This patch series fix a wrong power gain calibration in ath5k for entire 5GHz band and enables consistent usage of tx_power specified by the user. (1) Function ath5k_eeprom_read_target_rate_pwr_info() read 10 lines of rate_pcal_info[i].target_power which worngly produces e.g. from Winstron CM9: [ 38.988000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[0].freq = 4920, rate_pcal_info[0].target_power_6to24 = 36, rate_pcal_info[0].target_power_54 = 24 [ 39.004000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[1].freq = 5040, rate_pcal_info[1].target_power_6to24 = 36, rate_pcal_info[1].target_power_54 = 24 [ 39.020000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[2].freq = 5170, rate_pcal_info[2].target_power_6to24 = 36, rate_pcal_info[2].target_power_54 = 26 [ 39.040000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[3].freq = 5240, rate_pcal_info[3].target_power_6to24 = 38, rate_pcal_info[3].target_power_54 = 26 [ 39.056000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[4].freq = 5320, rate_pcal_info[4].target_power_6to24 = 38, rate_pcal_info[4].target_power_54 = 26 [ 39.076000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[5].freq = 5500, rate_pcal_info[5].target_power_6to24 = 38, rate_pcal_info[5].target_power_54 = 26 [ 39.092000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[6].freq = 5700, rate_pcal_info[6].target_power_6to24 = 36, rate_pcal_info[6].target_power_54 = 26 [ 39.112000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[7].freq = 5825, rate_pcal_info[7].target_power_6to24 = 34, rate_pcal_info[7].target_power_54 = 24 [ 39.128000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[8].freq = 5360, rate_pcal_info[8].target_power_6to24 = 39, rate_pcal_info[8].target_power_54 = 39 [ 39.144000] [ath5k_eeprom_read_target_rate_pwr_info] rate_pcal_info[9].freq = 5720, rate_pcal_info[9].target_power_6to24 = 38, rate_pcal_info[9].target_power_54 = 38 Beside that both last 2 lines hold invalid power gain values the last line is responsible that subsequent interpolation functions apply full power gain to all rates at all frequencies greater rate_pcal_info[max].freq, i.e. > 5720 MHz. This leads to spectral distortion when then global tx_power is higher than the expected rate_pcal_info[xy].target_power_54 and therfore gives rates above 24MBit a very poor performance in therms of throughput. The first patch reduces the eeprom calibration reads to 8. A validation against madwifi showed 8x eeprom reads at 5GHz as well. (2) Furthermore the tx_power specified by the user is not respected in cases where function phy_init() got called: tx_power is always set to ah->ah_txpower.txp_cur_pwr which is never updated with the txpower value specified by the user instead it equales max chan power and is even incremented. So this wrong behavior amplifies the occurrence of the performance problems coming from the wrong calibration in cases the user set a low power. The second patch fixes this by properly introduce and respect user tx_power. Greetings Thomas -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
