Return-path: Received: from mail.candelatech.com ([208.74.158.172]:37512 "EHLO ns3.lanforge.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751683Ab1ASA1B (ORCPT ); Tue, 18 Jan 2011 19:27:01 -0500 Message-ID: <4D362FC5.3090003@candelatech.com> Date: Tue, 18 Jan 2011 16:26:45 -0800 From: Ben Greear MIME-Version: 1.0 To: Felix Fietkau CC: linux-wireless@vger.kernel.org, linville@tuxdriver.com, lrodriguez@atheros.com Subject: Re: [PATCH 1/2] ath9k: try more than one tid when scheduling a new aggregate References: <1295098215-98910-1-git-send-email-nbd@openwrt.org> In-Reply-To: <1295098215-98910-1-git-send-email-nbd@openwrt.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Sender: linux-wireless-owner@vger.kernel.org List-ID: On 01/15/2011 05:30 AM, Felix Fietkau wrote: > Sometimes the first TID in the first AC's list is not available for forming > a new aggregate (the BAW might not allow it), however other TIDs may have > data available for sending. > Prevent a slowdown of other TIDs by going through multiple entries until > we've either hit the last one or enough AMPDUs are pending in the hardware > queue. I had attempted a similar patch, but I was thinking we should iterate through all of the txq->axq_acq entries, not just use the first one? If you agree, I can merge my patch with your patch below and re-post. Thanks, Ben > > Signed-off-by: Felix Fietkau > --- > drivers/net/wireless/ath/ath9k/xmit.c | 9 ++++++--- > 1 files changed, 6 insertions(+), 3 deletions(-) > > diff --git a/drivers/net/wireless/ath/ath9k/xmit.c b/drivers/net/wireless/ath/ath9k/xmit.c > index ab4f7b4..fffd13d 100644 > --- a/drivers/net/wireless/ath/ath9k/xmit.c > +++ b/drivers/net/wireless/ath/ath9k/xmit.c > @@ -1224,12 +1224,14 @@ void ath_tx_cleanupq(struct ath_softc *sc, struct ath_txq *txq) > void ath_txq_schedule(struct ath_softc *sc, struct ath_txq *txq) > { > struct ath_atx_ac *ac; > - struct ath_atx_tid *tid; > + struct ath_atx_tid *tid, *last; > > - if (list_empty(&txq->axq_acq)) > + if (list_empty(&txq->axq_acq) || > + txq->axq_ampdu_depth>= ATH_AGGR_MIN_QDEPTH) > return; > > ac = list_first_entry(&txq->axq_acq, struct ath_atx_ac, list); > + last = list_entry(ac->tid_q.prev, struct ath_atx_tid, list); > list_del(&ac->list); > ac->sched = false; > > @@ -1253,7 +1255,8 @@ void ath_txq_schedule(struct ath_softc *sc, struct ath_txq *txq) > if (!list_empty(&tid->buf_q)) > ath_tx_queue_tid(txq, tid); > > - break; > + if (tid == last || txq->axq_ampdu_depth>= ATH_AGGR_MIN_QDEPTH) > + break; > } while (!list_empty(&ac->tid_q)); > > if (!list_empty(&ac->tid_q)) { Thanks, Ben -- Ben Greear Candela Technologies Inc http://www.candelatech.com