kmemleak: Do not corrupt the object_list during clean-up
authorCatalin Marinas <catalin.marinas@arm.com>
Fri, 4 Oct 2019 13:46:24 +0000 (14:46 +0100)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 14 Oct 2019 15:56:16 +0000 (08:56 -0700)
commit2abd839aa7e615f2bbc50c8ba7deb9e40d186768
tree1e72fe29e969967bfdbf547edc1f98ff8a1002fd
parent4f5cafb5cb8471e54afdc9054d973535614f7675
kmemleak: Do not corrupt the object_list during clean-up

In case of an error (e.g. memory pool too small), kmemleak disables
itself and cleans up the already allocated metadata objects. However, if
this happens early before the RCU callback mechanism is available,
put_object() skips call_rcu() and frees the object directly. This is not
safe with the RCU list traversal in __kmemleak_do_cleanup().

Change the list traversal in __kmemleak_do_cleanup() to
list_for_each_entry_safe() and remove the rcu_read_{lock,unlock} since
the kmemleak is already disabled at this point. In addition, avoid an
unnecessary metadata object rb-tree look-up since it already has the
struct kmemleak_object pointer.

Fixes: c5665868183f ("mm: kmemleak: use the memory pool for early allocations")
Reported-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reported-by: Marc Dionne <marc.c.dionne@gmail.com>
Reported-by: Ted Ts'o <tytso@mit.edu>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/kmemleak.c