Hi Jason, On Wed, Jan 23, 2013 at 6:41 AM, Jason Flatt <jflatt@xxxxxxx> wrote: > I am trying to add the newer Sony Google TV remote, and am looking for some > advice on how to get this implemented. This is my first experience with hid > drivers and multitouch. > > Some base info: > Bluetooth only > Keyboard > Multitouch trackpad > Accelerometer > http://discover.store.sony.com/googletv/#remote Nice, seems pretty interesting. > > Below is the original report descriptor with my annotations. In kernel 3.6.11 > the keyboard works great, but the trackpad and accelerometer don't work out of > the box. I can get the trackpad to work as a one-button mouse by rewriting > the report descriptor in a custom hid device module. Ok, I have several comments on the report descriptor. See later. > > I am thinking the 0xff01 usage page is what keeps it from being detected as a > pointing device to begin with, requiring a rewrite of the report descriptor. > Is this kind of approach frowned upon? You won't be able to activate both touches with just a rewrite of the report descriptor. The protocol in use here is not standard (this is why they put the vendor usage page). > > There are some quirks that I believe might make this device unusual compared > to others. The entire trackpad surface acts as a mechanical button, i.e. the > trackpad physically pushes into the remote body, so using tapping movements > may be confusing to the user. Some synaptics devices and all the new trackpads from apple are using this kind of "clickpads". So it's not new here, and distributions already now how to handle them. > The remote also has a 'drag' button that acts > like a mouse button but is only active when held down and there is trackpad > motion. I think it was intended for one-handed dragging, so would have to > utilize some kind of dragging-on/dragging-off state. I understand the > individual mouse reports, which always contain X and Y absolute values for > both contact points. There are two values that seem to be some sort of > contact identifier, we'll call them a and b. With no fingers touching, a=0, > b=0. With just the first finger touching, a=1, b=0. With both the first and > second finger touching, a=2, b=2. Then, with just the second finger touching, > a=0, b=1. Has this been seen before? Would that be used to translate into a > contact identifier? definitively not a contact identifier. It could be translated to a contact count. But as it's a proprietary protocol, you can rely on it to know which contact is active and which is not. > > The accelerometer doesn't seem to show any events, I saw the sixaxis driver > needs some kind of set_operational command before it begins working, I'd have > to investigate if that sort of thing is necessary for this remote control. If > I get the accelerometer working, do I leave it as-is, and leave the > implementation up to userspace? What would the accelerometer do? Mimic a > mouse? Mimic a joystick? I still have not investigated accelerometers, so I'm not 100% sure of my answer. I don't think accelerometers mimic a mouse or a joystick. There is a work on sensor hub this time, but I'm not sure about the protocol they use. > > Also, the remote has a power button and when it is pressed the bluetooth > connection is dropped, just the same as when it goes to sleep. Is there a way > of handling the power button? Is that something that would have to happen on > the bluez, or bluetooth side of things? If you don't see any hid events, maybe not. This is something the bluetooth guys could answer. > > One more thing, the report ids 5, 6, and 8. Any idea what those could be? > They have output values, and there are 3 LEDs on the device, including a data > indicator and two Fn lights, but the amount of data in those report > descriptors are quite large. Yes, they contain output reports. It means that they are used to send data back to the device. As the usage page is set to "Generic Device Reserved", we can not guess what they are related to without capturing some frames of the device. > > Does anybody have an example rdesc of a similar multitouch trackpad that I > could use in my custom rdesc to enable multitouch? No rdesc I know will work with your device. Moreover, you are encountering the current limitation of the hid parser which is not able to split the different reportID into several hid devices, which means that currently, you will have to write one big hid kernel driver that will handle all the different reportIDs. > > Thanks for the help. > > > 05 01 // USAGE_PAGE (Generic Desktop) > 09 06 // USAGE (Keyboard) > a1 01 // COLLECTION (Application) > 85 01 // REPORT_ID (1) > 75 01 // REPORT_SIZE (1) > 95 08 // REPORT_COUNT (8) > 05 07 // USAGE_PAGE (Keyboard/Keypad) > 19 e0 // USAGE_MINIMUM (224) <- Left ctrl > 29 e7 // USAGE_MAXIMIUM (231) <- Right GUI > 15 00 // LOGICAL_MINIMUM (0) > 25 01 // LOGICAL_MAXIMUM (1) > 81 02 // INPUT (Data,Variable,Absolute) <- Usual modifier keys > 95 01 // REPORT_COUNT (1) > 75 08 // REPORT_SIZE (8) > 81 03 // INPUT (Constant,Variable,Absolute) > 95 06 // REPORT_COUNT (6) > 75 08 // REPORT_SIZE (8) > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 05 07 // USAGE_PAGE (Keyboard/Keypad) > 19 00 // USAGE_MINIMUM (0) > 29 ff // USAGE_MAXIMIUM (255) > 81 00 // INPUT (Data,Array,Absolute) <- Regular keyboard keys > c0 // END_COLLECTION > 06 01 ff // USAGE_PAGE (Generic Device Reserved) <- Vendor specific? yep > 09 01 // USAGE (Pointer) <- A multitouch trackpad > a1 01 // COLLECTION (Application) > 85 02 // REPORT_ID (2) > 19 01 // USAGE_MINIMUM (1) <- Main mechanical touchpad button > 29 04 // USAGE_MAXIMUM (4) <- Button 3 is drag, other values appear > constant > 15 00 // LOGICAL_MINIMUM (0) > 25 01 // LOGICAL_MAXIMUM (1) > 75 01 // REPORT_SIZE (1) > 95 04 // REPORT_COUNT (4) > 81 02 // INPUT (Data,Variable,Absolute) > 19 00 // USAGE_MINIMUM (0) > 29 01 // USAGE_MAXIMUM (1) > 15 00 // LOGICAL_MINIMUM (0) > 25 03 // LOGICAL_MAXIMUM (3) > 75 02 // REPORT_SIZE (2) > 95 02 // REPORT_COUNT (2) > 81 02 // INPUT (Data,Variable,Absolute) <- Contact identifiers > 09 30 // USAGE (X) > 09 31 // USAGE (Y) > 16 01 f8 // LOGICAL_MINIMUM (-2047) > 26 ff 07 // LOGICAL_MAXIMUM (2047) > 75 0c // REPORT_SIZE (12) > 95 02 // REPORT_COUNT(2) > 81 02 // INPUT (Data,Variable,Absolute) <- First contact coordinates > 19 00 // USAGE_MINIMUM (0) > 29 0f // USAGE_MAXIMUM (15) > 15 00 // LOGICAL_MINIMUM (0) > 25 0f // LOGICAL_MAXIMUM (15) > 75 04 // REPORT_SIZE (4) > 95 02 // REPORT_COUNT (2) > 81 02 // INPUT (Data,Variable,Absolute) <- First contact width and > height > 09 30 // USAGE (X) > 15 81 // LOGICAL_MINIMUM (-127) > 25 7f // LOGICAL_MAXIMUM (127) > 75 08 // REPORT_SIZE (8) > 95 01 // REPORT_COUNT (1) > 81 06 // INPUT (Data,Variable,Relative) <- X axis relative This is kind of new: this can be calculated quite easily, why do they bother to add it... > 09 30 // USAGE (X) > 09 31 // USAGE (Y) > 16 01 f8 // LOGICAL_MINIMUM (-2047) > 26 ff 07 // LOGICAL_MAXIMUM (2047) > 75 0c // REPORT_SIZE (12) > 95 02 // REPORT_COUNT (2) > 81 02 // INPUT (Data,Variable,Absolute) <- Second contact coordinates > 19 00 // USAGE_MINIMUM (0) > 29 0f // USAGE_MAXIMUM (15) > 15 00 // LOGICAL_MINIMUM (0) > 25 0f // LOGICAL_MAXIMUM (15) > 75 04 // REPORT_SIZE (4) > 95 02 // REPORT_COUNT (2) > 81 02 // INPUT (Data,Variable,Absolute) <- Second contact width and > height > 09 31 // USAGE (Y) > 15 81 // LOGICAL_MINIMUM (-127) > 25 7f // LOGICAL_MAXIMUM (127) > 75 08 // REPORT_SIZE (8) > 95 01 // REPORT_COUNT (1) > 81 06 // INPUT (Data,Variable,Relative) <- Y axis relative > c0 // END_COLLECTION > 05 0c // USAGE_PAGE (Consumer Device) > 09 01 // USAGE (Consumer Control) > a1 01 // COLLECTION (Application) > 85 03 // REPORT_ID (3) > 19 00 // USAGE_MINIMUM (0) > 2a 9c 02 // USAGE_MAXIMUM (668) > 15 00 // LOGICAL_MINIMUM (0) > 26 9c 02 // LOGICAL_MAXIMUM (668) > 75 10 // REPORT_SIZE (16) > 95 01 // REPORT_COUNT (1) > 81 00 // INPUT (Data,Array,Absolute) <- Additional remote buttons, on, > off, etc. > c0 // END_COLLECTION > 05 01 // USAGE_PAGE (Generic Desktop) > 09 08 // USAGE (Multi-axis Controller) > a1 01 // COLLECTION (Application) > 85 04 // REPORT_ID (4) > 09 40 // USAGE (Vx) > 09 41 // USAGE (Vy) > 09 42 // USAGE (Vz) > 66 11 e0 // UNIT (SI Linear: Centimeter*Seconds^-2) > 16 00 80 // LOGICAL_MINIMUM (-32768) > 26 ff 7f // LOGICAL_MAXIMUM (32767) > 75 10 // REPORT_SIZE (16) > 95 03 // REPORT_COUNT (3) > 81 02 // INPUT (Data,Variable,Absolute) <- 3-axis accelerometer > 09 00 // USAGE () > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 75 08 // REPORT_SIZE (8) > 95 01 // REPORT_COUNT (1) > 81 02 // INPUT (Data,Variable,Absolute) <- motion intensity? > c0 // END_COLLECTION > 06 01 ff // USAGE_PAGE (Generic Device Reserved) <- Vendor specific yep, so no idea on the protocol. This can be anything. > 09 02 // USAGE () <- Trying to look like a Generic Desktop Mouse? no. The Mice don't need to present output reports. This is a bidirectional communication of 2*6 bytes. > a1 01 // COLLECTION (Application) > 85 05 // REPORT_ID (5) > 19 01 // USAGE_MINIMUM (1) > 29 ff // USAGE_MAXIMUM (255) > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 75 08 // REPORT_SIZE (8) > 95 06 // REPORT_COUNT (6) > 81 00 // INPUT (Data,Array,Absolute) > 19 01 // USAGE_MINIMUM (1) > 29 ff // USAGE_MAXIMUM (255) > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 75 08 // REPORT_SIZE (8) > 95 06 // REPORT_COUNT (6) > 91 00 // OUTPUT (Data,Array,Absolute) <- eh? > c0 // END_COLLECTION > 06 01 ff // USAGE_PAGE (Generic Device Reserved) <- Vendor specific > 09 03 // USAGE () <- Generic Desktop Reserved? 2*15 bytes communication. > a1 01 // COLLECTION (Application) > 85 06 // REPORT_ID (6) > 19 01 // USAGE_MINIMUM (1) > 29 ff // USAGE_MAXIMUM (255) > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 75 08 // REPORT_SIZE (8) > 95 0f // REPORT_COUNT (15) > 81 00 // INPUT (Data,Array,Absolute) > 19 01 // USAGE_MINIMUM (1) > 29 ff // USAGE_MAXIMUM (255) > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 75 08 // REPORT_SIZE (8) > 95 0f // REPORT_COUNT (15) > 91 00 // OUTPUT (Data,Array,Absolute) > c0 // END_COLLECTION > 06 01 ff // USAGE_PAGE (Generic Device Reserved) <- Vendor specific > 09 05 // USAGE () <- Trying to look like a Generic Desktop Game Pad? Again, 2*6 bytes of communication... > a1 01 // COLLECTION (Application) > 85 08 // REPORT_ID (8) <- Where did report id #7 go? > 19 01 // USAGE_MINIMUM (1) > 29 ff // USAGE_MAXIMUM (255) > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 75 08 // REPORT_SIZE (8) > 95 06 // REPORT_COUNT (6) > 81 00 // INPUT (Data,Array,Absolute) > 19 01 // USAGE_MINIMUM (1) > 29 ff // USAGE_MAXIMUM (255) > 15 00 // LOGICAL_MINIMUM (0) > 26 ff 00 // LOGICAL_MAXIMUM (255) > 75 08 // REPORT_SIZE (8) > 95 06 // REPORT_COUNT (6) > 91 00 // OUTPUT (Data,Array,Absolute) > c0 // END_COLLECTION > 00 // <- Extra byte, similar to sixaxis report descriptor > Cheers, Benjamin -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
