Re: [PATCH] USB: N-trig Finger Pen Multitouch fix

Re: [PATCH] USB: N-trig Finger Pen Multitouch fix

[Rafi Rubin]

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

First let me thank you for contributing to the development of this driver.


I suggest you use scripts/checkpatch.pl it will catch a number of style issues.
 Dmitry was kind enough to point it out when I sent in my recent set of patches.


Mostly it looks like this patch is the delta from an almost completely
independently developed driver.  That makes it considerably more work to analyze
the fundamental differences.  I would have preferred to see a set of patches
which first mutate the existing code to match your naming conventions (if that
is so desired) and then patches that show how things are actually different (A
lesson I've learned the hard way).

I see quite a number of changes that handle the protocol better than my version,
would you be offended if I begin to absorb those and submit joint patches?  I'm
not sure about the proper etiquette for that.


On 03/07/10 14:04, Micki Balanga wrote:
> The purpose of this patch is to enable Linux users to experience the
> complete N-trig DuoSense pen and true multi-touch solution capabilities
> over a Linux platform. This driver has been carefully and thoroughly
> tested by N-trig, over a period of several months.

What user space tools have you tested this driver with?  It does not seem to
work well with the wacom and evdev drivers.

What improvements beyond the driver that's in 2.6.33 are you targeting?


> + * The driver will support MTM firwmare Pen, Finger (Up to 6)

Is this code intended to support older firmwares as well?  It would be best not
to break the driver for users with legacy firmwares without reason.


> +#define PEN_REPORT_SIZE                0x48
> +#define MAX_FINGERS_SUPPORT    0x06

Are these guaranteed to be kept constant for the foreseeable future?  I've
already seen hints that N-Trig may move to 10+ fingers in the not to distant future.

http://www.n-trig.com/Content.aspx?Page=PressReleases&PressReleaseId=541

I was starting to move to non-fixed finger limits before I pulled the finger
tracking code.

> -       if (field->application == HID_DG_PEN)
> -               return 0;

> +               if (PEN_REPORT_SIZE == field->report->size) {

What was wrong with using field->application to detect the pen?  On a side note,
after splitting the touch and pen events to separate input devices, I found that
the pen events conform quite well to the hid spec and don't need special
handling.  Is there something I missed?



> +#define MODULE_NAME "hid_ntrig"
> +
> +
> +#define info(format, arg...) \
> +       printk(KERN_INFO "%s: " format , MODULE_NAME , ## arg)
> +#define ntrig_dbg(format, arg...) \
> +       do { \
> +               if (debug) \
> +                       printk(KERN_DEBUG "%s: " format, \
> +                       MODULE_NAME , ## arg); \
> +       } while (0)

>  static struct hid_driver ntrig_driver = {
>         .name = "ntrig",

Might want to use the #define for that too.


> +                       case HID_DG_CONTACTCOUNT:
> +                               nt_rpt->real_fingers = value;
> +                               nt_rpt->fake_fingers =
> MAX_FINGERS_SUPPORT - value;

> +                                               if (MAX_FINGERS_SUPPORT
> == nt_rpt->fake_fingers--) {

Would you care to explain "fake_fingers"?  I'm not really seeing why it helps
when you already have real_fingers.  It looks like you use that to detect some
ghost events, but really again, fingers should be fine at that point.

That does remind me that I pulled that special case when I cleaned up my last
patch set and forgot to put it back.

Are the fake finger events intentional or just a bug?

> -               case HID_DG_PEN:
> -                       input->name = "N-Trig Pen";
> -                       break;
> -               case HID_DG_TOUCHSCREEN:
> -                       input->name =
> -                               (hidinput->report->field[0]
> -                                ->physical) ?
> -                               "N-Trig Touchscreen" :
> -                               "N-Trig MultiTouch";

Do you have something against the device naming, or did you delete that
unintentionally?

> +                     /*
> +                      * These command will implement later accroding to
> demand
> +                      */
> +               case REPORTID_SET_MODE_DUAL:    /* FALLTHRU */
> +               case REPORTID_CALIBRATION:      /* FALLTHRU */
> +               case REPORTID_GET_VERSION:      /* FALLTHRU */
> +               case REPORTID_GET_MODE:         /* FALLTHRU */
> +               case REPORTID_SET_MODE_PEN:     /* FALLTHRU */
> +               case REPORTID_SET_MODE_TOUCH:   /* FALLTHRU */
> +               case REPORTID_CALIBRATION_RESPOND:/* FALLTHRU */
> +               case HID_CAPACITORS_CALIB:      /* FALLTHRU */
> +               case HID_GET_CAPACITORS_CALIB_DONE:

Getting the firmware version would be my top priority, its been on my list of
things to write for months.  I'd like to see the firmware version printed when
the module is loaded and available through sysfs.

I've heard a few people complain about calibration, but most users, including
myself, have not seen any calibration issues.

Any chance of getting a tool to install firmware updates?



Rafi
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkuUVBkACgkQwuRiAT9o60+htACgt+S+ON/9GWwXP+dliEU9GdQN
TaoAoPUW4Ndlfg0MAWnoEcg4H1ZHU1VO
=NQkb
-----END PGP SIGNATURE-----

Re: [PATCH] USB: N-trig Finger Pen Multitouch fix

[Rafi Rubin]

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

One other note:

> +                                               input_report_abs(input,
> ABS_MT_TRACKING_ID, nt_rpt->finger_id);

I pulled the tracking id to conform to
	Documentation/input/multi-touch-protocol.txt
which states that unreliable tracking ids should not be passed to the user
space.  So unless you have done something to fix the broken tracking that I've
been seeing, we should leave it out for now.

Rafi
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkuUVdIACgkQwuRiAT9o608ymwCgpGXXnsWcEd2C+WaBxVpnla/e
xYgAoK68Yyvl2RLzqhWVGk9tKOFpsEwI
=O+8/
-----END PGP SIGNATURE-----

Re: [PATCH] USB: N-trig Finger Pen Multitouch fix

[Stéphane Chatty]

Le 8 mars 10 à 02:34, Rafi Rubin a écrit :

> (...)

> I see quite a number of changes that handle the protocol better  
> than my version,
> would you be offended if I begin to absorb those and submit joint  
> patches?  I'm
> not sure about the proper etiquette for that.

I would support this option.

St.

--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] USB: N-trig Finger Pen Multitouch fix

[Jiri Kosina]

On Sun, 7 Mar 2010, Micki Balanga wrote:

> The purpose of this patch is to enable Linux users to experience the 
> complete N-trig DuoSense pen and true multi-touch solution capabilities 
> over a Linux platform. This driver has been carefully and thoroughly 
> tested by N-trig, over a period of several months.
> 
> N-trig is changing the way people interact with computers by providing a 
> dual-mode pen and true multi-touch input device, specifically designed 
> for today's advanced computing world. N-trig DuoSense® solution provides 
> a real Hands-on computing® experience, and sets the stage for OEMs and 
> ISVs to introduce innovative computer products and applications for an 
> intuitive, Hands-on® experience directly onscreen. DuoSense digitizers 
> are easily integrated into existing technologies, support all LCDs, keep 
> devices slim and light, and can be implemented in a broad range of 
> products, ranging from small notebooks to large LCDs. N-trig has offices 
> in Israel, the US, Taiwan and Japan.
> 
> Signed-off-by: Micki Balanga <micki@n-trig.com>

Thanks for the patch.

As you are very well aware, we already have n-trig driver written by Rafi 
Rubin. I understand that your patch is implementing new features that 
Rafi's patch is not able to handle, and also re-factoring the driver in a 
non-trivial way.

Therefore it would be nice, if you could separate out the individual 
changes/additions to be based on top of the driver that we already have, 
and accompany them with nice changelog describing what changes are being 
done and why 

Or even Rafi volunteered to do this, as I can see -- thanks for that! It 
would still be nice to get Acked-by from the N-Trig guys for the changes 
at least.

Thanks,

-- 
Jiri Kosina
SUSE Labs, Novell Inc.
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [PATCH] USB: N-trig Finger Pen Multitouch fix

[Micki Balanga]

-----Original Message-----
From: Rafi Rubin [mailto:rafi@seas.upenn.edu] 
Sent: Monday, March 08, 2010 3:34 AM
To: Micki Balanga
Cc: jkosina@suse.cz; chatty@enac.fr; peterhuewe@gmx.de;
linux-input@vger.kernel.org; linux-kernel@vger.kernel.org
Subject: Re: [PATCH] USB: N-trig Finger Pen Multitouch fix

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

First let me thank you for contributing to the development of this
driver.


I suggest you use scripts/checkpatch.pl it will catch a number of style
issues.
 Dmitry was kind enough to point it out when I sent in my recent set of
patches.

[Micki] I used the script to check the patch

Mostly it looks like this patch is the delta from an almost completely
independently developed driver.  That makes it considerably more work to
analyze
the fundamental differences.  I would have preferred to see a set of
patches
which first mutate the existing code to match your naming conventions
(if that
is so desired) and then patches that show how things are actually
different (A
lesson I've learned the hard way).

I see quite a number of changes that handle the protocol better than my
version,
would you be offended if I begin to absorb those and submit joint
patches?  I'm
not sure about the proper etiquette for that.

[Micki] As mentioned in my mail we have tested the driver with
application develop by N-trig. 
[Micki] Our user space application rely on these events, so if you want
to make any change I will be happy if you 
[Micki] consult me.

On 03/07/10 14:04, Micki Balanga wrote:
> The purpose of this patch is to enable Linux users to experience the
> complete N-trig DuoSense pen and true multi-touch solution
capabilities
> over a Linux platform. This driver has been carefully and thoroughly
> tested by N-trig, over a period of several months.

What user space tools have you tested this driver with?  It does not
seem to
work well with the wacom and evdev drivers.

[Micki] We will put on N-trig web site a DEB package

What improvements beyond the driver that's in 2.6.33 are you targeting?

[Micki] We will update the driver and user space application with new
features.

> + * The driver will support MTM firwmare Pen, Finger (Up to 6)

Is this code intended to support older firmwares as well?  It would be
best not
to break the driver for users with legacy firmwares without reason.

[Micki] There will be a DEB package for firmware upgrade.

> +#define PEN_REPORT_SIZE                0x48
> +#define MAX_FINGERS_SUPPORT    0x06

Are these guaranteed to be kept constant for the foreseeable future?
I've
already seen hints that N-Trig may move to 10+ fingers in the not to
distant future.

http://www.n-trig.com/Content.aspx?Page=PressReleases&PressReleaseId=541

I was starting to move to non-fixed finger limits before I pulled the
finger
tracking code.

> -       if (field->application == HID_DG_PEN)
> -               return 0;

> +               if (PEN_REPORT_SIZE == field->report->size) {

What was wrong with using field->application to detect the pen?  On a
side note,
after splitting the touch and pen events to separate input devices, I
found that
the pen events conform quite well to the hid spec and don't need special
handling.  Is there something I missed?

[Micki] Report Pen is constant , but I will look into it, as you
mentioned in your mail 

> +#define MODULE_NAME "hid_ntrig"
> +
> +
> +#define info(format, arg...) \
> +       printk(KERN_INFO "%s: " format , MODULE_NAME , ## arg)
> +#define ntrig_dbg(format, arg...) \
> +       do { \
> +               if (debug) \
> +                       printk(KERN_DEBUG "%s: " format, \
> +                       MODULE_NAME , ## arg); \
> +       } while (0)

>  static struct hid_driver ntrig_driver = {
>         .name = "ntrig",

Might want to use the #define for that too.

[Micki] : Not A Problem I will change it. 


> +                       case HID_DG_CONTACTCOUNT:
> +                               nt_rpt->real_fingers = value;
> +                               nt_rpt->fake_fingers =
> MAX_FINGERS_SUPPORT - value;

> +                                               if
(MAX_FINGERS_SUPPORT
> == nt_rpt->fake_fingers--) {

Would you care to explain "fake_fingers"?  I'm not really seeing why it
helps
when you already have real_fingers.  It looks like you use that to
detect some
ghost events, but really again, fingers should be fine at that point.

That does remind me that I pulled that special case when I cleaned up my
last
patch set and forgot to put it back.

Are the fake finger events intentional or just a bug?
[Micki] As I mentioned before we developed a user space application
which analyze the data 
        Transferred from driver and the driver need to report about fake
fingers.

> -               case HID_DG_PEN:
> -                       input->name = "N-Trig Pen";
> -                       break;
> -               case HID_DG_TOUCHSCREEN:
> -                       input->name =
> -                               (hidinput->report->field[0]
> -                                ->physical) ?
> -                               "N-Trig Touchscreen" :
> -                               "N-Trig MultiTouch";

Do you have something against the device naming, or did you delete that
unintentionally?
[Micki] When I download a git branch It didn't include this change, I
will update the patch with this new change
> +                     /*
> +                      * These command will implement later accroding
to
> demand
> +                      */
> +               case REPORTID_SET_MODE_DUAL:    /* FALLTHRU */
> +               case REPORTID_CALIBRATION:      /* FALLTHRU */
> +               case REPORTID_GET_VERSION:      /* FALLTHRU */
> +               case REPORTID_GET_MODE:         /* FALLTHRU */
> +               case REPORTID_SET_MODE_PEN:     /* FALLTHRU */
> +               case REPORTID_SET_MODE_TOUCH:   /* FALLTHRU */
> +               case REPORTID_CALIBRATION_RESPOND:/* FALLTHRU */
> +               case HID_CAPACITORS_CALIB:      /* FALLTHRU */
> +               case HID_GET_CAPACITORS_CALIB_DONE:

Getting the firmware version would be my top priority, its been on my
list of
things to write for months.  I'd like to see the firmware version
printed when
the module is loaded and available through sysfs.

[Micki] Later version we will support Get Version Number

I've heard a few people complain about calibration, but most users,
including
myself, have not seen any calibration issues.

Any chance of getting a tool to install firmware updates?

[Micki] Later There will be a DEB package on the site you can download
and make firmware updates.
[Micki] I will update The patch with your new version and will send it
again.


Rafi
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkuUVBkACgkQwuRiAT9o60+htACgt+S+ON/9GWwXP+dliEU9GdQN
TaoAoPUW4Ndlfg0MAWnoEcg4H1ZHU1VO
=NQkb
-----END PGP SIGNATURE-----

Re: [PATCH] USB: N-trig Finger Pen Multitouch fix

[Dmitry Torokhov]

Hi Micki,

On Mon, Mar 08, 2010 at 03:11:08PM +0200, Micki Balanga wrote:
> 
> 
> -----Original Message-----
> From: Rafi Rubin [mailto:rafi@seas.upenn.edu] 
> Sent: Monday, March 08, 2010 3:34 AM
> To: Micki Balanga
> Cc: jkosina@suse.cz; chatty@enac.fr; peterhuewe@gmx.de;
> linux-input@vger.kernel.org; linux-kernel@vger.kernel.org
> Subject: Re: [PATCH] USB: N-trig Finger Pen Multitouch fix
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> First let me thank you for contributing to the development of this
> driver.
> 
> 
> I suggest you use scripts/checkpatch.pl it will catch a number of style
> issues.
>  Dmitry was kind enough to point it out when I sent in my recent set of
> patches.
> 
> [Micki] I used the script to check the patch
> 
> Mostly it looks like this patch is the delta from an almost completely
> independently developed driver.  That makes it considerably more work to
> analyze
> the fundamental differences.  I would have preferred to see a set of
> patches
> which first mutate the existing code to match your naming conventions
> (if that
> is so desired) and then patches that show how things are actually
> different (A
> lesson I've learned the hard way).
> 
> I see quite a number of changes that handle the protocol better than my
> version,
> would you be offended if I begin to absorb those and submit joint
> patches?  I'm
> not sure about the proper etiquette for that.
> 
> [Micki] As mentioned in my mail we have tested the driver with
> application develop by N-trig. 
> [Micki] Our user space application rely on these events, so if you want
> to make any change I will be happy if you 
> [Micki] consult me.
> 

No, this is not how it works. There are already existing userspace
drivers/applications working with current n-trig driver, the fact that
your changes happen to work with your userspace is not enough.


> On 03/07/10 14:04, Micki Balanga wrote:
> > The purpose of this patch is to enable Linux users to experience the
> > complete N-trig DuoSense pen and true multi-touch solution
> capabilities
> > over a Linux platform. This driver has been carefully and thoroughly
> > tested by N-trig, over a period of several months.
> 
> What user space tools have you tested this driver with?  It does not
> seem to
> work well with the wacom and evdev drivers.
> 
> [Micki] We will put on N-trig web site a DEB package
> 

What about distributions not user deb?

> What improvements beyond the driver that's in 2.6.33 are you targeting?
> 
> [Micki] We will update the driver and user space application with new
> features.

What about other applications? You are not working in a vacuum. Please
consider working with X evdev and wacom developers - I am pretty sure
distributions would prefer a single userspace component supporting wide
range of multitouch devieces.  Consider how Synaptics X driver
(coincidentially not developed by Synaptisc) happens to support pretty
much every non-multitouch touchpad that is supported by Linux.

> 
> > + * The driver will support MTM firwmare Pen, Finger (Up to 6)
> 
> Is this code intended to support older firmwares as well?  It would be
> best not
> to break the driver for users with legacy firmwares without reason.
> 
> [Micki] There will be a DEB package for firmware upgrade.

You surely do not expect users to rush and upgrade their fimwares
en-masse? The dirver should support devices with both older and newer
firmwares (obviously devices with newer firmware will have wider set of
features).

> 
> > +#define PEN_REPORT_SIZE                0x48
> > +#define MAX_FINGERS_SUPPORT    0x06
> 
> Are these guaranteed to be kept constant for the foreseeable future?
> I've
> already seen hints that N-Trig may move to 10+ fingers in the not to
> distant future.
> 
> http://www.n-trig.com/Content.aspx?Page=PressReleases&PressReleaseId=541
> 
> I was starting to move to non-fixed finger limits before I pulled the
> finger
> tracking code.
> 
> > -       if (field->application == HID_DG_PEN)
> > -               return 0;
> 
> > +               if (PEN_REPORT_SIZE == field->report->size) {
> 
> What was wrong with using field->application to detect the pen?  On a
> side note,
> after splitting the touch and pen events to separate input devices, I
> found that
> the pen events conform quite well to the hid spec and don't need special
> handling.  Is there something I missed?
> 
> [Micki] Report Pen is constant , but I will look into it, as you
> mentioned in your mail 
> 
> > +#define MODULE_NAME "hid_ntrig"
> > +
> > +
> > +#define info(format, arg...) \
> > +       printk(KERN_INFO "%s: " format , MODULE_NAME , ## arg)
> > +#define ntrig_dbg(format, arg...) \
> > +       do { \
> > +               if (debug) \
> > +                       printk(KERN_DEBUG "%s: " format, \
> > +                       MODULE_NAME , ## arg); \
> > +       } while (0)
> 
> >  static struct hid_driver ntrig_driver = {
> >         .name = "ntrig",
> 
> Might want to use the #define for that too.
> 
> [Micki] : Not A Problem I will change it. 
> 
> 
> > +                       case HID_DG_CONTACTCOUNT:
> > +                               nt_rpt->real_fingers = value;
> > +                               nt_rpt->fake_fingers =
> > MAX_FINGERS_SUPPORT - value;
> 
> > +                                               if
> (MAX_FINGERS_SUPPORT
> > == nt_rpt->fake_fingers--) {
> 
> Would you care to explain "fake_fingers"?  I'm not really seeing why it
> helps
> when you already have real_fingers.  It looks like you use that to
> detect some
> ghost events, but really again, fingers should be fine at that point.
> 
> That does remind me that I pulled that special case when I cleaned up my
> last
> patch set and forgot to put it back.
> 
> Are the fake finger events intentional or just a bug?
> [Micki] As I mentioned before we developed a user space application
> which analyze the data 
>         Transferred from driver and the driver need to report about fake
> fingers.

This is not a technical argument ;(

-- 
Dmitry

RE: [PATCH] USB: N-trig Finger Pen Multitouch fix

[Jiri Kosina]

On Mon, 8 Mar 2010, Micki Balanga wrote:

> I see quite a number of changes that handle the protocol better than my 
> version, would you be offended if I begin to absorb those and submit 
> joint patches?  I'm not sure about the proper etiquette for that.
> 
> [Micki] As mentioned in my mail we have tested the driver with
> application develop by N-trig. 
> [Micki] Our user space application rely on these events, so if you want
> to make any change I will be happy if you 
> [Micki] consult me.

Please keep in mind that we don't want to break backwards compatibility 
for existing users of the current in-kernel n-trig driver.

> What user space tools have you tested this driver with?  It does not 
> seem to work well with the wacom and evdev drivers.
> 
> [Micki] We will put on N-trig web site a DEB package

Distributions based on Debian packaging system are not the only existing 
ones. What's usually better is to release source code package, so that 
distributions could eventually package it themselves for their users, if 
applicable.

> What improvements beyond the driver that's in 2.6.33 are you targeting?
> 
> [Micki] We will update the driver and user space application with new
> features.
>
> > + * The driver will support MTM firwmare Pen, Finger (Up to 6)
> 
> Is this code intended to support older firmwares as well?  It would be 
> best not to break the driver for users with legacy firmwares without 
> reason.
> 
> [Micki] There will be a DEB package for firmware upgrade.

The in-kernel driver needs to support both older and newer firmware at the 
same time, otherwise it would bring a lot of pain to users.

> > +                       case HID_DG_CONTACTCOUNT:
> > +                               nt_rpt->real_fingers = value;
> > +                               nt_rpt->fake_fingers =
> > MAX_FINGERS_SUPPORT - value;
> 
> > +                                               if
> (MAX_FINGERS_SUPPORT
> > == nt_rpt->fake_fingers--) {
> 
> Would you care to explain "fake_fingers"?  I'm not really seeing why it
> helps
> when you already have real_fingers.  It looks like you use that to
> detect some
> ghost events, but really again, fingers should be fine at that point.
> 
> That does remind me that I pulled that special case when I cleaned up my
> last
> patch set and forgot to put it back.
> 
> Are the fake finger events intentional or just a bug?
> [Micki] As I mentioned before we developed a user space application
> which analyze the data 
>         Transferred from driver and the driver need to report about fake
> fingers.

I am afraid this will need much more explanation.

Also, have you considered wrokign with X evdev people, so that ideally no 
other userspace software would eventually be needed?

Thanks,

-- 
Jiri Kosina
SUSE Labs, Novell Inc.