memory.c 19.7 KB
Newer Older
1
/*
2
 * Memory subsystem support
3 4 5 6 7 8 9 10 11 12 13 14 15
 *
 * Written by Matt Tolentino <matthew.e.tolentino@intel.com>
 *            Dave Hansen <haveblue@us.ibm.com>
 *
 * This file provides the necessary infrastructure to represent
 * a SPARSEMEM-memory-model system's physical memory in /sysfs.
 * All arch-independent code that assumes MEMORY_HOTPLUG requires
 * SPARSEMEM should be contained here, or in mm/memory_hotplug.c.
 */

#include <linux/module.h>
#include <linux/init.h>
#include <linux/topology.h>
16
#include <linux/capability.h>
17 18 19 20
#include <linux/device.h>
#include <linux/memory.h>
#include <linux/memory_hotplug.h>
#include <linux/mm.h>
21
#include <linux/mutex.h>
22
#include <linux/stat.h>
23
#include <linux/slab.h>
24

Arun Sharma's avatar
Arun Sharma committed
25
#include <linux/atomic.h>
26 27
#include <asm/uaccess.h>

28 29
static DEFINE_MUTEX(mem_sysfs_mutex);

30
#define MEMORY_CLASS_NAME	"memory"
31

32 33
#define to_memory_block(dev) container_of(dev, struct memory_block, dev)

34 35 36 37 38 39
static int sections_per_block;

static inline int base_memory_block_id(int section_nr)
{
	return section_nr / sections_per_block;
}
40

41 42 43
static int memory_subsys_online(struct device *dev);
static int memory_subsys_offline(struct device *dev);

44
static struct bus_type memory_subsys = {
45
	.name = MEMORY_CLASS_NAME,
46
	.dev_name = MEMORY_CLASS_NAME,
47 48
	.online = memory_subsys_online,
	.offline = memory_subsys_offline,
49 50
};

51
static BLOCKING_NOTIFIER_HEAD(memory_chain);
52

53
int register_memory_notifier(struct notifier_block *nb)
54
{
55
	return blocking_notifier_chain_register(&memory_chain, nb);
56
}
57
EXPORT_SYMBOL(register_memory_notifier);
58

59
void unregister_memory_notifier(struct notifier_block *nb)
60
{
61
	blocking_notifier_chain_unregister(&memory_chain, nb);
62
}
63
EXPORT_SYMBOL(unregister_memory_notifier);
64

65 66 67 68 69 70 71 72 73 74 75 76 77 78
static ATOMIC_NOTIFIER_HEAD(memory_isolate_chain);

int register_memory_isolate_notifier(struct notifier_block *nb)
{
	return atomic_notifier_chain_register(&memory_isolate_chain, nb);
}
EXPORT_SYMBOL(register_memory_isolate_notifier);

void unregister_memory_isolate_notifier(struct notifier_block *nb)
{
	atomic_notifier_chain_unregister(&memory_isolate_chain, nb);
}
EXPORT_SYMBOL(unregister_memory_isolate_notifier);

79 80
static void memory_block_release(struct device *dev)
{
81
	struct memory_block *mem = to_memory_block(dev);
82 83 84 85

	kfree(mem);
}

86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
unsigned long __weak memory_block_size_bytes(void)
{
	return MIN_MEMORY_BLOCK_SIZE;
}

static unsigned long get_memory_block_size(void)
{
	unsigned long block_sz;

	block_sz = memory_block_size_bytes();

	/* Validate blk_sz is a power of 2 and not less than section size */
	if ((block_sz & (block_sz - 1)) || (block_sz < MIN_MEMORY_BLOCK_SIZE)) {
		WARN_ON(1);
		block_sz = MIN_MEMORY_BLOCK_SIZE;
	}

	return block_sz;
}

106 107 108 109 110
/*
 * use this as the physical section index that this memsection
 * uses.
 */

111 112
static ssize_t show_mem_start_phys_index(struct device *dev,
			struct device_attribute *attr, char *buf)
113
{
114
	struct memory_block *mem = to_memory_block(dev);
115 116 117 118 119 120
	unsigned long phys_index;

	phys_index = mem->start_section_nr / sections_per_block;
	return sprintf(buf, "%08lx\n", phys_index);
}

121 122 123
/*
 * Show whether the section of memory is likely to be hot-removable
 */
124 125
static ssize_t show_mem_removable(struct device *dev,
			struct device_attribute *attr, char *buf)
126
{
127 128
	unsigned long i, pfn;
	int ret = 1;
129
	struct memory_block *mem = to_memory_block(dev);
130

131
	for (i = 0; i < sections_per_block; i++) {
132 133
		if (!present_section_nr(mem->start_section_nr + i))
			continue;
134
		pfn = section_nr_to_pfn(mem->start_section_nr + i);
135 136 137
		ret &= is_mem_section_removable(pfn, PAGES_PER_SECTION);
	}

138 139 140
	return sprintf(buf, "%d\n", ret);
}

141 142 143
/*
 * online, offline, going offline, etc.
 */
144 145
static ssize_t show_mem_state(struct device *dev,
			struct device_attribute *attr, char *buf)
146
{
147
	struct memory_block *mem = to_memory_block(dev);
148 149 150 151 152 153 154
	ssize_t len = 0;

	/*
	 * We can probably put these states in a nice little array
	 * so that they're not open-coded
	 */
	switch (mem->state) {
155 156 157 158 159 160 161 162 163 164 165 166 167 168
	case MEM_ONLINE:
		len = sprintf(buf, "online\n");
		break;
	case MEM_OFFLINE:
		len = sprintf(buf, "offline\n");
		break;
	case MEM_GOING_OFFLINE:
		len = sprintf(buf, "going-offline\n");
		break;
	default:
		len = sprintf(buf, "ERROR-UNKNOWN-%ld\n",
				mem->state);
		WARN_ON(1);
		break;
169 170 171 172 173
	}

	return len;
}

174
int memory_notify(unsigned long val, void *v)
175
{
176
	return blocking_notifier_call_chain(&memory_chain, val, v);
177 178
}

179 180 181 182 183
int memory_isolate_notify(unsigned long val, void *v)
{
	return atomic_notifier_call_chain(&memory_isolate_chain, val, v);
}

184 185 186 187
/*
 * The probe routines leave the pages reserved, just as the bootmem code does.
 * Make sure they're still that way.
 */
188
static bool pages_correctly_reserved(unsigned long start_pfn)
189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218
{
	int i, j;
	struct page *page;
	unsigned long pfn = start_pfn;

	/*
	 * memmap between sections is not contiguous except with
	 * SPARSEMEM_VMEMMAP. We lookup the page once per section
	 * and assume memmap is contiguous within each section
	 */
	for (i = 0; i < sections_per_block; i++, pfn += PAGES_PER_SECTION) {
		if (WARN_ON_ONCE(!pfn_valid(pfn)))
			return false;
		page = pfn_to_page(pfn);

		for (j = 0; j < PAGES_PER_SECTION; j++) {
			if (PageReserved(page + j))
				continue;

			printk(KERN_WARNING "section number %ld page number %d "
				"not reserved, was it already online?\n",
				pfn_to_section_nr(pfn), j);

			return false;
		}
	}

	return true;
}

219 220 221
/*
 * MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
 * OK to have direct references to sparsemem variables in here.
222
 * Must already be protected by mem_hotplug_begin().
223 224
 */
static int
225
memory_block_action(unsigned long phys_index, unsigned long action, int online_type)
226
{
227
	unsigned long start_pfn;
228
	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
229
	struct page *first_page;
230 231
	int ret;

232
	start_pfn = section_nr_to_pfn(phys_index);
233
	first_page = pfn_to_page(start_pfn);
234

235
	switch (action) {
236 237 238 239 240 241 242 243 244 245 246 247 248
	case MEM_ONLINE:
		if (!pages_correctly_reserved(start_pfn))
			return -EBUSY;

		ret = online_pages(start_pfn, nr_pages, online_type);
		break;
	case MEM_OFFLINE:
		ret = offline_pages(start_pfn, nr_pages);
		break;
	default:
		WARN(1, KERN_WARNING "%s(%ld, %ld) unknown action: "
		     "%ld\n", __func__, phys_index, action, action);
		ret = -EINVAL;
249 250 251 252 253
	}

	return ret;
}

254
int memory_block_change_state(struct memory_block *mem,
255
		unsigned long to_state, unsigned long from_state_req)
256
{
257
	int ret = 0;
258

259 260
	if (mem->state != from_state_req)
		return -EINVAL;
261

262 263 264
	if (to_state == MEM_OFFLINE)
		mem->state = MEM_GOING_OFFLINE;

265 266 267
	ret = memory_block_action(mem->start_section_nr, to_state,
				mem->online_type);

268
	mem->state = ret ? from_state_req : to_state;
269

270 271
	return ret;
}
272

273
/* The device lock serializes operations on memory_subsys_[online|offline] */
274 275
static int memory_subsys_online(struct device *dev)
{
276
	struct memory_block *mem = to_memory_block(dev);
277
	int ret;
278

279 280
	if (mem->state == MEM_ONLINE)
		return 0;
281

282 283 284 285 286 287
	/*
	 * If we are called from store_mem_state(), online_type will be
	 * set >= 0 Otherwise we were called from the device online
	 * attribute and need to set the online_type.
	 */
	if (mem->online_type < 0)
288
		mem->online_type = MMOP_ONLINE_KEEP;
289

290
	/* Already under protection of mem_hotplug_begin() */
291
	ret = memory_block_change_state(mem, MEM_ONLINE, MEM_OFFLINE);
292

293 294
	/* clear online_type */
	mem->online_type = -1;
295 296 297 298 299

	return ret;
}

static int memory_subsys_offline(struct device *dev)
300
{
301
	struct memory_block *mem = to_memory_block(dev);
302

303 304
	if (mem->state == MEM_OFFLINE)
		return 0;
305

306 307 308 309
	/* Can't offline block with non-present sections */
	if (mem->section_count != sections_per_block)
		return -EINVAL;

310
	return memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
311
}
312

313
static ssize_t
314 315
store_mem_state(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
316
{
317
	struct memory_block *mem = to_memory_block(dev);
318
	int ret, online_type;
319

320 321 322
	ret = lock_device_hotplug_sysfs();
	if (ret)
		return ret;
323

324
	if (sysfs_streq(buf, "online_kernel"))
325
		online_type = MMOP_ONLINE_KERNEL;
326
	else if (sysfs_streq(buf, "online_movable"))
327
		online_type = MMOP_ONLINE_MOVABLE;
328
	else if (sysfs_streq(buf, "online"))
329
		online_type = MMOP_ONLINE_KEEP;
330
	else if (sysfs_streq(buf, "offline"))
331
		online_type = MMOP_OFFLINE;
332 333 334 335
	else {
		ret = -EINVAL;
		goto err;
	}
336

337 338 339 340 341 342 343 344 345
	/*
	 * Memory hotplug needs to hold mem_hotplug_begin() for probe to find
	 * the correct memory block to online before doing device_online(dev),
	 * which will take dev->mutex.  Take the lock early to prevent an
	 * inversion, memory_subsys_online() callbacks will be implemented by
	 * assuming it's already protected.
	 */
	mem_hotplug_begin();

346
	switch (online_type) {
347 348 349
	case MMOP_ONLINE_KERNEL:
	case MMOP_ONLINE_MOVABLE:
	case MMOP_ONLINE_KEEP:
350 351 352
		mem->online_type = online_type;
		ret = device_online(&mem->dev);
		break;
353
	case MMOP_OFFLINE:
354 355 356 357
		ret = device_offline(&mem->dev);
		break;
	default:
		ret = -EINVAL; /* should never happen */
358 359
	}

360
	mem_hotplug_done();
361
err:
362
	unlock_device_hotplug();
363

364 365 366 367 368 369 370 371 372 373 374 375 376 377
	if (ret)
		return ret;
	return count;
}

/*
 * phys_device is a bad name for this.  What I really want
 * is a way to differentiate between memory ranges that
 * are part of physical devices that constitute
 * a complete removable unit or fru.
 * i.e. do these ranges belong to the same physical device,
 * s.t. if I offline all of these sections I can then
 * remove the physical device?
 */
378 379
static ssize_t show_phys_device(struct device *dev,
				struct device_attribute *attr, char *buf)
380
{
381
	struct memory_block *mem = to_memory_block(dev);
382 383 384
	return sprintf(buf, "%d\n", mem->phys_device);
}

385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423
#ifdef CONFIG_MEMORY_HOTREMOVE
static ssize_t show_valid_zones(struct device *dev,
				struct device_attribute *attr, char *buf)
{
	struct memory_block *mem = to_memory_block(dev);
	unsigned long start_pfn, end_pfn;
	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
	struct page *first_page;
	struct zone *zone;

	start_pfn = section_nr_to_pfn(mem->start_section_nr);
	end_pfn = start_pfn + nr_pages;
	first_page = pfn_to_page(start_pfn);

	/* The block contains more than one zone can not be offlined. */
	if (!test_pages_in_a_zone(start_pfn, end_pfn))
		return sprintf(buf, "none\n");

	zone = page_zone(first_page);

	if (zone_idx(zone) == ZONE_MOVABLE - 1) {
		/*The mem block is the last memoryblock of this zone.*/
		if (end_pfn == zone_end_pfn(zone))
			return sprintf(buf, "%s %s\n",
					zone->name, (zone + 1)->name);
	}

	if (zone_idx(zone) == ZONE_MOVABLE) {
		/*The mem block is the first memoryblock of ZONE_MOVABLE.*/
		if (start_pfn == zone->zone_start_pfn)
			return sprintf(buf, "%s %s\n",
					zone->name, (zone - 1)->name);
	}

	return sprintf(buf, "%s\n", zone->name);
}
static DEVICE_ATTR(valid_zones, 0444, show_valid_zones, NULL);
#endif

424 425 426 427
static DEVICE_ATTR(phys_index, 0444, show_mem_start_phys_index, NULL);
static DEVICE_ATTR(state, 0644, show_mem_state, store_mem_state);
static DEVICE_ATTR(phys_device, 0444, show_phys_device, NULL);
static DEVICE_ATTR(removable, 0444, show_mem_removable, NULL);
428 429 430 431 432

/*
 * Block size attribute stuff
 */
static ssize_t
433
print_block_size(struct device *dev, struct device_attribute *attr,
434
		 char *buf)
435
{
436
	return sprintf(buf, "%lx\n", get_memory_block_size());
437 438
}

439
static DEVICE_ATTR(block_size_bytes, 0444, print_block_size, NULL);
440

441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471
/*
 * Memory auto online policy.
 */

static ssize_t
show_auto_online_blocks(struct device *dev, struct device_attribute *attr,
			char *buf)
{
	if (memhp_auto_online)
		return sprintf(buf, "online\n");
	else
		return sprintf(buf, "offline\n");
}

static ssize_t
store_auto_online_blocks(struct device *dev, struct device_attribute *attr,
			 const char *buf, size_t count)
{
	if (sysfs_streq(buf, "online"))
		memhp_auto_online = true;
	else if (sysfs_streq(buf, "offline"))
		memhp_auto_online = false;
	else
		return -EINVAL;

	return count;
}

static DEVICE_ATTR(auto_online_blocks, 0644, show_auto_online_blocks,
		   store_auto_online_blocks);

472 473 474 475 476 477 478 479
/*
 * Some architectures will have custom drivers to do this, and
 * will not need to do it from userspace.  The fake hot-add code
 * as well as ppc64 will do all of their discovery in userspace
 * and will require this interface.
 */
#ifdef CONFIG_ARCH_MEMORY_PROBE
static ssize_t
480
memory_probe_store(struct device *dev, struct device_attribute *attr,
481
		   const char *buf, size_t count)
482 483
{
	u64 phys_addr;
484
	int nid, ret;
485
	unsigned long pages_per_block = PAGES_PER_SECTION * sections_per_block;
486

487 488 489
	ret = kstrtoull(buf, 0, &phys_addr);
	if (ret)
		return ret;
490

491 492 493
	if (phys_addr & ((pages_per_block << PAGE_SHIFT) - 1))
		return -EINVAL;

494 495 496
	nid = memory_add_physaddr_to_nid(phys_addr);
	ret = add_memory(nid, phys_addr,
			 MIN_MEMORY_BLOCK_SIZE * sections_per_block);
497

498 499
	if (ret)
		goto out;
500

501 502 503
	ret = count;
out:
	return ret;
504 505
}

506
static DEVICE_ATTR(probe, S_IWUSR, NULL, memory_probe_store);
507 508
#endif

509 510 511 512 513 514 515
#ifdef CONFIG_MEMORY_FAILURE
/*
 * Support for offlining pages of memory
 */

/* Soft offline a page */
static ssize_t
516 517
store_soft_offline_page(struct device *dev,
			struct device_attribute *attr,
518
			const char *buf, size_t count)
519 520 521 522 523
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
524
	if (kstrtoull(buf, 0, &pfn) < 0)
525 526 527 528 529 530 531 532 533 534
		return -EINVAL;
	pfn >>= PAGE_SHIFT;
	if (!pfn_valid(pfn))
		return -ENXIO;
	ret = soft_offline_page(pfn_to_page(pfn), 0);
	return ret == 0 ? count : ret;
}

/* Forcibly offline a page, including killing processes. */
static ssize_t
535 536
store_hard_offline_page(struct device *dev,
			struct device_attribute *attr,
537
			const char *buf, size_t count)
538 539 540 541 542
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
543
	if (kstrtoull(buf, 0, &pfn) < 0)
544 545
		return -EINVAL;
	pfn >>= PAGE_SHIFT;
546
	ret = memory_failure(pfn, 0, 0);
547 548 549
	return ret ? ret : count;
}

550 551
static DEVICE_ATTR(soft_offline_page, S_IWUSR, NULL, store_soft_offline_page);
static DEVICE_ATTR(hard_offline_page, S_IWUSR, NULL, store_hard_offline_page);
552 553
#endif

554 555 556 557 558
/*
 * Note that phys_device is optional.  It is here to allow for
 * differentiation between which *physical* devices each
 * section belongs to...
 */
559 560 561 562
int __weak arch_get_memory_phys_device(unsigned long start_pfn)
{
	return 0;
}
563

564 565 566 567
/*
 * A reference for the returned object is held and the reference for the
 * hinted object is released.
 */
568 569
struct memory_block *find_memory_block_hinted(struct mem_section *section,
					      struct memory_block *hint)
570
{
571
	int block_id = base_memory_block_id(__section_nr(section));
572 573
	struct device *hintdev = hint ? &hint->dev : NULL;
	struct device *dev;
574

575 576 577 578
	dev = subsys_find_device_by_id(&memory_subsys, block_id, hintdev);
	if (hint)
		put_device(&hint->dev);
	if (!dev)
579
		return NULL;
580
	return to_memory_block(dev);
581 582
}

583 584 585 586 587 588
/*
 * For now, we have a linear search to go find the appropriate
 * memory_block corresponding to a particular phys_index. If
 * this gets to be a real problem, we can always use a radix
 * tree or something here.
 *
589
 * This could be made generic for all device subsystems.
590 591 592 593 594 595
 */
struct memory_block *find_memory_block(struct mem_section *section)
{
	return find_memory_block_hinted(section, NULL);
}

596 597 598 599 600
static struct attribute *memory_memblk_attrs[] = {
	&dev_attr_phys_index.attr,
	&dev_attr_state.attr,
	&dev_attr_phys_device.attr,
	&dev_attr_removable.attr,
601 602 603
#ifdef CONFIG_MEMORY_HOTREMOVE
	&dev_attr_valid_zones.attr,
#endif
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625
	NULL
};

static struct attribute_group memory_memblk_attr_group = {
	.attrs = memory_memblk_attrs,
};

static const struct attribute_group *memory_memblk_attr_groups[] = {
	&memory_memblk_attr_group,
	NULL,
};

/*
 * register_memory - Setup a sysfs device for a memory block
 */
static
int register_memory(struct memory_block *memory)
{
	memory->dev.bus = &memory_subsys;
	memory->dev.id = memory->start_section_nr / sections_per_block;
	memory->dev.release = memory_block_release;
	memory->dev.groups = memory_memblk_attr_groups;
626
	memory->dev.offline = memory->state == MEM_OFFLINE;
627

628
	return device_register(&memory->dev);
629 630
}

631 632
static int init_memory_block(struct memory_block **memory,
			     struct mem_section *section, unsigned long state)
633
{
634
	struct memory_block *mem;
635
	unsigned long start_pfn;
636
	int scn_nr;
637 638
	int ret = 0;

639
	mem = kzalloc(sizeof(*mem), GFP_KERNEL);
640 641 642
	if (!mem)
		return -ENOMEM;

643
	scn_nr = __section_nr(section);
644 645 646
	mem->start_section_nr =
			base_memory_block_id(scn_nr) * sections_per_block;
	mem->end_section_nr = mem->start_section_nr + sections_per_block - 1;
647
	mem->state = state;
648
	start_pfn = section_nr_to_pfn(mem->start_section_nr);
649 650
	mem->phys_device = arch_get_memory_phys_device(start_pfn);

651 652 653 654 655 656
	ret = register_memory(mem);

	*memory = mem;
	return ret;
}

657
static int add_memory_block(int base_section_nr)
658
{
659 660
	struct memory_block *mem;
	int i, ret, section_count = 0, section_nr;
661

662 663 664 665 666 667 668 669
	for (i = base_section_nr;
	     (i < base_section_nr + sections_per_block) && i < NR_MEM_SECTIONS;
	     i++) {
		if (!present_section_nr(i))
			continue;
		if (section_count == 0)
			section_nr = i;
		section_count++;
670 671
	}

672 673 674 675 676 677 678
	if (section_count == 0)
		return 0;
	ret = init_memory_block(&mem, __nr_to_section(section_nr), MEM_ONLINE);
	if (ret)
		return ret;
	mem->section_count = section_count;
	return 0;
679 680
}

681 682 683 684 685 686 687
static bool is_zone_device_section(struct mem_section *ms)
{
	struct page *page;

	page = sparse_decode_mem_map(ms->section_mem_map, __section_nr(ms));
	return is_zone_device_page(page);
}
688

689 690 691 692 693 694
/*
 * need an interface for the VM to add new memory regions,
 * but without onlining it.
 */
int register_new_memory(int nid, struct mem_section *section)
{
695 696
	int ret = 0;
	struct memory_block *mem;
697

698 699 700
	if (is_zone_device_section(section))
		return 0;

701 702
	mutex_lock(&mem_sysfs_mutex);

703 704 705 706 707 708 709 710
	mem = find_memory_block(section);
	if (mem) {
		mem->section_count++;
		put_device(&mem->dev);
	} else {
		ret = init_memory_block(&mem, section, MEM_OFFLINE);
		if (ret)
			goto out;
711
		mem->section_count++;
712 713 714 715 716 717
	}

	if (mem->section_count == sections_per_block)
		ret = register_mem_sect_under_node(mem, nid);
out:
	mutex_unlock(&mem_sysfs_mutex);
718
	return ret;
719 720 721 722 723 724 725 726 727
}

#ifdef CONFIG_MEMORY_HOTREMOVE
static void
unregister_memory(struct memory_block *memory)
{
	BUG_ON(memory->dev.bus != &memory_subsys);

	/* drop the ref. we got in remove_memory_block() */
728
	put_device(&memory->dev);
729 730 731
	device_unregister(&memory->dev);
}

732
static int remove_memory_section(unsigned long node_id,
733
			       struct mem_section *section, int phys_device)
734 735 736
{
	struct memory_block *mem;

737 738 739
	if (is_zone_device_section(section))
		return 0;

740
	mutex_lock(&mem_sysfs_mutex);
741
	mem = find_memory_block(section);
742
	unregister_mem_sect_under_nodes(mem, __section_nr(section));
743 744

	mem->section_count--;
745
	if (mem->section_count == 0)
746
		unregister_memory(mem);
747
	else
748
		put_device(&mem->dev);
749

750
	mutex_unlock(&mem_sysfs_mutex);
751 752 753 754 755
	return 0;
}

int unregister_memory_section(struct mem_section *section)
{
756
	if (!present_section(section))
757 758
		return -EINVAL;

759
	return remove_memory_section(0, section, 0);
760
}
761
#endif /* CONFIG_MEMORY_HOTREMOVE */
762

763 764 765 766 767 768
/* return true if the memory block is offlined, otherwise, return false */
bool is_memblock_offlined(struct memory_block *mem)
{
	return mem->state == MEM_OFFLINE;
}

769 770 771 772 773 774 775 776 777 778 779
static struct attribute *memory_root_attrs[] = {
#ifdef CONFIG_ARCH_MEMORY_PROBE
	&dev_attr_probe.attr,
#endif

#ifdef CONFIG_MEMORY_FAILURE
	&dev_attr_soft_offline_page.attr,
	&dev_attr_hard_offline_page.attr,
#endif

	&dev_attr_block_size_bytes.attr,
780
	&dev_attr_auto_online_blocks.attr,
781 782 783 784 785 786 787 788 789 790 791 792
	NULL
};

static struct attribute_group memory_root_attr_group = {
	.attrs = memory_root_attrs,
};

static const struct attribute_group *memory_root_attr_groups[] = {
	&memory_root_attr_group,
	NULL,
};

793 794 795 796 797 798 799
/*
 * Initialize the sysfs support for memory devices...
 */
int __init memory_dev_init(void)
{
	unsigned int i;
	int ret;
800
	int err;
801
	unsigned long block_sz;
802

803
	ret = subsys_system_register(&memory_subsys, memory_root_attr_groups);
804 805
	if (ret)
		goto out;
806

807 808 809
	block_sz = get_memory_block_size();
	sections_per_block = block_sz / MIN_MEMORY_BLOCK_SIZE;

810 811 812 813
	/*
	 * Create entries for memory sections that were found
	 * during boot and have been initialized
	 */
814
	mutex_lock(&mem_sysfs_mutex);
815 816
	for (i = 0; i < NR_MEM_SECTIONS; i += sections_per_block) {
		err = add_memory_block(i);
817 818
		if (!ret)
			ret = err;
819
	}
820
	mutex_unlock(&mem_sysfs_mutex);
821

822 823
out:
	if (ret)
824
		printk(KERN_ERR "%s() failed: %d\n", __func__, ret);
825 826
	return ret;
}