dev.c 23.7 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
{
	sigprocmask(SIG_SETMASK, oldset, NULL);
}

67
68
69
70
71
72
/*
 * Reset request, so that it can be reused
 *
 * The caller must be _very_ careful to make sure, that it is holding
 * the only reference to req
 */
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
void fuse_reset_request(struct fuse_req *req)
{
	BUG_ON(atomic_read(&req->count) != 1);
	fuse_request_init(req);
}

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);
}

91
struct fuse_req *fuse_get_req(struct fuse_conn *fc)
92
{
93
94
	struct fuse_req *req;
	sigset_t oldset;
95
	int intr;
96
97
	int err;

98
	atomic_inc(&fc->num_waiting);
99
	block_sigs(&oldset);
100
	intr = wait_event_interruptible(fc->blocked_waitq, !fc->blocked);
101
	restore_sigs(&oldset);
102
103
104
	err = -EINTR;
	if (intr)
		goto out;
105
106

	req = fuse_request_alloc();
107
	err = -ENOMEM;
108
	if (!req)
109
		goto out;
110
111
112
113
114

	fuse_request_init(req);
	req->in.h.uid = current->fsuid;
	req->in.h.gid = current->fsgid;
	req->in.h.pid = current->pid;
115
	req->waiting = 1;
116
	return req;
117
118
119
120

 out:
	atomic_dec(&fc->num_waiting);
	return ERR_PTR(err);
121
122
123
}

void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req)
124
125
{
	if (atomic_dec_and_test(&req->count)) {
126
127
		if (req->waiting)
			atomic_dec(&fc->num_waiting);
128
		fuse_request_free(req);
129
130
131
	}
}

132
void fuse_remove_background(struct fuse_conn *fc, struct fuse_req *req)
Miklos Szeredi's avatar
Miklos Szeredi committed
133
{
134
	list_del_init(&req->bg_entry);
135
136
137
138
139
	if (fc->num_background == FUSE_MAX_BACKGROUND) {
		fc->blocked = 0;
		wake_up_all(&fc->blocked_waitq);
	}
	fc->num_background--;
Miklos Szeredi's avatar
Miklos Szeredi committed
140
141
}

142
143
144
/*
 * 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
145
146
147
 * occurred during communication with userspace, or the device file
 * was closed.  In case of a background request the reference to the
 * stored objects are released.  The requester thread is woken up (if
148
149
 * still waiting), the 'end' callback is called if given, else the
 * reference to the request is released
150
 *
151
152
153
154
155
156
157
 * Releasing extra reference for foreground requests must be done
 * within the same locked region as setting state to finished.  This
 * is because fuse_reset_request() may be called after request is
 * finished and it must be the sole possessor.  If request is
 * interrupted and put in the background, it will return with an error
 * and hence never be reset and reused.
 *
158
 * Called with fc->lock, unlocks it
159
160
161
 */
static void request_end(struct fuse_conn *fc, struct fuse_req *req)
{
162
	list_del(&req->list);
163
	req->state = FUSE_REQ_FINISHED;
164
	if (!req->background) {
165
		spin_unlock(&fc->lock);
166
167
		wake_up(&req->waitq);
		fuse_put_request(fc, req);
168
	} else {
169
170
171
		struct inode *inode = req->inode;
		struct inode *inode2 = req->inode2;
		struct file *file = req->file;
172
173
		void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;
		req->end = NULL;
174
175
176
177
178
		req->inode = NULL;
		req->inode2 = NULL;
		req->file = NULL;
		if (!list_empty(&req->bg_entry))
			fuse_remove_background(fc, req);
179
		spin_unlock(&fc->lock);
180

181
182
183
184
		if (end)
			end(fc, req);
		else
			fuse_put_request(fc, req);
185
186
187
188
189

		if (file)
			fput(file);
		iput(inode);
		iput(inode2);
190
191
192
	}
}

Miklos Szeredi's avatar
Miklos Szeredi committed
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
/*
 * Unfortunately request interruption not just solves the deadlock
 * problem, it causes problems too.  These stem from the fact, that an
 * interrupted request is continued to be processed in userspace,
 * while all the locks and object references (inode and file) held
 * during the operation are released.
 *
 * To release the locks is exactly why there's a need to interrupt the
 * request, so there's not a lot that can be done about this, except
 * introduce additional locking in userspace.
 *
 * More important is to keep inode and file references until userspace
 * has replied, otherwise FORGET and RELEASE could be sent while the
 * inode/file is still used by the filesystem.
 *
 * For this reason the concept of "background" request is introduced.
 * An interrupted request is backgrounded if it has been already sent
 * to userspace.  Backgrounding involves getting an extra reference to
 * inode(s) or file used in the request, and adding the request to
 * fc->background list.  When a reply is received for a background
 * request, the object references are released, and the request is
 * removed from the list.  If the filesystem is unmounted while there
 * are still background requests, the list is walked and references
 * are released as if a reply was received.
 *
 * There's one more use for a background request.  The RELEASE message is
 * always sent as background, since it doesn't return an error or
 * data.
 */
static void background_request(struct fuse_conn *fc, struct fuse_req *req)
223
224
{
	req->background = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
225
	list_add(&req->bg_entry, &fc->background);
226
227
228
	fc->num_background++;
	if (fc->num_background == FUSE_MAX_BACKGROUND)
		fc->blocked = 1;
229
230
231
232
233
234
235
236
	if (req->inode)
		req->inode = igrab(req->inode);
	if (req->inode2)
		req->inode2 = igrab(req->inode2);
	if (req->file)
		get_file(req->file);
}

237
/* Called with fc->lock held.  Releases, and then reacquires it. */
238
static void request_wait_answer(struct fuse_conn *fc, struct fuse_req *req)
239
{
240
	sigset_t oldset;
241

242
	spin_unlock(&fc->lock);
243
	block_sigs(&oldset);
244
	wait_event_interruptible(req->waitq, req->state == FUSE_REQ_FINISHED);
245
	restore_sigs(&oldset);
246
	spin_lock(&fc->lock);
247
	if (req->state == FUSE_REQ_FINISHED && !req->interrupted)
248
249
		return;

250
251
252
253
	if (!req->interrupted) {
		req->out.h.error = -EINTR;
		req->interrupted = 1;
	}
254
255
256
257
258
259
	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 */
260
		spin_unlock(&fc->lock);
261
		wait_event(req->waitq, !req->locked);
262
		spin_lock(&fc->lock);
263
	}
264
	if (req->state == FUSE_REQ_PENDING) {
265
266
		list_del(&req->list);
		__fuse_put_request(req);
267
	} else if (req->state == FUSE_REQ_SENT)
Miklos Szeredi's avatar
Miklos Szeredi committed
268
		background_request(fc, req);
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
}

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);
292
	req->state = FUSE_REQ_PENDING;
293
294
295
296
	if (!req->waiting) {
		req->waiting = 1;
		atomic_inc(&fc->num_waiting);
	}
297
	wake_up(&fc->waitq);
298
	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
299
300
}

301
302
303
304
/*
 * This can only be interrupted by a SIGKILL
 */
void request_send(struct fuse_conn *fc, struct fuse_req *req)
305
306
{
	req->isreply = 1;
307
	spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
308
	if (!fc->connected)
309
310
311
312
313
314
315
316
317
		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);

318
		request_wait_answer(fc, req);
319
	}
320
	spin_unlock(&fc->lock);
321
322
323
324
}

static void request_send_nowait(struct fuse_conn *fc, struct fuse_req *req)
{
325
	spin_lock(&fc->lock);
326
	background_request(fc, req);
Miklos Szeredi's avatar
Miklos Szeredi committed
327
	if (fc->connected) {
328
		queue_request(fc, req);
329
		spin_unlock(&fc->lock);
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
	} 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.
 */
353
static int lock_request(struct fuse_conn *fc, struct fuse_req *req)
354
355
356
{
	int err = 0;
	if (req) {
357
		spin_lock(&fc->lock);
358
359
360
361
		if (req->interrupted)
			err = -ENOENT;
		else
			req->locked = 1;
362
		spin_unlock(&fc->lock);
363
364
365
366
367
368
369
370
371
	}
	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.
 */
372
static void unlock_request(struct fuse_conn *fc, struct fuse_req *req)
373
374
{
	if (req) {
375
		spin_lock(&fc->lock);
376
377
378
		req->locked = 0;
		if (req->interrupted)
			wake_up(&req->waitq);
379
		spin_unlock(&fc->lock);
380
381
382
383
	}
}

struct fuse_copy_state {
384
	struct fuse_conn *fc;
385
386
387
388
389
390
391
392
393
394
395
396
	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;
};

397
398
399
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)
400
401
{
	memset(cs, 0, sizeof(*cs));
402
	cs->fc = fc;
403
404
405
406
407
408
409
	cs->write = write;
	cs->req = req;
	cs->iov = iov;
	cs->nr_segs = nr_segs;
}

/* Unmap and put previous page of userspace buffer */
410
static void fuse_copy_finish(struct fuse_copy_state *cs)
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
{
	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;

432
	unlock_request(cs->fc, cs->req);
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
	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;

455
	return lock_request(cs->fc, cs->req);
456
457
458
}

/* Do as much copy to/from userspace buffer as we can */
459
static int fuse_copy_do(struct fuse_copy_state *cs, void **val, unsigned *size)
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
{
	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
 */
479
480
static int fuse_copy_page(struct fuse_copy_state *cs, struct page *page,
			  unsigned offset, unsigned count, int zeroing)
481
482
483
484
485
486
487
488
489
490
491
492
493
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
{
	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);
562
	while (fc->connected && list_empty(&fc->pending)) {
563
564
565
566
		set_current_state(TASK_INTERRUPTIBLE);
		if (signal_pending(current))
			break;

567
		spin_unlock(&fc->lock);
568
		schedule();
569
		spin_lock(&fc->lock);
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
	}
	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;
592
593
594
	struct fuse_conn *fc = fuse_get_conn(file);
	if (!fc)
		return -EPERM;
595

596
 restart:
597
	spin_lock(&fc->lock);
598
599
600
601
602
	err = -EAGAIN;
	if ((file->f_flags & O_NONBLOCK) && fc->connected &&
	    list_empty(&fc->pending))
		goto err_unlock;

603
604
	request_wait(fc);
	err = -ENODEV;
605
	if (!fc->connected)
606
607
608
609
610
611
		goto err_unlock;
	err = -ERESTARTSYS;
	if (list_empty(&fc->pending))
		goto err_unlock;

	req = list_entry(fc->pending.next, struct fuse_req, list);
612
	req->state = FUSE_REQ_READING;
613
	list_move(&req->list, &fc->io);
614
615

	in = &req->in;
616
617
618
619
620
621
622
623
624
	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;
625
	}
626
627
	spin_unlock(&fc->lock);
	fuse_copy_init(&cs, fc, 1, req, iov, nr_segs);
628
629
630
631
	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);
632
	fuse_copy_finish(&cs);
633
	spin_lock(&fc->lock);
634
635
636
637
638
639
640
641
642
643
644
645
	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 {
646
		req->state = FUSE_REQ_SENT;
647
		list_move_tail(&req->list, &fc->processing);
648
		spin_unlock(&fc->lock);
649
650
651
652
	}
	return reqsize;

 err_unlock:
653
	spin_unlock(&fc->lock);
654
655
656
657
658
659
660
661
662
663
664
665
666
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
	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)
720
		return -EPERM;
721

722
	fuse_copy_init(&cs, fc, 0, NULL, iov, nr_segs);
723
724
725
726
727
728
729
730
731
732
733
	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;

734
	spin_lock(&fc->lock);
735
736
737
738
	err = -ENOENT;
	if (!fc->connected)
		goto err_unlock;

739
740
741
742
743
744
	req = request_find(fc, oh.unique);
	err = -EINVAL;
	if (!req)
		goto err_unlock;

	if (req->interrupted) {
745
		spin_unlock(&fc->lock);
746
		fuse_copy_finish(&cs);
747
		spin_lock(&fc->lock);
748
		request_end(fc, req);
749
750
		return -ENOENT;
	}
751
	list_move(&req->list, &fc->io);
752
753
754
	req->out.h = oh;
	req->locked = 1;
	cs.req = req;
755
	spin_unlock(&fc->lock);
756
757
758
759

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

760
	spin_lock(&fc->lock);
761
762
763
764
765
766
767
768
769
770
771
	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:
772
	spin_unlock(&fc->lock);
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
 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;
790
	struct fuse_conn *fc = fuse_get_conn(file);
791
	if (!fc)
792
		return POLLERR;
793
794
795

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

796
	spin_lock(&fc->lock);
797
798
799
800
	if (!fc->connected)
		mask = POLLERR;
	else if (!list_empty(&fc->pending))
		mask |= POLLIN | POLLRDNORM;
801
	spin_unlock(&fc->lock);
802
803
804
805

	return mask;
}

806
807
808
/*
 * Abort all requests on the given list (pending or processing)
 *
809
 * This function releases and reacquires fc->lock
810
 */
811
812
813
814
815
816
817
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);
818
		spin_lock(&fc->lock);
819
820
821
	}
}

822
823
824
825
826
827
/*
 * 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.
828
829
830
831
 *
 * 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).
832
833
834
835
 */
static void end_io_requests(struct fuse_conn *fc)
{
	while (!list_empty(&fc->io)) {
836
837
838
839
		struct fuse_req *req =
			list_entry(fc->io.next, struct fuse_req, list);
		void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;

840
841
842
843
844
		req->interrupted = 1;
		req->out.h.error = -ECONNABORTED;
		req->state = FUSE_REQ_FINISHED;
		list_del_init(&req->list);
		wake_up(&req->waitq);
845
846
847
848
		if (end) {
			req->end = NULL;
			/* The end function will consume this reference */
			__fuse_get_request(req);
849
			spin_unlock(&fc->lock);
850
851
			wait_event(req->waitq, !req->locked);
			end(fc, req);
852
			spin_lock(&fc->lock);
853
		}
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
	}
}

/*
 * 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)
{
879
	spin_lock(&fc->lock);
880
881
882
883
884
885
	if (fc->connected) {
		fc->connected = 0;
		end_io_requests(fc);
		end_requests(fc, &fc->pending);
		end_requests(fc, &fc->processing);
		wake_up_all(&fc->waitq);
886
		kill_fasync(&fc->fasync, SIGIO, POLL_IN);
887
	}
888
	spin_unlock(&fc->lock);
889
890
}

891
892
static int fuse_dev_release(struct inode *inode, struct file *file)
{
893
	struct fuse_conn *fc = fuse_get_conn(file);
894
	if (fc) {
895
		spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
896
		fc->connected = 0;
897
898
		end_requests(fc, &fc->pending);
		end_requests(fc, &fc->processing);
899
		spin_unlock(&fc->lock);
900
		fasync_helper(-1, file, 0, &fc->fasync);
901
		kobject_put(&fc->kobj);
902
	}
903

904
905
906
	return 0;
}

907
908
909
910
static int fuse_dev_fasync(int fd, struct file *file, int on)
{
	struct fuse_conn *fc = fuse_get_conn(file);
	if (!fc)
911
		return -EPERM;
912
913
914
915
916

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

917
const struct file_operations fuse_dev_operations = {
918
919
920
921
922
923
924
925
	.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,
926
	.fasync		= fuse_dev_fasync,
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
};

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);
}