| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | 		      Joystick API Documentation                -*-Text-*- | 
 | 2 |  | 
 | 3 | 		        Ragnar Hojland Espinosa | 
 | 4 | 			  <ragnar@macula.net> | 
 | 5 |  | 
 | 6 | 			      7 Aug 1998 | 
 | 7 |  | 
 | 8 | 	$Id: joystick-api.txt,v 1.2 2001/05/08 21:21:23 vojtech Exp $ | 
 | 9 |  | 
 | 10 | 1. Initialization | 
 | 11 | ~~~~~~~~~~~~~~~~~ | 
 | 12 |  | 
 | 13 | Open the joystick device following the usual semantics (that is, with open). | 
 | 14 | Since the driver now reports events instead of polling for changes, | 
 | 15 | immediately after the open it will issue a series of synthetic events | 
 | 16 | (JS_EVENT_INIT) that you can read to check the initial state of the | 
 | 17 | joystick. | 
 | 18 |  | 
 | 19 | By default, the device is opened in blocking mode. | 
 | 20 |  | 
 | 21 | 	int fd = open ("/dev/js0", O_RDONLY); | 
 | 22 |  | 
 | 23 |  | 
 | 24 | 2. Event Reading | 
 | 25 | ~~~~~~~~~~~~~~~~ | 
 | 26 |  | 
 | 27 | 	struct js_event e; | 
 | 28 | 	read (fd, &e, sizeof(struct js_event)); | 
 | 29 |  | 
 | 30 | where js_event is defined as | 
 | 31 |  | 
 | 32 | 	struct js_event { | 
 | 33 | 		__u32 time;     /* event timestamp in milliseconds */ | 
 | 34 | 		__s16 value;    /* value */ | 
 | 35 | 		__u8 type;      /* event type */ | 
 | 36 | 		__u8 number;    /* axis/button number */ | 
 | 37 | 	}; | 
 | 38 |  | 
 | 39 | If the read is successful, it will return sizeof(struct js_event), unless | 
 | 40 | you wanted to read more than one event per read as described in section 3.1. | 
 | 41 |  | 
 | 42 |  | 
 | 43 | 2.1 js_event.type | 
 | 44 | ~~~~~~~~~~~~~~~~~ | 
 | 45 |  | 
 | 46 | The possible values of ``type'' are | 
 | 47 |  | 
 | 48 | 	#define JS_EVENT_BUTTON         0x01    /* button pressed/released */ | 
 | 49 | 	#define JS_EVENT_AXIS           0x02    /* joystick moved */ | 
 | 50 | 	#define JS_EVENT_INIT           0x80    /* initial state of device */ | 
 | 51 |  | 
 | 52 | As mentioned above, the driver will issue synthetic JS_EVENT_INIT ORed | 
 | 53 | events on open. That is, if it's issuing a INIT BUTTON event, the | 
 | 54 | current type value will be | 
 | 55 |  | 
 | 56 | 	int type = JS_EVENT_BUTTON | JS_EVENT_INIT;	/* 0x81 */ | 
 | 57 |  | 
 | 58 | If you choose not to differentiate between synthetic or real events | 
 | 59 | you can turn off the JS_EVENT_INIT bits | 
 | 60 |  | 
 | 61 | 	type &= ~JS_EVENT_INIT;				/* 0x01 */ | 
 | 62 |  | 
 | 63 |  | 
 | 64 | 2.2 js_event.number | 
 | 65 | ~~~~~~~~~~~~~~~~~~~ | 
 | 66 |  | 
 | 67 | The values of ``number'' correspond to the axis or button that | 
 | 68 | generated the event. Note that they carry separate numeration (that | 
 | 69 | is, you have both an axis 0 and a button 0). Generally, | 
 | 70 |  | 
 | 71 | 			number | 
 | 72 | 	1st Axis X	0 | 
 | 73 | 	1st Axis Y	1 | 
 | 74 | 	2nd Axis X	2 | 
 | 75 | 	2nd Axis Y	3 | 
 | 76 | 	...and so on | 
 | 77 |  | 
 | 78 | Hats vary from one joystick type to another. Some can be moved in 8 | 
 | 79 | directions, some only in 4, The driver, however, always reports a hat as two | 
 | 80 | independent axis, even if the hardware doesn't allow independent movement. | 
 | 81 |  | 
 | 82 |  | 
 | 83 | 2.3 js_event.value | 
 | 84 | ~~~~~~~~~~~~~~~~~~ | 
 | 85 |  | 
 | 86 | For an axis, ``value'' is a signed integer between -32767 and +32767 | 
 | 87 | representing the position of the joystick along that axis. If you | 
 | 88 | don't read a 0 when the joystick is `dead', or if it doesn't span the | 
 | 89 | full range, you should recalibrate it (with, for example, jscal). | 
 | 90 |  | 
 | 91 | For a button, ``value'' for a press button event is 1 and for a release | 
 | 92 | button event is 0. | 
 | 93 |  | 
 | 94 | Though this | 
 | 95 |  | 
 | 96 | 	if (js_event.type == JS_EVENT_BUTTON) { | 
 | 97 | 		buttons_state ^= (1 << js_event.number); | 
 | 98 | 	} | 
 | 99 |  | 
 | 100 | may work well if you handle JS_EVENT_INIT events separately, | 
 | 101 |  | 
 | 102 | 	if ((js_event.type & ~JS_EVENT_INIT) == JS_EVENT_BUTTON) { | 
 | 103 | 		if (js_event.value) | 
 | 104 | 	        	buttons_state |= (1 << js_event.number); | 
 | 105 | 	   	else | 
 | 106 | 	      		buttons_state &= ~(1 << js_event.number); | 
 | 107 | 	} | 
 | 108 |  | 
 | 109 | is much safer since it can't lose sync with the driver. As you would | 
 | 110 | have to write a separate handler for JS_EVENT_INIT events in the first | 
 | 111 | snippet, this ends up being shorter. | 
 | 112 |  | 
 | 113 |  | 
 | 114 | 2.4 js_event.time | 
 | 115 | ~~~~~~~~~~~~~~~~~ | 
 | 116 |  | 
 | 117 | The time an event was generated is stored in ``js_event.time''. It's a time | 
 | 118 | in milliseconds since ... well, since sometime in the past.  This eases the | 
 | 119 | task of detecting double clicks, figuring out if movement of axis and button | 
 | 120 | presses happened at the same time, and similar. | 
 | 121 |  | 
 | 122 |  | 
 | 123 | 3. Reading | 
 | 124 | ~~~~~~~~~~ | 
 | 125 |  | 
 | 126 | If you open the device in blocking mode, a read will block (that is, | 
 | 127 | wait) forever until an event is generated and effectively read. There | 
 | 128 | are two alternatives if you can't afford to wait forever (which is, | 
 | 129 | admittedly, a long time;) | 
 | 130 |  | 
 | 131 | 	a) use select to wait until there's data to be read on fd, or | 
 | 132 | 	   until it timeouts. There's a good example on the select(2) | 
 | 133 | 	   man page. | 
 | 134 |  | 
 | 135 | 	b) open the device in non-blocking mode (O_NONBLOCK) | 
 | 136 |  | 
 | 137 |  | 
 | 138 | 3.1 O_NONBLOCK | 
 | 139 | ~~~~~~~~~~~~~~ | 
 | 140 |  | 
 | 141 | If read returns -1 when reading in O_NONBLOCK mode, this isn't | 
 | 142 | necessarily a "real" error (check errno(3)); it can just mean there | 
 | 143 | are no events pending to be read on the driver queue. You should read | 
 | 144 | all events on the queue (that is, until you get a -1). | 
 | 145 |  | 
 | 146 | For example, | 
 | 147 |  | 
 | 148 | 	while (1) { | 
 | 149 | 		while (read (fd, &e, sizeof(struct js_event)) > 0) { | 
 | 150 | 	        	process_event (e); | 
 | 151 | 	   	} | 
 | 152 | 	   	/* EAGAIN is returned when the queue is empty */ | 
 | 153 | 	   	if (errno != EAGAIN) { | 
 | 154 | 	      		/* error */ | 
 | 155 | 	   	} | 
 | 156 | 	   	/* do something interesting with processed events */ | 
 | 157 | 	} | 
 | 158 |  | 
 | 159 | One reason for emptying the queue is that if it gets full you'll start | 
 | 160 | missing events since the queue is finite, and older events will get | 
 | 161 | overwritten. | 
 | 162 |  | 
 | 163 | The other reason is that you want to know all what happened, and not | 
 | 164 | delay the processing till later. | 
 | 165 |  | 
 | 166 | Why can get the queue full? Because you don't empty the queue as | 
 | 167 | mentioned, or because too much time elapses from one read to another | 
 | 168 | and too many events to store in the queue get generated. Note that | 
 | 169 | high system load may contribute to space those reads even more. | 
 | 170 |  | 
 | 171 | If time between reads is enough to fill the queue and lose an event, | 
 | 172 | the driver will switch to startup mode and next time you read it, | 
 | 173 | synthetic events (JS_EVENT_INIT) will be generated to inform you of | 
 | 174 | the actual state of the joystick. | 
 | 175 |  | 
 | 176 | [As for version 1.2.8, the queue is circular and able to hold 64 | 
 | 177 |  events. You can increment this size bumping up JS_BUFF_SIZE in | 
 | 178 |  joystick.h and recompiling the driver.] | 
 | 179 |  | 
 | 180 |  | 
 | 181 | In the above code, you might as well want to read more than one event | 
 | 182 | at a time using the typical read(2) functionality. For that, you would | 
 | 183 | replace the read above with something like | 
 | 184 |  | 
 | 185 | 	struct js_event mybuffer[0xff]; | 
 | 186 | 	int i = read (fd, mybuffer, sizeof(struct mybuffer)); | 
 | 187 |  | 
 | 188 | In this case, read would return -1 if the queue was empty, or some | 
 | 189 | other value in which the number of events read would be i / | 
 | 190 | sizeof(js_event)  Again, if the buffer was full, it's a good idea to | 
 | 191 | process the events and keep reading it until you empty the driver queue. | 
 | 192 |  | 
 | 193 |  | 
 | 194 | 4. IOCTLs | 
 | 195 | ~~~~~~~~~ | 
 | 196 |  | 
 | 197 | The joystick driver defines the following ioctl(2) operations. | 
 | 198 |  | 
 | 199 | 				/* function			3rd arg  */ | 
 | 200 | 	#define JSIOCGAXES	/* get number of axes		char	 */ | 
 | 201 | 	#define JSIOCGBUTTONS	/* get number of buttons	char	 */ | 
 | 202 | 	#define JSIOCGVERSION	/* get driver version		int	 */ | 
 | 203 | 	#define JSIOCGNAME(len) /* get identifier string	char	 */ | 
 | 204 | 	#define JSIOCSCORR	/* set correction values	&js_corr */ | 
 | 205 | 	#define JSIOCGCORR	/* get correction values	&js_corr */ | 
 | 206 |  | 
 | 207 | For example, to read the number of axes | 
 | 208 |  | 
 | 209 | 	char number_of_axes; | 
 | 210 | 	ioctl (fd, JSIOCGAXES, &number_of_axes); | 
 | 211 |  | 
 | 212 |  | 
 | 213 | 4.1 JSIOGCVERSION | 
 | 214 | ~~~~~~~~~~~~~~~~~ | 
 | 215 |  | 
 | 216 | JSIOGCVERSION is a good way to check in run-time whether the running | 
 | 217 | driver is 1.0+ and supports the event interface. If it is not, the | 
 | 218 | IOCTL will fail. For a compile-time decision, you can test the | 
 | 219 | JS_VERSION symbol | 
 | 220 |  | 
 | 221 | 	#ifdef JS_VERSION | 
 | 222 | 	#if JS_VERSION > 0xsomething | 
 | 223 |  | 
 | 224 |  | 
 | 225 | 4.2 JSIOCGNAME | 
 | 226 | ~~~~~~~~~~~~~~ | 
 | 227 |  | 
 | 228 | JSIOCGNAME(len) allows you to get the name string of the joystick - the same | 
 | 229 | as is being printed at boot time. The 'len' argument is the length of the | 
 | 230 | buffer provided by the application asking for the name. It is used to avoid | 
 | 231 | possible overrun should the name be too long. | 
 | 232 |  | 
 | 233 | 	char name[128]; | 
 | 234 | 	if (ioctl(fd, JSIOCGNAME(sizeof(name)), name) < 0) | 
 | 235 | 		strncpy(name, "Unknown", sizeof(name)); | 
 | 236 | 	printf("Name: %s\n", name); | 
 | 237 |  | 
 | 238 |  | 
 | 239 | 4.3 JSIOC[SG]CORR | 
 | 240 | ~~~~~~~~~~~~~~~~~ | 
 | 241 |  | 
 | 242 | For usage on JSIOC[SG]CORR I suggest you to look into jscal.c  They are | 
 | 243 | not needed in a normal program, only in joystick calibration software | 
 | 244 | such as jscal or kcmjoy. These IOCTLs and data types aren't considered | 
 | 245 | to be in the stable part of the API, and therefore may change without | 
 | 246 | warning in following releases of the driver. | 
 | 247 |  | 
 | 248 | Both JSIOCSCORR and JSIOCGCORR expect &js_corr to be able to hold | 
 | 249 | information for all axis. That is, struct js_corr corr[MAX_AXIS]; | 
 | 250 |  | 
 | 251 | struct js_corr is defined as | 
 | 252 |  | 
 | 253 | 	struct js_corr { | 
 | 254 | 		__s32 coef[8]; | 
 | 255 | 		__u16 prec; | 
 | 256 | 		__u16 type; | 
 | 257 | 	}; | 
 | 258 |  | 
 | 259 | and ``type'' | 
 | 260 |  | 
 | 261 | 	#define JS_CORR_NONE            0x00    /* returns raw values */ | 
 | 262 | 	#define JS_CORR_BROKEN          0x01    /* broken line */ | 
 | 263 |  | 
 | 264 |  | 
 | 265 | 5. Backward compatibility | 
 | 266 | ~~~~~~~~~~~~~~~~~~~~~~~~~ | 
 | 267 |  | 
 | 268 | The 0.x joystick driver API is quite limited and its usage is deprecated. | 
 | 269 | The driver offers backward compatibility, though. Here's a quick summary: | 
 | 270 |  | 
 | 271 | 	struct JS_DATA_TYPE js; | 
 | 272 | 	while (1) { | 
 | 273 | 		if (read (fd, &js, JS_RETURN) != JS_RETURN) { | 
 | 274 | 	      		/* error */ | 
 | 275 | 	   	} | 
 | 276 | 	   	usleep (1000); | 
 | 277 | 	} | 
 | 278 |  | 
 | 279 | As you can figure out from the example, the read returns immediately, | 
 | 280 | with the actual state of the joystick. | 
 | 281 |  | 
 | 282 | 	struct JS_DATA_TYPE { | 
 | 283 | 		int buttons;    /* immediate button state */ | 
 | 284 | 		int x;          /* immediate x axis value */ | 
 | 285 | 		int y;          /* immediate y axis value */ | 
 | 286 | 	}; | 
 | 287 |  | 
 | 288 | and JS_RETURN is defined as | 
 | 289 |  | 
 | 290 | 	#define JS_RETURN       sizeof(struct JS_DATA_TYPE) | 
 | 291 |  | 
 | 292 | To test the state of the buttons, | 
 | 293 |  | 
 | 294 | 	first_button_state  = js.buttons & 1; | 
 | 295 | 	second_button_state = js.buttons & 2; | 
 | 296 |  | 
 | 297 | The axis values do not have a defined range in the original 0.x driver, | 
 | 298 | except for that the values are non-negative. The 1.2.8+ drivers use a | 
 | 299 | fixed range for reporting the values, 1 being the minimum, 128 the | 
 | 300 | center, and 255 maximum value. | 
 | 301 |  | 
 | 302 | The v0.8.0.2 driver also had an interface for 'digital joysticks', (now | 
 | 303 | called Multisystem joysticks in this driver), under /dev/djsX. This driver | 
 | 304 | doesn't try to be compatible with that interface. | 
 | 305 |  | 
 | 306 |  | 
 | 307 | 6. Final Notes | 
 | 308 | ~~~~~~~~~~~~~~ | 
 | 309 |  | 
 | 310 | ____/|	Comments, additions, and specially corrections are welcome. | 
 | 311 | \ o.O|	Documentation valid for at least version 1.2.8 of the joystick | 
 | 312 |  =(_)=	driver and as usual, the ultimate source for documentation is | 
 | 313 |    U	to "Use The Source Luke" or, at your convenience, Vojtech ;) | 
 | 314 |  | 
 | 315 | 					- Ragnar | 
 | 316 | EOF |