projects/linux·blob master · refs

732 lines · 16.5 KiB

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
61struct 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
83struct user_evtchn {
84 struct rb_node node;
85 struct per_user_data *user;
86 evtchn_port_t port;
87 bool enabled;
88 bool unbinding;
89};
90
91static void evtchn_free_ring(evtchn_port_t *ring)
92{
93 kvfree(ring);
94}
95
96static unsigned int evtchn_ring_offset(struct per_user_data *u,
97 unsigned int idx)
98{
99 return idx & (u->ring_size - 1);
100}
101
102static evtchn_port_t *evtchn_ring_entry(struct per_user_data *u,
103 unsigned int idx)
104{
105 return u->ring + evtchn_ring_offset(u, idx);
106}
107
108static int add_evtchn(struct per_user_data *u, struct user_evtchn *evtchn)
109{
110 struct rb_node **new = &(u->evtchns.rb_node), *parent = NULL;
111
112 u->nr_evtchns++;
113
114 while (*new) {
115 struct user_evtchn *this;
116
117 this = rb_entry(*new, struct user_evtchn, node);
118
119 parent = *new;
120 if (this->port < evtchn->port)
121 new = &((*new)->rb_left);
122 else if (this->port > evtchn->port)
123 new = &((*new)->rb_right);
124 else
125 return -EEXIST;
126 }
127
128 /* Add new node and rebalance tree. */
129 rb_link_node(&evtchn->node, parent, new);
130 rb_insert_color(&evtchn->node, &u->evtchns);
131
132 return 0;
133}
134
135static void del_evtchn(struct per_user_data *u, struct user_evtchn *evtchn)
136{
137 u->nr_evtchns--;
138 rb_erase(&evtchn->node, &u->evtchns);
139 kfree(evtchn);
140}
141
142static struct user_evtchn *find_evtchn(struct per_user_data *u,
143 evtchn_port_t port)
144{
145 struct rb_node *node = u->evtchns.rb_node;
146
147 while (node) {
148 struct user_evtchn *evtchn;
149
150 evtchn = rb_entry(node, struct user_evtchn, node);
151
152 if (evtchn->port < port)
153 node = node->rb_left;
154 else if (evtchn->port > port)
155 node = node->rb_right;
156 else
157 return evtchn;
158 }
159 return NULL;
160}
161
162static irqreturn_t evtchn_interrupt(int irq, void *data)
163{
164 struct user_evtchn *evtchn = data;
165 struct per_user_data *u = evtchn->user;
166 unsigned int prod, cons;
167
168 /* Handler might be called when tearing down the IRQ. */
169 if (evtchn->unbinding)
170 return IRQ_HANDLED;
171
172 WARN(!evtchn->enabled,
173 "Interrupt for port %u, but apparently not enabled; per-user %p\n",
174 evtchn->port, u);
175
176 evtchn->enabled = false;
177
178 spin_lock(&u->ring_prod_lock);
179
180 prod = READ_ONCE(u->ring_prod);
181 cons = READ_ONCE(u->ring_cons);
182
183 if ((prod - cons) < u->ring_size) {
184 *evtchn_ring_entry(u, prod) = evtchn->port;
185 smp_wmb(); /* Ensure ring contents visible */
186 WRITE_ONCE(u->ring_prod, prod + 1);
187 if (cons == prod) {
188 wake_up_interruptible(&u->evtchn_wait);
189 kill_fasync(&u->evtchn_async_queue,
190 SIGIO, POLL_IN);
191 }
192 } else
193 u->ring_overflow = 1;
194
195 spin_unlock(&u->ring_prod_lock);
196
197 return IRQ_HANDLED;
198}
199
200static ssize_t evtchn_read(struct file *file, char __user *buf,
201 size_t count, loff_t *ppos)
202{
203 int rc;
204 unsigned int c, p, bytes1 = 0, bytes2 = 0;
205 struct per_user_data *u = file->private_data;
206
207 /* Whole number of ports. */
208 count &= ~(sizeof(evtchn_port_t)-1);
209
210 if (count == 0)
211 return 0;
212
213 if (count > PAGE_SIZE)
214 count = PAGE_SIZE;
215
216 for (;;) {
217 mutex_lock(&u->ring_cons_mutex);
218
219 rc = -EFBIG;
220 if (u->ring_overflow)
221 goto unlock_out;
222
223 c = READ_ONCE(u->ring_cons);
224 p = READ_ONCE(u->ring_prod);
225 if (c != p)
226 break;
227
228 mutex_unlock(&u->ring_cons_mutex);
229
230 if (file->f_flags & O_NONBLOCK)
231 return -EAGAIN;
232
233 rc = wait_event_interruptible(u->evtchn_wait,
234 READ_ONCE(u->ring_cons) != READ_ONCE(u->ring_prod));
235 if (rc)
236 return rc;
237 }
238
239 /* Byte lengths of two chunks. Chunk split (if any) is at ring wrap. */
240 if (((c ^ p) & u->ring_size) != 0) {
241 bytes1 = (u->ring_size - evtchn_ring_offset(u, c)) *
242 sizeof(evtchn_port_t);
243 bytes2 = evtchn_ring_offset(u, p) * sizeof(evtchn_port_t);
244 } else {
245 bytes1 = (p - c) * sizeof(evtchn_port_t);
246 bytes2 = 0;
247 }
248
249 /* Truncate chunks according to caller's maximum byte count. */
250 if (bytes1 > count) {
251 bytes1 = count;
252 bytes2 = 0;
253 } else if ((bytes1 + bytes2) > count) {
254 bytes2 = count - bytes1;
255 }
256
257 rc = -EFAULT;
258 smp_rmb(); /* Ensure that we see the port before we copy it. */
259 if (copy_to_user(buf, evtchn_ring_entry(u, c), bytes1) ||
260 ((bytes2 != 0) &&
261 copy_to_user(&buf[bytes1], &u->ring[0], bytes2)))
262 goto unlock_out;
263
264 WRITE_ONCE(u->ring_cons, c + (bytes1 + bytes2) / sizeof(evtchn_port_t));
265 rc = bytes1 + bytes2;
266
267 unlock_out:
268 mutex_unlock(&u->ring_cons_mutex);
269 return rc;
270}
271
272static ssize_t evtchn_write(struct file *file, const char __user *buf,
273 size_t count, loff_t *ppos)
274{
275 int rc, i;
276 evtchn_port_t *kbuf = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
277 struct per_user_data *u = file->private_data;
278
279 if (kbuf == NULL)
280 return -ENOMEM;
281
282 /* Whole number of ports. */
283 count &= ~(sizeof(evtchn_port_t)-1);
284
285 rc = 0;
286 if (count == 0)
287 goto out;
288
289 if (count > PAGE_SIZE)
290 count = PAGE_SIZE;
291
292 rc = -EFAULT;
293 if (copy_from_user(kbuf, buf, count) != 0)
294 goto out;
295
296 mutex_lock(&u->bind_mutex);
297
298 for (i = 0; i < (count/sizeof(evtchn_port_t)); i++) {
299 evtchn_port_t port = kbuf[i];
300 struct user_evtchn *evtchn;
301
302 evtchn = find_evtchn(u, port);
303 if (evtchn && !evtchn->enabled) {
304 evtchn->enabled = true;
305 xen_irq_lateeoi(irq_from_evtchn(port), 0);
306 }
307 }
308
309 mutex_unlock(&u->bind_mutex);
310
311 rc = count;
312
313 out:
314 free_page((unsigned long)kbuf);
315 return rc;
316}
317
318static int evtchn_resize_ring(struct per_user_data *u)
319{
320 unsigned int new_size;
321 evtchn_port_t *new_ring, *old_ring;
322
323 /*
324 * Ensure the ring is large enough to capture all possible
325 * events. i.e., one free slot for each bound event.
326 */
327 if (u->nr_evtchns <= u->ring_size)
328 return 0;
329
330 if (u->ring_size == 0)
331 new_size = 64;
332 else
333 new_size = 2 * u->ring_size;
334
335 new_ring = kvmalloc_objs(*new_ring, new_size);
336 if (!new_ring)
337 return -ENOMEM;
338
339 old_ring = u->ring;
340
341 /*
342 * Access to the ring contents is serialized by either the
343 * prod /or/ cons lock so take both when resizing.
344 */
345 mutex_lock(&u->ring_cons_mutex);
346 spin_lock_irq(&u->ring_prod_lock);
347
348 /*
349 * Copy the old ring contents to the new ring.
350 *
351 * To take care of wrapping, a full ring, and the new index
352 * pointing into the second half, simply copy the old contents
353 * twice.
354 *
355 * +---------+ +------------------+
356 * |34567 12| -> |34567 1234567 12|
357 * +-----p-c-+ +-------c------p---+
358 */
359 memcpy(new_ring, old_ring, u->ring_size * sizeof(*u->ring));
360 memcpy(new_ring + u->ring_size, old_ring,
361 u->ring_size * sizeof(*u->ring));
362
363 u->ring = new_ring;
364 u->ring_size = new_size;
365
366 spin_unlock_irq(&u->ring_prod_lock);
367 mutex_unlock(&u->ring_cons_mutex);
368
369 evtchn_free_ring(old_ring);
370
371 return 0;
372}
373
374static int evtchn_bind_to_user(struct per_user_data *u, evtchn_port_t port,
375 bool is_static)
376{
377 struct user_evtchn *evtchn;
378 int rc = 0;
379
380 /*
381 * Ports are never reused, so every caller should pass in a
382 * unique port.
383 *
384 * (Locking not necessary because we haven't registered the
385 * interrupt handler yet, and our caller has already
386 * serialized bind operations.)
387 */
388
389 evtchn = kzalloc_obj(*evtchn);
390 if (!evtchn)
391 return -ENOMEM;
392
393 evtchn->user = u;
394 evtchn->port = port;
395 evtchn->enabled = true; /* start enabled */
396
397 rc = add_evtchn(u, evtchn);
398 if (rc < 0)
399 goto err;
400
401 rc = evtchn_resize_ring(u);
402 if (rc < 0)
403 goto err;
404
405 rc = bind_evtchn_to_irqhandler_lateeoi(port, evtchn_interrupt, IRQF_SHARED,
406 u->name, evtchn);
407 if (rc < 0)
408 goto err;
409
410 rc = evtchn_make_refcounted(port, is_static);
411 return rc;
412
413err:
414 /* bind failed, should close the port now */
415 if (!is_static)
416 xen_evtchn_close(port);
417
418 del_evtchn(u, evtchn);
419 return rc;
420}
421
422static void evtchn_unbind_from_user(struct per_user_data *u,
423 struct user_evtchn *evtchn)
424{
425 int irq = irq_from_evtchn(evtchn->port);
426
427 BUG_ON(irq < 0);
428
429 evtchn->unbinding = true;
430 unbind_from_irqhandler(irq, evtchn);
431
432 del_evtchn(u, evtchn);
433}
434
435static long evtchn_ioctl(struct file *file,
436 unsigned int cmd, unsigned long arg)
437{
438 int rc;
439 struct per_user_data *u = file->private_data;
440 void __user *uarg = (void __user *) arg;
441
442 /* Prevent bind from racing with unbind */
443 mutex_lock(&u->bind_mutex);
444
445 switch (cmd) {
446 case IOCTL_EVTCHN_BIND_VIRQ: {
447 struct ioctl_evtchn_bind_virq bind;
448 struct evtchn_bind_virq bind_virq;
449
450 rc = -EACCES;
451 if (u->restrict_domid != UNRESTRICTED_DOMID)
452 break;
453
454 rc = -EFAULT;
455 if (copy_from_user(&bind, uarg, sizeof(bind)))
456 break;
457
458 bind_virq.virq = bind.virq;
459 bind_virq.vcpu = xen_vcpu_nr(0);
460 rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
461 &bind_virq);
462 if (rc != 0)
463 break;
464
465 rc = evtchn_bind_to_user(u, bind_virq.port, false);
466 if (rc == 0)
467 rc = bind_virq.port;
468 break;
469 }
470
471 case IOCTL_EVTCHN_BIND_INTERDOMAIN: {
472 struct ioctl_evtchn_bind_interdomain bind;
473 struct evtchn_bind_interdomain bind_interdomain;
474
475 rc = -EFAULT;
476 if (copy_from_user(&bind, uarg, sizeof(bind)))
477 break;
478
479 rc = -EACCES;
480 if (u->restrict_domid != UNRESTRICTED_DOMID &&
481 u->restrict_domid != bind.remote_domain)
482 break;
483
484 bind_interdomain.remote_dom = bind.remote_domain;
485 bind_interdomain.remote_port = bind.remote_port;
486 rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
487 &bind_interdomain);
488 if (rc != 0)
489 break;
490
491 rc = evtchn_bind_to_user(u, bind_interdomain.local_port, false);
492 if (rc == 0)
493 rc = bind_interdomain.local_port;
494 break;
495 }
496
497 case IOCTL_EVTCHN_BIND_UNBOUND_PORT: {
498 struct ioctl_evtchn_bind_unbound_port bind;
499 struct evtchn_alloc_unbound alloc_unbound;
500
501 rc = -EACCES;
502 if (u->restrict_domid != UNRESTRICTED_DOMID)
503 break;
504
505 rc = -EFAULT;
506 if (copy_from_user(&bind, uarg, sizeof(bind)))
507 break;
508
509 alloc_unbound.dom = DOMID_SELF;
510 alloc_unbound.remote_dom = bind.remote_domain;
511 rc = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
512 &alloc_unbound);
513 if (rc != 0)
514 break;
515
516 rc = evtchn_bind_to_user(u, alloc_unbound.port, false);
517 if (rc == 0)
518 rc = alloc_unbound.port;
519 break;
520 }
521
522 case IOCTL_EVTCHN_UNBIND: {
523 struct ioctl_evtchn_unbind unbind;
524 struct user_evtchn *evtchn;
525
526 rc = -EFAULT;
527 if (copy_from_user(&unbind, uarg, sizeof(unbind)))
528 break;
529
530 rc = -EINVAL;
531 if (unbind.port >= xen_evtchn_nr_channels())
532 break;
533
534 rc = -ENOTCONN;
535 evtchn = find_evtchn(u, unbind.port);
536 if (!evtchn)
537 break;
538
539 disable_irq(irq_from_evtchn(unbind.port));
540 evtchn_unbind_from_user(u, evtchn);
541 rc = 0;
542 break;
543 }
544
545 case IOCTL_EVTCHN_BIND_STATIC: {
546 struct ioctl_evtchn_bind bind;
547 struct user_evtchn *evtchn;
548
549 rc = -EFAULT;
550 if (copy_from_user(&bind, uarg, sizeof(bind)))
551 break;
552
553 rc = -EISCONN;
554 evtchn = find_evtchn(u, bind.port);
555 if (evtchn)
556 break;
557
558 rc = evtchn_bind_to_user(u, bind.port, true);
559 break;
560 }
561
562 case IOCTL_EVTCHN_NOTIFY: {
563 struct ioctl_evtchn_notify notify;
564 struct user_evtchn *evtchn;
565
566 rc = -EFAULT;
567 if (copy_from_user(&notify, uarg, sizeof(notify)))
568 break;
569
570 rc = -ENOTCONN;
571 evtchn = find_evtchn(u, notify.port);
572 if (evtchn) {
573 notify_remote_via_evtchn(notify.port);
574 rc = 0;
575 }
576 break;
577 }
578
579 case IOCTL_EVTCHN_RESET: {
580 /* Initialise the ring to empty. Clear errors. */
581 mutex_lock(&u->ring_cons_mutex);
582 spin_lock_irq(&u->ring_prod_lock);
583 WRITE_ONCE(u->ring_cons, 0);
584 WRITE_ONCE(u->ring_prod, 0);
585 u->ring_overflow = 0;
586 spin_unlock_irq(&u->ring_prod_lock);
587 mutex_unlock(&u->ring_cons_mutex);
588 rc = 0;
589 break;
590 }
591
592 case IOCTL_EVTCHN_RESTRICT_DOMID: {
593 struct ioctl_evtchn_restrict_domid ierd;
594
595 rc = -EACCES;
596 if (u->restrict_domid != UNRESTRICTED_DOMID)
597 break;
598
599 rc = -EFAULT;
600 if (copy_from_user(&ierd, uarg, sizeof(ierd)))
601 break;
602
603 rc = -EINVAL;
604 if (ierd.domid == 0 || ierd.domid >= DOMID_FIRST_RESERVED)
605 break;
606
607 u->restrict_domid = ierd.domid;
608 rc = 0;
609
610 break;
611 }
612
613 default:
614 rc = -ENOSYS;
615 break;
616 }
617 mutex_unlock(&u->bind_mutex);
618
619 return rc;
620}
621
622static __poll_t evtchn_poll(struct file *file, poll_table *wait)
623{
624 __poll_t mask = EPOLLOUT | EPOLLWRNORM;
625 struct per_user_data *u = file->private_data;
626
627 poll_wait(file, &u->evtchn_wait, wait);
628 if (READ_ONCE(u->ring_cons) != READ_ONCE(u->ring_prod))
629 mask |= EPOLLIN | EPOLLRDNORM;
630 if (u->ring_overflow)
631 mask = EPOLLERR;
632 return mask;
633}
634
635static int evtchn_fasync(int fd, struct file *filp, int on)
636{
637 struct per_user_data *u = filp->private_data;
638 return fasync_helper(fd, filp, on, &u->evtchn_async_queue);
639}
640
641static int evtchn_open(struct inode *inode, struct file *filp)
642{
643 struct per_user_data *u;
644
645 u = kzalloc_obj(*u);
646 if (u == NULL)
647 return -ENOMEM;
648
649 u->name = kasprintf(GFP_KERNEL, "evtchn:%s", current->comm);
650 if (u->name == NULL) {
651 kfree(u);
652 return -ENOMEM;
653 }
654
655 init_waitqueue_head(&u->evtchn_wait);
656
657 mutex_init(&u->bind_mutex);
658 mutex_init(&u->ring_cons_mutex);
659 spin_lock_init(&u->ring_prod_lock);
660
661 u->restrict_domid = UNRESTRICTED_DOMID;
662
663 filp->private_data = u;
664
665 return stream_open(inode, filp);
666}
667
668static int evtchn_release(struct inode *inode, struct file *filp)
669{
670 struct per_user_data *u = filp->private_data;
671 struct rb_node *node;
672
673 while ((node = u->evtchns.rb_node)) {
674 struct user_evtchn *evtchn;
675
676 evtchn = rb_entry(node, struct user_evtchn, node);
677 disable_irq(irq_from_evtchn(evtchn->port));
678 evtchn_unbind_from_user(u, evtchn);
679 }
680
681 evtchn_free_ring(u->ring);
682 kfree(u->name);
683 kfree(u);
684
685 return 0;
686}
687
688static const struct file_operations evtchn_fops = {
689 .owner = THIS_MODULE,
690 .read = evtchn_read,
691 .write = evtchn_write,
692 .unlocked_ioctl = evtchn_ioctl,
693 .poll = evtchn_poll,
694 .fasync = evtchn_fasync,
695 .open = evtchn_open,
696 .release = evtchn_release,
697};
698
699static struct miscdevice evtchn_miscdev = {
700 .minor = MISC_DYNAMIC_MINOR,
701 .name = "xen/evtchn",
702 .fops = &evtchn_fops,
703};
704static int __init evtchn_init(void)
705{
706 int err;
707
708 if (!xen_domain())
709 return -ENODEV;
710
711 /* Create '/dev/xen/evtchn'. */
712 err = misc_register(&evtchn_miscdev);
713 if (err != 0) {
714 pr_err("Could not register /dev/xen/evtchn\n");
715 return err;
716 }
717
718 pr_info("Event-channel device installed\n");
719
720 return 0;
721}
722
723static void __exit evtchn_cleanup(void)
724{
725 misc_deregister(&evtchn_miscdev);
726}
727
728module_init(evtchn_init);
729module_exit(evtchn_cleanup);
730
731MODULE_DESCRIPTION("Xen /dev/xen/evtchn device driver");
732MODULE_LICENSE("GPL");