From: Jan Kiszka <jan.kiszka@web.de>
To: Christian Lamparter <chunkeey@googlemail.com>
Cc: linux-wireless@vger.kernel.org,
"John W. Linville" <linville@tuxdriver.com>
Subject: Re: [PATCH 2/5] ar9170: improve network latency
Date: Sun, 06 Dec 2009 19:16:31 +0100 [thread overview]
Message-ID: <4B1BF4FF.2040707@web.de> (raw)
In-Reply-To: <200911290052.51819.chunkeey@googlemail.com>
[-- Attachment #1: Type: text/plain, Size: 2983 bytes --]
Christian Lamparter wrote:
> AR9170 has the bad habit of choking when traffic builds up.
> Tests have shown that this can partially be attributed to
> a huge buildup of backlogged frames.
>
> Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
> ---
> low priority.
I disagree about this classification. :)
I'm running your series for two days now on my ar9170-based AP, and
obviously this patch improved the "chokiness" under suboptimal reception
conditions greatly. No regressions noticed so far. Please merge.
Thanks a lot,
Jan
> ---
> diff --git a/drivers/net/wireless/ath/ar9170/ar9170.h b/drivers/net/wireless/ath/ar9170/ar9170.h
> index d95a7d4..44c54b8 100644
> --- a/drivers/net/wireless/ath/ar9170/ar9170.h
> +++ b/drivers/net/wireless/ath/ar9170/ar9170.h
> @@ -152,6 +152,8 @@ struct ar9170_sta_tid {
>
> #define AR9170_NUM_TX_STATUS 128
> #define AR9170_NUM_TX_AGG_MAX 30
> +#define AR9170_NUM_TX_LIMIT_HARD AR9170_TXQ_DEPTH
> +#define AR9170_NUM_TX_LIMIT_SOFT (AR9170_TXQ_DEPTH - 10)
>
> struct ar9170 {
> struct ieee80211_hw *hw;
> diff --git a/drivers/net/wireless/ath/ar9170/main.c b/drivers/net/wireless/ath/ar9170/main.c
> index cbf59b1..3cec2f1 100644
> --- a/drivers/net/wireless/ath/ar9170/main.c
> +++ b/drivers/net/wireless/ath/ar9170/main.c
> @@ -430,7 +430,7 @@ void ar9170_tx_callback(struct ar9170 *ar, struct sk_buff *skb)
> spin_lock_irqsave(&ar->tx_stats_lock, flags);
> ar->tx_stats[queue].len--;
>
> - if (skb_queue_empty(&ar->tx_pending[queue])) {
> + if (ar->tx_stats[queue].len < AR9170_NUM_TX_LIMIT_SOFT) {
> #ifdef AR9170_QUEUE_STOP_DEBUG
> printk(KERN_DEBUG "%s: wake queue %d\n",
> wiphy_name(ar->hw->wiphy), queue);
> @@ -1716,6 +1716,21 @@ static void ar9170_tx(struct ar9170 *ar)
>
> for (i = 0; i < __AR9170_NUM_TXQ; i++) {
> spin_lock_irqsave(&ar->tx_stats_lock, flags);
> + frames = min(ar->tx_stats[i].limit - ar->tx_stats[i].len,
> + skb_queue_len(&ar->tx_pending[i]));
> +
> + if (remaining_space < frames) {
> +#ifdef AR9170_QUEUE_DEBUG
> + printk(KERN_DEBUG "%s: tx quota reached queue:%d, "
> + "remaining slots:%d, needed:%d\n",
> + wiphy_name(ar->hw->wiphy), i, remaining_space,
> + frames);
> +#endif /* AR9170_QUEUE_DEBUG */
> + frames = remaining_space;
> + }
> +
> + ar->tx_stats[i].len += frames;
> + ar->tx_stats[i].count += frames;
> if (ar->tx_stats[i].len >= ar->tx_stats[i].limit) {
> #ifdef AR9170_QUEUE_DEBUG
> printk(KERN_DEBUG "%s: queue %d full\n",
> @@ -1733,25 +1748,8 @@ static void ar9170_tx(struct ar9170 *ar)
> __ar9170_dump_txstats(ar);
> #endif /* AR9170_QUEUE_STOP_DEBUG */
> ieee80211_stop_queue(ar->hw, i);
> - spin_unlock_irqrestore(&ar->tx_stats_lock, flags);
> - continue;
> - }
> -
> - frames = min(ar->tx_stats[i].limit - ar->tx_stats[i].len,
> - skb_queue_len(&ar->tx_pending[i]));
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 257 bytes --]
prev parent reply other threads:[~2009-12-06 18:16 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-11-28 23:52 [PATCH 2/5] ar9170: improve network latency Christian Lamparter
2009-12-06 18:16 ` Jan Kiszka [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4B1BF4FF.2040707@web.de \
--to=jan.kiszka@web.de \
--cc=chunkeey@googlemail.com \
--cc=linux-wireless@vger.kernel.org \
--cc=linville@tuxdriver.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).