2016-03-03 22:25:45

by Felix Fietkau

Subject: [PATCH] mac80211: minstrel_ht: improve sample rate skip logic

There were a few issues that were slowing down the process of finding
the optimal rate, especially on devices with multi-rate retry
limitations:

When max_tp_rate[0] was slower than max_tp_rate[1], the code did not
sample max_tp_rate[1], which would often allow it to switch places with
max_tp_rate[0] (e.g. if only the first sampling attempts were bad, but the
rate is otherwise good).

Also, sample attempts of rates between max_tp_rate[0] and [1] were being
ignored in this case, because the code only checked if the rate was
slower than [1].

Fix this by checking against the fastest / second fastest max_tp_rate
instead of assuming a specific order between the two.

In my tests this patch significantly reduces the time until minstrel_ht
finds the optimal rate right after assoc

Signed-off-by: Felix Fietkau <[email protected]>
---
net/mac80211/rc80211_minstrel_ht.c | 23 ++++++++++++++++-------
1 file changed, 16 insertions(+), 7 deletions(-)

diff --git a/net/mac80211/rc80211_minstrel_ht.c b/net/mac80211/rc80211_minstrel_ht.c
index 3928dbd..9197191 100644
--- a/net/mac80211/rc80211_minstrel_ht.c
+++ b/net/mac80211/rc80211_minstrel_ht.c
@@ -923,6 +923,7 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
struct minstrel_rate_stats *mrs;
struct minstrel_mcs_group_data *mg;
unsigned int sample_dur, sample_group, cur_max_tp_streams;
+ int tp_rate1, tp_rate2;
int sample_idx = 0;

if (mi->sample_wait > 0) {
@@ -944,14 +945,22 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
mrs = &mg->rates[sample_idx];
sample_idx += sample_group * MCS_GROUP_RATES;

+ /* Set tp_rate1, tp_rate2 to the highest / second highest max_tp_rate */
+ if (minstrel_get_duration(mi->max_tp_rate[0]) >
+ minstrel_get_duration(mi->max_tp_rate[1])) {
+ tp_rate1 = mi->max_tp_rate[1];
+ tp_rate2 = mi->max_tp_rate[0];
+ } else {
+ tp_rate1 = mi->max_tp_rate[0];
+ tp_rate2 = mi->max_tp_rate[1];
+ }
+
/*
* Sampling might add some overhead (RTS, no aggregation)
- * to the frame. Hence, don't use sampling for the currently
- * used rates.
+ * to the frame. Hence, don't use sampling for the highest currently
+ * used highest throughput or probability rate.
*/
- if (sample_idx == mi->max_tp_rate[0] ||
- sample_idx == mi->max_tp_rate[1] ||
- sample_idx == mi->max_prob_rate)
+ if (sample_idx == mi->max_tp_rate[0] || sample_idx == mi->max_prob_rate)
return -1;

/*
@@ -966,10 +975,10 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
* if the link is working perfectly.
*/

- cur_max_tp_streams = minstrel_mcs_groups[mi->max_tp_rate[0] /
+ cur_max_tp_streams = minstrel_mcs_groups[tp_rate1 /
MCS_GROUP_RATES].streams;
sample_dur = minstrel_get_duration(sample_idx);
- if (sample_dur >= minstrel_get_duration(mi->max_tp_rate[1]) &&
+ if (sample_dur >= minstrel_get_duration(tp_rate2) &&
(cur_max_tp_streams - 1 <
minstrel_mcs_groups[sample_group].streams ||
sample_dur >= minstrel_get_duration(mi->max_prob_rate))) {
--
2.2.2

2016-04-05 09:40:27

by Johannes Berg

Subject: Re: [PATCH] mac80211: minstrel_ht: improve sample rate skip logic

On Thu, 2016-03-03 at 23:25 +0100, Felix Fietkau wrote:
> There were a few issues that were slowing down the process of finding
> the optimal rate, especially on devices with multi-rate retry
> limitations:
>
[...]

applied.

johannes