file.c 50.6 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

Miklos Szeredi's avatar
Miklos Szeredi committed
248
249
250
251
252
253
	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
254
	/* Hold vfsmount and dentry until release is finished */
255
256
	path_get(&file->f_path);
	req->misc.release.path = file->f_path;
Tejun Heo's avatar
Tejun Heo committed
257
258
259
260
261

	/*
	 * Normally this will send the RELEASE request, however if
	 * some asynchronous READ or WRITE requests are outstanding,
	 * the sending will be delayed.
262
263
264
265
	 *
	 * 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
266
	 */
267
	fuse_file_put(ff, ff->fc->destroy_req != NULL);
268
269
}

270
271
static int fuse_open(struct inode *inode, struct file *file)
{
272
	return fuse_open_common(inode, file, false);
273
274
275
276
}

static int fuse_release(struct inode *inode, struct file *file)
{
277
278
279
280
281
282
283
284
285
286
287
288
289
290
	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);
291
}
292
EXPORT_SYMBOL_GPL(fuse_sync_release);
293

294
/*
295
296
 * Scramble the ID space with XTEA, so that the value of the files_struct
 * pointer is not exposed to userspace.
297
 */
298
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id)
299
{
300
301
302
303
304
305
306
307
308
309
310
311
312
313
	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);
314
315
}

Miklos Szeredi's avatar
Miklos Szeredi committed
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
/*
 * 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;
}

359
static int fuse_flush(struct file *file, fl_owner_t id)
360
{
361
	struct inode *inode = file->f_path.dentry->d_inode;
362
363
364
365
366
367
	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;

368
369
370
	if (is_bad_inode(inode))
		return -EIO;

371
372
373
	if (fc->no_flush)
		return 0;

374
	req = fuse_get_req_nofail(fc, file);
375
376
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
377
	inarg.lock_owner = fuse_lock_owner_id(fc, id);
378
379
380
381
382
	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;
383
	req->force = 1;
384
	fuse_request_send(fc, req);
385
386
387
388
389
390
391
392
393
	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
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
/*
 * 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);
}

409
int fuse_fsync_common(struct file *file, int datasync, int isdir)
410
{
411
	struct inode *inode = file->f_mapping->host;
412
413
414
415
416
417
	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;

418
419
420
	if (is_bad_inode(inode))
		return -EIO;

421
	if ((!isdir && fc->no_fsync) || (isdir && fc->no_fsyncdir))
422
423
		return 0;

Miklos Szeredi's avatar
Miklos Szeredi committed
424
425
426
427
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)
		return err;

	fuse_sync_writes(inode);

435
436
437
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
438
439
440
441

	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
	inarg.fsync_flags = datasync ? 1 : 0;
442
	req->in.h.opcode = isdir ? FUSE_FSYNCDIR : FUSE_FSYNC;
443
444
445
446
	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;
447
	fuse_request_send(fc, req);
448
449
450
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
451
452
453
454
		if (isdir)
			fc->no_fsyncdir = 1;
		else
			fc->no_fsync = 1;
455
456
457
458
459
		err = 0;
	}
	return err;
}

460
static int fuse_fsync(struct file *file, int datasync)
461
{
462
	return fuse_fsync_common(file, datasync, 0);
463
464
}

465
466
void fuse_read_fill(struct fuse_req *req, struct file *file, loff_t pos,
		    size_t count, int opcode)
467
{
468
	struct fuse_read_in *inarg = &req->misc.read.in;
469
	struct fuse_file *ff = file->private_data;
470

471
472
473
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
474
	inarg->flags = file->f_flags;
475
	req->in.h.opcode = opcode;
476
	req->in.h.nodeid = ff->nodeid;
477
478
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_read_in);
479
	req->in.args[0].value = inarg;
480
481
482
483
484
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = count;
}

485
static size_t fuse_send_read(struct fuse_req *req, struct file *file,
486
			     loff_t pos, size_t count, fl_owner_t owner)
487
{
488
489
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
490

491
	fuse_read_fill(req, file, pos, count, FUSE_READ);
492
	if (owner != NULL) {
493
		struct fuse_read_in *inarg = &req->misc.read.in;
494
495
496
497

		inarg->read_flags |= FUSE_READ_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
498
	fuse_request_send(fc, req);
499
	return req->out.args[0].size;
500
501
}

502
503
504
505
506
507
508
509
510
511
512
513
514
515
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);
}

516
517
518
519
static int fuse_readpage(struct file *file, struct page *page)
{
	struct inode *inode = page->mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
520
	struct fuse_req *req;
521
522
523
524
	size_t num_read;
	loff_t pos = page_offset(page);
	size_t count = PAGE_CACHE_SIZE;
	u64 attr_ver;
525
526
527
528
529
530
	int err;

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

Miklos Szeredi's avatar
Miklos Szeredi committed
531
	/*
Lucas De Marchi's avatar
Lucas De Marchi committed
532
	 * Page writeback can extend beyond the lifetime of the
Miklos Szeredi's avatar
Miklos Szeredi committed
533
534
535
536
537
	 * page-cache page, so make sure we read a properly synced
	 * page.
	 */
	fuse_wait_on_page_writeback(inode, page->index);

538
539
540
	req = fuse_get_req(fc);
	err = PTR_ERR(req);
	if (IS_ERR(req))
541
542
		goto out;

543
544
	attr_ver = fuse_get_attr_version(fc);

545
	req->out.page_zeroing = 1;
546
	req->out.argpages = 1;
547
548
	req->num_pages = 1;
	req->pages[0] = page;
549
	num_read = fuse_send_read(req, file, pos, count, NULL);
550
551
	err = req->out.h.error;
	fuse_put_request(fc, req);
552
553
554
555
556
557
558
559

	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);

560
		SetPageUptodate(page);
561
562
	}

563
	fuse_invalidate_attr(inode); /* atime changed */
564
565
566
567
568
 out:
	unlock_page(page);
	return err;
}

569
static void fuse_readpages_end(struct fuse_conn *fc, struct fuse_req *req)
570
{
571
	int i;
572
573
	size_t count = req->misc.read.in.size;
	size_t num_read = req->out.args[0].size;
574
	struct address_space *mapping = NULL;
575

576
577
	for (i = 0; mapping == NULL && i < req->num_pages; i++)
		mapping = req->pages[i]->mapping;
578

579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
	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 */
	}
594

595
596
597
598
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (!req->out.h.error)
			SetPageUptodate(page);
599
600
		else
			SetPageError(page);
601
		unlock_page(page);
602
		page_cache_release(page);
603
	}
604
	if (req->ff)
605
		fuse_file_put(req->ff, false);
606
607
}

608
static void fuse_send_readpages(struct fuse_req *req, struct file *file)
609
{
610
611
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
612
613
	loff_t pos = page_offset(req->pages[0]);
	size_t count = req->num_pages << PAGE_CACHE_SHIFT;
614
615

	req->out.argpages = 1;
616
	req->out.page_zeroing = 1;
617
	req->out.page_replace = 1;
618
	fuse_read_fill(req, file, pos, count, FUSE_READ);
619
	req->misc.read.attr_ver = fuse_get_attr_version(fc);
620
	if (fc->async_read) {
621
		req->ff = fuse_file_get(ff);
622
		req->end = fuse_readpages_end;
623
		fuse_request_send_background(fc, req);
624
	} else {
625
		fuse_request_send(fc, req);
626
		fuse_readpages_end(fc, req);
627
		fuse_put_request(fc, req);
628
	}
629
630
}

631
struct fuse_fill_data {
632
	struct fuse_req *req;
633
	struct file *file;
634
635
636
637
638
	struct inode *inode;
};

static int fuse_readpages_fill(void *_data, struct page *page)
{
639
	struct fuse_fill_data *data = _data;
640
641
642
643
	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
644
645
	fuse_wait_on_page_writeback(inode, page->index);

646
647
648
649
	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)) {
650
		fuse_send_readpages(req, data->file);
651
652
		data->req = req = fuse_get_req(fc);
		if (IS_ERR(req)) {
653
			unlock_page(page);
654
			return PTR_ERR(req);
655
656
		}
	}
657
	page_cache_get(page);
658
	req->pages[req->num_pages] = page;
Miklos Szeredi's avatar
Miklos Szeredi committed
659
	req->num_pages++;
660
661
662
663
664
665
666
667
	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);
668
	struct fuse_fill_data data;
669
	int err;
670

671
	err = -EIO;
672
	if (is_bad_inode(inode))
673
		goto out;
674

675
	data.file = file;
676
	data.inode = inode;
677
	data.req = fuse_get_req(fc);
678
	err = PTR_ERR(data.req);
679
	if (IS_ERR(data.req))
680
		goto out;
681
682

	err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
683
684
	if (!err) {
		if (data.req->num_pages)
685
			fuse_send_readpages(data.req, file);
686
687
688
		else
			fuse_put_request(fc, data.req);
	}
689
out:
690
	return err;
691
692
}

Miklos Szeredi's avatar
Miklos Szeredi committed
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
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);
}

712
static void fuse_write_fill(struct fuse_req *req, struct fuse_file *ff,
713
			    loff_t pos, size_t count)
714
{
715
716
	struct fuse_write_in *inarg = &req->misc.write.in;
	struct fuse_write_out *outarg = &req->misc.write.out;
717

718
719
720
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
721
	req->in.h.opcode = FUSE_WRITE;
722
	req->in.h.nodeid = ff->nodeid;
723
	req->in.numargs = 2;
724
	if (ff->fc->minor < 9)
725
726
727
		req->in.args[0].size = FUSE_COMPAT_WRITE_IN_SIZE;
	else
		req->in.args[0].size = sizeof(struct fuse_write_in);
728
	req->in.args[0].value = inarg;
729
730
731
	req->in.args[1].size = count;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(struct fuse_write_out);
732
733
734
735
	req->out.args[0].value = outarg;
}

static size_t fuse_send_write(struct fuse_req *req, struct file *file,
736
			      loff_t pos, size_t count, fl_owner_t owner)
737
{
738
739
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
740
741
	struct fuse_write_in *inarg = &req->misc.write.in;

742
	fuse_write_fill(req, ff, pos, count);
743
	inarg->flags = file->f_flags;
744
745
746
747
	if (owner != NULL) {
		inarg->write_flags |= FUSE_WRITE_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
748
	fuse_request_send(fc, req);
749
	return req->misc.write.out.size;
750
751
}

Miklos Szeredi's avatar
Miklos Szeredi committed
752
void fuse_write_update_size(struct inode *inode, loff_t pos)
753
754
755
756
757
758
759
760
761
762
763
{
	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
764
765
766
767
768
769
770
771
772
773
774
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);

775
	res = fuse_send_write(req, file, pos, count, NULL);
Nick Piggin's avatar
Nick Piggin committed
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806

	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;

807
	req->in.argpages = 1;
Nick Piggin's avatar
Nick Piggin committed
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
	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;
825
		page = grab_cache_page_write_begin(mapping, index, 0);
Nick Piggin's avatar
Nick Piggin committed
826
827
828
		if (!page)
			break;

829
830
831
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

Nick Piggin's avatar
Nick Piggin committed
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
		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;

855
856
		if (!fc->big_writes)
			break;
Nick Piggin's avatar
Nick Piggin committed
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
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
	} 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;

944
	err = file_remove_suid(file);
Nick Piggin's avatar
Nick Piggin committed
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
	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
962
963
964
965
966
967
968
969
970
971
972
973
974
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,
975
			       size_t *nbytesp, int write)
Miklos Szeredi's avatar
Miklos Szeredi committed
976
{
977
	size_t nbytes = *nbytesp;
Miklos Szeredi's avatar
Miklos Szeredi committed
978
979
980
981
	unsigned long user_addr = (unsigned long) buf;
	unsigned offset = user_addr & ~PAGE_MASK;
	int npages;

982
983
984
985
986
987
988
989
990
	/* 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
991

992
	nbytes = min_t(size_t, nbytes, FUSE_MAX_PAGES_PER_REQ << PAGE_SHIFT);
Miklos Szeredi's avatar
Miklos Szeredi committed
993
	npages = (nbytes + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
994
	npages = clamp(npages, 1, FUSE_MAX_PAGES_PER_REQ);
995
	npages = get_user_pages_fast(user_addr, npages, !write, req->pages);
Miklos Szeredi's avatar
Miklos Szeredi committed
996
997
998
999
1000
	if (npages < 0)
		return npages;

	req->num_pages = npages;
	req->page_offset = offset;
1001
1002
1003
1004
1005
1006
1007
1008
1009

	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
1010
1011
1012
	return 0;
}

1013
1014
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
1015
{
1016
1017
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
Miklos Szeredi's avatar
Miklos Szeredi committed
1018
1019
1020
	size_t nmax = write ? fc->max_write : fc->max_read;
	loff_t pos = *ppos;
	ssize_t res = 0;
1021
1022
	struct fuse_req *req;

1023
1024
1025
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1026
1027
1028

	while (count) {
		size_t nres;
1029
		fl_owner_t owner = current->files;
1030
1031
		size_t nbytes = min(count, nmax);
		int err = fuse_get_user_pages(req, buf, &nbytes, write);
Miklos Szeredi's avatar
Miklos Szeredi committed
1032
1033
1034
1035
		if (err) {
			res = err;
			break;
		}
1036

Miklos Szeredi's avatar
Miklos Szeredi committed
1037
		if (write)
1038
			nres = fuse_send_write(req, file, pos, nbytes, owner);
Miklos Szeredi's avatar
Miklos Szeredi committed
1039
		else
1040
1041
			nres = fuse_send_read(req, file, pos, nbytes, owner);

Miklos Szeredi's avatar
Miklos Szeredi committed
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
		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;
1057
1058
1059
1060
1061
1062
		if (count) {
			fuse_put_request(fc, req);
			req = fuse_get_req(fc);
			if (IS_ERR(req))
				break;
		}
Miklos Szeredi's avatar
Miklos Szeredi committed
1063
	}
1064
1065
	if (!IS_ERR(req))
		fuse_put_request(fc, req);
1066
	if (res > 0)
Miklos Szeredi's avatar
Miklos Szeredi committed
1067
1068
1069
1070
		*ppos = pos;

	return res;
}
1071
EXPORT_SYMBOL_GPL(fuse_direct_io);
Miklos Szeredi's avatar
Miklos Szeredi committed
1072
1073
1074
1075

static ssize_t fuse_direct_read(struct file *file, char __user *buf,
				     size_t count, loff_t *ppos)
{
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
	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
1087
1088
1089
1090
1091
}

static ssize_t fuse_direct_write(struct file *file, const char __user *buf,
				 size_t count, loff_t *ppos)
{
1092
	struct inode *inode = file->f_path.dentry->d_inode;
Miklos Szeredi's avatar
Miklos Szeredi committed
1093
	ssize_t res;
1094
1095
1096
1097

	if (is_bad_inode(inode))
		return -EIO;

Miklos Szeredi's avatar
Miklos Szeredi committed
1098
	/* Don't allow parallel writes to the same file */
1099
	mutex_lock(&inode->i_mutex);
1100
	res = generic_write_checks(file, ppos, &count, 0);
1101
	if (!res) {
1102
		res = fuse_direct_io(file, buf, count, ppos, 1);
1103
1104
1105
		if (res > 0)
			fuse_write_update_size(inode, *ppos);
	}
1106
	mutex_unlock(&inode->i_mutex);
1107
1108
1109

	fuse_invalidate_attr(inode);

Miklos Szeredi's avatar
Miklos Szeredi committed
1110
1111
1112
	return res;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1113
static void fuse_writepage_free(struct fuse_conn *fc, struct fuse_req *req)
1114
{
Miklos Szeredi's avatar
Miklos Szeredi committed
1115
	__free_page(req->pages[0]);
1116
	fuse_file_put(req->ff, false);
Miklos Szeredi's avatar
Miklos Szeredi committed
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
}

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
1134
1135
__releases(fc->lock)
__acquires(fc->lock)
Miklos Szeredi's avatar
Miklos Szeredi committed
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
{
	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;
1151
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1152
1153
1154

	req->in.args[1].size = inarg->size;
	fi->writectr++;
1155
	fuse_request_send_background_locked(fc, req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1156
1157
1158
1159
1160
1161
	return;

 out_free:
	fuse_writepage_finish(fc, req);
	spin_unlock(&fc->lock);
	fuse_writepage_free(fc, req);
1162
	fuse_put_request(fc, req);
Miklos Szeredi's avatar
Miklos Szeredi committed
1163
	spin_lock(&fc->lock);
1164
1165
}

Miklos Szeredi's avatar
Miklos Szeredi committed
1166
1167
1168
1169
1170
1171
1172
/*
 * 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
1173
1174
__releases(fc->lock)
__acquires(fc->lock)
1175
{
Miklos Szeredi's avatar
Miklos Szeredi committed
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
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
	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);

	req = fuse_request_alloc_nofs();
	if (!req)
		goto err;

	tmp_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
	if (!tmp_page)
		goto err_free;

	spin_lock(&fc->lock);
	BUG_ON(list_empty(&fi->write_files));
	ff = list_entry(fi->write_files.next, struct fuse_file, write_entry);
	req->ff = fuse_file_get(ff);
	spin_unlock(&fc->lock);

1226
	fuse_write_fill(req, ff, page_offset(page), 0);
Miklos Szeredi's avatar
Miklos Szeredi committed
1227
1228

	copy_highpage(tmp_page, page);
1229
	req->misc.write.in.write_flags |= FUSE_WRITE_CACHE;
1230
	req->in.argpages = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
	req->num_pages = 1;
	req->pages[0] = tmp_page;
	req->page_offset = 0;
	req->end = fuse_writepage_end;
	req->inode = inode;

	inc_bdi_stat(mapping->backing_dev_info, BDI_WRITEBACK);
	inc_zone_page_state(tmp_page, NR_WRITEBACK_TEMP);
	end_page_writeback(page);

	spin_lock(&fc->lock);
	list_add(&req->writepages_entry, &fi->writepages);
	list_add_tail(&req->list, &fi->queued_writes);
	fuse_flush_writepages(inode);
	spin_unlock(&fc->lock);

	return 0;

err_free:
	fuse_request_free