memory.c 18.6 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 21
#include <linux/device.h>
#include <linux/memory.h>
#include <linux/kobject.h>
#include <linux/memory_hotplug.h>
#include <linux/mm.h>
22
#include <linux/mutex.h>
23
#include <linux/stat.h>
24
#include <linux/slab.h>
25

Arun Sharma's avatar
Arun Sharma committed
26
#include <linux/atomic.h>
27 28
#include <asm/uaccess.h>

29 30
static DEFINE_MUTEX(mem_sysfs_mutex);

31
#define MEMORY_CLASS_NAME	"memory"
32 33 34 35 36 37 38

static int sections_per_block;

static inline int base_memory_block_id(int section_nr)
{
	return section_nr / sections_per_block;
}
39

40 41 42
static int memory_subsys_online(struct device *dev);
static int memory_subsys_offline(struct device *dev);

43
static struct bus_type memory_subsys = {
44
	.name = MEMORY_CLASS_NAME,
45
	.dev_name = MEMORY_CLASS_NAME,
46 47
	.online = memory_subsys_online,
	.offline = memory_subsys_offline,
48 49
};

50
static BLOCKING_NOTIFIER_HEAD(memory_chain);
51

52
int register_memory_notifier(struct notifier_block *nb)
53
{
54
        return blocking_notifier_chain_register(&memory_chain, nb);
55
}
56
EXPORT_SYMBOL(register_memory_notifier);
57

58
void unregister_memory_notifier(struct notifier_block *nb)
59
{
60
        blocking_notifier_chain_unregister(&memory_chain, nb);
61
}
62
EXPORT_SYMBOL(unregister_memory_notifier);
63

64 65 66 67 68 69 70 71 72 73 74 75 76 77
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);

78 79 80 81 82 83 84
static void memory_block_release(struct device *dev)
{
	struct memory_block *mem = container_of(dev, struct memory_block, dev);

	kfree(mem);
}

85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
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;
}

105 106 107 108 109
/*
 * use this as the physical section index that this memsection
 * uses.
 */

110 111
static ssize_t show_mem_start_phys_index(struct device *dev,
			struct device_attribute *attr, char *buf)
112 113
{
	struct memory_block *mem =
114
		container_of(dev, struct 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
static ssize_t show_mem_end_phys_index(struct device *dev,
			struct device_attribute *attr, char *buf)
123 124
{
	struct memory_block *mem =
125
		container_of(dev, struct memory_block, dev);
126 127 128 129
	unsigned long phys_index;

	phys_index = mem->end_section_nr / sections_per_block;
	return sprintf(buf, "%08lx\n", phys_index);
130 131
}

132 133 134
/*
 * Show whether the section of memory is likely to be hot-removable
 */
135 136
static ssize_t show_mem_removable(struct device *dev,
			struct device_attribute *attr, char *buf)
137
{
138 139
	unsigned long i, pfn;
	int ret = 1;
140
	struct memory_block *mem =
141
		container_of(dev, struct memory_block, dev);
142

143
	for (i = 0; i < sections_per_block; i++) {
144
		pfn = section_nr_to_pfn(mem->start_section_nr + i);
145 146 147
		ret &= is_mem_section_removable(pfn, PAGES_PER_SECTION);
	}

148 149 150
	return sprintf(buf, "%d\n", ret);
}

151 152 153
/*
 * online, offline, going offline, etc.
 */
154 155
static ssize_t show_mem_state(struct device *dev,
			struct device_attribute *attr, char *buf)
156 157
{
	struct memory_block *mem =
158
		container_of(dev, struct memory_block, dev);
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
	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) {
		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;
	}

	return len;
}

185
int memory_notify(unsigned long val, void *v)
186
{
187
	return blocking_notifier_call_chain(&memory_chain, val, v);
188 189
}

190 191 192 193 194
int memory_isolate_notify(unsigned long val, void *v)
{
	return atomic_notifier_call_chain(&memory_isolate_chain, val, v);
}

195 196 197 198
/*
 * The probe routines leave the pages reserved, just as the bootmem code does.
 * Make sure they're still that way.
 */
199
static bool pages_correctly_reserved(unsigned long start_pfn)
200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
{
	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;
}

230 231 232 233 234
/*
 * MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
 * OK to have direct references to sparsemem variables in here.
 */
static int
235
memory_block_action(unsigned long phys_index, unsigned long action, int online_type)
236
{
237
	unsigned long start_pfn;
238
	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
239
	struct page *first_page;
240 241
	int ret;

242
	first_page = pfn_to_page(phys_index << PFN_SECTION_SHIFT);
243
	start_pfn = page_to_pfn(first_page);
244

245 246
	switch (action) {
		case MEM_ONLINE:
247
			if (!pages_correctly_reserved(start_pfn))
248 249
				return -EBUSY;

250
			ret = online_pages(start_pfn, nr_pages, online_type);
251 252
			break;
		case MEM_OFFLINE:
253
			ret = offline_pages(start_pfn, nr_pages);
254 255
			break;
		default:
256 257
			WARN(1, KERN_WARNING "%s(%ld, %ld) unknown action: "
			     "%ld\n", __func__, phys_index, action, action);
258 259 260 261 262 263
			ret = -EINVAL;
	}

	return ret;
}

264
static int __memory_block_change_state(struct memory_block *mem,
265 266
		unsigned long to_state, unsigned long from_state_req,
		int online_type)
267
{
268
	int ret = 0;
269

270 271
	if (mem->state != from_state_req)
		return -EINVAL;
272

273 274 275
	if (to_state == MEM_OFFLINE)
		mem->state = MEM_GOING_OFFLINE;

276
	ret = memory_block_action(mem->start_section_nr, to_state, online_type);
277
	mem->state = ret ? from_state_req : to_state;
278 279
	return ret;
}
280

281 282 283 284
static int memory_subsys_online(struct device *dev)
{
	struct memory_block *mem = container_of(dev, struct memory_block, dev);
	int ret;
285

286 287 288 289
	mutex_lock(&mem->state_mutex);

	ret = mem->state == MEM_ONLINE ? 0 :
		__memory_block_change_state(mem, MEM_ONLINE, MEM_OFFLINE,
290
					    ONLINE_KEEP);
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326

	mutex_unlock(&mem->state_mutex);
	return ret;
}

static int memory_subsys_offline(struct device *dev)
{
	struct memory_block *mem = container_of(dev, struct memory_block, dev);
	int ret;

	mutex_lock(&mem->state_mutex);

	ret = mem->state == MEM_OFFLINE ? 0 :
		__memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE, -1);

	mutex_unlock(&mem->state_mutex);
	return ret;
}

static int __memory_block_change_state_uevent(struct memory_block *mem,
		unsigned long to_state, unsigned long from_state_req,
		int online_type)
{
	int ret = __memory_block_change_state(mem, to_state, from_state_req,
					      online_type);
	if (!ret) {
		switch (mem->state) {
		case MEM_OFFLINE:
			kobject_uevent(&mem->dev.kobj, KOBJ_OFFLINE);
			break;
		case MEM_ONLINE:
			kobject_uevent(&mem->dev.kobj, KOBJ_ONLINE);
			break;
		default:
			break;
		}
327
	}
328 329 330
	return ret;
}

331
static int memory_block_change_state(struct memory_block *mem,
332 333
		unsigned long to_state, unsigned long from_state_req,
		int online_type)
334 335 336 337
{
	int ret;

	mutex_lock(&mem->state_mutex);
338 339
	ret = __memory_block_change_state_uevent(mem, to_state, from_state_req,
						 online_type);
340 341 342 343
	mutex_unlock(&mem->state_mutex);

	return ret;
}
344
static ssize_t
345 346
store_mem_state(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
347 348
{
	struct memory_block *mem;
349
	bool offline;
350 351
	int ret = -EINVAL;

352
	mem = container_of(dev, struct memory_block, dev);
353

354 355 356 357
	lock_device_hotplug();

	if (!strncmp(buf, "online_kernel", min_t(int, count, 13))) {
		offline = false;
358 359
		ret = memory_block_change_state(mem, MEM_ONLINE,
						MEM_OFFLINE, ONLINE_KERNEL);
360 361
	} else if (!strncmp(buf, "online_movable", min_t(int, count, 14))) {
		offline = false;
362 363
		ret = memory_block_change_state(mem, MEM_ONLINE,
						MEM_OFFLINE, ONLINE_MOVABLE);
364 365
	} else if (!strncmp(buf, "online", min_t(int, count, 6))) {
		offline = false;
366 367
		ret = memory_block_change_state(mem, MEM_ONLINE,
						MEM_OFFLINE, ONLINE_KEEP);
368 369
	} else if(!strncmp(buf, "offline", min_t(int, count, 7))) {
		offline = true;
370 371
		ret = memory_block_change_state(mem, MEM_OFFLINE,
						MEM_ONLINE, -1);
372 373 374 375 376
	}
	if (!ret)
		dev->offline = offline;

	unlock_device_hotplug();
377

378 379 380 381 382 383 384 385 386 387 388 389 390 391
	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?
 */
392 393
static ssize_t show_phys_device(struct device *dev,
				struct device_attribute *attr, char *buf)
394 395
{
	struct memory_block *mem =
396
		container_of(dev, struct memory_block, dev);
397 398 399
	return sprintf(buf, "%d\n", mem->phys_device);
}

400 401 402 403 404
static DEVICE_ATTR(phys_index, 0444, show_mem_start_phys_index, NULL);
static DEVICE_ATTR(end_phys_index, 0444, show_mem_end_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);
405 406 407 408 409

/*
 * Block size attribute stuff
 */
static ssize_t
410
print_block_size(struct device *dev, struct device_attribute *attr,
411
		 char *buf)
412
{
413
	return sprintf(buf, "%lx\n", get_memory_block_size());
414 415
}

416
static DEVICE_ATTR(block_size_bytes, 0444, print_block_size, NULL);
417 418 419 420 421 422 423 424 425

/*
 * 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
426
memory_probe_store(struct device *dev, struct device_attribute *attr,
427
		   const char *buf, size_t count)
428 429
{
	u64 phys_addr;
430
	int nid;
431
	int i, ret;
432
	unsigned long pages_per_block = PAGES_PER_SECTION * sections_per_block;
433 434 435

	phys_addr = simple_strtoull(buf, NULL, 0);

436 437 438
	if (phys_addr & ((pages_per_block << PAGE_SHIFT) - 1))
		return -EINVAL;

439 440 441 442 443
	for (i = 0; i < sections_per_block; i++) {
		nid = memory_add_physaddr_to_nid(phys_addr);
		ret = add_memory(nid, phys_addr,
				 PAGES_PER_SECTION << PAGE_SHIFT);
		if (ret)
444
			goto out;
445 446 447

		phys_addr += MIN_MEMORY_BLOCK_SIZE;
	}
448

449 450 451
	ret = count;
out:
	return ret;
452 453
}

454
static DEVICE_ATTR(probe, S_IWUSR, NULL, memory_probe_store);
455 456
#endif

457 458 459 460 461 462 463
#ifdef CONFIG_MEMORY_FAILURE
/*
 * Support for offlining pages of memory
 */

/* Soft offline a page */
static ssize_t
464 465
store_soft_offline_page(struct device *dev,
			struct device_attribute *attr,
466
			const char *buf, size_t count)
467 468 469 470 471
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
472
	if (kstrtoull(buf, 0, &pfn) < 0)
473 474 475 476 477 478 479 480 481 482
		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
483 484
store_hard_offline_page(struct device *dev,
			struct device_attribute *attr,
485
			const char *buf, size_t count)
486 487 488 489 490
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
491
	if (kstrtoull(buf, 0, &pfn) < 0)
492 493
		return -EINVAL;
	pfn >>= PAGE_SHIFT;
494
	ret = memory_failure(pfn, 0, 0);
495 496 497
	return ret ? ret : count;
}

498 499
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);
500 501
#endif

502 503 504 505 506
/*
 * Note that phys_device is optional.  It is here to allow for
 * differentiation between which *physical* devices each
 * section belongs to...
 */
507 508 509 510
int __weak arch_get_memory_phys_device(unsigned long start_pfn)
{
	return 0;
}
511

512 513 514 515
/*
 * A reference for the returned object is held and the reference for the
 * hinted object is released.
 */
516 517
struct memory_block *find_memory_block_hinted(struct mem_section *section,
					      struct memory_block *hint)
518
{
519
	int block_id = base_memory_block_id(__section_nr(section));
520 521
	struct device *hintdev = hint ? &hint->dev : NULL;
	struct device *dev;
522

523 524 525 526
	dev = subsys_find_device_by_id(&memory_subsys, block_id, hintdev);
	if (hint)
		put_device(&hint->dev);
	if (!dev)
527
		return NULL;
528
	return container_of(dev, struct memory_block, dev);
529 530
}

531 532 533 534 535 536
/*
 * 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.
 *
537
 * This could be made generic for all device subsystems.
538 539 540 541 542 543
 */
struct memory_block *find_memory_block(struct mem_section *section)
{
	return find_memory_block_hinted(section, NULL);
}

544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571
static struct attribute *memory_memblk_attrs[] = {
	&dev_attr_phys_index.attr,
	&dev_attr_end_phys_index.attr,
	&dev_attr_state.attr,
	&dev_attr_phys_device.attr,
	&dev_attr_removable.attr,
	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;
572
	memory->dev.offline = memory->state == MEM_OFFLINE;
573

574
	return device_register(&memory->dev);
575 576
}

577 578
static int init_memory_block(struct memory_block **memory,
			     struct mem_section *section, unsigned long state)
579
{
580
	struct memory_block *mem;
581
	unsigned long start_pfn;
582
	int scn_nr;
583 584
	int ret = 0;

585
	mem = kzalloc(sizeof(*mem), GFP_KERNEL);
586 587 588
	if (!mem)
		return -ENOMEM;

589
	scn_nr = __section_nr(section);
590 591 592
	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;
593
	mem->state = state;
594
	mem->section_count++;
595
	mutex_init(&mem->state_mutex);
596
	start_pfn = section_nr_to_pfn(mem->start_section_nr);
597 598
	mem->phys_device = arch_get_memory_phys_device(start_pfn);

599 600 601 602 603 604 605
	ret = register_memory(mem);

	*memory = mem;
	return ret;
}

static int add_memory_section(int nid, struct mem_section *section,
606
			struct memory_block **mem_p,
607 608
			unsigned long state, enum mem_add_context context)
{
609 610
	struct memory_block *mem = NULL;
	int scn_nr = __section_nr(section);
611 612
	int ret = 0;

613 614 615 616 617 618 619 620 621 622 623
	if (context == BOOT) {
		/* same memory block ? */
		if (mem_p && *mem_p)
			if (scn_nr >= (*mem_p)->start_section_nr &&
			    scn_nr <= (*mem_p)->end_section_nr) {
				mem = *mem_p;
				kobject_get(&mem->dev.kobj);
			}
	} else
		mem = find_memory_block(section);

624 625
	if (mem) {
		mem->section_count++;
626
		kobject_put(&mem->dev.kobj);
627
	} else {
628
		ret = init_memory_block(&mem, section, state);
629 630 631 632 633
		/* store memory_block pointer for next loop */
		if (!ret && context == BOOT)
			if (mem_p)
				*mem_p = mem;
	}
634

635
	if (!ret) {
636 637
		if (context == HOTPLUG &&
		    mem->section_count == sections_per_block)
638 639 640 641 642 643
			ret = register_mem_sect_under_node(mem, nid);
	}

	return ret;
}

644 645 646 647 648 649
/*
 * 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)
{
650 651 652 653 654 655 656
	int ret;

	mutex_lock(&mem_sysfs_mutex);
	ret = add_memory_section(nid, section, NULL, MEM_OFFLINE, HOTPLUG);
	mutex_unlock(&mem_sysfs_mutex);

	return ret;
657 658 659 660 661 662 663 664 665 666 667 668 669 670 671
}

#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() */
	kobject_put(&memory->dev.kobj);
	device_unregister(&memory->dev);
}

static int remove_memory_block(unsigned long node_id,
			       struct mem_section *section, int phys_device)
672 673 674
{
	struct memory_block *mem;

675
	mutex_lock(&mem_sysfs_mutex);
676
	mem = find_memory_block(section);
677
	unregister_mem_sect_under_nodes(mem, __section_nr(section));
678 679

	mem->section_count--;
680
	if (mem->section_count == 0)
681
		unregister_memory(mem);
682
	else
683
		kobject_put(&mem->dev.kobj);
684

685
	mutex_unlock(&mem_sysfs_mutex);
686 687 688 689 690
	return 0;
}

int unregister_memory_section(struct mem_section *section)
{
691
	if (!present_section(section))
692 693 694 695
		return -EINVAL;

	return remove_memory_block(0, section, 0);
}
696
#endif /* CONFIG_MEMORY_HOTREMOVE */
697

698 699 700 701 702 703
/* return true if the memory block is offlined, otherwise, return false */
bool is_memblock_offlined(struct memory_block *mem)
{
	return mem->state == MEM_OFFLINE;
}

704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726
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,
	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,
};

727 728 729 730 731 732 733
/*
 * Initialize the sysfs support for memory devices...
 */
int __init memory_dev_init(void)
{
	unsigned int i;
	int ret;
734
	int err;
735
	unsigned long block_sz;
736
	struct memory_block *mem = NULL;
737

738
	ret = subsys_system_register(&memory_subsys, memory_root_attr_groups);
739 740
	if (ret)
		goto out;
741

742 743 744
	block_sz = get_memory_block_size();
	sections_per_block = block_sz / MIN_MEMORY_BLOCK_SIZE;

745 746 747 748
	/*
	 * Create entries for memory sections that were found
	 * during boot and have been initialized
	 */
749
	mutex_lock(&mem_sysfs_mutex);
750
	for (i = 0; i < NR_MEM_SECTIONS; i++) {
751
		if (!present_section_nr(i))
752
			continue;
753 754 755 756
		/* don't need to reuse memory_block if only one per block */
		err = add_memory_section(0, __nr_to_section(i),
				 (sections_per_block == 1) ? NULL : &mem,
					 MEM_ONLINE,
757
					 BOOT);
758 759
		if (!ret)
			ret = err;
760
	}
761
	mutex_unlock(&mem_sysfs_mutex);
762

763 764
out:
	if (ret)
765
		printk(KERN_ERR "%s() failed: %d\n", __func__, ret);
766 767
	return ret;
}