memory.c 14.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
/*
 * drivers/base/memory.c - basic Memory class support
 *
 * 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/sysdev.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/topology.h>
17
#include <linux/capability.h>
18 19 20 21 22
#include <linux/device.h>
#include <linux/memory.h>
#include <linux/kobject.h>
#include <linux/memory_hotplug.h>
#include <linux/mm.h>
23
#include <linux/mutex.h>
24
#include <linux/stat.h>
25
#include <linux/slab.h>
26

27 28 29
#include <asm/atomic.h>
#include <asm/uaccess.h>

30 31
static DEFINE_MUTEX(mem_sysfs_mutex);

32 33 34
#define MEMORY_CLASS_NAME	"memory"

static struct sysdev_class memory_sysdev_class = {
35
	.name = MEMORY_CLASS_NAME,
36 37
};

38
static const char *memory_uevent_name(struct kset *kset, struct kobject *kobj)
39 40 41 42
{
	return MEMORY_CLASS_NAME;
}

43
static int memory_uevent(struct kset *kset, struct kobject *obj, struct kobj_uevent_env *env)
44 45 46 47 48 49
{
	int retval = 0;

	return retval;
}

50
static const struct kset_uevent_ops memory_uevent_ops = {
51 52
	.name		= memory_uevent_name,
	.uevent		= memory_uevent,
53 54
};

55
static BLOCKING_NOTIFIER_HEAD(memory_chain);
56

57
int register_memory_notifier(struct notifier_block *nb)
58
{
59
        return blocking_notifier_chain_register(&memory_chain, nb);
60
}
61
EXPORT_SYMBOL(register_memory_notifier);
62

63
void unregister_memory_notifier(struct notifier_block *nb)
64
{
65
        blocking_notifier_chain_unregister(&memory_chain, nb);
66
}
67
EXPORT_SYMBOL(unregister_memory_notifier);
68

69 70 71 72 73 74 75 76 77 78 79 80 81 82
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);

83 84 85
/*
 * register_memory - Setup a sysfs device for a memory block
 */
86 87
static
int register_memory(struct memory_block *memory, struct mem_section *section)
88 89 90 91 92 93 94 95 96 97 98
{
	int error;

	memory->sysdev.cls = &memory_sysdev_class;
	memory->sysdev.id = __section_nr(section);

	error = sysdev_register(&memory->sysdev);
	return error;
}

static void
99
unregister_memory(struct memory_block *memory, struct mem_section *section)
100 101 102 103
{
	BUG_ON(memory->sysdev.cls != &memory_sysdev_class);
	BUG_ON(memory->sysdev.id != __section_nr(section));

104 105
	/* drop the ref. we got in remove_memory_block() */
	kobject_put(&memory->sysdev.kobj);
106 107 108 109 110 111 112 113
	sysdev_unregister(&memory->sysdev);
}

/*
 * use this as the physical section index that this memsection
 * uses.
 */

114 115
static ssize_t show_mem_phys_index(struct sys_device *dev,
			struct sysdev_attribute *attr, char *buf)
116 117 118 119 120 121
{
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);
	return sprintf(buf, "%08lx\n", mem->phys_index);
}

122 123 124
/*
 * Show whether the section of memory is likely to be hot-removable
 */
125 126
static ssize_t show_mem_removable(struct sys_device *dev,
			struct sysdev_attribute *attr, char *buf)
127 128 129 130 131 132 133 134 135 136 137
{
	unsigned long start_pfn;
	int ret;
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);

	start_pfn = section_nr_to_pfn(mem->phys_index);
	ret = is_mem_section_removable(start_pfn, PAGES_PER_SECTION);
	return sprintf(buf, "%d\n", ret);
}

138 139 140
/*
 * online, offline, going offline, etc.
 */
141 142
static ssize_t show_mem_state(struct sys_device *dev,
			struct sysdev_attribute *attr, char *buf)
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
{
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);
	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;
}

172
int memory_notify(unsigned long val, void *v)
173
{
174
	return blocking_notifier_call_chain(&memory_chain, val, v);
175 176
}

177 178 179 180 181
int memory_isolate_notify(unsigned long val, void *v)
{
	return atomic_notifier_call_chain(&memory_isolate_chain, val, v);
}

182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231
/*
 * MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
 * OK to have direct references to sparsemem variables in here.
 */
static int
memory_block_action(struct memory_block *mem, unsigned long action)
{
	int i;
	unsigned long psection;
	unsigned long start_pfn, start_paddr;
	struct page *first_page;
	int ret;
	int old_state = mem->state;

	psection = mem->phys_index;
	first_page = pfn_to_page(psection << PFN_SECTION_SHIFT);

	/*
	 * The probe routines leave the pages reserved, just
	 * as the bootmem code does.  Make sure they're still
	 * that way.
	 */
	if (action == MEM_ONLINE) {
		for (i = 0; i < PAGES_PER_SECTION; i++) {
			if (PageReserved(first_page+i))
				continue;

			printk(KERN_WARNING "section number %ld page number %d "
				"not reserved, was it already online? \n",
				psection, i);
			return -EBUSY;
		}
	}

	switch (action) {
		case MEM_ONLINE:
			start_pfn = page_to_pfn(first_page);
			ret = online_pages(start_pfn, PAGES_PER_SECTION);
			break;
		case MEM_OFFLINE:
			mem->state = MEM_GOING_OFFLINE;
			start_paddr = page_to_pfn(first_page) << PAGE_SHIFT;
			ret = remove_memory(start_paddr,
					    PAGES_PER_SECTION << PAGE_SHIFT);
			if (ret) {
				mem->state = old_state;
				break;
			}
			break;
		default:
Arjan van de Ven's avatar
Arjan van de Ven committed
232
			WARN(1, KERN_WARNING "%s(%p, %ld) unknown action: %ld\n",
233
					__func__, mem, action, action);
234 235 236 237 238 239 240 241 242 243
			ret = -EINVAL;
	}

	return ret;
}

static int memory_block_change_state(struct memory_block *mem,
		unsigned long to_state, unsigned long from_state_req)
{
	int ret = 0;
244
	mutex_lock(&mem->state_mutex);
245 246 247 248 249 250 251 252 253 254 255

	if (mem->state != from_state_req) {
		ret = -EINVAL;
		goto out;
	}

	ret = memory_block_action(mem, to_state);
	if (!ret)
		mem->state = to_state;

out:
256
	mutex_unlock(&mem->state_mutex);
257 258 259 260
	return ret;
}

static ssize_t
261 262
store_mem_state(struct sys_device *dev,
		struct sysdev_attribute *attr, const char *buf, size_t count)
263 264 265 266 267 268 269 270
{
	struct memory_block *mem;
	unsigned int phys_section_nr;
	int ret = -EINVAL;

	mem = container_of(dev, struct memory_block, sysdev);
	phys_section_nr = mem->phys_index;

271
	if (!present_section_nr(phys_section_nr))
272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292
		goto out;

	if (!strncmp(buf, "online", min((int)count, 6)))
		ret = memory_block_change_state(mem, MEM_ONLINE, MEM_OFFLINE);
	else if(!strncmp(buf, "offline", min((int)count, 7)))
		ret = memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
out:
	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?
 */
293 294
static ssize_t show_phys_device(struct sys_device *dev,
				struct sysdev_attribute *attr, char *buf)
295 296 297 298 299 300 301 302 303
{
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);
	return sprintf(buf, "%d\n", mem->phys_device);
}

static SYSDEV_ATTR(phys_index, 0444, show_mem_phys_index, NULL);
static SYSDEV_ATTR(state, 0644, show_mem_state, store_mem_state);
static SYSDEV_ATTR(phys_device, 0444, show_phys_device, NULL);
304
static SYSDEV_ATTR(removable, 0444, show_mem_removable, NULL);
305 306 307 308 309 310 311 312 313 314

#define mem_create_simple_file(mem, attr_name)	\
	sysdev_create_file(&mem->sysdev, &attr_##attr_name)
#define mem_remove_simple_file(mem, attr_name)	\
	sysdev_remove_file(&mem->sysdev, &attr_##attr_name)

/*
 * Block size attribute stuff
 */
static ssize_t
315 316
print_block_size(struct sysdev_class *class, struct sysdev_class_attribute *attr,
		 char *buf)
317
{
318
	return sprintf(buf, "%lx\n", (unsigned long)PAGES_PER_SECTION * PAGE_SIZE);
319 320
}

321
static SYSDEV_CLASS_ATTR(block_size_bytes, 0444, print_block_size, NULL);
322 323 324

static int block_size_init(void)
{
325
	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
326
				&attr_block_size_bytes.attr);
327 328 329 330 331 332 333 334 335 336
}

/*
 * 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
337 338
memory_probe_store(struct class *class, struct class_attribute *attr,
		   const char *buf, size_t count)
339 340
{
	u64 phys_addr;
341
	int nid;
342 343 344 345
	int ret;

	phys_addr = simple_strtoull(buf, NULL, 0);

346 347
	nid = memory_add_physaddr_to_nid(phys_addr);
	ret = add_memory(nid, phys_addr, PAGES_PER_SECTION << PAGE_SHIFT);
348 349 350 351 352 353

	if (ret)
		count = ret;

	return count;
}
354
static CLASS_ATTR(probe, S_IWUSR, NULL, memory_probe_store);
355 356 357

static int memory_probe_init(void)
{
358
	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
359
				&class_attr_probe.attr);
360 361
}
#else
362 363 364 365
static inline int memory_probe_init(void)
{
	return 0;
}
366 367
#endif

368 369 370 371 372 373 374
#ifdef CONFIG_MEMORY_FAILURE
/*
 * Support for offlining pages of memory
 */

/* Soft offline a page */
static ssize_t
375 376 377
store_soft_offline_page(struct class *class,
			struct class_attribute *attr,
			const char *buf, size_t count)
378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	if (strict_strtoull(buf, 0, &pfn) < 0)
		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
394 395 396
store_hard_offline_page(struct class *class,
			struct class_attribute *attr,
			const char *buf, size_t count)
397 398 399 400 401 402 403 404 405 406 407 408
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	if (strict_strtoull(buf, 0, &pfn) < 0)
		return -EINVAL;
	pfn >>= PAGE_SHIFT;
	ret = __memory_failure(pfn, 0, 0);
	return ret ? ret : count;
}

409 410
static CLASS_ATTR(soft_offline_page, 0644, NULL, store_soft_offline_page);
static CLASS_ATTR(hard_offline_page, 0644, NULL, store_hard_offline_page);
411 412 413 414 415 416

static __init int memory_fail_init(void)
{
	int err;

	err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
417
				&class_attr_soft_offline_page.attr);
418 419
	if (!err)
		err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
420
				&class_attr_hard_offline_page.attr);
421 422 423 424 425 426 427 428 429
	return err;
}
#else
static inline int memory_fail_init(void)
{
	return 0;
}
#endif

430 431 432 433 434
/*
 * Note that phys_device is optional.  It is here to allow for
 * differentiation between which *physical* devices each
 * section belongs to...
 */
435 436 437 438
int __weak arch_get_memory_phys_device(unsigned long start_pfn)
{
	return 0;
}
439

440 441
struct memory_block *find_memory_block_hinted(struct mem_section *section,
					      struct memory_block *hint)
442 443 444 445 446 447
{
	struct kobject *kobj;
	struct sys_device *sysdev;
	struct memory_block *mem;
	char name[sizeof(MEMORY_CLASS_NAME) + 9 + 1];

448 449
	kobj = hint ? &hint->sysdev.kobj : NULL;

450 451 452 453 454 455
	/*
	 * This only works because we know that section == sysdev->id
	 * slightly redundant with sysdev_register()
	 */
	sprintf(&name[0], "%s%d", MEMORY_CLASS_NAME, __section_nr(section));

456
	kobj = kset_find_obj_hinted(&memory_sysdev_class.kset, name, kobj);
457 458 459 460 461 462 463 464 465
	if (!kobj)
		return NULL;

	sysdev = container_of(kobj, struct sys_device, kobj);
	mem = container_of(sysdev, struct memory_block, sysdev);

	return mem;
}

466 467 468 469 470 471 472 473 474 475 476 477 478
/*
 * 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.
 *
 * This could be made generic for all sysdev classes.
 */
struct memory_block *find_memory_block(struct mem_section *section)
{
	return find_memory_block_hinted(section, NULL);
}

479 480 481 482 483 484 485 486 487 488
static int add_memory_block(int nid, struct mem_section *section,
			unsigned long state, enum mem_add_context context)
{
	struct memory_block *mem = kzalloc(sizeof(*mem), GFP_KERNEL);
	unsigned long start_pfn;
	int ret = 0;

	if (!mem)
		return -ENOMEM;

489 490
	mutex_lock(&mem_sysfs_mutex);

491 492
	mem->phys_index = __section_nr(section);
	mem->state = state;
493
	mem->section_count++;
494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511
	mutex_init(&mem->state_mutex);
	start_pfn = section_nr_to_pfn(mem->phys_index);
	mem->phys_device = arch_get_memory_phys_device(start_pfn);

	ret = register_memory(mem, section);
	if (!ret)
		ret = mem_create_simple_file(mem, phys_index);
	if (!ret)
		ret = mem_create_simple_file(mem, state);
	if (!ret)
		ret = mem_create_simple_file(mem, phys_device);
	if (!ret)
		ret = mem_create_simple_file(mem, removable);
	if (!ret) {
		if (context == HOTPLUG)
			ret = register_mem_sect_under_node(mem, nid);
	}

512
	mutex_unlock(&mem_sysfs_mutex);
513 514 515
	return ret;
}

516 517 518 519 520
int remove_memory_block(unsigned long node_id, struct mem_section *section,
		int phys_device)
{
	struct memory_block *mem;

521
	mutex_lock(&mem_sysfs_mutex);
522
	mem = find_memory_block(section);
523 524 525 526 527 528 529 530 531 532

	mem->section_count--;
	if (mem->section_count == 0) {
		unregister_mem_sect_under_nodes(mem);
		mem_remove_simple_file(mem, phys_index);
		mem_remove_simple_file(mem, state);
		mem_remove_simple_file(mem, phys_device);
		mem_remove_simple_file(mem, removable);
		unregister_memory(mem, section);
	}
533

534
	mutex_unlock(&mem_sysfs_mutex);
535 536 537 538 539 540 541
	return 0;
}

/*
 * need an interface for the VM to add new memory regions,
 * but without onlining it.
 */
542
int register_new_memory(int nid, struct mem_section *section)
543
{
544
	return add_memory_block(nid, section, MEM_OFFLINE, HOTPLUG);
545 546 547 548
}

int unregister_memory_section(struct mem_section *section)
{
549
	if (!present_section(section))
550 551 552 553 554 555 556 557 558 559 560 561
		return -EINVAL;

	return remove_memory_block(0, section, 0);
}

/*
 * Initialize the sysfs support for memory devices...
 */
int __init memory_dev_init(void)
{
	unsigned int i;
	int ret;
562
	int err;
563

564
	memory_sysdev_class.kset.uevent_ops = &memory_uevent_ops;
565
	ret = sysdev_class_register(&memory_sysdev_class);
566 567
	if (ret)
		goto out;
568 569 570 571 572 573

	/*
	 * Create entries for memory sections that were found
	 * during boot and have been initialized
	 */
	for (i = 0; i < NR_MEM_SECTIONS; i++) {
574
		if (!present_section_nr(i))
575
			continue;
576
		err = add_memory_block(0, __nr_to_section(i), MEM_ONLINE,
577
				       BOOT);
578 579
		if (!ret)
			ret = err;
580 581
	}

582
	err = memory_probe_init();
583 584 585
	if (!ret)
		ret = err;
	err = memory_fail_init();
586 587 588 589 590 591 592
	if (!ret)
		ret = err;
	err = block_size_init();
	if (!ret)
		ret = err;
out:
	if (ret)
593
		printk(KERN_ERR "%s() failed: %d\n", __func__, ret);
594 595
	return ret;
}