On Mon, Aug 13, 2012 at 5:42 AM, Henrik Rydberg <rydberg@xxxxxxxxxxx> wrote:
> By sending a full frame of events at the same time, the irqsoff
> latency at heavy load is brought down from 200 us to 100 us.
>
> Signed-off-by: Henrik Rydberg <rydberg@xxxxxxxxxxx>
> ---
> drivers/input/evdev.c | 68 +++++++++++++++++++++++++++++++++++----------------
> 1 file changed, 47 insertions(+), 21 deletions(-)
>
> diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c
> index a0692c5..1bf4ce5 100644
> --- a/drivers/input/evdev.c
> +++ b/drivers/input/evdev.c
> @@ -54,16 +54,9 @@ struct evdev_client {
> static struct evdev *evdev_table[EVDEV_MINORS];
> static DEFINE_MUTEX(evdev_table_mutex);
>
> -static void evdev_pass_event(struct evdev_client *client,
> - struct input_event *event,
> - ktime_t mono, ktime_t real)
> +static void __pass_event(struct evdev_client *client,
> + const struct input_event *event)
> {
> - event->time = ktime_to_timeval(client->clkid == CLOCK_MONOTONIC ?
> - mono : real);
> -
> - /* Interrupts are disabled, just acquire the lock. */
> - spin_lock(&client->buffer_lock);
> -
> client->buffer[client->head++] = *event;
> client->head &= client->bufsize - 1;
>
> @@ -86,42 +79,74 @@ static void evdev_pass_event(struct evdev_client *client,
> client->packet_head = client->head;
> kill_fasync(&client->fasync, SIGIO, POLL_IN);
> }
> +}
> +
> +static void evdev_pass_values(struct evdev_client *client,
> + const struct input_value *vals, size_t count,
> + ktime_t mono, ktime_t real)
> +{
> + struct evdev *evdev = client->evdev;
> + const struct input_value *v;
> + struct input_event event;
> + bool wakeup = false;
> +
> + event.time = ktime_to_timeval(client->clkid == CLOCK_MONOTONIC ?
> + mono : real);
> +
> + /* Interrupts are disabled, just acquire the lock. */
> + spin_lock(&client->buffer_lock);
> +
> + for (v = vals; v != vals + count; v++) {
> + event.type = v->type;
> + event.code = v->code;
> + event.value = v->value;
> + __pass_event(client, &event);
> + if (v->type == EV_SYN && v->code == SYN_REPORT)
> + wakeup = true;
> + }
>
> spin_unlock(&client->buffer_lock);
> +
> + if (wakeup)
> + wake_up_interruptible(&evdev->wait);
> }
>
> /*
> - * Pass incoming event to all connected clients.
> + * Pass incoming events to all connected clients.
> */
> -static void evdev_event(struct input_handle *handle,
> - unsigned int type, unsigned int code, int value)
> +static void evdev_events(struct input_handle *handle,
> + const struct input_value *vals, size_t count)
> {
> struct evdev *evdev = handle->private;
> struct evdev_client *client;
> - struct input_event event;
> ktime_t time_mono, time_real;
>
> time_mono = ktime_get();
> time_real = ktime_sub(time_mono, ktime_get_monotonic_offset());
>
> - event.type = type;
> - event.code = code;
> - event.value = value;
> -
> rcu_read_lock();
>
> client = rcu_dereference(evdev->grab);
>
> if (client)
> - evdev_pass_event(client, &event, time_mono, time_real);
> + evdev_pass_values(client, vals, count, time_mono, time_real);
> else
> list_for_each_entry_rcu(client, &evdev->client_list, node)
> - evdev_pass_event(client, &event, time_mono, time_real);
> + evdev_pass_values(client, vals, count,
> + time_mono, time_real);
Hi Henrik,
Reading the time just once and applying it as the timestamp to an
entire frame is very nice.
However, is it ever possible for the SYN_REPORT to get delayed until
the next batch of input_values, therefore breaking the assumption that
the SYN_REPORT timestamp applies to the rest of the input_values for
its frame?
Also, bonus points if the input driver could set this input frame
timestamp based on when it first saw a hardware interrupt rather then
when evdev gets around to sending the frame to userspace. This could
potentially remove a lot of the timing jitter userspace sees when
computing ballistics based on input event timestamps.
Thanks!
-Daniel
>
> rcu_read_unlock();
> +}
>
> - if (type == EV_SYN && code == SYN_REPORT)
> - wake_up_interruptible(&evdev->wait);
> +/*
> + * Pass incoming event to all connected clients.
> + */
> +static void evdev_event(struct input_handle *handle,
> + unsigned int type, unsigned int code, int value)
> +{
> + struct input_value vals[] = { { type, code, value } };
> +
> + evdev_events(handle, vals, 1);
> }
>
> static int evdev_fasync(int fd, struct file *file, int on)
> @@ -1050,6 +1075,7 @@ MODULE_DEVICE_TABLE(input, evdev_ids);
>
> static struct input_handler evdev_handler = {
> .event = evdev_event,
> + .events = evdev_events,
> .connect = evdev_connect,
> .disconnect = evdev_disconnect,
> .fops = &evdev_fops,
> --
> 1.7.11.4
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-input" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
