[asterisk-commits] tilghman: branch 1.6.2 r268496 - in /branches/1.6.2: ./ channels/chan_iax2.c

[SVN commits to the Asterisk project]

Author: tilghman
Date: Sat Jun  5 19:38:16 2010
New Revision: 268496

URL: http://svnview.digium.com/svn/asterisk?view=rev&rev=268496
Log:
Merged revisions 268495 via svnmerge from 
https://origsvn.digium.com/svn/asterisk/trunk

........
  r268495 | tilghman | 2010-06-05 19:37:30 -0500 (Sat, 05 Jun 2010) | 2 lines
  
  Finally track down and eliminate the "FRACK! warnings from chan_iax2".
........

Modified:
    branches/1.6.2/   (props changed)
    branches/1.6.2/channels/chan_iax2.c

Propchange: branches/1.6.2/
------------------------------------------------------------------------------
Binary property 'trunk-merged' - no diff available.

Modified: branches/1.6.2/channels/chan_iax2.c
URL: http://svnview.digium.com/svn/asterisk/branches/1.6.2/channels/chan_iax2.c?view=diff&rev=268496&r1=268495&r2=268496
==============================================================================
--- branches/1.6.2/channels/chan_iax2.c (original)
+++ branches/1.6.2/channels/chan_iax2.c Sat Jun  5 19:38:16 2010
@@ -3227,7 +3227,16 @@
 
 retry:
 	if ((pvt = iaxs[callno])) {
+#if 0
+		/* iax2_destroy_helper gets called from this function later on.  When
+		 * called twice, we get the (previously) familiar FRACK! errors in
+		 * devmode, from the scheduler.  An alternative to this approach is to
+		 * reset the scheduler entries to -1 when they're deleted in
+		 * iax2_destroy_helper().  That approach was previously decided to be
+		 * "wrong" because "the memory is going to be deallocated anyway.  Why
+		 * should we be resetting those values?" */
 		iax2_destroy_helper(pvt);
+#endif
 	}
 
 	owner = pvt ? pvt->owner : NULL;