rcu: rcu_read_lock_bh_held(): disabling irqs also disables bh
authorEric Dumazet <eric.dumazet@gmail.com>
Wed, 22 Sep 2010 06:38:44 +0000 (08:38 +0200)
committerPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Thu, 23 Sep 2010 15:25:17 +0000 (08:25 -0700)
rcu_dereference_bh() doesnt know yet about hard irq being disabled, so
lockdep can trigger in netpoll_rx() after commit f0f9deae9e7c4 (netpoll:
Disable IRQ around RCU dereference in netpoll_rx)

Reported-by: Miles Lane <miles.lane@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Tested-by: Miles Lane <miles.lane@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
include/linux/rcupdate.h

index 9fbc54a2585d42cb9276adf2c2d168f53e883f63..83af1f8d8b746cde3b8369d7125f3cdfe07da6d3 100644 (file)
@@ -454,7 +454,7 @@ static inline notrace void rcu_read_unlock_sched_notrace(void)
  * Makes rcu_dereference_check() do the dirty work.
  */
 #define rcu_dereference_bh(p) \
-               rcu_dereference_check(p, rcu_read_lock_bh_held())
+               rcu_dereference_check(p, rcu_read_lock_bh_held() || irqs_disabled())
 
 /**
  * rcu_dereference_sched - fetch RCU-protected pointer, checking for RCU-sched