Re: [PATCH] kthread_work: add cancel_kthread_work[_sync]()

[Tejun Heo <tj@kernel.org>]

On Sat, Jul 26, 2014 at 12:04:01PM +0800, Lai Jiangshan wrote:
> When an object or a subsystem quits, we need to destroy the kthread_work
> which is used by the object or the subsystem.  We used to use
> flush_kthread_work().  But flush_kthread_work() has not any guarantee
> about the suspension of the work, this duty is pushed to the users.
> 
> So we introduce the cancel_kthread_work_sync() with a strict guarantee
> like cancel_work_sync() (workqueue).  We also introduce cancel_kthread_work()
> which can be used by users on some conditions.  And it is required for
> making the implementation of the cancel_kthread_work_sync() simpler.
> kthread_flush_work_fn() owns the running state of the kthread_worker
> and calls cancel_kthread_work() to cancel the possible requeued work.
> 
> Both cancel_kthread_work_sync() and cancel_kthread_work() share the
> code of flush_kthread_work() which also make the implementation simpler.
> 
> Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
> ---
>  include/linux/kthread.h |    2 +
>  kernel/kthread.c        |   78 ++++++++++++++++++++++++++++++++++++++--------
>  2 files changed, 66 insertions(+), 14 deletions(-)

We don't have any user.  Let's not implement features which aren't
used just for completeness.  If you can spot and convert users, please
be my guest.

Thanks.

-- 
tejun