---

yaml
---
r: 349820
b: refs/heads/master
c: 34a64b6
h: refs/heads/master
v: v3

[refs]

@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 511a0868bed6694512348fc177cdfaf3fd97d0bb
+refs/heads/master: 34a64b6bb64b5cf193932e2b4394c5284732e008

trunk/kernel/srcu.c

@@ -450,10 +450,12 @@ static void __synchronize_srcu(struct srcu_struct *sp, int trycount)
  * synchronize_srcu - wait for prior SRCU read-side critical-section completion
  * @sp: srcu_struct with which to synchronize.
  *
- * Flip the completed counter, and wait for the old count to drain to zero.
- * As with classic RCU, the updater must use some separate means of
- * synchronizing concurrent updates.  Can block; must be called from
- * process context.
+ * Wait for the count to drain to zero of both indexes. To avoid the
+ * possible starvation of synchronize_srcu(), it waits for the count of
+ * the index=((->completed & 1) ^ 1) to drain to zero at first,
+ * and then flip the completed and wait for the count of the other index.
+ *
+ * Can block; must be called from process context.
  *
  * Note that it is illegal to call synchronize_srcu() from the corresponding
  * SRCU read-side critical section; doing so will result in deadlock.