Re: [PATCH 0/2] io_uring: add a CQ ring flag to enable/disable eventfd notification

[Stefano Garzarella]

On Fri, May 15, 2020 at 09:13:33AM -0600, Jens Axboe wrote:
> On 5/15/20 8:34 AM, Stefano Garzarella wrote:
> > On Fri, May 15, 2020 at 08:24:58AM -0600, Jens Axboe wrote:
> >> On 5/15/20 4:54 AM, Stefano Garzarella wrote:
> >>> The first patch adds the new 'cq_flags' field for the CQ ring. It
> >>> should be written by the application and read by the kernel.
> >>>
> >>> The second patch adds a new IORING_CQ_NEED_WAKEUP flag that can be
> >>> used by the application to enable/disable eventfd notifications.
> >>>
> >>> I'm not sure the name is the best one, an alternative could be
> >>> IORING_CQ_NEED_EVENT.
> >>>
> >>> This feature can be useful if the application are using eventfd to be
> >>> notified when requests are completed, but they don't want a notification
> >>> for every request.
> >>> Of course the application can already remove the eventfd from the event
> >>> loop, but as soon as it adds the eventfd again, it will be notified,
> >>> even if it has already handled all the completed requests.
> >>>
> >>> The most important use case is when the registered eventfd is used to
> >>> notify a KVM guest through irqfd and we want a mechanism to
> >>> enable/disable interrupts.
> >>>
> >>> I also extended liburing API and added a test case here:
> >>> https://github.com/stefano-garzarella/liburing/tree/eventfd-disable
> >>
> >> Don't mind the feature, and I think the patches look fine. But the name
> >> is really horrible, I'd have no idea what that flag does without looking
> >> at the code or a man page. Why not call it IORING_CQ_EVENTFD_ENABLED or
> >> something like that? Or maybe IORING_CQ_EVENTFD_DISABLED, and then you
> >> don't have to muck with the default value either. The app would set the
> >> flag to disable eventfd, temporarily, and clear it again when it wants
> >> notifications again.
> > 
> > You're clearly right! :-) The name was horrible.
> 
> Sometimes you go down that path on naming and just can't think of
> the right one. I think we've all been there.

:-)

> 
> > I agree that IORING_CQ_EVENTFD_DISABLED should be the best.
> > I'll send a v2 changing the name and removing the default value.
> 
> Great thanks, and please do queue a pull for the liburing side too.

For the liburing side do you prefer a PR on github or posting the
patches on io-uring@xxxxxxxxxxxxxxx with 'liburing' tag?

Thanks,
Stefano