Messages in this thread |  | | From | Catalin Marinas <> | Subject | [PATCH v2 0/4] Fix kmemleak races on the disable/error path | Date | Mon, 8 Jun 2015 15:29:14 +0100 |
| |
This is a follow-up to the initial patch fixing the kmemleak_scan page fault during kmemleak disabling:
http://article.gmane.org/gmane.linux.kernel.mm/133707
The first patch is pretty much the same, apart from a clearer (hopefully) commit log and code comment regarding the ordering of kmemleak_free_enabled with the scanning thread. The race between kmemleak_free() and kmemleak_do_cleanup() is fixed in the second patch.
The third patch fixes a potential deadlock on scan_mutex between the kmemleak_scan_thread and kmemleak_do_cleanup().
The fourth patch is more of a theoretical scenario but worth fixing the lock acquiring order.
Catalin Marinas (4): mm: kmemleak: Allow safe memory scanning during kmemleak disabling mm: kmemleak: Fix delete_object_*() race when called on the same memory block mm: kmemleak: Do not acquire scan_mutex in kmemleak_do_cleanup() mm: kmemleak: Avoid deadlock on the kmemleak object insertion error path
mm/kmemleak.c | 75 +++++++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 53 insertions(+), 22 deletions(-)
|  |