Linux Input drivers v0.9 (c) 1999 Vojtech Pavlik Sponsored by SuSE ---------------------------------------------------------------------------- 0. Disclaimer ~~~~~~~~~~~~~ This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA Should you need to contact me, the author, you can do so either by e-mail - mail your message to , or by paper mail: Vojtech Pavlik, Ucitelska 1576, Prague 8, 182 00 Czech Republic For your convenience, the GNU General Public License version 2 is included in the package: See the file COPYING. 1. Introduction ~~~~~~~~~~~~~~~ This is a collection of drivers that is designed to support all input devices under Linux. However, in the current kernels, although it's possibilities are much bigger, it's limited to USB devices only. This is also why it resides in the drivers/usb subdirectory. The center of the input drivers is the input.o module, which must be loaded before any other of the input modules - it serves as a way of communication between two groups of modules: 1.1 Device drivers ~~~~~~~~~~~~~~~~~~ These modules talk to the hardware (for example via USB), and provide events (keystrokes, mouse movements) to the input.o module. 1.2 Event handlers ~~~~~~~~~~~~~~~~~~ These modules get events from input.o and pass them where needed via various interfaces - keystrokes to the kernel, mouse movements via a simulated PS/2 interface to GPM and X and so on. 2. Simple Usage ~~~~~~~~~~~~~~~ For the most usual configuration, with one USB mouse and one USB keyboard, you'll have to load the following modules (or have them built in to the kernel): input.o mousedev.o keybdev.o usbcore.o usb-[uo]hci.o hid.o After this, the USB keyboard will work straight away, and the USB mouse will be available as a character device on major 13, minor 32: crw-r--r-- 1 root root 13, 32 Mar 28 22:45 mouse0 This device, has to be created, unless you use devfs, in which case it's created automatically. The commands to do that are: cd /dev mkdir input mknod input/mouse0 c 13 32 After that you have to point GPM (the textmode mouse cut&paste tool) and XFree to this device to use it - GPM should be called like: gpm -t ps2 -m /dev/input/mouse0 And in X: Section "Pointer" Protocol "ImPS/2" Device "/dev/input/mouse0" ZAxisMapping 4 5 EndSection When you do all of the above, you can use your USB mouse and keyboard. 3. Detailed Description ~~~~~~~~~~~~~~~~~~~~~~~ 3.1 Device drivers ~~~~~~~~~~~~~~~~~~ Device drivers are the modules that generate events. The events are however not useful without being handled, so you also will need to use some of the modules from section 3.2. 3.1.1 hid.c ~~~~~~~~~~~ Hid.c is the largest and most complex driver of the whole suite. It handles all HID devices, and because there is a very wide variety of them, and because the USB HID specification isn't simple, it needs to be this big. Currently, it handles USB mice, joysticks, gamepads, steering wheels keyboards, trackballs and digitizers. However, USB uses HID also for monitor controls, speaker controls, UPSs, LCDs and many other purposes. The monitor and speaker controls should be easy to add to the hid/input interface, but for the UPSs and LCDs it doesn't make much sense. The driver doesn't support these yet, and a new, non-event interface should be designed for them. If you have any of these devices (I don't), feel free to design something and if it's good, it'll get into the driver. The usage of the hid.o module is very simple, it takes no parameters, detects everything automatically and when a HID device is inserted, it detects it appropriately. However, because the devices vary wildly, you might happen to have a device that doesn't work well. In that case #define DEBUG at the beginning of hid.c and send me the syslog traces. 3.1.2 usbmouse.c ~~~~~~~~~~~~~~~~ For embedded systems, for mice with broken HID descriptors and just any other use when the big hid.c wouldn't be a good choice, there is the usbmouse.c driver. It handles USB mice only. It uses a simpler HIDBP protocol. This also means the mice must support this simpler protocol. Not all do. If you don't have any strong reason to use this module, use hid.c instead. 3.1.3 usbkbd.c ~~~~~~~~~~~~~~ Much like usbmouse.c, this module talks to keyboards with a simpplified HIDBP protocol. It's smaller, but doesn't support any extra special keys. Use hid.c instead if there isn't any special reason to use this. 3.1.4 wacom.c ~~~~~~~~~~~~~ This is a driver for Wacom Graphire and Intuos tablets. Not for Wacom PenPartner, that one is handled by the HID driver. Although the Intuos and Graphire tablets claim that they are HID tablets as well, they are not and thus need this specific driver. 3.1.5 wmforce.c ~~~~~~~~~~~~~~~ A driver for the Logitech WingMan Force joysticks, when connected via the USB port. It works quite well, but there is no force feedback support yet, because the interface to do that is a trade secret of Immersion Corp, and even Logitech can't disclose it. Support for Logitech WingMan Force Wheel isn't present in this driver, but if someone has the device, and is willing to cooperate, it should be a matter of a couple days to add it. 3.2 Event handlers ~~~~~~~~~~~~~~~~~~ Event handlers distrubite the events from the devices to userland and kernel, as needed. 3.2.1 keybdev.c ~~~~~~~~~~~~~~~ Keybdev is currently a rather ugly hack that translates the input events into architecture-specific keyboard raw mode (Xlated AT Set2 on x86), and passes them into the handle_scancode function of the keyboard.c module. This works well enough on all architectures that keybdev can generate rawmode on, other architectures can be added to it. The right way would be to pass the events to keyboard.c directly, best if keyboard.c would itself be an event handler. This is done in the input patch, available on the webpage mentioned below. 3.2.2 mousedev.c ~~~~~~~~~~~~~~~~ Mousedev is also a hack to make programs that use mouse input work. It takes events from either mice or digitizers/tablets and makes a PS/2-style (a la /dev/psaux) mouse device available to the userland. Ideally, the programs could use a more reasonable interface, for example evdev.c Mousedev devices in /dev/input (as shown above) are: crw-r--r-- 1 root root 13, 32 Mar 28 22:45 mouse0 crw-r--r-- 1 root root 13, 33 Mar 29 00:41 mouse1 crw-r--r-- 1 root root 13, 34 Mar 29 00:41 mouse2 crw-r--r-- 1 root root 13, 35 Apr 1 10:50 mouse3 and so on, up to mouse31. Each is assigned to a single mouse or digitizer, unless CONFIG_INPUT_MOUSEDEV_MIX is set. In that case all mice and digitizers share a single character device, mouse0, and even when none are connected, mouse0 is present. This is useful for hotplugging USB mice, so that programs can open the device even when no mice are present. CONFIG_INPUT_MOUSEDEV_SCREEN_[XY] in the kernel configuration are the size of your screen (in pixels) in XFree86. This is needed if you want to use your digitizer in X, because it's movement is sent to X via a virtual PS/2 mouse. Mousedev.c will generate either PS/2, ImPS/2 (microsoft intellimouse) or GenPS/2 (genius netmouse/netscroll) protocols, depending on what the program wishes. You can set GPM and X to any of these. You'll need ImPS/2 if you want to make use of a wheel on a USB mouse and GenPS/2 if you want to use extra (up to 5) buttons. I'm not sure how much is GenPS/2 supported in X, though. 3.2.3 joydev.c ~~~~~~~~~~~~~~ Joydev implements v0.x and v1.x Linux joystick api, much like drivers/char/joystick/joystick.c. See joystick-api.txt in the Documentation subdirectory for details. Joydev does it on top of the input subsystem, though. As soon as any USB joystick is connected, it can be accessed in /dev/input on: crw-r--r-- 1 root root 13, 0 Apr 1 10:50 js0 crw-r--r-- 1 root root 13, 1 Apr 1 10:50 js1 crw-r--r-- 1 root root 13, 2 Apr 1 10:50 js2 crw-r--r-- 1 root root 13, 3 Apr 1 10:50 js3 And so on up to js31. 3.2.4 evdev.c ~~~~~~~~~~~~~ Evdev is the generic input event interface. It passes the events generated in the kernel straight to the program, with timestamps. The API is still evolving, but should be useable now. It's described in section 5. This should be the way for GPM and X to get keyboard and mouse mouse events. It allows for multihead in X without any specific multihead kernel support. The event codes are the same on all architectures and are hardware independent. The devices are in /dev/input: crw-r--r-- 1 root root 13, 64 Apr 1 10:49 event0 crw-r--r-- 1 root root 13, 65 Apr 1 10:50 event1 crw-r--r-- 1 root root 13, 66 Apr 1 10:50 event2 crw-r--r-- 1 root root 13, 67 Apr 1 10:50 event3 3. Contacts ~~~~~~~~~~~ This effort has it's home page at: http://www.suse.cz/development/input/ You'll find both the latest HID driver and the complete Input driver there. There is also a mailing list for this: majordomo@atrey.karlin.mff.cuni.cz Send "subscribe linux-input" to subscribe to it. 4. Verifying if it works ~~~~~~~~~~~~~~~~~~~~~~~~ Typing a couple keys on the keyboard should be enough to check that a USB keyboard works and is correctly connected to the kernel keyboard driver. Doing a cat /dev/input/mouse0 (c, 13, 32) will verify that a mouse is also emulated, characters should appear if you move it. You can test the joystick emulation with the 'jstest' utility, available in the joystick package (see Documentation/joystick.txt). You can test the event devics with the 'evtest' utitily available on the input driver homepage (see the URL above). 5. Event interface ~~~~~~~~~~~~~~~~~~ Should you want to add event device support into any application (X, gpm, svgalib ...) I will be happy to provide you any help I can. Here goes a description of the current state of things, which is going to be extended, but not changed incompatibly as time goes: You can use blocking and nonblocking reads, also select() on the /dev/inputX devices, and you'll always get a whole number of input events on a read. Their layout is: struct input_event { struct timeval time; unsigned short type; unsigned short code; unsigned int value; }; 'time' is the timestamp, it returns the time at which the event happened. Type is for example EV_REL for relative momement, REL_KEY for a keypress or release. More types are defined in include/linux/input.h. 'code' is event code, for example REL_X or KEY_BACKSPACE, again a complete list is in include/linux/input.h. 'value' is the value the event carries. Either a relative change for EV_REL, absolute new value for EV_ABS (joysticks ...), or 0 for EV_KEY for release, 1 for keypress and 2 for autorepeat.