From: Christian Lamparter <chunkeey@googlemail.com>
To: Johannes Berg <johannes@sipsolutions.net>
Cc: "Sean Patrick Santos" <quantheory@gmail.com>,
linux-wireless@vger.kernel.org, nbd@openwrt.org,
helmut.schaa@googlemail.com, mar.kolya@gmail.com,
"Per-Erik Westerberg" <per-erik.westerberg@bredband.net>,
"Mikołaj Kuligowski" <mikolaj.q@wp.pl>
Subject: Re: [RFC v2] mac80211: follow 802.11-2007 11.5.3 Error recovery upon HT BA failure
Date: Sat, 9 Jun 2012 14:14:41 +0200 [thread overview]
Message-ID: <201206091414.41940.chunkeey@googlemail.com> (raw)
In-Reply-To: <1339228767.4539.3.camel@jlt3.sipsolutions.net>
On Saturday 09 June 2012 09:59:27 Johannes Berg wrote:
> > introduced a TX BA session timeout timer. However the
> > timer was reset for each outgoing transmission instead
> > what 802.11-2007 11.5.3 specifies:
>
> You should probably reference 802.11-2012 now since some
> things were renumbered.
yeah, I should. But 802.11-2012 is not yet available for
free like the old 802.11-2007 through IEEE 802 Get.
> > "The inactivity timer at the originator is reset when a
> > BlockAck frame corresponding to the TID for which the
> > Block Ack policy is set is received."
>
> > This patch was only compile-tested and the design has
> > some rather big problems:
>
> Maybe then we should just revert Nikolay's patch?
(I was referring to this RFC patch and not his patch)
I don't think we can/should revert it, just change it
a bit so the timeout reset is triggered by a received
BA from the peer.
> > - The spec says we should test for BlockAcks, however
> > that is not feasible because it is a control frame
> > (so FIF_CONTROL might filter it on some hardware).
>
> In some way, all devices are going to have to report these
> frames. Maybe not as the frame itself, but as some other
> notification, to allow cleaning up the TX queues
> accordingly, I think? There's a BA notification in iwlwifi
> for example.
Alright, I guess this means that we probably need a new HW
feature flag like:
IEEE80211_HW_REPORTS_BA_NOTIFICATION
for hardware/firmware which filter BA frames, but have a BA
notification trap. And every other driver need to pass BA
to the stack where a new rx handler will update the last_tx
accordingly.
(yeah, that IEEE80211_HW_HAS_RATE_CONTROL vs. ACK_REPORT mix-up
was a stupid error - anyway here's the last updated version
of this approach - v3 will use IEEE80211_HW_REPORTS_BA_NOTIFICATION)
---
diff --git a/net/mac80211/sta_info.h b/net/mac80211/sta_info.h
index 3bb24a1..fbe84a7 100644
--- a/net/mac80211/sta_info.h
+++ b/net/mac80211/sta_info.h
@@ -103,7 +103,7 @@ enum ieee80211_sta_info_flags {
* @dialog_token: dialog token for aggregation session
* @timeout: session timeout value to be filled in ADDBA requests
* @state: session state (see above)
- * @last_tx: jiffies of last tx activity
+ * @last_tx: jiffies of last successful tx activity
* @stop_initiator: initiator of a session stop
* @tx_stop: TX DelBA frame when stopping
* @buf_size: reorder buffer size at receiver
diff --git a/net/mac80211/status.c b/net/mac80211/status.c
index 6b4f425..dfc61de 100644
--- a/net/mac80211/status.c
+++ b/net/mac80211/status.c
@@ -439,6 +439,28 @@ void ieee80211_tx_status(struct ieee80211_hw *hw, struct sk_buff *skb)
}
}
+ if (ieee80211_is_data_qos(fc) && acked &&
+ info->flags & IEEE80211_TX_CTL_AMPDU &&
+ local->hw.flags & IEEE80211_HW_REPORTS_TX_ACK_STATUS) {
+ struct tid_ampdu_tx *tid_tx;
+ u8 *qc;
+ u16 tid;
+
+ qc = ieee80211_get_qos_ctl(hdr);
+ tid = qc[0] & 0xf;
+ tid_tx = rcu_dereference(sta->ampdu_mlme.tid_tx[tid]);
+ if (tid_tx) {
+ /*
+ * 802.11-2007 11.5.3 Error recovery upon a peer failure
+ *
+ * The inactivity timer at the originator is reset when
+ * a BlockAck frame corresponding to the TID for which
+ * the Block Ack policy is set is received.
+ */
+ tid_tx->last_tx = jiffies;
+ }
+ }
+
if (info->flags & IEEE80211_TX_STAT_TX_FILTERED) {
ieee80211_handle_filtered_frame(local, sta, skb);
rcu_read_unlock();
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index af25c4e..67887c5 100644
--- a/net/mac80211/tx.c
+++ b/net/mac80211/tx.c
@@ -1098,7 +1098,9 @@ static bool ieee80211_tx_prep_agg(struct ieee80211_tx_data *tx,
/* do nothing, let packet pass through */
} else if (test_bit(HT_AGG_STATE_OPERATIONAL, &tid_tx->state)) {
info->flags |= IEEE80211_TX_CTL_AMPDU;
- reset_agg_timer = true;
+ if (!(tx->local->hw.flags &
+ IEEE80211_HW_REPORTS_TX_ACK_STATUS))
+ reset_agg_timer = true;
} else {
queued = true;
info->control.vif = &tx->sdata->vif;
next prev parent reply other threads:[~2012-06-09 12:14 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-06-07 6:53 Sean Patrick Santos
2012-06-07 13:18 ` carl9170 issue Christian Lamparter
2012-06-07 19:31 ` Sean Patrick Santos
2012-06-08 7:57 ` Sean Patrick Santos
2012-06-09 0:18 ` BA session issue due to old BARs? Christian Lamparter
2012-06-09 1:11 ` [RFC] mac80211: follow 802.11-2007 11.5.3 Error recovery upon HT BA failure Christian Lamparter
2012-06-09 7:59 ` Johannes Berg
2012-06-09 12:14 ` Christian Lamparter [this message]
2012-06-09 12:28 ` [RFC v2] " Johannes Berg
2012-06-09 14:01 ` [RFC v3] " Christian Lamparter
2012-06-10 5:13 ` Emmanuel Grumbach
2012-06-10 12:20 ` Christian Lamparter
2012-06-10 18:55 ` Emmanuel Grumbach
2012-06-09 12:20 ` BA session issue due to old BARs? Helmut Schaa
2012-06-09 14:23 ` Christian Lamparter
2012-06-09 14:55 ` Felix Fietkau
2012-06-09 17:40 ` Christian Lamparter
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=201206091414.41940.chunkeey@googlemail.com \
--to=chunkeey@googlemail.com \
--cc=helmut.schaa@googlemail.com \
--cc=johannes@sipsolutions.net \
--cc=linux-wireless@vger.kernel.org \
--cc=mar.kolya@gmail.com \
--cc=mikolaj.q@wp.pl \
--cc=nbd@openwrt.org \
--cc=per-erik.westerberg@bredband.net \
--cc=quantheory@gmail.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).