file.c 56.2 KB
Newer Older
1
2
/*
  FUSE: Filesystem in Userspace
Miklos Szeredi's avatar
Miklos Szeredi committed
3
  Copyright (C) 2001-2008  Miklos Szeredi <miklos@szeredi.hu>
4
5
6
7
8
9
10
11
12
13

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

#include "fuse_i.h"

#include <linux/pagemap.h>
#include <linux/slab.h>
#include <linux/kernel.h>
Alexey Dobriyan's avatar
Alexey Dobriyan committed
14
#include <linux/sched.h>
15
#include <linux/module.h>
16
#include <linux/compat.h>
17
#include <linux/swap.h>
18
#include <linux/aio.h>
19

20
static const struct file_operations fuse_direct_io_file_operations;
21

22
23
static int fuse_send_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
			  int opcode, struct fuse_open_out *outargp)
24
25
{
	struct fuse_open_in inarg;
26
27
28
	struct fuse_req *req;
	int err;

29
	req = fuse_get_req_nopages(fc);
30
31
	if (IS_ERR(req))
		return PTR_ERR(req);
32
33

	memset(&inarg, 0, sizeof(inarg));
34
35
36
	inarg.flags = file->f_flags & ~(O_CREAT | O_EXCL | O_NOCTTY);
	if (!fc->atomic_o_trunc)
		inarg.flags &= ~O_TRUNC;
37
38
	req->in.h.opcode = opcode;
	req->in.h.nodeid = nodeid;
39
40
41
42
43
44
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(*outargp);
	req->out.args[0].value = outargp;
45
	fuse_request_send(fc, req);
46
47
48
49
50
51
	err = req->out.h.error;
	fuse_put_request(fc, req);

	return err;
}

Tejun Heo's avatar
Tejun Heo committed
52
struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
53
54
{
	struct fuse_file *ff;
Tejun Heo's avatar
Tejun Heo committed
55

56
	ff = kmalloc(sizeof(struct fuse_file), GFP_KERNEL);
Tejun Heo's avatar
Tejun Heo committed
57
58
59
	if (unlikely(!ff))
		return NULL;

60
	ff->fc = fc;
61
	ff->reserved_req = fuse_request_alloc(0);
Tejun Heo's avatar
Tejun Heo committed
62
63
64
	if (unlikely(!ff->reserved_req)) {
		kfree(ff);
		return NULL;
65
	}
Tejun Heo's avatar
Tejun Heo committed
66
67
68
69
70
71
72
73
74
75

	INIT_LIST_HEAD(&ff->write_entry);
	atomic_set(&ff->count, 0);
	RB_CLEAR_NODE(&ff->polled_node);
	init_waitqueue_head(&ff->poll_wait);

	spin_lock(&fc->lock);
	ff->kh = ++fc->khctr;
	spin_unlock(&fc->lock);

76
77
78
79
80
	return ff;
}

void fuse_file_free(struct fuse_file *ff)
{
81
	fuse_request_free(ff->reserved_req);
82
83
84
	kfree(ff);
}

85
struct fuse_file *fuse_file_get(struct fuse_file *ff)
86
87
88
89
90
{
	atomic_inc(&ff->count);
	return ff;
}

91
92
93
94
95
96
97
98
99
100
101
102
103
104
static void fuse_release_async(struct work_struct *work)
{
	struct fuse_req *req;
	struct fuse_conn *fc;
	struct path path;

	req = container_of(work, struct fuse_req, misc.release.work);
	path = req->misc.release.path;
	fc = get_fuse_conn(path.dentry->d_inode);

	fuse_put_request(fc, req);
	path_put(&path);
}

Miklos Szeredi's avatar
Miklos Szeredi committed
105
106
static void fuse_release_end(struct fuse_conn *fc, struct fuse_req *req)
{
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
	if (fc->destroy_req) {
		/*
		 * If this is a fuseblk mount, then it's possible that
		 * releasing the path will result in releasing the
		 * super block and sending the DESTROY request.  If
		 * the server is single threaded, this would hang.
		 * For this reason do the path_put() in a separate
		 * thread.
		 */
		atomic_inc(&req->count);
		INIT_WORK(&req->misc.release.work, fuse_release_async);
		schedule_work(&req->misc.release.work);
	} else {
		path_put(&req->misc.release.path);
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
122
123
}

124
static void fuse_file_put(struct fuse_file *ff, bool sync)
125
126
127
{
	if (atomic_dec_and_test(&ff->count)) {
		struct fuse_req *req = ff->reserved_req;
128

129
130
131
132
133
134
135
136
		if (sync) {
			fuse_request_send(ff->fc, req);
			path_put(&req->misc.release.path);
			fuse_put_request(ff->fc, req);
		} else {
			req->end = fuse_release_end;
			fuse_request_send_background(ff->fc, req);
		}
137
138
139
140
		kfree(ff);
	}
}

141
142
int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
		 bool isdir)
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
{
	struct fuse_open_out outarg;
	struct fuse_file *ff;
	int err;
	int opcode = isdir ? FUSE_OPENDIR : FUSE_OPEN;

	ff = fuse_file_alloc(fc);
	if (!ff)
		return -ENOMEM;

	err = fuse_send_open(fc, nodeid, file, opcode, &outarg);
	if (err) {
		fuse_file_free(ff);
		return err;
	}

	if (isdir)
		outarg.open_flags &= ~FOPEN_DIRECT_IO;

	ff->fh = outarg.fh;
	ff->nodeid = nodeid;
	ff->open_flags = outarg.open_flags;
	file->private_data = fuse_file_get(ff);

	return 0;
}
169
EXPORT_SYMBOL_GPL(fuse_do_open);
170

171
void fuse_finish_open(struct inode *inode, struct file *file)
172
{
173
	struct fuse_file *ff = file->private_data;
174
	struct fuse_conn *fc = get_fuse_conn(inode);
175
176

	if (ff->open_flags & FOPEN_DIRECT_IO)
177
		file->f_op = &fuse_direct_io_file_operations;
178
	if (!(ff->open_flags & FOPEN_KEEP_CACHE))
Miklos Szeredi's avatar
Miklos Szeredi committed
179
		invalidate_inode_pages2(inode->i_mapping);
180
	if (ff->open_flags & FOPEN_NONSEEKABLE)
Tejun Heo's avatar
Tejun Heo committed
181
		nonseekable_open(inode, file);
182
183
184
185
186
187
188
189
190
	if (fc->atomic_o_trunc && (file->f_flags & O_TRUNC)) {
		struct fuse_inode *fi = get_fuse_inode(inode);

		spin_lock(&fc->lock);
		fi->attr_version = ++fc->attr_version;
		i_size_write(inode, 0);
		spin_unlock(&fc->lock);
		fuse_invalidate_attr(inode);
	}
191
192
}

193
int fuse_open_common(struct inode *inode, struct file *file, bool isdir)
194
{
Tejun Heo's avatar
Tejun Heo committed
195
	struct fuse_conn *fc = get_fuse_conn(inode);
196
197
198
199
200
201
	int err;

	err = generic_file_open(inode, file);
	if (err)
		return err;

202
	err = fuse_do_open(fc, get_node_id(inode), file, isdir);
203
	if (err)
204
		return err;
205

206
207
208
	fuse_finish_open(inode, file);

	return 0;
209
210
}

211
static void fuse_prepare_release(struct fuse_file *ff, int flags, int opcode)
212
{
213
	struct fuse_conn *fc = ff->fc;
214
	struct fuse_req *req = ff->reserved_req;
215
	struct fuse_release_in *inarg = &req->misc.release.in;
216

217
218
219
220
221
222
	spin_lock(&fc->lock);
	list_del(&ff->write_entry);
	if (!RB_EMPTY_NODE(&ff->polled_node))
		rb_erase(&ff->polled_node, &fc->polled_files);
	spin_unlock(&fc->lock);

223
	wake_up_interruptible_all(&ff->poll_wait);
224

225
	inarg->fh = ff->fh;
226
	inarg->flags = flags;
227
	req->in.h.opcode = opcode;
228
	req->in.h.nodeid = ff->nodeid;
229
230
231
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_release_in);
	req->in.args[0].value = inarg;
232
233
}

234
void fuse_release_common(struct file *file, int opcode)
235
{
Tejun Heo's avatar
Tejun Heo committed
236
237
	struct fuse_file *ff;
	struct fuse_req *req;
238

Tejun Heo's avatar
Tejun Heo committed
239
240
	ff = file->private_data;
	if (unlikely(!ff))
241
		return;
Tejun Heo's avatar
Tejun Heo committed
242
243

	req = ff->reserved_req;
244
	fuse_prepare_release(ff, file->f_flags, opcode);
Tejun Heo's avatar
Tejun Heo committed
245

Miklos Szeredi's avatar
Miklos Szeredi committed
246
247
248
249
250
251
	if (ff->flock) {
		struct fuse_release_in *inarg = &req->misc.release.in;
		inarg->release_flags |= FUSE_RELEASE_FLOCK_UNLOCK;
		inarg->lock_owner = fuse_lock_owner_id(ff->fc,
						       (fl_owner_t) file);
	}
Tejun Heo's avatar
Tejun Heo committed
252
	/* Hold vfsmount and dentry until release is finished */
253
254
	path_get(&file->f_path);
	req->misc.release.path = file->f_path;
Tejun Heo's avatar
Tejun Heo committed
255
256
257
258
259

	/*
	 * Normally this will send the RELEASE request, however if
	 * some asynchronous READ or WRITE requests are outstanding,
	 * the sending will be delayed.
260
261
262
263
	 *
	 * Make the release synchronous if this is a fuseblk mount,
	 * synchronous RELEASE is allowed (and desirable) in this case
	 * because the server can be trusted not to screw up.
Tejun Heo's avatar
Tejun Heo committed
264
	 */
265
	fuse_file_put(ff, ff->fc->destroy_req != NULL);
266
267
}

268
269
static int fuse_open(struct inode *inode, struct file *file)
{
270
	return fuse_open_common(inode, file, false);
271
272
273
274
}

static int fuse_release(struct inode *inode, struct file *file)
{
275
276
277
278
279
280
281
282
283
284
285
286
287
288
	fuse_release_common(file, FUSE_RELEASE);

	/* return value is ignored by VFS */
	return 0;
}

void fuse_sync_release(struct fuse_file *ff, int flags)
{
	WARN_ON(atomic_read(&ff->count) > 1);
	fuse_prepare_release(ff, flags, FUSE_RELEASE);
	ff->reserved_req->force = 1;
	fuse_request_send(ff->fc, ff->reserved_req);
	fuse_put_request(ff->fc, ff->reserved_req);
	kfree(ff);
289
}
290
EXPORT_SYMBOL_GPL(fuse_sync_release);
291

292
/*
293
294
 * Scramble the ID space with XTEA, so that the value of the files_struct
 * pointer is not exposed to userspace.
295
 */
296
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id)
297
{
298
299
300
301
302
303
304
305
306
307
308
309
310
311
	u32 *k = fc->scramble_key;
	u64 v = (unsigned long) id;
	u32 v0 = v;
	u32 v1 = v >> 32;
	u32 sum = 0;
	int i;

	for (i = 0; i < 32; i++) {
		v0 += ((v1 << 4 ^ v1 >> 5) + v1) ^ (sum + k[sum & 3]);
		sum += 0x9E3779B9;
		v1 += ((v0 << 4 ^ v0 >> 5) + v0) ^ (sum + k[sum>>11 & 3]);
	}

	return (u64) v0 + ((u64) v1 << 32);
312
313
}

Miklos Szeredi's avatar
Miklos Szeredi committed
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
/*
 * Check if page is under writeback
 *
 * This is currently done by walking the list of writepage requests
 * for the inode, which can be pretty inefficient.
 */
static bool fuse_page_is_writeback(struct inode *inode, pgoff_t index)
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct fuse_req *req;
	bool found = false;

	spin_lock(&fc->lock);
	list_for_each_entry(req, &fi->writepages, writepages_entry) {
		pgoff_t curr_index;

		BUG_ON(req->inode != inode);
		curr_index = req->misc.write.in.offset >> PAGE_CACHE_SHIFT;
		if (curr_index == index) {
			found = true;
			break;
		}
	}
	spin_unlock(&fc->lock);

	return found;
}

/*
 * Wait for page writeback to be completed.
 *
 * Since fuse doesn't rely on the VM writeback tracking, this has to
 * use some other means.
 */
static int fuse_wait_on_page_writeback(struct inode *inode, pgoff_t index)
{
	struct fuse_inode *fi = get_fuse_inode(inode);

	wait_event(fi->page_waitq, !fuse_page_is_writeback(inode, index));
	return 0;
}

357
static int fuse_flush(struct file *file, fl_owner_t id)
358
{
359
	struct inode *inode = file_inode(file);
360
361
362
363
364
365
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_file *ff = file->private_data;
	struct fuse_req *req;
	struct fuse_flush_in inarg;
	int err;

366
367
368
	if (is_bad_inode(inode))
		return -EIO;

369
370
371
	if (fc->no_flush)
		return 0;

372
	req = fuse_get_req_nofail_nopages(fc, file);
373
374
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
375
	inarg.lock_owner = fuse_lock_owner_id(fc, id);
376
377
378
379
380
	req->in.h.opcode = FUSE_FLUSH;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
381
	req->force = 1;
382
	fuse_request_send(fc, req);
383
384
385
386
387
388
389
390
391
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
		fc->no_flush = 1;
		err = 0;
	}
	return err;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
/*
 * Wait for all pending writepages on the inode to finish.
 *
 * This is currently done by blocking further writes with FUSE_NOWRITE
 * and waiting for all sent writes to complete.
 *
 * This must be called under i_mutex, otherwise the FUSE_NOWRITE usage
 * could conflict with truncation.
 */
static void fuse_sync_writes(struct inode *inode)
{
	fuse_set_nowrite(inode);
	fuse_release_nowrite(inode);
}

407
408
int fuse_fsync_common(struct file *file, loff_t start, loff_t end,
		      int datasync, int isdir)
409
{
410
	struct inode *inode = file->f_mapping->host;
411
412
413
414
415
416
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_file *ff = file->private_data;
	struct fuse_req *req;
	struct fuse_fsync_in inarg;
	int err;

417
418
419
	if (is_bad_inode(inode))
		return -EIO;

420
421
422
423
	err = filemap_write_and_wait_range(inode->i_mapping, start, end);
	if (err)
		return err;

424
	if ((!isdir && fc->no_fsync) || (isdir && fc->no_fsyncdir))
425
426
		return 0;

427
428
	mutex_lock(&inode->i_mutex);

Miklos Szeredi's avatar
Miklos Szeredi committed
429
430
431
432
433
434
435
	/*
	 * Start writeback against all dirty pages of the inode, then
	 * wait for all outstanding writes, before sending the FSYNC
	 * request.
	 */
	err = write_inode_now(inode, 0);
	if (err)
436
		goto out;
Miklos Szeredi's avatar
Miklos Szeredi committed
437
438
439

	fuse_sync_writes(inode);

440
	req = fuse_get_req_nopages(fc);
441
442
443
444
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		goto out;
	}
445
446
447
448

	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
	inarg.fsync_flags = datasync ? 1 : 0;
449
	req->in.h.opcode = isdir ? FUSE_FSYNCDIR : FUSE_FSYNC;
450
451
452
453
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
454
	fuse_request_send(fc, req);
455
456
457
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
458
459
460
461
		if (isdir)
			fc->no_fsyncdir = 1;
		else
			fc->no_fsync = 1;
462
463
		err = 0;
	}
464
465
out:
	mutex_unlock(&inode->i_mutex);
466
467
468
	return err;
}

469
470
static int fuse_fsync(struct file *file, loff_t start, loff_t end,
		      int datasync)
471
{
472
	return fuse_fsync_common(file, start, end, datasync, 0);
473
474
}

475
476
void fuse_read_fill(struct fuse_req *req, struct file *file, loff_t pos,
		    size_t count, int opcode)
477
{
478
	struct fuse_read_in *inarg = &req->misc.read.in;
479
	struct fuse_file *ff = file->private_data;
480

481
482
483
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
484
	inarg->flags = file->f_flags;
485
	req->in.h.opcode = opcode;
486
	req->in.h.nodeid = ff->nodeid;
487
488
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_read_in);
489
	req->in.args[0].value = inarg;
490
491
492
493
494
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = count;
}

495
static size_t fuse_send_read(struct fuse_req *req, struct file *file,
496
			     loff_t pos, size_t count, fl_owner_t owner)
497
{
498
499
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
500

501
	fuse_read_fill(req, file, pos, count, FUSE_READ);
502
	if (owner != NULL) {
503
		struct fuse_read_in *inarg = &req->misc.read.in;
504
505
506
507

		inarg->read_flags |= FUSE_READ_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
508
	fuse_request_send(fc, req);
509
	return req->out.args[0].size;
510
511
}

512
513
514
515
516
517
518
519
520
521
522
523
524
525
static void fuse_read_update_size(struct inode *inode, loff_t size,
				  u64 attr_ver)
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);

	spin_lock(&fc->lock);
	if (attr_ver == fi->attr_version && size < inode->i_size) {
		fi->attr_version = ++fc->attr_version;
		i_size_write(inode, size);
	}
	spin_unlock(&fc->lock);
}

526
527
528
529
static int fuse_readpage(struct file *file, struct page *page)
{
	struct inode *inode = page->mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
530
	struct fuse_req *req;
531
532
533
534
	size_t num_read;
	loff_t pos = page_offset(page);
	size_t count = PAGE_CACHE_SIZE;
	u64 attr_ver;
535
536
537
538
539
540
	int err;

	err = -EIO;
	if (is_bad_inode(inode))
		goto out;

Miklos Szeredi's avatar
Miklos Szeredi committed
541
	/*
Lucas De Marchi's avatar
Lucas De Marchi committed
542
	 * Page writeback can extend beyond the lifetime of the
Miklos Szeredi's avatar
Miklos Szeredi committed
543
544
545
546
547
	 * page-cache page, so make sure we read a properly synced
	 * page.
	 */
	fuse_wait_on_page_writeback(inode, page->index);

548
	req = fuse_get_req(fc, 1);
549
550
	err = PTR_ERR(req);
	if (IS_ERR(req))
551
552
		goto out;

553
554
	attr_ver = fuse_get_attr_version(fc);

555
	req->out.page_zeroing = 1;
556
	req->out.argpages = 1;
557
558
	req->num_pages = 1;
	req->pages[0] = page;
559
	req->page_descs[0].length = count;
560
	num_read = fuse_send_read(req, file, pos, count, NULL);
561
562
	err = req->out.h.error;
	fuse_put_request(fc, req);
563
564
565
566
567
568
569
570

	if (!err) {
		/*
		 * Short read means EOF.  If file size is larger, truncate it
		 */
		if (num_read < count)
			fuse_read_update_size(inode, pos + num_read, attr_ver);

571
		SetPageUptodate(page);
572
573
	}

574
	fuse_invalidate_attr(inode); /* atime changed */
575
576
577
578
579
 out:
	unlock_page(page);
	return err;
}

580
static void fuse_readpages_end(struct fuse_conn *fc, struct fuse_req *req)
581
{
582
	int i;
583
584
	size_t count = req->misc.read.in.size;
	size_t num_read = req->out.args[0].size;
585
	struct address_space *mapping = NULL;
586

587
588
	for (i = 0; mapping == NULL && i < req->num_pages; i++)
		mapping = req->pages[i]->mapping;
589

590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
	if (mapping) {
		struct inode *inode = mapping->host;

		/*
		 * Short read means EOF. If file size is larger, truncate it
		 */
		if (!req->out.h.error && num_read < count) {
			loff_t pos;

			pos = page_offset(req->pages[0]) + num_read;
			fuse_read_update_size(inode, pos,
					      req->misc.read.attr_ver);
		}
		fuse_invalidate_attr(inode); /* atime changed */
	}
605

606
607
608
609
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (!req->out.h.error)
			SetPageUptodate(page);
610
611
		else
			SetPageError(page);
612
		unlock_page(page);
613
		page_cache_release(page);
614
	}
615
	if (req->ff)
616
		fuse_file_put(req->ff, false);
617
618
}

619
static void fuse_send_readpages(struct fuse_req *req, struct file *file)
620
{
621
622
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
623
624
	loff_t pos = page_offset(req->pages[0]);
	size_t count = req->num_pages << PAGE_CACHE_SHIFT;
625
626

	req->out.argpages = 1;
627
	req->out.page_zeroing = 1;
628
	req->out.page_replace = 1;
629
	fuse_read_fill(req, file, pos, count, FUSE_READ);
630
	req->misc.read.attr_ver = fuse_get_attr_version(fc);
631
	if (fc->async_read) {
632
		req->ff = fuse_file_get(ff);
633
		req->end = fuse_readpages_end;
634
		fuse_request_send_background(fc, req);
635
	} else {
636
		fuse_request_send(fc, req);
637
		fuse_readpages_end(fc, req);
638
		fuse_put_request(fc, req);
639
	}
640
641
}

642
struct fuse_fill_data {
643
	struct fuse_req *req;
644
	struct file *file;
645
	struct inode *inode;
Maxim Patlasov's avatar
Maxim Patlasov committed
646
	unsigned nr_pages;
647
648
649
650
};

static int fuse_readpages_fill(void *_data, struct page *page)
{
651
	struct fuse_fill_data *data = _data;
652
653
654
655
	struct fuse_req *req = data->req;
	struct inode *inode = data->inode;
	struct fuse_conn *fc = get_fuse_conn(inode);

Miklos Szeredi's avatar
Miklos Szeredi committed
656
657
	fuse_wait_on_page_writeback(inode, page->index);

658
659
660
661
	if (req->num_pages &&
	    (req->num_pages == FUSE_MAX_PAGES_PER_REQ ||
	     (req->num_pages + 1) * PAGE_CACHE_SIZE > fc->max_read ||
	     req->pages[req->num_pages - 1]->index + 1 != page->index)) {
Maxim Patlasov's avatar
Maxim Patlasov committed
662
663
		int nr_alloc = min_t(unsigned, data->nr_pages,
				     FUSE_MAX_PAGES_PER_REQ);
664
		fuse_send_readpages(req, data->file);
Maxim Patlasov's avatar
Maxim Patlasov committed
665
		data->req = req = fuse_get_req(fc, nr_alloc);
666
		if (IS_ERR(req)) {
667
			unlock_page(page);
668
			return PTR_ERR(req);
669
670
		}
	}
Maxim Patlasov's avatar
Maxim Patlasov committed
671
672
673
674
675
676

	if (WARN_ON(req->num_pages >= req->max_pages)) {
		fuse_put_request(fc, req);
		return -EIO;
	}

677
	page_cache_get(page);
678
	req->pages[req->num_pages] = page;
679
	req->page_descs[req->num_pages].length = PAGE_SIZE;
Miklos Szeredi's avatar
Miklos Szeredi committed
680
	req->num_pages++;
Maxim Patlasov's avatar
Maxim Patlasov committed
681
	data->nr_pages--;
682
683
684
685
686
687
688
689
	return 0;
}

static int fuse_readpages(struct file *file, struct address_space *mapping,
			  struct list_head *pages, unsigned nr_pages)
{
	struct inode *inode = mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
690
	struct fuse_fill_data data;
691
	int err;
Maxim Patlasov's avatar
Maxim Patlasov committed
692
	int nr_alloc = min_t(unsigned, nr_pages, FUSE_MAX_PAGES_PER_REQ);
693

694
	err = -EIO;
695
	if (is_bad_inode(inode))
696
		goto out;
697

698
	data.file = file;
699
	data.inode = inode;
Maxim Patlasov's avatar
Maxim Patlasov committed
700
701
	data.req = fuse_get_req(fc, nr_alloc);
	data.nr_pages = nr_pages;
702
	err = PTR_ERR(data.req);
703
	if (IS_ERR(data.req))
704
		goto out;
705
706

	err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
707
708
	if (!err) {
		if (data.req->num_pages)
709
			fuse_send_readpages(data.req, file);
710
711
712
		else
			fuse_put_request(fc, data.req);
	}
713
out:
714
	return err;
715
716
}

Miklos Szeredi's avatar
Miklos Szeredi committed
717
718
719
720
static ssize_t fuse_file_aio_read(struct kiocb *iocb, const struct iovec *iov,
				  unsigned long nr_segs, loff_t pos)
{
	struct inode *inode = iocb->ki_filp->f_mapping->host;
721
	struct fuse_conn *fc = get_fuse_conn(inode);
Miklos Szeredi's avatar
Miklos Szeredi committed
722

723
724
725
726
727
728
729
	/*
	 * In auto invalidate mode, always update attributes on read.
	 * Otherwise, only update if we attempt to read past EOF (to ensure
	 * i_size is up to date).
	 */
	if (fc->auto_inval_data ||
	    (pos + iov_length(iov, nr_segs) > i_size_read(inode))) {
Miklos Szeredi's avatar
Miklos Szeredi committed
730
731
732
733
734
735
736
737
738
		int err;
		err = fuse_update_attributes(inode, NULL, iocb->ki_filp, NULL);
		if (err)
			return err;
	}

	return generic_file_aio_read(iocb, iov, nr_segs, pos);
}

739
static void fuse_write_fill(struct fuse_req *req, struct fuse_file *ff,
740
			    loff_t pos, size_t count)
741
{
742
743
	struct fuse_write_in *inarg = &req->misc.write.in;
	struct fuse_write_out *outarg = &req->misc.write.out;
744

745
746
747
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
748
	req->in.h.opcode = FUSE_WRITE;
749
	req->in.h.nodeid = ff->nodeid;
750
	req->in.numargs = 2;
751
	if (ff->fc->minor < 9)
752
753
754
		req->in.args[0].size = FUSE_COMPAT_WRITE_IN_SIZE;
	else
		req->in.args[0].size = sizeof(struct fuse_write_in);
755
	req->in.args[0].value = inarg;
756
757
758
	req->in.args[1].size = count;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(struct fuse_write_out);
759
760
761
762
	req->out.args[0].value = outarg;
}

static size_t fuse_send_write(struct fuse_req *req, struct file *file,
763
			      loff_t pos, size_t count, fl_owner_t owner)
764
{
765
766
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
767
768
	struct fuse_write_in *inarg = &req->misc.write.in;

769
	fuse_write_fill(req, ff, pos, count);
770
	inarg->flags = file->f_flags;
771
772
773
774
	if (owner != NULL) {
		inarg->write_flags |= FUSE_WRITE_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
775
	fuse_request_send(fc, req);
776
	return req->misc.write.out.size;
777
778
}

Miklos Szeredi's avatar
Miklos Szeredi committed
779
void fuse_write_update_size(struct inode *inode, loff_t pos)
780
781
782
783
784
785
786
787
788
789
790
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);

	spin_lock(&fc->lock);
	fi->attr_version = ++fc->attr_version;
	if (pos > inode->i_size)
		i_size_write(inode, pos);
	spin_unlock(&fc->lock);
}

Nick Piggin's avatar
Nick Piggin committed
791
792
793
794
795
796
797
798
799
800
801
static size_t fuse_send_write_pages(struct fuse_req *req, struct file *file,
				    struct inode *inode, loff_t pos,
				    size_t count)
{
	size_t res;
	unsigned offset;
	unsigned i;

	for (i = 0; i < req->num_pages; i++)
		fuse_wait_on_page_writeback(inode, req->pages[i]->index);

802
	res = fuse_send_write(req, file, pos, count, NULL);
Nick Piggin's avatar
Nick Piggin committed
803

804
	offset = req->page_descs[0].offset;
Nick Piggin's avatar
Nick Piggin committed
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
	count = res;
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];

		if (!req->out.h.error && !offset && count >= PAGE_CACHE_SIZE)
			SetPageUptodate(page);

		if (count > PAGE_CACHE_SIZE - offset)
			count -= PAGE_CACHE_SIZE - offset;
		else
			count = 0;
		offset = 0;

		unlock_page(page);
		page_cache_release(page);
	}

	return res;
}

static ssize_t fuse_fill_write_pages(struct fuse_req *req,
			       struct address_space *mapping,
			       struct iov_iter *ii, loff_t pos)
{
	struct fuse_conn *fc = get_fuse_conn(mapping->host);
	unsigned offset = pos & (PAGE_CACHE_SIZE - 1);
	size_t count = 0;
	int err;

834
	req->in.argpages = 1;
835
	req->page_descs[0].offset = offset;
Nick Piggin's avatar
Nick Piggin committed
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851

	do {
		size_t tmp;
		struct page *page;
		pgoff_t index = pos >> PAGE_CACHE_SHIFT;
		size_t bytes = min_t(size_t, PAGE_CACHE_SIZE - offset,
				     iov_iter_count(ii));

		bytes = min_t(size_t, bytes, fc->max_write - count);

 again:
		err = -EFAULT;
		if (iov_iter_fault_in_readable(ii, bytes))
			break;

		err = -ENOMEM;
852
		page = grab_cache_page_write_begin(mapping, index, 0);
Nick Piggin's avatar
Nick Piggin committed
853
854
855
		if (!page)
			break;

856
857
858
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

Nick Piggin's avatar
Nick Piggin committed
859
860
861
862
863
		pagefault_disable();
		tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
		pagefault_enable();
		flush_dcache_page(page);

864
865
		mark_page_accessed(page);

Nick Piggin's avatar
Nick Piggin committed
866
867
868
869
870
871
872
873
874
		if (!tmp) {
			unlock_page(page);
			page_cache_release(page);
			bytes = min(bytes, iov_iter_single_seg_count(ii));
			goto again;
		}

		err = 0;
		req->pages[req->num_pages] = page;
875
		req->page_descs[req->num_pages].length = tmp;
Nick Piggin's avatar
Nick Piggin committed
876
877
878
879
880
881
882
883
884
		req->num_pages++;

		iov_iter_advance(ii, tmp);
		count += tmp;
		pos += tmp;
		offset += tmp;
		if (offset == PAGE_CACHE_SIZE)
			offset = 0;

885
886
		if (!fc->big_writes)
			break;
Nick Piggin's avatar
Nick Piggin committed
887
	} while (iov_iter_count(ii) && count < fc->max_write &&
888
		 req->num_pages < req->max_pages && offset == 0);
Nick Piggin's avatar
Nick Piggin committed
889
890
891
892

	return count > 0 ? count : err;
}

893
894
895
896
897
898
899
900
static inline unsigned fuse_wr_pages(loff_t pos, size_t len)
{
	return min_t(unsigned,
		     ((pos + len - 1) >> PAGE_CACHE_SHIFT) -
		     (pos >> PAGE_CACHE_SHIFT) + 1,
		     FUSE_MAX_PAGES_PER_REQ);
}

Nick Piggin's avatar
Nick Piggin committed
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
static ssize_t fuse_perform_write(struct file *file,
				  struct address_space *mapping,
				  struct iov_iter *ii, loff_t pos)
{
	struct inode *inode = mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
	int err = 0;
	ssize_t res = 0;

	if (is_bad_inode(inode))
		return -EIO;

	do {
		struct fuse_req *req;
		ssize_t count;
916
		unsigned nr_pages = fuse_wr_pages(pos, iov_iter_count(ii));
Nick Piggin's avatar
Nick Piggin committed
917

918
		req = fuse_get_req(fc, nr_pages);
Nick Piggin's avatar
Nick Piggin committed
919
920
921
922
923
924
925
926
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
		if (IS_ERR(req)) {
			err = PTR_ERR(req);
			break;
		}

		count = fuse_fill_write_pages(req, mapping, ii, pos);
		if (count <= 0) {
			err = count;
		} else {
			size_t num_written;

			num_written = fuse_send_write_pages(req, file, inode,
							    pos, count);
			err = req->out.h.error;
			if (!err) {
				res += num_written;
				pos += num_written;

				/* break out of the loop on short write */
				if (num_written != count)
					err = -EIO;
			}
		}
		fuse_put_request(fc, req);
	} while (!err && iov_iter_count(ii));

	if (res > 0)
		fuse_write_update_size(inode, pos);

	fuse_invalidate_attr(inode);

	return res > 0 ? res : err;
}

static ssize_t fuse_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
				   unsigned long nr_segs, loff_t pos)
{
	struct file *file = iocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
	size_t count = 0;
Anand Avati's avatar
Anand Avati committed
959
	size_t ocount = 0;
Nick Piggin's avatar
Nick Piggin committed
960
	ssize_t written = 0;
Anand Avati's avatar
Anand Avati committed
961
	ssize_t written_buffered = 0;
Nick Piggin's avatar
Nick Piggin committed
962
963
964
	struct inode *inode = mapping->host;
	ssize_t err;
	struct iov_iter i;
Anand Avati's avatar
Anand Avati committed
965
	loff_t endbyte = 0;
Nick Piggin's avatar
Nick Piggin committed
966
967
968

	WARN_ON(iocb->ki_pos != pos);

Anand Avati's avatar
Anand Avati committed
969
970
	ocount = 0;
	err = generic_segment_checks(iov, &nr_segs, &ocount, VERIFY_READ);
Nick Piggin's avatar
Nick Piggin committed
971
972
973
	if (err)
		return err;

Anand Avati's avatar
Anand Avati committed
974
	count = ocount;
Nick Piggin's avatar
Nick Piggin committed
975
976
977
978
979
980
981
982
983
984
985
986
	mutex_lock(&inode->i_mutex);

	/* We can write back this queue in page reclaim */
	current->backing_dev_info = mapping->backing_dev_info;

	err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode));
	if (err)
		goto out;

	if (count == 0)
		goto out;

987
	err = file_remove_suid(file);
Nick Piggin's avatar
Nick Piggin committed
988
989
990
	if (err)
		goto out;

991
992
993
	err = file_update_time(file);
	if (err)
		goto out;
Nick Piggin's avatar
Nick Piggin committed
994

Anand Avati's avatar
Anand Avati committed
995
996
997
998
999
1000
1001
1002
1003
	if (file->f_flags & O_DIRECT) {
		written = generic_file_direct_write(iocb, iov, &nr_segs,
						    pos, &iocb->ki_pos,
						    count, ocount);
		if (written < 0 || written == count)
			goto out;

		pos += written;
		count -= written;
Nick Piggin's avatar
Nick Piggin committed
1004

Anand Avati's avatar
Anand Avati committed
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
		iov_iter_init(&i, iov, nr_segs, count, written);
		written_buffered = fuse_perform_write(file, mapping, &i, pos);
		if (written_buffered < 0) {
			err = written_buffered;
			goto out;
		}
		endbyte = pos + written_buffered - 1;

		err = filemap_write_and_wait_range(file->f_mapping, pos,
						   endbyte);
		if (err)
			goto out;

		invalidate_mapping_pages(file->f_mapping,
					 pos >> PAGE_CACHE_SHIFT,
					 endbyte >> PAGE_CACHE_SHIFT);

		written += written_buffered;
		iocb->ki_pos = pos + written_buffered;
	} else {
		iov_iter_init(&i, iov, nr_segs, count, 0);
		written = fuse_perform_write(file, mapping, &i, pos);
		if (written >= 0)
			iocb->ki_pos = pos + written;
	}
Nick Piggin's avatar
Nick Piggin committed
1030
1031
1032
1033
1034
1035
1036
out:
	current->backing_dev_info = NULL;
	mutex_unlock(&inode->i_mutex);

	return written ? written : err;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
static void fuse_release_user_pages(struct fuse_req *req, int write)
{
	unsigned i;

	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (write)
			set_page_dirty_lock(page);
		put_page(page);
	}
}

1049
1050
static inline void fuse_page_descs_length_init(struct fuse_req *req,
		unsigned index, unsigned nr_pages)
1051
1052
1053
{
	int i;

1054
	for (i = index; i < index + nr_pages; i++)
1055
1056
1057
1058
		req->page_descs[i].length = PAGE_SIZE -
			req->page_descs[i].offset;
}

1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
static inline unsigned long fuse_get_user_addr(const struct iov_iter *ii)
{
	return (unsigned long)ii->iov->iov_base + ii->iov_offset;
}

static inline size_t fuse_get_frag_size(const struct iov_iter *ii,
					size_t max_size)
{
	return min(iov_iter_single_seg_count(ii), max_size);
}

1070
static int fuse_get_user_pages(struct fuse_req *req, struct iov_iter *ii,
1071
			       size_t *nbytesp, int write)
Miklos Szeredi's avatar
Miklos Szeredi committed
1072
{
1073
	size_t nbytes = 0;  /* # bytes already packed in req */
1074

1075
1076
	/* Special case for kernel I/O: can copy directly into the buffer */
	if (segment_eq(get_fs(), KERNEL_DS)) {
1077
1078
1079
		unsigned long user_addr = fuse_get_user_addr(ii);
		size_t frag_size = fuse_get_frag_size(ii, *nbytesp);

1080
1081
1082
1083
1084
		if (write)
			req->in.args[1].value = (void *) user_addr;
		else
			req->out.args[0].value = (void *) user_addr;

1085
1086
		iov_iter_advance(ii, frag_size);
		*nbytesp = frag_size;
1087
1088
		return 0;
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1089

1090
	while (nbytes < *nbytesp && req->num_pages < req->max_pages) {
1091
1092
1093
1094
1095
		unsigned npages;
		unsigned long user_addr = fuse_get_user_addr(ii);
		unsigned offset = user_addr & ~PAGE_MASK;
		size_t frag_size = fuse_get_frag_size(ii, *nbytesp - nbytes);
		int ret;
Miklos Szeredi's avatar
Miklos Szeredi committed
1096

1097
		unsigned n = req->max_pages - req->num_pages;
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
		frag_size = min_t(size_t, frag_size, n << PAGE_SHIFT);

		npages = (frag_size + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
		npages = clamp(npages, 1U, n);

		ret = get_user_pages_fast(user_addr, npages, !write,
					  &req->pages[req->num_pages]);
		if (ret < 0)
			return ret;

		npages = ret;
		frag_size = min_t(size_t, frag_size,
				  (npages << PAGE_SHIFT) - offset);
		iov_iter_advance(ii, frag_size);

		req->page_descs[req->num_pages].offset = offset;
		fuse_page_descs_length_init(req, req->num_pages, npages);

		req->num_pages += npages;
		req->page_descs[req->num_pages - 1].length -=
			(npages << PAGE_SHIFT) - offset - frag_size;

		nbytes += frag_size;
	}
1122
1123
1124
1125
1126
1127

	if (write)
		req->in.argpages = 1;
	else
		req->out.argpages = 1;

1128
	*nbytesp = nbytes;
1129

Miklos Szeredi's avatar
Miklos Szeredi committed
1130
1131
1132
	return 0;
}

1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
static inline int fuse_iter_npages(const struct iov_iter *ii_p)
{
	struct iov_iter ii = *ii_p;
	int npages = 0;

	while (iov_iter_count(&ii) && npages < FUSE_MAX_PAGES_PER_REQ) {
		unsigned long user_addr = fuse_get_user_addr(&ii);
		unsigned offset = user_addr & ~PAGE_MASK;
		size_t frag_size = iov_iter_single_seg_count(&ii);

		npages += (frag_size + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
		iov_iter_advance(&ii, frag_size);
	}

	return min(npages, FUSE_MAX_PAGES_PER_REQ);
}

1150
1151
1152
ssize_t fuse_direct_io(struct file *file, const struct iovec *iov,
		       unsigned long nr_segs, size_t count, loff_t *ppos,
		       int write)
Miklos Szeredi's avatar
Miklos Szeredi committed
1153
{
1154
1155
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
Miklos Szeredi's avatar
Miklos Szeredi committed
1156
1157
1158
	size_t nmax = write ? fc->max_write : fc->max_read;
	loff_t pos = *ppos;
	ssize_t res = 0;
1159
	struct fuse_req *req;
1160
1161
1162
	struct iov_iter ii;

	iov_iter_init(&ii, iov, nr_segs, count, 0);
1163

1164
	req = fuse_get_req(fc, fuse_iter_npages(&ii));
1165
1166
	if (IS_ERR(req))
		return PTR_ERR(req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1167
1168
1169

	while (count) {
		size_t nres;
1170
		fl_owner_t owner = current->files;
1171
		size_t nbytes = min(count, nmax);
1172
		int err = fuse_get_user_pages(req, &ii, &nbytes, write);
Miklos Szeredi's avatar
Miklos Szeredi committed
1173
1174
1175
1176
		if (err) {
			res = err;
			break;
		}
1177