xref: /freebsd/sys/dev/xen/xenstore/xenstore_dev.c (revision 2e3507c25e42292b45a5482e116d278f5515d04d)
1 /*
2  * xenstore_dev.c
3  *
4  * Driver giving user-space access to the kernel's connection to the
5  * XenStore service.
6  *
7  * Copyright (c) 2005, Christian Limpach
8  * Copyright (c) 2005, Rusty Russell, IBM Corporation
9  *
10  * This file may be distributed separately from the Linux kernel, or
11  * incorporated into other software packages, subject to the following license:
12  *
13  * Permission is hereby granted, free of charge, to any person obtaining a copy
14  * of this source file (the "Software"), to deal in the Software without
15  * restriction, including without limitation the rights to use, copy, modify,
16  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
17  * and to permit persons to whom the Software is furnished to do so, subject to
18  * the following conditions:
19  *
20  * The above copyright notice and this permission notice shall be included in
21  * all copies or substantial portions of the Software.
22  *
23  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
24  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
25  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
26  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
27  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
28  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
29  * IN THE SOFTWARE.
30  */
31 
32 #include <sys/types.h>
33 #include <sys/cdefs.h>
34 #include <sys/errno.h>
35 #include <sys/uio.h>
36 #include <sys/param.h>
37 #include <sys/systm.h>
38 #include <sys/proc.h>
39 #include <sys/kernel.h>
40 #include <sys/malloc.h>
41 #include <sys/conf.h>
42 #include <sys/module.h>
43 #include <sys/selinfo.h>
44 #include <sys/sysctl.h>
45 #include <sys/poll.h>
46 
47 #include <xen/xen-os.h>
48 
49 #include <xen/hypervisor.h>
50 #include <xen/xenstore/xenstorevar.h>
51 #include <xen/xenstore/xenstore_internal.h>
52 
53 static unsigned int max_pending_watches = 1000;
54 
55 struct xs_dev_transaction {
56 	LIST_ENTRY(xs_dev_transaction) list;
57 	struct xs_transaction handle;
58 };
59 
60 struct xs_dev_watch {
61 	LIST_ENTRY(xs_dev_watch) list;
62 	struct xs_watch watch;
63 	char *token;
64 	struct xs_dev_data *user;
65 };
66 
67 struct xs_dev_data {
68 	/* In-progress transaction. */
69 	LIST_HEAD(, xs_dev_transaction) transactions;
70 
71 	/* Active watches. */
72 	LIST_HEAD(, xs_dev_watch) watches;
73 
74 	/* Partial request. */
75 	unsigned int len;
76 	union {
77 		struct xsd_sockmsg msg;
78 		char buffer[PAGE_SIZE];
79 	} u;
80 
81 	/* Response queue. */
82 #define MASK_READ_IDX(idx) ((idx)&(PAGE_SIZE-1))
83 	char read_buffer[PAGE_SIZE];
84 	unsigned int read_cons, read_prod;
85 
86 	/* Serializes writes to the read buffer. */
87 	struct mtx lock;
88 
89 	/* Polling structure (for reads only ATM). */
90 	struct selinfo ev_rsel;
91 };
92 
93 static void
94 xs_queue_reply(struct xs_dev_data *u, const char *data, unsigned int len)
95 {
96 	unsigned int i;
97 
98 	for (i = 0; i < len; i++, u->read_prod++)
99 		u->read_buffer[MASK_READ_IDX(u->read_prod)] = data[i];
100 
101 	KASSERT((u->read_prod - u->read_cons) <= sizeof(u->read_buffer),
102 	    ("xenstore reply too big"));
103 
104 	wakeup(u);
105 	selwakeup(&u->ev_rsel);
106 }
107 
108 static const char *
109 xs_dev_error_to_string(int error)
110 {
111 	unsigned int i;
112 
113 	for (i = 0; i < nitems(xsd_errors); i++)
114 		if (xsd_errors[i].errnum == error)
115 			return (xsd_errors[i].errstring);
116 
117 	return (NULL);
118 }
119 
120 static void
121 xs_dev_return_error(struct xs_dev_data *u, int error, int req_id, int tx_id)
122 {
123 	struct xsd_sockmsg msg;
124 	const char *payload;
125 
126 	msg.type = XS_ERROR;
127 	msg.req_id = req_id;
128 	msg.tx_id = tx_id;
129 	payload = NULL;
130 
131 	payload = xs_dev_error_to_string(error);
132 	if (payload == NULL)
133 		payload = xs_dev_error_to_string(EINVAL);
134 	KASSERT(payload != NULL, ("Unable to find string for EINVAL errno"));
135 
136 	msg.len = strlen(payload) + 1;
137 
138 	mtx_lock(&u->lock);
139 	xs_queue_reply(u, (char *)&msg, sizeof(msg));
140 	xs_queue_reply(u, payload, msg.len);
141 	mtx_unlock(&u->lock);
142 }
143 
144 static int
145 xs_dev_watch_message_parse_string(const char **p, const char *end,
146     const char **string_r)
147 {
148 	const char *nul;
149 
150 	nul = memchr(*p, 0, end - *p);
151 	if (!nul)
152 		return (EINVAL);
153 
154 	*string_r = *p;
155 	*p = nul+1;
156 
157 	return (0);
158 }
159 
160 static int
161 xs_dev_watch_message_parse(const struct xsd_sockmsg *msg, const char **path_r,
162     const char **token_r)
163 {
164 	const char *p, *end;
165 	int error;
166 
167 	p = (const char *)msg + sizeof(*msg);
168 	end = p + msg->len;
169 	KASSERT(p <= end, ("payload overflow"));
170 
171 	error = xs_dev_watch_message_parse_string(&p, end, path_r);
172 	if (error)
173 		return (error);
174 	error = xs_dev_watch_message_parse_string(&p, end, token_r);
175 	if (error)
176 		return (error);
177 
178 	return (0);
179 }
180 
181 static struct xs_dev_watch *
182 xs_dev_find_watch(struct xs_dev_data *u, const char *token)
183 {
184 	struct xs_dev_watch *watch;
185 
186 	LIST_FOREACH(watch, &u->watches, list)
187 		if (strcmp(watch->token, token) == 0)
188 			return (watch);
189 
190 	return (NULL);
191 }
192 
193 static void
194 xs_dev_watch_cb(struct xs_watch *watch, const char **vec, unsigned int len)
195 {
196 	struct xs_dev_watch *dwatch;
197 	struct xsd_sockmsg msg;
198 	char *payload;
199 
200 	dwatch = (struct xs_dev_watch *)watch->callback_data;
201 	msg.type = XS_WATCH_EVENT;
202 	msg.req_id = msg.tx_id = 0;
203 	msg.len = strlen(vec[XS_WATCH_PATH]) + strlen(dwatch->token) + 2;
204 
205 	payload = malloc(msg.len, M_XENSTORE, M_WAITOK);
206 	strcpy(payload, vec[XS_WATCH_PATH]);
207 	strcpy(&payload[strlen(vec[XS_WATCH_PATH]) + 1], dwatch->token);
208 	mtx_lock(&dwatch->user->lock);
209 	xs_queue_reply(dwatch->user, (char *)&msg, sizeof(msg));
210 	xs_queue_reply(dwatch->user, payload, msg.len);
211 	mtx_unlock(&dwatch->user->lock);
212 	free(payload, M_XENSTORE);
213 }
214 
215 static struct xs_dev_transaction *
216 xs_dev_find_transaction(struct xs_dev_data *u, uint32_t tx_id)
217 {
218 	struct xs_dev_transaction *trans;
219 
220 	LIST_FOREACH(trans, &u->transactions, list)
221 		if (trans->handle.id == tx_id)
222 			return (trans);
223 
224 	return (NULL);
225 }
226 
227 static int
228 xs_dev_read(struct cdev *dev, struct uio *uio, int ioflag)
229 {
230 	int error;
231 	struct xs_dev_data *u;
232 
233 	error = devfs_get_cdevpriv((void **)&u);
234 	if (error != 0)
235 		return (error);
236 
237 	while (u->read_prod == u->read_cons) {
238 		error = tsleep(u, PCATCH, "xsdread", hz/10);
239 		if (error && error != EWOULDBLOCK)
240 			return (error);
241 	}
242 
243 	while (uio->uio_resid > 0) {
244 		if (u->read_cons == u->read_prod)
245 			break;
246 		error = uiomove(&u->read_buffer[MASK_READ_IDX(u->read_cons)],
247 		    1, uio);
248 		if (error)
249 			return (error);
250 		u->read_cons++;
251 	}
252 	return (0);
253 }
254 
255 static int
256 xs_dev_write(struct cdev *dev, struct uio *uio, int ioflag)
257 {
258 	int error;
259 	const char *wpath, *wtoken;
260 	struct xs_dev_data *u;
261 	struct xs_dev_transaction *trans;
262 	struct xs_dev_watch *watch;
263 	void *reply;
264 	static const char *ok = "OK";
265 	int len = uio->uio_resid;
266 
267 	error = devfs_get_cdevpriv((void **)&u);
268 	if (error != 0)
269 		return (error);
270 
271 	if ((len + u->len) > sizeof(u->u.buffer))
272 		return (EINVAL);
273 
274 	error = uiomove(u->u.buffer + u->len, len, uio);
275 	if (error)
276 		return (error);
277 
278 	u->len += len;
279 	if (u->len < (sizeof(u->u.msg) + u->u.msg.len))
280 		return (0);
281 
282 	switch (u->u.msg.type) {
283 	case XS_TRANSACTION_START:
284 	case XS_TRANSACTION_END:
285 	case XS_DIRECTORY:
286 	case XS_READ:
287 	case XS_GET_PERMS:
288 	case XS_RELEASE:
289 	case XS_GET_DOMAIN_PATH:
290 	case XS_WRITE:
291 	case XS_MKDIR:
292 	case XS_RM:
293 	case XS_SET_PERMS:
294 		/* Check that this transaction id is not hijacked. */
295 		if (u->u.msg.tx_id != 0 &&
296 		    xs_dev_find_transaction(u, u->u.msg.tx_id) == NULL) {
297 			error = EINVAL;
298 			break;
299 		}
300 		error = xs_dev_request_and_reply(&u->u.msg, &reply);
301 		if (!error) {
302 			if (u->u.msg.type == XS_TRANSACTION_START) {
303 				trans = malloc(sizeof(*trans), M_XENSTORE,
304 				    M_WAITOK);
305 				trans->handle.id = strtoul(reply, NULL, 0);
306 				LIST_INSERT_HEAD(&u->transactions, trans, list);
307 			} else if (u->u.msg.type == XS_TRANSACTION_END) {
308 				trans = xs_dev_find_transaction(u,
309 				    u->u.msg.tx_id);
310 				KASSERT(trans != NULL,
311 				    ("Unable to find transaction"));
312 				LIST_REMOVE(trans, list);
313 				free(trans, M_XENSTORE);
314 			}
315 			mtx_lock(&u->lock);
316 			xs_queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg));
317 			xs_queue_reply(u, (char *)reply, u->u.msg.len);
318 			mtx_unlock(&u->lock);
319 			free(reply, M_XENSTORE);
320 		}
321 		break;
322 	case XS_WATCH:
323 		u->u.msg.tx_id = 0;
324 		error = xs_dev_watch_message_parse(&u->u.msg, &wpath, &wtoken);
325 		if (error)
326 			break;
327 		if (xs_dev_find_watch(u, wtoken) != NULL) {
328 			error = EINVAL;
329 			break;
330 		}
331 
332 		watch = malloc(sizeof(*watch), M_XENSTORE, M_WAITOK);
333 		watch->watch.node = strdup(wpath, M_XENSTORE);
334 		watch->watch.callback = xs_dev_watch_cb;
335 		watch->watch.callback_data = (uintptr_t)watch;
336 		watch->watch.max_pending = max_pending_watches;
337 		watch->token = strdup(wtoken, M_XENSTORE);
338 		watch->user = u;
339 
340 		error = xs_register_watch(&watch->watch);
341 		if (error != 0) {
342 			free(watch->token, M_XENSTORE);
343 			free(watch->watch.node, M_XENSTORE);
344 			free(watch, M_XENSTORE);
345 			break;
346 		}
347 
348 		LIST_INSERT_HEAD(&u->watches, watch, list);
349 		u->u.msg.len = sizeof(ok);
350 		mtx_lock(&u->lock);
351 		xs_queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg));
352 		xs_queue_reply(u, ok, sizeof(ok));
353 		mtx_unlock(&u->lock);
354 		break;
355 	case XS_UNWATCH:
356 		u->u.msg.tx_id = 0;
357 		error = xs_dev_watch_message_parse(&u->u.msg, &wpath, &wtoken);
358 		if (error)
359 			break;
360 		watch = xs_dev_find_watch(u, wtoken);
361 		if (watch == NULL) {
362 			error = EINVAL;
363 			break;
364 		}
365 
366 		LIST_REMOVE(watch, list);
367 		xs_unregister_watch(&watch->watch);
368 		free(watch->watch.node, M_XENSTORE);
369 		free(watch->token, M_XENSTORE);
370 		free(watch, M_XENSTORE);
371 		u->u.msg.len = sizeof(ok);
372 		mtx_lock(&u->lock);
373 		xs_queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg));
374 		xs_queue_reply(u, ok, sizeof(ok));
375 		mtx_unlock(&u->lock);
376 		break;
377 	default:
378 		error = EINVAL;
379 		break;
380 	}
381 
382 	if (error != 0)
383 		xs_dev_return_error(u, error, u->u.msg.req_id, u->u.msg.tx_id);
384 
385 	/* Reset the write buffer. */
386 	u->len = 0;
387 
388 	return (0);
389 }
390 
391 static int
392 xs_dev_poll(struct cdev *dev, int events, struct thread *td)
393 {
394 	struct xs_dev_data *u;
395 	int error, mask;
396 
397 	error = devfs_get_cdevpriv((void **)&u);
398 	if (error != 0)
399 		return (POLLERR);
400 
401 	/* we can always write */
402 	mask = events & (POLLOUT | POLLWRNORM);
403 
404 	if (events & (POLLIN | POLLRDNORM)) {
405 		if (u->read_cons != u->read_prod) {
406 			mask |= events & (POLLIN | POLLRDNORM);
407 		} else {
408 			/* Record that someone is waiting */
409 			selrecord(td, &u->ev_rsel);
410 		}
411 	}
412 
413 	return (mask);
414 }
415 
416 static void
417 xs_dev_dtor(void *arg)
418 {
419 	struct xs_dev_data *u = arg;
420 	struct xs_dev_transaction *trans, *tmpt;
421 	struct xs_dev_watch *watch, *tmpw;
422 
423 	seldrain(&u->ev_rsel);
424 
425 	LIST_FOREACH_SAFE(trans, &u->transactions, list, tmpt) {
426 		xs_transaction_end(trans->handle, 1);
427 		LIST_REMOVE(trans, list);
428 		free(trans, M_XENSTORE);
429 	}
430 
431 	LIST_FOREACH_SAFE(watch, &u->watches, list, tmpw) {
432 		LIST_REMOVE(watch, list);
433 		xs_unregister_watch(&watch->watch);
434 		free(watch->watch.node, M_XENSTORE);
435 		free(watch->token, M_XENSTORE);
436 		free(watch, M_XENSTORE);
437 	}
438 	mtx_destroy(&u->lock);
439 
440 	free(u, M_XENSTORE);
441 }
442 
443 static int
444 xs_dev_open(struct cdev *dev, int oflags, int devtype, struct thread *td)
445 {
446 	struct xs_dev_data *u;
447 	int error;
448 
449 	u = malloc(sizeof(*u), M_XENSTORE, M_WAITOK|M_ZERO);
450 	mtx_init(&u->lock, "xsdev_lock", NULL, MTX_DEF);
451 	LIST_INIT(&u->transactions);
452 	LIST_INIT(&u->watches);
453 	error = devfs_set_cdevpriv(u, xs_dev_dtor);
454 	if (error != 0)
455 		free(u, M_XENSTORE);
456 
457 	return (error);
458 }
459 
460 static struct cdevsw xs_dev_cdevsw = {
461 	.d_version = D_VERSION,
462 	.d_read = xs_dev_read,
463 	.d_write = xs_dev_write,
464 	.d_open = xs_dev_open,
465 	.d_poll = xs_dev_poll,
466 	.d_name = "xs_dev",
467 };
468 
469 /*------------------ Private Device Attachment Functions  --------------------*/
470 /**
471  * \brief Identify instances of this device type in the system.
472  *
473  * \param driver  The driver performing this identify action.
474  * \param parent  The NewBus parent device for any devices this method adds.
475  */
476 static void
477 xs_dev_identify(driver_t *driver, device_t parent)
478 {
479 	/*
480 	 * A single device instance for our driver is always present
481 	 * in a system operating under Xen.
482 	 */
483 	BUS_ADD_CHILD(parent, 0, driver->name, 0);
484 }
485 
486 /**
487  * \brief Probe for the existence of the Xenstore device
488  *
489  * \param dev  NewBus device_t for this instance.
490  *
491  * \return  Always returns 0 indicating success.
492  */
493 static int
494 xs_dev_probe(device_t dev)
495 {
496 
497 	device_set_desc(dev, "Xenstore user-space device");
498 	return (0);
499 }
500 
501 /**
502  * \brief Attach the Xenstore device.
503  *
504  * \param dev  NewBus device_t for this instance.
505  *
506  * \return  On success, 0. Otherwise an errno value indicating the
507  *          type of failure.
508  */
509 static int
510 xs_dev_attach(device_t dev)
511 {
512 	struct cdev *xs_cdev;
513 	struct sysctl_ctx_list *sysctl_ctx;
514 	struct sysctl_oid *sysctl_tree;
515 
516 	sysctl_ctx = device_get_sysctl_ctx(dev);
517 	sysctl_tree = device_get_sysctl_tree(dev);
518 	if (sysctl_ctx == NULL || sysctl_tree == NULL)
519 	    return (EINVAL);
520 
521 	SYSCTL_ADD_UINT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO,
522 	    "max_pending_watch_events", CTLFLAG_RW, &max_pending_watches, 0,
523 	    "maximum amount of pending watch events to be delivered");
524 
525 	xs_cdev = make_dev_credf(MAKEDEV_ETERNAL, &xs_dev_cdevsw, 0, NULL,
526 	    UID_ROOT, GID_WHEEL, 0400, "xen/xenstore");
527 	if (xs_cdev == NULL)
528 		return (EINVAL);
529 
530 	return (0);
531 }
532 
533 /*-------------------- Private Device Attachment Data  -----------------------*/
534 static device_method_t xs_dev_methods[] = {
535 	/* Device interface */
536 	DEVMETHOD(device_identify,	xs_dev_identify),
537 	DEVMETHOD(device_probe,         xs_dev_probe),
538 	DEVMETHOD(device_attach,        xs_dev_attach),
539 
540 	DEVMETHOD_END
541 };
542 
543 DEFINE_CLASS_0(xs_dev, xs_dev_driver, xs_dev_methods, 0);
544 
545 DRIVER_MODULE(xs_dev, xenstore, xs_dev_driver, NULL, NULL);
546