xref: /openbmc/linux/drivers/xen/evtchn.c (revision 930beb5a) 
1 /******************************************************************************
2  * evtchn.c
3  *
4  * Driver for receiving and demuxing event-channel signals.
5  *
6  * Copyright (c) 2004-2005, K A Fraser
7  * Multi-process extensions Copyright (c) 2004, Steven Smith
8  *
9  * This program is free software; you can redistribute it and/or
10  * modify it under the terms of the GNU General Public License version 2
11  * as published by the Free Software Foundation; or, when distributed
12  * separately from the Linux kernel or incorporated into other
13  * software packages, subject to the following license:
14  *
15  * Permission is hereby granted, free of charge, to any person obtaining a copy
16  * of this source file (the "Software"), to deal in the Software without
17  * restriction, including without limitation the rights to use, copy, modify,
18  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
19  * and to permit persons to whom the Software is furnished to do so, subject to
20  * the following conditions:
21  *
22  * The above copyright notice and this permission notice shall be included in
23  * all copies or substantial portions of the Software.
24  *
25  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
26  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
27  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
28  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
29  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
30  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
31  * IN THE SOFTWARE.
32  */
33 
34 #define pr_fmt(fmt) "xen:" KBUILD_MODNAME ": " fmt
35 
36 #include <linux/module.h>
37 #include <linux/kernel.h>
38 #include <linux/sched.h>
39 #include <linux/slab.h>
40 #include <linux/string.h>
41 #include <linux/errno.h>
42 #include <linux/fs.h>
43 #include <linux/miscdevice.h>
44 #include <linux/major.h>
45 #include <linux/proc_fs.h>
46 #include <linux/stat.h>
47 #include <linux/poll.h>
48 #include <linux/irq.h>
49 #include <linux/init.h>
50 #include <linux/mutex.h>
51 #include <linux/cpu.h>
52 
53 #include <xen/xen.h>
54 #include <xen/events.h>
55 #include <xen/evtchn.h>
56 #include <asm/xen/hypervisor.h>
57 
58 struct per_user_data {
59 	struct mutex bind_mutex; /* serialize bind/unbind operations */
60 	struct rb_root evtchns;
61 
62 	/* Notification ring, accessed via /dev/xen/evtchn. */
63 #define EVTCHN_RING_SIZE     (PAGE_SIZE / sizeof(evtchn_port_t))
64 #define EVTCHN_RING_MASK(_i) ((_i)&(EVTCHN_RING_SIZE-1))
65 	evtchn_port_t *ring;
66 	unsigned int ring_cons, ring_prod, ring_overflow;
67 	struct mutex ring_cons_mutex; /* protect against concurrent readers */
68 	spinlock_t ring_prod_lock; /* product against concurrent interrupts */
69 
70 	/* Processes wait on this queue when ring is empty. */
71 	wait_queue_head_t evtchn_wait;
72 	struct fasync_struct *evtchn_async_queue;
73 	const char *name;
74 };
75 
76 struct user_evtchn {
77 	struct rb_node node;
78 	struct per_user_data *user;
79 	unsigned port;
80 	bool enabled;
81 };
82 
83 static int add_evtchn(struct per_user_data *u, struct user_evtchn *evtchn)
84 {
85 	struct rb_node **new = &(u->evtchns.rb_node), *parent = NULL;
86 
87 	while (*new) {
88 		struct user_evtchn *this;
89 
90 		this = container_of(*new, struct user_evtchn, node);
91 
92 		parent = *new;
93 		if (this->port < evtchn->port)
94 			new = &((*new)->rb_left);
95 		else if (this->port > evtchn->port)
96 			new = &((*new)->rb_right);
97 		else
98 			return -EEXIST;
99 	}
100 
101 	/* Add new node and rebalance tree. */
102 	rb_link_node(&evtchn->node, parent, new);
103 	rb_insert_color(&evtchn->node, &u->evtchns);
104 
105 	return 0;
106 }
107 
108 static void del_evtchn(struct per_user_data *u, struct user_evtchn *evtchn)
109 {
110 	rb_erase(&evtchn->node, &u->evtchns);
111 	kfree(evtchn);
112 }
113 
114 static struct user_evtchn *find_evtchn(struct per_user_data *u, unsigned port)
115 {
116 	struct rb_node *node = u->evtchns.rb_node;
117 
118 	while (node) {
119 		struct user_evtchn *evtchn;
120 
121 		evtchn = container_of(node, struct user_evtchn, node);
122 
123 		if (evtchn->port < port)
124 			node = node->rb_left;
125 		else if (evtchn->port > port)
126 			node = node->rb_right;
127 		else
128 			return evtchn;
129 	}
130 	return NULL;
131 }
132 
133 static irqreturn_t evtchn_interrupt(int irq, void *data)
134 {
135 	struct user_evtchn *evtchn = data;
136 	struct per_user_data *u = evtchn->user;
137 
138 	WARN(!evtchn->enabled,
139 	     "Interrupt for port %d, but apparently not enabled; per-user %p\n",
140 	     evtchn->port, u);
141 
142 	disable_irq_nosync(irq);
143 	evtchn->enabled = false;
144 
145 	spin_lock(&u->ring_prod_lock);
146 
147 	if ((u->ring_prod - u->ring_cons) < EVTCHN_RING_SIZE) {
148 		u->ring[EVTCHN_RING_MASK(u->ring_prod)] = evtchn->port;
149 		wmb(); /* Ensure ring contents visible */
150 		if (u->ring_cons == u->ring_prod++) {
151 			wake_up_interruptible(&u->evtchn_wait);
152 			kill_fasync(&u->evtchn_async_queue,
153 				    SIGIO, POLL_IN);
154 		}
155 	} else
156 		u->ring_overflow = 1;
157 
158 	spin_unlock(&u->ring_prod_lock);
159 
160 	return IRQ_HANDLED;
161 }
162 
163 static ssize_t evtchn_read(struct file *file, char __user *buf,
164 			   size_t count, loff_t *ppos)
165 {
166 	int rc;
167 	unsigned int c, p, bytes1 = 0, bytes2 = 0;
168 	struct per_user_data *u = file->private_data;
169 
170 	/* Whole number of ports. */
171 	count &= ~(sizeof(evtchn_port_t)-1);
172 
173 	if (count == 0)
174 		return 0;
175 
176 	if (count > PAGE_SIZE)
177 		count = PAGE_SIZE;
178 
179 	for (;;) {
180 		mutex_lock(&u->ring_cons_mutex);
181 
182 		rc = -EFBIG;
183 		if (u->ring_overflow)
184 			goto unlock_out;
185 
186 		c = u->ring_cons;
187 		p = u->ring_prod;
188 		if (c != p)
189 			break;
190 
191 		mutex_unlock(&u->ring_cons_mutex);
192 
193 		if (file->f_flags & O_NONBLOCK)
194 			return -EAGAIN;
195 
196 		rc = wait_event_interruptible(u->evtchn_wait,
197 					      u->ring_cons != u->ring_prod);
198 		if (rc)
199 			return rc;
200 	}
201 
202 	/* Byte lengths of two chunks. Chunk split (if any) is at ring wrap. */
203 	if (((c ^ p) & EVTCHN_RING_SIZE) != 0) {
204 		bytes1 = (EVTCHN_RING_SIZE - EVTCHN_RING_MASK(c)) *
205 			sizeof(evtchn_port_t);
206 		bytes2 = EVTCHN_RING_MASK(p) * sizeof(evtchn_port_t);
207 	} else {
208 		bytes1 = (p - c) * sizeof(evtchn_port_t);
209 		bytes2 = 0;
210 	}
211 
212 	/* Truncate chunks according to caller's maximum byte count. */
213 	if (bytes1 > count) {
214 		bytes1 = count;
215 		bytes2 = 0;
216 	} else if ((bytes1 + bytes2) > count) {
217 		bytes2 = count - bytes1;
218 	}
219 
220 	rc = -EFAULT;
221 	rmb(); /* Ensure that we see the port before we copy it. */
222 	if (copy_to_user(buf, &u->ring[EVTCHN_RING_MASK(c)], bytes1) ||
223 	    ((bytes2 != 0) &&
224 	     copy_to_user(&buf[bytes1], &u->ring[0], bytes2)))
225 		goto unlock_out;
226 
227 	u->ring_cons += (bytes1 + bytes2) / sizeof(evtchn_port_t);
228 	rc = bytes1 + bytes2;
229 
230  unlock_out:
231 	mutex_unlock(&u->ring_cons_mutex);
232 	return rc;
233 }
234 
235 static ssize_t evtchn_write(struct file *file, const char __user *buf,
236 			    size_t count, loff_t *ppos)
237 {
238 	int rc, i;
239 	evtchn_port_t *kbuf = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
240 	struct per_user_data *u = file->private_data;
241 
242 	if (kbuf == NULL)
243 		return -ENOMEM;
244 
245 	/* Whole number of ports. */
246 	count &= ~(sizeof(evtchn_port_t)-1);
247 
248 	rc = 0;
249 	if (count == 0)
250 		goto out;
251 
252 	if (count > PAGE_SIZE)
253 		count = PAGE_SIZE;
254 
255 	rc = -EFAULT;
256 	if (copy_from_user(kbuf, buf, count) != 0)
257 		goto out;
258 
259 	mutex_lock(&u->bind_mutex);
260 
261 	for (i = 0; i < (count/sizeof(evtchn_port_t)); i++) {
262 		unsigned port = kbuf[i];
263 		struct user_evtchn *evtchn;
264 
265 		evtchn = find_evtchn(u, port);
266 		if (evtchn && !evtchn->enabled) {
267 			evtchn->enabled = true;
268 			enable_irq(irq_from_evtchn(port));
269 		}
270 	}
271 
272 	mutex_unlock(&u->bind_mutex);
273 
274 	rc = count;
275 
276  out:
277 	free_page((unsigned long)kbuf);
278 	return rc;
279 }
280 
281 static int evtchn_bind_to_user(struct per_user_data *u, int port)
282 {
283 	struct user_evtchn *evtchn;
284 	struct evtchn_close close;
285 	int rc = 0;
286 
287 	/*
288 	 * Ports are never reused, so every caller should pass in a
289 	 * unique port.
290 	 *
291 	 * (Locking not necessary because we haven't registered the
292 	 * interrupt handler yet, and our caller has already
293 	 * serialized bind operations.)
294 	 */
295 
296 	evtchn = kzalloc(sizeof(*evtchn), GFP_KERNEL);
297 	if (!evtchn)
298 		return -ENOMEM;
299 
300 	evtchn->user = u;
301 	evtchn->port = port;
302 	evtchn->enabled = true; /* start enabled */
303 
304 	rc = add_evtchn(u, evtchn);
305 	if (rc < 0)
306 		goto err;
307 
308 	rc = bind_evtchn_to_irqhandler(port, evtchn_interrupt, 0,
309 				       u->name, evtchn);
310 	if (rc < 0)
311 		goto err;
312 
313 	rc = evtchn_make_refcounted(port);
314 	return rc;
315 
316 err:
317 	/* bind failed, should close the port now */
318 	close.port = port;
319 	if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
320 		BUG();
321 	del_evtchn(u, evtchn);
322 	return rc;
323 }
324 
325 static void evtchn_unbind_from_user(struct per_user_data *u,
326 				    struct user_evtchn *evtchn)
327 {
328 	int irq = irq_from_evtchn(evtchn->port);
329 
330 	BUG_ON(irq < 0);
331 
332 	unbind_from_irqhandler(irq, evtchn);
333 
334 	del_evtchn(u, evtchn);
335 }
336 
337 static long evtchn_ioctl(struct file *file,
338 			 unsigned int cmd, unsigned long arg)
339 {
340 	int rc;
341 	struct per_user_data *u = file->private_data;
342 	void __user *uarg = (void __user *) arg;
343 
344 	/* Prevent bind from racing with unbind */
345 	mutex_lock(&u->bind_mutex);
346 
347 	switch (cmd) {
348 	case IOCTL_EVTCHN_BIND_VIRQ: {
349 		struct ioctl_evtchn_bind_virq bind;
350 		struct evtchn_bind_virq bind_virq;
351 
352 		rc = -EFAULT;
353 		if (copy_from_user(&bind, uarg, sizeof(bind)))
354 			break;
355 
356 		bind_virq.virq = bind.virq;
357 		bind_virq.vcpu = 0;
358 		rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
359 						 &bind_virq);
360 		if (rc != 0)
361 			break;
362 
363 		rc = evtchn_bind_to_user(u, bind_virq.port);
364 		if (rc == 0)
365 			rc = bind_virq.port;
366 		break;
367 	}
368 
369 	case IOCTL_EVTCHN_BIND_INTERDOMAIN: {
370 		struct ioctl_evtchn_bind_interdomain bind;
371 		struct evtchn_bind_interdomain bind_interdomain;
372 
373 		rc = -EFAULT;
374 		if (copy_from_user(&bind, uarg, sizeof(bind)))
375 			break;
376 
377 		bind_interdomain.remote_dom  = bind.remote_domain;
378 		bind_interdomain.remote_port = bind.remote_port;
379 		rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
380 						 &bind_interdomain);
381 		if (rc != 0)
382 			break;
383 
384 		rc = evtchn_bind_to_user(u, bind_interdomain.local_port);
385 		if (rc == 0)
386 			rc = bind_interdomain.local_port;
387 		break;
388 	}
389 
390 	case IOCTL_EVTCHN_BIND_UNBOUND_PORT: {
391 		struct ioctl_evtchn_bind_unbound_port bind;
392 		struct evtchn_alloc_unbound alloc_unbound;
393 
394 		rc = -EFAULT;
395 		if (copy_from_user(&bind, uarg, sizeof(bind)))
396 			break;
397 
398 		alloc_unbound.dom        = DOMID_SELF;
399 		alloc_unbound.remote_dom = bind.remote_domain;
400 		rc = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
401 						 &alloc_unbound);
402 		if (rc != 0)
403 			break;
404 
405 		rc = evtchn_bind_to_user(u, alloc_unbound.port);
406 		if (rc == 0)
407 			rc = alloc_unbound.port;
408 		break;
409 	}
410 
411 	case IOCTL_EVTCHN_UNBIND: {
412 		struct ioctl_evtchn_unbind unbind;
413 		struct user_evtchn *evtchn;
414 
415 		rc = -EFAULT;
416 		if (copy_from_user(&unbind, uarg, sizeof(unbind)))
417 			break;
418 
419 		rc = -EINVAL;
420 		if (unbind.port >= NR_EVENT_CHANNELS)
421 			break;
422 
423 		rc = -ENOTCONN;
424 		evtchn = find_evtchn(u, unbind.port);
425 		if (!evtchn)
426 			break;
427 
428 		disable_irq(irq_from_evtchn(unbind.port));
429 		evtchn_unbind_from_user(u, evtchn);
430 		rc = 0;
431 		break;
432 	}
433 
434 	case IOCTL_EVTCHN_NOTIFY: {
435 		struct ioctl_evtchn_notify notify;
436 		struct user_evtchn *evtchn;
437 
438 		rc = -EFAULT;
439 		if (copy_from_user(&notify, uarg, sizeof(notify)))
440 			break;
441 
442 		rc = -ENOTCONN;
443 		evtchn = find_evtchn(u, notify.port);
444 		if (evtchn) {
445 			notify_remote_via_evtchn(notify.port);
446 			rc = 0;
447 		}
448 		break;
449 	}
450 
451 	case IOCTL_EVTCHN_RESET: {
452 		/* Initialise the ring to empty. Clear errors. */
453 		mutex_lock(&u->ring_cons_mutex);
454 		spin_lock_irq(&u->ring_prod_lock);
455 		u->ring_cons = u->ring_prod = u->ring_overflow = 0;
456 		spin_unlock_irq(&u->ring_prod_lock);
457 		mutex_unlock(&u->ring_cons_mutex);
458 		rc = 0;
459 		break;
460 	}
461 
462 	default:
463 		rc = -ENOSYS;
464 		break;
465 	}
466 	mutex_unlock(&u->bind_mutex);
467 
468 	return rc;
469 }
470 
471 static unsigned int evtchn_poll(struct file *file, poll_table *wait)
472 {
473 	unsigned int mask = POLLOUT | POLLWRNORM;
474 	struct per_user_data *u = file->private_data;
475 
476 	poll_wait(file, &u->evtchn_wait, wait);
477 	if (u->ring_cons != u->ring_prod)
478 		mask |= POLLIN | POLLRDNORM;
479 	if (u->ring_overflow)
480 		mask = POLLERR;
481 	return mask;
482 }
483 
484 static int evtchn_fasync(int fd, struct file *filp, int on)
485 {
486 	struct per_user_data *u = filp->private_data;
487 	return fasync_helper(fd, filp, on, &u->evtchn_async_queue);
488 }
489 
490 static int evtchn_open(struct inode *inode, struct file *filp)
491 {
492 	struct per_user_data *u;
493 
494 	u = kzalloc(sizeof(*u), GFP_KERNEL);
495 	if (u == NULL)
496 		return -ENOMEM;
497 
498 	u->name = kasprintf(GFP_KERNEL, "evtchn:%s", current->comm);
499 	if (u->name == NULL) {
500 		kfree(u);
501 		return -ENOMEM;
502 	}
503 
504 	init_waitqueue_head(&u->evtchn_wait);
505 
506 	u->ring = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
507 	if (u->ring == NULL) {
508 		kfree(u->name);
509 		kfree(u);
510 		return -ENOMEM;
511 	}
512 
513 	mutex_init(&u->bind_mutex);
514 	mutex_init(&u->ring_cons_mutex);
515 	spin_lock_init(&u->ring_prod_lock);
516 
517 	filp->private_data = u;
518 
519 	return nonseekable_open(inode, filp);
520 }
521 
522 static int evtchn_release(struct inode *inode, struct file *filp)
523 {
524 	struct per_user_data *u = filp->private_data;
525 	struct rb_node *node;
526 
527 	while ((node = u->evtchns.rb_node)) {
528 		struct user_evtchn *evtchn;
529 
530 		evtchn = rb_entry(node, struct user_evtchn, node);
531 		disable_irq(irq_from_evtchn(evtchn->port));
532 		evtchn_unbind_from_user(u, evtchn);
533 	}
534 
535 	free_page((unsigned long)u->ring);
536 	kfree(u->name);
537 	kfree(u);
538 
539 	return 0;
540 }
541 
542 static const struct file_operations evtchn_fops = {
543 	.owner   = THIS_MODULE,
544 	.read    = evtchn_read,
545 	.write   = evtchn_write,
546 	.unlocked_ioctl = evtchn_ioctl,
547 	.poll    = evtchn_poll,
548 	.fasync  = evtchn_fasync,
549 	.open    = evtchn_open,
550 	.release = evtchn_release,
551 	.llseek	 = no_llseek,
552 };
553 
554 static struct miscdevice evtchn_miscdev = {
555 	.minor        = MISC_DYNAMIC_MINOR,
556 	.name         = "xen/evtchn",
557 	.fops         = &evtchn_fops,
558 };
559 static int __init evtchn_init(void)
560 {
561 	int err;
562 
563 	if (!xen_domain())
564 		return -ENODEV;
565 
566 	/* Create '/dev/xen/evtchn'. */
567 	err = misc_register(&evtchn_miscdev);
568 	if (err != 0) {
569 		pr_err("Could not register /dev/xen/evtchn\n");
570 		return err;
571 	}
572 
573 	pr_info("Event-channel device installed\n");
574 
575 	return 0;
576 }
577 
578 static void __exit evtchn_cleanup(void)
579 {
580 	misc_deregister(&evtchn_miscdev);
581 }
582 
583 module_init(evtchn_init);
584 module_exit(evtchn_cleanup);
585 
586 MODULE_LICENSE("GPL");
587