Browse Source

Effectively revert 85bfc4cf7a59fff661cb7b732fb5a030a00f309d

The idea behind the reverting commit was that sqm will
be started several times during boot-up if it is instantiated
on an interface that receives hotplug events. At the point
before 85bfc4cf7a59fff661cb7b732fb5a030a00f309d sqm was started
three times, ones by init.d? and twice by hotplug. This seemed
quite involved and wasteful, so we tried to reduce the idle
work to the two hotplug restarts (guided by the observation that
the two hotplug restarts together took almost as long as the
initial start). The assumption was that all interfaces valid for
sqm would receive hotplug events dring start-up.

See https://forum.openwrt.org/viewtopic.php?id=60903 for an example
of a valid sqm configuration that does not receive hotplug events.
This configuration is described and advertised in the openwrt
wiki as bridged AP, see https://wiki.openwrt.org/doc/recipes/bridgedap
With this commit reverted sqm will increase start-up times by
roughly 10 seconds, a small price to pay to have sqm work after a
reboot on valid, recommended openwrt configurations.

Since a real revert did not work this effectively undoes
85bfc4cf7a59fff661cb7b732fb5a030a00f309d manually.

Signed-off-by: Sebastian Moeller <moeller0@gmx.de>
Sebastian Moeller 8 năm trước cách đây
mục cha
commit
90399ffc4d
1 tập tin đã thay đổi với 0 bổ sung7 xóa
  1. 0 7
      platform/openwrt/sqm-init

+ 0 - 7
platform/openwrt/sqm-init

@@ -21,10 +21,3 @@ stop()
 {
     /usr/lib/sqm/run.sh stop
 }
-
-boot()
-{
-    # We are being run through hotplug scripts, don't do anything at boot
-    logger -t SQM -s "sqm-scripts do nothing at boot, waiting for hotplug events."
-    return 0
-}