vhost.c 42.3 KB
Newer Older
1 2 3 4 5 6
/* Copyright (C) 2009 Red Hat, Inc.
 * Copyright (C) 2006 Rusty Russell IBM Corporation
 *
 * Author: Michael S. Tsirkin <mst@redhat.com>
 *
 * Inspiration, some code, and most witty comments come from
Rob Landley's avatar
Rob Landley committed
7
 * Documentation/virtual/lguest/lguest.c, by Rusty Russell
8 9 10 11 12 13 14 15
 *
 * This work is licensed under the terms of the GNU GPL, version 2.
 *
 * Generic code for virtio server in host kernel.
 */

#include <linux/eventfd.h>
#include <linux/vhost.h>
16
#include <linux/uio.h>
17
#include <linux/mm.h>
18
#include <linux/mmu_context.h>
19 20 21 22 23
#include <linux/miscdevice.h>
#include <linux/mutex.h>
#include <linux/poll.h>
#include <linux/file.h>
#include <linux/highmem.h>
24
#include <linux/slab.h>
25
#include <linux/kthread.h>
26
#include <linux/cgroup.h>
27
#include <linux/module.h>
28
#include <linux/sort.h>
29 30 31 32 33 34 35 36

#include "vhost.h"

enum {
	VHOST_MEMORY_MAX_NREGIONS = 64,
	VHOST_MEMORY_F_LOG = 0x1,
};

37 38
#define vhost_used_event(vq) ((__virtio16 __user *)&vq->avail->ring[vq->num])
#define vhost_avail_event(vq) ((__virtio16 __user *)&vq->used->ring[vq->num])
39

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
#ifdef CONFIG_VHOST_CROSS_ENDIAN_LEGACY
static void vhost_vq_reset_user_be(struct vhost_virtqueue *vq)
{
	vq->user_be = !virtio_legacy_is_little_endian();
}

static long vhost_set_vring_endian(struct vhost_virtqueue *vq, int __user *argp)
{
	struct vhost_vring_state s;

	if (vq->private_data)
		return -EBUSY;

	if (copy_from_user(&s, argp, sizeof(s)))
		return -EFAULT;

	if (s.num != VHOST_VRING_LITTLE_ENDIAN &&
	    s.num != VHOST_VRING_BIG_ENDIAN)
		return -EINVAL;

	vq->user_be = s.num;

	return 0;
}

static long vhost_get_vring_endian(struct vhost_virtqueue *vq, u32 idx,
				   int __user *argp)
{
	struct vhost_vring_state s = {
		.index = idx,
		.num = vq->user_be
	};

	if (copy_to_user(argp, &s, sizeof(s)))
		return -EFAULT;

	return 0;
}

static void vhost_init_is_le(struct vhost_virtqueue *vq)
{
	/* Note for legacy virtio: user_be is initialized at reset time
	 * according to the host endianness. If userspace does not set an
	 * explicit endianness, the default behavior is native endian, as
	 * expected by legacy virtio.
	 */
	vq->is_le = vhost_has_feature(vq, VIRTIO_F_VERSION_1) || !vq->user_be;
}
#else
static void vhost_vq_reset_user_be(struct vhost_virtqueue *vq)
{
}

static long vhost_set_vring_endian(struct vhost_virtqueue *vq, int __user *argp)
{
	return -ENOIOCTLCMD;
}

static long vhost_get_vring_endian(struct vhost_virtqueue *vq, u32 idx,
				   int __user *argp)
{
	return -ENOIOCTLCMD;
}

static void vhost_init_is_le(struct vhost_virtqueue *vq)
{
	if (vhost_has_feature(vq, VIRTIO_F_VERSION_1))
		vq->is_le = true;
}
#endif /* CONFIG_VHOST_CROSS_ENDIAN_LEGACY */

111 112 113 114 115
static void vhost_poll_func(struct file *file, wait_queue_head_t *wqh,
			    poll_table *pt)
{
	struct vhost_poll *poll;

116
	poll = container_of(pt, struct vhost_poll, table);
117 118 119 120 121 122 123
	poll->wqh = wqh;
	add_wait_queue(wqh, &poll->wait);
}

static int vhost_poll_wakeup(wait_queue_t *wait, unsigned mode, int sync,
			     void *key)
{
124 125
	struct vhost_poll *poll = container_of(wait, struct vhost_poll, wait);

126 127 128
	if (!((unsigned long)key & poll->mask))
		return 0;

129
	vhost_poll_queue(poll);
130 131 132
	return 0;
}

133
void vhost_work_init(struct vhost_work *work, vhost_work_fn_t fn)
134 135 136 137 138 139 140
{
	INIT_LIST_HEAD(&work->node);
	work->fn = fn;
	init_waitqueue_head(&work->done);
	work->flushing = 0;
	work->queue_seq = work->done_seq = 0;
}
141
EXPORT_SYMBOL_GPL(vhost_work_init);
142

143
/* Init poll structure */
144 145
void vhost_poll_init(struct vhost_poll *poll, vhost_work_fn_t fn,
		     unsigned long mask, struct vhost_dev *dev)
146 147 148 149
{
	init_waitqueue_func_entry(&poll->wait, vhost_poll_wakeup);
	init_poll_funcptr(&poll->table, vhost_poll_func);
	poll->mask = mask;
150
	poll->dev = dev;
151
	poll->wqh = NULL;
152

153
	vhost_work_init(&poll->work, fn);
154
}
155
EXPORT_SYMBOL_GPL(vhost_poll_init);
156 157 158

/* Start polling a file. We add ourselves to file's wait queue. The caller must
 * keep a reference to a file until after vhost_poll_stop is called. */
159
int vhost_poll_start(struct vhost_poll *poll, struct file *file)
160 161
{
	unsigned long mask;
162
	int ret = 0;
163

164 165 166
	if (poll->wqh)
		return 0;

167 168 169
	mask = file->f_op->poll(file, &poll->table);
	if (mask)
		vhost_poll_wakeup(&poll->wait, 0, 0, (void *)mask);
170 171 172 173 174 175 176
	if (mask & POLLERR) {
		if (poll->wqh)
			remove_wait_queue(poll->wqh, &poll->wait);
		ret = -EINVAL;
	}

	return ret;
177
}
178
EXPORT_SYMBOL_GPL(vhost_poll_start);
179 180 181 182 183

/* Stop polling a file. After this function returns, it becomes safe to drop the
 * file reference. You must also flush afterwards. */
void vhost_poll_stop(struct vhost_poll *poll)
{
184 185 186 187
	if (poll->wqh) {
		remove_wait_queue(poll->wqh, &poll->wait);
		poll->wqh = NULL;
	}
188
}
189
EXPORT_SYMBOL_GPL(vhost_poll_stop);
190

191 192 193 194
static bool vhost_work_seq_done(struct vhost_dev *dev, struct vhost_work *work,
				unsigned seq)
{
	int left;
195

196 197 198 199 200 201
	spin_lock_irq(&dev->work_lock);
	left = seq - work->done_seq;
	spin_unlock_irq(&dev->work_lock);
	return left <= 0;
}

202
void vhost_work_flush(struct vhost_dev *dev, struct vhost_work *work)
203
{
204 205 206
	unsigned seq;
	int flushing;

207
	spin_lock_irq(&dev->work_lock);
208 209
	seq = work->queue_seq;
	work->flushing++;
210
	spin_unlock_irq(&dev->work_lock);
211
	wait_event(work->done, vhost_work_seq_done(dev, work, seq));
212
	spin_lock_irq(&dev->work_lock);
213
	flushing = --work->flushing;
214
	spin_unlock_irq(&dev->work_lock);
215
	BUG_ON(flushing < 0);
216
}
217
EXPORT_SYMBOL_GPL(vhost_work_flush);
218

219 220 221 222 223 224
/* Flush any work that has been scheduled. When calling this, don't hold any
 * locks that are also used by the callback. */
void vhost_poll_flush(struct vhost_poll *poll)
{
	vhost_work_flush(poll->dev, &poll->work);
}
225
EXPORT_SYMBOL_GPL(vhost_poll_flush);
226

227
void vhost_work_queue(struct vhost_dev *dev, struct vhost_work *work)
228
{
229 230 231 232 233 234
	unsigned long flags;

	spin_lock_irqsave(&dev->work_lock, flags);
	if (list_empty(&work->node)) {
		list_add_tail(&work->node, &dev->work_list);
		work->queue_seq++;
235
		spin_unlock_irqrestore(&dev->work_lock, flags);
236
		wake_up_process(dev->worker);
237 238
	} else {
		spin_unlock_irqrestore(&dev->work_lock, flags);
239
	}
240
}
241
EXPORT_SYMBOL_GPL(vhost_work_queue);
242

243 244 245 246
void vhost_poll_queue(struct vhost_poll *poll)
{
	vhost_work_queue(poll->dev, &poll->work);
}
247
EXPORT_SYMBOL_GPL(vhost_poll_queue);
248

249 250 251 252 253 254 255 256 257 258
static void vhost_vq_reset(struct vhost_dev *dev,
			   struct vhost_virtqueue *vq)
{
	vq->num = 1;
	vq->desc = NULL;
	vq->avail = NULL;
	vq->used = NULL;
	vq->last_avail_idx = 0;
	vq->avail_idx = 0;
	vq->last_used_idx = 0;
259 260
	vq->signalled_used = 0;
	vq->signalled_used_valid = false;
261 262 263 264
	vq->used_flags = 0;
	vq->log_used = false;
	vq->log_addr = -1ull;
	vq->private_data = NULL;
265
	vq->acked_features = 0;
266 267 268 269 270 271
	vq->log_base = NULL;
	vq->error_ctx = NULL;
	vq->error = NULL;
	vq->kick = NULL;
	vq->call_ctx = NULL;
	vq->call = NULL;
272
	vq->log_ctx = NULL;
273
	vq->memory = NULL;
274 275
	vq->is_le = virtio_legacy_is_little_endian();
	vhost_vq_reset_user_be(vq);
276 277
}

278 279 280 281 282
static int vhost_worker(void *data)
{
	struct vhost_dev *dev = data;
	struct vhost_work *work = NULL;
	unsigned uninitialized_var(seq);
283
	mm_segment_t oldfs = get_fs();
284

285
	set_fs(USER_DS);
286 287
	use_mm(dev->mm);

288 289 290 291 292 293 294 295 296 297 298 299 300 301
	for (;;) {
		/* mb paired w/ kthread_stop */
		set_current_state(TASK_INTERRUPTIBLE);

		spin_lock_irq(&dev->work_lock);
		if (work) {
			work->done_seq = seq;
			if (work->flushing)
				wake_up_all(&work->done);
		}

		if (kthread_should_stop()) {
			spin_unlock_irq(&dev->work_lock);
			__set_current_state(TASK_RUNNING);
302
			break;
303 304 305 306 307 308 309 310 311 312 313 314 315
		}
		if (!list_empty(&dev->work_list)) {
			work = list_first_entry(&dev->work_list,
						struct vhost_work, node);
			list_del_init(&work->node);
			seq = work->queue_seq;
		} else
			work = NULL;
		spin_unlock_irq(&dev->work_lock);

		if (work) {
			__set_current_state(TASK_RUNNING);
			work->fn(work);
316 317
			if (need_resched())
				schedule();
318 319 320 321
		} else
			schedule();

	}
322
	unuse_mm(dev->mm);
323
	set_fs(oldfs);
324
	return 0;
325 326
}

327 328 329 330 331 332 333 334 335 336
static void vhost_vq_free_iovecs(struct vhost_virtqueue *vq)
{
	kfree(vq->indirect);
	vq->indirect = NULL;
	kfree(vq->log);
	vq->log = NULL;
	kfree(vq->heads);
	vq->heads = NULL;
}

Jason Wang's avatar
Jason Wang committed
337 338 339
/* Helper to allocate iovec buffers for all vqs. */
static long vhost_dev_alloc_iovecs(struct vhost_dev *dev)
{
340
	struct vhost_virtqueue *vq;
Jason Wang's avatar
Jason Wang committed
341
	int i;
342

Jason Wang's avatar
Jason Wang committed
343
	for (i = 0; i < dev->nvqs; ++i) {
344 345 346 347 348 349
		vq = dev->vqs[i];
		vq->indirect = kmalloc(sizeof *vq->indirect * UIO_MAXIOV,
				       GFP_KERNEL);
		vq->log = kmalloc(sizeof *vq->log * UIO_MAXIOV, GFP_KERNEL);
		vq->heads = kmalloc(sizeof *vq->heads * UIO_MAXIOV, GFP_KERNEL);
		if (!vq->indirect || !vq->log || !vq->heads)
Jason Wang's avatar
Jason Wang committed
350 351 352
			goto err_nomem;
	}
	return 0;
353

Jason Wang's avatar
Jason Wang committed
354
err_nomem:
355
	for (; i >= 0; --i)
356
		vhost_vq_free_iovecs(dev->vqs[i]);
Jason Wang's avatar
Jason Wang committed
357 358 359 360 361 362
	return -ENOMEM;
}

static void vhost_dev_free_iovecs(struct vhost_dev *dev)
{
	int i;
363

364
	for (i = 0; i < dev->nvqs; ++i)
365
		vhost_vq_free_iovecs(dev->vqs[i]);
Jason Wang's avatar
Jason Wang committed
366 367
}

Zhi Yong Wu's avatar
Zhi Yong Wu committed
368
void vhost_dev_init(struct vhost_dev *dev,
369
		    struct vhost_virtqueue **vqs, int nvqs)
370
{
371
	struct vhost_virtqueue *vq;
372
	int i;
373

374 375 376 377 378 379 380
	dev->vqs = vqs;
	dev->nvqs = nvqs;
	mutex_init(&dev->mutex);
	dev->log_ctx = NULL;
	dev->log_file = NULL;
	dev->memory = NULL;
	dev->mm = NULL;
381 382 383
	spin_lock_init(&dev->work_lock);
	INIT_LIST_HEAD(&dev->work_list);
	dev->worker = NULL;
384 385

	for (i = 0; i < dev->nvqs; ++i) {
386 387 388 389 390 391 392 393 394 395
		vq = dev->vqs[i];
		vq->log = NULL;
		vq->indirect = NULL;
		vq->heads = NULL;
		vq->dev = dev;
		mutex_init(&vq->mutex);
		vhost_vq_reset(dev, vq);
		if (vq->handle_kick)
			vhost_poll_init(&vq->poll, vq->handle_kick,
					POLLIN, dev);
396 397
	}
}
398
EXPORT_SYMBOL_GPL(vhost_dev_init);
399 400 401 402 403 404 405

/* Caller should have device mutex */
long vhost_dev_check_owner(struct vhost_dev *dev)
{
	/* Are you the owner? If not, I don't think you mean to do that */
	return dev->mm == current->mm ? 0 : -EPERM;
}
406
EXPORT_SYMBOL_GPL(vhost_dev_check_owner);
407

408
struct vhost_attach_cgroups_struct {
409 410 411
	struct vhost_work work;
	struct task_struct *owner;
	int ret;
412 413 414 415
};

static void vhost_attach_cgroups_work(struct vhost_work *work)
{
416 417 418 419
	struct vhost_attach_cgroups_struct *s;

	s = container_of(work, struct vhost_attach_cgroups_struct, work);
	s->ret = cgroup_attach_task_all(s->owner, current);
420 421 422 423
}

static int vhost_attach_cgroups(struct vhost_dev *dev)
{
424 425 426 427 428 429 430
	struct vhost_attach_cgroups_struct attach;

	attach.owner = current;
	vhost_work_init(&attach.work, vhost_attach_cgroups_work);
	vhost_work_queue(dev, &attach.work);
	vhost_work_flush(dev, &attach.work);
	return attach.ret;
431 432
}

433 434 435 436 437
/* Caller should have device mutex */
bool vhost_dev_has_owner(struct vhost_dev *dev)
{
	return dev->mm;
}
438
EXPORT_SYMBOL_GPL(vhost_dev_has_owner);
439

440
/* Caller should have device mutex */
Asias He's avatar
Asias He committed
441
long vhost_dev_set_owner(struct vhost_dev *dev)
442
{
443 444
	struct task_struct *worker;
	int err;
445

446
	/* Is there an owner already? */
447
	if (vhost_dev_has_owner(dev)) {
448 449 450
		err = -EBUSY;
		goto err_mm;
	}
451

452 453
	/* No owner, become one */
	dev->mm = get_task_mm(current);
454 455 456 457 458 459 460
	worker = kthread_create(vhost_worker, dev, "vhost-%d", current->pid);
	if (IS_ERR(worker)) {
		err = PTR_ERR(worker);
		goto err_worker;
	}

	dev->worker = worker;
461 462 463
	wake_up_process(worker);	/* avoid contributing to loadavg */

	err = vhost_attach_cgroups(dev);
464 465
	if (err)
		goto err_cgroup;
466

Jason Wang's avatar
Jason Wang committed
467 468 469 470
	err = vhost_dev_alloc_iovecs(dev);
	if (err)
		goto err_cgroup;

471
	return 0;
472 473
err_cgroup:
	kthread_stop(worker);
474
	dev->worker = NULL;
475 476 477 478 479 480
err_worker:
	if (dev->mm)
		mmput(dev->mm);
	dev->mm = NULL;
err_mm:
	return err;
481
}
482
EXPORT_SYMBOL_GPL(vhost_dev_set_owner);
483

484
struct vhost_memory *vhost_dev_reset_owner_prepare(void)
485
{
486 487
	return kmalloc(offsetof(struct vhost_memory, regions), GFP_KERNEL);
}
488
EXPORT_SYMBOL_GPL(vhost_dev_reset_owner_prepare);
489

490 491 492
/* Caller should have device mutex */
void vhost_dev_reset_owner(struct vhost_dev *dev, struct vhost_memory *memory)
{
493 494
	int i;

495
	vhost_dev_cleanup(dev, true);
496

497
	/* Restore memory to default empty mapping. */
498
	memory->nregions = 0;
499 500 501 502 503 504
	dev->memory = memory;
	/* We don't need VQ locks below since vhost_dev_cleanup makes sure
	 * VQs aren't running.
	 */
	for (i = 0; i < dev->nvqs; ++i)
		dev->vqs[i]->memory = memory;
505
}
506
EXPORT_SYMBOL_GPL(vhost_dev_reset_owner);
507

508
void vhost_dev_stop(struct vhost_dev *dev)
509 510
{
	int i;
511 512

	for (i = 0; i < dev->nvqs; ++i) {
513 514 515
		if (dev->vqs[i]->kick && dev->vqs[i]->handle_kick) {
			vhost_poll_stop(&dev->vqs[i]->poll);
			vhost_poll_flush(&dev->vqs[i]->poll);
516
		}
517 518
	}
}
519
EXPORT_SYMBOL_GPL(vhost_dev_stop);
520

521 522
/* Caller should have device mutex if and only if locked is set */
void vhost_dev_cleanup(struct vhost_dev *dev, bool locked)
523 524
{
	int i;
525

526
	for (i = 0; i < dev->nvqs; ++i) {
527 528 529 530 531 532 533 534 535 536 537
		if (dev->vqs[i]->error_ctx)
			eventfd_ctx_put(dev->vqs[i]->error_ctx);
		if (dev->vqs[i]->error)
			fput(dev->vqs[i]->error);
		if (dev->vqs[i]->kick)
			fput(dev->vqs[i]->kick);
		if (dev->vqs[i]->call_ctx)
			eventfd_ctx_put(dev->vqs[i]->call_ctx);
		if (dev->vqs[i]->call)
			fput(dev->vqs[i]->call);
		vhost_vq_reset(dev, dev->vqs[i]);
538
	}
Jason Wang's avatar
Jason Wang committed
539
	vhost_dev_free_iovecs(dev);
540 541 542 543 544 545 546
	if (dev->log_ctx)
		eventfd_ctx_put(dev->log_ctx);
	dev->log_ctx = NULL;
	if (dev->log_file)
		fput(dev->log_file);
	dev->log_file = NULL;
	/* No one will access memory at this point */
547 548
	kfree(dev->memory);
	dev->memory = NULL;
549
	WARN_ON(!list_empty(&dev->work_list));
550 551 552 553
	if (dev->worker) {
		kthread_stop(dev->worker);
		dev->worker = NULL;
	}
554 555 556
	if (dev->mm)
		mmput(dev->mm);
	dev->mm = NULL;
557
}
558
EXPORT_SYMBOL_GPL(vhost_dev_cleanup);
559 560 561 562

static int log_access_ok(void __user *log_base, u64 addr, unsigned long sz)
{
	u64 a = addr / VHOST_PAGE_SIZE / 8;
563

564 565 566
	/* Make sure 64 bit math will not overflow. */
	if (a > ULONG_MAX - (unsigned long)log_base ||
	    a + (unsigned long)log_base > ULONG_MAX)
567
		return 0;
568 569 570 571 572 573 574 575 576 577

	return access_ok(VERIFY_WRITE, log_base + a,
			 (sz + VHOST_PAGE_SIZE * 8 - 1) / VHOST_PAGE_SIZE / 8);
}

/* Caller should have vq mutex and device mutex. */
static int vq_memory_access_ok(void __user *log_base, struct vhost_memory *mem,
			       int log_all)
{
	int i;
578

Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
579 580
	if (!mem)
		return 0;
581

582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603
	for (i = 0; i < mem->nregions; ++i) {
		struct vhost_memory_region *m = mem->regions + i;
		unsigned long a = m->userspace_addr;
		if (m->memory_size > ULONG_MAX)
			return 0;
		else if (!access_ok(VERIFY_WRITE, (void __user *)a,
				    m->memory_size))
			return 0;
		else if (log_all && !log_access_ok(log_base,
						   m->guest_phys_addr,
						   m->memory_size))
			return 0;
	}
	return 1;
}

/* Can we switch to this memory table? */
/* Caller should have device mutex but not vq mutex */
static int memory_access_ok(struct vhost_dev *d, struct vhost_memory *mem,
			    int log_all)
{
	int i;
604

605 606
	for (i = 0; i < d->nvqs; ++i) {
		int ok;
607 608
		bool log;

609
		mutex_lock(&d->vqs[i]->mutex);
610
		log = log_all || vhost_has_feature(d->vqs[i], VHOST_F_LOG_ALL);
611
		/* If ring is inactive, will check when it's enabled. */
612
		if (d->vqs[i]->private_data)
613
			ok = vq_memory_access_ok(d->vqs[i]->log_base, mem, log);
614 615
		else
			ok = 1;
616
		mutex_unlock(&d->vqs[i]->mutex);
617 618 619 620 621 622
		if (!ok)
			return 0;
	}
	return 1;
}

623
static int vq_access_ok(struct vhost_virtqueue *vq, unsigned int num,
624 625 626 627
			struct vring_desc __user *desc,
			struct vring_avail __user *avail,
			struct vring_used __user *used)
{
628
	size_t s = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
629 630
	return access_ok(VERIFY_READ, desc, num * sizeof *desc) &&
	       access_ok(VERIFY_READ, avail,
631
			 sizeof *avail + num * sizeof *avail->ring + s) &&
632
	       access_ok(VERIFY_WRITE, used,
633
			sizeof *used + num * sizeof *used->ring + s);
634 635 636 637 638 639
}

/* Can we log writes? */
/* Caller should have device mutex but not vq mutex */
int vhost_log_access_ok(struct vhost_dev *dev)
{
640
	return memory_access_ok(dev, dev->memory, 1);
641
}
642
EXPORT_SYMBOL_GPL(vhost_log_access_ok);
643 644 645

/* Verify access for write logging. */
/* Caller should have vq mutex and device mutex */
646
static int vq_log_access_ok(struct vhost_virtqueue *vq,
647
			    void __user *log_base)
648
{
649
	size_t s = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
Arnd Bergmann's avatar
Arnd Bergmann committed
650

651
	return vq_memory_access_ok(log_base, vq->memory,
652
				   vhost_has_feature(vq, VHOST_F_LOG_ALL)) &&
653 654
		(!vq->log_used || log_access_ok(log_base, vq->log_addr,
					sizeof *vq->used +
655
					vq->num * sizeof *vq->used->ring + s));
656 657 658 659 660 661
}

/* Can we start vq? */
/* Caller should have vq mutex and device mutex */
int vhost_vq_access_ok(struct vhost_virtqueue *vq)
{
662 663
	return vq_access_ok(vq, vq->num, vq->desc, vq->avail, vq->used) &&
		vq_log_access_ok(vq, vq->log_base);
664
}
665
EXPORT_SYMBOL_GPL(vhost_vq_access_ok);
666

667 668 669 670 671 672 673 674 675 676
static int vhost_memory_reg_sort_cmp(const void *p1, const void *p2)
{
	const struct vhost_memory_region *r1 = p1, *r2 = p2;
	if (r1->guest_phys_addr < r2->guest_phys_addr)
		return 1;
	if (r1->guest_phys_addr > r2->guest_phys_addr)
		return -1;
	return 0;
}

677 678 679 680
static long vhost_set_memory(struct vhost_dev *d, struct vhost_memory __user *m)
{
	struct vhost_memory mem, *newmem, *oldmem;
	unsigned long size = offsetof(struct vhost_memory, regions);
681
	int i;
682

683 684
	if (copy_from_user(&mem, m, size))
		return -EFAULT;
685 686 687 688 689 690 691 692 693
	if (mem.padding)
		return -EOPNOTSUPP;
	if (mem.nregions > VHOST_MEMORY_MAX_NREGIONS)
		return -E2BIG;
	newmem = kmalloc(size + mem.nregions * sizeof *m->regions, GFP_KERNEL);
	if (!newmem)
		return -ENOMEM;

	memcpy(newmem, &mem, size);
694 695
	if (copy_from_user(newmem->regions, m->regions,
			   mem.nregions * sizeof *m->regions)) {
696
		kvfree(newmem);
697
		return -EFAULT;
698
	}
699 700
	sort(newmem->regions, newmem->nregions, sizeof(*newmem->regions),
		vhost_memory_reg_sort_cmp, NULL);
701

702
	if (!memory_access_ok(d, newmem, 0)) {
703
		kfree(newmem);
704
		return -EFAULT;
705
	}
706 707
	oldmem = d->memory;
	d->memory = newmem;
708

709
	/* All memory accesses are done under some VQ mutex. */
710 711
	for (i = 0; i < d->nvqs; ++i) {
		mutex_lock(&d->vqs[i]->mutex);
712
		d->vqs[i]->memory = newmem;
713 714
		mutex_unlock(&d->vqs[i]->mutex);
	}
715 716 717 718
	kfree(oldmem);
	return 0;
}

719
long vhost_vring_ioctl(struct vhost_dev *d, int ioctl, void __user *argp)
720
{
721 722
	struct file *eventfp, *filep = NULL;
	bool pollstart = false, pollstop = false;
723 724 725 726 727 728 729 730 731 732 733 734
	struct eventfd_ctx *ctx = NULL;
	u32 __user *idxp = argp;
	struct vhost_virtqueue *vq;
	struct vhost_vring_state s;
	struct vhost_vring_file f;
	struct vhost_vring_addr a;
	u32 idx;
	long r;

	r = get_user(idx, idxp);
	if (r < 0)
		return r;
735
	if (idx >= d->nvqs)
736 737
		return -ENOBUFS;

738
	vq = d->vqs[idx];
739 740 741 742 743 744 745 746 747 748 749

	mutex_lock(&vq->mutex);

	switch (ioctl) {
	case VHOST_SET_VRING_NUM:
		/* Resizing ring with an active backend?
		 * You don't want to do that. */
		if (vq->private_data) {
			r = -EBUSY;
			break;
		}
750 751
		if (copy_from_user(&s, argp, sizeof s)) {
			r = -EFAULT;
752
			break;
753
		}
754 755 756 757 758 759 760 761 762 763 764 765 766
		if (!s.num || s.num > 0xffff || (s.num & (s.num - 1))) {
			r = -EINVAL;
			break;
		}
		vq->num = s.num;
		break;
	case VHOST_SET_VRING_BASE:
		/* Moving base with an active backend?
		 * You don't want to do that. */
		if (vq->private_data) {
			r = -EBUSY;
			break;
		}
767 768
		if (copy_from_user(&s, argp, sizeof s)) {
			r = -EFAULT;
769
			break;
770
		}
771 772 773 774 775 776 777 778 779 780 781
		if (s.num > 0xffff) {
			r = -EINVAL;
			break;
		}
		vq->last_avail_idx = s.num;
		/* Forget the cached index value. */
		vq->avail_idx = vq->last_avail_idx;
		break;
	case VHOST_GET_VRING_BASE:
		s.index = idx;
		s.num = vq->last_avail_idx;
782 783
		if (copy_to_user(argp, &s, sizeof s))
			r = -EFAULT;
784 785
		break;
	case VHOST_SET_VRING_ADDR:
786 787
		if (copy_from_user(&a, argp, sizeof a)) {
			r = -EFAULT;
788
			break;
789
		}
790 791 792 793 794 795 796 797 798 799 800 801
		if (a.flags & ~(0x1 << VHOST_VRING_F_LOG)) {
			r = -EOPNOTSUPP;
			break;
		}
		/* For 32bit, verify that the top 32bits of the user
		   data are set to zero. */
		if ((u64)(unsigned long)a.desc_user_addr != a.desc_user_addr ||
		    (u64)(unsigned long)a.used_user_addr != a.used_user_addr ||
		    (u64)(unsigned long)a.avail_user_addr != a.avail_user_addr) {
			r = -EFAULT;
			break;
		}
802 803 804 805 806 807 808

		/* Make sure it's safe to cast pointers to vring types. */
		BUILD_BUG_ON(__alignof__ *vq->avail > VRING_AVAIL_ALIGN_SIZE);
		BUILD_BUG_ON(__alignof__ *vq->used > VRING_USED_ALIGN_SIZE);
		if ((a.avail_user_addr & (VRING_AVAIL_ALIGN_SIZE - 1)) ||
		    (a.used_user_addr & (VRING_USED_ALIGN_SIZE - 1)) ||
		    (a.log_guest_addr & (sizeof(u64) - 1))) {
809 810 811 812 813 814 815 816
			r = -EINVAL;
			break;
		}

		/* We only verify access here if backend is configured.
		 * If it is not, we don't as size might not have been setup.
		 * We will verify when backend is configured. */
		if (vq->private_data) {
817
			if (!vq_access_ok(vq, vq->num,
818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841
				(void __user *)(unsigned long)a.desc_user_addr,
				(void __user *)(unsigned long)a.avail_user_addr,
				(void __user *)(unsigned long)a.used_user_addr)) {
				r = -EINVAL;
				break;
			}

			/* Also validate log access for used ring if enabled. */
			if ((a.flags & (0x1 << VHOST_VRING_F_LOG)) &&
			    !log_access_ok(vq->log_base, a.log_guest_addr,
					   sizeof *vq->used +
					   vq->num * sizeof *vq->used->ring)) {
				r = -EINVAL;
				break;
			}
		}

		vq->log_used = !!(a.flags & (0x1 << VHOST_VRING_F_LOG));
		vq->desc = (void __user *)(unsigned long)a.desc_user_addr;
		vq->avail = (void __user *)(unsigned long)a.avail_user_addr;
		vq->log_addr = a.log_guest_addr;
		vq->used = (void __user *)(unsigned long)a.used_user_addr;
		break;
	case VHOST_SET_VRING_KICK:
842 843
		if (copy_from_user(&f, argp, sizeof f)) {
			r = -EFAULT;
844
			break;
845
		}
846
		eventfp = f.fd == -1 ? NULL : eventfd_fget(f.fd);
847 848 849 850
		if (IS_ERR(eventfp)) {
			r = PTR_ERR(eventfp);
			break;
		}
851
		if (eventfp != vq->kick) {
852 853
			pollstop = (filep = vq->kick) != NULL;
			pollstart = (vq->kick = eventfp) != NULL;
854 855 856 857
		} else
			filep = eventfp;
		break;
	case VHOST_SET_VRING_CALL:
858 859
		if (copy_from_user(&f, argp, sizeof f)) {
			r = -EFAULT;
860
			break;
861
		}
862
		eventfp = f.fd == -1 ? NULL : eventfd_fget(f.fd);
863 864 865 866
		if (IS_ERR(eventfp)) {
			r = PTR_ERR(eventfp);
			break;
		}
867 868 869 870 871 872 873 874 875 876
		if (eventfp != vq->call) {
			filep = vq->call;
			ctx = vq->call_ctx;
			vq->call = eventfp;
			vq->call_ctx = eventfp ?
				eventfd_ctx_fileget(eventfp) : NULL;
		} else
			filep = eventfp;
		break;
	case VHOST_SET_VRING_ERR:
877 878
		if (copy_from_user(&f, argp, sizeof f)) {
			r = -EFAULT;
879
			break;
880
		}
881
		eventfp = f.fd == -1 ? NULL : eventfd_fget(f.fd);
882 883 884 885
		if (IS_ERR(eventfp)) {
			r = PTR_ERR(eventfp);
			break;
		}
886 887 888 889 890 891 892 893 894
		if (eventfp != vq->error) {
			filep = vq->error;
			vq->error = eventfp;
			ctx = vq->error_ctx;
			vq->error_ctx = eventfp ?
				eventfd_ctx_fileget(eventfp) : NULL;
		} else
			filep = eventfp;
		break;
895 896 897 898 899 900
	case VHOST_SET_VRING_ENDIAN:
		r = vhost_set_vring_endian(vq, argp);
		break;
	case VHOST_GET_VRING_ENDIAN:
		r = vhost_get_vring_endian(vq, idx, argp);
		break;
901 902 903 904 905 906 907 908 909 910 911 912 913
	default:
		r = -ENOIOCTLCMD;
	}

	if (pollstop && vq->handle_kick)
		vhost_poll_stop(&vq->poll);

	if (ctx)
		eventfd_ctx_put(ctx);
	if (filep)
		fput(filep);

	if (pollstart && vq->handle_kick)
914
		r = vhost_poll_start(&vq->poll, vq->kick);
915 916 917 918 919 920 921

	mutex_unlock(&vq->mutex);

	if (pollstop && vq->handle_kick)
		vhost_poll_flush(&vq->poll);
	return r;
}
922
EXPORT_SYMBOL_GPL(vhost_vring_ioctl);
923 924

/* Caller must have device mutex */
925
long vhost_dev_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *argp)
926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948
{
	struct file *eventfp, *filep = NULL;
	struct eventfd_ctx *ctx = NULL;
	u64 p;
	long r;
	int i, fd;

	/* If you are not the owner, you can become one */
	if (ioctl == VHOST_SET_OWNER) {
		r = vhost_dev_set_owner(d);
		goto done;
	}

	/* You must be the owner to do anything else */
	r = vhost_dev_check_owner(d);
	if (r)
		goto done;

	switch (ioctl) {
	case VHOST_SET_MEM_TABLE:
		r = vhost_set_memory(d, argp);
		break;
	case VHOST_SET_LOG_BASE:
949 950
		if (copy_from_user(&p, argp, sizeof p)) {
			r = -EFAULT;
951
			break;
952
		}
953 954 955 956 957 958 959
		if ((u64)(unsigned long)p != p) {
			r = -EFAULT;
			break;
		}
		for (i = 0; i < d->nvqs; ++i) {
			struct vhost_virtqueue *vq;
			void __user *base = (void __user *)(unsigned long)p;
960
			vq = d->vqs[i];
961 962
			mutex_lock(&vq->mutex);
			/* If ring is inactive, will check when it's enabled. */
963
			if (vq->private_data && !vq_log_access_ok(vq, base))
964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986
				r = -EFAULT;
			else
				vq->log_base = base;
			mutex_unlock(&vq->mutex);
		}
		break;
	case VHOST_SET_LOG_FD:
		r = get_user(fd, (int __user *)argp);
		if (r < 0)
			break;
		eventfp = fd == -1 ? NULL : eventfd_fget(fd);
		if (IS_ERR(eventfp)) {
			r = PTR_ERR(eventfp);
			break;
		}
		if (eventfp != d->log_file) {
			filep = d->log_file;
			ctx = d->log_ctx;
			d->log_ctx = eventfp ?
				eventfd_ctx_fileget(eventfp) : NULL;
		} else
			filep = eventfp;
		for (i = 0; i < d->nvqs; ++i) {
987 988 989
			mutex_lock(&d->vqs[i]->mutex);
			d->vqs[i]->log_ctx = d->log_ctx;
			mutex_unlock(&d->vqs[i]->mutex);
990 991 992 993 994 995 996
		}
		if (ctx)
			eventfd_ctx_put(ctx);
		if (filep)
			fput(filep);
		break;
	default:
997
		r = -ENOIOCTLCMD;
998 999 1000 1001 1002
		break;
	}
done:
	return r;
}
1003
EXPORT_SYMBOL_GPL(vhost_dev_ioctl);
1004 1005 1006 1007

static const struct vhost_memory_region *find_region(struct vhost_memory *mem,
						     __u64 addr, __u32 len)
{
1008 1009 1010 1011 1012 1013 1014 1015 1016 1017
	const struct vhost_memory_region *reg;
	int start = 0, end = mem->nregions;

	while (start < end) {
		int slot = start + (end - start) / 2;
		reg = mem->regions + slot;
		if (addr >= reg->guest_phys_addr)
			end = slot;
		else
			start = slot + 1;
1018
	}
1019 1020 1021 1022 1023

	reg = mem->regions + start;
	if (addr >= reg->guest_phys_addr &&
		reg->guest_phys_addr + reg->memory_size > addr)
		return reg;
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037
	return NULL;
}

/* TODO: This is really inefficient.  We need something like get_user()
 * (instruction directly accesses the data, with an exception table entry
 * returning -EFAULT). See Documentation/x86/exception-tables.txt.
 */
static int set_bit_to_user(int nr, void __user *addr)
{
	unsigned long log = (unsigned long)addr;
	struct page *page;
	void *base;
	int bit = nr + (log % PAGE_SIZE) * 8;
	int r;
1038

1039
	r = get_user_pages_fast(log, 1, 1, &page);
1040
	if (r < 0)
1041
		return r;
1042
	BUG_ON(r != 1);
1043
	base = kmap_atomic(page);
1044
	set_bit(bit, base);
1045
	kunmap_atomic(base);
1046 1047 1048 1049 1050 1051 1052 1053
	set_page_dirty_lock(page);
	put_page(page);
	return 0;
}

static int log_write(void __user *log_base,
		     u64 write_address, u64 write_length)
{
1054
	u64 write_page = write_address / VHOST_PAGE_SIZE;
1055
	int r;
1056

1057 1058
	if (!write_length)
		return 0;
1059
	write_length += write_address % VHOST_PAGE_SIZE;
1060 1061
	for (;;) {
		u64 base = (u64)(unsigned long)log_base;
1062 1063
		u64 log = base + write_page / 8;
		int bit = write_page % 8;
1064 1065 1066 1067 1068 1069 1070 1071
		if ((u64)(unsigned long)log != log)
			return -EFAULT;
		r = set_bit_to_user(bit, (void __user *)(unsigned long)log);
		if (r < 0)
			return r;
		if (write_length <= VHOST_PAGE_SIZE)
			break;
		write_length -= VHOST_PAGE_SIZE;
1072
		write_page += 1;
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
	}
	return r;
}

int vhost_log_write(struct vhost_virtqueue *vq, struct vhost_log *log,
		    unsigned int log_num, u64 len)
{
	int i, r;

	/* Make sure data written is seen before log. */
1083
	smp_wmb();
1084 1085 1086 1087 1088 1089
	for (i = 0; i < log_num; ++i) {
		u64 l = min(log[i].len, len);
		r = log_write(vq->log_base, log[i].addr, l);
		if (r < 0)
			return r;
		len -= l;
1090 1091 1092
		if (!len) {
			if (vq->log_ctx)
				eventfd_signal(vq->log_ctx, 1);
1093
			return 0;
1094
		}
1095 1096 1097 1098 1099
	}
	/* Length written exceeds what we have stored. This is a bug. */
	BUG();
	return 0;
}
1100
EXPORT_SYMBOL_GPL(vhost_log_write);
1101

1102 1103 1104
static int vhost_update_used_flags(struct vhost_virtqueue *vq)
{
	void __user *used;
1105
	if (__put_user(cpu_to_vhost16(vq, vq->used_flags), &vq->used->flags) < 0)
1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122
		return -EFAULT;
	if (unlikely(vq->log_used)) {
		/* Make sure the flag is seen before log. */
		smp_wmb();
		/* Log used flag write. */
		used = &vq->used->flags;
		log_write(vq->log_base, vq->log_addr +
			  (used - (void __user *)vq->used),
			  sizeof vq->used->flags);
		if (vq->log_ctx)
			eventfd_signal(vq->log_ctx, 1);
	}
	return 0;
}

static int vhost_update_avail_event(struct vhost_virtqueue *vq, u16 avail_event)
{
1123
	if (__put_user(cpu_to_vhost16(vq, vq->avail_idx), vhost_avail_event(vq)))
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
		return -EFAULT;
	if (unlikely(vq->log_used)) {
		void __user *used;
		/* Make sure the event is seen before log. */
		smp_wmb();
		/* Log avail event write */
		used = vhost_avail_event(vq);
		log_write(vq->log_base, vq->log_addr +
			  (used - (void __user *)vq->used),
			  sizeof *vhost_avail_event(vq));
		if (vq->log_ctx)
			eventfd_signal(vq->log_ctx, 1);
	}
	return 0;
}

int vhost_init_used(struct vhost_virtqueue *vq)
{
1142
	__virtio16 last_used_idx;
1143
	int r;
1144 1145
	if (!vq->private_data) {
		vq->is_le = virtio_legacy_is_little_endian();
1146
		return 0;
1147 1148 1149
	}

	vhost_init_is_le(vq);
1150 1151 1152 1153 1154

	r = vhost_update_used_flags(vq);
	if (r)
		return r;
	vq->signalled_used_valid = false;
1155 1156 1157 1158 1159
	if (!access_ok(VERIFY_READ, &vq->used->idx, sizeof vq->used->idx))
		return -EFAULT;
	r = __get_user(last_used_idx, &vq->used->idx);
	if (r)
		return r;
1160
	vq->last_used_idx = vhost16_to_cpu(vq, last_used_idx);
1161
	return 0;
1162
}
1163
EXPORT_SYMBOL_GPL(vhost_init_used);
1164

1165
static int translate_desc(struct vhost_virtqueue *vq, u64 addr, u32 len,
1166
			  struct iovec iov[], int iov_size)
1167 1168 1169 1170 1171 1172 1173
{
	const struct vhost_memory_region *reg;
	struct vhost_memory *mem;
	struct iovec *_iov;
	u64 s = 0;
	int ret = 0;

1174
	mem = vq->memory;
1175 1176
	while ((u64)len > s) {
		u64 size;
1177
		if (unlikely(ret >= iov_size)) {
1178 1179 1180 1181
			ret = -ENOBUFS;
			break;
		}
		reg = find_region(mem, addr, len);
1182
		if (unlikely(!reg)) {
1183 1184 1185 1186 1187
			ret = -EFAULT;
			break;
		}
		_iov = iov + ret;
		size = reg->memory_size - addr + reg->guest_phys_addr;
1188
		_iov->iov_len = min((u64)len - s, size);
1189
		_iov->iov_base = (void __user *)(unsigned long)
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
			(reg->userspace_addr + addr - reg->guest_phys_addr);
		s += size;
		addr += size;
		++ret;
	}

	return ret;
}

/* Each buffer in the virtqueues is actually a chain of descriptors.  This
 * function returns the next descriptor in the chain,
 * or -1U if we're at the end. */
1202
static unsigned next_desc(struct vhost_virtqueue *vq, struct vring_desc *desc)
1203 1204 1205 1206
{
	unsigned int next;

	/* If this descriptor says it doesn't chain, we're done. */
1207
	if (!(desc->flags & cpu_to_vhost16(vq, VRING_DESC_F_NEXT)))
1208 1209 1210
		return -1U;

	/* Check they're not leading us off end of descriptors. */
1211
	next = vhost16_to_cpu(vq, desc->next);
1212 1213 1214 1215 1216 1217 1218 1219
	/* Make sure compiler knows to grab that: we don't want it changing! */
	/* We will use the result as an index in an array, so most
	 * architectures only need a compiler barrier here. */
	read_barrier_depends();

	return next;
}

1220
static int get_indirect(struct vhost_virtqueue *vq,
1221 1222 1223 1224
			struct iovec iov[], unsigned int iov_size,
			unsigned int *out_num, unsigned int *in_num,
			struct vhost_log *log, unsigned int *log_num,
			struct vring_desc *indirect)
1225 1226 1227
{
	struct vring_desc desc;
	unsigned int i = 0, count, found = 0;