Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1470125 > unrolled thread

Re: [PATCH V4] leds: trigger: Introduce an USB port trigger

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Greg KH <gregkh@linuxfoundation.org> - 2016-08-25 15:00 +0200
    Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Rafał Miłecki <zajec5@gmail.com> - 2016-08-26 17:40 +0200
      Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Greg KH <gregkh@linuxfoundation.org> - 2016-08-30 14:10 +0200
        Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Rafał Miłecki <zajec5@gmail.com> - 2016-08-30 22:30 +0200
          Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Alan Stern <stern@rowland.harvard.edu> - 2016-08-30 23:00 +0200
            Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Rafał Miłecki <zajec5@gmail.com> - 2016-08-30 23:20 +0200
              Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Alan Stern <stern@rowland.harvard.edu> - 2016-08-31 20:30 +0200
                Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Rafał Miłecki <zajec5@gmail.com> - 2016-08-31 21:20 +0200
                  Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Rafał Miłecki <zajec5@gmail.com> - 2016-09-01 07:30 +0200
                    Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Jacek Anaszewski <j.anaszewski@samsung.com> - 2016-09-01 09:30 +0200
                      Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Alan Stern <stern@rowland.harvard.edu> - 2016-09-01 16:40 +0200
                        Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Jacek Anaszewski <j.anaszewski@samsung.com> - 2016-09-02 09:00 +0200
                          Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Alan Stern <stern@rowland.harvard.edu> - 2016-09-02 16:40 +0200
                            Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Jacek Anaszewski <jacek.anaszewski@gmail.com> - 2016-09-03 17:20 +0200
                              Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Alan Stern <stern@rowland.harvard.edu> - 2016-09-03 17:20 +0200
                                Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Jacek Anaszewski <jacek.anaszewski@gmail.com> - 2016-09-03 21:20 +0200
                                  Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Alan Stern <stern@rowland.harvard.edu> - 2016-09-04 02:30 +0200
                                    Re: [PATCH V4] leds: trigger: Introduce an USB port trigger Rafał Miłecki <zajec5@gmail.com> - 2016-09-05 11:30 +0200

#1470125 — Re: [PATCH V4] leds: trigger: Introduce an USB port trigger

On Thu, Aug 25, 2016 at 10:03:52AM +0200, Rafał Miłecki wrote:
> +static void usbport_trig_activate(struct led_classdev *led_cdev)
> +{
> +	struct usbport_trig_data *usbport_data;
> +	int err;
> +
> +	usbport_data = kzalloc(sizeof(*usbport_data), GFP_KERNEL);
> +	if (!usbport_data)
> +		return;
> +	usbport_data->led_cdev = led_cdev;
> +
> +	/* Storing ports */
> +	INIT_LIST_HEAD(&usbport_data->ports);
> +	usbport_data->ports_dir = kobject_create_and_add("ports",
> +							 &led_cdev->dev->kobj);

If you _ever_ find yourself in a driver having to use kobject calls,
it's a HUGE hint that you are doing something wrong.

Hint, you are doing this wrong :)

Use an attribute group if you need a subdirectory in sysfs, using a
"raw" kobject like this hides it from all userspace tools and so no
userspace program can see it (hint, try using libudev to access these
files attached to the device...)

> +	if (!usbport_data->ports_dir)
> +		goto err_free;
> +
> +	/* API for ports management */
> +	err = device_create_file(led_cdev->dev, &dev_attr_new_port);
> +	if (err)
> +		goto err_put_ports;
> +	err = device_create_file(led_cdev->dev, &dev_attr_remove_port);
> +	if (err)
> +		goto err_remove_new_port;

Doesn't this race with userspace and fail?  Shouldn't the led core be
creating your leds for you?

> +
> +	/* Notifications */
> +	usbport_data->nb.notifier_call = usbport_trig_notify,
> +	led_cdev->trigger_data = usbport_data;
> +	usb_register_notify(&usbport_data->nb);

Don't abuse the USB notifier chain with stuff like this please, is that
really necessary?  Why can't your hardware just tell you what USB ports
are in use "out of band"?


> +
> +	led_cdev->activated = true;
> +	return;
> +
> +err_remove_new_port:
> +	device_remove_file(led_cdev->dev, &dev_attr_new_port);
> +err_put_ports:
> +	kobject_put(usbport_data->ports_dir);
> +err_free:
> +	kfree(usbport_data);
> +}

And again, why is this USB specific?  Why can't you use this same
userspace api and codebase for PCI ports?  For a sdcard port?  For a
thunderbolt port?

thanks,

greg k-h

[toc] | [next] | [standalone]

#1470830

On 25 August 2016 at 14:49, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Thu, Aug 25, 2016 at 10:03:52AM +0200, Rafał Miłecki wrote:
>> +static void usbport_trig_activate(struct led_classdev *led_cdev)
>> +{
>> +     struct usbport_trig_data *usbport_data;
>> +     int err;
>> +
>> +     usbport_data = kzalloc(sizeof(*usbport_data), GFP_KERNEL);
>> +     if (!usbport_data)
>> +             return;
>> +     usbport_data->led_cdev = led_cdev;
>> +
>> +     /* Storing ports */
>> +     INIT_LIST_HEAD(&usbport_data->ports);
>> +     usbport_data->ports_dir = kobject_create_and_add("ports",
>> +                                                      &led_cdev->dev->kobj);
>
> If you _ever_ find yourself in a driver having to use kobject calls,
> it's a HUGE hint that you are doing something wrong.
>
> Hint, you are doing this wrong :)
>
> Use an attribute group if you need a subdirectory in sysfs, using a
> "raw" kobject like this hides it from all userspace tools and so no
> userspace program can see it (hint, try using libudev to access these
> files attached to the device...)

Oops. Thanks for pointing groups to me. I was looking at sysfs.h
earlier but I didn't realize group can be a subdirectory. I can see
these sysfs_create_group(s) and friends now, thanks.


>> +     if (!usbport_data->ports_dir)
>> +             goto err_free;
>> +
>> +     /* API for ports management */
>> +     err = device_create_file(led_cdev->dev, &dev_attr_new_port);
>> +     if (err)
>> +             goto err_put_ports;
>> +     err = device_create_file(led_cdev->dev, &dev_attr_remove_port);
>> +     if (err)
>> +             goto err_remove_new_port;
>
> Doesn't this race with userspace and fail?  Shouldn't the led core be
> creating your leds for you?

These questions aren't clear to me. What kind of race? Doing
echo usbport > trigger
results in trigger core calling usbport_trig_activate. We create new
attributes and then we return.

I'm also not creating any leds there. This already has to be LED
available if you want to assign some trigger to it.


>> +
>> +     /* Notifications */
>> +     usbport_data->nb.notifier_call = usbport_trig_notify,
>> +     led_cdev->trigger_data = usbport_data;
>> +     usb_register_notify(&usbport_data->nb);
>
> Don't abuse the USB notifier chain with stuff like this please, is that
> really necessary?  Why can't your hardware just tell you what USB ports
> are in use "out of band"?

I totally don't understand this part. This LED trigger is supposed to
react to USB devices appearing at specified ports. How else can I know
there is a new USB device if not by notifications?
I'm wondering if we are on the same page. Could it be my idea of this
LED trigger is not clear at all? Could you take a look at commit
message again, please? Mostly this part:
> This commit adds a new trigger responsible for turning on LED when USB
> device gets connected to the specified USB port. This can can useful for
> various home routers that have USB port(s) and a proper LED telling user
> a device is connected.

Can I add something more to make it clear what this trigger is responsible for?


>> +
>> +     led_cdev->activated = true;
>> +     return;
>> +
>> +err_remove_new_port:
>> +     device_remove_file(led_cdev->dev, &dev_attr_new_port);
>> +err_put_ports:
>> +     kobject_put(usbport_data->ports_dir);
>> +err_free:
>> +     kfree(usbport_data);
>> +}
>
> And again, why is this USB specific?  Why can't you use this same
> userspace api and codebase for PCI ports?  For a sdcard port?  For a
> thunderbolt port?

I'm leaving this one unanswered as discussion on this continued in
V3.5 thread below my reply:
On 25 August 2016 at 07:14, Rafał Miłecki <zajec5@gmail.com> wrote:
> Good question. I would like to extend this USB port trigger in the
> future by reacting to USB activity. This involves playing with URBs
> and I believe that at that point it'd be getting too much USB specific
> to /rule them all/.

-- 
Rafał

[toc] | [prev] | [next] | [standalone]

#1472441

On Fri, Aug 26, 2016 at 05:38:05PM +0200, Rafał Miłecki wrote:
> On 25 August 2016 at 14:49, Greg KH <gregkh@linuxfoundation.org> wrote:
> > On Thu, Aug 25, 2016 at 10:03:52AM +0200, Rafał Miłecki wrote:
> >> +static void usbport_trig_activate(struct led_classdev *led_cdev)
> >> +{
> >> +     struct usbport_trig_data *usbport_data;
> >> +     int err;
> >> +
> >> +     usbport_data = kzalloc(sizeof(*usbport_data), GFP_KERNEL);
> >> +     if (!usbport_data)
> >> +             return;
> >> +     usbport_data->led_cdev = led_cdev;
> >> +
> >> +     /* Storing ports */
> >> +     INIT_LIST_HEAD(&usbport_data->ports);
> >> +     usbport_data->ports_dir = kobject_create_and_add("ports",
> >> +                                                      &led_cdev->dev->kobj);
> >
> > If you _ever_ find yourself in a driver having to use kobject calls,
> > it's a HUGE hint that you are doing something wrong.
> >
> > Hint, you are doing this wrong :)
> >
> > Use an attribute group if you need a subdirectory in sysfs, using a
> > "raw" kobject like this hides it from all userspace tools and so no
> > userspace program can see it (hint, try using libudev to access these
> > files attached to the device...)
> 
> Oops. Thanks for pointing groups to me. I was looking at sysfs.h
> earlier but I didn't realize group can be a subdirectory. I can see
> these sysfs_create_group(s) and friends now, thanks.

No, use an attribute group, and name it, and the subdir will be created
automagically for you.

> >> +     if (!usbport_data->ports_dir)
> >> +             goto err_free;
> >> +
> >> +     /* API for ports management */
> >> +     err = device_create_file(led_cdev->dev, &dev_attr_new_port);
> >> +     if (err)
> >> +             goto err_put_ports;
> >> +     err = device_create_file(led_cdev->dev, &dev_attr_remove_port);
> >> +     if (err)
> >> +             goto err_remove_new_port;
> >
> > Doesn't this race with userspace and fail?  Shouldn't the led core be
> > creating your leds for you?
> 
> These questions aren't clear to me. What kind of race? Doing
> echo usbport > trigger
> results in trigger core calling usbport_trig_activate. We create new
> attributes and then we return.

Why aren't the attributes present when the device is found?  What is
"trigger" for?

> I'm also not creating any leds there. This already has to be LED
> available if you want to assign some trigger to it.

That sounds really odd...

> >> +
> >> +     /* Notifications */
> >> +     usbport_data->nb.notifier_call = usbport_trig_notify,
> >> +     led_cdev->trigger_data = usbport_data;
> >> +     usb_register_notify(&usbport_data->nb);
> >
> > Don't abuse the USB notifier chain with stuff like this please, is that
> > really necessary?  Why can't your hardware just tell you what USB ports
> > are in use "out of band"?
> 
> I totally don't understand this part. This LED trigger is supposed to
> react to USB devices appearing at specified ports. How else can I know
> there is a new USB device if not by notifications?
> I'm wondering if we are on the same page. Could it be my idea of this
> LED trigger is not clear at all? Could you take a look at commit
> message again, please? Mostly this part:
> > This commit adds a new trigger responsible for turning on LED when USB
> > device gets connected to the specified USB port. This can can useful for
> > various home routers that have USB port(s) and a proper LED telling user
> > a device is connected.
> 
> Can I add something more to make it clear what this trigger is responsible for?

Yes please, as I totally missed that.

And the USB notifier was created for some "special" parts of the USB
core to use, this feels like an abuse of that in some way.  But it's
hard to define, I know...

> >> +
> >> +     led_cdev->activated = true;
> >> +     return;
> >> +
> >> +err_remove_new_port:
> >> +     device_remove_file(led_cdev->dev, &dev_attr_new_port);
> >> +err_put_ports:
> >> +     kobject_put(usbport_data->ports_dir);
> >> +err_free:
> >> +     kfree(usbport_data);
> >> +}
> >
> > And again, why is this USB specific?  Why can't you use this same
> > userspace api and codebase for PCI ports?  For a sdcard port?  For a
> > thunderbolt port?
> 
> I'm leaving this one unanswered as discussion on this continued in
> V3.5 thread below my reply:
> On 25 August 2016 at 07:14, Rafał Miłecki <zajec5@gmail.com> wrote:
> > Good question. I would like to extend this USB port trigger in the
> > future by reacting to USB activity. This involves playing with URBs
> > and I believe that at that point it'd be getting too much USB specific
> > to /rule them all/.

No, I mean why not have this specific activity LED work for all types of
devices, not just USB.

Not that this specific LED works for other types of USB activity.

thanks,

greg k-h

[toc] | [prev] | [next] | [standalone]

#1472846

On 30 August 2016 at 14:05, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Fri, Aug 26, 2016 at 05:38:05PM +0200, Rafał Miłecki wrote:
>> On 25 August 2016 at 14:49, Greg KH <gregkh@linuxfoundation.org> wrote:
>> > On Thu, Aug 25, 2016 at 10:03:52AM +0200, Rafał Miłecki wrote:
>> >> +static void usbport_trig_activate(struct led_classdev *led_cdev)
>> >> +{
>> >> +     struct usbport_trig_data *usbport_data;
>> >> +     int err;
>> >> +
>> >> +     usbport_data = kzalloc(sizeof(*usbport_data), GFP_KERNEL);
>> >> +     if (!usbport_data)
>> >> +             return;
>> >> +     usbport_data->led_cdev = led_cdev;
>> >> +
>> >> +     /* Storing ports */
>> >> +     INIT_LIST_HEAD(&usbport_data->ports);
>> >> +     usbport_data->ports_dir = kobject_create_and_add("ports",
>> >> +                                                      &led_cdev->dev->kobj);
>> >
>> > If you _ever_ find yourself in a driver having to use kobject calls,
>> > it's a HUGE hint that you are doing something wrong.
>> >
>> > Hint, you are doing this wrong :)
>> >
>> > Use an attribute group if you need a subdirectory in sysfs, using a
>> > "raw" kobject like this hides it from all userspace tools and so no
>> > userspace program can see it (hint, try using libudev to access these
>> > files attached to the device...)
>>
>> Oops. Thanks for pointing groups to me. I was looking at sysfs.h
>> earlier but I didn't realize group can be a subdirectory. I can see
>> these sysfs_create_group(s) and friends now, thanks.
>
> No, use an attribute group, and name it, and the subdir will be created
> automagically for you.

Oh, so I got it wrong again. Thanks for explaining.


>> >> +     if (!usbport_data->ports_dir)
>> >> +             goto err_free;
>> >> +
>> >> +     /* API for ports management */
>> >> +     err = device_create_file(led_cdev->dev, &dev_attr_new_port);
>> >> +     if (err)
>> >> +             goto err_put_ports;
>> >> +     err = device_create_file(led_cdev->dev, &dev_attr_remove_port);
>> >> +     if (err)
>> >> +             goto err_remove_new_port;
>> >
>> > Doesn't this race with userspace and fail?  Shouldn't the led core be
>> > creating your leds for you?
>>
>> These questions aren't clear to me. What kind of race? Doing
>> echo usbport > trigger
>> results in trigger core calling usbport_trig_activate. We create new
>> attributes and then we return.
>
> Why aren't the attributes present when the device is found?  What is
> "trigger" for?
>
>> I'm also not creating any leds there. This already has to be LED
>> available if you want to assign some trigger to it.
>
> That sounds really odd...

Please take a look at Documentation to get some idea of LED triggers:
Documentation/leds/leds-class.txt

Basically a LED (/sys/class/leds/foo/) can be controller with
"brightness" sysfs file like this:
echo 0 > brightness
echo 5 > brightness
echo 255 > brightness
(most LEDs react only 0 and non-0 values).

As you quite often need more complex LED management, there are
triggers that were introduced in 2006 by c3bc9956ec52f ("[PATCH] LED:
add LED trigger tupport"). Some triggers are trivial and could be
implemented in userspace as well (e.g. "timer"). Some had to be
implemented in kernelspace (CPU activity, MTD activity, etc.). Having
few triggers compiled, you can assign them to LEDs at it pleases you.
Your hardware may have generic LED (not labeled) and you can
dynamically assign various triggers to it, depending e.g. on user
actions. E.g. if user (using GUI or whatever) wants to see flash
activity, your userspace script should do:
echo mtd > /sys/class/leds/foo/trigger

I hope it explains things a big and you can see know why trigger code
is independent of creating LEDs. It's about layers and making things
generic I believe.

There is a place for LED driver that is responsible for creating
"struct led_classdev" with proper callback setting brightness. It
provides LED name, but is unaware of the way it should be
lighted/turned off.
There is LED subsystem.
There are triggers that are unaware of LED hardware details and only
set LED brightness using generic API.

I'm obviously not author of all of that and I can't explain some
design decisions, but I hope I gave you a clue how it works.


>> >> +
>> >> +     /* Notifications */
>> >> +     usbport_data->nb.notifier_call = usbport_trig_notify,
>> >> +     led_cdev->trigger_data = usbport_data;
>> >> +     usb_register_notify(&usbport_data->nb);
>> >
>> > Don't abuse the USB notifier chain with stuff like this please, is that
>> > really necessary?  Why can't your hardware just tell you what USB ports
>> > are in use "out of band"?
>>
>> I totally don't understand this part. This LED trigger is supposed to
>> react to USB devices appearing at specified ports. How else can I know
>> there is a new USB device if not by notifications?
>> I'm wondering if we are on the same page. Could it be my idea of this
>> LED trigger is not clear at all? Could you take a look at commit
>> message again, please? Mostly this part:
>> > This commit adds a new trigger responsible for turning on LED when USB
>> > device gets connected to the specified USB port. This can can useful for
>> > various home routers that have USB port(s) and a proper LED telling user
>> > a device is connected.
>>
>> Can I add something more to make it clear what this trigger is responsible for?
>
> Yes please, as I totally missed that.
>
> And the USB notifier was created for some "special" parts of the USB
> core to use, this feels like an abuse of that in some way.  But it's
> hard to define, I know...

I didn't mean to abuse this USB notifier, can you think of any other
way to achieve the same result? We could think about modifying USB
core to call some symbol from the trigger driver (see usage of
ledtrig_mtd_activity symbol) but is it any better than using
notification block?


>> >> +
>> >> +     led_cdev->activated = true;
>> >> +     return;
>> >> +
>> >> +err_remove_new_port:
>> >> +     device_remove_file(led_cdev->dev, &dev_attr_new_port);
>> >> +err_put_ports:
>> >> +     kobject_put(usbport_data->ports_dir);
>> >> +err_free:
>> >> +     kfree(usbport_data);
>> >> +}
>> >
>> > And again, why is this USB specific?  Why can't you use this same
>> > userspace api and codebase for PCI ports?  For a sdcard port?  For a
>> > thunderbolt port?
>>
>> I'm leaving this one unanswered as discussion on this continued in
>> V3.5 thread below my reply:
>> On 25 August 2016 at 07:14, Rafał Miłecki <zajec5@gmail.com> wrote:
>> > Good question. I would like to extend this USB port trigger in the
>> > future by reacting to USB activity. This involves playing with URBs
>> > and I believe that at that point it'd be getting too much USB specific
>> > to /rule them all/.
>
> No, I mean why not have this specific activity LED work for all types of
> devices, not just USB.
>
> Not that this specific LED works for other types of USB activity.

Sorry, I'm really trying my best, but I may be missing some point. I'm
not sure if I'll answer your question as you expected.

This patch adds only 1 trigger some users may be interested in, user
is still free to assign any trigger he wants to any LED it has on his
hardware. One day I plan to work on more triggers, so user can pick
the one he wants (e.g. when using generic not-labeled LED):
echo none > trigger
echo usbport > trigger
echo foo > trigger

-- 
Rafał

[toc] | [prev] | [next] | [standalone]

#1472867

On Tue, 30 Aug 2016, Rafał Miłecki wrote:

> Please take a look at Documentation to get some idea of LED triggers:
> Documentation/leds/leds-class.txt
> 
> Basically a LED (/sys/class/leds/foo/) can be controller with
> "brightness" sysfs file like this:
> echo 0 > brightness
> echo 5 > brightness
> echo 255 > brightness
> (most LEDs react only 0 and non-0 values).
> 
> As you quite often need more complex LED management, there are
> triggers that were introduced in 2006 by c3bc9956ec52f ("[PATCH] LED:
> add LED trigger tupport"). Some triggers are trivial and could be
> implemented in userspace as well (e.g. "timer"). Some had to be
> implemented in kernelspace (CPU activity, MTD activity, etc.). Having
> few triggers compiled, you can assign them to LEDs at it pleases you.
> Your hardware may have generic LED (not labeled) and you can
> dynamically assign various triggers to it, depending e.g. on user
> actions. E.g. if user (using GUI or whatever) wants to see flash
> activity, your userspace script should do:
> echo mtd > /sys/class/leds/foo/trigger

So for example, you might want to do:

	echo usb1-4 >/sys/class/leds/foo/trigger

and then have the "foo" LED toggle whenever an URB was submitted or 
completed for a device attached to the 1-4 port.  Right?

> I hope it explains things a big and you can see know why trigger code
> is independent of creating LEDs. It's about layers and making things
> generic I believe.
> 
> There is a place for LED driver that is responsible for creating
> "struct led_classdev" with proper callback setting brightness. It
> provides LED name, but is unaware of the way it should be
> lighted/turned off.
> There is LED subsystem.
> There are triggers that are unaware of LED hardware details and only
> set LED brightness using generic API.
> 
> I'm obviously not author of all of that and I can't explain some
> design decisions, but I hope I gave you a clue how it works.

> >> >> +     /* Notifications */
> >> >> +     usbport_data->nb.notifier_call = usbport_trig_notify,
> >> >> +     led_cdev->trigger_data = usbport_data;
> >> >> +     usb_register_notify(&usbport_data->nb);
> >> >
> >> > Don't abuse the USB notifier chain with stuff like this please, is that
> >> > really necessary?  Why can't your hardware just tell you what USB ports
> >> > are in use "out of band"?
> >>
> >> I totally don't understand this part. This LED trigger is supposed to
> >> react to USB devices appearing at specified ports. How else can I know
> >> there is a new USB device if not by notifications?
> >> I'm wondering if we are on the same page. Could it be my idea of this
> >> LED trigger is not clear at all? Could you take a look at commit
> >> message again, please? Mostly this part:
> >> > This commit adds a new trigger responsible for turning on LED when USB
> >> > device gets connected to the specified USB port. This can can useful for
> >> > various home routers that have USB port(s) and a proper LED telling user
> >> > a device is connected.
> >>
> >> Can I add something more to make it clear what this trigger is responsible for?
> >
> > Yes please, as I totally missed that.
> >
> > And the USB notifier was created for some "special" parts of the USB
> > core to use, this feels like an abuse of that in some way.  But it's
> > hard to define, I know...
> 
> I didn't mean to abuse this USB notifier, can you think of any other
> way to achieve the same result? We could think about modifying USB
> core to call some symbol from the trigger driver (see usage of
> ledtrig_mtd_activity symbol) but is it any better than using
> notification block?

I don't think this is such a bad use of the USB notifier.  All you need 
to know is when a device is attached to or unplugged from an LED's 
port.  Neither of these is a very frequent event.

However, there is still the question of how to know when an URB is 
submitted or completed for the device in question.  Obviously the USB 
core would have to call an LED routine.  But how would you determine 
which LED(s) to toggle?  Go through the entire list, looking for ones 
that are bound to the USB device in question?  This seems inefficient.  
Use a hash table?

Alan Stern

[toc] | [prev] | [next] | [standalone]

#1472885

On 30 August 2016 at 22:54, Alan Stern <stern@rowland.harvard.edu> wrote:
> On Tue, 30 Aug 2016, Rafał Miłecki wrote:
>
>> Please take a look at Documentation to get some idea of LED triggers:
>> Documentation/leds/leds-class.txt
>>
>> Basically a LED (/sys/class/leds/foo/) can be controller with
>> "brightness" sysfs file like this:
>> echo 0 > brightness
>> echo 5 > brightness
>> echo 255 > brightness
>> (most LEDs react only 0 and non-0 values).
>>
>> As you quite often need more complex LED management, there are
>> triggers that were introduced in 2006 by c3bc9956ec52f ("[PATCH] LED:
>> add LED trigger tupport"). Some triggers are trivial and could be
>> implemented in userspace as well (e.g. "timer"). Some had to be
>> implemented in kernelspace (CPU activity, MTD activity, etc.). Having
>> few triggers compiled, you can assign them to LEDs at it pleases you.
>> Your hardware may have generic LED (not labeled) and you can
>> dynamically assign various triggers to it, depending e.g. on user
>> actions. E.g. if user (using GUI or whatever) wants to see flash
>> activity, your userspace script should do:
>> echo mtd > /sys/class/leds/foo/trigger
>
> So for example, you might want to do:
>
>         echo usb1-4 >/sys/class/leds/foo/trigger
>
> and then have the "foo" LED toggle whenever an URB was submitted or
> completed for a device attached to the 1-4 port.  Right?

Not really as it won't cover some pretty common use cases. Many home
routers have few USB ports (2-5) and only 1 USB LED. It has to be
possible to assign few USB ports to a single LED (trigger). That way
LED should be turned on (and kept on) if there is at least 1 USB
device connected. You obviously can't do:
echo "usb1-1 usb1-2 usb2-1" > /sys/class/leds/foo/trigger

This was already brought up by Rob (who mentioned CPU trigger) and I
replied him pretty much the same way in:
https://lkml.org/lkml/2016/7/29/38
(reply starts with "Anyway, the serious limitation I see").


>> I hope it explains things a big and you can see know why trigger code
>> is independent of creating LEDs. It's about layers and making things
>> generic I believe.
>>
>> There is a place for LED driver that is responsible for creating
>> "struct led_classdev" with proper callback setting brightness. It
>> provides LED name, but is unaware of the way it should be
>> lighted/turned off.
>> There is LED subsystem.
>> There are triggers that are unaware of LED hardware details and only
>> set LED brightness using generic API.
>>
>> I'm obviously not author of all of that and I can't explain some
>> design decisions, but I hope I gave you a clue how it works.
>
>> >> >> +     /* Notifications */
>> >> >> +     usbport_data->nb.notifier_call = usbport_trig_notify,
>> >> >> +     led_cdev->trigger_data = usbport_data;
>> >> >> +     usb_register_notify(&usbport_data->nb);
>> >> >
>> >> > Don't abuse the USB notifier chain with stuff like this please, is that
>> >> > really necessary?  Why can't your hardware just tell you what USB ports
>> >> > are in use "out of band"?
>> >>
>> >> I totally don't understand this part. This LED trigger is supposed to
>> >> react to USB devices appearing at specified ports. How else can I know
>> >> there is a new USB device if not by notifications?
>> >> I'm wondering if we are on the same page. Could it be my idea of this
>> >> LED trigger is not clear at all? Could you take a look at commit
>> >> message again, please? Mostly this part:
>> >> > This commit adds a new trigger responsible for turning on LED when USB
>> >> > device gets connected to the specified USB port. This can can useful for
>> >> > various home routers that have USB port(s) and a proper LED telling user
>> >> > a device is connected.
>> >>
>> >> Can I add something more to make it clear what this trigger is responsible for?
>> >
>> > Yes please, as I totally missed that.
>> >
>> > And the USB notifier was created for some "special" parts of the USB
>> > core to use, this feels like an abuse of that in some way.  But it's
>> > hard to define, I know...
>>
>> I didn't mean to abuse this USB notifier, can you think of any other
>> way to achieve the same result? We could think about modifying USB
>> core to call some symbol from the trigger driver (see usage of
>> ledtrig_mtd_activity symbol) but is it any better than using
>> notification block?
>
> I don't think this is such a bad use of the USB notifier.  All you need
> to know is when a device is attached to or unplugged from an LED's
> port.  Neither of these is a very frequent event.
>
> However, there is still the question of how to know when an URB is
> submitted or completed for the device in question.  Obviously the USB
> core would have to call an LED routine.  But how would you determine
> which LED(s) to toggle?  Go through the entire list, looking for ones
> that are bound to the USB device in question?  This seems inefficient.
> Use a hash table?

I was hoping to bring it to discussion later, as it seems even the
basic version of this trigger causes many design problems.

-- 
Rafał

[toc] | [prev] | [next] | [standalone]

#1473783

On Tue, 30 Aug 2016, Rafał Miłecki wrote:

> >> As you quite often need more complex LED management, there are
> >> triggers that were introduced in 2006 by c3bc9956ec52f ("[PATCH] LED:
> >> add LED trigger tupport"). Some triggers are trivial and could be
> >> implemented in userspace as well (e.g. "timer"). Some had to be
> >> implemented in kernelspace (CPU activity, MTD activity, etc.). Having
> >> few triggers compiled, you can assign them to LEDs at it pleases you.
> >> Your hardware may have generic LED (not labeled) and you can
> >> dynamically assign various triggers to it, depending e.g. on user
> >> actions. E.g. if user (using GUI or whatever) wants to see flash
> >> activity, your userspace script should do:
> >> echo mtd > /sys/class/leds/foo/trigger
> >
> > So for example, you might want to do:
> >
> >         echo usb1-4 >/sys/class/leds/foo/trigger
> >
> > and then have the "foo" LED toggle whenever an URB was submitted or
> > completed for a device attached to the 1-4 port.  Right?
> 
> Not really as it won't cover some pretty common use cases. Many home
> routers have few USB ports (2-5) and only 1 USB LED. It has to be
> possible to assign few USB ports to a single LED (trigger). That way
> LED should be turned on (and kept on) if there is at least 1 USB
> device connected. You obviously can't do:
> echo "usb1-1 usb1-2 usb2-1" > /sys/class/leds/foo/trigger
> 
> This was already brought up by Rob (who mentioned CPU trigger) and I
> replied him pretty much the same way in:
> https://lkml.org/lkml/2016/7/29/38
> (reply starts with "Anyway, the serious limitation I see").

The code for a bunch of triggers must already be written.  What would 
the user do if he wanted to flash a single LED in response to both
CPU activity and MTD activity?  If not

	echo "cpu mtd" >/sys/class/leds/foo/trigger

then what?

Alan Stern

[toc] | [prev] | [next] | [standalone]

#1473797

On 31 August 2016 at 20:23, Alan Stern <stern@rowland.harvard.edu> wrote:
> On Tue, 30 Aug 2016, Rafał Miłecki wrote:
>
>> >> As you quite often need more complex LED management, there are
>> >> triggers that were introduced in 2006 by c3bc9956ec52f ("[PATCH] LED:
>> >> add LED trigger tupport"). Some triggers are trivial and could be
>> >> implemented in userspace as well (e.g. "timer"). Some had to be
>> >> implemented in kernelspace (CPU activity, MTD activity, etc.). Having
>> >> few triggers compiled, you can assign them to LEDs at it pleases you.
>> >> Your hardware may have generic LED (not labeled) and you can
>> >> dynamically assign various triggers to it, depending e.g. on user
>> >> actions. E.g. if user (using GUI or whatever) wants to see flash
>> >> activity, your userspace script should do:
>> >> echo mtd > /sys/class/leds/foo/trigger
>> >
>> > So for example, you might want to do:
>> >
>> >         echo usb1-4 >/sys/class/leds/foo/trigger
>> >
>> > and then have the "foo" LED toggle whenever an URB was submitted or
>> > completed for a device attached to the 1-4 port.  Right?
>>
>> Not really as it won't cover some pretty common use cases. Many home
>> routers have few USB ports (2-5) and only 1 USB LED. It has to be
>> possible to assign few USB ports to a single LED (trigger). That way
>> LED should be turned on (and kept on) if there is at least 1 USB
>> device connected. You obviously can't do:
>> echo "usb1-1 usb1-2 usb2-1" > /sys/class/leds/foo/trigger
>>
>> This was already brought up by Rob (who mentioned CPU trigger) and I
>> replied him pretty much the same way in:
>> https://lkml.org/lkml/2016/7/29/38
>> (reply starts with "Anyway, the serious limitation I see").
>
> The code for a bunch of triggers must already be written.  What would
> the user do if he wanted to flash a single LED in response to both
> CPU activity and MTD activity?  If not
>
>         echo "cpu mtd" >/sys/class/leds/foo/trigger
>
> then what?

Well, it sounds like a new feature then. Shall we add an extra API
with a request function for turning LED on? It could internally count
how many requests were raised and keep LED on as long as there is at
least 1 left. I guess we should implement it in trigger "subsystem"
(if I can call it so). Does it sound like a good plan?

-- 
Rafał

[toc] | [prev] | [next] | [standalone]

#1474065

On 31 August 2016 at 21:00, Rafał Miłecki <zajec5@gmail.com> wrote:
> On 31 August 2016 at 20:23, Alan Stern <stern@rowland.harvard.edu> wrote:
>> On Tue, 30 Aug 2016, Rafał Miłecki wrote:
>>> Not really as it won't cover some pretty common use cases. Many home
>>> routers have few USB ports (2-5) and only 1 USB LED. It has to be
>>> possible to assign few USB ports to a single LED (trigger). That way
>>> LED should be turned on (and kept on) if there is at least 1 USB
>>> device connected. You obviously can't do:
>>> echo "usb1-1 usb1-2 usb2-1" > /sys/class/leds/foo/trigger
>>>
>>> This was already brought up by Rob (who mentioned CPU trigger) and I
>>> replied him pretty much the same way in:
>>> https://lkml.org/lkml/2016/7/29/38
>>> (reply starts with "Anyway, the serious limitation I see").
>>
>> The code for a bunch of triggers must already be written.  What would
>> the user do if he wanted to flash a single LED in response to both
>> CPU activity and MTD activity?  If not
>>
>>         echo "cpu mtd" >/sys/class/leds/foo/trigger
>>
>> then what?
>
> Well, it sounds like a new feature then. Shall we add an extra API
> with a request function for turning LED on? It could internally count
> how many requests were raised and keep LED on as long as there is at
> least 1 left. I guess we should implement it in trigger "subsystem"
> (if I can call it so). Does it sound like a good plan?

I'm pretty sure noone ever planned to have more than 1 trigger
assigned to a single LED. I just realized there will be a problem with
proposed solution: sysfs files conflict.

Consider 2 existing triggers for a moment:
1) oneshot: it creates following sysfs files:
/sys/class/leds/foo/delay_on
/sys/class/leds/foo/delay_off
/sys/class/leds/foo/invert
/sys/class/leds/foo/shot
2) timer: it creates following sysfs files:
/sys/class/leds/foo/delay_on
/sys/class/leds/foo/delay_off

Activating both of them will probably cause a WARNING in sysfs. They
can't coexist :(

We should probably have per-trigger subdirs, e.g.:
/sys/class/leds/foo/trigger-oneshot/delay_on
/sys/class/leds/foo/trigger-oneshot/delay_off
/sys/class/leds/foo/trigger-oneshot/invert
/sys/class/leds/foo/trigger-oneshot/shot
/sys/class/leds/foo/trigger-timer/delay_on
/sys/class/leds/foo/trigger-timer/delay_off
but implementing it now would break the ABI.

One workaround I can see is doing triggers V2, they:
1) Would put sysfs files in /sys/class/leds/foo/trigger-bar/
2) Use new API for *requesting* LED to be on/off
3) There would be a counter of requests in V2 API
4) Multiple triggers V2 would be allowed to be used (assigned) at the same time

-- 
Rafał

[toc] | [prev] | [next] | [standalone]

#1474135

On 09/01/2016 07:25 AM, Rafał Miłecki wrote:
> On 31 August 2016 at 21:00, Rafał Miłecki <zajec5@gmail.com> wrote:
>> On 31 August 2016 at 20:23, Alan Stern <stern@rowland.harvard.edu> wrote:
>>> On Tue, 30 Aug 2016, Rafał Miłecki wrote:
>>>> Not really as it won't cover some pretty common use cases. Many home
>>>> routers have few USB ports (2-5) and only 1 USB LED. It has to be
>>>> possible to assign few USB ports to a single LED (trigger). That way
>>>> LED should be turned on (and kept on) if there is at least 1 USB
>>>> device connected. You obviously can't do:
>>>> echo "usb1-1 usb1-2 usb2-1" > /sys/class/leds/foo/trigger
>>>>
>>>> This was already brought up by Rob (who mentioned CPU trigger) and I
>>>> replied him pretty much the same way in:
>>>> https://lkml.org/lkml/2016/7/29/38
>>>> (reply starts with "Anyway, the serious limitation I see").
>>>
>>> The code for a bunch of triggers must already be written.  What would
>>> the user do if he wanted to flash a single LED in response to both
>>> CPU activity and MTD activity?  If not
>>>
>>>         echo "cpu mtd" >/sys/class/leds/foo/trigger
>>>
>>> then what?
>>
>> Well, it sounds like a new feature then. Shall we add an extra API
>> with a request function for turning LED on? It could internally count
>> how many requests were raised and keep LED on as long as there is at
>> least 1 left. I guess we should implement it in trigger "subsystem"
>> (if I can call it so). Does it sound like a good plan?
>
> I'm pretty sure noone ever planned to have more than 1 trigger
> assigned to a single LED. I just realized there will be a problem with
> proposed solution: sysfs files conflict.
>
> Consider 2 existing triggers for a moment:
> 1) oneshot: it creates following sysfs files:
> /sys/class/leds/foo/delay_on
> /sys/class/leds/foo/delay_off
> /sys/class/leds/foo/invert
> /sys/class/leds/foo/shot
> 2) timer: it creates following sysfs files:
> /sys/class/leds/foo/delay_on
> /sys/class/leds/foo/delay_off
>
> Activating both of them will probably cause a WARNING in sysfs. They
> can't coexist :(
>
> We should probably have per-trigger subdirs, e.g.:
> /sys/class/leds/foo/trigger-oneshot/delay_on
> /sys/class/leds/foo/trigger-oneshot/delay_off
> /sys/class/leds/foo/trigger-oneshot/invert
> /sys/class/leds/foo/trigger-oneshot/shot
> /sys/class/leds/foo/trigger-timer/delay_on
> /sys/class/leds/foo/trigger-timer/delay_off
> but implementing it now would break the ABI.
>
> One workaround I can see is doing triggers V2, they:
> 1) Would put sysfs files in /sys/class/leds/foo/trigger-bar/
> 2) Use new API for *requesting* LED to be on/off
> 3) There would be a counter of requests in V2 API
> 4) Multiple triggers V2 would be allowed to be used (assigned) at the same time

Currently we support only triggers dedicated to specific type of
devices. Even in case of triggers that don't expose custom sysfs
attributes, registered with led_trigger_register_simple(), device
drivers have to generate trigger event with dedicated function, e.g:

- ledtrig-cpu: void ledtrig_cpu(enum cpu_led_event ledevt)
- ledtrig-disk: void ledtrig_disk_activity(void)
- ledtrig-mtd: void ledtrig_mtd_activity(void)

If one wanted to have the LED notified by different type of devices,
then they would have to implement a trigger that would exposed all
required types of API.

Unfortunately, there are many possible combinations of
triggers and that doesn't sound sane to add a new one when someone
will find it useful. Of course it would entail adding a call to the
new trigger API in the drivers, which doesn't seem like something
acceptable in the mainline.

-- 
Best regards,
Jacek Anaszewski

[toc] | [prev] | [next] | [standalone]

#1474439

On Thu, 1 Sep 2016, Jacek Anaszewski wrote:

> On 09/01/2016 07:25 AM, Rafał Miłecki wrote:
> > On 31 August 2016 at 21:00, Rafał Miłecki <zajec5@gmail.com> wrote:
> >> On 31 August 2016 at 20:23, Alan Stern <stern@rowland.harvard.edu> wrote:
> >>> On Tue, 30 Aug 2016, Rafał Miłecki wrote:
> >>>> Not really as it won't cover some pretty common use cases. Many home
> >>>> routers have few USB ports (2-5) and only 1 USB LED. It has to be
> >>>> possible to assign few USB ports to a single LED (trigger). That way
> >>>> LED should be turned on (and kept on) if there is at least 1 USB
> >>>> device connected. You obviously can't do:
> >>>> echo "usb1-1 usb1-2 usb2-1" > /sys/class/leds/foo/trigger
> >>>>
> >>>> This was already brought up by Rob (who mentioned CPU trigger) and I
> >>>> replied him pretty much the same way in:
> >>>> https://lkml.org/lkml/2016/7/29/38
> >>>> (reply starts with "Anyway, the serious limitation I see").
> >>>
> >>> The code for a bunch of triggers must already be written.  What would
> >>> the user do if he wanted to flash a single LED in response to both
> >>> CPU activity and MTD activity?  If not
> >>>
> >>>         echo "cpu mtd" >/sys/class/leds/foo/trigger
> >>>
> >>> then what?
> >>
> >> Well, it sounds like a new feature then. Shall we add an extra API
> >> with a request function for turning LED on? It could internally count
> >> how many requests were raised and keep LED on as long as there is at
> >> least 1 left. I guess we should implement it in trigger "subsystem"
> >> (if I can call it so). Does it sound like a good plan?
> >
> > I'm pretty sure noone ever planned to have more than 1 trigger
> > assigned to a single LED. I just realized there will be a problem with
> > proposed solution: sysfs files conflict.
> >
> > Consider 2 existing triggers for a moment:
> > 1) oneshot: it creates following sysfs files:
> > /sys/class/leds/foo/delay_on
> > /sys/class/leds/foo/delay_off
> > /sys/class/leds/foo/invert
> > /sys/class/leds/foo/shot
> > 2) timer: it creates following sysfs files:
> > /sys/class/leds/foo/delay_on
> > /sys/class/leds/foo/delay_off
> >
> > Activating both of them will probably cause a WARNING in sysfs. They
> > can't coexist :(
> >
> > We should probably have per-trigger subdirs, e.g.:
> > /sys/class/leds/foo/trigger-oneshot/delay_on
> > /sys/class/leds/foo/trigger-oneshot/delay_off
> > /sys/class/leds/foo/trigger-oneshot/invert
> > /sys/class/leds/foo/trigger-oneshot/shot
> > /sys/class/leds/foo/trigger-timer/delay_on
> > /sys/class/leds/foo/trigger-timer/delay_off
> > but implementing it now would break the ABI.
> >
> > One workaround I can see is doing triggers V2, they:
> > 1) Would put sysfs files in /sys/class/leds/foo/trigger-bar/
> > 2) Use new API for *requesting* LED to be on/off
> > 3) There would be a counter of requests in V2 API
> > 4) Multiple triggers V2 would be allowed to be used (assigned) at the same time
> 
> Currently we support only triggers dedicated to specific type of
> devices. Even in case of triggers that don't expose custom sysfs
> attributes, registered with led_trigger_register_simple(), device
> drivers have to generate trigger event with dedicated function, e.g:
> 
> - ledtrig-cpu: void ledtrig_cpu(enum cpu_led_event ledevt)
> - ledtrig-disk: void ledtrig_disk_activity(void)
> - ledtrig-mtd: void ledtrig_mtd_activity(void)
> 
> If one wanted to have the LED notified by different type of devices,
> then they would have to implement a trigger that would exposed all
> required types of API.
> 
> Unfortunately, there are many possible combinations of
> triggers and that doesn't sound sane to add a new one when someone
> will find it useful. Of course it would entail adding a call to the
> new trigger API in the drivers, which doesn't seem like something
> acceptable in the mainline.

Maybe it would make more sense, in this case, to allow only three 
possibilities for a USB port activity trigger.  Toggle the LED 
whenever:

	There is activity on the specified port, or

	There is any activity on any port on the specified hub, or

	There is any USB activity on any port.

That ought to cover most of the normal use cases, and it would be
simple enough to implement.

Alan Stern

[toc] | [prev] | [next] | [standalone]

#1474921

On 09/01/2016 04:36 PM, Alan Stern wrote:
> On Thu, 1 Sep 2016, Jacek Anaszewski wrote:
>
>> On 09/01/2016 07:25 AM, Rafał Miłecki wrote:
>>> On 31 August 2016 at 21:00, Rafał Miłecki <zajec5@gmail.com> wrote:
>>>> On 31 August 2016 at 20:23, Alan Stern <stern@rowland.harvard.edu> wrote:
>>>>> On Tue, 30 Aug 2016, Rafał Miłecki wrote:
>>>>>> Not really as it won't cover some pretty common use cases. Many home
>>>>>> routers have few USB ports (2-5) and only 1 USB LED. It has to be
>>>>>> possible to assign few USB ports to a single LED (trigger). That way
>>>>>> LED should be turned on (and kept on) if there is at least 1 USB
>>>>>> device connected. You obviously can't do:
>>>>>> echo "usb1-1 usb1-2 usb2-1" > /sys/class/leds/foo/trigger
>>>>>>
>>>>>> This was already brought up by Rob (who mentioned CPU trigger) and I
>>>>>> replied him pretty much the same way in:
>>>>>> https://lkml.org/lkml/2016/7/29/38
>>>>>> (reply starts with "Anyway, the serious limitation I see").
>>>>>
>>>>> The code for a bunch of triggers must already be written.  What would
>>>>> the user do if he wanted to flash a single LED in response to both
>>>>> CPU activity and MTD activity?  If not
>>>>>
>>>>>         echo "cpu mtd" >/sys/class/leds/foo/trigger
>>>>>
>>>>> then what?
>>>>
>>>> Well, it sounds like a new feature then. Shall we add an extra API
>>>> with a request function for turning LED on? It could internally count
>>>> how many requests were raised and keep LED on as long as there is at
>>>> least 1 left. I guess we should implement it in trigger "subsystem"
>>>> (if I can call it so). Does it sound like a good plan?
>>>
>>> I'm pretty sure noone ever planned to have more than 1 trigger
>>> assigned to a single LED. I just realized there will be a problem with
>>> proposed solution: sysfs files conflict.
>>>
>>> Consider 2 existing triggers for a moment:
>>> 1) oneshot: it creates following sysfs files:
>>> /sys/class/leds/foo/delay_on
>>> /sys/class/leds/foo/delay_off
>>> /sys/class/leds/foo/invert
>>> /sys/class/leds/foo/shot
>>> 2) timer: it creates following sysfs files:
>>> /sys/class/leds/foo/delay_on
>>> /sys/class/leds/foo/delay_off
>>>
>>> Activating both of them will probably cause a WARNING in sysfs. They
>>> can't coexist :(
>>>
>>> We should probably have per-trigger subdirs, e.g.:
>>> /sys/class/leds/foo/trigger-oneshot/delay_on
>>> /sys/class/leds/foo/trigger-oneshot/delay_off
>>> /sys/class/leds/foo/trigger-oneshot/invert
>>> /sys/class/leds/foo/trigger-oneshot/shot
>>> /sys/class/leds/foo/trigger-timer/delay_on
>>> /sys/class/leds/foo/trigger-timer/delay_off
>>> but implementing it now would break the ABI.
>>>
>>> One workaround I can see is doing triggers V2, they:
>>> 1) Would put sysfs files in /sys/class/leds/foo/trigger-bar/
>>> 2) Use new API for *requesting* LED to be on/off
>>> 3) There would be a counter of requests in V2 API
>>> 4) Multiple triggers V2 would be allowed to be used (assigned) at the same time
>>
>> Currently we support only triggers dedicated to specific type of
>> devices. Even in case of triggers that don't expose custom sysfs
>> attributes, registered with led_trigger_register_simple(), device
>> drivers have to generate trigger event with dedicated function, e.g:
>>
>> - ledtrig-cpu: void ledtrig_cpu(enum cpu_led_event ledevt)
>> - ledtrig-disk: void ledtrig_disk_activity(void)
>> - ledtrig-mtd: void ledtrig_mtd_activity(void)
>>
>> If one wanted to have the LED notified by different type of devices,
>> then they would have to implement a trigger that would exposed all
>> required types of API.
>>
>> Unfortunately, there are many possible combinations of
>> triggers and that doesn't sound sane to add a new one when someone
>> will find it useful. Of course it would entail adding a call to the
>> new trigger API in the drivers, which doesn't seem like something
>> acceptable in the mainline.
>
> Maybe it would make more sense, in this case, to allow only three
> possibilities for a USB port activity trigger.  Toggle the LED
> whenever:
>
> 	There is activity on the specified port, or
>
> 	There is any activity on any port on the specified hub, or
>
> 	There is any USB activity on any port.
>
> That ought to cover most of the normal use cases, and it would be
> simple enough to implement.

What would be the benefit of having a USB port activity trigger,
for which we would be specifying the port to observe, but in the same
time we would react on any activity on any port (cases 1 and 3)?

-- 
Best regards,
Jacek Anaszewski

[toc] | [prev] | [next] | [standalone]

#1475204

On Fri, 2 Sep 2016, Jacek Anaszewski wrote:

> >>> I'm pretty sure noone ever planned to have more than 1 trigger
> >>> assigned to a single LED. I just realized there will be a problem with
> >>> proposed solution: sysfs files conflict.

...

> >> Currently we support only triggers dedicated to specific type of
> >> devices. Even in case of triggers that don't expose custom sysfs
> >> attributes, registered with led_trigger_register_simple(), device
> >> drivers have to generate trigger event with dedicated function, e.g:
> >>
> >> - ledtrig-cpu: void ledtrig_cpu(enum cpu_led_event ledevt)
> >> - ledtrig-disk: void ledtrig_disk_activity(void)
> >> - ledtrig-mtd: void ledtrig_mtd_activity(void)
> >>
> >> If one wanted to have the LED notified by different type of devices,
> >> then they would have to implement a trigger that would exposed all
> >> required types of API.
> >>
> >> Unfortunately, there are many possible combinations of
> >> triggers and that doesn't sound sane to add a new one when someone
> >> will find it useful. Of course it would entail adding a call to the
> >> new trigger API in the drivers, which doesn't seem like something
> >> acceptable in the mainline.
> >
> > Maybe it would make more sense, in this case, to allow only three
> > possibilities for a USB port activity trigger.  Toggle the LED
> > whenever:
> >
> > 	There is activity on the specified port, or
> >
> > 	There is any activity on any port on the specified hub, or
> >
> > 	There is any USB activity on any port.
> >
> > That ought to cover most of the normal use cases, and it would be
> > simple enough to implement.
> 
> What would be the benefit of having a USB port activity trigger,
> for which we would be specifying the port to observe, but in the same
> time we would react on any activity on any port (cases 1 and 3)?

I meant these three cases to be mutually exclusive.  For a given LED,
you could have only one of those trigger types (like mentioned above,
only one trigger per LED).  For example, you might accept any one of:

	echo usb1-4.2 >/sys/class/led/foo/trigger

	echo hub1-4 >/sys/class/led/foo/trigger

	echo usb >/sys/class/led/foo/trigger

Yes, it would be possible to have a port-specific trigger for one LED
and an overall USB activity trigger for another LED.  I don't know how
useful this would be -- you could probably imagine some unlikely
scenario.

The point is that doing things this way wouldn't require any API
violations, and it would allow users to do almost all of the things
they are likely to want.

Alan Stern

[toc] | [prev] | [next] | [standalone]

#1475660

On 09/02/2016 04:33 PM, Alan Stern wrote:
> On Fri, 2 Sep 2016, Jacek Anaszewski wrote:
>
>>>>> I'm pretty sure noone ever planned to have more than 1 trigger
>>>>> assigned to a single LED. I just realized there will be a problem with
>>>>> proposed solution: sysfs files conflict.
>
> ...
>
>>>> Currently we support only triggers dedicated to specific type of
>>>> devices. Even in case of triggers that don't expose custom sysfs
>>>> attributes, registered with led_trigger_register_simple(), device
>>>> drivers have to generate trigger event with dedicated function, e.g:
>>>>
>>>> - ledtrig-cpu: void ledtrig_cpu(enum cpu_led_event ledevt)
>>>> - ledtrig-disk: void ledtrig_disk_activity(void)
>>>> - ledtrig-mtd: void ledtrig_mtd_activity(void)
>>>>
>>>> If one wanted to have the LED notified by different type of devices,
>>>> then they would have to implement a trigger that would exposed all
>>>> required types of API.
>>>>
>>>> Unfortunately, there are many possible combinations of
>>>> triggers and that doesn't sound sane to add a new one when someone
>>>> will find it useful. Of course it would entail adding a call to the
>>>> new trigger API in the drivers, which doesn't seem like something
>>>> acceptable in the mainline.
>>>
>>> Maybe it would make more sense, in this case, to allow only three
>>> possibilities for a USB port activity trigger.  Toggle the LED
>>> whenever:
>>>
>>> 	There is activity on the specified port, or
>>>
>>> 	There is any activity on any port on the specified hub, or
>>>
>>> 	There is any USB activity on any port.
>>>
>>> That ought to cover most of the normal use cases, and it would be
>>> simple enough to implement.
>>
>> What would be the benefit of having a USB port activity trigger,
>> for which we would be specifying the port to observe, but in the same
>> time we would react on any activity on any port (cases 1 and 3)?
>
> I meant these three cases to be mutually exclusive.  For a given LED,
> you could have only one of those trigger types (like mentioned above,
> only one trigger per LED).  For example, you might accept any one of:
>
> 	echo usb1-4.2 >/sys/class/led/foo/trigger
>
> 	echo hub1-4 >/sys/class/led/foo/trigger
>
> 	echo usb >/sys/class/led/foo/trigger
>
> Yes, it would be possible to have a port-specific trigger for one LED
> and an overall USB activity trigger for another LED.  I don't know how
> useful this would be -- you could probably imagine some unlikely
> scenario.
>
> The point is that doing things this way wouldn't require any API
> violations, and it would allow users to do almost all of the things
> they are likely to want.

We'd have to define single API for generating USB trigger event,
so as not enforce addition of three different API calls to the USB
drivers. The type of USB events that the LED should react upon could be
defined by parsing the value written to the sysfs file.
This of course implies, that we should have single LED USB port trigger.

The remaining issue is the sysfs interface design for defining and
presenting multiple USB ports. I'm still in favour of a single
attribute with space separated list. This scheme is commonly used
in existing interfaces.

-- 
Best regards,
Jacek Anaszewski

[toc] | [prev] | [next] | [standalone]

#1475663

On Sat, 3 Sep 2016, Jacek Anaszewski wrote:

> >>> Maybe it would make more sense, in this case, to allow only three
> >>> possibilities for a USB port activity trigger.  Toggle the LED
> >>> whenever:
> >>>
> >>> 	There is activity on the specified port, or
> >>>
> >>> 	There is any activity on any port on the specified hub, or
> >>>
> >>> 	There is any USB activity on any port.
> >>>
> >>> That ought to cover most of the normal use cases, and it would be
> >>> simple enough to implement.
> >>
> >> What would be the benefit of having a USB port activity trigger,
> >> for which we would be specifying the port to observe, but in the same
> >> time we would react on any activity on any port (cases 1 and 3)?
> >
> > I meant these three cases to be mutually exclusive.  For a given LED,
> > you could have only one of those trigger types (like mentioned above,
> > only one trigger per LED).  For example, you might accept any one of:
> >
> > 	echo usb1-4.2 >/sys/class/led/foo/trigger
> >
> > 	echo hub1-4 >/sys/class/led/foo/trigger
> >
> > 	echo usb >/sys/class/led/foo/trigger
> >
> > Yes, it would be possible to have a port-specific trigger for one LED
> > and an overall USB activity trigger for another LED.  I don't know how
> > useful this would be -- you could probably imagine some unlikely
> > scenario.
> >
> > The point is that doing things this way wouldn't require any API
> > violations, and it would allow users to do almost all of the things
> > they are likely to want.
> 
> We'd have to define single API for generating USB trigger event,
> so as not enforce addition of three different API calls to the USB
> drivers.

The USB core would need only one LED-API call, which it would make upon
the completion of an URB.  The trigger code should be able to handle
all the rest (i.e., see which LEDs should be triggered by that URB).

>  The type of USB events that the LED should react upon could be
> defined by parsing the value written to the sysfs file.

There is only one type of event: completion of an URB.  Triggers would
differ depending only on the device/port that the URB was aimed at.  
_That_ information could be defined by parsing the value written to the 
sysfs file.

> This of course implies, that we should have single LED USB port trigger.
> 
> The remaining issue is the sysfs interface design for defining and
> presenting multiple USB ports. I'm still in favour of a single
> attribute with space separated list. This scheme is commonly used
> in existing interfaces.

No such interface is needed if you do things the way I outlined above.  
Each trigger would require the user to specify either one port, one 
hub, or nothing at all.  Multiple ports would not be used.

Alan Stern

[toc] | [prev] | [next] | [standalone]

#1475779

On 09/03/2016 05:17 PM, Alan Stern wrote:
> On Sat, 3 Sep 2016, Jacek Anaszewski wrote:
>
>>>>> Maybe it would make more sense, in this case, to allow only three
>>>>> possibilities for a USB port activity trigger.  Toggle the LED
>>>>> whenever:
>>>>>
>>>>> 	There is activity on the specified port, or
>>>>>
>>>>> 	There is any activity on any port on the specified hub, or
>>>>>
>>>>> 	There is any USB activity on any port.
>>>>>
>>>>> That ought to cover most of the normal use cases, and it would be
>>>>> simple enough to implement.
>>>>
>>>> What would be the benefit of having a USB port activity trigger,
>>>> for which we would be specifying the port to observe, but in the same
>>>> time we would react on any activity on any port (cases 1 and 3)?
>>>
>>> I meant these three cases to be mutually exclusive.  For a given LED,
>>> you could have only one of those trigger types (like mentioned above,
>>> only one trigger per LED).  For example, you might accept any one of:
>>>
>>> 	echo usb1-4.2 >/sys/class/led/foo/trigger
>>>
>>> 	echo hub1-4 >/sys/class/led/foo/trigger
>>>
>>> 	echo usb >/sys/class/led/foo/trigger
>>>
>>> Yes, it would be possible to have a port-specific trigger for one LED
>>> and an overall USB activity trigger for another LED.  I don't know how
>>> useful this would be -- you could probably imagine some unlikely
>>> scenario.
>>>
>>> The point is that doing things this way wouldn't require any API
>>> violations, and it would allow users to do almost all of the things
>>> they are likely to want.
>>
>> We'd have to define single API for generating USB trigger event,
>> so as not enforce addition of three different API calls to the USB
>> drivers.

Of course this trigger represents yet another type of triggers, that
don't require exposing an API for generating events, but instead
register as listeners to some notifiers. I missed that initially.

> The USB core would need only one LED-API call, which it would make upon
> the completion of an URB.  The trigger code should be able to handle
> all the rest (i.e., see which LEDs should be triggered by that URB).

This is assured by the LED trigger core.

>
>>  The type of USB events that the LED should react upon could be
>> defined by parsing the value written to the sysfs file.
>
> There is only one type of event: completion of an URB.  Triggers would
> differ depending only on the device/port that the URB was aimed at.
> _That_ information could be defined by parsing the value written to the
> sysfs file.
>
>> This of course implies, that we should have single LED USB port trigger.
>>
>> The remaining issue is the sysfs interface design for defining and
>> presenting multiple USB ports. I'm still in favour of a single
>> attribute with space separated list. This scheme is commonly used
>> in existing interfaces.
>
> No such interface is needed if you do things the way I outlined above.
> Each trigger would require the user to specify either one port, one
> hub, or nothing at all.  Multiple ports would not be used.

The patch assumes that it is possible to register trigger for many
ports.

-- 
Best regards,
Jacek Anaszewski

[toc] | [prev] | [next] | [standalone]

#1475807

On Sat, 3 Sep 2016, Jacek Anaszewski wrote:

> >> The remaining issue is the sysfs interface design for defining and
> >> presenting multiple USB ports. I'm still in favour of a single
> >> attribute with space separated list. This scheme is commonly used
> >> in existing interfaces.
> >
> > No such interface is needed if you do things the way I outlined above.
> > Each trigger would require the user to specify either one port, one
> > hub, or nothing at all.  Multiple ports would not be used.
> 
> The patch assumes that it is possible to register trigger for many
> ports.

The patch could be changed to remove that assumption.

Alan Stern

[toc] | [prev] | [next] | [standalone]

#1476227

On 4 September 2016 at 02:24, Alan Stern <stern@rowland.harvard.edu> wrote:
> On Sat, 3 Sep 2016, Jacek Anaszewski wrote:
>
>> >> The remaining issue is the sysfs interface design for defining and
>> >> presenting multiple USB ports. I'm still in favour of a single
>> >> attribute with space separated list. This scheme is commonly used
>> >> in existing interfaces.
>> >
>> > No such interface is needed if you do things the way I outlined above.
>> > Each trigger would require the user to specify either one port, one
>> > hub, or nothing at all.  Multiple ports would not be used.
>>
>> The patch assumes that it is possible to register trigger for many
>> ports.
>
> The patch could be changed to remove that assumption.

I did this assumption for a reason. There are many devices with more
than 1 USB port and these ports are handled by the same controllers,
just different ports. Let me share USB details of my SmartRG SR400ac
device with 2 physical USB ports:

Physical USB 3.0 port is handled by following controllers and their
root hub ports:
1) OHCI's port 1
2) EHCI's port 1
3) XHCI's port 1

Physical USB 2.0 port is handled by following controllers and their
root hub ports:
1) OHCI's port 2
2) EHCI's port 2


This devices has 3 separated LEDs. I'll list them and tell which ports
should be assigned to them:

bcm53xx:green:usb3 should have assigned:
1) OHCI's port 1
2) EHCI's port 1

bcm53xx:white:usb3 should have assigned:
1) XHCI's port 1

bcm53xx:white:usb2 should have assigned:
1) OHCI's port 2
2) EHCI's port 2


As you can see, I need to specify ports precisely. Assigning a single
port is not enough. Assigning whole hub is too much. So none of:
>         echo usb1-4.2 >/sys/class/led/foo/trigger
>         echo hub1-4 >/sys/class/led/foo/trigger
>         echo usb >/sys/class/led/foo/trigger
would be sufficient here :(

-- 
Rafał

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web