* [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
@ 2013-03-03 11:49 Felix Fietkau
2013-03-03 11:49 ` [PATCH 2/2] mac80211/minstrel_ht: fix spacing between sample attempts Felix Fietkau
` (2 more replies)
0 siblings, 3 replies; 9+ messages in thread
From: Felix Fietkau @ 2013-03-03 11:49 UTC (permalink / raw)
To: linux-wireless; +Cc: johannes
If a rate is below the max_tp_rate, sample it frequently if:
- it is above max_tp_rate2, or
- it is above max_prob_rate and is a candidate for max_prob_rate
(has fewer streams than max_tp_rate).
This helps the retry chain recover more quickly from bad statistics
caused by collisions or interference, and slightly reduces throughput
fluctuations with higher rates.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
---
net/mac80211/rc80211_minstrel_ht.c | 20 ++++++++++++--------
net/mac80211/rc80211_minstrel_ht.h | 1 +
2 files changed, 13 insertions(+), 8 deletions(-)
diff --git a/net/mac80211/rc80211_minstrel_ht.c b/net/mac80211/rc80211_minstrel_ht.c
index 4d35bc5..1b69924 100644
--- a/net/mac80211/rc80211_minstrel_ht.c
+++ b/net/mac80211/rc80211_minstrel_ht.c
@@ -246,7 +246,6 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
struct minstrel_rate_stats *mr;
int cur_prob, cur_prob_tp, cur_tp, cur_tp2;
int group, i, index;
- int prob_max_streams = 1;
if (mi->ampdu_packets > 0) {
mi->avg_ampdu_len = minstrel_ewma(mi->avg_ampdu_len,
@@ -330,7 +329,7 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
cur_tp2 = cur_tp;
mi->max_tp_rate = mg->max_tp_rate;
cur_tp = mr->cur_tp;
- prob_max_streams = minstrel_mcs_groups[group].streams - 1;
+ mi->max_prob_streams = minstrel_mcs_groups[group].streams - 1;
}
mr = minstrel_get_ratestats(mi, mg->max_tp_rate2);
@@ -340,8 +339,8 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
}
}
- if (prob_max_streams < 1)
- prob_max_streams = 1;
+ if (mi->max_prob_streams < 1)
+ mi->max_prob_streams = 1;
for (group = 0; group < ARRAY_SIZE(minstrel_mcs_groups); group++) {
mg = &mi->groups[group];
@@ -349,7 +348,7 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
continue;
mr = minstrel_get_ratestats(mi, mg->max_prob_rate);
if (cur_prob_tp < mr->cur_tp &&
- minstrel_mcs_groups[group].streams <= prob_max_streams) {
+ minstrel_mcs_groups[group].streams <= mi->max_prob_streams) {
mi->max_prob_rate = mg->max_prob_rate;
cur_prob = mr->cur_prob;
cur_prob_tp = mr->cur_tp;
@@ -630,6 +629,7 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
{
struct minstrel_rate_stats *mr;
struct minstrel_mcs_group_data *mg;
+ unsigned int sample_dur, sample_group;
int sample_idx = 0;
if (mi->sample_wait > 0) {
@@ -644,7 +644,8 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
mg = &mi->groups[mi->sample_group];
sample_idx = sample_table[mg->column][mg->index];
mr = &mg->rates[sample_idx];
- sample_idx += mi->sample_group * MCS_GROUP_RATES;
+ sample_group = mi->sample_group;
+ sample_idx += sample_group * MCS_GROUP_RATES;
minstrel_next_sample_idx(mi);
/*
@@ -665,8 +666,11 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
* Make sure that lower rates get sampled only occasionally,
* if the link is working perfectly.
*/
- if (minstrel_get_duration(sample_idx) >
- minstrel_get_duration(mi->max_tp_rate)) {
+ sample_dur = minstrel_get_duration(sample_idx);
+ if (sample_dur >= minstrel_get_duration(mi->max_tp_rate2) &&
+ (mi->max_prob_streams <
+ minstrel_mcs_groups[sample_group].streams ||
+ sample_dur >= minstrel_get_duration(mi->max_prob_rate))) {
if (mr->sample_skipped < 20)
return -1;
diff --git a/net/mac80211/rc80211_minstrel_ht.h b/net/mac80211/rc80211_minstrel_ht.h
index 302dbd5..c6d6a0d 100644
--- a/net/mac80211/rc80211_minstrel_ht.h
+++ b/net/mac80211/rc80211_minstrel_ht.h
@@ -85,6 +85,7 @@ struct minstrel_ht_sta {
/* best probability rate */
unsigned int max_prob_rate;
+ unsigned int max_prob_streams;
/* time of last status update */
unsigned long stats_update;
--
1.8.0.2
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH 2/2] mac80211/minstrel_ht: fix spacing between sample attempts
2013-03-03 11:49 [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates Felix Fietkau
@ 2013-03-03 11:49 ` Felix Fietkau
2013-03-04 15:07 ` [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates Johannes Berg
2013-03-05 10:35 ` Georgiewskiy Yuriy
2 siblings, 0 replies; 9+ messages in thread
From: Felix Fietkau @ 2013-03-03 11:49 UTC (permalink / raw)
To: linux-wireless; +Cc: johannes
A sample attempt should only count in mi->sample_tries if the sample
attempt wasn't skipped based on slower rate criteria.
This patch increases the sampling frequency for potentially desirable
rates and thus enables faster recovery from interference or collisions.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
---
net/mac80211/rc80211_minstrel_ht.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/mac80211/rc80211_minstrel_ht.c b/net/mac80211/rc80211_minstrel_ht.c
index 1b69924..da4ec73 100644
--- a/net/mac80211/rc80211_minstrel_ht.c
+++ b/net/mac80211/rc80211_minstrel_ht.c
@@ -640,7 +640,6 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
if (!mi->sample_tries)
return -1;
- mi->sample_tries--;
mg = &mi->groups[mi->sample_group];
sample_idx = sample_table[mg->column][mg->index];
mr = &mg->rates[sample_idx];
@@ -677,6 +676,7 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
if (mi->sample_slow++ > 2)
return -1;
}
+ mi->sample_tries--;
return sample_idx;
}
--
1.8.0.2
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
2013-03-03 11:49 [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates Felix Fietkau
2013-03-03 11:49 ` [PATCH 2/2] mac80211/minstrel_ht: fix spacing between sample attempts Felix Fietkau
@ 2013-03-04 15:07 ` Johannes Berg
2013-03-05 10:35 ` Georgiewskiy Yuriy
2 siblings, 0 replies; 9+ messages in thread
From: Johannes Berg @ 2013-03-04 15:07 UTC (permalink / raw)
To: Felix Fietkau; +Cc: linux-wireless
On Sun, 2013-03-03 at 12:49 +0100, Felix Fietkau wrote:
> If a rate is below the max_tp_rate, sample it frequently if:
> - it is above max_tp_rate2, or
> - it is above max_prob_rate and is a candidate for max_prob_rate
> (has fewer streams than max_tp_rate).
> This helps the retry chain recover more quickly from bad statistics
> caused by collisions or interference, and slightly reduces throughput
> fluctuations with higher rates.
Applied both.
johannes
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
2013-03-03 11:49 [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates Felix Fietkau
2013-03-03 11:49 ` [PATCH 2/2] mac80211/minstrel_ht: fix spacing between sample attempts Felix Fietkau
2013-03-04 15:07 ` [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates Johannes Berg
@ 2013-03-05 10:35 ` Georgiewskiy Yuriy
2013-03-05 11:49 ` Felix Fietkau
2 siblings, 1 reply; 9+ messages in thread
From: Georgiewskiy Yuriy @ 2013-03-05 10:35 UTC (permalink / raw)
To: Felix Fietkau; +Cc: linux-wireless, johannes
[-- Attachment #1: Type: TEXT/PLAIN, Size: 4804 bytes --]
On 2013-03-03 12:49 +0100, Felix Fietkau wrote linux-wireless@vger.kernel.org:
Hi, in my case this two patches cause throughput drop, withous it i have 50-55Mbs
with it applied only 35-40.
FF>If a rate is below the max_tp_rate, sample it frequently if:
FF>- it is above max_tp_rate2, or
FF>- it is above max_prob_rate and is a candidate for max_prob_rate
FF> (has fewer streams than max_tp_rate).
FF>This helps the retry chain recover more quickly from bad statistics
FF>caused by collisions or interference, and slightly reduces throughput
FF>fluctuations with higher rates.
FF>
FF>Signed-off-by: Felix Fietkau <nbd@openwrt.org>
FF>---
FF> net/mac80211/rc80211_minstrel_ht.c | 20 ++++++++++++--------
FF> net/mac80211/rc80211_minstrel_ht.h | 1 +
FF> 2 files changed, 13 insertions(+), 8 deletions(-)
FF>
FF>diff --git a/net/mac80211/rc80211_minstrel_ht.c b/net/mac80211/rc80211_minstrel_ht.c
FF>index 4d35bc5..1b69924 100644
FF>--- a/net/mac80211/rc80211_minstrel_ht.c
FF>+++ b/net/mac80211/rc80211_minstrel_ht.c
FF>@@ -246,7 +246,6 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
FF> struct minstrel_rate_stats *mr;
FF> int cur_prob, cur_prob_tp, cur_tp, cur_tp2;
FF> int group, i, index;
FF>- int prob_max_streams = 1;
FF>
FF> if (mi->ampdu_packets > 0) {
FF> mi->avg_ampdu_len = minstrel_ewma(mi->avg_ampdu_len,
FF>@@ -330,7 +329,7 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
FF> cur_tp2 = cur_tp;
FF> mi->max_tp_rate = mg->max_tp_rate;
FF> cur_tp = mr->cur_tp;
FF>- prob_max_streams = minstrel_mcs_groups[group].streams - 1;
FF>+ mi->max_prob_streams = minstrel_mcs_groups[group].streams - 1;
FF> }
FF>
FF> mr = minstrel_get_ratestats(mi, mg->max_tp_rate2);
FF>@@ -340,8 +339,8 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
FF> }
FF> }
FF>
FF>- if (prob_max_streams < 1)
FF>- prob_max_streams = 1;
FF>+ if (mi->max_prob_streams < 1)
FF>+ mi->max_prob_streams = 1;
FF>
FF> for (group = 0; group < ARRAY_SIZE(minstrel_mcs_groups); group++) {
FF> mg = &mi->groups[group];
FF>@@ -349,7 +348,7 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
FF> continue;
FF> mr = minstrel_get_ratestats(mi, mg->max_prob_rate);
FF> if (cur_prob_tp < mr->cur_tp &&
FF>- minstrel_mcs_groups[group].streams <= prob_max_streams) {
FF>+ minstrel_mcs_groups[group].streams <= mi->max_prob_streams) {
FF> mi->max_prob_rate = mg->max_prob_rate;
FF> cur_prob = mr->cur_prob;
FF> cur_prob_tp = mr->cur_tp;
FF>@@ -630,6 +629,7 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
FF> {
FF> struct minstrel_rate_stats *mr;
FF> struct minstrel_mcs_group_data *mg;
FF>+ unsigned int sample_dur, sample_group;
FF> int sample_idx = 0;
FF>
FF> if (mi->sample_wait > 0) {
FF>@@ -644,7 +644,8 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
FF> mg = &mi->groups[mi->sample_group];
FF> sample_idx = sample_table[mg->column][mg->index];
FF> mr = &mg->rates[sample_idx];
FF>- sample_idx += mi->sample_group * MCS_GROUP_RATES;
FF>+ sample_group = mi->sample_group;
FF>+ sample_idx += sample_group * MCS_GROUP_RATES;
FF> minstrel_next_sample_idx(mi);
FF>
FF> /*
FF>@@ -665,8 +666,11 @@ minstrel_get_sample_rate(struct minstrel_priv *mp, struct minstrel_ht_sta *mi)
FF> * Make sure that lower rates get sampled only occasionally,
FF> * if the link is working perfectly.
FF> */
FF>- if (minstrel_get_duration(sample_idx) >
FF>- minstrel_get_duration(mi->max_tp_rate)) {
FF>+ sample_dur = minstrel_get_duration(sample_idx);
FF>+ if (sample_dur >= minstrel_get_duration(mi->max_tp_rate2) &&
FF>+ (mi->max_prob_streams <
FF>+ minstrel_mcs_groups[sample_group].streams ||
FF>+ sample_dur >= minstrel_get_duration(mi->max_prob_rate))) {
FF> if (mr->sample_skipped < 20)
FF> return -1;
FF>
FF>diff --git a/net/mac80211/rc80211_minstrel_ht.h b/net/mac80211/rc80211_minstrel_ht.h
FF>index 302dbd5..c6d6a0d 100644
FF>--- a/net/mac80211/rc80211_minstrel_ht.h
FF>+++ b/net/mac80211/rc80211_minstrel_ht.h
FF>@@ -85,6 +85,7 @@ struct minstrel_ht_sta {
FF>
FF> /* best probability rate */
FF> unsigned int max_prob_rate;
FF>+ unsigned int max_prob_streams;
FF>
FF> /* time of last status update */
FF> unsigned long stats_update;
FF>
C уважением With Best Regards
Георгиевский Юрий. Georgiewskiy Yuriy
+7 4872 711666 +7 4872 711666
факс +7 4872 711143 fax +7 4872 711143
Компания ООО "Ай Ти Сервис" IT Service Ltd
http://nkoort.ru http://nkoort.ru
JID: GHhost@icf.org.ru JID: GHhost@icf.org.ru
YG129-RIPE YG129-RIPE
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
2013-03-05 10:35 ` Georgiewskiy Yuriy
@ 2013-03-05 11:49 ` Felix Fietkau
2013-03-05 12:08 ` Georgiewskiy Yuriy
0 siblings, 1 reply; 9+ messages in thread
From: Felix Fietkau @ 2013-03-05 11:49 UTC (permalink / raw)
To: Georgiewskiy Yuriy; +Cc: linux-wireless, johannes
On 2013-03-05 11:35 AM, Georgiewskiy Yuriy wrote:
> On 2013-03-03 12:49 +0100, Felix Fietkau wrote linux-wireless@vger.kernel.org:
>
> Hi, in my case this two patches cause throughput drop, withous it i have 50-55Mbs
> with it applied only 35-40.
Please show me the rate control statistics from debugfs.
- Felix
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
2013-03-05 11:49 ` Felix Fietkau
@ 2013-03-05 12:08 ` Georgiewskiy Yuriy
2013-03-05 12:26 ` Felix Fietkau
0 siblings, 1 reply; 9+ messages in thread
From: Georgiewskiy Yuriy @ 2013-03-05 12:08 UTC (permalink / raw)
To: Felix Fietkau; +Cc: linux-wireless, johannes
[-- Attachment #1: Type: TEXT/PLAIN, Size: 815 bytes --]
On 2013-03-05 12:49 +0100, Felix Fietkau wrote Georgiewskiy Yuriy:
FF>On 2013-03-05 11:35 AM, Georgiewskiy Yuriy wrote:
FF>> On 2013-03-03 12:49 +0100, Felix Fietkau wrote linux-wireless@vger.kernel.org:
FF>>
FF>> Hi, in my case this two patches cause throughput drop, withous it i have 50-55Mbs
FF>> with it applied only 35-40.
FF>Please show me the rate control statistics from debugfs.
in attachment
C уважением With Best Regards
Георгиевский Юрий. Georgiewskiy Yuriy
+7 4872 711666 +7 4872 711666
факс +7 4872 711143 fax +7 4872 711143
Компания ООО "Ай Ти Сервис" IT Service Ltd
http://nkoort.ru http://nkoort.ru
JID: GHhost@icf.org.ru JID: GHhost@icf.org.ru
YG129-RIPE YG129-RIPE
[-- Attachment #2: Type: APPLICATION/octet-stream, Size: 2348 bytes --]
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
2013-03-05 12:08 ` Georgiewskiy Yuriy
@ 2013-03-05 12:26 ` Felix Fietkau
2013-03-05 12:55 ` Georgiewskiy Yuriy
0 siblings, 1 reply; 9+ messages in thread
From: Felix Fietkau @ 2013-03-05 12:26 UTC (permalink / raw)
To: Georgiewskiy Yuriy; +Cc: linux-wireless, johannes
On 2013-03-05 1:08 PM, Georgiewskiy Yuriy wrote:
> On 2013-03-05 12:49 +0100, Felix Fietkau wrote Georgiewskiy Yuriy:
>
> FF>On 2013-03-05 11:35 AM, Georgiewskiy Yuriy wrote:
> FF>> On 2013-03-03 12:49 +0100, Felix Fietkau wrote linux-wireless@vger.kernel.org:
> FF>>
> FF>> Hi, in my case this two patches cause throughput drop, withous it i have 50-55Mbs
> FF>> with it applied only 35-40.
> FF>Please show me the rate control statistics from debugfs.
>
> in attachment
Please test if this patch helps:
---
--- a/net/mac80211/rc80211_minstrel_ht.c
+++ b/net/mac80211/rc80211_minstrel_ht.c
@@ -480,7 +480,7 @@ minstrel_ht_tx_status(void *priv, struct
if (!mi->sample_wait && !mi->sample_tries && mi->sample_count > 0) {
mi->sample_wait = 16 + 2 * MINSTREL_TRUNC(mi->avg_ampdu_len);
- mi->sample_tries = 2;
+ mi->sample_tries = 1;
mi->sample_count--;
}
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
2013-03-05 12:26 ` Felix Fietkau
@ 2013-03-05 12:55 ` Georgiewskiy Yuriy
2013-03-05 13:05 ` Felix Fietkau
0 siblings, 1 reply; 9+ messages in thread
From: Georgiewskiy Yuriy @ 2013-03-05 12:55 UTC (permalink / raw)
To: Felix Fietkau; +Cc: linux-wireless, johannes
[-- Attachment #1: Type: TEXT/PLAIN, Size: 1000 bytes --]
On 2013-03-05 13:26 +0100, Felix Fietkau wrote Georgiewskiy Yuriy:
FF>Please test if this patch helps:
FF>---
FF>--- a/net/mac80211/rc80211_minstrel_ht.c
FF>+++ b/net/mac80211/rc80211_minstrel_ht.c
FF>@@ -480,7 +480,7 @@ minstrel_ht_tx_status(void *priv, struct
FF>
FF> if (!mi->sample_wait && !mi->sample_tries && mi->sample_count > 0) {
FF> mi->sample_wait = 16 + 2 * MINSTREL_TRUNC(mi->avg_ampdu_len);
FF>- mi->sample_tries = 2;
FF>+ mi->sample_tries = 1;
FF> mi->sample_count--;
FF> }
Seems this helps, with this patch throughput is near as without this two patches applied.
C уважением With Best Regards
Георгиевский Юрий. Georgiewskiy Yuriy
+7 4872 711666 +7 4872 711666
факс +7 4872 711143 fax +7 4872 711143
Компания ООО "Ай Ти Сервис" IT Service Ltd
http://nkoort.ru http://nkoort.ru
JID: GHhost@icf.org.ru JID: GHhost@icf.org.ru
YG129-RIPE YG129-RIPE
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates
2013-03-05 12:55 ` Georgiewskiy Yuriy
@ 2013-03-05 13:05 ` Felix Fietkau
0 siblings, 0 replies; 9+ messages in thread
From: Felix Fietkau @ 2013-03-05 13:05 UTC (permalink / raw)
To: Georgiewskiy Yuriy; +Cc: linux-wireless, johannes
On 2013-03-05 1:55 PM, Georgiewskiy Yuriy wrote:
> On 2013-03-05 13:26 +0100, Felix Fietkau wrote Georgiewskiy Yuriy:
>
>
> FF>Please test if this patch helps:
> FF>---
> FF>--- a/net/mac80211/rc80211_minstrel_ht.c
> FF>+++ b/net/mac80211/rc80211_minstrel_ht.c
> FF>@@ -480,7 +480,7 @@ minstrel_ht_tx_status(void *priv, struct
> FF>
> FF> if (!mi->sample_wait && !mi->sample_tries && mi->sample_count > 0) {
> FF> mi->sample_wait = 16 + 2 * MINSTREL_TRUNC(mi->avg_ampdu_len);
> FF>- mi->sample_tries = 2;
> FF>+ mi->sample_tries = 1;
> FF> mi->sample_count--;
> FF> }
>
> Seems this helps, with this patch throughput is near as without this two patches applied.
OK, thanks for testing. I will submit it now.
- Felix
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2013-03-05 13:06 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-03-03 11:49 [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates Felix Fietkau
2013-03-03 11:49 ` [PATCH 2/2] mac80211/minstrel_ht: fix spacing between sample attempts Felix Fietkau
2013-03-04 15:07 ` [PATCH 1/2] mac80211/minstrel_ht: increase sampling frequency of some slower rates Johannes Berg
2013-03-05 10:35 ` Georgiewskiy Yuriy
2013-03-05 11:49 ` Felix Fietkau
2013-03-05 12:08 ` Georgiewskiy Yuriy
2013-03-05 12:26 ` Felix Fietkau
2013-03-05 12:55 ` Georgiewskiy Yuriy
2013-03-05 13:05 ` Felix Fietkau
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).