-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/4389/
-----------------------------------------------------------
(Updated Jan. 29, 2015, 7:24 p.m.)
Review request for Asterisk Developers.
Changes
-------
John added one more change.
In res_pjsip_refer, some ordering was modified to avoid reading from a
refer_progress_notification after it had been freed. This was because the
queued task could sometimes run and complete before the check of the
notification state.
Bugs: ASTERISK-24736
https://issues.asterisk.org/jira/browse/ASTERISK-24736
Repository: Asterisk
Description
-------
John Hardin of Digium did some investigation into memory leaks he was seeing
and created a set of patches that fixes them. I've already given these patches
a look and have made a few small adjustments to them where necessary for coding
guidelines reasons. Most of the leak fixes are pretty straightforward, but one
of these bears a bit of explanation:
Prior to Asterisk 12, originating calls with channel variables required that
you transfer ownership of the created ast_variables structure away when calling
ast_pbx_outgoing_exten() or ast_pbx_outgoing_app(). This was presumably because
these functions could spawn threads that required these variables to not be
freed yet. However, in Asterisk 12, ast_pbx_outgoing_exten() and
ast_pbx_outgoing_app() were rewritten to not use the variables in the spawned
threads. However, we also stopped actually destroying the variables there, too.
So with this patch, responsibility for freeing the variables lies with the
original creator of the variables.
Diffs (updated)
-----
/branches/13/res/res_pjsip_refer.c 431302
/branches/13/pbx/pbx_spool.c 431302
/branches/13/main/xmldoc.c 431302
/branches/13/main/stasis_channels.c 431302
/branches/13/main/pbx.c 431302
/branches/13/main/manager.c 431302
/branches/13/main/bridge_after.c 431302
/branches/13/channels/chan_pjsip.c 431302
Diff: https://reviewboard.asterisk.org/r/4389/diff/
Testing
-------
John Hardin's testing with valgrind has shown that the memory leaks it had been
reported are no longer present.
Thanks,
Mark Michelson
--
_____________________________________________________________________
-- Bandwidth and Colocation Provided by http://www.api-digital.com --
asterisk-dev mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-dev
