Paul Querna wrote:
[EMAIL PROTECTED] wrote:
Author: mturk
Date: Sun Apr 13 01:31:03 2008
New Revision: 647540
....
Modified: apr/apr/trunk/poll/unix/epoll.c
URL:
http://svn.apache.org/viewvc/apr/apr/trunk/poll/unix/epoll.c?rev=647540&r1=647539&r2=647540&view=diff
==============================================================================
--- apr/apr/trunk/poll/unix/epoll.c (original)
+++ apr/apr/trunk/poll/unix/epoll.c Sun Apr 13 01:31:03 2008
@@ -68,6 +68,8 @@
#if APR_HAS_THREADS
/* A thread mutex to protect operations on the rings */
apr_thread_mutex_t *ring_lock;
+ /* Pipe descriptors used for wakeup */
+ apr_file_t *wakeup_pipe[2]; #endif
Why are all of these changes hidden behind APR_HAS_THREADS?
I don't think they should be.
The idea behind wakeup is that you unblock some thread
that is blocked on the pollset_poll call, so threading
is presumed and only then it makes sense.
However I'm not sure if this can be used from parent/child
calling wakeup in parent, causing unblocking in child.
If it can, then it could be compiled even without
threading support.
Regards
--
(TM)
