On 2018-09-28 04:02, Toke Høiland-Jørgensen wrote:
On 28 September 2018 12:47:51 CEST, Rajkumar Manoharan
<rmanohar@xxxxxxxxxxxxxx> wrote:
On 2018-09-28 03:35, Jonathan Morton wrote:
On 28 Sep, 2018, at 1:19 pm, Rajkumar Manoharan
<rmanohar@xxxxxxxxxxxxxx> wrote:
This is going to break fairness; we only want to increase deficits
when
all stations' deficits are negative. Hence the two loops. Did you
see
any problems with those specifically?
No. I didn't see any issue but the loop won't exit until one txq
becomes positive.
Till then the lock won't be released. Hence I converged into single
iteration.
The problem is that the fairness behaviour is changed when there are
already positive txqs, but the first one happens to not be positive.
That's why there were two loops.
Yeah.. Understood. we can ignore the cleanup change.
Great! I'll fold in the rest, test it with ath9k and submit as a proper
patch :)
Thanks. I am seeing higher CPU load. Did you observe similar behavior
with ath9k.
-Rajkumar
