[svn-commits] rmudgett: branch group/issue14068 r199816 - /team/group/issue14068/channels/

[SVN commits to the Digium repositories]

Author: rmudgett
Date: Tue Jun  9 13:43:16 2009
New Revision: 199816

URL: http://svn.asterisk.org/svn-view/asterisk?view=rev&rev=199816
Log:
Added theoretical deadlock possibility ignored comment.

Modified:
    team/group/issue14068/channels/chan_dahdi.c

Modified: team/group/issue14068/channels/chan_dahdi.c
URL: http://svn.asterisk.org/svn-view/asterisk/team/group/issue14068/channels/chan_dahdi.c?view=diff&rev=199816&r1=199815&r2=199816
==============================================================================
--- team/group/issue14068/channels/chan_dahdi.c (original)
+++ team/group/issue14068/channels/chan_dahdi.c Tue Jun  9 13:43:16 2009
@@ -13780,6 +13780,16 @@
 							c = dahdi_new(pri->pvts[chanpos], AST_STATE_RING, 0, SUB_REAL, law, e->ring.ctype);
 							ast_mutex_lock(&pri->lock);
 							if (c) {
+								/*
+								 * It is reasonably safe to set the following
+								 * channel variables while the PRI and DAHDI private
+								 * structures are locked.  The PBX has not been
+								 * started yet and it is unlikely that any other task
+								 * will do anything with the channel we have just
+								 * created.  To fix the locking so there is no
+								 * possibility of a deadlock would require some
+								 * redesign of this channel driver.
+								 */
 								if (!ast_strlen_zero(e->ring.callingsubaddr)) {
 									pbx_builtin_setvar_helper(c, "CALLINGSUBADDR", e->ring.callingsubaddr);
 								}