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

From: Dmitry Torokhov
Date: Tue Mar 09 2010 - 03:06:34 EST


Hi Micki,

On Mon, Mar 08, 2010 at 03:11:08PM +0200, Micki Balanga wrote:
>
>
> -----Original Message-----
> From: Rafi Rubin [mailto:rafi@xxxxxxxxxxxxxx]
> Sent: Monday, March 08, 2010 3:34 AM
> To: Micki Balanga
> Cc: jkosina@xxxxxxx; chatty@xxxxxxx; peterhuewe@xxxxxx;
> linux-input@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> 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
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/