xref: /openbmc/linux/drivers/xen/evtchn.c (revision 0b26ca68) 
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 #include <linux/mm.h>
53 #include <linux/vmalloc.h>
54 
55 #include <xen/xen.h>
56 #include <xen/events.h>
57 #include <xen/evtchn.h>
58 #include <xen/xen-ops.h>
59 #include <asm/xen/hypervisor.h>
60 
61 struct per_user_data {
62 	struct mutex bind_mutex; /* serialize bind/unbind operations */
63 	struct rb_root evtchns;
64 	unsigned int nr_evtchns;
65 
66 	/* Notification ring, accessed via /dev/xen/evtchn. */
67 	unsigned int ring_size;
68 	evtchn_port_t *ring;
69 	unsigned int ring_cons, ring_prod, ring_overflow;
70 	struct mutex ring_cons_mutex; /* protect against concurrent readers */
71 	spinlock_t ring_prod_lock; /* product against concurrent interrupts */
72 
73 	/* Processes wait on this queue when ring is empty. */
74 	wait_queue_head_t evtchn_wait;
75 	struct fasync_struct *evtchn_async_queue;
76 	const char *name;
77 
78 	domid_t restrict_domid;
79 };
80 
81 #define UNRESTRICTED_DOMID ((domid_t)-1)
82 
83 struct user_evtchn {
84 	struct rb_node node;
85 	struct per_user_data *user;
86 	evtchn_port_t port;
87 	bool enabled;
88 };
89 
90 static void evtchn_free_ring(evtchn_port_t *ring)
91 {
92 	kvfree(ring);
93 }
94 
95 static unsigned int evtchn_ring_offset(struct per_user_data *u,
96 				       unsigned int idx)
97 {
98 	return idx & (u->ring_size - 1);
99 }
100 
101 static evtchn_port_t *evtchn_ring_entry(struct per_user_data *u,
102 					unsigned int idx)
103 {
104 	return u->ring + evtchn_ring_offset(u, idx);
105 }
106 
107 static int add_evtchn(struct per_user_data *u, struct user_evtchn *evtchn)
108 {
109 	struct rb_node **new = &(u->evtchns.rb_node), *parent = NULL;
110 
111 	u->nr_evtchns++;
112 
113 	while (*new) {
114 		struct user_evtchn *this;
115 
116 		this = rb_entry(*new, struct user_evtchn, node);
117 
118 		parent = *new;
119 		if (this->port < evtchn->port)
120 			new = &((*new)->rb_left);
121 		else if (this->port > evtchn->port)
122 			new = &((*new)->rb_right);
123 		else
124 			return -EEXIST;
125 	}
126 
127 	/* Add new node and rebalance tree. */
128 	rb_link_node(&evtchn->node, parent, new);
129 	rb_insert_color(&evtchn->node, &u->evtchns);
130 
131 	return 0;
132 }
133 
134 static void del_evtchn(struct per_user_data *u, struct user_evtchn *evtchn)
135 {
136 	u->nr_evtchns--;
137 	rb_erase(&evtchn->node, &u->evtchns);
138 	kfree(evtchn);
139 }
140 
141 static struct user_evtchn *find_evtchn(struct per_user_data *u,
142 				       evtchn_port_t port)
143 {
144 	struct rb_node *node = u->evtchns.rb_node;
145 
146 	while (node) {
147 		struct user_evtchn *evtchn;
148 
149 		evtchn = rb_entry(node, struct user_evtchn, node);
150 
151 		if (evtchn->port < port)
152 			node = node->rb_left;
153 		else if (evtchn->port > port)
154 			node = node->rb_right;
155 		else
156 			return evtchn;
157 	}
158 	return NULL;
159 }
160 
161 static irqreturn_t evtchn_interrupt(int irq, void *data)
162 {
163 	struct user_evtchn *evtchn = data;
164 	struct per_user_data *u = evtchn->user;
165 
166 	WARN(!evtchn->enabled,
167 	     "Interrupt for port %u, but apparently not enabled; per-user %p\n",
168 	     evtchn->port, u);
169 
170 	evtchn->enabled = false;
171 
172 	spin_lock(&u->ring_prod_lock);
173 
174 	if ((u->ring_prod - u->ring_cons) < u->ring_size) {
175 		*evtchn_ring_entry(u, u->ring_prod) = evtchn->port;
176 		wmb(); /* Ensure ring contents visible */
177 		if (u->ring_cons == u->ring_prod++) {
178 			wake_up_interruptible(&u->evtchn_wait);
179 			kill_fasync(&u->evtchn_async_queue,
180 				    SIGIO, POLL_IN);
181 		}
182 	} else
183 		u->ring_overflow = 1;
184 
185 	spin_unlock(&u->ring_prod_lock);
186 
187 	return IRQ_HANDLED;
188 }
189 
190 static ssize_t evtchn_read(struct file *file, char __user *buf,
191 			   size_t count, loff_t *ppos)
192 {
193 	int rc;
194 	unsigned int c, p, bytes1 = 0, bytes2 = 0;
195 	struct per_user_data *u = file->private_data;
196 
197 	/* Whole number of ports. */
198 	count &= ~(sizeof(evtchn_port_t)-1);
199 
200 	if (count == 0)
201 		return 0;
202 
203 	if (count > PAGE_SIZE)
204 		count = PAGE_SIZE;
205 
206 	for (;;) {
207 		mutex_lock(&u->ring_cons_mutex);
208 
209 		rc = -EFBIG;
210 		if (u->ring_overflow)
211 			goto unlock_out;
212 
213 		c = u->ring_cons;
214 		p = u->ring_prod;
215 		if (c != p)
216 			break;
217 
218 		mutex_unlock(&u->ring_cons_mutex);
219 
220 		if (file->f_flags & O_NONBLOCK)
221 			return -EAGAIN;
222 
223 		rc = wait_event_interruptible(u->evtchn_wait,
224 					      u->ring_cons != u->ring_prod);
225 		if (rc)
226 			return rc;
227 	}
228 
229 	/* Byte lengths of two chunks. Chunk split (if any) is at ring wrap. */
230 	if (((c ^ p) & u->ring_size) != 0) {
231 		bytes1 = (u->ring_size - evtchn_ring_offset(u, c)) *
232 			sizeof(evtchn_port_t);
233 		bytes2 = evtchn_ring_offset(u, p) * sizeof(evtchn_port_t);
234 	} else {
235 		bytes1 = (p - c) * sizeof(evtchn_port_t);
236 		bytes2 = 0;
237 	}
238 
239 	/* Truncate chunks according to caller's maximum byte count. */
240 	if (bytes1 > count) {
241 		bytes1 = count;
242 		bytes2 = 0;
243 	} else if ((bytes1 + bytes2) > count) {
244 		bytes2 = count - bytes1;
245 	}
246 
247 	rc = -EFAULT;
248 	rmb(); /* Ensure that we see the port before we copy it. */
249 	if (copy_to_user(buf, evtchn_ring_entry(u, c), bytes1) ||
250 	    ((bytes2 != 0) &&
251 	     copy_to_user(&buf[bytes1], &u->ring[0], bytes2)))
252 		goto unlock_out;
253 
254 	u->ring_cons += (bytes1 + bytes2) / sizeof(evtchn_port_t);
255 	rc = bytes1 + bytes2;
256 
257  unlock_out:
258 	mutex_unlock(&u->ring_cons_mutex);
259 	return rc;
260 }
261 
262 static ssize_t evtchn_write(struct file *file, const char __user *buf,
263 			    size_t count, loff_t *ppos)
264 {
265 	int rc, i;
266 	evtchn_port_t *kbuf = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
267 	struct per_user_data *u = file->private_data;
268 
269 	if (kbuf == NULL)
270 		return -ENOMEM;
271 
272 	/* Whole number of ports. */
273 	count &= ~(sizeof(evtchn_port_t)-1);
274 
275 	rc = 0;
276 	if (count == 0)
277 		goto out;
278 
279 	if (count > PAGE_SIZE)
280 		count = PAGE_SIZE;
281 
282 	rc = -EFAULT;
283 	if (copy_from_user(kbuf, buf, count) != 0)
284 		goto out;
285 
286 	mutex_lock(&u->bind_mutex);
287 
288 	for (i = 0; i < (count/sizeof(evtchn_port_t)); i++) {
289 		evtchn_port_t port = kbuf[i];
290 		struct user_evtchn *evtchn;
291 
292 		evtchn = find_evtchn(u, port);
293 		if (evtchn && !evtchn->enabled) {
294 			evtchn->enabled = true;
295 			xen_irq_lateeoi(irq_from_evtchn(port), 0);
296 		}
297 	}
298 
299 	mutex_unlock(&u->bind_mutex);
300 
301 	rc = count;
302 
303  out:
304 	free_page((unsigned long)kbuf);
305 	return rc;
306 }
307 
308 static int evtchn_resize_ring(struct per_user_data *u)
309 {
310 	unsigned int new_size;
311 	evtchn_port_t *new_ring, *old_ring;
312 
313 	/*
314 	 * Ensure the ring is large enough to capture all possible
315 	 * events. i.e., one free slot for each bound event.
316 	 */
317 	if (u->nr_evtchns <= u->ring_size)
318 		return 0;
319 
320 	if (u->ring_size == 0)
321 		new_size = 64;
322 	else
323 		new_size = 2 * u->ring_size;
324 
325 	new_ring = kvmalloc_array(new_size, sizeof(*new_ring), GFP_KERNEL);
326 	if (!new_ring)
327 		return -ENOMEM;
328 
329 	old_ring = u->ring;
330 
331 	/*
332 	 * Access to the ring contents is serialized by either the
333 	 * prod /or/ cons lock so take both when resizing.
334 	 */
335 	mutex_lock(&u->ring_cons_mutex);
336 	spin_lock_irq(&u->ring_prod_lock);
337 
338 	/*
339 	 * Copy the old ring contents to the new ring.
340 	 *
341 	 * To take care of wrapping, a full ring, and the new index
342 	 * pointing into the second half, simply copy the old contents
343 	 * twice.
344 	 *
345 	 * +---------+    +------------------+
346 	 * |34567  12| -> |34567  1234567  12|
347 	 * +-----p-c-+    +-------c------p---+
348 	 */
349 	memcpy(new_ring, old_ring, u->ring_size * sizeof(*u->ring));
350 	memcpy(new_ring + u->ring_size, old_ring,
351 	       u->ring_size * sizeof(*u->ring));
352 
353 	u->ring = new_ring;
354 	u->ring_size = new_size;
355 
356 	spin_unlock_irq(&u->ring_prod_lock);
357 	mutex_unlock(&u->ring_cons_mutex);
358 
359 	evtchn_free_ring(old_ring);
360 
361 	return 0;
362 }
363 
364 static int evtchn_bind_to_user(struct per_user_data *u, evtchn_port_t port)
365 {
366 	struct user_evtchn *evtchn;
367 	struct evtchn_close close;
368 	int rc = 0;
369 
370 	/*
371 	 * Ports are never reused, so every caller should pass in a
372 	 * unique port.
373 	 *
374 	 * (Locking not necessary because we haven't registered the
375 	 * interrupt handler yet, and our caller has already
376 	 * serialized bind operations.)
377 	 */
378 
379 	evtchn = kzalloc(sizeof(*evtchn), GFP_KERNEL);
380 	if (!evtchn)
381 		return -ENOMEM;
382 
383 	evtchn->user = u;
384 	evtchn->port = port;
385 	evtchn->enabled = true; /* start enabled */
386 
387 	rc = add_evtchn(u, evtchn);
388 	if (rc < 0)
389 		goto err;
390 
391 	rc = evtchn_resize_ring(u);
392 	if (rc < 0)
393 		goto err;
394 
395 	rc = bind_evtchn_to_irqhandler_lateeoi(port, evtchn_interrupt, 0,
396 					       u->name, evtchn);
397 	if (rc < 0)
398 		goto err;
399 
400 	rc = evtchn_make_refcounted(port);
401 	return rc;
402 
403 err:
404 	/* bind failed, should close the port now */
405 	close.port = port;
406 	if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
407 		BUG();
408 	del_evtchn(u, evtchn);
409 	return rc;
410 }
411 
412 static void evtchn_unbind_from_user(struct per_user_data *u,
413 				    struct user_evtchn *evtchn)
414 {
415 	int irq = irq_from_evtchn(evtchn->port);
416 
417 	BUG_ON(irq < 0);
418 
419 	unbind_from_irqhandler(irq, evtchn);
420 
421 	del_evtchn(u, evtchn);
422 }
423 
424 static long evtchn_ioctl(struct file *file,
425 			 unsigned int cmd, unsigned long arg)
426 {
427 	int rc;
428 	struct per_user_data *u = file->private_data;
429 	void __user *uarg = (void __user *) arg;
430 
431 	/* Prevent bind from racing with unbind */
432 	mutex_lock(&u->bind_mutex);
433 
434 	switch (cmd) {
435 	case IOCTL_EVTCHN_BIND_VIRQ: {
436 		struct ioctl_evtchn_bind_virq bind;
437 		struct evtchn_bind_virq bind_virq;
438 
439 		rc = -EACCES;
440 		if (u->restrict_domid != UNRESTRICTED_DOMID)
441 			break;
442 
443 		rc = -EFAULT;
444 		if (copy_from_user(&bind, uarg, sizeof(bind)))
445 			break;
446 
447 		bind_virq.virq = bind.virq;
448 		bind_virq.vcpu = xen_vcpu_nr(0);
449 		rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
450 						 &bind_virq);
451 		if (rc != 0)
452 			break;
453 
454 		rc = evtchn_bind_to_user(u, bind_virq.port);
455 		if (rc == 0)
456 			rc = bind_virq.port;
457 		break;
458 	}
459 
460 	case IOCTL_EVTCHN_BIND_INTERDOMAIN: {
461 		struct ioctl_evtchn_bind_interdomain bind;
462 		struct evtchn_bind_interdomain bind_interdomain;
463 
464 		rc = -EFAULT;
465 		if (copy_from_user(&bind, uarg, sizeof(bind)))
466 			break;
467 
468 		rc = -EACCES;
469 		if (u->restrict_domid != UNRESTRICTED_DOMID &&
470 		    u->restrict_domid != bind.remote_domain)
471 			break;
472 
473 		bind_interdomain.remote_dom  = bind.remote_domain;
474 		bind_interdomain.remote_port = bind.remote_port;
475 		rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
476 						 &bind_interdomain);
477 		if (rc != 0)
478 			break;
479 
480 		rc = evtchn_bind_to_user(u, bind_interdomain.local_port);
481 		if (rc == 0)
482 			rc = bind_interdomain.local_port;
483 		break;
484 	}
485 
486 	case IOCTL_EVTCHN_BIND_UNBOUND_PORT: {
487 		struct ioctl_evtchn_bind_unbound_port bind;
488 		struct evtchn_alloc_unbound alloc_unbound;
489 
490 		rc = -EACCES;
491 		if (u->restrict_domid != UNRESTRICTED_DOMID)
492 			break;
493 
494 		rc = -EFAULT;
495 		if (copy_from_user(&bind, uarg, sizeof(bind)))
496 			break;
497 
498 		alloc_unbound.dom        = DOMID_SELF;
499 		alloc_unbound.remote_dom = bind.remote_domain;
500 		rc = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
501 						 &alloc_unbound);
502 		if (rc != 0)
503 			break;
504 
505 		rc = evtchn_bind_to_user(u, alloc_unbound.port);
506 		if (rc == 0)
507 			rc = alloc_unbound.port;
508 		break;
509 	}
510 
511 	case IOCTL_EVTCHN_UNBIND: {
512 		struct ioctl_evtchn_unbind unbind;
513 		struct user_evtchn *evtchn;
514 
515 		rc = -EFAULT;
516 		if (copy_from_user(&unbind, uarg, sizeof(unbind)))
517 			break;
518 
519 		rc = -EINVAL;
520 		if (unbind.port >= xen_evtchn_nr_channels())
521 			break;
522 
523 		rc = -ENOTCONN;
524 		evtchn = find_evtchn(u, unbind.port);
525 		if (!evtchn)
526 			break;
527 
528 		disable_irq(irq_from_evtchn(unbind.port));
529 		evtchn_unbind_from_user(u, evtchn);
530 		rc = 0;
531 		break;
532 	}
533 
534 	case IOCTL_EVTCHN_NOTIFY: {
535 		struct ioctl_evtchn_notify notify;
536 		struct user_evtchn *evtchn;
537 
538 		rc = -EFAULT;
539 		if (copy_from_user(&notify, uarg, sizeof(notify)))
540 			break;
541 
542 		rc = -ENOTCONN;
543 		evtchn = find_evtchn(u, notify.port);
544 		if (evtchn) {
545 			notify_remote_via_evtchn(notify.port);
546 			rc = 0;
547 		}
548 		break;
549 	}
550 
551 	case IOCTL_EVTCHN_RESET: {
552 		/* Initialise the ring to empty. Clear errors. */
553 		mutex_lock(&u->ring_cons_mutex);
554 		spin_lock_irq(&u->ring_prod_lock);
555 		u->ring_cons = u->ring_prod = u->ring_overflow = 0;
556 		spin_unlock_irq(&u->ring_prod_lock);
557 		mutex_unlock(&u->ring_cons_mutex);
558 		rc = 0;
559 		break;
560 	}
561 
562 	case IOCTL_EVTCHN_RESTRICT_DOMID: {
563 		struct ioctl_evtchn_restrict_domid ierd;
564 
565 		rc = -EACCES;
566 		if (u->restrict_domid != UNRESTRICTED_DOMID)
567 			break;
568 
569 		rc = -EFAULT;
570 		if (copy_from_user(&ierd, uarg, sizeof(ierd)))
571 		    break;
572 
573 		rc = -EINVAL;
574 		if (ierd.domid == 0 || ierd.domid >= DOMID_FIRST_RESERVED)
575 			break;
576 
577 		u->restrict_domid = ierd.domid;
578 		rc = 0;
579 
580 		break;
581 	}
582 
583 	default:
584 		rc = -ENOSYS;
585 		break;
586 	}
587 	mutex_unlock(&u->bind_mutex);
588 
589 	return rc;
590 }
591 
592 static __poll_t evtchn_poll(struct file *file, poll_table *wait)
593 {
594 	__poll_t mask = EPOLLOUT | EPOLLWRNORM;
595 	struct per_user_data *u = file->private_data;
596 
597 	poll_wait(file, &u->evtchn_wait, wait);
598 	if (u->ring_cons != u->ring_prod)
599 		mask |= EPOLLIN | EPOLLRDNORM;
600 	if (u->ring_overflow)
601 		mask = EPOLLERR;
602 	return mask;
603 }
604 
605 static int evtchn_fasync(int fd, struct file *filp, int on)
606 {
607 	struct per_user_data *u = filp->private_data;
608 	return fasync_helper(fd, filp, on, &u->evtchn_async_queue);
609 }
610 
611 static int evtchn_open(struct inode *inode, struct file *filp)
612 {
613 	struct per_user_data *u;
614 
615 	u = kzalloc(sizeof(*u), GFP_KERNEL);
616 	if (u == NULL)
617 		return -ENOMEM;
618 
619 	u->name = kasprintf(GFP_KERNEL, "evtchn:%s", current->comm);
620 	if (u->name == NULL) {
621 		kfree(u);
622 		return -ENOMEM;
623 	}
624 
625 	init_waitqueue_head(&u->evtchn_wait);
626 
627 	mutex_init(&u->bind_mutex);
628 	mutex_init(&u->ring_cons_mutex);
629 	spin_lock_init(&u->ring_prod_lock);
630 
631 	u->restrict_domid = UNRESTRICTED_DOMID;
632 
633 	filp->private_data = u;
634 
635 	return stream_open(inode, filp);
636 }
637 
638 static int evtchn_release(struct inode *inode, struct file *filp)
639 {
640 	struct per_user_data *u = filp->private_data;
641 	struct rb_node *node;
642 
643 	while ((node = u->evtchns.rb_node)) {
644 		struct user_evtchn *evtchn;
645 
646 		evtchn = rb_entry(node, struct user_evtchn, node);
647 		disable_irq(irq_from_evtchn(evtchn->port));
648 		evtchn_unbind_from_user(u, evtchn);
649 	}
650 
651 	evtchn_free_ring(u->ring);
652 	kfree(u->name);
653 	kfree(u);
654 
655 	return 0;
656 }
657 
658 static const struct file_operations evtchn_fops = {
659 	.owner   = THIS_MODULE,
660 	.read    = evtchn_read,
661 	.write   = evtchn_write,
662 	.unlocked_ioctl = evtchn_ioctl,
663 	.poll    = evtchn_poll,
664 	.fasync  = evtchn_fasync,
665 	.open    = evtchn_open,
666 	.release = evtchn_release,
667 	.llseek	 = no_llseek,
668 };
669 
670 static struct miscdevice evtchn_miscdev = {
671 	.minor        = MISC_DYNAMIC_MINOR,
672 	.name         = "xen/evtchn",
673 	.fops         = &evtchn_fops,
674 };
675 static int __init evtchn_init(void)
676 {
677 	int err;
678 
679 	if (!xen_domain())
680 		return -ENODEV;
681 
682 	/* Create '/dev/xen/evtchn'. */
683 	err = misc_register(&evtchn_miscdev);
684 	if (err != 0) {
685 		pr_err("Could not register /dev/xen/evtchn\n");
686 		return err;
687 	}
688 
689 	pr_info("Event-channel device installed\n");
690 
691 	return 0;
692 }
693 
694 static void __exit evtchn_cleanup(void)
695 {
696 	misc_deregister(&evtchn_miscdev);
697 }
698 
699 module_init(evtchn_init);
700 module_exit(evtchn_cleanup);
701 
702 MODULE_LICENSE("GPL");
703