvell 8686 - why do they
> differ so much... almost exactly 2x difference

Maybe one has a good antenna and the other one a bad antenna?