relay.c 32.1 KB
Newer Older
1 2 3
/*
 * Public API and common code for kernel->userspace relay file support.
 *
4
 * See Documentation/filesystems/relay.txt for an overview.
5 6 7 8 9
 *
 * Copyright (C) 2002-2005 - Tom Zanussi (zanussi@us.ibm.com), IBM Corp
 * Copyright (C) 1999-2005 - Karim Yaghmour (karim@opersys.com)
 *
 * Moved to kernel/relay.c by Paul Mundt, 2006.
10 11
 * November 2006 - CPU hotplug support by Mathieu Desnoyers
 * 	(mathieu.desnoyers@polymtl.ca)
12 13 14 15 16 17
 *
 * This file is released under the GPL.
 */
#include <linux/errno.h>
#include <linux/stddef.h>
#include <linux/slab.h>
18
#include <linux/export.h>
19 20 21 22
#include <linux/string.h>
#include <linux/relay.h>
#include <linux/vmalloc.h>
#include <linux/mm.h>
23
#include <linux/cpu.h>
24
#include <linux/splice.h>
25 26 27 28

/* list of open channels, for cpu hotplug */
static DEFINE_MUTEX(relay_channels_mutex);
static LIST_HEAD(relay_channels);
29 30 31 32 33 34 35 36 37 38 39

/*
 * close() vm_op implementation for relay file mapping.
 */
static void relay_file_mmap_close(struct vm_area_struct *vma)
{
	struct rchan_buf *buf = vma->vm_private_data;
	buf->chan->cb->buf_unmapped(buf, vma->vm_file);
}

/*
N
Nick Piggin 已提交
40
 * fault() vm_op implementation for relay file mapping.
41
 */
42
static int relay_buf_fault(struct vm_fault *vmf)
43 44
{
	struct page *page;
45
	struct rchan_buf *buf = vmf->vma->vm_private_data;
N
Nick Piggin 已提交
46
	pgoff_t pgoff = vmf->pgoff;
47 48

	if (!buf)
N
Nick Piggin 已提交
49
		return VM_FAULT_OOM;
50

N
Nick Piggin 已提交
51
	page = vmalloc_to_page(buf->start + (pgoff << PAGE_SHIFT));
52
	if (!page)
N
Nick Piggin 已提交
53
		return VM_FAULT_SIGBUS;
54
	get_page(page);
N
Nick Piggin 已提交
55
	vmf->page = page;
56

N
Nick Piggin 已提交
57
	return 0;
58 59 60 61 62
}

/*
 * vm_ops for relay file mappings.
 */
63
static const struct vm_operations_struct relay_file_mmap_ops = {
N
Nick Piggin 已提交
64
	.fault = relay_buf_fault,
65 66 67
	.close = relay_file_mmap_close,
};

68 69 70 71 72
/*
 * allocate an array of pointers of struct page
 */
static struct page **relay_alloc_page_array(unsigned int n_pages)
{
73 74 75 76
	const size_t pa_size = n_pages * sizeof(struct page *);
	if (pa_size > PAGE_SIZE)
		return vzalloc(pa_size);
	return kzalloc(pa_size, GFP_KERNEL);
77 78 79 80 81 82 83
}

/*
 * free an array of pointers of struct page
 */
static void relay_free_page_array(struct page **array)
{
84
	kvfree(array);
85 86
}

87 88 89 90 91 92 93 94 95
/**
 *	relay_mmap_buf: - mmap channel buffer to process address space
 *	@buf: relay channel buffer
 *	@vma: vm_area_struct describing memory to be mapped
 *
 *	Returns 0 if ok, negative on error
 *
 *	Caller should already have grabbed mmap_sem.
 */
96
static int relay_mmap_buf(struct rchan_buf *buf, struct vm_area_struct *vma)
97 98 99 100 101 102 103 104 105 106 107
{
	unsigned long length = vma->vm_end - vma->vm_start;
	struct file *filp = vma->vm_file;

	if (!buf)
		return -EBADF;

	if (length != (unsigned long)buf->chan->alloc_size)
		return -EINVAL;

	vma->vm_ops = &relay_file_mmap_ops;
108
	vma->vm_flags |= VM_DONTEXPAND;
109 110 111 112 113 114 115 116 117 118 119
	vma->vm_private_data = buf;
	buf->chan->cb->buf_mapped(buf, filp);

	return 0;
}

/**
 *	relay_alloc_buf - allocate a channel buffer
 *	@buf: the buffer struct
 *	@size: total size of the buffer
 *
120
 *	Returns a pointer to the resulting buffer, %NULL if unsuccessful. The
121
 *	passed in size will get page aligned, if it isn't already.
122
 */
123
static void *relay_alloc_buf(struct rchan_buf *buf, size_t *size)
124 125 126 127
{
	void *mem;
	unsigned int i, j, n_pages;

128 129
	*size = PAGE_ALIGN(*size);
	n_pages = *size >> PAGE_SHIFT;
130

131
	buf->page_array = relay_alloc_page_array(n_pages);
132 133 134 135 136 137 138
	if (!buf->page_array)
		return NULL;

	for (i = 0; i < n_pages; i++) {
		buf->page_array[i] = alloc_page(GFP_KERNEL);
		if (unlikely(!buf->page_array[i]))
			goto depopulate;
T
Tom Zanussi 已提交
139
		set_page_private(buf->page_array[i], (unsigned long)buf);
140 141 142 143 144
	}
	mem = vmap(buf->page_array, n_pages, VM_MAP, PAGE_KERNEL);
	if (!mem)
		goto depopulate;

145
	memset(mem, 0, *size);
146 147 148 149 150 151
	buf->page_count = n_pages;
	return mem;

depopulate:
	for (j = 0; j < i; j++)
		__free_page(buf->page_array[j]);
152
	relay_free_page_array(buf->page_array);
153 154 155 156 157
	return NULL;
}

/**
 *	relay_create_buf - allocate and initialize a channel buffer
158
 *	@chan: the relay channel
159
 *
160
 *	Returns channel buffer if successful, %NULL otherwise.
161
 */
162
static struct rchan_buf *relay_create_buf(struct rchan *chan)
163
{
164 165 166
	struct rchan_buf *buf;

	if (chan->n_subbufs > UINT_MAX / sizeof(size_t *))
167 168
		return NULL;

169 170 171
	buf = kzalloc(sizeof(struct rchan_buf), GFP_KERNEL);
	if (!buf)
		return NULL;
172 173 174 175
	buf->padding = kmalloc(chan->n_subbufs * sizeof(size_t *), GFP_KERNEL);
	if (!buf->padding)
		goto free_buf;

176
	buf->start = relay_alloc_buf(buf, &chan->alloc_size);
177 178 179 180 181 182 183 184 185 186 187 188 189 190 191
	if (!buf->start)
		goto free_buf;

	buf->chan = chan;
	kref_get(&buf->chan->kref);
	return buf;

free_buf:
	kfree(buf->padding);
	kfree(buf);
	return NULL;
}

/**
 *	relay_destroy_channel - free the channel struct
192
 *	@kref: target kernel reference that contains the relay channel
193 194 195
 *
 *	Should only be called from kref_put().
 */
196
static void relay_destroy_channel(struct kref *kref)
197 198 199 200 201 202 203 204 205
{
	struct rchan *chan = container_of(kref, struct rchan, kref);
	kfree(chan);
}

/**
 *	relay_destroy_buf - destroy an rchan_buf struct and associated buffer
 *	@buf: the buffer struct
 */
206
static void relay_destroy_buf(struct rchan_buf *buf)
207 208 209 210 211 212 213 214
{
	struct rchan *chan = buf->chan;
	unsigned int i;

	if (likely(buf->start)) {
		vunmap(buf->start);
		for (i = 0; i < buf->page_count; i++)
			__free_page(buf->page_array[i]);
215
		relay_free_page_array(buf->page_array);
216
	}
217
	*per_cpu_ptr(chan->buf, buf->cpu) = NULL;
218 219 220 221 222 223 224
	kfree(buf->padding);
	kfree(buf);
	kref_put(&chan->kref, relay_destroy_channel);
}

/**
 *	relay_remove_buf - remove a channel buffer
225
 *	@kref: target kernel reference that contains the relay buffer
226
 *
227
 *	Removes the file from the filesystem, which also frees the
228 229 230
 *	rchan_buf_struct and the channel buffer.  Should only be called from
 *	kref_put().
 */
231
static void relay_remove_buf(struct kref *kref)
232 233 234 235 236 237 238 239 240 241 242
{
	struct rchan_buf *buf = container_of(kref, struct rchan_buf, kref);
	relay_destroy_buf(buf);
}

/**
 *	relay_buf_empty - boolean, is the channel buffer empty?
 *	@buf: channel buffer
 *
 *	Returns 1 if the buffer is empty, 0 otherwise.
 */
243
static int relay_buf_empty(struct rchan_buf *buf)
244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304
{
	return (buf->subbufs_produced - buf->subbufs_consumed) ? 0 : 1;
}

/**
 *	relay_buf_full - boolean, is the channel buffer full?
 *	@buf: channel buffer
 *
 *	Returns 1 if the buffer is full, 0 otherwise.
 */
int relay_buf_full(struct rchan_buf *buf)
{
	size_t ready = buf->subbufs_produced - buf->subbufs_consumed;
	return (ready >= buf->chan->n_subbufs) ? 1 : 0;
}
EXPORT_SYMBOL_GPL(relay_buf_full);

/*
 * High-level relay kernel API and associated functions.
 */

/*
 * rchan_callback implementations defining default channel behavior.  Used
 * in place of corresponding NULL values in client callback struct.
 */

/*
 * subbuf_start() default callback.  Does nothing.
 */
static int subbuf_start_default_callback (struct rchan_buf *buf,
					  void *subbuf,
					  void *prev_subbuf,
					  size_t prev_padding)
{
	if (relay_buf_full(buf))
		return 0;

	return 1;
}

/*
 * buf_mapped() default callback.  Does nothing.
 */
static void buf_mapped_default_callback(struct rchan_buf *buf,
					struct file *filp)
{
}

/*
 * buf_unmapped() default callback.  Does nothing.
 */
static void buf_unmapped_default_callback(struct rchan_buf *buf,
					  struct file *filp)
{
}

/*
 * create_buf_file_create() default callback.  Does nothing.
 */
static struct dentry *create_buf_file_default_callback(const char *filename,
						       struct dentry *parent,
A
Al Viro 已提交
305
						       umode_t mode,
306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
						       struct rchan_buf *buf,
						       int *is_global)
{
	return NULL;
}

/*
 * remove_buf_file() default callback.  Does nothing.
 */
static int remove_buf_file_default_callback(struct dentry *dentry)
{
	return -EINVAL;
}

/* relay channel default callbacks */
static struct rchan_callbacks default_channel_callbacks = {
	.subbuf_start = subbuf_start_default_callback,
	.buf_mapped = buf_mapped_default_callback,
	.buf_unmapped = buf_unmapped_default_callback,
	.create_buf_file = create_buf_file_default_callback,
	.remove_buf_file = remove_buf_file_default_callback,
};

/**
 *	wakeup_readers - wake up readers waiting on a channel
331
 *	@work: contains the channel buffer
332
 *
333
 *	This is the function used to defer reader waking
334
 */
335
static void wakeup_readers(struct irq_work *work)
336
{
337 338 339
	struct rchan_buf *buf;

	buf = container_of(work, struct rchan_buf, wakeup_work);
340 341 342 343 344 345 346 347
	wake_up_interruptible(&buf->read_wait);
}

/**
 *	__relay_reset - reset a channel buffer
 *	@buf: the channel buffer
 *	@init: 1 if this is a first-time initialization
 *
348
 *	See relay_reset() for description of effect.
349
 */
A
Andrew Morton 已提交
350
static void __relay_reset(struct rchan_buf *buf, unsigned int init)
351 352 353 354 355 356
{
	size_t i;

	if (init) {
		init_waitqueue_head(&buf->read_wait);
		kref_init(&buf->kref);
357 358 359 360
		init_irq_work(&buf->wakeup_work, wakeup_readers);
	} else {
		irq_work_sync(&buf->wakeup_work);
	}
361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382

	buf->subbufs_produced = 0;
	buf->subbufs_consumed = 0;
	buf->bytes_consumed = 0;
	buf->finalized = 0;
	buf->data = buf->start;
	buf->offset = 0;

	for (i = 0; i < buf->chan->n_subbufs; i++)
		buf->padding[i] = 0;

	buf->chan->cb->subbuf_start(buf, buf->data, NULL, 0);
}

/**
 *	relay_reset - reset the channel
 *	@chan: the channel
 *
 *	This has the effect of erasing all data from all channel buffers
 *	and restarting the channel in its initial state.  The buffers
 *	are not freed, so any mappings are still in effect.
 *
383
 *	NOTE. Care should be taken that the channel isn't actually
384 385 386 387
 *	being used by anything when this call is made.
 */
void relay_reset(struct rchan *chan)
{
388
	struct rchan_buf *buf;
389 390 391 392 393
	unsigned int i;

	if (!chan)
		return;

394 395
	if (chan->is_global && (buf = *per_cpu_ptr(chan->buf, 0))) {
		__relay_reset(buf, 0);
396
		return;
397
	}
398 399

	mutex_lock(&relay_channels_mutex);
L
Lai Jiangshan 已提交
400
	for_each_possible_cpu(i)
401 402
		if ((buf = *per_cpu_ptr(chan->buf, i)))
			__relay_reset(buf, 0);
403
	mutex_unlock(&relay_channels_mutex);
404 405 406
}
EXPORT_SYMBOL_GPL(relay_reset);

407 408 409 410
static inline void relay_set_buf_dentry(struct rchan_buf *buf,
					struct dentry *dentry)
{
	buf->dentry = dentry;
411
	d_inode(buf->dentry)->i_size = buf->early_bytes;
412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
}

static struct dentry *relay_create_buf_file(struct rchan *chan,
					    struct rchan_buf *buf,
					    unsigned int cpu)
{
	struct dentry *dentry;
	char *tmpname;

	tmpname = kzalloc(NAME_MAX + 1, GFP_KERNEL);
	if (!tmpname)
		return NULL;
	snprintf(tmpname, NAME_MAX, "%s%d", chan->base_filename, cpu);

	/* Create file in fs */
	dentry = chan->cb->create_buf_file(tmpname, chan->parent,
					   S_IRUSR, buf,
					   &chan->is_global);

	kfree(tmpname);

	return dentry;
}

436
/*
437 438
 *	relay_open_buf - create a new relay channel buffer
 *
439
 *	used by relay_open() and CPU hotplug.
440
 */
441
static struct rchan_buf *relay_open_buf(struct rchan *chan, unsigned int cpu)
442
{
443
 	struct rchan_buf *buf = NULL;
444 445
	struct dentry *dentry;

446
 	if (chan->is_global)
447
		return *per_cpu_ptr(chan->buf, 0);
448 449 450

	buf = relay_create_buf(chan);
	if (!buf)
451 452 453 454 455 456 457
		return NULL;

	if (chan->has_base_filename) {
		dentry = relay_create_buf_file(chan, buf, cpu);
		if (!dentry)
			goto free_buf;
		relay_set_buf_dentry(buf, dentry);
458 459 460 461 462 463 464
	} else {
		/* Only retrieve global info, nothing more, nothing less */
		dentry = chan->cb->create_buf_file(NULL, NULL,
						   S_IRUSR, buf,
						   &chan->is_global);
		if (WARN_ON(dentry))
			goto free_buf;
465
	}
466 467 468

 	buf->cpu = cpu;
 	__relay_reset(buf, 1);
469

470
 	if(chan->is_global) {
471
		*per_cpu_ptr(chan->buf, 0) = buf;
472 473 474
 		buf->cpu = 0;
  	}

475
	return buf;
476 477 478

free_buf:
 	relay_destroy_buf(buf);
479
	return NULL;
480 481 482 483 484 485 486 487 488 489
}

/**
 *	relay_close_buf - close a channel buffer
 *	@buf: channel buffer
 *
 *	Marks the buffer finalized and restores the default callbacks.
 *	The channel buffer and channel buffer data structure are then freed
 *	automatically when the last reference is given up.
 */
A
Andrew Morton 已提交
490
static void relay_close_buf(struct rchan_buf *buf)
491 492
{
	buf->finalized = 1;
493
	irq_work_sync(&buf->wakeup_work);
494
	buf->chan->cb->remove_buf_file(buf->dentry);
495 496 497
	kref_put(&buf->kref, relay_remove_buf);
}

A
Andrew Morton 已提交
498
static void setup_callbacks(struct rchan *chan,
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518
				   struct rchan_callbacks *cb)
{
	if (!cb) {
		chan->cb = &default_channel_callbacks;
		return;
	}

	if (!cb->subbuf_start)
		cb->subbuf_start = subbuf_start_default_callback;
	if (!cb->buf_mapped)
		cb->buf_mapped = buf_mapped_default_callback;
	if (!cb->buf_unmapped)
		cb->buf_unmapped = buf_unmapped_default_callback;
	if (!cb->create_buf_file)
		cb->create_buf_file = create_buf_file_default_callback;
	if (!cb->remove_buf_file)
		cb->remove_buf_file = remove_buf_file_default_callback;
	chan->cb = cb;
}

519
int relay_prepare_cpu(unsigned int cpu)
520 521
{
	struct rchan *chan;
522
	struct rchan_buf *buf;
523

524 525 526 527 528 529 530 531 532
	mutex_lock(&relay_channels_mutex);
	list_for_each_entry(chan, &relay_channels, list) {
		if ((buf = *per_cpu_ptr(chan->buf, cpu)))
			continue;
		buf = relay_open_buf(chan, cpu);
		if (!buf) {
			pr_err("relay: cpu %d buffer creation failed\n", cpu);
			mutex_unlock(&relay_channels_mutex);
			return -ENOMEM;
533
		}
534
		*per_cpu_ptr(chan->buf, cpu) = buf;
535
	}
536 537
	mutex_unlock(&relay_channels_mutex);
	return 0;
538 539
}

540 541
/**
 *	relay_open - create a new relay channel
542 543
 *	@base_filename: base name of files to create, %NULL for buffering only
 *	@parent: dentry of parent directory, %NULL for root directory or buffer
544 545 546
 *	@subbuf_size: size of sub-buffers
 *	@n_subbufs: number of sub-buffers
 *	@cb: client callback functions
547
 *	@private_data: user-defined data
548
 *
549
 *	Returns channel pointer if successful, %NULL otherwise.
550 551 552 553
 *
 *	Creates a channel buffer for each cpu using the sizes and
 *	attributes specified.  The created channel buffer files
 *	will be named base_filename0...base_filenameN-1.  File
554
 *	permissions will be %S_IRUSR.
555 556 557 558
 *
 *	If opening a buffer (@parent = NULL) that you later wish to register
 *	in a filesystem, call relay_late_setup_files() once the @parent dentry
 *	is available.
559 560 561 562 563
 */
struct rchan *relay_open(const char *base_filename,
			 struct dentry *parent,
			 size_t subbuf_size,
			 size_t n_subbufs,
564 565
			 struct rchan_callbacks *cb,
			 void *private_data)
566 567 568
{
	unsigned int i;
	struct rchan *chan;
569
	struct rchan_buf *buf;
570 571 572

	if (!(subbuf_size && n_subbufs))
		return NULL;
573 574
	if (subbuf_size > UINT_MAX / n_subbufs)
		return NULL;
575

576
	chan = kzalloc(sizeof(struct rchan), GFP_KERNEL);
577 578 579
	if (!chan)
		return NULL;

580
	chan->buf = alloc_percpu(struct rchan_buf *);
581 582 583
	chan->version = RELAYFS_CHANNEL_VERSION;
	chan->n_subbufs = n_subbufs;
	chan->subbuf_size = subbuf_size;
584
	chan->alloc_size = PAGE_ALIGN(subbuf_size * n_subbufs);
585 586
	chan->parent = parent;
	chan->private_data = private_data;
587 588 589 590
	if (base_filename) {
		chan->has_base_filename = 1;
		strlcpy(chan->base_filename, base_filename, NAME_MAX);
	}
591 592 593
	setup_callbacks(chan, cb);
	kref_init(&chan->kref);

594
	mutex_lock(&relay_channels_mutex);
595
	for_each_online_cpu(i) {
596 597
		buf = relay_open_buf(chan, i);
		if (!buf)
598
			goto free_bufs;
599
		*per_cpu_ptr(chan->buf, i) = buf;
600
	}
601 602
	list_add(&chan->list, &relay_channels);
	mutex_unlock(&relay_channels_mutex);
603 604 605 606

	return chan;

free_bufs:
L
Lai Jiangshan 已提交
607
	for_each_possible_cpu(i) {
608 609
		if ((buf = *per_cpu_ptr(chan->buf, i)))
			relay_close_buf(buf);
610 611 612
	}

	kref_put(&chan->kref, relay_destroy_channel);
613
	mutex_unlock(&relay_channels_mutex);
614
	kfree(chan);
615 616 617 618
	return NULL;
}
EXPORT_SYMBOL_GPL(relay_open);

619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639
struct rchan_percpu_buf_dispatcher {
	struct rchan_buf *buf;
	struct dentry *dentry;
};

/* Called in atomic context. */
static void __relay_set_buf_dentry(void *info)
{
	struct rchan_percpu_buf_dispatcher *p = info;

	relay_set_buf_dentry(p->buf, p->dentry);
}

/**
 *	relay_late_setup_files - triggers file creation
 *	@chan: channel to operate on
 *	@base_filename: base name of files to create
 *	@parent: dentry of parent directory, %NULL for root directory
 *
 *	Returns 0 if successful, non-zero otherwise.
 *
640 641 642 643 644 645
 *	Use to setup files for a previously buffer-only channel created
 *	by relay_open() with a NULL parent dentry.
 *
 *	For example, this is useful for perfomring early tracing in kernel,
 *	before VFS is up and then exposing the early results once the dentry
 *	is available.
646 647 648 649 650 651 652 653 654
 */
int relay_late_setup_files(struct rchan *chan,
			   const char *base_filename,
			   struct dentry *parent)
{
	int err = 0;
	unsigned int i, curr_cpu;
	unsigned long flags;
	struct dentry *dentry;
655
	struct rchan_buf *buf;
656 657 658 659 660 661 662 663 664
	struct rchan_percpu_buf_dispatcher disp;

	if (!chan || !base_filename)
		return -EINVAL;

	strlcpy(chan->base_filename, base_filename, NAME_MAX);

	mutex_lock(&relay_channels_mutex);
	/* Is chan already set up? */
665 666
	if (unlikely(chan->has_base_filename)) {
		mutex_unlock(&relay_channels_mutex);
667
		return -EEXIST;
668
	}
669 670
	chan->has_base_filename = 1;
	chan->parent = parent;
671 672 673

	if (chan->is_global) {
		err = -EINVAL;
674 675 676
		buf = *per_cpu_ptr(chan->buf, 0);
		if (!WARN_ON_ONCE(!buf)) {
			dentry = relay_create_buf_file(chan, buf, 0);
677
			if (dentry && !WARN_ON_ONCE(!chan->is_global)) {
678
				relay_set_buf_dentry(buf, dentry);
679 680 681 682 683 684 685
				err = 0;
			}
		}
		mutex_unlock(&relay_channels_mutex);
		return err;
	}

686 687 688 689 690 691 692
	curr_cpu = get_cpu();
	/*
	 * The CPU hotplug notifier ran before us and created buffers with
	 * no files associated. So it's safe to call relay_setup_buf_file()
	 * on all currently online CPUs.
	 */
	for_each_online_cpu(i) {
693 694
		buf = *per_cpu_ptr(chan->buf, i);
		if (unlikely(!buf)) {
695
			WARN_ONCE(1, KERN_ERR "CPU has no buffer!\n");
696 697 698 699
			err = -EINVAL;
			break;
		}

700
		dentry = relay_create_buf_file(chan, buf, i);
701 702 703 704 705 706 707
		if (unlikely(!dentry)) {
			err = -EINVAL;
			break;
		}

		if (curr_cpu == i) {
			local_irq_save(flags);
708
			relay_set_buf_dentry(buf, dentry);
709 710
			local_irq_restore(flags);
		} else {
711
			disp.buf = buf;
712 713 714 715 716 717 718 719 720 721 722 723 724 725 726
			disp.dentry = dentry;
			smp_mb();
			/* relay_channels_mutex must be held, so wait. */
			err = smp_call_function_single(i,
						       __relay_set_buf_dentry,
						       &disp, 1);
		}
		if (unlikely(err))
			break;
	}
	put_cpu();
	mutex_unlock(&relay_channels_mutex);

	return err;
}
727
EXPORT_SYMBOL_GPL(relay_late_setup_files);
728

729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751
/**
 *	relay_switch_subbuf - switch to a new sub-buffer
 *	@buf: channel buffer
 *	@length: size of current event
 *
 *	Returns either the length passed in or 0 if full.
 *
 *	Performs sub-buffer-switch tasks such as invoking callbacks,
 *	updating padding counts, waking up readers, etc.
 */
size_t relay_switch_subbuf(struct rchan_buf *buf, size_t length)
{
	void *old, *new;
	size_t old_subbuf, new_subbuf;

	if (unlikely(length > buf->chan->subbuf_size))
		goto toobig;

	if (buf->offset != buf->chan->subbuf_size + 1) {
		buf->prev_padding = buf->chan->subbuf_size - buf->offset;
		old_subbuf = buf->subbufs_produced % buf->chan->n_subbufs;
		buf->padding[old_subbuf] = buf->prev_padding;
		buf->subbufs_produced++;
752
		if (buf->dentry)
753
			d_inode(buf->dentry)->i_size +=
754 755 756 757 758
				buf->chan->subbuf_size -
				buf->padding[old_subbuf];
		else
			buf->early_bytes += buf->chan->subbuf_size -
					    buf->padding[old_subbuf];
759
		smp_mb();
760
		if (waitqueue_active(&buf->read_wait)) {
761 762 763 764 765 766
			/*
			 * Calling wake_up_interruptible() from here
			 * will deadlock if we happen to be logging
			 * from the scheduler (trying to re-grab
			 * rq->lock), so defer it.
			 */
767 768
			irq_work_queue(&buf->wakeup_work);
		}
769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802
	}

	old = buf->data;
	new_subbuf = buf->subbufs_produced % buf->chan->n_subbufs;
	new = buf->start + new_subbuf * buf->chan->subbuf_size;
	buf->offset = 0;
	if (!buf->chan->cb->subbuf_start(buf, new, old, buf->prev_padding)) {
		buf->offset = buf->chan->subbuf_size + 1;
		return 0;
	}
	buf->data = new;
	buf->padding[new_subbuf] = 0;

	if (unlikely(length + buf->offset > buf->chan->subbuf_size))
		goto toobig;

	return length;

toobig:
	buf->chan->last_toobig = length;
	return 0;
}
EXPORT_SYMBOL_GPL(relay_switch_subbuf);

/**
 *	relay_subbufs_consumed - update the buffer's sub-buffers-consumed count
 *	@chan: the channel
 *	@cpu: the cpu associated with the channel buffer to update
 *	@subbufs_consumed: number of sub-buffers to add to current buf's count
 *
 *	Adds to the channel buffer's consumed sub-buffer count.
 *	subbufs_consumed should be the number of sub-buffers newly consumed,
 *	not the total consumed.
 *
803
 *	NOTE. Kernel clients don't need to call this function if the channel
804 805 806 807 808 809 810 811
 *	mode is 'overwrite'.
 */
void relay_subbufs_consumed(struct rchan *chan,
			    unsigned int cpu,
			    size_t subbufs_consumed)
{
	struct rchan_buf *buf;

812
	if (!chan || cpu >= NR_CPUS)
813 814
		return;

815
	buf = *per_cpu_ptr(chan->buf, cpu);
816
	if (!buf || subbufs_consumed > chan->n_subbufs)
817 818
		return;

819
	if (subbufs_consumed > buf->subbufs_produced - buf->subbufs_consumed)
820
		buf->subbufs_consumed = buf->subbufs_produced;
821 822
	else
		buf->subbufs_consumed += subbufs_consumed;
823 824 825 826 827 828 829 830 831 832 833
}
EXPORT_SYMBOL_GPL(relay_subbufs_consumed);

/**
 *	relay_close - close the channel
 *	@chan: the channel
 *
 *	Closes all channel buffers and frees the channel.
 */
void relay_close(struct rchan *chan)
{
834
	struct rchan_buf *buf;
835 836 837 838 839
	unsigned int i;

	if (!chan)
		return;

840
	mutex_lock(&relay_channels_mutex);
841 842
	if (chan->is_global && (buf = *per_cpu_ptr(chan->buf, 0)))
		relay_close_buf(buf);
843 844
	else
		for_each_possible_cpu(i)
845 846
			if ((buf = *per_cpu_ptr(chan->buf, i)))
				relay_close_buf(buf);
847 848 849 850 851 852

	if (chan->last_toobig)
		printk(KERN_WARNING "relay: one or more items not logged "
		       "[item size (%Zd) > sub-buffer size (%Zd)]\n",
		       chan->last_toobig, chan->subbuf_size);

853
	list_del(&chan->list);
854
	kref_put(&chan->kref, relay_destroy_channel);
855
	mutex_unlock(&relay_channels_mutex);
856 857 858 859 860 861 862
}
EXPORT_SYMBOL_GPL(relay_close);

/**
 *	relay_flush - close the channel
 *	@chan: the channel
 *
863
 *	Flushes all channel buffers, i.e. forces buffer switch.
864 865 866
 */
void relay_flush(struct rchan *chan)
{
867
	struct rchan_buf *buf;
868 869 870 871 872
	unsigned int i;

	if (!chan)
		return;

873 874
	if (chan->is_global && (buf = *per_cpu_ptr(chan->buf, 0))) {
		relay_switch_subbuf(buf, 0);
875
		return;
876
	}
877 878 879

	mutex_lock(&relay_channels_mutex);
	for_each_possible_cpu(i)
880 881
		if ((buf = *per_cpu_ptr(chan->buf, i)))
			relay_switch_subbuf(buf, 0);
882
	mutex_unlock(&relay_channels_mutex);
883 884 885 886 887 888 889 890 891 892 893 894
}
EXPORT_SYMBOL_GPL(relay_flush);

/**
 *	relay_file_open - open file op for relay files
 *	@inode: the inode
 *	@filp: the file
 *
 *	Increments the channel buffer refcount.
 */
static int relay_file_open(struct inode *inode, struct file *filp)
{
895
	struct rchan_buf *buf = inode->i_private;
896 897 898
	kref_get(&buf->kref);
	filp->private_data = buf;

899
	return nonseekable_open(inode, filp);
900 901 902 903 904 905 906
}

/**
 *	relay_file_mmap - mmap file op for relay files
 *	@filp: the file
 *	@vma: the vma describing what to map
 *
907
 *	Calls upon relay_mmap_buf() to map the file into user space.
908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954
 */
static int relay_file_mmap(struct file *filp, struct vm_area_struct *vma)
{
	struct rchan_buf *buf = filp->private_data;
	return relay_mmap_buf(buf, vma);
}

/**
 *	relay_file_poll - poll file op for relay files
 *	@filp: the file
 *	@wait: poll table
 *
 *	Poll implemention.
 */
static unsigned int relay_file_poll(struct file *filp, poll_table *wait)
{
	unsigned int mask = 0;
	struct rchan_buf *buf = filp->private_data;

	if (buf->finalized)
		return POLLERR;

	if (filp->f_mode & FMODE_READ) {
		poll_wait(filp, &buf->read_wait, wait);
		if (!relay_buf_empty(buf))
			mask |= POLLIN | POLLRDNORM;
	}

	return mask;
}

/**
 *	relay_file_release - release file op for relay files
 *	@inode: the inode
 *	@filp: the file
 *
 *	Decrements the channel refcount, as the filesystem is
 *	no longer using it.
 */
static int relay_file_release(struct inode *inode, struct file *filp)
{
	struct rchan_buf *buf = filp->private_data;
	kref_put(&buf->kref, relay_remove_buf);

	return 0;
}

955
/*
956 957 958 959 960 961 962 963 964 965
 *	relay_file_read_consume - update the consumed count for the buffer
 */
static void relay_file_read_consume(struct rchan_buf *buf,
				    size_t read_pos,
				    size_t bytes_consumed)
{
	size_t subbuf_size = buf->chan->subbuf_size;
	size_t n_subbufs = buf->chan->n_subbufs;
	size_t read_subbuf;

966 967 968 969
	if (buf->subbufs_produced == buf->subbufs_consumed &&
	    buf->offset == buf->bytes_consumed)
		return;

970 971 972 973 974 975
	if (buf->bytes_consumed + bytes_consumed > subbuf_size) {
		relay_subbufs_consumed(buf->chan, buf->cpu, 1);
		buf->bytes_consumed = 0;
	}

	buf->bytes_consumed += bytes_consumed;
M
Masami Hiramatsu 已提交
976 977 978 979
	if (!read_pos)
		read_subbuf = buf->subbufs_consumed % n_subbufs;
	else
		read_subbuf = read_pos / buf->chan->subbuf_size;
980 981 982 983 984 985 986 987 988
	if (buf->bytes_consumed + buf->padding[read_subbuf] == subbuf_size) {
		if ((read_subbuf == buf->subbufs_produced % n_subbufs) &&
		    (buf->offset == subbuf_size))
			return;
		relay_subbufs_consumed(buf->chan, buf->cpu, 1);
		buf->bytes_consumed = 0;
	}
}

989
/*
990 991 992 993 994 995
 *	relay_file_read_avail - boolean, are there unconsumed bytes available?
 */
static int relay_file_read_avail(struct rchan_buf *buf, size_t read_pos)
{
	size_t subbuf_size = buf->chan->subbuf_size;
	size_t n_subbufs = buf->chan->n_subbufs;
996 997
	size_t produced = buf->subbufs_produced;
	size_t consumed = buf->subbufs_consumed;
998

999
	relay_file_read_consume(buf, read_pos, 0);
1000

1001 1002
	consumed = buf->subbufs_consumed;

1003 1004 1005 1006
	if (unlikely(buf->offset > subbuf_size)) {
		if (produced == consumed)
			return 0;
		return 1;
1007 1008
	}

1009
	if (unlikely(produced - consumed >= n_subbufs)) {
M
Masami Hiramatsu 已提交
1010
		consumed = produced - n_subbufs + 1;
1011
		buf->subbufs_consumed = consumed;
M
Masami Hiramatsu 已提交
1012
		buf->bytes_consumed = 0;
1013
	}
D
Daniel Walker 已提交
1014

1015 1016 1017 1018 1019
	produced = (produced % n_subbufs) * subbuf_size + buf->offset;
	consumed = (consumed % n_subbufs) * subbuf_size + buf->bytes_consumed;

	if (consumed > produced)
		produced += n_subbufs * subbuf_size;
D
Daniel Walker 已提交
1020

1021 1022 1023 1024
	if (consumed == produced) {
		if (buf->offset == subbuf_size &&
		    buf->subbufs_produced > buf->subbufs_consumed)
			return 1;
1025
		return 0;
1026
	}
1027 1028 1029 1030 1031 1032

	return 1;
}

/**
 *	relay_file_read_subbuf_avail - return bytes available in sub-buffer
1033 1034
 *	@read_pos: file read position
 *	@buf: relay channel buffer
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059
 */
static size_t relay_file_read_subbuf_avail(size_t read_pos,
					   struct rchan_buf *buf)
{
	size_t padding, avail = 0;
	size_t read_subbuf, read_offset, write_subbuf, write_offset;
	size_t subbuf_size = buf->chan->subbuf_size;

	write_subbuf = (buf->data - buf->start) / subbuf_size;
	write_offset = buf->offset > subbuf_size ? subbuf_size : buf->offset;
	read_subbuf = read_pos / subbuf_size;
	read_offset = read_pos % subbuf_size;
	padding = buf->padding[read_subbuf];

	if (read_subbuf == write_subbuf) {
		if (read_offset + padding < write_offset)
			avail = write_offset - (read_offset + padding);
	} else
		avail = (subbuf_size - padding) - read_offset;

	return avail;
}

/**
 *	relay_file_read_start_pos - find the first available byte to read
1060 1061
 *	@read_pos: file read position
 *	@buf: relay channel buffer
1062
 *
1063
 *	If the @read_pos is in the middle of padding, return the
1064 1065 1066 1067 1068 1069 1070 1071 1072
 *	position of the first actually available byte, otherwise
 *	return the original value.
 */
static size_t relay_file_read_start_pos(size_t read_pos,
					struct rchan_buf *buf)
{
	size_t read_subbuf, padding, padding_start, padding_end;
	size_t subbuf_size = buf->chan->subbuf_size;
	size_t n_subbufs = buf->chan->n_subbufs;
D
David Wilder 已提交
1073
	size_t consumed = buf->subbufs_consumed % n_subbufs;
1074

D
David Wilder 已提交
1075 1076
	if (!read_pos)
		read_pos = consumed * subbuf_size + buf->bytes_consumed;
1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
	read_subbuf = read_pos / subbuf_size;
	padding = buf->padding[read_subbuf];
	padding_start = (read_subbuf + 1) * subbuf_size - padding;
	padding_end = (read_subbuf + 1) * subbuf_size;
	if (read_pos >= padding_start && read_pos < padding_end) {
		read_subbuf = (read_subbuf + 1) % n_subbufs;
		read_pos = read_subbuf * subbuf_size;
	}

	return read_pos;
}

/**
 *	relay_file_read_end_pos - return the new read position
1091 1092 1093
 *	@read_pos: file read position
 *	@buf: relay channel buffer
 *	@count: number of bytes to be read
1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
 */
static size_t relay_file_read_end_pos(struct rchan_buf *buf,
				      size_t read_pos,
				      size_t count)
{
	size_t read_subbuf, padding, end_pos;
	size_t subbuf_size = buf->chan->subbuf_size;
	size_t n_subbufs = buf->chan->n_subbufs;

	read_subbuf = read_pos / subbuf_size;
	padding = buf->padding[read_subbuf];
	if (read_pos % subbuf_size + count + padding == subbuf_size)
		end_pos = (read_subbuf + 1) * subbuf_size;
	else
		end_pos = read_pos + count;
	if (end_pos >= subbuf_size * n_subbufs)
		end_pos = 0;

	return end_pos;
}

A
Al Viro 已提交
1115 1116 1117 1118
static ssize_t relay_file_read(struct file *filp,
			       char __user *buffer,
			       size_t count,
			       loff_t *ppos)
1119
{
1120 1121
	struct rchan_buf *buf = filp->private_data;
	size_t read_start, avail;
A
Al Viro 已提交
1122
	size_t written = 0;
1123
	int ret;
1124

A
Al Viro 已提交
1125
	if (!count)
1126 1127
		return 0;

A
Al Viro 已提交
1128
	inode_lock(file_inode(filp));
1129
	do {
A
Al Viro 已提交
1130 1131
		void *from;

1132 1133 1134 1135 1136 1137
		if (!relay_file_read_avail(buf, *ppos))
			break;

		read_start = relay_file_read_start_pos(*ppos, buf);
		avail = relay_file_read_subbuf_avail(read_start, buf);
		if (!avail)
1138 1139
			break;

A
Al Viro 已提交
1140 1141 1142 1143
		avail = min(count, avail);
		from = buf->start + read_start;
		ret = avail;
		if (copy_to_user(buffer, from, avail))
1144 1145
			break;

A
Al Viro 已提交
1146 1147 1148
		buffer += ret;
		written += ret;
		count -= ret;
1149

A
Al Viro 已提交
1150 1151 1152 1153
		relay_file_read_consume(buf, read_start, ret);
		*ppos = relay_file_read_end_pos(buf, read_start, ret);
	} while (count);
	inode_unlock(file_inode(filp));
1154

A
Al Viro 已提交
1155
	return written;
1156 1157
}

1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
static void relay_consume_bytes(struct rchan_buf *rbuf, int bytes_consumed)
{
	rbuf->bytes_consumed += bytes_consumed;

	if (rbuf->bytes_consumed >= rbuf->chan->subbuf_size) {
		relay_subbufs_consumed(rbuf->chan, rbuf->cpu, 1);
		rbuf->bytes_consumed %= rbuf->chan->subbuf_size;
	}
}

T
Tom Zanussi 已提交
1168 1169
static void relay_pipe_buf_release(struct pipe_inode_info *pipe,
				   struct pipe_buffer *buf)
1170
{
T
Tom Zanussi 已提交
1171 1172 1173
	struct rchan_buf *rbuf;

	rbuf = (struct rchan_buf *)page_private(buf->page);
1174
	relay_consume_bytes(rbuf, buf->private);
T
Tom Zanussi 已提交
1175 1176
}

1177
static const struct pipe_buf_operations relay_pipe_buf_ops = {
T
Tom Zanussi 已提交
1178
	.can_merge = 0,
1179
	.confirm = generic_pipe_buf_confirm,
T
Tom Zanussi 已提交
1180 1181 1182 1183 1184
	.release = relay_pipe_buf_release,
	.steal = generic_pipe_buf_steal,
	.get = generic_pipe_buf_get,
};

1185 1186 1187 1188
static void relay_page_release(struct splice_pipe_desc *spd, unsigned int i)
{
}

T
Tom Zanussi 已提交
1189
/*
T
Tom Zanussi 已提交
1190 1191
 *	subbuf_splice_actor - splice up to one subbuf's worth of data
 */
1192
static ssize_t subbuf_splice_actor(struct file *in,
T
Tom Zanussi 已提交
1193 1194 1195 1196 1197 1198
			       loff_t *ppos,
			       struct pipe_inode_info *pipe,
			       size_t len,
			       unsigned int flags,
			       int *nonpad_ret)
{
1199
	unsigned int pidx, poff, total_len, subbuf_pages, nr_pages;
T
Tom Zanussi 已提交
1200 1201
	struct rchan_buf *rbuf = in->private_data;
	unsigned int subbuf_size = rbuf->chan->subbuf_size;
1202 1203 1204
	uint64_t pos = (uint64_t) *ppos;
	uint32_t alloc_size = (uint32_t) rbuf->chan->alloc_size;
	size_t read_start = (size_t) do_div(pos, alloc_size);
T
Tom Zanussi 已提交
1205 1206 1207
	size_t read_subbuf = read_start / subbuf_size;
	size_t padding = rbuf->padding[read_subbuf];
	size_t nonpad_end = read_subbuf * subbuf_size + subbuf_size - padding;
1208 1209
	struct page *pages[PIPE_DEF_BUFFERS];
	struct partial_page partial[PIPE_DEF_BUFFERS];
1210 1211 1212
	struct splice_pipe_desc spd = {
		.pages = pages,
		.nr_pages = 0,
1213
		.nr_pages_max = PIPE_DEF_BUFFERS,
1214 1215 1216
		.partial = partial,
		.flags = flags,
		.ops = &relay_pipe_buf_ops,
1217
		.spd_release = relay_page_release,
1218
	};
1219
	ssize_t ret;
T
Tom Zanussi 已提交
1220 1221 1222

	if (rbuf->subbufs_produced == rbuf->subbufs_consumed)
		return 0;
1223 1224
	if (splice_grow_spd(pipe, &spd))
		return -ENOMEM;
T
Tom Zanussi 已提交
1225

1226 1227 1228 1229 1230
	/*
	 * Adjust read len, if longer than what is available
	 */
	if (len > (subbuf_size - read_start % subbuf_size))
		len = subbuf_size - read_start % subbuf_size;
T
Tom Zanussi 已提交
1231 1232 1233 1234

	subbuf_pages = rbuf->chan->alloc_size >> PAGE_SHIFT;
	pidx = (read_start / PAGE_SIZE) % subbuf_pages;
	poff = read_start & ~PAGE_MASK;
1235
	nr_pages = min_t(unsigned int, subbuf_pages, spd.nr_pages_max);
T
Tom Zanussi 已提交
1236

1237
	for (total_len = 0; spd.nr_pages < nr_pages; spd.nr_pages++) {
1238 1239
		unsigned int this_len, this_end, private;
		unsigned int cur_pos = read_start + total_len;
T
Tom Zanussi 已提交
1240

1241
		if (!len)
T
Tom Zanussi 已提交
1242 1243
			break;

1244 1245
		this_len = min_t(unsigned long, len, PAGE_SIZE - poff);
		private = this_len;
T
Tom Zanussi 已提交
1246

1247 1248
		spd.pages[spd.nr_pages] = rbuf->page_array[pidx];
		spd.partial[spd.nr_pages].offset = poff;
T
Tom Zanussi 已提交
1249

1250 1251 1252 1253
		this_end = cur_pos + this_len;
		if (this_end >= nonpad_end) {
			this_len = nonpad_end - cur_pos;
			private = this_len + padding;
T
Tom Zanussi 已提交
1254
		}
1255 1256
		spd.partial[spd.nr_pages].len = this_len;
		spd.partial[spd.nr_pages].private = private;
T
Tom Zanussi 已提交
1257

1258 1259 1260 1261
		len -= this_len;
		total_len += this_len;
		poff = 0;
		pidx = (pidx + 1) % subbuf_pages;
T
Tom Zanussi 已提交
1262

1263 1264
		if (this_end >= nonpad_end) {
			spd.nr_pages++;
T
Tom Zanussi 已提交
1265 1266 1267 1268
			break;
		}
	}

1269
	ret = 0;
1270
	if (!spd.nr_pages)
1271
		goto out;
T
Tom Zanussi 已提交
1272

1273 1274
	ret = *nonpad_ret = splice_to_pipe(pipe, &spd);
	if (ret < 0 || ret < total_len)
1275
		goto out;
T
Tom Zanussi 已提交
1276

1277 1278 1279
        if (read_start + ret == nonpad_end)
                ret += padding;

1280
out:
1281 1282
	splice_shrink_spd(&spd);
	return ret;
T
Tom Zanussi 已提交
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
}

static ssize_t relay_file_splice_read(struct file *in,
				      loff_t *ppos,
				      struct pipe_inode_info *pipe,
				      size_t len,
				      unsigned int flags)
{
	ssize_t spliced;
	int ret;
	int nonpad_ret = 0;

	ret = 0;
	spliced = 0;

1298
	while (len && !spliced) {
T
Tom Zanussi 已提交
1299 1300 1301 1302
		ret = subbuf_splice_actor(in, ppos, pipe, len, flags, &nonpad_ret);
		if (ret < 0)
			break;
		else if (!ret) {
1303
			if (flags & SPLICE_F_NONBLOCK)
T
Tom Zanussi 已提交
1304
				ret = -EAGAIN;
1305
			break;
T
Tom Zanussi 已提交
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320
		}

		*ppos += ret;
		if (ret > len)
			len = 0;
		else
			len -= ret;
		spliced += nonpad_ret;
		nonpad_ret = 0;
	}

	if (spliced)
		return spliced;

	return ret;
1321 1322
}

1323
const struct file_operations relay_file_operations = {
1324 1325 1326 1327 1328 1329
	.open		= relay_file_open,
	.poll		= relay_file_poll,
	.mmap		= relay_file_mmap,
	.read		= relay_file_read,
	.llseek		= no_llseek,
	.release	= relay_file_release,
T
Tom Zanussi 已提交
1330
	.splice_read	= relay_file_splice_read,
1331 1332
};
EXPORT_SYMBOL_GPL(relay_file_operations);