PM regression with LED changes in next-20161109

PM regression with LED changes in next-20161109

[Tony Lindgren]

Hi,

Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
the sysfs brightness attr for changes.") breaks runtime PM for me.

On my omap dm3730 based test system, idle power consumption is over 70
times higher now with this patch! It goes from about 6mW for the core
system to over 440mW during idle meaning there's some busy timer now
active.

Reverting this patch fixes the issue. Any ideas?

Regards,

Tony

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi Tony,

On 11/09/2016 08:23 PM, Tony Lindgren wrote:
> Hi,
>
> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
> the sysfs brightness attr for changes.") breaks runtime PM for me.
>
> On my omap dm3730 based test system, idle power consumption is over 70
> times higher now with this patch! It goes from about 6mW for the core
> system to over 440mW during idle meaning there's some busy timer now
> active.
>
> Reverting this patch fixes the issue. Any ideas?

Thanks for the report. This is probably caused by sysfs_notify_dirent().
I'm afraid that we can't keep this feature in the current shape.
Hans, I'm dropping the patch. We probably will have to delegate this
call to a workqueue task. Think about use cases when the LED is blinked
with high frequency e.g. from ledtrig-disk.c.

Also, IMHO the notifications should be enabled only if explicitly
selected in the kernel config.

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 09-11-16 21:45, Jacek Anaszewski wrote:
> Hi Tony,
>
> On 11/09/2016 08:23 PM, Tony Lindgren wrote:
>> Hi,
>>
>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>
>> On my omap dm3730 based test system, idle power consumption is over 70
>> times higher now with this patch! It goes from about 6mW for the core
>> system to over 440mW during idle meaning there's some busy timer now
>> active.
>>
>> Reverting this patch fixes the issue. Any ideas?
>
> Thanks for the report. This is probably caused by sysfs_notify_dirent().
> I'm afraid that we can't keep this feature in the current shape.
> Hans, I'm dropping the patch. We probably will have to delegate this
> call to a workqueue task. Think about use cases when the LED is blinked
> with high frequency e.g. from ledtrig-disk.c.

sysfs_notify_dirent() already uses a workqueue, here is the actual
implementation of it (from fs/kernfs/file.c) :

void kernfs_notify(struct kernfs_node *kn)
{
         static DECLARE_WORK(kernfs_notify_work, kernfs_notify_workfn);
         unsigned long flags;

         if (WARN_ON(kernfs_type(kn) != KERNFS_FILE))
                 return;

         spin_lock_irqsave(&kernfs_notify_lock, flags);
         if (!kn->attr.notify_next) {
                 kernfs_get(kn);
                 kn->attr.notify_next = kernfs_notify_list;
                 kernfs_notify_list = kn;
                 schedule_work(&kernfs_notify_work);
         }
         spin_unlock_irqrestore(&kernfs_notify_lock, flags);
}

So using a workqueue is not going to help. Note that I already
feared this, which is why my initial implementation only called
sysfs_notify_dirent() for user initiated changes and not for
triggers / blinking.

I think we may need to reconsider what getting the brightness
sysfs atrribute actually returns. Currently when a LED is
blinking it will return 0 resp. the actual brightness depending
on when in the blink cycle the user reads the brightness
sysfs atrribute.

So a user can do "echo 128 > brightness && cat brightness" and
get out 0, or 128, depending purely on timing.

This seems to contradict what Documentation/ABI/testing/sysfs-class-led
has to say:

What:           /sys/class/leds/<led>/brightness
Date:           March 2006
KernelVersion:  2.6.17
Contact:        Richard Purdie <rpurdie@rpsys.net>
Description:
                 Set the brightness of the LED. Most LEDs don't
                 have hardware brightness support, so will just be turned on for
                 non-zero brightness settings. The value is between 0 and
                 /sys/class/leds/<led>/max_brightness.

                 Writing 0 to this file clears active trigger.

                 Writing non-zero to this file while trigger is active changes the
                 top brightness trigger is going to use.

Even though it only talks about writing, the logical thing would be for
reading to be the exact opposite of writing, so we would get:

                 Reading from this file while a trigger is active returns the
                 top brightness trigger is going to use.

The current docs say not about (sw) blinking, but that should be treated just
like a trigger IMHO.

If we can get consensus on what the read behavior for the brightness attribute
should be, then I think that a better poll() behavior will automatically follow
from that.

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/10/2016 09:49 AM, Hans de Goede wrote:
> Hi,
>
> On 09-11-16 21:45, Jacek Anaszewski wrote:
>> Hi Tony,
>>
>> On 11/09/2016 08:23 PM, Tony Lindgren wrote:
>>> Hi,
>>>
>>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>>
>>> On my omap dm3730 based test system, idle power consumption is over 70
>>> times higher now with this patch! It goes from about 6mW for the core
>>> system to over 440mW during idle meaning there's some busy timer now
>>> active.
>>>
>>> Reverting this patch fixes the issue. Any ideas?
>>
>> Thanks for the report. This is probably caused by sysfs_notify_dirent().
>> I'm afraid that we can't keep this feature in the current shape.
>> Hans, I'm dropping the patch. We probably will have to delegate this
>> call to a workqueue task. Think about use cases when the LED is blinked
>> with high frequency e.g. from ledtrig-disk.c.
>
> sysfs_notify_dirent() already uses a workqueue, here is the actual
> implementation of it (from fs/kernfs/file.c) :
>
> void kernfs_notify(struct kernfs_node *kn)
> {
>         static DECLARE_WORK(kernfs_notify_work, kernfs_notify_workfn);
>         unsigned long flags;
>
>         if (WARN_ON(kernfs_type(kn) != KERNFS_FILE))
>                 return;
>
>         spin_lock_irqsave(&kernfs_notify_lock, flags);
>         if (!kn->attr.notify_next) {
>                 kernfs_get(kn);
>                 kn->attr.notify_next = kernfs_notify_list;
>                 kernfs_notify_list = kn;
>                 schedule_work(&kernfs_notify_work);
>         }
>         spin_unlock_irqrestore(&kernfs_notify_lock, flags);
> }

Indeed. As a next step of this investigation Tony could disable
particular calls made in kernfs_notify_workfn to check what
exactly causes excessive power consumption.

> So using a workqueue is not going to help. Note that I already
> feared this, which is why my initial implementation only called
> sysfs_notify_dirent() for user initiated changes and not for
> triggers / blinking.

AFAIR there were no calls to led_notify_brightness_change() in
the initial implementation and it was entirely predestined for
being called by LED class drivers on brightness changes made
by firmware.

> I think we may need to reconsider what getting the brightness
> sysfs atrribute actually returns. Currently when a LED is
> blinking it will return 0 resp. the actual brightness depending
> on when in the blink cycle the user reads the brightness
> sysfs atrribute.
>
> So a user can do "echo 128 > brightness && cat brightness" and
> get out 0, or 128, depending purely on timing.
>
> This seems to contradict what Documentation/ABI/testing/sysfs-class-led
> has to say:
>
> What:           /sys/class/leds/<led>/brightness
> Date:           March 2006
> KernelVersion:  2.6.17
> Contact:        Richard Purdie <rpurdie@rpsys.net>
> Description:
>                 Set the brightness of the LED. Most LEDs don't
>                 have hardware brightness support, so will just be turned
> on for
>                 non-zero brightness settings. The value is between 0 and
>                 /sys/class/leds/<led>/max_brightness.
>
>                 Writing 0 to this file clears active trigger.
>
>                 Writing non-zero to this file while trigger is active
> changes the
>                 top brightness trigger is going to use.
>
> Even though it only talks about writing, the logical thing would be for
> reading to be the exact opposite of writing, so we would get:
>
>                 Reading from this file while a trigger is active returns
> the
>                 top brightness trigger is going to use.
>
> The current docs say not about (sw) blinking, but that should be treated
> just
> like a trigger IMHO.

You'r right, we should describe the semantics on reading, but it would
have to be as follows:

Reading from this file returns LED brightness at given moment, i.e.
even though LED class device brightness setting is greater than 0, the
momentary brightness can be 0 if the readout occurred during low phase
of blink cycle.

> If we can get consensus on what the read behavior for the brightness
> attribute
> should be, then I think that a better poll() behavior will automatically
> follow
> from that.

It seems that we should get back to your initial approach. i.e. only
brightness changes caused by hardware should be reported.

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 10-11-16 13:56, Jacek Anaszewski wrote:
> Hi,
>
> On 11/10/2016 09:49 AM, Hans de Goede wrote:
>> Hi,
>>
>> On 09-11-16 21:45, Jacek Anaszewski wrote:
>>> Hi Tony,
>>>
>>> On 11/09/2016 08:23 PM, Tony Lindgren wrote:
>>>> Hi,
>>>>
>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>>>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>>>
>>>> On my omap dm3730 based test system, idle power consumption is over 70
>>>> times higher now with this patch! It goes from about 6mW for the core
>>>> system to over 440mW during idle meaning there's some busy timer now
>>>> active.
>>>>
>>>> Reverting this patch fixes the issue. Any ideas?
>>>
>>> Thanks for the report. This is probably caused by sysfs_notify_dirent().
>>> I'm afraid that we can't keep this feature in the current shape.
>>> Hans, I'm dropping the patch. We probably will have to delegate this
>>> call to a workqueue task. Think about use cases when the LED is blinked
>>> with high frequency e.g. from ledtrig-disk.c.
>>
>> sysfs_notify_dirent() already uses a workqueue, here is the actual
>> implementation of it (from fs/kernfs/file.c) :
>>
>> void kernfs_notify(struct kernfs_node *kn)
>> {
>>         static DECLARE_WORK(kernfs_notify_work, kernfs_notify_workfn);
>>         unsigned long flags;
>>
>>         if (WARN_ON(kernfs_type(kn) != KERNFS_FILE))
>>                 return;
>>
>>         spin_lock_irqsave(&kernfs_notify_lock, flags);
>>         if (!kn->attr.notify_next) {
>>                 kernfs_get(kn);
>>                 kn->attr.notify_next = kernfs_notify_list;
>>                 kernfs_notify_list = kn;
>>                 schedule_work(&kernfs_notify_work);
>>         }
>>         spin_unlock_irqrestore(&kernfs_notify_lock, flags);
>> }
>
> Indeed. As a next step of this investigation Tony could disable
> particular calls made in kernfs_notify_workfn to check what
> exactly causes excessive power consumption.
>
>> So using a workqueue is not going to help. Note that I already
>> feared this, which is why my initial implementation only called
>> sysfs_notify_dirent() for user initiated changes and not for
>> triggers / blinking.
>
> AFAIR there were no calls to led_notify_brightness_change() in
> the initial implementation and it was entirely predestined for
> being called by LED class drivers on brightness changes made
> by firmware.
>
>> I think we may need to reconsider what getting the brightness
>> sysfs atrribute actually returns. Currently when a LED is
>> blinking it will return 0 resp. the actual brightness depending
>> on when in the blink cycle the user reads the brightness
>> sysfs atrribute.
>>
>> So a user can do "echo 128 > brightness && cat brightness" and
>> get out 0, or 128, depending purely on timing.
>>
>> This seems to contradict what Documentation/ABI/testing/sysfs-class-led
>> has to say:
>>
>> What:           /sys/class/leds/<led>/brightness
>> Date:           March 2006
>> KernelVersion:  2.6.17
>> Contact:        Richard Purdie <rpurdie@rpsys.net>
>> Description:
>>                 Set the brightness of the LED. Most LEDs don't
>>                 have hardware brightness support, so will just be turned
>> on for
>>                 non-zero brightness settings. The value is between 0 and
>>                 /sys/class/leds/<led>/max_brightness.
>>
>>                 Writing 0 to this file clears active trigger.
>>
>>                 Writing non-zero to this file while trigger is active
>> changes the
>>                 top brightness trigger is going to use.
>>
>> Even though it only talks about writing, the logical thing would be for
>> reading to be the exact opposite of writing, so we would get:
>>
>>                 Reading from this file while a trigger is active returns
>> the
>>                 top brightness trigger is going to use.
>>
>> The current docs say not about (sw) blinking, but that should be treated
>> just
>> like a trigger IMHO.
>
> You'r right, we should describe the semantics on reading, but it would
> have to be as follows:
>
> Reading from this file returns LED brightness at given moment, i.e.
> even though LED class device brightness setting is greater than 0, the
> momentary brightness can be 0 if the readout occurred during low phase
> of blink cycle.

Why would it need to read like this, because this is the current behavior ?

I doubt anyone is relying on this current behavior because it is really
unpredictable which value one can get.

I believe it would be better to change the read semantics to follow
the write semantics, this would be much more consistent.

Making the read behavior match the write behavior should be easy I would
be happy to write a patch for this.

>> If we can get consensus on what the read behavior for the brightness
>> attribute
>> should be, then I think that a better poll() behavior will automatically
>> follow
>> from that.
>
> It seems that we should get back to your initial approach. i.e. only
> brightness changes caused by hardware should be reported.

Ok, if you really want to keep the read behavior as is, I can provide
an updated patch for this.

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/10/2016 02:04 PM, Hans de Goede wrote:
> Hi,
>
> On 10-11-16 13:56, Jacek Anaszewski wrote:
>> Hi,
>>
>> On 11/10/2016 09:49 AM, Hans de Goede wrote:
>>> Hi,
>>>
>>> On 09-11-16 21:45, Jacek Anaszewski wrote:
>>>> Hi Tony,
>>>>
>>>> On 11/09/2016 08:23 PM, Tony Lindgren wrote:
>>>>> Hi,
>>>>>
>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>>>>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>>>>
>>>>> On my omap dm3730 based test system, idle power consumption is over 70
>>>>> times higher now with this patch! It goes from about 6mW for the core
>>>>> system to over 440mW during idle meaning there's some busy timer now
>>>>> active.
>>>>>
>>>>> Reverting this patch fixes the issue. Any ideas?
>>>>
>>>> Thanks for the report. This is probably caused by
>>>> sysfs_notify_dirent().
>>>> I'm afraid that we can't keep this feature in the current shape.
>>>> Hans, I'm dropping the patch. We probably will have to delegate this
>>>> call to a workqueue task. Think about use cases when the LED is blinked
>>>> with high frequency e.g. from ledtrig-disk.c.
>>>
>>> sysfs_notify_dirent() already uses a workqueue, here is the actual
>>> implementation of it (from fs/kernfs/file.c) :
>>>
>>> void kernfs_notify(struct kernfs_node *kn)
>>> {
>>>         static DECLARE_WORK(kernfs_notify_work, kernfs_notify_workfn);
>>>         unsigned long flags;
>>>
>>>         if (WARN_ON(kernfs_type(kn) != KERNFS_FILE))
>>>                 return;
>>>
>>>         spin_lock_irqsave(&kernfs_notify_lock, flags);
>>>         if (!kn->attr.notify_next) {
>>>                 kernfs_get(kn);
>>>                 kn->attr.notify_next = kernfs_notify_list;
>>>                 kernfs_notify_list = kn;
>>>                 schedule_work(&kernfs_notify_work);
>>>         }
>>>         spin_unlock_irqrestore(&kernfs_notify_lock, flags);
>>> }
>>
>> Indeed. As a next step of this investigation Tony could disable
>> particular calls made in kernfs_notify_workfn to check what
>> exactly causes excessive power consumption.
>>
>>> So using a workqueue is not going to help. Note that I already
>>> feared this, which is why my initial implementation only called
>>> sysfs_notify_dirent() for user initiated changes and not for
>>> triggers / blinking.
>>
>> AFAIR there were no calls to led_notify_brightness_change() in
>> the initial implementation and it was entirely predestined for
>> being called by LED class drivers on brightness changes made
>> by firmware.
>>
>>> I think we may need to reconsider what getting the brightness
>>> sysfs atrribute actually returns. Currently when a LED is
>>> blinking it will return 0 resp. the actual brightness depending
>>> on when in the blink cycle the user reads the brightness
>>> sysfs atrribute.
>>>
>>> So a user can do "echo 128 > brightness && cat brightness" and
>>> get out 0, or 128, depending purely on timing.
>>>
>>> This seems to contradict what Documentation/ABI/testing/sysfs-class-led
>>> has to say:
>>>
>>> What:           /sys/class/leds/<led>/brightness
>>> Date:           March 2006
>>> KernelVersion:  2.6.17
>>> Contact:        Richard Purdie <rpurdie@rpsys.net>
>>> Description:
>>>                 Set the brightness of the LED. Most LEDs don't
>>>                 have hardware brightness support, so will just be turned
>>> on for
>>>                 non-zero brightness settings. The value is between 0 and
>>>                 /sys/class/leds/<led>/max_brightness.
>>>
>>>                 Writing 0 to this file clears active trigger.
>>>
>>>                 Writing non-zero to this file while trigger is active
>>> changes the
>>>                 top brightness trigger is going to use.
>>>
>>> Even though it only talks about writing, the logical thing would be for
>>> reading to be the exact opposite of writing, so we would get:
>>>
>>>                 Reading from this file while a trigger is active returns
>>> the
>>>                 top brightness trigger is going to use.
>>>
>>> The current docs say not about (sw) blinking, but that should be treated
>>> just
>>> like a trigger IMHO.
>>
>> You'r right, we should describe the semantics on reading, but it would
>> have to be as follows:
>>
>> Reading from this file returns LED brightness at given moment, i.e.
>> even though LED class device brightness setting is greater than 0, the
>> momentary brightness can be 0 if the readout occurred during low phase
>> of blink cycle.
>
> Why would it need to read like this, because this is the current behavior ?

We have led_update_brightness() which was introduced long time ago and
is used in brightness_show(). Note that if LED controller changed
actual LED brightness e.g. due to battery voltage dropping below
certain threshold, we wouldn't be able to find it out otherwise
(except if we added separate sysfs file for that).

>
> I doubt anyone is relying on this current behavior because it is really
> unpredictable which value one can get.
>
> I believe it would be better to change the read semantics to follow
> the write semantics, this would be much more consistent.
>
> Making the read behavior match the write behavior should be easy I would
> be happy to write a patch for this.

Let's better agree on the description of the current semantics.
It has been around for a long time.

>>> If we can get consensus on what the read behavior for the brightness
>>> attribute
>>> should be, then I think that a better poll() behavior will automatically
>>> follow
>>> from that.
>>
>> It seems that we should get back to your initial approach. i.e. only
>> brightness changes caused by hardware should be reported.
>
> Ok, if you really want to keep the read behavior as is, I can provide
> an updated patch for this.

Yes please.

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 1361 bytes --]

Hi!


> >>>The current docs say not about (sw) blinking, but that should be treated
> >>>just
> >>>like a trigger IMHO.
> >>
> >>You'r right, we should describe the semantics on reading, but it would
> >>have to be as follows:
> >>
> >>Reading from this file returns LED brightness at given moment, i.e.
> >>even though LED class device brightness setting is greater than 0, the
> >>momentary brightness can be 0 if the readout occurred during low phase
> >>of blink cycle.
> >
> >Why would it need to read like this, because this is the current behavior ?
> 
> We have led_update_brightness() which was introduced long time ago and
> is used in brightness_show(). Note that if LED controller changed
> actual LED brightness e.g. due to battery voltage dropping below
> certain threshold, we wouldn't be able to find it out otherwise
> (except if we added separate sysfs file for that).

And we should have a separate sysfs file for that. Note that on some
hardware leds, you are able to let hardware control them, but if you
do, you can't really tell the current state. Examples are n900 and
thinkpad-acpi. So it is better we don't pretend we can get that value
for userspace.
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 1374 bytes --]

Hi!

> >>>Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
> >>>the sysfs brightness attr for changes.") breaks runtime PM for me.
> >>>
> >>>On my omap dm3730 based test system, idle power consumption is over 70
> >>>times higher now with this patch! It goes from about 6mW for the core
> >>>system to over 440mW during idle meaning there's some busy timer now
> >>>active.
> >>>
> >>>Reverting this patch fixes the issue. Any ideas?

Are you using any LED that toggles with high frequency? Like perhaps
LED that is lit when CPU is active?

> >So a user can do "echo 128 > brightness && cat brightness" and
> >get out 0, or 128, depending purely on timing.
...
> >                Reading from this file while a trigger is active returns
> >the
> >                top brightness trigger is going to use.

Yes, that sounds sane.

> It seems that we should get back to your initial approach. i.e. only
> brightness changes caused by hardware should be reported.

I don't think enabling poll() here is good idea. Some hardware won't
be able to tell you that it changed the state. Returning maximum
brightness trigger is going to use seems easier/better.

Best regards,
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 10-11-16 17:29, Pavel Machek wrote:
> Hi!
>
>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>>>>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>>>>
>>>>> On my omap dm3730 based test system, idle power consumption is over 70
>>>>> times higher now with this patch! It goes from about 6mW for the core
>>>>> system to over 440mW during idle meaning there's some busy timer now
>>>>> active.
>>>>>
>>>>> Reverting this patch fixes the issue. Any ideas?
>
> Are you using any LED that toggles with high frequency? Like perhaps
> LED that is lit when CPU is active?
>
>>> So a user can do "echo 128 > brightness && cat brightness" and
>>> get out 0, or 128, depending purely on timing.
> ...
>>>                Reading from this file while a trigger is active returns
>>> the
>>>                top brightness trigger is going to use.
>
> Yes, that sounds sane.
>
>> It seems that we should get back to your initial approach. i.e. only
>> brightness changes caused by hardware should be reported.
>
> I don't think enabling poll() here is good idea. Some hardware won't
> be able to tell you that it changed the state. Returning maximum
> brightness trigger is going to use seems easier/better.

The idea here is to allow userspace to poll() on the brightness
sysfs atrribute to detect changes autonomously done by the hardware,
such as e.g. happens on both Dell and Thinkpad laptops when pressing
the keyboard backlight cycle hotkey. Note that these keys do not
generate key-press events, the cycling through the brightness levels
(including off) is done entirely in firmware.

But we do get other ACPI events for this which we can use to let
userspace know this happens, which is something which user-
interfaces which allow control over the kbd backlight want to know.

I understand that we will not always be able to do this, here is the
Documentation/ABI/testing/sysfs-class-led text I have in mind:

		The file supports poll() to detect changes, changes are only
		signalled when this file is written or when the hardware /
		firmware changes the brightness itself and the driver can detect
		this. Changes done by kernel triggers / software blinking are
		not signalled.

Note the "and the driver can detect this" language, that has been there
since v1 of the poll() notification patch since I already expected not
all hardware to be able to signal this.

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 2557 bytes --]

Hi!

> >>It seems that we should get back to your initial approach. i.e. only
> >>brightness changes caused by hardware should be reported.
> >
> >I don't think enabling poll() here is good idea. Some hardware won't
> >be able to tell you that it changed the state. Returning maximum
> >brightness trigger is going to use seems easier/better.
> 
> The idea here is to allow userspace to poll() on the brightness
> sysfs atrribute to detect changes autonomously done by the hardware,
> such as e.g. happens on both Dell and Thinkpad laptops when pressing
> the keyboard backlight cycle hotkey. Note that these keys do not
> generate key-press events, the cycling through the brightness levels
> (including off) is done entirely in firmware.

Ok, so you can do that for keyboard backlight on thinkpad... I guess
you handle that as a special trigger on the keyboard leds? Can other
triggers, such as heartbeat, be assigned to that "led"? 

> But we do get other ACPI events for this which we can use to let
> userspace know this happens, which is something which user-
> interfaces which allow control over the kbd backlight want to know.

Yes, you can do that for keyboard backlight... but on thinkpads there
are more leds, such as battery led. That can blink on battery low, and
I don't think you can read the current status from hardware.

Getting current state of led blinking with cpu trigger is also not
quite a good idea. 
So IMO this should not be done in generic code. Instead,
kbd-backlight trigger should have special attribute, and that one
should be pollable.

> I understand that we will not always be able to do this, here is the
> Documentation/ABI/testing/sysfs-class-led text I have in mind:
> 
> 		The file supports poll() to detect changes, changes are only
> 		signalled when this file is written or when the hardware /
> 		firmware changes the brightness itself and the driver can detect
> 		this. Changes done by kernel triggers / software blinking are
> 		not signalled.
> 
> Note the "and the driver can detect this" language, that has been there
> since v1 of the poll() notification patch since I already expected not
> all hardware to be able to signal this.

Lets move it to separate attribute, for triggers that can do that,
please.

We do want a way to read maximum brightness for the heartbeat trigger,
for example..

									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 10-11-16 21:48, Pavel Machek wrote:
> Hi!
>
>>>> It seems that we should get back to your initial approach. i.e. only
>>>> brightness changes caused by hardware should be reported.
>>>
>>> I don't think enabling poll() here is good idea. Some hardware won't
>>> be able to tell you that it changed the state. Returning maximum
>>> brightness trigger is going to use seems easier/better.
>>
>> The idea here is to allow userspace to poll() on the brightness
>> sysfs atrribute to detect changes autonomously done by the hardware,
>> such as e.g. happens on both Dell and Thinkpad laptops when pressing
>> the keyboard backlight cycle hotkey. Note that these keys do not
>> generate key-press events, the cycling through the brightness levels
>> (including off) is done entirely in firmware.
>
> Ok, so you can do that for keyboard backlight on thinkpad... I guess
> you handle that as a special trigger on the keyboard leds?

No, as said this is all done in firmware, as in this is all dealt
with by (presumably) the acpi-ec (acpi-embedded-controller) the kernel
does not do anything here, the key is "hardwired" to control the
keyboard backlight from the kernels pov.

 > Can other
> triggers, such as heartbeat, be assigned to that "led"?
>
>> But we do get other ACPI events for this which we can use to let
>> userspace know this happens, which is something which user-
>> interfaces which allow control over the kbd backlight want to know.
>
> Yes, you can do that for keyboard backlight... but on thinkpads there
> are more leds, such as battery led. That can blink on battery low, and
> I don't think you can read the current status from hardware.

Well the battery LED does not show up under /sys/class/led so that
is not relevant for this situation, anyways ...

> Getting current state of led blinking with cpu trigger is also not
> quite a good idea.

I agree with you that it would be better if reading the brightness
sysfs attribute would always return the max brightness for LEDs
which are blinking or have a trigger set. But it seems that Jacek
disagrees, I will leave further discussion of this up to you and
Jacek.

> So IMO this should not be done in generic code. Instead,
> kbd-backlight trigger should have special attribute, and that one
> should be pollable.

Again there is no kbd-backlight trigger.

>> I understand that we will not always be able to do this, here is the
>> Documentation/ABI/testing/sysfs-class-led text I have in mind:
>>
>> 		The file supports poll() to detect changes, changes are only
>> 		signalled when this file is written or when the hardware /
>> 		firmware changes the brightness itself and the driver can detect
>> 		this. Changes done by kernel triggers / software blinking are
>> 		not signalled.
>>
>> Note the "and the driver can detect this" language, that has been there
>> since v1 of the poll() notification patch since I already expected not
>> all hardware to be able to signal this.
>
> Lets move it to separate attribute, for triggers that can do that,
> please.

As explained above this has nothing to do with triggers...

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Tony Lindgren]

* Pavel Machek <pavel@ucw.cz> [161110 09:29]:
> Hi!
> 
> > >>>Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
> > >>>the sysfs brightness attr for changes.") breaks runtime PM for me.
> > >>>
> > >>>On my omap dm3730 based test system, idle power consumption is over 70
> > >>>times higher now with this patch! It goes from about 6mW for the core
> > >>>system to over 440mW during idle meaning there's some busy timer now
> > >>>active.
> > >>>
> > >>>Reverting this patch fixes the issue. Any ideas?
> 
> Are you using any LED that toggles with high frequency? Like perhaps
> LED that is lit when CPU is active?

Yeah one of them seems to have cpu0 as the default trigger.

Tony

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 1222 bytes --]

On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
> * Pavel Machek <pavel@ucw.cz> [161110 09:29]:
> > Hi!
> > 
> > > >>>Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
> > > >>>the sysfs brightness attr for changes.") breaks runtime PM for me.
> > > >>>
> > > >>>On my omap dm3730 based test system, idle power consumption is over 70
> > > >>>times higher now with this patch! It goes from about 6mW for the core
> > > >>>system to over 440mW during idle meaning there's some busy timer now
> > > >>>active.
> > > >>>
> > > >>>Reverting this patch fixes the issue. Any ideas?
> > 
> > Are you using any LED that toggles with high frequency? Like perhaps
> > LED that is lit when CPU is active?
> 
> Yeah one of them seems to have cpu0 as the default trigger.

Aha. Its quite obvious we don't want to notify sysfs each time that
one is toggled, right?

IMO brightness should display max brightness for the trigger, as Hans
suggested, anything else is madness for trigger such as cpu activity.

Thanks and best regards,

								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/10/2016 09:29 PM, Pavel Machek wrote:
> On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
>> * Pavel Machek <pavel@ucw.cz> [161110 09:29]:
>>> Hi!
>>>
>>>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>>>>>>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>>>>>>
>>>>>>> On my omap dm3730 based test system, idle power consumption is over 70
>>>>>>> times higher now with this patch! It goes from about 6mW for the core
>>>>>>> system to over 440mW during idle meaning there's some busy timer now
>>>>>>> active.
>>>>>>>
>>>>>>> Reverting this patch fixes the issue. Any ideas?
>>>
>>> Are you using any LED that toggles with high frequency? Like perhaps
>>> LED that is lit when CPU is active?
>>
>> Yeah one of them seems to have cpu0 as the default trigger.
>
> Aha. Its quite obvious we don't want to notify sysfs each time that
> one is toggled, right?
>
> IMO brightness should display max brightness for the trigger, as Hans
> suggested, anything else is madness for trigger such as cpu activity.

Are you suggesting that we should revert changes introduced
by below patch?

commit 29d76dfa29fe22583aefddccda0bc56aa81035dc
Author: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Date:   Tue Mar 18 09:47:48 2008 +0000

     leds: Add support to leds with readable status

     Some led hardware allows drivers to query the led state, and this patch
     adds a hook to let the led class take advantage of that information 
when
     available.

     Without this functionality, when access to the led hardware is not
     exclusive (i.e. firmware or hardware might change its state behind the
     kernel's back), reality goes out of sync with the led class' idea 
of what
     the led is doing, which is annoying at best.

     Behaviour for drivers that do not or cannot read the led status is
     unchanged.

     Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
     Signed-off-by: Richard Purdie <rpurdie@rpsys.net>


-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 2619 bytes --]

On Thu 2016-11-10 22:34:07, Jacek Anaszewski wrote:
> Hi,
> 
> On 11/10/2016 09:29 PM, Pavel Machek wrote:
> >On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
> >>* Pavel Machek <pavel@ucw.cz> [161110 09:29]:
> >>>Hi!
> >>>
> >>>>>>>Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
> >>>>>>>the sysfs brightness attr for changes.") breaks runtime PM for me.
> >>>>>>>
> >>>>>>>On my omap dm3730 based test system, idle power consumption is over 70
> >>>>>>>times higher now with this patch! It goes from about 6mW for the core
> >>>>>>>system to over 440mW during idle meaning there's some busy timer now
> >>>>>>>active.
> >>>>>>>
> >>>>>>>Reverting this patch fixes the issue. Any ideas?
> >>>
> >>>Are you using any LED that toggles with high frequency? Like perhaps
> >>>LED that is lit when CPU is active?
> >>
> >>Yeah one of them seems to have cpu0 as the default trigger.
> >
> >Aha. Its quite obvious we don't want to notify sysfs each time that
> >one is toggled, right?
> >
> >IMO brightness should display max brightness for the trigger, as Hans
> >suggested, anything else is madness for trigger such as cpu activity.
> 
> Are you suggesting that we should revert changes introduced
> by below patch?
> 
> commit 29d76dfa29fe22583aefddccda0bc56aa81035dc
> Author: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
> Date:   Tue Mar 18 09:47:48 2008 +0000
> 
>     leds: Add support to leds with readable status
> 
>     Some led hardware allows drivers to query the led state, and this patch
>     adds a hook to let the led class take advantage of that information when
>     available.
> 
>     Without this functionality, when access to the led hardware is not
>     exclusive (i.e. firmware or hardware might change its state behind the
>     kernel's back), reality goes out of sync with the led class' idea of
> what
>     the led is doing, which is annoying at best.

Hmm. So userland can read the LED state, and it can get _some_ value
back, but it can not know if it is current state or not.

I don't think that's a good interface. I see it is from 2008... is
someone using it? Maybe it is too late for revert.

But I'd certainly not extend it with poll.

IMO reading/polling should only be available with some triggers. It
does not make sense with "CPU load" trigger. It makes sense with
"keyboard light changeable by hardware" trigger.

Best regards,
									
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

On 11/11/2016 01:01 PM, Pavel Machek wrote:
> On Thu 2016-11-10 22:34:07, Jacek Anaszewski wrote:
>> Hi,
>>
>> On 11/10/2016 09:29 PM, Pavel Machek wrote:
>>> On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
>>>> * Pavel Machek <pavel@ucw.cz> [161110 09:29]:
>>>>> Hi!
>>>>>
>>>>>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>>>>>>>>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>>>>>>>>
>>>>>>>>> On my omap dm3730 based test system, idle power consumption is over 70
>>>>>>>>> times higher now with this patch! It goes from about 6mW for the core
>>>>>>>>> system to over 440mW during idle meaning there's some busy timer now
>>>>>>>>> active.
>>>>>>>>>
>>>>>>>>> Reverting this patch fixes the issue. Any ideas?
>>>>>
>>>>> Are you using any LED that toggles with high frequency? Like perhaps
>>>>> LED that is lit when CPU is active?
>>>>
>>>> Yeah one of them seems to have cpu0 as the default trigger.
>>>
>>> Aha. Its quite obvious we don't want to notify sysfs each time that
>>> one is toggled, right?
>>>
>>> IMO brightness should display max brightness for the trigger, as Hans
>>> suggested, anything else is madness for trigger such as cpu activity.
>>
>> Are you suggesting that we should revert changes introduced
>> by below patch?
>>
>> commit 29d76dfa29fe22583aefddccda0bc56aa81035dc
>> Author: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
>> Date:   Tue Mar 18 09:47:48 2008 +0000
>>
>>     leds: Add support to leds with readable status
>>
>>     Some led hardware allows drivers to query the led state, and this patch
>>     adds a hook to let the led class take advantage of that information when
>>     available.
>>
>>     Without this functionality, when access to the led hardware is not
>>     exclusive (i.e. firmware or hardware might change its state behind the
>>     kernel's back), reality goes out of sync with the led class' idea of
>> what
>>     the led is doing, which is annoying at best.
>
> Hmm. So userland can read the LED state, and it can get _some_ value
> back, but it can not know if it is current state or not.
>
> I don't think that's a good interface. I see it is from 2008... is
> someone using it? Maybe it is too late for revert.

I can imagine it being used in flash LED use case. E.g. one
could use oneshot trigger to trigger flash strobe, and then
he could periodically read brightness file to check, for whatever
reason, whether the flash is strobing.

> But I'd certainly not extend it with poll.

We could add a dedicated file e.g. hw_brightness_change for that
(maybe someone will have a better candidate for the file name).

This way it would be semantically consistent to report only
hardware originating brightness changes on it, which was the
initial reason for adding the brightness change notification
feature.

Moreover, LED class drivers could report on this file the
brightness level which was set by the firmware, which wouldn't
affect current LED class device brightness setting, unless
brightness file is read (and brightness_get op is supported).

> IMO reading/polling should only be available with some triggers. It
> does not make sense with "CPU load" trigger. It makes sense with
> "keyboard light changeable by hardware" trigger.


-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 11-11-16 18:03, Jacek Anaszewski wrote:
> On 11/11/2016 01:01 PM, Pavel Machek wrote:
>> On Thu 2016-11-10 22:34:07, Jacek Anaszewski wrote:
>>> Hi,
>>>
>>> On 11/10/2016 09:29 PM, Pavel Machek wrote:
>>>> On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
>>>>> * Pavel Machek <pavel@ucw.cz> [161110 09:29]:
>>>>>> Hi!
>>>>>>
>>>>>>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
>>>>>>>>>> the sysfs brightness attr for changes.") breaks runtime PM for me.
>>>>>>>>>>
>>>>>>>>>> On my omap dm3730 based test system, idle power consumption is over 70
>>>>>>>>>> times higher now with this patch! It goes from about 6mW for the core
>>>>>>>>>> system to over 440mW during idle meaning there's some busy timer now
>>>>>>>>>> active.
>>>>>>>>>>
>>>>>>>>>> Reverting this patch fixes the issue. Any ideas?
>>>>>>
>>>>>> Are you using any LED that toggles with high frequency? Like perhaps
>>>>>> LED that is lit when CPU is active?
>>>>>
>>>>> Yeah one of them seems to have cpu0 as the default trigger.
>>>>
>>>> Aha. Its quite obvious we don't want to notify sysfs each time that
>>>> one is toggled, right?
>>>>
>>>> IMO brightness should display max brightness for the trigger, as Hans
>>>> suggested, anything else is madness for trigger such as cpu activity.
>>>
>>> Are you suggesting that we should revert changes introduced
>>> by below patch?
>>>
>>> commit 29d76dfa29fe22583aefddccda0bc56aa81035dc
>>> Author: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
>>> Date:   Tue Mar 18 09:47:48 2008 +0000
>>>
>>>     leds: Add support to leds with readable status
>>>
>>>     Some led hardware allows drivers to query the led state, and this patch
>>>     adds a hook to let the led class take advantage of that information when
>>>     available.
>>>
>>>     Without this functionality, when access to the led hardware is not
>>>     exclusive (i.e. firmware or hardware might change its state behind the
>>>     kernel's back), reality goes out of sync with the led class' idea of
>>> what
>>>     the led is doing, which is annoying at best.
>>
>> Hmm. So userland can read the LED state, and it can get _some_ value
>> back, but it can not know if it is current state or not.
>>
>> I don't think that's a good interface. I see it is from 2008... is
>> someone using it? Maybe it is too late for revert.
>
> I can imagine it being used in flash LED use case. E.g. one
> could use oneshot trigger to trigger flash strobe, and then
> he could periodically read brightness file to check, for whatever
> reason, whether the flash is strobing.
>
>> But I'd certainly not extend it with poll.
>
> We could add a dedicated file e.g. hw_brightness_change for that
> (maybe someone will have a better candidate for the file name).

Why a dedicated file? Are we going to mirror brightness here
wrt r/w (show/store) behavior ? If not userspace now needs
2 open fds which is not really nice. If we are and we are
not going to use poll for something else on brightness itself
then why not just poll directly on brightness ?

Thinking more about this, I'm strongly against having to do
poll on /sys/.../bar to detect changes on /sys/.../foo that
is something which no other interface does. So my vote on this
is NACK for the having a separate file for this.

Regards,

Hans



>
> This way it would be semantically consistent to report only
> hardware originating brightness changes on it, which was the
> initial reason for adding the brightness change notification
> feature.
>
> Moreover, LED class drivers could report on this file the
> brightness level which was set by the firmware, which wouldn't
> affect current LED class device brightness setting, unless
> brightness file is read (and brightness_get op is supported).
>
>> IMO reading/polling should only be available with some triggers. It
>> does not make sense with "CPU load" trigger. It makes sense with
>> "keyboard light changeable by hardware" trigger.
>
>

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 1194 bytes --]

Hi!

Reason #1:

> >>Hmm. So userland can read the LED state, and it can get _some_ value
> >>back, but it can not know if it is current state or not.

> Why a dedicated file? Are we going to mirror brightness here
> wrt r/w (show/store) behavior ? If not userspace now needs
> 2 open fds which is not really nice. If we are and we are
> not going to use poll for something else on brightness itself
> then why not just poll directly on brightness ?

Reason #1 is above.

Reason #2 is "if userspace sees brightness file, it can not know if
the notifications on change actually work or not".

Reason #3 is that you broke Tony's system. Polling does not make sense
when trigger such as "CPU in use" is active.

Reason #4 is that there are really two brightnesses:

1) maximum brightness trigger is going to use

2) current brightness

Currently writing to "brightness" file changes 1), but reading returns
2) when available.

So, feel free to propose better interface. One that solves #1..#4
above.

Thanks,
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 11-11-16 23:12, Pavel Machek wrote:
> Hi!
>
> Reason #1:
>
>>>> Hmm. So userland can read the LED state, and it can get _some_ value
>>>> back, but it can not know if it is current state or not.

That is not correct, the current behavior for eading the brightness
atrribute is to always return the current state.

>> Why a dedicated file? Are we going to mirror brightness here
>> wrt r/w (show/store) behavior ? If not userspace now needs
>> 2 open fds which is not really nice. If we are and we are
>> not going to use poll for something else on brightness itself
>> then why not just poll directly on brightness ?
>
> Reason #1 is above.

See my reply above.

> Reason #2 is "if userspace sees brightness file, it can not know if
> the notifications on change actually work or not".

If it needs to know that it can simply check the kernel version.

> Reason #3 is that you broke Tony's system. Polling does not make sense
> when trigger such as "CPU in use" is active.

Have you seen v4 of my patch? It fixes this while keeping the
polling on the brightness attribute itself, it basically goes
back (more or less) to v1 of my patch which did not have this
problem. I never wanted notification of trigger / blinking
changes because I already feared Tony's problem would happen.

> Reason #4 is that there are really two brightnesses:
>
> 1) maximum brightness trigger is going to use
>
> 2) current brightness
>
> Currently writing to "brightness" file changes 1), but reading returns
> 2) when available.

Right and Jacek has already said that we cannot change the
reading behavior on the brightness file because of ABI concerns.

So if anything we need a new blink_brightness file or such,
which when read shows the maximum brightness when blinking or
triggers are active. Note that we already have a max_brightness
file which is the actual maximum brightness the led supports.

Since the existing ABI behavior is for the existing brightness
file to return the *current* brightness, please explain to me
how polling on say the new blink_brightness file would make
sense to detect changes in the current brightness ?

> So, feel free to propose better interface. One that solves #1..#4
> above.

Proposal 1:

v4 of my patch, see the list. It solves all but #4, which
is out of scope for my patch, feel free to submit a patch to
solve #4 (with a new sysfs attr).

Proposal 2:

Add a new "user_brightness" file, which shows the last brightness
as set by the user, this would show the read behavior we really
want of brightness: show the real brightness when not blinking /
triggers are active, show the brightness used when on when
blinking / triggers are active.

And then we could add poll support on this new user_brightness
file, thus avoiding the problem with the extra cpu-load on
notifications on blinking / triggers.

Regards,

Hans

Three different LED brightnesses (was Re: PM regression with LED changes in next-20161109)

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 4433 bytes --]

On Sat 2016-11-12 09:03:42, Hans de Goede wrote:
> Hi,
> 
> On 11-11-16 23:12, Pavel Machek wrote:
> >Hi!
> >
> >Reason #1:
> >
> >>>>Hmm. So userland can read the LED state, and it can get _some_ value
> >>>>back, but it can not know if it is current state or not.
> 
> That is not correct, the current behavior for eading the brightness
> atrribute is to always return the current state.

No. (Because some hardware can't get back current state of
hardware-controlled leds, and because of blinking).

> >>Why a dedicated file? Are we going to mirror brightness here
> >>wrt r/w (show/store) behavior ? If not userspace now needs
> >>2 open fds which is not really nice. If we are and we are
> >>not going to use poll for something else on brightness itself
> >>then why not just poll directly on brightness ?
> >
> >Reason #1 is above.
> 
> See my reply above.
> 
> >Reason #2 is "if userspace sees brightness file, it can not know if
> >the notifications on change actually work or not".
> 
> If it needs to know that it can simply check the kernel version.

No. Because in case of hardware blinking we can't provide poll()
functionality.

Plus, saying "simply check the kernel version" simply means you should
not be submitting patches to kernel... at all. (Hint... it also does
not work.)

> >Reason #3 is that you broke Tony's system. Polling does not make sense
> >when trigger such as "CPU in use" is active.
> 
> Have you seen v4 of my patch? It fixes this while keeping the
> polling on the brightness attribute itself, it basically goes
> back (more or less) to v1 of my patch which did not have this
> problem. I never wanted notification of trigger / blinking
> changes because I already feared Tony's problem would happen.

Have you seen v67123 of my latest facebook post? It explains why you
are completely wrong.

> >Reason #4 is that there are really two brightnesses:
> >
> >1) maximum brightness trigger is going to use
> >
> >2) current brightness
> >
> >Currently writing to "brightness" file changes 1), but reading returns
> >2) when available.
> 
> Right and Jacek has already said that we cannot change the
> reading behavior on the brightness file because of ABI concerns.

Until there's user that actually reads that, ABI can be fixed. Given
that it basically returns random value,

> >So, feel free to propose better interface. One that solves #1..#4
> >above.
> 
> Proposal 1:
> 
> v4 of my patch, see the list. It solves all but #4, which
> is out of scope for my patch, feel free to submit a patch to
> solve #4 (with a new sysfs attr).

NAK on that. (And it does not solve #1 and #2 at least.)

> Proposal 2:
> 
> Add a new "user_brightness" file, which shows the last brightness
> as set by the user, this would show the read behavior we really
> want of brightness: show the real brightness when not blinking /
> triggers are active, show the brightness used when on when
> blinking / triggers are active.

No, that's just adding more mess on the system.

Here's better proposal:

brightness (write): what we do today. (Mess, but too late to change it)
	   (read): -Esomething or what we do today (if someone
	   	   	       	       	  	   acutally uses it)
           (poll): -Esomething

current_brightness (write): -Esomething, or maybe change brightness
		   for triggers that can work with that
              (read, poll): if the current trigger can get current
	             state of led, do it, otherwise -Esomething...
		     or maybe file should be simply hidden from sysfs.

trigger_max_brightness (read,write): change the maximum brightness for
		       a trigger.
	       (poll): -Esomething

If you have hardware changing the brightness behind kernel's back,
that should be modelled as a trigger. Userspace should know
there's hardware changing it autonomously ... there should be
"hardware-keylight-brightness" trigger, probably impossible to change
(depends on hardware behaviour).

On thinkpad, for example, for many LEDs kernel can select either
"hardware drives the LED", but then current_brightness is unavailable,
or "kernel drives the LED", but then hardware does not touch the led
at all.

Best regards,

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: Three different LED brightnesses (was Re: PM regression with LED changes in next-20161109)

[Hans de Goede]

Hi,

On 13-11-16 10:10, Pavel Machek wrote:
> On Sat 2016-11-12 09:03:42, Hans de Goede wrote:
>> Hi,
>>
>> On 11-11-16 23:12, Pavel Machek wrote:
>>> Hi!
>>>
>>> Reason #1:
>>>
>>>>>> Hmm. So userland can read the LED state, and it can get _some_ value
>>>>>> back, but it can not know if it is current state or not.
>>
>> That is not correct, the current behavior for eading the brightness
>> atrribute is to always return the current state.
>
> No. (Because some hardware can't get back current state of
> hardware-controlled leds, and because of blinking).
>
>>>> Why a dedicated file? Are we going to mirror brightness here
>>>> wrt r/w (show/store) behavior ? If not userspace now needs
>>>> 2 open fds which is not really nice. If we are and we are
>>>> not going to use poll for something else on brightness itself
>>>> then why not just poll directly on brightness ?
>>>
>>> Reason #1 is above.
>>
>> See my reply above.
>>
>>> Reason #2 is "if userspace sees brightness file, it can not know if
>>> the notifications on change actually work or not".
>>
>> If it needs to know that it can simply check the kernel version.
>
> No. Because in case of hardware blinking we can't provide poll()
> functionality.

We have already decided that we do not want to wakeup poll()
on blinking because it causes to many wakeups, and there is
no use-case for waking poll for blinking (or for triggers).

> Plus, saying "simply check the kernel version" simply means you should
> not be submitting patches to kernel... at all. (Hint... it also does
> not work.)

Lets keep things civil please.

>>> Reason #3 is that you broke Tony's system. Polling does not make sense
>>> when trigger such as "CPU in use" is active.
>>
>> Have you seen v4 of my patch? It fixes this while keeping the
>> polling on the brightness attribute itself, it basically goes
>> back (more or less) to v1 of my patch which did not have this
>> problem. I never wanted notification of trigger / blinking
>> changes because I already feared Tony's problem would happen.
>
> Have you seen v67123 of my latest facebook post? It explains why you
> are completely wrong.

Lets keep things civil please.

>>> Reason #4 is that there are really two brightnesses:
>>>
>>> 1) maximum brightness trigger is going to use
>>>
>>> 2) current brightness
>>>
>>> Currently writing to "brightness" file changes 1), but reading returns
>>> 2) when available.
>>
>> Right and Jacek has already said that we cannot change the
>> reading behavior on the brightness file because of ABI concerns.
>
> Until there's user that actually reads that, ABI can be fixed. Given
> that it basically returns random value,

Jacek has pretty much nacked fixing this because of ABI concerns,
so I see no use in further discussing changing the read behavior
of the existing brightness sysfs attribute.

>>> So, feel free to propose better interface. One that solves #1..#4
>>> above.
>>
>> Proposal 1:
>>
>> v4 of my patch, see the list. It solves all but #4, which
>> is out of scope for my patch, feel free to submit a patch to
>> solve #4 (with a new sysfs attr).
>
> NAK on that. (And it does not solve #1 and #2 at least.)
>
>> Proposal 2:
>>
>> Add a new "user_brightness" file, which shows the last brightness
>> as set by the user, this would show the read behavior we really
>> want of brightness: show the real brightness when not blinking /
>> triggers are active, show the brightness used when on when
>> blinking / triggers are active.
>
> No, that's just adding more mess on the system.
>
> Here's better proposal:
>
> brightness (write): what we do today. (Mess, but too late to change it)
> 	   (read): -Esomething or what we do today (if someone
> 	   	   	       	       	  	   acutally uses it)

As said Jacek has already nacked any changes to read behavior.

>            (poll): -Esomething

Making poll() on sysfs attributes return -Esomething is not possible,
the internal sysfs API does not allow this.

> current_brightness (write): -Esomething, or maybe change brightness
> 		   for triggers that can work with that
>               (read, poll): if the current trigger can get current
> 	             state of led, do it, otherwise -Esomething...
> 		     or maybe file should be simply hidden from sysfs.

So write is -EINVAL and read is the same as what brightness currently does,
so I see no use in introducing this new file.

Also this reintroduces all the issues of v2 of my poll() patch since the
CPU load problem is not actually in waking up userspace, it is in
even checking if there are any userspace waiters. TLDR we simply cannot
have poll behavior where we need to try and wakeup userspace on
every blink / every time a trigger triggers.

> trigger_max_brightness (read,write): change the maximum brightness for
> 		       a trigger.
> 	       (poll): -Esomething

The write behavior here is the same as what brightness currently does
and the read behavior is that of what I suggested for user_brightness,
minus that you've not definied the read behavior for when no trigger /
blinking is active.

In itself I would be fine with this file to work around the read
behavior of trigger_max_brightness, but you've not solved the
polling problem.

We've 2 sorts of brightness really:

1) transient brightness, aka current brightness, when blinking or
triggers are used this will switch many times a second
between off and some on level.

2) non-transient brightness, for non blinking leds this is the
actual brightness, for blinking leds this is the brightness
level used when the led is on.

Now we want to have a sysfs attribute reflecting 2, so that
userspace can poll on that, both for my use-case as well as so
that userspace process a can detect changes made by writing to
the brightness file by process b.

So maybe we need to simply call the new attribute
non_transient_brightness instead of user_brightness?

> If you have hardware changing the brightness behind kernel's back,
> that should be modelled as a trigger. Userspace should know
> there's hardware changing it autonomously ... there should be
> "hardware-keylight-brightness" trigger, probably impossible to change
> (depends on hardware behaviour).
>
> On thinkpad, for example, for many LEDs kernel can select either
> "hardware drives the LED", but then current_brightness is unavailable,
> or "kernel drives the LED", but then hardware does not touch the led
> at all.

Whether or not the hardware changing the brightness behind kernel's
should be modeled as a trigger is really outside of the scope of
this discussion, as it is not related to the issues with the
brightness atrribute / polling on the brightness attribute.

Regards,

Hans

Re: Three different LED brightnesses (was Re: PM regression with LED changes in next-20161109)

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 2171 bytes --]

Hi!

> >No, that's just adding more mess on the system.
> >
> >Here's better proposal:
> >
> >brightness (write): what we do today. (Mess, but too late to change it)
> >	   (read): -Esomething or what we do today (if someone
> >	   	   	       	       	  	   acutally uses it)
> 
> As said Jacek has already nacked any changes to read behavior.
> 
> >           (poll): -Esomething
> 
> Making poll() on sysfs attributes return -Esomething is not possible,
> the internal sysfs API does not allow this.
> 
> >current_brightness (write): -Esomething, or maybe change brightness
> >		   for triggers that can work with that
> >              (read, poll): if the current trigger can get current
> >	             state of led, do it, otherwise -Esomething...
> >		     or maybe file should be simply hidden from sysfs.
> 
> So write is -EINVAL and read is the same as what brightness currently does,
> so I see no use in introducing this new file.

No, read is not same as current brightness file. Currently, reading
brightness file returns either current brightness or maximum
brightness, and userspace can't tell which is which.

> >trigger_max_brightness (read,write): change the maximum brightness for
> >		       a trigger.
> >	       (poll): -Esomething
> 
> The write behavior here is the same as what brightness currently does
> and the read behavior is that of what I suggested for

No, it is not. "brightness" behaviour is currently broken, as it sets
either current brightness or maximum trigger brightness.

> We've 2 sorts of brightness really:
> 
> 1) transient brightness, aka current brightness, when blinking or
> triggers are used this will switch many times a second
> between off and some on level.
> 
> 2) non-transient brightness, for non blinking leds this is the
> actual brightness, for blinking leds this is the brightness
> level used when the led is on.

Yes, and we want reasonable interface where userspace sees those two
brightnesses.

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/11/2016 08:28 PM, Hans de Goede wrote:
> Hi,
>
> On 11-11-16 18:03, Jacek Anaszewski wrote:
>> On 11/11/2016 01:01 PM, Pavel Machek wrote:
>>> On Thu 2016-11-10 22:34:07, Jacek Anaszewski wrote:
>>>> Hi,
>>>>
>>>> On 11/10/2016 09:29 PM, Pavel Machek wrote:
>>>>> On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
>>>>>> * Pavel Machek <pavel@ucw.cz> [161110 09:29]:
>>>>>>> Hi!
>>>>>>>
>>>>>>>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for
>>>>>>>>>>> poll()ing
>>>>>>>>>>> the sysfs brightness attr for changes.") breaks runtime PM
>>>>>>>>>>> for me.
>>>>>>>>>>>
>>>>>>>>>>> On my omap dm3730 based test system, idle power consumption
>>>>>>>>>>> is over 70
>>>>>>>>>>> times higher now with this patch! It goes from about 6mW for
>>>>>>>>>>> the core
>>>>>>>>>>> system to over 440mW during idle meaning there's some busy
>>>>>>>>>>> timer now
>>>>>>>>>>> active.
>>>>>>>>>>>
>>>>>>>>>>> Reverting this patch fixes the issue. Any ideas?
>>>>>>>
>>>>>>> Are you using any LED that toggles with high frequency? Like perhaps
>>>>>>> LED that is lit when CPU is active?
>>>>>>
>>>>>> Yeah one of them seems to have cpu0 as the default trigger.
>>>>>
>>>>> Aha. Its quite obvious we don't want to notify sysfs each time that
>>>>> one is toggled, right?
>>>>>
>>>>> IMO brightness should display max brightness for the trigger, as Hans
>>>>> suggested, anything else is madness for trigger such as cpu activity.
>>>>
>>>> Are you suggesting that we should revert changes introduced
>>>> by below patch?
>>>>
>>>> commit 29d76dfa29fe22583aefddccda0bc56aa81035dc
>>>> Author: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
>>>> Date:   Tue Mar 18 09:47:48 2008 +0000
>>>>
>>>>     leds: Add support to leds with readable status
>>>>
>>>>     Some led hardware allows drivers to query the led state, and
>>>> this patch
>>>>     adds a hook to let the led class take advantage of that
>>>> information when
>>>>     available.
>>>>
>>>>     Without this functionality, when access to the led hardware is not
>>>>     exclusive (i.e. firmware or hardware might change its state
>>>> behind the
>>>>     kernel's back), reality goes out of sync with the led class'
>>>> idea of
>>>> what
>>>>     the led is doing, which is annoying at best.
>>>
>>> Hmm. So userland can read the LED state, and it can get _some_ value
>>> back, but it can not know if it is current state or not.
>>>
>>> I don't think that's a good interface. I see it is from 2008... is
>>> someone using it? Maybe it is too late for revert.
>>
>> I can imagine it being used in flash LED use case. E.g. one
>> could use oneshot trigger to trigger flash strobe, and then
>> he could periodically read brightness file to check, for whatever
>> reason, whether the flash is strobing.
>>
>>> But I'd certainly not extend it with poll.
>>
>> We could add a dedicated file e.g. hw_brightness_change for that
>> (maybe someone will have a better candidate for the file name).
>
> Why a dedicated file? Are we going to mirror brightness here
> wrt r/w (show/store) behavior ? If not userspace now needs
> 2 open fds which is not really nice. If we are and we are
> not going to use poll for something else on brightness itself
> then why not just poll directly on brightness ?

My main concern is that reporting only hw brightness changes
wouldn't be consistent with general brightness file purpose.
One could expect that brightness changes made by triggers
should be also reported.

I'd make it only readable, so it wouldn't mirror brightness
file behavior.

Its purpose would be clear: notify hw brightness changes
and provide the brightness value that was set by the hardware
last time. It implies that this value could be different from
the one the brightness file reports. E.g. hw could have changed
brightness, which could be later updated through brightness
file, but hw_brightness_change would still report brightness level
that was set by the hardware last time. It could be useful
e.g. in case of showing the difference between the desired
value and the currently allowed configuration (e.g. if the
firmware automatically adjusted the value set by the user).

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 12-11-16 11:24, Jacek Anaszewski wrote:
> Hi,
>
> On 11/11/2016 08:28 PM, Hans de Goede wrote:
>> Hi,
>>
>> On 11-11-16 18:03, Jacek Anaszewski wrote:
>>> On 11/11/2016 01:01 PM, Pavel Machek wrote:
>>>> On Thu 2016-11-10 22:34:07, Jacek Anaszewski wrote:
>>>>> Hi,
>>>>>
>>>>> On 11/10/2016 09:29 PM, Pavel Machek wrote:
>>>>>> On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
>>>>>>> * Pavel Machek <pavel@ucw.cz> [161110 09:29]:
>>>>>>>> Hi!
>>>>>>>>
>>>>>>>>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for
>>>>>>>>>>>> poll()ing
>>>>>>>>>>>> the sysfs brightness attr for changes.") breaks runtime PM
>>>>>>>>>>>> for me.
>>>>>>>>>>>>
>>>>>>>>>>>> On my omap dm3730 based test system, idle power consumption
>>>>>>>>>>>> is over 70
>>>>>>>>>>>> times higher now with this patch! It goes from about 6mW for
>>>>>>>>>>>> the core
>>>>>>>>>>>> system to over 440mW during idle meaning there's some busy
>>>>>>>>>>>> timer now
>>>>>>>>>>>> active.
>>>>>>>>>>>>
>>>>>>>>>>>> Reverting this patch fixes the issue. Any ideas?
>>>>>>>>
>>>>>>>> Are you using any LED that toggles with high frequency? Like perhaps
>>>>>>>> LED that is lit when CPU is active?
>>>>>>>
>>>>>>> Yeah one of them seems to have cpu0 as the default trigger.
>>>>>>
>>>>>> Aha. Its quite obvious we don't want to notify sysfs each time that
>>>>>> one is toggled, right?
>>>>>>
>>>>>> IMO brightness should display max brightness for the trigger, as Hans
>>>>>> suggested, anything else is madness for trigger such as cpu activity.
>>>>>
>>>>> Are you suggesting that we should revert changes introduced
>>>>> by below patch?
>>>>>
>>>>> commit 29d76dfa29fe22583aefddccda0bc56aa81035dc
>>>>> Author: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
>>>>> Date:   Tue Mar 18 09:47:48 2008 +0000
>>>>>
>>>>>     leds: Add support to leds with readable status
>>>>>
>>>>>     Some led hardware allows drivers to query the led state, and
>>>>> this patch
>>>>>     adds a hook to let the led class take advantage of that
>>>>> information when
>>>>>     available.
>>>>>
>>>>>     Without this functionality, when access to the led hardware is not
>>>>>     exclusive (i.e. firmware or hardware might change its state
>>>>> behind the
>>>>>     kernel's back), reality goes out of sync with the led class'
>>>>> idea of
>>>>> what
>>>>>     the led is doing, which is annoying at best.
>>>>
>>>> Hmm. So userland can read the LED state, and it can get _some_ value
>>>> back, but it can not know if it is current state or not.
>>>>
>>>> I don't think that's a good interface. I see it is from 2008... is
>>>> someone using it? Maybe it is too late for revert.
>>>
>>> I can imagine it being used in flash LED use case. E.g. one
>>> could use oneshot trigger to trigger flash strobe, and then
>>> he could periodically read brightness file to check, for whatever
>>> reason, whether the flash is strobing.
>>>
>>>> But I'd certainly not extend it with poll.
>>>
>>> We could add a dedicated file e.g. hw_brightness_change for that
>>> (maybe someone will have a better candidate for the file name).
>>
>> Why a dedicated file? Are we going to mirror brightness here
>> wrt r/w (show/store) behavior ? If not userspace now needs
>> 2 open fds which is not really nice. If we are and we are
>> not going to use poll for something else on brightness itself
>> then why not just poll directly on brightness ?
>
> My main concern is that reporting only hw brightness changes
> wouldn't be consistent with general brightness file purpose.
> One could expect that brightness changes made by triggers
> should be also reported.

Ok, I agree that not notifying poll() while an actual
read() would result in a different value is not really good
semantics.

I don't like to call it hw_brightness_change though, as
mentioned before I believe that if we were to start with
a clean slate we would make the brightness file's read/write
behavior more a mirror of itself.

So I would like to propose creating a new read-write
user_brightness file.

The write behavior would be 100% identical to the brightness
file (in code terms it will call the same store function).

The the read behavior otoh will be different: it will shows
the last brightness as set by the user, this would show the
read behavior we really want of brightness: show the real
brightness when not blinking / triggers are active and show
the brightness used when on when blinking / triggers are active.

We could then add poll support on this new user_brightness
file, thus avoiding the problem with the extra cpu-load on
notifications on blinking / triggers.

> I'd make it only readable, so it wouldn't mirror brightness
> file behavior.

Then userspace which wants to be able to read + write + poll
the brightness again needs to open 2 fds, as suggested
above for the new user_brightness file it will be easy
to just make it mimic the brightness file write behavior
and then userspace only needs to open one fd.

Regards,

Hans




>
> Its purpose would be clear: notify hw brightness changes
> and provide the brightness value that was set by the hardware
> last time. It implies that this value could be different from
> the one the brightness file reports. E.g. hw could have changed
> brightness, which could be later updated through brightness
> file, but hw_brightness_change would still report brightness level
> that was set by the hardware last time. It could be useful
> e.g. in case of showing the difference between the desired
> value and the currently allowed configuration (e.g. if the
> firmware automatically adjusted the value set by the user).
>

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/12/2016 11:33 AM, Hans de Goede wrote:
> Hi,
>
> On 12-11-16 11:24, Jacek Anaszewski wrote:
>> Hi,
>>
>> On 11/11/2016 08:28 PM, Hans de Goede wrote:
>>> Hi,
>>>
>>> On 11-11-16 18:03, Jacek Anaszewski wrote:
>>>> On 11/11/2016 01:01 PM, Pavel Machek wrote:
>>>>> On Thu 2016-11-10 22:34:07, Jacek Anaszewski wrote:
>>>>>> Hi,
>>>>>>
>>>>>> On 11/10/2016 09:29 PM, Pavel Machek wrote:
>>>>>>> On Thu 2016-11-10 10:55:37, Tony Lindgren wrote:
>>>>>>>> * Pavel Machek <pavel@ucw.cz> [161110 09:29]:
>>>>>>>>> Hi!
>>>>>>>>>
>>>>>>>>>>>>> Looks like commit 883d32ce3385 ("leds: core: Add support for
>>>>>>>>>>>>> poll()ing
>>>>>>>>>>>>> the sysfs brightness attr for changes.") breaks runtime PM
>>>>>>>>>>>>> for me.
>>>>>>>>>>>>>
>>>>>>>>>>>>> On my omap dm3730 based test system, idle power consumption
>>>>>>>>>>>>> is over 70
>>>>>>>>>>>>> times higher now with this patch! It goes from about 6mW for
>>>>>>>>>>>>> the core
>>>>>>>>>>>>> system to over 440mW during idle meaning there's some busy
>>>>>>>>>>>>> timer now
>>>>>>>>>>>>> active.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Reverting this patch fixes the issue. Any ideas?
>>>>>>>>>
>>>>>>>>> Are you using any LED that toggles with high frequency? Like
>>>>>>>>> perhaps
>>>>>>>>> LED that is lit when CPU is active?
>>>>>>>>
>>>>>>>> Yeah one of them seems to have cpu0 as the default trigger.
>>>>>>>
>>>>>>> Aha. Its quite obvious we don't want to notify sysfs each time that
>>>>>>> one is toggled, right?
>>>>>>>
>>>>>>> IMO brightness should display max brightness for the trigger, as
>>>>>>> Hans
>>>>>>> suggested, anything else is madness for trigger such as cpu
>>>>>>> activity.
>>>>>>
>>>>>> Are you suggesting that we should revert changes introduced
>>>>>> by below patch?
>>>>>>
>>>>>> commit 29d76dfa29fe22583aefddccda0bc56aa81035dc
>>>>>> Author: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
>>>>>> Date:   Tue Mar 18 09:47:48 2008 +0000
>>>>>>
>>>>>>     leds: Add support to leds with readable status
>>>>>>
>>>>>>     Some led hardware allows drivers to query the led state, and
>>>>>> this patch
>>>>>>     adds a hook to let the led class take advantage of that
>>>>>> information when
>>>>>>     available.
>>>>>>
>>>>>>     Without this functionality, when access to the led hardware is
>>>>>> not
>>>>>>     exclusive (i.e. firmware or hardware might change its state
>>>>>> behind the
>>>>>>     kernel's back), reality goes out of sync with the led class'
>>>>>> idea of
>>>>>> what
>>>>>>     the led is doing, which is annoying at best.
>>>>>
>>>>> Hmm. So userland can read the LED state, and it can get _some_ value
>>>>> back, but it can not know if it is current state or not.
>>>>>
>>>>> I don't think that's a good interface. I see it is from 2008... is
>>>>> someone using it? Maybe it is too late for revert.
>>>>
>>>> I can imagine it being used in flash LED use case. E.g. one
>>>> could use oneshot trigger to trigger flash strobe, and then
>>>> he could periodically read brightness file to check, for whatever
>>>> reason, whether the flash is strobing.
>>>>
>>>>> But I'd certainly not extend it with poll.
>>>>
>>>> We could add a dedicated file e.g. hw_brightness_change for that
>>>> (maybe someone will have a better candidate for the file name).
>>>
>>> Why a dedicated file? Are we going to mirror brightness here
>>> wrt r/w (show/store) behavior ? If not userspace now needs
>>> 2 open fds which is not really nice. If we are and we are
>>> not going to use poll for something else on brightness itself
>>> then why not just poll directly on brightness ?
>>
>> My main concern is that reporting only hw brightness changes
>> wouldn't be consistent with general brightness file purpose.
>> One could expect that brightness changes made by triggers
>> should be also reported.
>
> Ok, I agree that not notifying poll() while an actual
> read() would result in a different value is not really good
> semantics.
>
> I don't like to call it hw_brightness_change though, as
> mentioned before I believe that if we were to start with
> a clean slate we would make the brightness file's read/write
> behavior more a mirror of itself.
>
> So I would like to propose creating a new read-write
> user_brightness file.
>
> The write behavior would be 100% identical to the brightness
> file (in code terms it will call the same store function).
>
> The the read behavior otoh will be different: it will shows
> the last brightness as set by the user, this would show the
> read behavior we really want of brightness: show the real
> brightness when not blinking / triggers are active and show
> the brightness used when on when blinking / triggers are active.
>
> We could then add poll support on this new user_brightness
> file, thus avoiding the problem with the extra cpu-load on
> notifications on blinking / triggers.

I agree that user_brightness allows to solve the issues you raised
about inconsistent write and read brightness' semantics
(which is not that painful IMHO).

Reporting non-user brightness changes on user_brightness file
doesn't sound reasonable though. Also, how would we read the
brightness set by the firmware? We'd have to read brightness
file, so still two files would have to be opened which is
a second drawback of this approach.

Having no difference in this area between the two approaches
I'm still in favour of the read-only file for notifying
brightness changes procured by hardware.

>> I'd make it only readable, so it wouldn't mirror brightness
>> file behavior.
>
> Then userspace which wants to be able to read + write + poll
> the brightness again needs to open 2 fds, as suggested
> above for the new user_brightness file it will be easy
> to just make it mimic the brightness file write behavior
> and then userspace only needs to open one fd.
>
> Regards,
>
> Hans
>
>
>
>
>>
>> Its purpose would be clear: notify hw brightness changes
>> and provide the brightness value that was set by the hardware
>> last time. It implies that this value could be different from
>> the one the brightness file reports. E.g. hw could have changed
>> brightness, which could be later updated through brightness
>> file, but hw_brightness_change would still report brightness level
>> that was set by the hardware last time. It could be useful
>> e.g. in case of showing the difference between the desired
>> value and the currently allowed configuration (e.g. if the
>> firmware automatically adjusted the value set by the user).
>>
>

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 12-11-16 20:14, Jacek Anaszewski wrote:

<snip>

>>>> Why a dedicated file? Are we going to mirror brightness here
>>>> wrt r/w (show/store) behavior ? If not userspace now needs
>>>> 2 open fds which is not really nice. If we are and we are
>>>> not going to use poll for something else on brightness itself
>>>> then why not just poll directly on brightness ?
>>>
>>> My main concern is that reporting only hw brightness changes
>>> wouldn't be consistent with general brightness file purpose.
>>> One could expect that brightness changes made by triggers
>>> should be also reported.
>>
>> Ok, I agree that not notifying poll() while an actual
>> read() would result in a different value is not really good
>> semantics.
>>
>> I don't like to call it hw_brightness_change though, as
>> mentioned before I believe that if we were to start with
>> a clean slate we would make the brightness file's read/write
>> behavior more a mirror of itself.
>>
>> So I would like to propose creating a new read-write
>> user_brightness file.
>>
>> The write behavior would be 100% identical to the brightness
>> file (in code terms it will call the same store function).
>>
>> The the read behavior otoh will be different: it will shows
>> the last brightness as set by the user, this would show the
>> read behavior we really want of brightness: show the real
>> brightness when not blinking / triggers are active and show
>> the brightness used when on when blinking / triggers are active.
>>
>> We could then add poll support on this new user_brightness
>> file, thus avoiding the problem with the extra cpu-load on
>> notifications on blinking / triggers.
>
> I agree that user_brightness allows to solve the issues you raised
> about inconsistent write and read brightness' semantics
> (which is not that painful IMHO).
>
> Reporting non-user brightness changes on user_brightness file
> doesn't sound reasonable though.

The changes I'm interested in are user brightness changes they
are just not done through sysfs, but through a hardwired hotkey,
they are however very much done by the user.

> Also, how would we read the
> brightness set by the firmware? We'd have to read brightness
> file, so still two files would have to be opened which is
> a second drawback of this approach.

No, look carefully at the definition of the read behavior
I plan to put in the ABI doc:

"Reading this file will return the actual led brightness
when not blinking and no triggers are active; reading this
file will return the brightness used when the led is on
when blinking or triggers are active."

So for e.g. the backlit keyboard case reading this single
file will return the actual brightness of the backlight,
since this does not involve blinking or triggers.

Basically the idea is that the user_brightness file
will have the semantics which IMHO the brightness file
itself should have had from the beginning, but which
we can't change now due to ABI reasons.

> Having no difference in this area between the two approaches
> I'm still in favour of the read-only file for notifying
> brightness changes procured by hardware.

That brings back the needing 2 fds problem; and does
not solve userspace not being able to reliably read
the led on brightness when blinking or using triggers.

And this also has the issue that one is doing poll() on
one fd to detect changes on another fd, which is completely
unheard of in any kernel API, so I still vote NACK for the
entire idea of having a different file purely for notifying
changes. The way unix defines poll to work means that the
poll() and read() must be on the same fd, anything else
does not make sense.

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/12/2016 10:14 PM, Hans de Goede wrote:
> Hi,
>
> On 12-11-16 20:14, Jacek Anaszewski wrote:
>
> <snip>
>
>>>>> Why a dedicated file? Are we going to mirror brightness here
>>>>> wrt r/w (show/store) behavior ? If not userspace now needs
>>>>> 2 open fds which is not really nice. If we are and we are
>>>>> not going to use poll for something else on brightness itself
>>>>> then why not just poll directly on brightness ?
>>>>
>>>> My main concern is that reporting only hw brightness changes
>>>> wouldn't be consistent with general brightness file purpose.
>>>> One could expect that brightness changes made by triggers
>>>> should be also reported.
>>>
>>> Ok, I agree that not notifying poll() while an actual
>>> read() would result in a different value is not really good
>>> semantics.
>>>
>>> I don't like to call it hw_brightness_change though, as
>>> mentioned before I believe that if we were to start with
>>> a clean slate we would make the brightness file's read/write
>>> behavior more a mirror of itself.
>>>
>>> So I would like to propose creating a new read-write
>>> user_brightness file.
>>>
>>> The write behavior would be 100% identical to the brightness
>>> file (in code terms it will call the same store function).
>>>
>>> The the read behavior otoh will be different: it will shows
>>> the last brightness as set by the user, this would show the
>>> read behavior we really want of brightness: show the real
>>> brightness when not blinking / triggers are active and show
>>> the brightness used when on when blinking / triggers are active.
>>>
>>> We could then add poll support on this new user_brightness
>>> file, thus avoiding the problem with the extra cpu-load on
>>> notifications on blinking / triggers.
>>
>> I agree that user_brightness allows to solve the issues you raised
>> about inconsistent write and read brightness' semantics
>> (which is not that painful IMHO).
>>
>> Reporting non-user brightness changes on user_brightness file
>> doesn't sound reasonable though.
>
> The changes I'm interested in are user brightness changes they
> are just not done through sysfs, but through a hardwired hotkey,
> they are however very much done by the user.

Ah, so this file name would be misleading especially taking into account
the context in which "user" is used in kernel, which predominantly
means "userspace", e.g. copy_to_user(), copy_from_user().

>> Also, how would we read the
>> brightness set by the firmware? We'd have to read brightness
>> file, so still two files would have to be opened which is
>> a second drawback of this approach.
>
> No, look carefully at the definition of the read behavior
> I plan to put in the ABI doc:

OK, "user" was what confused me. So in this case changes made
by the firmware even if in a result of user activity
(pressing hardware key) obviously cannot be treated similarly
to the changes made from the userspace context.

Unless you're able to give references to the kernel code which
contradict my judgement.

>
> "Reading this file will return the actual led brightness
> when not blinking and no triggers are active; reading this
> file will return the brightness used when the led is on
> when blinking or triggers are active."

This is unnecessarily entangled. Blinking means timer trigger
is active.

> So for e.g. the backlit keyboard case reading this single
> file will return the actual brightness of the backlight,
> since this does not involve blinking or triggers.
>
> Basically the idea is that the user_brightness file
> will have the semantics which IMHO the brightness file
> itself should have had from the beginning, but which
> we can't change now due to ABI reasons.

And in fact introducing user_brightness file would indeed
fix that shortcoming. However without providing notifications
of hw brightness changes on it.

>> Having no difference in this area between the two approaches
>> I'm still in favour of the read-only file for notifying
>> brightness changes procured by hardware.
>
> That brings back the needing 2 fds problem; and does
> not solve userspace not being able to reliably read
> the led on brightness when blinking or using triggers.
>
> And this also has the issue that one is doing poll() on
> one fd to detect changes on another fd,

It is not necessarily true. We can treat the polling on
hw_brightness_change file as a means to detect brightness
changes procured by hardware and we can read that brightness
by executing read on this same fd. It could return -ENODATA
if no such an event has occurred so far.

> which is completely
> unheard of in any kernel API, so I still vote NACK for the
> entire idea of having a different file purely for notifying
> changes. The way unix defines poll to work means that the
> poll() and read() must be on the same fd, anything else
> does not make sense.

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 13-11-16 12:44, Jacek Anaszewski wrote:
> Hi,
>
> On 11/12/2016 10:14 PM, Hans de Goede wrote:

<snip>

>>>> So I would like to propose creating a new read-write
>>>> user_brightness file.
>>>>
>>>> The write behavior would be 100% identical to the brightness
>>>> file (in code terms it will call the same store function).
>>>>
>>>> The the read behavior otoh will be different: it will shows
>>>> the last brightness as set by the user, this would show the
>>>> read behavior we really want of brightness: show the real
>>>> brightness when not blinking / triggers are active and show
>>>> the brightness used when on when blinking / triggers are active.
>>>>
>>>> We could then add poll support on this new user_brightness
>>>> file, thus avoiding the problem with the extra cpu-load on
>>>> notifications on blinking / triggers.
>>>
>>> I agree that user_brightness allows to solve the issues you raised
>>> about inconsistent write and read brightness' semantics
>>> (which is not that painful IMHO).
>>>
>>> Reporting non-user brightness changes on user_brightness file
>>> doesn't sound reasonable though.
>>
>> The changes I'm interested in are user brightness changes they
>> are just not done through sysfs, but through a hardwired hotkey,
>> they are however very much done by the user.
>
> Ah, so this file name would be misleading especially taking into account
> the context in which "user" is used in kernel, which predominantly
> means "userspace", e.g. copy_to_user(), copy_from_user().
>
>>> Also, how would we read the
>>> brightness set by the firmware? We'd have to read brightness
>>> file, so still two files would have to be opened which is
>>> a second drawback of this approach.
>>
>> No, look carefully at the definition of the read behavior
>> I plan to put in the ABI doc:
>
> OK, "user" was what confused me. So in this case changes made
> by the firmware even if in a result of user activity
> (pressing hardware key) obviously cannot be treated similarly
> to the changes made from the userspace context.

In the end both result on the brightness of the device
changing, so any userspace process interested in monitoring
the brightness will want to know about both type of changes.

> Unless you're able to give references to the kernel code which
> contradict my judgement.

AFAIK the audio code will signal volume changes done by
hardwired buttons the same way as audio changes done
by userspace calling into the kernel. This also makes
sense because in the end, what is interesting for a
mixer app, is that the volume changed, and what the
new volume is.

>> "Reading this file will return the actual led brightness
>> when not blinking and no triggers are active; reading this
>> file will return the brightness used when the led is on
>> when blinking or triggers are active."
>
> This is unnecessarily entangled. Blinking means timer trigger
> is active.

Ok.

>> So for e.g. the backlit keyboard case reading this single
>> file will return the actual brightness of the backlight,
>> since this does not involve blinking or triggers.
>>
>> Basically the idea is that the user_brightness file
>> will have the semantics which IMHO the brightness file
>> itself should have had from the beginning, but which
>> we can't change now due to ABI reasons.
>
> And in fact introducing user_brightness file would indeed
> fix that shortcoming. However without providing notifications
> of hw brightness changes on it.

See above, I believe such a file should report any
changes in brightness, except those caused by triggers,
so it would report hw brightness changes.

Anyways if you're not interested in fixing the
shortcomings of the current read behavior on the
brightness file (I'm fine with that, I can live
with the shortcomings) I suggest that we simply go
with v2 of my poll() patch.

>>> Having no difference in this area between the two approaches
>>> I'm still in favour of the read-only file for notifying
>>> brightness changes procured by hardware.
>>
>> That brings back the needing 2 fds problem; and does
>> not solve userspace not being able to reliably read
>> the led on brightness when blinking or using triggers.
>>
>> And this also has the issue that one is doing poll() on
>> one fd to detect changes on another fd,
>
> It is not necessarily true. We can treat the polling on
> hw_brightness_change file as a means to detect brightness
> changes procured by hardware and we can read that brightness
> by executing read on this same fd. It could return -ENODATA
> if no such an event has occurred so far.

That would still require 2 fds as userspace also wants to
be able to set the keyboard backlight, but allowing read()
on the hw_brightness_change file at least fixes the weirdness
where userspace gets woken from poll() without being able to
read. So if you insist on going the hw_brightness_change file
route, then I can live with that (and upower will simply
need to open 2 fds, that is doable).

But, BUT, I would greatly prefer to just go for v4 of my
patch, which fixes the only real problem we've seen with
my patch as original merged without adding a new, somewhat
convoluted sysfs attribute.

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/13/2016 02:52 PM, Hans de Goede wrote:
> Hi,
>
> On 13-11-16 12:44, Jacek Anaszewski wrote:
>> Hi,
>>
>> On 11/12/2016 10:14 PM, Hans de Goede wrote:
>
> <snip>
>
>>>>> So I would like to propose creating a new read-write
>>>>> user_brightness file.
>>>>>
>>>>> The write behavior would be 100% identical to the brightness
>>>>> file (in code terms it will call the same store function).
>>>>>
>>>>> The the read behavior otoh will be different: it will shows
>>>>> the last brightness as set by the user, this would show the
>>>>> read behavior we really want of brightness: show the real
>>>>> brightness when not blinking / triggers are active and show
>>>>> the brightness used when on when blinking / triggers are active.
>>>>>
>>>>> We could then add poll support on this new user_brightness
>>>>> file, thus avoiding the problem with the extra cpu-load on
>>>>> notifications on blinking / triggers.
>>>>
>>>> I agree that user_brightness allows to solve the issues you raised
>>>> about inconsistent write and read brightness' semantics
>>>> (which is not that painful IMHO).
>>>>
>>>> Reporting non-user brightness changes on user_brightness file
>>>> doesn't sound reasonable though.
>>>
>>> The changes I'm interested in are user brightness changes they
>>> are just not done through sysfs, but through a hardwired hotkey,
>>> they are however very much done by the user.
>>
>> Ah, so this file name would be misleading especially taking into account
>> the context in which "user" is used in kernel, which predominantly
>> means "userspace", e.g. copy_to_user(), copy_from_user().
>>
>>>> Also, how would we read the
>>>> brightness set by the firmware? We'd have to read brightness
>>>> file, so still two files would have to be opened which is
>>>> a second drawback of this approach.
>>>
>>> No, look carefully at the definition of the read behavior
>>> I plan to put in the ABI doc:
>>
>> OK, "user" was what confused me. So in this case changes made
>> by the firmware even if in a result of user activity
>> (pressing hardware key) obviously cannot be treated similarly
>> to the changes made from the userspace context.
>
> In the end both result on the brightness of the device
> changing, so any userspace process interested in monitoring
> the brightness will want to know about both type of changes.
>
>> Unless you're able to give references to the kernel code which
>> contradict my judgement.
>
> AFAIK the audio code will signal volume changes done by
> hardwired buttons the same way as audio changes done
> by userspace calling into the kernel. This also makes
> sense because in the end, what is interesting for a
> mixer app, is that the volume changed, and what the
> new volume is.

OK, so it is indeed similar to your LED use case. Nonetheless
in case of LED controllers it is also possible that hardware
adjusts LED brightness in case of low battery voltage.

If a device is able e.g. to generate an interrupt to notify this
kind of event, then we would like also to be able to notify the client
about that. It wouldn't be user generated brightness change though.

>>> "Reading this file will return the actual led brightness
>>> when not blinking and no triggers are active; reading this
>>> file will return the brightness used when the led is on
>>> when blinking or triggers are active."
>>
>> This is unnecessarily entangled. Blinking means timer trigger
>> is active.
>
> Ok.
>
>>> So for e.g. the backlit keyboard case reading this single
>>> file will return the actual brightness of the backlight,
>>> since this does not involve blinking or triggers.
>>>
>>> Basically the idea is that the user_brightness file
>>> will have the semantics which IMHO the brightness file
>>> itself should have had from the beginning, but which
>>> we can't change now due to ABI reasons.
>>
>> And in fact introducing user_brightness file would indeed
>> fix that shortcoming. However without providing notifications
>> of hw brightness changes on it.
>
> See above, I believe such a file should report any
> changes in brightness, except those caused by triggers,
> so it would report hw brightness changes.
>
> Anyways if you're not interested in fixing the
> shortcomings of the current read behavior on the
> brightness file (I'm fine with that, I can live
> with the shortcomings) I suggest that we simply go
> with v2 of my poll() patch.

v2 entails power consumption related issues.

Generally I think that we could add the file you proposed,
however it would be good to devise a name which will cover
also the cases when brightness is changed by firmware without
user interaction.

>>>> Having no difference in this area between the two approaches
>>>> I'm still in favour of the read-only file for notifying
>>>> brightness changes procured by hardware.
>>>
>>> That brings back the needing 2 fds problem; and does
>>> not solve userspace not being able to reliably read
>>> the led on brightness when blinking or using triggers.
>>>
>>> And this also has the issue that one is doing poll() on
>>> one fd to detect changes on another fd,
>>
>> It is not necessarily true. We can treat the polling on
>> hw_brightness_change file as a means to detect brightness
>> changes procured by hardware and we can read that brightness
>> by executing read on this same fd. It could return -ENODATA
>> if no such an event has occurred so far.
>
> That would still require 2 fds as userspace also wants to
> be able to set the keyboard backlight, but allowing read()
> on the hw_brightness_change file at least fixes the weirdness
> where userspace gets woken from poll() without being able to
> read. So if you insist on going the hw_brightness_change file
> route, then I can live with that (and upower will simply
> need to open 2 fds, that is doable).
>
> But, BUT, I would greatly prefer to just go for v4 of my
> patch, which fixes the only real problem we've seen with
> my patch as original merged without adding a new, somewhat
> convoluted sysfs attribute.

Hmm, v4 still calls led_notify_brightness_change(led_cdev)
from both __led_set_brightness() and __led_set_brightness_blocking().

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 14-11-16 10:12, Jacek Anaszewski wrote:
> Hi,
>
> On 11/13/2016 02:52 PM, Hans de Goede wrote:
>> Hi,
>>
>> On 13-11-16 12:44, Jacek Anaszewski wrote:
>>> Hi,
>>>
>>> On 11/12/2016 10:14 PM, Hans de Goede wrote:
>>
>> <snip>
>>
>>>>>> So I would like to propose creating a new read-write
>>>>>> user_brightness file.
>>>>>>
>>>>>> The write behavior would be 100% identical to the brightness
>>>>>> file (in code terms it will call the same store function).
>>>>>>
>>>>>> The the read behavior otoh will be different: it will shows
>>>>>> the last brightness as set by the user, this would show the
>>>>>> read behavior we really want of brightness: show the real
>>>>>> brightness when not blinking / triggers are active and show
>>>>>> the brightness used when on when blinking / triggers are active.
>>>>>>
>>>>>> We could then add poll support on this new user_brightness
>>>>>> file, thus avoiding the problem with the extra cpu-load on
>>>>>> notifications on blinking / triggers.
>>>>>
>>>>> I agree that user_brightness allows to solve the issues you raised
>>>>> about inconsistent write and read brightness' semantics
>>>>> (which is not that painful IMHO).
>>>>>
>>>>> Reporting non-user brightness changes on user_brightness file
>>>>> doesn't sound reasonable though.
>>>>
>>>> The changes I'm interested in are user brightness changes they
>>>> are just not done through sysfs, but through a hardwired hotkey,
>>>> they are however very much done by the user.
>>>
>>> Ah, so this file name would be misleading especially taking into account
>>> the context in which "user" is used in kernel, which predominantly
>>> means "userspace", e.g. copy_to_user(), copy_from_user().
>>>
>>>>> Also, how would we read the
>>>>> brightness set by the firmware? We'd have to read brightness
>>>>> file, so still two files would have to be opened which is
>>>>> a second drawback of this approach.
>>>>
>>>> No, look carefully at the definition of the read behavior
>>>> I plan to put in the ABI doc:
>>>
>>> OK, "user" was what confused me. So in this case changes made
>>> by the firmware even if in a result of user activity
>>> (pressing hardware key) obviously cannot be treated similarly
>>> to the changes made from the userspace context.
>>
>> In the end both result on the brightness of the device
>> changing, so any userspace process interested in monitoring
>> the brightness will want to know about both type of changes.
>>
>>> Unless you're able to give references to the kernel code which
>>> contradict my judgement.
>>
>> AFAIK the audio code will signal volume changes done by
>> hardwired buttons the same way as audio changes done
>> by userspace calling into the kernel. This also makes
>> sense because in the end, what is interesting for a
>> mixer app, is that the volume changed, and what the
>> new volume is.
>
> OK, so it is indeed similar to your LED use case. Nonetheless
> in case of LED controllers it is also possible that hardware
> adjusts LED brightness in case of low battery voltage.
>
> If a device is able e.g. to generate an interrupt to notify this
> kind of event, then we would like also to be able to notify the client
> about that. It wouldn't be user generated brightness change though.
>
>>>> "Reading this file will return the actual led brightness
>>>> when not blinking and no triggers are active; reading this
>>>> file will return the brightness used when the led is on
>>>> when blinking or triggers are active."
>>>
>>> This is unnecessarily entangled. Blinking means timer trigger
>>> is active.
>>
>> Ok.
>>
>>>> So for e.g. the backlit keyboard case reading this single
>>>> file will return the actual brightness of the backlight,
>>>> since this does not involve blinking or triggers.
>>>>
>>>> Basically the idea is that the user_brightness file
>>>> will have the semantics which IMHO the brightness file
>>>> itself should have had from the beginning, but which
>>>> we can't change now due to ABI reasons.
>>>
>>> And in fact introducing user_brightness file would indeed
>>> fix that shortcoming. However without providing notifications
>>> of hw brightness changes on it.
>>
>> See above, I believe such a file should report any
>> changes in brightness, except those caused by triggers,
>> so it would report hw brightness changes.
>>
>> Anyways if you're not interested in fixing the
>> shortcomings of the current read behavior on the
>> brightness file (I'm fine with that, I can live
>> with the shortcomings) I suggest that we simply go
>> with v2 of my poll() patch.
>
> v2 entails power consumption related issues.
>
> Generally I think that we could add the file you proposed,
> however it would be good to devise a name which will cover
> also the cases when brightness is changed by firmware without
> user interaction.
>
>>>>> Having no difference in this area between the two approaches
>>>>> I'm still in favour of the read-only file for notifying
>>>>> brightness changes procured by hardware.
>>>>
>>>> That brings back the needing 2 fds problem; and does
>>>> not solve userspace not being able to reliably read
>>>> the led on brightness when blinking or using triggers.
>>>>
>>>> And this also has the issue that one is doing poll() on
>>>> one fd to detect changes on another fd,
>>>
>>> It is not necessarily true. We can treat the polling on
>>> hw_brightness_change file as a means to detect brightness
>>> changes procured by hardware and we can read that brightness
>>> by executing read on this same fd. It could return -ENODATA
>>> if no such an event has occurred so far.
>>
>> That would still require 2 fds as userspace also wants to
>> be able to set the keyboard backlight, but allowing read()
>> on the hw_brightness_change file at least fixes the weirdness
>> where userspace gets woken from poll() without being able to
>> read. So if you insist on going the hw_brightness_change file
>> route, then I can live with that (and upower will simply
>> need to open 2 fds, that is doable).
>>
>> But, BUT, I would greatly prefer to just go for v4 of my
>> patch, which fixes the only real problem we've seen with
>> my patch as original merged without adding a new, somewhat
>> convoluted sysfs attribute.
>
> Hmm, v4 still calls led_notify_brightness_change(led_cdev)
> from both __led_set_brightness() and __led_set_brightness_blocking().

Ugh, I see I accidentally send a v4 twice, instead of
calling the version which dropped those called v5 as
I should have, sorry.

The v4 which I would like to see merged, the one with
those calls dropped, is here:

https://patchwork.kernel.org/patch/9423093/

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

Hi,

On 11/14/2016 01:51 PM, Hans de Goede wrote:
> Hi,
>
> On 14-11-16 10:12, Jacek Anaszewski wrote:
>> Hi,
>>
>> On 11/13/2016 02:52 PM, Hans de Goede wrote:
>>> Hi,
>>>
>>> On 13-11-16 12:44, Jacek Anaszewski wrote:
>>>> Hi,
>>>>
>>>> On 11/12/2016 10:14 PM, Hans de Goede wrote:
>>>
>>> <snip>
>>>
>>>>>>> So I would like to propose creating a new read-write
>>>>>>> user_brightness file.
>>>>>>>
>>>>>>> The write behavior would be 100% identical to the brightness
>>>>>>> file (in code terms it will call the same store function).
>>>>>>>
>>>>>>> The the read behavior otoh will be different: it will shows
>>>>>>> the last brightness as set by the user, this would show the
>>>>>>> read behavior we really want of brightness: show the real
>>>>>>> brightness when not blinking / triggers are active and show
>>>>>>> the brightness used when on when blinking / triggers are active.
>>>>>>>
>>>>>>> We could then add poll support on this new user_brightness
>>>>>>> file, thus avoiding the problem with the extra cpu-load on
>>>>>>> notifications on blinking / triggers.
>>>>>>
>>>>>> I agree that user_brightness allows to solve the issues you raised
>>>>>> about inconsistent write and read brightness' semantics
>>>>>> (which is not that painful IMHO).
>>>>>>
>>>>>> Reporting non-user brightness changes on user_brightness file
>>>>>> doesn't sound reasonable though.
>>>>>
>>>>> The changes I'm interested in are user brightness changes they
>>>>> are just not done through sysfs, but through a hardwired hotkey,
>>>>> they are however very much done by the user.
>>>>
>>>> Ah, so this file name would be misleading especially taking into
>>>> account
>>>> the context in which "user" is used in kernel, which predominantly
>>>> means "userspace", e.g. copy_to_user(), copy_from_user().
>>>>
>>>>>> Also, how would we read the
>>>>>> brightness set by the firmware? We'd have to read brightness
>>>>>> file, so still two files would have to be opened which is
>>>>>> a second drawback of this approach.
>>>>>
>>>>> No, look carefully at the definition of the read behavior
>>>>> I plan to put in the ABI doc:
>>>>
>>>> OK, "user" was what confused me. So in this case changes made
>>>> by the firmware even if in a result of user activity
>>>> (pressing hardware key) obviously cannot be treated similarly
>>>> to the changes made from the userspace context.
>>>
>>> In the end both result on the brightness of the device
>>> changing, so any userspace process interested in monitoring
>>> the brightness will want to know about both type of changes.
>>>
>>>> Unless you're able to give references to the kernel code which
>>>> contradict my judgement.
>>>
>>> AFAIK the audio code will signal volume changes done by
>>> hardwired buttons the same way as audio changes done
>>> by userspace calling into the kernel. This also makes
>>> sense because in the end, what is interesting for a
>>> mixer app, is that the volume changed, and what the
>>> new volume is.
>>
>> OK, so it is indeed similar to your LED use case. Nonetheless
>> in case of LED controllers it is also possible that hardware
>> adjusts LED brightness in case of low battery voltage.
>>
>> If a device is able e.g. to generate an interrupt to notify this
>> kind of event, then we would like also to be able to notify the client
>> about that. It wouldn't be user generated brightness change though.
>>
>>>>> "Reading this file will return the actual led brightness
>>>>> when not blinking and no triggers are active; reading this
>>>>> file will return the brightness used when the led is on
>>>>> when blinking or triggers are active."
>>>>
>>>> This is unnecessarily entangled. Blinking means timer trigger
>>>> is active.
>>>
>>> Ok.
>>>
>>>>> So for e.g. the backlit keyboard case reading this single
>>>>> file will return the actual brightness of the backlight,
>>>>> since this does not involve blinking or triggers.
>>>>>
>>>>> Basically the idea is that the user_brightness file
>>>>> will have the semantics which IMHO the brightness file
>>>>> itself should have had from the beginning, but which
>>>>> we can't change now due to ABI reasons.
>>>>
>>>> And in fact introducing user_brightness file would indeed
>>>> fix that shortcoming. However without providing notifications
>>>> of hw brightness changes on it.
>>>
>>> See above, I believe such a file should report any
>>> changes in brightness, except those caused by triggers,
>>> so it would report hw brightness changes.
>>>
>>> Anyways if you're not interested in fixing the
>>> shortcomings of the current read behavior on the
>>> brightness file (I'm fine with that, I can live
>>> with the shortcomings) I suggest that we simply go
>>> with v2 of my poll() patch.
>>
>> v2 entails power consumption related issues.
>>
>> Generally I think that we could add the file you proposed,
>> however it would be good to devise a name which will cover
>> also the cases when brightness is changed by firmware without
>> user interaction.
>>
>>>>>> Having no difference in this area between the two approaches
>>>>>> I'm still in favour of the read-only file for notifying
>>>>>> brightness changes procured by hardware.
>>>>>
>>>>> That brings back the needing 2 fds problem; and does
>>>>> not solve userspace not being able to reliably read
>>>>> the led on brightness when blinking or using triggers.
>>>>>
>>>>> And this also has the issue that one is doing poll() on
>>>>> one fd to detect changes on another fd,
>>>>
>>>> It is not necessarily true. We can treat the polling on
>>>> hw_brightness_change file as a means to detect brightness
>>>> changes procured by hardware and we can read that brightness
>>>> by executing read on this same fd. It could return -ENODATA
>>>> if no such an event has occurred so far.
>>>
>>> That would still require 2 fds as userspace also wants to
>>> be able to set the keyboard backlight, but allowing read()
>>> on the hw_brightness_change file at least fixes the weirdness
>>> where userspace gets woken from poll() without being able to
>>> read. So if you insist on going the hw_brightness_change file
>>> route, then I can live with that (and upower will simply
>>> need to open 2 fds, that is doable).
>>>
>>> But, BUT, I would greatly prefer to just go for v4 of my
>>> patch, which fixes the only real problem we've seen with
>>> my patch as original merged without adding a new, somewhat
>>> convoluted sysfs attribute.
>>
>> Hmm, v4 still calls led_notify_brightness_change(led_cdev)
>> from both __led_set_brightness() and __led_set_brightness_blocking().
>
> Ugh, I see I accidentally send a v4 twice, instead of
> calling the version which dropped those called v5 as
> I should have, sorry.
>
> The v4 which I would like to see merged, the one with
> those calls dropped, is here:
>
> https://patchwork.kernel.org/patch/9423093/

Right I've had an impression that I've already seen something
different than "first" v4.

Regarding the patch - adding led_notify_brightness_change() to
brightness_store() can have similar power consumption related
implications if brightness is set frequently via sysfs. I'm leaning
towards adding a new brightness file similar to user_brightness
discussed in this thread.

It would cover shortcomings and read/write inconsistencies that
brightness file currently has, but without breaking existing users.

I'd not however go for "user_brightness" name due to the possible
brightness adjustments made autonomously by firmware. I'm afraid
that devising a meaningful name for the new file will be hard,
so the simplest would be just brighntess2. Dedicated section
in leds-class.txt should be devoted to it.

-- 
Best regards,
Jacek Anaszewski

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 15-11-16 11:01, Jacek Anaszewski wrote:
> Hi,
>
> On 11/14/2016 01:51 PM, Hans de Goede wrote:

<snip>

>> Ugh, I see I accidentally send a v4 twice, instead of
>> calling the version which dropped those called v5 as
>> I should have, sorry.
>>
>> The v4 which I would like to see merged, the one with
>> those calls dropped, is here:
>>
>> https://patchwork.kernel.org/patch/9423093/
>
> Right I've had an impression that I've already seen something
> different than "first" v4.
>
> Regarding the patch - adding led_notify_brightness_change() to
> brightness_store() can have similar power consumption related
> implications if brightness is set frequently via sysfs.

That means that userspace is waking up frequently to write the
sysfs file, so in that case userspace is already draining
a lot of energy, so I don't think that is something we need to
worry about.

> I'm leaning
> towards adding a new brightness file similar to user_brightness
> discussed in this thread.
>
> It would cover shortcomings and read/write inconsistencies that
> brightness file currently has, but without breaking existing users.
>
> I'd not however go for "user_brightness" name due to the possible
> brightness adjustments made autonomously by firmware. I'm afraid
> that devising a meaningful name for the new file will be hard,
> so the simplest would be just brighntess2. Dedicated section
> in leds-class.txt should be devoted to it.

Ok, let me quote myself from another part of this thread:

We've 2 sorts of brightness really:

1) transient brightness, aka current brightness, when blinking or
triggers are used this will switch many times a second
between off and some on level.

2) non-transient brightness, for non blinking leds this is the
actual brightness, for blinking leds this is the brightness
level used when the led is on.

Now we want to have a sysfs attribute reflecting 2, so that
userspace can poll on that, both for my use-case as well as so
that userspace process a can detect changes made by writing to
the brightness file by process b.

So maybe we need to simply call the new attribute
non_transient_brightness instead of user_brightness?

non_transient_brightness certainly seems like a better name
then brightness2 ?

Regards,

Hans

LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 994 bytes --]

Hi!

> >Hmm, v4 still calls led_notify_brightness_change(led_cdev)
> >from both __led_set_brightness() and __led_set_brightness_blocking().
> 
> Ugh, I see I accidentally send a v4 twice, instead of
> calling the version which dropped those called v5 as
> I should have, sorry.
> 
> The v4 which I would like to see merged, the one with
> those calls dropped, is here:
> 
> https://patchwork.kernel.org/patch/9423093/

Please, lets fix this properly.

The LED you are talking about _has_ a trigger, implemented in
hardware. That trigger can change LED brightness behind kernel's (and
userspace's) back. Don't pretend the trigger does not exist, it does.

And when you do that, you'll have nice place to report changes to
userspace -- trigger can now export that information, and offer poll()
interface.

Thanks,
									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

On 11/15/2016 11:31 AM, Pavel Machek wrote:
> Hi!
>
>>> Hmm, v4 still calls led_notify_brightness_change(led_cdev)
>> >from both __led_set_brightness() and __led_set_brightness_blocking().
>>
>> Ugh, I see I accidentally send a v4 twice, instead of
>> calling the version which dropped those called v5 as
>> I should have, sorry.
>>
>> The v4 which I would like to see merged, the one with
>> those calls dropped, is here:
>>
>> https://patchwork.kernel.org/patch/9423093/
>
> Please, lets fix this properly.
>
> The LED you are talking about _has_ a trigger, implemented in
> hardware. That trigger can change LED brightness behind kernel's (and
> userspace's) back. Don't pretend the trigger does not exist, it does.
>
> And when you do that, you'll have nice place to report changes to
> userspace -- trigger can now export that information, and offer poll()
> interface.

Well, that sounds interesting. It is logically justifiable.
I initially proposed exactly this solution, with recently
added userspace LED being a trigger listener. It seems a bit
awkward though. How would you listen to the trigger events?

-- 
Best regards,
Jacek Anaszewski

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 1708 bytes --]

On Tue 2016-11-15 11:58:06, Jacek Anaszewski wrote:
> On 11/15/2016 11:31 AM, Pavel Machek wrote:
> >Hi!
> >
> >>>Hmm, v4 still calls led_notify_brightness_change(led_cdev)
> >>>from both __led_set_brightness() and __led_set_brightness_blocking().
> >>
> >>Ugh, I see I accidentally send a v4 twice, instead of
> >>calling the version which dropped those called v5 as
> >>I should have, sorry.
> >>
> >>The v4 which I would like to see merged, the one with
> >>those calls dropped, is here:
> >>
> >>https://patchwork.kernel.org/patch/9423093/
> >
> >Please, lets fix this properly.
> >
> >The LED you are talking about _has_ a trigger, implemented in
> >hardware. That trigger can change LED brightness behind kernel's (and
> >userspace's) back. Don't pretend the trigger does not exist, it does.
> >
> >And when you do that, you'll have nice place to report changes to
> >userspace -- trigger can now export that information, and offer poll()
> >interface.
> 
> Well, that sounds interesting. It is logically justifiable.

Thanks.

> I initially proposed exactly this solution, with recently
> added userspace LED being a trigger listener. It seems a bit
> awkward though. How would you listen to the trigger events?

Trigger exposes a file in sysfs, with poll() working on that file (and
probably read exposing the current brightness).

Key difference is that only triggers where this makes sense (keyboard
backlight) expose it and carry the overhead. CPU trigger would
definitely not do this.

Best regards,
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 15-11-16 12:11, Pavel Machek wrote:
> On Tue 2016-11-15 11:58:06, Jacek Anaszewski wrote:
>> On 11/15/2016 11:31 AM, Pavel Machek wrote:
>>> Hi!
>>>
>>>>> Hmm, v4 still calls led_notify_brightness_change(led_cdev)
>>>> >from both __led_set_brightness() and __led_set_brightness_blocking().
>>>>
>>>> Ugh, I see I accidentally send a v4 twice, instead of
>>>> calling the version which dropped those called v5 as
>>>> I should have, sorry.
>>>>
>>>> The v4 which I would like to see merged, the one with
>>>> those calls dropped, is here:
>>>>
>>>> https://patchwork.kernel.org/patch/9423093/
>>>
>>> Please, lets fix this properly.
>>>
>>> The LED you are talking about _has_ a trigger, implemented in
>>> hardware. That trigger can change LED brightness behind kernel's (and
>>> userspace's) back. Don't pretend the trigger does not exist, it does.
>>>
>>> And when you do that, you'll have nice place to report changes to
>>> userspace -- trigger can now export that information, and offer poll()
>>> interface.
>>
>> Well, that sounds interesting. It is logically justifiable.
>
> Thanks.
>
>> I initially proposed exactly this solution, with recently
>> added userspace LED being a trigger listener. It seems a bit
>> awkward though. How would you listen to the trigger events?
>
> Trigger exposes a file in sysfs, with poll() working on that file

Hmm, a new file would give the advantage of making it easy for
userspace to see if the trigger is poll-able, this is likely
better then my own proposal I just send.

> (and
> probably read exposing the current brightness).

If we do this, can we please make it mirror brightness, iow
also make it writable, that will make it easier for userspace
to deal with it. We can simply re-use the existing show / store
methods for brightness for this.

I suggest we call it:

trigger_brightness

And only register it when a poll-able trigger is present.


> Key difference is that only triggers where this makes sense (keyboard
> backlight) expose it and carry the overhead. CPU trigger would
> definitely not do this.

Ack only having some triggers pollable is important.

Regards,

Hans

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 2081 bytes --]

Hi!

> >>>The LED you are talking about _has_ a trigger, implemented in
> >>>hardware. That trigger can change LED brightness behind kernel's (and
> >>>userspace's) back. Don't pretend the trigger does not exist, it does.
> >>>
> >>>And when you do that, you'll have nice place to report changes to
> >>>userspace -- trigger can now export that information, and offer poll()
> >>>interface.
> >>
> >>Well, that sounds interesting. It is logically justifiable.
> >
> >Thanks.
> >
> >>I initially proposed exactly this solution, with recently
> >>added userspace LED being a trigger listener. It seems a bit
> >>awkward though. How would you listen to the trigger events?
> >
> >Trigger exposes a file in sysfs, with poll() working on that file
> 
> Hmm, a new file would give the advantage of making it easy for
> userspace to see if the trigger is poll-able, this is likely
> better then my own proposal I just send.

Good.

> >(and
> >probably read exposing the current brightness).
> 
> If we do this, can we please make it mirror brightness, iow
> also make it writable, that will make it easier for userspace
> to deal with it. We can simply re-use the existing show / store
> methods for brightness for this.

Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
that here, you want to be able to turn off the backlight but still
keep the trigger (and be notified of future changes).

> I suggest we call it:
> 
> trigger_brightness
> 
> And only register it when a poll-able trigger is present.

I'd call it 'current_brightness', but that's no big deal. Yes, only
registering it for poll-able triggers makes sense.

> >Key difference is that only triggers where this makes sense (keyboard
> >backlight) expose it and carry the overhead. CPU trigger would
> >definitely not do this.
> 
> Ack only having some triggers pollable is important.

Thanks,
									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 15-11-16 12:48, Pavel Machek wrote:
> Hi!
>
>>>>> The LED you are talking about _has_ a trigger, implemented in
>>>>> hardware. That trigger can change LED brightness behind kernel's (and
>>>>> userspace's) back. Don't pretend the trigger does not exist, it does.
>>>>>
>>>>> And when you do that, you'll have nice place to report changes to
>>>>> userspace -- trigger can now export that information, and offer poll()
>>>>> interface.
>>>>
>>>> Well, that sounds interesting. It is logically justifiable.
>>>
>>> Thanks.
>>>
>>>> I initially proposed exactly this solution, with recently
>>>> added userspace LED being a trigger listener. It seems a bit
>>>> awkward though. How would you listen to the trigger events?
>>>
>>> Trigger exposes a file in sysfs, with poll() working on that file
>>
>> Hmm, a new file would give the advantage of making it easy for
>> userspace to see if the trigger is poll-able, this is likely
>> better then my own proposal I just send.
>
> Good.
>
>>> (and
>>> probably read exposing the current brightness).
>>
>> If we do this, can we please make it mirror brightness, iow
>> also make it writable, that will make it easier for userspace
>> to deal with it. We can simply re-use the existing show / store
>> methods for brightness for this.
>
> Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
> that here, you want to be able to turn off the backlight but still
> keep the trigger (and be notified of future changes).

True, that is easy to do the store method will just need to call
led_set_brightness_nosleep instead of led_set_brightness, this
will skip the checks to stop blinking in led_set_brightness and
otherwise is equivalent.

>> I suggest we call it:
>>
>> trigger_brightness
>>
>> And only register it when a poll-able trigger is present.
>
> I'd call it 'current_brightness', but that's no big deal. Yes, only
> registering it for poll-able triggers makes sense.

current_brightness works for me. I will take a shot a patch-set
implementing this.

Regards,

Hans



>
>>> Key difference is that only triggers where this makes sense (keyboard
>>> backlight) expose it and carry the overhead. CPU trigger would
>>> definitely not do this.
>>
>> Ack only having some triggers pollable is important.
>
> Thanks,
> 									Pavel
>

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 2390 bytes --]

On Tue 2016-11-15 13:06:14, Hans de Goede wrote:
> Hi,
> 
> On 15-11-16 12:48, Pavel Machek wrote:
> >Hi!
> >
> >>>>>The LED you are talking about _has_ a trigger, implemented in
> >>>>>hardware. That trigger can change LED brightness behind kernel's (and
> >>>>>userspace's) back. Don't pretend the trigger does not exist, it does.
> >>>>>
> >>>>>And when you do that, you'll have nice place to report changes to
> >>>>>userspace -- trigger can now export that information, and offer poll()
> >>>>>interface.
> >>>>
> >>>>Well, that sounds interesting. It is logically justifiable.
> >>>
> >>>Thanks.
> >>>
> >>>>I initially proposed exactly this solution, with recently
> >>>>added userspace LED being a trigger listener. It seems a bit
> >>>>awkward though. How would you listen to the trigger events?
> >>>
> >>>Trigger exposes a file in sysfs, with poll() working on that file
> >>
> >>Hmm, a new file would give the advantage of making it easy for
> >>userspace to see if the trigger is poll-able, this is likely
> >>better then my own proposal I just send.
> >
> >Good.
> >
> >>>(and
> >>>probably read exposing the current brightness).
> >>
> >>If we do this, can we please make it mirror brightness, iow
> >>also make it writable, that will make it easier for userspace
> >>to deal with it. We can simply re-use the existing show / store
> >>methods for brightness for this.
> >
> >Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
> >that here, you want to be able to turn off the backlight but still
> >keep the trigger (and be notified of future changes).
> 
> True, that is easy to do the store method will just need to call
> led_set_brightness_nosleep instead of led_set_brightness, this
> will skip the checks to stop blinking in led_set_brightness and
> otherwise is equivalent.
> 
> >>I suggest we call it:
> >>
> >>trigger_brightness
> >>
> >>And only register it when a poll-able trigger is present.
> >
> >I'd call it 'current_brightness', but that's no big deal. Yes, only
> >registering it for poll-able triggers makes sense.
> 
> current_brightness works for me. I will take a shot a patch-set
> implementing this.

Thanks!
								Pavel
								
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

On 11/15/2016 01:06 PM, Hans de Goede wrote:
> Hi,
>
> On 15-11-16 12:48, Pavel Machek wrote:
>> Hi!
>>
>>>>>> The LED you are talking about _has_ a trigger, implemented in
>>>>>> hardware. That trigger can change LED brightness behind kernel's (and
>>>>>> userspace's) back. Don't pretend the trigger does not exist, it does.
>>>>>>
>>>>>> And when you do that, you'll have nice place to report changes to
>>>>>> userspace -- trigger can now export that information, and offer
>>>>>> poll()
>>>>>> interface.
>>>>>
>>>>> Well, that sounds interesting. It is logically justifiable.
>>>>
>>>> Thanks.
>>>>
>>>>> I initially proposed exactly this solution, with recently
>>>>> added userspace LED being a trigger listener. It seems a bit
>>>>> awkward though. How would you listen to the trigger events?
>>>>
>>>> Trigger exposes a file in sysfs, with poll() working on that file
>>>
>>> Hmm, a new file would give the advantage of making it easy for
>>> userspace to see if the trigger is poll-able, this is likely
>>> better then my own proposal I just send.
>>
>> Good.
>>
>>>> (and
>>>> probably read exposing the current brightness).
>>>
>>> If we do this, can we please make it mirror brightness, iow
>>> also make it writable, that will make it easier for userspace
>>> to deal with it. We can simply re-use the existing show / store
>>> methods for brightness for this.
>>
>> Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
>> that here, you want to be able to turn off the backlight but still
>> keep the trigger (and be notified of future changes).
>
> True, that is easy to do the store method will just need to call
> led_set_brightness_nosleep instead of led_set_brightness, this
> will skip the checks to stop blinking in led_set_brightness and
> otherwise is equivalent.
>
>>> I suggest we call it:
>>>
>>> trigger_brightness
>>>
>>> And only register it when a poll-able trigger is present.
>>
>> I'd call it 'current_brightness', but that's no big deal. Yes, only
>> registering it for poll-able triggers makes sense.
>
> current_brightness works for me. I will take a shot a patch-set
> implementing this.

Word "current" is not precise here.

It can be thought of as either last brightness set by the
user or the brightness currently written to the device
(returned by brightness file).

There is a semantic discrepancy in our requirements -
we want the file representing both permanent brightness
set by the user and brightness set by the hardware.

The two stand in contradiction to each other since
brightness set by the user can be adjusted by the hardware.

Reading the file shouldn't update brightness property of
struct led_classdev, so it shouldn't call led_update_brightness()
but it still should allow reading brightness set by the
hardware, as a result of each POLLPRI event. So in fact in
the same time it should report both according to our requirements
which is impossible. Do we need three brightness files ?

-- 
Best regards,
Jacek Anaszewski

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 15-11-16 14:28, Jacek Anaszewski wrote:
> On 11/15/2016 01:06 PM, Hans de Goede wrote:
>> Hi,
>>
>> On 15-11-16 12:48, Pavel Machek wrote:
>>> Hi!
>>>
>>>>>>> The LED you are talking about _has_ a trigger, implemented in
>>>>>>> hardware. That trigger can change LED brightness behind kernel's (and
>>>>>>> userspace's) back. Don't pretend the trigger does not exist, it does.
>>>>>>>
>>>>>>> And when you do that, you'll have nice place to report changes to
>>>>>>> userspace -- trigger can now export that information, and offer
>>>>>>> poll()
>>>>>>> interface.
>>>>>>
>>>>>> Well, that sounds interesting. It is logically justifiable.
>>>>>
>>>>> Thanks.
>>>>>
>>>>>> I initially proposed exactly this solution, with recently
>>>>>> added userspace LED being a trigger listener. It seems a bit
>>>>>> awkward though. How would you listen to the trigger events?
>>>>>
>>>>> Trigger exposes a file in sysfs, with poll() working on that file
>>>>
>>>> Hmm, a new file would give the advantage of making it easy for
>>>> userspace to see if the trigger is poll-able, this is likely
>>>> better then my own proposal I just send.
>>>
>>> Good.
>>>
>>>>> (and
>>>>> probably read exposing the current brightness).
>>>>
>>>> If we do this, can we please make it mirror brightness, iow
>>>> also make it writable, that will make it easier for userspace
>>>> to deal with it. We can simply re-use the existing show / store
>>>> methods for brightness for this.
>>>
>>> Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
>>> that here, you want to be able to turn off the backlight but still
>>> keep the trigger (and be notified of future changes).
>>
>> True, that is easy to do the store method will just need to call
>> led_set_brightness_nosleep instead of led_set_brightness, this
>> will skip the checks to stop blinking in led_set_brightness and
>> otherwise is equivalent.
>>
>>>> I suggest we call it:
>>>>
>>>> trigger_brightness
>>>>
>>>> And only register it when a poll-able trigger is present.
>>>
>>> I'd call it 'current_brightness', but that's no big deal. Yes, only
>>> registering it for poll-able triggers makes sense.
>>
>> current_brightness works for me. I will take a shot a patch-set
>> implementing this.
>
> Word "current" is not precise here.
>
> It can be thought of as either last brightness set by the
> user or the brightness currently written to the device
> (returned by brightness file).
>
> There is a semantic discrepancy in our requirements -
> we want the file representing both permanent brightness
> set by the user and brightness set by the hardware.
>
> The two stand in contradiction to each other since
> brightness set by the user can be adjusted by the hardware.
>
> Reading the file shouldn't update brightness property of
> struct led_classdev, so it shouldn't call led_update_brightness()
> but it still should allow reading brightness set by the
> hardware, as a result of each POLLPRI event. So in fact in
> the same time it should report both according to our requirements
> which is impossible. Do we need three brightness files ?

I don't think so, current_brightness actually is an accurate
name, if the brightness was last changed by writing from
sysfs, the keyboard backlight will honor that and the current_brightness
attribute will show the brightness last set through writing it,
which matches the actual current brightness of the keyboard backlight.

Likewise if it was changed with the hotkey last then the keyboard
backlight brightness will be changed and reading from current_brightness
will return the new actual brightness. Basically reading from this
file will be no different then reading from the normal "brightness"
file the difference will be in that it is poll-able and that
writing 0 turns off the LED without stopping blinking.

Regards,

Hans

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

On 11/15/2016 02:48 PM, Hans de Goede wrote:
> Hi,
>
> On 15-11-16 14:28, Jacek Anaszewski wrote:
>> On 11/15/2016 01:06 PM, Hans de Goede wrote:
>>> Hi,
>>>
>>> On 15-11-16 12:48, Pavel Machek wrote:
>>>> Hi!
>>>>
>>>>>>>> The LED you are talking about _has_ a trigger, implemented in
>>>>>>>> hardware. That trigger can change LED brightness behind kernel's
>>>>>>>> (and
>>>>>>>> userspace's) back. Don't pretend the trigger does not exist, it
>>>>>>>> does.
>>>>>>>>
>>>>>>>> And when you do that, you'll have nice place to report changes to
>>>>>>>> userspace -- trigger can now export that information, and offer
>>>>>>>> poll()
>>>>>>>> interface.
>>>>>>>
>>>>>>> Well, that sounds interesting. It is logically justifiable.
>>>>>>
>>>>>> Thanks.
>>>>>>
>>>>>>> I initially proposed exactly this solution, with recently
>>>>>>> added userspace LED being a trigger listener. It seems a bit
>>>>>>> awkward though. How would you listen to the trigger events?
>>>>>>
>>>>>> Trigger exposes a file in sysfs, with poll() working on that file
>>>>>
>>>>> Hmm, a new file would give the advantage of making it easy for
>>>>> userspace to see if the trigger is poll-able, this is likely
>>>>> better then my own proposal I just send.
>>>>
>>>> Good.
>>>>
>>>>>> (and
>>>>>> probably read exposing the current brightness).
>>>>>
>>>>> If we do this, can we please make it mirror brightness, iow
>>>>> also make it writable, that will make it easier for userspace
>>>>> to deal with it. We can simply re-use the existing show / store
>>>>> methods for brightness for this.
>>>>
>>>> Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
>>>> that here, you want to be able to turn off the backlight but still
>>>> keep the trigger (and be notified of future changes).
>>>
>>> True, that is easy to do the store method will just need to call
>>> led_set_brightness_nosleep instead of led_set_brightness, this
>>> will skip the checks to stop blinking in led_set_brightness and
>>> otherwise is equivalent.
>>>
>>>>> I suggest we call it:
>>>>>
>>>>> trigger_brightness
>>>>>
>>>>> And only register it when a poll-able trigger is present.
>>>>
>>>> I'd call it 'current_brightness', but that's no big deal. Yes, only
>>>> registering it for poll-able triggers makes sense.
>>>
>>> current_brightness works for me. I will take a shot a patch-set
>>> implementing this.
>>
>> Word "current" is not precise here.
>>
>> It can be thought of as either last brightness set by the
>> user or the brightness currently written to the device
>> (returned by brightness file).
>>
>> There is a semantic discrepancy in our requirements -
>> we want the file representing both permanent brightness
>> set by the user and brightness set by the hardware.
>>
>> The two stand in contradiction to each other since
>> brightness set by the user can be adjusted by the hardware.
>>
>> Reading the file shouldn't update brightness property of
>> struct led_classdev, so it shouldn't call led_update_brightness()
>> but it still should allow reading brightness set by the
>> hardware, as a result of each POLLPRI event. So in fact in
>> the same time it should report both according to our requirements
>> which is impossible. Do we need three brightness files ?
>
> I don't think so, current_brightness actually is an accurate
> name, if the brightness was last changed by writing from
> sysfs, the keyboard backlight will honor that and the current_brightness
> attribute will show the brightness last set through writing it,
> which matches the actual current brightness of the keyboard backlight.
>
> Likewise if it was changed with the hotkey last then the keyboard
> backlight brightness will be changed and reading from current_brightness
> will return the new actual brightness. Basically reading from this
> file will be no different then reading from the normal "brightness"
> file the difference will be in that it is poll-able and that
> writing 0 turns off the LED without stopping blinking.

If so then when software blinking enabled, it will return 0 on low
blink cycle no matter what current brightness level is.

-- 
Best regards,
Jacek Anaszewski

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 15-11-16 15:04, Jacek Anaszewski wrote:
> On 11/15/2016 02:48 PM, Hans de Goede wrote:
>> Hi,
>>
>> On 15-11-16 14:28, Jacek Anaszewski wrote:
>>> On 11/15/2016 01:06 PM, Hans de Goede wrote:
>>>> Hi,
>>>>
>>>> On 15-11-16 12:48, Pavel Machek wrote:
>>>>> Hi!
>>>>>
>>>>>>>>> The LED you are talking about _has_ a trigger, implemented in
>>>>>>>>> hardware. That trigger can change LED brightness behind kernel's
>>>>>>>>> (and
>>>>>>>>> userspace's) back. Don't pretend the trigger does not exist, it
>>>>>>>>> does.
>>>>>>>>>
>>>>>>>>> And when you do that, you'll have nice place to report changes to
>>>>>>>>> userspace -- trigger can now export that information, and offer
>>>>>>>>> poll()
>>>>>>>>> interface.
>>>>>>>>
>>>>>>>> Well, that sounds interesting. It is logically justifiable.
>>>>>>>
>>>>>>> Thanks.
>>>>>>>
>>>>>>>> I initially proposed exactly this solution, with recently
>>>>>>>> added userspace LED being a trigger listener. It seems a bit
>>>>>>>> awkward though. How would you listen to the trigger events?
>>>>>>>
>>>>>>> Trigger exposes a file in sysfs, with poll() working on that file
>>>>>>
>>>>>> Hmm, a new file would give the advantage of making it easy for
>>>>>> userspace to see if the trigger is poll-able, this is likely
>>>>>> better then my own proposal I just send.
>>>>>
>>>>> Good.
>>>>>
>>>>>>> (and
>>>>>>> probably read exposing the current brightness).
>>>>>>
>>>>>> If we do this, can we please make it mirror brightness, iow
>>>>>> also make it writable, that will make it easier for userspace
>>>>>> to deal with it. We can simply re-use the existing show / store
>>>>>> methods for brightness for this.
>>>>>
>>>>> Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
>>>>> that here, you want to be able to turn off the backlight but still
>>>>> keep the trigger (and be notified of future changes).
>>>>
>>>> True, that is easy to do the store method will just need to call
>>>> led_set_brightness_nosleep instead of led_set_brightness, this
>>>> will skip the checks to stop blinking in led_set_brightness and
>>>> otherwise is equivalent.
>>>>
>>>>>> I suggest we call it:
>>>>>>
>>>>>> trigger_brightness
>>>>>>
>>>>>> And only register it when a poll-able trigger is present.
>>>>>
>>>>> I'd call it 'current_brightness', but that's no big deal. Yes, only
>>>>> registering it for poll-able triggers makes sense.
>>>>
>>>> current_brightness works for me. I will take a shot a patch-set
>>>> implementing this.
>>>
>>> Word "current" is not precise here.
>>>
>>> It can be thought of as either last brightness set by the
>>> user or the brightness currently written to the device
>>> (returned by brightness file).
>>>
>>> There is a semantic discrepancy in our requirements -
>>> we want the file representing both permanent brightness
>>> set by the user and brightness set by the hardware.
>>>
>>> The two stand in contradiction to each other since
>>> brightness set by the user can be adjusted by the hardware.
>>>
>>> Reading the file shouldn't update brightness property of
>>> struct led_classdev, so it shouldn't call led_update_brightness()
>>> but it still should allow reading brightness set by the
>>> hardware, as a result of each POLLPRI event. So in fact in
>>> the same time it should report both according to our requirements
>>> which is impossible. Do we need three brightness files ?
>>
>> I don't think so, current_brightness actually is an accurate
>> name, if the brightness was last changed by writing from
>> sysfs, the keyboard backlight will honor that and the current_brightness
>> attribute will show the brightness last set through writing it,
>> which matches the actual current brightness of the keyboard backlight.
>>
>> Likewise if it was changed with the hotkey last then the keyboard
>> backlight brightness will be changed and reading from current_brightness
>> will return the new actual brightness. Basically reading from this
>> file will be no different then reading from the normal "brightness"
>> file the difference will be in that it is poll-able and that
>> writing 0 turns off the LED without stopping blinking.
>
> If so then when software blinking enabled, it will return 0 on low
> blink cycle no matter what current brightness level is.

For software blinking there will not be a current_brightness atrribute,
as we will only add that for LEDs with poll-able triggers.

Also during the low cycle the LED is off, so its brightness at the
moment of reading (iow currently) is 0.

Regards,

Hans

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Jacek Anaszewski]

On 11/15/2016 03:30 PM, Hans de Goede wrote:
> Hi,
>
> On 15-11-16 15:04, Jacek Anaszewski wrote:
>> On 11/15/2016 02:48 PM, Hans de Goede wrote:
>>> Hi,
>>>
>>> On 15-11-16 14:28, Jacek Anaszewski wrote:
>>>> On 11/15/2016 01:06 PM, Hans de Goede wrote:
>>>>> Hi,
>>>>>
>>>>> On 15-11-16 12:48, Pavel Machek wrote:
>>>>>> Hi!
>>>>>>
>>>>>>>>>> The LED you are talking about _has_ a trigger, implemented in
>>>>>>>>>> hardware. That trigger can change LED brightness behind kernel's
>>>>>>>>>> (and
>>>>>>>>>> userspace's) back. Don't pretend the trigger does not exist, it
>>>>>>>>>> does.
>>>>>>>>>>
>>>>>>>>>> And when you do that, you'll have nice place to report changes to
>>>>>>>>>> userspace -- trigger can now export that information, and offer
>>>>>>>>>> poll()
>>>>>>>>>> interface.
>>>>>>>>>
>>>>>>>>> Well, that sounds interesting. It is logically justifiable.
>>>>>>>>
>>>>>>>> Thanks.
>>>>>>>>
>>>>>>>>> I initially proposed exactly this solution, with recently
>>>>>>>>> added userspace LED being a trigger listener. It seems a bit
>>>>>>>>> awkward though. How would you listen to the trigger events?
>>>>>>>>
>>>>>>>> Trigger exposes a file in sysfs, with poll() working on that file
>>>>>>>
>>>>>>> Hmm, a new file would give the advantage of making it easy for
>>>>>>> userspace to see if the trigger is poll-able, this is likely
>>>>>>> better then my own proposal I just send.
>>>>>>
>>>>>> Good.
>>>>>>
>>>>>>>> (and
>>>>>>>> probably read exposing the current brightness).
>>>>>>>
>>>>>>> If we do this, can we please make it mirror brightness, iow
>>>>>>> also make it writable, that will make it easier for userspace
>>>>>>> to deal with it. We can simply re-use the existing show / store
>>>>>>> methods for brightness for this.
>>>>>>
>>>>>> Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
>>>>>> that here, you want to be able to turn off the backlight but still
>>>>>> keep the trigger (and be notified of future changes).
>>>>>
>>>>> True, that is easy to do the store method will just need to call
>>>>> led_set_brightness_nosleep instead of led_set_brightness, this
>>>>> will skip the checks to stop blinking in led_set_brightness and
>>>>> otherwise is equivalent.
>>>>>
>>>>>>> I suggest we call it:
>>>>>>>
>>>>>>> trigger_brightness
>>>>>>>
>>>>>>> And only register it when a poll-able trigger is present.
>>>>>>
>>>>>> I'd call it 'current_brightness', but that's no big deal. Yes, only
>>>>>> registering it for poll-able triggers makes sense.
>>>>>
>>>>> current_brightness works for me. I will take a shot a patch-set
>>>>> implementing this.
>>>>
>>>> Word "current" is not precise here.
>>>>
>>>> It can be thought of as either last brightness set by the
>>>> user or the brightness currently written to the device
>>>> (returned by brightness file).
>>>>
>>>> There is a semantic discrepancy in our requirements -
>>>> we want the file representing both permanent brightness
>>>> set by the user and brightness set by the hardware.
>>>>
>>>> The two stand in contradiction to each other since
>>>> brightness set by the user can be adjusted by the hardware.
>>>>
>>>> Reading the file shouldn't update brightness property of
>>>> struct led_classdev, so it shouldn't call led_update_brightness()
>>>> but it still should allow reading brightness set by the
>>>> hardware, as a result of each POLLPRI event. So in fact in
>>>> the same time it should report both according to our requirements
>>>> which is impossible. Do we need three brightness files ?
>>>
>>> I don't think so, current_brightness actually is an accurate
>>> name, if the brightness was last changed by writing from
>>> sysfs, the keyboard backlight will honor that and the current_brightness
>>> attribute will show the brightness last set through writing it,
>>> which matches the actual current brightness of the keyboard backlight.
>>>
>>> Likewise if it was changed with the hotkey last then the keyboard
>>> backlight brightness will be changed and reading from current_brightness
>>> will return the new actual brightness. Basically reading from this
>>> file will be no different then reading from the normal "brightness"
>>> file the difference will be in that it is poll-able and that
>>> writing 0 turns off the LED without stopping blinking.
>>
>> If so then when software blinking enabled, it will return 0 on low
>> blink cycle no matter what current brightness level is.
>
> For software blinking there will not be a current_brightness atrribute,
> as we will only add that for LEDs with poll-able triggers.
>
> Also during the low cycle the LED is off, so its brightness at the
> moment of reading (iow currently) is 0.

OK, I wanted to make sure that we know what we've agreed on.
Earlier there were doubts raised about brightness during
software blinking being periodically reported 0, but actually
your reasoning seems to be the best answer to that.

-- 
Best regards,
Jacek Anaszewski

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 15-11-16 13:06, Hans de Goede wrote:
> Hi,
>
> On 15-11-16 12:48, Pavel Machek wrote:
>> Hi!
>>
>>>>>> The LED you are talking about _has_ a trigger, implemented in
>>>>>> hardware. That trigger can change LED brightness behind kernel's (and
>>>>>> userspace's) back. Don't pretend the trigger does not exist, it does.
>>>>>>
>>>>>> And when you do that, you'll have nice place to report changes to
>>>>>> userspace -- trigger can now export that information, and offer poll()
>>>>>> interface.
>>>>>
>>>>> Well, that sounds interesting. It is logically justifiable.
>>>>
>>>> Thanks.
>>>>
>>>>> I initially proposed exactly this solution, with recently
>>>>> added userspace LED being a trigger listener. It seems a bit
>>>>> awkward though. How would you listen to the trigger events?
>>>>
>>>> Trigger exposes a file in sysfs, with poll() working on that file
>>>
>>> Hmm, a new file would give the advantage of making it easy for
>>> userspace to see if the trigger is poll-able, this is likely
>>> better then my own proposal I just send.
>>
>> Good.
>>
>>>> (and
>>>> probably read exposing the current brightness).
>>>
>>> If we do this, can we please make it mirror brightness, iow
>>> also make it writable, that will make it easier for userspace
>>> to deal with it. We can simply re-use the existing show / store
>>> methods for brightness for this.
>>
>> Actually, echo 0 > brightness disables the trigger, IIRC. I'd avoid
>> that here, you want to be able to turn off the backlight but still
>> keep the trigger (and be notified of future changes).
>
> True, that is easy to do the store method will just need to call
> led_set_brightness_nosleep instead of led_set_brightness, this
> will skip the checks to stop blinking in led_set_brightness and
> otherwise is equivalent.
>
>>> I suggest we call it:
>>>
>>> trigger_brightness
>>>
>>> And only register it when a poll-able trigger is present.
>>
>> I'd call it 'current_brightness', but that's no big deal. Yes, only
>> registering it for poll-able triggers makes sense.
>
> current_brightness works for me. I will take a shot a patch-set
> implementing this.

Done, this actually turned out pretty nice, the trigger also helps
in propagating the change events from dell-wmi to the led-classdev
in dell-laptop without needing the ugly hacks I needed before.

v5 coming up.

Regards,

Hans



>
>
>
>>
>>>> Key difference is that only triggers where this makes sense (keyboard
>>>> backlight) expose it and carry the overhead. CPU trigger would
>>>> definitely not do this.
>>>
>>> Ack only having some triggers pollable is important.
>>
>> Thanks,
>>                                     Pavel
>>

Re: LEDs that change brightness "itself" -- that's a trigger. Re: PM regression with LED changes in next-20161109

[Hans de Goede]

HI,

On 15-11-16 11:58, Jacek Anaszewski wrote:
> On 11/15/2016 11:31 AM, Pavel Machek wrote:
>> Hi!
>>
>>>> Hmm, v4 still calls led_notify_brightness_change(led_cdev)
>>> >from both __led_set_brightness() and __led_set_brightness_blocking().
>>>
>>> Ugh, I see I accidentally send a v4 twice, instead of
>>> calling the version which dropped those called v5 as
>>> I should have, sorry.
>>>
>>> The v4 which I would like to see merged, the one with
>>> those calls dropped, is here:
>>>
>>> https://patchwork.kernel.org/patch/9423093/
>>
>> Please, lets fix this properly.
>>
>> The LED you are talking about _has_ a trigger, implemented in
>> hardware. That trigger can change LED brightness behind kernel's (and
>> userspace's) back. Don't pretend the trigger does not exist, it does.
>>
>> And when you do that, you'll have nice place to report changes to
>> userspace -- trigger can now export that information, and offer poll()
>> interface.
>
> Well, that sounds interesting. It is logically justifiable.
> I initially proposed exactly this solution, with recently
> added userspace LED being a trigger listener. It seems a bit
> awkward though. How would you listen to the trigger events?

We could make the trigger sysfs attribute poll()-able, but only
for select triggers, e.g.:

Documentation/ABI/testing/sysfs-class-led

What:           /sys/class/leds/<led>/trigger
Date:           March 2006
KernelVersion:  2.6.17
Contact:        Richard Purdie <rpurdie@rpsys.net>
Description:
                 Set the trigger for this LED. A trigger is a kernel based source
                 of led events.
                 You can change triggers in a similar manner to the way an IO
                 scheduler is chosen. Trigger specific parameters can appear in
                 /sys/class/leds/<led> once a given trigger is selected. For
                 their documentation see sysfs-class-led-trigger-*.
+
+		For some triggers userspace my poll() this file, watching for
+		POLL_PRI to detect when the trigger triggers. This is only
+		supported if this is explicitly mentioned as supported in
+		sysfs-class-led-trigger-* for the selected trigger.

The reason for making this only supported for select triggers is to
avoid getting the whole power-consumption issue from triggers which fire
frequently again.

And then we could add a new:

Documentation/ABI/testing/sysfs-class-led-trigger-kbd-backlight-change

File which documents that the new to be added kbd-backlight-change
trigger is poll-able.

We would also need:

--- a/include/linux/leds.h
+++ b/include/linux/leds.h
@@ -47,6 +47,7 @@ struct led_classdev {
  #define LED_DEV_CAP_FLASH      (1 << 18)
  #define LED_HW_PLUGGABLE       (1 << 19)
  #define LED_PANIC_INDICATOR    (1 << 20)
+#define LED_TRIGGER_READ_ONLY   (1 << 21)

         /* set_brightness_work / blink_timer flags, atomic, private. */
         unsigned long           work_flags;

To allow led drivers to indicate that there trigger is hardwired.

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 1568 bytes --]

Hi!

> >Also, how would we read the
> >brightness set by the firmware? We'd have to read brightness
> >file, so still two files would have to be opened which is
> >a second drawback of this approach.
> 
> No, look carefully at the definition of the read behavior
> I plan to put in the ABI doc:
> 
> "Reading this file will return the actual led brightness
> when not blinking and no triggers are active; reading this
> file will return the brightness used when the led is on
> when blinking or triggers are active."

That's not sane semantics. Userspace would have to read three files
(racy) to find out about triggers etc.

It also prevents modelling
"hardware-changes-brightness-behind-kernel's-back" as a trigger.

> So for e.g. the backlit keyboard case reading this single
> file will return the actual brightness of the backlight,
> since this does not involve blinking or triggers.

Stop obsessing about "ingle file". FDs are pretty cheap. This is sysfs.

> >Having no difference in this area between the two approaches
> >I'm still in favour of the read-only file for notifying
> >brightness changes procured by hardware.
> 
> That brings back the needing 2 fds problem; and does

Actually you have turned "2 fds" into "3 fds", as userspace now needs
to check for trigger _and_ blinking _and_ actuall brightness.

fds are cheap, but that is still not nice design.

								Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 723 bytes --]

Hi!

> >Hmm. So userland can read the LED state, and it can get _some_ value
> >back, but it can not know if it is current state or not.
> >
> >I don't think that's a good interface. I see it is from 2008... is
> >someone using it? Maybe it is too late for revert.
> 
> I can imagine it being used in flash LED use case. E.g. one
> could use oneshot trigger to trigger flash strobe, and then
> he could periodically read brightness file to check, for whatever
> reason, whether the flash is strobing.

I'm pretty sure nobody does that for flah strobe.
									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: PM regression with LED changes in next-20161109

[Hans de Goede]

Hi,

On 09-11-16 20:23, Tony Lindgren wrote:
> Hi,
>
> Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
> the sysfs brightness attr for changes.") breaks runtime PM for me.
>
> On my omap dm3730 based test system, idle power consumption is over 70
> times higher now with this patch! It goes from about 6mW for the core
> system to over 440mW during idle meaning there's some busy timer now
> active.

Do you have any blinking LEDs or LED triggers defined on the system ?

> Reverting this patch fixes the issue. Any ideas?

All I can think of is something calling led_set_brightness quite often,
the patch in question makes led_set_brightness somewhat more expensive,
but it should not cause such a big difference unless something is
really calling led_set_brightness quite often maybe something is calling
it with the same value all the time ?

Regards,

Hans

Re: PM regression with LED changes in next-20161109

[Tony Lindgren]

* Hans de Goede <hdegoede@redhat.com> [161110 01:35]:
> Hi,
> 
> On 09-11-16 20:23, Tony Lindgren wrote:
> > Hi,
> > 
> > Looks like commit 883d32ce3385 ("leds: core: Add support for poll()ing
> > the sysfs brightness attr for changes.") breaks runtime PM for me.
> > 
> > On my omap dm3730 based test system, idle power consumption is over 70
> > times higher now with this patch! It goes from about 6mW for the core
> > system to over 440mW during idle meaning there's some busy timer now
> > active.
> 
> Do you have any blinking LEDs or LED triggers defined on the system ?

There are some configured in the dts file:

$ grep -i led arch/arm/boot/dts/*torpedo*.dts*

And the gpio controlled led1 is configured to blink with
linux,default-trigger = "cpu0".

> > Reverting this patch fixes the issue. Any ideas?
> 
> All I can think of is something calling led_set_brightness quite often,
> the patch in question makes led_set_brightness somewhat more expensive,
> but it should not cause such a big difference unless something is
> really calling led_set_brightness quite often maybe something is calling
> it with the same value all the time ?

I don't think this one has any brightness control.

Regards,

Tony