On Thu, 1 Aug 2024 15:00:23 -0700 Allen wrote:

quoted

Could you shed some light in the cover letter or this patch why
tasklet_enable() is converted to enable_and_queue_work() at
the face of it those two do not appear to do the same thing?  

With the transition to workqueues, the implementation on the workqueue side is:

tasklet_enable() -> enable_work() + queue_work()

Ref: https://lore.kernel.org/all/20240227172852.2386358-7-tj@kernel.org/ (local)

enable_and_queue_work() is a helper which combines the two calls.

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=474a549ff4c989427a14fdab851e562c8a63fe24

Hope this answers your question.

To an extent. tj says "unconditionally scheduling the work item after
enable_work() returns %true should work for most users." 
You need to include the explanation of the conversion not being 1:1 
in the commit message, and provide some analysis why it's fine for this
user.