Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Waiman Long <longman@redhat.com>
To:     Jens Axboe <axboe@kernel.dk>, Tejun Heo <tj@kernel.org>,
        Josef Bacik <josef@toxicpanda.com>,
        Zefan Li <lizefan.x@bytedance.com>,
        Johannes Weiner <hannes@cmpxchg.org>,
        Andrew Morton <akpm@linux-foundation.org>
Cc:     cgroups@vger.kernel.org, linux-block@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        =?UTF-8?q?Michal=20Koutn=C3=BD?= <mkoutny@suse.com>,
        "Dennis Zhou (Facebook)" <dennisszhou@gmail.com>,
        Waiman Long <longman@redhat.com>
Subject: [PATCH-block v3 0/2] blk-cgroup: Fix potential UAF & flush rstat at blkgs destruction path
Date:   Tue, 13 Dec 2022 13:44:44 -0500
Message-Id: <20221213184446.50181-1-longman@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

 v3:
  - Drop v2 patch 2 as it may not be needed.
  - Replace css_tryget() with percpu_ref_is_zero() in patch 1 as
    suggested by Tejun.
  - Expand comment on patch 2 to elaborate the reason for this patch.

 v2:
  - Remove unnecessary rcu_read_{lock|unlock} from
    cgroup_rstat_css_cpu_flush() in patch 3.

It was found that blkcg_destroy_blkgs() may be called with all blkcg
references gone. This may potentially cause user-after-free and so should
be fixed. The second patch flushes rstat when blkcg_destroy_blkgs().

Waiman Long (2):
  bdi, blk-cgroup: Fix potential UAF of blkcg
  blk-cgroup: Flush stats at blkgs destruction path

 block/blk-cgroup.c     | 22 ++++++++++++++++++++++
 include/linux/cgroup.h |  1 +
 kernel/cgroup/rstat.c  | 18 ++++++++++++++++++
 mm/backing-dev.c       |  8 ++++++--
 4 files changed, 47 insertions(+), 2 deletions(-)

-- 
2.31.1