[PATCH] locking/lockdep: Fix expected depth value in __lock_release()

[j.glisse@gmail.com]

From: Jérôme Glisse <jglisse@redhat.com>

In __lock_release() we are removing one entry from the stack and
rebuilding the hash chain by re-adding entry above the entry we
just removed. If the entry removed was between 2 entry of same
class then this 2 entry might be coalesced into one single entry
which in turns means that the lockdep_depth value will not be
incremented and thus the expected lockdep_depth value after this
operation will be wrong triggering an unjustified WARN_ONCE() at
the end of __lock_release().

This patch adjust the expect depth value by decrementing it if
what was previously 2 entry inside the stack are coalesced into
only one entry.

Note that __lock_set_class() does not suffer from same issue as
it adds a new class and thus can not lead to coalescing of stack
entry.

Signed-off-by: Jérôme Glisse <jglisse@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sasha Levin <sasha.levin@oracle.com>
---
 kernel/locking/lockdep.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
index 4e49cc4..cac5e21 100644
--- a/kernel/locking/lockdep.c
+++ b/kernel/locking/lockdep.c
@@ -3428,6 +3428,8 @@ found_it:
 	curr->curr_chain_key = hlock->prev_chain_key;
 
 	for (i++; i < depth; i++) {
+		int tmp = curr->lockdep_depth;
+
 		hlock = curr->held_locks + i;
 		if (!__lock_acquire(hlock->instance,
 			hlock_class(hlock)->subclass, hlock->trylock,
@@ -3435,6 +3437,13 @@ found_it:
 				hlock->nest_lock, hlock->acquire_ip,
 				hlock->references, hlock->pin_count))
 			return 0;
+		/*
+		 * If nest_lock is true and the lock we just removed allow two
+		 * lock of same class to be consolidated in only one held_lock
+		 * then the lockdep_depth count will not increase as we expect
+		 * it to. So adjust the expected depth value accordingly.
+		 */
+		depth -= (curr->lockdep_depth == tmp);
 	}
 
 	/*
-- 
1.8.3.1