Fixing the Tx rate

[Yariv Bash]

On Sun, Nov 2, 2014 at 11:02 PM, Oleksij Rempel <linux at rempel-privat.de> wrote:
> Am 02.11.2014 um 16:17 schrieb Yariv Bash:
>> Hi guys,
>>
>> I'm trying to fix (as in fixate or choose only one) the Tx rate of a TP
>> LINK TL-WN722N device, containing the Atheros AR9271 chip.
>> Specifically I want it fixed on a low rate - I want to play with the
>> 1,2,5.5 and 6 MBPS rates.
>> Yes, 6 belongs to 802.11g, I know :)
>>
>> I've tried variations of the:
>> sudo iw dev wlan0 set bitrates legacy-2.4 6
>> Command, while the device is down, up, configured and what not, and
>> unfortunately, this does not work too well - The device still falls to
>> 1MBit when the connection becomes unstable, and almost never returns to a
>> higher rate (or takes a lot of time)
>>
>> I'm working with a raspberry pi, already compiled the ath9k_htc driver (FW
>> Version 1.4), and I already have 2 raspberry pi's talking directly - one as
>> the host of course (hostapd 2.3) each with its own WN722N device.
>> Oh, and I'm working with the Minstrel rate algorithm, but could of course
>> change to PID if it helps.
>>
>> One thing I thought about was that 1MBit is a default fallback with
>> minstrel, so it might be hardwired. But again, even when the link shows
>> really good reception (-40-50dBm), the link stays at 1MBPS after it falls
>> there on a bad connection.
>>
>> I'm considering tweaking the firmware code to just one rate, not sure if
>> this is the best option, but would love any directions as to where in the
>> code that can be done, or what am I doing wrong...
>
> I think this question should go in to FAQ :)
> There firmware has own, may be buggy, internal rate controller. Linux
> driver ath9k_htc can only suggest the driver to use some rate.
> Any one who cares is free and _welcome_ to investigate it :) And provide
> patches.
>
> --
> Regards,
> Oleksij
>

Thanks for the tip !
If I'm compiling the ar9271.fw file using
https://github.com/qca/open-ath9k-htc-firmware, don't I have access to
the code of the internal rate controller ?
Or is this file (and source code) just another layer to a closed
binary I cannot explore ?
Thanks.