[PATCH 2/5] lockdep: don't create the wrong dependency on hlock->check == 0

[Oleg Nesterov]

Test-case:

        DEFINE_MUTEX(m1);
        DEFINE_MUTEX(m2);
        DEFINE_MUTEX(mx);

        void lockdep_should_complain(void)
        {
                lockdep_set_novalidate_class(&mx);

                // m1 -> mx -> m2
                mutex_lock(&m1);
                mutex_lock(&mx);
                mutex_lock(&m2);
                mutex_unlock(&m2);
                mutex_unlock(&mx);
                mutex_unlock(&m1);

                // m2 -> m1 ; should trigger the warning
                mutex_lock(&m2);
                mutex_lock(&m1);
                mutex_unlock(&m1);
                mutex_unlock(&m2);
        }

this doesn't trigger any warning, lockdep can't detect the trivial
deadlock.

This is because lock(&mx) correctly avoids m1 -> mx dependency, it
skips validate_chain() due to mx->check == 0. But lock(&m2) wrongly
adds mx -> m2 and thus m1 -> m2 is not created.

rcu_lock_acquire()->lock_acquire(check => 0) is fine due to read == 2,
so currently only __lockdep_no_validate__ can trigger this problem.

Signed-off-by: Oleg Nesterov <o...@redhat.com>
---
 kernel/locking/lockdep.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
index c6a7d9d..543e120 100644
--- a/kernel/locking/lockdep.c
+++ b/kernel/locking/lockdep.c
@@ -1934,12 +1934,12 @@ check_prevs_add(struct task_struct *curr, struct 
held_lock *next)
 
        for (;;) {
                int distance = curr->lockdep_depth - depth + 1;
-               hlock = curr->held_locks + depth-1;
+               hlock = curr->held_locks + depth - 1;
                /*
                 * Only non-recursive-read entries get new dependencies
                 * added:
                 */
-               if (hlock->read != 2) {
+               if (hlock->read != 2 && hlock->check) {
                        if (!check_prev_add(curr, hlock, next,
                                                distance, trylock_loop))
                                return 0;
-- 
1.5.5.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/