On Fri, May 15, 2020 at 10:53:50AM -0600, Jens Axboe wrote:
> On 5/15/20 10:43 AM, Stefano Garzarella wrote:
> > This patch adds the new IORING_CQ_EVENTFD_DISABLED flag. It can be
> > used to disable/enable notifications from the kernel when a
> > request is completed and queued to the CQ ring.
> >
> > We also add two helpers function to check if the notifications are
> > enabled and to enable/disable them.
> >
> > If the kernel doesn't provide CQ ring flags, the notifications are
> > always enabled if an eventfd is registered.
> >
> > Signed-off-by: Stefano Garzarella <sgarzare@xxxxxxxxxx>
> > ---
> > src/include/liburing.h | 30 ++++++++++++++++++++++++++++++
> > src/include/liburing/io_uring.h | 7 +++++++
> > 2 files changed, 37 insertions(+)
> >
> > diff --git a/src/include/liburing.h b/src/include/liburing.h
> > index ea596f6..fe03547 100644
> > --- a/src/include/liburing.h
> > +++ b/src/include/liburing.h
> > @@ -9,7 +9,9 @@ extern "C" {
> > #include <sys/socket.h>
> > #include <sys/uio.h>
> > #include <sys/stat.h>
> > +#include <errno.h>
> > #include <signal.h>
> > +#include <stdbool.h>
> > #include <inttypes.h>
> > #include <time.h>
> > #include "liburing/compat.h"
> > @@ -445,6 +447,34 @@ static inline unsigned io_uring_cq_ready(struct io_uring *ring)
> > return io_uring_smp_load_acquire(ring->cq.ktail) - *ring->cq.khead;
> > }
> >
> > +static inline int io_uring_cq_eventfd_enable(struct io_uring *ring,
> > + bool enabled)
> > +{
> > + uint32_t flags;
> > +
> > + if (!ring->cq.kflags)
> > + return -ENOTSUP;
> > +
> > + flags = *ring->cq.kflags;
> > +
> > + if (enabled)
> > + flags &= ~IORING_CQ_EVENTFD_DISABLED;
> > + else
> > + flags |= IORING_CQ_EVENTFD_DISABLED;
> > +
> > + IO_URING_WRITE_ONCE(*ring->cq.kflags, flags);
> > +
> > + return 0;
> > +}
>
> The -ENOTSUP seems a bit odd, I wonder if we should even flag that as an
> error.
Do you think it's better to ignore the enabling/disabling if we don't have
the flag field available?
>
> The function should probably also be io_uring_cq_eventfd_toggle() or
> something like that, as it does both enable and disable.
>
> Either that, or have two functions, and enable and disable.
Okay, I'll change it in io_uring_cq_eventfd_toggle().
>
> The bigger question is probably how to handle kernels that don't
> have this feature. It'll succeed, but we'll still post events. Maybe
> the kernel side should have a feature flag that we can test?
I thought about that, and initially I added a
IORING_FEAT_EVENTFD_DISABLE, but then I realized that we are adding
the CQ 'flags' field together with the eventfd disabling feature.
So I supposed that if 'p->cq_off.flags' is not zero, than the kernel
supports CQ flags and also the IORING_CQ_EVENTFD_DISABLED bit.
Do you think that's okay, or should we add IORING_FEAT_EVENTFD_DISABLE
(or something similar)?
Thanks,
Stefano
