PERFORCE change 156304 for review

[Robert Watson]

http://perforce.freebsd.org/chv.cgi?CH=156304

Change 156304 by rwatson at rwatson_freebsd_capabilities on 2009/01/17 22:37:14

	When procdesc decides the parent shouldn't be signaled on exit, we
	still need to lock the parent for the benefits of the scheduler,
	so do so.

Affected files ...

.. //depot/projects/trustedbsd/capabilities/src/sys/kern/kern_exit.c#7 edit

Differences ...

==== //depot/projects/trustedbsd/capabilities/src/sys/kern/kern_exit.c#7 (text+ko) ====

@@ -536,6 +536,8 @@
 				psignal(p->p_pptr, p->p_sigparent);
 		}
 #ifdef PROCDESC
+	} else {
+		PROC_LOCK(p->p_pptr);
 	}
 #endif
 	sx_xunlock(&proctree_lock);