process-shared barrier support, based on discussion with bdonlan

this implementation is rather heavy-weight, but it's the first
solution i've found that's actually correct. all waiters actually wait
twice at the barrier so that they can synchronize exit, and they hold
a "vm lock" that prevents changes to virtual memory mappings (and
blocks pthread_barrier_destroy) until all waiters are finished
inspecting the barrier.

thus, it is safe for any thread to destroy and/or unmap the barrier's
memory as soon as pthread_barrier_wait returns, without further
synchronization.

1 files changed, 5 insertions, 3 deletions

diff --git a/src/internal/pthread_impl.h b/src/internal/pthread_impl.h
index bbb4502f..82d62426 100644
--- a/src/internal/pthread_impl.h
+++ b/src/internal/pthread_impl.h
@@ -73,10 +73,12 @@ struct __timer {
 #define _c_waiters2 __u.__i[7]
 #define _rw_lock __u.__i[0]
 #define _rw_waiters __u.__i[1]
-#define _b_inst __u.__p[0]
+#define _b_lock __u.__i[0]
+#define _b_waiters __u.__i[1]
 #define _b_limit __u.__i[2]
-#define _b_lock __u.__i[3]
-#define _b_waiters __u.__i[4]
+#define _b_count __u.__i[3]
+#define _b_seq __u.__i[4]
+#define _b_inst __u.__p[4]
 
 #include "pthread_arch.h"