mempolicy.c 54.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4
/*
 * Simple NUMA memory policy for the Linux kernel.
 *
 * Copyright 2003,2004 Andi Kleen, SuSE Labs.
5
 * (C) Copyright 2005 Christoph Lameter, Silicon Graphics, Inc.
L
Linus Torvalds 已提交
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
 * Subject to the GNU Public License, version 2.
 *
 * NUMA policy allows the user to give hints in which node(s) memory should
 * be allocated.
 *
 * Support four policies per VMA and per process:
 *
 * The VMA policy has priority over the process policy for a page fault.
 *
 * interleave     Allocate memory interleaved over a set of nodes,
 *                with normal fallback if it fails.
 *                For VMA based allocations this interleaves based on the
 *                offset into the backing object or offset into the mapping
 *                for anonymous memory. For process policy an process counter
 *                is used.
21
 *
L
Linus Torvalds 已提交
22 23
 * bind           Only allocate memory on a specific set of nodes,
 *                no fallback.
24 25 26 27
 *                FIXME: memory is allocated starting with the first node
 *                to the last. It would be better if bind would truly restrict
 *                the allocation to memory nodes instead
 *
L
Linus Torvalds 已提交
28 29 30 31 32
 * preferred       Try a specific node first before normal fallback.
 *                As a special case node -1 here means do the allocation
 *                on the local CPU. This is normally identical to default,
 *                but useful to set in a VMA when you have a non default
 *                process policy.
33
 *
L
Linus Torvalds 已提交
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79
 * default        Allocate on the local node first, or when on a VMA
 *                use the process policy. This is what Linux always did
 *		  in a NUMA aware kernel and still does by, ahem, default.
 *
 * The process policy is applied for most non interrupt memory allocations
 * in that process' context. Interrupts ignore the policies and always
 * try to allocate on the local CPU. The VMA policy is only applied for memory
 * allocations for a VMA in the VM.
 *
 * Currently there are a few corner cases in swapping where the policy
 * is not applied, but the majority should be handled. When process policy
 * is used it is not remembered over swap outs/swap ins.
 *
 * Only the highest zone in the zone hierarchy gets policied. Allocations
 * requesting a lower zone just use default policy. This implies that
 * on systems with highmem kernel lowmem allocation don't get policied.
 * Same with GFP_DMA allocations.
 *
 * For shmfs/tmpfs/hugetlbfs shared memory the policy is shared between
 * all users and remembered even when nobody has memory mapped.
 */

/* Notebook:
   fix mmap readahead to honour policy and enable policy for any page cache
   object
   statistics for bigpages
   global policy for page cache? currently it uses process policy. Requires
   first item above.
   handle mremap for shared memory (currently ignored for the policy)
   grows down?
   make bind policy root only? It can trigger oom much faster and the
   kernel is not always grateful with that.
*/

#include <linux/mempolicy.h>
#include <linux/mm.h>
#include <linux/highmem.h>
#include <linux/hugetlb.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/nodemask.h>
#include <linux/cpuset.h>
#include <linux/gfp.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/module.h>
80
#include <linux/nsproxy.h>
L
Linus Torvalds 已提交
81 82 83
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/compat.h>
84
#include <linux/swap.h>
85 86
#include <linux/seq_file.h>
#include <linux/proc_fs.h>
C
Christoph Lameter 已提交
87
#include <linux/migrate.h>
88
#include <linux/rmap.h>
89
#include <linux/security.h>
A
Adrian Bunk 已提交
90
#include <linux/syscalls.h>
91

L
Linus Torvalds 已提交
92 93 94
#include <asm/tlbflush.h>
#include <asm/uaccess.h>

95
/* Internal flags */
96
#define MPOL_MF_DISCONTIG_OK (MPOL_MF_INTERNAL << 0)	/* Skip checks for continuous vmas */
97
#define MPOL_MF_INVERT (MPOL_MF_INTERNAL << 1)		/* Invert check for nodemask */
98
#define MPOL_MF_STATS (MPOL_MF_INTERNAL << 2)		/* Gather statistics */
99

100 101
static struct kmem_cache *policy_cache;
static struct kmem_cache *sn_cache;
L
Linus Torvalds 已提交
102 103 104

/* Highest zone. An specific allocation for a zone below that is not
   policied. */
105
enum zone_type policy_zone = 0;
L
Linus Torvalds 已提交
106

107 108 109
/*
 * run-time system-wide default policy => local allocation
 */
110
struct mempolicy default_policy = {
L
Linus Torvalds 已提交
111
	.refcnt = ATOMIC_INIT(1), /* never free it */
112 113
	.mode = MPOL_PREFERRED,
	.v =  { .preferred_node =  -1 },
L
Linus Torvalds 已提交
114 115
};

116 117 118 119 120
static const struct mempolicy_operations {
	int (*create)(struct mempolicy *pol, const nodemask_t *nodes);
	void (*rebind)(struct mempolicy *pol, const nodemask_t *nodes);
} mpol_ops[MPOL_MAX];

121
/* Check that the nodemask contains at least one populated zone */
122
static int is_valid_nodemask(const nodemask_t *nodemask)
L
Linus Torvalds 已提交
123
{
124
	int nd, k;
L
Linus Torvalds 已提交
125

126 127 128 129 130 131 132 133 134 135
	/* Check that there is something useful in this mask */
	k = policy_zone;

	for_each_node_mask(nd, *nodemask) {
		struct zone *z;

		for (k = 0; k <= policy_zone; k++) {
			z = &NODE_DATA(nd)->node_zones[k];
			if (z->present_pages > 0)
				return 1;
136
		}
137
	}
138 139

	return 0;
L
Linus Torvalds 已提交
140 141
}

142 143
static inline int mpol_store_user_nodemask(const struct mempolicy *pol)
{
144 145 146 147 148 149 150 151 152
	return pol->flags & (MPOL_F_STATIC_NODES | MPOL_F_RELATIVE_NODES);
}

static void mpol_relative_nodemask(nodemask_t *ret, const nodemask_t *orig,
				   const nodemask_t *rel)
{
	nodemask_t tmp;
	nodes_fold(tmp, *orig, nodes_weight(*rel));
	nodes_onto(*ret, tmp, *rel);
153 154
}

155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181
static int mpol_new_interleave(struct mempolicy *pol, const nodemask_t *nodes)
{
	if (nodes_empty(*nodes))
		return -EINVAL;
	pol->v.nodes = *nodes;
	return 0;
}

static int mpol_new_preferred(struct mempolicy *pol, const nodemask_t *nodes)
{
	if (!nodes)
		pol->v.preferred_node = -1;	/* local allocation */
	else if (nodes_empty(*nodes))
		return -EINVAL;			/*  no allowed nodes */
	else
		pol->v.preferred_node = first_node(*nodes);
	return 0;
}

static int mpol_new_bind(struct mempolicy *pol, const nodemask_t *nodes)
{
	if (!is_valid_nodemask(nodes))
		return -EINVAL;
	pol->v.nodes = *nodes;
	return 0;
}

L
Linus Torvalds 已提交
182
/* Create a new policy */
183 184
static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
				  nodemask_t *nodes)
L
Linus Torvalds 已提交
185 186
{
	struct mempolicy *policy;
187
	nodemask_t cpuset_context_nmask;
188
	int ret;
L
Linus Torvalds 已提交
189

190 191
	pr_debug("setting mode %d flags %d nodes[0] %lx\n",
		 mode, flags, nodes ? nodes_addr(*nodes)[0] : -1);
192

193 194
	if (mode == MPOL_DEFAULT) {
		if (nodes && !nodes_empty(*nodes))
195
			return ERR_PTR(-EINVAL);
196
		return NULL;	/* simply delete any existing policy */
197
	}
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213
	VM_BUG_ON(!nodes);

	/*
	 * MPOL_PREFERRED cannot be used with MPOL_F_STATIC_NODES or
	 * MPOL_F_RELATIVE_NODES if the nodemask is empty (local allocation).
	 * All other modes require a valid pointer to a non-empty nodemask.
	 */
	if (mode == MPOL_PREFERRED) {
		if (nodes_empty(*nodes)) {
			if (((flags & MPOL_F_STATIC_NODES) ||
			     (flags & MPOL_F_RELATIVE_NODES)))
				return ERR_PTR(-EINVAL);
			nodes = NULL;	/* flag local alloc */
		}
	} else if (nodes_empty(*nodes))
		return ERR_PTR(-EINVAL);
L
Linus Torvalds 已提交
214 215 216 217
	policy = kmem_cache_alloc(policy_cache, GFP_KERNEL);
	if (!policy)
		return ERR_PTR(-ENOMEM);
	atomic_set(&policy->refcnt, 1);
218
	policy->mode = mode;
219
	policy->flags = flags;
220

221 222 223 224
	if (nodes) {
		/*
		 * cpuset related setup doesn't apply to local allocation
		 */
225 226 227 228 229 230 231 232 233 234 235 236 237 238 239
		cpuset_update_task_memory_state();
		if (flags & MPOL_F_RELATIVE_NODES)
			mpol_relative_nodemask(&cpuset_context_nmask, nodes,
					       &cpuset_current_mems_allowed);
		else
			nodes_and(cpuset_context_nmask, *nodes,
				  cpuset_current_mems_allowed);
		if (mpol_store_user_nodemask(policy))
			policy->w.user_nodemask = *nodes;
		else
			policy->w.cpuset_mems_allowed =
						cpuset_mems_allowed(current);
	}

	ret = mpol_ops[mode].create(policy,
240
				nodes ? &cpuset_context_nmask : NULL);
241 242 243 244
	if (ret < 0) {
		kmem_cache_free(policy_cache, policy);
		return ERR_PTR(ret);
	}
L
Linus Torvalds 已提交
245
	return policy;
246 247
}

248 249 250 251 252 253 254 255
/* Slow path of a mpol destructor. */
void __mpol_put(struct mempolicy *p)
{
	if (!atomic_dec_and_test(&p->refcnt))
		return;
	kmem_cache_free(policy_cache, p);
}

256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
static void mpol_rebind_default(struct mempolicy *pol, const nodemask_t *nodes)
{
}

static void mpol_rebind_nodemask(struct mempolicy *pol,
				 const nodemask_t *nodes)
{
	nodemask_t tmp;

	if (pol->flags & MPOL_F_STATIC_NODES)
		nodes_and(tmp, pol->w.user_nodemask, *nodes);
	else if (pol->flags & MPOL_F_RELATIVE_NODES)
		mpol_relative_nodemask(&tmp, &pol->w.user_nodemask, nodes);
	else {
		nodes_remap(tmp, pol->v.nodes, pol->w.cpuset_mems_allowed,
			    *nodes);
		pol->w.cpuset_mems_allowed = *nodes;
	}
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
	pol->v.nodes = tmp;
	if (!node_isset(current->il_next, tmp)) {
		current->il_next = next_node(current->il_next, tmp);
		if (current->il_next >= MAX_NUMNODES)
			current->il_next = first_node(tmp);
		if (current->il_next >= MAX_NUMNODES)
			current->il_next = numa_node_id();
	}
}

static void mpol_rebind_preferred(struct mempolicy *pol,
				  const nodemask_t *nodes)
{
	nodemask_t tmp;

	if (pol->flags & MPOL_F_STATIC_NODES) {
		int node = first_node(pol->w.user_nodemask);

		if (node_isset(node, *nodes))
			pol->v.preferred_node = node;
		else
			pol->v.preferred_node = -1;
	} else if (pol->flags & MPOL_F_RELATIVE_NODES) {
		mpol_relative_nodemask(&tmp, &pol->w.user_nodemask, nodes);
		pol->v.preferred_node = first_node(tmp);
300
	} else if (pol->v.preferred_node != -1) {
301 302 303 304 305
		pol->v.preferred_node = node_remap(pol->v.preferred_node,
						   pol->w.cpuset_mems_allowed,
						   *nodes);
		pol->w.cpuset_mems_allowed = *nodes;
	}
L
Linus Torvalds 已提交
306 307
}

308 309 310 311 312 313 314 315 316
/* Migrate a policy to a different set of nodes */
static void mpol_rebind_policy(struct mempolicy *pol,
			       const nodemask_t *newmask)
{
	if (!pol)
		return;
	if (!mpol_store_user_nodemask(pol) &&
	    nodes_equal(pol->w.cpuset_mems_allowed, *newmask))
		return;
317
	mpol_ops[pol->mode].rebind(pol, newmask);
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345
}

/*
 * Wrapper for mpol_rebind_policy() that just requires task
 * pointer, and updates task mempolicy.
 */

void mpol_rebind_task(struct task_struct *tsk, const nodemask_t *new)
{
	mpol_rebind_policy(tsk->mempolicy, new);
}

/*
 * Rebind each vma in mm to new nodemask.
 *
 * Call holding a reference to mm.  Takes mm->mmap_sem during call.
 */

void mpol_rebind_mm(struct mm_struct *mm, nodemask_t *new)
{
	struct vm_area_struct *vma;

	down_write(&mm->mmap_sem);
	for (vma = mm->mmap; vma; vma = vma->vm_next)
		mpol_rebind_policy(vma->vm_policy, new);
	up_write(&mm->mmap_sem);
}

346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363
static const struct mempolicy_operations mpol_ops[MPOL_MAX] = {
	[MPOL_DEFAULT] = {
		.rebind = mpol_rebind_default,
	},
	[MPOL_INTERLEAVE] = {
		.create = mpol_new_interleave,
		.rebind = mpol_rebind_nodemask,
	},
	[MPOL_PREFERRED] = {
		.create = mpol_new_preferred,
		.rebind = mpol_rebind_preferred,
	},
	[MPOL_BIND] = {
		.create = mpol_new_bind,
		.rebind = mpol_rebind_nodemask,
	},
};

C
Christoph Lameter 已提交
364
static void gather_stats(struct page *, void *, int pte_dirty);
365 366
static void migrate_page_add(struct page *page, struct list_head *pagelist,
				unsigned long flags);
367

368
/* Scan through pages checking if pages follow certain conditions. */
N
Nick Piggin 已提交
369
static int check_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
370 371
		unsigned long addr, unsigned long end,
		const nodemask_t *nodes, unsigned long flags,
372
		void *private)
L
Linus Torvalds 已提交
373
{
374 375
	pte_t *orig_pte;
	pte_t *pte;
376
	spinlock_t *ptl;
377

378
	orig_pte = pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
379
	do {
380
		struct page *page;
381
		int nid;
382 383

		if (!pte_present(*pte))
L
Linus Torvalds 已提交
384
			continue;
385 386
		page = vm_normal_page(vma, addr, *pte);
		if (!page)
L
Linus Torvalds 已提交
387
			continue;
388 389 390 391 392 393 394 395 396 397 398
		/*
		 * The check for PageReserved here is important to avoid
		 * handling zero pages and other pages that may have been
		 * marked special by the system.
		 *
		 * If the PageReserved would not be checked here then f.e.
		 * the location of the zero page could have an influence
		 * on MPOL_MF_STRICT, zero pages would be counted for
		 * the per node stats, and there would be useless attempts
		 * to put zero pages on the migration list.
		 */
399 400
		if (PageReserved(page))
			continue;
401
		nid = page_to_nid(page);
402 403 404
		if (node_isset(nid, *nodes) == !!(flags & MPOL_MF_INVERT))
			continue;

405
		if (flags & MPOL_MF_STATS)
C
Christoph Lameter 已提交
406
			gather_stats(page, private, pte_dirty(*pte));
407
		else if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
408
			migrate_page_add(page, private, flags);
409 410
		else
			break;
411
	} while (pte++, addr += PAGE_SIZE, addr != end);
412
	pte_unmap_unlock(orig_pte, ptl);
413 414 415
	return addr != end;
}

N
Nick Piggin 已提交
416
static inline int check_pmd_range(struct vm_area_struct *vma, pud_t *pud,
417 418
		unsigned long addr, unsigned long end,
		const nodemask_t *nodes, unsigned long flags,
419
		void *private)
420 421 422 423 424 425 426 427 428
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
429
		if (check_pte_range(vma, pmd, addr, next, nodes,
430
				    flags, private))
431 432 433 434 435
			return -EIO;
	} while (pmd++, addr = next, addr != end);
	return 0;
}

N
Nick Piggin 已提交
436
static inline int check_pud_range(struct vm_area_struct *vma, pgd_t *pgd,
437 438
		unsigned long addr, unsigned long end,
		const nodemask_t *nodes, unsigned long flags,
439
		void *private)
440 441 442 443 444 445 446 447 448
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
449
		if (check_pmd_range(vma, pud, addr, next, nodes,
450
				    flags, private))
451 452 453 454 455
			return -EIO;
	} while (pud++, addr = next, addr != end);
	return 0;
}

N
Nick Piggin 已提交
456
static inline int check_pgd_range(struct vm_area_struct *vma,
457 458
		unsigned long addr, unsigned long end,
		const nodemask_t *nodes, unsigned long flags,
459
		void *private)
460 461 462 463
{
	pgd_t *pgd;
	unsigned long next;

N
Nick Piggin 已提交
464
	pgd = pgd_offset(vma->vm_mm, addr);
465 466 467 468
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
469
		if (check_pud_range(vma, pgd, addr, next, nodes,
470
				    flags, private))
471 472 473
			return -EIO;
	} while (pgd++, addr = next, addr != end);
	return 0;
L
Linus Torvalds 已提交
474 475
}

476 477 478 479 480
/*
 * Check if all pages in a range are on a set of nodes.
 * If pagelist != NULL then isolate pages from the LRU and
 * put them on the pagelist.
 */
L
Linus Torvalds 已提交
481 482
static struct vm_area_struct *
check_range(struct mm_struct *mm, unsigned long start, unsigned long end,
483
		const nodemask_t *nodes, unsigned long flags, void *private)
L
Linus Torvalds 已提交
484 485 486 487
{
	int err;
	struct vm_area_struct *first, *vma, *prev;

488 489
	if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) {

C
Christoph Lameter 已提交
490 491 492
		err = migrate_prep();
		if (err)
			return ERR_PTR(err);
493
	}
494

L
Linus Torvalds 已提交
495 496 497 498 499
	first = find_vma(mm, start);
	if (!first)
		return ERR_PTR(-EFAULT);
	prev = NULL;
	for (vma = first; vma && vma->vm_start < end; vma = vma->vm_next) {
500 501 502 503 504 505 506 507 508 509
		if (!(flags & MPOL_MF_DISCONTIG_OK)) {
			if (!vma->vm_next && vma->vm_end < end)
				return ERR_PTR(-EFAULT);
			if (prev && prev->vm_end < vma->vm_start)
				return ERR_PTR(-EFAULT);
		}
		if (!is_vm_hugetlb_page(vma) &&
		    ((flags & MPOL_MF_STRICT) ||
		     ((flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) &&
				vma_migratable(vma)))) {
A
Andi Kleen 已提交
510
			unsigned long endvma = vma->vm_end;
511

A
Andi Kleen 已提交
512 513 514 515
			if (endvma > end)
				endvma = end;
			if (vma->vm_start > start)
				start = vma->vm_start;
516
			err = check_pgd_range(vma, start, endvma, nodes,
517
						flags, private);
L
Linus Torvalds 已提交
518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533
			if (err) {
				first = ERR_PTR(err);
				break;
			}
		}
		prev = vma;
	}
	return first;
}

/* Apply policy to a single VMA */
static int policy_vma(struct vm_area_struct *vma, struct mempolicy *new)
{
	int err = 0;
	struct mempolicy *old = vma->vm_policy;

534
	pr_debug("vma %lx-%lx/%lx vm_ops %p vm_file %p set_policy %p\n",
L
Linus Torvalds 已提交
535 536 537 538 539 540 541 542 543
		 vma->vm_start, vma->vm_end, vma->vm_pgoff,
		 vma->vm_ops, vma->vm_file,
		 vma->vm_ops ? vma->vm_ops->set_policy : NULL);

	if (vma->vm_ops && vma->vm_ops->set_policy)
		err = vma->vm_ops->set_policy(vma, new);
	if (!err) {
		mpol_get(new);
		vma->vm_policy = new;
544
		mpol_put(old);
L
Linus Torvalds 已提交
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
	}
	return err;
}

/* Step 2: apply policy to a range and do splits. */
static int mbind_range(struct vm_area_struct *vma, unsigned long start,
		       unsigned long end, struct mempolicy *new)
{
	struct vm_area_struct *next;
	int err;

	err = 0;
	for (; vma && vma->vm_start < end; vma = next) {
		next = vma->vm_next;
		if (vma->vm_start < start)
			err = split_vma(vma->vm_mm, vma, start, 1);
		if (!err && vma->vm_end > end)
			err = split_vma(vma->vm_mm, vma, end, 0);
		if (!err)
			err = policy_vma(vma, new);
		if (err)
			break;
	}
	return err;
}

571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600
/*
 * Update task->flags PF_MEMPOLICY bit: set iff non-default
 * mempolicy.  Allows more rapid checking of this (combined perhaps
 * with other PF_* flag bits) on memory allocation hot code paths.
 *
 * If called from outside this file, the task 'p' should -only- be
 * a newly forked child not yet visible on the task list, because
 * manipulating the task flags of a visible task is not safe.
 *
 * The above limitation is why this routine has the funny name
 * mpol_fix_fork_child_flag().
 *
 * It is also safe to call this with a task pointer of current,
 * which the static wrapper mpol_set_task_struct_flag() does,
 * for use within this file.
 */

void mpol_fix_fork_child_flag(struct task_struct *p)
{
	if (p->mempolicy)
		p->flags |= PF_MEMPOLICY;
	else
		p->flags &= ~PF_MEMPOLICY;
}

static void mpol_set_task_struct_flag(void)
{
	mpol_fix_fork_child_flag(current);
}

L
Linus Torvalds 已提交
601
/* Set the process memory policy */
602 603
static long do_set_mempolicy(unsigned short mode, unsigned short flags,
			     nodemask_t *nodes)
L
Linus Torvalds 已提交
604 605
{
	struct mempolicy *new;
606
	struct mm_struct *mm = current->mm;
L
Linus Torvalds 已提交
607

608
	new = mpol_new(mode, flags, nodes);
L
Linus Torvalds 已提交
609 610
	if (IS_ERR(new))
		return PTR_ERR(new);
611 612 613 614 615 616 617 618 619

	/*
	 * prevent changing our mempolicy while show_numa_maps()
	 * is using it.
	 * Note:  do_set_mempolicy() can be called at init time
	 * with no 'mm'.
	 */
	if (mm)
		down_write(&mm->mmap_sem);
620
	mpol_put(current->mempolicy);
L
Linus Torvalds 已提交
621
	current->mempolicy = new;
622
	mpol_set_task_struct_flag();
623
	if (new && new->mode == MPOL_INTERLEAVE &&
624
	    nodes_weight(new->v.nodes))
625
		current->il_next = first_node(new->v.nodes);
626 627 628
	if (mm)
		up_write(&mm->mmap_sem);

L
Linus Torvalds 已提交
629 630 631
	return 0;
}

632 633 634 635
/*
 * Return nodemask for policy for get_mempolicy() query
 */
static void get_policy_nodemask(struct mempolicy *p, nodemask_t *nodes)
L
Linus Torvalds 已提交
636
{
637
	nodes_clear(*nodes);
638 639 640
	if (p == &default_policy)
		return;

641
	switch (p->mode) {
642 643
	case MPOL_BIND:
		/* Fall through */
L
Linus Torvalds 已提交
644
	case MPOL_INTERLEAVE:
645
		*nodes = p->v.nodes;
L
Linus Torvalds 已提交
646 647
		break;
	case MPOL_PREFERRED:
648
		/* or use current node instead of memory_map? */
L
Linus Torvalds 已提交
649
		if (p->v.preferred_node < 0)
650
			*nodes = node_states[N_HIGH_MEMORY];
L
Linus Torvalds 已提交
651
		else
652
			node_set(p->v.preferred_node, *nodes);
L
Linus Torvalds 已提交
653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672
		break;
	default:
		BUG();
	}
}

static int lookup_node(struct mm_struct *mm, unsigned long addr)
{
	struct page *p;
	int err;

	err = get_user_pages(current, mm, addr & PAGE_MASK, 1, 0, 0, &p, NULL);
	if (err >= 0) {
		err = page_to_nid(p);
		put_page(p);
	}
	return err;
}

/* Retrieve NUMA policy */
A
Adrian Bunk 已提交
673 674
static long do_get_mempolicy(int *policy, nodemask_t *nmask,
			     unsigned long addr, unsigned long flags)
L
Linus Torvalds 已提交
675
{
676
	int err;
L
Linus Torvalds 已提交
677 678 679 680
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma = NULL;
	struct mempolicy *pol = current->mempolicy;

681
	cpuset_update_task_memory_state();
682 683
	if (flags &
		~(unsigned long)(MPOL_F_NODE|MPOL_F_ADDR|MPOL_F_MEMS_ALLOWED))
L
Linus Torvalds 已提交
684
		return -EINVAL;
685 686 687 688 689 690 691 692 693

	if (flags & MPOL_F_MEMS_ALLOWED) {
		if (flags & (MPOL_F_NODE|MPOL_F_ADDR))
			return -EINVAL;
		*policy = 0;	/* just so it's initialized */
		*nmask  = cpuset_current_mems_allowed;
		return 0;
	}

L
Linus Torvalds 已提交
694
	if (flags & MPOL_F_ADDR) {
695 696 697 698 699
		/*
		 * Do NOT fall back to task policy if the
		 * vma/shared policy at addr is NULL.  We
		 * want to return MPOL_DEFAULT in this case.
		 */
L
Linus Torvalds 已提交
700 701 702 703 704 705 706 707 708 709 710 711 712 713
		down_read(&mm->mmap_sem);
		vma = find_vma_intersection(mm, addr, addr+1);
		if (!vma) {
			up_read(&mm->mmap_sem);
			return -EFAULT;
		}
		if (vma->vm_ops && vma->vm_ops->get_policy)
			pol = vma->vm_ops->get_policy(vma, addr);
		else
			pol = vma->vm_policy;
	} else if (addr)
		return -EINVAL;

	if (!pol)
714
		pol = &default_policy;	/* indicates default behavior */
L
Linus Torvalds 已提交
715 716 717 718 719 720

	if (flags & MPOL_F_NODE) {
		if (flags & MPOL_F_ADDR) {
			err = lookup_node(mm, addr);
			if (err < 0)
				goto out;
721
			*policy = err;
L
Linus Torvalds 已提交
722
		} else if (pol == current->mempolicy &&
723
				pol->mode == MPOL_INTERLEAVE) {
724
			*policy = current->il_next;
L
Linus Torvalds 已提交
725 726 727 728
		} else {
			err = -EINVAL;
			goto out;
		}
729 730 731 732 733
	} else {
		*policy = pol == &default_policy ? MPOL_DEFAULT :
						pol->mode;
		*policy |= pol->flags;
	}
L
Linus Torvalds 已提交
734 735 736 737 738 739 740

	if (vma) {
		up_read(&current->mm->mmap_sem);
		vma = NULL;
	}

	err = 0;
741
	if (nmask)
742
		get_policy_nodemask(pol, nmask);
L
Linus Torvalds 已提交
743 744

 out:
745
	mpol_cond_put(pol);
L
Linus Torvalds 已提交
746 747 748 749 750
	if (vma)
		up_read(&current->mm->mmap_sem);
	return err;
}

C
Christoph Lameter 已提交
751
#ifdef CONFIG_MIGRATION
752 753 754
/*
 * page migration
 */
755 756
static void migrate_page_add(struct page *page, struct list_head *pagelist,
				unsigned long flags)
757 758
{
	/*
759
	 * Avoid migrating a page that is shared with others.
760
	 */
C
Christoph Lameter 已提交
761 762
	if ((flags & MPOL_MF_MOVE_ALL) || page_mapcount(page) == 1)
		isolate_lru_page(page, pagelist);
763
}
764

765
static struct page *new_node_page(struct page *page, unsigned long node, int **x)
766
{
767
	return alloc_pages_node(node, GFP_HIGHUSER_MOVABLE, 0);
768 769
}

770 771 772 773
/*
 * Migrate pages from one node to a target node.
 * Returns error or the number of pages not migrated.
 */
A
Adrian Bunk 已提交
774 775
static int migrate_to_node(struct mm_struct *mm, int source, int dest,
			   int flags)
776 777 778 779 780 781 782
{
	nodemask_t nmask;
	LIST_HEAD(pagelist);
	int err = 0;

	nodes_clear(nmask);
	node_set(source, nmask);
783

784 785 786
	check_range(mm, mm->mmap->vm_start, TASK_SIZE, &nmask,
			flags | MPOL_MF_DISCONTIG_OK, &pagelist);

787
	if (!list_empty(&pagelist))
788 789
		err = migrate_pages(&pagelist, new_node_page, dest);

790
	return err;
791 792
}

793
/*
794 795
 * Move pages between the two nodesets so as to preserve the physical
 * layout as much as possible.
796 797 798 799 800 801 802
 *
 * Returns the number of page that could not be moved.
 */
int do_migrate_pages(struct mm_struct *mm,
	const nodemask_t *from_nodes, const nodemask_t *to_nodes, int flags)
{
	LIST_HEAD(pagelist);
803 804 805
	int busy = 0;
	int err = 0;
	nodemask_t tmp;
806

807
  	down_read(&mm->mmap_sem);
808

809 810 811 812
	err = migrate_vmas(mm, from_nodes, to_nodes, flags);
	if (err)
		goto out;

813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842
/*
 * Find a 'source' bit set in 'tmp' whose corresponding 'dest'
 * bit in 'to' is not also set in 'tmp'.  Clear the found 'source'
 * bit in 'tmp', and return that <source, dest> pair for migration.
 * The pair of nodemasks 'to' and 'from' define the map.
 *
 * If no pair of bits is found that way, fallback to picking some
 * pair of 'source' and 'dest' bits that are not the same.  If the
 * 'source' and 'dest' bits are the same, this represents a node
 * that will be migrating to itself, so no pages need move.
 *
 * If no bits are left in 'tmp', or if all remaining bits left
 * in 'tmp' correspond to the same bit in 'to', return false
 * (nothing left to migrate).
 *
 * This lets us pick a pair of nodes to migrate between, such that
 * if possible the dest node is not already occupied by some other
 * source node, minimizing the risk of overloading the memory on a
 * node that would happen if we migrated incoming memory to a node
 * before migrating outgoing memory source that same node.
 *
 * A single scan of tmp is sufficient.  As we go, we remember the
 * most recent <s, d> pair that moved (s != d).  If we find a pair
 * that not only moved, but what's better, moved to an empty slot
 * (d is not set in tmp), then we break out then, with that pair.
 * Otherwise when we finish scannng from_tmp, we at least have the
 * most recent <s, d> pair that moved.  If we get all the way through
 * the scan of tmp without finding any node that moved, much less
 * moved to an empty node, then there is nothing left worth migrating.
 */
843

844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
	tmp = *from_nodes;
	while (!nodes_empty(tmp)) {
		int s,d;
		int source = -1;
		int dest = 0;

		for_each_node_mask(s, tmp) {
			d = node_remap(s, *from_nodes, *to_nodes);
			if (s == d)
				continue;

			source = s;	/* Node moved. Memorize */
			dest = d;

			/* dest not in remaining from nodes? */
			if (!node_isset(dest, tmp))
				break;
		}
		if (source == -1)
			break;

		node_clear(source, tmp);
		err = migrate_to_node(mm, source, dest, flags);
		if (err > 0)
			busy += err;
		if (err < 0)
			break;
871
	}
872
out:
873
	up_read(&mm->mmap_sem);
874 875 876
	if (err < 0)
		return err;
	return busy;
C
Christoph Lameter 已提交
877 878 879

}

880 881 882 883 884 885 886
/*
 * Allocate a new page for page migration based on vma policy.
 * Start assuming that page is mapped by vma pointed to by @private.
 * Search forward from there, if not.  N.B., this assumes that the
 * list of pages handed to migrate_pages()--which is how we get here--
 * is in virtual address order.
 */
887
static struct page *new_vma_page(struct page *page, unsigned long private, int **x)
888 889
{
	struct vm_area_struct *vma = (struct vm_area_struct *)private;
890
	unsigned long uninitialized_var(address);
891

892 893 894 895 896 897 898 899 900 901 902
	while (vma) {
		address = page_address_in_vma(page, vma);
		if (address != -EFAULT)
			break;
		vma = vma->vm_next;
	}

	/*
	 * if !vma, alloc_page_vma() will use task or system default policy
	 */
	return alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
903
}
C
Christoph Lameter 已提交
904 905 906 907 908
#else

static void migrate_page_add(struct page *page, struct list_head *pagelist,
				unsigned long flags)
{
909 910
}

C
Christoph Lameter 已提交
911 912 913 914 915
int do_migrate_pages(struct mm_struct *mm,
	const nodemask_t *from_nodes, const nodemask_t *to_nodes, int flags)
{
	return -ENOSYS;
}
916

917
static struct page *new_vma_page(struct page *page, unsigned long private, int **x)
918 919 920
{
	return NULL;
}
C
Christoph Lameter 已提交
921 922
#endif

A
Adrian Bunk 已提交
923
static long do_mbind(unsigned long start, unsigned long len,
924 925
		     unsigned short mode, unsigned short mode_flags,
		     nodemask_t *nmask, unsigned long flags)
926 927 928 929 930 931 932 933
{
	struct vm_area_struct *vma;
	struct mm_struct *mm = current->mm;
	struct mempolicy *new;
	unsigned long end;
	int err;
	LIST_HEAD(pagelist);

934 935
	if (flags & ~(unsigned long)(MPOL_MF_STRICT |
				     MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
936
		return -EINVAL;
937
	if ((flags & MPOL_MF_MOVE_ALL) && !capable(CAP_SYS_NICE))
938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953
		return -EPERM;

	if (start & ~PAGE_MASK)
		return -EINVAL;

	if (mode == MPOL_DEFAULT)
		flags &= ~MPOL_MF_STRICT;

	len = (len + PAGE_SIZE - 1) & PAGE_MASK;
	end = start + len;

	if (end < start)
		return -EINVAL;
	if (end == start)
		return 0;

954
	new = mpol_new(mode, mode_flags, nmask);
955 956 957 958 959 960 961 962 963 964
	if (IS_ERR(new))
		return PTR_ERR(new);

	/*
	 * If we are using the default policy then operation
	 * on discontinuous address spaces is okay after all
	 */
	if (!new)
		flags |= MPOL_MF_DISCONTIG_OK;

965 966 967
	pr_debug("mbind %lx-%lx mode:%d flags:%d nodes:%lx\n",
		 start, start + len, mode, mode_flags,
		 nmask ? nodes_addr(*nmask)[0] : -1);
968 969 970 971 972 973 974 975 976 977

	down_write(&mm->mmap_sem);
	vma = check_range(mm, start, end, nmask,
			  flags | MPOL_MF_INVERT, &pagelist);

	err = PTR_ERR(vma);
	if (!IS_ERR(vma)) {
		int nr_failed = 0;

		err = mbind_range(vma, start, end, new);
978

979
		if (!list_empty(&pagelist))
980 981
			nr_failed = migrate_pages(&pagelist, new_vma_page,
						(unsigned long)vma);
982 983 984 985

		if (!err && nr_failed && (flags & MPOL_MF_STRICT))
			err = -EIO;
	}
C
Christoph Lameter 已提交
986

987
	up_write(&mm->mmap_sem);
988
	mpol_put(new);
989 990 991
	return err;
}

992 993 994 995 996
/*
 * User space interface with variable sized bitmaps for nodelists.
 */

/* Copy a node mask from user space. */
997
static int get_nodes(nodemask_t *nodes, const unsigned long __user *nmask,
998 999 1000 1001 1002 1003 1004 1005 1006 1007
		     unsigned long maxnode)
{
	unsigned long k;
	unsigned long nlongs;
	unsigned long endmask;

	--maxnode;
	nodes_clear(*nodes);
	if (maxnode == 0 || !nmask)
		return 0;
A
Andi Kleen 已提交
1008
	if (maxnode > PAGE_SIZE*BITS_PER_BYTE)
C
Chris Wright 已提交
1009
		return -EINVAL;
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 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 1060 1061 1062 1063 1064 1065

	nlongs = BITS_TO_LONGS(maxnode);
	if ((maxnode % BITS_PER_LONG) == 0)
		endmask = ~0UL;
	else
		endmask = (1UL << (maxnode % BITS_PER_LONG)) - 1;

	/* When the user specified more nodes than supported just check
	   if the non supported part is all zero. */
	if (nlongs > BITS_TO_LONGS(MAX_NUMNODES)) {
		if (nlongs > PAGE_SIZE/sizeof(long))
			return -EINVAL;
		for (k = BITS_TO_LONGS(MAX_NUMNODES); k < nlongs; k++) {
			unsigned long t;
			if (get_user(t, nmask + k))
				return -EFAULT;
			if (k == nlongs - 1) {
				if (t & endmask)
					return -EINVAL;
			} else if (t)
				return -EINVAL;
		}
		nlongs = BITS_TO_LONGS(MAX_NUMNODES);
		endmask = ~0UL;
	}

	if (copy_from_user(nodes_addr(*nodes), nmask, nlongs*sizeof(unsigned long)))
		return -EFAULT;
	nodes_addr(*nodes)[nlongs-1] &= endmask;
	return 0;
}

/* Copy a kernel node mask to user space */
static int copy_nodes_to_user(unsigned long __user *mask, unsigned long maxnode,
			      nodemask_t *nodes)
{
	unsigned long copy = ALIGN(maxnode-1, 64) / 8;
	const int nbytes = BITS_TO_LONGS(MAX_NUMNODES) * sizeof(long);

	if (copy > nbytes) {
		if (copy > PAGE_SIZE)
			return -EINVAL;
		if (clear_user((char __user *)mask + nbytes, copy - nbytes))
			return -EFAULT;
		copy = nbytes;
	}
	return copy_to_user(mask, nodes_addr(*nodes), copy) ? -EFAULT : 0;
}

asmlinkage long sys_mbind(unsigned long start, unsigned long len,
			unsigned long mode,
			unsigned long __user *nmask, unsigned long maxnode,
			unsigned flags)
{
	nodemask_t nodes;
	int err;
1066
	unsigned short mode_flags;
1067

1068 1069
	mode_flags = mode & MPOL_MODE_FLAGS;
	mode &= ~MPOL_MODE_FLAGS;
1070 1071
	if (mode >= MPOL_MAX)
		return -EINVAL;
1072 1073 1074
	if ((mode_flags & MPOL_F_STATIC_NODES) &&
	    (mode_flags & MPOL_F_RELATIVE_NODES))
		return -EINVAL;
1075 1076 1077
	err = get_nodes(&nodes, nmask, maxnode);
	if (err)
		return err;
1078
	return do_mbind(start, len, mode, mode_flags, &nodes, flags);
1079 1080 1081 1082 1083 1084 1085 1086
}

/* Set the process memory policy */
asmlinkage long sys_set_mempolicy(int mode, unsigned long __user *nmask,
		unsigned long maxnode)
{
	int err;
	nodemask_t nodes;
1087
	unsigned short flags;
1088

1089 1090 1091
	flags = mode & MPOL_MODE_FLAGS;
	mode &= ~MPOL_MODE_FLAGS;
	if ((unsigned int)mode >= MPOL_MAX)
1092
		return -EINVAL;
1093 1094
	if ((flags & MPOL_F_STATIC_NODES) && (flags & MPOL_F_RELATIVE_NODES))
		return -EINVAL;
1095 1096 1097
	err = get_nodes(&nodes, nmask, maxnode);
	if (err)
		return err;
1098
	return do_set_mempolicy(mode, flags, &nodes);
1099 1100
}

1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
asmlinkage long sys_migrate_pages(pid_t pid, unsigned long maxnode,
		const unsigned long __user *old_nodes,
		const unsigned long __user *new_nodes)
{
	struct mm_struct *mm;
	struct task_struct *task;
	nodemask_t old;
	nodemask_t new;
	nodemask_t task_nodes;
	int err;

	err = get_nodes(&old, old_nodes, maxnode);
	if (err)
		return err;

	err = get_nodes(&new, new_nodes, maxnode);
	if (err)
		return err;

	/* Find the mm_struct */
	read_lock(&tasklist_lock);
1122
	task = pid ? find_task_by_vpid(pid) : current;
1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
	if (!task) {
		read_unlock(&tasklist_lock);
		return -ESRCH;
	}
	mm = get_task_mm(task);
	read_unlock(&tasklist_lock);

	if (!mm)
		return -EINVAL;

	/*
	 * Check if this process has the right to modify the specified
	 * process. The right exists if the process has administrative
A
Alexey Dobriyan 已提交
1136
	 * capabilities, superuser privileges or the same
1137 1138 1139 1140
	 * userid as the target process.
	 */
	if ((current->euid != task->suid) && (current->euid != task->uid) &&
	    (current->uid != task->suid) && (current->uid != task->uid) &&
1141
	    !capable(CAP_SYS_NICE)) {
1142 1143 1144 1145 1146 1147
		err = -EPERM;
		goto out;
	}

	task_nodes = cpuset_mems_allowed(task);
	/* Is the user allowed to access the target nodes? */
1148
	if (!nodes_subset(new, task_nodes) && !capable(CAP_SYS_NICE)) {
1149 1150 1151 1152
		err = -EPERM;
		goto out;
	}

1153
	if (!nodes_subset(new, node_states[N_HIGH_MEMORY])) {
1154 1155 1156 1157
		err = -EINVAL;
		goto out;
	}

1158 1159 1160 1161
	err = security_task_movememory(task);
	if (err)
		goto out;

1162
	err = do_migrate_pages(mm, &old, &new,
1163
		capable(CAP_SYS_NICE) ? MPOL_MF_MOVE_ALL : MPOL_MF_MOVE);
1164 1165 1166 1167 1168 1169
out:
	mmput(mm);
	return err;
}


1170 1171 1172 1173 1174 1175
/* Retrieve NUMA policy */
asmlinkage long sys_get_mempolicy(int __user *policy,
				unsigned long __user *nmask,
				unsigned long maxnode,
				unsigned long addr, unsigned long flags)
{
A
Adrian Bunk 已提交
1176 1177
	int err;
	int uninitialized_var(pval);
1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
	nodemask_t nodes;

	if (nmask != NULL && maxnode < MAX_NUMNODES)
		return -EINVAL;

	err = do_get_mempolicy(&pval, &nodes, addr, flags);

	if (err)
		return err;

	if (policy && put_user(pval, policy))
		return -EFAULT;

	if (nmask)
		err = copy_nodes_to_user(nmask, maxnode, &nodes);

	return err;
}

L
Linus Torvalds 已提交
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256
#ifdef CONFIG_COMPAT

asmlinkage long compat_sys_get_mempolicy(int __user *policy,
				     compat_ulong_t __user *nmask,
				     compat_ulong_t maxnode,
				     compat_ulong_t addr, compat_ulong_t flags)
{
	long err;
	unsigned long __user *nm = NULL;
	unsigned long nr_bits, alloc_size;
	DECLARE_BITMAP(bm, MAX_NUMNODES);

	nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
	alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;

	if (nmask)
		nm = compat_alloc_user_space(alloc_size);

	err = sys_get_mempolicy(policy, nm, nr_bits+1, addr, flags);

	if (!err && nmask) {
		err = copy_from_user(bm, nm, alloc_size);
		/* ensure entire bitmap is zeroed */
		err |= clear_user(nmask, ALIGN(maxnode-1, 8) / 8);
		err |= compat_put_bitmap(nmask, bm, nr_bits);
	}

	return err;
}

asmlinkage long compat_sys_set_mempolicy(int mode, compat_ulong_t __user *nmask,
				     compat_ulong_t maxnode)
{
	long err = 0;
	unsigned long __user *nm = NULL;
	unsigned long nr_bits, alloc_size;
	DECLARE_BITMAP(bm, MAX_NUMNODES);

	nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
	alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;

	if (nmask) {
		err = compat_get_bitmap(bm, nmask, nr_bits);
		nm = compat_alloc_user_space(alloc_size);
		err |= copy_to_user(nm, bm, alloc_size);
	}

	if (err)
		return -EFAULT;

	return sys_set_mempolicy(mode, nm, nr_bits+1);
}

asmlinkage long compat_sys_mbind(compat_ulong_t start, compat_ulong_t len,
			     compat_ulong_t mode, compat_ulong_t __user *nmask,
			     compat_ulong_t maxnode, compat_ulong_t flags)
{
	long err = 0;
	unsigned long __user *nm = NULL;
	unsigned long nr_bits, alloc_size;
1257
	nodemask_t bm;
L
Linus Torvalds 已提交
1258 1259 1260 1261 1262

	nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
	alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;

	if (nmask) {
1263
		err = compat_get_bitmap(nodes_addr(bm), nmask, nr_bits);
L
Linus Torvalds 已提交
1264
		nm = compat_alloc_user_space(alloc_size);
1265
		err |= copy_to_user(nm, nodes_addr(bm), alloc_size);
L
Linus Torvalds 已提交
1266 1267 1268 1269 1270 1271 1272 1273 1274 1275
	}

	if (err)
		return -EFAULT;

	return sys_mbind(start, len, mode, nm, nr_bits+1, flags);
}

#endif

1276 1277 1278 1279 1280 1281 1282 1283
/*
 * get_vma_policy(@task, @vma, @addr)
 * @task - task for fallback if vma policy == default
 * @vma   - virtual memory area whose policy is sought
 * @addr  - address in @vma for shared policy lookup
 *
 * Returns effective policy for a VMA at specified address.
 * Falls back to @task or system default policy, as necessary.
1284 1285 1286 1287 1288 1289 1290
 * Current or other task's task mempolicy and non-shared vma policies
 * are protected by the task's mmap_sem, which must be held for read by
 * the caller.
 * Shared policies [those marked as MPOL_F_SHARED] require an extra reference
 * count--added by the get_policy() vm_op, as appropriate--to protect against
 * freeing by another task.  It is the caller's responsibility to free the
 * extra reference for shared policies.
1291
 */
1292
static struct mempolicy *get_vma_policy(struct task_struct *task,
1293
		struct vm_area_struct *vma, unsigned long addr)
L
Linus Torvalds 已提交
1294
{
1295
	struct mempolicy *pol = task->mempolicy;
L
Linus Torvalds 已提交
1296 1297

	if (vma) {
1298
		if (vma->vm_ops && vma->vm_ops->get_policy) {
1299 1300 1301 1302
			struct mempolicy *vpol = vma->vm_ops->get_policy(vma,
									addr);
			if (vpol)
				pol = vpol;
1303
		} else if (vma->vm_policy)
L
Linus Torvalds 已提交
1304 1305 1306 1307 1308 1309 1310
			pol = vma->vm_policy;
	}
	if (!pol)
		pol = &default_policy;
	return pol;
}

1311 1312 1313 1314 1315
/*
 * Return a nodemask representing a mempolicy for filtering nodes for
 * page allocation
 */
static nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *policy)
1316 1317
{
	/* Lower zones don't get a nodemask applied for MPOL_BIND */
1318
	if (unlikely(policy->mode == MPOL_BIND) &&
1319 1320 1321 1322 1323 1324 1325
			gfp_zone(gfp) >= policy_zone &&
			cpuset_nodemask_valid_mems_allowed(&policy->v.nodes))
		return &policy->v.nodes;

	return NULL;
}

1326 1327
/* Return a zonelist indicated by gfp for node representing a mempolicy */
static struct zonelist *policy_zonelist(gfp_t gfp, struct mempolicy *policy)
L
Linus Torvalds 已提交
1328 1329 1330
{
	int nd;

1331
	switch (policy->mode) {
L
Linus Torvalds 已提交
1332 1333 1334 1335 1336 1337
	case MPOL_PREFERRED:
		nd = policy->v.preferred_node;
		if (nd < 0)
			nd = numa_node_id();
		break;
	case MPOL_BIND:
1338
		/*
1339 1340 1341 1342
		 * Normally, MPOL_BIND allocations are node-local within the
		 * allowed nodemask.  However, if __GFP_THISNODE is set and the
		 * current node is part of the mask, we use the zonelist for
		 * the first node in the mask instead.
1343 1344 1345 1346 1347 1348
		 */
		nd = numa_node_id();
		if (unlikely(gfp & __GFP_THISNODE) &&
				unlikely(!node_isset(nd, policy->v.nodes)))
			nd = first_node(policy->v.nodes);
		break;
L
Linus Torvalds 已提交
1349 1350 1351 1352 1353 1354 1355
	case MPOL_INTERLEAVE: /* should not happen */
		nd = numa_node_id();
		break;
	default:
		nd = 0;
		BUG();
	}
1356
	return node_zonelist(nd, gfp);
L
Linus Torvalds 已提交
1357 1358 1359 1360 1361 1362 1363 1364 1365
}

/* Do dynamic interleaving for a process */
static unsigned interleave_nodes(struct mempolicy *policy)
{
	unsigned nid, next;
	struct task_struct *me = current;

	nid = me->il_next;
1366
	next = next_node(nid, policy->v.nodes);
L
Linus Torvalds 已提交
1367
	if (next >= MAX_NUMNODES)
1368
		next = first_node(policy->v.nodes);
1369 1370
	if (next < MAX_NUMNODES)
		me->il_next = next;
L
Linus Torvalds 已提交
1371 1372 1373
	return nid;
}

1374 1375 1376
/*
 * Depending on the memory policy provide a node from which to allocate the
 * next slab entry.
1377 1378 1379 1380
 * @policy must be protected by freeing by the caller.  If @policy is
 * the current task's mempolicy, this protection is implicit, as only the
 * task can change it's policy.  The system default policy requires no
 * such protection.
1381 1382 1383
 */
unsigned slab_node(struct mempolicy *policy)
{
1384 1385 1386 1387 1388 1389 1390 1391
	if (!policy)
		return numa_node_id();

	switch (policy->mode) {
	case MPOL_PREFERRED:
		if (unlikely(policy->v.preferred_node >= 0))
			return policy->v.preferred_node;
		return numa_node_id();
1392

1393 1394 1395
	case MPOL_INTERLEAVE:
		return interleave_nodes(policy);

1396
	case MPOL_BIND: {
1397 1398 1399 1400
		/*
		 * Follow bind policy behavior and start allocation at the
		 * first node.
		 */
1401 1402 1403 1404 1405 1406 1407 1408
		struct zonelist *zonelist;
		struct zone *zone;
		enum zone_type highest_zoneidx = gfp_zone(GFP_KERNEL);
		zonelist = &NODE_DATA(numa_node_id())->node_zonelists[0];
		(void)first_zones_zonelist(zonelist, highest_zoneidx,
							&policy->v.nodes,
							&zone);
		return zone->node;
1409
	}
1410 1411

	default:
1412
		BUG();
1413 1414 1415
	}
}

L
Linus Torvalds 已提交
1416 1417 1418 1419
/* Do static interleaving for a VMA with known offset. */
static unsigned offset_il_node(struct mempolicy *pol,
		struct vm_area_struct *vma, unsigned long off)
{
1420
	unsigned nnodes = nodes_weight(pol->v.nodes);
1421
	unsigned target;
L
Linus Torvalds 已提交
1422 1423 1424
	int c;
	int nid = -1;

1425 1426 1427
	if (!nnodes)
		return numa_node_id();
	target = (unsigned int)off % nnodes;
L
Linus Torvalds 已提交
1428 1429
	c = 0;
	do {
1430
		nid = next_node(nid, pol->v.nodes);
L
Linus Torvalds 已提交
1431 1432 1433 1434 1435
		c++;
	} while (c <= target);
	return nid;
}

1436 1437 1438 1439 1440 1441 1442
/* Determine a node number for interleave */
static inline unsigned interleave_nid(struct mempolicy *pol,
		 struct vm_area_struct *vma, unsigned long addr, int shift)
{
	if (vma) {
		unsigned long off;

1443 1444 1445 1446 1447 1448 1449 1450 1451
		/*
		 * for small pages, there is no difference between
		 * shift and PAGE_SHIFT, so the bit-shift is safe.
		 * for huge pages, since vm_pgoff is in units of small
		 * pages, we need to shift off the always 0 bits to get
		 * a useful offset.
		 */
		BUG_ON(shift < PAGE_SHIFT);
		off = vma->vm_pgoff >> (shift - PAGE_SHIFT);
1452 1453 1454 1455 1456 1457
		off += (addr - vma->vm_start) >> shift;
		return offset_il_node(pol, vma, off);
	} else
		return interleave_nodes(pol);
}

1458
#ifdef CONFIG_HUGETLBFS
1459 1460 1461 1462 1463
/*
 * huge_zonelist(@vma, @addr, @gfp_flags, @mpol)
 * @vma = virtual memory area whose policy is sought
 * @addr = address in @vma for shared policy lookup and interleave policy
 * @gfp_flags = for requested zone
1464 1465
 * @mpol = pointer to mempolicy pointer for reference counted mempolicy
 * @nodemask = pointer to nodemask pointer for MPOL_BIND nodemask
1466
 *
1467 1468 1469 1470
 * Returns a zonelist suitable for a huge page allocation and a pointer
 * to the struct mempolicy for conditional unref after allocation.
 * If the effective policy is 'BIND, returns a pointer to the mempolicy's
 * @nodemask for filtering the zonelist.
1471
 */
1472
struct zonelist *huge_zonelist(struct vm_area_struct *vma, unsigned long addr,
1473 1474
				gfp_t gfp_flags, struct mempolicy **mpol,
				nodemask_t **nodemask)
1475
{
1476
	struct zonelist *zl;
1477

1478
	*mpol = get_vma_policy(current, vma, addr);
1479
	*nodemask = NULL;	/* assume !MPOL_BIND */
1480

1481 1482 1483 1484 1485 1486 1487
	if (unlikely((*mpol)->mode == MPOL_INTERLEAVE)) {
		zl = node_zonelist(interleave_nid(*mpol, vma, addr,
						HPAGE_SHIFT), gfp_flags);
	} else {
		zl = policy_zonelist(gfp_flags, *mpol);
		if ((*mpol)->mode == MPOL_BIND)
			*nodemask = &(*mpol)->v.nodes;
1488 1489
	}
	return zl;
1490
}
1491
#endif
1492

L
Linus Torvalds 已提交
1493 1494
/* Allocate a page in interleaved policy.
   Own path because it needs to do special accounting. */
1495 1496
static struct page *alloc_page_interleave(gfp_t gfp, unsigned order,
					unsigned nid)
L
Linus Torvalds 已提交
1497 1498 1499 1500
{
	struct zonelist *zl;
	struct page *page;

1501
	zl = node_zonelist(nid, gfp);
L
Linus Torvalds 已提交
1502
	page = __alloc_pages(gfp, order, zl);
1503
	if (page && page_zone(page) == zonelist_zone(&zl->_zonerefs[0]))
1504
		inc_zone_page_state(page, NUMA_INTERLEAVE_HIT);
L
Linus Torvalds 已提交
1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530
	return page;
}

/**
 * 	alloc_page_vma	- Allocate a page for a VMA.
 *
 * 	@gfp:
 *      %GFP_USER    user allocation.
 *      %GFP_KERNEL  kernel allocations,
 *      %GFP_HIGHMEM highmem/user allocations,
 *      %GFP_FS      allocation should not call back into a file system.
 *      %GFP_ATOMIC  don't sleep.
 *
 * 	@vma:  Pointer to VMA or NULL if not available.
 *	@addr: Virtual Address of the allocation. Must be inside the VMA.
 *
 * 	This function allocates a page from the kernel page pool and applies
 *	a NUMA policy associated with the VMA or the current process.
 *	When VMA is not NULL caller must hold down_read on the mmap_sem of the
 *	mm_struct of the VMA to prevent it from going away. Should be used for
 *	all allocations for pages that will be mapped into
 * 	user space. Returns NULL when no page can be allocated.
 *
 *	Should be called with the mm_sem of the vma hold.
 */
struct page *
A
Al Viro 已提交
1531
alloc_page_vma(gfp_t gfp, struct vm_area_struct *vma, unsigned long addr)
L
Linus Torvalds 已提交
1532
{
1533
	struct mempolicy *pol = get_vma_policy(current, vma, addr);
1534
	struct zonelist *zl;
L
Linus Torvalds 已提交
1535

1536
	cpuset_update_task_memory_state();
L
Linus Torvalds 已提交
1537

1538
	if (unlikely(pol->mode == MPOL_INTERLEAVE)) {
L
Linus Torvalds 已提交
1539
		unsigned nid;
1540 1541

		nid = interleave_nid(pol, vma, addr, PAGE_SHIFT);
1542
		mpol_cond_put(pol);
L
Linus Torvalds 已提交
1543 1544
		return alloc_page_interleave(gfp, 0, nid);
	}
1545 1546
	zl = policy_zonelist(gfp, pol);
	if (unlikely(mpol_needs_cond_ref(pol))) {
1547
		/*
1548
		 * slow path: ref counted shared policy
1549
		 */
1550
		struct page *page =  __alloc_pages_nodemask(gfp, 0,
1551
						zl, policy_nodemask(gfp, pol));
1552
		__mpol_put(pol);
1553 1554 1555 1556 1557
		return page;
	}
	/*
	 * fast path:  default or task policy
	 */
1558
	return __alloc_pages_nodemask(gfp, 0, zl, policy_nodemask(gfp, pol));
L
Linus Torvalds 已提交
1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
}

/**
 * 	alloc_pages_current - Allocate pages.
 *
 *	@gfp:
 *		%GFP_USER   user allocation,
 *      	%GFP_KERNEL kernel allocation,
 *      	%GFP_HIGHMEM highmem allocation,
 *      	%GFP_FS     don't call back into a file system.
 *      	%GFP_ATOMIC don't sleep.
 *	@order: Power of two of allocation size in pages. 0 is a single page.
 *
 *	Allocate a page from the kernel page pool.  When not in
 *	interrupt context and apply the current process NUMA policy.
 *	Returns NULL when no page can be allocated.
 *
1576
 *	Don't call cpuset_update_task_memory_state() unless
L
Linus Torvalds 已提交
1577 1578 1579
 *	1) it's ok to take cpuset_sem (can WAIT), and
 *	2) allocating for current task (not interrupt).
 */
A
Al Viro 已提交
1580
struct page *alloc_pages_current(gfp_t gfp, unsigned order)
L
Linus Torvalds 已提交
1581 1582 1583 1584
{
	struct mempolicy *pol = current->mempolicy;

	if ((gfp & __GFP_WAIT) && !in_interrupt())
1585
		cpuset_update_task_memory_state();
1586
	if (!pol || in_interrupt() || (gfp & __GFP_THISNODE))
L
Linus Torvalds 已提交
1587
		pol = &default_policy;
1588 1589 1590 1591 1592

	/*
	 * No reference counting needed for current->mempolicy
	 * nor system default_policy
	 */
1593
	if (pol->mode == MPOL_INTERLEAVE)
L
Linus Torvalds 已提交
1594
		return alloc_page_interleave(gfp, order, interleave_nodes(pol));
1595
	return __alloc_pages_nodemask(gfp, order,
1596
			policy_zonelist(gfp, pol), policy_nodemask(gfp, pol));
L
Linus Torvalds 已提交
1597 1598 1599
}
EXPORT_SYMBOL(alloc_pages_current);

1600
/*
1601
 * If mpol_dup() sees current->cpuset == cpuset_being_rebound, then it
1602 1603 1604 1605 1606 1607
 * rebinds the mempolicy its copying by calling mpol_rebind_policy()
 * with the mems_allowed returned by cpuset_mems_allowed().  This
 * keeps mempolicies cpuset relative after its cpuset moves.  See
 * further kernel/cpuset.c update_nodemask().
 */

1608 1609
/* Slow path of a mempolicy duplicate */
struct mempolicy *__mpol_dup(struct mempolicy *old)
L
Linus Torvalds 已提交
1610 1611 1612 1613 1614
{
	struct mempolicy *new = kmem_cache_alloc(policy_cache, GFP_KERNEL);

	if (!new)
		return ERR_PTR(-ENOMEM);
1615 1616 1617 1618
	if (current_cpuset_is_being_rebound()) {
		nodemask_t mems = cpuset_mems_allowed(current);
		mpol_rebind_policy(old, &mems);
	}
L
Linus Torvalds 已提交
1619 1620 1621 1622 1623
	*new = *old;
	atomic_set(&new->refcnt, 1);
	return new;
}

1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645
/*
 * If *frompol needs [has] an extra ref, copy *frompol to *tompol ,
 * eliminate the * MPOL_F_* flags that require conditional ref and
 * [NOTE!!!] drop the extra ref.  Not safe to reference *frompol directly
 * after return.  Use the returned value.
 *
 * Allows use of a mempolicy for, e.g., multiple allocations with a single
 * policy lookup, even if the policy needs/has extra ref on lookup.
 * shmem_readahead needs this.
 */
struct mempolicy *__mpol_cond_copy(struct mempolicy *tompol,
						struct mempolicy *frompol)
{
	if (!mpol_needs_cond_ref(frompol))
		return frompol;

	*tompol = *frompol;
	tompol->flags &= ~MPOL_F_SHARED;	/* copy doesn't need unref */
	__mpol_put(frompol);
	return tompol;
}

1646 1647 1648 1649 1650 1651 1652 1653 1654 1655
static int mpol_match_intent(const struct mempolicy *a,
			     const struct mempolicy *b)
{
	if (a->flags != b->flags)
		return 0;
	if (!mpol_store_user_nodemask(a))
		return 1;
	return nodes_equal(a->w.user_nodemask, b->w.user_nodemask);
}

L
Linus Torvalds 已提交
1656 1657 1658 1659 1660
/* Slow path of a mempolicy comparison */
int __mpol_equal(struct mempolicy *a, struct mempolicy *b)
{
	if (!a || !b)
		return 0;
1661
	if (a->mode != b->mode)
L
Linus Torvalds 已提交
1662
		return 0;
1663
	if (a->mode != MPOL_DEFAULT && !mpol_match_intent(a, b))
1664
		return 0;
1665
	switch (a->mode) {
1666 1667
	case MPOL_BIND:
		/* Fall through */
L
Linus Torvalds 已提交
1668
	case MPOL_INTERLEAVE:
1669
		return nodes_equal(a->v.nodes, b->v.nodes);
L
Linus Torvalds 已提交
1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738
	case MPOL_PREFERRED:
		return a->v.preferred_node == b->v.preferred_node;
	default:
		BUG();
		return 0;
	}
}

/*
 * Shared memory backing store policy support.
 *
 * Remember policies even when nobody has shared memory mapped.
 * The policies are kept in Red-Black tree linked from the inode.
 * They are protected by the sp->lock spinlock, which should be held
 * for any accesses to the tree.
 */

/* lookup first element intersecting start-end */
/* Caller holds sp->lock */
static struct sp_node *
sp_lookup(struct shared_policy *sp, unsigned long start, unsigned long end)
{
	struct rb_node *n = sp->root.rb_node;

	while (n) {
		struct sp_node *p = rb_entry(n, struct sp_node, nd);

		if (start >= p->end)
			n = n->rb_right;
		else if (end <= p->start)
			n = n->rb_left;
		else
			break;
	}
	if (!n)
		return NULL;
	for (;;) {
		struct sp_node *w = NULL;
		struct rb_node *prev = rb_prev(n);
		if (!prev)
			break;
		w = rb_entry(prev, struct sp_node, nd);
		if (w->end <= start)
			break;
		n = prev;
	}
	return rb_entry(n, struct sp_node, nd);
}

/* Insert a new shared policy into the list. */
/* Caller holds sp->lock */
static void sp_insert(struct shared_policy *sp, struct sp_node *new)
{
	struct rb_node **p = &sp->root.rb_node;
	struct rb_node *parent = NULL;
	struct sp_node *nd;

	while (*p) {
		parent = *p;
		nd = rb_entry(parent, struct sp_node, nd);
		if (new->start < nd->start)
			p = &(*p)->rb_left;
		else if (new->end > nd->end)
			p = &(*p)->rb_right;
		else
			BUG();
	}
	rb_link_node(&new->nd, parent, p);
	rb_insert_color(&new->nd, &sp->root);
1739
	pr_debug("inserting %lx-%lx: %d\n", new->start, new->end,
1740
		 new->policy ? new->policy->mode : 0);
L
Linus Torvalds 已提交
1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
}

/* Find shared policy intersecting idx */
struct mempolicy *
mpol_shared_policy_lookup(struct shared_policy *sp, unsigned long idx)
{
	struct mempolicy *pol = NULL;
	struct sp_node *sn;

	if (!sp->root.rb_node)
		return NULL;
	spin_lock(&sp->lock);
	sn = sp_lookup(sp, idx, idx+1);
	if (sn) {
		mpol_get(sn->policy);
		pol = sn->policy;
	}
	spin_unlock(&sp->lock);
	return pol;
}

static void sp_delete(struct shared_policy *sp, struct sp_node *n)
{
1764
	pr_debug("deleting %lx-l%lx\n", n->start, n->end);
L
Linus Torvalds 已提交
1765
	rb_erase(&n->nd, &sp->root);
1766
	mpol_put(n->policy);
L
Linus Torvalds 已提交
1767 1768 1769
	kmem_cache_free(sn_cache, n);
}

A
Adrian Bunk 已提交
1770 1771
static struct sp_node *sp_alloc(unsigned long start, unsigned long end,
				struct mempolicy *pol)
L
Linus Torvalds 已提交
1772 1773 1774 1775 1776 1777 1778 1779
{
	struct sp_node *n = kmem_cache_alloc(sn_cache, GFP_KERNEL);

	if (!n)
		return NULL;
	n->start = start;
	n->end = end;
	mpol_get(pol);
1780
	pol->flags |= MPOL_F_SHARED;	/* for unref */
L
Linus Torvalds 已提交
1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
	n->policy = pol;
	return n;
}

/* Replace a policy range. */
static int shared_policy_replace(struct shared_policy *sp, unsigned long start,
				 unsigned long end, struct sp_node *new)
{
	struct sp_node *n, *new2 = NULL;

restart:
	spin_lock(&sp->lock);
	n = sp_lookup(sp, start, end);
	/* Take care of old policies in the same range. */
	while (n && n->start < end) {
		struct rb_node *next = rb_next(&n->nd);
		if (n->start >= start) {
			if (n->end <= end)
				sp_delete(sp, n);
			else
				n->start = end;
		} else {
			/* Old policy spanning whole new range. */
			if (n->end > end) {
				if (!new2) {
					spin_unlock(&sp->lock);
					new2 = sp_alloc(end, n->end, n->policy);
					if (!new2)
						return -ENOMEM;
					goto restart;
				}
				n->end = start;
				sp_insert(sp, new2);
				new2 = NULL;
				break;
			} else
				n->end = start;
		}
		if (!next)
			break;
		n = rb_entry(next, struct sp_node, nd);
	}
	if (new)
		sp_insert(sp, new);
	spin_unlock(&sp->lock);
	if (new2) {
1827
		mpol_put(new2->policy);
L
Linus Torvalds 已提交
1828 1829 1830 1831 1832
		kmem_cache_free(sn_cache, new2);
	}
	return 0;
}

1833
void mpol_shared_policy_init(struct shared_policy *info, unsigned short policy,
1834
			unsigned short flags, nodemask_t *policy_nodes)
1835 1836 1837 1838 1839 1840 1841
{
	info->root = RB_ROOT;
	spin_lock_init(&info->lock);

	if (policy != MPOL_DEFAULT) {
		struct mempolicy *newpol;

1842
		/* Falls back to NULL policy [MPOL_DEFAULT] on any error */
1843
		newpol = mpol_new(policy, flags, policy_nodes);
1844 1845 1846 1847 1848 1849 1850 1851
		if (!IS_ERR(newpol)) {
			/* Create pseudo-vma that contains just the policy */
			struct vm_area_struct pvma;

			memset(&pvma, 0, sizeof(struct vm_area_struct));
			/* Policy covers entire file */
			pvma.vm_end = TASK_SIZE;
			mpol_set_shared_policy(info, &pvma, newpol);
1852
			mpol_put(newpol);
1853 1854 1855 1856
		}
	}
}

L
Linus Torvalds 已提交
1857 1858 1859 1860 1861 1862 1863
int mpol_set_shared_policy(struct shared_policy *info,
			struct vm_area_struct *vma, struct mempolicy *npol)
{
	int err;
	struct sp_node *new = NULL;
	unsigned long sz = vma_pages(vma);

1864
	pr_debug("set_shared_policy %lx sz %lu %d %d %lx\n",
L
Linus Torvalds 已提交
1865
		 vma->vm_pgoff,
1866
		 sz, npol ? npol->mode : -1,
1867
		 npol ? npol->flags : -1,
1868
		 npol ? nodes_addr(npol->v.nodes)[0] : -1);
L
Linus Torvalds 已提交
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893

	if (npol) {
		new = sp_alloc(vma->vm_pgoff, vma->vm_pgoff + sz, npol);
		if (!new)
			return -ENOMEM;
	}
	err = shared_policy_replace(info, vma->vm_pgoff, vma->vm_pgoff+sz, new);
	if (err && new)
		kmem_cache_free(sn_cache, new);
	return err;
}

/* Free a backing policy store on inode delete. */
void mpol_free_shared_policy(struct shared_policy *p)
{
	struct sp_node *n;
	struct rb_node *next;

	if (!p->root.rb_node)
		return;
	spin_lock(&p->lock);
	next = rb_first(&p->root);
	while (next) {
		n = rb_entry(next, struct sp_node, nd);
		next = rb_next(&n->nd);
1894
		rb_erase(&n->nd, &p->root);
1895
		mpol_put(n->policy);
L
Linus Torvalds 已提交
1896 1897 1898 1899 1900 1901 1902 1903
		kmem_cache_free(sn_cache, n);
	}
	spin_unlock(&p->lock);
}

/* assumes fs == KERNEL_DS */
void __init numa_policy_init(void)
{
1904 1905 1906 1907
	nodemask_t interleave_nodes;
	unsigned long largest = 0;
	int nid, prefer = 0;

L
Linus Torvalds 已提交
1908 1909
	policy_cache = kmem_cache_create("numa_policy",
					 sizeof(struct mempolicy),
1910
					 0, SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
1911 1912 1913

	sn_cache = kmem_cache_create("shared_policy_node",
				     sizeof(struct sp_node),
1914
				     0, SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
1915

1916 1917 1918 1919 1920 1921
	/*
	 * Set interleaving policy for system init. Interleaving is only
	 * enabled across suitably sized nodes (default is >= 16MB), or
	 * fall back to the largest node if they're all smaller.
	 */
	nodes_clear(interleave_nodes);
1922
	for_each_node_state(nid, N_HIGH_MEMORY) {
1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938
		unsigned long total_pages = node_present_pages(nid);

		/* Preserve the largest node */
		if (largest < total_pages) {
			largest = total_pages;
			prefer = nid;
		}

		/* Interleave this node? */
		if ((total_pages << PAGE_SHIFT) >= (16 << 20))
			node_set(nid, interleave_nodes);
	}

	/* All too small, use the largest */
	if (unlikely(nodes_empty(interleave_nodes)))
		node_set(prefer, interleave_nodes);
L
Linus Torvalds 已提交
1939

1940
	if (do_set_mempolicy(MPOL_INTERLEAVE, 0, &interleave_nodes))
L
Linus Torvalds 已提交
1941 1942 1943
		printk("numa_policy_init: interleaving failed\n");
}

1944
/* Reset policy of current process to default */
L
Linus Torvalds 已提交
1945 1946
void numa_default_policy(void)
{
1947
	do_set_mempolicy(MPOL_DEFAULT, 0, NULL);
L
Linus Torvalds 已提交
1948
}
1949

1950 1951 1952
/*
 * Display pages allocated per node and memory policy via /proc.
 */
1953 1954
static const char * const policy_types[] =
	{ "default", "prefer", "bind", "interleave" };
1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965

/*
 * Convert a mempolicy into a string.
 * Returns the number of characters in buffer (if positive)
 * or an error (negative)
 */
static inline int mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
{
	char *p = buffer;
	int l;
	nodemask_t nodes;
1966
	unsigned short mode;
1967
	unsigned short flags = pol ? pol->flags : 0;
1968

1969 1970 1971 1972 1973
	if (!pol || pol == &default_policy)
		mode = MPOL_DEFAULT;
	else
		mode = pol->mode;

1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984
	switch (mode) {
	case MPOL_DEFAULT:
		nodes_clear(nodes);
		break;

	case MPOL_PREFERRED:
		nodes_clear(nodes);
		node_set(pol->v.preferred_node, nodes);
		break;

	case MPOL_BIND:
1985
		/* Fall through */
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
	case MPOL_INTERLEAVE:
		nodes = pol->v.nodes;
		break;

	default:
		BUG();
		return -EFAULT;
	}

	l = strlen(policy_types[mode]);
 	if (buffer + maxlen < p + l + 1)
 		return -ENOSPC;

	strcpy(p, policy_types[mode]);
	p += l;

2002 2003 2004 2005 2006 2007 2008 2009 2010
	if (flags) {
		int need_bar = 0;

		if (buffer + maxlen < p + 2)
			return -ENOSPC;
		*p++ = '=';

		if (flags & MPOL_F_STATIC_NODES)
			p += sprintf(p, "%sstatic", need_bar++ ? "|" : "");
2011 2012
		if (flags & MPOL_F_RELATIVE_NODES)
			p += sprintf(p, "%srelative", need_bar++ ? "|" : "");
2013 2014
	}

2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
	if (!nodes_empty(nodes)) {
		if (buffer + maxlen < p + 2)
			return -ENOSPC;
		*p++ = '=';
	 	p += nodelist_scnprintf(p, buffer + maxlen - p, nodes);
	}
	return p - buffer;
}

struct numa_maps {
	unsigned long pages;
	unsigned long anon;
C
Christoph Lameter 已提交
2027 2028
	unsigned long active;
	unsigned long writeback;
2029
	unsigned long mapcount_max;
C
Christoph Lameter 已提交
2030 2031
	unsigned long dirty;
	unsigned long swapcache;
2032 2033 2034
	unsigned long node[MAX_NUMNODES];
};

C
Christoph Lameter 已提交
2035
static void gather_stats(struct page *page, void *private, int pte_dirty)
2036 2037 2038 2039
{
	struct numa_maps *md = private;
	int count = page_mapcount(page);

C
Christoph Lameter 已提交
2040 2041 2042
	md->pages++;
	if (pte_dirty || PageDirty(page))
		md->dirty++;
2043

C
Christoph Lameter 已提交
2044 2045
	if (PageSwapCache(page))
		md->swapcache++;
2046

C
Christoph Lameter 已提交
2047 2048 2049 2050 2051
	if (PageActive(page))
		md->active++;

	if (PageWriteback(page))
		md->writeback++;
2052 2053 2054 2055

	if (PageAnon(page))
		md->anon++;

C
Christoph Lameter 已提交
2056 2057 2058
	if (count > md->mapcount_max)
		md->mapcount_max = count;

2059 2060 2061
	md->node[page_to_nid(page)]++;
}

A
Andrew Morton 已提交
2062
#ifdef CONFIG_HUGETLB_PAGE
C
Christoph Lameter 已提交
2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
static void check_huge_range(struct vm_area_struct *vma,
		unsigned long start, unsigned long end,
		struct numa_maps *md)
{
	unsigned long addr;
	struct page *page;

	for (addr = start; addr < end; addr += HPAGE_SIZE) {
		pte_t *ptep = huge_pte_offset(vma->vm_mm, addr & HPAGE_MASK);
		pte_t pte;

		if (!ptep)
			continue;

		pte = *ptep;
		if (pte_none(pte))
			continue;

		page = pte_page(pte);
		if (!page)
			continue;

		gather_stats(page, md, pte_dirty(*ptep));
	}
}
A
Andrew Morton 已提交
2088 2089 2090 2091 2092 2093 2094
#else
static inline void check_huge_range(struct vm_area_struct *vma,
		unsigned long start, unsigned long end,
		struct numa_maps *md)
{
}
#endif
C
Christoph Lameter 已提交
2095

2096 2097
int show_numa_map(struct seq_file *m, void *v)
{
2098
	struct proc_maps_private *priv = m->private;
2099 2100
	struct vm_area_struct *vma = v;
	struct numa_maps *md;
C
Christoph Lameter 已提交
2101 2102
	struct file *file = vma->vm_file;
	struct mm_struct *mm = vma->vm_mm;
2103
	struct mempolicy *pol;
2104 2105 2106
	int n;
	char buffer[50];

C
Christoph Lameter 已提交
2107
	if (!mm)
2108 2109 2110 2111 2112 2113
		return 0;

	md = kzalloc(sizeof(struct numa_maps), GFP_KERNEL);
	if (!md)
		return 0;

2114 2115
	pol = get_vma_policy(priv->task, vma, vma->vm_start);
	mpol_to_str(buffer, sizeof(buffer), pol);
2116
	mpol_cond_put(pol);
C
Christoph Lameter 已提交
2117 2118 2119 2120 2121

	seq_printf(m, "%08lx %s", vma->vm_start, buffer);

	if (file) {
		seq_printf(m, " file=");
2122
		seq_path(m, &file->f_path, "\n\t= ");
C
Christoph Lameter 已提交
2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133
	} else if (vma->vm_start <= mm->brk && vma->vm_end >= mm->start_brk) {
		seq_printf(m, " heap");
	} else if (vma->vm_start <= mm->start_stack &&
			vma->vm_end >= mm->start_stack) {
		seq_printf(m, " stack");
	}

	if (is_vm_hugetlb_page(vma)) {
		check_huge_range(vma, vma->vm_start, vma->vm_end, md);
		seq_printf(m, " huge");
	} else {
2134
		check_pgd_range(vma, vma->vm_start, vma->vm_end,
2135
			&node_states[N_HIGH_MEMORY], MPOL_MF_STATS, md);
C
Christoph Lameter 已提交
2136 2137 2138 2139
	}

	if (!md->pages)
		goto out;
2140

C
Christoph Lameter 已提交
2141 2142
	if (md->anon)
		seq_printf(m," anon=%lu",md->anon);
2143

C
Christoph Lameter 已提交
2144 2145
	if (md->dirty)
		seq_printf(m," dirty=%lu",md->dirty);
2146

C
Christoph Lameter 已提交
2147 2148
	if (md->pages != md->anon && md->pages != md->dirty)
		seq_printf(m, " mapped=%lu", md->pages);
2149

C
Christoph Lameter 已提交
2150 2151
	if (md->mapcount_max > 1)
		seq_printf(m, " mapmax=%lu", md->mapcount_max);
2152

C
Christoph Lameter 已提交
2153 2154 2155 2156 2157 2158 2159 2160 2161
	if (md->swapcache)
		seq_printf(m," swapcache=%lu", md->swapcache);

	if (md->active < md->pages && !is_vm_hugetlb_page(vma))
		seq_printf(m," active=%lu", md->active);

	if (md->writeback)
		seq_printf(m," writeback=%lu", md->writeback);

2162
	for_each_node_state(n, N_HIGH_MEMORY)
C
Christoph Lameter 已提交
2163 2164 2165 2166
		if (md->node[n])
			seq_printf(m, " N%d=%lu", n, md->node[n]);
out:
	seq_putc(m, '\n');
2167 2168 2169
	kfree(md);

	if (m->count < m->size)
2170
		m->version = (vma != priv->tail_vma) ? vma->vm_start : 0;
2171 2172
	return 0;
}