rcu: Warn on allegedly impossible rcu_read_unlock_special() from irq
authorLai Jiangshan <laijs@cn.fujitsu.com>
Wed, 30 Oct 2013 11:13:22 +0000 (04:13 -0700)
committerPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Thu, 12 Dec 2013 20:34:00 +0000 (12:34 -0800)
commit79a62f957e0b37c59610a96d018cc341aebb48f4
treeade9134ca3e64d87f7debbb102e9d1902763ac6b
parent24ef659a857c3cba40b64ea51ea4fce8d2fb7bbc
rcu: Warn on allegedly impossible rcu_read_unlock_special() from irq

After commit #10f39bb1b2c1 (rcu: protect __rcu_read_unlock() against
scheduler-using irq handlers), it is no longer possible to enter
the main body of rcu_read_lock_special() from an NMI, interrupt, or
softirq handler.  In theory, this implies that the check for "in_irq()
|| in_serving_softirq()" must always fail, so that in theory this check
could be removed entirely.

In practice, this commit wraps this condition with a WARN_ON_ONCE().
If this warning never triggers, then the condition will be removed
entirely.

[ paulmck: And one way of triggering the WARN_ON() is if a scheduling
  clock interrupt occurs in an RCU read-side critical section, setting
  RCU_READ_UNLOCK_NEED_QS, which is handled by rcu_read_unlock_special().
  Updated this commit to return if only that bit was set. ]

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
kernel/rcu/tree_plugin.h