*NEW* ACPI driver to support App Hot Startup AKA PNP0C32

If the laptop is turned off and a hot button is pressed, the computer turns on and boots. This module gets a notification from ACPI which button was pressed. Then the driver writes to the proc fs which button was pressed allowing userspace apps to read that files and behave accordingly. Proc files are also writable.

Angelo,

Quoted 5 lines · On Saturday 10 November 2007 01:44:48 Angelo Miguel Arrifano wrote:

If the laptop is turned off and a hot button is pressed, the computer turns on and boots. This module gets a notification from ACPI which button was pressed. Then the driver writes to the proc fs which button was pressed allowing userspace apps to read that files and behave accordingly. Proc files are also writable.

Please use sysfs, not proc (even if you aren't considering submitting this upstream) - /proc/acpi is deprecated and going away (slowly but surely).

-Carlos

Signature · 3 lines

E-Mail: [email protected]
Web: strangeworlds.co.uk
GPG Key ID: 0x23EE722D

Thanks! The driver is now using sysfs.
What should I do to submit the driver properly to upstream?

Hello all,

I opened a SF.net project to support devices found under PNP0C32.
These devices are known as Direct App Launch Buttons and are used on Vista Hotstart(TM) technology. I believe most of Vista "ready" laptops have this.

The project aims to provide similar functionality to Vista Hotstart(TM) in linux. You can read more at http://quickstart.sourceforge.net/

Greetings,

Signature · 11 lines

Angelo Arrifano AKA MiKNiX
CSE Student at UBI, Portugal
Gentoo Linux AMD64 Arch Tester
[email protected]
http://miknix.homelinux.com
PGP Pubkey online

- -

Q: How many Chuck Norris' does it take to change a light bulb?
A: None, Chuck Norris prefers to kill in the dark. 

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of ?ngelo Miguel Arrifano
Sent: Saturday, January 19, 2008 11:19 PM
To: [email protected]
Cc: [email protected]
Subject: Re: *NEW* ACPI driver to support App Hot Startup AKA PNP0C32 - please apply

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 4 lines · On Fri, 23 Nov 2007 15:24:46 +0000 Ângelo Miguel Arrifano <[email protected]>…

Hello all,

I opened a SF.net project to support devices found under PNP0C32. These devices are known as Direct App Launch Buttons and are used on Vista

Hotstart(TM) technology.

Quoted 3 lines

I believe most of Vista "ready" laptops have this.

The project aims to provide similar functionality to Vista Hotstart(TM) in

linux.

Quoted 11 lines

You can read more at http://quickstart.sourceforge.net/

Greetings,
--
Angelo Arrifano AKA MiKNiX
CSE Student at UBI, Portugal
Gentoo Linux AMD64 Arch Tester
[email protected]
http://miknix.homelinux.com
PGP Pubkey online

Quoted 4 lines

- -

Q: How many Chuck Norris' does it take to change a light bulb?
A: None, Chuck Norris prefers to kill in the dark.

The project is running for some time now. During this time there were no reports of people getting it trouble with it.

Attached is the patch.

Thanks,
- --
Angelo Arrifano AKA MiKNiX
CSE Student at UBI, Portugal
Gentoo Linux AMD64 Arch Tester
Linwizard Developer
[email protected]
http://miknix.homelinux.com
PGP Pubkey 0x3D92BB0B
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.7 (GNU/Linux)

iD8DBQFHkhUBNahyoD2SuwsRAn8gAJ4yj/kQl3RyAbBw3s9/2R04rEjFbQCfc1fC /HlHqqB8B9my/vP67QMWrBM=
=nVRr
-----END PGP SIGNATURE-----

Quoted 9 lines · Rui

Hi, Angelo,

For a typical ACPI device driver, it usually works like this:
1. Register the ACPI device driver is the module_init
2. Enable the device, create the sys I/F in the drivers .add method.
3. Remove the device I/F in .remove method.
4. unregister the ACPI device driver in module_exit.

With your patch applied, the sys I/F will be created even if no

PNP0C32 device exists in the ACPI namespace. IMO, moving this piece of code to the .add method will be much better.

That's right, but the sys I/F being created are not in a per-device basis. I mean:

Suppose PNP0C32 has three devices:
ABTN
BBTN
CBTN

The sys files are created for PNP0C32 and not for every [ABC]BTN device.
So, moving the
sys I/F code into .add, AFAIK, will get the sys files created several times which is not
the purpose.

The sys files created are "buttons" and "pressed_button". "buttons" displays all devices (buttons) under PNP0C32 for, IMHO, easier enumeration by
userspace; instead of having a file for each device. "pressed_buttons" displays the button used to turn on/resume the computer.

But you are still right, the driver loads successfully (although displaying "none" on "buttons" sys file) when no PNP0C32 is present as we can see on this user post which doesn't have PNP0C32 at all:
http://sourceforge.net/forum/forum.php?thread_id=1959640&forum_id=754264

On the code below, shouldn't (status < 0) be true when no PNP0C32 is present?

        /* ACPI driver register */
        status = acpi_bus_register_driver(&quickstart_acpi_driver);
        if (status < 0)
                return -ENODEV;

This was supposed to exit with -ENODEV before creating sys I/F.

Hah, I nearly missed this email.

Angelo,
Can you please re-send the refreshed patch so that I can add the comments in-line? Better kick off a new thread with the title like "[PATCH] ACPI: ..."

BTW: please don't send it as an attachment. :)

Thanks
rui

Quoted 167 lines

-----Original Message-----
From: Ângelo Miguel Arrifano [mailto:[email protected]]
Sent: Monday, March 10, 2008 8:36 PM
To: Zhang, Rui
Cc: linux-acpi
Subject: Re: *NEW* ACPI driver to support App Hot Startup AKA PNP0C32 - please apply

Quoted 1 line · Rui

As there is nothing private, add linux-acpi in the discussion. :)

Ah, sorry. I was distracted.

Quoted 19 lines · Ângelo Miguel Arrifano

Quoted 18 lines

Quoted 9 lines · Rui

Hi, Angelo,

For a typical ACPI device driver, it usually works like this:
1. Register the ACPI device driver is the module_init
2. Enable the device, create the sys I/F in the drivers .add method.
3. Remove the device I/F in .remove method.
4. unregister the ACPI device driver in module_exit.

With your patch applied, the sys I/F will be created even if no

PNP0C32 device exists in the ACPI namespace. IMO, moving this piece of code to the .add method will be much better.

That's right, but the sys I/F being created are not in a per-device basis. I mean:

Quoted 6 lines

Quoted 4 lines

Suppose PNP0C32 has three devices:
ABTN
BBTN
CBTN

You mean three devices with _HID(PNP0C32), or three devices enumerated under a PNP0C32 device? could you please attach the acpidump output?

(...)

Device (QBTN)

{
   Name (_HID, EisaId ("PNP0C32"))
   Name (_UID, 0x01)
   Method (_STA, 0, NotSerialized)
   {
	If (LEqual (OSYS, 0x07D6))
	{
	    Return (0x0F)
	}
	Else
	{
	    Return (0x00)
	}
   }

   Method (GHID, 0, NotSerialized)
   {
	If (LEqual (HOTB, 0x04))
	{
	    Notify (QBTN, 0x02)
	    Store (0x00, HOTB)
	}

	Return (Buffer (0x01)
	{
	    /* 0000 */    0x01
	})
   }
}

Device (DBTN)

{
   Name (_HID, EisaId ("PNP0C32"))
   Name (_UID, 0x02)
   Method (_STA, 0, NotSerialized)
   {
	If (LEqual (OSYS, 0x07D6))
	{
	    Return (0x0F)
	}
	Else
	{
	    Return (0x00)
	}
   }

   Method (GHID, 0, NotSerialized)
   {
	If (LEqual (HOTB, 0x05))
	{
	    Notify (DBTN, 0x02)
	    Store (0x00, HOTB)
	}

	Return (Buffer (0x01)
	{
	    /* 0000 */    0x02
	})
   }
}

(...)

So the devices above are those whose I was referring as buttons. The first one is the QuickPlay button the second DVD Button, etc..

Quoted 8 lines

Quoted 6 lines

The sys files are created for PNP0C32 and not for every [ABC]BTN device.
So, moving the
sys I/F code into .add, AFAIK, will get the sys files created several times which is not
the purpose.

.add method is called for each PNP0C32 device,not for each button. so in this case, I think it will only create two sys files...

Yeah, but PNP0C32 devices are the buttons.

Quoted 15 lines

Quoted 15 lines

The sys files created are "buttons" and "pressed_button". "buttons" displays all devices (buttons) under PNP0C32 for, IMHO, easier enumeration by
userspace; instead of having a file for each device. "pressed_buttons" displays the button used to turn on/resume the computer.

But you are still right, the driver loads successfully (although displaying "none" on "buttons" sys file) when no PNP0C32 is present as we can see on this user post which doesn't have PNP0C32 at all:
http://sourceforge.net/forum/forum.php?thread_id=1959640&forum_id=754264

On the code below, shouldn't (status < 0) be true when no PNP0C32 is present?

Quoted 8 lines

Quoted 6 lines

        /* ACPI driver register */
        status = acpi_bus_register_driver(&quickstart_acpi_driver);
        if (status < 0)
                return -ENODEV;

This was supposed to exit with -ENODEV before creating sys I/F.

No, acpi_bus_register_driver returns 0 if no matched device is found, in order to support device hotplug.

Hum.. Then this patch should prevent the module to load if no devices are present.

quickstart.c 2008-03-10 12:27:15.223132836 +0000 renamed from quickstart.c 2008-03-06 00:23:20.000000000 +0000 +1 -1

@@ -362,7 +362,7 @@ static int __init quickstart_init(void)
 
       /* ACPI driver register */
       status = acpi_bus_register_driver(&quickstart_acpi_driver);
-       if (status < 0)
+       if (!quickstart_data.btn_lst || status < 0)
               return -ENODEV;
 
       /* Platform driver register */

Quoted 4 lines

thanks,
rui

Best regards,
--
Angelo Arrifano AKA MiKNiX
CSE Student at UBI, Portugal
Gentoo Linux AMD64 Arch Tester
Linwizard Developer
http://miknix.homelinux.com
PGP Pubkey 0x3D92BB0B