mempolicy.c 70.4 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
 * preferred       Try a specific node first before normal fallback.
D
David Rientjes 已提交
29
 *                As a special case NUMA_NO_NODE here means do the allocation
L
Linus Torvalds 已提交
30 31 32
 *                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
 * 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.
*/

68 69
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

L
Linus Torvalds 已提交
70 71 72 73 74 75 76 77 78 79
#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/slab.h>
#include <linux/string.h>
80
#include <linux/export.h>
81
#include <linux/nsproxy.h>
L
Linus Torvalds 已提交
82 83 84
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/compat.h>
85
#include <linux/swap.h>
86 87
#include <linux/seq_file.h>
#include <linux/proc_fs.h>
C
Christoph Lameter 已提交
88
#include <linux/migrate.h>
89
#include <linux/ksm.h>
90
#include <linux/rmap.h>
91
#include <linux/security.h>
A
Adrian Bunk 已提交
92
#include <linux/syscalls.h>
93
#include <linux/ctype.h>
94
#include <linux/mm_inline.h>
L
Lee Schermerhorn 已提交
95
#include <linux/mmu_notifier.h>
96
#include <linux/printk.h>
97

L
Linus Torvalds 已提交
98 99
#include <asm/tlbflush.h>
#include <asm/uaccess.h>
100
#include <linux/random.h>
L
Linus Torvalds 已提交
101

102 103
#include "internal.h"

104
/* Internal flags */
105
#define MPOL_MF_DISCONTIG_OK (MPOL_MF_INTERNAL << 0)	/* Skip checks for continuous vmas */
106
#define MPOL_MF_INVERT (MPOL_MF_INTERNAL << 1)		/* Invert check for nodemask */
107

108 109
static struct kmem_cache *policy_cache;
static struct kmem_cache *sn_cache;
L
Linus Torvalds 已提交
110 111 112

/* Highest zone. An specific allocation for a zone below that is not
   policied. */
113
enum zone_type policy_zone = 0;
L
Linus Torvalds 已提交
114

115 116 117
/*
 * run-time system-wide default policy => local allocation
 */
118
static struct mempolicy default_policy = {
L
Linus Torvalds 已提交
119
	.refcnt = ATOMIC_INIT(1), /* never free it */
120
	.mode = MPOL_PREFERRED,
121
	.flags = MPOL_F_LOCAL,
L
Linus Torvalds 已提交
122 123
};

M
Mel Gorman 已提交
124 125
static struct mempolicy preferred_node_policy[MAX_NUMNODES];

126
struct mempolicy *get_task_policy(struct task_struct *p)
M
Mel Gorman 已提交
127 128
{
	struct mempolicy *pol = p->mempolicy;
129
	int node;
M
Mel Gorman 已提交
130

131 132
	if (pol)
		return pol;
M
Mel Gorman 已提交
133

134 135 136 137 138 139
	node = numa_node_id();
	if (node != NUMA_NO_NODE) {
		pol = &preferred_node_policy[node];
		/* preferred_node_policy is not initialised early in boot */
		if (pol->mode)
			return pol;
M
Mel Gorman 已提交
140 141
	}

142
	return &default_policy;
M
Mel Gorman 已提交
143 144
}

145 146
static const struct mempolicy_operations {
	int (*create)(struct mempolicy *pol, const nodemask_t *nodes);
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
	/*
	 * If read-side task has no lock to protect task->mempolicy, write-side
	 * task will rebind the task->mempolicy by two step. The first step is
	 * setting all the newly nodes, and the second step is cleaning all the
	 * disallowed nodes. In this way, we can avoid finding no node to alloc
	 * page.
	 * If we have a lock to protect task->mempolicy in read-side, we do
	 * rebind directly.
	 *
	 * step:
	 * 	MPOL_REBIND_ONCE - do rebind work at once
	 * 	MPOL_REBIND_STEP1 - set all the newly nodes
	 * 	MPOL_REBIND_STEP2 - clean all the disallowed nodes
	 */
	void (*rebind)(struct mempolicy *pol, const nodemask_t *nodes,
			enum mpol_rebind_step step);
163 164
} mpol_ops[MPOL_MAX];

165 166
static inline int mpol_store_user_nodemask(const struct mempolicy *pol)
{
B
Bob Liu 已提交
167
	return pol->flags & MPOL_MODE_FLAGS;
168 169 170 171 172 173 174 175
}

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);
176 177
}

178 179 180 181 182 183 184 185 186 187 188
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)
189
		pol->flags |= MPOL_F_LOCAL;	/* local allocation */
190 191 192 193 194 195 196 197 198
	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)
{
199
	if (nodes_empty(*nodes))
200 201 202 203 204
		return -EINVAL;
	pol->v.nodes = *nodes;
	return 0;
}

205 206 207 208 209 210 211 212 213
/*
 * mpol_set_nodemask is called after mpol_new() to set up the nodemask, if
 * any, for the new policy.  mpol_new() has already validated the nodes
 * parameter with respect to the policy mode and flags.  But, we need to
 * handle an empty nodemask with MPOL_PREFERRED here.
 *
 * Must be called holding task's alloc_lock to protect task's mems_allowed
 * and mempolicy.  May also be called holding the mmap_semaphore for write.
 */
214 215
static int mpol_set_nodemask(struct mempolicy *pol,
		     const nodemask_t *nodes, struct nodemask_scratch *nsc)
216 217 218 219 220 221
{
	int ret;

	/* if mode is MPOL_DEFAULT, pol is NULL. This is right. */
	if (pol == NULL)
		return 0;
222
	/* Check N_MEMORY */
223
	nodes_and(nsc->mask1,
224
		  cpuset_current_mems_allowed, node_states[N_MEMORY]);
225 226 227 228 229 230

	VM_BUG_ON(!nodes);
	if (pol->mode == MPOL_PREFERRED && nodes_empty(*nodes))
		nodes = NULL;	/* explicit local allocation */
	else {
		if (pol->flags & MPOL_F_RELATIVE_NODES)
231
			mpol_relative_nodemask(&nsc->mask2, nodes, &nsc->mask1);
232
		else
233 234
			nodes_and(nsc->mask2, *nodes, nsc->mask1);

235 236 237 238 239 240 241
		if (mpol_store_user_nodemask(pol))
			pol->w.user_nodemask = *nodes;
		else
			pol->w.cpuset_mems_allowed =
						cpuset_current_mems_allowed;
	}

242 243 244 245
	if (nodes)
		ret = mpol_ops[pol->mode].create(pol, &nsc->mask2);
	else
		ret = mpol_ops[pol->mode].create(pol, NULL);
246 247 248 249 250 251 252
	return ret;
}

/*
 * This function just creates a new policy, does some check and simple
 * initialization. You must invoke mpol_set_nodemask() to set nodes.
 */
253 254
static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
				  nodemask_t *nodes)
L
Linus Torvalds 已提交
255 256 257
{
	struct mempolicy *policy;

258
	pr_debug("setting mode %d flags %d nodes[0] %lx\n",
D
David Rientjes 已提交
259
		 mode, flags, nodes ? nodes_addr(*nodes)[0] : NUMA_NO_NODE);
260

261 262
	if (mode == MPOL_DEFAULT) {
		if (nodes && !nodes_empty(*nodes))
263
			return ERR_PTR(-EINVAL);
L
Lee Schermerhorn 已提交
264
		return NULL;
265
	}
266 267 268 269 270 271 272 273 274 275 276 277 278
	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);
		}
279 280 281 282
	} else if (mode == MPOL_LOCAL) {
		if (!nodes_empty(*nodes))
			return ERR_PTR(-EINVAL);
		mode = MPOL_PREFERRED;
283 284
	} else if (nodes_empty(*nodes))
		return ERR_PTR(-EINVAL);
L
Linus Torvalds 已提交
285 286 287 288
	policy = kmem_cache_alloc(policy_cache, GFP_KERNEL);
	if (!policy)
		return ERR_PTR(-ENOMEM);
	atomic_set(&policy->refcnt, 1);
289
	policy->mode = mode;
290
	policy->flags = flags;
291

L
Linus Torvalds 已提交
292
	return policy;
293 294
}

295 296 297 298 299 300 301 302
/* 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);
}

303 304
static void mpol_rebind_default(struct mempolicy *pol, const nodemask_t *nodes,
				enum mpol_rebind_step step)
305 306 307
{
}

308 309 310 311 312 313 314 315
/*
 * step:
 * 	MPOL_REBIND_ONCE  - do rebind work at once
 * 	MPOL_REBIND_STEP1 - set all the newly nodes
 * 	MPOL_REBIND_STEP2 - clean all the disallowed nodes
 */
static void mpol_rebind_nodemask(struct mempolicy *pol, const nodemask_t *nodes,
				 enum mpol_rebind_step step)
316 317 318 319 320 321 322 323
{
	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 {
324 325 326 327 328 329 330 331 332 333 334 335 336
		/*
		 * if step == 1, we use ->w.cpuset_mems_allowed to cache the
		 * result
		 */
		if (step == MPOL_REBIND_ONCE || step == MPOL_REBIND_STEP1) {
			nodes_remap(tmp, pol->v.nodes,
					pol->w.cpuset_mems_allowed, *nodes);
			pol->w.cpuset_mems_allowed = step ? tmp : *nodes;
		} else if (step == MPOL_REBIND_STEP2) {
			tmp = pol->w.cpuset_mems_allowed;
			pol->w.cpuset_mems_allowed = *nodes;
		} else
			BUG();
337
	}
338

339 340 341 342 343 344 345 346 347 348
	if (nodes_empty(tmp))
		tmp = *nodes;

	if (step == MPOL_REBIND_STEP1)
		nodes_or(pol->v.nodes, pol->v.nodes, tmp);
	else if (step == MPOL_REBIND_ONCE || step == MPOL_REBIND_STEP2)
		pol->v.nodes = tmp;
	else
		BUG();

349 350 351 352 353 354 355 356 357 358
	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,
359 360
				  const nodemask_t *nodes,
				  enum mpol_rebind_step step)
361 362 363 364 365 366
{
	nodemask_t tmp;

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

367
		if (node_isset(node, *nodes)) {
368
			pol->v.preferred_node = node;
369 370 371
			pol->flags &= ~MPOL_F_LOCAL;
		} else
			pol->flags |= MPOL_F_LOCAL;
372 373 374
	} else if (pol->flags & MPOL_F_RELATIVE_NODES) {
		mpol_relative_nodemask(&tmp, &pol->w.user_nodemask, nodes);
		pol->v.preferred_node = first_node(tmp);
375
	} else if (!(pol->flags & MPOL_F_LOCAL)) {
376 377 378 379 380
		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
381 382
}

383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
/*
 * mpol_rebind_policy - Migrate a policy to a different set of nodes
 *
 * If read-side task has no lock to protect task->mempolicy, write-side
 * task will rebind the task->mempolicy by two step. The first step is
 * setting all the newly nodes, and the second step is cleaning all the
 * disallowed nodes. In this way, we can avoid finding no node to alloc
 * page.
 * If we have a lock to protect task->mempolicy in read-side, we do
 * rebind directly.
 *
 * step:
 * 	MPOL_REBIND_ONCE  - do rebind work at once
 * 	MPOL_REBIND_STEP1 - set all the newly nodes
 * 	MPOL_REBIND_STEP2 - clean all the disallowed nodes
 */
static void mpol_rebind_policy(struct mempolicy *pol, const nodemask_t *newmask,
				enum mpol_rebind_step step)
401 402 403
{
	if (!pol)
		return;
404
	if (!mpol_store_user_nodemask(pol) && step == MPOL_REBIND_ONCE &&
405 406
	    nodes_equal(pol->w.cpuset_mems_allowed, *newmask))
		return;
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421

	if (step == MPOL_REBIND_STEP1 && (pol->flags & MPOL_F_REBINDING))
		return;

	if (step == MPOL_REBIND_STEP2 && !(pol->flags & MPOL_F_REBINDING))
		BUG();

	if (step == MPOL_REBIND_STEP1)
		pol->flags |= MPOL_F_REBINDING;
	else if (step == MPOL_REBIND_STEP2)
		pol->flags &= ~MPOL_F_REBINDING;
	else if (step >= MPOL_REBIND_NSTEP)
		BUG();

	mpol_ops[pol->mode].rebind(pol, newmask, step);
422 423 424 425 426
}

/*
 * Wrapper for mpol_rebind_policy() that just requires task
 * pointer, and updates task mempolicy.
427 428
 *
 * Called with task's alloc_lock held.
429 430
 */

431 432
void mpol_rebind_task(struct task_struct *tsk, const nodemask_t *new,
			enum mpol_rebind_step step)
433
{
434
	mpol_rebind_policy(tsk->mempolicy, new, step);
435 436 437 438 439 440 441 442 443 444 445 446 447 448
}

/*
 * 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)
449
		mpol_rebind_policy(vma->vm_policy, new, MPOL_REBIND_ONCE);
450 451 452
	up_write(&mm->mmap_sem);
}

453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470
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,
	},
};

471 472
static void migrate_page_add(struct page *page, struct list_head *pagelist,
				unsigned long flags);
473

474 475 476 477 478 479 480
struct queue_pages {
	struct list_head *pagelist;
	unsigned long flags;
	nodemask_t *nmask;
	struct vm_area_struct *prev;
};

481 482 483 484
/*
 * Scan through pages checking if pages follow certain conditions,
 * and move them to the pagelist if they do.
 */
485 486
static int queue_pages_pte_range(pmd_t *pmd, unsigned long addr,
			unsigned long end, struct mm_walk *walk)
L
Linus Torvalds 已提交
487
{
488 489 490 491 492
	struct vm_area_struct *vma = walk->vma;
	struct page *page;
	struct queue_pages *qp = walk->private;
	unsigned long flags = qp->flags;
	int nid;
493
	pte_t *pte;
494
	spinlock_t *ptl;
495

496 497 498
	split_huge_page_pmd(vma, addr, pmd);
	if (pmd_trans_unstable(pmd))
		return 0;
499

500 501
	pte = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
	for (; addr != end; pte++, addr += PAGE_SIZE) {
502
		if (!pte_present(*pte))
L
Linus Torvalds 已提交
503
			continue;
504 505
		page = vm_normal_page(vma, addr, *pte);
		if (!page)
L
Linus Torvalds 已提交
506
			continue;
507
		/*
508 509
		 * vm_normal_page() filters out zero pages, but there might
		 * still be PageReserved pages to skip, perhaps in a VDSO.
510
		 */
H
Hugh Dickins 已提交
511
		if (PageReserved(page))
512
			continue;
513
		nid = page_to_nid(page);
514
		if (node_isset(nid, *qp->nmask) == !!(flags & MPOL_MF_INVERT))
515 516
			continue;

S
Stephen Wilson 已提交
517
		if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
518 519 520 521 522
			migrate_page_add(page, qp->pagelist, flags);
	}
	pte_unmap_unlock(pte - 1, ptl);
	cond_resched();
	return 0;
523 524
}

525 526 527
static int queue_pages_hugetlb(pte_t *pte, unsigned long hmask,
			       unsigned long addr, unsigned long end,
			       struct mm_walk *walk)
528 529
{
#ifdef CONFIG_HUGETLB_PAGE
530 531
	struct queue_pages *qp = walk->private;
	unsigned long flags = qp->flags;
532 533
	int nid;
	struct page *page;
534
	spinlock_t *ptl;
535
	pte_t entry;
536

537 538
	ptl = huge_pte_lock(hstate_vma(walk->vma), walk->mm, pte);
	entry = huge_ptep_get(pte);
539 540 541
	if (!pte_present(entry))
		goto unlock;
	page = pte_page(entry);
542
	nid = page_to_nid(page);
543
	if (node_isset(nid, *qp->nmask) == !!(flags & MPOL_MF_INVERT))
544 545 546 547
		goto unlock;
	/* With MPOL_MF_MOVE, we migrate only unshared hugepage. */
	if (flags & (MPOL_MF_MOVE_ALL) ||
	    (flags & MPOL_MF_MOVE && page_mapcount(page) == 1))
548
		isolate_huge_page(page, qp->pagelist);
549
unlock:
550
	spin_unlock(ptl);
551 552 553
#else
	BUG();
#endif
554
	return 0;
L
Linus Torvalds 已提交
555 556
}

557
#ifdef CONFIG_NUMA_BALANCING
L
Lee Schermerhorn 已提交
558
/*
559 560 561 562 563 564 565
 * This is used to mark a range of virtual addresses to be inaccessible.
 * These are later cleared by a NUMA hinting fault. Depending on these
 * faults, pages may be migrated for better NUMA placement.
 *
 * This is assuming that NUMA faults are handled using PROT_NONE. If
 * an architecture makes a different choice, it will need further
 * changes to the core.
L
Lee Schermerhorn 已提交
566
 */
567 568
unsigned long change_prot_numa(struct vm_area_struct *vma,
			unsigned long addr, unsigned long end)
L
Lee Schermerhorn 已提交
569
{
570
	int nr_updated;
L
Lee Schermerhorn 已提交
571

572
	nr_updated = change_protection(vma, addr, end, PAGE_NONE, 0, 1);
573 574
	if (nr_updated)
		count_vm_numa_events(NUMA_PTE_UPDATES, nr_updated);
L
Lee Schermerhorn 已提交
575

576
	return nr_updated;
L
Lee Schermerhorn 已提交
577 578 579 580 581 582 583
}
#else
static unsigned long change_prot_numa(struct vm_area_struct *vma,
			unsigned long addr, unsigned long end)
{
	return 0;
}
584
#endif /* CONFIG_NUMA_BALANCING */
L
Lee Schermerhorn 已提交
585

586 587 588 589 590 591 592 593
static int queue_pages_test_walk(unsigned long start, unsigned long end,
				struct mm_walk *walk)
{
	struct vm_area_struct *vma = walk->vma;
	struct queue_pages *qp = walk->private;
	unsigned long endvma = vma->vm_end;
	unsigned long flags = qp->flags;

594 595 596
	if (vma->vm_flags & VM_PFNMAP)
		return 1;

597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628
	if (endvma > end)
		endvma = end;
	if (vma->vm_start > start)
		start = vma->vm_start;

	if (!(flags & MPOL_MF_DISCONTIG_OK)) {
		if (!vma->vm_next && vma->vm_end < end)
			return -EFAULT;
		if (qp->prev && qp->prev->vm_end < vma->vm_start)
			return -EFAULT;
	}

	qp->prev = vma;

	if (vma->vm_flags & VM_PFNMAP)
		return 1;

	if (flags & MPOL_MF_LAZY) {
		/* Similar to task_numa_work, skip inaccessible VMAs */
		if (vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE))
			change_prot_numa(vma, start, endvma);
		return 1;
	}

	if ((flags & MPOL_MF_STRICT) ||
	    ((flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) &&
	     vma_migratable(vma)))
		/* queue pages from current vma */
		return 0;
	return 1;
}

629
/*
630 631 632 633 634
 * Walk through page tables and collect pages to be migrated.
 *
 * If pages found in a given range are on a set of nodes (determined by
 * @nodes and @flags,) it's isolated and queued to the pagelist which is
 * passed via @private.)
635
 */
636
static int
637
queue_pages_range(struct mm_struct *mm, unsigned long start, unsigned long end,
638 639
		nodemask_t *nodes, unsigned long flags,
		struct list_head *pagelist)
L
Linus Torvalds 已提交
640
{
641 642 643 644 645 646 647 648 649 650 651 652 653 654 655
	struct queue_pages qp = {
		.pagelist = pagelist,
		.flags = flags,
		.nmask = nodes,
		.prev = NULL,
	};
	struct mm_walk queue_pages_walk = {
		.hugetlb_entry = queue_pages_hugetlb,
		.pmd_entry = queue_pages_pte_range,
		.test_walk = queue_pages_test_walk,
		.mm = mm,
		.private = &qp,
	};

	return walk_page_range(start, end, &queue_pages_walk);
L
Linus Torvalds 已提交
656 657
}

658 659 660 661 662 663
/*
 * Apply policy to a single VMA
 * This must be called with the mmap_sem held for writing.
 */
static int vma_replace_policy(struct vm_area_struct *vma,
						struct mempolicy *pol)
664
{
665 666 667
	int err;
	struct mempolicy *old;
	struct mempolicy *new;
668 669 670 671 672 673

	pr_debug("vma %lx-%lx/%lx vm_ops %p vm_file %p set_policy %p\n",
		 vma->vm_start, vma->vm_end, vma->vm_pgoff,
		 vma->vm_ops, vma->vm_file,
		 vma->vm_ops ? vma->vm_ops->set_policy : NULL);

674 675 676 677 678
	new = mpol_dup(pol);
	if (IS_ERR(new))
		return PTR_ERR(new);

	if (vma->vm_ops && vma->vm_ops->set_policy) {
679
		err = vma->vm_ops->set_policy(vma, new);
680 681
		if (err)
			goto err_out;
682
	}
683 684 685 686 687 688 689 690

	old = vma->vm_policy;
	vma->vm_policy = new; /* protected by mmap_sem */
	mpol_put(old);

	return 0;
 err_out:
	mpol_put(new);
691 692 693
	return err;
}

L
Linus Torvalds 已提交
694
/* Step 2: apply policy to a range and do splits. */
695 696
static int mbind_range(struct mm_struct *mm, unsigned long start,
		       unsigned long end, struct mempolicy *new_pol)
L
Linus Torvalds 已提交
697 698
{
	struct vm_area_struct *next;
699 700 701
	struct vm_area_struct *prev;
	struct vm_area_struct *vma;
	int err = 0;
702
	pgoff_t pgoff;
703 704
	unsigned long vmstart;
	unsigned long vmend;
L
Linus Torvalds 已提交
705

706
	vma = find_vma(mm, start);
707 708 709
	if (!vma || vma->vm_start > start)
		return -EFAULT;

710
	prev = vma->vm_prev;
711 712 713
	if (start > vma->vm_start)
		prev = vma;

714
	for (; vma && vma->vm_start < end; prev = vma, vma = next) {
L
Linus Torvalds 已提交
715
		next = vma->vm_next;
716 717 718
		vmstart = max(start, vma->vm_start);
		vmend   = min(end, vma->vm_end);

719 720 721 722 723
		if (mpol_equal(vma_policy(vma), new_pol))
			continue;

		pgoff = vma->vm_pgoff +
			((vmstart - vma->vm_start) >> PAGE_SHIFT);
724
		prev = vma_merge(mm, prev, vmstart, vmend, vma->vm_flags,
725
				  vma->anon_vma, vma->vm_file, pgoff,
726
				  new_pol);
727 728 729
		if (prev) {
			vma = prev;
			next = vma->vm_next;
730 731 732 733
			if (mpol_equal(vma_policy(vma), new_pol))
				continue;
			/* vma_merge() joined vma && vma->next, case 8 */
			goto replace;
734 735 736 737 738 739 740 741 742 743 744
		}
		if (vma->vm_start != vmstart) {
			err = split_vma(vma->vm_mm, vma, vmstart, 1);
			if (err)
				goto out;
		}
		if (vma->vm_end != vmend) {
			err = split_vma(vma->vm_mm, vma, vmend, 0);
			if (err)
				goto out;
		}
745
 replace:
746
		err = vma_replace_policy(vma, new_pol);
747 748
		if (err)
			goto out;
L
Linus Torvalds 已提交
749
	}
750 751

 out:
L
Linus Torvalds 已提交
752 753 754 755
	return err;
}

/* Set the process memory policy */
756 757
static long do_set_mempolicy(unsigned short mode, unsigned short flags,
			     nodemask_t *nodes)
L
Linus Torvalds 已提交
758
{
759
	struct mempolicy *new, *old;
760
	NODEMASK_SCRATCH(scratch);
761
	int ret;
L
Linus Torvalds 已提交
762

763 764
	if (!scratch)
		return -ENOMEM;
765

766 767 768 769 770
	new = mpol_new(mode, flags, nodes);
	if (IS_ERR(new)) {
		ret = PTR_ERR(new);
		goto out;
	}
771

772
	task_lock(current);
773
	ret = mpol_set_nodemask(new, nodes, scratch);
774 775 776
	if (ret) {
		task_unlock(current);
		mpol_put(new);
777
		goto out;
778 779
	}
	old = current->mempolicy;
L
Linus Torvalds 已提交
780
	current->mempolicy = new;
781
	if (new && new->mode == MPOL_INTERLEAVE &&
782
	    nodes_weight(new->v.nodes))
783
		current->il_next = first_node(new->v.nodes);
784 785
	task_unlock(current);
	mpol_put(old);
786 787 788 789
	ret = 0;
out:
	NODEMASK_SCRATCH_FREE(scratch);
	return ret;
L
Linus Torvalds 已提交
790 791
}

792 793
/*
 * Return nodemask for policy for get_mempolicy() query
794 795
 *
 * Called with task's alloc_lock held
796 797
 */
static void get_policy_nodemask(struct mempolicy *p, nodemask_t *nodes)
L
Linus Torvalds 已提交
798
{
799
	nodes_clear(*nodes);
800 801 802
	if (p == &default_policy)
		return;

803
	switch (p->mode) {
804 805
	case MPOL_BIND:
		/* Fall through */
L
Linus Torvalds 已提交
806
	case MPOL_INTERLEAVE:
807
		*nodes = p->v.nodes;
L
Linus Torvalds 已提交
808 809
		break;
	case MPOL_PREFERRED:
810
		if (!(p->flags & MPOL_F_LOCAL))
811
			node_set(p->v.preferred_node, *nodes);
812
		/* else return empty node mask for local allocation */
L
Linus Torvalds 已提交
813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832
		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
833 834
static long do_get_mempolicy(int *policy, nodemask_t *nmask,
			     unsigned long addr, unsigned long flags)
L
Linus Torvalds 已提交
835
{
836
	int err;
L
Linus Torvalds 已提交
837 838 839 840
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma = NULL;
	struct mempolicy *pol = current->mempolicy;

841 842
	if (flags &
		~(unsigned long)(MPOL_F_NODE|MPOL_F_ADDR|MPOL_F_MEMS_ALLOWED))
L
Linus Torvalds 已提交
843
		return -EINVAL;
844 845 846 847 848

	if (flags & MPOL_F_MEMS_ALLOWED) {
		if (flags & (MPOL_F_NODE|MPOL_F_ADDR))
			return -EINVAL;
		*policy = 0;	/* just so it's initialized */
849
		task_lock(current);
850
		*nmask  = cpuset_current_mems_allowed;
851
		task_unlock(current);
852 853 854
		return 0;
	}

L
Linus Torvalds 已提交
855
	if (flags & MPOL_F_ADDR) {
856 857 858 859 860
		/*
		 * 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
861 862 863 864 865 866 867 868 869 870 871 872 873 874
		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)
875
		pol = &default_policy;	/* indicates default behavior */
L
Linus Torvalds 已提交
876 877 878 879 880 881

	if (flags & MPOL_F_NODE) {
		if (flags & MPOL_F_ADDR) {
			err = lookup_node(mm, addr);
			if (err < 0)
				goto out;
882
			*policy = err;
L
Linus Torvalds 已提交
883
		} else if (pol == current->mempolicy &&
884
				pol->mode == MPOL_INTERLEAVE) {
885
			*policy = current->il_next;
L
Linus Torvalds 已提交
886 887 888 889
		} else {
			err = -EINVAL;
			goto out;
		}
890 891 892
	} else {
		*policy = pol == &default_policy ? MPOL_DEFAULT :
						pol->mode;
893 894 895 896 897
		/*
		 * Internal mempolicy flags must be masked off before exposing
		 * the policy to userspace.
		 */
		*policy |= (pol->flags & MPOL_MODE_FLAGS);
898
	}
L
Linus Torvalds 已提交
899 900 901 902 903 904 905

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

	err = 0;
906
	if (nmask) {
907 908 909 910 911 912 913
		if (mpol_store_user_nodemask(pol)) {
			*nmask = pol->w.user_nodemask;
		} else {
			task_lock(current);
			get_policy_nodemask(pol, nmask);
			task_unlock(current);
		}
914
	}
L
Linus Torvalds 已提交
915 916

 out:
917
	mpol_cond_put(pol);
L
Linus Torvalds 已提交
918 919 920 921 922
	if (vma)
		up_read(&current->mm->mmap_sem);
	return err;
}

C
Christoph Lameter 已提交
923
#ifdef CONFIG_MIGRATION
924 925 926
/*
 * page migration
 */
927 928
static void migrate_page_add(struct page *page, struct list_head *pagelist,
				unsigned long flags)
929 930
{
	/*
931
	 * Avoid migrating a page that is shared with others.
932
	 */
933 934 935
	if ((flags & MPOL_MF_MOVE_ALL) || page_mapcount(page) == 1) {
		if (!isolate_lru_page(page)) {
			list_add_tail(&page->lru, pagelist);
936 937
			inc_zone_page_state(page, NR_ISOLATED_ANON +
					    page_is_file_cache(page));
938 939
		}
	}
940
}
941

942
static struct page *new_node_page(struct page *page, unsigned long node, int **x)
943
{
944 945 946 947 948
	if (PageHuge(page))
		return alloc_huge_page_node(page_hstate(compound_head(page)),
					node);
	else
		return alloc_pages_exact_node(node, GFP_HIGHUSER_MOVABLE, 0);
949 950
}

951 952 953 954
/*
 * Migrate pages from one node to a target node.
 * Returns error or the number of pages not migrated.
 */
A
Adrian Bunk 已提交
955 956
static int migrate_to_node(struct mm_struct *mm, int source, int dest,
			   int flags)
957 958 959 960 961 962 963
{
	nodemask_t nmask;
	LIST_HEAD(pagelist);
	int err = 0;

	nodes_clear(nmask);
	node_set(source, nmask);
964

965 966 967 968 969 970
	/*
	 * This does not "check" the range but isolates all pages that
	 * need migration.  Between passing in the full user address
	 * space range and MPOL_MF_DISCONTIG_OK, this call can not fail.
	 */
	VM_BUG_ON(!(flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)));
971
	queue_pages_range(mm, mm->mmap->vm_start, mm->task_size, &nmask,
972 973
			flags | MPOL_MF_DISCONTIG_OK, &pagelist);

974
	if (!list_empty(&pagelist)) {
975
		err = migrate_pages(&pagelist, new_node_page, NULL, dest,
976
					MIGRATE_SYNC, MR_SYSCALL);
977
		if (err)
978
			putback_movable_pages(&pagelist);
979
	}
980

981
	return err;
982 983
}

984
/*
985 986
 * Move pages between the two nodesets so as to preserve the physical
 * layout as much as possible.
987 988 989
 *
 * Returns the number of page that could not be moved.
 */
990 991
int do_migrate_pages(struct mm_struct *mm, const nodemask_t *from,
		     const nodemask_t *to, int flags)
992
{
993
	int busy = 0;
994
	int err;
995
	nodemask_t tmp;
996

997 998 999 1000
	err = migrate_prep();
	if (err)
		return err;

1001
	down_read(&mm->mmap_sem);
1002

1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027
	/*
	 * 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.
1028
	 * Otherwise when we finish scanning from_tmp, we at least have the
1029 1030 1031 1032
	 * 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.
	 */
1033

1034
	tmp = *from;
1035 1036
	while (!nodes_empty(tmp)) {
		int s,d;
J
Jianguo Wu 已提交
1037
		int source = NUMA_NO_NODE;
1038 1039 1040
		int dest = 0;

		for_each_node_mask(s, tmp) {
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056

			/*
			 * do_migrate_pages() tries to maintain the relative
			 * node relationship of the pages established between
			 * threads and memory areas.
                         *
			 * However if the number of source nodes is not equal to
			 * the number of destination nodes we can not preserve
			 * this node relative relationship.  In that case, skip
			 * copying memory from a node that is in the destination
			 * mask.
			 *
			 * Example: [2,3,4] -> [3,4,5] moves everything.
			 *          [0-7] - > [3,4,5] moves only 0,1,2,6,7.
			 */

1057 1058
			if ((nodes_weight(*from) != nodes_weight(*to)) &&
						(node_isset(s, *to)))
1059 1060
				continue;

1061
			d = node_remap(s, *from, *to);
1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
			if (s == d)
				continue;

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

			/* dest not in remaining from nodes? */
			if (!node_isset(dest, tmp))
				break;
		}
J
Jianguo Wu 已提交
1072
		if (source == NUMA_NO_NODE)
1073 1074 1075 1076 1077 1078 1079 1080
			break;

		node_clear(source, tmp);
		err = migrate_to_node(mm, source, dest, flags);
		if (err > 0)
			busy += err;
		if (err < 0)
			break;
1081 1082
	}
	up_read(&mm->mmap_sem);
1083 1084 1085
	if (err < 0)
		return err;
	return busy;
C
Christoph Lameter 已提交
1086 1087 1088

}

1089 1090
/*
 * Allocate a new page for page migration based on vma policy.
1091
 * Start by assuming the page is mapped by the same vma as contains @start.
1092 1093 1094 1095
 * 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.
 */
1096
static struct page *new_page(struct page *page, unsigned long start, int **x)
1097
{
1098
	struct vm_area_struct *vma;
1099
	unsigned long uninitialized_var(address);
1100

1101
	vma = find_vma(current->mm, start);
1102 1103 1104 1105 1106 1107
	while (vma) {
		address = page_address_in_vma(page, vma);
		if (address != -EFAULT)
			break;
		vma = vma->vm_next;
	}
1108 1109

	if (PageHuge(page)) {
1110 1111
		BUG_ON(!vma);
		return alloc_huge_page_noerr(vma, address, 1);
1112
	}
1113
	/*
1114
	 * if !vma, alloc_page_vma() will use task or system default policy
1115
	 */
1116
	return alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
1117
}
C
Christoph Lameter 已提交
1118 1119 1120 1121 1122
#else

static void migrate_page_add(struct page *page, struct list_head *pagelist,
				unsigned long flags)
{
1123 1124
}

1125 1126
int do_migrate_pages(struct mm_struct *mm, const nodemask_t *from,
		     const nodemask_t *to, int flags)
C
Christoph Lameter 已提交
1127 1128 1129
{
	return -ENOSYS;
}
1130

1131
static struct page *new_page(struct page *page, unsigned long start, int **x)
1132 1133 1134
{
	return NULL;
}
C
Christoph Lameter 已提交
1135 1136
#endif

A
Adrian Bunk 已提交
1137
static long do_mbind(unsigned long start, unsigned long len,
1138 1139
		     unsigned short mode, unsigned short mode_flags,
		     nodemask_t *nmask, unsigned long flags)
1140 1141 1142 1143 1144 1145 1146
{
	struct mm_struct *mm = current->mm;
	struct mempolicy *new;
	unsigned long end;
	int err;
	LIST_HEAD(pagelist);

L
Lee Schermerhorn 已提交
1147
	if (flags & ~(unsigned long)MPOL_MF_VALID)
1148
		return -EINVAL;
1149
	if ((flags & MPOL_MF_MOVE_ALL) && !capable(CAP_SYS_NICE))
1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
		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;

1166
	new = mpol_new(mode, mode_flags, nmask);
1167 1168 1169
	if (IS_ERR(new))
		return PTR_ERR(new);

L
Lee Schermerhorn 已提交
1170 1171 1172
	if (flags & MPOL_MF_LAZY)
		new->flags |= MPOL_F_MOF;

1173 1174 1175 1176 1177 1178 1179
	/*
	 * If we are using the default policy then operation
	 * on discontinuous address spaces is okay after all
	 */
	if (!new)
		flags |= MPOL_MF_DISCONTIG_OK;

1180 1181
	pr_debug("mbind %lx-%lx mode:%d flags:%d nodes:%lx\n",
		 start, start + len, mode, mode_flags,
D
David Rientjes 已提交
1182
		 nmask ? nodes_addr(*nmask)[0] : NUMA_NO_NODE);
1183

1184 1185 1186 1187
	if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) {

		err = migrate_prep();
		if (err)
K
KOSAKI Motohiro 已提交
1188
			goto mpol_out;
1189
	}
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202
	{
		NODEMASK_SCRATCH(scratch);
		if (scratch) {
			down_write(&mm->mmap_sem);
			task_lock(current);
			err = mpol_set_nodemask(new, nmask, scratch);
			task_unlock(current);
			if (err)
				up_write(&mm->mmap_sem);
		} else
			err = -ENOMEM;
		NODEMASK_SCRATCH_FREE(scratch);
	}
K
KOSAKI Motohiro 已提交
1203 1204 1205
	if (err)
		goto mpol_out;

1206
	err = queue_pages_range(mm, start, end, nmask,
1207
			  flags | MPOL_MF_INVERT, &pagelist);
1208
	if (!err)
1209
		err = mbind_range(mm, start, end, new);
1210

L
Lee Schermerhorn 已提交
1211 1212 1213
	if (!err) {
		int nr_failed = 0;

1214
		if (!list_empty(&pagelist)) {
L
Lee Schermerhorn 已提交
1215
			WARN_ON_ONCE(flags & MPOL_MF_LAZY);
1216 1217
			nr_failed = migrate_pages(&pagelist, new_page, NULL,
				start, MIGRATE_SYNC, MR_MEMPOLICY_MBIND);
1218
			if (nr_failed)
1219
				putback_movable_pages(&pagelist);
1220
		}
1221

L
Lee Schermerhorn 已提交
1222
		if (nr_failed && (flags & MPOL_MF_STRICT))
1223
			err = -EIO;
1224
	} else
1225
		putback_movable_pages(&pagelist);
C
Christoph Lameter 已提交
1226

1227
	up_write(&mm->mmap_sem);
K
KOSAKI Motohiro 已提交
1228
 mpol_out:
1229
	mpol_put(new);
1230 1231 1232
	return err;
}

1233 1234 1235 1236 1237
/*
 * User space interface with variable sized bitmaps for nodelists.
 */

/* Copy a node mask from user space. */
1238
static int get_nodes(nodemask_t *nodes, const unsigned long __user *nmask,
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
		     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
1249
	if (maxnode > PAGE_SIZE*BITS_PER_BYTE)
C
Chris Wright 已提交
1250
		return -EINVAL;
1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299

	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;
}

1300
SYSCALL_DEFINE6(mbind, unsigned long, start, unsigned long, len,
1301
		unsigned long, mode, const unsigned long __user *, nmask,
1302
		unsigned long, maxnode, unsigned, flags)
1303 1304 1305
{
	nodemask_t nodes;
	int err;
1306
	unsigned short mode_flags;
1307

1308 1309
	mode_flags = mode & MPOL_MODE_FLAGS;
	mode &= ~MPOL_MODE_FLAGS;
1310 1311
	if (mode >= MPOL_MAX)
		return -EINVAL;
1312 1313 1314
	if ((mode_flags & MPOL_F_STATIC_NODES) &&
	    (mode_flags & MPOL_F_RELATIVE_NODES))
		return -EINVAL;
1315 1316 1317
	err = get_nodes(&nodes, nmask, maxnode);
	if (err)
		return err;
1318
	return do_mbind(start, len, mode, mode_flags, &nodes, flags);
1319 1320 1321
}

/* Set the process memory policy */
1322
SYSCALL_DEFINE3(set_mempolicy, int, mode, const unsigned long __user *, nmask,
1323
		unsigned long, maxnode)
1324 1325 1326
{
	int err;
	nodemask_t nodes;
1327
	unsigned short flags;
1328

1329 1330 1331
	flags = mode & MPOL_MODE_FLAGS;
	mode &= ~MPOL_MODE_FLAGS;
	if ((unsigned int)mode >= MPOL_MAX)
1332
		return -EINVAL;
1333 1334
	if ((flags & MPOL_F_STATIC_NODES) && (flags & MPOL_F_RELATIVE_NODES))
		return -EINVAL;
1335 1336 1337
	err = get_nodes(&nodes, nmask, maxnode);
	if (err)
		return err;
1338
	return do_set_mempolicy(mode, flags, &nodes);
1339 1340
}

1341 1342 1343
SYSCALL_DEFINE4(migrate_pages, pid_t, pid, unsigned long, maxnode,
		const unsigned long __user *, old_nodes,
		const unsigned long __user *, new_nodes)
1344
{
1345
	const struct cred *cred = current_cred(), *tcred;
1346
	struct mm_struct *mm = NULL;
1347 1348 1349
	struct task_struct *task;
	nodemask_t task_nodes;
	int err;
1350 1351 1352 1353 1354 1355
	nodemask_t *old;
	nodemask_t *new;
	NODEMASK_SCRATCH(scratch);

	if (!scratch)
		return -ENOMEM;
1356

1357 1358 1359 1360
	old = &scratch->mask1;
	new = &scratch->mask2;

	err = get_nodes(old, old_nodes, maxnode);
1361
	if (err)
1362
		goto out;
1363

1364
	err = get_nodes(new, new_nodes, maxnode);
1365
	if (err)
1366
		goto out;
1367 1368

	/* Find the mm_struct */
1369
	rcu_read_lock();
1370
	task = pid ? find_task_by_vpid(pid) : current;
1371
	if (!task) {
1372
		rcu_read_unlock();
1373 1374
		err = -ESRCH;
		goto out;
1375
	}
1376
	get_task_struct(task);
1377

1378
	err = -EINVAL;
1379 1380 1381 1382

	/*
	 * 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
1383
	 * capabilities, superuser privileges or the same
1384 1385
	 * userid as the target process.
	 */
1386
	tcred = __task_cred(task);
1387 1388
	if (!uid_eq(cred->euid, tcred->suid) && !uid_eq(cred->euid, tcred->uid) &&
	    !uid_eq(cred->uid,  tcred->suid) && !uid_eq(cred->uid,  tcred->uid) &&
1389
	    !capable(CAP_SYS_NICE)) {
1390
		rcu_read_unlock();
1391
		err = -EPERM;
1392
		goto out_put;
1393
	}
1394
	rcu_read_unlock();
1395 1396 1397

	task_nodes = cpuset_mems_allowed(task);
	/* Is the user allowed to access the target nodes? */
1398
	if (!nodes_subset(*new, task_nodes) && !capable(CAP_SYS_NICE)) {
1399
		err = -EPERM;
1400
		goto out_put;
1401 1402
	}

1403
	if (!nodes_subset(*new, node_states[N_MEMORY])) {
1404
		err = -EINVAL;
1405
		goto out_put;
1406 1407
	}

1408 1409
	err = security_task_movememory(task);
	if (err)
1410
		goto out_put;
1411

1412 1413
	mm = get_task_mm(task);
	put_task_struct(task);
1414 1415

	if (!mm) {
1416
		err = -EINVAL;
1417 1418 1419 1420 1421
		goto out;
	}

	err = do_migrate_pages(mm, old, new,
		capable(CAP_SYS_NICE) ? MPOL_MF_MOVE_ALL : MPOL_MF_MOVE);
1422 1423 1424

	mmput(mm);
out:
1425 1426
	NODEMASK_SCRATCH_FREE(scratch);

1427
	return err;
1428 1429 1430 1431 1432

out_put:
	put_task_struct(task);
	goto out;

1433 1434 1435
}


1436
/* Retrieve NUMA policy */
1437 1438 1439
SYSCALL_DEFINE5(get_mempolicy, int __user *, policy,
		unsigned long __user *, nmask, unsigned long, maxnode,
		unsigned long, addr, unsigned long, flags)
1440
{
A
Adrian Bunk 已提交
1441 1442
	int err;
	int uninitialized_var(pval);
1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461
	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
1462 1463
#ifdef CONFIG_COMPAT

1464 1465 1466 1467
COMPAT_SYSCALL_DEFINE5(get_mempolicy, int __user *, policy,
		       compat_ulong_t __user *, nmask,
		       compat_ulong_t, maxnode,
		       compat_ulong_t, addr, compat_ulong_t, flags)
L
Linus Torvalds 已提交
1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482
{
	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) {
1483 1484 1485
		unsigned long copy_size;
		copy_size = min_t(unsigned long, sizeof(bm), alloc_size);
		err = copy_from_user(bm, nm, copy_size);
L
Linus Torvalds 已提交
1486 1487 1488 1489 1490 1491 1492 1493
		/* ensure entire bitmap is zeroed */
		err |= clear_user(nmask, ALIGN(maxnode-1, 8) / 8);
		err |= compat_put_bitmap(nmask, bm, nr_bits);
	}

	return err;
}

1494 1495
COMPAT_SYSCALL_DEFINE3(set_mempolicy, int, mode, compat_ulong_t __user *, nmask,
		       compat_ulong_t, maxnode)
L
Linus Torvalds 已提交
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
{
	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);
}

1517 1518 1519
COMPAT_SYSCALL_DEFINE6(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)
L
Linus Torvalds 已提交
1520 1521 1522 1523
{
	long err = 0;
	unsigned long __user *nm = NULL;
	unsigned long nr_bits, alloc_size;
1524
	nodemask_t bm;
L
Linus Torvalds 已提交
1525 1526 1527 1528 1529

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

	if (nmask) {
1530
		err = compat_get_bitmap(nodes_addr(bm), nmask, nr_bits);
L
Linus Torvalds 已提交
1531
		nm = compat_alloc_user_space(alloc_size);
1532
		err |= copy_to_user(nm, nodes_addr(bm), alloc_size);
L
Linus Torvalds 已提交
1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
	}

	if (err)
		return -EFAULT;

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

#endif

1543 1544
struct mempolicy *__get_vma_policy(struct vm_area_struct *vma,
						unsigned long addr)
L
Linus Torvalds 已提交
1545
{
1546
	struct mempolicy *pol = NULL;
L
Linus Torvalds 已提交
1547 1548

	if (vma) {
1549
		if (vma->vm_ops && vma->vm_ops->get_policy) {
1550
			pol = vma->vm_ops->get_policy(vma, addr);
1551
		} else if (vma->vm_policy) {
L
Linus Torvalds 已提交
1552
			pol = vma->vm_policy;
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562

			/*
			 * shmem_alloc_page() passes MPOL_F_SHARED policy with
			 * a pseudo vma whose vma->vm_ops=NULL. Take a reference
			 * count on these policies which will be dropped by
			 * mpol_cond_put() later
			 */
			if (mpol_needs_cond_ref(pol))
				mpol_get(pol);
		}
L
Linus Torvalds 已提交
1563
	}
1564

1565 1566 1567 1568
	return pol;
}

/*
1569
 * get_vma_policy(@vma, @addr)
1570 1571 1572 1573
 * @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.
1574
 * Falls back to current->mempolicy or system default policy, as necessary.
1575 1576 1577 1578 1579
 * 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.
 */
1580 1581
static struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
						unsigned long addr)
1582 1583 1584
{
	struct mempolicy *pol = __get_vma_policy(vma, addr);

1585
	if (!pol)
1586
		pol = get_task_policy(current);
1587

L
Linus Torvalds 已提交
1588 1589 1590
	return pol;
}

1591
bool vma_policy_mof(struct vm_area_struct *vma)
1592
{
1593
	struct mempolicy *pol;
1594

1595 1596
	if (vma->vm_ops && vma->vm_ops->get_policy) {
		bool ret = false;
1597

1598 1599 1600 1601
		pol = vma->vm_ops->get_policy(vma, vma->vm_start);
		if (pol && (pol->flags & MPOL_F_MOF))
			ret = true;
		mpol_cond_put(pol);
1602

1603
		return ret;
1604 1605
	}

1606
	pol = vma->vm_policy;
1607
	if (!pol)
1608
		pol = get_task_policy(current);
1609

1610 1611 1612
	return pol->flags & MPOL_F_MOF;
}

1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632
static int apply_policy_zone(struct mempolicy *policy, enum zone_type zone)
{
	enum zone_type dynamic_policy_zone = policy_zone;

	BUG_ON(dynamic_policy_zone == ZONE_MOVABLE);

	/*
	 * if policy->v.nodes has movable memory only,
	 * we apply policy when gfp_zone(gfp) = ZONE_MOVABLE only.
	 *
	 * policy->v.nodes is intersect with node_states[N_MEMORY].
	 * so if the following test faile, it implies
	 * policy->v.nodes has movable memory only.
	 */
	if (!nodes_intersects(policy->v.nodes, node_states[N_HIGH_MEMORY]))
		dynamic_policy_zone = ZONE_MOVABLE;

	return zone >= dynamic_policy_zone;
}

1633 1634 1635 1636 1637
/*
 * Return a nodemask representing a mempolicy for filtering nodes for
 * page allocation
 */
static nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *policy)
1638 1639
{
	/* Lower zones don't get a nodemask applied for MPOL_BIND */
1640
	if (unlikely(policy->mode == MPOL_BIND) &&
1641
			apply_policy_zone(policy, gfp_zone(gfp)) &&
1642 1643 1644 1645 1646 1647
			cpuset_nodemask_valid_mems_allowed(&policy->v.nodes))
		return &policy->v.nodes;

	return NULL;
}

1648
/* Return a zonelist indicated by gfp for node representing a mempolicy */
1649 1650
static struct zonelist *policy_zonelist(gfp_t gfp, struct mempolicy *policy,
	int nd)
L
Linus Torvalds 已提交
1651
{
1652
	switch (policy->mode) {
L
Linus Torvalds 已提交
1653
	case MPOL_PREFERRED:
1654 1655
		if (!(policy->flags & MPOL_F_LOCAL))
			nd = policy->v.preferred_node;
L
Linus Torvalds 已提交
1656 1657
		break;
	case MPOL_BIND:
1658
		/*
1659 1660
		 * Normally, MPOL_BIND allocations are node-local within the
		 * allowed nodemask.  However, if __GFP_THISNODE is set and the
1661
		 * current node isn't part of the mask, we use the zonelist for
1662
		 * the first node in the mask instead.
1663 1664 1665 1666 1667
		 */
		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
1668 1669 1670
	default:
		BUG();
	}
1671
	return node_zonelist(nd, gfp);
L
Linus Torvalds 已提交
1672 1673 1674 1675 1676 1677 1678 1679 1680
}

/* 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;
1681
	next = next_node(nid, policy->v.nodes);
L
Linus Torvalds 已提交
1682
	if (next >= MAX_NUMNODES)
1683
		next = first_node(policy->v.nodes);
1684 1685
	if (next < MAX_NUMNODES)
		me->il_next = next;
L
Linus Torvalds 已提交
1686 1687 1688
	return nid;
}

1689 1690 1691 1692
/*
 * Depending on the memory policy provide a node from which to allocate the
 * next slab entry.
 */
1693
unsigned int mempolicy_slab_node(void)
1694
{
1695
	struct mempolicy *policy;
1696
	int node = numa_mem_id();
1697 1698

	if (in_interrupt())
1699
		return node;
1700 1701

	policy = current->mempolicy;
1702
	if (!policy || policy->flags & MPOL_F_LOCAL)
1703
		return node;
1704 1705 1706

	switch (policy->mode) {
	case MPOL_PREFERRED:
1707 1708 1709 1710
		/*
		 * handled MPOL_F_LOCAL above
		 */
		return policy->v.preferred_node;
1711

1712 1713 1714
	case MPOL_INTERLEAVE:
		return interleave_nodes(policy);

1715
	case MPOL_BIND: {
1716 1717 1718 1719
		/*
		 * Follow bind policy behavior and start allocation at the
		 * first node.
		 */
1720 1721 1722
		struct zonelist *zonelist;
		struct zone *zone;
		enum zone_type highest_zoneidx = gfp_zone(GFP_KERNEL);
1723
		zonelist = &NODE_DATA(node)->node_zonelists[0];
1724 1725 1726
		(void)first_zones_zonelist(zonelist, highest_zoneidx,
							&policy->v.nodes,
							&zone);
1727
		return zone ? zone->node : node;
1728
	}
1729 1730

	default:
1731
		BUG();
1732 1733 1734
	}
}

L
Linus Torvalds 已提交
1735 1736 1737 1738
/* 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)
{
1739
	unsigned nnodes = nodes_weight(pol->v.nodes);
1740
	unsigned target;
L
Linus Torvalds 已提交
1741
	int c;
J
Jianguo Wu 已提交
1742
	int nid = NUMA_NO_NODE;
L
Linus Torvalds 已提交
1743

1744 1745 1746
	if (!nnodes)
		return numa_node_id();
	target = (unsigned int)off % nnodes;
L
Linus Torvalds 已提交
1747 1748
	c = 0;
	do {
1749
		nid = next_node(nid, pol->v.nodes);
L
Linus Torvalds 已提交
1750 1751 1752 1753 1754
		c++;
	} while (c <= target);
	return nid;
}

1755 1756 1757 1758 1759 1760 1761
/* 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;

1762 1763 1764 1765 1766 1767 1768 1769 1770
		/*
		 * 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);
1771 1772 1773 1774 1775 1776
		off += (addr - vma->vm_start) >> shift;
		return offset_il_node(pol, vma, off);
	} else
		return interleave_nodes(pol);
}

1777 1778
/*
 * Return the bit number of a random bit set in the nodemask.
J
Jianguo Wu 已提交
1779
 * (returns NUMA_NO_NODE if nodemask is empty)
1780 1781 1782
 */
int node_random(const nodemask_t *maskp)
{
J
Jianguo Wu 已提交
1783
	int w, bit = NUMA_NO_NODE;
1784 1785 1786 1787 1788 1789 1790 1791

	w = nodes_weight(*maskp);
	if (w)
		bit = bitmap_ord_to_pos(maskp->bits,
			get_random_int() % w, MAX_NUMNODES);
	return bit;
}

1792
#ifdef CONFIG_HUGETLBFS
1793 1794
/*
 * huge_zonelist(@vma, @addr, @gfp_flags, @mpol)
1795 1796 1797 1798 1799
 * @vma: virtual memory area whose policy is sought
 * @addr: address in @vma for shared policy lookup and interleave policy
 * @gfp_flags: for requested zone
 * @mpol: pointer to mempolicy pointer for reference counted mempolicy
 * @nodemask: pointer to nodemask pointer for MPOL_BIND nodemask
1800
 *
1801 1802 1803 1804
 * 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.
1805
 *
1806
 * Must be protected by read_mems_allowed_begin()
1807
 */
1808
struct zonelist *huge_zonelist(struct vm_area_struct *vma, unsigned long addr,
1809 1810
				gfp_t gfp_flags, struct mempolicy **mpol,
				nodemask_t **nodemask)
1811
{
1812
	struct zonelist *zl;
1813

1814
	*mpol = get_vma_policy(vma, addr);
1815
	*nodemask = NULL;	/* assume !MPOL_BIND */
1816

1817 1818
	if (unlikely((*mpol)->mode == MPOL_INTERLEAVE)) {
		zl = node_zonelist(interleave_nid(*mpol, vma, addr,
1819
				huge_page_shift(hstate_vma(vma))), gfp_flags);
1820
	} else {
1821
		zl = policy_zonelist(gfp_flags, *mpol, numa_node_id());
1822 1823
		if ((*mpol)->mode == MPOL_BIND)
			*nodemask = &(*mpol)->v.nodes;
1824 1825
	}
	return zl;
1826
}
1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851

/*
 * init_nodemask_of_mempolicy
 *
 * If the current task's mempolicy is "default" [NULL], return 'false'
 * to indicate default policy.  Otherwise, extract the policy nodemask
 * for 'bind' or 'interleave' policy into the argument nodemask, or
 * initialize the argument nodemask to contain the single node for
 * 'preferred' or 'local' policy and return 'true' to indicate presence
 * of non-default mempolicy.
 *
 * We don't bother with reference counting the mempolicy [mpol_get/put]
 * because the current task is examining it's own mempolicy and a task's
 * mempolicy is only ever changed by the task itself.
 *
 * N.B., it is the caller's responsibility to free a returned nodemask.
 */
bool init_nodemask_of_mempolicy(nodemask_t *mask)
{
	struct mempolicy *mempolicy;
	int nid;

	if (!(mask && current->mempolicy))
		return false;

1852
	task_lock(current);
1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
	mempolicy = current->mempolicy;
	switch (mempolicy->mode) {
	case MPOL_PREFERRED:
		if (mempolicy->flags & MPOL_F_LOCAL)
			nid = numa_node_id();
		else
			nid = mempolicy->v.preferred_node;
		init_nodemask_of_node(mask, nid);
		break;

	case MPOL_BIND:
		/* Fall through */
	case MPOL_INTERLEAVE:
		*mask =  mempolicy->v.nodes;
		break;

	default:
		BUG();
	}
1872
	task_unlock(current);
1873 1874 1875

	return true;
}
1876
#endif
1877

1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921
/*
 * mempolicy_nodemask_intersects
 *
 * If tsk's mempolicy is "default" [NULL], return 'true' to indicate default
 * policy.  Otherwise, check for intersection between mask and the policy
 * nodemask for 'bind' or 'interleave' policy.  For 'perferred' or 'local'
 * policy, always return true since it may allocate elsewhere on fallback.
 *
 * Takes task_lock(tsk) to prevent freeing of its mempolicy.
 */
bool mempolicy_nodemask_intersects(struct task_struct *tsk,
					const nodemask_t *mask)
{
	struct mempolicy *mempolicy;
	bool ret = true;

	if (!mask)
		return ret;
	task_lock(tsk);
	mempolicy = tsk->mempolicy;
	if (!mempolicy)
		goto out;

	switch (mempolicy->mode) {
	case MPOL_PREFERRED:
		/*
		 * MPOL_PREFERRED and MPOL_F_LOCAL are only preferred nodes to
		 * allocate from, they may fallback to other nodes when oom.
		 * Thus, it's possible for tsk to have allocated memory from
		 * nodes in mask.
		 */
		break;
	case MPOL_BIND:
	case MPOL_INTERLEAVE:
		ret = nodes_intersects(mempolicy->v.nodes, *mask);
		break;
	default:
		BUG();
	}
out:
	task_unlock(tsk);
	return ret;
}

L
Linus Torvalds 已提交
1922 1923
/* Allocate a page in interleaved policy.
   Own path because it needs to do special accounting. */
1924 1925
static struct page *alloc_page_interleave(gfp_t gfp, unsigned order,
					unsigned nid)
L
Linus Torvalds 已提交
1926 1927 1928 1929
{
	struct zonelist *zl;
	struct page *page;

1930
	zl = node_zonelist(nid, gfp);
L
Linus Torvalds 已提交
1931
	page = __alloc_pages(gfp, order, zl);
1932
	if (page && page_zone(page) == zonelist_zone(&zl->_zonerefs[0]))
1933
		inc_zone_page_state(page, NUMA_INTERLEAVE_HIT);
L
Linus Torvalds 已提交
1934 1935 1936 1937
	return page;
}

/**
1938
 * 	alloc_pages_vma	- Allocate a page for a VMA.
L
Linus Torvalds 已提交
1939 1940 1941 1942 1943 1944 1945 1946
 *
 * 	@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.
 *
1947
 *	@order:Order of the GFP allocation.
L
Linus Torvalds 已提交
1948 1949
 * 	@vma:  Pointer to VMA or NULL if not available.
 *	@addr: Virtual Address of the allocation. Must be inside the VMA.
1950 1951
 *	@node: Which node to prefer for allocation (modulo policy).
 *	@hugepage: for hugepages try only the preferred node if possible
L
Linus Torvalds 已提交
1952 1953 1954 1955 1956
 *
 * 	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
1957 1958
 *	all allocations for pages that will be mapped into user space. Returns
 *	NULL when no page can be allocated.
L
Linus Torvalds 已提交
1959 1960
 */
struct page *
1961
alloc_pages_vma(gfp_t gfp, int order, struct vm_area_struct *vma,
1962
		unsigned long addr, int node, bool hugepage)
L
Linus Torvalds 已提交
1963
{
1964
	struct mempolicy *pol;
1965
	struct page *page;
1966
	unsigned int cpuset_mems_cookie;
1967 1968
	struct zonelist *zl;
	nodemask_t *nmask;
1969 1970

retry_cpuset:
1971
	pol = get_vma_policy(vma, addr);
1972
	cpuset_mems_cookie = read_mems_allowed_begin();
L
Linus Torvalds 已提交
1973

1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
	if (unlikely(IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) && hugepage &&
					pol->mode != MPOL_INTERLEAVE)) {
		/*
		 * For hugepage allocation and non-interleave policy which
		 * allows the current node, we only try to allocate from the
		 * current node and don't fall back to other nodes, as the
		 * cost of remote accesses would likely offset THP benefits.
		 *
		 * If the policy is interleave, or does not allow the current
		 * node in its nodemask, we allocate the standard way.
		 */
		nmask = policy_nodemask(gfp, pol);
		if (!nmask || node_isset(node, *nmask)) {
			mpol_cond_put(pol);
			page = alloc_pages_exact_node(node, gfp, order);
			goto out;
		}
	}

	if (pol->mode == MPOL_INTERLEAVE) {
L
Linus Torvalds 已提交
1994
		unsigned nid;
1995

1996
		nid = interleave_nid(pol, vma, addr, PAGE_SHIFT + order);
1997
		mpol_cond_put(pol);
1998
		page = alloc_page_interleave(gfp, order, nid);
1999
		goto out;
L
Linus Torvalds 已提交
2000
	}
2001 2002 2003

	nmask = policy_nodemask(gfp, pol);
	zl = policy_zonelist(gfp, pol, node);
2004
	mpol_cond_put(pol);
2005 2006
	page = __alloc_pages_nodemask(gfp, order, zl, nmask);
out:
2007
	if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie)))
2008
		goto retry_cpuset;
2009
	return page;
L
Linus Torvalds 已提交
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
}

/**
 * 	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.
 *
2027
 *	Don't call cpuset_update_task_memory_state() unless
L
Linus Torvalds 已提交
2028 2029 2030
 *	1) it's ok to take cpuset_sem (can WAIT), and
 *	2) allocating for current task (not interrupt).
 */
A
Al Viro 已提交
2031
struct page *alloc_pages_current(gfp_t gfp, unsigned order)
L
Linus Torvalds 已提交
2032
{
2033
	struct mempolicy *pol = &default_policy;
2034
	struct page *page;
2035
	unsigned int cpuset_mems_cookie;
L
Linus Torvalds 已提交
2036

2037 2038
	if (!in_interrupt() && !(gfp & __GFP_THISNODE))
		pol = get_task_policy(current);
2039

2040
retry_cpuset:
2041
	cpuset_mems_cookie = read_mems_allowed_begin();
2042

2043 2044 2045 2046
	/*
	 * No reference counting needed for current->mempolicy
	 * nor system default_policy
	 */
2047
	if (pol->mode == MPOL_INTERLEAVE)
2048 2049 2050
		page = alloc_page_interleave(gfp, order, interleave_nodes(pol));
	else
		page = __alloc_pages_nodemask(gfp, order,
2051 2052
				policy_zonelist(gfp, pol, numa_node_id()),
				policy_nodemask(gfp, pol));
2053

2054
	if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie)))
2055 2056
		goto retry_cpuset;

2057
	return page;
L
Linus Torvalds 已提交
2058 2059 2060
}
EXPORT_SYMBOL(alloc_pages_current);

2061 2062 2063 2064 2065 2066 2067 2068 2069 2070
int vma_dup_policy(struct vm_area_struct *src, struct vm_area_struct *dst)
{
	struct mempolicy *pol = mpol_dup(vma_policy(src));

	if (IS_ERR(pol))
		return PTR_ERR(pol);
	dst->vm_policy = pol;
	return 0;
}

2071
/*
2072
 * If mpol_dup() sees current->cpuset == cpuset_being_rebound, then it
2073 2074 2075 2076
 * 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().
2077 2078 2079
 *
 * current's mempolicy may be rebinded by the other task(the task that changes
 * cpuset's mems), so we needn't do rebind work for current task.
2080 2081
 */

2082 2083
/* Slow path of a mempolicy duplicate */
struct mempolicy *__mpol_dup(struct mempolicy *old)
L
Linus Torvalds 已提交
2084 2085 2086 2087 2088
{
	struct mempolicy *new = kmem_cache_alloc(policy_cache, GFP_KERNEL);

	if (!new)
		return ERR_PTR(-ENOMEM);
2089 2090 2091 2092 2093 2094 2095 2096 2097

	/* task's mempolicy is protected by alloc_lock */
	if (old == current->mempolicy) {
		task_lock(current);
		*new = *old;
		task_unlock(current);
	} else
		*new = *old;

2098 2099
	if (current_cpuset_is_being_rebound()) {
		nodemask_t mems = cpuset_mems_allowed(current);
2100 2101 2102 2103
		if (new->flags & MPOL_F_REBINDING)
			mpol_rebind_policy(new, &mems, MPOL_REBIND_STEP2);
		else
			mpol_rebind_policy(new, &mems, MPOL_REBIND_ONCE);
2104
	}
L
Linus Torvalds 已提交
2105 2106 2107 2108 2109
	atomic_set(&new->refcnt, 1);
	return new;
}

/* Slow path of a mempolicy comparison */
2110
bool __mpol_equal(struct mempolicy *a, struct mempolicy *b)
L
Linus Torvalds 已提交
2111 2112
{
	if (!a || !b)
2113
		return false;
2114
	if (a->mode != b->mode)
2115
		return false;
B
Bob Liu 已提交
2116
	if (a->flags != b->flags)
2117
		return false;
B
Bob Liu 已提交
2118 2119
	if (mpol_store_user_nodemask(a))
		if (!nodes_equal(a->w.user_nodemask, b->w.user_nodemask))
2120
			return false;
B
Bob Liu 已提交
2121

2122
	switch (a->mode) {
2123 2124
	case MPOL_BIND:
		/* Fall through */
L
Linus Torvalds 已提交
2125
	case MPOL_INTERLEAVE:
2126
		return !!nodes_equal(a->v.nodes, b->v.nodes);
L
Linus Torvalds 已提交
2127
	case MPOL_PREFERRED:
2128
		return a->v.preferred_node == b->v.preferred_node;
L
Linus Torvalds 已提交
2129 2130
	default:
		BUG();
2131
		return false;
L
Linus Torvalds 已提交
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144
	}
}

/*
 * 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 */
2145
/* Caller holds sp->lock */
L
Linus Torvalds 已提交
2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195
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);
2196
	pr_debug("inserting %lx-%lx: %d\n", new->start, new->end,
2197
		 new->policy ? new->policy->mode : 0);
L
Linus Torvalds 已提交
2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208
}

/* 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;
2209
	spin_lock(&sp->lock);
L
Linus Torvalds 已提交
2210 2211 2212 2213 2214
	sn = sp_lookup(sp, idx, idx+1);
	if (sn) {
		mpol_get(sn->policy);
		pol = sn->policy;
	}
2215
	spin_unlock(&sp->lock);
L
Linus Torvalds 已提交
2216 2217 2218
	return pol;
}

2219 2220 2221 2222 2223 2224
static void sp_free(struct sp_node *n)
{
	mpol_put(n->policy);
	kmem_cache_free(sn_cache, n);
}

2225 2226 2227
/**
 * mpol_misplaced - check whether current page node is valid in policy
 *
2228 2229 2230
 * @page: page to be checked
 * @vma: vm area where page mapped
 * @addr: virtual address where page mapped
2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247
 *
 * Lookup current policy node id for vma,addr and "compare to" page's
 * node id.
 *
 * Returns:
 *	-1	- not misplaced, page is in the right node
 *	node	- node id where the page should be
 *
 * Policy determination "mimics" alloc_page_vma().
 * Called from fault path where we know the vma and faulting address.
 */
int mpol_misplaced(struct page *page, struct vm_area_struct *vma, unsigned long addr)
{
	struct mempolicy *pol;
	struct zone *zone;
	int curnid = page_to_nid(page);
	unsigned long pgoff;
2248 2249
	int thiscpu = raw_smp_processor_id();
	int thisnid = cpu_to_node(thiscpu);
2250 2251 2252 2253 2254
	int polnid = -1;
	int ret = -1;

	BUG_ON(!vma);

2255
	pol = get_vma_policy(vma, addr);
2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294
	if (!(pol->flags & MPOL_F_MOF))
		goto out;

	switch (pol->mode) {
	case MPOL_INTERLEAVE:
		BUG_ON(addr >= vma->vm_end);
		BUG_ON(addr < vma->vm_start);

		pgoff = vma->vm_pgoff;
		pgoff += (addr - vma->vm_start) >> PAGE_SHIFT;
		polnid = offset_il_node(pol, vma, pgoff);
		break;

	case MPOL_PREFERRED:
		if (pol->flags & MPOL_F_LOCAL)
			polnid = numa_node_id();
		else
			polnid = pol->v.preferred_node;
		break;

	case MPOL_BIND:
		/*
		 * allows binding to multiple nodes.
		 * use current page if in policy nodemask,
		 * else select nearest allowed node, if any.
		 * If no allowed nodes, use current [!misplaced].
		 */
		if (node_isset(curnid, pol->v.nodes))
			goto out;
		(void)first_zones_zonelist(
				node_zonelist(numa_node_id(), GFP_HIGHUSER),
				gfp_zone(GFP_HIGHUSER),
				&pol->v.nodes, &zone);
		polnid = zone->node;
		break;

	default:
		BUG();
	}
M
Mel Gorman 已提交
2295 2296

	/* Migrate the page towards the node whose CPU is referencing it */
2297
	if (pol->flags & MPOL_F_MORON) {
2298
		polnid = thisnid;
M
Mel Gorman 已提交
2299

2300
		if (!should_numa_migrate_memory(current, page, curnid, thiscpu))
2301
			goto out;
2302 2303
	}

2304 2305 2306 2307 2308 2309 2310 2311
	if (curnid != polnid)
		ret = polnid;
out:
	mpol_cond_put(pol);

	return ret;
}

L
Linus Torvalds 已提交
2312 2313
static void sp_delete(struct shared_policy *sp, struct sp_node *n)
{
2314
	pr_debug("deleting %lx-l%lx\n", n->start, n->end);
L
Linus Torvalds 已提交
2315
	rb_erase(&n->nd, &sp->root);
2316
	sp_free(n);
L
Linus Torvalds 已提交
2317 2318
}

2319 2320 2321 2322 2323 2324 2325 2326
static void sp_node_init(struct sp_node *node, unsigned long start,
			unsigned long end, struct mempolicy *pol)
{
	node->start = start;
	node->end = end;
	node->policy = pol;
}

A
Adrian Bunk 已提交
2327 2328
static struct sp_node *sp_alloc(unsigned long start, unsigned long end,
				struct mempolicy *pol)
L
Linus Torvalds 已提交
2329
{
2330 2331
	struct sp_node *n;
	struct mempolicy *newpol;
L
Linus Torvalds 已提交
2332

2333
	n = kmem_cache_alloc(sn_cache, GFP_KERNEL);
L
Linus Torvalds 已提交
2334 2335
	if (!n)
		return NULL;
2336 2337 2338 2339 2340 2341 2342

	newpol = mpol_dup(pol);
	if (IS_ERR(newpol)) {
		kmem_cache_free(sn_cache, n);
		return NULL;
	}
	newpol->flags |= MPOL_F_SHARED;
2343
	sp_node_init(n, start, end, newpol);
2344

L
Linus Torvalds 已提交
2345 2346 2347 2348 2349 2350 2351
	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)
{
2352
	struct sp_node *n;
2353 2354
	struct sp_node *n_new = NULL;
	struct mempolicy *mpol_new = NULL;
2355
	int ret = 0;
L
Linus Torvalds 已提交
2356

2357 2358
restart:
	spin_lock(&sp->lock);
L
Linus Torvalds 已提交
2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370
	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) {
2371 2372 2373 2374 2375
				if (!n_new)
					goto alloc_new;

				*mpol_new = *n->policy;
				atomic_set(&mpol_new->refcnt, 1);
2376
				sp_node_init(n_new, end, n->end, mpol_new);
L
Linus Torvalds 已提交
2377
				n->end = start;
2378
				sp_insert(sp, n_new);
2379 2380
				n_new = NULL;
				mpol_new = NULL;
L
Linus Torvalds 已提交
2381 2382 2383 2384 2385 2386 2387 2388 2389 2390
				break;
			} else
				n->end = start;
		}
		if (!next)
			break;
		n = rb_entry(next, struct sp_node, nd);
	}
	if (new)
		sp_insert(sp, new);
2391 2392 2393 2394 2395 2396 2397 2398 2399
	spin_unlock(&sp->lock);
	ret = 0;

err_out:
	if (mpol_new)
		mpol_put(mpol_new);
	if (n_new)
		kmem_cache_free(sn_cache, n_new);

2400
	return ret;
2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411

alloc_new:
	spin_unlock(&sp->lock);
	ret = -ENOMEM;
	n_new = kmem_cache_alloc(sn_cache, GFP_KERNEL);
	if (!n_new)
		goto err_out;
	mpol_new = kmem_cache_alloc(policy_cache, GFP_KERNEL);
	if (!mpol_new)
		goto err_out;
	goto restart;
L
Linus Torvalds 已提交
2412 2413
}

2414 2415 2416 2417 2418 2419 2420 2421
/**
 * mpol_shared_policy_init - initialize shared policy for inode
 * @sp: pointer to inode shared policy
 * @mpol:  struct mempolicy to install
 *
 * Install non-NULL @mpol in inode's shared policy rb-tree.
 * On entry, the current task has a reference on a non-NULL @mpol.
 * This must be released on exit.
2422
 * This is called at get_inode() calls and we can use GFP_KERNEL.
2423 2424 2425
 */
void mpol_shared_policy_init(struct shared_policy *sp, struct mempolicy *mpol)
{
2426 2427
	int ret;

2428
	sp->root = RB_ROOT;		/* empty tree == default mempolicy */
2429
	spin_lock_init(&sp->lock);
2430 2431 2432 2433

	if (mpol) {
		struct vm_area_struct pvma;
		struct mempolicy *new;
2434
		NODEMASK_SCRATCH(scratch);
2435

2436
		if (!scratch)
2437
			goto put_mpol;
2438 2439
		/* contextualize the tmpfs mount point mempolicy */
		new = mpol_new(mpol->mode, mpol->flags, &mpol->w.user_nodemask);
2440
		if (IS_ERR(new))
2441
			goto free_scratch; /* no valid nodemask intersection */
2442 2443

		task_lock(current);
2444
		ret = mpol_set_nodemask(new, &mpol->w.user_nodemask, scratch);
2445
		task_unlock(current);
2446
		if (ret)
2447
			goto put_new;
2448 2449 2450 2451 2452

		/* Create pseudo-vma that contains just the policy */
		memset(&pvma, 0, sizeof(struct vm_area_struct));
		pvma.vm_end = TASK_SIZE;	/* policy covers entire file */
		mpol_set_shared_policy(sp, &pvma, new); /* adds ref */
2453

2454
put_new:
2455
		mpol_put(new);			/* drop initial ref */
2456
free_scratch:
2457
		NODEMASK_SCRATCH_FREE(scratch);
2458 2459
put_mpol:
		mpol_put(mpol);	/* drop our incoming ref on sb mpol */
2460 2461 2462
	}
}

L
Linus Torvalds 已提交
2463 2464 2465 2466 2467 2468 2469
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);

2470
	pr_debug("set_shared_policy %lx sz %lu %d %d %lx\n",
L
Linus Torvalds 已提交
2471
		 vma->vm_pgoff,
2472
		 sz, npol ? npol->mode : -1,
2473
		 npol ? npol->flags : -1,
D
David Rientjes 已提交
2474
		 npol ? nodes_addr(npol->v.nodes)[0] : NUMA_NO_NODE);
L
Linus Torvalds 已提交
2475 2476 2477 2478 2479 2480 2481 2482

	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)
2483
		sp_free(new);
L
Linus Torvalds 已提交
2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494
	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;
2495
	spin_lock(&p->lock);
L
Linus Torvalds 已提交
2496 2497 2498 2499
	next = rb_first(&p->root);
	while (next) {
		n = rb_entry(next, struct sp_node, nd);
		next = rb_next(&n->nd);
2500
		sp_delete(p, n);
L
Linus Torvalds 已提交
2501
	}
2502
	spin_unlock(&p->lock);
L
Linus Torvalds 已提交
2503 2504
}

2505
#ifdef CONFIG_NUMA_BALANCING
2506
static int __initdata numabalancing_override;
2507 2508 2509 2510 2511 2512 2513 2514

static void __init check_numabalancing_enable(void)
{
	bool numabalancing_default = false;

	if (IS_ENABLED(CONFIG_NUMA_BALANCING_DEFAULT_ENABLED))
		numabalancing_default = true;

2515 2516 2517 2518
	/* Parsed by setup_numabalancing. override == 1 enables, -1 disables */
	if (numabalancing_override)
		set_numabalancing_state(numabalancing_override == 1);

2519
	if (nr_node_ids > 1 && !numabalancing_override) {
2520
		pr_info("%s automatic NUMA balancing. "
2521 2522 2523
			"Configure with numa_balancing= or the "
			"kernel.numa_balancing sysctl",
			numabalancing_default ? "Enabling" : "Disabling");
2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534
		set_numabalancing_state(numabalancing_default);
	}
}

static int __init setup_numabalancing(char *str)
{
	int ret = 0;
	if (!str)
		goto out;

	if (!strcmp(str, "enable")) {
2535
		numabalancing_override = 1;
2536 2537
		ret = 1;
	} else if (!strcmp(str, "disable")) {
2538
		numabalancing_override = -1;
2539 2540 2541 2542
		ret = 1;
	}
out:
	if (!ret)
2543
		pr_warn("Unable to parse numa_balancing=\n");
2544 2545 2546 2547 2548 2549 2550 2551 2552 2553

	return ret;
}
__setup("numa_balancing=", setup_numabalancing);
#else
static inline void __init check_numabalancing_enable(void)
{
}
#endif /* CONFIG_NUMA_BALANCING */

L
Linus Torvalds 已提交
2554 2555 2556
/* assumes fs == KERNEL_DS */
void __init numa_policy_init(void)
{
2557 2558 2559 2560
	nodemask_t interleave_nodes;
	unsigned long largest = 0;
	int nid, prefer = 0;

L
Linus Torvalds 已提交
2561 2562
	policy_cache = kmem_cache_create("numa_policy",
					 sizeof(struct mempolicy),
2563
					 0, SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
2564 2565 2566

	sn_cache = kmem_cache_create("shared_policy_node",
				     sizeof(struct sp_node),
2567
				     0, SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
2568

M
Mel Gorman 已提交
2569 2570 2571 2572 2573 2574 2575 2576 2577
	for_each_node(nid) {
		preferred_node_policy[nid] = (struct mempolicy) {
			.refcnt = ATOMIC_INIT(1),
			.mode = MPOL_PREFERRED,
			.flags = MPOL_F_MOF | MPOL_F_MORON,
			.v = { .preferred_node = nid, },
		};
	}

2578 2579 2580 2581 2582 2583
	/*
	 * 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);
2584
	for_each_node_state(nid, N_MEMORY) {
2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600
		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
2601

2602
	if (do_set_mempolicy(MPOL_INTERLEAVE, 0, &interleave_nodes))
2603
		pr_err("%s: interleaving failed\n", __func__);
2604 2605

	check_numabalancing_enable();
L
Linus Torvalds 已提交
2606 2607
}

2608
/* Reset policy of current process to default */
L
Linus Torvalds 已提交
2609 2610
void numa_default_policy(void)
{
2611
	do_set_mempolicy(MPOL_DEFAULT, 0, NULL);
L
Linus Torvalds 已提交
2612
}
2613

2614 2615 2616 2617
/*
 * Parse and format mempolicy from/to strings
 */

2618
/*
2619
 * "local" is implemented internally by MPOL_PREFERRED with MPOL_F_LOCAL flag.
2620
 */
2621 2622 2623 2624 2625 2626
static const char * const policy_modes[] =
{
	[MPOL_DEFAULT]    = "default",
	[MPOL_PREFERRED]  = "prefer",
	[MPOL_BIND]       = "bind",
	[MPOL_INTERLEAVE] = "interleave",
L
Lee Schermerhorn 已提交
2627
	[MPOL_LOCAL]      = "local",
2628
};
2629

2630 2631 2632

#ifdef CONFIG_TMPFS
/**
2633
 * mpol_parse_str - parse string to mempolicy, for tmpfs mpol mount option.
2634
 * @str:  string containing mempolicy to parse
2635
 * @mpol:  pointer to struct mempolicy pointer, returned on success.
2636 2637 2638 2639
 *
 * Format of input:
 *	<mode>[=<flags>][:<nodelist>]
 *
2640
 * On success, returns 0, else 1
2641
 */
2642
int mpol_parse_str(char *str, struct mempolicy **mpol)
2643
{
2644
	struct mempolicy *new = NULL;
2645
	unsigned short mode;
2646
	unsigned short mode_flags;
2647
	nodemask_t nodes;
2648 2649 2650 2651 2652 2653 2654
	char *nodelist = strchr(str, ':');
	char *flags = strchr(str, '=');
	int err = 1;

	if (nodelist) {
		/* NUL-terminate mode or flags string */
		*nodelist++ = '\0';
2655
		if (nodelist_parse(nodelist, nodes))
2656
			goto out;
2657
		if (!nodes_subset(nodes, node_states[N_MEMORY]))
2658
			goto out;
2659 2660 2661
	} else
		nodes_clear(nodes);

2662 2663 2664
	if (flags)
		*flags++ = '\0';	/* terminate mode string */

2665
	for (mode = 0; mode < MPOL_MAX; mode++) {
2666
		if (!strcmp(str, policy_modes[mode])) {
2667 2668 2669
			break;
		}
	}
2670
	if (mode >= MPOL_MAX)
2671 2672
		goto out;

2673
	switch (mode) {
2674
	case MPOL_PREFERRED:
2675 2676 2677
		/*
		 * Insist on a nodelist of one node only
		 */
2678 2679 2680 2681
		if (nodelist) {
			char *rest = nodelist;
			while (isdigit(*rest))
				rest++;
2682 2683
			if (*rest)
				goto out;
2684 2685 2686 2687 2688 2689 2690
		}
		break;
	case MPOL_INTERLEAVE:
		/*
		 * Default to online nodes with memory if no nodelist
		 */
		if (!nodelist)
2691
			nodes = node_states[N_MEMORY];
2692
		break;
2693
	case MPOL_LOCAL:
2694
		/*
2695
		 * Don't allow a nodelist;  mpol_new() checks flags
2696
		 */
2697
		if (nodelist)
2698
			goto out;
2699
		mode = MPOL_PREFERRED;
2700
		break;
2701 2702 2703 2704 2705 2706 2707
	case MPOL_DEFAULT:
		/*
		 * Insist on a empty nodelist
		 */
		if (!nodelist)
			err = 0;
		goto out;
2708 2709 2710 2711 2712 2713
	case MPOL_BIND:
		/*
		 * Insist on a nodelist
		 */
		if (!nodelist)
			goto out;
2714 2715
	}

2716
	mode_flags = 0;
2717 2718 2719 2720 2721 2722
	if (flags) {
		/*
		 * Currently, we only support two mutually exclusive
		 * mode flags.
		 */
		if (!strcmp(flags, "static"))
2723
			mode_flags |= MPOL_F_STATIC_NODES;
2724
		else if (!strcmp(flags, "relative"))
2725
			mode_flags |= MPOL_F_RELATIVE_NODES;
2726
		else
2727
			goto out;
2728
	}
2729 2730 2731

	new = mpol_new(mode, mode_flags, &nodes);
	if (IS_ERR(new))
2732 2733
		goto out;

2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750
	/*
	 * Save nodes for mpol_to_str() to show the tmpfs mount options
	 * for /proc/mounts, /proc/pid/mounts and /proc/pid/mountinfo.
	 */
	if (mode != MPOL_PREFERRED)
		new->v.nodes = nodes;
	else if (nodelist)
		new->v.preferred_node = first_node(nodes);
	else
		new->flags |= MPOL_F_LOCAL;

	/*
	 * Save nodes for contextualization: this will be used to "clone"
	 * the mempolicy in a specific context [cpuset] at a later time.
	 */
	new->w.user_nodemask = nodes;

2751
	err = 0;
2752

2753 2754 2755 2756 2757 2758
out:
	/* Restore string for error message */
	if (nodelist)
		*--nodelist = ':';
	if (flags)
		*--flags = '=';
2759 2760
	if (!err)
		*mpol = new;
2761 2762 2763 2764
	return err;
}
#endif /* CONFIG_TMPFS */

2765 2766 2767 2768 2769 2770
/**
 * mpol_to_str - format a mempolicy structure for printing
 * @buffer:  to contain formatted mempolicy string
 * @maxlen:  length of @buffer
 * @pol:  pointer to mempolicy to be formatted
 *
2771 2772 2773
 * Convert @pol into a string.  If @buffer is too short, truncate the string.
 * Recommend a @maxlen of at least 32 for the longest mode, "interleave", the
 * longest flag, "relative", and to display at least a few node ids.
2774
 */
2775
void mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
2776 2777
{
	char *p = buffer;
2778 2779 2780
	nodemask_t nodes = NODE_MASK_NONE;
	unsigned short mode = MPOL_DEFAULT;
	unsigned short flags = 0;
2781

2782
	if (pol && pol != &default_policy && !(pol->flags & MPOL_F_MORON)) {
2783
		mode = pol->mode;
2784 2785
		flags = pol->flags;
	}
2786

2787 2788 2789 2790
	switch (mode) {
	case MPOL_DEFAULT:
		break;
	case MPOL_PREFERRED:
2791
		if (flags & MPOL_F_LOCAL)
2792
			mode = MPOL_LOCAL;
2793
		else
2794
			node_set(pol->v.preferred_node, nodes);
2795 2796 2797
		break;
	case MPOL_BIND:
	case MPOL_INTERLEAVE:
2798
		nodes = pol->v.nodes;
2799 2800
		break;
	default:
2801 2802 2803
		WARN_ON_ONCE(1);
		snprintf(p, maxlen, "unknown");
		return;
2804 2805
	}

2806
	p += snprintf(p, maxlen, "%s", policy_modes[mode]);
2807

2808
	if (flags & MPOL_MODE_FLAGS) {
2809
		p += snprintf(p, buffer + maxlen - p, "=");
2810

2811 2812 2813
		/*
		 * Currently, the only defined flags are mutually exclusive
		 */
2814
		if (flags & MPOL_F_STATIC_NODES)
2815 2816 2817
			p += snprintf(p, buffer + maxlen - p, "static");
		else if (flags & MPOL_F_RELATIVE_NODES)
			p += snprintf(p, buffer + maxlen - p, "relative");
2818 2819
	}

2820
	if (!nodes_empty(nodes)) {
2821
		p += snprintf(p, buffer + maxlen - p, ":");
2822 2823 2824
	 	p += nodelist_scnprintf(p, buffer + maxlen - p, nodes);
	}
}