dev.c 23 KB
Newer Older
1
2
/*
  FUSE: Filesystem in Userspace
3
  Copyright (C) 2001-2006  Miklos Szeredi <miklos@szeredi.hu>
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

  This program can be distributed under the terms of the GNU GPL.
  See the file COPYING.
*/

#include "fuse_i.h"

#include <linux/init.h>
#include <linux/module.h>
#include <linux/poll.h>
#include <linux/uio.h>
#include <linux/miscdevice.h>
#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/slab.h>

MODULE_ALIAS_MISCDEV(FUSE_MINOR);

static kmem_cache_t *fuse_req_cachep;

24
static struct fuse_conn *fuse_get_conn(struct file *file)
25
{
26
27
28
29
30
	/*
	 * Lockless access is OK, because file->private data is set
	 * once during mount and is valid until the file is released.
	 */
	return file->private_data;
31
32
}

33
static void fuse_request_init(struct fuse_req *req)
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
{
	memset(req, 0, sizeof(*req));
	INIT_LIST_HEAD(&req->list);
	init_waitqueue_head(&req->waitq);
	atomic_set(&req->count, 1);
}

struct fuse_req *fuse_request_alloc(void)
{
	struct fuse_req *req = kmem_cache_alloc(fuse_req_cachep, SLAB_KERNEL);
	if (req)
		fuse_request_init(req);
	return req;
}

void fuse_request_free(struct fuse_req *req)
{
	kmem_cache_free(fuse_req_cachep, req);
}

54
static void block_sigs(sigset_t *oldset)
55
56
57
58
59
60
61
{
	sigset_t mask;

	siginitsetinv(&mask, sigmask(SIGKILL));
	sigprocmask(SIG_BLOCK, &mask, oldset);
}

62
static void restore_sigs(sigset_t *oldset)
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
{
	sigprocmask(SIG_SETMASK, oldset, NULL);
}

static void __fuse_get_request(struct fuse_req *req)
{
	atomic_inc(&req->count);
}

/* Must be called with > 1 refcount */
static void __fuse_put_request(struct fuse_req *req)
{
	BUG_ON(atomic_read(&req->count) < 2);
	atomic_dec(&req->count);
}

79
80
81
82
83
84
85
static void fuse_req_init_context(struct fuse_req *req)
{
	req->in.h.uid = current->fsuid;
	req->in.h.gid = current->fsgid;
	req->in.h.pid = current->pid;
}

86
struct fuse_req *fuse_get_req(struct fuse_conn *fc)
87
{
88
89
	struct fuse_req *req;
	sigset_t oldset;
90
	int intr;
91
92
	int err;

93
	atomic_inc(&fc->num_waiting);
94
	block_sigs(&oldset);
95
	intr = wait_event_interruptible(fc->blocked_waitq, !fc->blocked);
96
	restore_sigs(&oldset);
97
98
99
	err = -EINTR;
	if (intr)
		goto out;
100

101
102
103
104
	err = -ENOTCONN;
	if (!fc->connected)
		goto out;

105
	req = fuse_request_alloc();
106
	err = -ENOMEM;
107
	if (!req)
108
		goto out;
109

110
	fuse_req_init_context(req);
111
	req->waiting = 1;
112
	return req;
113
114
115
116

 out:
	atomic_dec(&fc->num_waiting);
	return ERR_PTR(err);
117
118
}

119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
/*
 * Return request in fuse_file->reserved_req.  However that may
 * currently be in use.  If that is the case, wait for it to become
 * available.
 */
static struct fuse_req *get_reserved_req(struct fuse_conn *fc,
					 struct file *file)
{
	struct fuse_req *req = NULL;
	struct fuse_file *ff = file->private_data;

	do {
		wait_event(fc->blocked_waitq, ff->reserved_req);
		spin_lock(&fc->lock);
		if (ff->reserved_req) {
			req = ff->reserved_req;
			ff->reserved_req = NULL;
			get_file(file);
			req->stolen_file = file;
		}
		spin_unlock(&fc->lock);
	} while (!req);

	return req;
}

/*
 * Put stolen request back into fuse_file->reserved_req
 */
static void put_reserved_req(struct fuse_conn *fc, struct fuse_req *req)
{
	struct file *file = req->stolen_file;
	struct fuse_file *ff = file->private_data;

	spin_lock(&fc->lock);
	fuse_request_init(req);
	BUG_ON(ff->reserved_req);
	ff->reserved_req = req;
	wake_up(&fc->blocked_waitq);
	spin_unlock(&fc->lock);
	fput(file);
}

/*
 * Gets a requests for a file operation, always succeeds
 *
 * This is used for sending the FLUSH request, which must get to
 * userspace, due to POSIX locks which may need to be unlocked.
 *
 * If allocation fails due to OOM, use the reserved request in
 * fuse_file.
 *
 * This is very unlikely to deadlock accidentally, since the
 * filesystem should not have it's own file open.  If deadlock is
 * intentional, it can still be broken by "aborting" the filesystem.
 */
struct fuse_req *fuse_get_req_nofail(struct fuse_conn *fc, struct file *file)
{
	struct fuse_req *req;

	atomic_inc(&fc->num_waiting);
	wait_event(fc->blocked_waitq, !fc->blocked);
	req = fuse_request_alloc();
	if (!req)
		req = get_reserved_req(fc, file);

	fuse_req_init_context(req);
	req->waiting = 1;
	return req;
}

190
void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req)
191
192
{
	if (atomic_dec_and_test(&req->count)) {
193
194
		if (req->waiting)
			atomic_dec(&fc->num_waiting);
195
196
197
198
199

		if (req->stolen_file)
			put_reserved_req(fc, req);
		else
			fuse_request_free(req);
200
201
202
	}
}

203
204
205
/*
 * This function is called when a request is finished.  Either a reply
 * has arrived or it was interrupted (and not yet sent) or some error
206
 * occurred during communication with userspace, or the device file
207
208
209
 * was closed.  The requester thread is woken up (if still waiting),
 * the 'end' callback is called if given, else the reference to the
 * request is released
210
 *
211
 * Called with fc->lock, unlocks it
212
213
214
 */
static void request_end(struct fuse_conn *fc, struct fuse_req *req)
{
215
216
	void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;
	req->end = NULL;
217
	list_del(&req->list);
218
	req->state = FUSE_REQ_FINISHED;
219
220
221
222
223
224
	if (req->background) {
		if (fc->num_background == FUSE_MAX_BACKGROUND) {
			fc->blocked = 0;
			wake_up_all(&fc->blocked_waitq);
		}
		fc->num_background--;
225
	}
226
227
228
	spin_unlock(&fc->lock);
	dput(req->dentry);
	mntput(req->vfsmount);
229
	if (req->file)
230
231
232
233
234
235
		fput(req->file);
	wake_up(&req->waitq);
	if (end)
		end(fc, req);
	else
		fuse_put_request(fc, req);
236
237
}

238
/* Called with fc->lock held.  Releases, and then reacquires it. */
239
static void request_wait_answer(struct fuse_conn *fc, struct fuse_req *req)
240
{
241
	sigset_t oldset;
242

243
	spin_unlock(&fc->lock);
244
245
246
247
248
249
250
251
	if (req->force)
		wait_event(req->waitq, req->state == FUSE_REQ_FINISHED);
	else {
		block_sigs(&oldset);
		wait_event_interruptible(req->waitq,
					 req->state == FUSE_REQ_FINISHED);
		restore_sigs(&oldset);
	}
252
	spin_lock(&fc->lock);
253
	if (req->state == FUSE_REQ_FINISHED && !req->interrupted)
254
255
		return;

256
257
258
259
	if (!req->interrupted) {
		req->out.h.error = -EINTR;
		req->interrupted = 1;
	}
260
261
262
263
264
265
	if (req->locked) {
		/* This is uninterruptible sleep, because data is
		   being copied to/from the buffers of req.  During
		   locked state, there mustn't be any filesystem
		   operation (e.g. page fault), since that could lead
		   to deadlock */
266
		spin_unlock(&fc->lock);
267
		wait_event(req->waitq, !req->locked);
268
		spin_lock(&fc->lock);
269
	}
270
	if (req->state == FUSE_REQ_PENDING) {
271
272
		list_del(&req->list);
		__fuse_put_request(req);
273
274
275
276
277
	} else if (req->state == FUSE_REQ_SENT) {
		spin_unlock(&fc->lock);
		wait_event(req->waitq, req->state == FUSE_REQ_FINISHED);
		spin_lock(&fc->lock);
	}
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
}

static unsigned len_args(unsigned numargs, struct fuse_arg *args)
{
	unsigned nbytes = 0;
	unsigned i;

	for (i = 0; i < numargs; i++)
		nbytes += args[i].size;

	return nbytes;
}

static void queue_request(struct fuse_conn *fc, struct fuse_req *req)
{
	fc->reqctr++;
	/* zero is special */
	if (fc->reqctr == 0)
		fc->reqctr = 1;
	req->in.h.unique = fc->reqctr;
	req->in.h.len = sizeof(struct fuse_in_header) +
		len_args(req->in.numargs, (struct fuse_arg *) req->in.args);
	list_add_tail(&req->list, &fc->pending);
301
	req->state = FUSE_REQ_PENDING;
302
303
304
305
	if (!req->waiting) {
		req->waiting = 1;
		atomic_inc(&fc->num_waiting);
	}
306
	wake_up(&fc->waitq);
307
	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
308
309
}

310
311
312
313
/*
 * This can only be interrupted by a SIGKILL
 */
void request_send(struct fuse_conn *fc, struct fuse_req *req)
314
315
{
	req->isreply = 1;
316
	spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
317
	if (!fc->connected)
318
319
320
321
322
323
324
325
326
		req->out.h.error = -ENOTCONN;
	else if (fc->conn_error)
		req->out.h.error = -ECONNREFUSED;
	else {
		queue_request(fc, req);
		/* acquire extra reference, since request is still needed
		   after request_end() */
		__fuse_get_request(req);

327
		request_wait_answer(fc, req);
328
	}
329
	spin_unlock(&fc->lock);
330
331
332
333
}

static void request_send_nowait(struct fuse_conn *fc, struct fuse_req *req)
{
334
	spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
335
	if (fc->connected) {
336
337
338
339
340
		req->background = 1;
		fc->num_background++;
		if (fc->num_background == FUSE_MAX_BACKGROUND)
			fc->blocked = 1;

341
		queue_request(fc, req);
342
		spin_unlock(&fc->lock);
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
	} else {
		req->out.h.error = -ENOTCONN;
		request_end(fc, req);
	}
}

void request_send_noreply(struct fuse_conn *fc, struct fuse_req *req)
{
	req->isreply = 0;
	request_send_nowait(fc, req);
}

void request_send_background(struct fuse_conn *fc, struct fuse_req *req)
{
	req->isreply = 1;
	request_send_nowait(fc, req);
}

/*
 * Lock the request.  Up to the next unlock_request() there mustn't be
 * anything that could cause a page-fault.  If the request was already
 * interrupted bail out.
 */
366
static int lock_request(struct fuse_conn *fc, struct fuse_req *req)
367
368
369
{
	int err = 0;
	if (req) {
370
		spin_lock(&fc->lock);
371
372
373
374
		if (req->interrupted)
			err = -ENOENT;
		else
			req->locked = 1;
375
		spin_unlock(&fc->lock);
376
377
378
379
380
381
382
383
384
	}
	return err;
}

/*
 * Unlock request.  If it was interrupted during being locked, the
 * requester thread is currently waiting for it to be unlocked, so
 * wake it up.
 */
385
static void unlock_request(struct fuse_conn *fc, struct fuse_req *req)
386
387
{
	if (req) {
388
		spin_lock(&fc->lock);
389
390
391
		req->locked = 0;
		if (req->interrupted)
			wake_up(&req->waitq);
392
		spin_unlock(&fc->lock);
393
394
395
396
	}
}

struct fuse_copy_state {
397
	struct fuse_conn *fc;
398
399
400
401
402
403
404
405
406
407
408
409
	int write;
	struct fuse_req *req;
	const struct iovec *iov;
	unsigned long nr_segs;
	unsigned long seglen;
	unsigned long addr;
	struct page *pg;
	void *mapaddr;
	void *buf;
	unsigned len;
};

410
411
412
static void fuse_copy_init(struct fuse_copy_state *cs, struct fuse_conn *fc,
			   int write, struct fuse_req *req,
			   const struct iovec *iov, unsigned long nr_segs)
413
414
{
	memset(cs, 0, sizeof(*cs));
415
	cs->fc = fc;
416
417
418
419
420
421
422
	cs->write = write;
	cs->req = req;
	cs->iov = iov;
	cs->nr_segs = nr_segs;
}

/* Unmap and put previous page of userspace buffer */
423
static void fuse_copy_finish(struct fuse_copy_state *cs)
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
{
	if (cs->mapaddr) {
		kunmap_atomic(cs->mapaddr, KM_USER0);
		if (cs->write) {
			flush_dcache_page(cs->pg);
			set_page_dirty_lock(cs->pg);
		}
		put_page(cs->pg);
		cs->mapaddr = NULL;
	}
}

/*
 * Get another pagefull of userspace buffer, and map it to kernel
 * address space, and lock request
 */
static int fuse_copy_fill(struct fuse_copy_state *cs)
{
	unsigned long offset;
	int err;

445
	unlock_request(cs->fc, cs->req);
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
	fuse_copy_finish(cs);
	if (!cs->seglen) {
		BUG_ON(!cs->nr_segs);
		cs->seglen = cs->iov[0].iov_len;
		cs->addr = (unsigned long) cs->iov[0].iov_base;
		cs->iov ++;
		cs->nr_segs --;
	}
	down_read(&current->mm->mmap_sem);
	err = get_user_pages(current, current->mm, cs->addr, 1, cs->write, 0,
			     &cs->pg, NULL);
	up_read(&current->mm->mmap_sem);
	if (err < 0)
		return err;
	BUG_ON(err != 1);
	offset = cs->addr % PAGE_SIZE;
	cs->mapaddr = kmap_atomic(cs->pg, KM_USER0);
	cs->buf = cs->mapaddr + offset;
	cs->len = min(PAGE_SIZE - offset, cs->seglen);
	cs->seglen -= cs->len;
	cs->addr += cs->len;

468
	return lock_request(cs->fc, cs->req);
469
470
471
}

/* Do as much copy to/from userspace buffer as we can */
472
static int fuse_copy_do(struct fuse_copy_state *cs, void **val, unsigned *size)
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
{
	unsigned ncpy = min(*size, cs->len);
	if (val) {
		if (cs->write)
			memcpy(cs->buf, *val, ncpy);
		else
			memcpy(*val, cs->buf, ncpy);
		*val += ncpy;
	}
	*size -= ncpy;
	cs->len -= ncpy;
	cs->buf += ncpy;
	return ncpy;
}

/*
 * Copy a page in the request to/from the userspace buffer.  Must be
 * done atomically
 */
492
493
static int fuse_copy_page(struct fuse_copy_state *cs, struct page *page,
			  unsigned offset, unsigned count, int zeroing)
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
{
	if (page && zeroing && count < PAGE_SIZE) {
		void *mapaddr = kmap_atomic(page, KM_USER1);
		memset(mapaddr, 0, PAGE_SIZE);
		kunmap_atomic(mapaddr, KM_USER1);
	}
	while (count) {
		int err;
		if (!cs->len && (err = fuse_copy_fill(cs)))
			return err;
		if (page) {
			void *mapaddr = kmap_atomic(page, KM_USER1);
			void *buf = mapaddr + offset;
			offset += fuse_copy_do(cs, &buf, &count);
			kunmap_atomic(mapaddr, KM_USER1);
		} else
			offset += fuse_copy_do(cs, NULL, &count);
	}
	if (page && !cs->write)
		flush_dcache_page(page);
	return 0;
}

/* Copy pages in the request to/from userspace buffer */
static int fuse_copy_pages(struct fuse_copy_state *cs, unsigned nbytes,
			   int zeroing)
{
	unsigned i;
	struct fuse_req *req = cs->req;
	unsigned offset = req->page_offset;
	unsigned count = min(nbytes, (unsigned) PAGE_SIZE - offset);

	for (i = 0; i < req->num_pages && (nbytes || zeroing); i++) {
		struct page *page = req->pages[i];
		int err = fuse_copy_page(cs, page, offset, count, zeroing);
		if (err)
			return err;

		nbytes -= count;
		count = min(nbytes, (unsigned) PAGE_SIZE);
		offset = 0;
	}
	return 0;
}

/* Copy a single argument in the request to/from userspace buffer */
static int fuse_copy_one(struct fuse_copy_state *cs, void *val, unsigned size)
{
	while (size) {
		int err;
		if (!cs->len && (err = fuse_copy_fill(cs)))
			return err;
		fuse_copy_do(cs, &val, &size);
	}
	return 0;
}

/* Copy request arguments to/from userspace buffer */
static int fuse_copy_args(struct fuse_copy_state *cs, unsigned numargs,
			  unsigned argpages, struct fuse_arg *args,
			  int zeroing)
{
	int err = 0;
	unsigned i;

	for (i = 0; !err && i < numargs; i++)  {
		struct fuse_arg *arg = &args[i];
		if (i == numargs - 1 && argpages)
			err = fuse_copy_pages(cs, arg->size, zeroing);
		else
			err = fuse_copy_one(cs, arg->value, arg->size);
	}
	return err;
}

/* Wait until a request is available on the pending list */
static void request_wait(struct fuse_conn *fc)
{
	DECLARE_WAITQUEUE(wait, current);

	add_wait_queue_exclusive(&fc->waitq, &wait);
575
	while (fc->connected && list_empty(&fc->pending)) {
576
577
578
579
		set_current_state(TASK_INTERRUPTIBLE);
		if (signal_pending(current))
			break;

580
		spin_unlock(&fc->lock);
581
		schedule();
582
		spin_lock(&fc->lock);
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
	}
	set_current_state(TASK_RUNNING);
	remove_wait_queue(&fc->waitq, &wait);
}

/*
 * Read a single request into the userspace filesystem's buffer.  This
 * function waits until a request is available, then removes it from
 * the pending list and copies request data to userspace buffer.  If
 * no reply is needed (FORGET) or request has been interrupted or
 * there was an error during the copying then it's finished by calling
 * request_end().  Otherwise add it to the processing list, and set
 * the 'sent' flag.
 */
static ssize_t fuse_dev_readv(struct file *file, const struct iovec *iov,
			      unsigned long nr_segs, loff_t *off)
{
	int err;
	struct fuse_req *req;
	struct fuse_in *in;
	struct fuse_copy_state cs;
	unsigned reqsize;
605
606
607
	struct fuse_conn *fc = fuse_get_conn(file);
	if (!fc)
		return -EPERM;
608

609
 restart:
610
	spin_lock(&fc->lock);
611
612
613
614
615
	err = -EAGAIN;
	if ((file->f_flags & O_NONBLOCK) && fc->connected &&
	    list_empty(&fc->pending))
		goto err_unlock;

616
617
	request_wait(fc);
	err = -ENODEV;
618
	if (!fc->connected)
619
620
621
622
623
624
		goto err_unlock;
	err = -ERESTARTSYS;
	if (list_empty(&fc->pending))
		goto err_unlock;

	req = list_entry(fc->pending.next, struct fuse_req, list);
625
	req->state = FUSE_REQ_READING;
626
	list_move(&req->list, &fc->io);
627
628

	in = &req->in;
629
630
631
632
633
634
635
636
637
	reqsize = in->h.len;
	/* If request is too large, reply with an error and restart the read */
	if (iov_length(iov, nr_segs) < reqsize) {
		req->out.h.error = -EIO;
		/* SETXATTR is special, since it may contain too large data */
		if (in->h.opcode == FUSE_SETXATTR)
			req->out.h.error = -E2BIG;
		request_end(fc, req);
		goto restart;
638
	}
639
640
	spin_unlock(&fc->lock);
	fuse_copy_init(&cs, fc, 1, req, iov, nr_segs);
641
642
643
644
	err = fuse_copy_one(&cs, &in->h, sizeof(in->h));
	if (!err)
		err = fuse_copy_args(&cs, in->numargs, in->argpages,
				     (struct fuse_arg *) in->args, 0);
645
	fuse_copy_finish(&cs);
646
	spin_lock(&fc->lock);
647
648
649
650
651
652
653
654
655
656
657
658
	req->locked = 0;
	if (!err && req->interrupted)
		err = -ENOENT;
	if (err) {
		if (!req->interrupted)
			req->out.h.error = -EIO;
		request_end(fc, req);
		return err;
	}
	if (!req->isreply)
		request_end(fc, req);
	else {
659
		req->state = FUSE_REQ_SENT;
660
		list_move_tail(&req->list, &fc->processing);
661
		spin_unlock(&fc->lock);
662
663
664
665
	}
	return reqsize;

 err_unlock:
666
	spin_unlock(&fc->lock);
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
	return err;
}

static ssize_t fuse_dev_read(struct file *file, char __user *buf,
			     size_t nbytes, loff_t *off)
{
	struct iovec iov;
	iov.iov_len = nbytes;
	iov.iov_base = buf;
	return fuse_dev_readv(file, &iov, 1, off);
}

/* Look up request on processing list by unique ID */
static struct fuse_req *request_find(struct fuse_conn *fc, u64 unique)
{
	struct list_head *entry;

	list_for_each(entry, &fc->processing) {
		struct fuse_req *req;
		req = list_entry(entry, struct fuse_req, list);
		if (req->in.h.unique == unique)
			return req;
	}
	return NULL;
}

static int copy_out_args(struct fuse_copy_state *cs, struct fuse_out *out,
			 unsigned nbytes)
{
	unsigned reqsize = sizeof(struct fuse_out_header);

	if (out->h.error)
		return nbytes != reqsize ? -EINVAL : 0;

	reqsize += len_args(out->numargs, out->args);

	if (reqsize < nbytes || (reqsize > nbytes && !out->argvar))
		return -EINVAL;
	else if (reqsize > nbytes) {
		struct fuse_arg *lastarg = &out->args[out->numargs-1];
		unsigned diffsize = reqsize - nbytes;
		if (diffsize > lastarg->size)
			return -EINVAL;
		lastarg->size -= diffsize;
	}
	return fuse_copy_args(cs, out->numargs, out->argpages, out->args,
			      out->page_zeroing);
}

/*
 * Write a single reply to a request.  First the header is copied from
 * the write buffer.  The request is then searched on the processing
 * list by the unique ID found in the header.  If found, then remove
 * it from the list and copy the rest of the buffer to the request.
 * The request is finished by calling request_end()
 */
static ssize_t fuse_dev_writev(struct file *file, const struct iovec *iov,
			       unsigned long nr_segs, loff_t *off)
{
	int err;
	unsigned nbytes = iov_length(iov, nr_segs);
	struct fuse_req *req;
	struct fuse_out_header oh;
	struct fuse_copy_state cs;
	struct fuse_conn *fc = fuse_get_conn(file);
	if (!fc)
733
		return -EPERM;
734

735
	fuse_copy_init(&cs, fc, 0, NULL, iov, nr_segs);
736
737
738
739
740
741
742
743
744
745
746
	if (nbytes < sizeof(struct fuse_out_header))
		return -EINVAL;

	err = fuse_copy_one(&cs, &oh, sizeof(oh));
	if (err)
		goto err_finish;
	err = -EINVAL;
	if (!oh.unique || oh.error <= -1000 || oh.error > 0 ||
	    oh.len != nbytes)
		goto err_finish;

747
	spin_lock(&fc->lock);
748
749
750
751
	err = -ENOENT;
	if (!fc->connected)
		goto err_unlock;

752
753
754
755
756
757
	req = request_find(fc, oh.unique);
	err = -EINVAL;
	if (!req)
		goto err_unlock;

	if (req->interrupted) {
758
		spin_unlock(&fc->lock);
759
		fuse_copy_finish(&cs);
760
		spin_lock(&fc->lock);
761
		request_end(fc, req);
762
763
		return -ENOENT;
	}
764
	list_move(&req->list, &fc->io);
765
766
767
	req->out.h = oh;
	req->locked = 1;
	cs.req = req;
768
	spin_unlock(&fc->lock);
769
770
771
772

	err = copy_out_args(&cs, &req->out, nbytes);
	fuse_copy_finish(&cs);

773
	spin_lock(&fc->lock);
774
775
776
777
778
779
780
781
782
783
784
	req->locked = 0;
	if (!err) {
		if (req->interrupted)
			err = -ENOENT;
	} else if (!req->interrupted)
		req->out.h.error = -EIO;
	request_end(fc, req);

	return err ? err : nbytes;

 err_unlock:
785
	spin_unlock(&fc->lock);
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
 err_finish:
	fuse_copy_finish(&cs);
	return err;
}

static ssize_t fuse_dev_write(struct file *file, const char __user *buf,
			      size_t nbytes, loff_t *off)
{
	struct iovec iov;
	iov.iov_len = nbytes;
	iov.iov_base = (char __user *) buf;
	return fuse_dev_writev(file, &iov, 1, off);
}

static unsigned fuse_dev_poll(struct file *file, poll_table *wait)
{
	unsigned mask = POLLOUT | POLLWRNORM;
803
	struct fuse_conn *fc = fuse_get_conn(file);
804
	if (!fc)
805
		return POLLERR;
806
807
808

	poll_wait(file, &fc->waitq, wait);

809
	spin_lock(&fc->lock);
810
811
812
813
	if (!fc->connected)
		mask = POLLERR;
	else if (!list_empty(&fc->pending))
		mask |= POLLIN | POLLRDNORM;
814
	spin_unlock(&fc->lock);
815
816
817
818

	return mask;
}

819
820
821
/*
 * Abort all requests on the given list (pending or processing)
 *
822
 * This function releases and reacquires fc->lock
823
 */
824
825
826
827
828
829
830
static void end_requests(struct fuse_conn *fc, struct list_head *head)
{
	while (!list_empty(head)) {
		struct fuse_req *req;
		req = list_entry(head->next, struct fuse_req, list);
		req->out.h.error = -ECONNABORTED;
		request_end(fc, req);
831
		spin_lock(&fc->lock);
832
833
834
	}
}

835
836
837
838
839
840
/*
 * Abort requests under I/O
 *
 * The requests are set to interrupted and finished, and the request
 * waiter is woken up.  This will make request_wait_answer() wait
 * until the request is unlocked and then return.
841
842
843
844
 *
 * If the request is asynchronous, then the end function needs to be
 * called after waiting for the request to be unlocked (if it was
 * locked).
845
846
847
848
 */
static void end_io_requests(struct fuse_conn *fc)
{
	while (!list_empty(&fc->io)) {
849
850
851
852
		struct fuse_req *req =
			list_entry(fc->io.next, struct fuse_req, list);
		void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;

853
854
855
856
857
		req->interrupted = 1;
		req->out.h.error = -ECONNABORTED;
		req->state = FUSE_REQ_FINISHED;
		list_del_init(&req->list);
		wake_up(&req->waitq);
858
859
860
861
		if (end) {
			req->end = NULL;
			/* The end function will consume this reference */
			__fuse_get_request(req);
862
			spin_unlock(&fc->lock);
863
864
			wait_event(req->waitq, !req->locked);
			end(fc, req);
865
			spin_lock(&fc->lock);
866
		}
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
	}
}

/*
 * Abort all requests.
 *
 * Emergency exit in case of a malicious or accidental deadlock, or
 * just a hung filesystem.
 *
 * The same effect is usually achievable through killing the
 * filesystem daemon and all users of the filesystem.  The exception
 * is the combination of an asynchronous request and the tricky
 * deadlock (see Documentation/filesystems/fuse.txt).
 *
 * During the aborting, progression of requests from the pending and
 * processing lists onto the io list, and progression of new requests
 * onto the pending list is prevented by req->connected being false.
 *
 * Progression of requests under I/O to the processing list is
 * prevented by the req->interrupted flag being true for these
 * requests.  For this reason requests on the io list must be aborted
 * first.
 */
void fuse_abort_conn(struct fuse_conn *fc)
{
892
	spin_lock(&fc->lock);
893
894
	if (fc->connected) {
		fc->connected = 0;
895
		fc->blocked = 0;
896
897
898
899
		end_io_requests(fc);
		end_requests(fc, &fc->pending);
		end_requests(fc, &fc->processing);
		wake_up_all(&fc->waitq);
900
		wake_up_all(&fc->blocked_waitq);
901
		kill_fasync(&fc->fasync, SIGIO, POLL_IN);
902
	}
903
	spin_unlock(&fc->lock);
904
905
}

906
907
static int fuse_dev_release(struct inode *inode, struct file *file)
{
908
	struct fuse_conn *fc = fuse_get_conn(file);
909
	if (fc) {
910
		spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
911
		fc->connected = 0;
912
913
		end_requests(fc, &fc->pending);
		end_requests(fc, &fc->processing);
914
		spin_unlock(&fc->lock);
915
		fasync_helper(-1, file, 0, &fc->fasync);
916
		fuse_conn_put(fc);
917
	}
918

919
920
921
	return 0;
}

922
923
924
925
static int fuse_dev_fasync(int fd, struct file *file, int on)
{
	struct fuse_conn *fc = fuse_get_conn(file);
	if (!fc)
926
		return -EPERM;
927
928
929
930
931

	/* No locking - fasync_helper does its own locking */
	return fasync_helper(fd, file, on, &fc->fasync);
}

932
const struct file_operations fuse_dev_operations = {
933
934
935
936
937
938
939
940
	.owner		= THIS_MODULE,
	.llseek		= no_llseek,
	.read		= fuse_dev_read,
	.readv		= fuse_dev_readv,
	.write		= fuse_dev_write,
	.writev		= fuse_dev_writev,
	.poll		= fuse_dev_poll,
	.release	= fuse_dev_release,
941
	.fasync		= fuse_dev_fasync,
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
};

static struct miscdevice fuse_miscdevice = {
	.minor = FUSE_MINOR,
	.name  = "fuse",
	.fops = &fuse_dev_operations,
};

int __init fuse_dev_init(void)
{
	int err = -ENOMEM;
	fuse_req_cachep = kmem_cache_create("fuse_request",
					    sizeof(struct fuse_req),
					    0, 0, NULL, NULL);
	if (!fuse_req_cachep)
		goto out;

	err = misc_register(&fuse_miscdevice);
	if (err)
		goto out_cache_clean;

	return 0;

 out_cache_clean:
	kmem_cache_destroy(fuse_req_cachep);
 out:
	return err;
}

void fuse_dev_cleanup(void)
{
	misc_deregister(&fuse_miscdevice);
	kmem_cache_destroy(fuse_req_cachep);
}