cgroups: fix probable race with put_css_set[_taskexit] and find_css_set
authorLai Jiangshan <laijs@cn.fujitsu.com>
Sun, 19 Oct 2008 03:28:03 +0000 (20:28 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 20 Oct 2008 15:52:38 +0000 (08:52 -0700)
commit146aa1bd0511f88ddb4e92fafa2b8aad4f2f65f3
treed7deb46b9a38f82f109b2126317899efbbce41c2
parent248736c2a57206388c86f8cdd3392ee986e84f9f
cgroups: fix probable race with put_css_set[_taskexit] and find_css_set

put_css_set_taskexit may be called when find_css_set is called on other
cpu.  And the race will occur:

put_css_set_taskexit side                    find_css_set side

                                        |
atomic_dec_and_test(&kref->refcount)    |
    /* kref->refcount = 0 */            |
....................................................................
                                        |  read_lock(&css_set_lock)
                                        |  find_existing_css_set
                                        |  get_css_set
                                        |  read_unlock(&css_set_lock);
....................................................................
__release_css_set                       |
....................................................................
                                        | /* use a released css_set */
                                        |

[put_css_set is the same. But in the current code, all put_css_set are
put into cgroup mutex critical region as the same as find_css_set.]

[akpm@linux-foundation.org: repair comments]
[menage@google.com: eliminate race in css_set refcounting]
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/cgroup.h
kernel/cgroup.c
kernel/cgroup_debug.c