splice.c 36.3 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 23 24
 *
 */
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/pagemap.h>
#include <linux/pipe_fs_i.h>
#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

32 33 34 35 36 37
struct partial_page {
	unsigned int offset;
	unsigned int len;
};

/*
38
 * Passed to splice_to_pipe
39 40 41 42 43 44 45 46 47
 */
struct splice_pipe_desc {
	struct page **pages;		/* page map */
	struct partial_page *partial;	/* pages[] may not be contig */
	int nr_pages;			/* number of pages in map */
	unsigned int flags;		/* splice flags */
	struct pipe_buf_operations *ops;/* ops associated with output pipe */
};

48 49 50 51 52 53
/*
 * 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.
 */
54
static int page_cache_pipe_buf_steal(struct pipe_inode_info *pipe,
55 56 57
				     struct pipe_buffer *buf)
{
	struct page *page = buf->page;
58
	struct address_space *mapping;
59

60 61
	lock_page(page);

62 63 64
	mapping = page_mapping(page);
	if (mapping) {
		WARN_ON(!PageUptodate(page));
65

66 67 68 69 70 71 72 73 74
		/*
		 * 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);
75

76
		if (PagePrivate(page))
77
			try_to_release_page(page, GFP_KERNEL);
78

79 80 81 82 83 84 85 86
		/*
		 * 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;
		}
87
	}
88

89 90 91 92 93 94
	/*
	 * Raced with truncate or failed to remove page from current
	 * address space, unlock and return failure.
	 */
	unlock_page(page);
	return 1;
95 96
}

97
static void page_cache_pipe_buf_release(struct pipe_inode_info *pipe,
98 99 100
					struct pipe_buffer *buf)
{
	page_cache_release(buf->page);
J
Jens Axboe 已提交
101
	buf->flags &= ~PIPE_BUF_FLAG_LRU;
102 103
}

104
static int page_cache_pipe_buf_pin(struct pipe_inode_info *pipe,
105
				   struct pipe_buffer *buf)
106 107
{
	struct page *page = buf->page;
108
	int err;
109 110

	if (!PageUptodate(page)) {
111 112 113 114
		lock_page(page);

		/*
		 * Page got truncated/unhashed. This will cause a 0-byte
I
Ingo Molnar 已提交
115
		 * splice, if this is the first page.
116 117 118 119 120
		 */
		if (!page->mapping) {
			err = -ENODATA;
			goto error;
		}
121

122
		/*
I
Ingo Molnar 已提交
123
		 * Uh oh, read-error from disk.
124 125 126 127 128 129 130
		 */
		if (!PageUptodate(page)) {
			err = -EIO;
			goto error;
		}

		/*
131
		 * Page is ok afterall, we are done.
132
		 */
133 134 135
		unlock_page(page);
	}

136
	return 0;
137 138
error:
	unlock_page(page);
139
	return err;
140 141
}

142 143
static struct pipe_buf_operations page_cache_pipe_buf_ops = {
	.can_merge = 0,
144 145 146
	.map = generic_pipe_buf_map,
	.unmap = generic_pipe_buf_unmap,
	.pin = page_cache_pipe_buf_pin,
147
	.release = page_cache_pipe_buf_release,
148
	.steal = page_cache_pipe_buf_steal,
149
	.get = generic_pipe_buf_get,
150 151
};

152 153 154
static int user_page_pipe_buf_steal(struct pipe_inode_info *pipe,
				    struct pipe_buffer *buf)
{
155 156 157
	if (!(buf->flags & PIPE_BUF_FLAG_GIFT))
		return 1;

J
Jens Axboe 已提交
158
	buf->flags |= PIPE_BUF_FLAG_LRU;
159
	return generic_pipe_buf_steal(pipe, buf);
160 161 162 163
}

static struct pipe_buf_operations user_page_pipe_buf_ops = {
	.can_merge = 0,
164 165 166
	.map = generic_pipe_buf_map,
	.unmap = generic_pipe_buf_unmap,
	.pin = generic_pipe_buf_pin,
167 168
	.release = page_cache_pipe_buf_release,
	.steal = user_page_pipe_buf_steal,
169
	.get = generic_pipe_buf_get,
170 171
};

172 173 174 175
/*
 * Pipe output worker. This sets up our pipe format with the page cache
 * pipe buffer operations. Otherwise very similar to the regular pipe_writev().
 */
176 177
static ssize_t splice_to_pipe(struct pipe_inode_info *pipe,
			      struct splice_pipe_desc *spd)
178
{
179
	int ret, do_wakeup, page_nr;
180 181 182

	ret = 0;
	do_wakeup = 0;
183
	page_nr = 0;
184

185 186
	if (pipe->inode)
		mutex_lock(&pipe->inode->i_mutex);
187 188

	for (;;) {
189
		if (!pipe->readers) {
190 191 192 193 194 195
			send_sig(SIGPIPE, current, 0);
			if (!ret)
				ret = -EPIPE;
			break;
		}

196 197
		if (pipe->nrbufs < PIPE_BUFFERS) {
			int newbuf = (pipe->curbuf + pipe->nrbufs) & (PIPE_BUFFERS - 1);
198
			struct pipe_buffer *buf = pipe->bufs + newbuf;
199

200 201 202 203
			buf->page = spd->pages[page_nr];
			buf->offset = spd->partial[page_nr].offset;
			buf->len = spd->partial[page_nr].len;
			buf->ops = spd->ops;
204 205 206
			if (spd->flags & SPLICE_F_GIFT)
				buf->flags |= PIPE_BUF_FLAG_GIFT;

207
			pipe->nrbufs++;
208 209 210
			page_nr++;
			ret += buf->len;

211 212
			if (pipe->inode)
				do_wakeup = 1;
213

214
			if (!--spd->nr_pages)
215
				break;
216
			if (pipe->nrbufs < PIPE_BUFFERS)
217 218 219 220 221
				continue;

			break;
		}

222
		if (spd->flags & SPLICE_F_NONBLOCK) {
223 224 225 226 227
			if (!ret)
				ret = -EAGAIN;
			break;
		}

228 229 230 231 232 233 234
		if (signal_pending(current)) {
			if (!ret)
				ret = -ERESTARTSYS;
			break;
		}

		if (do_wakeup) {
235
			smp_mb();
236 237 238
			if (waitqueue_active(&pipe->wait))
				wake_up_interruptible_sync(&pipe->wait);
			kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
239 240 241
			do_wakeup = 0;
		}

242 243 244
		pipe->waiting_writers++;
		pipe_wait(pipe);
		pipe->waiting_writers--;
245 246
	}

247 248
	if (pipe->inode)
		mutex_unlock(&pipe->inode->i_mutex);
249 250

	if (do_wakeup) {
251
		smp_mb();
252 253 254
		if (waitqueue_active(&pipe->wait))
			wake_up_interruptible(&pipe->wait);
		kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
255 256
	}

257 258
	while (page_nr < spd->nr_pages)
		page_cache_release(spd->pages[page_nr++]);
259 260 261 262

	return ret;
}

263
static int
264 265 266
__generic_file_splice_read(struct file *in, loff_t *ppos,
			   struct pipe_inode_info *pipe, size_t len,
			   unsigned int flags)
267 268
{
	struct address_space *mapping = in->f_mapping;
269
	unsigned int loff, nr_pages;
270
	struct page *pages[PIPE_BUFFERS];
271
	struct partial_page partial[PIPE_BUFFERS];
272
	struct page *page;
273 274
	pgoff_t index, end_index;
	loff_t isize;
275
	size_t total_len;
276
	int error, page_nr;
277 278 279 280 281 282
	struct splice_pipe_desc spd = {
		.pages = pages,
		.partial = partial,
		.flags = flags,
		.ops = &page_cache_pipe_buf_ops,
	};
283

284
	index = *ppos >> PAGE_CACHE_SHIFT;
285 286
	loff = *ppos & ~PAGE_CACHE_MASK;
	nr_pages = (len + loff + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
287 288 289 290 291

	if (nr_pages > PIPE_BUFFERS)
		nr_pages = PIPE_BUFFERS;

	/*
I
Ingo Molnar 已提交
292
	 * Initiate read-ahead on this page range. however, don't call into
293 294
	 * read-ahead if this is a non-zero offset (we are likely doing small
	 * chunk splice and the page is already there) for a single page.
295
	 */
296 297
	if (!loff || nr_pages > 1)
		page_cache_readahead(mapping, &in->f_ra, in, index, nr_pages);
298 299

	/*
I
Ingo Molnar 已提交
300
	 * Now fill in the holes:
301
	 */
302
	error = 0;
303
	total_len = 0;
304

305 306 307 308
	/*
	 * Lookup the (hopefully) full range of pages we need.
	 */
	spd.nr_pages = find_get_pages_contig(mapping, index, nr_pages, pages);
309

310 311 312 313 314 315
	/*
	 * If find_get_pages_contig() returned fewer pages than we needed,
	 * allocate the rest.
	 */
	index += spd.nr_pages;
	while (spd.nr_pages < nr_pages) {
316
		/*
317 318
		 * Page could be there, find_get_pages_contig() breaks on
		 * the first hole.
319
		 */
320 321
		page = find_get_page(mapping, index);
		if (!page) {
322 323 324 325 326 327
			/*
			 * Make sure the read-ahead engine is notified
			 * about this failure.
			 */
			handle_ra_miss(mapping, &in->f_ra, index);

328
			/*
329
			 * page didn't exist, allocate one.
330 331 332 333 334 335
			 */
			page = page_cache_alloc_cold(mapping);
			if (!page)
				break;

			error = add_to_page_cache_lru(page, mapping, index,
336
					      GFP_KERNEL);
337 338
			if (unlikely(error)) {
				page_cache_release(page);
339 340
				if (error == -EEXIST)
					continue;
341 342
				break;
			}
343 344 345 346 347
			/*
			 * add_to_page_cache() locks the page, unlock it
			 * to avoid convoluting the logic below even more.
			 */
			unlock_page(page);
348 349
		}

350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372
		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];

373 374 375 376
		/*
		 * If the page isn't uptodate, we may need to start io on it
		 */
		if (!PageUptodate(page)) {
377 378 379 380 381 382 383
			/*
			 * If in nonblock mode then dont block on waiting
			 * for an in-flight io page
			 */
			if (flags & SPLICE_F_NONBLOCK)
				break;

384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401
			lock_page(page);

			/*
			 * 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;
			}
402

403 404 405 406
			/*
			 * need to read in the page
			 */
			error = mapping->a_ops->readpage(in, page);
407
			if (unlikely(error)) {
408 409 410 411 412 413
				/*
				 * 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.
				 */
414
				if (error == AOP_TRUNCATED_PAGE)
415 416
					error = 0;

417 418
				break;
			}
419 420 421 422 423 424

			/*
			 * i_size must be checked after ->readpage().
			 */
			isize = i_size_read(mapping->host);
			end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
425
			if (unlikely(!isize || index > end_index))
426 427 428 429 430 431 432 433
				break;

			/*
			 * if this is the last page, see if we need to shrink
			 * the length and stop
			 */
			if (end_index == index) {
				loff = PAGE_CACHE_SIZE - (isize & ~PAGE_CACHE_MASK);
434
				if (total_len + loff > isize)
435 436 437 438
					break;
				/*
				 * force quit after adding this page
				 */
439
				len = this_len;
440
				this_len = min(this_len, loff);
441
				loff = 0;
442
			}
443
		}
444
fill_it:
445 446
		partial[page_nr].offset = loff;
		partial[page_nr].len = this_len;
447
		len -= this_len;
448
		total_len += this_len;
449
		loff = 0;
450 451
		spd.nr_pages++;
		index++;
452 453
	}

454 455 456 457 458 459 460
	/*
	 * Release any pages at the end, if we quit early. 'i' is how far
	 * we got, 'nr_pages' is how many pages are in the map.
	 */
	while (page_nr < nr_pages)
		page_cache_release(pages[page_nr++]);

461
	if (spd.nr_pages)
462
		return splice_to_pipe(pipe, &spd);
463

464
	return error;
465 466
}

467 468 469 470 471 472 473 474 475
/**
 * generic_file_splice_read - splice data from file to a pipe
 * @in:		file to splice from
 * @pipe:	pipe to splice to
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
 * Will read pages from given file and fill them into a pipe.
 */
476 477 478
ssize_t generic_file_splice_read(struct file *in, loff_t *ppos,
				 struct pipe_inode_info *pipe, size_t len,
				 unsigned int flags)
479 480 481 482 483 484
{
	ssize_t spliced;
	int ret;

	ret = 0;
	spliced = 0;
485

486
	while (len) {
487
		ret = __generic_file_splice_read(in, ppos, pipe, len, flags);
488

489
		if (ret < 0)
490
			break;
491 492 493 494 495 496 497 498
		else if (!ret) {
			if (spliced)
				break;
			if (flags & SPLICE_F_NONBLOCK) {
				ret = -EAGAIN;
				break;
			}
		}
499

500
		*ppos += ret;
501 502 503 504 505 506 507 508 509 510
		len -= ret;
		spliced += ret;
	}

	if (spliced)
		return spliced;

	return ret;
}

511 512
EXPORT_SYMBOL(generic_file_splice_read);

513
/*
514
 * Send 'sd->len' bytes to socket from 'sd->file' at position 'sd->pos'
J
Jens Axboe 已提交
515
 * using sendpage(). Return the number of bytes sent.
516
 */
517
static int pipe_to_sendpage(struct pipe_inode_info *pipe,
518 519 520 521
			    struct pipe_buffer *buf, struct splice_desc *sd)
{
	struct file *file = sd->file;
	loff_t pos = sd->pos;
522
	int ret, more;
523

524
	ret = buf->ops->pin(pipe, buf);
525 526
	if (!ret) {
		more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
527

528 529 530
		ret = file->f_op->sendpage(file, buf->page, buf->offset,
					   sd->len, &pos, more);
	}
531

J
Jens Axboe 已提交
532
	return ret;
533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
}

/*
 * 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.
 *
548 549 550 551 552 553
 * 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.
554
 */
555
static int pipe_to_file(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
556 557 558 559
			struct splice_desc *sd)
{
	struct file *file = sd->file;
	struct address_space *mapping = file->f_mapping;
J
Jens Axboe 已提交
560
	unsigned int offset, this_len;
561 562
	struct page *page;
	pgoff_t index;
563
	int ret;
564 565

	/*
566
	 * make sure the data in this buffer is uptodate
567
	 */
568
	ret = buf->ops->pin(pipe, buf);
569 570
	if (unlikely(ret))
		return ret;
571 572 573 574

	index = sd->pos >> PAGE_CACHE_SHIFT;
	offset = sd->pos & ~PAGE_CACHE_MASK;

J
Jens Axboe 已提交
575 576 577 578
	this_len = sd->len;
	if (this_len + offset > PAGE_CACHE_SIZE)
		this_len = PAGE_CACHE_SIZE - offset;

579
	/*
580 581
	 * Reuse buf page, if SPLICE_F_MOVE is set and we are doing a full
	 * page.
582
	 */
583
	if ((sd->flags & SPLICE_F_MOVE) && this_len == PAGE_CACHE_SIZE) {
584
		/*
J
Jens Axboe 已提交
585 586 587
		 * If steal succeeds, buf->page is now pruned from the
		 * pagecache and we can reuse it. The page will also be
		 * locked on successful return.
588
		 */
589
		if (buf->ops->steal(pipe, buf))
590 591 592
			goto find_page;

		page = buf->page;
593
		if (add_to_page_cache(page, mapping, index, GFP_KERNEL)) {
594
			unlock_page(page);
595
			goto find_page;
596
		}
J
Jens Axboe 已提交
597 598 599 600 601

		page_cache_get(page);

		if (!(buf->flags & PIPE_BUF_FLAG_LRU))
			lru_cache_add(page);
602 603
	} else {
find_page:
604 605 606 607 608
		page = find_lock_page(mapping, index);
		if (!page) {
			ret = -ENOMEM;
			page = page_cache_alloc_cold(mapping);
			if (unlikely(!page))
609
				goto out_ret;
610 611 612 613 614

			/*
			 * This will also lock the page
			 */
			ret = add_to_page_cache_lru(page, mapping, index,
615
						    GFP_KERNEL);
616 617 618
			if (unlikely(ret))
				goto out;
		}
619 620

		/*
621 622 623 624
		 * We get here with the page locked. If the page is also
		 * uptodate, we don't need to do more. If it isn't, we
		 * may need to bring it in if we are not going to overwrite
		 * the full page.
625 626
		 */
		if (!PageUptodate(page)) {
J
Jens Axboe 已提交
627
			if (this_len < PAGE_CACHE_SIZE) {
628 629 630 631 632 633 634 635
				ret = mapping->a_ops->readpage(file, page);
				if (unlikely(ret))
					goto out;

				lock_page(page);

				if (!PageUptodate(page)) {
					/*
I
Ingo Molnar 已提交
636
					 * Page got invalidated, repeat.
637 638 639 640 641 642 643 644
					 */
					if (!page->mapping) {
						unlock_page(page);
						page_cache_release(page);
						goto find_page;
					}
					ret = -EIO;
					goto out;
645
				}
646
			} else
647
				SetPageUptodate(page);
648 649 650
		}
	}

J
Jens Axboe 已提交
651
	ret = mapping->a_ops->prepare_write(file, page, offset, offset+this_len);
652 653 654 655 656
	if (unlikely(ret)) {
		loff_t isize = i_size_read(mapping->host);

		if (ret != AOP_TRUNCATED_PAGE)
			unlock_page(page);
657
		page_cache_release(page);
658 659 660 661 662 663 664 665 666 667
		if (ret == AOP_TRUNCATED_PAGE)
			goto find_page;

		/*
		 * prepare_write() may have instantiated a few blocks
		 * outside i_size.  Trim these off again.
		 */
		if (sd->pos + this_len > isize)
			vmtruncate(mapping->host, isize);

668
		goto out_ret;
669
	}
670

671
	if (buf->page != page) {
672 673 674
		/*
		 * Careful, ->map() uses KM_USER0!
		 */
675
		char *src = buf->ops->map(pipe, buf, 1);
676
		char *dst = kmap_atomic(page, KM_USER1);
677

J
Jens Axboe 已提交
678
		memcpy(dst + offset, src + buf->offset, this_len);
679
		flush_dcache_page(page);
680
		kunmap_atomic(dst, KM_USER1);
681
		buf->ops->unmap(pipe, buf, src);
682
	}
683

J
Jens Axboe 已提交
684
	ret = mapping->a_ops->commit_write(file, page, offset, offset+this_len);
685 686 687 688 689 690 691 692 693
	if (!ret) {
		/*
		 * Return the number of bytes written and mark page as
		 * accessed, we are now done!
		 */
		ret = this_len;
		mark_page_accessed(page);
		balance_dirty_pages_ratelimited(mapping);
	} else if (ret == AOP_TRUNCATED_PAGE) {
694 695
		page_cache_release(page);
		goto find_page;
696
	}
697
out:
698
	page_cache_release(page);
699
	unlock_page(page);
700
out_ret:
701 702 703
	return ret;
}

704 705 706 707 708
/*
 * Pipe input worker. Most of this logic works like a regular pipe, the
 * key here is the 'actor' worker passed in that actually moves the data
 * to the wanted destination. See pipe_to_file/pipe_to_sendpage above.
 */
709 710 711
static 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)
712 713 714 715 716 717 718 719 720 721
{
	int ret, do_wakeup, err;
	struct splice_desc sd;

	ret = 0;
	do_wakeup = 0;

	sd.total_len = len;
	sd.flags = flags;
	sd.file = out;
722
	sd.pos = *ppos;
723 724

	for (;;) {
725 726
		if (pipe->nrbufs) {
			struct pipe_buffer *buf = pipe->bufs + pipe->curbuf;
727 728 729 730 731 732
			struct pipe_buf_operations *ops = buf->ops;

			sd.len = buf->len;
			if (sd.len > sd.total_len)
				sd.len = sd.total_len;

733
			err = actor(pipe, buf, &sd);
J
Jens Axboe 已提交
734
			if (err <= 0) {
735 736 737 738 739 740
				if (!ret && err != -ENODATA)
					ret = err;

				break;
			}

J
Jens Axboe 已提交
741 742 743 744 745 746 747 748 749
			ret += err;
			buf->offset += err;
			buf->len -= err;

			sd.len -= err;
			sd.pos += err;
			sd.total_len -= err;
			if (sd.len)
				continue;
I
Ingo Molnar 已提交
750

751 752
			if (!buf->len) {
				buf->ops = NULL;
753
				ops->release(pipe, buf);
754 755 756 757
				pipe->curbuf = (pipe->curbuf + 1) & (PIPE_BUFFERS - 1);
				pipe->nrbufs--;
				if (pipe->inode)
					do_wakeup = 1;
758 759 760 761 762 763
			}

			if (!sd.total_len)
				break;
		}

764
		if (pipe->nrbufs)
765
			continue;
766
		if (!pipe->writers)
767
			break;
768
		if (!pipe->waiting_writers) {
769 770 771 772
			if (ret)
				break;
		}

773 774 775 776 777 778
		if (flags & SPLICE_F_NONBLOCK) {
			if (!ret)
				ret = -EAGAIN;
			break;
		}

779 780 781 782 783 784 785
		if (signal_pending(current)) {
			if (!ret)
				ret = -ERESTARTSYS;
			break;
		}

		if (do_wakeup) {
786
			smp_mb();
787 788 789
			if (waitqueue_active(&pipe->wait))
				wake_up_interruptible_sync(&pipe->wait);
			kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT);
790 791 792
			do_wakeup = 0;
		}

793
		pipe_wait(pipe);
794 795 796
	}

	if (do_wakeup) {
797
		smp_mb();
798 799 800
		if (waitqueue_active(&pipe->wait))
			wake_up_interruptible(&pipe->wait);
		kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT);
801 802 803 804 805
	}

	return ret;
}

806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
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;

	/*
	 * 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);
	ret = __splice_from_pipe(pipe, out, ppos, len, flags, actor);
	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
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
 * 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.
 *
 */
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;
	ssize_t ret;
	int err;

847
	err = remove_suid(out->f_path.dentry);
848 849 850
	if (unlikely(err))
		return err;

851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872
	ret = __splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
	if (ret > 0) {
		*ppos += ret;

		/*
		 * 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;
		}
	}

	return ret;
}

EXPORT_SYMBOL(generic_file_splice_write_nolock);

873 874
/**
 * generic_file_splice_write - splice data from a pipe to a file
875
 * @pipe:	pipe info
876 877 878 879 880 881 882 883
 * @out:	file to write to
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
 * Will either move or copy pages (determined by @flags options) from
 * the given pipe inode to the given file.
 *
 */
884 885
ssize_t
generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out,
886
			  loff_t *ppos, size_t len, unsigned int flags)
887
{
888
	struct address_space *mapping = out->f_mapping;
889
	struct inode *inode = mapping->host;
890
	ssize_t ret;
891 892
	int err;

893
	err = should_remove_suid(out->f_path.dentry);
894 895
	if (unlikely(err)) {
		mutex_lock(&inode->i_mutex);
896
		err = __remove_suid(out->f_path.dentry, err);
897 898 899 900
		mutex_unlock(&inode->i_mutex);
		if (err)
			return err;
	}
901

902
	ret = splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
J
Jens Axboe 已提交
903 904 905 906 907 908 909 910 911 912 913 914
	if (ret > 0) {
		*ppos += ret;

		/*
		 * 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);
915

J
Jens Axboe 已提交
916 917 918
			if (err)
				ret = err;
		}
919 920 921
	}

	return ret;
922 923
}

924 925
EXPORT_SYMBOL(generic_file_splice_write);

926 927 928 929 930 931 932 933 934 935 936
/**
 * generic_splice_sendpage - splice data from a pipe to a socket
 * @inode:	pipe inode
 * @out:	socket to write to
 * @len:	number of bytes to splice
 * @flags:	splice modifier flags
 *
 * Will send @len bytes from the pipe to a network socket. No data copying
 * is involved.
 *
 */
937
ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, struct file *out,
938
				loff_t *ppos, size_t len, unsigned int flags)
939
{
940
	return splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_sendpage);
941 942
}

943
EXPORT_SYMBOL(generic_splice_sendpage);
J
Jeff Garzik 已提交
944

945 946 947
/*
 * Attempt to initiate a splice from pipe to file.
 */
948
static long do_splice_from(struct pipe_inode_info *pipe, struct file *out,
949
			   loff_t *ppos, size_t len, unsigned int flags)
950 951 952
{
	int ret;

953
	if (unlikely(!out->f_op || !out->f_op->splice_write))
954 955
		return -EINVAL;

956
	if (unlikely(!(out->f_mode & FMODE_WRITE)))
957 958
		return -EBADF;

959
	ret = rw_verify_area(WRITE, out, ppos, len);
960 961 962
	if (unlikely(ret < 0))
		return ret;

963
	return out->f_op->splice_write(pipe, out, ppos, len, flags);
964 965
}

966 967 968
/*
 * Attempt to initiate a splice from a file to a pipe.
 */
969 970 971
static long do_splice_to(struct file *in, loff_t *ppos,
			 struct pipe_inode_info *pipe, size_t len,
			 unsigned int flags)
972
{
973
	loff_t isize, left;
974 975
	int ret;

976
	if (unlikely(!in->f_op || !in->f_op->splice_read))
977 978
		return -EINVAL;

979
	if (unlikely(!(in->f_mode & FMODE_READ)))
980 981
		return -EBADF;

982
	ret = rw_verify_area(READ, in, ppos, len);
983 984 985 986
	if (unlikely(ret < 0))
		return ret;

	isize = i_size_read(in->f_mapping->host);
987
	if (unlikely(*ppos >= isize))
988 989
		return 0;
	
990
	left = isize - *ppos;
991
	if (unlikely(left < len))
992 993
		len = left;

994
	return in->f_op->splice_read(in, ppos, pipe, len, flags);
995 996
}

997 998
long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
		      size_t len, unsigned int flags)
999 1000 1001
{
	struct pipe_inode_info *pipe;
	long ret, bytes;
1002
	loff_t out_off;
1003 1004 1005 1006 1007 1008 1009 1010
	umode_t i_mode;
	int i;

	/*
	 * 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!
	 */
1011
	i_mode = in->f_path.dentry->d_inode->i_mode;
1012 1013 1014 1015 1016 1017 1018 1019
	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;
1020
	if (unlikely(!pipe)) {
1021 1022 1023 1024 1025 1026
		pipe = alloc_pipe_info(NULL);
		if (!pipe)
			return -ENOMEM;

		/*
		 * We don't have an immediate reader, but we'll read the stuff
1027
		 * out of the pipe right after the splice_to_pipe(). So set
1028 1029 1030 1031 1032 1033 1034 1035
		 * PIPE_READERS appropriately.
		 */
		pipe->readers = 1;

		current->splice_pipe = pipe;
	}

	/*
I
Ingo Molnar 已提交
1036
	 * Do the splice.
1037 1038 1039
	 */
	ret = 0;
	bytes = 0;
1040
	out_off = 0;
1041 1042 1043 1044 1045 1046 1047 1048 1049

	while (len) {
		size_t read_len, max_read_len;

		/*
		 * Do at most PIPE_BUFFERS pages worth of transfer:
		 */
		max_read_len = min(len, (size_t)(PIPE_BUFFERS*PAGE_SIZE));

1050
		ret = do_splice_to(in, ppos, pipe, max_read_len, flags);
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
		if (unlikely(ret < 0))
			goto out_release;

		read_len = ret;

		/*
		 * 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:
		 */
1061
		ret = do_splice_from(pipe, out, &out_off, read_len,
1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110
				     flags & ~SPLICE_F_NONBLOCK);
		if (unlikely(ret < 0))
			goto out_release;

		bytes += ret;
		len -= ret;

		/*
		 * In nonblocking mode, if we got back a short read then
		 * that was due to either an IO error or due to the
		 * pagecache entry not being there. In the IO error case
		 * the _next_ splice attempt will produce a clean IO error
		 * return value (not a short read), so in both cases it's
		 * correct to break out of the loop here:
		 */
		if ((flags & SPLICE_F_NONBLOCK) && (read_len < max_read_len))
			break;
	}

	pipe->nrbufs = pipe->curbuf = 0;

	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;
		}
	}
	pipe->nrbufs = pipe->curbuf = 0;

	/*
	 * If we transferred some data, return the number of bytes:
	 */
	if (bytes > 0)
		return bytes;

	return ret;
}

EXPORT_SYMBOL(do_splice_direct);

1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
/*
 * 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;
}

1124 1125 1126
/*
 * Determine where to splice to/from.
 */
1127 1128 1129
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)
1130
{
1131
	struct pipe_inode_info *pipe;
1132
	loff_t offset, *off;
J
Jens Axboe 已提交
1133
	long ret;
1134

1135
	pipe = pipe_info(in->f_path.dentry->d_inode);
1136 1137 1138
	if (pipe) {
		if (off_in)
			return -ESPIPE;
1139 1140 1141
		if (off_out) {
			if (out->f_op->llseek == no_llseek)
				return -EINVAL;
1142
			if (copy_from_user(&offset, off_out, sizeof(loff_t)))
1143
				return -EFAULT;
1144 1145 1146
			off = &offset;
		} else
			off = &out->f_pos;
1147

J
Jens Axboe 已提交
1148 1149 1150 1151 1152 1153
		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;
1154
	}
1155

1156
	pipe = pipe_info(out->f_path.dentry->d_inode);
1157 1158 1159
	if (pipe) {
		if (off_out)
			return -ESPIPE;
1160 1161 1162
		if (off_in) {
			if (in->f_op->llseek == no_llseek)
				return -EINVAL;
1163
			if (copy_from_user(&offset, off_in, sizeof(loff_t)))
1164
				return -EFAULT;
1165 1166 1167
			off = &offset;
		} else
			off = &in->f_pos;
1168

J
Jens Axboe 已提交
1169 1170 1171 1172 1173 1174
		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;
1175
	}
1176 1177 1178 1179

	return -EINVAL;
}

1180 1181 1182 1183 1184 1185 1186 1187 1188
/*
 * 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,
1189
				struct partial_page *partial, int aligned)
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
{
	int buffers = 0, error = 0;

	/*
	 * It's ok to take the mmap_sem for reading, even
	 * across a "get_user()".
	 */
	down_read(&current->mm->mmap_sem);

	while (nr_vecs) {
		unsigned long off, npages;
		void __user *base;
		size_t len;
		int i;

		/*
		 * 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;
		error = -EFAULT;
		if (unlikely(!base))
			break;

		/*
		 * Get this base offset and number of pages, then map
		 * in the user pages.
		 */
		off = (unsigned long) base & ~PAGE_MASK;
1229 1230 1231 1232 1233 1234 1235 1236 1237

		/*
		 * 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;

1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252
		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++) {
1253
			const int plen = min_t(size_t, len, PAGE_SIZE - off);
1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313

			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;
}

/*
 * 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.
 *
 * Note that vmsplice only supports 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).
 *
 * Alas, it isn't here.
 *
 */
static long do_vmsplice(struct file *file, const struct iovec __user *iov,
			unsigned long nr_segs, unsigned int flags)
{
1314
	struct pipe_inode_info *pipe;
1315 1316 1317 1318 1319 1320 1321 1322 1323
	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,
	};

1324
	pipe = pipe_info(file->f_path.dentry->d_inode);
1325
	if (!pipe)
1326 1327 1328 1329 1330 1331
		return -EBADF;
	if (unlikely(nr_segs > UIO_MAXIOV))
		return -EINVAL;
	else if (unlikely(!nr_segs))
		return 0;

1332 1333
	spd.nr_pages = get_iovec_page_array(iov, nr_segs, pages, partial,
					    flags & SPLICE_F_GIFT);
1334 1335 1336
	if (spd.nr_pages <= 0)
		return spd.nr_pages;

1337
	return splice_to_pipe(pipe, &spd);
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
}

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;

	error = -EBADF;
	file = fget_light(fd, &fput);
	if (file) {
		if (file->f_mode & FMODE_WRITE)
			error = do_vmsplice(file, iov, nr_segs, flags);

		fput_light(file, fput);
	}

	return error;
}

1359 1360 1361
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)
1362 1363 1364 1365 1366 1367 1368 1369 1370
{
	long error;
	struct file *in, *out;
	int fput_in, fput_out;

	if (unlikely(!len))
		return 0;

	error = -EBADF;
1371
	in = fget_light(fd_in, &fput_in);
1372 1373
	if (in) {
		if (in->f_mode & FMODE_READ) {
1374
			out = fget_light(fd_out, &fput_out);
1375 1376
			if (out) {
				if (out->f_mode & FMODE_WRITE)
1377 1378 1379
					error = do_splice(in, off_in,
							  out, off_out,
							  len, flags);
1380 1381 1382 1383 1384 1385 1386 1387 1388
				fput_light(out, fput_out);
			}
		}

		fput_light(in, fput_in);
	}

	return error;
}
1389

1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
/*
 * 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;
}

1469 1470 1471 1472 1473 1474 1475 1476
/*
 * 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;
1477
	int ret = 0, i = 0, nbuf;
1478 1479 1480 1481 1482 1483

	/*
	 * 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).
	 */
1484
	inode_double_lock(ipipe->inode, opipe->inode);
1485

1486
	do {
1487 1488 1489 1490 1491 1492 1493
		if (!opipe->readers) {
			send_sig(SIGPIPE, current, 0);
			if (!ret)
				ret = -EPIPE;
			break;
		}

1494 1495 1496 1497 1498 1499
		/*
		 * If we have iterated all input buffers or ran out of
		 * output room, break.
		 */
		if (i >= ipipe->nrbufs || opipe->nrbufs >= PIPE_BUFFERS)
			break;
1500

1501 1502
		ibuf = ipipe->bufs + ((ipipe->curbuf + i) & (PIPE_BUFFERS - 1));
		nbuf = (opipe->curbuf + opipe->nrbufs) & (PIPE_BUFFERS - 1);
1503 1504

		/*
1505 1506
		 * Get a reference to this pipe buffer,
		 * so we can copy the contents over.
1507
		 */
1508 1509 1510 1511 1512
		ibuf->ops->get(ipipe, ibuf);

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

1513
		/*
1514 1515
		 * Don't inherit the gift flag, we need to
		 * prevent multiple steals of this page.
1516
		 */
1517
		obuf->flags &= ~PIPE_BUF_FLAG_GIFT;
1518

1519 1520
		if (obuf->len > len)
			obuf->len = len;
1521

1522 1523 1524 1525 1526
		opipe->nrbufs++;
		ret += obuf->len;
		len -= obuf->len;
		i++;
	} while (len);
1527

1528
	inode_double_unlock(ipipe->inode, opipe->inode);
1529

1530 1531 1532 1533
	/*
	 * If we put data in the output pipe, wakeup any potential readers.
	 */
	if (ret > 0) {
1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
		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)
{
1552 1553
	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);
1554
	int ret = -EINVAL;
1555 1556

	/*
1557 1558
	 * Duplicate the contents of ipipe to opipe without actually
	 * copying the data.
1559
	 */
1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
	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);
			if (!ret) {
				ret = link_pipe(ipipe, opipe, len, flags);
				if (!ret && (flags & SPLICE_F_NONBLOCK))
					ret = -EAGAIN;
			}
		}
	}
1575

1576
	return ret;
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
}

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;
}