[ath9k-devel] About A-MPDU aggregation. ATH_AMPDU_SUBFRAME_DEFAULT flag.

[ath9k-devel] About A-MPDU aggregation. ATH_AMPDU_SUBFRAME_DEFAULT flag.

[Rakesh Kumar]

Hi,

I notice this parameter ATH_AMPDU_SUBFRAME_DEFAULT and in the code in
xmit.c, function ath_tx_form_aggr limits the number of sub-frames that can
be included in a A-MPDU to half the total size. The accompanying comments
say:

                /* do not exceed subframe limit */

But why halve the baw_size for that? I looked around the standard documents,
couldn't find any reference to halving the station-specfic baw_size
(negotiated during association) to half.

--
Rakesh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20100405/62ea2339/attachment.htm 

[ath9k-devel] About A-MPDU aggregation. ATH_AMPDU_SUBFRAME_DEFAULT flag.

[Felix Fietkau]

On 2010-04-05 4:11 PM, Rakesh Kumar wrote:
> Hi,
> 
> I notice this parameter ATH_AMPDU_SUBFRAME_DEFAULT and in the code in
> xmit.c, function ath_tx_form_aggr limits the number of sub-frames that
> can be included in a A-MPDU to half the total size. The accompanying
> comments say:
> 
>                 /* do not exceed subframe limit */
> 
> But why halve the baw_size for that? I looked around the standard
> documents, couldn't find any reference to halving the station-specfic
> baw_size (negotiated during association) to half.
Maybe because ath9k tries to keep two A-MPDU aggregates in the xmit
queue and has to handle retransmission of failed subframes in software.
IMHO it makes sense to use half of the BAW for each of the queue entries
under those conditions.

- Felix

[ath9k-devel] About A-MPDU aggregation. ATH_AMPDU_SUBFRAME_DEFAULT flag.

[Rakesh Kumar]

>
>
> IMHO it makes sense to use half of the BAW for each of the queue entries
> under those conditions.
>
>
Can you point where (which function(s)/file(s)) the two A-MPDU keeping logic
is implemented?

Thanks.

--
Rakesh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20100405/b7911812/attachment.htm