[asterisk-dev] [Code Review] 931: Remove 'chans' array from dahdi-base.c

Shaun Ruffell reviewboard at asterisk.org
Thu Apr 11 09:19:31 CDT 2013


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/931/
-----------------------------------------------------------

(Updated April 11, 2013, 2:19 p.m.)


Status
------

This change has been marked as submitted.


Review request for Asterisk Developers and Tzafrir Cohen.


Repository: DAHDI


Description
-------

This review is for the 11 commits on http://github.com/sruffell/dahdi-linux/tree/chan_list (b07adb15484398f8fb37...8425d20d6e84766590ec)  It completely removes the 'chans' array from dahdi-base.c, but keeps the external behavior the same, namely channels are numbered in registration order.

The point of these changes is to eventually support renumbering spans / channels on the fly as part of supporting hot pluggable spans, but this change set has the affect of reducing memory usage for small/normal installations (since arrays no longer need to be setup for spans and chans).


Diffs
-----

  linux/trunk/drivers/dahdi/dahdi-base.c 9390 
  linux/trunk/include/dahdi/kernel.h 9390 

Diff: https://reviewboard.asterisk.org/r/931/diff/


Testing
-------

Again, hasn't gone through a complete regression, but I've been using it as the base for most of my work in order to ensure that there aren't any performance regressions.

With DAHDI_CHUNKSIZE == 8, the practical limit is still something under 1024 channels due to the masterspan processing overhead, but this change, in conjunction with some improvements/changes there (default CHUNKSIZE something other than 8? Reduced locking?) this could be part of a solution to allow more channels on a single system without recompiling.


Thanks,

Shaun Ruffell

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20130411/ae6827e1/attachment.htm>


More information about the asterisk-dev mailing list