file.c 73.5 KB
Newer Older
M
Miklos Szeredi 已提交
1 2
/*
  FUSE: Filesystem in Userspace
M
Miklos Szeredi 已提交
3
  Copyright (C) 2001-2008  Miklos Szeredi <miklos@szeredi.hu>
M
Miklos Szeredi 已提交
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>
A
Alexey Dobriyan 已提交
14
#include <linux/sched.h>
15
#include <linux/module.h>
16
#include <linux/compat.h>
17
#include <linux/swap.h>
18
#include <linux/aio.h>
19
#include <linux/falloc.h>
M
Miklos Szeredi 已提交
20

21
static const struct file_operations fuse_direct_io_file_operations;
22

23 24
static int fuse_send_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
			  int opcode, struct fuse_open_out *outargp)
M
Miklos Szeredi 已提交
25 26
{
	struct fuse_open_in inarg;
27
	FUSE_ARGS(args);
M
Miklos Szeredi 已提交
28 29

	memset(&inarg, 0, sizeof(inarg));
30 31 32
	inarg.flags = file->f_flags & ~(O_CREAT | O_EXCL | O_NOCTTY);
	if (!fc->atomic_o_trunc)
		inarg.flags &= ~O_TRUNC;
33 34 35 36 37 38 39 40
	args.in.h.opcode = opcode;
	args.in.h.nodeid = nodeid;
	args.in.numargs = 1;
	args.in.args[0].size = sizeof(inarg);
	args.in.args[0].value = &inarg;
	args.out.numargs = 1;
	args.out.args[0].size = sizeof(*outargp);
	args.out.args[0].value = outargp;
M
Miklos Szeredi 已提交
41

42
	return fuse_simple_request(fc, &args);
M
Miklos Szeredi 已提交
43 44
}

T
Tejun Heo 已提交
45
struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
M
Miklos Szeredi 已提交
46 47
{
	struct fuse_file *ff;
T
Tejun Heo 已提交
48

M
Miklos Szeredi 已提交
49
	ff = kmalloc(sizeof(struct fuse_file), GFP_KERNEL);
T
Tejun Heo 已提交
50 51 52
	if (unlikely(!ff))
		return NULL;

53
	ff->fc = fc;
54
	ff->reserved_req = fuse_request_alloc(0);
T
Tejun Heo 已提交
55 56 57
	if (unlikely(!ff->reserved_req)) {
		kfree(ff);
		return NULL;
M
Miklos Szeredi 已提交
58
	}
T
Tejun Heo 已提交
59 60 61 62 63 64 65 66 67 68

	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);

M
Miklos Szeredi 已提交
69 70 71 72 73
	return ff;
}

void fuse_file_free(struct fuse_file *ff)
{
74
	fuse_request_free(ff->reserved_req);
M
Miklos Szeredi 已提交
75 76 77
	kfree(ff);
}

78
struct fuse_file *fuse_file_get(struct fuse_file *ff)
79 80 81 82 83
{
	atomic_inc(&ff->count);
	return ff;
}

M
Miklos Szeredi 已提交
84 85
static void fuse_release_end(struct fuse_conn *fc, struct fuse_req *req)
{
86
	iput(req->misc.release.inode);
M
Miklos Szeredi 已提交
87 88
}

89
static void fuse_file_put(struct fuse_file *ff, bool sync)
90 91 92
{
	if (atomic_dec_and_test(&ff->count)) {
		struct fuse_req *req = ff->reserved_req;
93

94 95 96 97 98 99
		if (ff->fc->no_open) {
			/*
			 * Drop the release request when client does not
			 * implement 'open'
			 */
			req->background = 0;
100
			iput(req->misc.release.inode);
101 102
			fuse_put_request(ff->fc, req);
		} else if (sync) {
103
			req->background = 0;
104
			fuse_request_send(ff->fc, req);
105
			iput(req->misc.release.inode);
106 107 108
			fuse_put_request(ff->fc, req);
		} else {
			req->end = fuse_release_end;
109
			req->background = 1;
110 111
			fuse_request_send_background(ff->fc, req);
		}
112 113 114 115
		kfree(ff);
	}
}

116 117
int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
		 bool isdir)
118 119 120 121 122 123 124 125
{
	struct fuse_file *ff;
	int opcode = isdir ? FUSE_OPENDIR : FUSE_OPEN;

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

126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142
	ff->fh = 0;
	ff->open_flags = FOPEN_KEEP_CACHE; /* Default for no-open */
	if (!fc->no_open || isdir) {
		struct fuse_open_out outarg;
		int err;

		err = fuse_send_open(fc, nodeid, file, opcode, &outarg);
		if (!err) {
			ff->fh = outarg.fh;
			ff->open_flags = outarg.open_flags;

		} else if (err != -ENOSYS || isdir) {
			fuse_file_free(ff);
			return err;
		} else {
			fc->no_open = 1;
		}
143 144 145
	}

	if (isdir)
146
		ff->open_flags &= ~FOPEN_DIRECT_IO;
147 148 149 150 151 152

	ff->nodeid = nodeid;
	file->private_data = fuse_file_get(ff);

	return 0;
}
153
EXPORT_SYMBOL_GPL(fuse_do_open);
154

155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
static void fuse_link_write_file(struct file *file)
{
	struct inode *inode = file_inode(file);
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct fuse_file *ff = file->private_data;
	/*
	 * file may be written through mmap, so chain it onto the
	 * inodes's write_file list
	 */
	spin_lock(&fc->lock);
	if (list_empty(&ff->write_entry))
		list_add(&ff->write_entry, &fi->write_files);
	spin_unlock(&fc->lock);
}

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

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

		spin_lock(&fc->lock);
		fi->attr_version = ++fc->attr_version;
		i_size_write(inode, 0);
		spin_unlock(&fc->lock);
		fuse_invalidate_attr(inode);
190 191
		if (fc->writeback_cache)
			file_update_time(file);
192
	}
P
Pavel Emelyanov 已提交
193 194
	if ((file->f_mode & FMODE_WRITE) && fc->writeback_cache)
		fuse_link_write_file(file);
M
Miklos Szeredi 已提交
195 196
}

197
int fuse_open_common(struct inode *inode, struct file *file, bool isdir)
M
Miklos Szeredi 已提交
198
{
T
Tejun Heo 已提交
199
	struct fuse_conn *fc = get_fuse_conn(inode);
M
Miklos Szeredi 已提交
200
	int err;
201 202 203
	bool lock_inode = (file->f_flags & O_TRUNC) &&
			  fc->atomic_o_trunc &&
			  fc->writeback_cache;
M
Miklos Szeredi 已提交
204 205 206 207 208

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

209 210 211
	if (lock_inode)
		mutex_lock(&inode->i_mutex);

212
	err = fuse_do_open(fc, get_node_id(inode), file, isdir);
M
Miklos Szeredi 已提交
213

214 215
	if (!err)
		fuse_finish_open(inode, file);
216

217 218 219 220
	if (lock_inode)
		mutex_unlock(&inode->i_mutex);

	return err;
M
Miklos Szeredi 已提交
221 222
}

223
static void fuse_prepare_release(struct fuse_file *ff, int flags, int opcode)
224
{
225
	struct fuse_conn *fc = ff->fc;
226
	struct fuse_req *req = ff->reserved_req;
227
	struct fuse_release_in *inarg = &req->misc.release.in;
M
Miklos Szeredi 已提交
228

229 230 231 232 233 234
	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);

235
	wake_up_interruptible_all(&ff->poll_wait);
236

M
Miklos Szeredi 已提交
237
	inarg->fh = ff->fh;
M
Miklos Szeredi 已提交
238
	inarg->flags = flags;
239
	req->in.h.opcode = opcode;
240
	req->in.h.nodeid = ff->nodeid;
M
Miklos Szeredi 已提交
241 242 243
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_release_in);
	req->in.args[0].value = inarg;
M
Miklos Szeredi 已提交
244 245
}

246
void fuse_release_common(struct file *file, int opcode)
M
Miklos Szeredi 已提交
247
{
T
Tejun Heo 已提交
248 249
	struct fuse_file *ff;
	struct fuse_req *req;
M
Miklos Szeredi 已提交
250

T
Tejun Heo 已提交
251 252
	ff = file->private_data;
	if (unlikely(!ff))
253
		return;
T
Tejun Heo 已提交
254 255

	req = ff->reserved_req;
256
	fuse_prepare_release(ff, file->f_flags, opcode);
T
Tejun Heo 已提交
257

M
Miklos Szeredi 已提交
258 259 260 261 262 263
	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);
	}
264 265
	/* Hold inode until release is finished */
	req->misc.release.inode = igrab(file_inode(file));
T
Tejun Heo 已提交
266 267 268 269 270

	/*
	 * Normally this will send the RELEASE request, however if
	 * some asynchronous READ or WRITE requests are outstanding,
	 * the sending will be delayed.
271 272 273 274
	 *
	 * 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.
T
Tejun Heo 已提交
275
	 */
276
	fuse_file_put(ff, ff->fc->destroy_req != NULL);
M
Miklos Szeredi 已提交
277 278
}

279 280
static int fuse_open(struct inode *inode, struct file *file)
{
281
	return fuse_open_common(inode, file, false);
282 283 284 285
}

static int fuse_release(struct inode *inode, struct file *file)
{
P
Pavel Emelyanov 已提交
286 287 288 289
	struct fuse_conn *fc = get_fuse_conn(inode);

	/* see fuse_vma_close() for !writeback_cache case */
	if (fc->writeback_cache)
M
Miklos Szeredi 已提交
290
		write_inode_now(inode, 1);
M
Maxim Patlasov 已提交
291

292 293 294 295 296 297 298 299 300 301 302
	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;
303
	ff->reserved_req->background = 0;
304 305 306
	fuse_request_send(ff->fc, ff->reserved_req);
	fuse_put_request(ff->fc, ff->reserved_req);
	kfree(ff);
307
}
308
EXPORT_SYMBOL_GPL(fuse_sync_release);
309

310
/*
311 312
 * Scramble the ID space with XTEA, so that the value of the files_struct
 * pointer is not exposed to userspace.
313
 */
314
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id)
315
{
316 317 318 319 320 321 322 323 324 325 326 327 328 329
	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);
330 331
}

M
Miklos Szeredi 已提交
332
/*
333
 * Check if any page in a range is under writeback
M
Miklos Szeredi 已提交
334 335 336 337
 *
 * This is currently done by walking the list of writepage requests
 * for the inode, which can be pretty inefficient.
 */
338 339
static bool fuse_range_is_writeback(struct inode *inode, pgoff_t idx_from,
				   pgoff_t idx_to)
M
Miklos Szeredi 已提交
340 341 342 343 344 345 346 347 348 349 350 351
{
	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;
352 353
		if (idx_from < curr_index + req->num_pages &&
		    curr_index <= idx_to) {
M
Miklos Szeredi 已提交
354 355 356 357 358 359 360 361 362
			found = true;
			break;
		}
	}
	spin_unlock(&fc->lock);

	return found;
}

363 364 365 366 367
static inline bool fuse_page_is_writeback(struct inode *inode, pgoff_t index)
{
	return fuse_range_is_writeback(inode, index, index);
}

M
Miklos Szeredi 已提交
368 369 370 371 372 373 374 375 376 377 378 379 380 381
/*
 * 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;
}

382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
/*
 * 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);
}

397
static int fuse_flush(struct file *file, fl_owner_t id)
M
Miklos Szeredi 已提交
398
{
A
Al Viro 已提交
399
	struct inode *inode = file_inode(file);
M
Miklos Szeredi 已提交
400 401 402 403 404 405
	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;

406 407 408
	if (is_bad_inode(inode))
		return -EIO;

M
Miklos Szeredi 已提交
409 410 411
	if (fc->no_flush)
		return 0;

M
Miklos Szeredi 已提交
412
	err = write_inode_now(inode, 1);
413 414 415 416 417 418 419
	if (err)
		return err;

	mutex_lock(&inode->i_mutex);
	fuse_sync_writes(inode);
	mutex_unlock(&inode->i_mutex);

M
Maxim Patlasov 已提交
420
	req = fuse_get_req_nofail_nopages(fc, file);
M
Miklos Szeredi 已提交
421 422
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
423
	inarg.lock_owner = fuse_lock_owner_id(fc, id);
M
Miklos Szeredi 已提交
424 425 426 427 428
	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;
429
	req->force = 1;
430
	fuse_request_send(fc, req);
M
Miklos Szeredi 已提交
431 432 433 434 435 436 437 438 439
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
		fc->no_flush = 1;
		err = 0;
	}
	return err;
}

440 441
int fuse_fsync_common(struct file *file, loff_t start, loff_t end,
		      int datasync, int isdir)
M
Miklos Szeredi 已提交
442
{
443
	struct inode *inode = file->f_mapping->host;
M
Miklos Szeredi 已提交
444 445
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_file *ff = file->private_data;
446
	FUSE_ARGS(args);
M
Miklos Szeredi 已提交
447 448 449
	struct fuse_fsync_in inarg;
	int err;

450 451 452
	if (is_bad_inode(inode))
		return -EIO;

453 454
	mutex_lock(&inode->i_mutex);

M
Miklos Szeredi 已提交
455 456 457 458 459
	/*
	 * Start writeback against all dirty pages of the inode, then
	 * wait for all outstanding writes, before sending the FSYNC
	 * request.
	 */
M
Miklos Szeredi 已提交
460
	err = filemap_write_and_wait_range(inode->i_mapping, start, end);
M
Miklos Szeredi 已提交
461
	if (err)
462
		goto out;
M
Miklos Szeredi 已提交
463 464

	fuse_sync_writes(inode);
M
Miklos Szeredi 已提交
465 466 467
	err = sync_inode_metadata(inode, 1);
	if (err)
		goto out;
M
Miklos Szeredi 已提交
468

M
Miklos Szeredi 已提交
469 470
	if ((!isdir && fc->no_fsync) || (isdir && fc->no_fsyncdir))
		goto out;
M
Maxim Patlasov 已提交
471

M
Miklos Szeredi 已提交
472 473 474
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
	inarg.fsync_flags = datasync ? 1 : 0;
475 476 477 478 479 480
	args.in.h.opcode = isdir ? FUSE_FSYNCDIR : FUSE_FSYNC;
	args.in.h.nodeid = get_node_id(inode);
	args.in.numargs = 1;
	args.in.args[0].size = sizeof(inarg);
	args.in.args[0].value = &inarg;
	err = fuse_simple_request(fc, &args);
M
Miklos Szeredi 已提交
481
	if (err == -ENOSYS) {
482 483 484 485
		if (isdir)
			fc->no_fsyncdir = 1;
		else
			fc->no_fsync = 1;
M
Miklos Szeredi 已提交
486 487
		err = 0;
	}
488 489
out:
	mutex_unlock(&inode->i_mutex);
M
Miklos Szeredi 已提交
490 491 492
	return err;
}

493 494
static int fuse_fsync(struct file *file, loff_t start, loff_t end,
		      int datasync)
495
{
496
	return fuse_fsync_common(file, start, end, datasync, 0);
497 498
}

499 500
void fuse_read_fill(struct fuse_req *req, struct file *file, loff_t pos,
		    size_t count, int opcode)
M
Miklos Szeredi 已提交
501
{
502
	struct fuse_read_in *inarg = &req->misc.read.in;
503
	struct fuse_file *ff = file->private_data;
M
Miklos Szeredi 已提交
504

505 506 507
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
508
	inarg->flags = file->f_flags;
509
	req->in.h.opcode = opcode;
510
	req->in.h.nodeid = ff->nodeid;
M
Miklos Szeredi 已提交
511 512
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_read_in);
513
	req->in.args[0].value = inarg;
M
Miklos Szeredi 已提交
514 515 516 517 518
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = count;
}

519 520 521 522 523 524 525 526 527 528 529 530
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);
	}
}

M
Maxim Patlasov 已提交
531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570
/**
 * In case of short read, the caller sets 'pos' to the position of
 * actual end of fuse request in IO request. Otherwise, if bytes_requested
 * == bytes_transferred or rw == WRITE, the caller sets 'pos' to -1.
 *
 * An example:
 * User requested DIO read of 64K. It was splitted into two 32K fuse requests,
 * both submitted asynchronously. The first of them was ACKed by userspace as
 * fully completed (req->out.args[0].size == 32K) resulting in pos == -1. The
 * second request was ACKed as short, e.g. only 1K was read, resulting in
 * pos == 33K.
 *
 * Thus, when all fuse requests are completed, the minimal non-negative 'pos'
 * will be equal to the length of the longest contiguous fragment of
 * transferred data starting from the beginning of IO request.
 */
static void fuse_aio_complete(struct fuse_io_priv *io, int err, ssize_t pos)
{
	int left;

	spin_lock(&io->lock);
	if (err)
		io->err = io->err ? : err;
	else if (pos >= 0 && (io->bytes < 0 || pos < io->bytes))
		io->bytes = pos;

	left = --io->reqs;
	spin_unlock(&io->lock);

	if (!left) {
		long res;

		if (io->err)
			res = io->err;
		else if (io->bytes >= 0 && io->write)
			res = -EIO;
		else {
			res = io->bytes < 0 ? io->size : io->bytes;

			if (!is_sync_kiocb(io->iocb)) {
A
Al Viro 已提交
571
				struct inode *inode = file_inode(io->iocb->ki_filp);
M
Maxim Patlasov 已提交
572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616
				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;
				spin_unlock(&fc->lock);
			}
		}

		aio_complete(io->iocb, res, 0);
		kfree(io);
	}
}

static void fuse_aio_complete_req(struct fuse_conn *fc, struct fuse_req *req)
{
	struct fuse_io_priv *io = req->io;
	ssize_t pos = -1;

	fuse_release_user_pages(req, !io->write);

	if (io->write) {
		if (req->misc.write.in.size != req->misc.write.out.size)
			pos = req->misc.write.in.offset - io->offset +
				req->misc.write.out.size;
	} else {
		if (req->misc.read.in.size != req->out.args[0].size)
			pos = req->misc.read.in.offset - io->offset +
				req->out.args[0].size;
	}

	fuse_aio_complete(io, req->out.h.error, pos);
}

static size_t fuse_async_req_send(struct fuse_conn *fc, struct fuse_req *req,
		size_t num_bytes, struct fuse_io_priv *io)
{
	spin_lock(&io->lock);
	io->size += num_bytes;
	io->reqs++;
	spin_unlock(&io->lock);

	req->io = io;
	req->end = fuse_aio_complete_req;

617
	__fuse_get_request(req);
M
Maxim Patlasov 已提交
618 619 620 621 622
	fuse_request_send_background(fc, req);

	return num_bytes;
}

623
static size_t fuse_send_read(struct fuse_req *req, struct fuse_io_priv *io,
624
			     loff_t pos, size_t count, fl_owner_t owner)
625
{
626
	struct file *file = io->file;
627 628
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
629

630
	fuse_read_fill(req, file, pos, count, FUSE_READ);
631
	if (owner != NULL) {
632
		struct fuse_read_in *inarg = &req->misc.read.in;
633 634 635 636

		inarg->read_flags |= FUSE_READ_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
637 638 639 640

	if (io->async)
		return fuse_async_req_send(fc, req, count, io);

641
	fuse_request_send(fc, req);
642
	return req->out.args[0].size;
643 644
}

645 646 647 648 649 650 651
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);
652 653
	if (attr_ver == fi->attr_version && size < inode->i_size &&
	    !test_bit(FUSE_I_SIZE_UNSTABLE, &fi->state)) {
654 655 656 657 658 659
		fi->attr_version = ++fc->attr_version;
		i_size_write(inode, size);
	}
	spin_unlock(&fc->lock);
}

660 661 662 663
static void fuse_short_read(struct fuse_req *req, struct inode *inode,
			    u64 attr_ver)
{
	size_t num_read = req->out.args[0].size;
P
Pavel Emelyanov 已提交
664 665 666 667 668 669 670 671 672 673 674
	struct fuse_conn *fc = get_fuse_conn(inode);

	if (fc->writeback_cache) {
		/*
		 * A hole in a file. Some data after the hole are in page cache,
		 * but have not reached the client fs yet. So, the hole is not
		 * present there.
		 */
		int i;
		int start_idx = num_read >> PAGE_CACHE_SHIFT;
		size_t off = num_read & (PAGE_CACHE_SIZE - 1);
675

P
Pavel Emelyanov 已提交
676 677 678 679 680 681 682 683
		for (i = start_idx; i < req->num_pages; i++) {
			zero_user_segment(req->pages[i], off, PAGE_CACHE_SIZE);
			off = 0;
		}
	} else {
		loff_t pos = page_offset(req->pages[0]) + num_read;
		fuse_read_update_size(inode, pos, attr_ver);
	}
684 685
}

686
static int fuse_do_readpage(struct file *file, struct page *page)
M
Miklos Szeredi 已提交
687
{
688
	struct fuse_io_priv io = { .async = 0, .file = file };
M
Miklos Szeredi 已提交
689 690
	struct inode *inode = page->mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
691
	struct fuse_req *req;
692 693 694 695
	size_t num_read;
	loff_t pos = page_offset(page);
	size_t count = PAGE_CACHE_SIZE;
	u64 attr_ver;
696 697
	int err;

M
Miklos Szeredi 已提交
698
	/*
L
Lucas De Marchi 已提交
699
	 * Page writeback can extend beyond the lifetime of the
M
Miklos Szeredi 已提交
700 701 702 703 704
	 * page-cache page, so make sure we read a properly synced
	 * page.
	 */
	fuse_wait_on_page_writeback(inode, page->index);

M
Maxim Patlasov 已提交
705
	req = fuse_get_req(fc, 1);
706
	if (IS_ERR(req))
707
		return PTR_ERR(req);
M
Miklos Szeredi 已提交
708

709 710
	attr_ver = fuse_get_attr_version(fc);

M
Miklos Szeredi 已提交
711
	req->out.page_zeroing = 1;
712
	req->out.argpages = 1;
M
Miklos Szeredi 已提交
713 714
	req->num_pages = 1;
	req->pages[0] = page;
715
	req->page_descs[0].length = count;
716
	num_read = fuse_send_read(req, &io, pos, count, NULL);
M
Miklos Szeredi 已提交
717
	err = req->out.h.error;
718 719 720 721 722 723

	if (!err) {
		/*
		 * Short read means EOF.  If file size is larger, truncate it
		 */
		if (num_read < count)
724
			fuse_short_read(req, inode, attr_ver);
725

M
Miklos Szeredi 已提交
726
		SetPageUptodate(page);
727 728
	}

729
	fuse_put_request(fc, req);
730 731 732 733 734 735 736 737 738 739 740 741 742 743

	return err;
}

static int fuse_readpage(struct file *file, struct page *page)
{
	struct inode *inode = page->mapping->host;
	int err;

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

	err = fuse_do_readpage(file, page);
744
	fuse_invalidate_atime(inode);
M
Miklos Szeredi 已提交
745 746 747 748 749
 out:
	unlock_page(page);
	return err;
}

750
static void fuse_readpages_end(struct fuse_conn *fc, struct fuse_req *req)
751
{
752
	int i;
753 754
	size_t count = req->misc.read.in.size;
	size_t num_read = req->out.args[0].size;
755
	struct address_space *mapping = NULL;
756

757 758
	for (i = 0; mapping == NULL && i < req->num_pages; i++)
		mapping = req->pages[i]->mapping;
759

760 761 762 763 764 765
	if (mapping) {
		struct inode *inode = mapping->host;

		/*
		 * Short read means EOF. If file size is larger, truncate it
		 */
766 767
		if (!req->out.h.error && num_read < count)
			fuse_short_read(req, inode, req->misc.read.attr_ver);
768

769
		fuse_invalidate_atime(inode);
770
	}
771

772 773 774 775
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (!req->out.h.error)
			SetPageUptodate(page);
776 777
		else
			SetPageError(page);
778
		unlock_page(page);
779
		page_cache_release(page);
780
	}
781
	if (req->ff)
782
		fuse_file_put(req->ff, false);
783 784
}

785
static void fuse_send_readpages(struct fuse_req *req, struct file *file)
786
{
787 788
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
789 790
	loff_t pos = page_offset(req->pages[0]);
	size_t count = req->num_pages << PAGE_CACHE_SHIFT;
791 792

	req->out.argpages = 1;
793
	req->out.page_zeroing = 1;
794
	req->out.page_replace = 1;
795
	fuse_read_fill(req, file, pos, count, FUSE_READ);
796
	req->misc.read.attr_ver = fuse_get_attr_version(fc);
797
	if (fc->async_read) {
798
		req->ff = fuse_file_get(ff);
799
		req->end = fuse_readpages_end;
800
		fuse_request_send_background(fc, req);
801
	} else {
802
		fuse_request_send(fc, req);
803
		fuse_readpages_end(fc, req);
804
		fuse_put_request(fc, req);
805
	}
806 807
}

808
struct fuse_fill_data {
809
	struct fuse_req *req;
810
	struct file *file;
811
	struct inode *inode;
M
Maxim Patlasov 已提交
812
	unsigned nr_pages;
813 814 815 816
};

static int fuse_readpages_fill(void *_data, struct page *page)
{
817
	struct fuse_fill_data *data = _data;
818 819 820 821
	struct fuse_req *req = data->req;
	struct inode *inode = data->inode;
	struct fuse_conn *fc = get_fuse_conn(inode);

M
Miklos Szeredi 已提交
822 823
	fuse_wait_on_page_writeback(inode, page->index);

824 825 826 827
	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)) {
M
Maxim Patlasov 已提交
828 829
		int nr_alloc = min_t(unsigned, data->nr_pages,
				     FUSE_MAX_PAGES_PER_REQ);
830
		fuse_send_readpages(req, data->file);
831 832 833 834 835 836
		if (fc->async_read)
			req = fuse_get_req_for_background(fc, nr_alloc);
		else
			req = fuse_get_req(fc, nr_alloc);

		data->req = req;
837
		if (IS_ERR(req)) {
838
			unlock_page(page);
839
			return PTR_ERR(req);
840 841
		}
	}
M
Maxim Patlasov 已提交
842 843 844 845 846 847

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

848
	page_cache_get(page);
849
	req->pages[req->num_pages] = page;
850
	req->page_descs[req->num_pages].length = PAGE_SIZE;
M
Miklos Szeredi 已提交
851
	req->num_pages++;
M
Maxim Patlasov 已提交
852
	data->nr_pages--;
853 854 855 856 857 858 859 860
	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);
861
	struct fuse_fill_data data;
862
	int err;
M
Maxim Patlasov 已提交
863
	int nr_alloc = min_t(unsigned, nr_pages, FUSE_MAX_PAGES_PER_REQ);
864

865
	err = -EIO;
866
	if (is_bad_inode(inode))
867
		goto out;
868

869
	data.file = file;
870
	data.inode = inode;
871 872 873 874
	if (fc->async_read)
		data.req = fuse_get_req_for_background(fc, nr_alloc);
	else
		data.req = fuse_get_req(fc, nr_alloc);
M
Maxim Patlasov 已提交
875
	data.nr_pages = nr_pages;
876
	err = PTR_ERR(data.req);
877
	if (IS_ERR(data.req))
878
		goto out;
879 880

	err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
881 882
	if (!err) {
		if (data.req->num_pages)
883
			fuse_send_readpages(data.req, file);
884 885 886
		else
			fuse_put_request(fc, data.req);
	}
887
out:
888
	return err;
889 890
}

891
static ssize_t fuse_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
M
Miklos Szeredi 已提交
892 893
{
	struct inode *inode = iocb->ki_filp->f_mapping->host;
894
	struct fuse_conn *fc = get_fuse_conn(inode);
M
Miklos Szeredi 已提交
895

896 897 898 899 900 901
	/*
	 * In auto invalidate mode, always update attributes on read.
	 * Otherwise, only update if we attempt to read past EOF (to ensure
	 * i_size is up to date).
	 */
	if (fc->auto_inval_data ||
902
	    (iocb->ki_pos + iov_iter_count(to) > i_size_read(inode))) {
M
Miklos Szeredi 已提交
903 904 905 906 907 908
		int err;
		err = fuse_update_attributes(inode, NULL, iocb->ki_filp, NULL);
		if (err)
			return err;
	}

909
	return generic_file_read_iter(iocb, to);
M
Miklos Szeredi 已提交
910 911
}

912
static void fuse_write_fill(struct fuse_req *req, struct fuse_file *ff,
913
			    loff_t pos, size_t count)
M
Miklos Szeredi 已提交
914
{
915 916
	struct fuse_write_in *inarg = &req->misc.write.in;
	struct fuse_write_out *outarg = &req->misc.write.out;
M
Miklos Szeredi 已提交
917

918 919 920
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
M
Miklos Szeredi 已提交
921
	req->in.h.opcode = FUSE_WRITE;
922
	req->in.h.nodeid = ff->nodeid;
M
Miklos Szeredi 已提交
923
	req->in.numargs = 2;
924
	if (ff->fc->minor < 9)
925 926 927
		req->in.args[0].size = FUSE_COMPAT_WRITE_IN_SIZE;
	else
		req->in.args[0].size = sizeof(struct fuse_write_in);
928
	req->in.args[0].value = inarg;
M
Miklos Szeredi 已提交
929 930 931
	req->in.args[1].size = count;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(struct fuse_write_out);
932 933 934
	req->out.args[0].value = outarg;
}

935
static size_t fuse_send_write(struct fuse_req *req, struct fuse_io_priv *io,
936
			      loff_t pos, size_t count, fl_owner_t owner)
937
{
938
	struct file *file = io->file;
939 940
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
941 942
	struct fuse_write_in *inarg = &req->misc.write.in;

943
	fuse_write_fill(req, ff, pos, count);
944
	inarg->flags = file->f_flags;
945 946 947 948
	if (owner != NULL) {
		inarg->write_flags |= FUSE_WRITE_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
949 950 951 952

	if (io->async)
		return fuse_async_req_send(fc, req, count, io);

953
	fuse_request_send(fc, req);
954
	return req->misc.write.out.size;
M
Miklos Szeredi 已提交
955 956
}

M
Maxim Patlasov 已提交
957
bool fuse_write_update_size(struct inode *inode, loff_t pos)
958 959 960
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
M
Maxim Patlasov 已提交
961
	bool ret = false;
962 963 964

	spin_lock(&fc->lock);
	fi->attr_version = ++fc->attr_version;
M
Maxim Patlasov 已提交
965
	if (pos > inode->i_size) {
966
		i_size_write(inode, pos);
M
Maxim Patlasov 已提交
967 968
		ret = true;
	}
969
	spin_unlock(&fc->lock);
M
Maxim Patlasov 已提交
970 971

	return ret;
972 973
}

N
Nick Piggin 已提交
974 975 976 977 978 979 980
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;
981
	struct fuse_io_priv io = { .async = 0, .file = file };
N
Nick Piggin 已提交
982 983 984 985

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

986
	res = fuse_send_write(req, &io, pos, count, NULL);
N
Nick Piggin 已提交
987

988
	offset = req->page_descs[0].offset;
N
Nick Piggin 已提交
989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017
	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;

1018
	req->in.argpages = 1;
1019
	req->page_descs[0].offset = offset;
N
Nick Piggin 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035

	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;
1036
		page = grab_cache_page_write_begin(mapping, index, 0);
N
Nick Piggin 已提交
1037 1038 1039
		if (!page)
			break;

1040 1041 1042
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

N
Nick Piggin 已提交
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
		tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
		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;
1055
		req->page_descs[req->num_pages].length = tmp;
N
Nick Piggin 已提交
1056 1057 1058 1059 1060 1061 1062 1063 1064
		req->num_pages++;

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

1065 1066
		if (!fc->big_writes)
			break;
N
Nick Piggin 已提交
1067
	} while (iov_iter_count(ii) && count < fc->max_write &&
1068
		 req->num_pages < req->max_pages && offset == 0);
N
Nick Piggin 已提交
1069 1070 1071 1072

	return count > 0 ? count : err;
}

1073 1074 1075 1076 1077 1078 1079 1080
static inline unsigned fuse_wr_pages(loff_t pos, size_t len)
{
	return min_t(unsigned,
		     ((pos + len - 1) >> PAGE_CACHE_SHIFT) -
		     (pos >> PAGE_CACHE_SHIFT) + 1,
		     FUSE_MAX_PAGES_PER_REQ);
}

N
Nick Piggin 已提交
1081 1082 1083 1084 1085 1086
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);
1087
	struct fuse_inode *fi = get_fuse_inode(inode);
N
Nick Piggin 已提交
1088 1089 1090 1091 1092 1093
	int err = 0;
	ssize_t res = 0;

	if (is_bad_inode(inode))
		return -EIO;

1094 1095 1096
	if (inode->i_size < pos + iov_iter_count(ii))
		set_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);

N
Nick Piggin 已提交
1097 1098 1099
	do {
		struct fuse_req *req;
		ssize_t count;
1100
		unsigned nr_pages = fuse_wr_pages(pos, iov_iter_count(ii));
N
Nick Piggin 已提交
1101

1102
		req = fuse_get_req(fc, nr_pages);
N
Nick Piggin 已提交
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131
		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);

1132
	clear_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);
N
Nick Piggin 已提交
1133 1134 1135 1136 1137
	fuse_invalidate_attr(inode);

	return res > 0 ? res : err;
}

A
Al Viro 已提交
1138
static ssize_t fuse_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
N
Nick Piggin 已提交
1139 1140 1141
{
	struct file *file = iocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
A
Al Viro 已提交
1142
	size_t count = iov_iter_count(from);
N
Nick Piggin 已提交
1143
	ssize_t written = 0;
A
Anand Avati 已提交
1144
	ssize_t written_buffered = 0;
N
Nick Piggin 已提交
1145 1146
	struct inode *inode = mapping->host;
	ssize_t err;
A
Anand Avati 已提交
1147
	loff_t endbyte = 0;
A
Al Viro 已提交
1148
	loff_t pos = iocb->ki_pos;
N
Nick Piggin 已提交
1149

P
Pavel Emelyanov 已提交
1150 1151 1152 1153 1154 1155
	if (get_fuse_conn(inode)->writeback_cache) {
		/* Update size (EOF optimization) and mode (SUID clearing) */
		err = fuse_update_attributes(mapping->host, NULL, file, NULL);
		if (err)
			return err;

A
Al Viro 已提交
1156
		return generic_file_write_iter(iocb, from);
P
Pavel Emelyanov 已提交
1157 1158
	}

N
Nick Piggin 已提交
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
	mutex_lock(&inode->i_mutex);

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

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

	if (count == 0)
		goto out;

A
Al Viro 已提交
1171
	iov_iter_truncate(from, count);
1172
	err = file_remove_suid(file);
N
Nick Piggin 已提交
1173 1174 1175
	if (err)
		goto out;

1176 1177 1178
	err = file_update_time(file);
	if (err)
		goto out;
N
Nick Piggin 已提交
1179

A
Anand Avati 已提交
1180
	if (file->f_flags & O_DIRECT) {
A
Al Viro 已提交
1181 1182
		written = generic_file_direct_write(iocb, from, pos);
		if (written < 0 || !iov_iter_count(from))
A
Anand Avati 已提交
1183 1184 1185
			goto out;

		pos += written;
N
Nick Piggin 已提交
1186

A
Al Viro 已提交
1187
		written_buffered = fuse_perform_write(file, mapping, from, pos);
A
Anand Avati 已提交
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
		if (written_buffered < 0) {
			err = written_buffered;
			goto out;
		}
		endbyte = pos + written_buffered - 1;

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

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

		written += written_buffered;
		iocb->ki_pos = pos + written_buffered;
	} else {
A
Al Viro 已提交
1206
		written = fuse_perform_write(file, mapping, from, pos);
A
Anand Avati 已提交
1207 1208 1209
		if (written >= 0)
			iocb->ki_pos = pos + written;
	}
N
Nick Piggin 已提交
1210 1211 1212 1213 1214 1215 1216
out:
	current->backing_dev_info = NULL;
	mutex_unlock(&inode->i_mutex);

	return written ? written : err;
}

1217 1218
static inline void fuse_page_descs_length_init(struct fuse_req *req,
		unsigned index, unsigned nr_pages)
1219 1220 1221
{
	int i;

1222
	for (i = index; i < index + nr_pages; i++)
1223 1224 1225 1226
		req->page_descs[i].length = PAGE_SIZE -
			req->page_descs[i].offset;
}

1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
static inline unsigned long fuse_get_user_addr(const struct iov_iter *ii)
{
	return (unsigned long)ii->iov->iov_base + ii->iov_offset;
}

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

1238
static int fuse_get_user_pages(struct fuse_req *req, struct iov_iter *ii,
1239
			       size_t *nbytesp, int write)
M
Miklos Szeredi 已提交
1240
{
1241
	size_t nbytes = 0;  /* # bytes already packed in req */
1242

1243
	/* Special case for kernel I/O: can copy directly into the buffer */
A
Al Viro 已提交
1244
	if (ii->type & ITER_KVEC) {
1245 1246 1247
		unsigned long user_addr = fuse_get_user_addr(ii);
		size_t frag_size = fuse_get_frag_size(ii, *nbytesp);

1248 1249 1250 1251 1252
		if (write)
			req->in.args[1].value = (void *) user_addr;
		else
			req->out.args[0].value = (void *) user_addr;

1253 1254
		iov_iter_advance(ii, frag_size);
		*nbytesp = frag_size;
1255 1256
		return 0;
	}
M
Miklos Szeredi 已提交
1257

1258
	while (nbytes < *nbytesp && req->num_pages < req->max_pages) {
1259
		unsigned npages;
A
Al Viro 已提交
1260
		size_t start;
A
Al Viro 已提交
1261 1262
		ssize_t ret = iov_iter_get_pages(ii,
					&req->pages[req->num_pages],
1263
					*nbytesp - nbytes,
1264 1265
					req->max_pages - req->num_pages,
					&start);
1266 1267 1268
		if (ret < 0)
			return ret;

A
Al Viro 已提交
1269 1270
		iov_iter_advance(ii, ret);
		nbytes += ret;
1271

A
Al Viro 已提交
1272 1273
		ret += start;
		npages = (ret + PAGE_SIZE - 1) / PAGE_SIZE;
1274

A
Al Viro 已提交
1275
		req->page_descs[req->num_pages].offset = start;
1276 1277 1278 1279
		fuse_page_descs_length_init(req, req->num_pages, npages);

		req->num_pages += npages;
		req->page_descs[req->num_pages - 1].length -=
A
Al Viro 已提交
1280
			(PAGE_SIZE - ret) & (PAGE_SIZE - 1);
1281
	}
1282 1283 1284 1285 1286 1287

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

1288
	*nbytesp = nbytes;
1289

M
Miklos Szeredi 已提交
1290 1291 1292
	return 0;
}

1293 1294
static inline int fuse_iter_npages(const struct iov_iter *ii_p)
{
A
Al Viro 已提交
1295
	return iov_iter_npages(ii_p, FUSE_MAX_PAGES_PER_REQ);
1296 1297
}

A
Al Viro 已提交
1298 1299
ssize_t fuse_direct_io(struct fuse_io_priv *io, struct iov_iter *iter,
		       loff_t *ppos, int flags)
M
Miklos Szeredi 已提交
1300
{
1301 1302
	int write = flags & FUSE_DIO_WRITE;
	int cuse = flags & FUSE_DIO_CUSE;
1303
	struct file *file = io->file;
1304
	struct inode *inode = file->f_mapping->host;
1305 1306
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
M
Miklos Szeredi 已提交
1307 1308
	size_t nmax = write ? fc->max_write : fc->max_read;
	loff_t pos = *ppos;
A
Al Viro 已提交
1309
	size_t count = iov_iter_count(iter);
1310 1311
	pgoff_t idx_from = pos >> PAGE_CACHE_SHIFT;
	pgoff_t idx_to = (pos + count - 1) >> PAGE_CACHE_SHIFT;
M
Miklos Szeredi 已提交
1312
	ssize_t res = 0;
1313 1314
	struct fuse_req *req;

1315
	if (io->async)
A
Al Viro 已提交
1316
		req = fuse_get_req_for_background(fc, fuse_iter_npages(iter));
1317
	else
A
Al Viro 已提交
1318
		req = fuse_get_req(fc, fuse_iter_npages(iter));
1319 1320
	if (IS_ERR(req))
		return PTR_ERR(req);
M
Miklos Szeredi 已提交
1321

1322 1323 1324 1325 1326 1327 1328 1329
	if (!cuse && fuse_range_is_writeback(inode, idx_from, idx_to)) {
		if (!write)
			mutex_lock(&inode->i_mutex);
		fuse_sync_writes(inode);
		if (!write)
			mutex_unlock(&inode->i_mutex);
	}

M
Miklos Szeredi 已提交
1330 1331
	while (count) {
		size_t nres;
1332
		fl_owner_t owner = current->files;
1333
		size_t nbytes = min(count, nmax);
A
Al Viro 已提交
1334
		int err = fuse_get_user_pages(req, iter, &nbytes, write);
M
Miklos Szeredi 已提交
1335 1336 1337 1338
		if (err) {
			res = err;
			break;
		}
1339

M
Miklos Szeredi 已提交
1340
		if (write)
1341
			nres = fuse_send_write(req, io, pos, nbytes, owner);
M
Miklos Szeredi 已提交
1342
		else
1343
			nres = fuse_send_read(req, io, pos, nbytes, owner);
1344

1345 1346
		if (!io->async)
			fuse_release_user_pages(req, !write);
M
Miklos Szeredi 已提交
1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
		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;
		if (nres != nbytes)
			break;
1360 1361
		if (count) {
			fuse_put_request(fc, req);
1362 1363
			if (io->async)
				req = fuse_get_req_for_background(fc,
A
Al Viro 已提交
1364
					fuse_iter_npages(iter));
1365
			else
A
Al Viro 已提交
1366
				req = fuse_get_req(fc, fuse_iter_npages(iter));
1367 1368 1369
			if (IS_ERR(req))
				break;
		}
M
Miklos Szeredi 已提交
1370
	}
1371 1372
	if (!IS_ERR(req))
		fuse_put_request(fc, req);
1373
	if (res > 0)
M
Miklos Szeredi 已提交
1374 1375 1376 1377
		*ppos = pos;

	return res;
}
1378
EXPORT_SYMBOL_GPL(fuse_direct_io);
M
Miklos Szeredi 已提交
1379

1380
static ssize_t __fuse_direct_read(struct fuse_io_priv *io,
A
Al Viro 已提交
1381 1382
				  struct iov_iter *iter,
				  loff_t *ppos)
M
Miklos Szeredi 已提交
1383
{
1384
	ssize_t res;
1385
	struct file *file = io->file;
A
Al Viro 已提交
1386
	struct inode *inode = file_inode(file);
1387 1388 1389 1390

	if (is_bad_inode(inode))
		return -EIO;

A
Al Viro 已提交
1391
	res = fuse_direct_io(io, iter, ppos, 0);
1392 1393 1394 1395

	fuse_invalidate_attr(inode);

	return res;
M
Miklos Szeredi 已提交
1396 1397
}

1398 1399 1400
static ssize_t fuse_direct_read(struct file *file, char __user *buf,
				     size_t count, loff_t *ppos)
{
1401
	struct fuse_io_priv io = { .async = 0, .file = file };
M
Miklos Szeredi 已提交
1402
	struct iovec iov = { .iov_base = buf, .iov_len = count };
A
Al Viro 已提交
1403 1404 1405
	struct iov_iter ii;
	iov_iter_init(&ii, READ, &iov, 1, count);
	return __fuse_direct_read(&io, &ii, ppos);
1406 1407
}

1408
static ssize_t __fuse_direct_write(struct fuse_io_priv *io,
A
Al Viro 已提交
1409 1410
				   struct iov_iter *iter,
				   loff_t *ppos)
M
Miklos Szeredi 已提交
1411
{
1412
	struct file *file = io->file;
A
Al Viro 已提交
1413
	struct inode *inode = file_inode(file);
A
Al Viro 已提交
1414
	size_t count = iov_iter_count(iter);
M
Miklos Szeredi 已提交
1415
	ssize_t res;
1416

A
Al Viro 已提交
1417

1418
	res = generic_write_checks(file, ppos, &count, 0);
A
Al Viro 已提交
1419
	if (!res) {
A
Al Viro 已提交
1420
		iov_iter_truncate(iter, count);
A
Al Viro 已提交
1421 1422
		res = fuse_direct_io(io, iter, ppos, FUSE_DIO_WRITE);
	}
1423 1424 1425

	fuse_invalidate_attr(inode);

M
Miklos Szeredi 已提交
1426 1427 1428
	return res;
}

A
Anand Avati 已提交
1429 1430 1431
static ssize_t fuse_direct_write(struct file *file, const char __user *buf,
				 size_t count, loff_t *ppos)
{
M
Miklos Szeredi 已提交
1432
	struct iovec iov = { .iov_base = (void __user *)buf, .iov_len = count };
A
Al Viro 已提交
1433
	struct inode *inode = file_inode(file);
A
Anand Avati 已提交
1434
	ssize_t res;
1435
	struct fuse_io_priv io = { .async = 0, .file = file };
A
Al Viro 已提交
1436 1437
	struct iov_iter ii;
	iov_iter_init(&ii, WRITE, &iov, 1, count);
A
Anand Avati 已提交
1438 1439 1440 1441 1442 1443

	if (is_bad_inode(inode))
		return -EIO;

	/* Don't allow parallel writes to the same file */
	mutex_lock(&inode->i_mutex);
A
Al Viro 已提交
1444
	res = __fuse_direct_write(&io, &ii, ppos);
1445 1446
	if (res > 0)
		fuse_write_update_size(inode, *ppos);
A
Anand Avati 已提交
1447 1448 1449 1450 1451
	mutex_unlock(&inode->i_mutex);

	return res;
}

M
Miklos Szeredi 已提交
1452
static void fuse_writepage_free(struct fuse_conn *fc, struct fuse_req *req)
M
Miklos Szeredi 已提交
1453
{
1454 1455 1456 1457
	int i;

	for (i = 0; i < req->num_pages; i++)
		__free_page(req->pages[i]);
1458 1459 1460

	if (req->ff)
		fuse_file_put(req->ff, false);
M
Miklos Szeredi 已提交
1461 1462 1463 1464 1465 1466 1467
}

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;
1468
	int i;
M
Miklos Szeredi 已提交
1469 1470

	list_del(&req->writepages_entry);
1471 1472 1473 1474 1475
	for (i = 0; i < req->num_pages; i++) {
		dec_bdi_stat(bdi, BDI_WRITEBACK);
		dec_zone_page_state(req->pages[i], NR_WRITEBACK_TEMP);
		bdi_writeout_inc(bdi);
	}
M
Miklos Szeredi 已提交
1476 1477 1478 1479
	wake_up(&fi->page_waitq);
}

/* Called under fc->lock, may release and reacquire it */
1480 1481
static void fuse_send_writepage(struct fuse_conn *fc, struct fuse_req *req,
				loff_t size)
M
Miklos Szeredi 已提交
1482 1483
__releases(fc->lock)
__acquires(fc->lock)
M
Miklos Szeredi 已提交
1484 1485 1486
{
	struct fuse_inode *fi = get_fuse_inode(req->inode);
	struct fuse_write_in *inarg = &req->misc.write.in;
1487
	__u64 data_size = req->num_pages * PAGE_CACHE_SIZE;
M
Miklos Szeredi 已提交
1488 1489 1490 1491

	if (!fc->connected)
		goto out_free;

1492 1493
	if (inarg->offset + data_size <= size) {
		inarg->size = data_size;
M
Miklos Szeredi 已提交
1494
	} else if (inarg->offset < size) {
1495
		inarg->size = size - inarg->offset;
M
Miklos Szeredi 已提交
1496 1497 1498
	} else {
		/* Got truncated off completely */
		goto out_free;
M
Miklos Szeredi 已提交
1499
	}
M
Miklos Szeredi 已提交
1500 1501 1502

	req->in.args[1].size = inarg->size;
	fi->writectr++;
1503
	fuse_request_send_background_locked(fc, req);
M
Miklos Szeredi 已提交
1504 1505 1506 1507 1508 1509
	return;

 out_free:
	fuse_writepage_finish(fc, req);
	spin_unlock(&fc->lock);
	fuse_writepage_free(fc, req);
1510
	fuse_put_request(fc, req);
M
Miklos Szeredi 已提交
1511
	spin_lock(&fc->lock);
M
Miklos Szeredi 已提交
1512 1513
}

M
Miklos Szeredi 已提交
1514 1515 1516 1517 1518 1519 1520
/*
 * 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)
M
Miklos Szeredi 已提交
1521 1522
__releases(fc->lock)
__acquires(fc->lock)
M
Miklos Szeredi 已提交
1523
{
M
Miklos Szeredi 已提交
1524 1525
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
1526
	size_t crop = i_size_read(inode);
M
Miklos Szeredi 已提交
1527 1528 1529 1530 1531
	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);
1532
		fuse_send_writepage(fc, req, crop);
M
Miklos Szeredi 已提交
1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
	}
}

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);
1543
	while (req->misc.write.next) {
1544 1545
		struct fuse_conn *fc = get_fuse_conn(inode);
		struct fuse_write_in *inarg = &req->misc.write.in;
1546 1547 1548
		struct fuse_req *next = req->misc.write.next;
		req->misc.write.next = next->misc.write.next;
		next->misc.write.next = NULL;
1549
		next->ff = fuse_file_get(req->ff);
1550
		list_add(&next->writepages_entry, &fi->writepages);
1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575

		/*
		 * Skip fuse_flush_writepages() to make it easy to crop requests
		 * based on primary request size.
		 *
		 * 1st case (trivial): there are no concurrent activities using
		 * fuse_set/release_nowrite.  Then we're on safe side because
		 * fuse_flush_writepages() would call fuse_send_writepage()
		 * anyway.
		 *
		 * 2nd case: someone called fuse_set_nowrite and it is waiting
		 * now for completion of all in-flight requests.  This happens
		 * rarely and no more than once per page, so this should be
		 * okay.
		 *
		 * 3rd case: someone (e.g. fuse_do_setattr()) is in the middle
		 * of fuse_set_nowrite..fuse_release_nowrite section.  The fact
		 * that fuse_set_nowrite returned implies that all in-flight
		 * requests were completed along with all of their secondary
		 * requests.  Further primary requests are blocked by negative
		 * writectr.  Hence there cannot be any in-flight requests and
		 * no invocations of fuse_writepage_end() while we're in
		 * fuse_set_nowrite..fuse_release_nowrite section.
		 */
		fuse_send_writepage(fc, next, inarg->offset + inarg->size);
1576
	}
M
Miklos Szeredi 已提交
1577 1578 1579 1580 1581 1582
	fi->writectr--;
	fuse_writepage_finish(fc, req);
	spin_unlock(&fc->lock);
	fuse_writepage_free(fc, req);
}

M
Miklos Szeredi 已提交
1583 1584
static struct fuse_file *__fuse_write_file_get(struct fuse_conn *fc,
					       struct fuse_inode *fi)
1585
{
M
Miklos Szeredi 已提交
1586
	struct fuse_file *ff = NULL;
1587 1588

	spin_lock(&fc->lock);
M
Miklos Szeredi 已提交
1589
	if (!list_empty(&fi->write_files)) {
M
Miklos Szeredi 已提交
1590 1591 1592 1593
		ff = list_entry(fi->write_files.next, struct fuse_file,
				write_entry);
		fuse_file_get(ff);
	}
1594 1595 1596 1597 1598
	spin_unlock(&fc->lock);

	return ff;
}

M
Miklos Szeredi 已提交
1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614
static struct fuse_file *fuse_write_file_get(struct fuse_conn *fc,
					     struct fuse_inode *fi)
{
	struct fuse_file *ff = __fuse_write_file_get(fc, fi);
	WARN_ON(!ff);
	return ff;
}

int fuse_write_inode(struct inode *inode, struct writeback_control *wbc)
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct fuse_file *ff;
	int err;

	ff = __fuse_write_file_get(fc, fi);
1615
	err = fuse_flush_times(inode, ff);
M
Miklos Szeredi 已提交
1616 1617 1618 1619 1620 1621
	if (ff)
		fuse_file_put(ff, 0);

	return err;
}

M
Miklos Szeredi 已提交
1622 1623 1624 1625 1626 1627 1628 1629
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 page *tmp_page;
M
Miklos Szeredi 已提交
1630
	int error = -ENOMEM;
M
Miklos Szeredi 已提交
1631 1632 1633

	set_page_writeback(page);

1634
	req = fuse_request_alloc_nofs(1);
M
Miklos Szeredi 已提交
1635 1636 1637
	if (!req)
		goto err;

1638
	req->background = 1; /* writeback always goes to bg_queue */
M
Miklos Szeredi 已提交
1639 1640 1641 1642
	tmp_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
	if (!tmp_page)
		goto err_free;

M
Miklos Szeredi 已提交
1643
	error = -EIO;
1644
	req->ff = fuse_write_file_get(fc, fi);
M
Miklos Szeredi 已提交
1645
	if (!req->ff)
1646
		goto err_nofile;
M
Miklos Szeredi 已提交
1647

1648
	fuse_write_fill(req, req->ff, page_offset(page), 0);
M
Miklos Szeredi 已提交
1649 1650

	copy_highpage(tmp_page, page);
1651
	req->misc.write.in.write_flags |= FUSE_WRITE_CACHE;
1652
	req->misc.write.next = NULL;
1653
	req->in.argpages = 1;
M
Miklos Szeredi 已提交
1654 1655
	req->num_pages = 1;
	req->pages[0] = tmp_page;
1656
	req->page_descs[0].offset = 0;
1657
	req->page_descs[0].length = PAGE_SIZE;
M
Miklos Szeredi 已提交
1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669
	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);

	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);

1670 1671
	end_page_writeback(page);

M
Miklos Szeredi 已提交
1672 1673
	return 0;

1674 1675
err_nofile:
	__free_page(tmp_page);
M
Miklos Szeredi 已提交
1676 1677 1678 1679
err_free:
	fuse_request_free(req);
err:
	end_page_writeback(page);
M
Miklos Szeredi 已提交
1680
	return error;
M
Miklos Szeredi 已提交
1681 1682 1683 1684 1685 1686
}

static int fuse_writepage(struct page *page, struct writeback_control *wbc)
{
	int err;

1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
	if (fuse_page_is_writeback(page->mapping->host, page->index)) {
		/*
		 * ->writepages() should be called for sync() and friends.  We
		 * should only get here on direct reclaim and then we are
		 * allowed to skip a page which is already in flight
		 */
		WARN_ON(wbc->sync_mode == WB_SYNC_ALL);

		redirty_page_for_writepage(wbc, page);
		return 0;
	}

M
Miklos Szeredi 已提交
1699 1700 1701 1702 1703 1704
	err = fuse_writepage_locked(page);
	unlock_page(page);

	return err;
}

1705 1706 1707 1708
struct fuse_fill_wb_data {
	struct fuse_req *req;
	struct fuse_file *ff;
	struct inode *inode;
1709
	struct page **orig_pages;
1710 1711 1712 1713 1714 1715 1716 1717
};

static void fuse_writepages_send(struct fuse_fill_wb_data *data)
{
	struct fuse_req *req = data->req;
	struct inode *inode = data->inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
1718 1719
	int num_pages = req->num_pages;
	int i;
1720 1721 1722 1723 1724 1725

	req->ff = fuse_file_get(data->ff);
	spin_lock(&fc->lock);
	list_add_tail(&req->list, &fi->queued_writes);
	fuse_flush_writepages(inode);
	spin_unlock(&fc->lock);
1726 1727 1728

	for (i = 0; i < num_pages; i++)
		end_page_writeback(data->orig_pages[i]);
1729 1730
}

1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
static bool fuse_writepage_in_flight(struct fuse_req *new_req,
				     struct page *page)
{
	struct fuse_conn *fc = get_fuse_conn(new_req->inode);
	struct fuse_inode *fi = get_fuse_inode(new_req->inode);
	struct fuse_req *tmp;
	struct fuse_req *old_req;
	bool found = false;
	pgoff_t curr_index;

	BUG_ON(new_req->num_pages != 0);

	spin_lock(&fc->lock);
	list_del(&new_req->writepages_entry);
	list_for_each_entry(old_req, &fi->writepages, writepages_entry) {
		BUG_ON(old_req->inode != new_req->inode);
		curr_index = old_req->misc.write.in.offset >> PAGE_CACHE_SHIFT;
		if (curr_index <= page->index &&
		    page->index < curr_index + old_req->num_pages) {
			found = true;
			break;
		}
	}
1754 1755
	if (!found) {
		list_add(&new_req->writepages_entry, &fi->writepages);
1756
		goto out_unlock;
1757
	}
1758

1759
	new_req->num_pages = 1;
1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770
	for (tmp = old_req; tmp != NULL; tmp = tmp->misc.write.next) {
		BUG_ON(tmp->inode != new_req->inode);
		curr_index = tmp->misc.write.in.offset >> PAGE_CACHE_SHIFT;
		if (tmp->num_pages == 1 &&
		    curr_index == page->index) {
			old_req = tmp;
		}
	}

	if (old_req->num_pages == 1 && (old_req->state == FUSE_REQ_INIT ||
					old_req->state == FUSE_REQ_PENDING)) {
1771 1772
		struct backing_dev_info *bdi = page->mapping->backing_dev_info;

1773 1774 1775
		copy_highpage(old_req->pages[0], page);
		spin_unlock(&fc->lock);

1776
		dec_bdi_stat(bdi, BDI_WRITEBACK);
1777
		dec_zone_page_state(page, NR_WRITEBACK_TEMP);
1778
		bdi_writeout_inc(bdi);
1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791
		fuse_writepage_free(fc, new_req);
		fuse_request_free(new_req);
		goto out;
	} else {
		new_req->misc.write.next = old_req->misc.write.next;
		old_req->misc.write.next = new_req;
	}
out_unlock:
	spin_unlock(&fc->lock);
out:
	return found;
}

1792 1793 1794 1795 1796 1797 1798 1799
static int fuse_writepages_fill(struct page *page,
		struct writeback_control *wbc, void *_data)
{
	struct fuse_fill_wb_data *data = _data;
	struct fuse_req *req = data->req;
	struct inode *inode = data->inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct page *tmp_page;
1800
	bool is_writeback;
1801 1802 1803 1804 1805 1806 1807 1808 1809
	int err;

	if (!data->ff) {
		err = -EIO;
		data->ff = fuse_write_file_get(fc, get_fuse_inode(inode));
		if (!data->ff)
			goto out_unlock;
	}

1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823
	/*
	 * Being under writeback is unlikely but possible.  For example direct
	 * read to an mmaped fuse file will set the page dirty twice; once when
	 * the pages are faulted with get_user_pages(), and then after the read
	 * completed.
	 */
	is_writeback = fuse_page_is_writeback(inode, page->index);

	if (req && req->num_pages &&
	    (is_writeback || req->num_pages == FUSE_MAX_PAGES_PER_REQ ||
	     (req->num_pages + 1) * PAGE_CACHE_SIZE > fc->max_write ||
	     data->orig_pages[req->num_pages - 1]->index + 1 != page->index)) {
		fuse_writepages_send(data);
		data->req = NULL;
1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
	}
	err = -ENOMEM;
	tmp_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
	if (!tmp_page)
		goto out_unlock;

	/*
	 * The page must not be redirtied until the writeout is completed
	 * (i.e. userspace has sent a reply to the write request).  Otherwise
	 * there could be more than one temporary page instance for each real
	 * page.
	 *
	 * This is ensured by holding the page lock in page_mkwrite() while
	 * checking fuse_page_is_writeback().  We already hold the page lock
	 * since clear_page_dirty_for_io() and keep it held until we add the
	 * request to the fi->writepages list and increment req->num_pages.
	 * After this fuse_page_is_writeback() will indicate that the page is
	 * under writeback, so we can release the page lock.
	 */
	if (data->req == NULL) {
		struct fuse_inode *fi = get_fuse_inode(inode);

		err = -ENOMEM;
		req = fuse_request_alloc_nofs(FUSE_MAX_PAGES_PER_REQ);
		if (!req) {
			__free_page(tmp_page);
			goto out_unlock;
		}

		fuse_write_fill(req, data->ff, page_offset(page), 0);
		req->misc.write.in.write_flags |= FUSE_WRITE_CACHE;
1855
		req->misc.write.next = NULL;
1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876
		req->in.argpages = 1;
		req->background = 1;
		req->num_pages = 0;
		req->end = fuse_writepage_end;
		req->inode = inode;

		spin_lock(&fc->lock);
		list_add(&req->writepages_entry, &fi->writepages);
		spin_unlock(&fc->lock);

		data->req = req;
	}
	set_page_writeback(page);

	copy_highpage(tmp_page, page);
	req->pages[req->num_pages] = tmp_page;
	req->page_descs[req->num_pages].offset = 0;
	req->page_descs[req->num_pages].length = PAGE_SIZE;

	inc_bdi_stat(page->mapping->backing_dev_info, BDI_WRITEBACK);
	inc_zone_page_state(tmp_page, NR_WRITEBACK_TEMP);
1877 1878 1879 1880 1881 1882 1883

	err = 0;
	if (is_writeback && fuse_writepage_in_flight(req, page)) {
		end_page_writeback(page);
		data->req = NULL;
		goto out_unlock;
	}
1884
	data->orig_pages[req->num_pages] = page;
1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914

	/*
	 * Protected by fc->lock against concurrent access by
	 * fuse_page_is_writeback().
	 */
	spin_lock(&fc->lock);
	req->num_pages++;
	spin_unlock(&fc->lock);

out_unlock:
	unlock_page(page);

	return err;
}

static int fuse_writepages(struct address_space *mapping,
			   struct writeback_control *wbc)
{
	struct inode *inode = mapping->host;
	struct fuse_fill_wb_data data;
	int err;

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

	data.inode = inode;
	data.req = NULL;
	data.ff = NULL;

1915
	err = -ENOMEM;
1916 1917
	data.orig_pages = kcalloc(FUSE_MAX_PAGES_PER_REQ,
				  sizeof(struct page *),
1918 1919 1920 1921
				  GFP_NOFS);
	if (!data.orig_pages)
		goto out;

1922 1923 1924 1925 1926 1927 1928 1929 1930
	err = write_cache_pages(mapping, wbc, fuse_writepages_fill, &data);
	if (data.req) {
		/* Ignore errors if we can write at least one page */
		BUG_ON(!data.req->num_pages);
		fuse_writepages_send(&data);
		err = 0;
	}
	if (data.ff)
		fuse_file_put(data.ff, false);
1931 1932

	kfree(data.orig_pages);
1933 1934 1935 1936
out:
	return err;
}

1937 1938 1939 1940 1941 1942 1943 1944 1945
/*
 * It's worthy to make sure that space is reserved on disk for the write,
 * but how to implement it without killing performance need more thinking.
 */
static int fuse_write_begin(struct file *file, struct address_space *mapping,
		loff_t pos, unsigned len, unsigned flags,
		struct page **pagep, void **fsdata)
{
	pgoff_t index = pos >> PAGE_CACHE_SHIFT;
A
Al Viro 已提交
1946
	struct fuse_conn *fc = get_fuse_conn(file_inode(file));
1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
	struct page *page;
	loff_t fsize;
	int err = -ENOMEM;

	WARN_ON(!fc->writeback_cache);

	page = grab_cache_page_write_begin(mapping, index, flags);
	if (!page)
		goto error;

	fuse_wait_on_page_writeback(mapping->host, page->index);

	if (PageUptodate(page) || len == PAGE_CACHE_SIZE)
		goto success;
	/*
	 * Check if the start this page comes after the end of file, in which
	 * case the readpage can be optimized away.
	 */
	fsize = i_size_read(mapping->host);
	if (fsize <= (pos & PAGE_CACHE_MASK)) {
		size_t off = pos & ~PAGE_CACHE_MASK;
		if (off)
			zero_user_segment(page, 0, off);
		goto success;
	}
	err = fuse_do_readpage(file, page);
	if (err)
		goto cleanup;
success:
	*pagep = page;
	return 0;

cleanup:
	unlock_page(page);
	page_cache_release(page);
error:
	return err;
}

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 = page->mapping->host;

	if (!PageUptodate(page)) {
		/* Zero any unwritten bytes at the end of the page */
		size_t endoff = (pos + copied) & ~PAGE_CACHE_MASK;
		if (endoff)
			zero_user_segment(page, endoff, PAGE_CACHE_SIZE);
		SetPageUptodate(page);
	}

	fuse_write_update_size(inode, pos + copied);
	set_page_dirty(page);
	unlock_page(page);
	page_cache_release(page);

	return copied;
}

M
Miklos Szeredi 已提交
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043
static int fuse_launder_page(struct page *page)
{
	int err = 0;
	if (clear_page_dirty_for_io(page)) {
		struct inode *inode = page->mapping->host;
		err = fuse_writepage_locked(page);
		if (!err)
			fuse_wait_on_page_writeback(inode, page->index);
	}
	return err;
}

/*
 * Write back dirty pages now, because there may not be any suitable
 * open files later
 */
static void fuse_vma_close(struct vm_area_struct *vma)
{
	filemap_write_and_wait(vma->vm_file->f_mapping);
}

/*
 * Wait for writeback against this page to complete before allowing it
 * to be marked dirty again, and hence written back again, possibly
 * before the previous writepage completed.
 *
 * Block here, instead of in ->writepage(), so that the userspace fs
 * can only block processes actually operating on the filesystem.
 *
 * Otherwise unprivileged userspace fs would be able to block
 * unrelated:
 *
 * - page migration
 * - sync(2)
 * - try_to_free_pages() with order > PAGE_ALLOC_COSTLY_ORDER
 */
2044
static int fuse_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
M
Miklos Szeredi 已提交
2045
{
2046
	struct page *page = vmf->page;
M
Miklos Szeredi 已提交
2047 2048 2049 2050 2051 2052 2053 2054
	struct inode *inode = file_inode(vma->vm_file);

	file_update_time(vma->vm_file);
	lock_page(page);
	if (page->mapping != inode->i_mapping) {
		unlock_page(page);
		return VM_FAULT_NOPAGE;
	}
M
Miklos Szeredi 已提交
2055 2056

	fuse_wait_on_page_writeback(inode, page->index);
M
Miklos Szeredi 已提交
2057
	return VM_FAULT_LOCKED;
M
Miklos Szeredi 已提交
2058 2059
}

2060
static const struct vm_operations_struct fuse_file_vm_ops = {
M
Miklos Szeredi 已提交
2061 2062
	.close		= fuse_vma_close,
	.fault		= filemap_fault,
2063
	.map_pages	= filemap_map_pages,
M
Miklos Szeredi 已提交
2064 2065 2066 2067 2068
	.page_mkwrite	= fuse_page_mkwrite,
};

static int fuse_file_mmap(struct file *file, struct vm_area_struct *vma)
{
2069 2070 2071
	if ((vma->vm_flags & VM_SHARED) && (vma->vm_flags & VM_MAYWRITE))
		fuse_link_write_file(file);

M
Miklos Szeredi 已提交
2072 2073
	file_accessed(file);
	vma->vm_ops = &fuse_file_vm_ops;
M
Miklos Szeredi 已提交
2074 2075 2076
	return 0;
}

2077 2078 2079 2080 2081 2082
static int fuse_direct_mmap(struct file *file, struct vm_area_struct *vma)
{
	/* Can't provide the coherency needed for MAP_SHARED */
	if (vma->vm_flags & VM_MAYSHARE)
		return -ENODEV;

2083 2084
	invalidate_inode_pages2(file->f_mapping);

2085 2086 2087
	return generic_file_mmap(file, vma);
}

2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112
static int convert_fuse_file_lock(const struct fuse_file_lock *ffl,
				  struct file_lock *fl)
{
	switch (ffl->type) {
	case F_UNLCK:
		break;

	case F_RDLCK:
	case F_WRLCK:
		if (ffl->start > OFFSET_MAX || ffl->end > OFFSET_MAX ||
		    ffl->end < ffl->start)
			return -EIO;

		fl->fl_start = ffl->start;
		fl->fl_end = ffl->end;
		fl->fl_pid = ffl->pid;
		break;

	default:
		return -EIO;
	}
	fl->fl_type = ffl->type;
	return 0;
}

2113
static void fuse_lk_fill(struct fuse_args *args, struct file *file,
2114
			 const struct file_lock *fl, int opcode, pid_t pid,
2115
			 int flock, struct fuse_lk_in *inarg)
2116
{
A
Al Viro 已提交
2117
	struct inode *inode = file_inode(file);
2118
	struct fuse_conn *fc = get_fuse_conn(inode);
2119
	struct fuse_file *ff = file->private_data;
2120 2121 2122 2123 2124 2125 2126 2127

	memset(inarg, 0, sizeof(*inarg));
	inarg->fh = ff->fh;
	inarg->owner = fuse_lock_owner_id(fc, fl->fl_owner);
	inarg->lk.start = fl->fl_start;
	inarg->lk.end = fl->fl_end;
	inarg->lk.type = fl->fl_type;
	inarg->lk.pid = pid;
2128
	if (flock)
2129 2130 2131 2132 2133 2134
		inarg->lk_flags |= FUSE_LK_FLOCK;
	args->in.h.opcode = opcode;
	args->in.h.nodeid = get_node_id(inode);
	args->in.numargs = 1;
	args->in.args[0].size = sizeof(*inarg);
	args->in.args[0].value = inarg;
2135 2136 2137 2138
}

static int fuse_getlk(struct file *file, struct file_lock *fl)
{
A
Al Viro 已提交
2139
	struct inode *inode = file_inode(file);
2140
	struct fuse_conn *fc = get_fuse_conn(inode);
2141 2142
	FUSE_ARGS(args);
	struct fuse_lk_in inarg;
2143 2144 2145
	struct fuse_lk_out outarg;
	int err;

2146 2147 2148 2149 2150
	fuse_lk_fill(&args, file, fl, FUSE_GETLK, 0, 0, &inarg);
	args.out.numargs = 1;
	args.out.args[0].size = sizeof(outarg);
	args.out.args[0].value = &outarg;
	err = fuse_simple_request(fc, &args);
2151 2152 2153 2154 2155 2156
	if (!err)
		err = convert_fuse_file_lock(&outarg.lk, fl);

	return err;
}

2157
static int fuse_setlk(struct file *file, struct file_lock *fl, int flock)
2158
{
A
Al Viro 已提交
2159
	struct inode *inode = file_inode(file);
2160
	struct fuse_conn *fc = get_fuse_conn(inode);
2161 2162
	FUSE_ARGS(args);
	struct fuse_lk_in inarg;
2163 2164 2165 2166
	int opcode = (fl->fl_flags & FL_SLEEP) ? FUSE_SETLKW : FUSE_SETLK;
	pid_t pid = fl->fl_type != F_UNLCK ? current->tgid : 0;
	int err;

J
J. Bruce Fields 已提交
2167
	if (fl->fl_lmops && fl->fl_lmops->lm_grant) {
M
Miklos Szeredi 已提交
2168 2169 2170 2171
		/* NLM needs asynchronous locks, which we don't support yet */
		return -ENOLCK;
	}

2172 2173 2174 2175
	/* Unlock on close is handled by the flush method */
	if (fl->fl_flags & FL_CLOSE)
		return 0;

2176 2177
	fuse_lk_fill(&args, file, fl, opcode, pid, flock, &inarg);
	err = fuse_simple_request(fc, &args);
2178

2179 2180 2181
	/* locking is restartable */
	if (err == -EINTR)
		err = -ERESTARTSYS;
2182

2183 2184 2185 2186 2187
	return err;
}

static int fuse_file_lock(struct file *file, int cmd, struct file_lock *fl)
{
A
Al Viro 已提交
2188
	struct inode *inode = file_inode(file);
2189 2190 2191
	struct fuse_conn *fc = get_fuse_conn(inode);
	int err;

M
Miklos Szeredi 已提交
2192 2193 2194
	if (cmd == F_CANCELLK) {
		err = 0;
	} else if (cmd == F_GETLK) {
2195
		if (fc->no_lock) {
2196
			posix_test_lock(file, fl);
2197 2198 2199 2200 2201
			err = 0;
		} else
			err = fuse_getlk(file, fl);
	} else {
		if (fc->no_lock)
M
Miklos Szeredi 已提交
2202
			err = posix_lock_file(file, fl, NULL);
2203
		else
2204
			err = fuse_setlk(file, fl, 0);
2205 2206 2207 2208
	}
	return err;
}

2209 2210
static int fuse_file_flock(struct file *file, int cmd, struct file_lock *fl)
{
A
Al Viro 已提交
2211
	struct inode *inode = file_inode(file);
2212 2213 2214
	struct fuse_conn *fc = get_fuse_conn(inode);
	int err;

M
Miklos Szeredi 已提交
2215
	if (fc->no_flock) {
2216 2217
		err = flock_lock_file_wait(file, fl);
	} else {
M
Miklos Szeredi 已提交
2218 2219
		struct fuse_file *ff = file->private_data;

2220
		/* emulate flock with POSIX locks */
M
Miklos Szeredi 已提交
2221
		ff->flock = true;
2222 2223 2224 2225 2226 2227
		err = fuse_setlk(file, fl, 1);
	}

	return err;
}

M
Miklos Szeredi 已提交
2228 2229 2230 2231
static sector_t fuse_bmap(struct address_space *mapping, sector_t block)
{
	struct inode *inode = mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
2232
	FUSE_ARGS(args);
M
Miklos Szeredi 已提交
2233 2234 2235 2236 2237 2238 2239 2240 2241 2242
	struct fuse_bmap_in inarg;
	struct fuse_bmap_out outarg;
	int err;

	if (!inode->i_sb->s_bdev || fc->no_bmap)
		return 0;

	memset(&inarg, 0, sizeof(inarg));
	inarg.block = block;
	inarg.blocksize = inode->i_sb->s_blocksize;
2243 2244 2245 2246 2247 2248 2249 2250 2251
	args.in.h.opcode = FUSE_BMAP;
	args.in.h.nodeid = get_node_id(inode);
	args.in.numargs = 1;
	args.in.args[0].size = sizeof(inarg);
	args.in.args[0].value = &inarg;
	args.out.numargs = 1;
	args.out.args[0].size = sizeof(outarg);
	args.out.args[0].value = &outarg;
	err = fuse_simple_request(fc, &args);
M
Miklos Szeredi 已提交
2252 2253 2254 2255 2256 2257
	if (err == -ENOSYS)
		fc->no_bmap = 1;

	return err ? 0 : outarg.block;
}

2258
static loff_t fuse_file_llseek(struct file *file, loff_t offset, int whence)
M
Miklos Szeredi 已提交
2259 2260
{
	loff_t retval;
A
Al Viro 已提交
2261
	struct inode *inode = file_inode(file);
M
Miklos Szeredi 已提交
2262

2263
	/* No i_mutex protection necessary for SEEK_CUR and SEEK_SET */
2264 2265
	if (whence == SEEK_CUR || whence == SEEK_SET)
		return generic_file_llseek(file, offset, whence);
2266

2267 2268 2269
	mutex_lock(&inode->i_mutex);
	retval = fuse_update_attributes(inode, NULL, file, NULL);
	if (!retval)
2270
		retval = generic_file_llseek(file, offset, whence);
M
Miklos Szeredi 已提交
2271
	mutex_unlock(&inode->i_mutex);
2272

M
Miklos Szeredi 已提交
2273 2274 2275
	return retval;
}

T
Tejun Heo 已提交
2276 2277 2278 2279 2280 2281 2282 2283 2284
static int fuse_ioctl_copy_user(struct page **pages, struct iovec *iov,
			unsigned int nr_segs, size_t bytes, bool to_user)
{
	struct iov_iter ii;
	int page_idx = 0;

	if (!bytes)
		return 0;

A
Al Viro 已提交
2285
	iov_iter_init(&ii, to_user ? READ : WRITE, iov, nr_segs, bytes);
T
Tejun Heo 已提交
2286 2287 2288 2289

	while (iov_iter_count(&ii)) {
		struct page *page = pages[page_idx++];
		size_t todo = min_t(size_t, PAGE_SIZE, iov_iter_count(&ii));
D
Dan Carpenter 已提交
2290
		void *kaddr;
T
Tejun Heo 已提交
2291

D
Dan Carpenter 已提交
2292
		kaddr = kmap(page);
T
Tejun Heo 已提交
2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312

		while (todo) {
			char __user *uaddr = ii.iov->iov_base + ii.iov_offset;
			size_t iov_len = ii.iov->iov_len - ii.iov_offset;
			size_t copy = min(todo, iov_len);
			size_t left;

			if (!to_user)
				left = copy_from_user(kaddr, uaddr, copy);
			else
				left = copy_to_user(uaddr, kaddr, copy);

			if (unlikely(left))
				return -EFAULT;

			iov_iter_advance(&ii, copy);
			todo -= copy;
			kaddr += copy;
		}

2313
		kunmap(page);
T
Tejun Heo 已提交
2314 2315 2316 2317 2318
	}

	return 0;
}

2319 2320 2321 2322 2323 2324
/*
 * CUSE servers compiled on 32bit broke on 64bit kernels because the
 * ABI was defined to be 'struct iovec' which is different on 32bit
 * and 64bit.  Fortunately we can determine which structure the server
 * used from the size of the reply.
 */
M
Miklos Szeredi 已提交
2325 2326 2327
static int fuse_copy_ioctl_iovec_old(struct iovec *dst, void *src,
				     size_t transferred, unsigned count,
				     bool is_compat)
2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356
{
#ifdef CONFIG_COMPAT
	if (count * sizeof(struct compat_iovec) == transferred) {
		struct compat_iovec *ciov = src;
		unsigned i;

		/*
		 * With this interface a 32bit server cannot support
		 * non-compat (i.e. ones coming from 64bit apps) ioctl
		 * requests
		 */
		if (!is_compat)
			return -EINVAL;

		for (i = 0; i < count; i++) {
			dst[i].iov_base = compat_ptr(ciov[i].iov_base);
			dst[i].iov_len = ciov[i].iov_len;
		}
		return 0;
	}
#endif

	if (count * sizeof(struct iovec) != transferred)
		return -EIO;

	memcpy(dst, src, transferred);
	return 0;
}

M
Miklos Szeredi 已提交
2357 2358 2359 2360 2361 2362
/* Make sure iov_length() won't overflow */
static int fuse_verify_ioctl_iov(struct iovec *iov, size_t count)
{
	size_t n;
	u32 max = FUSE_MAX_PAGES_PER_REQ << PAGE_SHIFT;

2363
	for (n = 0; n < count; n++, iov++) {
M
Miklos Szeredi 已提交
2364 2365 2366 2367 2368 2369 2370
		if (iov->iov_len > (size_t) max)
			return -ENOMEM;
		max -= iov->iov_len;
	}
	return 0;
}

M
Miklos Szeredi 已提交
2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406
static int fuse_copy_ioctl_iovec(struct fuse_conn *fc, struct iovec *dst,
				 void *src, size_t transferred, unsigned count,
				 bool is_compat)
{
	unsigned i;
	struct fuse_ioctl_iovec *fiov = src;

	if (fc->minor < 16) {
		return fuse_copy_ioctl_iovec_old(dst, src, transferred,
						 count, is_compat);
	}

	if (count * sizeof(struct fuse_ioctl_iovec) != transferred)
		return -EIO;

	for (i = 0; i < count; i++) {
		/* Did the server supply an inappropriate value? */
		if (fiov[i].base != (unsigned long) fiov[i].base ||
		    fiov[i].len != (unsigned long) fiov[i].len)
			return -EIO;

		dst[i].iov_base = (void __user *) (unsigned long) fiov[i].base;
		dst[i].iov_len = (size_t) fiov[i].len;

#ifdef CONFIG_COMPAT
		if (is_compat &&
		    (ptr_to_compat(dst[i].iov_base) != fiov[i].base ||
		     (compat_size_t) dst[i].iov_len != fiov[i].len))
			return -EIO;
#endif
	}

	return 0;
}


T
Tejun Heo 已提交
2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452
/*
 * For ioctls, there is no generic way to determine how much memory
 * needs to be read and/or written.  Furthermore, ioctls are allowed
 * to dereference the passed pointer, so the parameter requires deep
 * copying but FUSE has no idea whatsoever about what to copy in or
 * out.
 *
 * This is solved by allowing FUSE server to retry ioctl with
 * necessary in/out iovecs.  Let's assume the ioctl implementation
 * needs to read in the following structure.
 *
 * struct a {
 *	char	*buf;
 *	size_t	buflen;
 * }
 *
 * On the first callout to FUSE server, inarg->in_size and
 * inarg->out_size will be NULL; then, the server completes the ioctl
 * with FUSE_IOCTL_RETRY set in out->flags, out->in_iovs set to 1 and
 * the actual iov array to
 *
 * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a) } }
 *
 * which tells FUSE to copy in the requested area and retry the ioctl.
 * On the second round, the server has access to the structure and
 * from that it can tell what to look for next, so on the invocation,
 * it sets FUSE_IOCTL_RETRY, out->in_iovs to 2 and iov array to
 *
 * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a)	},
 *   { .iov_base = a.buf,	.iov_len = a.buflen		} }
 *
 * FUSE will copy both struct a and the pointed buffer from the
 * process doing the ioctl and retry ioctl with both struct a and the
 * buffer.
 *
 * This time, FUSE server has everything it needs and completes ioctl
 * without FUSE_IOCTL_RETRY which finishes the ioctl call.
 *
 * Copying data out works the same way.
 *
 * Note that if FUSE_IOCTL_UNRESTRICTED is clear, the kernel
 * automatically initializes in and out iovs by decoding @cmd with
 * _IOC_* macros and the server is not allowed to request RETRY.  This
 * limits ioctl data transfers to well-formed ioctls and is the forced
 * behavior for all FUSE servers.
 */
2453 2454
long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
		   unsigned int flags)
T
Tejun Heo 已提交
2455 2456
{
	struct fuse_file *ff = file->private_data;
2457
	struct fuse_conn *fc = ff->fc;
T
Tejun Heo 已提交
2458 2459 2460 2461 2462 2463 2464 2465 2466
	struct fuse_ioctl_in inarg = {
		.fh = ff->fh,
		.cmd = cmd,
		.arg = arg,
		.flags = flags
	};
	struct fuse_ioctl_out outarg;
	struct fuse_req *req = NULL;
	struct page **pages = NULL;
M
Miklos Szeredi 已提交
2467
	struct iovec *iov_page = NULL;
T
Tejun Heo 已提交
2468 2469 2470 2471 2472
	struct iovec *in_iov = NULL, *out_iov = NULL;
	unsigned int in_iovs = 0, out_iovs = 0, num_pages = 0, max_pages;
	size_t in_size, out_size, transferred;
	int err;

M
Miklos Szeredi 已提交
2473 2474 2475 2476 2477 2478 2479
#if BITS_PER_LONG == 32
	inarg.flags |= FUSE_IOCTL_32BIT;
#else
	if (flags & FUSE_IOCTL_COMPAT)
		inarg.flags |= FUSE_IOCTL_32BIT;
#endif

T
Tejun Heo 已提交
2480
	/* assume all the iovs returned by client always fits in a page */
M
Miklos Szeredi 已提交
2481
	BUILD_BUG_ON(sizeof(struct fuse_ioctl_iovec) * FUSE_IOCTL_MAX_IOV > PAGE_SIZE);
T
Tejun Heo 已提交
2482 2483

	err = -ENOMEM;
2484
	pages = kcalloc(FUSE_MAX_PAGES_PER_REQ, sizeof(pages[0]), GFP_KERNEL);
M
Miklos Szeredi 已提交
2485
	iov_page = (struct iovec *) __get_free_page(GFP_KERNEL);
T
Tejun Heo 已提交
2486 2487 2488 2489 2490 2491 2492 2493
	if (!pages || !iov_page)
		goto out;

	/*
	 * If restricted, initialize IO parameters as encoded in @cmd.
	 * RETRY from server is not allowed.
	 */
	if (!(flags & FUSE_IOCTL_UNRESTRICTED)) {
M
Miklos Szeredi 已提交
2494
		struct iovec *iov = iov_page;
T
Tejun Heo 已提交
2495

M
Miklos Szeredi 已提交
2496
		iov->iov_base = (void __user *)arg;
T
Tejun Heo 已提交
2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531
		iov->iov_len = _IOC_SIZE(cmd);

		if (_IOC_DIR(cmd) & _IOC_WRITE) {
			in_iov = iov;
			in_iovs = 1;
		}

		if (_IOC_DIR(cmd) & _IOC_READ) {
			out_iov = iov;
			out_iovs = 1;
		}
	}

 retry:
	inarg.in_size = in_size = iov_length(in_iov, in_iovs);
	inarg.out_size = out_size = iov_length(out_iov, out_iovs);

	/*
	 * Out data can be used either for actual out data or iovs,
	 * make sure there always is at least one page.
	 */
	out_size = max_t(size_t, out_size, PAGE_SIZE);
	max_pages = DIV_ROUND_UP(max(in_size, out_size), PAGE_SIZE);

	/* make sure there are enough buffer pages and init request with them */
	err = -ENOMEM;
	if (max_pages > FUSE_MAX_PAGES_PER_REQ)
		goto out;
	while (num_pages < max_pages) {
		pages[num_pages] = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
		if (!pages[num_pages])
			goto out;
		num_pages++;
	}

M
Maxim Patlasov 已提交
2532
	req = fuse_get_req(fc, num_pages);
T
Tejun Heo 已提交
2533 2534 2535 2536 2537 2538 2539
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		req = NULL;
		goto out;
	}
	memcpy(req->pages, pages, sizeof(req->pages[0]) * num_pages);
	req->num_pages = num_pages;
2540
	fuse_page_descs_length_init(req, 0, req->num_pages);
T
Tejun Heo 已提交
2541 2542 2543

	/* okay, let's send it to the client */
	req->in.h.opcode = FUSE_IOCTL;
2544
	req->in.h.nodeid = ff->nodeid;
T
Tejun Heo 已提交
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	if (in_size) {
		req->in.numargs++;
		req->in.args[1].size = in_size;
		req->in.argpages = 1;

		err = fuse_ioctl_copy_user(pages, in_iov, in_iovs, in_size,
					   false);
		if (err)
			goto out;
	}

	req->out.numargs = 2;
	req->out.args[0].size = sizeof(outarg);
	req->out.args[0].value = &outarg;
	req->out.args[1].size = out_size;
	req->out.argpages = 1;
	req->out.argvar = 1;

2566
	fuse_request_send(fc, req);
T
Tejun Heo 已提交
2567 2568 2569 2570 2571 2572 2573 2574 2575
	err = req->out.h.error;
	transferred = req->out.args[1].size;
	fuse_put_request(fc, req);
	req = NULL;
	if (err)
		goto out;

	/* did it ask for retry? */
	if (outarg.flags & FUSE_IOCTL_RETRY) {
M
Miklos Szeredi 已提交
2576
		void *vaddr;
T
Tejun Heo 已提交
2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595

		/* no retry if in restricted mode */
		err = -EIO;
		if (!(flags & FUSE_IOCTL_UNRESTRICTED))
			goto out;

		in_iovs = outarg.in_iovs;
		out_iovs = outarg.out_iovs;

		/*
		 * Make sure things are in boundary, separate checks
		 * are to protect against overflow.
		 */
		err = -ENOMEM;
		if (in_iovs > FUSE_IOCTL_MAX_IOV ||
		    out_iovs > FUSE_IOCTL_MAX_IOV ||
		    in_iovs + out_iovs > FUSE_IOCTL_MAX_IOV)
			goto out;

2596
		vaddr = kmap_atomic(pages[0]);
M
Miklos Szeredi 已提交
2597
		err = fuse_copy_ioctl_iovec(fc, iov_page, vaddr,
2598 2599
					    transferred, in_iovs + out_iovs,
					    (flags & FUSE_IOCTL_COMPAT) != 0);
2600
		kunmap_atomic(vaddr);
2601 2602
		if (err)
			goto out;
T
Tejun Heo 已提交
2603

M
Miklos Szeredi 已提交
2604
		in_iov = iov_page;
T
Tejun Heo 已提交
2605 2606
		out_iov = in_iov + in_iovs;

M
Miklos Szeredi 已提交
2607 2608 2609 2610 2611 2612 2613 2614
		err = fuse_verify_ioctl_iov(in_iov, in_iovs);
		if (err)
			goto out;

		err = fuse_verify_ioctl_iov(out_iov, out_iovs);
		if (err)
			goto out;

T
Tejun Heo 已提交
2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625
		goto retry;
	}

	err = -EIO;
	if (transferred > inarg.out_size)
		goto out;

	err = fuse_ioctl_copy_user(pages, out_iov, out_iovs, transferred, true);
 out:
	if (req)
		fuse_put_request(fc, req);
M
Miklos Szeredi 已提交
2626
	free_page((unsigned long) iov_page);
T
Tejun Heo 已提交
2627 2628 2629 2630 2631 2632
	while (num_pages)
		__free_page(pages[--num_pages]);
	kfree(pages);

	return err ? err : outarg.result;
}
2633
EXPORT_SYMBOL_GPL(fuse_do_ioctl);
T
Tejun Heo 已提交
2634

2635 2636
long fuse_ioctl_common(struct file *file, unsigned int cmd,
		       unsigned long arg, unsigned int flags)
2637
{
A
Al Viro 已提交
2638
	struct inode *inode = file_inode(file);
2639 2640
	struct fuse_conn *fc = get_fuse_conn(inode);

2641
	if (!fuse_allow_current_process(fc))
2642 2643 2644 2645 2646 2647 2648 2649
		return -EACCES;

	if (is_bad_inode(inode))
		return -EIO;

	return fuse_do_ioctl(file, cmd, arg, flags);
}

T
Tejun Heo 已提交
2650 2651 2652
static long fuse_file_ioctl(struct file *file, unsigned int cmd,
			    unsigned long arg)
{
2653
	return fuse_ioctl_common(file, cmd, arg, 0);
T
Tejun Heo 已提交
2654 2655 2656 2657 2658
}

static long fuse_file_compat_ioctl(struct file *file, unsigned int cmd,
				   unsigned long arg)
{
2659
	return fuse_ioctl_common(file, cmd, arg, FUSE_IOCTL_COMPAT);
T
Tejun Heo 已提交
2660 2661
}

T
Tejun Heo 已提交
2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702
/*
 * All files which have been polled are linked to RB tree
 * fuse_conn->polled_files which is indexed by kh.  Walk the tree and
 * find the matching one.
 */
static struct rb_node **fuse_find_polled_node(struct fuse_conn *fc, u64 kh,
					      struct rb_node **parent_out)
{
	struct rb_node **link = &fc->polled_files.rb_node;
	struct rb_node *last = NULL;

	while (*link) {
		struct fuse_file *ff;

		last = *link;
		ff = rb_entry(last, struct fuse_file, polled_node);

		if (kh < ff->kh)
			link = &last->rb_left;
		else if (kh > ff->kh)
			link = &last->rb_right;
		else
			return link;
	}

	if (parent_out)
		*parent_out = last;
	return link;
}

/*
 * The file is about to be polled.  Make sure it's on the polled_files
 * RB tree.  Note that files once added to the polled_files tree are
 * not removed before the file is released.  This is because a file
 * polled once is likely to be polled again.
 */
static void fuse_register_polled_file(struct fuse_conn *fc,
				      struct fuse_file *ff)
{
	spin_lock(&fc->lock);
	if (RB_EMPTY_NODE(&ff->polled_node)) {
2703
		struct rb_node **link, *uninitialized_var(parent);
T
Tejun Heo 已提交
2704 2705 2706 2707 2708 2709 2710 2711 2712

		link = fuse_find_polled_node(fc, ff->kh, &parent);
		BUG_ON(*link);
		rb_link_node(&ff->polled_node, parent, link);
		rb_insert_color(&ff->polled_node, &fc->polled_files);
	}
	spin_unlock(&fc->lock);
}

2713
unsigned fuse_file_poll(struct file *file, poll_table *wait)
T
Tejun Heo 已提交
2714 2715
{
	struct fuse_file *ff = file->private_data;
2716
	struct fuse_conn *fc = ff->fc;
T
Tejun Heo 已提交
2717 2718
	struct fuse_poll_in inarg = { .fh = ff->fh, .kh = ff->kh };
	struct fuse_poll_out outarg;
2719
	FUSE_ARGS(args);
T
Tejun Heo 已提交
2720 2721 2722 2723 2724 2725
	int err;

	if (fc->no_poll)
		return DEFAULT_POLLMASK;

	poll_wait(file, &ff->poll_wait, wait);
E
Enke Chen 已提交
2726
	inarg.events = (__u32)poll_requested_events(wait);
T
Tejun Heo 已提交
2727 2728 2729 2730 2731 2732 2733 2734 2735 2736

	/*
	 * Ask for notification iff there's someone waiting for it.
	 * The client may ignore the flag and always notify.
	 */
	if (waitqueue_active(&ff->poll_wait)) {
		inarg.flags |= FUSE_POLL_SCHEDULE_NOTIFY;
		fuse_register_polled_file(fc, ff);
	}

2737 2738 2739 2740 2741 2742 2743 2744 2745
	args.in.h.opcode = FUSE_POLL;
	args.in.h.nodeid = ff->nodeid;
	args.in.numargs = 1;
	args.in.args[0].size = sizeof(inarg);
	args.in.args[0].value = &inarg;
	args.out.numargs = 1;
	args.out.args[0].size = sizeof(outarg);
	args.out.args[0].value = &outarg;
	err = fuse_simple_request(fc, &args);
T
Tejun Heo 已提交
2746 2747 2748 2749 2750 2751 2752 2753 2754

	if (!err)
		return outarg.revents;
	if (err == -ENOSYS) {
		fc->no_poll = 1;
		return DEFAULT_POLLMASK;
	}
	return POLLERR;
}
2755
EXPORT_SYMBOL_GPL(fuse_file_poll);
T
Tejun Heo 已提交
2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780

/*
 * This is called from fuse_handle_notify() on FUSE_NOTIFY_POLL and
 * wakes up the poll waiters.
 */
int fuse_notify_poll_wakeup(struct fuse_conn *fc,
			    struct fuse_notify_poll_wakeup_out *outarg)
{
	u64 kh = outarg->kh;
	struct rb_node **link;

	spin_lock(&fc->lock);

	link = fuse_find_polled_node(fc, kh, NULL);
	if (*link) {
		struct fuse_file *ff;

		ff = rb_entry(*link, struct fuse_file, polled_node);
		wake_up_interruptible_sync(&ff->poll_wait);
	}

	spin_unlock(&fc->lock);
	return 0;
}

2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794
static void fuse_do_truncate(struct file *file)
{
	struct inode *inode = file->f_mapping->host;
	struct iattr attr;

	attr.ia_valid = ATTR_SIZE;
	attr.ia_size = i_size_read(inode);

	attr.ia_file = file;
	attr.ia_valid |= ATTR_FILE;

	fuse_do_setattr(inode, &attr, file);
}

2795 2796 2797 2798 2799
static inline loff_t fuse_round_up(loff_t off)
{
	return round_up(off, FUSE_MAX_PAGES_PER_REQ << PAGE_SHIFT);
}

A
Anand Avati 已提交
2800
static ssize_t
A
Al Viro 已提交
2801 2802
fuse_direct_IO(int rw, struct kiocb *iocb, struct iov_iter *iter,
			loff_t offset)
A
Anand Avati 已提交
2803 2804
{
	ssize_t ret = 0;
2805 2806
	struct file *file = iocb->ki_filp;
	struct fuse_file *ff = file->private_data;
2807
	bool async_dio = ff->fc->async_dio;
A
Anand Avati 已提交
2808
	loff_t pos = 0;
2809 2810
	struct inode *inode;
	loff_t i_size;
2811
	size_t count = iov_iter_count(iter);
2812
	struct fuse_io_priv *io;
A
Anand Avati 已提交
2813 2814

	pos = offset;
2815 2816
	inode = file->f_mapping->host;
	i_size = i_size_read(inode);
A
Anand Avati 已提交
2817

2818 2819 2820
	if ((rw == READ) && (offset > i_size))
		return 0;

2821
	/* optimization for short read */
2822
	if (async_dio && rw != WRITE && offset + count > i_size) {
2823 2824
		if (offset >= i_size)
			return 0;
2825
		count = min_t(loff_t, count, fuse_round_up(i_size - offset));
A
Al Viro 已提交
2826
		iov_iter_truncate(iter, count);
2827 2828
	}

2829
	io = kmalloc(sizeof(struct fuse_io_priv), GFP_KERNEL);
2830 2831
	if (!io)
		return -ENOMEM;
2832 2833 2834 2835 2836 2837 2838
	spin_lock_init(&io->lock);
	io->reqs = 1;
	io->bytes = -1;
	io->size = 0;
	io->offset = offset;
	io->write = (rw == WRITE);
	io->err = 0;
2839
	io->file = file;
2840 2841
	/*
	 * By default, we want to optimize all I/Os with async request
2842
	 * submission to the client filesystem if supported.
2843
	 */
2844
	io->async = async_dio;
2845 2846 2847 2848 2849 2850 2851
	io->iocb = iocb;

	/*
	 * We cannot asynchronously extend the size of a file. We have no method
	 * to wait on real async I/O requests, so we must submit this request
	 * synchronously.
	 */
2852
	if (!is_sync_kiocb(iocb) && (offset + count > i_size) && rw == WRITE)
2853
		io->async = false;
A
Anand Avati 已提交
2854

2855
	if (rw == WRITE)
A
Al Viro 已提交
2856
		ret = __fuse_direct_write(io, iter, &pos);
2857
	else
A
Al Viro 已提交
2858
		ret = __fuse_direct_read(io, iter, &pos);
2859

2860 2861 2862 2863
	if (io->async) {
		fuse_aio_complete(io, ret < 0 ? ret : 0, -1);

		/* we have a non-extending, async request, so return */
2864
		if (!is_sync_kiocb(iocb))
2865 2866 2867 2868 2869 2870 2871
			return -EIOCBQUEUED;

		ret = wait_on_sync_kiocb(iocb);
	} else {
		kfree(io);
	}

2872 2873 2874 2875 2876 2877
	if (rw == WRITE) {
		if (ret > 0)
			fuse_write_update_size(inode, pos);
		else if (ret < 0 && offset + count > i_size)
			fuse_do_truncate(file);
	}
A
Anand Avati 已提交
2878 2879 2880 2881

	return ret;
}

2882 2883
static long fuse_file_fallocate(struct file *file, int mode, loff_t offset,
				loff_t length)
A
Anatol Pomozov 已提交
2884 2885
{
	struct fuse_file *ff = file->private_data;
2886
	struct inode *inode = file_inode(file);
2887
	struct fuse_inode *fi = get_fuse_inode(inode);
A
Anatol Pomozov 已提交
2888
	struct fuse_conn *fc = ff->fc;
2889
	FUSE_ARGS(args);
A
Anatol Pomozov 已提交
2890 2891 2892 2893 2894 2895 2896
	struct fuse_fallocate_in inarg = {
		.fh = ff->fh,
		.offset = offset,
		.length = length,
		.mode = mode
	};
	int err;
2897 2898
	bool lock_inode = !(mode & FALLOC_FL_KEEP_SIZE) ||
			   (mode & FALLOC_FL_PUNCH_HOLE);
A
Anatol Pomozov 已提交
2899

M
Miklos Szeredi 已提交
2900 2901 2902
	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
		return -EOPNOTSUPP;

2903 2904 2905
	if (fc->no_fallocate)
		return -EOPNOTSUPP;

2906
	if (lock_inode) {
2907
		mutex_lock(&inode->i_mutex);
2908 2909 2910 2911 2912 2913 2914 2915 2916
		if (mode & FALLOC_FL_PUNCH_HOLE) {
			loff_t endbyte = offset + length - 1;
			err = filemap_write_and_wait_range(inode->i_mapping,
							   offset, endbyte);
			if (err)
				goto out;

			fuse_sync_writes(inode);
		}
2917 2918
	}

2919 2920 2921
	if (!(mode & FALLOC_FL_KEEP_SIZE))
		set_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);

2922 2923 2924 2925 2926 2927
	args.in.h.opcode = FUSE_FALLOCATE;
	args.in.h.nodeid = ff->nodeid;
	args.in.numargs = 1;
	args.in.args[0].size = sizeof(inarg);
	args.in.args[0].value = &inarg;
	err = fuse_simple_request(fc, &args);
2928 2929 2930 2931
	if (err == -ENOSYS) {
		fc->no_fallocate = 1;
		err = -EOPNOTSUPP;
	}
2932 2933 2934 2935
	if (err)
		goto out;

	/* we could have extended the file */
M
Maxim Patlasov 已提交
2936 2937 2938
	if (!(mode & FALLOC_FL_KEEP_SIZE)) {
		bool changed = fuse_write_update_size(inode, offset + length);

2939 2940
		if (changed && fc->writeback_cache)
			file_update_time(file);
M
Maxim Patlasov 已提交
2941
	}
2942 2943 2944 2945 2946 2947

	if (mode & FALLOC_FL_PUNCH_HOLE)
		truncate_pagecache_range(inode, offset, offset + length - 1);

	fuse_invalidate_attr(inode);

2948
out:
2949 2950 2951
	if (!(mode & FALLOC_FL_KEEP_SIZE))
		clear_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);

2952
	if (lock_inode)
2953 2954
		mutex_unlock(&inode->i_mutex);

A
Anatol Pomozov 已提交
2955 2956 2957
	return err;
}

2958
static const struct file_operations fuse_file_operations = {
M
Miklos Szeredi 已提交
2959
	.llseek		= fuse_file_llseek,
2960 2961
	.read		= new_sync_read,
	.read_iter	= fuse_file_read_iter,
A
Al Viro 已提交
2962 2963
	.write		= new_sync_write,
	.write_iter	= fuse_file_write_iter,
M
Miklos Szeredi 已提交
2964 2965 2966 2967 2968
	.mmap		= fuse_file_mmap,
	.open		= fuse_open,
	.flush		= fuse_flush,
	.release	= fuse_release,
	.fsync		= fuse_fsync,
2969
	.lock		= fuse_file_lock,
2970
	.flock		= fuse_file_flock,
2971
	.splice_read	= generic_file_splice_read,
T
Tejun Heo 已提交
2972 2973
	.unlocked_ioctl	= fuse_file_ioctl,
	.compat_ioctl	= fuse_file_compat_ioctl,
T
Tejun Heo 已提交
2974
	.poll		= fuse_file_poll,
A
Anatol Pomozov 已提交
2975
	.fallocate	= fuse_file_fallocate,
M
Miklos Szeredi 已提交
2976 2977
};

2978
static const struct file_operations fuse_direct_io_file_operations = {
M
Miklos Szeredi 已提交
2979
	.llseek		= fuse_file_llseek,
M
Miklos Szeredi 已提交
2980 2981
	.read		= fuse_direct_read,
	.write		= fuse_direct_write,
2982
	.mmap		= fuse_direct_mmap,
M
Miklos Szeredi 已提交
2983 2984 2985 2986
	.open		= fuse_open,
	.flush		= fuse_flush,
	.release	= fuse_release,
	.fsync		= fuse_fsync,
2987
	.lock		= fuse_file_lock,
2988
	.flock		= fuse_file_flock,
T
Tejun Heo 已提交
2989 2990
	.unlocked_ioctl	= fuse_file_ioctl,
	.compat_ioctl	= fuse_file_compat_ioctl,
T
Tejun Heo 已提交
2991
	.poll		= fuse_file_poll,
A
Anatol Pomozov 已提交
2992
	.fallocate	= fuse_file_fallocate,
2993
	/* no splice_read */
M
Miklos Szeredi 已提交
2994 2995
};

2996
static const struct address_space_operations fuse_file_aops  = {
M
Miklos Szeredi 已提交
2997
	.readpage	= fuse_readpage,
M
Miklos Szeredi 已提交
2998
	.writepage	= fuse_writepage,
2999
	.writepages	= fuse_writepages,
M
Miklos Szeredi 已提交
3000
	.launder_page	= fuse_launder_page,
3001
	.readpages	= fuse_readpages,
M
Miklos Szeredi 已提交
3002
	.set_page_dirty	= __set_page_dirty_nobuffers,
M
Miklos Szeredi 已提交
3003
	.bmap		= fuse_bmap,
A
Anand Avati 已提交
3004
	.direct_IO	= fuse_direct_IO,
3005 3006
	.write_begin	= fuse_write_begin,
	.write_end	= fuse_write_end,
M
Miklos Szeredi 已提交
3007 3008 3009 3010
};

void fuse_init_file_inode(struct inode *inode)
{
3011 3012
	inode->i_fop = &fuse_file_operations;
	inode->i_data.a_ops = &fuse_file_aops;
M
Miklos Szeredi 已提交
3013
}