file.c 56.4 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

19
static const struct file_operations fuse_direct_io_file_operations;
20

21
22
static int fuse_send_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
			  int opcode, struct fuse_open_out *outargp)
23
24
{
	struct fuse_open_in inarg;
25
26
27
	struct fuse_req *req;
	int err;

28
	req = fuse_get_req_nopages(fc);
29
30
	if (IS_ERR(req))
		return PTR_ERR(req);
31
32

	memset(&inarg, 0, sizeof(inarg));
33
34
35
	inarg.flags = file->f_flags & ~(O_CREAT | O_EXCL | O_NOCTTY);
	if (!fc->atomic_o_trunc)
		inarg.flags &= ~O_TRUNC;
36
37
	req->in.h.opcode = opcode;
	req->in.h.nodeid = nodeid;
38
39
40
41
42
43
	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;
44
	fuse_request_send(fc, req);
45
46
47
48
49
50
	err = req->out.h.error;
	fuse_put_request(fc, req);

	return err;
}

Tejun Heo's avatar
Tejun Heo committed
51
struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
52
53
{
	struct fuse_file *ff;
Tejun Heo's avatar
Tejun Heo committed
54

55
	ff = kmalloc(sizeof(struct fuse_file), GFP_KERNEL);
Tejun Heo's avatar
Tejun Heo committed
56
57
58
	if (unlikely(!ff))
		return NULL;

59
	ff->fc = fc;
60
	ff->reserved_req = fuse_request_alloc(0);
Tejun Heo's avatar
Tejun Heo committed
61
62
63
	if (unlikely(!ff->reserved_req)) {
		kfree(ff);
		return NULL;
64
	}
Tejun Heo's avatar
Tejun Heo committed
65
66
67
68
69
70
71
72
73
74

	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);

75
76
77
78
79
	return ff;
}

void fuse_file_free(struct fuse_file *ff)
{
80
	fuse_request_free(ff->reserved_req);
81
82
83
	kfree(ff);
}

84
struct fuse_file *fuse_file_get(struct fuse_file *ff)
85
86
87
88
89
{
	atomic_inc(&ff->count);
	return ff;
}

90
91
92
93
94
95
96
97
98
99
100
101
102
103
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
104
105
static void fuse_release_end(struct fuse_conn *fc, struct fuse_req *req)
{
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
	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
121
122
}

123
static void fuse_file_put(struct fuse_file *ff, bool sync)
124
125
126
{
	if (atomic_dec_and_test(&ff->count)) {
		struct fuse_req *req = ff->reserved_req;
127

128
129
130
131
132
133
134
135
		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);
		}
136
137
138
139
		kfree(ff);
	}
}

140
141
int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
		 bool isdir)
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
{
	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;
}
168
EXPORT_SYMBOL_GPL(fuse_do_open);
169

170
void fuse_finish_open(struct inode *inode, struct file *file)
171
{
172
	struct fuse_file *ff = file->private_data;
173
	struct fuse_conn *fc = get_fuse_conn(inode);
174
175

	if (ff->open_flags & FOPEN_DIRECT_IO)
176
		file->f_op = &fuse_direct_io_file_operations;
177
	if (!(ff->open_flags & FOPEN_KEEP_CACHE))
Miklos Szeredi's avatar
Miklos Szeredi committed
178
		invalidate_inode_pages2(inode->i_mapping);
179
	if (ff->open_flags & FOPEN_NONSEEKABLE)
Tejun Heo's avatar
Tejun Heo committed
180
		nonseekable_open(inode, file);
181
182
183
184
185
186
187
188
189
	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);
	}
190
191
}

192
int fuse_open_common(struct inode *inode, struct file *file, bool isdir)
193
{
Tejun Heo's avatar
Tejun Heo committed
194
	struct fuse_conn *fc = get_fuse_conn(inode);
195
196
197
198
199
200
	int err;

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

201
	err = fuse_do_open(fc, get_node_id(inode), file, isdir);
202
	if (err)
203
		return err;
204

205
206
207
	fuse_finish_open(inode, file);

	return 0;
208
209
}

210
static void fuse_prepare_release(struct fuse_file *ff, int flags, int opcode)
211
{
212
	struct fuse_conn *fc = ff->fc;
213
	struct fuse_req *req = ff->reserved_req;
214
	struct fuse_release_in *inarg = &req->misc.release.in;
215

216
217
218
219
220
221
	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);

222
	wake_up_interruptible_all(&ff->poll_wait);
223

224
	inarg->fh = ff->fh;
225
	inarg->flags = flags;
226
	req->in.h.opcode = opcode;
227
	req->in.h.nodeid = ff->nodeid;
228
229
230
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_release_in);
	req->in.args[0].value = inarg;
231
232
}

233
void fuse_release_common(struct file *file, int opcode)
234
{
Tejun Heo's avatar
Tejun Heo committed
235
236
	struct fuse_file *ff;
	struct fuse_req *req;
237

Tejun Heo's avatar
Tejun Heo committed
238
239
	ff = file->private_data;
	if (unlikely(!ff))
240
		return;
Tejun Heo's avatar
Tejun Heo committed
241
242

	req = ff->reserved_req;
243
	fuse_prepare_release(ff, file->f_flags, opcode);
Tejun Heo's avatar
Tejun Heo committed
244

Miklos Szeredi's avatar
Miklos Szeredi committed
245
246
247
248
249
250
	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
251
	/* Hold vfsmount and dentry until release is finished */
252
253
	path_get(&file->f_path);
	req->misc.release.path = file->f_path;
Tejun Heo's avatar
Tejun Heo committed
254
255
256
257
258

	/*
	 * Normally this will send the RELEASE request, however if
	 * some asynchronous READ or WRITE requests are outstanding,
	 * the sending will be delayed.
259
260
261
262
	 *
	 * 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
263
	 */
264
	fuse_file_put(ff, ff->fc->destroy_req != NULL);
265
266
}

267
268
static int fuse_open(struct inode *inode, struct file *file)
{
269
	return fuse_open_common(inode, file, false);
270
271
272
273
}

static int fuse_release(struct inode *inode, struct file *file)
{
274
275
276
277
278
279
280
281
282
283
284
285
286
287
	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);
288
}
289
EXPORT_SYMBOL_GPL(fuse_sync_release);
290

291
/*
292
293
 * Scramble the ID space with XTEA, so that the value of the files_struct
 * pointer is not exposed to userspace.
294
 */
295
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id)
296
{
297
298
299
300
301
302
303
304
305
306
307
308
309
310
	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);
311
312
}

Miklos Szeredi's avatar
Miklos Szeredi committed
313
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
/*
 * 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;
}

356
static int fuse_flush(struct file *file, fl_owner_t id)
357
{
358
	struct inode *inode = file->f_path.dentry->d_inode;
359
360
361
362
363
364
	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;

365
366
367
	if (is_bad_inode(inode))
		return -EIO;

368
369
370
	if (fc->no_flush)
		return 0;

371
	req = fuse_get_req_nofail_nopages(fc, file);
372
373
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
374
	inarg.lock_owner = fuse_lock_owner_id(fc, id);
375
376
377
378
379
	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;
380
	req->force = 1;
381
	fuse_request_send(fc, req);
382
383
384
385
386
387
388
389
390
	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
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
/*
 * 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);
}

406
407
int fuse_fsync_common(struct file *file, loff_t start, loff_t end,
		      int datasync, int isdir)
408
{
409
	struct inode *inode = file->f_mapping->host;
410
411
412
413
414
415
	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;

416
417
418
	if (is_bad_inode(inode))
		return -EIO;

419
420
421
422
	err = filemap_write_and_wait_range(inode->i_mapping, start, end);
	if (err)
		return err;

423
	if ((!isdir && fc->no_fsync) || (isdir && fc->no_fsyncdir))
424
425
		return 0;

426
427
	mutex_lock(&inode->i_mutex);

Miklos Szeredi's avatar
Miklos Szeredi committed
428
429
430
431
432
433
434
	/*
	 * 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)
435
		goto out;
Miklos Szeredi's avatar
Miklos Szeredi committed
436
437
438

	fuse_sync_writes(inode);

439
	req = fuse_get_req_nopages(fc);
440
441
442
443
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		goto out;
	}
444
445
446
447

	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
	inarg.fsync_flags = datasync ? 1 : 0;
448
	req->in.h.opcode = isdir ? FUSE_FSYNCDIR : FUSE_FSYNC;
449
450
451
452
	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;
453
	fuse_request_send(fc, req);
454
455
456
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
457
458
459
460
		if (isdir)
			fc->no_fsyncdir = 1;
		else
			fc->no_fsync = 1;
461
462
		err = 0;
	}
463
464
out:
	mutex_unlock(&inode->i_mutex);
465
466
467
	return err;
}

468
469
static int fuse_fsync(struct file *file, loff_t start, loff_t end,
		      int datasync)
470
{
471
	return fuse_fsync_common(file, start, end, datasync, 0);
472
473
}

474
475
void fuse_read_fill(struct fuse_req *req, struct file *file, loff_t pos,
		    size_t count, int opcode)
476
{
477
	struct fuse_read_in *inarg = &req->misc.read.in;
478
	struct fuse_file *ff = file->private_data;
479

480
481
482
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
483
	inarg->flags = file->f_flags;
484
	req->in.h.opcode = opcode;
485
	req->in.h.nodeid = ff->nodeid;
486
487
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_read_in);
488
	req->in.args[0].value = inarg;
489
490
491
492
493
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = count;
}

494
static size_t fuse_send_read(struct fuse_req *req, struct file *file,
495
			     loff_t pos, size_t count, fl_owner_t owner)
496
{
497
498
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
499

500
	fuse_read_fill(req, file, pos, count, FUSE_READ);
501
	if (owner != NULL) {
502
		struct fuse_read_in *inarg = &req->misc.read.in;
503
504
505
506

		inarg->read_flags |= FUSE_READ_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
507
	fuse_request_send(fc, req);
508
	return req->out.args[0].size;
509
510
}

511
512
513
514
515
516
517
518
519
520
521
522
523
524
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);
}

525
526
527
528
static int fuse_readpage(struct file *file, struct page *page)
{
	struct inode *inode = page->mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
529
	struct fuse_req *req;
530
531
532
533
	size_t num_read;
	loff_t pos = page_offset(page);
	size_t count = PAGE_CACHE_SIZE;
	u64 attr_ver;
534
535
536
537
538
539
	int err;

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

Miklos Szeredi's avatar
Miklos Szeredi committed
540
	/*
Lucas De Marchi's avatar
Lucas De Marchi committed
541
	 * Page writeback can extend beyond the lifetime of the
Miklos Szeredi's avatar
Miklos Szeredi committed
542
543
544
545
546
	 * page-cache page, so make sure we read a properly synced
	 * page.
	 */
	fuse_wait_on_page_writeback(inode, page->index);

547
	req = fuse_get_req(fc, 1);
548
549
	err = PTR_ERR(req);
	if (IS_ERR(req))
550
551
		goto out;

552
553
	attr_ver = fuse_get_attr_version(fc);

554
	req->out.page_zeroing = 1;
555
	req->out.argpages = 1;
556
557
	req->num_pages = 1;
	req->pages[0] = page;
558
	req->page_descs[0].length = count;
559
	num_read = fuse_send_read(req, file, pos, count, NULL);
560
561
	err = req->out.h.error;
	fuse_put_request(fc, req);
562
563
564
565
566
567
568
569

	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);

570
		SetPageUptodate(page);
571
572
	}

573
	fuse_invalidate_attr(inode); /* atime changed */
574
575
576
577
578
 out:
	unlock_page(page);
	return err;
}

579
static void fuse_readpages_end(struct fuse_conn *fc, struct fuse_req *req)
580
{
581
	int i;
582
583
	size_t count = req->misc.read.in.size;
	size_t num_read = req->out.args[0].size;
584
	struct address_space *mapping = NULL;
585

586
587
	for (i = 0; mapping == NULL && i < req->num_pages; i++)
		mapping = req->pages[i]->mapping;
588

589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
	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 */
	}
604

605
606
607
608
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (!req->out.h.error)
			SetPageUptodate(page);
609
610
		else
			SetPageError(page);
611
		unlock_page(page);
612
		page_cache_release(page);
613
	}
614
	if (req->ff)
615
		fuse_file_put(req->ff, false);
616
617
}

618
static void fuse_send_readpages(struct fuse_req *req, struct file *file)
619
{
620
621
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
622
623
	loff_t pos = page_offset(req->pages[0]);
	size_t count = req->num_pages << PAGE_CACHE_SHIFT;
624
625

	req->out.argpages = 1;
626
	req->out.page_zeroing = 1;
627
	req->out.page_replace = 1;
628
	fuse_read_fill(req, file, pos, count, FUSE_READ);
629
	req->misc.read.attr_ver = fuse_get_attr_version(fc);
630
	if (fc->async_read) {
631
		req->ff = fuse_file_get(ff);
632
		req->end = fuse_readpages_end;
633
		fuse_request_send_background(fc, req);
634
	} else {
635
		fuse_request_send(fc, req);
636
		fuse_readpages_end(fc, req);
637
		fuse_put_request(fc, req);
638
	}
639
640
}

641
struct fuse_fill_data {
642
	struct fuse_req *req;
643
	struct file *file;
644
	struct inode *inode;
Maxim Patlasov's avatar
Maxim Patlasov committed
645
	unsigned nr_pages;
646
647
648
649
};

static int fuse_readpages_fill(void *_data, struct page *page)
{
650
	struct fuse_fill_data *data = _data;
651
652
653
654
	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
655
656
	fuse_wait_on_page_writeback(inode, page->index);

657
658
659
660
	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
661
662
		int nr_alloc = min_t(unsigned, data->nr_pages,
				     FUSE_MAX_PAGES_PER_REQ);
663
		fuse_send_readpages(req, data->file);
Maxim Patlasov's avatar
Maxim Patlasov committed
664
		data->req = req = fuse_get_req(fc, nr_alloc);
665
		if (IS_ERR(req)) {
666
			unlock_page(page);
667
			return PTR_ERR(req);
668
669
		}
	}
Maxim Patlasov's avatar
Maxim Patlasov committed
670
671
672
673
674
675

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

676
	page_cache_get(page);
677
	req->pages[req->num_pages] = page;
678
	req->page_descs[req->num_pages].length = PAGE_SIZE;
Miklos Szeredi's avatar
Miklos Szeredi committed
679
	req->num_pages++;
Maxim Patlasov's avatar
Maxim Patlasov committed
680
	data->nr_pages--;
681
682
683
684
685
686
687
688
	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);
689
	struct fuse_fill_data data;
690
	int err;
Maxim Patlasov's avatar
Maxim Patlasov committed
691
	int nr_alloc = min_t(unsigned, nr_pages, FUSE_MAX_PAGES_PER_REQ);
692

693
	err = -EIO;
694
	if (is_bad_inode(inode))
695
		goto out;
696

697
	data.file = file;
698
	data.inode = inode;
Maxim Patlasov's avatar
Maxim Patlasov committed
699
700
	data.req = fuse_get_req(fc, nr_alloc);
	data.nr_pages = nr_pages;
701
	err = PTR_ERR(data.req);
702
	if (IS_ERR(data.req))
703
		goto out;
704
705

	err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
706
707
	if (!err) {
		if (data.req->num_pages)
708
			fuse_send_readpages(data.req, file);
709
710
711
		else
			fuse_put_request(fc, data.req);
	}
712
out:
713
	return err;
714
715
}

Miklos Szeredi's avatar
Miklos Szeredi committed
716
717
718
719
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;
720
	struct fuse_conn *fc = get_fuse_conn(inode);
Miklos Szeredi's avatar
Miklos Szeredi committed
721

722
723
724
725
726
727
728
	/*
	 * 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
729
730
731
732
733
734
735
736
737
		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);
}

738
static void fuse_write_fill(struct fuse_req *req, struct fuse_file *ff,
739
			    loff_t pos, size_t count)
740
{
741
742
	struct fuse_write_in *inarg = &req->misc.write.in;
	struct fuse_write_out *outarg = &req->misc.write.out;
743

744
745
746
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
747
	req->in.h.opcode = FUSE_WRITE;
748
	req->in.h.nodeid = ff->nodeid;
749
	req->in.numargs = 2;
750
	if (ff->fc->minor < 9)
751
752
753
		req->in.args[0].size = FUSE_COMPAT_WRITE_IN_SIZE;
	else
		req->in.args[0].size = sizeof(struct fuse_write_in);
754
	req->in.args[0].value = inarg;
755
756
757
	req->in.args[1].size = count;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(struct fuse_write_out);
758
759
760
761
	req->out.args[0].value = outarg;
}

static size_t fuse_send_write(struct fuse_req *req, struct file *file,
762
			      loff_t pos, size_t count, fl_owner_t owner)
763
{
764
765
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
766
767
	struct fuse_write_in *inarg = &req->misc.write.in;

768
	fuse_write_fill(req, ff, pos, count);
769
	inarg->flags = file->f_flags;
770
771
772
773
	if (owner != NULL) {
		inarg->write_flags |= FUSE_WRITE_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
774
	fuse_request_send(fc, req);
775
	return req->misc.write.out.size;
776
777
}

Miklos Szeredi's avatar
Miklos Szeredi committed
778
void fuse_write_update_size(struct inode *inode, loff_t pos)
779
780
781
782
783
784
785
786
787
788
789
{
	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
790
791
792
793
794
795
796
797
798
799
800
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);

801
	res = fuse_send_write(req, file, pos, count, NULL);
Nick Piggin's avatar
Nick Piggin committed
802

803
	offset = req->page_descs[0].offset;
Nick Piggin's avatar
Nick Piggin committed
804
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
	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;

833
	req->in.argpages = 1;
834
	req->page_descs[0].offset = offset;
Nick Piggin's avatar
Nick Piggin committed
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850

	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;
851
		page = grab_cache_page_write_begin(mapping, index, 0);
Nick Piggin's avatar
Nick Piggin committed
852
853
854
		if (!page)
			break;

855
856
857
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

Nick Piggin's avatar
Nick Piggin committed
858
859
860
861
862
		pagefault_disable();
		tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
		pagefault_enable();
		flush_dcache_page(page);

863
864
		mark_page_accessed(page);

Nick Piggin's avatar
Nick Piggin committed
865
866
867
868
869
870
871
872
873
		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;
874
		req->page_descs[req->num_pages].length = tmp;
Nick Piggin's avatar
Nick Piggin committed
875
876
877
878
879
880
881
882
883
		req->num_pages++;

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

884
885
		if (!fc->big_writes)
			break;
Nick Piggin's avatar
Nick Piggin committed
886
	} while (iov_iter_count(ii) && count < fc->max_write &&
887
		 req->num_pages < req->max_pages && offset == 0);
Nick Piggin's avatar
Nick Piggin committed
888
889
890
891

	return count > 0 ? count : err;
}

892
893
894
895
896
897
898
899
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
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
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;
915
		unsigned nr_pages = fuse_wr_pages(pos, iov_iter_count(ii));
Nick Piggin's avatar
Nick Piggin committed
916

917
		req = fuse_get_req(fc, nr_pages);
Nick Piggin's avatar
Nick Piggin committed
918
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
		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
958
	size_t ocount = 0;
Nick Piggin's avatar
Nick Piggin committed
959
	ssize_t written = 0;
Anand Avati's avatar
Anand Avati committed
960
	ssize_t written_buffered = 0;
Nick Piggin's avatar
Nick Piggin committed
961
962
963
	struct inode *inode = mapping->host;
	ssize_t err;
	struct iov_iter i;
Anand Avati's avatar
Anand Avati committed
964
	loff_t endbyte = 0;
Nick Piggin's avatar
Nick Piggin committed
965
966
967

	WARN_ON(iocb->ki_pos != pos);

Anand Avati's avatar
Anand Avati committed
968
969
	ocount = 0;
	err = generic_segment_checks(iov, &nr_segs, &ocount, VERIFY_READ);
Nick Piggin's avatar
Nick Piggin committed
970
971
972
	if (err)
		return err;

Anand Avati's avatar
Anand Avati committed
973
	count = ocount;
974
	sb_start_write(inode->i_sb);
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
out:
	current->backing_dev_info = NULL;
	mutex_unlock(&inode->i_mutex);
1033
	sb_end_write(inode->i_sb);
Nick Piggin's avatar
Nick Piggin committed
1034
1035
1036
1037

	return written ? written : err;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
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);
	}
}

1050
1051
static inline void fuse_page_descs_length_init(struct fuse_req *req,
		unsigned index, unsigned nr_pages)
1052
1053
1054
{
	int i;

1055
	for (i = index; i < index + nr_pages; i++)
1056
1057
1058
1059
		req->page_descs[i].length = PAGE_SIZE -
			req->page_descs[i].offset;
}

1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
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);
}

1071
static int fuse_get_user_pages(struct fuse_req *req, struct iov_iter *ii,
1072
			       size_t *nbytesp, int write)
Miklos Szeredi's avatar
Miklos Szeredi committed
1073
{
1074
	size_t nbytes = 0;  /* # bytes already packed in req */
1075

1076
1077
	/* Special case for kernel I/O: can copy directly into the buffer */
	if (segment_eq(get_fs(), KERNEL_DS)) {
1078
1079
1080
		unsigned long user_addr = fuse_get_user_addr(ii);
		size_t frag_size = fuse_get_frag_size(ii, *nbytesp);

1081
1082
1083
1084
1085
		if (write)
			req->in.args[1].value = (void *) user_addr;
		else
			req->out.args[0].value = (void *) user_addr;

1086
1087
		iov_iter_advance(ii, frag_size);
		*nbytesp = frag_size;
1088
1089
		return 0;
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1090

1091
	while (nbytes < *nbytesp && req->num_pages < req->max_pages) {
1092
1093
1094
1095
1096
		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
1097

1098
		unsigned n = req->max_pages - req->num_pages;
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
		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;
	}
1123
1124
1125
1126
1127
1128

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

1129
	*nbytesp = nbytes;
1130

Miklos Szeredi's avatar
Miklos Szeredi committed
1131
1132
1133
	return 0;
}

1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
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);
}

1151
1152
1153
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
1154
{
1155
1156
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
Miklos Szeredi's avatar
Miklos Szeredi committed
1157
1158
1159
	size_t nmax = write ? fc->max_write : fc->max_read;
	loff_t pos = *ppos;
	ssize_t res = 0;
1160
	struct fuse_req *req;
1161
1162
1163
	struct iov_iter ii;

	iov_iter_init(&ii, iov, nr_segs, count, 0);
1164

1165
	req = fuse_get_req(fc, fuse_iter_npages(&ii));
1166
1167
	if (IS_ERR(req))
		return PTR_ERR(req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1168
1169
1170

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