Merge branch 'core-futexes-for-linus' of...

Merge branch 'core-futexes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'core-futexes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  futex: Detect mismatched requeue targets
  futex: Correct futex_wait_requeue_pi() commentary

kernel/futex.c

@@ -115,6 +115,9 @@ struct futex_q {
 	/* rt_waiter storage for requeue_pi: */
 	struct rt_mutex_waiter *rt_waiter;
 
+	/* The expected requeue pi target futex key: */
+	union futex_key *requeue_pi_key;
+
 	/* Bitset for the optional bitmasked wakeup */
 	u32 bitset;
 };
@@ -1089,6 +1092,10 @@ static int futex_proxy_trylock_atomic(u32 __user *pifutex,
 	if (!top_waiter)
 		return 0;
 
+	/* Ensure we requeue to the expected futex. */
+	if (!match_futex(top_waiter->requeue_pi_key, key2))
+		return -EINVAL;
+
 	/*
 	 * Try to take the lock for top_waiter.  Set the FUTEX_WAITERS bit in
 	 * the contended case or if set_waiters is 1.  The pi_state is returned
@@ -1276,6 +1283,12 @@ retry_private:
 			continue;
 		}
 
+		/* Ensure we requeue to the expected futex for requeue_pi. */
+		if (requeue_pi && !match_futex(this->requeue_pi_key, &key2)) {
+			ret = -EINVAL;
+			break;
+		}
+
 		/*
 		 * Requeue nr_requeue waiters and possibly one more in the case
 		 * of requeue_pi if we couldn't acquire the lock atomically.
@@ -1751,6 +1764,7 @@ static int futex_wait(u32 __user *uaddr, int fshared,
 	q.pi_state = NULL;
 	q.bitset = bitset;
 	q.rt_waiter = NULL;
+	q.requeue_pi_key = NULL;
 
 	if (abs_time) {
 		to = &timeout;
@@ -1858,6 +1872,7 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared,
 
 	q.pi_state = NULL;
 	q.rt_waiter = NULL;
+	q.requeue_pi_key = NULL;
 retry:
 	q.key = FUTEX_KEY_INIT;
 	ret = get_futex_key(uaddr, fshared, &q.key, VERIFY_WRITE);
@@ -2118,11 +2133,11 @@ int handle_early_requeue_pi_wakeup(struct futex_hash_bucket *hb,
  * We call schedule in futex_wait_queue_me() when we enqueue and return there
  * via the following:
  * 1) wakeup on uaddr2 after an atomic lock acquisition by futex_requeue()
- * 2) wakeup on uaddr2 after a requeue and subsequent unlock
- * 3) signal (before or after requeue)
- * 4) timeout (before or after requeue)
+ * 2) wakeup on uaddr2 after a requeue
+ * 3) signal
+ * 4) timeout
  *
- * If 3, we setup a restart_block with futex_wait_requeue_pi() as the function.
+ * If 3, cleanup and return -ERESTARTNOINTR.
  *
  * If 2, we may then block on trying to take the rt_mutex and return via:
  * 5) successful lock
@@ -2130,7 +2145,7 @@ int handle_early_requeue_pi_wakeup(struct futex_hash_bucket *hb,
  * 7) timeout
  * 8) other lock acquisition failure
  *
- * If 6, we setup a restart_block with futex_lock_pi() as the function.
+ * If 6, return -EWOULDBLOCK (restarting the syscall would do the same).
  *
  * If 4 or 7, we cleanup and return with -ETIMEDOUT.
  *
@@ -2169,15 +2184,16 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared,
 	debug_rt_mutex_init_waiter(&rt_waiter);
 	rt_waiter.task = NULL;
 
-	q.pi_state = NULL;
-	q.bitset = bitset;
-	q.rt_waiter = &rt_waiter;
-
 	key2 = FUTEX_KEY_INIT;
 	ret = get_futex_key(uaddr2, fshared, &key2, VERIFY_WRITE);
 	if (unlikely(ret != 0))
 		goto out;
 
+	q.pi_state = NULL;
+	q.bitset = bitset;
+	q.rt_waiter = &rt_waiter;
+	q.requeue_pi_key = &key2;
+
 	/* Prepare to wait on uaddr. */
 	ret = futex_wait_setup(uaddr, val, fshared, &q, &hb);
 	if (ret)
@@ -2248,14 +2264,11 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared,
 			rt_mutex_unlock(pi_mutex);
 	} else if (ret == -EINTR) {
 		/*
-		 * We've already been requeued, but we have no way to
-		 * restart by calling futex_lock_pi() directly. We
-		 * could restart the syscall, but that will look at
-		 * the user space value and return right away. So we
-		 * drop back with EWOULDBLOCK to tell user space that
-		 * "val" has been changed. That's the same what the
-		 * restart of the syscall would do in
-		 * futex_wait_setup().
+		 * We've already been requeued, but cannot restart by calling
+		 * futex_lock_pi() directly. We could restart this syscall, but
+		 * it would detect that the user space "val" changed and return
+		 * -EWOULDBLOCK.  Save the overhead of the restart and return
+		 * -EWOULDBLOCK directly.
 		 */
 		ret = -EWOULDBLOCK;
 	}