[ATH9K_HTC] Debug files

Matteo Danieletto matteo.danieletto at gmail.com
Tue Aug 6 12:47:10 EDT 2013


Hi Sujith,
how you suggested me few days ago I tried to modify the ath9k_htc rate control to find the total frame retransmission .

I modified the struct ath_softc_tgt and  sc_tx_stats adding a new field: retransmission to report this value in tgt_tx_status.

However, I am doing something wrong because I modified this function

void rcUpdate_11n(struct ath_softc_tgt *sc, struct ath_node_target *an,
		  A_UINT8 curTxAnt, 
		  int finalTSIdx, int Xretries,
		  struct ath_rc_series rcs[], int nFrames, 
		  int nBad, int long_retry){

and I added sc->sc_tx_stats.retransmission +=long_retry ;

But my new field retransmission is equal to sc->sc_tx_stats.ast_tx_longretry .

So the problem is that  I don't know how work 4 series multi rate. I tried to understand reading the sourcecode without good results.
Do you any suggestion about that ? I found there are initialised 4 series but I don't know how I can exploit them.


Thanks a lot.
Matteo


On Jul 29, 2013, at 4:31 AM, Sujith Manoharan <sujith at msujith.org> wrote:

> Matteo wrote:
>> Now I have another doubt, What does it mean "not filtered"?
> 
> A TX frame is returned with the filtered status if the destination
> mask is set in the station table maintained in the HW. This happens
> if a previous frame wasn't acknowledged. See the usage of
> ATH9K_TXDESC_CLRDMASK, ATH9K_TXERR_FILT etc.
> 
>> However, If I wish to know the global number retransmission in a finite
>> interval, could I extract long retries plus no filtered ?
> 
> Not really. The long_retries corresponds to the final rate index that was
> used to transmit the frame. In a 4-series Multi Rate algorithm, the other
> entries which were attempted and failed have to be taken into account too.
> 
> Accurate statistics for retransmission, xretries etc. should be available in
> the rate control algorithm. For ath9k_htc, the RC is in the firmware.
> 
> Sujith
> 




More information about the ath9k_htc_fw mailing list