On 12/14/22 11:54, Jens Axboe wrote:
On 12/13/22 12:53 PM, Waiman Long wrote:
On 12/13/22 14:29, Tejun Heo wrote:
On Tue, Dec 13, 2022 at 01:44:45PM -0500, Waiman Long wrote:
Commit 59b57717fff8 ("blkcg: delay blkg destruction until after
writeback has finished") delayed call to blkcg_destroy_blkgs() to
cgwb_release_workfn(). However, it is done after a css_put() of blkcg
which may be the final put that causes the blkcg to be freed as RCU
read lock isn't held.
Another place where blkcg_destroy_blkgs() can be called indirectly via
blkcg_unpin_online() is from the offline_css() function called from
css_killed_work_fn(). Over there, the potentially final css_put() call
is issued after offline_css().
By adding a css_tryget() into blkcg_destroy_blkgs() and warning its
failure, the following stack trace was produced in a test system on
bootup.
This doesn't agree with the code anymore. Otherwise
Acked-by: Tejun Heo <tj@xxxxxxxxxx>
Sorry, I overlooked the commit log in my update. I will update it if I need another version, or Jens can make the following edit:
css_tryget() -> percpu_ref_is_zero().
Since the other one also needs an edit, would be great if you could
just send out a v4.
Sure, will do that.
Cheers,
Longman
