On Thu, Sep 1, 2011 at 7:13 AM, John W. Linville [off-list ref] wrote:

On Wed, Aug 31, 2011 at 01:50:48PM -0700, Luis R. Rodriguez wrote:

quoted

On Wed, Aug 31, 2011 at 6:28 AM, Jim Gettys [off-list ref] wrote:

quoted

quoted

such as wireless, or even possibly modern broadband with
PowerBoost, classic RED or similar algorithms that do not take the
buffer drain rate cannot possibly hack it properly.

Understood, just curious if anyone has tried a Minstrel approach.

FWIW, eBDP and the related algorithms from Tianji Li's paper are
philosophically similar to minstrel.

Oh look at that, awesome!!!

 They depend on measuring recent
conditions and modifying the current queue length accordingly.

       http://www.hamilton.ie/tianji_li/buffersizing.pdf

The hack I added in debloat-testing is based on my understanding
of eBDP.  It timestamps the SKBs when they are handed to the driver
for Tx and then checks the timestamp when the SKB is orphaned.  It is
a bit crude and is an abuse of the skb_orphan API.

Neat!

 Also while it
accounts for the 802.11e queues separately, it doesn't account for
802.11n aggregation.

I see..

 Still, it seems to improve latency w/o hugely
impacting throughput in at least some environments -- YMMV!

Sweet dude. For aggregation it seems the way to go is to get some
helpers as Andrew has suggested. Andrew, can you elaborate a little on
that? If feasible, then maybe then we can add it to the TODO list
page:

http://wireless.kernel.org/en/developers/todo-list

and when one of us gets to it, we get cranking on it.

  Luis