splice.c 39.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11
/*
 * "splice": joining two ropes together by interweaving their strands.
 *
 * This is the "extended pipe" functionality, where a pipe is used as
 * an arbitrary in-memory buffer. Think of a pipe as a small kernel
 * buffer that you can use to transfer data from one end to the other.
 *
 * The traditional unix read/write is extended with a "splice()" operation
 * that transfers data buffers to or from a pipe buffer.
 *
 * Named by Larry McVoy, original implementation from Linus, extended by
12 13
 * Jens to support splicing to files, network, direct splicing, etc and
 * fixing lots of bugs.
14
 *
15
 * Copyright (C) 2005-2006 Jens Axboe <axboe@kernel.dk>
16 17
 * Copyright (C) 2005-2006 Linus Torvalds <torvalds@osdl.org>
 * Copyright (C) 2006 Ingo Molnar <mingo@elte.hu>
18 19 20 21 22
 *
 */
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/pagemap.h>
23
#include <linux/splice.h>
24
#include <linux/mm_inline.h>
25
#include <linux/swap.h>
26 27
#include <linux/writeback.h>
#include <linux/buffer_head.h>
J
Jeff Garzik 已提交
28
#include <linux/module.h>
29
#include <linux/syscalls.h>
30
#include <linux/uio.h>
31
#include <linux/security.h>
32

33 34 35 36 37 38
/*
 * Attempt to steal a page from a pipe buffer. This should perhaps go into
 * a vm helper function, it's already simplified quite a bit by the
 * addition of remove_mapping(). If success is returned, the caller may
 * attempt to reuse this page for another destination.
 */
39
static int page_cache_pipe_buf_steal(struct pipe_inode_info *pipe,
40 41 42
				     struct pipe_buffer *buf)
{
	struct page *page = buf->page;
43
	struct address_space *mapping;
44

45 46
	lock_page(page);

47 48 49
	mapping = page_mapping(page);
	if (mapping) {
		WARN_ON(!PageUptodate(page));
50

51 52 53 54 55 56 57 58 59
		/*
		 * At least for ext2 with nobh option, we need to wait on
		 * writeback completing on this page, since we'll remove it
		 * from the pagecache.  Otherwise truncate wont wait on the
		 * page, allowing the disk blocks to be reused by someone else
		 * before we actually wrote our data to them. fs corruption
		 * ensues.
		 */
		wait_on_page_writeback(page);
60

61
		if (PagePrivate(page))
62
			try_to_release_page(page, GFP_KERNEL);
63

64 65 66 67 68 69 70 71
		/*
		 * If we succeeded in removing the mapping, set LRU flag
		 * and return good.
		 */
		if (remove_mapping(mapping, page)) {
			buf->flags |= PIPE_BUF_FLAG_LRU;
			return 0;
		}
72
	}
73

74 75 76 77 78 79
	/*
	 * Raced with truncate or failed to remove page from current
	 * address space, unlock and return failure.
	 */
	unlock_page(page);
	return 1;
80 81
}

82
static void page_cache_pipe_buf_release(struct pipe_inode_info *pipe,
83 84 85
					struct pipe_buffer *buf)
{
	page_cache_release(buf->page);
J
Jens Axboe 已提交
86
	buf->flags &= ~PIPE_BUF_FLAG_LRU;
87 88
}

89 90 91 92
/*
 * Check whether the contents of buf is OK to access. Since the content
 * is a page cache page, IO may be in flight.
 */
93 94
static int page_cache_pipe_buf_confirm(struct pipe_inode_info *pipe,
				       struct pipe_buffer *buf)
95 96
{
	struct page *page = buf->page;
97
	int err;
98 99

	if (!PageUptodate(page)) {
100 101 102 103
		lock_page(page);

		/*
		 * Page got truncated/unhashed. This will cause a 0-byte
I
Ingo Molnar 已提交
104
		 * splice, if this is the first page.
105 106 107 108 109
		 */
		if (!page->mapping) {
			err = -ENODATA;
			goto error;
		}
110

111
		/*
I
Ingo Molnar 已提交
112
		 * Uh oh, read-error from disk.
113 114 115 116 117 118 119
		 */
		if (!PageUptodate(page)) {
			err = -EIO;
			goto error;
		}

		/*
120
		 * Page is ok afterall, we are done.
121
		 */
122 123 124
		unlock_page(page);
	}

125
	return 0;
126 127
error:
	unlock_page(page);
128
	return err;
129 130
}

131
static const struct pipe_buf_operations page_cache_pipe_buf_ops = {
132
	.can_merge = 0,
133 134
	.map = generic_pipe_buf_map,
	.unmap = generic_pipe_buf_unmap,
135
	.confirm = page_cache_pipe_buf_confirm,
136
	.release = page_cache_pipe_buf_release,
137
	.steal = page_cache_pipe_buf_steal,
138
	.get = generic_pipe_buf_get,
139 140
};

141 142 143
static int user_page_pipe_buf_steal(struct pipe_inode_info *pipe,
				    struct pipe_buffer *buf)
{
144 145 146
	if (!(buf->flags & PIPE_BUF_FLAG_GIFT))
		return 1;

J
Jens Axboe 已提交
147
	buf->flags |= PIPE_BUF_FLAG_LRU;
148
	return generic_pipe_buf_steal(pipe, buf);
149 150
}

151
static const struct pipe_buf_operations user_page_pipe_buf_ops = {
152
	.can_merge = 0,
153 154
	.map = generic_pipe_buf_map,
	.unmap = generic_pipe_buf_unmap,
155
	.confirm = generic_pipe_buf_confirm,
156 157
	.release = page_cache_pipe_buf_release,
	.steal = user_page_pipe_buf_steal,
158
	.get = generic_pipe_buf_get,
159 160
};

161 162 163 164 165 166
/**
 * splice_to_pipe - fill passed data into a pipe
 * @pipe:	pipe to fill
 * @spd:	data to fill
 *
 * Description:
R
Randy Dunlap 已提交
167
 *    @spd contains a map of pages and len/offset tuples, along with
168 169 170
 *    the struct pipe_buf_operations associated with these pages. This
 *    function will link that data to the pipe.
 *
171
 */
172 173
ssize_t splice_to_pipe(struct pipe_inode_info *pipe,
		       struct splice_pipe_desc *spd)
174
{
175
	unsigned int spd_pages = spd->nr_pages;
176
	int ret, do_wakeup, page_nr;
177 178 179

	ret = 0;
	do_wakeup = 0;
180
	page_nr = 0;
181

182 183
	if (pipe->inode)
		mutex_lock(&pipe->inode->i_mutex);
184 185

	for (;;) {
186
		if (!pipe->readers) {
187 188 189 190 191 192
			send_sig(SIGPIPE, current, 0);
			if (!ret)
				ret = -EPIPE;
			break;
		}

193 194
		if (pipe->nrbufs < PIPE_BUFFERS) {
			int newbuf = (pipe->curbuf + pipe->nrbufs) & (PIPE_BUFFERS - 1);
195
			struct pipe_buffer *buf = pipe->bufs + newbuf;
196

197 198 199
			buf->page = spd->pages[page_nr];
			buf->offset = spd->partial[page_nr].offset;
			buf->len = spd->partial[page_nr].len;
200
			buf->private = spd->partial[page_nr].private;
201
			buf->ops = spd->ops;
202 203 204
			if (spd->flags & SPLICE_F_GIFT)
				buf->flags |= PIPE_BUF_FLAG_GIFT;

205
			pipe->nrbufs++;
206 207 208
			page_nr++;
			ret += buf->len;

209 210
			if (pipe->inode)
				do_wakeup = 1;
211

212
			if (!--spd->nr_pages)
213
				break;
214
			if (pipe->nrbufs < PIPE_BUFFERS)
215 216 217 218 219
				continue;

			break;
		}

220
		if (spd->flags & SPLICE_F_NONBLOCK) {
221 222 223 224 225
			if (!ret)
				ret = -EAGAIN;
			break;
		}

226 227 228 229 230 231 232
		if (signal_pending(current)) {
			if (!ret)
				ret = -ERESTARTSYS;
			break;
		}

		if (do_wakeup) {
233
			smp_mb();
234 235 236
			if (waitqueue_active(&pipe->wait))
				wake_up_interruptible_sync(&pipe->wait);
			kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
237 238 239
			do_wakeup = 0;
		}

240 241 242
		pipe->waiting_writers++;
		pipe_wait(pipe);
		pipe->waiting_writers--;
243 244
	}

245
	if (pipe->inode) {
246
		mutex_unlock(&pipe->inode->i_mutex);
247

248 249 250 251 252 253
		if (do_wakeup) {
			smp_mb();
			if (waitqueue_active(&pipe->wait))
				wake_up_interruptible(&pipe->wait);
			kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
		}
254 255
	}

256
	while (page_nr < spd_pages)
257
		spd->spd_release(spd, page_nr++);
258 259 260 261

	return ret;
}

262 263 264 265 266
static void spd_release_page(struct splice_pipe_desc *spd, unsigned int i)
{
	page_cache_release(spd->pages[i]);
}

267
static int
268 269 270
__generic_file_splice_read(struct file *in, loff_t *ppos,
			   struct pipe_inode_info *pipe, size_t len,
			   unsigned int flags)
271 272
{
	struct address_space *mapping = in->f_mapping;
F
Fengguang Wu 已提交
273
	unsigned int loff, nr_pages, req_pages;
274
	struct page *pages[PIPE_BUFFERS];
275
	struct partial_page partial[PIPE_BUFFERS];
276
	struct page *page;
277 278
	pgoff_t index, end_index;
	loff_t isize;
279
	int error, page_nr;
280 281 282 283 284
	struct splice_pipe_desc spd = {
		.pages = pages,
		.partial = partial,
		.flags = flags,
		.ops = &page_cache_pipe_buf_ops,
285
		.spd_release = spd_release_page,
286
	};
287

288
	index = *ppos >> PAGE_CACHE_SHIFT;
289
	loff = *ppos & ~PAGE_CACHE_MASK;
F
Fengguang Wu 已提交
290 291
	req_pages = (len + loff + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
	nr_pages = min(req_pages, (unsigned)PIPE_BUFFERS);
292

293 294 295 296
	/*
	 * Lookup the (hopefully) full range of pages we need.
	 */
	spd.nr_pages = find_get_pages_contig(mapping, index, nr_pages, pages);
297
	index += spd.nr_pages;
298

299 300
	/*
	 * If find_get_pages_contig() returned fewer pages than we needed,
301
	 * readahead/allocate the rest and fill in the holes.
302
	 */
303
	if (spd.nr_pages < nr_pages)
304 305
		page_cache_sync_readahead(mapping, &in->f_ra, in,
				index, req_pages - spd.nr_pages);
306

307
	error = 0;
308
	while (spd.nr_pages < nr_pages) {
309
		/*
310 311
		 * Page could be there, find_get_pages_contig() breaks on
		 * the first hole.
312
		 */
313 314 315
		page = find_get_page(mapping, index);
		if (!page) {
			/*
316
			 * page didn't exist, allocate one.
317 318 319 320 321 322
			 */
			page = page_cache_alloc_cold(mapping);
			if (!page)
				break;

			error = add_to_page_cache_lru(page, mapping, index,
H
Hugh Dickins 已提交
323
						mapping_gfp_mask(mapping));
324 325
			if (unlikely(error)) {
				page_cache_release(page);
326 327
				if (error == -EEXIST)
					continue;
328 329
				break;
			}
330 331 332 333 334
			/*
			 * add_to_page_cache() locks the page, unlock it
			 * to avoid convoluting the logic below even more.
			 */
			unlock_page(page);
335 336
		}

337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359
		pages[spd.nr_pages++] = page;
		index++;
	}

	/*
	 * Now loop over the map and see if we need to start IO on any
	 * pages, fill in the partial map, etc.
	 */
	index = *ppos >> PAGE_CACHE_SHIFT;
	nr_pages = spd.nr_pages;
	spd.nr_pages = 0;
	for (page_nr = 0; page_nr < nr_pages; page_nr++) {
		unsigned int this_len;

		if (!len)
			break;

		/*
		 * this_len is the max we'll use from this page
		 */
		this_len = min_t(unsigned long, len, PAGE_CACHE_SIZE - loff);
		page = pages[page_nr];

360
		if (PageReadahead(page))
361
			page_cache_async_readahead(mapping, &in->f_ra, in,
F
Fengguang Wu 已提交
362
					page, index, req_pages - page_nr);
363

364 365 366 367
		/*
		 * If the page isn't uptodate, we may need to start io on it
		 */
		if (!PageUptodate(page)) {
368 369 370 371
			/*
			 * If in nonblock mode then dont block on waiting
			 * for an in-flight io page
			 */
372
			if (flags & SPLICE_F_NONBLOCK) {
373 374
				if (TestSetPageLocked(page)) {
					error = -EAGAIN;
375
					break;
376
				}
377 378
			} else
				lock_page(page);
379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395

			/*
			 * page was truncated, stop here. if this isn't the
			 * first page, we'll just complete what we already
			 * added
			 */
			if (!page->mapping) {
				unlock_page(page);
				break;
			}
			/*
			 * page was already under io and is now done, great
			 */
			if (PageUptodate(page)) {
				unlock_page(page);
				goto fill_it;
			}
396

397 398 399 400
			/*
			 * need to read in the page
			 */
			error = mapping->a_ops->readpage(in, page);
401
			if (unlikely(error)) {
402 403 404 405 406 407
				/*
				 * We really should re-lookup the page here,
				 * but it complicates things a lot. Instead
				 * lets just do what we already stored, and
				 * we'll get it the next time we are called.
				 */
408
				if (error == AOP_TRUNCATED_PAGE)
409 410
					error = 0;

411 412
				break;
			}
413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
		}
fill_it:
		/*
		 * i_size must be checked after PageUptodate.
		 */
		isize = i_size_read(mapping->host);
		end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
		if (unlikely(!isize || index > end_index))
			break;

		/*
		 * if this is the last page, see if we need to shrink
		 * the length and stop
		 */
		if (end_index == index) {
			unsigned int plen;
429 430

			/*
431
			 * max good bytes in this page
432
			 */
433 434
			plen = ((isize - 1) & ~PAGE_CACHE_MASK) + 1;
			if (plen <= loff)
435 436 437
				break;

			/*
438
			 * force quit after adding this page
439
			 */
440 441
			this_len = min(this_len, plen - loff);
			len = this_len;
442
		}
443

444 445
		partial[page_nr].offset = loff;
		partial[page_nr].len = this_len;
446
		len -= this_len;
447
		loff = 0;
448 449
		spd.nr_pages++;
		index++;
450 451
	}

452
	/*
453
	 * Release any pages at the end, if we quit early. 'page_nr' is how far
454 455 456 457
	 * we got, 'nr_pages' is how many pages are in the map.
	 */
	while (page_nr < nr_pages)
		page_cache_release(pages[page_nr++]);
458
	in->f_ra.prev_pos = (loff_t)index << PAGE_CACHE_SHIFT;
459

460
	if (spd.nr_pages)
461
		return splice_to_pipe(pipe, &spd);
462

463
	return error;
464 465
}

466 467 468
/**
 * generic_file_splice_read - splice data from file to a pipe
 * @in:		file to splice from
469
 * @ppos:	position in @in
470 471 472 473
 * @pipe:	pipe to splice to
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
474 475 476 477 478
 * Description:
 *    Will read pages from given file and fill them into a pipe. Can be
 *    used as long as the address_space operations for the source implements
 *    a readpage() hook.
 *
479
 */
480 481 482
ssize_t generic_file_splice_read(struct file *in, loff_t *ppos,
				 struct pipe_inode_info *pipe, size_t len,
				 unsigned int flags)
483
{
484
	loff_t isize, left;
485
	int ret;
486 487 488 489 490 491 492 493

	isize = i_size_read(in->f_mapping->host);
	if (unlikely(*ppos >= isize))
		return 0;

	left = isize - *ppos;
	if (unlikely(left < len))
		len = left;
494

495 496
	ret = __generic_file_splice_read(in, ppos, pipe, len, flags);
	if (ret > 0)
497
		*ppos += ret;
498 499 500 501

	return ret;
}

502 503
EXPORT_SYMBOL(generic_file_splice_read);

504
/*
505
 * Send 'sd->len' bytes to socket from 'sd->file' at position 'sd->pos'
J
Jens Axboe 已提交
506
 * using sendpage(). Return the number of bytes sent.
507
 */
508
static int pipe_to_sendpage(struct pipe_inode_info *pipe,
509 510
			    struct pipe_buffer *buf, struct splice_desc *sd)
{
511
	struct file *file = sd->u.file;
512
	loff_t pos = sd->pos;
513
	int ret, more;
514

515
	ret = buf->ops->confirm(pipe, buf);
516 517
	if (!ret) {
		more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
518

519 520 521
		ret = file->f_op->sendpage(file, buf->page, buf->offset,
					   sd->len, &pos, more);
	}
522

J
Jens Axboe 已提交
523
	return ret;
524 525 526 527 528 529 530 531 532 533 534 535 536 537 538
}

/*
 * This is a little more tricky than the file -> pipe splicing. There are
 * basically three cases:
 *
 *	- Destination page already exists in the address space and there
 *	  are users of it. For that case we have no other option that
 *	  copying the data. Tough luck.
 *	- Destination page already exists in the address space, but there
 *	  are no users of it. Make sure it's uptodate, then drop it. Fall
 *	  through to last case.
 *	- Destination page does not exist, we can add the pipe page to
 *	  the page cache and avoid the copy.
 *
539 540 541 542 543 544
 * If asked to move pages to the output file (SPLICE_F_MOVE is set in
 * sd->flags), we attempt to migrate pages from the pipe to the output
 * file address space page cache. This is possible if no one else has
 * the pipe page referenced outside of the pipe and page cache. If
 * SPLICE_F_MOVE isn't set, or we cannot move the page, we simply create
 * a new page in the output file page cache and fill/dirty that.
545
 */
546
static int pipe_to_file(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
547 548
			struct splice_desc *sd)
{
549
	struct file *file = sd->u.file;
550
	struct address_space *mapping = file->f_mapping;
J
Jens Axboe 已提交
551
	unsigned int offset, this_len;
552
	struct page *page;
553
	void *fsdata;
554
	int ret;
555 556

	/*
557
	 * make sure the data in this buffer is uptodate
558
	 */
559
	ret = buf->ops->confirm(pipe, buf);
560 561
	if (unlikely(ret))
		return ret;
562 563 564

	offset = sd->pos & ~PAGE_CACHE_MASK;

J
Jens Axboe 已提交
565 566 567 568
	this_len = sd->len;
	if (this_len + offset > PAGE_CACHE_SIZE)
		this_len = PAGE_CACHE_SIZE - offset;

569 570 571 572
	ret = pagecache_write_begin(file, mapping, sd->pos, this_len,
				AOP_FLAG_UNINTERRUPTIBLE, &page, &fsdata);
	if (unlikely(ret))
		goto out;
573

574
	if (buf->page != page) {
575 576 577
		/*
		 * Careful, ->map() uses KM_USER0!
		 */
578
		char *src = buf->ops->map(pipe, buf, 1);
579
		char *dst = kmap_atomic(page, KM_USER1);
580

J
Jens Axboe 已提交
581
		memcpy(dst + offset, src + buf->offset, this_len);
582
		flush_dcache_page(page);
583
		kunmap_atomic(dst, KM_USER1);
584
		buf->ops->unmap(pipe, buf, src);
585
	}
586 587
	ret = pagecache_write_end(file, mapping, sd->pos, this_len, this_len,
				page, fsdata);
588 589 590 591
out:
	return ret;
}

592 593 594 595 596 597 598 599 600 601 602 603
/**
 * __splice_from_pipe - splice data from a pipe to given actor
 * @pipe:	pipe to splice from
 * @sd:		information to @actor
 * @actor:	handler that splices the data
 *
 * Description:
 *    This function does little more than loop over the pipe and call
 *    @actor to do the actual moving of a single struct pipe_buffer to
 *    the desired destination. See pipe_to_file, pipe_to_sendpage, or
 *    pipe_to_user.
 *
604
 */
J
Jens Axboe 已提交
605 606
ssize_t __splice_from_pipe(struct pipe_inode_info *pipe, struct splice_desc *sd,
			   splice_actor *actor)
607 608 609 610 611 612 613
{
	int ret, do_wakeup, err;

	ret = 0;
	do_wakeup = 0;

	for (;;) {
614 615
		if (pipe->nrbufs) {
			struct pipe_buffer *buf = pipe->bufs + pipe->curbuf;
616
			const struct pipe_buf_operations *ops = buf->ops;
617

J
Jens Axboe 已提交
618 619 620
			sd->len = buf->len;
			if (sd->len > sd->total_len)
				sd->len = sd->total_len;
621

J
Jens Axboe 已提交
622
			err = actor(pipe, buf, sd);
J
Jens Axboe 已提交
623
			if (err <= 0) {
624 625 626 627 628 629
				if (!ret && err != -ENODATA)
					ret = err;

				break;
			}

J
Jens Axboe 已提交
630 631 632 633
			ret += err;
			buf->offset += err;
			buf->len -= err;

J
Jens Axboe 已提交
634 635 636 637
			sd->len -= err;
			sd->pos += err;
			sd->total_len -= err;
			if (sd->len)
J
Jens Axboe 已提交
638
				continue;
I
Ingo Molnar 已提交
639

640 641
			if (!buf->len) {
				buf->ops = NULL;
642
				ops->release(pipe, buf);
643 644 645 646
				pipe->curbuf = (pipe->curbuf + 1) & (PIPE_BUFFERS - 1);
				pipe->nrbufs--;
				if (pipe->inode)
					do_wakeup = 1;
647 648
			}

J
Jens Axboe 已提交
649
			if (!sd->total_len)
650 651 652
				break;
		}

653
		if (pipe->nrbufs)
654
			continue;
655
		if (!pipe->writers)
656
			break;
657
		if (!pipe->waiting_writers) {
658 659 660 661
			if (ret)
				break;
		}

J
Jens Axboe 已提交
662
		if (sd->flags & SPLICE_F_NONBLOCK) {
663 664 665 666 667
			if (!ret)
				ret = -EAGAIN;
			break;
		}

668 669 670 671 672 673 674
		if (signal_pending(current)) {
			if (!ret)
				ret = -ERESTARTSYS;
			break;
		}

		if (do_wakeup) {
675
			smp_mb();
676 677 678
			if (waitqueue_active(&pipe->wait))
				wake_up_interruptible_sync(&pipe->wait);
			kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT);
679 680 681
			do_wakeup = 0;
		}

682
		pipe_wait(pipe);
683 684 685
	}

	if (do_wakeup) {
686
		smp_mb();
687 688 689
		if (waitqueue_active(&pipe->wait))
			wake_up_interruptible(&pipe->wait);
		kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT);
690 691 692 693
	}

	return ret;
}
M
Mark Fasheh 已提交
694
EXPORT_SYMBOL(__splice_from_pipe);
695

696 697 698 699 700 701 702 703 704 705 706 707 708 709
/**
 * splice_from_pipe - splice data from a pipe to a file
 * @pipe:	pipe to splice from
 * @out:	file to splice to
 * @ppos:	position in @out
 * @len:	how many bytes to splice
 * @flags:	splice modifier flags
 * @actor:	handler that splices the data
 *
 * Description:
 *    See __splice_from_pipe. This function locks the input and output inodes,
 *    otherwise it's identical to __splice_from_pipe().
 *
 */
710 711 712 713 714 715
ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
			 loff_t *ppos, size_t len, unsigned int flags,
			 splice_actor *actor)
{
	ssize_t ret;
	struct inode *inode = out->f_mapping->host;
J
Jens Axboe 已提交
716 717 718 719
	struct splice_desc sd = {
		.total_len = len,
		.flags = flags,
		.pos = *ppos,
720
		.u.file = out,
J
Jens Axboe 已提交
721
	};
722 723 724 725 726 727 728 729

	/*
	 * The actor worker might be calling ->prepare_write and
	 * ->commit_write. Most of the time, these expect i_mutex to
	 * be held. Since this may result in an ABBA deadlock with
	 * pipe->inode, we have to order lock acquiry here.
	 */
	inode_double_lock(inode, pipe->inode);
J
Jens Axboe 已提交
730
	ret = __splice_from_pipe(pipe, &sd, actor);
731 732 733 734 735 736 737 738 739
	inode_double_unlock(inode, pipe->inode);

	return ret;
}

/**
 * generic_file_splice_write_nolock - generic_file_splice_write without mutexes
 * @pipe:	pipe info
 * @out:	file to write to
740
 * @ppos:	position in @out
741 742 743
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
744 745 746 747
 * Description:
 *    Will either move or copy pages (determined by @flags options) from
 *    the given pipe inode to the given file. The caller is responsible
 *    for acquiring i_mutex on both inodes.
748 749 750 751 752 753 754 755
 *
 */
ssize_t
generic_file_splice_write_nolock(struct pipe_inode_info *pipe, struct file *out,
				 loff_t *ppos, size_t len, unsigned int flags)
{
	struct address_space *mapping = out->f_mapping;
	struct inode *inode = mapping->host;
J
Jens Axboe 已提交
756 757 758 759
	struct splice_desc sd = {
		.total_len = len,
		.flags = flags,
		.pos = *ppos,
760
		.u.file = out,
J
Jens Axboe 已提交
761
	};
762 763 764
	ssize_t ret;
	int err;

765
	err = remove_suid(out->f_path.dentry);
766 767 768
	if (unlikely(err))
		return err;

J
Jens Axboe 已提交
769
	ret = __splice_from_pipe(pipe, &sd, pipe_to_file);
770
	if (ret > 0) {
771 772
		unsigned long nr_pages;

773
		*ppos += ret;
774
		nr_pages = (ret + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
775 776 777 778 779 780 781 782 783 784 785 786

		/*
		 * If file or inode is SYNC and we actually wrote some data,
		 * sync it.
		 */
		if (unlikely((out->f_flags & O_SYNC) || IS_SYNC(inode))) {
			err = generic_osync_inode(inode, mapping,
						  OSYNC_METADATA|OSYNC_DATA);

			if (err)
				ret = err;
		}
787
		balance_dirty_pages_ratelimited_nr(mapping, nr_pages);
788 789 790 791 792 793 794
	}

	return ret;
}

EXPORT_SYMBOL(generic_file_splice_write_nolock);

795 796
/**
 * generic_file_splice_write - splice data from a pipe to a file
797
 * @pipe:	pipe info
798
 * @out:	file to write to
799
 * @ppos:	position in @out
800 801 802
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
803 804 805
 * Description:
 *    Will either move or copy pages (determined by @flags options) from
 *    the given pipe inode to the given file.
806 807
 *
 */
808 809
ssize_t
generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out,
810
			  loff_t *ppos, size_t len, unsigned int flags)
811
{
812
	struct address_space *mapping = out->f_mapping;
813
	struct inode *inode = mapping->host;
814
	int killsuid, killpriv;
815
	ssize_t ret;
816
	int err = 0;
817

818 819 820
	killpriv = security_inode_need_killpriv(out->f_path.dentry);
	killsuid = should_remove_suid(out->f_path.dentry);
	if (unlikely(killsuid || killpriv)) {
821
		mutex_lock(&inode->i_mutex);
822 823 824 825
		if (killpriv)
			err = security_inode_killpriv(out->f_path.dentry);
		if (!err && killsuid)
			err = __remove_suid(out->f_path.dentry, killsuid);
826 827 828 829
		mutex_unlock(&inode->i_mutex);
		if (err)
			return err;
	}
830

831
	ret = splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
J
Jens Axboe 已提交
832
	if (ret > 0) {
833 834
		unsigned long nr_pages;

J
Jens Axboe 已提交
835
		*ppos += ret;
836
		nr_pages = (ret + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
J
Jens Axboe 已提交
837 838 839 840 841 842 843 844 845 846

		/*
		 * If file or inode is SYNC and we actually wrote some data,
		 * sync it.
		 */
		if (unlikely((out->f_flags & O_SYNC) || IS_SYNC(inode))) {
			mutex_lock(&inode->i_mutex);
			err = generic_osync_inode(inode, mapping,
						  OSYNC_METADATA|OSYNC_DATA);
			mutex_unlock(&inode->i_mutex);
847

J
Jens Axboe 已提交
848 849 850
			if (err)
				ret = err;
		}
851
		balance_dirty_pages_ratelimited_nr(mapping, nr_pages);
852 853 854
	}

	return ret;
855 856
}

857 858
EXPORT_SYMBOL(generic_file_splice_write);

859 860
/**
 * generic_splice_sendpage - splice data from a pipe to a socket
861
 * @pipe:	pipe to splice from
862
 * @out:	socket to write to
863
 * @ppos:	position in @out
864 865 866
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
867 868 869
 * Description:
 *    Will send @len bytes from the pipe to a network socket. No data copying
 *    is involved.
870 871
 *
 */
872
ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, struct file *out,
873
				loff_t *ppos, size_t len, unsigned int flags)
874
{
875
	return splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_sendpage);
876 877
}

878
EXPORT_SYMBOL(generic_splice_sendpage);
J
Jeff Garzik 已提交
879

880 881 882
/*
 * Attempt to initiate a splice from pipe to file.
 */
883
static long do_splice_from(struct pipe_inode_info *pipe, struct file *out,
884
			   loff_t *ppos, size_t len, unsigned int flags)
885 886 887
{
	int ret;

888
	if (unlikely(!out->f_op || !out->f_op->splice_write))
889 890
		return -EINVAL;

891
	if (unlikely(!(out->f_mode & FMODE_WRITE)))
892 893
		return -EBADF;

894
	ret = rw_verify_area(WRITE, out, ppos, len);
895 896 897
	if (unlikely(ret < 0))
		return ret;

898
	return out->f_op->splice_write(pipe, out, ppos, len, flags);
899 900
}

901 902 903
/*
 * Attempt to initiate a splice from a file to a pipe.
 */
904 905 906
static long do_splice_to(struct file *in, loff_t *ppos,
			 struct pipe_inode_info *pipe, size_t len,
			 unsigned int flags)
907 908 909
{
	int ret;

910
	if (unlikely(!in->f_op || !in->f_op->splice_read))
911 912
		return -EINVAL;

913
	if (unlikely(!(in->f_mode & FMODE_READ)))
914 915
		return -EBADF;

916
	ret = rw_verify_area(READ, in, ppos, len);
917 918 919
	if (unlikely(ret < 0))
		return ret;

920
	return in->f_op->splice_read(in, ppos, pipe, len, flags);
921 922
}

923 924 925 926 927 928 929 930 931
/**
 * splice_direct_to_actor - splices data directly between two non-pipes
 * @in:		file to splice from
 * @sd:		actor information on where to splice to
 * @actor:	handles the data splicing
 *
 * Description:
 *    This is a special case helper to splice directly between two
 *    points, without requiring an explicit pipe. Internally an allocated
R
Randy Dunlap 已提交
932
 *    pipe is cached in the process, and reused during the lifetime of
933 934
 *    that process.
 *
J
Jens Axboe 已提交
935 936 937
 */
ssize_t splice_direct_to_actor(struct file *in, struct splice_desc *sd,
			       splice_direct_actor *actor)
938 939 940 941
{
	struct pipe_inode_info *pipe;
	long ret, bytes;
	umode_t i_mode;
J
Jens Axboe 已提交
942 943
	size_t len;
	int i, flags;
944 945 946 947 948 949

	/*
	 * We require the input being a regular file, as we don't want to
	 * randomly drop data for eg socket -> socket splicing. Use the
	 * piped splicing for that!
	 */
950
	i_mode = in->f_path.dentry->d_inode->i_mode;
951 952 953 954 955 956 957 958
	if (unlikely(!S_ISREG(i_mode) && !S_ISBLK(i_mode)))
		return -EINVAL;

	/*
	 * neither in nor out is a pipe, setup an internal pipe attached to
	 * 'out' and transfer the wanted data from 'in' to 'out' through that
	 */
	pipe = current->splice_pipe;
959
	if (unlikely(!pipe)) {
960 961 962 963 964 965
		pipe = alloc_pipe_info(NULL);
		if (!pipe)
			return -ENOMEM;

		/*
		 * We don't have an immediate reader, but we'll read the stuff
966
		 * out of the pipe right after the splice_to_pipe(). So set
967 968 969 970 971 972 973 974
		 * PIPE_READERS appropriately.
		 */
		pipe->readers = 1;

		current->splice_pipe = pipe;
	}

	/*
I
Ingo Molnar 已提交
975
	 * Do the splice.
976 977 978
	 */
	ret = 0;
	bytes = 0;
J
Jens Axboe 已提交
979 980 981 982 983 984 985
	len = sd->total_len;
	flags = sd->flags;

	/*
	 * Don't block on output, we have to drain the direct pipe.
	 */
	sd->flags &= ~SPLICE_F_NONBLOCK;
986 987

	while (len) {
988
		size_t read_len;
989
		loff_t pos = sd->pos;
990

991
		ret = do_splice_to(in, &pos, pipe, len, flags);
992
		if (unlikely(ret <= 0))
993 994 995
			goto out_release;

		read_len = ret;
J
Jens Axboe 已提交
996
		sd->total_len = read_len;
997 998 999 1000 1001 1002

		/*
		 * NOTE: nonblocking mode only applies to the input. We
		 * must not do the output in nonblocking mode as then we
		 * could get stuck data in the internal pipe:
		 */
J
Jens Axboe 已提交
1003
		ret = actor(pipe, sd);
1004
		if (unlikely(ret <= 0))
1005 1006 1007 1008
			goto out_release;

		bytes += ret;
		len -= ret;
1009
		sd->pos = pos;
1010

1011 1012
		if (ret < read_len)
			goto out_release;
1013 1014
	}

1015
done:
1016
	pipe->nrbufs = pipe->curbuf = 0;
1017
	file_accessed(in);
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
	return bytes;

out_release:
	/*
	 * If we did an incomplete transfer we must release
	 * the pipe buffers in question:
	 */
	for (i = 0; i < PIPE_BUFFERS; i++) {
		struct pipe_buffer *buf = pipe->bufs + i;

		if (buf->ops) {
			buf->ops->release(pipe, buf);
			buf->ops = NULL;
		}
	}

1034 1035
	if (!bytes)
		bytes = ret;
J
Jens Axboe 已提交
1036

1037
	goto done;
J
Jens Axboe 已提交
1038 1039 1040 1041 1042 1043
}
EXPORT_SYMBOL(splice_direct_to_actor);

static int direct_splice_actor(struct pipe_inode_info *pipe,
			       struct splice_desc *sd)
{
1044
	struct file *file = sd->u.file;
J
Jens Axboe 已提交
1045 1046 1047 1048

	return do_splice_from(pipe, file, &sd->pos, sd->total_len, sd->flags);
}

1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
/**
 * do_splice_direct - splices data directly between two files
 * @in:		file to splice from
 * @ppos:	input file offset
 * @out:	file to splice to
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
 * Description:
 *    For use by do_sendfile(). splice can easily emulate sendfile, but
 *    doing it in the application would incur an extra system call
 *    (splice in + splice out, as compared to just sendfile()). So this helper
 *    can splice directly through a process-private pipe.
 *
 */
J
Jens Axboe 已提交
1064 1065 1066 1067 1068 1069 1070 1071
long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
		      size_t len, unsigned int flags)
{
	struct splice_desc sd = {
		.len		= len,
		.total_len	= len,
		.flags		= flags,
		.pos		= *ppos,
1072
		.u.file		= out,
J
Jens Axboe 已提交
1073
	};
1074
	long ret;
J
Jens Axboe 已提交
1075 1076

	ret = splice_direct_to_actor(in, &sd, direct_splice_actor);
1077
	if (ret > 0)
T
Tom Zanussi 已提交
1078
		*ppos = sd.pos;
1079

J
Jens Axboe 已提交
1080
	return ret;
1081 1082
}

1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095
/*
 * After the inode slimming patch, i_pipe/i_bdev/i_cdev share the same
 * location, so checking ->i_pipe is not enough to verify that this is a
 * pipe.
 */
static inline struct pipe_inode_info *pipe_info(struct inode *inode)
{
	if (S_ISFIFO(inode->i_mode))
		return inode->i_pipe;

	return NULL;
}

1096 1097 1098
/*
 * Determine where to splice to/from.
 */
1099 1100 1101
static long do_splice(struct file *in, loff_t __user *off_in,
		      struct file *out, loff_t __user *off_out,
		      size_t len, unsigned int flags)
1102
{
1103
	struct pipe_inode_info *pipe;
1104
	loff_t offset, *off;
J
Jens Axboe 已提交
1105
	long ret;
1106

1107
	pipe = pipe_info(in->f_path.dentry->d_inode);
1108 1109 1110
	if (pipe) {
		if (off_in)
			return -ESPIPE;
1111 1112 1113
		if (off_out) {
			if (out->f_op->llseek == no_llseek)
				return -EINVAL;
1114
			if (copy_from_user(&offset, off_out, sizeof(loff_t)))
1115
				return -EFAULT;
1116 1117 1118
			off = &offset;
		} else
			off = &out->f_pos;
1119

J
Jens Axboe 已提交
1120 1121 1122 1123 1124 1125
		ret = do_splice_from(pipe, out, off, len, flags);

		if (off_out && copy_to_user(off_out, off, sizeof(loff_t)))
			ret = -EFAULT;

		return ret;
1126
	}
1127

1128
	pipe = pipe_info(out->f_path.dentry->d_inode);
1129 1130 1131
	if (pipe) {
		if (off_out)
			return -ESPIPE;
1132 1133 1134
		if (off_in) {
			if (in->f_op->llseek == no_llseek)
				return -EINVAL;
1135
			if (copy_from_user(&offset, off_in, sizeof(loff_t)))
1136
				return -EFAULT;
1137 1138 1139
			off = &offset;
		} else
			off = &in->f_pos;
1140

J
Jens Axboe 已提交
1141 1142 1143 1144 1145 1146
		ret = do_splice_to(in, off, pipe, len, flags);

		if (off_in && copy_to_user(off_in, off, sizeof(loff_t)))
			ret = -EFAULT;

		return ret;
1147
	}
1148 1149 1150 1151

	return -EINVAL;
}

1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
/*
 * Do a copy-from-user while holding the mmap_semaphore for reading, in a
 * manner safe from deadlocking with simultaneous mmap() (grabbing mmap_sem
 * for writing) and page faulting on the user memory pointed to by src.
 * This assumes that we will very rarely hit the partial != 0 path, or this
 * will not be a win.
 */
static int copy_from_user_mmap_sem(void *dst, const void __user *src, size_t n)
{
	int partial;

1163 1164 1165
	if (!access_ok(VERIFY_READ, src, n))
		return -EFAULT;

1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
	pagefault_disable();
	partial = __copy_from_user_inatomic(dst, src, n);
	pagefault_enable();

	/*
	 * Didn't copy everything, drop the mmap_sem and do a faulting copy
	 */
	if (unlikely(partial)) {
		up_read(&current->mm->mmap_sem);
		partial = copy_from_user(dst, src, n);
		down_read(&current->mm->mmap_sem);
	}

	return partial;
}

1182 1183 1184 1185 1186 1187 1188 1189 1190
/*
 * Map an iov into an array of pages and offset/length tupples. With the
 * partial_page structure, we can map several non-contiguous ranges into
 * our ones pages[] map instead of splitting that operation into pieces.
 * Could easily be exported as a generic helper for other users, in which
 * case one would probably want to add a 'max_nr_pages' parameter as well.
 */
static int get_iovec_page_array(const struct iovec __user *iov,
				unsigned int nr_vecs, struct page **pages,
1191
				struct partial_page *partial, int aligned)
1192 1193 1194 1195 1196 1197 1198
{
	int buffers = 0, error = 0;

	down_read(&current->mm->mmap_sem);

	while (nr_vecs) {
		unsigned long off, npages;
1199
		struct iovec entry;
1200 1201 1202 1203
		void __user *base;
		size_t len;
		int i;

1204 1205
		error = -EFAULT;
		if (copy_from_user_mmap_sem(&entry, iov, sizeof(entry)))
1206 1207
			break;

1208 1209 1210
		base = entry.iov_base;
		len = entry.iov_len;

1211 1212 1213
		/*
		 * Sanity check this iovec. 0 read succeeds.
		 */
1214
		error = 0;
1215 1216 1217
		if (unlikely(!len))
			break;
		error = -EFAULT;
1218
		if (!access_ok(VERIFY_READ, base, len))
1219 1220 1221 1222 1223 1224 1225
			break;

		/*
		 * Get this base offset and number of pages, then map
		 * in the user pages.
		 */
		off = (unsigned long) base & ~PAGE_MASK;
1226 1227 1228 1229 1230 1231 1232 1233 1234

		/*
		 * If asked for alignment, the offset must be zero and the
		 * length a multiple of the PAGE_SIZE.
		 */
		error = -EINVAL;
		if (aligned && (off || len & ~PAGE_MASK))
			break;

1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
		npages = (off + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
		if (npages > PIPE_BUFFERS - buffers)
			npages = PIPE_BUFFERS - buffers;

		error = get_user_pages(current, current->mm,
				       (unsigned long) base, npages, 0, 0,
				       &pages[buffers], NULL);

		if (unlikely(error <= 0))
			break;

		/*
		 * Fill this contiguous range into the partial page map.
		 */
		for (i = 0; i < error; i++) {
1250
			const int plen = min_t(size_t, len, PAGE_SIZE - off);
1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287

			partial[buffers].offset = off;
			partial[buffers].len = plen;

			off = 0;
			len -= plen;
			buffers++;
		}

		/*
		 * We didn't complete this iov, stop here since it probably
		 * means we have to move some of this into a pipe to
		 * be able to continue.
		 */
		if (len)
			break;

		/*
		 * Don't continue if we mapped fewer pages than we asked for,
		 * or if we mapped the max number of pages that we have
		 * room for.
		 */
		if (error < npages || buffers == PIPE_BUFFERS)
			break;

		nr_vecs--;
		iov++;
	}

	up_read(&current->mm->mmap_sem);

	if (buffers)
		return buffers;

	return error;
}

1288 1289 1290 1291 1292 1293
static int pipe_to_user(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
			struct splice_desc *sd)
{
	char *src;
	int ret;

1294
	ret = buf->ops->confirm(pipe, buf);
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
	if (unlikely(ret))
		return ret;

	/*
	 * See if we can use the atomic maps, by prefaulting in the
	 * pages and doing an atomic copy
	 */
	if (!fault_in_pages_writeable(sd->u.userptr, sd->len)) {
		src = buf->ops->map(pipe, buf, 1);
		ret = __copy_to_user_inatomic(sd->u.userptr, src + buf->offset,
							sd->len);
		buf->ops->unmap(pipe, buf, src);
		if (!ret) {
			ret = sd->len;
			goto out;
		}
	}

	/*
	 * No dice, use slow non-atomic map and copy
 	 */
	src = buf->ops->map(pipe, buf, 0);

	ret = sd->len;
	if (copy_to_user(sd->u.userptr, src + buf->offset, sd->len))
		ret = -EFAULT;

1322
	buf->ops->unmap(pipe, buf, src);
1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
out:
	if (ret > 0)
		sd->u.userptr += ret;
	return ret;
}

/*
 * For lack of a better implementation, implement vmsplice() to userspace
 * as a simple copy of the pipes pages to the user iov.
 */
static long vmsplice_to_user(struct file *file, const struct iovec __user *iov,
			     unsigned long nr_segs, unsigned int flags)
{
	struct pipe_inode_info *pipe;
	struct splice_desc sd;
	ssize_t size;
	int error;
	long ret;

	pipe = pipe_info(file->f_path.dentry->d_inode);
	if (!pipe)
		return -EBADF;

	if (pipe->inode)
		mutex_lock(&pipe->inode->i_mutex);

	error = ret = 0;
	while (nr_segs) {
		void __user *base;
		size_t len;

		/*
		 * Get user address base and length for this iovec.
		 */
		error = get_user(base, &iov->iov_base);
		if (unlikely(error))
			break;
		error = get_user(len, &iov->iov_len);
		if (unlikely(error))
			break;

		/*
		 * Sanity check this iovec. 0 read succeeds.
		 */
		if (unlikely(!len))
			break;
		if (unlikely(!base)) {
			error = -EFAULT;
			break;
		}

1374 1375 1376 1377 1378
		if (unlikely(!access_ok(VERIFY_WRITE, base, len))) {
			error = -EFAULT;
			break;
		}

1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410
		sd.len = 0;
		sd.total_len = len;
		sd.flags = flags;
		sd.u.userptr = base;
		sd.pos = 0;

		size = __splice_from_pipe(pipe, &sd, pipe_to_user);
		if (size < 0) {
			if (!ret)
				ret = size;

			break;
		}

		ret += size;

		if (size < len)
			break;

		nr_segs--;
		iov++;
	}

	if (pipe->inode)
		mutex_unlock(&pipe->inode->i_mutex);

	if (!ret)
		ret = error;

	return ret;
}

1411 1412 1413 1414 1415
/*
 * vmsplice splices a user address range into a pipe. It can be thought of
 * as splice-from-memory, where the regular splice is splice-from-file (or
 * to file). In both cases the output is a pipe, naturally.
 */
1416 1417
static long vmsplice_to_pipe(struct file *file, const struct iovec __user *iov,
			     unsigned long nr_segs, unsigned int flags)
1418
{
1419
	struct pipe_inode_info *pipe;
1420 1421 1422 1423 1424 1425 1426
	struct page *pages[PIPE_BUFFERS];
	struct partial_page partial[PIPE_BUFFERS];
	struct splice_pipe_desc spd = {
		.pages = pages,
		.partial = partial,
		.flags = flags,
		.ops = &user_page_pipe_buf_ops,
1427
		.spd_release = spd_release_page,
1428 1429
	};

1430
	pipe = pipe_info(file->f_path.dentry->d_inode);
1431
	if (!pipe)
1432 1433
		return -EBADF;

1434 1435
	spd.nr_pages = get_iovec_page_array(iov, nr_segs, pages, partial,
					    flags & SPLICE_F_GIFT);
1436 1437 1438
	if (spd.nr_pages <= 0)
		return spd.nr_pages;

1439
	return splice_to_pipe(pipe, &spd);
1440 1441
}

1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
/*
 * Note that vmsplice only really supports true splicing _from_ user memory
 * to a pipe, not the other way around. Splicing from user memory is a simple
 * operation that can be supported without any funky alignment restrictions
 * or nasty vm tricks. We simply map in the user memory and fill them into
 * a pipe. The reverse isn't quite as easy, though. There are two possible
 * solutions for that:
 *
 *	- memcpy() the data internally, at which point we might as well just
 *	  do a regular read() on the buffer anyway.
 *	- Lots of nasty vm tricks, that are neither fast nor flexible (it
 *	  has restriction limitations on both ends of the pipe).
 *
 * Currently we punt and implement it as a normal copy, see pipe_to_user().
 *
 */
1458 1459 1460 1461 1462 1463 1464
asmlinkage long sys_vmsplice(int fd, const struct iovec __user *iov,
			     unsigned long nr_segs, unsigned int flags)
{
	struct file *file;
	long error;
	int fput;

1465 1466 1467 1468 1469
	if (unlikely(nr_segs > UIO_MAXIOV))
		return -EINVAL;
	else if (unlikely(!nr_segs))
		return 0;

1470 1471 1472 1473
	error = -EBADF;
	file = fget_light(fd, &fput);
	if (file) {
		if (file->f_mode & FMODE_WRITE)
1474 1475 1476
			error = vmsplice_to_pipe(file, iov, nr_segs, flags);
		else if (file->f_mode & FMODE_READ)
			error = vmsplice_to_user(file, iov, nr_segs, flags);
1477 1478 1479 1480 1481 1482 1483

		fput_light(file, fput);
	}

	return error;
}

1484 1485 1486
asmlinkage long sys_splice(int fd_in, loff_t __user *off_in,
			   int fd_out, loff_t __user *off_out,
			   size_t len, unsigned int flags)
1487 1488 1489 1490 1491 1492 1493 1494 1495
{
	long error;
	struct file *in, *out;
	int fput_in, fput_out;

	if (unlikely(!len))
		return 0;

	error = -EBADF;
1496
	in = fget_light(fd_in, &fput_in);
1497 1498
	if (in) {
		if (in->f_mode & FMODE_READ) {
1499
			out = fget_light(fd_out, &fput_out);
1500 1501
			if (out) {
				if (out->f_mode & FMODE_WRITE)
1502 1503 1504
					error = do_splice(in, off_in,
							  out, off_out,
							  len, flags);
1505 1506 1507 1508 1509 1510 1511 1512 1513
				fput_light(out, fput_out);
			}
		}

		fput_light(in, fput_in);
	}

	return error;
}
1514

1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 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 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593
/*
 * Make sure there's data to read. Wait for input if we can, otherwise
 * return an appropriate error.
 */
static int link_ipipe_prep(struct pipe_inode_info *pipe, unsigned int flags)
{
	int ret;

	/*
	 * Check ->nrbufs without the inode lock first. This function
	 * is speculative anyways, so missing one is ok.
	 */
	if (pipe->nrbufs)
		return 0;

	ret = 0;
	mutex_lock(&pipe->inode->i_mutex);

	while (!pipe->nrbufs) {
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
		if (!pipe->writers)
			break;
		if (!pipe->waiting_writers) {
			if (flags & SPLICE_F_NONBLOCK) {
				ret = -EAGAIN;
				break;
			}
		}
		pipe_wait(pipe);
	}

	mutex_unlock(&pipe->inode->i_mutex);
	return ret;
}

/*
 * Make sure there's writeable room. Wait for room if we can, otherwise
 * return an appropriate error.
 */
static int link_opipe_prep(struct pipe_inode_info *pipe, unsigned int flags)
{
	int ret;

	/*
	 * Check ->nrbufs without the inode lock first. This function
	 * is speculative anyways, so missing one is ok.
	 */
	if (pipe->nrbufs < PIPE_BUFFERS)
		return 0;

	ret = 0;
	mutex_lock(&pipe->inode->i_mutex);

	while (pipe->nrbufs >= PIPE_BUFFERS) {
		if (!pipe->readers) {
			send_sig(SIGPIPE, current, 0);
			ret = -EPIPE;
			break;
		}
		if (flags & SPLICE_F_NONBLOCK) {
			ret = -EAGAIN;
			break;
		}
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
		pipe->waiting_writers++;
		pipe_wait(pipe);
		pipe->waiting_writers--;
	}

	mutex_unlock(&pipe->inode->i_mutex);
	return ret;
}

1594 1595 1596 1597 1598 1599 1600 1601
/*
 * Link contents of ipipe to opipe.
 */
static int link_pipe(struct pipe_inode_info *ipipe,
		     struct pipe_inode_info *opipe,
		     size_t len, unsigned int flags)
{
	struct pipe_buffer *ibuf, *obuf;
1602
	int ret = 0, i = 0, nbuf;
1603 1604 1605 1606 1607 1608

	/*
	 * Potential ABBA deadlock, work around it by ordering lock
	 * grabbing by inode address. Otherwise two different processes
	 * could deadlock (one doing tee from A -> B, the other from B -> A).
	 */
1609
	inode_double_lock(ipipe->inode, opipe->inode);
1610

1611
	do {
1612 1613 1614 1615 1616 1617 1618
		if (!opipe->readers) {
			send_sig(SIGPIPE, current, 0);
			if (!ret)
				ret = -EPIPE;
			break;
		}

1619 1620 1621 1622 1623 1624
		/*
		 * If we have iterated all input buffers or ran out of
		 * output room, break.
		 */
		if (i >= ipipe->nrbufs || opipe->nrbufs >= PIPE_BUFFERS)
			break;
1625

1626 1627
		ibuf = ipipe->bufs + ((ipipe->curbuf + i) & (PIPE_BUFFERS - 1));
		nbuf = (opipe->curbuf + opipe->nrbufs) & (PIPE_BUFFERS - 1);
1628 1629

		/*
1630 1631
		 * Get a reference to this pipe buffer,
		 * so we can copy the contents over.
1632
		 */
1633 1634 1635 1636 1637
		ibuf->ops->get(ipipe, ibuf);

		obuf = opipe->bufs + nbuf;
		*obuf = *ibuf;

1638
		/*
1639 1640
		 * Don't inherit the gift flag, we need to
		 * prevent multiple steals of this page.
1641
		 */
1642
		obuf->flags &= ~PIPE_BUF_FLAG_GIFT;
1643

1644 1645
		if (obuf->len > len)
			obuf->len = len;
1646

1647 1648 1649 1650 1651
		opipe->nrbufs++;
		ret += obuf->len;
		len -= obuf->len;
		i++;
	} while (len);
1652

1653 1654 1655 1656 1657 1658 1659
	/*
	 * return EAGAIN if we have the potential of some data in the
	 * future, otherwise just return 0
	 */
	if (!ret && ipipe->waiting_writers && (flags & SPLICE_F_NONBLOCK))
		ret = -EAGAIN;

1660
	inode_double_unlock(ipipe->inode, opipe->inode);
1661

1662 1663 1664 1665
	/*
	 * If we put data in the output pipe, wakeup any potential readers.
	 */
	if (ret > 0) {
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683
		smp_mb();
		if (waitqueue_active(&opipe->wait))
			wake_up_interruptible(&opipe->wait);
		kill_fasync(&opipe->fasync_readers, SIGIO, POLL_IN);
	}

	return ret;
}

/*
 * This is a tee(1) implementation that works on pipes. It doesn't copy
 * any data, it simply references the 'in' pages on the 'out' pipe.
 * The 'flags' used are the SPLICE_F_* variants, currently the only
 * applicable one is SPLICE_F_NONBLOCK.
 */
static long do_tee(struct file *in, struct file *out, size_t len,
		   unsigned int flags)
{
1684 1685
	struct pipe_inode_info *ipipe = pipe_info(in->f_path.dentry->d_inode);
	struct pipe_inode_info *opipe = pipe_info(out->f_path.dentry->d_inode);
1686
	int ret = -EINVAL;
1687 1688

	/*
1689 1690
	 * Duplicate the contents of ipipe to opipe without actually
	 * copying the data.
1691
	 */
1692 1693 1694 1695 1696 1697 1698 1699
	if (ipipe && opipe && ipipe != opipe) {
		/*
		 * Keep going, unless we encounter an error. The ipipe/opipe
		 * ordering doesn't really matter.
		 */
		ret = link_ipipe_prep(ipipe, flags);
		if (!ret) {
			ret = link_opipe_prep(opipe, flags);
1700
			if (!ret)
1701 1702 1703
				ret = link_pipe(ipipe, opipe, len, flags);
		}
	}
1704

1705
	return ret;
1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
}

asmlinkage long sys_tee(int fdin, int fdout, size_t len, unsigned int flags)
{
	struct file *in;
	int error, fput_in;

	if (unlikely(!len))
		return 0;

	error = -EBADF;
	in = fget_light(fdin, &fput_in);
	if (in) {
		if (in->f_mode & FMODE_READ) {
			int fput_out;
			struct file *out = fget_light(fdout, &fput_out);

			if (out) {
				if (out->f_mode & FMODE_WRITE)
					error = do_tee(in, out, len, flags);
				fput_light(out, fput_out);
			}
		}
 		fput_light(in, fput_in);
 	}

	return error;
}