Re: [PATCH] workqueue: rust: add creation of workqueues

[Tejun Heo <tj@kernel.org>]

On Wed, Apr 16, 2025 at 10:10:41AM -1000, Tejun Heo wrote:
> On Wed, Apr 16, 2025 at 10:08:35PM +0200, Alice Ryhl wrote:
> ...
> > > This should be pretty cheap and we can probably enable this for everyone,
> > > but if the overhead is noticeable, this can be an optional behavior
> > > depending on a workqueue flag.
> > 
> > My only concern is that we're executing work items *before* the
> > deadline they specified. There could be work items that assume this
> > doesn't happen? But maybe it's okay. Otherwise, what you suggest seems
> > reasonable enough to me.
> 
> That's already what flush_delayed_work() does, so I don't think it'd be too
> surprising. Alternatively, we can go for canceling on draining/destruction
> but that'd be more surprising I think. As long as the behavior is documented
> clearly, I don't see problems with running and flushing them.

Also, note that self-requeueing work items may still be pending after
draining a workqueue as the draining is best effort. This is considered a
bug in the caller and, we trigger a warn and just skip freeing the
workqueue. This is again not great but may be acceptable for rust too. If
one wants to improve this, now that we have disable_work(), we can probably
trigger warn after X retries and then switch to disabling & flushing
afterwards.

Thanks.

-- 
tejun