Hi Nick

According to EEPROM docs there are 10 calibration piers on all chips
after 5111 for 11a but some of them are not used and the way to test
this is if frequency is zero. Now here is a bug in the EEPROM code
that might result the wrong reads you get, check this out:

on ath5k_eeprom_read_freq_list we do this

if (!freq1)
 break;

instead of this

if (!freq1)
 continue;

That did the trick !
So reading 10 calibration piers and using the fix from above works in
terms of tx_power changes on CM9 and DCMA82 are properly used by the
hardware.

I will send a patch for that.


Greetings Thomas