linux-wireless.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Johannes Berg <johannes@sipsolutions.net>
To: linux-wireless <linux-wireless@vger.kernel.org>
Subject: converting mac80211 to TXQs entirely
Date: Thu, 05 Oct 2017 14:13:38 +0200	[thread overview]
Message-ID: <1507205618.2387.19.camel@sipsolutions.net> (raw)

Hi,

Part 1 is just a dump of my notes analysing the current TX scheme.


driver setup
============

non-QUEUE_CONTROL drivers
 * have >= 4 queues
   - per-AC queues [0-3]
 * have < 4 queues
   - all goes to queue 0

QUEUE_CONTROL drivers
 * hwsim (doesn't handle CAB correctly)
   - each vif: 0...3
   - each cab: 0
   - offchannel: 4
 * TI (doesn't set HOST_BROADCAST_PS_BUFFERING)
   - each vif: 4 queues (separate)
   - each cab: separate queue
   - offchannel: separate from all others
 * iwldvm/iwlmvm (doesn't set HOST_BROADCAST_PS_BUFFERING)
   - each vif: 4 queues (separate)
   - each cab: separate queue
   - offchannel: separate from all others (AUX)
 * ath9k (uses TXQ, sets HOST_BROADCAST_PS_BUFFERING)
   - each vif: 4 queues (shared based on chanctx)
   - each cab: all the same (# queues - 2)
   - offchannel: separate from all others
 * ath10k (may use TXQ, doesn't set HOST_BROADCAST_PS_BUFFERING)
   - each vif: 1 queue for all ACs
   - each cab: same queue as for ACs
   - offchannel: separate from all others


current TX scheme
=================
HOST_BROADCAST_PS_BUFFERING && IEEE80211_TX_CTL_SEND_AFTER_DTIM
 --> queue for ieee80211_get_buffered_bc()

!AP_LINK_PS && sta sleeping
 --> queue on sta->ps_tx_buf[ac] for wakeup/poll
 --> send on poll with IEEE80211_TX_CTRL_PS_RESPONSE
 --> send on wakeup as normal frame (with or without TXQ)
     [NB: with TXQs, this is buggy due to waking old TXQ before tx_filtered]

finally
 --> send directly (with or without TXQ)

if filtered TX status
 --> append to tx_filtered[ac] and use that before ps_tx_buf[ac]

TXQ scheme (where used)
=======================

MONITOR || IEEE80211_TX_CTL_SEND_AFTER_DTIM || IEEE80211_TX_CTRL_PS_RESPONSE || non-data
 --> send directly (to TX queue number as given above)
have STA
 --> per-STA/TID TXQ
otherwise
 --> per-VIF TXQ (for VLAN use AP instead)

             reply	other threads:[~2017-10-05 12:13 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-10-05 12:13 Johannes Berg [this message]
2017-10-05 15:39 ` converting mac80211 to TXQs entirely Johannes Berg
2017-10-05 16:28   ` Toke Høiland-Jørgensen
2017-10-05 16:43     ` Johannes Berg
2017-10-05 21:52       ` Toke Høiland-Jørgensen
2017-10-06  9:44         ` Johannes Berg
2017-10-06 10:17           ` Toke Høiland-Jørgensen
2017-10-06 10:18             ` Johannes Berg

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=1507205618.2387.19.camel@sipsolutions.net \
    --to=johannes@sipsolutions.net \
    --cc=linux-wireless@vger.kernel.org \
    /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).