rcu: kmemleak: Ignore kmemleak false positives when RCU-freeing objects
authorCatalin Marinas <catalin.marinas@arm.com>
Sat, 30 Sep 2023 17:46:56 +0000 (17:46 +0000)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 28 Nov 2023 17:20:02 +0000 (17:20 +0000)
commitb88cc37a8208d17a0298ec5c0c5bd519435aefee
treece5890b1d33daf1ae8aea88b56c7df2f6fb8400a
parent0756504578f28f3814219f414708536a16967ad7
rcu: kmemleak: Ignore kmemleak false positives when RCU-freeing objects

commit 5f98fd034ca6fd1ab8c91a3488968a0e9caaabf6 upstream.

Since the actual slab freeing is deferred when calling kvfree_rcu(), so
is the kmemleak_free() callback informing kmemleak of the object
deletion. From the perspective of the kvfree_rcu() caller, the object is
freed and it may remove any references to it. Since kmemleak does not
scan RCU internal data storing the pointer, it will report such objects
as leaks during the grace period.

Tell kmemleak to ignore such objects on the kvfree_call_rcu() path. Note
that the tiny RCU implementation does not have such issue since the
objects can be tracked from the rcu_ctrlblk structure.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Reported-by: Christoph Paasch <cpaasch@apple.com>
Closes: https://lore.kernel.org/all/F903A825-F05F-4B77-A2B5-7356282FBA2C@apple.com/
Cc: <stable@vger.kernel.org>
Tested-by: Christoph Paasch <cpaasch@apple.com>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
kernel/rcu/tree.c