Erik Stromdahl <erik.stromdahl@xxxxxxxxx> writes: > On 4/16/19 9:07 PM, Toke Høiland-Jørgensen wrote: >> Erik Stromdahl <erik.stromdahl@xxxxxxxxx> writes: >> >>> On 4/1/19 1:05 PM, Toke Høiland-Jørgensen wrote: >>>> Erik Stromdahl <erik.stromdahl@xxxxxxxxx> writes: >>>> >>>>> Iterating the TX queue and thereby dequeuing all available packets in the >>>>> queue could result in performance penalties on some SMP systems. >>>>> >>>>> The reason for this is most likely that the per-ac lock (active_txq_lock) >>>>> in mac80211 will be held by the CPU iterating the current queue. >>>>> >>>>> This will lock up other CPUs trying to push new messages on the TX >>>>> queue. >>>>> >>>>> Instead of iterating the queue we fetch just one packet at the time, >>>>> resulting in minimal starvation of the other CPUs. >>>> >>>> Did you test this with Felix' patches reducing the time the lock is held >>>> in mac80211? >>>> >>>> -Toke >>>> >>> Hi Toke, >>> >>> I am not aware of these patches. Can you please point them out for me? >> >> They've already been merged. Commits dcec1d9bc8a7 and 7ef769459f14 in >> mac80211-next :) >> >> -Toke >> > > I see. I am using the ath tree and I couldn't find them there. > I can cherry-pick them to my own tree and try them out > (or wait until Kalle updates ath.git). It will take a while before these commits trickle down to ath-next branch, most likely after v5.2-rc1 is released. -- Kalle Valo
