file.c 48.5 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

17
static const struct file_operations fuse_direct_io_file_operations;
18

19
20
static int fuse_send_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
			  int opcode, struct fuse_open_out *outargp)
21
22
{
	struct fuse_open_in inarg;
23
24
25
	struct fuse_req *req;
	int err;

26
27
28
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
29
30

	memset(&inarg, 0, sizeof(inarg));
31
32
33
	inarg.flags = file->f_flags & ~(O_CREAT | O_EXCL | O_NOCTTY);
	if (!fc->atomic_o_trunc)
		inarg.flags &= ~O_TRUNC;
34
35
	req->in.h.opcode = opcode;
	req->in.h.nodeid = nodeid;
36
37
38
39
40
41
	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;
42
	fuse_request_send(fc, req);
43
44
45
46
47
48
	err = req->out.h.error;
	fuse_put_request(fc, req);

	return err;
}

Tejun Heo's avatar
Tejun Heo committed
49
struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
50
51
{
	struct fuse_file *ff;
Tejun Heo's avatar
Tejun Heo committed
52

53
	ff = kmalloc(sizeof(struct fuse_file), GFP_KERNEL);
Tejun Heo's avatar
Tejun Heo committed
54
55
56
	if (unlikely(!ff))
		return NULL;

57
	ff->fc = fc;
Tejun Heo's avatar
Tejun Heo committed
58
59
60
61
	ff->reserved_req = fuse_request_alloc();
	if (unlikely(!ff->reserved_req)) {
		kfree(ff);
		return NULL;
62
	}
Tejun Heo's avatar
Tejun Heo committed
63
64
65
66
67
68
69
70
71
72

	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);

73
74
75
76
77
	return ff;
}

void fuse_file_free(struct fuse_file *ff)
{
78
	fuse_request_free(ff->reserved_req);
79
80
81
	kfree(ff);
}

82
struct fuse_file *fuse_file_get(struct fuse_file *ff)
83
84
85
86
87
{
	atomic_inc(&ff->count);
	return ff;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
88
89
static void fuse_release_end(struct fuse_conn *fc, struct fuse_req *req)
{
90
	path_put(&req->misc.release.path);
Miklos Szeredi's avatar
Miklos Szeredi committed
91
92
}

93
94
95
96
static void fuse_file_put(struct fuse_file *ff)
{
	if (atomic_dec_and_test(&ff->count)) {
		struct fuse_req *req = ff->reserved_req;
97

Miklos Szeredi's avatar
Miklos Szeredi committed
98
		req->end = fuse_release_end;
99
		fuse_request_send_background(ff->fc, req);
100
101
102
103
		kfree(ff);
	}
}

104
105
int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
		 bool isdir)
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
{
	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;
}
132
EXPORT_SYMBOL_GPL(fuse_do_open);
133

134
void fuse_finish_open(struct inode *inode, struct file *file)
135
{
136
	struct fuse_file *ff = file->private_data;
137
	struct fuse_conn *fc = get_fuse_conn(inode);
138
139

	if (ff->open_flags & FOPEN_DIRECT_IO)
140
		file->f_op = &fuse_direct_io_file_operations;
141
	if (!(ff->open_flags & FOPEN_KEEP_CACHE))
Miklos Szeredi's avatar
Miklos Szeredi committed
142
		invalidate_inode_pages2(inode->i_mapping);
143
	if (ff->open_flags & FOPEN_NONSEEKABLE)
Tejun Heo's avatar
Tejun Heo committed
144
		nonseekable_open(inode, file);
145
146
147
148
149
150
151
152
153
	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);
	}
154
155
}

156
int fuse_open_common(struct inode *inode, struct file *file, bool isdir)
157
{
Tejun Heo's avatar
Tejun Heo committed
158
	struct fuse_conn *fc = get_fuse_conn(inode);
159
160
	int err;

Miklos Szeredi's avatar
Miklos Szeredi committed
161
162
163
164
	/* VFS checks this, but only _after_ ->open() */
	if (file->f_flags & O_DIRECT)
		return -EINVAL;

165
166
167
168
	err = generic_file_open(inode, file);
	if (err)
		return err;

169
	err = fuse_do_open(fc, get_node_id(inode), file, isdir);
170
	if (err)
171
		return err;
172

173
174
175
	fuse_finish_open(inode, file);

	return 0;
176
177
}

178
static void fuse_prepare_release(struct fuse_file *ff, int flags, int opcode)
179
{
180
	struct fuse_conn *fc = ff->fc;
181
	struct fuse_req *req = ff->reserved_req;
182
	struct fuse_release_in *inarg = &req->misc.release.in;
183

184
185
186
187
188
189
190
191
	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);

	wake_up_interruptible_sync(&ff->poll_wait);

192
	inarg->fh = ff->fh;
193
	inarg->flags = flags;
194
	req->in.h.opcode = opcode;
195
	req->in.h.nodeid = ff->nodeid;
196
197
198
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_release_in);
	req->in.args[0].value = inarg;
199
200
}

201
void fuse_release_common(struct file *file, int opcode)
202
{
Tejun Heo's avatar
Tejun Heo committed
203
204
	struct fuse_file *ff;
	struct fuse_req *req;
205

Tejun Heo's avatar
Tejun Heo committed
206
207
	ff = file->private_data;
	if (unlikely(!ff))
208
		return;
Tejun Heo's avatar
Tejun Heo committed
209
210

	req = ff->reserved_req;
211
	fuse_prepare_release(ff, file->f_flags, opcode);
Tejun Heo's avatar
Tejun Heo committed
212
213

	/* Hold vfsmount and dentry until release is finished */
214
215
	path_get(&file->f_path);
	req->misc.release.path = file->f_path;
Tejun Heo's avatar
Tejun Heo committed
216
217
218
219
220
221
222

	/*
	 * Normally this will send the RELEASE request, however if
	 * some asynchronous READ or WRITE requests are outstanding,
	 * the sending will be delayed.
	 */
	fuse_file_put(ff);
223
224
}

225
226
static int fuse_open(struct inode *inode, struct file *file)
{
227
	return fuse_open_common(inode, file, false);
228
229
230
231
}

static int fuse_release(struct inode *inode, struct file *file)
{
232
233
234
235
236
237
238
239
240
241
242
243
244
245
	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);
246
}
247
EXPORT_SYMBOL_GPL(fuse_sync_release);
248

249
/*
250
251
 * Scramble the ID space with XTEA, so that the value of the files_struct
 * pointer is not exposed to userspace.
252
 */
253
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id)
254
{
255
256
257
258
259
260
261
262
263
264
265
266
267
268
	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);
269
270
}

Miklos Szeredi's avatar
Miklos Szeredi committed
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
/*
 * 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;
}

314
static int fuse_flush(struct file *file, fl_owner_t id)
315
{
316
	struct inode *inode = file->f_path.dentry->d_inode;
317
318
319
320
321
322
	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;

323
324
325
	if (is_bad_inode(inode))
		return -EIO;

326
327
328
	if (fc->no_flush)
		return 0;

329
	req = fuse_get_req_nofail(fc, file);
330
331
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
332
	inarg.lock_owner = fuse_lock_owner_id(fc, id);
333
334
335
336
337
	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;
338
	req->force = 1;
339
	fuse_request_send(fc, req);
340
341
342
343
344
345
346
347
348
	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
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
/*
 * 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);
}

364
int fuse_fsync_common(struct file *file, int datasync, int isdir)
365
{
366
	struct inode *inode = file->f_mapping->host;
367
368
369
370
371
372
	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;

373
374
375
	if (is_bad_inode(inode))
		return -EIO;

376
	if ((!isdir && fc->no_fsync) || (isdir && fc->no_fsyncdir))
377
378
		return 0;

Miklos Szeredi's avatar
Miklos Szeredi committed
379
380
381
382
383
384
385
386
387
388
389
	/*
	 * 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);

390
391
392
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
393
394
395
396

	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
	inarg.fsync_flags = datasync ? 1 : 0;
397
	req->in.h.opcode = isdir ? FUSE_FSYNCDIR : FUSE_FSYNC;
398
399
400
401
	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;
402
	fuse_request_send(fc, req);
403
404
405
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
406
407
408
409
		if (isdir)
			fc->no_fsyncdir = 1;
		else
			fc->no_fsync = 1;
410
411
412
413
414
		err = 0;
	}
	return err;
}

415
static int fuse_fsync(struct file *file, int datasync)
416
{
417
	return fuse_fsync_common(file, datasync, 0);
418
419
}

420
421
void fuse_read_fill(struct fuse_req *req, struct file *file, loff_t pos,
		    size_t count, int opcode)
422
{
423
	struct fuse_read_in *inarg = &req->misc.read.in;
424
	struct fuse_file *ff = file->private_data;
425

426
427
428
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
429
	inarg->flags = file->f_flags;
430
	req->in.h.opcode = opcode;
431
	req->in.h.nodeid = ff->nodeid;
432
433
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_read_in);
434
	req->in.args[0].value = inarg;
435
436
437
438
439
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = count;
}

440
static size_t fuse_send_read(struct fuse_req *req, struct file *file,
441
			     loff_t pos, size_t count, fl_owner_t owner)
442
{
443
444
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
445

446
	fuse_read_fill(req, file, pos, count, FUSE_READ);
447
	if (owner != NULL) {
448
		struct fuse_read_in *inarg = &req->misc.read.in;
449
450
451
452

		inarg->read_flags |= FUSE_READ_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
453
	fuse_request_send(fc, req);
454
	return req->out.args[0].size;
455
456
}

457
458
459
460
461
462
463
464
465
466
467
468
469
470
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);
}

471
472
473
474
static int fuse_readpage(struct file *file, struct page *page)
{
	struct inode *inode = page->mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
475
	struct fuse_req *req;
476
477
478
479
	size_t num_read;
	loff_t pos = page_offset(page);
	size_t count = PAGE_CACHE_SIZE;
	u64 attr_ver;
480
481
482
483
484
485
	int err;

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

Miklos Szeredi's avatar
Miklos Szeredi committed
486
487
488
489
490
491
492
	/*
	 * 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);

493
494
495
	req = fuse_get_req(fc);
	err = PTR_ERR(req);
	if (IS_ERR(req))
496
497
		goto out;

498
499
	attr_ver = fuse_get_attr_version(fc);

500
	req->out.page_zeroing = 1;
501
	req->out.argpages = 1;
502
503
	req->num_pages = 1;
	req->pages[0] = page;
504
	num_read = fuse_send_read(req, file, pos, count, NULL);
505
506
	err = req->out.h.error;
	fuse_put_request(fc, req);
507
508
509
510
511
512
513
514

	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);

515
		SetPageUptodate(page);
516
517
	}

518
	fuse_invalidate_attr(inode); /* atime changed */
519
520
521
522
523
 out:
	unlock_page(page);
	return err;
}

524
static void fuse_readpages_end(struct fuse_conn *fc, struct fuse_req *req)
525
{
526
	int i;
527
528
	size_t count = req->misc.read.in.size;
	size_t num_read = req->out.args[0].size;
529
	struct address_space *mapping = NULL;
530

531
532
	for (i = 0; mapping == NULL && i < req->num_pages; i++)
		mapping = req->pages[i]->mapping;
533

534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
	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 */
	}
549

550
551
552
553
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (!req->out.h.error)
			SetPageUptodate(page);
554
555
		else
			SetPageError(page);
556
		unlock_page(page);
557
		page_cache_release(page);
558
	}
559
560
	if (req->ff)
		fuse_file_put(req->ff);
561
562
}

563
static void fuse_send_readpages(struct fuse_req *req, struct file *file)
564
{
565
566
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
567
568
	loff_t pos = page_offset(req->pages[0]);
	size_t count = req->num_pages << PAGE_CACHE_SHIFT;
569
570

	req->out.argpages = 1;
571
	req->out.page_zeroing = 1;
572
	req->out.page_replace = 1;
573
	fuse_read_fill(req, file, pos, count, FUSE_READ);
574
	req->misc.read.attr_ver = fuse_get_attr_version(fc);
575
	if (fc->async_read) {
576
		req->ff = fuse_file_get(ff);
577
		req->end = fuse_readpages_end;
578
		fuse_request_send_background(fc, req);
579
	} else {
580
		fuse_request_send(fc, req);
581
		fuse_readpages_end(fc, req);
582
		fuse_put_request(fc, req);
583
	}
584
585
}

586
struct fuse_fill_data {
587
	struct fuse_req *req;
588
	struct file *file;
589
590
591
592
593
	struct inode *inode;
};

static int fuse_readpages_fill(void *_data, struct page *page)
{
594
	struct fuse_fill_data *data = _data;
595
596
597
598
	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
599
600
	fuse_wait_on_page_writeback(inode, page->index);

601
602
603
604
	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)) {
605
		fuse_send_readpages(req, data->file);
606
607
		data->req = req = fuse_get_req(fc);
		if (IS_ERR(req)) {
608
			unlock_page(page);
609
			return PTR_ERR(req);
610
611
		}
	}
612
	page_cache_get(page);
613
	req->pages[req->num_pages] = page;
Miklos Szeredi's avatar
Miklos Szeredi committed
614
	req->num_pages++;
615
616
617
618
619
620
621
622
	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);
623
	struct fuse_fill_data data;
624
	int err;
625

626
	err = -EIO;
627
	if (is_bad_inode(inode))
628
		goto out;
629

630
	data.file = file;
631
	data.inode = inode;
632
	data.req = fuse_get_req(fc);
633
	err = PTR_ERR(data.req);
634
	if (IS_ERR(data.req))
635
		goto out;
636
637

	err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
638
639
	if (!err) {
		if (data.req->num_pages)
640
			fuse_send_readpages(data.req, file);
641
642
643
		else
			fuse_put_request(fc, data.req);
	}
644
out:
645
	return err;
646
647
}

Miklos Szeredi's avatar
Miklos Szeredi committed
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
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);
}

667
static void fuse_write_fill(struct fuse_req *req, struct fuse_file *ff,
668
			    loff_t pos, size_t count)
669
{
670
671
	struct fuse_write_in *inarg = &req->misc.write.in;
	struct fuse_write_out *outarg = &req->misc.write.out;
672

673
674
675
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
676
	req->in.h.opcode = FUSE_WRITE;
677
	req->in.h.nodeid = ff->nodeid;
678
	req->in.numargs = 2;
679
	if (ff->fc->minor < 9)
680
681
682
		req->in.args[0].size = FUSE_COMPAT_WRITE_IN_SIZE;
	else
		req->in.args[0].size = sizeof(struct fuse_write_in);
683
	req->in.args[0].value = inarg;
684
685
686
	req->in.args[1].size = count;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(struct fuse_write_out);
687
688
689
690
	req->out.args[0].value = outarg;
}

static size_t fuse_send_write(struct fuse_req *req, struct file *file,
691
			      loff_t pos, size_t count, fl_owner_t owner)
692
{
693
694
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
695
696
	struct fuse_write_in *inarg = &req->misc.write.in;

697
	fuse_write_fill(req, ff, pos, count);
698
	inarg->flags = file->f_flags;
699
700
701
702
	if (owner != NULL) {
		inarg->write_flags |= FUSE_WRITE_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
703
	fuse_request_send(fc, req);
704
	return req->misc.write.out.size;
705
706
}

Nick Piggin's avatar
Nick Piggin committed
707
708
709
static int fuse_write_begin(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned flags,
			struct page **pagep, void **fsdata)
710
{
Nick Piggin's avatar
Nick Piggin committed
711
712
	pgoff_t index = pos >> PAGE_CACHE_SHIFT;

713
	*pagep = grab_cache_page_write_begin(mapping, index, flags);
Nick Piggin's avatar
Nick Piggin committed
714
715
	if (!*pagep)
		return -ENOMEM;
716
717
718
	return 0;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
719
void fuse_write_update_size(struct inode *inode, loff_t pos)
720
721
722
723
724
725
726
727
728
729
730
{
	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
731
732
static int fuse_buffered_write(struct file *file, struct inode *inode,
			       loff_t pos, unsigned count, struct page *page)
733
734
{
	int err;
735
	size_t nres;
736
	struct fuse_conn *fc = get_fuse_conn(inode);
Nick Piggin's avatar
Nick Piggin committed
737
	unsigned offset = pos & (PAGE_CACHE_SIZE - 1);
738
739
740
741
742
	struct fuse_req *req;

	if (is_bad_inode(inode))
		return -EIO;

Miklos Szeredi's avatar
Miklos Szeredi committed
743
744
745
746
747
748
	/*
	 * Make sure writepages on the same page are not mixed up with
	 * plain writes.
	 */
	fuse_wait_on_page_writeback(inode, page->index);

749
750
751
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
752

753
	req->in.argpages = 1;
754
755
756
	req->num_pages = 1;
	req->pages[0] = page;
	req->page_offset = offset;
757
	nres = fuse_send_write(req, file, pos, count, NULL);
758
759
	err = req->out.h.error;
	fuse_put_request(fc, req);
Nick Piggin's avatar
Nick Piggin committed
760
	if (!err && !nres)
761
762
		err = -EIO;
	if (!err) {
Nick Piggin's avatar
Nick Piggin committed
763
		pos += nres;
764
		fuse_write_update_size(inode, pos);
Nick Piggin's avatar
Nick Piggin committed
765
		if (count == PAGE_CACHE_SIZE)
766
			SetPageUptodate(page);
767
768
	}
	fuse_invalidate_attr(inode);
Nick Piggin's avatar
Nick Piggin committed
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
	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;
785
786
}

Nick Piggin's avatar
Nick Piggin committed
787
788
789
790
791
792
793
794
795
796
797
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);

798
	res = fuse_send_write(req, file, pos, count, NULL);
Nick Piggin's avatar
Nick Piggin committed
799
800
801
802
803
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

	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;

830
	req->in.argpages = 1;
Nick Piggin's avatar
Nick Piggin committed
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
	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;
848
		page = grab_cache_page_write_begin(mapping, index, 0);
Nick Piggin's avatar
Nick Piggin committed
849
850
851
		if (!page)
			break;

852
853
854
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

Nick Piggin's avatar
Nick Piggin committed
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
		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;

878
879
		if (!fc->big_writes)
			break;
Nick Piggin's avatar
Nick Piggin committed
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
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
958
959
960
961
962
963
964
965
966
	} 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;

967
	err = file_remove_suid(file);
Nick Piggin's avatar
Nick Piggin committed
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
	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
985
986
987
988
989
990
991
992
993
994
995
996
997
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,
998
			       size_t *nbytesp, int write)
Miklos Szeredi's avatar
Miklos Szeredi committed
999
{
1000
	size_t nbytes = *nbytesp;
Miklos Szeredi's avatar
Miklos Szeredi committed
1001
1002
1003
1004
	unsigned long user_addr = (unsigned long) buf;
	unsigned offset = user_addr & ~PAGE_MASK;
	int npages;

1005
1006
1007
1008
1009
1010
1011
1012
1013
	/* 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
1014

1015
	nbytes = min_t(size_t, nbytes, FUSE_MAX_PAGES_PER_REQ << PAGE_SHIFT);
Miklos Szeredi's avatar
Miklos Szeredi committed
1016
	npages = (nbytes + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
1017
	npages = clamp(npages, 1, FUSE_MAX_PAGES_PER_REQ);
1018
	npages = get_user_pages_fast(user_addr, npages, !write, req->pages);
Miklos Szeredi's avatar
Miklos Szeredi committed
1019
1020
1021
1022
1023
	if (npages < 0)
		return npages;

	req->num_pages = npages;
	req->page_offset = offset;
1024
1025
1026
1027
1028
1029
1030
1031
1032

	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
1033
1034
1035
	return 0;
}

1036
1037
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
1038
{
1039
1040
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
Miklos Szeredi's avatar
Miklos Szeredi committed
1041
1042
1043
	size_t nmax = write ? fc->max_write : fc->max_read;
	loff_t pos = *ppos;
	ssize_t res = 0;
1044
1045
	struct fuse_req *req;

1046
1047
1048
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1049
1050
1051

	while (count) {
		size_t nres;
1052
		fl_owner_t owner = current->files;
1053
1054
		size_t nbytes = min(count, nmax);
		int err = fuse_get_user_pages(req, buf, &nbytes, write);
Miklos Szeredi's avatar
Miklos Szeredi committed
1055
1056
1057
1058
		if (err) {
			res = err;
			break;
		}
1059

Miklos Szeredi's avatar
Miklos Szeredi committed
1060
		if (write)
1061
			nres = fuse_send_write(req, file, pos, nbytes, owner);
Miklos Szeredi's avatar
Miklos Szeredi committed
1062
		else
1063
1064
			nres = fuse_send_read(req, file, pos, nbytes, owner);

Miklos Szeredi's avatar
Miklos Szeredi committed
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
		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;
1080
1081
1082
1083
1084
1085
		if (count) {
			fuse_put_request(fc, req);
			req = fuse_get_req(fc);
			if (IS_ERR(req))
				break;
		}
Miklos Szeredi's avatar
Miklos Szeredi committed
1086
	}
1087
1088
	if (!IS_ERR(req))
		fuse_put_request(fc, req);
1089
	if (res > 0)
Miklos Szeredi's avatar
Miklos Szeredi committed
1090
1091
1092
1093
		*ppos = pos;

	return res;
}
1094
EXPORT_SYMBOL_GPL(fuse_direct_io);
Miklos Szeredi's avatar
Miklos Szeredi committed
1095
1096
1097
1098

static ssize_t fuse_direct_read(struct file *file, char __user *buf,
				     size_t count, loff_t *ppos)
{
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
	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
1110
1111
1112
1113
1114
}

static ssize_t fuse_direct_write(struct file *file, const char __user *buf,
				 size_t count, loff_t *ppos)
{
1115
	struct inode *inode = file->f_path.dentry->d_inode;
Miklos Szeredi's avatar
Miklos Szeredi committed
1116
	ssize_t res;
1117
1118
1119
1120

	if (is_bad_inode(inode))
		return -EIO;

Miklos Szeredi's avatar
Miklos Szeredi committed
1121
	/* Don't allow parallel writes to the same file */
1122
	mutex_lock(&inode->i_mutex);
1123
	res = generic_write_checks(file, ppos, &count, 0);
1124
	if (!res) {
1125
		res = fuse_direct_io(file, buf, count, ppos, 1);
1126
1127
1128
		if (res > 0)
			fuse_write_update_size(inode, *ppos);
	}
1129
	mutex_unlock(&inode->i_mutex);
1130
1131
1132

	fuse_invalidate_attr(inode);

Miklos Szeredi's avatar
Miklos Szeredi committed
1133
1134
1135
	return res;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1136
static void fuse_writepage_free(struct fuse_conn *fc, struct fuse_req *req)
1137
{
Miklos Szeredi's avatar
Miklos Szeredi committed
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
	__free_page(req->pages[0]);
	fuse_file_put(req->ff);
}

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
1157
1158
__releases(fc->lock)
__acquires(fc->lock)
Miklos Szeredi's avatar
Miklos Szeredi committed
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
{
	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;
1174
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1175
1176
1177

	req->in.args[1].size = inarg->size;
	fi->writectr++;
1178
	fuse_request_send_background_locked(fc, req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1179
1180
1181
1182
1183
1184
	return;

 out_free:
	fuse_writepage_finish(fc, req);
	spin_unlock(&fc->lock);
	fuse_writepage_free(fc, req);
1185
	fuse_put_request(fc, req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1186
	spin_lock(&fc->lock);
1187
1188
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1189
1190
1191
1192
1193
1194
1195
/*
 * If fi->writectr is positive (no truncate or fsync going on) send
 * all queued writepage requests.
 *
 * Called with fc->lock
 */
void fuse_flush_writepages(struct inode *inode)
Miklos Szeredi's avatar
Miklos Szeredi committed
1196
1197
__releases(fc->lock)
__acquires(fc->lock)
1198
{
Miklos Szeredi's avatar
Miklos Szeredi committed
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct fuse_req *req;

	while (fi->writectr >= 0 && !list_empty(&fi->queued_writes)) {
		req = list_entry(fi->queued_writes.next, struct fuse_req, list);
		list_del_init(&req->list);
		fuse_send_writepage(fc, req);
	}
}

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

	mapping_set_error(inode->i_mapping, req->out.h.error);
	spin_lock(&fc->lock);
	fi->writectr--;
	fuse_writepage_finish(fc, req);
	spin_unlock(&fc->lock);
	fuse_writepage_free(fc, req);
}

static int fuse_writepage_locked(struct page *page)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode = mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct fuse_req *req;
	struct fuse_file *ff;
	struct page *tmp_page;

	set_page_writeback(page