Recently, I got a new Macbook Pro, so I decided to give OpenSolaris
2008.11 a go on the old one. The old one is a 15" Core Duo (32bit),
first-generation MacBook Pro (2006).
OpenSolaris installed just fine. Most things work. The internal
keyboard, wireless (had to add an alias for ath, but it works ok), etc.
One thing that doesn't work is the internal mouse/trackpad.
From what I've read, in default mode, the trackpad should just work as
a normal USB mouse. I tried booting a Linux (OpenSUSE) live CD, and it
works fine there, except that Linux has a special 'appletouch' driver,
that puts the trackpad in a special, apple internal, mode (report id
0x5), so that more detailed information is given, and you can do things
like tap on the trackpad with two or three fingers.
So, to see if it will work as a normal USB mouse, I booted a NetBSD live
CD, since I know that NetBSD will treat it as a normal USB mouse. And,
sure enough, it worked fine.
However, no such luck under Solaris. I am unable to generate any
movement data moving my finger over the trackpad, and the mousebutton
does generate a click, but also generates +1/+1 x/y movement, which
isn't right.
I looked at the HID report structures, and they look normal. When I dump
the raw data from /dev/usb/hid?, I see the following:
* 4-byte messages: a report id of 0x2, and then 3 bytes of data. That's
consistent with the HID report structure, which says 5x1+3x1+2x8 == 24 bits.
* A continuous stream of messages with all zeros for the three byte data
when there's no activity. A comment in the Linux appletouch driver says
that the device tends to do that. While this is annoying, it shouldn't
stop the driver from working, since usbms ignores those messages.
* When the button is pressed, I see the following messages:
0x02 0x00 0x00 0x00
<press button>
0x02 0x00 0x00 0x01
0x02 0x00 0x01 0x01
0x02 0x00 0x01 0x01
0x02 0x01 0x01 0x01
<release button>
0x02 0x00 0x01 0x01
0x02 0x00 0x01 0x01
0x02 0x00 0x00 0x01
0x02 0x00 0x00 0x00
That's weird. The button press/release seems to be reflected correctly,
but then the data reflects +1 X and +1 Y movement, which isn't there at all.
And, again, moving my finger over the trackpad doesn't generate any data
(well, no data that isn't all zeros, anyway; it might be generating all
zero data, but that gets lost in all the other zero data packets).
Since NetBSD works fine with the device in plain USB mouse mode, there
must be something that the Solaris USB code does wrong. The problem does
not seem to be usbms itself; it is taking the correct actions based on
bad data passed to it from below. So, the hid code must be at fault here
somehow.
Ill post the relevant parts of the ::prtusb -t output below.
The confusing part is that it has two "mouse" instances. One comes with
the keyboard, and appears to be the one that can be switched to use
report id 5, with full tap/finger support. The other one looks like a
plain mouse.
- Frank
=====
mouse, instance #5 (driver name: hid)
/pci at 0,0/pci8086,7270 at 1d/device at 2/mouse at 1
dip: 0xd1baa170
hid_statep: 0xd3deb480
hid_errlevel: 0x4
usage page ( 0x01 )
usage ( 0x02 )
COLLECTION ( 0x01 )
usage page ( 0x01 )
usage ( 0x01 )
COLLECTION ( 0x00 )
usage page ( 0x09 )
logical minimum ( 0x00 )
logical maximum ( 0x01 )
report id ( 0x02 )
report count ( 0x03 )
report size ( 0x01 )
usage min ( 0x01 )
usage max ( 0x03 )
INPUT ( 0x02 )
usage page ( 0x09 )
logical minimum ( 0x00 )
logical maximum ( 0x01 )
report id ( 0x02 )
report count ( 0x01 )
report size ( 0x05 )
INPUT ( 0x01 )
report id ( 0x02 )
usage page ( 0x01 )
logical minimum ( 0x81 )
logical maximum ( 0x7f )
report size ( 0x08 )
report count ( 0x02 )
usage ( 0x30 )
usage ( 0x31 )
INPUT ( 0x06 )
END_COLLECTION ( )
END_COLLECTION ( )
report id ( 0x02 )
logical minimum ( 0x81 )
logical maximum ( 0x7f )
report size ( 0x08 )
report count ( 0x02 )
usage page ( 0xff )
usage ( 0x01 )
COLLECTION ( 0x00 )
usage page ( 0xff )
logical minimum ( 0x01 )
logical maximum ( 0x41 )
physical minimum ( 0x00 )
physical maximum ( 0xff )
report id ( 0x05 )
report size ( 0x08 )
report count ( 0x3f )
usage ( 0x02 )
INPUT ( 0x00 )
usage page ( 0xff )
physical minimum ( 0x00 )
physical maximum ( 0xff )
report id ( 0x05 )
logical minimum ( 0x80 )
logical maximum ( 0x7f )
report size ( 0x08 )
report count ( 0x10 )
usage ( 0x04 )
FEATURE ( 0x02 )
END_COLLECTION ( )
input, instance #6 (driver name: hid)
/pci at 0,0/pci8086,7270 at 1d/device at 2/input at 2
dip: 0xd1baa000
hid_statep: 0xdefb8d40
hid_errlevel: 0x4
usage page ( 0x0c )
usage ( 0x01 )
COLLECTION ( 0x01 )
usage page ( 0x0c )
report size ( 0x08 )
report count ( 0x01 )
logical minimum ( 0x00 )
logical maximum ( 0xff 0x00 )
usage ( 0xb8 )
INPUT ( 0x02 )
END_COLLECTION ( )
}
[...]
mouse, instance #1 (driver name: hid)
/pci at 0,0/pci8086,7270 at 1d,3/device at 1/mouse at 1
dip: 0xd1baa450
hid_statep: 0xd3465340
hid_errlevel: 0x4
usage page ( 0x01 )
usage ( 0x02 )
COLLECTION ( 0x01 )
usage page ( 0x01 )
usage ( 0x01 )
COLLECTION ( 0x00 )
usage page ( 0x09 )
logical minimum ( 0x00 )
logical maximum ( 0x01 )
report count ( 0x05 )
report size ( 0x01 )
usage min ( 0x01 )
usage max ( 0x05 )
INPUT ( 0x02 )
usage page ( 0x09 )
logical minimum ( 0x00 )
logical maximum ( 0x01 )
report count ( 0x01 )
report size ( 0x03 )
INPUT ( 0x03 )
usage page ( 0x01 )
logical minimum ( 0x81 )
logical maximum ( 0x7f )
report size ( 0x08 )
report count ( 0x02 )
usage ( 0x30 )
usage ( 0x31 )
INPUT ( 0x06 )
logical minimum ( 0x81 )
logical maximum ( 0x7f )
usage page ( 0x00 0xff )
report count ( 0x01 )
report size ( 0x08 )
usage ( 0x01 )
FEATURE ( 0x02 )
END_COLLECTION ( )
END_COLLECTION ( )
}
