file.c 52 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

18
static const struct file_operations fuse_direct_io_file_operations;
19

20
21
static int fuse_send_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
			  int opcode, struct fuse_open_out *outargp)
22
23
{
	struct fuse_open_in inarg;
24
25
26
	struct fuse_req *req;
	int err;

27
28
29
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
30
31

	memset(&inarg, 0, sizeof(inarg));
32
33
34
	inarg.flags = file->f_flags & ~(O_CREAT | O_EXCL | O_NOCTTY);
	if (!fc->atomic_o_trunc)
		inarg.flags &= ~O_TRUNC;
35
36
	req->in.h.opcode = opcode;
	req->in.h.nodeid = nodeid;
37
38
39
40
41
42
	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;
43
	fuse_request_send(fc, req);
44
45
46
47
48
49
	err = req->out.h.error;
	fuse_put_request(fc, req);

	return err;
}

Tejun Heo's avatar
Tejun Heo committed
50
struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
51
52
{
	struct fuse_file *ff;
Tejun Heo's avatar
Tejun Heo committed
53

54
	ff = kmalloc(sizeof(struct fuse_file), GFP_KERNEL);
Tejun Heo's avatar
Tejun Heo committed
55
56
57
	if (unlikely(!ff))
		return NULL;

58
	ff->fc = fc;
Tejun Heo's avatar
Tejun Heo committed
59
60
61
62
	ff->reserved_req = fuse_request_alloc();
	if (unlikely(!ff->reserved_req)) {
		kfree(ff);
		return NULL;
63
	}
Tejun Heo's avatar
Tejun Heo committed
64
65
66
67
68
69
70
71
72
73

	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);

74
75
76
77
78
	return ff;
}

void fuse_file_free(struct fuse_file *ff)
{
79
	fuse_request_free(ff->reserved_req);
80
81
82
	kfree(ff);
}

83
struct fuse_file *fuse_file_get(struct fuse_file *ff)
84
85
86
87
88
{
	atomic_inc(&ff->count);
	return ff;
}

89
90
91
92
93
94
95
96
97
98
99
100
101
102
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
103
104
static void fuse_release_end(struct fuse_conn *fc, struct fuse_req *req)
{
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
	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
120
121
}

122
static void fuse_file_put(struct fuse_file *ff, bool sync)
123
124
125
{
	if (atomic_dec_and_test(&ff->count)) {
		struct fuse_req *req = ff->reserved_req;
126

127
128
129
130
131
132
133
134
		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);
		}
135
136
137
138
		kfree(ff);
	}
}

139
140
int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
		 bool isdir)
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
{
	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;
}
167
EXPORT_SYMBOL_GPL(fuse_do_open);
168

169
void fuse_finish_open(struct inode *inode, struct file *file)
170
{
171
	struct fuse_file *ff = file->private_data;
172
	struct fuse_conn *fc = get_fuse_conn(inode);
173
174

	if (ff->open_flags & FOPEN_DIRECT_IO)
175
		file->f_op = &fuse_direct_io_file_operations;
176
	if (!(ff->open_flags & FOPEN_KEEP_CACHE))
Miklos Szeredi's avatar
Miklos Szeredi committed
177
		invalidate_inode_pages2(inode->i_mapping);
178
	if (ff->open_flags & FOPEN_NONSEEKABLE)
Tejun Heo's avatar
Tejun Heo committed
179
		nonseekable_open(inode, file);
180
181
182
183
184
185
186
187
188
	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);
	}
189
190
}

191
int fuse_open_common(struct inode *inode, struct file *file, bool isdir)
192
{
Tejun Heo's avatar
Tejun Heo committed
193
	struct fuse_conn *fc = get_fuse_conn(inode);
194
195
	int err;

Miklos Szeredi's avatar
Miklos Szeredi committed
196
197
198
199
	/* VFS checks this, but only _after_ ->open() */
	if (file->f_flags & O_DIRECT)
		return -EINVAL;

200
201
202
203
	err = generic_file_open(inode, file);
	if (err)
		return err;

204
	err = fuse_do_open(fc, get_node_id(inode), file, isdir);
205
	if (err)
206
		return err;
207

208
209
210
	fuse_finish_open(inode, file);

	return 0;
211
212
}

213
static void fuse_prepare_release(struct fuse_file *ff, int flags, int opcode)
214
{
215
	struct fuse_conn *fc = ff->fc;
216
	struct fuse_req *req = ff->reserved_req;
217
	struct fuse_release_in *inarg = &req->misc.release.in;
218

219
220
221
222
223
224
	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);

225
	wake_up_interruptible_all(&ff->poll_wait);
226

227
	inarg->fh = ff->fh;
228
	inarg->flags = flags;
229
	req->in.h.opcode = opcode;
230
	req->in.h.nodeid = ff->nodeid;
231
232
233
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_release_in);
	req->in.args[0].value = inarg;
234
235
}

236
void fuse_release_common(struct file *file, int opcode)
237
{
Tejun Heo's avatar
Tejun Heo committed
238
239
	struct fuse_file *ff;
	struct fuse_req *req;
240

Tejun Heo's avatar
Tejun Heo committed
241
242
	ff = file->private_data;
	if (unlikely(!ff))
243
		return;
Tejun Heo's avatar
Tejun Heo committed
244
245

	req = ff->reserved_req;
246
	fuse_prepare_release(ff, file->f_flags, opcode);
Tejun Heo's avatar
Tejun Heo committed
247
248

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

	/*
	 * Normally this will send the RELEASE request, however if
	 * some asynchronous READ or WRITE requests are outstanding,
	 * the sending will be delayed.
256
257
258
259
	 *
	 * 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
260
	 */
261
	fuse_file_put(ff, ff->fc->destroy_req != NULL);
262
263
}

264
265
static int fuse_open(struct inode *inode, struct file *file)
{
266
	return fuse_open_common(inode, file, false);
267
268
269
270
}

static int fuse_release(struct inode *inode, struct file *file)
{
271
272
273
274
275
276
277
278
279
280
281
282
283
284
	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);
285
}
286
EXPORT_SYMBOL_GPL(fuse_sync_release);
287

288
/*
289
290
 * Scramble the ID space with XTEA, so that the value of the files_struct
 * pointer is not exposed to userspace.
291
 */
292
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id)
293
{
294
295
296
297
298
299
300
301
302
303
304
305
306
307
	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);
308
309
}

Miklos Szeredi's avatar
Miklos Szeredi committed
310
311
312
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
/*
 * 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;
}

353
static int fuse_flush(struct file *file, fl_owner_t id)
354
{
355
	struct inode *inode = file->f_path.dentry->d_inode;
356
357
358
359
360
361
	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;

362
363
364
	if (is_bad_inode(inode))
		return -EIO;

365
366
367
	if (fc->no_flush)
		return 0;

368
	req = fuse_get_req_nofail(fc, file);
369
370
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
371
	inarg.lock_owner = fuse_lock_owner_id(fc, id);
372
373
374
375
376
	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;
377
	req->force = 1;
378
	fuse_request_send(fc, req);
379
380
381
382
383
384
385
386
387
	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
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
/*
 * 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);
}

403
int fuse_fsync_common(struct file *file, int datasync, int isdir)
404
{
405
	struct inode *inode = file->f_mapping->host;
406
407
408
409
410
411
	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;

412
413
414
	if (is_bad_inode(inode))
		return -EIO;

415
	if ((!isdir && fc->no_fsync) || (isdir && fc->no_fsyncdir))
416
417
		return 0;

Miklos Szeredi's avatar
Miklos Szeredi committed
418
419
420
421
422
423
424
425
426
427
428
	/*
	 * 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)
		return err;

	fuse_sync_writes(inode);

429
430
431
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
432
433
434
435

	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
	inarg.fsync_flags = datasync ? 1 : 0;
436
	req->in.h.opcode = isdir ? FUSE_FSYNCDIR : FUSE_FSYNC;
437
438
439
440
	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;
441
	fuse_request_send(fc, req);
442
443
444
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
445
446
447
448
		if (isdir)
			fc->no_fsyncdir = 1;
		else
			fc->no_fsync = 1;
449
450
451
452
453
		err = 0;
	}
	return err;
}

454
static int fuse_fsync(struct file *file, int datasync)
455
{
456
	return fuse_fsync_common(file, datasync, 0);
457
458
}

459
460
void fuse_read_fill(struct fuse_req *req, struct file *file, loff_t pos,
		    size_t count, int opcode)
461
{
462
	struct fuse_read_in *inarg = &req->misc.read.in;
463
	struct fuse_file *ff = file->private_data;
464

465
466
467
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
468
	inarg->flags = file->f_flags;
469
	req->in.h.opcode = opcode;
470
	req->in.h.nodeid = ff->nodeid;
471
472
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_read_in);
473
	req->in.args[0].value = inarg;
474
475
476
477
478
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = count;
}

479
static size_t fuse_send_read(struct fuse_req *req, struct file *file,
480
			     loff_t pos, size_t count, fl_owner_t owner)
481
{
482
483
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
484

485
	fuse_read_fill(req, file, pos, count, FUSE_READ);
486
	if (owner != NULL) {
487
		struct fuse_read_in *inarg = &req->misc.read.in;
488
489
490
491

		inarg->read_flags |= FUSE_READ_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
492
	fuse_request_send(fc, req);
493
	return req->out.args[0].size;
494
495
}

496
497
498
499
500
501
502
503
504
505
506
507
508
509
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);
}

510
511
512
513
static int fuse_readpage(struct file *file, struct page *page)
{
	struct inode *inode = page->mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
514
	struct fuse_req *req;
515
516
517
518
	size_t num_read;
	loff_t pos = page_offset(page);
	size_t count = PAGE_CACHE_SIZE;
	u64 attr_ver;
519
520
521
522
523
524
	int err;

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

Miklos Szeredi's avatar
Miklos Szeredi committed
525
526
527
528
529
530
531
	/*
	 * Page writeback can extend beyond the liftime of the
	 * page-cache page, so make sure we read a properly synced
	 * page.
	 */
	fuse_wait_on_page_writeback(inode, page->index);

532
533
534
	req = fuse_get_req(fc);
	err = PTR_ERR(req);
	if (IS_ERR(req))
535
536
		goto out;

537
538
	attr_ver = fuse_get_attr_version(fc);

539
	req->out.page_zeroing = 1;
540
	req->out.argpages = 1;
541
542
	req->num_pages = 1;
	req->pages[0] = page;
543
	num_read = fuse_send_read(req, file, pos, count, NULL);
544
545
	err = req->out.h.error;
	fuse_put_request(fc, req);
546
547
548
549
550
551
552
553

	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);

554
		SetPageUptodate(page);
555
556
	}

557
	fuse_invalidate_attr(inode); /* atime changed */
558
559
560
561
562
 out:
	unlock_page(page);
	return err;
}

563
static void fuse_readpages_end(struct fuse_conn *fc, struct fuse_req *req)
564
{
565
	int i;
566
567
	size_t count = req->misc.read.in.size;
	size_t num_read = req->out.args[0].size;
568
	struct address_space *mapping = NULL;
569

570
571
	for (i = 0; mapping == NULL && i < req->num_pages; i++)
		mapping = req->pages[i]->mapping;
572

573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
	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 */
	}
588

589
590
591
592
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (!req->out.h.error)
			SetPageUptodate(page);
593
594
		else
			SetPageError(page);
595
		unlock_page(page);
596
		page_cache_release(page);
597
	}
598
	if (req->ff)
599
		fuse_file_put(req->ff, false);
600
601
}

602
static void fuse_send_readpages(struct fuse_req *req, struct file *file)
603
{
604
605
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
606
607
	loff_t pos = page_offset(req->pages[0]);
	size_t count = req->num_pages << PAGE_CACHE_SHIFT;
608
609

	req->out.argpages = 1;
610
	req->out.page_zeroing = 1;
611
	req->out.page_replace = 1;
612
	fuse_read_fill(req, file, pos, count, FUSE_READ);
613
	req->misc.read.attr_ver = fuse_get_attr_version(fc);
614
	if (fc->async_read) {
615
		req->ff = fuse_file_get(ff);
616
		req->end = fuse_readpages_end;
617
		fuse_request_send_background(fc, req);
618
	} else {
619
		fuse_request_send(fc, req);
620
		fuse_readpages_end(fc, req);
621
		fuse_put_request(fc, req);
622
	}
623
624
}

625
struct fuse_fill_data {
626
	struct fuse_req *req;
627
	struct file *file;
628
629
630
631
632
	struct inode *inode;
};

static int fuse_readpages_fill(void *_data, struct page *page)
{
633
	struct fuse_fill_data *data = _data;
634
635
636
637
	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
638
639
	fuse_wait_on_page_writeback(inode, page->index);

640
641
642
643
	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)) {
644
		fuse_send_readpages(req, data->file);
645
646
		data->req = req = fuse_get_req(fc);
		if (IS_ERR(req)) {
647
			unlock_page(page);
648
			return PTR_ERR(req);
649
650
		}
	}
651
	page_cache_get(page);
652
	req->pages[req->num_pages] = page;
Miklos Szeredi's avatar
Miklos Szeredi committed
653
	req->num_pages++;
654
655
656
657
658
659
660
661
	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);
662
	struct fuse_fill_data data;
663
	int err;
664

665
	err = -EIO;
666
	if (is_bad_inode(inode))
667
		goto out;
668

669
	data.file = file;
670
	data.inode = inode;
671
	data.req = fuse_get_req(fc);
672
	err = PTR_ERR(data.req);
673
	if (IS_ERR(data.req))
674
		goto out;
675
676

	err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
677
678
	if (!err) {
		if (data.req->num_pages)
679
			fuse_send_readpages(data.req, file);
680
681
682
		else
			fuse_put_request(fc, data.req);
	}
683
out:
684
	return err;
685
686
}

Miklos Szeredi's avatar
Miklos Szeredi committed
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
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;

	if (pos + iov_length(iov, nr_segs) > i_size_read(inode)) {
		int err;
		/*
		 * If trying to read past EOF, make sure the i_size
		 * attribute is up-to-date.
		 */
		err = fuse_update_attributes(inode, NULL, iocb->ki_filp, NULL);
		if (err)
			return err;
	}

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

706
static void fuse_write_fill(struct fuse_req *req, struct fuse_file *ff,
707
			    loff_t pos, size_t count)
708
{
709
710
	struct fuse_write_in *inarg = &req->misc.write.in;
	struct fuse_write_out *outarg = &req->misc.write.out;
711

712
713
714
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
715
	req->in.h.opcode = FUSE_WRITE;
716
	req->in.h.nodeid = ff->nodeid;
717
	req->in.numargs = 2;
718
	if (ff->fc->minor < 9)
719
720
721
		req->in.args[0].size = FUSE_COMPAT_WRITE_IN_SIZE;
	else
		req->in.args[0].size = sizeof(struct fuse_write_in);
722
	req->in.args[0].value = inarg;
723
724
725
	req->in.args[1].size = count;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(struct fuse_write_out);
726
727
728
729
	req->out.args[0].value = outarg;
}

static size_t fuse_send_write(struct fuse_req *req, struct file *file,
730
			      loff_t pos, size_t count, fl_owner_t owner)
731
{
732
733
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
734
735
	struct fuse_write_in *inarg = &req->misc.write.in;

736
	fuse_write_fill(req, ff, pos, count);
737
	inarg->flags = file->f_flags;
738
739
740
741
	if (owner != NULL) {
		inarg->write_flags |= FUSE_WRITE_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
742
	fuse_request_send(fc, req);
743
	return req->misc.write.out.size;
744
745
}

Nick Piggin's avatar
Nick Piggin committed
746
747
748
static int fuse_write_begin(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned flags,
			struct page **pagep, void **fsdata)
749
{
Nick Piggin's avatar
Nick Piggin committed
750
751
	pgoff_t index = pos >> PAGE_CACHE_SHIFT;

752
	*pagep = grab_cache_page_write_begin(mapping, index, flags);
Nick Piggin's avatar
Nick Piggin committed
753
754
	if (!*pagep)
		return -ENOMEM;
755
756
757
	return 0;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
758
void fuse_write_update_size(struct inode *inode, loff_t pos)
759
760
761
762
763
764
765
766
767
768
769
{
	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
770
771
static int fuse_buffered_write(struct file *file, struct inode *inode,
			       loff_t pos, unsigned count, struct page *page)
772
773
{
	int err;
774
	size_t nres;
775
	struct fuse_conn *fc = get_fuse_conn(inode);
Nick Piggin's avatar
Nick Piggin committed
776
	unsigned offset = pos & (PAGE_CACHE_SIZE - 1);
777
778
779
780
781
	struct fuse_req *req;

	if (is_bad_inode(inode))
		return -EIO;

Miklos Szeredi's avatar
Miklos Szeredi committed
782
783
784
785
786
787
	/*
	 * Make sure writepages on the same page are not mixed up with
	 * plain writes.
	 */
	fuse_wait_on_page_writeback(inode, page->index);

788
789
790
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
791

792
	req->in.argpages = 1;
793
794
795
	req->num_pages = 1;
	req->pages[0] = page;
	req->page_offset = offset;
796
	nres = fuse_send_write(req, file, pos, count, NULL);
797
798
	err = req->out.h.error;
	fuse_put_request(fc, req);
Nick Piggin's avatar
Nick Piggin committed
799
	if (!err && !nres)
800
801
		err = -EIO;
	if (!err) {
Nick Piggin's avatar
Nick Piggin committed
802
		pos += nres;
803
		fuse_write_update_size(inode, pos);
Nick Piggin's avatar
Nick Piggin committed
804
		if (count == PAGE_CACHE_SIZE)
805
			SetPageUptodate(page);
806
807
	}
	fuse_invalidate_attr(inode);
Nick Piggin's avatar
Nick Piggin committed
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
	return err ? err : nres;
}

static int fuse_write_end(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned copied,
			struct page *page, void *fsdata)
{
	struct inode *inode = mapping->host;
	int res = 0;

	if (copied)
		res = fuse_buffered_write(file, inode, pos, copied, page);

	unlock_page(page);
	page_cache_release(page);
	return res;
824
825
}

Nick Piggin's avatar
Nick Piggin committed
826
827
828
829
830
831
832
833
834
835
836
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);

837
	res = fuse_send_write(req, file, pos, count, NULL);
Nick Piggin's avatar
Nick Piggin committed
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868

	offset = req->page_offset;
	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;

869
	req->in.argpages = 1;
Nick Piggin's avatar
Nick Piggin committed
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
	req->page_offset = offset;

	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;
887
		page = grab_cache_page_write_begin(mapping, index, 0);
Nick Piggin's avatar
Nick Piggin committed
888
889
890
		if (!page)
			break;

891
892
893
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

Nick Piggin's avatar
Nick Piggin committed
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
		pagefault_disable();
		tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
		pagefault_enable();
		flush_dcache_page(page);

		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;
		req->num_pages++;

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

917
918
		if (!fc->big_writes)
			break;
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
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
	} while (iov_iter_count(ii) && count < fc->max_write &&
		 req->num_pages < FUSE_MAX_PAGES_PER_REQ && offset == 0);

	return count > 0 ? count : err;
}

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;

		req = fuse_get_req(fc);
		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;
	ssize_t written = 0;
	struct inode *inode = mapping->host;
	ssize_t err;
	struct iov_iter i;

	WARN_ON(iocb->ki_pos != pos);

	err = generic_segment_checks(iov, &nr_segs, &count, VERIFY_READ);
	if (err)
		return err;

	mutex_lock(&inode->i_mutex);
	vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE);

	/* 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;

1006
	err = file_remove_suid(file);
Nick Piggin's avatar
Nick Piggin committed
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
	if (err)
		goto out;

	file_update_time(file);

	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;

out:
	current->backing_dev_info = NULL;
	mutex_unlock(&inode->i_mutex);

	return written ? written : err;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
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);
	}
}

static int fuse_get_user_pages(struct fuse_req *req, const char __user *buf,
1037
			       size_t *nbytesp, int write)
Miklos Szeredi's avatar
Miklos Szeredi committed
1038
{
1039
	size_t nbytes = *nbytesp;
Miklos Szeredi's avatar
Miklos Szeredi committed
1040
1041
1042
1043
	unsigned long user_addr = (unsigned long) buf;
	unsigned offset = user_addr & ~PAGE_MASK;
	int npages;

1044
1045
1046
1047
1048
1049
1050
1051
1052
	/* Special case for kernel I/O: can copy directly into the buffer */
	if (segment_eq(get_fs(), KERNEL_DS)) {
		if (write)
			req->in.args[1].value = (void *) user_addr;
		else
			req->out.args[0].value = (void *) user_addr;

		return 0;
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1053

1054
	nbytes = min_t(size_t, nbytes, FUSE_MAX_PAGES_PER_REQ << PAGE_SHIFT);
Miklos Szeredi's avatar
Miklos Szeredi committed
1055
	npages = (nbytes + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
1056
	npages = clamp(npages, 1, FUSE_MAX_PAGES_PER_REQ);
1057
	npages = get_user_pages_fast(user_addr, npages, !write, req->pages);
Miklos Szeredi's avatar
Miklos Szeredi committed
1058
1059
1060
1061
1062
	if (npages < 0)
		return npages;

	req->num_pages = npages;
	req->page_offset = offset;
1063
1064
1065
1066
1067
1068
1069
1070
1071

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

	nbytes = (req->num_pages << PAGE_SHIFT) - req->page_offset;
	*nbytesp = min(*nbytesp, nbytes);

Miklos Szeredi's avatar
Miklos Szeredi committed
1072
1073
1074
	return 0;
}

1075
1076
ssize_t fuse_direct_io(struct file *file, const char __user *buf,
		       size_t count, loff_t *ppos, int write)
Miklos Szeredi's avatar
Miklos Szeredi committed
1077
{
1078
1079
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
Miklos Szeredi's avatar
Miklos Szeredi committed
1080
1081
1082
	size_t nmax = write ? fc->max_write : fc->max_read;
	loff_t pos = *ppos;
	ssize_t res = 0;
1083
1084
	struct fuse_req *req;

1085
1086
1087
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1088
1089
1090

	while (count) {
		size_t nres;
1091
		fl_owner_t owner = current->files;
1092
1093
		size_t nbytes = min(count, nmax);
		int err = fuse_get_user_pages(req, buf, &nbytes, write);
Miklos Szeredi's avatar
Miklos Szeredi committed
1094
1095
1096
1097
		if (err) {
			res = err;
			break;
		}
1098

Miklos Szeredi's avatar
Miklos Szeredi committed
1099
		if (write)
1100
			nres = fuse_send_write(req, file, pos, nbytes, owner);
Miklos Szeredi's avatar
Miklos Szeredi committed
1101
		else
1102
1103
			nres = fuse_send_read(req, file, pos, nbytes, owner);

Miklos Szeredi's avatar
Miklos Szeredi committed
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
		fuse_release_user_pages(req, !write);
		if (req->out.h.error) {
			if (!res)
				res = req->out.h.error;
			break;
		} else if (nres > nbytes) {
			res = -EIO;
			break;
		}
		count -= nres;
		res += nres;
		pos += nres;
		buf += nres;
		if (nres != nbytes)
			break;
1119
1120
1121
1122
1123
1124
		if (count) {
			fuse_put_request(fc, req);
			req = fuse_get_req(fc);
			if (IS_ERR(req))
				break;
		}
Miklos Szeredi's avatar
Miklos Szeredi committed
1125
	}
1126
1127
	if (!IS_ERR(req))
		fuse_put_request(fc, req);
1128
	if (res > 0)
Miklos Szeredi's avatar
Miklos Szeredi committed
1129
1130
1131
1132
		*ppos = pos;

	return res;
}
1133
EXPORT_SYMBOL_GPL(fuse_direct_io);
Miklos Szeredi's avatar
Miklos Szeredi committed
1134
1135
1136
1137

static ssize_t fuse_direct_read(struct file *file, char __user *buf,
				     size_t count, loff_t *ppos)
{
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
	ssize_t res;
	struct inode *inode = file->f_path.dentry->d_inode;

	if (is_bad_inode(inode))
		return -EIO;

	res = fuse_direct_io(file, buf, count, ppos, 0);

	fuse_invalidate_attr(inode);

	return res;
Miklos Szeredi's avatar
Miklos Szeredi committed
1149
1150
1151
1152
1153
}

static ssize_t fuse_direct_write(struct file *file, const char __user *buf,
				 size_t count, loff_t *ppos)
{
1154
	struct inode *inode = file->f_path.dentry->d_inode;
Miklos Szeredi's avatar
Miklos Szeredi committed
1155
	ssize_t res;
1156
1157
1158
1159

	if (is_bad_inode(inode))
		return -EIO;

Miklos Szeredi's avatar
Miklos Szeredi committed
1160
	/* Don't allow parallel writes to the same file */
1161
	mutex_lock(&inode->i_mutex);
1162
	res = generic_write_checks(file, ppos, &count, 0);
1163
	if (!res) {
1164
		res = fuse_direct_io(file, buf, count, ppos, 1);
1165
1166
1167
		if (res > 0)
			fuse_write_update_size(inode, *ppos);
	}
1168
	mutex_unlock(&inode->i_mutex);
1169
1170
1171

	fuse_invalidate_attr(inode);

Miklos Szeredi's avatar
Miklos Szeredi committed
1172
1173
1174
	return res;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1175
static void fuse_writepage_free(struct fuse_conn *fc, struct fuse_req *req)
1176
{
Miklos Szeredi's avatar
Miklos Szeredi committed
1177
	__free_page(req->pages[0]);
1178
	fuse_file_put(req->ff, false);
Miklos Szeredi's avatar
Miklos Szeredi committed
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
}

static void fuse_writepage_finish(struct fuse_conn *fc, struct fuse_req *req)
{
	struct inode *inode = req->inode;
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct backing_dev_info *bdi = inode->i_mapping->backing_dev_info;

	list_del(&req->writepages_entry);
	dec_bdi_stat(bdi, BDI_WRITEBACK);
	dec_zone_page_state(req->pages[0], NR_WRITEBACK_TEMP);
	bdi_writeout_inc(bdi);
	wake_up(&fi->page_waitq);
}

/* Called under fc->lock, may release and reacquire it */
static void fuse_send_writepage(struct fuse_conn *fc, struct fuse_req *req)
Miklos Szeredi's avatar
Miklos Szeredi committed
1196
1197
__releases(fc->lock)
__acquires(fc->lock)
Miklos Szeredi's avatar
Miklos Szeredi committed
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
{
	struct fuse_inode *fi = get_fuse_inode(req->inode);
	loff_t size = i_size_read(req->inode);
	struct fuse_write_in *inarg = &req->misc.write.in;

	if (!fc->connected)
		goto out_free;

	if (inarg->offset + PAGE_CACHE_SIZE <= size) {
		inarg->size = PAGE_CACHE_SIZE;
	} else if (inarg->offset < size) {
		inarg->size = size & (PAGE_CACHE_SIZE - 1);
	} else {
		/* Got truncated off completely */
		goto out_free;
1213
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1214
1215
1216

	req->in.args[1].size = inarg->size;
	fi->writectr++;
1217
	fuse_request_send_background_locked(fc, req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1218
1219
1220
1221
1222
1223
	return;

 out_free:
	fuse_writepage_finish(fc, req);
	spin_unlock(&fc->lock);
	fuse_writepage_free(fc, req);
1224
	fuse_put_request(fc, req);