Messages in this thread Patch in this message | | | From | "Paul E. McKenney" <> | Subject | [PATCH tip/core/rcu 3/3] rcu: place root rcu_node structure in separate lockdep class | Date | Tue, 6 Oct 2009 21:48:18 -0700 |
| |
From: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Before this patch, all of the rcu_node structures were in the same lockdep class, so that lockdep would complain when rcu_preempt_offline_tasks() acquired the root rcu_node structure's lock while holding one of the leaf rcu_nodes' locks. This patch changes rcu_init_one() to use a separate spin_lock_init() for the root rcu_node structure's lock than is used for that of all of the rest of the rcu_node structures, which puts the root rcu_node structure's lock in its own lockdep class.
Suggested-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> --- kernel/rcutree.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-)
diff --git a/kernel/rcutree.c b/kernel/rcutree.c index d8d9865..705f02a 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -1647,7 +1647,8 @@ static void __init rcu_init_one(struct rcu_state *rsp) cpustride *= rsp->levelspread[i]; rnp = rsp->level[i]; for (j = 0; j < rsp->levelcnt[i]; j++, rnp++) { - spin_lock_init(&rnp->lock); + if (rnp != rcu_get_root(rsp)) + spin_lock_init(&rnp->lock); rnp->gpnum = 0; rnp->qsmask = 0; rnp->qsmaskinit = 0; @@ -1670,6 +1671,7 @@ static void __init rcu_init_one(struct rcu_state *rsp) INIT_LIST_HEAD(&rnp->blocked_tasks[1]); } } + spin_lock_init(&rcu_get_root(rsp)->lock); } /* -- 1.5.2.5
| |