From ad12d7d4b020e6be225a8d6494de69d051ec7f81 Mon Sep 17 00:00:00 2001
From: Lennart Poettering <lennart@poettering.net>
Date: Wed, 22 Apr 2009 19:10:09 +0200
Subject: memtrap,aupdate: split atomic update from memtrap into seperate
 aupdate framework

---
 src/pulsecore/aupdate.c | 129 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 129 insertions(+)
 create mode 100644 src/pulsecore/aupdate.c

(limited to 'src/pulsecore/aupdate.c')

diff --git a/src/pulsecore/aupdate.c b/src/pulsecore/aupdate.c
new file mode 100644
index 00000000..56ebb8e5
--- /dev/null
+++ b/src/pulsecore/aupdate.c
@@ -0,0 +1,129 @@
+/***
+  This file is part of PulseAudio.
+
+  Copyright 2009 Lennart Poettering
+
+  PulseAudio is free software; you can redistribute it and/or modify
+  it under the terms of the GNU Lesser General Public License as
+  published by the Free Software Foundation; either version 2.1 of the
+  License, or (at your option) any later version.
+
+  PulseAudio is distributed in the hope that it will be useful, but
+  WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+  General Public License for more details.
+
+  You should have received a copy of the GNU Lesser General Public
+  License along with PulseAudio; if not, write to the Free Software
+  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
+  USA.
+***/
+
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#include <pulse/xmalloc.h>
+
+#include <pulsecore/semaphore.h>
+#include <pulsecore/macro.h>
+#include <pulsecore/mutex.h>
+
+#include "aupdate.h"
+
+#define MSB (1U << (sizeof(unsigned)*8U-1))
+#define WHICH(n) (!!((n) & MSB))
+#define COUNTER(n) ((n) & ~MSB)
+
+struct pa_aupdate {
+    pa_atomic_t read_lock;
+    pa_mutex *write_lock;
+    pa_semaphore *semaphore;
+};
+
+pa_aupdate *pa_aupdate_new(void) {
+    pa_aupdate *a;
+
+    a = pa_xnew(pa_aupdate, 1);
+    pa_atomic_store(&a->read_lock, 0);
+    a->write_lock = pa_mutex_new(FALSE, FALSE);
+    a->semaphore = pa_semaphore_new(0);
+
+    return a;
+}
+
+void pa_aupdate_free(pa_aupdate *a) {
+    pa_assert(a);
+
+    pa_mutex_free(a->write_lock);
+    pa_semaphore_free(a->semaphore);
+
+    pa_xfree(a);
+}
+
+unsigned pa_aupdate_read_begin(pa_aupdate *a) {
+    unsigned n;
+
+    pa_assert(a);
+
+    /* Increase the lock counter */
+    n = (unsigned) pa_atomic_inc(&a->read_lock);
+
+    /* When n is 0 we have about 2^31 threads running that all try to
+     * access the data at the same time, oh my! */
+    pa_assert(COUNTER(n)+1 > 0);
+
+    /* The uppermost bit tells us which data to look at */
+    return WHICH(n);
+}
+
+void pa_aupdate_read_end(pa_aupdate *a) {
+    unsigned n;
+
+    pa_assert(a);
+
+    /* Decrease the lock counter */
+    n = (unsigned) pa_atomic_dec(&a->read_lock);
+
+    /* Make sure the counter was valid */
+    pa_assert(COUNTER(n) > 0);
+
+    /* Post the semaphore */
+    pa_semaphore_post(a->semaphore);
+}
+
+unsigned pa_aupdate_write_begin(pa_aupdate *a) {
+    unsigned n;
+
+    pa_assert(a);
+
+    pa_mutex_lock(a->write_lock);
+
+    n = (unsigned) pa_atomic_load(&a->read_lock);
+
+    return !WHICH(n);
+}
+
+unsigned pa_aupdate_write_swap(pa_aupdate *a) {
+    unsigned n;
+
+    pa_assert(a);
+
+    for (;;) {
+        n = (unsigned) pa_atomic_load(&a->read_lock);
+
+        /* If the read counter is > 0 wait; if it is 0 try to swap the lists */
+        if (COUNTER(n) > 0)
+            pa_semaphore_wait(a->semaphore);
+        else if (pa_atomic_cmpxchg(&a->read_lock, (int) n, (int) (n ^ MSB)))
+            break;
+    }
+
+    return WHICH(n);
+}
+
+void pa_aupdate_write_end(pa_aupdate *a) {
+    pa_assert(a);
+
+    pa_mutex_unlock(a->write_lock);
+}
-- 
cgit