2019-07-10 23:06:43

by Atif Niyaz

[permalink] [raw]
Subject: [PATCH] input: API for Setting a Timestamp from a Driver

Currently, evdev stamps time with timestamps acquired in
evdev_events. However, this timestamping may not be accurate in terms of
measuring when the actual event happened. This API allows any 3rd party
driver to be able to call input_set_timestamp, and provide a timestamp
that can be utilized in order to provide a more accurate sense of time
for the event

Signed-off-by: Atif Niyaz <[email protected]>
---
drivers/input/evdev.c | 42 ++++++++++++++++--------------------------
drivers/input/input.c | 17 +++++++++++++++++
include/linux/input.h | 38 ++++++++++++++++++++++++++++++++++++++
3 files changed, 71 insertions(+), 26 deletions(-)

diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c
index 867c2cfd0038..a331efa0a3f6 100644
--- a/drivers/input/evdev.c
+++ b/drivers/input/evdev.c
@@ -25,13 +25,6 @@
#include <linux/cdev.h>
#include "input-compat.h"

-enum evdev_clock_type {
- EV_CLK_REAL = 0,
- EV_CLK_MONO,
- EV_CLK_BOOT,
- EV_CLK_MAX
-};
-
struct evdev {
int open;
struct input_handle handle;
@@ -53,7 +46,7 @@ struct evdev_client {
struct fasync_struct *fasync;
struct evdev *evdev;
struct list_head node;
- unsigned int clk_type;
+ input_clk_t clk_type;
bool revoked;
unsigned long *evmasks[EV_CNT];
unsigned int bufsize;
@@ -150,16 +143,18 @@ static void __evdev_flush_queue(struct evdev_client *client, unsigned int type)
static void __evdev_queue_syn_dropped(struct evdev_client *client)
{
struct input_event ev;
- ktime_t time;
struct timespec64 ts;
+ ktime_t *time = input_get_timestamp(client->evdev->handle.dev);

- time = client->clk_type == EV_CLK_REAL ?
- ktime_get_real() :
- client->clk_type == EV_CLK_MONO ?
- ktime_get() :
- ktime_get_boottime();
+ switch (client->clk_type) {
+ case INPUT_CLK_REAL:
+ case INPUT_CLK_MONO:
+ ts = ktime_to_timespec64(time[client->clk_type]);
+ break;
+ default:
+ ts = ktime_to_timespec64(time[INPUT_CLK_BOOT]);
+ }

- ts = ktime_to_timespec64(time);
ev.input_event_sec = ts.tv_sec;
ev.input_event_usec = ts.tv_nsec / NSEC_PER_USEC;
ev.type = EV_SYN;
@@ -185,21 +180,21 @@ static void evdev_queue_syn_dropped(struct evdev_client *client)
spin_unlock_irqrestore(&client->buffer_lock, flags);
}

-static int evdev_set_clk_type(struct evdev_client *client, unsigned int clkid)
+static int evdev_set_clk_type(struct evdev_client *client, clockid_t clkid)
{
unsigned long flags;
- unsigned int clk_type;
+ input_clk_t clk_type;

switch (clkid) {

case CLOCK_REALTIME:
- clk_type = EV_CLK_REAL;
+ clk_type = INPUT_CLK_REAL;
break;
case CLOCK_MONOTONIC:
- clk_type = EV_CLK_MONO;
+ clk_type = INPUT_CLK_MONO;
break;
case CLOCK_BOOTTIME:
- clk_type = EV_CLK_BOOT;
+ clk_type = INPUT_CLK_BOOT;
break;
default:
return -EINVAL;
@@ -307,12 +302,7 @@ static void evdev_events(struct input_handle *handle,
{
struct evdev *evdev = handle->private;
struct evdev_client *client;
- ktime_t ev_time[EV_CLK_MAX];
-
- ev_time[EV_CLK_MONO] = ktime_get();
- ev_time[EV_CLK_REAL] = ktime_mono_to_real(ev_time[EV_CLK_MONO]);
- ev_time[EV_CLK_BOOT] = ktime_mono_to_any(ev_time[EV_CLK_MONO],
- TK_OFFS_BOOT);
+ ktime_t *ev_time = input_get_timestamp(handle->dev);

rcu_read_lock();

diff --git a/drivers/input/input.c b/drivers/input/input.c
index 7f3c5fcb9ed6..ae8b0ee58120 100644
--- a/drivers/input/input.c
+++ b/drivers/input/input.c
@@ -1894,6 +1894,23 @@ void input_free_device(struct input_dev *dev)
}
EXPORT_SYMBOL(input_free_device);

+/**
+ * input_get_timestamp - get timestamp for input events
+ * @dev: input device to get timestamp from
+ *
+ * A valid timestamp is a timestamp of non-zero value.
+ */
+ktime_t *input_get_timestamp(struct input_dev *dev)
+{
+ const ktime_t invalid_timestamp = ktime_set(0, 0);
+
+ if (!ktime_compare(dev->timestamp[INPUT_CLK_MONO], ktime_zero)) {
+ input_set_timestamp(dev, ktime_get());
+ }
+ return dev->timestamp;
+}
+EXPORT_SYMBOL(input_get_timestamp);
+
/**
* input_set_capability - mark device as capable of a certain event
* @dev: device that is capable of emitting or accepting event
diff --git a/include/linux/input.h b/include/linux/input.h
index 510e78558c10..3929b62ccbe5 100644
--- a/include/linux/input.h
+++ b/include/linux/input.h
@@ -33,6 +33,14 @@ struct input_value {
__s32 value;
};

+enum input_clock_type {
+ INPUT_CLK_REAL = 0,
+ INPUT_CLK_MONO,
+ INPUT_CLK_BOOT,
+ INPUT_CLK_MAX
+};
+typedef enum input_clock_type input_clk_t;
+
/**
* struct input_dev - represents an input device
* @name: name of the device
@@ -114,6 +122,8 @@ struct input_value {
* @vals: array of values queued in the current frame
* @devres_managed: indicates that devices is managed with devres framework
* and needs not be explicitly unregistered or freed.
+ * @timestamp: storage for a timestamp set by input_set_timestamp called
+ * by a driver
*/
struct input_dev {
const char *name;
@@ -184,6 +194,8 @@ struct input_dev {
struct input_value *vals;

bool devres_managed;
+
+ ktime_t timestamp[INPUT_CLK_MAX];
};
#define to_input_dev(d) container_of(d, struct input_dev, dev)

@@ -382,6 +394,32 @@ void input_close_device(struct input_handle *);

int input_flush_device(struct input_handle *handle, struct file *file);

+/**
+ * input_set_timestamp - set timestamp for input events
+ * @dev: input device to set timestamp for
+ * @timestamp: the time at which the event has occurred
+ * in CLOCK_MONOTONIC
+ *
+ * This function is intended to provide to the input system a more
+ * accurate time of when an event actually occurred. The driver should
+ * call this function as soon as a timestamp is acquired ensuring
+ * clock conversions in input_set_timestamp are done correctly.
+ *
+ * The system entering a suspend between timestamp acquisition and
+ * calling input_set_timestamp can result in inaccurate conversions.
+ *
+ */
+static inline void input_set_timestamp(struct input_dev *dev,
+ ktime_t timestamp)
+{
+ dev->timestamp[INPUT_CLK_MONO] = timestamp;
+ dev->timestamp[INPUT_CLK_REAL] = ktime_mono_to_real(timestamp);
+ dev->timestamp[INPUT_CLK_BOOT] = ktime_mono_to_any(
+ timestamp, TK_OFFS_BOOT);
+}
+
+ktime_t *input_get_timestamp(struct input_dev *dev);
+
void input_event(struct input_dev *dev, unsigned int type, unsigned int code, int value);
void input_inject_event(struct input_handle *handle, unsigned int type, unsigned int code, int value);

--
2.22.0.410.gd8fdbe21b5-goog


2019-07-12 06:42:14

by Dmitry Torokhov

[permalink] [raw]
Subject: Re: [PATCH] input: API for Setting a Timestamp from a Driver

Hi Atif,

On Wed, Jul 10, 2019 at 04:04:10PM -0700, Atif Niyaz wrote:
> Currently, evdev stamps time with timestamps acquired in
> evdev_events. However, this timestamping may not be accurate in terms of
> measuring when the actual event happened. This API allows any 3rd party
> driver to be able to call input_set_timestamp, and provide a timestamp
> that can be utilized in order to provide a more accurate sense of time
> for the event
>
> Signed-off-by: Atif Niyaz <[email protected]>

This looks OK to me. Benjamin, Peter, any concerns here?


> ---
> drivers/input/evdev.c | 42 ++++++++++++++++--------------------------
> drivers/input/input.c | 17 +++++++++++++++++
> include/linux/input.h | 38 ++++++++++++++++++++++++++++++++++++++
> 3 files changed, 71 insertions(+), 26 deletions(-)
>
> diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c
> index 867c2cfd0038..a331efa0a3f6 100644
> --- a/drivers/input/evdev.c
> +++ b/drivers/input/evdev.c
> @@ -25,13 +25,6 @@
> #include <linux/cdev.h>
> #include "input-compat.h"
>
> -enum evdev_clock_type {
> - EV_CLK_REAL = 0,
> - EV_CLK_MONO,
> - EV_CLK_BOOT,
> - EV_CLK_MAX
> -};
> -
> struct evdev {
> int open;
> struct input_handle handle;
> @@ -53,7 +46,7 @@ struct evdev_client {
> struct fasync_struct *fasync;
> struct evdev *evdev;
> struct list_head node;
> - unsigned int clk_type;
> + input_clk_t clk_type;
> bool revoked;
> unsigned long *evmasks[EV_CNT];
> unsigned int bufsize;
> @@ -150,16 +143,18 @@ static void __evdev_flush_queue(struct evdev_client *client, unsigned int type)
> static void __evdev_queue_syn_dropped(struct evdev_client *client)
> {
> struct input_event ev;
> - ktime_t time;
> struct timespec64 ts;
> + ktime_t *time = input_get_timestamp(client->evdev->handle.dev);
>
> - time = client->clk_type == EV_CLK_REAL ?
> - ktime_get_real() :
> - client->clk_type == EV_CLK_MONO ?
> - ktime_get() :
> - ktime_get_boottime();
> + switch (client->clk_type) {
> + case INPUT_CLK_REAL:
> + case INPUT_CLK_MONO:
> + ts = ktime_to_timespec64(time[client->clk_type]);
> + break;
> + default:
> + ts = ktime_to_timespec64(time[INPUT_CLK_BOOT]);

Add "break" here please.

> + }
>
> - ts = ktime_to_timespec64(time);
> ev.input_event_sec = ts.tv_sec;
> ev.input_event_usec = ts.tv_nsec / NSEC_PER_USEC;
> ev.type = EV_SYN;
> @@ -185,21 +180,21 @@ static void evdev_queue_syn_dropped(struct evdev_client *client)
> spin_unlock_irqrestore(&client->buffer_lock, flags);
> }
>
> -static int evdev_set_clk_type(struct evdev_client *client, unsigned int clkid)
> +static int evdev_set_clk_type(struct evdev_client *client, clockid_t clkid)
> {
> unsigned long flags;
> - unsigned int clk_type;
> + input_clk_t clk_type;
>
> switch (clkid) {
>
> case CLOCK_REALTIME:
> - clk_type = EV_CLK_REAL;
> + clk_type = INPUT_CLK_REAL;
> break;
> case CLOCK_MONOTONIC:
> - clk_type = EV_CLK_MONO;
> + clk_type = INPUT_CLK_MONO;
> break;
> case CLOCK_BOOTTIME:
> - clk_type = EV_CLK_BOOT;
> + clk_type = INPUT_CLK_BOOT;
> break;
> default:
> return -EINVAL;
> @@ -307,12 +302,7 @@ static void evdev_events(struct input_handle *handle,
> {
> struct evdev *evdev = handle->private;
> struct evdev_client *client;
> - ktime_t ev_time[EV_CLK_MAX];
> -
> - ev_time[EV_CLK_MONO] = ktime_get();
> - ev_time[EV_CLK_REAL] = ktime_mono_to_real(ev_time[EV_CLK_MONO]);
> - ev_time[EV_CLK_BOOT] = ktime_mono_to_any(ev_time[EV_CLK_MONO],
> - TK_OFFS_BOOT);
> + ktime_t *ev_time = input_get_timestamp(handle->dev);
>
> rcu_read_lock();
>
> diff --git a/drivers/input/input.c b/drivers/input/input.c
> index 7f3c5fcb9ed6..ae8b0ee58120 100644
> --- a/drivers/input/input.c
> +++ b/drivers/input/input.c
> @@ -1894,6 +1894,23 @@ void input_free_device(struct input_dev *dev)
> }
> EXPORT_SYMBOL(input_free_device);
>
> +/**
> + * input_get_timestamp - get timestamp for input events
> + * @dev: input device to get timestamp from
> + *
> + * A valid timestamp is a timestamp of non-zero value.
> + */
> +ktime_t *input_get_timestamp(struct input_dev *dev)
> +{
> + const ktime_t invalid_timestamp = ktime_set(0, 0);
> +
> + if (!ktime_compare(dev->timestamp[INPUT_CLK_MONO], ktime_zero)) {

You need to replace ktime_zero with invalid_timestamp here.

> + input_set_timestamp(dev, ktime_get());
> + }

No need for curly braces for 1-line body.

> + return dev->timestamp;
> +}
> +EXPORT_SYMBOL(input_get_timestamp);
> +
> /**
> * input_set_capability - mark device as capable of a certain event
> * @dev: device that is capable of emitting or accepting event
> diff --git a/include/linux/input.h b/include/linux/input.h
> index 510e78558c10..3929b62ccbe5 100644
> --- a/include/linux/input.h
> +++ b/include/linux/input.h
> @@ -33,6 +33,14 @@ struct input_value {
> __s32 value;
> };
>
> +enum input_clock_type {
> + INPUT_CLK_REAL = 0,
> + INPUT_CLK_MONO,
> + INPUT_CLK_BOOT,
> + INPUT_CLK_MAX
> +};
> +typedef enum input_clock_type input_clk_t;

We typically avoid typedefs unless we really want to hide kind of data
we are dealing with. Let's just use "enum input_clock_type" everywhere.

> +
> /**
> * struct input_dev - represents an input device
> * @name: name of the device
> @@ -114,6 +122,8 @@ struct input_value {
> * @vals: array of values queued in the current frame
> * @devres_managed: indicates that devices is managed with devres framework
> * and needs not be explicitly unregistered or freed.
> + * @timestamp: storage for a timestamp set by input_set_timestamp called
> + * by a driver
> */
> struct input_dev {
> const char *name;
> @@ -184,6 +194,8 @@ struct input_dev {
> struct input_value *vals;
>
> bool devres_managed;
> +
> + ktime_t timestamp[INPUT_CLK_MAX];
> };
> #define to_input_dev(d) container_of(d, struct input_dev, dev)
>
> @@ -382,6 +394,32 @@ void input_close_device(struct input_handle *);
>
> int input_flush_device(struct input_handle *handle, struct file *file);
>
> +/**
> + * input_set_timestamp - set timestamp for input events
> + * @dev: input device to set timestamp for
> + * @timestamp: the time at which the event has occurred
> + * in CLOCK_MONOTONIC
> + *
> + * This function is intended to provide to the input system a more
> + * accurate time of when an event actually occurred. The driver should
> + * call this function as soon as a timestamp is acquired ensuring
> + * clock conversions in input_set_timestamp are done correctly.
> + *
> + * The system entering a suspend between timestamp acquisition and
> + * calling input_set_timestamp can result in inaccurate conversions.
> + *
> + */
> +static inline void input_set_timestamp(struct input_dev *dev,
> + ktime_t timestamp)
> +{
> + dev->timestamp[INPUT_CLK_MONO] = timestamp;
> + dev->timestamp[INPUT_CLK_REAL] = ktime_mono_to_real(timestamp);
> + dev->timestamp[INPUT_CLK_BOOT] = ktime_mono_to_any(
> + timestamp, TK_OFFS_BOOT);
> +}
> +
> +ktime_t *input_get_timestamp(struct input_dev *dev);
> +
> void input_event(struct input_dev *dev, unsigned int type, unsigned int code, int value);
> void input_inject_event(struct input_handle *handle, unsigned int type, unsigned int code, int value);
>
> --
> 2.22.0.410.gd8fdbe21b5-goog
>

--
Dmitry

2019-07-12 07:24:21

by Benjamin Tissoires

[permalink] [raw]
Subject: Re: [PATCH] input: API for Setting a Timestamp from a Driver

On Fri, Jul 12, 2019 at 8:41 AM Dmitry Torokhov
<[email protected]> wrote:
>
> Hi Atif,
>
> On Wed, Jul 10, 2019 at 04:04:10PM -0700, Atif Niyaz wrote:
> > Currently, evdev stamps time with timestamps acquired in
> > evdev_events. However, this timestamping may not be accurate in terms of
> > measuring when the actual event happened. This API allows any 3rd party
> > driver to be able to call input_set_timestamp, and provide a timestamp
> > that can be utilized in order to provide a more accurate sense of time
> > for the event
> >
> > Signed-off-by: Atif Niyaz <[email protected]>
>
> This looks OK to me. Benjamin, Peter, any concerns here?
>

No red flags from me (though Peter is the one using all of this).

Just curious, which drivers do you think will be using this new API?
I can see that we might want to use hid-multitouch for it, with the
Scan Time forwarded by the device, but what do you have in mind?

Cheers,
Benjamin

>
> > ---
> > drivers/input/evdev.c | 42 ++++++++++++++++--------------------------
> > drivers/input/input.c | 17 +++++++++++++++++
> > include/linux/input.h | 38 ++++++++++++++++++++++++++++++++++++++
> > 3 files changed, 71 insertions(+), 26 deletions(-)
> >
> > diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c
> > index 867c2cfd0038..a331efa0a3f6 100644
> > --- a/drivers/input/evdev.c
> > +++ b/drivers/input/evdev.c
> > @@ -25,13 +25,6 @@
> > #include <linux/cdev.h>
> > #include "input-compat.h"
> >
> > -enum evdev_clock_type {
> > - EV_CLK_REAL = 0,
> > - EV_CLK_MONO,
> > - EV_CLK_BOOT,
> > - EV_CLK_MAX
> > -};
> > -
> > struct evdev {
> > int open;
> > struct input_handle handle;
> > @@ -53,7 +46,7 @@ struct evdev_client {
> > struct fasync_struct *fasync;
> > struct evdev *evdev;
> > struct list_head node;
> > - unsigned int clk_type;
> > + input_clk_t clk_type;
> > bool revoked;
> > unsigned long *evmasks[EV_CNT];
> > unsigned int bufsize;
> > @@ -150,16 +143,18 @@ static void __evdev_flush_queue(struct evdev_client *client, unsigned int type)
> > static void __evdev_queue_syn_dropped(struct evdev_client *client)
> > {
> > struct input_event ev;
> > - ktime_t time;
> > struct timespec64 ts;
> > + ktime_t *time = input_get_timestamp(client->evdev->handle.dev);
> >
> > - time = client->clk_type == EV_CLK_REAL ?
> > - ktime_get_real() :
> > - client->clk_type == EV_CLK_MONO ?
> > - ktime_get() :
> > - ktime_get_boottime();
> > + switch (client->clk_type) {
> > + case INPUT_CLK_REAL:
> > + case INPUT_CLK_MONO:
> > + ts = ktime_to_timespec64(time[client->clk_type]);
> > + break;
> > + default:
> > + ts = ktime_to_timespec64(time[INPUT_CLK_BOOT]);
>
> Add "break" here please.
>
> > + }
> >
> > - ts = ktime_to_timespec64(time);
> > ev.input_event_sec = ts.tv_sec;
> > ev.input_event_usec = ts.tv_nsec / NSEC_PER_USEC;
> > ev.type = EV_SYN;
> > @@ -185,21 +180,21 @@ static void evdev_queue_syn_dropped(struct evdev_client *client)
> > spin_unlock_irqrestore(&client->buffer_lock, flags);
> > }
> >
> > -static int evdev_set_clk_type(struct evdev_client *client, unsigned int clkid)
> > +static int evdev_set_clk_type(struct evdev_client *client, clockid_t clkid)
> > {
> > unsigned long flags;
> > - unsigned int clk_type;
> > + input_clk_t clk_type;
> >
> > switch (clkid) {
> >
> > case CLOCK_REALTIME:
> > - clk_type = EV_CLK_REAL;
> > + clk_type = INPUT_CLK_REAL;
> > break;
> > case CLOCK_MONOTONIC:
> > - clk_type = EV_CLK_MONO;
> > + clk_type = INPUT_CLK_MONO;
> > break;
> > case CLOCK_BOOTTIME:
> > - clk_type = EV_CLK_BOOT;
> > + clk_type = INPUT_CLK_BOOT;
> > break;
> > default:
> > return -EINVAL;
> > @@ -307,12 +302,7 @@ static void evdev_events(struct input_handle *handle,
> > {
> > struct evdev *evdev = handle->private;
> > struct evdev_client *client;
> > - ktime_t ev_time[EV_CLK_MAX];
> > -
> > - ev_time[EV_CLK_MONO] = ktime_get();
> > - ev_time[EV_CLK_REAL] = ktime_mono_to_real(ev_time[EV_CLK_MONO]);
> > - ev_time[EV_CLK_BOOT] = ktime_mono_to_any(ev_time[EV_CLK_MONO],
> > - TK_OFFS_BOOT);
> > + ktime_t *ev_time = input_get_timestamp(handle->dev);
> >
> > rcu_read_lock();
> >
> > diff --git a/drivers/input/input.c b/drivers/input/input.c
> > index 7f3c5fcb9ed6..ae8b0ee58120 100644
> > --- a/drivers/input/input.c
> > +++ b/drivers/input/input.c
> > @@ -1894,6 +1894,23 @@ void input_free_device(struct input_dev *dev)
> > }
> > EXPORT_SYMBOL(input_free_device);
> >
> > +/**
> > + * input_get_timestamp - get timestamp for input events
> > + * @dev: input device to get timestamp from
> > + *
> > + * A valid timestamp is a timestamp of non-zero value.
> > + */
> > +ktime_t *input_get_timestamp(struct input_dev *dev)
> > +{
> > + const ktime_t invalid_timestamp = ktime_set(0, 0);
> > +
> > + if (!ktime_compare(dev->timestamp[INPUT_CLK_MONO], ktime_zero)) {
>
> You need to replace ktime_zero with invalid_timestamp here.
>
> > + input_set_timestamp(dev, ktime_get());
> > + }
>
> No need for curly braces for 1-line body.
>
> > + return dev->timestamp;
> > +}
> > +EXPORT_SYMBOL(input_get_timestamp);
> > +
> > /**
> > * input_set_capability - mark device as capable of a certain event
> > * @dev: device that is capable of emitting or accepting event
> > diff --git a/include/linux/input.h b/include/linux/input.h
> > index 510e78558c10..3929b62ccbe5 100644
> > --- a/include/linux/input.h
> > +++ b/include/linux/input.h
> > @@ -33,6 +33,14 @@ struct input_value {
> > __s32 value;
> > };
> >
> > +enum input_clock_type {
> > + INPUT_CLK_REAL = 0,
> > + INPUT_CLK_MONO,
> > + INPUT_CLK_BOOT,
> > + INPUT_CLK_MAX
> > +};
> > +typedef enum input_clock_type input_clk_t;
>
> We typically avoid typedefs unless we really want to hide kind of data
> we are dealing with. Let's just use "enum input_clock_type" everywhere.
>
> > +
> > /**
> > * struct input_dev - represents an input device
> > * @name: name of the device
> > @@ -114,6 +122,8 @@ struct input_value {
> > * @vals: array of values queued in the current frame
> > * @devres_managed: indicates that devices is managed with devres framework
> > * and needs not be explicitly unregistered or freed.
> > + * @timestamp: storage for a timestamp set by input_set_timestamp called
> > + * by a driver
> > */
> > struct input_dev {
> > const char *name;
> > @@ -184,6 +194,8 @@ struct input_dev {
> > struct input_value *vals;
> >
> > bool devres_managed;
> > +
> > + ktime_t timestamp[INPUT_CLK_MAX];
> > };
> > #define to_input_dev(d) container_of(d, struct input_dev, dev)
> >
> > @@ -382,6 +394,32 @@ void input_close_device(struct input_handle *);
> >
> > int input_flush_device(struct input_handle *handle, struct file *file);
> >
> > +/**
> > + * input_set_timestamp - set timestamp for input events
> > + * @dev: input device to set timestamp for
> > + * @timestamp: the time at which the event has occurred
> > + * in CLOCK_MONOTONIC
> > + *
> > + * This function is intended to provide to the input system a more
> > + * accurate time of when an event actually occurred. The driver should
> > + * call this function as soon as a timestamp is acquired ensuring
> > + * clock conversions in input_set_timestamp are done correctly.
> > + *
> > + * The system entering a suspend between timestamp acquisition and
> > + * calling input_set_timestamp can result in inaccurate conversions.
> > + *
> > + */
> > +static inline void input_set_timestamp(struct input_dev *dev,
> > + ktime_t timestamp)
> > +{
> > + dev->timestamp[INPUT_CLK_MONO] = timestamp;
> > + dev->timestamp[INPUT_CLK_REAL] = ktime_mono_to_real(timestamp);
> > + dev->timestamp[INPUT_CLK_BOOT] = ktime_mono_to_any(
> > + timestamp, TK_OFFS_BOOT);
> > +}
> > +
> > +ktime_t *input_get_timestamp(struct input_dev *dev);
> > +
> > void input_event(struct input_dev *dev, unsigned int type, unsigned int code, int value);
> > void input_inject_event(struct input_handle *handle, unsigned int type, unsigned int code, int value);
> >
> > --
> > 2.22.0.410.gd8fdbe21b5-goog
> >
>
> --
> Dmitry

2019-07-12 11:48:26

by Peter Hutterer

[permalink] [raw]
Subject: Re: [PATCH] input: API for Setting a Timestamp from a Driver

On Fri, Jul 12, 2019 at 09:23:20AM +0200, Benjamin Tissoires wrote:
> On Fri, Jul 12, 2019 at 8:41 AM Dmitry Torokhov
> <[email protected]> wrote:
> >
> > Hi Atif,
> >
> > On Wed, Jul 10, 2019 at 04:04:10PM -0700, Atif Niyaz wrote:
> > > Currently, evdev stamps time with timestamps acquired in
> > > evdev_events. However, this timestamping may not be accurate in terms of
> > > measuring when the actual event happened. This API allows any 3rd party
> > > driver to be able to call input_set_timestamp, and provide a timestamp
> > > that can be utilized in order to provide a more accurate sense of time
> > > for the event
> > >
> > > Signed-off-by: Atif Niyaz <[email protected]>
> >
> > This looks OK to me. Benjamin, Peter, any concerns here?
> >
>
> No red flags from me (though Peter is the one using all of this).
>
> Just curious, which drivers do you think will be using this new API?
> I can see that we might want to use hid-multitouch for it, with the
> Scan Time forwarded by the device, but what do you have in mind?

that'd be my question as well. I'm all for more precise evdev timestamps but
there's some overlap with MSC_TIMESTAMP (which at least libinput isn't
handling well right now, with the exception of some quirk detection).

but yeah, overall this is a good solution from my POV.

Cheers,
Peter

> > > ---
> > > drivers/input/evdev.c | 42 ++++++++++++++++--------------------------
> > > drivers/input/input.c | 17 +++++++++++++++++
> > > include/linux/input.h | 38 ++++++++++++++++++++++++++++++++++++++
> > > 3 files changed, 71 insertions(+), 26 deletions(-)
> > >
> > > diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c
> > > index 867c2cfd0038..a331efa0a3f6 100644
> > > --- a/drivers/input/evdev.c
> > > +++ b/drivers/input/evdev.c
> > > @@ -25,13 +25,6 @@
> > > #include <linux/cdev.h>
> > > #include "input-compat.h"
> > >
> > > -enum evdev_clock_type {
> > > - EV_CLK_REAL = 0,
> > > - EV_CLK_MONO,
> > > - EV_CLK_BOOT,
> > > - EV_CLK_MAX
> > > -};
> > > -
> > > struct evdev {
> > > int open;
> > > struct input_handle handle;
> > > @@ -53,7 +46,7 @@ struct evdev_client {
> > > struct fasync_struct *fasync;
> > > struct evdev *evdev;
> > > struct list_head node;
> > > - unsigned int clk_type;
> > > + input_clk_t clk_type;
> > > bool revoked;
> > > unsigned long *evmasks[EV_CNT];
> > > unsigned int bufsize;
> > > @@ -150,16 +143,18 @@ static void __evdev_flush_queue(struct evdev_client *client, unsigned int type)
> > > static void __evdev_queue_syn_dropped(struct evdev_client *client)
> > > {
> > > struct input_event ev;
> > > - ktime_t time;
> > > struct timespec64 ts;
> > > + ktime_t *time = input_get_timestamp(client->evdev->handle.dev);
> > >
> > > - time = client->clk_type == EV_CLK_REAL ?
> > > - ktime_get_real() :
> > > - client->clk_type == EV_CLK_MONO ?
> > > - ktime_get() :
> > > - ktime_get_boottime();
> > > + switch (client->clk_type) {
> > > + case INPUT_CLK_REAL:
> > > + case INPUT_CLK_MONO:
> > > + ts = ktime_to_timespec64(time[client->clk_type]);
> > > + break;
> > > + default:
> > > + ts = ktime_to_timespec64(time[INPUT_CLK_BOOT]);
> >
> > Add "break" here please.
> >
> > > + }
> > >
> > > - ts = ktime_to_timespec64(time);
> > > ev.input_event_sec = ts.tv_sec;
> > > ev.input_event_usec = ts.tv_nsec / NSEC_PER_USEC;
> > > ev.type = EV_SYN;
> > > @@ -185,21 +180,21 @@ static void evdev_queue_syn_dropped(struct evdev_client *client)
> > > spin_unlock_irqrestore(&client->buffer_lock, flags);
> > > }
> > >
> > > -static int evdev_set_clk_type(struct evdev_client *client, unsigned int clkid)
> > > +static int evdev_set_clk_type(struct evdev_client *client, clockid_t clkid)
> > > {
> > > unsigned long flags;
> > > - unsigned int clk_type;
> > > + input_clk_t clk_type;
> > >
> > > switch (clkid) {
> > >
> > > case CLOCK_REALTIME:
> > > - clk_type = EV_CLK_REAL;
> > > + clk_type = INPUT_CLK_REAL;
> > > break;
> > > case CLOCK_MONOTONIC:
> > > - clk_type = EV_CLK_MONO;
> > > + clk_type = INPUT_CLK_MONO;
> > > break;
> > > case CLOCK_BOOTTIME:
> > > - clk_type = EV_CLK_BOOT;
> > > + clk_type = INPUT_CLK_BOOT;
> > > break;
> > > default:
> > > return -EINVAL;
> > > @@ -307,12 +302,7 @@ static void evdev_events(struct input_handle *handle,
> > > {
> > > struct evdev *evdev = handle->private;
> > > struct evdev_client *client;
> > > - ktime_t ev_time[EV_CLK_MAX];
> > > -
> > > - ev_time[EV_CLK_MONO] = ktime_get();
> > > - ev_time[EV_CLK_REAL] = ktime_mono_to_real(ev_time[EV_CLK_MONO]);
> > > - ev_time[EV_CLK_BOOT] = ktime_mono_to_any(ev_time[EV_CLK_MONO],
> > > - TK_OFFS_BOOT);
> > > + ktime_t *ev_time = input_get_timestamp(handle->dev);
> > >
> > > rcu_read_lock();
> > >
> > > diff --git a/drivers/input/input.c b/drivers/input/input.c
> > > index 7f3c5fcb9ed6..ae8b0ee58120 100644
> > > --- a/drivers/input/input.c
> > > +++ b/drivers/input/input.c
> > > @@ -1894,6 +1894,23 @@ void input_free_device(struct input_dev *dev)
> > > }
> > > EXPORT_SYMBOL(input_free_device);
> > >
> > > +/**
> > > + * input_get_timestamp - get timestamp for input events
> > > + * @dev: input device to get timestamp from
> > > + *
> > > + * A valid timestamp is a timestamp of non-zero value.
> > > + */
> > > +ktime_t *input_get_timestamp(struct input_dev *dev)
> > > +{
> > > + const ktime_t invalid_timestamp = ktime_set(0, 0);
> > > +
> > > + if (!ktime_compare(dev->timestamp[INPUT_CLK_MONO], ktime_zero)) {
> >
> > You need to replace ktime_zero with invalid_timestamp here.
> >
> > > + input_set_timestamp(dev, ktime_get());
> > > + }
> >
> > No need for curly braces for 1-line body.
> >
> > > + return dev->timestamp;
> > > +}
> > > +EXPORT_SYMBOL(input_get_timestamp);
> > > +
> > > /**
> > > * input_set_capability - mark device as capable of a certain event
> > > * @dev: device that is capable of emitting or accepting event
> > > diff --git a/include/linux/input.h b/include/linux/input.h
> > > index 510e78558c10..3929b62ccbe5 100644
> > > --- a/include/linux/input.h
> > > +++ b/include/linux/input.h
> > > @@ -33,6 +33,14 @@ struct input_value {
> > > __s32 value;
> > > };
> > >
> > > +enum input_clock_type {
> > > + INPUT_CLK_REAL = 0,
> > > + INPUT_CLK_MONO,
> > > + INPUT_CLK_BOOT,
> > > + INPUT_CLK_MAX
> > > +};
> > > +typedef enum input_clock_type input_clk_t;
> >
> > We typically avoid typedefs unless we really want to hide kind of data
> > we are dealing with. Let's just use "enum input_clock_type" everywhere.
> >
> > > +
> > > /**
> > > * struct input_dev - represents an input device
> > > * @name: name of the device
> > > @@ -114,6 +122,8 @@ struct input_value {
> > > * @vals: array of values queued in the current frame
> > > * @devres_managed: indicates that devices is managed with devres framework
> > > * and needs not be explicitly unregistered or freed.
> > > + * @timestamp: storage for a timestamp set by input_set_timestamp called
> > > + * by a driver
> > > */
> > > struct input_dev {
> > > const char *name;
> > > @@ -184,6 +194,8 @@ struct input_dev {
> > > struct input_value *vals;
> > >
> > > bool devres_managed;
> > > +
> > > + ktime_t timestamp[INPUT_CLK_MAX];
> > > };
> > > #define to_input_dev(d) container_of(d, struct input_dev, dev)
> > >
> > > @@ -382,6 +394,32 @@ void input_close_device(struct input_handle *);
> > >
> > > int input_flush_device(struct input_handle *handle, struct file *file);
> > >
> > > +/**
> > > + * input_set_timestamp - set timestamp for input events
> > > + * @dev: input device to set timestamp for
> > > + * @timestamp: the time at which the event has occurred
> > > + * in CLOCK_MONOTONIC
> > > + *
> > > + * This function is intended to provide to the input system a more
> > > + * accurate time of when an event actually occurred. The driver should
> > > + * call this function as soon as a timestamp is acquired ensuring
> > > + * clock conversions in input_set_timestamp are done correctly.
> > > + *
> > > + * The system entering a suspend between timestamp acquisition and
> > > + * calling input_set_timestamp can result in inaccurate conversions.
> > > + *
> > > + */
> > > +static inline void input_set_timestamp(struct input_dev *dev,
> > > + ktime_t timestamp)
> > > +{
> > > + dev->timestamp[INPUT_CLK_MONO] = timestamp;
> > > + dev->timestamp[INPUT_CLK_REAL] = ktime_mono_to_real(timestamp);
> > > + dev->timestamp[INPUT_CLK_BOOT] = ktime_mono_to_any(
> > > + timestamp, TK_OFFS_BOOT);
> > > +}
> > > +
> > > +ktime_t *input_get_timestamp(struct input_dev *dev);
> > > +
> > > void input_event(struct input_dev *dev, unsigned int type, unsigned int code, int value);
> > > void input_inject_event(struct input_handle *handle, unsigned int type, unsigned int code, int value);
> > >
> > > --
> > > 2.22.0.410.gd8fdbe21b5-goog
> > >
> >
> > --
> > Dmitry

2019-07-13 08:00:42

by Dmitry Torokhov

[permalink] [raw]
Subject: Re: [PATCH] input: API for Setting a Timestamp from a Driver

On Fri, Jul 12, 2019 at 09:46:19PM +1000, Peter Hutterer wrote:
> On Fri, Jul 12, 2019 at 09:23:20AM +0200, Benjamin Tissoires wrote:
> > On Fri, Jul 12, 2019 at 8:41 AM Dmitry Torokhov
> > <[email protected]> wrote:
> > >
> > > Hi Atif,
> > >
> > > On Wed, Jul 10, 2019 at 04:04:10PM -0700, Atif Niyaz wrote:
> > > > Currently, evdev stamps time with timestamps acquired in
> > > > evdev_events. However, this timestamping may not be accurate in terms of
> > > > measuring when the actual event happened. This API allows any 3rd party
> > > > driver to be able to call input_set_timestamp, and provide a timestamp
> > > > that can be utilized in order to provide a more accurate sense of time
> > > > for the event
> > > >
> > > > Signed-off-by: Atif Niyaz <[email protected]>
> > >
> > > This looks OK to me. Benjamin, Peter, any concerns here?
> > >
> >
> > No red flags from me (though Peter is the one using all of this).
> >
> > Just curious, which drivers do you think will be using this new API?
> > I can see that we might want to use hid-multitouch for it, with the
> > Scan Time forwarded by the device, but what do you have in mind?
>
> that'd be my question as well. I'm all for more precise evdev timestamps but
> there's some overlap with MSC_TIMESTAMP (which at least libinput isn't
> handling well right now, with the exception of some quirk detection).

I expect it will be used by drivers that use threaded interrupts to mark
the time in the hard interrupt and avoid the latency of scheduling the
thread, slow bus communication, etc.

This is not supposed to replace MSC_TIMESTAMP as MSC_TIMESTAMP carries
timestamp acquired by the device itself.

Thanks.

--
Dmitry