file.c 63.0 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;
M
Miklos Szeredi 已提交
27 28 29
	struct fuse_req *req;
	int err;

M
Maxim Patlasov 已提交
30
	req = fuse_get_req_nopages(fc);
31 32
	if (IS_ERR(req))
		return PTR_ERR(req);
M
Miklos Szeredi 已提交
33 34

	memset(&inarg, 0, sizeof(inarg));
35 36 37
	inarg.flags = file->f_flags & ~(O_CREAT | O_EXCL | O_NOCTTY);
	if (!fc->atomic_o_trunc)
		inarg.flags &= ~O_TRUNC;
38 39
	req->in.h.opcode = opcode;
	req->in.h.nodeid = nodeid;
M
Miklos Szeredi 已提交
40 41 42 43 44 45
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(*outargp);
	req->out.args[0].value = outargp;
46
	fuse_request_send(fc, req);
M
Miklos Szeredi 已提交
47 48 49 50 51 52
	err = req->out.h.error;
	fuse_put_request(fc, req);

	return err;
}

T
Tejun Heo 已提交
53
struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
M
Miklos Szeredi 已提交
54 55
{
	struct fuse_file *ff;
T
Tejun Heo 已提交
56

M
Miklos Szeredi 已提交
57
	ff = kmalloc(sizeof(struct fuse_file), GFP_KERNEL);
T
Tejun Heo 已提交
58 59 60
	if (unlikely(!ff))
		return NULL;

61
	ff->fc = fc;
62
	ff->reserved_req = fuse_request_alloc(0);
T
Tejun Heo 已提交
63 64 65
	if (unlikely(!ff->reserved_req)) {
		kfree(ff);
		return NULL;
M
Miklos Szeredi 已提交
66
	}
T
Tejun Heo 已提交
67 68 69 70 71 72 73 74 75 76

	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
77 78 79 80 81
	return ff;
}

void fuse_file_free(struct fuse_file *ff)
{
82
	fuse_request_free(ff->reserved_req);
M
Miklos Szeredi 已提交
83 84 85
	kfree(ff);
}

86
struct fuse_file *fuse_file_get(struct fuse_file *ff)
87 88 89 90 91
{
	atomic_inc(&ff->count);
	return ff;
}

92 93 94 95 96 97 98 99 100 101 102 103 104 105
static void fuse_release_async(struct work_struct *work)
{
	struct fuse_req *req;
	struct fuse_conn *fc;
	struct path path;

	req = container_of(work, struct fuse_req, misc.release.work);
	path = req->misc.release.path;
	fc = get_fuse_conn(path.dentry->d_inode);

	fuse_put_request(fc, req);
	path_put(&path);
}

M
Miklos Szeredi 已提交
106 107
static void fuse_release_end(struct fuse_conn *fc, struct fuse_req *req)
{
108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
	if (fc->destroy_req) {
		/*
		 * If this is a fuseblk mount, then it's possible that
		 * releasing the path will result in releasing the
		 * super block and sending the DESTROY request.  If
		 * the server is single threaded, this would hang.
		 * For this reason do the path_put() in a separate
		 * thread.
		 */
		atomic_inc(&req->count);
		INIT_WORK(&req->misc.release.work, fuse_release_async);
		schedule_work(&req->misc.release.work);
	} else {
		path_put(&req->misc.release.path);
	}
M
Miklos Szeredi 已提交
123 124
}

125
static void fuse_file_put(struct fuse_file *ff, bool sync)
126 127 128
{
	if (atomic_dec_and_test(&ff->count)) {
		struct fuse_req *req = ff->reserved_req;
129

130
		if (sync) {
131
			req->background = 0;
132 133 134 135 136
			fuse_request_send(ff->fc, req);
			path_put(&req->misc.release.path);
			fuse_put_request(ff->fc, req);
		} else {
			req->end = fuse_release_end;
137
			req->background = 1;
138 139
			fuse_request_send_background(ff->fc, req);
		}
140 141 142 143
		kfree(ff);
	}
}

144 145
int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
		 bool isdir)
146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
{
	struct fuse_open_out outarg;
	struct fuse_file *ff;
	int err;
	int opcode = isdir ? FUSE_OPENDIR : FUSE_OPEN;

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

	err = fuse_send_open(fc, nodeid, file, opcode, &outarg);
	if (err) {
		fuse_file_free(ff);
		return err;
	}

	if (isdir)
		outarg.open_flags &= ~FOPEN_DIRECT_IO;

	ff->fh = outarg.fh;
	ff->nodeid = nodeid;
	ff->open_flags = outarg.open_flags;
	file->private_data = fuse_file_get(ff);

	return 0;
}
172
EXPORT_SYMBOL_GPL(fuse_do_open);
173

174
void fuse_finish_open(struct inode *inode, struct file *file)
M
Miklos Szeredi 已提交
175
{
176
	struct fuse_file *ff = file->private_data;
177
	struct fuse_conn *fc = get_fuse_conn(inode);
178 179

	if (ff->open_flags & FOPEN_DIRECT_IO)
M
Miklos Szeredi 已提交
180
		file->f_op = &fuse_direct_io_file_operations;
181
	if (!(ff->open_flags & FOPEN_KEEP_CACHE))
M
Miklos Szeredi 已提交
182
		invalidate_inode_pages2(inode->i_mapping);
183
	if (ff->open_flags & FOPEN_NONSEEKABLE)
T
Tejun Heo 已提交
184
		nonseekable_open(inode, file);
185 186 187 188 189 190 191 192 193
	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);
	}
M
Miklos Szeredi 已提交
194 195
}

196
int fuse_open_common(struct inode *inode, struct file *file, bool isdir)
M
Miklos Szeredi 已提交
197
{
T
Tejun Heo 已提交
198
	struct fuse_conn *fc = get_fuse_conn(inode);
M
Miklos Szeredi 已提交
199 200 201 202 203 204
	int err;

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

205
	err = fuse_do_open(fc, get_node_id(inode), file, isdir);
M
Miklos Szeredi 已提交
206
	if (err)
207
		return err;
M
Miklos Szeredi 已提交
208

209 210 211
	fuse_finish_open(inode, file);

	return 0;
M
Miklos Szeredi 已提交
212 213
}

214
static void fuse_prepare_release(struct fuse_file *ff, int flags, int opcode)
215
{
216
	struct fuse_conn *fc = ff->fc;
217
	struct fuse_req *req = ff->reserved_req;
218
	struct fuse_release_in *inarg = &req->misc.release.in;
M
Miklos Szeredi 已提交
219

220 221 222 223 224 225
	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);

226
	wake_up_interruptible_all(&ff->poll_wait);
227

M
Miklos Szeredi 已提交
228
	inarg->fh = ff->fh;
M
Miklos Szeredi 已提交
229
	inarg->flags = flags;
230
	req->in.h.opcode = opcode;
231
	req->in.h.nodeid = ff->nodeid;
M
Miklos Szeredi 已提交
232 233 234
	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
235 236
}

237
void fuse_release_common(struct file *file, int opcode)
M
Miklos Szeredi 已提交
238
{
T
Tejun Heo 已提交
239 240
	struct fuse_file *ff;
	struct fuse_req *req;
M
Miklos Szeredi 已提交
241

T
Tejun Heo 已提交
242 243
	ff = file->private_data;
	if (unlikely(!ff))
244
		return;
T
Tejun Heo 已提交
245 246

	req = ff->reserved_req;
247
	fuse_prepare_release(ff, file->f_flags, opcode);
T
Tejun Heo 已提交
248

M
Miklos Szeredi 已提交
249 250 251 252 253 254
	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);
	}
T
Tejun Heo 已提交
255
	/* Hold vfsmount and dentry until release is finished */
256 257
	path_get(&file->f_path);
	req->misc.release.path = file->f_path;
T
Tejun Heo 已提交
258 259 260 261 262

	/*
	 * Normally this will send the RELEASE request, however if
	 * some asynchronous READ or WRITE requests are outstanding,
	 * the sending will be delayed.
263 264 265 266
	 *
	 * 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
267
	 */
268
	fuse_file_put(ff, ff->fc->destroy_req != NULL);
M
Miklos Szeredi 已提交
269 270
}

271 272
static int fuse_open(struct inode *inode, struct file *file)
{
273
	return fuse_open_common(inode, file, false);
274 275 276 277
}

static int fuse_release(struct inode *inode, struct file *file)
{
278 279 280 281 282 283 284 285 286 287 288
	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;
289
	ff->reserved_req->background = 0;
290 291 292
	fuse_request_send(ff->fc, ff->reserved_req);
	fuse_put_request(ff->fc, ff->reserved_req);
	kfree(ff);
293
}
294
EXPORT_SYMBOL_GPL(fuse_sync_release);
295

296
/*
297 298
 * Scramble the ID space with XTEA, so that the value of the files_struct
 * pointer is not exposed to userspace.
299
 */
300
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id)
301
{
302 303 304 305 306 307 308 309 310 311 312 313 314 315
	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);
316 317
}

M
Miklos Szeredi 已提交
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
/*
 * Check if page is under writeback
 *
 * This is currently done by walking the list of writepage requests
 * for the inode, which can be pretty inefficient.
 */
static bool fuse_page_is_writeback(struct inode *inode, pgoff_t index)
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct fuse_req *req;
	bool found = false;

	spin_lock(&fc->lock);
	list_for_each_entry(req, &fi->writepages, writepages_entry) {
		pgoff_t curr_index;

		BUG_ON(req->inode != inode);
		curr_index = req->misc.write.in.offset >> PAGE_CACHE_SHIFT;
337 338
		if (curr_index <= index &&
		    index < curr_index + req->num_pages) {
M
Miklos Szeredi 已提交
339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361
			found = true;
			break;
		}
	}
	spin_unlock(&fc->lock);

	return found;
}

/*
 * Wait for page writeback to be completed.
 *
 * Since fuse doesn't rely on the VM writeback tracking, this has to
 * use some other means.
 */
static int fuse_wait_on_page_writeback(struct inode *inode, pgoff_t index)
{
	struct fuse_inode *fi = get_fuse_inode(inode);

	wait_event(fi->page_waitq, !fuse_page_is_writeback(inode, index));
	return 0;
}

362
static int fuse_flush(struct file *file, fl_owner_t id)
M
Miklos Szeredi 已提交
363
{
A
Al Viro 已提交
364
	struct inode *inode = file_inode(file);
M
Miklos Szeredi 已提交
365 366 367 368 369 370
	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;

371 372 373
	if (is_bad_inode(inode))
		return -EIO;

M
Miklos Szeredi 已提交
374 375 376
	if (fc->no_flush)
		return 0;

M
Maxim Patlasov 已提交
377
	req = fuse_get_req_nofail_nopages(fc, file);
M
Miklos Szeredi 已提交
378 379
	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
380
	inarg.lock_owner = fuse_lock_owner_id(fc, id);
M
Miklos Szeredi 已提交
381 382 383 384 385
	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;
386
	req->force = 1;
387
	fuse_request_send(fc, req);
M
Miklos Szeredi 已提交
388 389 390 391 392 393 394 395 396
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
		fc->no_flush = 1;
		err = 0;
	}
	return err;
}

M
Miklos Szeredi 已提交
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
/*
 * 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);
}

412 413
int fuse_fsync_common(struct file *file, loff_t start, loff_t end,
		      int datasync, int isdir)
M
Miklos Szeredi 已提交
414
{
415
	struct inode *inode = file->f_mapping->host;
M
Miklos Szeredi 已提交
416 417 418 419 420 421
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_file *ff = file->private_data;
	struct fuse_req *req;
	struct fuse_fsync_in inarg;
	int err;

422 423 424
	if (is_bad_inode(inode))
		return -EIO;

425 426 427 428
	err = filemap_write_and_wait_range(inode->i_mapping, start, end);
	if (err)
		return err;

429
	if ((!isdir && fc->no_fsync) || (isdir && fc->no_fsyncdir))
M
Miklos Szeredi 已提交
430 431
		return 0;

432 433
	mutex_lock(&inode->i_mutex);

M
Miklos Szeredi 已提交
434 435 436 437 438 439 440
	/*
	 * Start writeback against all dirty pages of the inode, then
	 * wait for all outstanding writes, before sending the FSYNC
	 * request.
	 */
	err = write_inode_now(inode, 0);
	if (err)
441
		goto out;
M
Miklos Szeredi 已提交
442 443 444

	fuse_sync_writes(inode);

M
Maxim Patlasov 已提交
445
	req = fuse_get_req_nopages(fc);
446 447 448 449
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		goto out;
	}
M
Miklos Szeredi 已提交
450 451 452 453

	memset(&inarg, 0, sizeof(inarg));
	inarg.fh = ff->fh;
	inarg.fsync_flags = datasync ? 1 : 0;
454
	req->in.h.opcode = isdir ? FUSE_FSYNCDIR : FUSE_FSYNC;
M
Miklos Szeredi 已提交
455 456 457 458
	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;
459
	fuse_request_send(fc, req);
M
Miklos Szeredi 已提交
460 461 462
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
463 464 465 466
		if (isdir)
			fc->no_fsyncdir = 1;
		else
			fc->no_fsync = 1;
M
Miklos Szeredi 已提交
467 468
		err = 0;
	}
469 470
out:
	mutex_unlock(&inode->i_mutex);
M
Miklos Szeredi 已提交
471 472 473
	return err;
}

474 475
static int fuse_fsync(struct file *file, loff_t start, loff_t end,
		      int datasync)
476
{
477
	return fuse_fsync_common(file, start, end, datasync, 0);
478 479
}

480 481
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
482
{
483
	struct fuse_read_in *inarg = &req->misc.read.in;
484
	struct fuse_file *ff = file->private_data;
M
Miklos Szeredi 已提交
485

486 487 488
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
489
	inarg->flags = file->f_flags;
490
	req->in.h.opcode = opcode;
491
	req->in.h.nodeid = ff->nodeid;
M
Miklos Szeredi 已提交
492 493
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(struct fuse_read_in);
494
	req->in.args[0].value = inarg;
M
Miklos Szeredi 已提交
495 496 497 498 499
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = count;
}

500 501 502 503 504 505 506 507 508 509 510 511
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
512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551
/**
 * 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
552
				struct inode *inode = file_inode(io->iocb->ki_filp);
M
Maxim Patlasov 已提交
553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597
				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;

598
	__fuse_get_request(req);
M
Maxim Patlasov 已提交
599 600 601 602 603
	fuse_request_send_background(fc, req);

	return num_bytes;
}

604
static size_t fuse_send_read(struct fuse_req *req, struct fuse_io_priv *io,
605
			     loff_t pos, size_t count, fl_owner_t owner)
606
{
607
	struct file *file = io->file;
608 609
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
610

611
	fuse_read_fill(req, file, pos, count, FUSE_READ);
612
	if (owner != NULL) {
613
		struct fuse_read_in *inarg = &req->misc.read.in;
614 615 616 617

		inarg->read_flags |= FUSE_READ_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
618 619 620 621

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

622
	fuse_request_send(fc, req);
623
	return req->out.args[0].size;
624 625
}

626 627 628 629 630 631 632
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);
633 634
	if (attr_ver == fi->attr_version && size < inode->i_size &&
	    !test_bit(FUSE_I_SIZE_UNSTABLE, &fi->state)) {
635 636 637 638 639 640
		fi->attr_version = ++fc->attr_version;
		i_size_write(inode, size);
	}
	spin_unlock(&fc->lock);
}

M
Miklos Szeredi 已提交
641 642
static int fuse_readpage(struct file *file, struct page *page)
{
643
	struct fuse_io_priv io = { .async = 0, .file = file };
M
Miklos Szeredi 已提交
644 645
	struct inode *inode = page->mapping->host;
	struct fuse_conn *fc = get_fuse_conn(inode);
646
	struct fuse_req *req;
647 648 649 650
	size_t num_read;
	loff_t pos = page_offset(page);
	size_t count = PAGE_CACHE_SIZE;
	u64 attr_ver;
651 652 653 654 655 656
	int err;

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

M
Miklos Szeredi 已提交
657
	/*
L
Lucas De Marchi 已提交
658
	 * Page writeback can extend beyond the lifetime of the
M
Miklos Szeredi 已提交
659 660 661 662 663
	 * 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
664
	req = fuse_get_req(fc, 1);
665 666
	err = PTR_ERR(req);
	if (IS_ERR(req))
M
Miklos Szeredi 已提交
667 668
		goto out;

669 670
	attr_ver = fuse_get_attr_version(fc);

M
Miklos Szeredi 已提交
671
	req->out.page_zeroing = 1;
672
	req->out.argpages = 1;
M
Miklos Szeredi 已提交
673 674
	req->num_pages = 1;
	req->pages[0] = page;
675
	req->page_descs[0].length = count;
676
	num_read = fuse_send_read(req, &io, pos, count, NULL);
M
Miklos Szeredi 已提交
677 678
	err = req->out.h.error;
	fuse_put_request(fc, req);
679 680 681 682 683 684 685 686

	if (!err) {
		/*
		 * Short read means EOF.  If file size is larger, truncate it
		 */
		if (num_read < count)
			fuse_read_update_size(inode, pos + num_read, attr_ver);

M
Miklos Szeredi 已提交
687
		SetPageUptodate(page);
688 689
	}

690
	fuse_invalidate_attr(inode); /* atime changed */
M
Miklos Szeredi 已提交
691 692 693 694 695
 out:
	unlock_page(page);
	return err;
}

696
static void fuse_readpages_end(struct fuse_conn *fc, struct fuse_req *req)
697
{
698
	int i;
699 700
	size_t count = req->misc.read.in.size;
	size_t num_read = req->out.args[0].size;
701
	struct address_space *mapping = NULL;
702

703 704
	for (i = 0; mapping == NULL && i < req->num_pages; i++)
		mapping = req->pages[i]->mapping;
705

706 707 708 709 710 711 712 713 714 715 716 717 718 719 720
	if (mapping) {
		struct inode *inode = mapping->host;

		/*
		 * Short read means EOF. If file size is larger, truncate it
		 */
		if (!req->out.h.error && num_read < count) {
			loff_t pos;

			pos = page_offset(req->pages[0]) + num_read;
			fuse_read_update_size(inode, pos,
					      req->misc.read.attr_ver);
		}
		fuse_invalidate_attr(inode); /* atime changed */
	}
721

722 723 724 725
	for (i = 0; i < req->num_pages; i++) {
		struct page *page = req->pages[i];
		if (!req->out.h.error)
			SetPageUptodate(page);
726 727
		else
			SetPageError(page);
728
		unlock_page(page);
729
		page_cache_release(page);
730
	}
731
	if (req->ff)
732
		fuse_file_put(req->ff, false);
733 734
}

735
static void fuse_send_readpages(struct fuse_req *req, struct file *file)
736
{
737 738
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
739 740
	loff_t pos = page_offset(req->pages[0]);
	size_t count = req->num_pages << PAGE_CACHE_SHIFT;
741 742

	req->out.argpages = 1;
743
	req->out.page_zeroing = 1;
744
	req->out.page_replace = 1;
745
	fuse_read_fill(req, file, pos, count, FUSE_READ);
746
	req->misc.read.attr_ver = fuse_get_attr_version(fc);
747
	if (fc->async_read) {
748
		req->ff = fuse_file_get(ff);
749
		req->end = fuse_readpages_end;
750
		fuse_request_send_background(fc, req);
751
	} else {
752
		fuse_request_send(fc, req);
753
		fuse_readpages_end(fc, req);
754
		fuse_put_request(fc, req);
755
	}
756 757
}

758
struct fuse_fill_data {
759
	struct fuse_req *req;
760
	struct file *file;
761
	struct inode *inode;
M
Maxim Patlasov 已提交
762
	unsigned nr_pages;
763 764 765 766
};

static int fuse_readpages_fill(void *_data, struct page *page)
{
767
	struct fuse_fill_data *data = _data;
768 769 770 771
	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
772 773
	fuse_wait_on_page_writeback(inode, page->index);

774 775 776 777
	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
778 779
		int nr_alloc = min_t(unsigned, data->nr_pages,
				     FUSE_MAX_PAGES_PER_REQ);
780
		fuse_send_readpages(req, data->file);
781 782 783 784 785 786
		if (fc->async_read)
			req = fuse_get_req_for_background(fc, nr_alloc);
		else
			req = fuse_get_req(fc, nr_alloc);

		data->req = req;
787
		if (IS_ERR(req)) {
788
			unlock_page(page);
789
			return PTR_ERR(req);
790 791
		}
	}
M
Maxim Patlasov 已提交
792 793 794 795 796 797

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

798
	page_cache_get(page);
799
	req->pages[req->num_pages] = page;
800
	req->page_descs[req->num_pages].length = PAGE_SIZE;
M
Miklos Szeredi 已提交
801
	req->num_pages++;
M
Maxim Patlasov 已提交
802
	data->nr_pages--;
803 804 805 806 807 808 809 810
	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);
811
	struct fuse_fill_data data;
812
	int err;
M
Maxim Patlasov 已提交
813
	int nr_alloc = min_t(unsigned, nr_pages, FUSE_MAX_PAGES_PER_REQ);
814

815
	err = -EIO;
816
	if (is_bad_inode(inode))
817
		goto out;
818

819
	data.file = file;
820
	data.inode = inode;
821 822 823 824
	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
825
	data.nr_pages = nr_pages;
826
	err = PTR_ERR(data.req);
827
	if (IS_ERR(data.req))
828
		goto out;
829 830

	err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
831 832
	if (!err) {
		if (data.req->num_pages)
833
			fuse_send_readpages(data.req, file);
834 835 836
		else
			fuse_put_request(fc, data.req);
	}
837
out:
838
	return err;
839 840
}

M
Miklos Szeredi 已提交
841 842 843 844
static ssize_t fuse_file_aio_read(struct kiocb *iocb, const struct iovec *iov,
				  unsigned long nr_segs, loff_t pos)
{
	struct inode *inode = iocb->ki_filp->f_mapping->host;
845
	struct fuse_conn *fc = get_fuse_conn(inode);
M
Miklos Szeredi 已提交
846

847 848 849 850 851 852 853
	/*
	 * In auto invalidate mode, always update attributes on read.
	 * Otherwise, only update if we attempt to read past EOF (to ensure
	 * i_size is up to date).
	 */
	if (fc->auto_inval_data ||
	    (pos + iov_length(iov, nr_segs) > i_size_read(inode))) {
M
Miklos Szeredi 已提交
854 855 856 857 858 859 860 861 862
		int err;
		err = fuse_update_attributes(inode, NULL, iocb->ki_filp, NULL);
		if (err)
			return err;
	}

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

863
static void fuse_write_fill(struct fuse_req *req, struct fuse_file *ff,
864
			    loff_t pos, size_t count)
M
Miklos Szeredi 已提交
865
{
866 867
	struct fuse_write_in *inarg = &req->misc.write.in;
	struct fuse_write_out *outarg = &req->misc.write.out;
M
Miklos Szeredi 已提交
868

869 870 871
	inarg->fh = ff->fh;
	inarg->offset = pos;
	inarg->size = count;
M
Miklos Szeredi 已提交
872
	req->in.h.opcode = FUSE_WRITE;
873
	req->in.h.nodeid = ff->nodeid;
M
Miklos Szeredi 已提交
874
	req->in.numargs = 2;
875
	if (ff->fc->minor < 9)
876 877 878
		req->in.args[0].size = FUSE_COMPAT_WRITE_IN_SIZE;
	else
		req->in.args[0].size = sizeof(struct fuse_write_in);
879
	req->in.args[0].value = inarg;
M
Miklos Szeredi 已提交
880 881 882
	req->in.args[1].size = count;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(struct fuse_write_out);
883 884 885
	req->out.args[0].value = outarg;
}

886
static size_t fuse_send_write(struct fuse_req *req, struct fuse_io_priv *io,
887
			      loff_t pos, size_t count, fl_owner_t owner)
888
{
889
	struct file *file = io->file;
890 891
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
892 893
	struct fuse_write_in *inarg = &req->misc.write.in;

894
	fuse_write_fill(req, ff, pos, count);
895
	inarg->flags = file->f_flags;
896 897 898 899
	if (owner != NULL) {
		inarg->write_flags |= FUSE_WRITE_LOCKOWNER;
		inarg->lock_owner = fuse_lock_owner_id(fc, owner);
	}
900 901 902 903

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

904
	fuse_request_send(fc, req);
905
	return req->misc.write.out.size;
M
Miklos Szeredi 已提交
906 907
}

M
Miklos Szeredi 已提交
908
void fuse_write_update_size(struct inode *inode, loff_t pos)
909 910 911 912 913 914 915 916 917 918 919
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);

	spin_lock(&fc->lock);
	fi->attr_version = ++fc->attr_version;
	if (pos > inode->i_size)
		i_size_write(inode, pos);
	spin_unlock(&fc->lock);
}

N
Nick Piggin 已提交
920 921 922 923 924 925 926
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;
927
	struct fuse_io_priv io = { .async = 0, .file = file };
N
Nick Piggin 已提交
928 929 930 931

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

932
	res = fuse_send_write(req, &io, pos, count, NULL);
N
Nick Piggin 已提交
933

934
	offset = req->page_descs[0].offset;
N
Nick Piggin 已提交
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963
	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;

964
	req->in.argpages = 1;
965
	req->page_descs[0].offset = offset;
N
Nick Piggin 已提交
966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981

	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;
982
		page = grab_cache_page_write_begin(mapping, index, 0);
N
Nick Piggin 已提交
983 984 985
		if (!page)
			break;

986 987 988
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

N
Nick Piggin 已提交
989 990 991 992 993
		pagefault_disable();
		tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
		pagefault_enable();
		flush_dcache_page(page);

994 995
		mark_page_accessed(page);

N
Nick Piggin 已提交
996 997 998 999 1000 1001 1002 1003 1004
		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;
1005
		req->page_descs[req->num_pages].length = tmp;
N
Nick Piggin 已提交
1006 1007 1008 1009 1010 1011 1012 1013 1014
		req->num_pages++;

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

1015 1016
		if (!fc->big_writes)
			break;
N
Nick Piggin 已提交
1017
	} while (iov_iter_count(ii) && count < fc->max_write &&
1018
		 req->num_pages < req->max_pages && offset == 0);
N
Nick Piggin 已提交
1019 1020 1021 1022

	return count > 0 ? count : err;
}

1023 1024 1025 1026 1027 1028 1029 1030
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
1031 1032 1033 1034 1035 1036
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);
1037
	struct fuse_inode *fi = get_fuse_inode(inode);
N
Nick Piggin 已提交
1038 1039 1040 1041 1042 1043
	int err = 0;
	ssize_t res = 0;

	if (is_bad_inode(inode))
		return -EIO;

1044 1045 1046
	if (inode->i_size < pos + iov_iter_count(ii))
		set_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);

N
Nick Piggin 已提交
1047 1048 1049
	do {
		struct fuse_req *req;
		ssize_t count;
1050
		unsigned nr_pages = fuse_wr_pages(pos, iov_iter_count(ii));
N
Nick Piggin 已提交
1051

1052
		req = fuse_get_req(fc, nr_pages);
N
Nick Piggin 已提交
1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081
		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);

1082
	clear_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);
N
Nick Piggin 已提交
1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
	fuse_invalidate_attr(inode);

	return res > 0 ? res : err;
}

static ssize_t fuse_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
				   unsigned long nr_segs, loff_t pos)
{
	struct file *file = iocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
	size_t count = 0;
A
Anand Avati 已提交
1094
	size_t ocount = 0;
N
Nick Piggin 已提交
1095
	ssize_t written = 0;
A
Anand Avati 已提交
1096
	ssize_t written_buffered = 0;
N
Nick Piggin 已提交
1097 1098 1099
	struct inode *inode = mapping->host;
	ssize_t err;
	struct iov_iter i;
A
Anand Avati 已提交
1100
	loff_t endbyte = 0;
N
Nick Piggin 已提交
1101 1102 1103

	WARN_ON(iocb->ki_pos != pos);

A
Anand Avati 已提交
1104 1105
	ocount = 0;
	err = generic_segment_checks(iov, &nr_segs, &ocount, VERIFY_READ);
N
Nick Piggin 已提交
1106 1107 1108
	if (err)
		return err;

A
Anand Avati 已提交
1109
	count = ocount;
N
Nick Piggin 已提交
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
	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;

1122
	err = file_remove_suid(file);
N
Nick Piggin 已提交
1123 1124 1125
	if (err)
		goto out;

1126 1127 1128
	err = file_update_time(file);
	if (err)
		goto out;
N
Nick Piggin 已提交
1129

A
Anand Avati 已提交
1130 1131 1132 1133 1134 1135 1136 1137 1138
	if (file->f_flags & O_DIRECT) {
		written = generic_file_direct_write(iocb, iov, &nr_segs,
						    pos, &iocb->ki_pos,
						    count, ocount);
		if (written < 0 || written == count)
			goto out;

		pos += written;
		count -= written;
N
Nick Piggin 已提交
1139

A
Anand Avati 已提交
1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
		iov_iter_init(&i, iov, nr_segs, count, written);
		written_buffered = fuse_perform_write(file, mapping, &i, pos);
		if (written_buffered < 0) {
			err = written_buffered;
			goto out;
		}
		endbyte = pos + written_buffered - 1;

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

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

		written += written_buffered;
		iocb->ki_pos = pos + written_buffered;
	} else {
		iov_iter_init(&i, iov, nr_segs, count, 0);
		written = fuse_perform_write(file, mapping, &i, pos);
		if (written >= 0)
			iocb->ki_pos = pos + written;
	}
N
Nick Piggin 已提交
1165 1166 1167 1168 1169 1170 1171
out:
	current->backing_dev_info = NULL;
	mutex_unlock(&inode->i_mutex);

	return written ? written : err;
}

1172 1173
static inline void fuse_page_descs_length_init(struct fuse_req *req,
		unsigned index, unsigned nr_pages)
1174 1175 1176
{
	int i;

1177
	for (i = index; i < index + nr_pages; i++)
1178 1179 1180 1181
		req->page_descs[i].length = PAGE_SIZE -
			req->page_descs[i].offset;
}

1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192
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);
}

1193
static int fuse_get_user_pages(struct fuse_req *req, struct iov_iter *ii,
1194
			       size_t *nbytesp, int write)
M
Miklos Szeredi 已提交
1195
{
1196
	size_t nbytes = 0;  /* # bytes already packed in req */
1197

1198 1199
	/* Special case for kernel I/O: can copy directly into the buffer */
	if (segment_eq(get_fs(), KERNEL_DS)) {
1200 1201 1202
		unsigned long user_addr = fuse_get_user_addr(ii);
		size_t frag_size = fuse_get_frag_size(ii, *nbytesp);

1203 1204 1205 1206 1207
		if (write)
			req->in.args[1].value = (void *) user_addr;
		else
			req->out.args[0].value = (void *) user_addr;

1208 1209
		iov_iter_advance(ii, frag_size);
		*nbytesp = frag_size;
1210 1211
		return 0;
	}
M
Miklos Szeredi 已提交
1212

1213
	while (nbytes < *nbytesp && req->num_pages < req->max_pages) {
1214 1215 1216 1217 1218
		unsigned npages;
		unsigned long user_addr = fuse_get_user_addr(ii);
		unsigned offset = user_addr & ~PAGE_MASK;
		size_t frag_size = fuse_get_frag_size(ii, *nbytesp - nbytes);
		int ret;
M
Miklos Szeredi 已提交
1219

1220
		unsigned n = req->max_pages - req->num_pages;
1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244
		frag_size = min_t(size_t, frag_size, n << PAGE_SHIFT);

		npages = (frag_size + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
		npages = clamp(npages, 1U, n);

		ret = get_user_pages_fast(user_addr, npages, !write,
					  &req->pages[req->num_pages]);
		if (ret < 0)
			return ret;

		npages = ret;
		frag_size = min_t(size_t, frag_size,
				  (npages << PAGE_SHIFT) - offset);
		iov_iter_advance(ii, frag_size);

		req->page_descs[req->num_pages].offset = offset;
		fuse_page_descs_length_init(req, req->num_pages, npages);

		req->num_pages += npages;
		req->page_descs[req->num_pages - 1].length -=
			(npages << PAGE_SHIFT) - offset - frag_size;

		nbytes += frag_size;
	}
1245 1246 1247 1248 1249 1250

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

1251
	*nbytesp = nbytes;
1252

M
Miklos Szeredi 已提交
1253 1254 1255
	return 0;
}

1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272
static inline int fuse_iter_npages(const struct iov_iter *ii_p)
{
	struct iov_iter ii = *ii_p;
	int npages = 0;

	while (iov_iter_count(&ii) && npages < FUSE_MAX_PAGES_PER_REQ) {
		unsigned long user_addr = fuse_get_user_addr(&ii);
		unsigned offset = user_addr & ~PAGE_MASK;
		size_t frag_size = iov_iter_single_seg_count(&ii);

		npages += (frag_size + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
		iov_iter_advance(&ii, frag_size);
	}

	return min(npages, FUSE_MAX_PAGES_PER_REQ);
}

1273
ssize_t fuse_direct_io(struct fuse_io_priv *io, const struct iovec *iov,
M
Miklos Szeredi 已提交
1274 1275
		       unsigned long nr_segs, size_t count, loff_t *ppos,
		       int write)
M
Miklos Szeredi 已提交
1276
{
1277
	struct file *file = io->file;
1278 1279
	struct fuse_file *ff = file->private_data;
	struct fuse_conn *fc = ff->fc;
M
Miklos Szeredi 已提交
1280 1281 1282
	size_t nmax = write ? fc->max_write : fc->max_read;
	loff_t pos = *ppos;
	ssize_t res = 0;
1283
	struct fuse_req *req;
1284 1285 1286
	struct iov_iter ii;

	iov_iter_init(&ii, iov, nr_segs, count, 0);
1287

1288 1289 1290 1291
	if (io->async)
		req = fuse_get_req_for_background(fc, fuse_iter_npages(&ii));
	else
		req = fuse_get_req(fc, fuse_iter_npages(&ii));
1292 1293
	if (IS_ERR(req))
		return PTR_ERR(req);
M
Miklos Szeredi 已提交
1294 1295 1296

	while (count) {
		size_t nres;
1297
		fl_owner_t owner = current->files;
1298
		size_t nbytes = min(count, nmax);
1299
		int err = fuse_get_user_pages(req, &ii, &nbytes, write);
M
Miklos Szeredi 已提交
1300 1301 1302 1303
		if (err) {
			res = err;
			break;
		}
1304

M
Miklos Szeredi 已提交
1305
		if (write)
1306
			nres = fuse_send_write(req, io, pos, nbytes, owner);
M
Miklos Szeredi 已提交
1307
		else
1308
			nres = fuse_send_read(req, io, pos, nbytes, owner);
1309

1310 1311
		if (!io->async)
			fuse_release_user_pages(req, !write);
M
Miklos Szeredi 已提交
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324
		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;
1325 1326
		if (count) {
			fuse_put_request(fc, req);
1327 1328 1329 1330 1331
			if (io->async)
				req = fuse_get_req_for_background(fc,
					fuse_iter_npages(&ii));
			else
				req = fuse_get_req(fc, fuse_iter_npages(&ii));
1332 1333 1334
			if (IS_ERR(req))
				break;
		}
M
Miklos Szeredi 已提交
1335
	}
1336 1337
	if (!IS_ERR(req))
		fuse_put_request(fc, req);
1338
	if (res > 0)
M
Miklos Szeredi 已提交
1339 1340 1341 1342
		*ppos = pos;

	return res;
}
1343
EXPORT_SYMBOL_GPL(fuse_direct_io);
M
Miklos Szeredi 已提交
1344

1345 1346
static ssize_t __fuse_direct_read(struct fuse_io_priv *io,
				  const struct iovec *iov,
1347 1348
				  unsigned long nr_segs, loff_t *ppos,
				  size_t count)
M
Miklos Szeredi 已提交
1349
{
1350
	ssize_t res;
1351
	struct file *file = io->file;
A
Al Viro 已提交
1352
	struct inode *inode = file_inode(file);
1353 1354 1355 1356

	if (is_bad_inode(inode))
		return -EIO;

1357
	res = fuse_direct_io(io, iov, nr_segs, count, ppos, 0);
1358 1359 1360 1361

	fuse_invalidate_attr(inode);

	return res;
M
Miklos Szeredi 已提交
1362 1363
}

1364 1365 1366
static ssize_t fuse_direct_read(struct file *file, char __user *buf,
				     size_t count, loff_t *ppos)
{
1367
	struct fuse_io_priv io = { .async = 0, .file = file };
M
Miklos Szeredi 已提交
1368
	struct iovec iov = { .iov_base = buf, .iov_len = count };
1369
	return __fuse_direct_read(&io, &iov, 1, ppos, count);
1370 1371
}

1372 1373
static ssize_t __fuse_direct_write(struct fuse_io_priv *io,
				   const struct iovec *iov,
1374
				   unsigned long nr_segs, loff_t *ppos)
M
Miklos Szeredi 已提交
1375
{
1376
	struct file *file = io->file;
A
Al Viro 已提交
1377
	struct inode *inode = file_inode(file);
1378
	size_t count = iov_length(iov, nr_segs);
M
Miklos Szeredi 已提交
1379
	ssize_t res;
1380

1381
	res = generic_write_checks(file, ppos, &count, 0);
1382
	if (!res)
1383
		res = fuse_direct_io(io, iov, nr_segs, count, ppos, 1);
1384 1385 1386

	fuse_invalidate_attr(inode);

M
Miklos Szeredi 已提交
1387 1388 1389
	return res;
}

A
Anand Avati 已提交
1390 1391 1392
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
1393
	struct iovec iov = { .iov_base = (void __user *)buf, .iov_len = count };
A
Al Viro 已提交
1394
	struct inode *inode = file_inode(file);
A
Anand Avati 已提交
1395
	ssize_t res;
1396
	struct fuse_io_priv io = { .async = 0, .file = file };
A
Anand Avati 已提交
1397 1398 1399 1400 1401 1402

	if (is_bad_inode(inode))
		return -EIO;

	/* Don't allow parallel writes to the same file */
	mutex_lock(&inode->i_mutex);
1403
	res = __fuse_direct_write(&io, &iov, 1, ppos);
1404 1405
	if (res > 0)
		fuse_write_update_size(inode, *ppos);
A
Anand Avati 已提交
1406 1407 1408 1409 1410
	mutex_unlock(&inode->i_mutex);

	return res;
}

M
Miklos Szeredi 已提交
1411
static void fuse_writepage_free(struct fuse_conn *fc, struct fuse_req *req)
M
Miklos Szeredi 已提交
1412
{
1413 1414 1415 1416
	int i;

	for (i = 0; i < req->num_pages; i++)
		__free_page(req->pages[i]);
1417
	fuse_file_put(req->ff, false);
M
Miklos Szeredi 已提交
1418 1419 1420 1421 1422 1423 1424
}

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;
1425
	int i;
M
Miklos Szeredi 已提交
1426 1427

	list_del(&req->writepages_entry);
1428 1429 1430 1431 1432
	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
1433 1434 1435 1436 1437
	wake_up(&fi->page_waitq);
}

/* Called under fc->lock, may release and reacquire it */
static void fuse_send_writepage(struct fuse_conn *fc, struct fuse_req *req)
M
Miklos Szeredi 已提交
1438 1439
__releases(fc->lock)
__acquires(fc->lock)
M
Miklos Szeredi 已提交
1440 1441 1442 1443
{
	struct fuse_inode *fi = get_fuse_inode(req->inode);
	loff_t size = i_size_read(req->inode);
	struct fuse_write_in *inarg = &req->misc.write.in;
1444
	__u64 data_size = req->num_pages * PAGE_CACHE_SIZE;
M
Miklos Szeredi 已提交
1445 1446 1447 1448

	if (!fc->connected)
		goto out_free;

1449 1450
	if (inarg->offset + data_size <= size) {
		inarg->size = data_size;
M
Miklos Szeredi 已提交
1451
	} else if (inarg->offset < size) {
1452
		inarg->size = size - inarg->offset;
M
Miklos Szeredi 已提交
1453 1454 1455
	} else {
		/* Got truncated off completely */
		goto out_free;
M
Miklos Szeredi 已提交
1456
	}
M
Miklos Szeredi 已提交
1457 1458 1459

	req->in.args[1].size = inarg->size;
	fi->writectr++;
1460
	fuse_request_send_background_locked(fc, req);
M
Miklos Szeredi 已提交
1461 1462 1463 1464 1465 1466
	return;

 out_free:
	fuse_writepage_finish(fc, req);
	spin_unlock(&fc->lock);
	fuse_writepage_free(fc, req);
1467
	fuse_put_request(fc, req);
M
Miklos Szeredi 已提交
1468
	spin_lock(&fc->lock);
M
Miklos Szeredi 已提交
1469 1470
}

M
Miklos Szeredi 已提交
1471 1472 1473 1474 1475 1476 1477
/*
 * 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
1478 1479
__releases(fc->lock)
__acquires(fc->lock)
M
Miklos Szeredi 已提交
1480
{
M
Miklos Szeredi 已提交
1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);
	struct fuse_req *req;

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

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

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

1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518
static struct fuse_file *fuse_write_file(struct fuse_conn *fc,
					 struct fuse_inode *fi)
{
	struct fuse_file *ff;

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

	return ff;
}

M
Miklos Szeredi 已提交
1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529
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;

	set_page_writeback(page);

1530
	req = fuse_request_alloc_nofs(1);
M
Miklos Szeredi 已提交
1531 1532 1533
	if (!req)
		goto err;

1534
	req->background = 1; /* writeback always goes to bg_queue */
M
Miklos Szeredi 已提交
1535 1536 1537 1538
	tmp_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
	if (!tmp_page)
		goto err_free;

1539 1540
	req->ff = fuse_write_file(fc, fi);
	fuse_write_fill(req, req->ff, page_offset(page), 0);
M
Miklos Szeredi 已提交
1541 1542

	copy_highpage(tmp_page, page);
1543
	req->misc.write.in.write_flags |= FUSE_WRITE_CACHE;
1544
	req->in.argpages = 1;
M
Miklos Szeredi 已提交
1545 1546
	req->num_pages = 1;
	req->pages[0] = tmp_page;
1547
	req->page_descs[0].offset = 0;
1548
	req->page_descs[0].length = PAGE_SIZE;
M
Miklos Szeredi 已提交
1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
	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);

1561 1562
	end_page_writeback(page);

M
Miklos Szeredi 已提交
1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617
	return 0;

err_free:
	fuse_request_free(req);
err:
	end_page_writeback(page);
	return -ENOMEM;
}

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

	err = fuse_writepage_locked(page);
	unlock_page(page);

	return err;
}

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
 */
1618
static int fuse_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
M
Miklos Szeredi 已提交
1619
{
1620
	struct page *page = vmf->page;
M
Miklos Szeredi 已提交
1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
	/*
	 * Don't use page->mapping as it may become NULL from a
	 * concurrent truncate.
	 */
	struct inode *inode = vma->vm_file->f_mapping->host;

	fuse_wait_on_page_writeback(inode, page->index);
	return 0;
}

1631
static const struct vm_operations_struct fuse_file_vm_ops = {
M
Miklos Szeredi 已提交
1632 1633 1634
	.close		= fuse_vma_close,
	.fault		= filemap_fault,
	.page_mkwrite	= fuse_page_mkwrite,
1635
	.remap_pages	= generic_file_remap_pages,
M
Miklos Szeredi 已提交
1636 1637 1638 1639 1640
};

static int fuse_file_mmap(struct file *file, struct vm_area_struct *vma)
{
	if ((vma->vm_flags & VM_SHARED) && (vma->vm_flags & VM_MAYWRITE)) {
A
Al Viro 已提交
1641
		struct inode *inode = file_inode(file);
M
Miklos Szeredi 已提交
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655
		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);
	}
	file_accessed(file);
	vma->vm_ops = &fuse_file_vm_ops;
M
Miklos Szeredi 已提交
1656 1657 1658
	return 0;
}

1659 1660 1661 1662 1663 1664
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;

1665 1666
	invalidate_inode_pages2(file->f_mapping);

1667 1668 1669
	return generic_file_mmap(file, vma);
}

1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695
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;
}

static void fuse_lk_fill(struct fuse_req *req, struct file *file,
1696 1697
			 const struct file_lock *fl, int opcode, pid_t pid,
			 int flock)
1698
{
A
Al Viro 已提交
1699
	struct inode *inode = file_inode(file);
1700
	struct fuse_conn *fc = get_fuse_conn(inode);
1701 1702 1703 1704
	struct fuse_file *ff = file->private_data;
	struct fuse_lk_in *arg = &req->misc.lk_in;

	arg->fh = ff->fh;
1705
	arg->owner = fuse_lock_owner_id(fc, fl->fl_owner);
1706 1707 1708 1709
	arg->lk.start = fl->fl_start;
	arg->lk.end = fl->fl_end;
	arg->lk.type = fl->fl_type;
	arg->lk.pid = pid;
1710 1711
	if (flock)
		arg->lk_flags |= FUSE_LK_FLOCK;
1712 1713 1714 1715 1716 1717 1718 1719 1720
	req->in.h.opcode = opcode;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(*arg);
	req->in.args[0].value = arg;
}

static int fuse_getlk(struct file *file, struct file_lock *fl)
{
A
Al Viro 已提交
1721
	struct inode *inode = file_inode(file);
1722 1723 1724 1725 1726
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	struct fuse_lk_out outarg;
	int err;

M
Maxim Patlasov 已提交
1727
	req = fuse_get_req_nopages(fc);
1728 1729 1730
	if (IS_ERR(req))
		return PTR_ERR(req);

1731
	fuse_lk_fill(req, file, fl, FUSE_GETLK, 0, 0);
1732 1733 1734
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(outarg);
	req->out.args[0].value = &outarg;
1735
	fuse_request_send(fc, req);
1736 1737 1738 1739 1740 1741 1742 1743
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (!err)
		err = convert_fuse_file_lock(&outarg.lk, fl);

	return err;
}

1744
static int fuse_setlk(struct file *file, struct file_lock *fl, int flock)
1745
{
A
Al Viro 已提交
1746
	struct inode *inode = file_inode(file);
1747 1748 1749 1750 1751 1752
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	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
1753
	if (fl->fl_lmops && fl->fl_lmops->lm_grant) {
M
Miklos Szeredi 已提交
1754 1755 1756 1757
		/* NLM needs asynchronous locks, which we don't support yet */
		return -ENOLCK;
	}

1758 1759 1760 1761
	/* Unlock on close is handled by the flush method */
	if (fl->fl_flags & FL_CLOSE)
		return 0;

M
Maxim Patlasov 已提交
1762
	req = fuse_get_req_nopages(fc);
1763 1764 1765
	if (IS_ERR(req))
		return PTR_ERR(req);

1766
	fuse_lk_fill(req, file, fl, opcode, pid, flock);
1767
	fuse_request_send(fc, req);
1768
	err = req->out.h.error;
1769 1770 1771
	/* locking is restartable */
	if (err == -EINTR)
		err = -ERESTARTSYS;
1772 1773 1774 1775 1776 1777
	fuse_put_request(fc, req);
	return err;
}

static int fuse_file_lock(struct file *file, int cmd, struct file_lock *fl)
{
A
Al Viro 已提交
1778
	struct inode *inode = file_inode(file);
1779 1780 1781
	struct fuse_conn *fc = get_fuse_conn(inode);
	int err;

M
Miklos Szeredi 已提交
1782 1783 1784
	if (cmd == F_CANCELLK) {
		err = 0;
	} else if (cmd == F_GETLK) {
1785
		if (fc->no_lock) {
1786
			posix_test_lock(file, fl);
1787 1788 1789 1790 1791
			err = 0;
		} else
			err = fuse_getlk(file, fl);
	} else {
		if (fc->no_lock)
M
Miklos Szeredi 已提交
1792
			err = posix_lock_file(file, fl, NULL);
1793
		else
1794
			err = fuse_setlk(file, fl, 0);
1795 1796 1797 1798
	}
	return err;
}

1799 1800
static int fuse_file_flock(struct file *file, int cmd, struct file_lock *fl)
{
A
Al Viro 已提交
1801
	struct inode *inode = file_inode(file);
1802 1803 1804
	struct fuse_conn *fc = get_fuse_conn(inode);
	int err;

M
Miklos Szeredi 已提交
1805
	if (fc->no_flock) {
1806 1807
		err = flock_lock_file_wait(file, fl);
	} else {
M
Miklos Szeredi 已提交
1808 1809
		struct fuse_file *ff = file->private_data;

1810 1811
		/* emulate flock with POSIX locks */
		fl->fl_owner = (fl_owner_t) file;
M
Miklos Szeredi 已提交
1812
		ff->flock = true;
1813 1814 1815 1816 1817 1818
		err = fuse_setlk(file, fl, 1);
	}

	return err;
}

M
Miklos Szeredi 已提交
1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830
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);
	struct fuse_req *req;
	struct fuse_bmap_in inarg;
	struct fuse_bmap_out outarg;
	int err;

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

M
Maxim Patlasov 已提交
1831
	req = fuse_get_req_nopages(fc);
M
Miklos Szeredi 已提交
1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845
	if (IS_ERR(req))
		return 0;

	memset(&inarg, 0, sizeof(inarg));
	inarg.block = block;
	inarg.blocksize = inode->i_sb->s_blocksize;
	req->in.h.opcode = FUSE_BMAP;
	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;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(outarg);
	req->out.args[0].value = &outarg;
1846
	fuse_request_send(fc, req);
M
Miklos Szeredi 已提交
1847 1848 1849 1850 1851 1852 1853 1854
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS)
		fc->no_bmap = 1;

	return err ? 0 : outarg.block;
}

1855
static loff_t fuse_file_llseek(struct file *file, loff_t offset, int whence)
M
Miklos Szeredi 已提交
1856 1857
{
	loff_t retval;
A
Al Viro 已提交
1858
	struct inode *inode = file_inode(file);
M
Miklos Szeredi 已提交
1859

1860
	/* No i_mutex protection necessary for SEEK_CUR and SEEK_SET */
1861 1862
	if (whence == SEEK_CUR || whence == SEEK_SET)
		return generic_file_llseek(file, offset, whence);
1863

1864 1865 1866
	mutex_lock(&inode->i_mutex);
	retval = fuse_update_attributes(inode, NULL, file, NULL);
	if (!retval)
1867
		retval = generic_file_llseek(file, offset, whence);
M
Miklos Szeredi 已提交
1868
	mutex_unlock(&inode->i_mutex);
1869

M
Miklos Szeredi 已提交
1870 1871 1872
	return retval;
}

T
Tejun Heo 已提交
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886
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;

	iov_iter_init(&ii, iov, nr_segs, bytes, 0);

	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
1887
		void *kaddr;
T
Tejun Heo 已提交
1888

D
Dan Carpenter 已提交
1889
		kaddr = kmap(page);
T
Tejun Heo 已提交
1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909

		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;
		}

1910
		kunmap(page);
T
Tejun Heo 已提交
1911 1912 1913 1914 1915
	}

	return 0;
}

1916 1917 1918 1919 1920 1921
/*
 * 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
1922 1923 1924
static int fuse_copy_ioctl_iovec_old(struct iovec *dst, void *src,
				     size_t transferred, unsigned count,
				     bool is_compat)
1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953
{
#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
1954 1955 1956 1957 1958 1959
/* 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;

1960
	for (n = 0; n < count; n++, iov++) {
M
Miklos Szeredi 已提交
1961 1962 1963 1964 1965 1966 1967
		if (iov->iov_len > (size_t) max)
			return -ENOMEM;
		max -= iov->iov_len;
	}
	return 0;
}

M
Miklos Szeredi 已提交
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
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
2004 2005 2006 2007 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 2044 2045 2046 2047 2048 2049
/*
 * 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.
 */
2050 2051
long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
		   unsigned int flags)
T
Tejun Heo 已提交
2052 2053
{
	struct fuse_file *ff = file->private_data;
2054
	struct fuse_conn *fc = ff->fc;
T
Tejun Heo 已提交
2055 2056 2057 2058 2059 2060 2061 2062 2063
	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
2064
	struct iovec *iov_page = NULL;
T
Tejun Heo 已提交
2065 2066 2067 2068 2069
	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
2070 2071 2072 2073 2074 2075 2076
#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
2077
	/* assume all the iovs returned by client always fits in a page */
M
Miklos Szeredi 已提交
2078
	BUILD_BUG_ON(sizeof(struct fuse_ioctl_iovec) * FUSE_IOCTL_MAX_IOV > PAGE_SIZE);
T
Tejun Heo 已提交
2079 2080

	err = -ENOMEM;
2081
	pages = kcalloc(FUSE_MAX_PAGES_PER_REQ, sizeof(pages[0]), GFP_KERNEL);
M
Miklos Szeredi 已提交
2082
	iov_page = (struct iovec *) __get_free_page(GFP_KERNEL);
T
Tejun Heo 已提交
2083 2084 2085 2086 2087 2088 2089 2090
	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
2091
		struct iovec *iov = iov_page;
T
Tejun Heo 已提交
2092

M
Miklos Szeredi 已提交
2093
		iov->iov_base = (void __user *)arg;
T
Tejun Heo 已提交
2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128
		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
2129
	req = fuse_get_req(fc, num_pages);
T
Tejun Heo 已提交
2130 2131 2132 2133 2134 2135 2136
	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;
2137
	fuse_page_descs_length_init(req, 0, req->num_pages);
T
Tejun Heo 已提交
2138 2139 2140

	/* okay, let's send it to the client */
	req->in.h.opcode = FUSE_IOCTL;
2141
	req->in.h.nodeid = ff->nodeid;
T
Tejun Heo 已提交
2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162
	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;

2163
	fuse_request_send(fc, req);
T
Tejun Heo 已提交
2164 2165 2166 2167 2168 2169 2170 2171 2172
	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
2173
		void *vaddr;
T
Tejun Heo 已提交
2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192

		/* 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;

2193
		vaddr = kmap_atomic(pages[0]);
M
Miklos Szeredi 已提交
2194
		err = fuse_copy_ioctl_iovec(fc, iov_page, vaddr,
2195 2196
					    transferred, in_iovs + out_iovs,
					    (flags & FUSE_IOCTL_COMPAT) != 0);
2197
		kunmap_atomic(vaddr);
2198 2199
		if (err)
			goto out;
T
Tejun Heo 已提交
2200

M
Miklos Szeredi 已提交
2201
		in_iov = iov_page;
T
Tejun Heo 已提交
2202 2203
		out_iov = in_iov + in_iovs;

M
Miklos Szeredi 已提交
2204 2205 2206 2207 2208 2209 2210 2211
		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
2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222
		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
2223
	free_page((unsigned long) iov_page);
T
Tejun Heo 已提交
2224 2225 2226 2227 2228 2229
	while (num_pages)
		__free_page(pages[--num_pages]);
	kfree(pages);

	return err ? err : outarg.result;
}
2230
EXPORT_SYMBOL_GPL(fuse_do_ioctl);
T
Tejun Heo 已提交
2231

2232 2233
long fuse_ioctl_common(struct file *file, unsigned int cmd,
		       unsigned long arg, unsigned int flags)
2234
{
A
Al Viro 已提交
2235
	struct inode *inode = file_inode(file);
2236 2237
	struct fuse_conn *fc = get_fuse_conn(inode);

2238
	if (!fuse_allow_current_process(fc))
2239 2240 2241 2242 2243 2244 2245 2246
		return -EACCES;

	if (is_bad_inode(inode))
		return -EIO;

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

T
Tejun Heo 已提交
2247 2248 2249
static long fuse_file_ioctl(struct file *file, unsigned int cmd,
			    unsigned long arg)
{
2250
	return fuse_ioctl_common(file, cmd, arg, 0);
T
Tejun Heo 已提交
2251 2252 2253 2254 2255
}

static long fuse_file_compat_ioctl(struct file *file, unsigned int cmd,
				   unsigned long arg)
{
2256
	return fuse_ioctl_common(file, cmd, arg, FUSE_IOCTL_COMPAT);
T
Tejun Heo 已提交
2257 2258
}

T
Tejun Heo 已提交
2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309
/*
 * 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)) {
		struct rb_node **link, *parent;

		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);
}

2310
unsigned fuse_file_poll(struct file *file, poll_table *wait)
T
Tejun Heo 已提交
2311 2312
{
	struct fuse_file *ff = file->private_data;
2313
	struct fuse_conn *fc = ff->fc;
T
Tejun Heo 已提交
2314 2315 2316 2317 2318 2319 2320 2321 2322
	struct fuse_poll_in inarg = { .fh = ff->fh, .kh = ff->kh };
	struct fuse_poll_out outarg;
	struct fuse_req *req;
	int err;

	if (fc->no_poll)
		return DEFAULT_POLLMASK;

	poll_wait(file, &ff->poll_wait, wait);
E
Enke Chen 已提交
2323
	inarg.events = (__u32)poll_requested_events(wait);
T
Tejun Heo 已提交
2324 2325 2326 2327 2328 2329 2330 2331 2332 2333

	/*
	 * 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);
	}

M
Maxim Patlasov 已提交
2334
	req = fuse_get_req_nopages(fc);
T
Tejun Heo 已提交
2335
	if (IS_ERR(req))
2336
		return POLLERR;
T
Tejun Heo 已提交
2337 2338

	req->in.h.opcode = FUSE_POLL;
2339
	req->in.h.nodeid = ff->nodeid;
T
Tejun Heo 已提交
2340 2341 2342 2343 2344 2345
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->out.numargs = 1;
	req->out.args[0].size = sizeof(outarg);
	req->out.args[0].value = &outarg;
2346
	fuse_request_send(fc, req);
T
Tejun Heo 已提交
2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357
	err = req->out.h.error;
	fuse_put_request(fc, req);

	if (!err)
		return outarg.revents;
	if (err == -ENOSYS) {
		fc->no_poll = 1;
		return DEFAULT_POLLMASK;
	}
	return POLLERR;
}
2358
EXPORT_SYMBOL_GPL(fuse_file_poll);
T
Tejun Heo 已提交
2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383

/*
 * 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;
}

2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397
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);
}

2398 2399 2400 2401 2402
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
2403 2404 2405 2406 2407
static ssize_t
fuse_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
			loff_t offset, unsigned long nr_segs)
{
	ssize_t ret = 0;
2408 2409
	struct file *file = iocb->ki_filp;
	struct fuse_file *ff = file->private_data;
2410
	bool async_dio = ff->fc->async_dio;
A
Anand Avati 已提交
2411
	loff_t pos = 0;
2412 2413 2414
	struct inode *inode;
	loff_t i_size;
	size_t count = iov_length(iov, nr_segs);
2415
	struct fuse_io_priv *io;
A
Anand Avati 已提交
2416 2417

	pos = offset;
2418 2419
	inode = file->f_mapping->host;
	i_size = i_size_read(inode);
A
Anand Avati 已提交
2420

2421
	/* optimization for short read */
2422
	if (async_dio && rw != WRITE && offset + count > i_size) {
2423 2424
		if (offset >= i_size)
			return 0;
2425
		count = min_t(loff_t, count, fuse_round_up(i_size - offset));
2426 2427
	}

2428
	io = kmalloc(sizeof(struct fuse_io_priv), GFP_KERNEL);
2429 2430
	if (!io)
		return -ENOMEM;
2431 2432 2433 2434 2435 2436 2437
	spin_lock_init(&io->lock);
	io->reqs = 1;
	io->bytes = -1;
	io->size = 0;
	io->offset = offset;
	io->write = (rw == WRITE);
	io->err = 0;
2438
	io->file = file;
2439 2440
	/*
	 * By default, we want to optimize all I/Os with async request
2441
	 * submission to the client filesystem if supported.
2442
	 */
2443
	io->async = async_dio;
2444 2445 2446 2447 2448 2449 2450
	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.
	 */
2451
	if (!is_sync_kiocb(iocb) && (offset + count > i_size) && rw == WRITE)
2452
		io->async = false;
A
Anand Avati 已提交
2453

2454
	if (rw == WRITE)
2455
		ret = __fuse_direct_write(io, iov, nr_segs, &pos);
2456
	else
2457
		ret = __fuse_direct_read(io, iov, nr_segs, &pos, count);
2458

2459 2460 2461 2462
	if (io->async) {
		fuse_aio_complete(io, ret < 0 ? ret : 0, -1);

		/* we have a non-extending, async request, so return */
2463
		if (!is_sync_kiocb(iocb))
2464 2465 2466 2467 2468 2469 2470
			return -EIOCBQUEUED;

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

2471 2472 2473 2474 2475 2476
	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
2477 2478 2479 2480

	return ret;
}

2481 2482
static long fuse_file_fallocate(struct file *file, int mode, loff_t offset,
				loff_t length)
A
Anatol Pomozov 已提交
2483 2484
{
	struct fuse_file *ff = file->private_data;
2485
	struct inode *inode = file->f_inode;
2486
	struct fuse_inode *fi = get_fuse_inode(inode);
A
Anatol Pomozov 已提交
2487 2488 2489 2490 2491 2492 2493 2494 2495
	struct fuse_conn *fc = ff->fc;
	struct fuse_req *req;
	struct fuse_fallocate_in inarg = {
		.fh = ff->fh,
		.offset = offset,
		.length = length,
		.mode = mode
	};
	int err;
2496 2497
	bool lock_inode = !(mode & FALLOC_FL_KEEP_SIZE) ||
			   (mode & FALLOC_FL_PUNCH_HOLE);
A
Anatol Pomozov 已提交
2498

2499 2500 2501
	if (fc->no_fallocate)
		return -EOPNOTSUPP;

2502
	if (lock_inode) {
2503
		mutex_lock(&inode->i_mutex);
2504 2505 2506 2507 2508 2509 2510 2511 2512
		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);
		}
2513 2514
	}

2515 2516 2517
	if (!(mode & FALLOC_FL_KEEP_SIZE))
		set_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);

M
Maxim Patlasov 已提交
2518
	req = fuse_get_req_nopages(fc);
2519 2520 2521 2522
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		goto out;
	}
A
Anatol Pomozov 已提交
2523 2524 2525 2526 2527 2528 2529 2530

	req->in.h.opcode = FUSE_FALLOCATE;
	req->in.h.nodeid = ff->nodeid;
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	fuse_request_send(fc, req);
	err = req->out.h.error;
2531 2532 2533 2534
	if (err == -ENOSYS) {
		fc->no_fallocate = 1;
		err = -EOPNOTSUPP;
	}
A
Anatol Pomozov 已提交
2535 2536
	fuse_put_request(fc, req);

2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548
	if (err)
		goto out;

	/* we could have extended the file */
	if (!(mode & FALLOC_FL_KEEP_SIZE))
		fuse_write_update_size(inode, offset + length);

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

	fuse_invalidate_attr(inode);

2549
out:
2550 2551 2552
	if (!(mode & FALLOC_FL_KEEP_SIZE))
		clear_bit(FUSE_I_SIZE_UNSTABLE, &fi->state);

2553
	if (lock_inode)
2554 2555
		mutex_unlock(&inode->i_mutex);

A
Anatol Pomozov 已提交
2556 2557 2558
	return err;
}

2559
static const struct file_operations fuse_file_operations = {
M
Miklos Szeredi 已提交
2560
	.llseek		= fuse_file_llseek,
2561
	.read		= do_sync_read,
M
Miklos Szeredi 已提交
2562
	.aio_read	= fuse_file_aio_read,
2563
	.write		= do_sync_write,
N
Nick Piggin 已提交
2564
	.aio_write	= fuse_file_aio_write,
M
Miklos Szeredi 已提交
2565 2566 2567 2568 2569
	.mmap		= fuse_file_mmap,
	.open		= fuse_open,
	.flush		= fuse_flush,
	.release	= fuse_release,
	.fsync		= fuse_fsync,
2570
	.lock		= fuse_file_lock,
2571
	.flock		= fuse_file_flock,
2572
	.splice_read	= generic_file_splice_read,
T
Tejun Heo 已提交
2573 2574
	.unlocked_ioctl	= fuse_file_ioctl,
	.compat_ioctl	= fuse_file_compat_ioctl,
T
Tejun Heo 已提交
2575
	.poll		= fuse_file_poll,
A
Anatol Pomozov 已提交
2576
	.fallocate	= fuse_file_fallocate,
M
Miklos Szeredi 已提交
2577 2578
};

2579
static const struct file_operations fuse_direct_io_file_operations = {
M
Miklos Szeredi 已提交
2580
	.llseek		= fuse_file_llseek,
M
Miklos Szeredi 已提交
2581 2582
	.read		= fuse_direct_read,
	.write		= fuse_direct_write,
2583
	.mmap		= fuse_direct_mmap,
M
Miklos Szeredi 已提交
2584 2585 2586 2587
	.open		= fuse_open,
	.flush		= fuse_flush,
	.release	= fuse_release,
	.fsync		= fuse_fsync,
2588
	.lock		= fuse_file_lock,
2589
	.flock		= fuse_file_flock,
T
Tejun Heo 已提交
2590 2591
	.unlocked_ioctl	= fuse_file_ioctl,
	.compat_ioctl	= fuse_file_compat_ioctl,
T
Tejun Heo 已提交
2592
	.poll		= fuse_file_poll,
A
Anatol Pomozov 已提交
2593
	.fallocate	= fuse_file_fallocate,
2594
	/* no splice_read */
M
Miklos Szeredi 已提交
2595 2596
};

2597
static const struct address_space_operations fuse_file_aops  = {
M
Miklos Szeredi 已提交
2598
	.readpage	= fuse_readpage,
M
Miklos Szeredi 已提交
2599 2600
	.writepage	= fuse_writepage,
	.launder_page	= fuse_launder_page,
2601
	.readpages	= fuse_readpages,
M
Miklos Szeredi 已提交
2602
	.set_page_dirty	= __set_page_dirty_nobuffers,
M
Miklos Szeredi 已提交
2603
	.bmap		= fuse_bmap,
A
Anand Avati 已提交
2604
	.direct_IO	= fuse_direct_IO,
M
Miklos Szeredi 已提交
2605 2606 2607 2608
};

void fuse_init_file_inode(struct inode *inode)
{
2609 2610
	inode->i_fop = &fuse_file_operations;
	inode->i_data.a_ops = &fuse_file_aops;
M
Miklos Szeredi 已提交
2611
}