xref: /freebsd/sys/dev/xen/xenstore/xenstore_dev.c (revision 1603881667360c015f6685131f2f25474fa67a72)
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/cdefs.h>
33 __FBSDID("$FreeBSD$");
34 
35 #include <sys/types.h>
36 #include <sys/cdefs.h>
37 #include <sys/errno.h>
38 #include <sys/uio.h>
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/proc.h>
42 #include <sys/kernel.h>
43 #include <sys/malloc.h>
44 #include <sys/conf.h>
45 #include <sys/module.h>
46 #include <sys/selinfo.h>
47 #include <sys/sysctl.h>
48 #include <sys/poll.h>
49 
50 #include <xen/xen-os.h>
51 
52 #include <xen/hypervisor.h>
53 #include <xen/xenstore/xenstorevar.h>
54 #include <xen/xenstore/xenstore_internal.h>
55 
56 static unsigned int max_pending_watches = 1000;
57 
58 struct xs_dev_transaction {
59 	LIST_ENTRY(xs_dev_transaction) list;
60 	struct xs_transaction handle;
61 };
62 
63 struct xs_dev_watch {
64 	LIST_ENTRY(xs_dev_watch) list;
65 	struct xs_watch watch;
66 	char *token;
67 	struct xs_dev_data *user;
68 };
69 
70 struct xs_dev_data {
71 	/* In-progress transaction. */
72 	LIST_HEAD(, xs_dev_transaction) transactions;
73 
74 	/* Active watches. */
75 	LIST_HEAD(, xs_dev_watch) watches;
76 
77 	/* Partial request. */
78 	unsigned int len;
79 	union {
80 		struct xsd_sockmsg msg;
81 		char buffer[PAGE_SIZE];
82 	} u;
83 
84 	/* Response queue. */
85 #define MASK_READ_IDX(idx) ((idx)&(PAGE_SIZE-1))
86 	char read_buffer[PAGE_SIZE];
87 	unsigned int read_cons, read_prod;
88 
89 	/* Serializes writes to the read buffer. */
90 	struct mtx lock;
91 
92 	/* Polling structure (for reads only ATM). */
93 	struct selinfo ev_rsel;
94 };
95 
96 static void
97 xs_queue_reply(struct xs_dev_data *u, const char *data, unsigned int len)
98 {
99 	unsigned int i;
100 
101 	for (i = 0; i < len; i++, u->read_prod++)
102 		u->read_buffer[MASK_READ_IDX(u->read_prod)] = data[i];
103 
104 	KASSERT((u->read_prod - u->read_cons) <= sizeof(u->read_buffer),
105 	    ("xenstore reply too big"));
106 
107 	wakeup(u);
108 	selwakeup(&u->ev_rsel);
109 }
110 
111 static const char *
112 xs_dev_error_to_string(int error)
113 {
114 	unsigned int i;
115 
116 	for (i = 0; i < nitems(xsd_errors); i++)
117 		if (xsd_errors[i].errnum == error)
118 			return (xsd_errors[i].errstring);
119 
120 	return (NULL);
121 }
122 
123 static void
124 xs_dev_return_error(struct xs_dev_data *u, int error, int req_id, int tx_id)
125 {
126 	struct xsd_sockmsg msg;
127 	const char *payload;
128 
129 	msg.type = XS_ERROR;
130 	msg.req_id = req_id;
131 	msg.tx_id = tx_id;
132 	payload = NULL;
133 
134 	payload = xs_dev_error_to_string(error);
135 	if (payload == NULL)
136 		payload = xs_dev_error_to_string(EINVAL);
137 	KASSERT(payload != NULL, ("Unable to find string for EINVAL errno"));
138 
139 	msg.len = strlen(payload) + 1;
140 
141 	mtx_lock(&u->lock);
142 	xs_queue_reply(u, (char *)&msg, sizeof(msg));
143 	xs_queue_reply(u, payload, msg.len);
144 	mtx_unlock(&u->lock);
145 }
146 
147 static int
148 xs_dev_watch_message_parse_string(const char **p, const char *end,
149     const char **string_r)
150 {
151 	const char *nul;
152 
153 	nul = memchr(*p, 0, end - *p);
154 	if (!nul)
155 		return (EINVAL);
156 
157 	*string_r = *p;
158 	*p = nul+1;
159 
160 	return (0);
161 }
162 
163 static int
164 xs_dev_watch_message_parse(const struct xsd_sockmsg *msg, const char **path_r,
165     const char **token_r)
166 {
167 	const char *p, *end;
168 	int error;
169 
170 	p = (const char *)msg + sizeof(*msg);
171 	end = p + msg->len;
172 	KASSERT(p <= end, ("payload overflow"));
173 
174 	error = xs_dev_watch_message_parse_string(&p, end, path_r);
175 	if (error)
176 		return (error);
177 	error = xs_dev_watch_message_parse_string(&p, end, token_r);
178 	if (error)
179 		return (error);
180 
181 	return (0);
182 }
183 
184 static struct xs_dev_watch *
185 xs_dev_find_watch(struct xs_dev_data *u, const char *token)
186 {
187 	struct xs_dev_watch *watch;
188 
189 	LIST_FOREACH(watch, &u->watches, list)
190 		if (strcmp(watch->token, token) == 0)
191 			return (watch);
192 
193 	return (NULL);
194 }
195 
196 static void
197 xs_dev_watch_cb(struct xs_watch *watch, const char **vec, unsigned int len)
198 {
199 	struct xs_dev_watch *dwatch;
200 	struct xsd_sockmsg msg;
201 	char *payload;
202 
203 	dwatch = (struct xs_dev_watch *)watch->callback_data;
204 	msg.type = XS_WATCH_EVENT;
205 	msg.req_id = msg.tx_id = 0;
206 	msg.len = strlen(vec[XS_WATCH_PATH]) + strlen(dwatch->token) + 2;
207 
208 	payload = malloc(msg.len, M_XENSTORE, M_WAITOK);
209 	strcpy(payload, vec[XS_WATCH_PATH]);
210 	strcpy(&payload[strlen(vec[XS_WATCH_PATH]) + 1], dwatch->token);
211 	mtx_lock(&dwatch->user->lock);
212 	xs_queue_reply(dwatch->user, (char *)&msg, sizeof(msg));
213 	xs_queue_reply(dwatch->user, payload, msg.len);
214 	mtx_unlock(&dwatch->user->lock);
215 	free(payload, M_XENSTORE);
216 }
217 
218 static struct xs_dev_transaction *
219 xs_dev_find_transaction(struct xs_dev_data *u, uint32_t tx_id)
220 {
221 	struct xs_dev_transaction *trans;
222 
223 	LIST_FOREACH(trans, &u->transactions, list)
224 		if (trans->handle.id == tx_id)
225 			return (trans);
226 
227 	return (NULL);
228 }
229 
230 static int
231 xs_dev_read(struct cdev *dev, struct uio *uio, int ioflag)
232 {
233 	int error;
234 	struct xs_dev_data *u;
235 
236 	error = devfs_get_cdevpriv((void **)&u);
237 	if (error != 0)
238 		return (error);
239 
240 	while (u->read_prod == u->read_cons) {
241 		error = tsleep(u, PCATCH, "xsdread", hz/10);
242 		if (error && error != EWOULDBLOCK)
243 			return (error);
244 	}
245 
246 	while (uio->uio_resid > 0) {
247 		if (u->read_cons == u->read_prod)
248 			break;
249 		error = uiomove(&u->read_buffer[MASK_READ_IDX(u->read_cons)],
250 		    1, uio);
251 		if (error)
252 			return (error);
253 		u->read_cons++;
254 	}
255 	return (0);
256 }
257 
258 static int
259 xs_dev_write(struct cdev *dev, struct uio *uio, int ioflag)
260 {
261 	int error;
262 	const char *wpath, *wtoken;
263 	struct xs_dev_data *u;
264 	struct xs_dev_transaction *trans;
265 	struct xs_dev_watch *watch;
266 	void *reply;
267 	static const char *ok = "OK";
268 	int len = uio->uio_resid;
269 
270 	error = devfs_get_cdevpriv((void **)&u);
271 	if (error != 0)
272 		return (error);
273 
274 	if ((len + u->len) > sizeof(u->u.buffer))
275 		return (EINVAL);
276 
277 	error = uiomove(u->u.buffer + u->len, len, uio);
278 	if (error)
279 		return (error);
280 
281 	u->len += len;
282 	if (u->len < (sizeof(u->u.msg) + u->u.msg.len))
283 		return (0);
284 
285 	switch (u->u.msg.type) {
286 	case XS_TRANSACTION_START:
287 	case XS_TRANSACTION_END:
288 	case XS_DIRECTORY:
289 	case XS_READ:
290 	case XS_GET_PERMS:
291 	case XS_RELEASE:
292 	case XS_GET_DOMAIN_PATH:
293 	case XS_WRITE:
294 	case XS_MKDIR:
295 	case XS_RM:
296 	case XS_SET_PERMS:
297 		/* Check that this transaction id is not hijacked. */
298 		if (u->u.msg.tx_id != 0 &&
299 		    xs_dev_find_transaction(u, u->u.msg.tx_id) == NULL) {
300 			error = EINVAL;
301 			break;
302 		}
303 		error = xs_dev_request_and_reply(&u->u.msg, &reply);
304 		if (!error) {
305 			if (u->u.msg.type == XS_TRANSACTION_START) {
306 				trans = malloc(sizeof(*trans), M_XENSTORE,
307 				    M_WAITOK);
308 				trans->handle.id = strtoul(reply, NULL, 0);
309 				LIST_INSERT_HEAD(&u->transactions, trans, list);
310 			} else if (u->u.msg.type == XS_TRANSACTION_END) {
311 				trans = xs_dev_find_transaction(u,
312 				    u->u.msg.tx_id);
313 				KASSERT(trans != NULL,
314 				    ("Unable to find transaction"));
315 				LIST_REMOVE(trans, list);
316 				free(trans, M_XENSTORE);
317 			}
318 			mtx_lock(&u->lock);
319 			xs_queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg));
320 			xs_queue_reply(u, (char *)reply, u->u.msg.len);
321 			mtx_unlock(&u->lock);
322 			free(reply, M_XENSTORE);
323 		}
324 		break;
325 	case XS_WATCH:
326 		u->u.msg.tx_id = 0;
327 		error = xs_dev_watch_message_parse(&u->u.msg, &wpath, &wtoken);
328 		if (error)
329 			break;
330 		if (xs_dev_find_watch(u, wtoken) != NULL) {
331 			error = EINVAL;
332 			break;
333 		}
334 
335 		watch = malloc(sizeof(*watch), M_XENSTORE, M_WAITOK);
336 		watch->watch.node = strdup(wpath, M_XENSTORE);
337 		watch->watch.callback = xs_dev_watch_cb;
338 		watch->watch.callback_data = (uintptr_t)watch;
339 		watch->watch.max_pending = max_pending_watches;
340 		watch->token = strdup(wtoken, M_XENSTORE);
341 		watch->user = u;
342 
343 		error = xs_register_watch(&watch->watch);
344 		if (error != 0) {
345 			free(watch->token, M_XENSTORE);
346 			free(watch->watch.node, M_XENSTORE);
347 			free(watch, M_XENSTORE);
348 			break;
349 		}
350 
351 		LIST_INSERT_HEAD(&u->watches, watch, list);
352 		u->u.msg.len = sizeof(ok);
353 		mtx_lock(&u->lock);
354 		xs_queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg));
355 		xs_queue_reply(u, ok, sizeof(ok));
356 		mtx_unlock(&u->lock);
357 		break;
358 	case XS_UNWATCH:
359 		u->u.msg.tx_id = 0;
360 		error = xs_dev_watch_message_parse(&u->u.msg, &wpath, &wtoken);
361 		if (error)
362 			break;
363 		watch = xs_dev_find_watch(u, wtoken);
364 		if (watch == NULL) {
365 			error = EINVAL;
366 			break;
367 		}
368 
369 		LIST_REMOVE(watch, list);
370 		xs_unregister_watch(&watch->watch);
371 		free(watch->watch.node, M_XENSTORE);
372 		free(watch->token, M_XENSTORE);
373 		free(watch, M_XENSTORE);
374 		u->u.msg.len = sizeof(ok);
375 		mtx_lock(&u->lock);
376 		xs_queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg));
377 		xs_queue_reply(u, ok, sizeof(ok));
378 		mtx_unlock(&u->lock);
379 		break;
380 	default:
381 		error = EINVAL;
382 		break;
383 	}
384 
385 	if (error != 0)
386 		xs_dev_return_error(u, error, u->u.msg.req_id, u->u.msg.tx_id);
387 
388 	/* Reset the write buffer. */
389 	u->len = 0;
390 
391 	return (0);
392 }
393 
394 static int
395 xs_dev_poll(struct cdev *dev, int events, struct thread *td)
396 {
397 	struct xs_dev_data *u;
398 	int error, mask;
399 
400 	error = devfs_get_cdevpriv((void **)&u);
401 	if (error != 0)
402 		return (POLLERR);
403 
404 	/* we can always write */
405 	mask = events & (POLLOUT | POLLWRNORM);
406 
407 	if (events & (POLLIN | POLLRDNORM)) {
408 		if (u->read_cons != u->read_prod) {
409 			mask |= events & (POLLIN | POLLRDNORM);
410 		} else {
411 			/* Record that someone is waiting */
412 			selrecord(td, &u->ev_rsel);
413 		}
414 	}
415 
416 	return (mask);
417 }
418 
419 static void
420 xs_dev_dtor(void *arg)
421 {
422 	struct xs_dev_data *u = arg;
423 	struct xs_dev_transaction *trans, *tmpt;
424 	struct xs_dev_watch *watch, *tmpw;
425 
426 	seldrain(&u->ev_rsel);
427 
428 	LIST_FOREACH_SAFE(trans, &u->transactions, list, tmpt) {
429 		xs_transaction_end(trans->handle, 1);
430 		LIST_REMOVE(trans, list);
431 		free(trans, M_XENSTORE);
432 	}
433 
434 	LIST_FOREACH_SAFE(watch, &u->watches, list, tmpw) {
435 		LIST_REMOVE(watch, list);
436 		xs_unregister_watch(&watch->watch);
437 		free(watch->watch.node, M_XENSTORE);
438 		free(watch->token, M_XENSTORE);
439 		free(watch, M_XENSTORE);
440 	}
441 	mtx_destroy(&u->lock);
442 
443 	free(u, M_XENSTORE);
444 }
445 
446 static int
447 xs_dev_open(struct cdev *dev, int oflags, int devtype, struct thread *td)
448 {
449 	struct xs_dev_data *u;
450 	int error;
451 
452 	u = malloc(sizeof(*u), M_XENSTORE, M_WAITOK|M_ZERO);
453 	mtx_init(&u->lock, "xsdev_lock", NULL, MTX_DEF);
454 	LIST_INIT(&u->transactions);
455 	LIST_INIT(&u->watches);
456 	error = devfs_set_cdevpriv(u, xs_dev_dtor);
457 	if (error != 0)
458 		free(u, M_XENSTORE);
459 
460 	return (error);
461 }
462 
463 static struct cdevsw xs_dev_cdevsw = {
464 	.d_version = D_VERSION,
465 	.d_read = xs_dev_read,
466 	.d_write = xs_dev_write,
467 	.d_open = xs_dev_open,
468 	.d_poll = xs_dev_poll,
469 	.d_name = "xs_dev",
470 };
471 
472 /*------------------ Private Device Attachment Functions  --------------------*/
473 /**
474  * \brief Identify instances of this device type in the system.
475  *
476  * \param driver  The driver performing this identify action.
477  * \param parent  The NewBus parent device for any devices this method adds.
478  */
479 static void
480 xs_dev_identify(driver_t *driver __unused, device_t parent)
481 {
482 	/*
483 	 * A single device instance for our driver is always present
484 	 * in a system operating under Xen.
485 	 */
486 	BUS_ADD_CHILD(parent, 0, driver->name, 0);
487 }
488 
489 /**
490  * \brief Probe for the existence of the Xenstore device
491  *
492  * \param dev  NewBus device_t for this instance.
493  *
494  * \return  Always returns 0 indicating success.
495  */
496 static int
497 xs_dev_probe(device_t dev)
498 {
499 
500 	device_set_desc(dev, "Xenstore user-space device");
501 	return (0);
502 }
503 
504 /**
505  * \brief Attach the Xenstore device.
506  *
507  * \param dev  NewBus device_t for this instance.
508  *
509  * \return  On success, 0. Otherwise an errno value indicating the
510  *          type of failure.
511  */
512 static int
513 xs_dev_attach(device_t dev)
514 {
515 	struct cdev *xs_cdev;
516 	struct sysctl_ctx_list *sysctl_ctx;
517 	struct sysctl_oid *sysctl_tree;
518 
519 	sysctl_ctx = device_get_sysctl_ctx(dev);
520 	sysctl_tree = device_get_sysctl_tree(dev);
521 	if (sysctl_ctx == NULL || sysctl_tree == NULL)
522 	    return (EINVAL);
523 
524 	SYSCTL_ADD_UINT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO,
525 	    "max_pending_watch_events", CTLFLAG_RW, &max_pending_watches, 0,
526 	    "maximum amount of pending watch events to be delivered");
527 
528 	xs_cdev = make_dev_credf(MAKEDEV_ETERNAL, &xs_dev_cdevsw, 0, NULL,
529 	    UID_ROOT, GID_WHEEL, 0400, "xen/xenstore");
530 	if (xs_cdev == NULL)
531 		return (EINVAL);
532 
533 	return (0);
534 }
535 
536 /*-------------------- Private Device Attachment Data  -----------------------*/
537 static device_method_t xs_dev_methods[] = {
538 	/* Device interface */
539 	DEVMETHOD(device_identify,	xs_dev_identify),
540 	DEVMETHOD(device_probe,         xs_dev_probe),
541 	DEVMETHOD(device_attach,        xs_dev_attach),
542 
543 	DEVMETHOD_END
544 };
545 
546 DEFINE_CLASS_0(xs_dev, xs_dev_driver, xs_dev_methods, 0);
547 devclass_t xs_dev_devclass;
548 
549 DRIVER_MODULE(xs_dev, xenstore, xs_dev_driver, xs_dev_devclass,
550     NULL, NULL);
551