dev.c 25.4 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

  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);

22
static struct kmem_cache *fuse_req_cachep;
23

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
{
	memset(req, 0, sizeof(*req));
	INIT_LIST_HEAD(&req->list);
37
	INIT_LIST_HEAD(&req->intr_entry);
38
39
40
41
42
43
	init_waitqueue_head(&req->waitq);
	atomic_set(&req->count, 1);
}

struct fuse_req *fuse_request_alloc(void)
{
44
	struct fuse_req *req = kmem_cache_alloc(fuse_req_cachep, GFP_KERNEL);
45
46
47
48
49
50
51
52
53
54
	if (req)
		fuse_request_init(req);
	return req;
}

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

55
static void block_sigs(sigset_t *oldset)
56
57
58
59
60
61
62
{
	sigset_t mask;

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

63
static void restore_sigs(sigset_t *oldset)
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
{
	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);
}

80
81
82
83
84
85
86
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;
}

87
struct fuse_req *fuse_get_req(struct fuse_conn *fc)
88
{
89
90
	struct fuse_req *req;
	sigset_t oldset;
91
	int intr;
92
93
	int err;

94
	atomic_inc(&fc->num_waiting);
95
	block_sigs(&oldset);
96
	intr = wait_event_interruptible(fc->blocked_waitq, !fc->blocked);
97
	restore_sigs(&oldset);
98
99
100
	err = -EINTR;
	if (intr)
		goto out;
101

102
103
104
105
	err = -ENOTCONN;
	if (!fc->connected)
		goto out;

106
	req = fuse_request_alloc();
107
	err = -ENOMEM;
108
	if (!req)
109
		goto out;
110

111
	fuse_req_init_context(req);
112
	req->waiting = 1;
113
	return req;
114
115
116
117

 out:
	atomic_dec(&fc->num_waiting);
	return ERR_PTR(err);
118
119
}

120
121
122
123
124
125
126
127
128
129
130
131
/*
 * 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 {
132
		wait_event(fc->reserved_req_waitq, ff->reserved_req);
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
		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;
158
	wake_up_all(&fc->reserved_req_waitq);
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
190
	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;
}

191
void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req)
192
193
{
	if (atomic_dec_and_test(&req->count)) {
194
195
		if (req->waiting)
			atomic_dec(&fc->num_waiting);
196
197
198
199
200

		if (req->stolen_file)
			put_reserved_req(fc, req);
		else
			fuse_request_free(req);
201
202
203
	}
}

204
205
/*
 * This function is called when a request is finished.  Either a reply
206
 * has arrived or it was aborted (and not yet sent) or some error
207
 * occurred during communication with userspace, or the device file
208
209
210
 * 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
211
 *
212
 * Called with fc->lock, unlocks it
213
214
 */
static void request_end(struct fuse_conn *fc, struct fuse_req *req)
215
	__releases(fc->lock)
216
{
217
218
	void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;
	req->end = NULL;
219
	list_del(&req->list);
220
	list_del(&req->intr_entry);
221
	req->state = FUSE_REQ_FINISHED;
222
223
224
225
226
	if (req->background) {
		if (fc->num_background == FUSE_MAX_BACKGROUND) {
			fc->blocked = 0;
			wake_up_all(&fc->blocked_waitq);
		}
227
228
229
230
		if (fc->num_background == FUSE_CONGESTION_THRESHOLD) {
			clear_bdi_congested(&fc->bdi, READ);
			clear_bdi_congested(&fc->bdi, WRITE);
		}
231
		fc->num_background--;
232
	}
233
234
235
	spin_unlock(&fc->lock);
	dput(req->dentry);
	mntput(req->vfsmount);
236
	if (req->file)
237
238
239
240
241
242
		fput(req->file);
	wake_up(&req->waitq);
	if (end)
		end(fc, req);
	else
		fuse_put_request(fc, req);
243
244
}

245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
static void wait_answer_interruptible(struct fuse_conn *fc,
				      struct fuse_req *req)
{
	if (signal_pending(current))
		return;

	spin_unlock(&fc->lock);
	wait_event_interruptible(req->waitq, req->state == FUSE_REQ_FINISHED);
	spin_lock(&fc->lock);
}

static void queue_interrupt(struct fuse_conn *fc, struct fuse_req *req)
{
	list_add_tail(&req->intr_entry, &fc->interrupts);
	wake_up(&fc->waitq);
	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
}

263
/* Called with fc->lock held.  Releases, and then reacquires it. */
264
static void request_wait_answer(struct fuse_conn *fc, struct fuse_req *req)
265
{
266
267
268
	if (!fc->no_interrupt) {
		/* Any signal may interrupt this */
		wait_answer_interruptible(fc, req);
269

270
271
272
273
274
275
276
277
278
279
280
281
		if (req->aborted)
			goto aborted;
		if (req->state == FUSE_REQ_FINISHED)
			return;

		req->interrupted = 1;
		if (req->state == FUSE_REQ_SENT)
			queue_interrupt(fc, req);
	}

	if (req->force) {
		spin_unlock(&fc->lock);
282
		wait_event(req->waitq, req->state == FUSE_REQ_FINISHED);
283
284
285
286
287
		spin_lock(&fc->lock);
	} else {
		sigset_t oldset;

		/* Only fatal signals may interrupt this */
288
		block_sigs(&oldset);
289
		wait_answer_interruptible(fc, req);
290
291
		restore_sigs(&oldset);
	}
292

293
294
295
296
297
298
299
300
301
	if (req->aborted)
		goto aborted;
	if (req->state == FUSE_REQ_FINISHED)
 		return;

	req->out.h.error = -EINTR;
	req->aborted = 1;

 aborted:
302
303
304
305
306
307
	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 */
308
		spin_unlock(&fc->lock);
309
		wait_event(req->waitq, !req->locked);
310
		spin_lock(&fc->lock);
311
	}
312
	if (req->state == FUSE_REQ_PENDING) {
313
314
		list_del(&req->list);
		__fuse_put_request(req);
315
316
317
318
319
	} else if (req->state == FUSE_REQ_SENT) {
		spin_unlock(&fc->lock);
		wait_event(req->waitq, req->state == FUSE_REQ_FINISHED);
		spin_lock(&fc->lock);
	}
320
321
322
323
324
325
326
327
328
329
330
331
332
}

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;
}

333
334
335
336
337
338
339
340
341
342
static u64 fuse_get_unique(struct fuse_conn *fc)
 {
 	fc->reqctr++;
 	/* zero is special */
 	if (fc->reqctr == 0)
 		fc->reqctr = 1;

	return fc->reqctr;
}

343
344
static void queue_request(struct fuse_conn *fc, struct fuse_req *req)
{
345
	req->in.h.unique = fuse_get_unique(fc);
346
347
348
	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);
349
	req->state = FUSE_REQ_PENDING;
350
351
352
353
	if (!req->waiting) {
		req->waiting = 1;
		atomic_inc(&fc->num_waiting);
	}
354
	wake_up(&fc->waitq);
355
	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
356
357
}

358
void request_send(struct fuse_conn *fc, struct fuse_req *req)
359
360
{
	req->isreply = 1;
361
	spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
362
	if (!fc->connected)
363
364
365
366
367
368
369
370
371
		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);

372
		request_wait_answer(fc, req);
373
	}
374
	spin_unlock(&fc->lock);
375
376
377
378
}

static void request_send_nowait(struct fuse_conn *fc, struct fuse_req *req)
{
379
	spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
380
	if (fc->connected) {
381
382
383
384
		req->background = 1;
		fc->num_background++;
		if (fc->num_background == FUSE_MAX_BACKGROUND)
			fc->blocked = 1;
385
386
387
388
		if (fc->num_background == FUSE_CONGESTION_THRESHOLD) {
			set_bdi_congested(&fc->bdi, READ);
			set_bdi_congested(&fc->bdi, WRITE);
		}
389

390
		queue_request(fc, req);
391
		spin_unlock(&fc->lock);
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
	} 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
413
 * aborted bail out.
414
 */
415
static int lock_request(struct fuse_conn *fc, struct fuse_req *req)
416
417
418
{
	int err = 0;
	if (req) {
419
		spin_lock(&fc->lock);
420
		if (req->aborted)
421
422
423
			err = -ENOENT;
		else
			req->locked = 1;
424
		spin_unlock(&fc->lock);
425
426
427
428
429
	}
	return err;
}

/*
430
 * Unlock request.  If it was aborted during being locked, the
431
432
433
 * requester thread is currently waiting for it to be unlocked, so
 * wake it up.
 */
434
static void unlock_request(struct fuse_conn *fc, struct fuse_req *req)
435
436
{
	if (req) {
437
		spin_lock(&fc->lock);
438
		req->locked = 0;
439
		if (req->aborted)
440
			wake_up(&req->waitq);
441
		spin_unlock(&fc->lock);
442
443
444
445
	}
}

struct fuse_copy_state {
446
	struct fuse_conn *fc;
447
448
449
450
451
452
453
454
455
456
457
458
	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;
};

459
460
461
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)
462
463
{
	memset(cs, 0, sizeof(*cs));
464
	cs->fc = fc;
465
466
467
468
469
470
471
	cs->write = write;
	cs->req = req;
	cs->iov = iov;
	cs->nr_segs = nr_segs;
}

/* Unmap and put previous page of userspace buffer */
472
static void fuse_copy_finish(struct fuse_copy_state *cs)
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
{
	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;

494
	unlock_request(cs->fc, cs->req);
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
	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;

517
	return lock_request(cs->fc, cs->req);
518
519
520
}

/* Do as much copy to/from userspace buffer as we can */
521
static int fuse_copy_do(struct fuse_copy_state *cs, void **val, unsigned *size)
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
{
	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
 */
541
542
static int fuse_copy_page(struct fuse_copy_state *cs, struct page *page,
			  unsigned offset, unsigned count, int zeroing)
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
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
{
	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;
}

618
619
620
621
622
static int request_pending(struct fuse_conn *fc)
{
	return !list_empty(&fc->pending) || !list_empty(&fc->interrupts);
}

623
624
625
626
627
628
/* 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);
629
	while (fc->connected && !request_pending(fc)) {
630
631
632
633
		set_current_state(TASK_INTERRUPTIBLE);
		if (signal_pending(current))
			break;

634
		spin_unlock(&fc->lock);
635
		schedule();
636
		spin_lock(&fc->lock);
637
638
639
640
641
	}
	set_current_state(TASK_RUNNING);
	remove_wait_queue(&fc->waitq, &wait);
}

642
643
644
645
646
647
648
649
650
651
/*
 * Transfer an interrupt request to userspace
 *
 * Unlike other requests this is assembled on demand, without a need
 * to allocate a separate fuse_req structure.
 *
 * Called with fc->lock held, releases it
 */
static int fuse_read_interrupt(struct fuse_conn *fc, struct fuse_req *req,
			       const struct iovec *iov, unsigned long nr_segs)
652
	__releases(fc->lock)
653
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
{
	struct fuse_copy_state cs;
	struct fuse_in_header ih;
	struct fuse_interrupt_in arg;
	unsigned reqsize = sizeof(ih) + sizeof(arg);
	int err;

	list_del_init(&req->intr_entry);
	req->intr_unique = fuse_get_unique(fc);
	memset(&ih, 0, sizeof(ih));
	memset(&arg, 0, sizeof(arg));
	ih.len = reqsize;
	ih.opcode = FUSE_INTERRUPT;
	ih.unique = req->intr_unique;
	arg.unique = req->in.h.unique;

	spin_unlock(&fc->lock);
	if (iov_length(iov, nr_segs) < reqsize)
		return -EINVAL;

	fuse_copy_init(&cs, fc, 1, NULL, iov, nr_segs);
	err = fuse_copy_one(&cs, &ih, sizeof(ih));
	if (!err)
		err = fuse_copy_one(&cs, &arg, sizeof(arg));
	fuse_copy_finish(&cs);

	return err ? err : reqsize;
}

682
683
684
685
/*
 * 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
686
687
 * no reply is needed (FORGET) or request has been aborted or there
 * was an error during the copying then it's finished by calling
688
689
690
 * request_end().  Otherwise add it to the processing list, and set
 * the 'sent' flag.
 */
691
692
static ssize_t fuse_dev_read(struct kiocb *iocb, const struct iovec *iov,
			      unsigned long nr_segs, loff_t pos)
693
694
695
696
697
698
{
	int err;
	struct fuse_req *req;
	struct fuse_in *in;
	struct fuse_copy_state cs;
	unsigned reqsize;
699
	struct file *file = iocb->ki_filp;
700
701
702
	struct fuse_conn *fc = fuse_get_conn(file);
	if (!fc)
		return -EPERM;
703

704
 restart:
705
	spin_lock(&fc->lock);
706
707
	err = -EAGAIN;
	if ((file->f_flags & O_NONBLOCK) && fc->connected &&
708
	    !request_pending(fc))
709
710
		goto err_unlock;

711
712
	request_wait(fc);
	err = -ENODEV;
713
	if (!fc->connected)
714
715
		goto err_unlock;
	err = -ERESTARTSYS;
716
	if (!request_pending(fc))
717
718
		goto err_unlock;

719
720
721
722
723
724
	if (!list_empty(&fc->interrupts)) {
		req = list_entry(fc->interrupts.next, struct fuse_req,
				 intr_entry);
		return fuse_read_interrupt(fc, req, iov, nr_segs);
	}

725
	req = list_entry(fc->pending.next, struct fuse_req, list);
726
	req->state = FUSE_REQ_READING;
727
	list_move(&req->list, &fc->io);
728
729

	in = &req->in;
730
731
732
733
734
735
736
737
738
	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;
739
	}
740
741
	spin_unlock(&fc->lock);
	fuse_copy_init(&cs, fc, 1, req, iov, nr_segs);
742
743
744
745
	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);
746
	fuse_copy_finish(&cs);
747
	spin_lock(&fc->lock);
748
	req->locked = 0;
749
	if (!err && req->aborted)
750
751
		err = -ENOENT;
	if (err) {
752
		if (!req->aborted)
753
754
755
756
757
758
759
			req->out.h.error = -EIO;
		request_end(fc, req);
		return err;
	}
	if (!req->isreply)
		request_end(fc, req);
	else {
760
		req->state = FUSE_REQ_SENT;
761
		list_move_tail(&req->list, &fc->processing);
762
763
		if (req->interrupted)
			queue_interrupt(fc, req);
764
		spin_unlock(&fc->lock);
765
766
767
768
	}
	return reqsize;

 err_unlock:
769
	spin_unlock(&fc->lock);
770
771
772
773
774
775
776
777
778
779
780
	return err;
}

/* 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);
781
		if (req->in.h.unique == unique || req->intr_unique == unique)
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
			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()
 */
817
818
static ssize_t fuse_dev_write(struct kiocb *iocb, const struct iovec *iov,
			       unsigned long nr_segs, loff_t pos)
819
820
821
822
823
824
{
	int err;
	unsigned nbytes = iov_length(iov, nr_segs);
	struct fuse_req *req;
	struct fuse_out_header oh;
	struct fuse_copy_state cs;
825
	struct fuse_conn *fc = fuse_get_conn(iocb->ki_filp);
826
	if (!fc)
827
		return -EPERM;
828

829
	fuse_copy_init(&cs, fc, 0, NULL, iov, nr_segs);
830
831
832
833
834
835
836
837
838
839
840
	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;

841
	spin_lock(&fc->lock);
842
843
844
845
	err = -ENOENT;
	if (!fc->connected)
		goto err_unlock;

846
847
848
849
	req = request_find(fc, oh.unique);
	if (!req)
		goto err_unlock;

850
	if (req->aborted) {
851
		spin_unlock(&fc->lock);
852
		fuse_copy_finish(&cs);
853
		spin_lock(&fc->lock);
854
		request_end(fc, req);
855
856
		return -ENOENT;
	}
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
	/* Is it an interrupt reply? */
	if (req->intr_unique == oh.unique) {
		err = -EINVAL;
		if (nbytes != sizeof(struct fuse_out_header))
			goto err_unlock;

		if (oh.error == -ENOSYS)
			fc->no_interrupt = 1;
		else if (oh.error == -EAGAIN)
			queue_interrupt(fc, req);

		spin_unlock(&fc->lock);
		fuse_copy_finish(&cs);
		return nbytes;
	}

	req->state = FUSE_REQ_WRITING;
874
	list_move(&req->list, &fc->io);
875
876
877
	req->out.h = oh;
	req->locked = 1;
	cs.req = req;
878
	spin_unlock(&fc->lock);
879
880
881
882

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

883
	spin_lock(&fc->lock);
884
885
	req->locked = 0;
	if (!err) {
886
		if (req->aborted)
887
			err = -ENOENT;
888
	} else if (!req->aborted)
889
890
891
892
893
894
		req->out.h.error = -EIO;
	request_end(fc, req);

	return err ? err : nbytes;

 err_unlock:
895
	spin_unlock(&fc->lock);
896
897
898
899
900
901
902
903
 err_finish:
	fuse_copy_finish(&cs);
	return err;
}

static unsigned fuse_dev_poll(struct file *file, poll_table *wait)
{
	unsigned mask = POLLOUT | POLLWRNORM;
904
	struct fuse_conn *fc = fuse_get_conn(file);
905
	if (!fc)
906
		return POLLERR;
907
908
909

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

910
	spin_lock(&fc->lock);
911
912
	if (!fc->connected)
		mask = POLLERR;
913
	else if (request_pending(fc))
914
		mask |= POLLIN | POLLRDNORM;
915
	spin_unlock(&fc->lock);
916
917
918
919

	return mask;
}

920
921
922
/*
 * Abort all requests on the given list (pending or processing)
 *
923
 * This function releases and reacquires fc->lock
924
 */
925
926
927
928
929
930
931
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);
932
		spin_lock(&fc->lock);
933
934
935
	}
}

936
937
938
/*
 * Abort requests under I/O
 *
939
 * The requests are set to aborted and finished, and the request
940
941
 * waiter is woken up.  This will make request_wait_answer() wait
 * until the request is unlocked and then return.
942
943
944
945
 *
 * 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).
946
947
948
949
 */
static void end_io_requests(struct fuse_conn *fc)
{
	while (!list_empty(&fc->io)) {
950
951
952
953
		struct fuse_req *req =
			list_entry(fc->io.next, struct fuse_req, list);
		void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;

954
		req->aborted = 1;
955
956
957
958
		req->out.h.error = -ECONNABORTED;
		req->state = FUSE_REQ_FINISHED;
		list_del_init(&req->list);
		wake_up(&req->waitq);
959
960
961
962
		if (end) {
			req->end = NULL;
			/* The end function will consume this reference */
			__fuse_get_request(req);
963
			spin_unlock(&fc->lock);
964
965
			wait_event(req->waitq, !req->locked);
			end(fc, req);
966
			spin_lock(&fc->lock);
967
		}
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
	}
}

/*
 * 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
987
988
 * prevented by the req->aborted flag being true for these requests.
 * For this reason requests on the io list must be aborted first.
989
990
991
 */
void fuse_abort_conn(struct fuse_conn *fc)
{
992
	spin_lock(&fc->lock);
993
994
	if (fc->connected) {
		fc->connected = 0;
995
		fc->blocked = 0;
996
997
998
999
		end_io_requests(fc);
		end_requests(fc, &fc->pending);
		end_requests(fc, &fc->processing);
		wake_up_all(&fc->waitq);
1000
		wake_up_all(&fc->blocked_waitq);
1001
		kill_fasync(&fc->fasync, SIGIO, POLL_IN);
1002
	}
1003
	spin_unlock(&fc->lock);
1004
1005
}

1006
1007
static int fuse_dev_release(struct inode *inode, struct file *file)
{
1008
	struct fuse_conn *fc = fuse_get_conn(file);
1009
	if (fc) {
1010
		spin_lock(&fc->lock);
Miklos Szeredi's avatar
Miklos Szeredi committed
1011
		fc->connected = 0;
1012
1013
		end_requests(fc, &fc->pending);
		end_requests(fc, &fc->processing);
1014
		spin_unlock(&fc->lock);
1015
		fasync_helper(-1, file, 0, &fc->fasync);
1016
		fuse_conn_put(fc);
1017
	}
1018

1019
1020
1021
	return 0;
}

1022
1023
1024
1025
static int fuse_dev_fasync(int fd, struct file *file, int on)
{
	struct fuse_conn *fc = fuse_get_conn(file);
	if (!fc)
1026
		return -EPERM;
1027
1028
1029
1030
1031

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

1032
const struct file_operations fuse_dev_operations = {
1033
1034
	.owner		= THIS_MODULE,
	.llseek		= no_llseek,
1035
1036
1037
1038
	.read		= do_sync_read,
	.aio_read	= fuse_dev_read,
	.write		= do_sync_write,
	.aio_write	= fuse_dev_write,
1039
1040
	.poll		= fuse_dev_poll,
	.release	= fuse_dev_release,
1041
	.fasync		= fuse_dev_fasync,
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
};

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),
1055
					    0, 0, NULL);
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
	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);
}