Phantom button presses from trackpad

Date: Sun, 22 Oct 2006 19:09:45 -0400
Mime-Version: 1.0
Content-Type: text/plain; format=flowed

>>>I have an iBook G4 of early 2004, running
-current, but I haven't yet
>>>bothered to update X beyond 4.4.0.  I tried
changing the keyboard
>>>driver from "keyboard" to
"kbd", which gave positive results (more
>>>keys defined, less need for xmodmap), but also
made the trackpad
>>>behave strangely.  Whenever I put my finger on
it, it reports button
>>>179 being pressed:
>>
>>Whatever the cause was, the recent changes to ams.c
fixed it.  The
>>trackpad is now quite usable.  Thanks!

>Good, the point of the change was to make tapping the
pad work as button 
>one.
>About the phantom buttons - all Apple trackpads I know
report more than one 
>button ( two in the PB3400c, four in my iBook G4 ) - the
only one that's 
>corresponding to a real life button is button one. The
others are (ab)used 
>for stuff like signalling that there's a finger on the
pad ( the two-button 
>version uses button 2-4 for that, the four-button
version uses button 5 and 
>reports button four as always down).
>The button count reported is bogus and ought to be read
as protocol 
>revision or something.

I've also experienced the button 179 problem on my trusty
old clamshell
iBook SE, running 3.0.1 at present.  I grabbed v. 1.20 of
ams.c and 1.5
of amsvar.h and applied them against the 3.0 branch.  (I
just had to
change ams_ioctl() back to use the proc rather than lwp
struct, as
altered in v. 1.18 of ams.c.)  I assume that was safe; it
worked for me.

After the change, the trackpad is now completely usable. 
One thing
though -- not that it bothers me -- Michael, you've noted
the intent
was to make tapping the pad work as button one.  That
doesn't happen
for me.  As above, before the change if I used the trackpad
with xev(1)
it would report button 179 as soon as I touched it.  Now it
never
reports any button events when I touch it, whether that's a
tap or
something more prolonged.  Like I said, no worries for me, I
don't use
the trackpad to click (in fact, I had no idea one could do
that), but
thought I'd mention it.  (Perhaps it's something to do with
the older
branch I'm using?)

>So, has anyone ever seen an ADB trackpad that reports
something else than 
>two or four buttons?

My iBook reports two buttons.

Dave

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

Hello,

On Oct 22, 2006, at 19:09, David H. Gutteridge wrote:

>> Good, the point of the change was to make tapping
the pad work as 
>> button
>> one.
>> About the phantom buttons - all Apple trackpads I
know report more 
>> than one
>> button ( two in the PB3400c, four in my iBook G4 )
- the only one 
>> that's
>> corresponding to a real life button is button one.
The others are 
>> (ab)used
>> for stuff like signalling that there's a finger on
the pad ( the 
>> two-button
>> version uses button 2-4 for that, the four-button
version uses button 
>> 5 and
>> reports button four as always down).
>> The button count reported is bogus and ought to be
read as protocol
>> revision or something.
>
> I've also experienced the button 179 problem on my
trusty old clamshell
> iBook SE, running 3.0.1 at present.  I grabbed v. 1.20
of ams.c and 1.5
> of amsvar.h and applied them against the 3.0 branch. 
(I just had to
> change ams_ioctl() back to use the proc rather than lwp
struct, as
> altered in v. 1.18 of ams.c.)  I assume that was safe;
it worked for 
> me.

The code hasn't been touched for ages, it would probably
work with 2.0 
or even 1.6.

> After the change, the trackpad is now completely
usable.  One thing
> though -- not that it bothers me -- Michael, you've
noted the intent
> was to make tapping the pad work as button one.  That
doesn't happen
> for me.

So your pad appears to use yet another different protocol.

> As above, before the change if I used the trackpad with
xev(1)
> it would report button 179 as soon as I touched it.

Would it report /another/ button when you take the finger
from the pad? 
Or would it just report a button-up event for 179?

> Now it never reports any button events when I touch it,
whether that's 
> a tap or
> something more prolonged.  Like I said, no worries for
me, I don't use
> the trackpad to click (in fact, I had no idea one could
do that), but
> thought I'd mention it.  (Perhaps it's something to do
with the older
> branch I'm using?)

As far as I can tell your pad just uses another variant of
the 
protocol. Now we need to figure out how to recognize this
revision and 
which phantom buttons it uses how.

Would it be ok if I just send you a patch that reports
button events on 
the ADB level and you tell me what kind of events you get
when putting 
a finger on the pad, move it, lift it up again?

>> So, has anyone ever seen an ADB trackpad that
reports something else 
>> than
>> two or four buttons?
>
> My iBook reports two buttons.

It's listed ad EMP trackpad in dmesg, isn't it?

have fun
Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (Darwin)

iQEVAwUBRTwhWspnzkX8Yg2nAQJm0wf+P9unZjYbgYt6RyFvil+MUd1YCQCD
kTRJ
kjSgqirjg8VSgH+HucfIbIC/Lvf5eupWlHR/l2ENbP1EQ0Aml61RVTk53Pwe
sc3H
+09Wa6o3dTyPC5RQtwFRMGI5gfUUE3rWc6koCoSr1qvn1A+VSFmhOOc3AiTQ
F3Hq
e/Hz5URkdJm+LHoN1SSxuKy9EF5C3Y6lcA8dLRz9hRQPtYAiyq9uKjMTr3GH
Qvxm
4FcEoliHNCXgjq/hiDXS6O2l062NwhxxqrzZOu9BfCWe2lfoj8yQcD2Jheye
SBwF
Kr6yhEjFypKRVHBM5oSokKigdgPJ3kULZB7NnFRKc3ksEJaGGhEjxA==
=MAQJ
-----END PGP SIGNATURE-----

>>After the change, the trackpad is now completely
usable.  One thing
>>though -- not that it bothers me -- Michael, you've
noted the intent
>>was to make tapping the pad work as button one. 
That doesn't happen
>>for me.

>So your pad appears to use yet another different
protocol.

>>As above, before the change if I used the trackpad
with xev(1)
>>it would report button 179 as soon as I touched it.

>Would it report /another/ button when you take the
finger from the pad?
>Or would it just report a button-up event for 179?

It would report a button-up event for 179.

>>Now it never reports any button events when I touch
it, whether that's
>>a tap or something more prolonged.  Like I said, no
worries for me,
>>I don't use the trackpad to click (in fact, I had no
idea one could
>>do that), but thought I'd mention it.  (Perhaps it's
something to do
>>with the older branch I'm using?)

>As far as I can tell your pad just uses another variant
of the protocol.
>Now we need to figure out how to recognize this revision
and which
>phantom buttons it uses how.

>Would it be ok if I just send you a patch that reports
button events on
>the ADB level and you tell me what kind of events you
get when putting
>a finger on the pad, move it, lift it up again?

Sure, absolutely.

>>My iBook reports two buttons.

>It's listed ad EMP trackpad in dmesg, isn't it?

Yes, the dmesg says:

ams0 at adb0 addr 3: EMP trackpad <tpad> 2-button, 400
dpi

Dave