[dtaht/ceropackages] c02d8b: Working alternate congestion control selection scr...

[GitHub]

  Branch: refs/heads/master
  Home:   https://github.com/dtaht/ceropackages
  Commit: c02d8b8a663de3a8b51b71d9b892740b702daa99
      https://github.com/dtaht/ceropackages/commit/c02d8b8a663de3a8b51b71d9b892740b702daa99
  Author: Dave Taht <dave.taht at bufferbloat.net>
  Date:   2012-02-14 (Tue, 14 Feb 2012)

  Changed paths:
    M net/alttcp/files/alttcp.conf
  M net/alttcp/files/alttcp.init

  Log Message:
  -----------
  Working alternate congestion control selection script

there is now a config file and startup script that works
for selecting tcp congestion control algorithms via uci,
and ultimately, luci.


  Commit: 44f8febbd34686564516c3261a911bd7cffcf714
      https://github.com/dtaht/ceropackages/commit/44f8febbd34686564516c3261a911bd7cffcf714
  Author: Dave Taht <dave.taht at bufferbloat.net>
  Date:   2012-02-14 (Tue, 14 Feb 2012)

  Changed paths:
    M net/aqm-scripts/files/usr/lib/aqm/generate.sh

  Log Message:
  -----------
  Basic ipv6 support for openwrt modified shaper

This does nothing more than add ipv6 iptables calls to the
original qos script. In the long run (assuming this script is kept)
source and destination ip addresses need to be checked as
to whether they are ipv4 or ipv6 addresses.

Similarly, the conntrack support is rather suspect.


  Commit: be09b8c15b6dc6bf4cb7da3112c598138a9c77ef
      https://github.com/dtaht/ceropackages/commit/be09b8c15b6dc6bf4cb7da3112c598138a9c77ef
  Author: Dave Taht <dave.taht at bufferbloat.net>
  Date:   2012-02-14 (Tue, 14 Feb 2012)

  Changed paths:
    M net/aqm-scripts/files/usr/lib/aqm/tcrules.awk

  Log Message:
  -----------
  SFQ is limited in packets. RED is in bytes. tcrules.awk conflated these

The original openwrt shaper took the RED byte calculation...
and reused it to specify a limit for sfq. However, sfq uses
packets rather than bytes, so it was specifying, say 16000 bytes
and translating that to 16000 packets.

This was not an error I prior to 3.3, because SFQ had a
hard coded limit of 127 packets. It is now.

So this commit puts a lower and upper bound on the maximum packets
that is sane, but is not pre 3.2 compatible.


  Commit: efad3d4771c39ec678c12e350a69a02746dd4268
      https://github.com/dtaht/ceropackages/commit/efad3d4771c39ec678c12e350a69a02746dd4268
  Author: Dave Taht <dave.taht at bufferbloat.net>
  Date:   2012-02-14 (Tue, 14 Feb 2012)

  Changed paths:
    M net/aqm-scripts/files/usr/lib/aqm/tcrules.awk

  Log Message:
  -----------
  Switch to sfqred throughout

The math goes funky. The ipv6 support doesn't appear to do anything...
need to set up a test of ipv4 and ipv6 competing against each other.


  Commit: c0dc7e09b74370c6bed16d5161b2c5a59f60a823
      https://github.com/dtaht/ceropackages/commit/c0dc7e09b74370c6bed16d5161b2c5a59f60a823
  Author: Dave Taht <dave.taht at bufferbloat.net>
  Date:   2012-02-14 (Tue, 14 Feb 2012)

  Changed paths:
    M net/aqm-scripts/files/usr/lib/aqm/generate.sh
  M net/aqm-scripts/files/usr/lib/aqm/tcrules.awk

  Log Message:
  -----------
  ipv6 filtering added... Still seems to mistreat ipv4 vs ipv6

ipv6 and ipv4 end up in different hfsc buckets, and shouldn't.

[  5]  0.0-61.2 sec  4.00 MBytes   548 Kbits/sec
[  4]  0.0-63.4 sec  1.00 MBytes   132 Kbits/sec


Compare: https://github.com/dtaht/ceropackages/compare/2109066...c0dc7e0