[PATCH 1/2] Input: Create input notifier chain in input.c

Jonathan Denose posted 2 patches 3 months, 1 week ago
[PATCH 1/2] Input: Create input notifier chain in input.c
Posted by Jonathan Denose 3 months, 1 week ago
To expose input events to other kernel modules, add a blocking notifier
chain. Publish LID_SWITCH_OPEN/LID_SWITCH_CLOSE events through this
notifier chain when input_handle_event detects events signaling the lid
switch has opened or closed.

Additionally, export a function which allows other kernel modules to
register notifier_block structs against this notifier chain.

Signed-off-by: Jonathan Denose <jdenose@google.com>
---
 drivers/input/input.c | 13 +++++++++++++
 include/linux/input.h |  7 +++++++
 2 files changed, 20 insertions(+)

diff --git a/drivers/input/input.c b/drivers/input/input.c
index a500e1e276c211d1146dbfea421a3402084007f8..b342b1ff138ccc58d4623edcf1152bd85d7054bf 100644
--- a/drivers/input/input.c
+++ b/drivers/input/input.c
@@ -26,6 +26,7 @@
 #include <linux/kstrtox.h>
 #include <linux/mutex.h>
 #include <linux/rcupdate.h>
+#include <linux/notifier.h>
 #include "input-compat.h"
 #include "input-core-private.h"
 #include "input-poller.h"
@@ -62,6 +63,8 @@ static const unsigned int input_max_code[EV_CNT] = {
 	[EV_FF] = FF_MAX,
 };
 
+static struct blocking_notifier_head input_notifier_head;
+
 static inline int is_event_supported(unsigned int code,
 				     unsigned long *bm, unsigned int max)
 {
@@ -367,10 +370,20 @@ void input_handle_event(struct input_dev *dev,
 		if (type != EV_SYN)
 			add_input_randomness(type, code, value);
 
+		if (type == EV_SW && code == SW_LID && !value)
+			blocking_notifier_call_chain(&input_notifier_head, value ?
+				LID_SWITCH_CLOSE : LID_SWITCH_OPEN, dev);
+
 		input_event_dispose(dev, disposition, type, code, value);
 	}
 }
 
+int register_input_notifier(struct notifier_block *notifier)
+{
+	return blocking_notifier_chain_register(&input_notifier_head, notifier);
+}
+EXPORT_SYMBOL(register_input_notifier);
+
 /**
  * input_event() - report new input event
  * @dev: device that generated the event
diff --git a/include/linux/input.h b/include/linux/input.h
index 7d7cb0593a63e93c4906c49cde430188db2d1ab5..e940aff8843a0afc693c60a252d6b0dbcb3476c4 100644
--- a/include/linux/input.h
+++ b/include/linux/input.h
@@ -42,6 +42,11 @@ enum input_clock_type {
 	INPUT_CLK_MAX
 };
 
+enum input_notify_event_type {
+	LID_SWITCH_OPEN,
+	LID_SWITCH_CLOSE
+};
+
 /**
  * struct input_dev - represents an input device
  * @name: name of the device
@@ -431,6 +436,8 @@ int input_flush_device(struct input_handle *handle, struct file *file);
 void input_set_timestamp(struct input_dev *dev, ktime_t timestamp);
 ktime_t *input_get_timestamp(struct input_dev *dev);
 
+int register_input_notifier(struct notifier_block *notifier);
+
 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);
 

-- 
2.51.1.851.g4ebd6896fd-goog
Re: [PATCH 1/2] Input: Create input notifier chain in input.c
Posted by Dmitry Torokhov 3 months ago
Hi Jonathan,

On Thu, Oct 30, 2025 at 02:10:40PM +0000, Jonathan Denose wrote:
> To expose input events to other kernel modules, add a blocking notifier
> chain. Publish LID_SWITCH_OPEN/LID_SWITCH_CLOSE events through this
> notifier chain when input_handle_event detects events signaling the lid
> switch has opened or closed.
> 
> Additionally, export a function which allows other kernel modules to
> register notifier_block structs against this notifier chain.
> 
> Signed-off-by: Jonathan Denose <jdenose@google.com>
> ---
>  drivers/input/input.c | 13 +++++++++++++
>  include/linux/input.h |  7 +++++++
>  2 files changed, 20 insertions(+)
> 
> diff --git a/drivers/input/input.c b/drivers/input/input.c
> index a500e1e276c211d1146dbfea421a3402084007f8..b342b1ff138ccc58d4623edcf1152bd85d7054bf 100644
> --- a/drivers/input/input.c
> +++ b/drivers/input/input.c
> @@ -26,6 +26,7 @@
>  #include <linux/kstrtox.h>
>  #include <linux/mutex.h>
>  #include <linux/rcupdate.h>
> +#include <linux/notifier.h>
>  #include "input-compat.h"
>  #include "input-core-private.h"
>  #include "input-poller.h"
> @@ -62,6 +63,8 @@ static const unsigned int input_max_code[EV_CNT] = {
>  	[EV_FF] = FF_MAX,
>  };
>  
> +static struct blocking_notifier_head input_notifier_head;
> +
>  static inline int is_event_supported(unsigned int code,
>  				     unsigned long *bm, unsigned int max)
>  {
> @@ -367,10 +370,20 @@ void input_handle_event(struct input_dev *dev,
>  		if (type != EV_SYN)
>  			add_input_randomness(type, code, value);
>  
> +		if (type == EV_SW && code == SW_LID && !value)
> +			blocking_notifier_call_chain(&input_notifier_head, value ?
> +				LID_SWITCH_CLOSE : LID_SWITCH_OPEN, dev);

I would prefer not having this directly in the input core but rather
have a lid handler that can then use notifier chain to forward the
events further.

Also, here you are running in atomic context, so you need atomic
notifier, not blocking (or you need to involve a workqueue). 

Thanks.

-- 
Dmitry
Re: [PATCH 1/2] Input: Create input notifier chain in input.c
Posted by Jonathan Denose 3 months ago
Hi Dmitry,

Thanks for your feedback.

On Wed, Nov 5, 2025 at 3:55 PM Dmitry Torokhov
<dmitry.torokhov@gmail.com> wrote:
>
> Hi Jonathan,
>
> On Thu, Oct 30, 2025 at 02:10:40PM +0000, Jonathan Denose wrote:
> > To expose input events to other kernel modules, add a blocking notifier
> > chain. Publish LID_SWITCH_OPEN/LID_SWITCH_CLOSE events through this
> > notifier chain when input_handle_event detects events signaling the lid
> > switch has opened or closed.
> >
> > Additionally, export a function which allows other kernel modules to
> > register notifier_block structs against this notifier chain.
> >
> > Signed-off-by: Jonathan Denose <jdenose@google.com>
> > ---
> >  drivers/input/input.c | 13 +++++++++++++
> >  include/linux/input.h |  7 +++++++
> >  2 files changed, 20 insertions(+)
> >
> > diff --git a/drivers/input/input.c b/drivers/input/input.c
> > index a500e1e276c211d1146dbfea421a3402084007f8..b342b1ff138ccc58d4623edcf1152bd85d7054bf 100644
> > --- a/drivers/input/input.c
> > +++ b/drivers/input/input.c
> > @@ -26,6 +26,7 @@
> >  #include <linux/kstrtox.h>
> >  #include <linux/mutex.h>
> >  #include <linux/rcupdate.h>
> > +#include <linux/notifier.h>
> >  #include "input-compat.h"
> >  #include "input-core-private.h"
> >  #include "input-poller.h"
> > @@ -62,6 +63,8 @@ static const unsigned int input_max_code[EV_CNT] = {
> >       [EV_FF] = FF_MAX,
> >  };
> >
> > +static struct blocking_notifier_head input_notifier_head;
> > +
> >  static inline int is_event_supported(unsigned int code,
> >                                    unsigned long *bm, unsigned int max)
> >  {
> > @@ -367,10 +370,20 @@ void input_handle_event(struct input_dev *dev,
> >               if (type != EV_SYN)
> >                       add_input_randomness(type, code, value);
> >
> > +             if (type == EV_SW && code == SW_LID && !value)
> > +                     blocking_notifier_call_chain(&input_notifier_head, value ?
> > +                             LID_SWITCH_CLOSE : LID_SWITCH_OPEN, dev);
>
> I would prefer not having this directly in the input core but rather
> have a lid handler that can then use notifier chain to forward the
> events further.

Ok, that makes sense to me. In that case, do you have a recommendation
for where the lid handler should go?

It looks like drivers/acpi/button.c initializes and handles the lid switch, so
would it make sense for it to go there?

> Also, here you are running in atomic context, so you need atomic
> notifier, not blocking (or you need to involve a workqueue).

I'll use an atomic notifier in the next version.

> Thanks.
>
> --
> Dmitry
-- 
Jonathan
Re: [PATCH 1/2] Input: Create input notifier chain in input.c
Posted by Dmitry Torokhov 3 months ago
On Wed, Nov 05, 2025 at 04:40:29PM -0600, Jonathan Denose wrote:
> Hi Dmitry,
> 
> Thanks for your feedback.
> 
> On Wed, Nov 5, 2025 at 3:55 PM Dmitry Torokhov
> <dmitry.torokhov@gmail.com> wrote:
> >
> > Hi Jonathan,
> >
> > On Thu, Oct 30, 2025 at 02:10:40PM +0000, Jonathan Denose wrote:
> > > To expose input events to other kernel modules, add a blocking notifier
> > > chain. Publish LID_SWITCH_OPEN/LID_SWITCH_CLOSE events through this
> > > notifier chain when input_handle_event detects events signaling the lid
> > > switch has opened or closed.
> > >
> > > Additionally, export a function which allows other kernel modules to
> > > register notifier_block structs against this notifier chain.
> > >
> > > Signed-off-by: Jonathan Denose <jdenose@google.com>
> > > ---
> > >  drivers/input/input.c | 13 +++++++++++++
> > >  include/linux/input.h |  7 +++++++
> > >  2 files changed, 20 insertions(+)
> > >
> > > diff --git a/drivers/input/input.c b/drivers/input/input.c
> > > index a500e1e276c211d1146dbfea421a3402084007f8..b342b1ff138ccc58d4623edcf1152bd85d7054bf 100644
> > > --- a/drivers/input/input.c
> > > +++ b/drivers/input/input.c
> > > @@ -26,6 +26,7 @@
> > >  #include <linux/kstrtox.h>
> > >  #include <linux/mutex.h>
> > >  #include <linux/rcupdate.h>
> > > +#include <linux/notifier.h>
> > >  #include "input-compat.h"
> > >  #include "input-core-private.h"
> > >  #include "input-poller.h"
> > > @@ -62,6 +63,8 @@ static const unsigned int input_max_code[EV_CNT] = {
> > >       [EV_FF] = FF_MAX,
> > >  };
> > >
> > > +static struct blocking_notifier_head input_notifier_head;
> > > +
> > >  static inline int is_event_supported(unsigned int code,
> > >                                    unsigned long *bm, unsigned int max)
> > >  {
> > > @@ -367,10 +370,20 @@ void input_handle_event(struct input_dev *dev,
> > >               if (type != EV_SYN)
> > >                       add_input_randomness(type, code, value);
> > >
> > > +             if (type == EV_SW && code == SW_LID && !value)
> > > +                     blocking_notifier_call_chain(&input_notifier_head, value ?
> > > +                             LID_SWITCH_CLOSE : LID_SWITCH_OPEN, dev);
> >
> > I would prefer not having this directly in the input core but rather
> > have a lid handler that can then use notifier chain to forward the
> > events further.
> 
> Ok, that makes sense to me. In that case, do you have a recommendation
> for where the lid handler should go?
> 
> It looks like drivers/acpi/button.c initializes and handles the lid switch, so
> would it make sense for it to go there?

drivers/acpi/button.c is not the only source of SW_LID events (we also
have cros-ec-keyb.c and others), so I'd recommend putting it into
drivers/input, maybe as lid-notifier.c

> 
> > Also, here you are running in atomic context, so you need atomic
> > notifier, not blocking (or you need to involve a workqueue).
> 
> I'll use an atomic notifier in the next version.

Another option would be to schedule work and then use blocking notifier.

Thanks.

-- 
Dmitry