Lets make wifi fast again!
 help / color / mirror / Atom feed
From: Dave Taht <dave.taht@gmail.com>
To: Michael Braun <michael-dev@fami-braun.de>,
	make-wifi-fast@lists.bufferbloat.net
Subject: Re: [Make-wifi-fast] [PATCH 0/3] mac80211: multicast with AP_VLAN optimizations
Date: Sun, 25 Sep 2016 11:19:02 -0700	[thread overview]
Message-ID: <CAA93jw65cXUgjjZGxzBAvp8zb8+=LU0_4Uiy7Jj2PNUW_6DRfw@mail.gmail.com> (raw)
In-Reply-To: <1474821596-12155-1-git-send-email-michael-dev@fami-braun.de>

Dear Michael:

This looks pretty cool. I have a modified airtime analyzer over here,
(courtesy simon and sven) that counts all the airtime used in a
capture, that might be useful in analyzing the new behavior. We used
it to tear apart the airtime fairness stuff we're doing, and I'd
objected to the original code because it filtered out too much
multicast. Now (I think) - it counts all of it.

https://github.com/dtaht/airtime-pie-chart

One minor objection to what I see in this patch is that I've generally
thought (not backed by data!) that multicasting vs unicasting was not
a binary choice, that at some X number of stations it made sense to
fall back to multicast, and that maybe a better way was to chose the
highest possible (from data) multicast rate rather than a static one.

A possibly larger objection is that converting everything to
essentially "reliable multicast" is going to break the assumptions of
many a wifi routing protocol - which I don't mind, those assumptions
generally suck.

One of my tools for measuring multicast behaviors has been "uftp".

On Sun, Sep 25, 2016 at 9:39 AM, Michael Braun
<michael-dev@fami-braun.de> wrote:
> Hi,
>
> this series tries to optimize multicast delivery on access points with
> AP_VLAN interfaces.
>
> My setup is as follows: hostapd creates one AP_VLAN interface per station
> (per_sta_vif=1), which enables bridge igmp snooping to decide which
> stations need to receive a multicast packet.
>
> This series then
>  - avoids multicast packets by tracking the number of authenticated
>    stations per interface more accurately and drops packets if there is no
>    receiver connected,
>  - converts multicast packets to unicast packets for the most common
>    protocols, so they get delivered faster and more reliable. Additionally,
>    waking up non-receivers in other VLANs is avoided.
>  - speeds up station lookup if there is only one authenticated station
>    assigned to the AP_VLAN interface.
>
> Sincerely,
> M. Braun
>
> Michael Braun (3):
>   mac80211: filter multicast data packets on AP / AP_VLAN
>   mac80211: multicast to unicast conversion
>   mac80211: cache the only AP_VLAN station
>
>  net/mac80211/cfg.c            |  26 +++++-----
>  net/mac80211/debugfs_netdev.c |  38 ++++++++++++++
>  net/mac80211/ieee80211_i.h    |  61 ++++++++++++++++++++++
>  net/mac80211/rx.c             |   5 +-
>  net/mac80211/sta_info.c       |  41 +++++++++++----
>  net/mac80211/tx.c             | 115 ++++++++++++++++++++++++++++++++++++++++--
>  6 files changed, 258 insertions(+), 28 deletions(-)
>
> --
> 2.1.4
>



-- 
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org

           reply	other threads:[~2016-09-25 18:19 UTC|newest]

Thread overview: expand[flat|nested]  mbox.gz  Atom feed
 [parent not found: <1474821596-12155-1-git-send-email-michael-dev@fami-braun.de>]

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

  List information: https://lists.bufferbloat.net/postorius/lists/make-wifi-fast.lists.bufferbloat.net/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAA93jw65cXUgjjZGxzBAvp8zb8+=LU0_4Uiy7Jj2PNUW_6DRfw@mail.gmail.com' \
    --to=dave.taht@gmail.com \
    --cc=make-wifi-fast@lists.bufferbloat.net \
    --cc=michael-dev@fami-braun.de \
    /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