cpuset.c 70.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6
/*
 *  kernel/cpuset.c
 *
 *  Processor and Memory placement constraints for sets of tasks.
 *
 *  Copyright (C) 2003 BULL SA.
P
Paul Jackson 已提交
7
 *  Copyright (C) 2004-2007 Silicon Graphics, Inc.
8
 *  Copyright (C) 2006 Google, Inc
L
Linus Torvalds 已提交
9 10 11 12
 *
 *  Portions derived from Patrick Mochel's sysfs code.
 *  sysfs is Copyright (c) 2001-3 Patrick Mochel
 *
13
 *  2003-10-10 Written by Simon Derr.
L
Linus Torvalds 已提交
14
 *  2003-10-22 Updates by Stephen Hemminger.
15
 *  2004 May-July Rework by Paul Jackson.
16
 *  2006 Rework by Paul Menage to use generic cgroups
17 18
 *  2008 Rework of the scheduler domains and CPU hotplug handling
 *       by Max Krasnyansky
L
Linus Torvalds 已提交
19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
 *
 *  This file is subject to the terms and conditions of the GNU General Public
 *  License.  See the file COPYING in the main directory of the Linux
 *  distribution for more details.
 */

#include <linux/cpu.h>
#include <linux/cpumask.h>
#include <linux/cpuset.h>
#include <linux/err.h>
#include <linux/errno.h>
#include <linux/file.h>
#include <linux/fs.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/kernel.h>
#include <linux/kmod.h>
#include <linux/list.h>
37
#include <linux/mempolicy.h>
L
Linus Torvalds 已提交
38 39 40 41 42 43
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/mount.h>
#include <linux/namei.h>
#include <linux/pagemap.h>
#include <linux/proc_fs.h>
44
#include <linux/rcupdate.h>
L
Linus Torvalds 已提交
45 46
#include <linux/sched.h>
#include <linux/seq_file.h>
47
#include <linux/security.h>
L
Linus Torvalds 已提交
48 49 50 51 52 53 54 55 56 57
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/stat.h>
#include <linux/string.h>
#include <linux/time.h>
#include <linux/backing-dev.h>
#include <linux/sort.h>

#include <asm/uaccess.h>
#include <asm/atomic.h>
58
#include <linux/mutex.h>
59 60
#include <linux/workqueue.h>
#include <linux/cgroup.h>
L
Linus Torvalds 已提交
61

62 63 64 65 66
/*
 * Tracks how many cpusets are currently defined in system.
 * When there is only one cpuset (the root cpuset) we can
 * short circuit some hooks.
 */
67
int number_of_cpusets __read_mostly;
68

69
/* Forward declare cgroup structures */
70 71 72
struct cgroup_subsys cpuset_subsys;
struct cpuset;

73 74 75 76 77 78 79 80 81
/* See "Frequency meter" comments, below. */

struct fmeter {
	int cnt;		/* unprocessed events count */
	int val;		/* most recent output value */
	time_t time;		/* clock (secs) when val computed */
	spinlock_t lock;	/* guards read or write of above */
};

L
Linus Torvalds 已提交
82
struct cpuset {
83 84
	struct cgroup_subsys_state css;

L
Linus Torvalds 已提交
85 86 87 88 89 90 91 92 93 94
	unsigned long flags;		/* "unsigned long" so bitops work */
	cpumask_t cpus_allowed;		/* CPUs allowed to tasks in cpuset */
	nodemask_t mems_allowed;	/* Memory Nodes allowed to tasks */

	struct cpuset *parent;		/* my parent */

	/*
	 * Copy of global cpuset_mems_generation as of the most
	 * recent time this cpuset changed its mems_allowed.
	 */
95 96 97
	int mems_generation;

	struct fmeter fmeter;		/* memory_pressure filter */
P
Paul Jackson 已提交
98 99 100

	/* partition number for rebuild_sched_domains() */
	int pn;
101

102 103 104
	/* for custom sched domain */
	int relax_domain_level;

105 106
	/* used for walking a cpuset heirarchy */
	struct list_head stack_list;
L
Linus Torvalds 已提交
107 108
};

109 110 111 112 113 114 115 116 117 118 119 120 121
/* Retrieve the cpuset for a cgroup */
static inline struct cpuset *cgroup_cs(struct cgroup *cont)
{
	return container_of(cgroup_subsys_state(cont, cpuset_subsys_id),
			    struct cpuset, css);
}

/* Retrieve the cpuset for a task */
static inline struct cpuset *task_cs(struct task_struct *task)
{
	return container_of(task_subsys_state(task, cpuset_subsys_id),
			    struct cpuset, css);
}
122 123 124 125
struct cpuset_hotplug_scanner {
	struct cgroup_scanner scan;
	struct cgroup *to;
};
126

L
Linus Torvalds 已提交
127 128 129 130
/* bits in struct cpuset flags field */
typedef enum {
	CS_CPU_EXCLUSIVE,
	CS_MEM_EXCLUSIVE,
131
	CS_MEM_HARDWALL,
132
	CS_MEMORY_MIGRATE,
P
Paul Jackson 已提交
133
	CS_SCHED_LOAD_BALANCE,
134 135
	CS_SPREAD_PAGE,
	CS_SPREAD_SLAB,
L
Linus Torvalds 已提交
136 137 138 139 140
} cpuset_flagbits_t;

/* convenient tests for these bits */
static inline int is_cpu_exclusive(const struct cpuset *cs)
{
141
	return test_bit(CS_CPU_EXCLUSIVE, &cs->flags);
L
Linus Torvalds 已提交
142 143 144 145
}

static inline int is_mem_exclusive(const struct cpuset *cs)
{
146
	return test_bit(CS_MEM_EXCLUSIVE, &cs->flags);
L
Linus Torvalds 已提交
147 148
}

149 150 151 152 153
static inline int is_mem_hardwall(const struct cpuset *cs)
{
	return test_bit(CS_MEM_HARDWALL, &cs->flags);
}

P
Paul Jackson 已提交
154 155 156 157 158
static inline int is_sched_load_balance(const struct cpuset *cs)
{
	return test_bit(CS_SCHED_LOAD_BALANCE, &cs->flags);
}

159 160
static inline int is_memory_migrate(const struct cpuset *cs)
{
161
	return test_bit(CS_MEMORY_MIGRATE, &cs->flags);
162 163
}

164 165 166 167 168 169 170 171 172 173
static inline int is_spread_page(const struct cpuset *cs)
{
	return test_bit(CS_SPREAD_PAGE, &cs->flags);
}

static inline int is_spread_slab(const struct cpuset *cs)
{
	return test_bit(CS_SPREAD_SLAB, &cs->flags);
}

L
Linus Torvalds 已提交
174
/*
175
 * Increment this integer everytime any cpuset changes its
L
Linus Torvalds 已提交
176 177 178 179
 * mems_allowed value.  Users of cpusets can track this generation
 * number, and avoid having to lock and reload mems_allowed unless
 * the cpuset they're using changes generation.
 *
180
 * A single, global generation is needed because cpuset_attach_task() could
L
Linus Torvalds 已提交
181 182 183 184
 * reattach a task to a different cpuset, which must not have its
 * generation numbers aliased with those of that tasks previous cpuset.
 *
 * Generations are needed for mems_allowed because one task cannot
185
 * modify another's memory placement.  So we must enable every task,
L
Linus Torvalds 已提交
186 187 188
 * on every visit to __alloc_pages(), to efficiently check whether
 * its current->cpuset->mems_allowed has changed, requiring an update
 * of its current->mems_allowed.
189
 *
190
 * Since writes to cpuset_mems_generation are guarded by the cgroup lock
191
 * there is no need to mark it atomic.
L
Linus Torvalds 已提交
192
 */
193
static int cpuset_mems_generation;
L
Linus Torvalds 已提交
194 195 196 197 198 199 200 201

static struct cpuset top_cpuset = {
	.flags = ((1 << CS_CPU_EXCLUSIVE) | (1 << CS_MEM_EXCLUSIVE)),
	.cpus_allowed = CPU_MASK_ALL,
	.mems_allowed = NODE_MASK_ALL,
};

/*
202 203 204 205 206 207 208
 * There are two global mutexes guarding cpuset structures.  The first
 * is the main control groups cgroup_mutex, accessed via
 * cgroup_lock()/cgroup_unlock().  The second is the cpuset-specific
 * callback_mutex, below. They can nest.  It is ok to first take
 * cgroup_mutex, then nest callback_mutex.  We also require taking
 * task_lock() when dereferencing a task's cpuset pointer.  See "The
 * task_lock() exception", at the end of this comment.
209
 *
210
 * A task must hold both mutexes to modify cpusets.  If a task
211
 * holds cgroup_mutex, then it blocks others wanting that mutex,
212
 * ensuring that it is the only task able to also acquire callback_mutex
213 214
 * and be able to modify cpusets.  It can perform various checks on
 * the cpuset structure first, knowing nothing will change.  It can
215
 * also allocate memory while just holding cgroup_mutex.  While it is
216
 * performing these checks, various callback routines can briefly
217 218
 * acquire callback_mutex to query cpusets.  Once it is ready to make
 * the changes, it takes callback_mutex, blocking everyone else.
219 220
 *
 * Calls to the kernel memory allocator can not be made while holding
221
 * callback_mutex, as that would risk double tripping on callback_mutex
222 223 224
 * from one of the callbacks into the cpuset code from within
 * __alloc_pages().
 *
225
 * If a task is only holding callback_mutex, then it has read-only
226 227 228 229 230
 * access to cpusets.
 *
 * The task_struct fields mems_allowed and mems_generation may only
 * be accessed in the context of that task, so require no locks.
 *
231
 * The cpuset_common_file_read() handlers only hold callback_mutex across
232 233 234
 * small pieces of code, such as when reading out possibly multi-word
 * cpumasks and nodemasks.
 *
235 236
 * Accessing a task's cpuset should be done in accordance with the
 * guidelines for accessing subsystem state in kernel/cgroup.c
L
Linus Torvalds 已提交
237 238
 */

239
static DEFINE_MUTEX(callback_mutex);
240

241 242
/*
 * This is ugly, but preserves the userspace API for existing cpuset
243
 * users. If someone tries to mount the "cpuset" filesystem, we
244 245
 * silently switch it to mount "cgroup" instead
 */
246 247 248
static int cpuset_get_sb(struct file_system_type *fs_type,
			 int flags, const char *unused_dev_name,
			 void *data, struct vfsmount *mnt)
L
Linus Torvalds 已提交
249
{
250 251 252 253 254 255 256 257 258 259 260
	struct file_system_type *cgroup_fs = get_fs_type("cgroup");
	int ret = -ENODEV;
	if (cgroup_fs) {
		char mountopts[] =
			"cpuset,noprefix,"
			"release_agent=/sbin/cpuset_release_agent";
		ret = cgroup_fs->get_sb(cgroup_fs, flags,
					   unused_dev_name, mountopts, mnt);
		put_filesystem(cgroup_fs);
	}
	return ret;
L
Linus Torvalds 已提交
261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278
}

static struct file_system_type cpuset_fs_type = {
	.name = "cpuset",
	.get_sb = cpuset_get_sb,
};

/*
 * Return in *pmask the portion of a cpusets's cpus_allowed that
 * are online.  If none are online, walk up the cpuset hierarchy
 * until we find one that does have some online cpus.  If we get
 * all the way to the top and still haven't found any online cpus,
 * return cpu_online_map.  Or if passed a NULL cs from an exit'ing
 * task, return cpu_online_map.
 *
 * One way or another, we guarantee to return some non-empty subset
 * of cpu_online_map.
 *
279
 * Call with callback_mutex held.
L
Linus Torvalds 已提交
280 281 282 283 284 285 286 287 288 289 290 291 292 293 294
 */

static void guarantee_online_cpus(const struct cpuset *cs, cpumask_t *pmask)
{
	while (cs && !cpus_intersects(cs->cpus_allowed, cpu_online_map))
		cs = cs->parent;
	if (cs)
		cpus_and(*pmask, cs->cpus_allowed, cpu_online_map);
	else
		*pmask = cpu_online_map;
	BUG_ON(!cpus_intersects(*pmask, cpu_online_map));
}

/*
 * Return in *pmask the portion of a cpusets's mems_allowed that
295 296 297 298
 * are online, with memory.  If none are online with memory, walk
 * up the cpuset hierarchy until we find one that does have some
 * online mems.  If we get all the way to the top and still haven't
 * found any online mems, return node_states[N_HIGH_MEMORY].
L
Linus Torvalds 已提交
299 300
 *
 * One way or another, we guarantee to return some non-empty subset
301
 * of node_states[N_HIGH_MEMORY].
L
Linus Torvalds 已提交
302
 *
303
 * Call with callback_mutex held.
L
Linus Torvalds 已提交
304 305 306 307
 */

static void guarantee_online_mems(const struct cpuset *cs, nodemask_t *pmask)
{
308 309
	while (cs && !nodes_intersects(cs->mems_allowed,
					node_states[N_HIGH_MEMORY]))
L
Linus Torvalds 已提交
310 311
		cs = cs->parent;
	if (cs)
312 313
		nodes_and(*pmask, cs->mems_allowed,
					node_states[N_HIGH_MEMORY]);
L
Linus Torvalds 已提交
314
	else
315 316
		*pmask = node_states[N_HIGH_MEMORY];
	BUG_ON(!nodes_intersects(*pmask, node_states[N_HIGH_MEMORY]));
L
Linus Torvalds 已提交
317 318
}

319 320 321 322 323 324
/**
 * cpuset_update_task_memory_state - update task memory placement
 *
 * If the current tasks cpusets mems_allowed changed behind our
 * backs, update current->mems_allowed, mems_generation and task NUMA
 * mempolicy to the new value.
325
 *
326 327 328 329
 * Task mempolicy is updated by rebinding it relative to the
 * current->cpuset if a task has its memory placement changed.
 * Do not call this routine if in_interrupt().
 *
330
 * Call without callback_mutex or task_lock() held.  May be
331 332
 * called with or without cgroup_mutex held.  Thanks in part to
 * 'the_top_cpuset_hack', the task's cpuset pointer will never
D
David Rientjes 已提交
333 334
 * be NULL.  This routine also might acquire callback_mutex during
 * call.
335
 *
336 337
 * Reading current->cpuset->mems_generation doesn't need task_lock
 * to guard the current->cpuset derefence, because it is guarded
338
 * from concurrent freeing of current->cpuset using RCU.
339 340 341 342 343 344 345 346 347 348 349 350 351 352
 *
 * The rcu_dereference() is technically probably not needed,
 * as I don't actually mind if I see a new cpuset pointer but
 * an old value of mems_generation.  However this really only
 * matters on alpha systems using cpusets heavily.  If I dropped
 * that rcu_dereference(), it would save them a memory barrier.
 * For all other arch's, rcu_dereference is a no-op anyway, and for
 * alpha systems not using cpusets, another planned optimization,
 * avoiding the rcu critical section for tasks in the root cpuset
 * which is statically allocated, so can't vanish, will make this
 * irrelevant.  Better to use RCU as intended, than to engage in
 * some cute trick to save a memory barrier that is impossible to
 * test, for alpha systems using cpusets heavily, which might not
 * even exist.
353 354 355 356 357
 *
 * This routine is needed to update the per-task mems_allowed data,
 * within the tasks context, when it is trying to allocate memory
 * (in various mm/mempolicy.c routines) and notices that some other
 * task has been modifying its cpuset.
L
Linus Torvalds 已提交
358 359
 */

360
void cpuset_update_task_memory_state(void)
L
Linus Torvalds 已提交
361
{
362
	int my_cpusets_mem_gen;
363
	struct task_struct *tsk = current;
364
	struct cpuset *cs;
365

366
	if (task_cs(tsk) == &top_cpuset) {
367 368 369 370
		/* Don't need rcu for top_cpuset.  It's never freed. */
		my_cpusets_mem_gen = top_cpuset.mems_generation;
	} else {
		rcu_read_lock();
L
Lai Jiangshan 已提交
371
		my_cpusets_mem_gen = task_cs(tsk)->mems_generation;
372 373
		rcu_read_unlock();
	}
L
Linus Torvalds 已提交
374

375
	if (my_cpusets_mem_gen != tsk->cpuset_mems_generation) {
376
		mutex_lock(&callback_mutex);
377
		task_lock(tsk);
378
		cs = task_cs(tsk); /* Maybe changed when task not locked */
379 380
		guarantee_online_mems(cs, &tsk->mems_allowed);
		tsk->cpuset_mems_generation = cs->mems_generation;
381 382 383 384 385 386 387 388
		if (is_spread_page(cs))
			tsk->flags |= PF_SPREAD_PAGE;
		else
			tsk->flags &= ~PF_SPREAD_PAGE;
		if (is_spread_slab(cs))
			tsk->flags |= PF_SPREAD_SLAB;
		else
			tsk->flags &= ~PF_SPREAD_SLAB;
389
		task_unlock(tsk);
390
		mutex_unlock(&callback_mutex);
391
		mpol_rebind_task(tsk, &tsk->mems_allowed);
L
Linus Torvalds 已提交
392 393 394 395 396 397 398 399
	}
}

/*
 * is_cpuset_subset(p, q) - Is cpuset p a subset of cpuset q?
 *
 * One cpuset is a subset of another if all its allowed CPUs and
 * Memory Nodes are a subset of the other, and its exclusive flags
400
 * are only set if the other's are set.  Call holding cgroup_mutex.
L
Linus Torvalds 已提交
401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417
 */

static int is_cpuset_subset(const struct cpuset *p, const struct cpuset *q)
{
	return	cpus_subset(p->cpus_allowed, q->cpus_allowed) &&
		nodes_subset(p->mems_allowed, q->mems_allowed) &&
		is_cpu_exclusive(p) <= is_cpu_exclusive(q) &&
		is_mem_exclusive(p) <= is_mem_exclusive(q);
}

/*
 * validate_change() - Used to validate that any proposed cpuset change
 *		       follows the structural rules for cpusets.
 *
 * If we replaced the flag and mask values of the current cpuset
 * (cur) with those values in the trial cpuset (trial), would
 * our various subset and exclusive rules still be valid?  Presumes
418
 * cgroup_mutex held.
L
Linus Torvalds 已提交
419 420 421 422 423 424 425 426 427 428 429 430 431 432
 *
 * 'cur' is the address of an actual, in-use cpuset.  Operations
 * such as list traversal that depend on the actual address of the
 * cpuset in the list must use cur below, not trial.
 *
 * 'trial' is the address of bulk structure copy of cur, with
 * perhaps one or more of the fields cpus_allowed, mems_allowed,
 * or flags changed to new, trial values.
 *
 * Return 0 if valid, -errno if not.
 */

static int validate_change(const struct cpuset *cur, const struct cpuset *trial)
{
433
	struct cgroup *cont;
L
Linus Torvalds 已提交
434 435 436
	struct cpuset *c, *par;

	/* Each of our child cpusets must be a subset of us */
437 438
	list_for_each_entry(cont, &cur->css.cgroup->children, sibling) {
		if (!is_cpuset_subset(cgroup_cs(cont), trial))
L
Linus Torvalds 已提交
439 440 441 442
			return -EBUSY;
	}

	/* Remaining checks don't apply to root cpuset */
443
	if (cur == &top_cpuset)
L
Linus Torvalds 已提交
444 445
		return 0;

446 447
	par = cur->parent;

L
Linus Torvalds 已提交
448 449 450 451
	/* We must be a subset of our parent cpuset */
	if (!is_cpuset_subset(trial, par))
		return -EACCES;

452 453 454 455
	/*
	 * If either I or some sibling (!= me) is exclusive, we can't
	 * overlap
	 */
456 457
	list_for_each_entry(cont, &par->css.cgroup->children, sibling) {
		c = cgroup_cs(cont);
L
Linus Torvalds 已提交
458 459 460 461 462 463 464 465 466 467
		if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) &&
		    c != cur &&
		    cpus_intersects(trial->cpus_allowed, c->cpus_allowed))
			return -EINVAL;
		if ((is_mem_exclusive(trial) || is_mem_exclusive(c)) &&
		    c != cur &&
		    nodes_intersects(trial->mems_allowed, c->mems_allowed))
			return -EINVAL;
	}

468 469 470 471 472 473 474 475
	/* Cpusets with tasks can't have empty cpus_allowed or mems_allowed */
	if (cgroup_task_count(cur->css.cgroup)) {
		if (cpus_empty(trial->cpus_allowed) ||
		    nodes_empty(trial->mems_allowed)) {
			return -ENOSPC;
		}
	}

L
Linus Torvalds 已提交
476 477 478
	return 0;
}

P
Paul Jackson 已提交
479
/*
480
 * Helper routine for generate_sched_domains().
P
Paul Jackson 已提交
481 482 483 484 485 486 487
 * Do cpusets a, b have overlapping cpus_allowed masks?
 */
static int cpusets_overlap(struct cpuset *a, struct cpuset *b)
{
	return cpus_intersects(a->cpus_allowed, b->cpus_allowed);
}

488 489 490 491 492 493 494 495
static void
update_domain_attr(struct sched_domain_attr *dattr, struct cpuset *c)
{
	if (dattr->relax_domain_level < c->relax_domain_level)
		dattr->relax_domain_level = c->relax_domain_level;
	return;
}

496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
static void
update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *c)
{
	LIST_HEAD(q);

	list_add(&c->stack_list, &q);
	while (!list_empty(&q)) {
		struct cpuset *cp;
		struct cgroup *cont;
		struct cpuset *child;

		cp = list_first_entry(&q, struct cpuset, stack_list);
		list_del(q.next);

		if (cpus_empty(cp->cpus_allowed))
			continue;

		if (is_sched_load_balance(cp))
			update_domain_attr(dattr, cp);

		list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {
			child = cgroup_cs(cont);
			list_add_tail(&child->stack_list, &q);
		}
	}
}

P
Paul Jackson 已提交
523
/*
524 525 526 527 528 529 530 531 532
 * generate_sched_domains()
 *
 * This function builds a partial partition of the systems CPUs
 * A 'partial partition' is a set of non-overlapping subsets whose
 * union is a subset of that set.
 * The output of this function needs to be passed to kernel/sched.c
 * partition_sched_domains() routine, which will rebuild the scheduler's
 * load balancing domains (sched domains) as specified by that partial
 * partition.
P
Paul Jackson 已提交
533 534 535 536 537 538 539 540 541
 *
 * See "What is sched_load_balance" in Documentation/cpusets.txt
 * for a background explanation of this.
 *
 * Does not return errors, on the theory that the callers of this
 * routine would rather not worry about failures to rebuild sched
 * domains when operating in the severe memory shortage situations
 * that could cause allocation failures below.
 *
542
 * Must be called with cgroup_lock held.
P
Paul Jackson 已提交
543 544
 *
 * The three key local variables below are:
545
 *    q  - a linked-list queue of cpuset pointers, used to implement a
P
Paul Jackson 已提交
546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576
 *	   top-down scan of all cpusets.  This scan loads a pointer
 *	   to each cpuset marked is_sched_load_balance into the
 *	   array 'csa'.  For our purposes, rebuilding the schedulers
 *	   sched domains, we can ignore !is_sched_load_balance cpusets.
 *  csa  - (for CpuSet Array) Array of pointers to all the cpusets
 *	   that need to be load balanced, for convenient iterative
 *	   access by the subsequent code that finds the best partition,
 *	   i.e the set of domains (subsets) of CPUs such that the
 *	   cpus_allowed of every cpuset marked is_sched_load_balance
 *	   is a subset of one of these domains, while there are as
 *	   many such domains as possible, each as small as possible.
 * doms  - Conversion of 'csa' to an array of cpumasks, for passing to
 *	   the kernel/sched.c routine partition_sched_domains() in a
 *	   convenient format, that can be easily compared to the prior
 *	   value to determine what partition elements (sched domains)
 *	   were changed (added or removed.)
 *
 * Finding the best partition (set of domains):
 *	The triple nested loops below over i, j, k scan over the
 *	load balanced cpusets (using the array of cpuset pointers in
 *	csa[]) looking for pairs of cpusets that have overlapping
 *	cpus_allowed, but which don't have the same 'pn' partition
 *	number and gives them in the same partition number.  It keeps
 *	looping on the 'restart' label until it can no longer find
 *	any such pairs.
 *
 *	The union of the cpus_allowed masks from the set of
 *	all cpusets having the same 'pn' value then form the one
 *	element of the partition (one sched domain) to be passed to
 *	partition_sched_domains().
 */
577 578
static int generate_sched_domains(cpumask_t **domains,
			struct sched_domain_attr **attributes)
P
Paul Jackson 已提交
579
{
580
	LIST_HEAD(q);		/* queue of cpusets to be scanned */
P
Paul Jackson 已提交
581 582 583 584 585
	struct cpuset *cp;	/* scans q */
	struct cpuset **csa;	/* array of all cpuset ptrs */
	int csn;		/* how many cpuset ptrs in csa so far */
	int i, j, k;		/* indices for partition finding loops */
	cpumask_t *doms;	/* resulting partition; i.e. sched domains */
586
	struct sched_domain_attr *dattr;  /* attributes for custom domains */
P
Paul Jackson 已提交
587 588 589
	int ndoms;		/* number of sched domains in result */
	int nslot;		/* next empty doms[] cpumask_t slot */

590
	ndoms = 0;
P
Paul Jackson 已提交
591
	doms = NULL;
592
	dattr = NULL;
593
	csa = NULL;
P
Paul Jackson 已提交
594 595 596 597 598

	/* Special case for the 99% of systems with one, full, sched domain */
	if (is_sched_load_balance(&top_cpuset)) {
		doms = kmalloc(sizeof(cpumask_t), GFP_KERNEL);
		if (!doms)
599 600
			goto done;

601 602 603
		dattr = kmalloc(sizeof(struct sched_domain_attr), GFP_KERNEL);
		if (dattr) {
			*dattr = SD_ATTR_INIT;
604
			update_domain_attr_tree(dattr, &top_cpuset);
605
		}
P
Paul Jackson 已提交
606
		*doms = top_cpuset.cpus_allowed;
607 608 609

		ndoms = 1;
		goto done;
P
Paul Jackson 已提交
610 611 612 613 614 615 616
	}

	csa = kmalloc(number_of_cpusets * sizeof(cp), GFP_KERNEL);
	if (!csa)
		goto done;
	csn = 0;

617 618
	list_add(&top_cpuset.stack_list, &q);
	while (!list_empty(&q)) {
P
Paul Jackson 已提交
619 620
		struct cgroup *cont;
		struct cpuset *child;   /* scans child cpusets of cp */
621

622 623 624
		cp = list_first_entry(&q, struct cpuset, stack_list);
		list_del(q.next);

625 626 627
		if (cpus_empty(cp->cpus_allowed))
			continue;

628 629 630 631 632 633 634
		/*
		 * All child cpusets contain a subset of the parent's cpus, so
		 * just skip them, and then we call update_domain_attr_tree()
		 * to calc relax_domain_level of the corresponding sched
		 * domain.
		 */
		if (is_sched_load_balance(cp)) {
P
Paul Jackson 已提交
635
			csa[csn++] = cp;
636 637
			continue;
		}
638

P
Paul Jackson 已提交
639 640
		list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {
			child = cgroup_cs(cont);
641
			list_add_tail(&child->stack_list, &q);
P
Paul Jackson 已提交
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671
		}
  	}

	for (i = 0; i < csn; i++)
		csa[i]->pn = i;
	ndoms = csn;

restart:
	/* Find the best partition (set of sched domains) */
	for (i = 0; i < csn; i++) {
		struct cpuset *a = csa[i];
		int apn = a->pn;

		for (j = 0; j < csn; j++) {
			struct cpuset *b = csa[j];
			int bpn = b->pn;

			if (apn != bpn && cpusets_overlap(a, b)) {
				for (k = 0; k < csn; k++) {
					struct cpuset *c = csa[k];

					if (c->pn == bpn)
						c->pn = apn;
				}
				ndoms--;	/* one less element */
				goto restart;
			}
		}
	}

672 673 674 675
	/*
	 * Now we know how many domains to create.
	 * Convert <csn, csa> to <ndoms, doms> and populate cpu masks.
	 */
P
Paul Jackson 已提交
676
	doms = kmalloc(ndoms * sizeof(cpumask_t), GFP_KERNEL);
677 678 679 680 681 682 683 684 685
	if (!doms) {
		ndoms = 0;
		goto done;
	}

	/*
	 * The rest of the code, including the scheduler, can deal with
	 * dattr==NULL case. No need to abort if alloc fails.
	 */
686
	dattr = kmalloc(ndoms * sizeof(struct sched_domain_attr), GFP_KERNEL);
P
Paul Jackson 已提交
687 688 689

	for (nslot = 0, i = 0; i < csn; i++) {
		struct cpuset *a = csa[i];
690
		cpumask_t *dp;
P
Paul Jackson 已提交
691 692
		int apn = a->pn;

693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708
		if (apn < 0) {
			/* Skip completed partitions */
			continue;
		}

		dp = doms + nslot;

		if (nslot == ndoms) {
			static int warnings = 10;
			if (warnings) {
				printk(KERN_WARNING
				 "rebuild_sched_domains confused:"
				  " nslot %d, ndoms %d, csn %d, i %d,"
				  " apn %d\n",
				  nslot, ndoms, csn, i, apn);
				warnings--;
P
Paul Jackson 已提交
709
			}
710 711
			continue;
		}
P
Paul Jackson 已提交
712

713 714 715 716 717 718 719 720 721 722 723 724 725
		cpus_clear(*dp);
		if (dattr)
			*(dattr + nslot) = SD_ATTR_INIT;
		for (j = i; j < csn; j++) {
			struct cpuset *b = csa[j];

			if (apn == b->pn) {
				cpus_or(*dp, *dp, b->cpus_allowed);
				if (dattr)
					update_domain_attr_tree(dattr + nslot, b);

				/* Done with this partition */
				b->pn = -1;
P
Paul Jackson 已提交
726 727
			}
		}
728
		nslot++;
P
Paul Jackson 已提交
729 730 731
	}
	BUG_ON(nslot != ndoms);

732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755
done:
	kfree(csa);

	*domains    = doms;
	*attributes = dattr;
	return ndoms;
}

/*
 * Rebuild scheduler domains.
 *
 * Call with neither cgroup_mutex held nor within get_online_cpus().
 * Takes both cgroup_mutex and get_online_cpus().
 *
 * Cannot be directly called from cpuset code handling changes
 * to the cpuset pseudo-filesystem, because it cannot be called
 * from code that already holds cgroup_mutex.
 */
static void do_rebuild_sched_domains(struct work_struct *unused)
{
	struct sched_domain_attr *attr;
	cpumask_t *doms;
	int ndoms;

756
	get_online_cpus();
757 758 759 760 761 762 763 764 765

	/* Generate domain masks and attrs */
	cgroup_lock();
	ndoms = generate_sched_domains(&doms, &attr);
	cgroup_unlock();

	/* Have scheduler rebuild the domains */
	partition_sched_domains(ndoms, doms, attr);

766
	put_online_cpus();
767
}
P
Paul Jackson 已提交
768

769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
static DECLARE_WORK(rebuild_sched_domains_work, do_rebuild_sched_domains);

/*
 * Rebuild scheduler domains, asynchronously via workqueue.
 *
 * If the flag 'sched_load_balance' of any cpuset with non-empty
 * 'cpus' changes, or if the 'cpus' allowed changes in any cpuset
 * which has that flag enabled, or if any cpuset with a non-empty
 * 'cpus' is removed, then call this routine to rebuild the
 * scheduler's dynamic sched domains.
 *
 * The rebuild_sched_domains() and partition_sched_domains()
 * routines must nest cgroup_lock() inside get_online_cpus(),
 * but such cpuset changes as these must nest that locking the
 * other way, holding cgroup_lock() for much of the code.
 *
 * So in order to avoid an ABBA deadlock, the cpuset code handling
 * these user changes delegates the actual sched domain rebuilding
 * to a separate workqueue thread, which ends up processing the
 * above do_rebuild_sched_domains() function.
 */
static void async_rebuild_sched_domains(void)
{
	schedule_work(&rebuild_sched_domains_work);
}

/*
 * Accomplishes the same scheduler domain rebuild as the above
 * async_rebuild_sched_domains(), however it directly calls the
 * rebuild routine synchronously rather than calling it via an
 * asynchronous work thread.
 *
 * This can only be called from code that is not holding
 * cgroup_mutex (not nested in a cgroup_lock() call.)
 */
void rebuild_sched_domains(void)
{
	do_rebuild_sched_domains(NULL);
P
Paul Jackson 已提交
807 808
}

C
Cliff Wickman 已提交
809 810 811 812 813
/**
 * cpuset_test_cpumask - test a task's cpus_allowed versus its cpuset's
 * @tsk: task to test
 * @scan: struct cgroup_scanner contained in its struct cpuset_hotplug_scanner
 *
814
 * Call with cgroup_mutex held.  May take callback_mutex during call.
C
Cliff Wickman 已提交
815 816 817
 * Called for each task in a cgroup by cgroup_scan_tasks().
 * Return nonzero if this tasks's cpus_allowed mask should be changed (in other
 * words, if its mask is not equal to its cpuset's mask).
818
 */
819 820
static int cpuset_test_cpumask(struct task_struct *tsk,
			       struct cgroup_scanner *scan)
C
Cliff Wickman 已提交
821 822 823 824
{
	return !cpus_equal(tsk->cpus_allowed,
			(cgroup_cs(scan->cg))->cpus_allowed);
}
825

C
Cliff Wickman 已提交
826 827 828 829 830 831 832 833 834 835 836
/**
 * cpuset_change_cpumask - make a task's cpus_allowed the same as its cpuset's
 * @tsk: task to test
 * @scan: struct cgroup_scanner containing the cgroup of the task
 *
 * Called by cgroup_scan_tasks() for each task in a cgroup whose
 * cpus_allowed mask needs to be changed.
 *
 * We don't need to re-check for the cgroup/cpuset membership, since we're
 * holding cgroup_lock() at this point.
 */
837 838
static void cpuset_change_cpumask(struct task_struct *tsk,
				  struct cgroup_scanner *scan)
C
Cliff Wickman 已提交
839
{
840
	set_cpus_allowed_ptr(tsk, &((cgroup_cs(scan->cg))->cpus_allowed));
C
Cliff Wickman 已提交
841 842
}

843 844 845
/**
 * update_tasks_cpumask - Update the cpumasks of tasks in the cpuset.
 * @cs: the cpuset in which each task's cpus_allowed mask needs to be changed
846
 * @heap: if NULL, defer allocating heap memory to cgroup_scan_tasks()
847 848 849 850 851 852
 *
 * Called with cgroup_mutex held
 *
 * The cgroup_scan_tasks() function will scan all the tasks in a cgroup,
 * calling callback functions for each.
 *
853 854
 * No return value. It's guaranteed that cgroup_scan_tasks() always returns 0
 * if @heap != NULL.
855
 */
856
static void update_tasks_cpumask(struct cpuset *cs, struct ptr_heap *heap)
857 858 859 860 861 862
{
	struct cgroup_scanner scan;

	scan.cg = cs->css.cgroup;
	scan.test_task = cpuset_test_cpumask;
	scan.process_task = cpuset_change_cpumask;
863 864
	scan.heap = heap;
	cgroup_scan_tasks(&scan);
865 866
}

C
Cliff Wickman 已提交
867 868 869 870 871
/**
 * update_cpumask - update the cpus_allowed mask of a cpuset and all tasks in it
 * @cs: the cpuset to consider
 * @buf: buffer of cpu numbers written to this cpuset
 */
872
static int update_cpumask(struct cpuset *cs, const char *buf)
L
Linus Torvalds 已提交
873
{
874
	struct ptr_heap heap;
L
Linus Torvalds 已提交
875
	struct cpuset trialcs;
C
Cliff Wickman 已提交
876 877
	int retval;
	int is_load_balanced;
L
Linus Torvalds 已提交
878

879 880 881 882
	/* top_cpuset.cpus_allowed tracks cpu_online_map; it's read-only */
	if (cs == &top_cpuset)
		return -EACCES;

L
Linus Torvalds 已提交
883
	trialcs = *cs;
884 885

	/*
886
	 * An empty cpus_allowed is ok only if the cpuset has no tasks.
887 888 889
	 * Since cpulist_parse() fails on an empty mask, we special case
	 * that parsing.  The validate_change() call ensures that cpusets
	 * with tasks have cpus.
890
	 */
891
	if (!*buf) {
892 893 894 895 896
		cpus_clear(trialcs.cpus_allowed);
	} else {
		retval = cpulist_parse(buf, trialcs.cpus_allowed);
		if (retval < 0)
			return retval;
897 898 899

		if (!cpus_subset(trialcs.cpus_allowed, cpu_online_map))
			return -EINVAL;
900
	}
L
Linus Torvalds 已提交
901
	retval = validate_change(cs, &trialcs);
902 903
	if (retval < 0)
		return retval;
P
Paul Jackson 已提交
904

P
Paul Menage 已提交
905 906 907
	/* Nothing to do if the cpus didn't change */
	if (cpus_equal(cs->cpus_allowed, trialcs.cpus_allowed))
		return 0;
C
Cliff Wickman 已提交
908

909 910 911 912
	retval = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL);
	if (retval)
		return retval;

P
Paul Jackson 已提交
913 914
	is_load_balanced = is_sched_load_balance(&trialcs);

915
	mutex_lock(&callback_mutex);
916
	cs->cpus_allowed = trialcs.cpus_allowed;
917
	mutex_unlock(&callback_mutex);
P
Paul Jackson 已提交
918

P
Paul Menage 已提交
919 920
	/*
	 * Scan tasks in the cpuset, and update the cpumasks of any
C
Cliff Wickman 已提交
921
	 * that need an update.
P
Paul Menage 已提交
922
	 */
923 924 925
	update_tasks_cpumask(cs, &heap);

	heap_free(&heap);
C
Cliff Wickman 已提交
926

P
Paul Menage 已提交
927
	if (is_load_balanced)
928
		async_rebuild_sched_domains();
929
	return 0;
L
Linus Torvalds 已提交
930 931
}

932 933 934 935 936 937 938 939
/*
 * cpuset_migrate_mm
 *
 *    Migrate memory region from one set of nodes to another.
 *
 *    Temporarilly set tasks mems_allowed to target nodes of migration,
 *    so that the migration code can allocate pages on these nodes.
 *
940
 *    Call holding cgroup_mutex, so current's cpuset won't change
941
 *    during this call, as manage_mutex holds off any cpuset_attach()
942 943
 *    calls.  Therefore we don't need to take task_lock around the
 *    call to guarantee_online_mems(), as we know no one is changing
944
 *    our task's cpuset.
945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976
 *
 *    Hold callback_mutex around the two modifications of our tasks
 *    mems_allowed to synchronize with cpuset_mems_allowed().
 *
 *    While the mm_struct we are migrating is typically from some
 *    other task, the task_struct mems_allowed that we are hacking
 *    is for our current task, which must allocate new pages for that
 *    migrating memory region.
 *
 *    We call cpuset_update_task_memory_state() before hacking
 *    our tasks mems_allowed, so that we are assured of being in
 *    sync with our tasks cpuset, and in particular, callbacks to
 *    cpuset_update_task_memory_state() from nested page allocations
 *    won't see any mismatch of our cpuset and task mems_generation
 *    values, so won't overwrite our hacked tasks mems_allowed
 *    nodemask.
 */

static void cpuset_migrate_mm(struct mm_struct *mm, const nodemask_t *from,
							const nodemask_t *to)
{
	struct task_struct *tsk = current;

	cpuset_update_task_memory_state();

	mutex_lock(&callback_mutex);
	tsk->mems_allowed = *to;
	mutex_unlock(&callback_mutex);

	do_migrate_pages(mm, from, to, MPOL_MF_MOVE_ALL);

	mutex_lock(&callback_mutex);
977
	guarantee_online_mems(task_cs(tsk),&tsk->mems_allowed);
978 979 980
	mutex_unlock(&callback_mutex);
}

981 982
static void *cpuset_being_rebound;

983 984 985 986 987 988 989 990 991
/**
 * update_tasks_nodemask - Update the nodemasks of tasks in the cpuset.
 * @cs: the cpuset in which each task's mems_allowed mask needs to be changed
 * @oldmem: old mems_allowed of cpuset cs
 *
 * Called with cgroup_mutex held
 * Return 0 if successful, -errno if not.
 */
static int update_tasks_nodemask(struct cpuset *cs, const nodemask_t *oldmem)
L
Linus Torvalds 已提交
992
{
993
	struct task_struct *p;
994 995
	struct mm_struct **mmarray;
	int i, n, ntasks;
996
	int migrate;
997
	int fudge;
998
	struct cgroup_iter it;
999
	int retval;
1000

1001
	cpuset_being_rebound = cs;		/* causes mpol_dup() rebind */
1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014

	fudge = 10;				/* spare mmarray[] slots */
	fudge += cpus_weight(cs->cpus_allowed);	/* imagine one fork-bomb/cpu */
	retval = -ENOMEM;

	/*
	 * Allocate mmarray[] to hold mm reference for each task
	 * in cpuset cs.  Can't kmalloc GFP_KERNEL while holding
	 * tasklist_lock.  We could use GFP_ATOMIC, but with a
	 * few more lines of code, we can retry until we get a big
	 * enough mmarray[] w/o using GFP_ATOMIC.
	 */
	while (1) {
1015
		ntasks = cgroup_task_count(cs->css.cgroup);  /* guess */
1016 1017 1018 1019
		ntasks += fudge;
		mmarray = kmalloc(ntasks * sizeof(*mmarray), GFP_KERNEL);
		if (!mmarray)
			goto done;
1020
		read_lock(&tasklist_lock);		/* block fork */
1021
		if (cgroup_task_count(cs->css.cgroup) <= ntasks)
1022
			break;				/* got enough */
1023
		read_unlock(&tasklist_lock);		/* try again */
1024 1025 1026 1027 1028 1029
		kfree(mmarray);
	}

	n = 0;

	/* Load up mmarray[] with mm reference for each task in cpuset. */
1030 1031
	cgroup_iter_start(cs->css.cgroup, &it);
	while ((p = cgroup_iter_next(cs->css.cgroup, &it))) {
1032 1033 1034 1035 1036
		struct mm_struct *mm;

		if (n >= ntasks) {
			printk(KERN_WARNING
				"Cpuset mempolicy rebind incomplete.\n");
1037
			break;
1038 1039 1040 1041 1042
		}
		mm = get_task_mm(p);
		if (!mm)
			continue;
		mmarray[n++] = mm;
1043 1044
	}
	cgroup_iter_end(cs->css.cgroup, &it);
1045
	read_unlock(&tasklist_lock);
1046 1047 1048 1049 1050 1051

	/*
	 * Now that we've dropped the tasklist spinlock, we can
	 * rebind the vma mempolicies of each mm in mmarray[] to their
	 * new cpuset, and release that mm.  The mpol_rebind_mm()
	 * call takes mmap_sem, which we couldn't take while holding
1052
	 * tasklist_lock.  Forks can happen again now - the mpol_dup()
1053 1054
	 * cpuset_being_rebound check will catch such forks, and rebind
	 * their vma mempolicies too.  Because we still hold the global
1055
	 * cgroup_mutex, we know that no other rebind effort will
1056 1057
	 * be contending for the global variable cpuset_being_rebound.
	 * It's ok if we rebind the same mm twice; mpol_rebind_mm()
1058
	 * is idempotent.  Also migrate pages in each mm to new nodes.
1059
	 */
1060
	migrate = is_memory_migrate(cs);
1061 1062 1063 1064
	for (i = 0; i < n; i++) {
		struct mm_struct *mm = mmarray[i];

		mpol_rebind_mm(mm, &cs->mems_allowed);
1065
		if (migrate)
1066
			cpuset_migrate_mm(mm, oldmem, &cs->mems_allowed);
1067 1068 1069
		mmput(mm);
	}

1070
	/* We're done rebinding vmas to this cpuset's new mems_allowed. */
1071
	kfree(mmarray);
1072
	cpuset_being_rebound = NULL;
1073
	retval = 0;
1074
done:
L
Linus Torvalds 已提交
1075 1076 1077
	return retval;
}

1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
/*
 * Handle user request to change the 'mems' memory placement
 * of a cpuset.  Needs to validate the request, update the
 * cpusets mems_allowed and mems_generation, and for each
 * task in the cpuset, rebind any vma mempolicies and if
 * the cpuset is marked 'memory_migrate', migrate the tasks
 * pages to the new memory.
 *
 * Call with cgroup_mutex held.  May take callback_mutex during call.
 * Will take tasklist_lock, scan tasklist for tasks in cpuset cs,
 * lock each such tasks mm->mmap_sem, scan its vma's and rebind
 * their mempolicies to the cpusets new mems_allowed.
 */
static int update_nodemask(struct cpuset *cs, const char *buf)
{
	struct cpuset trialcs;
	nodemask_t oldmem;
	int retval;

	/*
	 * top_cpuset.mems_allowed tracks node_stats[N_HIGH_MEMORY];
	 * it's read-only
	 */
	if (cs == &top_cpuset)
		return -EACCES;

	trialcs = *cs;

	/*
	 * An empty mems_allowed is ok iff there are no tasks in the cpuset.
	 * Since nodelist_parse() fails on an empty mask, we special case
	 * that parsing.  The validate_change() call ensures that cpusets
	 * with tasks have memory.
	 */
	if (!*buf) {
		nodes_clear(trialcs.mems_allowed);
	} else {
		retval = nodelist_parse(buf, trialcs.mems_allowed);
		if (retval < 0)
			goto done;

		if (!nodes_subset(trialcs.mems_allowed,
				node_states[N_HIGH_MEMORY]))
			return -EINVAL;
	}
	oldmem = cs->mems_allowed;
	if (nodes_equal(oldmem, trialcs.mems_allowed)) {
		retval = 0;		/* Too easy - nothing to do */
		goto done;
	}
	retval = validate_change(cs, &trialcs);
	if (retval < 0)
		goto done;

	mutex_lock(&callback_mutex);
	cs->mems_allowed = trialcs.mems_allowed;
	cs->mems_generation = cpuset_mems_generation++;
	mutex_unlock(&callback_mutex);

	retval = update_tasks_nodemask(cs, &oldmem);
done:
	return retval;
}

1142 1143 1144 1145 1146
int current_cpuset_is_being_rebound(void)
{
	return task_cs(current) == cpuset_being_rebound;
}

1147
static int update_relax_domain_level(struct cpuset *cs, s64 val)
1148
{
1149 1150
	if (val < -1 || val >= SD_LV_MAX)
		return -EINVAL;
1151 1152 1153

	if (val != cs->relax_domain_level) {
		cs->relax_domain_level = val;
1154
		if (!cpus_empty(cs->cpus_allowed) && is_sched_load_balance(cs))
1155
			async_rebuild_sched_domains();
1156 1157 1158 1159 1160
	}

	return 0;
}

L
Linus Torvalds 已提交
1161 1162
/*
 * update_flag - read a 0 or a 1 in a file and update associated flag
1163 1164 1165
 * bit:		the bit to update (see cpuset_flagbits_t)
 * cs:		the cpuset to update
 * turning_on: 	whether the flag is being set or cleared
1166
 *
1167
 * Call with cgroup_mutex held.
L
Linus Torvalds 已提交
1168 1169
 */

1170 1171
static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs,
		       int turning_on)
L
Linus Torvalds 已提交
1172 1173
{
	struct cpuset trialcs;
1174
	int err;
R
Rakib Mullick 已提交
1175
	int balance_flag_changed;
L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181 1182 1183

	trialcs = *cs;
	if (turning_on)
		set_bit(bit, &trialcs.flags);
	else
		clear_bit(bit, &trialcs.flags);

	err = validate_change(cs, &trialcs);
1184 1185
	if (err < 0)
		return err;
P
Paul Jackson 已提交
1186 1187 1188 1189

	balance_flag_changed = (is_sched_load_balance(cs) !=
		 			is_sched_load_balance(&trialcs));

1190
	mutex_lock(&callback_mutex);
1191
	cs->flags = trialcs.flags;
1192
	mutex_unlock(&callback_mutex);
1193

R
Rakib Mullick 已提交
1194
	if (!cpus_empty(trialcs.cpus_allowed) && balance_flag_changed)
1195
		async_rebuild_sched_domains();
P
Paul Jackson 已提交
1196

1197
	return 0;
L
Linus Torvalds 已提交
1198 1199
}

1200
/*
A
Adrian Bunk 已提交
1201
 * Frequency meter - How fast is some event occurring?
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 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
 *
 * These routines manage a digitally filtered, constant time based,
 * event frequency meter.  There are four routines:
 *   fmeter_init() - initialize a frequency meter.
 *   fmeter_markevent() - called each time the event happens.
 *   fmeter_getrate() - returns the recent rate of such events.
 *   fmeter_update() - internal routine used to update fmeter.
 *
 * A common data structure is passed to each of these routines,
 * which is used to keep track of the state required to manage the
 * frequency meter and its digital filter.
 *
 * The filter works on the number of events marked per unit time.
 * The filter is single-pole low-pass recursive (IIR).  The time unit
 * is 1 second.  Arithmetic is done using 32-bit integers scaled to
 * simulate 3 decimal digits of precision (multiplied by 1000).
 *
 * With an FM_COEF of 933, and a time base of 1 second, the filter
 * has a half-life of 10 seconds, meaning that if the events quit
 * happening, then the rate returned from the fmeter_getrate()
 * will be cut in half each 10 seconds, until it converges to zero.
 *
 * It is not worth doing a real infinitely recursive filter.  If more
 * than FM_MAXTICKS ticks have elapsed since the last filter event,
 * just compute FM_MAXTICKS ticks worth, by which point the level
 * will be stable.
 *
 * Limit the count of unprocessed events to FM_MAXCNT, so as to avoid
 * arithmetic overflow in the fmeter_update() routine.
 *
 * Given the simple 32 bit integer arithmetic used, this meter works
 * best for reporting rates between one per millisecond (msec) and
 * one per 32 (approx) seconds.  At constant rates faster than one
 * per msec it maxes out at values just under 1,000,000.  At constant
 * rates between one per msec, and one per second it will stabilize
 * to a value N*1000, where N is the rate of events per second.
 * At constant rates between one per second and one per 32 seconds,
 * it will be choppy, moving up on the seconds that have an event,
 * and then decaying until the next event.  At rates slower than
 * about one in 32 seconds, it decays all the way back to zero between
 * each event.
 */

#define FM_COEF 933		/* coefficient for half-life of 10 secs */
#define FM_MAXTICKS ((time_t)99) /* useless computing more ticks than this */
#define FM_MAXCNT 1000000	/* limit cnt to avoid overflow */
#define FM_SCALE 1000		/* faux fixed point scale */

/* Initialize a frequency meter */
static void fmeter_init(struct fmeter *fmp)
{
	fmp->cnt = 0;
	fmp->val = 0;
	fmp->time = 0;
	spin_lock_init(&fmp->lock);
}

/* Internal meter update - process cnt events and update value */
static void fmeter_update(struct fmeter *fmp)
{
	time_t now = get_seconds();
	time_t ticks = now - fmp->time;

	if (ticks == 0)
		return;

	ticks = min(FM_MAXTICKS, ticks);
	while (ticks-- > 0)
		fmp->val = (FM_COEF * fmp->val) / FM_SCALE;
	fmp->time = now;

	fmp->val += ((FM_SCALE - FM_COEF) * fmp->cnt) / FM_SCALE;
	fmp->cnt = 0;
}

/* Process any previous ticks, then bump cnt by one (times scale). */
static void fmeter_markevent(struct fmeter *fmp)
{
	spin_lock(&fmp->lock);
	fmeter_update(fmp);
	fmp->cnt = min(FM_MAXCNT, fmp->cnt + FM_SCALE);
	spin_unlock(&fmp->lock);
}

/* Process any previous ticks, then return current value. */
static int fmeter_getrate(struct fmeter *fmp)
{
	int val;

	spin_lock(&fmp->lock);
	fmeter_update(fmp);
	val = fmp->val;
	spin_unlock(&fmp->lock);
	return val;
}

1298
/* Called by cgroups to determine if a cpuset is usable; cgroup_mutex held */
1299 1300
static int cpuset_can_attach(struct cgroup_subsys *ss,
			     struct cgroup *cont, struct task_struct *tsk)
L
Linus Torvalds 已提交
1301
{
1302
	struct cpuset *cs = cgroup_cs(cont);
L
Linus Torvalds 已提交
1303 1304 1305

	if (cpus_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed))
		return -ENOSPC;
1306 1307 1308 1309 1310 1311 1312 1313 1314
	if (tsk->flags & PF_THREAD_BOUND) {
		cpumask_t mask;

		mutex_lock(&callback_mutex);
		mask = cs->cpus_allowed;
		mutex_unlock(&callback_mutex);
		if (!cpus_equal(tsk->cpus_allowed, mask))
			return -EINVAL;
	}
L
Linus Torvalds 已提交
1315

1316 1317
	return security_task_setscheduler(tsk, 0, NULL);
}
L
Linus Torvalds 已提交
1318

1319 1320 1321 1322 1323 1324 1325 1326 1327
static void cpuset_attach(struct cgroup_subsys *ss,
			  struct cgroup *cont, struct cgroup *oldcont,
			  struct task_struct *tsk)
{
	cpumask_t cpus;
	nodemask_t from, to;
	struct mm_struct *mm;
	struct cpuset *cs = cgroup_cs(cont);
	struct cpuset *oldcs = cgroup_cs(oldcont);
1328
	int err;
1329

1330
	mutex_lock(&callback_mutex);
L
Linus Torvalds 已提交
1331
	guarantee_online_cpus(cs, &cpus);
1332
	err = set_cpus_allowed_ptr(tsk, &cpus);
1333
	mutex_unlock(&callback_mutex);
1334 1335
	if (err)
		return;
L
Linus Torvalds 已提交
1336

1337 1338
	from = oldcs->mems_allowed;
	to = cs->mems_allowed;
1339 1340 1341
	mm = get_task_mm(tsk);
	if (mm) {
		mpol_rebind_mm(mm, &to);
1342
		if (is_memory_migrate(cs))
1343
			cpuset_migrate_mm(mm, &from, &to);
1344 1345 1346
		mmput(mm);
	}

L
Linus Torvalds 已提交
1347 1348 1349 1350 1351
}

/* The various types of files and directories in a cpuset file system */

typedef enum {
1352
	FILE_MEMORY_MIGRATE,
L
Linus Torvalds 已提交
1353 1354 1355 1356
	FILE_CPULIST,
	FILE_MEMLIST,
	FILE_CPU_EXCLUSIVE,
	FILE_MEM_EXCLUSIVE,
1357
	FILE_MEM_HARDWALL,
P
Paul Jackson 已提交
1358
	FILE_SCHED_LOAD_BALANCE,
1359
	FILE_SCHED_RELAX_DOMAIN_LEVEL,
1360 1361
	FILE_MEMORY_PRESSURE_ENABLED,
	FILE_MEMORY_PRESSURE,
1362 1363
	FILE_SPREAD_PAGE,
	FILE_SPREAD_SLAB,
L
Linus Torvalds 已提交
1364 1365
} cpuset_filetype_t;

1366 1367 1368 1369 1370 1371
static int cpuset_write_u64(struct cgroup *cgrp, struct cftype *cft, u64 val)
{
	int retval = 0;
	struct cpuset *cs = cgroup_cs(cgrp);
	cpuset_filetype_t type = cft->private;

1372
	if (!cgroup_lock_live_group(cgrp))
1373 1374 1375
		return -ENODEV;

	switch (type) {
L
Linus Torvalds 已提交
1376
	case FILE_CPU_EXCLUSIVE:
1377
		retval = update_flag(CS_CPU_EXCLUSIVE, cs, val);
L
Linus Torvalds 已提交
1378 1379
		break;
	case FILE_MEM_EXCLUSIVE:
1380
		retval = update_flag(CS_MEM_EXCLUSIVE, cs, val);
L
Linus Torvalds 已提交
1381
		break;
1382 1383 1384
	case FILE_MEM_HARDWALL:
		retval = update_flag(CS_MEM_HARDWALL, cs, val);
		break;
P
Paul Jackson 已提交
1385
	case FILE_SCHED_LOAD_BALANCE:
1386
		retval = update_flag(CS_SCHED_LOAD_BALANCE, cs, val);
1387
		break;
1388
	case FILE_MEMORY_MIGRATE:
1389
		retval = update_flag(CS_MEMORY_MIGRATE, cs, val);
1390
		break;
1391
	case FILE_MEMORY_PRESSURE_ENABLED:
1392
		cpuset_memory_pressure_enabled = !!val;
1393 1394 1395 1396
		break;
	case FILE_MEMORY_PRESSURE:
		retval = -EACCES;
		break;
1397
	case FILE_SPREAD_PAGE:
1398
		retval = update_flag(CS_SPREAD_PAGE, cs, val);
1399
		cs->mems_generation = cpuset_mems_generation++;
1400 1401
		break;
	case FILE_SPREAD_SLAB:
1402
		retval = update_flag(CS_SPREAD_SLAB, cs, val);
1403
		cs->mems_generation = cpuset_mems_generation++;
1404
		break;
L
Linus Torvalds 已提交
1405 1406
	default:
		retval = -EINVAL;
1407
		break;
L
Linus Torvalds 已提交
1408
	}
1409
	cgroup_unlock();
L
Linus Torvalds 已提交
1410 1411 1412
	return retval;
}

1413 1414 1415 1416 1417 1418
static int cpuset_write_s64(struct cgroup *cgrp, struct cftype *cft, s64 val)
{
	int retval = 0;
	struct cpuset *cs = cgroup_cs(cgrp);
	cpuset_filetype_t type = cft->private;

1419
	if (!cgroup_lock_live_group(cgrp))
1420
		return -ENODEV;
1421

1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433
	switch (type) {
	case FILE_SCHED_RELAX_DOMAIN_LEVEL:
		retval = update_relax_domain_level(cs, val);
		break;
	default:
		retval = -EINVAL;
		break;
	}
	cgroup_unlock();
	return retval;
}

1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
/*
 * Common handling for a write to a "cpus" or "mems" file.
 */
static int cpuset_write_resmask(struct cgroup *cgrp, struct cftype *cft,
				const char *buf)
{
	int retval = 0;

	if (!cgroup_lock_live_group(cgrp))
		return -ENODEV;

	switch (cft->private) {
	case FILE_CPULIST:
		retval = update_cpumask(cgroup_cs(cgrp), buf);
		break;
	case FILE_MEMLIST:
		retval = update_nodemask(cgroup_cs(cgrp), buf);
		break;
	default:
		retval = -EINVAL;
		break;
	}
	cgroup_unlock();
	return retval;
}

L
Linus Torvalds 已提交
1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475
/*
 * These ascii lists should be read in a single call, by using a user
 * buffer large enough to hold the entire map.  If read in smaller
 * chunks, there is no guarantee of atomicity.  Since the display format
 * used, list of ranges of sequential numbers, is variable length,
 * and since these maps can change value dynamically, one could read
 * gibberish by doing partial reads while a list was changing.
 * A single large read to a buffer that crosses a page boundary is
 * ok, because the result being copied to user land is not recomputed
 * across a page fault.
 */

static int cpuset_sprintf_cpulist(char *page, struct cpuset *cs)
{
	cpumask_t mask;

1476
	mutex_lock(&callback_mutex);
L
Linus Torvalds 已提交
1477
	mask = cs->cpus_allowed;
1478
	mutex_unlock(&callback_mutex);
L
Linus Torvalds 已提交
1479 1480 1481 1482 1483 1484 1485 1486

	return cpulist_scnprintf(page, PAGE_SIZE, mask);
}

static int cpuset_sprintf_memlist(char *page, struct cpuset *cs)
{
	nodemask_t mask;

1487
	mutex_lock(&callback_mutex);
L
Linus Torvalds 已提交
1488
	mask = cs->mems_allowed;
1489
	mutex_unlock(&callback_mutex);
L
Linus Torvalds 已提交
1490 1491 1492 1493

	return nodelist_scnprintf(page, PAGE_SIZE, mask);
}

1494 1495 1496 1497 1498
static ssize_t cpuset_common_file_read(struct cgroup *cont,
				       struct cftype *cft,
				       struct file *file,
				       char __user *buf,
				       size_t nbytes, loff_t *ppos)
L
Linus Torvalds 已提交
1499
{
1500
	struct cpuset *cs = cgroup_cs(cont);
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505
	cpuset_filetype_t type = cft->private;
	char *page;
	ssize_t retval = 0;
	char *s;

1506
	if (!(page = (char *)__get_free_page(GFP_TEMPORARY)))
L
Linus Torvalds 已提交
1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523
		return -ENOMEM;

	s = page;

	switch (type) {
	case FILE_CPULIST:
		s += cpuset_sprintf_cpulist(s, cs);
		break;
	case FILE_MEMLIST:
		s += cpuset_sprintf_memlist(s, cs);
		break;
	default:
		retval = -EINVAL;
		goto out;
	}
	*s++ = '\n';

A
Al Viro 已提交
1524
	retval = simple_read_from_buffer(buf, nbytes, ppos, page, s - page);
L
Linus Torvalds 已提交
1525 1526 1527 1528 1529
out:
	free_page((unsigned long)page);
	return retval;
}

1530 1531 1532 1533 1534 1535 1536 1537 1538
static u64 cpuset_read_u64(struct cgroup *cont, struct cftype *cft)
{
	struct cpuset *cs = cgroup_cs(cont);
	cpuset_filetype_t type = cft->private;
	switch (type) {
	case FILE_CPU_EXCLUSIVE:
		return is_cpu_exclusive(cs);
	case FILE_MEM_EXCLUSIVE:
		return is_mem_exclusive(cs);
1539 1540
	case FILE_MEM_HARDWALL:
		return is_mem_hardwall(cs);
1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555
	case FILE_SCHED_LOAD_BALANCE:
		return is_sched_load_balance(cs);
	case FILE_MEMORY_MIGRATE:
		return is_memory_migrate(cs);
	case FILE_MEMORY_PRESSURE_ENABLED:
		return cpuset_memory_pressure_enabled;
	case FILE_MEMORY_PRESSURE:
		return fmeter_getrate(&cs->fmeter);
	case FILE_SPREAD_PAGE:
		return is_spread_page(cs);
	case FILE_SPREAD_SLAB:
		return is_spread_slab(cs);
	default:
		BUG();
	}
1556 1557 1558

	/* Unreachable but makes gcc happy */
	return 0;
1559
}
L
Linus Torvalds 已提交
1560

1561 1562 1563 1564 1565 1566 1567 1568 1569 1570
static s64 cpuset_read_s64(struct cgroup *cont, struct cftype *cft)
{
	struct cpuset *cs = cgroup_cs(cont);
	cpuset_filetype_t type = cft->private;
	switch (type) {
	case FILE_SCHED_RELAX_DOMAIN_LEVEL:
		return cs->relax_domain_level;
	default:
		BUG();
	}
1571 1572 1573

	/* Unrechable but makes gcc happy */
	return 0;
1574 1575
}

L
Linus Torvalds 已提交
1576 1577 1578 1579 1580

/*
 * for the common functions, 'private' gives the type of file
 */

1581 1582 1583 1584
static struct cftype files[] = {
	{
		.name = "cpus",
		.read = cpuset_common_file_read,
1585 1586
		.write_string = cpuset_write_resmask,
		.max_write_len = (100U + 6 * NR_CPUS),
1587 1588 1589 1590 1591 1592
		.private = FILE_CPULIST,
	},

	{
		.name = "mems",
		.read = cpuset_common_file_read,
1593 1594
		.write_string = cpuset_write_resmask,
		.max_write_len = (100U + 6 * MAX_NUMNODES),
1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
		.private = FILE_MEMLIST,
	},

	{
		.name = "cpu_exclusive",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_CPU_EXCLUSIVE,
	},

	{
		.name = "mem_exclusive",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_MEM_EXCLUSIVE,
	},

1612 1613 1614 1615 1616 1617 1618
	{
		.name = "mem_hardwall",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_MEM_HARDWALL,
	},

1619 1620 1621 1622 1623 1624 1625 1626 1627
	{
		.name = "sched_load_balance",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_SCHED_LOAD_BALANCE,
	},

	{
		.name = "sched_relax_domain_level",
1628 1629
		.read_s64 = cpuset_read_s64,
		.write_s64 = cpuset_write_s64,
1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659
		.private = FILE_SCHED_RELAX_DOMAIN_LEVEL,
	},

	{
		.name = "memory_migrate",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_MEMORY_MIGRATE,
	},

	{
		.name = "memory_pressure",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_MEMORY_PRESSURE,
	},

	{
		.name = "memory_spread_page",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_SPREAD_PAGE,
	},

	{
		.name = "memory_spread_slab",
		.read_u64 = cpuset_read_u64,
		.write_u64 = cpuset_write_u64,
		.private = FILE_SPREAD_SLAB,
	},
1660 1661
};

1662 1663
static struct cftype cft_memory_pressure_enabled = {
	.name = "memory_pressure_enabled",
1664 1665
	.read_u64 = cpuset_read_u64,
	.write_u64 = cpuset_write_u64,
1666 1667 1668
	.private = FILE_MEMORY_PRESSURE_ENABLED,
};

1669
static int cpuset_populate(struct cgroup_subsys *ss, struct cgroup *cont)
L
Linus Torvalds 已提交
1670 1671 1672
{
	int err;

1673 1674
	err = cgroup_add_files(cont, ss, files, ARRAY_SIZE(files));
	if (err)
L
Linus Torvalds 已提交
1675
		return err;
1676
	/* memory_pressure_enabled is in root cpuset only */
1677
	if (!cont->parent)
1678
		err = cgroup_add_file(cont, ss,
1679 1680
				      &cft_memory_pressure_enabled);
	return err;
L
Linus Torvalds 已提交
1681 1682
}

1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696
/*
 * post_clone() is called at the end of cgroup_clone().
 * 'cgroup' was just created automatically as a result of
 * a cgroup_clone(), and the current task is about to
 * be moved into 'cgroup'.
 *
 * Currently we refuse to set up the cgroup - thereby
 * refusing the task to be entered, and as a result refusing
 * the sys_unshare() or clone() which initiated it - if any
 * sibling cpusets have exclusive cpus or mem.
 *
 * If this becomes a problem for some users who wish to
 * allow that scenario, then cpuset_post_clone() could be
 * changed to grant parent->cpus_allowed-sibling_cpus_exclusive
1697 1698
 * (and likewise for mems) to the new cgroup. Called with cgroup_mutex
 * held.
1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719
 */
static void cpuset_post_clone(struct cgroup_subsys *ss,
			      struct cgroup *cgroup)
{
	struct cgroup *parent, *child;
	struct cpuset *cs, *parent_cs;

	parent = cgroup->parent;
	list_for_each_entry(child, &parent->children, sibling) {
		cs = cgroup_cs(child);
		if (is_mem_exclusive(cs) || is_cpu_exclusive(cs))
			return;
	}
	cs = cgroup_cs(cgroup);
	parent_cs = cgroup_cs(parent);

	cs->mems_allowed = parent_cs->mems_allowed;
	cs->cpus_allowed = parent_cs->cpus_allowed;
	return;
}

L
Linus Torvalds 已提交
1720 1721
/*
 *	cpuset_create - create a cpuset
1722 1723
 *	ss:	cpuset cgroup subsystem
 *	cont:	control group that the new cpuset will be part of
L
Linus Torvalds 已提交
1724 1725
 */

1726 1727 1728
static struct cgroup_subsys_state *cpuset_create(
	struct cgroup_subsys *ss,
	struct cgroup *cont)
L
Linus Torvalds 已提交
1729 1730
{
	struct cpuset *cs;
1731
	struct cpuset *parent;
L
Linus Torvalds 已提交
1732

1733 1734 1735 1736 1737 1738
	if (!cont->parent) {
		/* This is early initialization for the top cgroup */
		top_cpuset.mems_generation = cpuset_mems_generation++;
		return &top_cpuset.css;
	}
	parent = cgroup_cs(cont->parent);
L
Linus Torvalds 已提交
1739 1740
	cs = kmalloc(sizeof(*cs), GFP_KERNEL);
	if (!cs)
1741
		return ERR_PTR(-ENOMEM);
L
Linus Torvalds 已提交
1742

1743
	cpuset_update_task_memory_state();
L
Linus Torvalds 已提交
1744
	cs->flags = 0;
1745 1746 1747 1748
	if (is_spread_page(parent))
		set_bit(CS_SPREAD_PAGE, &cs->flags);
	if (is_spread_slab(parent))
		set_bit(CS_SPREAD_SLAB, &cs->flags);
P
Paul Jackson 已提交
1749
	set_bit(CS_SCHED_LOAD_BALANCE, &cs->flags);
1750 1751
	cpus_clear(cs->cpus_allowed);
	nodes_clear(cs->mems_allowed);
1752
	cs->mems_generation = cpuset_mems_generation++;
1753
	fmeter_init(&cs->fmeter);
1754
	cs->relax_domain_level = -1;
L
Linus Torvalds 已提交
1755 1756

	cs->parent = parent;
1757
	number_of_cpusets++;
1758
	return &cs->css ;
L
Linus Torvalds 已提交
1759 1760
}

P
Paul Jackson 已提交
1761 1762 1763
/*
 * If the cpuset being removed has its flag 'sched_load_balance'
 * enabled, then simulate turning sched_load_balance off, which
1764
 * will call async_rebuild_sched_domains().
P
Paul Jackson 已提交
1765 1766
 */

1767
static void cpuset_destroy(struct cgroup_subsys *ss, struct cgroup *cont)
L
Linus Torvalds 已提交
1768
{
1769
	struct cpuset *cs = cgroup_cs(cont);
L
Linus Torvalds 已提交
1770

1771
	cpuset_update_task_memory_state();
P
Paul Jackson 已提交
1772 1773

	if (is_sched_load_balance(cs))
1774
		update_flag(CS_SCHED_LOAD_BALANCE, cs, 0);
P
Paul Jackson 已提交
1775

1776
	number_of_cpusets--;
1777
	kfree(cs);
L
Linus Torvalds 已提交
1778 1779
}

1780 1781 1782
struct cgroup_subsys cpuset_subsys = {
	.name = "cpuset",
	.create = cpuset_create,
1783
	.destroy = cpuset_destroy,
1784 1785 1786 1787 1788 1789 1790 1791
	.can_attach = cpuset_can_attach,
	.attach = cpuset_attach,
	.populate = cpuset_populate,
	.post_clone = cpuset_post_clone,
	.subsys_id = cpuset_subsys_id,
	.early_init = 1,
};

1792 1793 1794 1795 1796 1797 1798 1799
/*
 * cpuset_init_early - just enough so that the calls to
 * cpuset_update_task_memory_state() in early init code
 * are harmless.
 */

int __init cpuset_init_early(void)
{
1800
	top_cpuset.mems_generation = cpuset_mems_generation++;
1801 1802 1803
	return 0;
}

1804

L
Linus Torvalds 已提交
1805 1806 1807 1808 1809 1810 1811 1812
/**
 * cpuset_init - initialize cpusets at system boot
 *
 * Description: Initialize top_cpuset and the cpuset internal file system,
 **/

int __init cpuset_init(void)
{
1813
	int err = 0;
L
Linus Torvalds 已提交
1814

1815 1816
	cpus_setall(top_cpuset.cpus_allowed);
	nodes_setall(top_cpuset.mems_allowed);
L
Linus Torvalds 已提交
1817

1818
	fmeter_init(&top_cpuset.fmeter);
1819
	top_cpuset.mems_generation = cpuset_mems_generation++;
P
Paul Jackson 已提交
1820
	set_bit(CS_SCHED_LOAD_BALANCE, &top_cpuset.flags);
1821
	top_cpuset.relax_domain_level = -1;
L
Linus Torvalds 已提交
1822 1823 1824

	err = register_filesystem(&cpuset_fs_type);
	if (err < 0)
1825 1826
		return err;

1827
	number_of_cpusets = 1;
1828
	return 0;
L
Linus Torvalds 已提交
1829 1830
}

1831 1832 1833 1834 1835 1836 1837 1838
/**
 * cpuset_do_move_task - move a given task to another cpuset
 * @tsk: pointer to task_struct the task to move
 * @scan: struct cgroup_scanner contained in its struct cpuset_hotplug_scanner
 *
 * Called by cgroup_scan_tasks() for each task in a cgroup.
 * Return nonzero to stop the walk through the tasks.
 */
1839 1840
static void cpuset_do_move_task(struct task_struct *tsk,
				struct cgroup_scanner *scan)
1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
{
	struct cpuset_hotplug_scanner *chsp;

	chsp = container_of(scan, struct cpuset_hotplug_scanner, scan);
	cgroup_attach_task(chsp->to, tsk);
}

/**
 * move_member_tasks_to_cpuset - move tasks from one cpuset to another
 * @from: cpuset in which the tasks currently reside
 * @to: cpuset to which the tasks will be moved
 *
1853 1854
 * Called with cgroup_mutex held
 * callback_mutex must not be held, as cpuset_attach() will take it.
1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868
 *
 * The cgroup_scan_tasks() function will scan all the tasks in a cgroup,
 * calling callback functions for each.
 */
static void move_member_tasks_to_cpuset(struct cpuset *from, struct cpuset *to)
{
	struct cpuset_hotplug_scanner scan;

	scan.scan.cg = from->css.cgroup;
	scan.scan.test_task = NULL; /* select all tasks in cgroup */
	scan.scan.process_task = cpuset_do_move_task;
	scan.scan.heap = NULL;
	scan.to = to->css.cgroup;

L
Lai Jiangshan 已提交
1869
	if (cgroup_scan_tasks(&scan.scan))
1870 1871 1872 1873
		printk(KERN_ERR "move_member_tasks_to_cpuset: "
				"cgroup_scan_tasks failed\n");
}

1874
/*
1875
 * If CPU and/or memory hotplug handlers, below, unplug any CPUs
1876 1877
 * or memory nodes, we need to walk over the cpuset hierarchy,
 * removing that CPU or node from all cpusets.  If this removes the
1878 1879
 * last CPU or node from a cpuset, then move the tasks in the empty
 * cpuset to its next-highest non-empty parent.
1880
 *
1881 1882
 * Called with cgroup_mutex held
 * callback_mutex must not be held, as cpuset_attach() will take it.
1883
 */
1884 1885 1886 1887
static void remove_tasks_in_empty_cpuset(struct cpuset *cs)
{
	struct cpuset *parent;

1888 1889 1890 1891 1892
	/*
	 * The cgroup's css_sets list is in use if there are tasks
	 * in the cpuset; the list is empty if there are none;
	 * the cs->css.refcnt seems always 0.
	 */
1893 1894
	if (list_empty(&cs->css.cgroup->css_sets))
		return;
1895

1896 1897 1898 1899 1900
	/*
	 * Find its next-highest non-empty parent, (top cpuset
	 * has online cpus, so can't be empty).
	 */
	parent = cs->parent;
1901 1902
	while (cpus_empty(parent->cpus_allowed) ||
			nodes_empty(parent->mems_allowed))
1903 1904 1905 1906 1907 1908 1909 1910 1911
		parent = parent->parent;

	move_member_tasks_to_cpuset(cs, parent);
}

/*
 * Walk the specified cpuset subtree and look for empty cpusets.
 * The tasks of such cpuset must be moved to a parent cpuset.
 *
1912
 * Called with cgroup_mutex held.  We take callback_mutex to modify
1913 1914 1915 1916 1917 1918 1919 1920 1921 1922
 * cpus_allowed and mems_allowed.
 *
 * This walk processes the tree from top to bottom, completing one layer
 * before dropping down to the next.  It always processes a node before
 * any of its children.
 *
 * For now, since we lack memory hot unplug, we'll never see a cpuset
 * that has tasks along with an empty 'mems'.  But if we did see such
 * a cpuset, we'd handle it just like we do if its 'cpus' was empty.
 */
1923
static void scan_for_empty_cpusets(struct cpuset *root)
1924
{
1925
	LIST_HEAD(queue);
1926 1927
	struct cpuset *cp;	/* scans cpusets being updated */
	struct cpuset *child;	/* scans child cpusets of cp */
1928
	struct cgroup *cont;
1929
	nodemask_t oldmems;
1930

1931 1932 1933
	list_add_tail((struct list_head *)&root->stack_list, &queue);

	while (!list_empty(&queue)) {
1934
		cp = list_first_entry(&queue, struct cpuset, stack_list);
1935 1936 1937 1938 1939
		list_del(queue.next);
		list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {
			child = cgroup_cs(cont);
			list_add_tail(&child->stack_list, &queue);
		}
1940 1941 1942 1943 1944 1945

		/* Continue past cpusets with all cpus, mems online */
		if (cpus_subset(cp->cpus_allowed, cpu_online_map) &&
		    nodes_subset(cp->mems_allowed, node_states[N_HIGH_MEMORY]))
			continue;

1946 1947
		oldmems = cp->mems_allowed;

1948
		/* Remove offline cpus and mems from this cpuset. */
1949
		mutex_lock(&callback_mutex);
1950 1951 1952
		cpus_and(cp->cpus_allowed, cp->cpus_allowed, cpu_online_map);
		nodes_and(cp->mems_allowed, cp->mems_allowed,
						node_states[N_HIGH_MEMORY]);
1953 1954 1955
		mutex_unlock(&callback_mutex);

		/* Move tasks from the empty cpuset to a parent */
1956
		if (cpus_empty(cp->cpus_allowed) ||
1957
		     nodes_empty(cp->mems_allowed))
1958
			remove_tasks_in_empty_cpuset(cp);
1959
		else {
1960
			update_tasks_cpumask(cp, NULL);
1961 1962
			update_tasks_nodemask(cp, &oldmems);
		}
1963 1964 1965
	}
}

1966 1967 1968 1969 1970 1971
/*
 * The top_cpuset tracks what CPUs and Memory Nodes are online,
 * period.  This is necessary in order to make cpusets transparent
 * (of no affect) on systems that are actively using CPU hotplug
 * but making no active use of cpusets.
 *
1972 1973
 * This routine ensures that top_cpuset.cpus_allowed tracks
 * cpu_online_map on each CPU hotplug (cpuhp) event.
1974 1975 1976
 *
 * Called within get_online_cpus().  Needs to call cgroup_lock()
 * before calling generate_sched_domains().
1977
 */
1978
static int cpuset_track_online_cpus(struct notifier_block *unused_nb,
P
Paul Jackson 已提交
1979
				unsigned long phase, void *unused_cpu)
1980
{
1981 1982 1983 1984
	struct sched_domain_attr *attr;
	cpumask_t *doms;
	int ndoms;

1985 1986 1987 1988 1989 1990
	switch (phase) {
	case CPU_ONLINE:
	case CPU_ONLINE_FROZEN:
	case CPU_DEAD:
	case CPU_DEAD_FROZEN:
		break;
1991

1992
	default:
1993
		return NOTIFY_DONE;
1994
	}
1995

1996 1997 1998 1999 2000 2001 2002 2003 2004
	cgroup_lock();
	top_cpuset.cpus_allowed = cpu_online_map;
	scan_for_empty_cpusets(&top_cpuset);
	ndoms = generate_sched_domains(&doms, &attr);
	cgroup_unlock();

	/* Have scheduler rebuild the domains */
	partition_sched_domains(ndoms, doms, attr);

2005
	return NOTIFY_OK;
2006 2007
}

2008
#ifdef CONFIG_MEMORY_HOTPLUG
2009
/*
2010
 * Keep top_cpuset.mems_allowed tracking node_states[N_HIGH_MEMORY].
2011 2012
 * Call this routine anytime after node_states[N_HIGH_MEMORY] changes.
 * See also the previous routine cpuset_track_online_cpus().
2013
 */
A
Al Viro 已提交
2014
void cpuset_track_online_nodes(void)
2015
{
2016 2017 2018 2019
	cgroup_lock();
	top_cpuset.mems_allowed = node_states[N_HIGH_MEMORY];
	scan_for_empty_cpusets(&top_cpuset);
	cgroup_unlock();
2020 2021 2022
}
#endif

L
Linus Torvalds 已提交
2023 2024 2025 2026 2027 2028 2029 2030 2031
/**
 * cpuset_init_smp - initialize cpus_allowed
 *
 * Description: Finish top cpuset after cpu, node maps are initialized
 **/

void __init cpuset_init_smp(void)
{
	top_cpuset.cpus_allowed = cpu_online_map;
2032
	top_cpuset.mems_allowed = node_states[N_HIGH_MEMORY];
2033

2034
	hotcpu_notifier(cpuset_track_online_cpus, 0);
L
Linus Torvalds 已提交
2035 2036 2037 2038 2039
}

/**
 * cpuset_cpus_allowed - return cpus_allowed mask from a tasks cpuset.
 * @tsk: pointer to task_struct from which to obtain cpuset->cpus_allowed.
2040
 * @pmask: pointer to cpumask_t variable to receive cpus_allowed set.
L
Linus Torvalds 已提交
2041 2042 2043 2044 2045 2046 2047
 *
 * Description: Returns the cpumask_t cpus_allowed of the cpuset
 * attached to the specified @tsk.  Guaranteed to return some non-empty
 * subset of cpu_online_map, even if this means going outside the
 * tasks cpuset.
 **/

2048
void cpuset_cpus_allowed(struct task_struct *tsk, cpumask_t *pmask)
L
Linus Torvalds 已提交
2049
{
2050
	mutex_lock(&callback_mutex);
2051
	cpuset_cpus_allowed_locked(tsk, pmask);
2052 2053 2054 2055 2056
	mutex_unlock(&callback_mutex);
}

/**
 * cpuset_cpus_allowed_locked - return cpus_allowed mask from a tasks cpuset.
2057
 * Must be called with callback_mutex held.
2058
 **/
2059
void cpuset_cpus_allowed_locked(struct task_struct *tsk, cpumask_t *pmask)
2060
{
2061
	task_lock(tsk);
2062
	guarantee_online_cpus(task_cs(tsk), pmask);
2063
	task_unlock(tsk);
L
Linus Torvalds 已提交
2064 2065 2066 2067
}

void cpuset_init_current_mems_allowed(void)
{
2068
	nodes_setall(current->mems_allowed);
L
Linus Torvalds 已提交
2069 2070
}

2071 2072 2073 2074 2075 2076
/**
 * cpuset_mems_allowed - return mems_allowed mask from a tasks cpuset.
 * @tsk: pointer to task_struct from which to obtain cpuset->mems_allowed.
 *
 * Description: Returns the nodemask_t mems_allowed of the cpuset
 * attached to the specified @tsk.  Guaranteed to return some non-empty
2077
 * subset of node_states[N_HIGH_MEMORY], even if this means going outside the
2078 2079 2080 2081 2082 2083 2084
 * tasks cpuset.
 **/

nodemask_t cpuset_mems_allowed(struct task_struct *tsk)
{
	nodemask_t mask;

2085
	mutex_lock(&callback_mutex);
2086
	task_lock(tsk);
2087
	guarantee_online_mems(task_cs(tsk), &mask);
2088
	task_unlock(tsk);
2089
	mutex_unlock(&callback_mutex);
2090 2091 2092 2093

	return mask;
}

2094
/**
2095 2096
 * cpuset_nodemask_valid_mems_allowed - check nodemask vs. curremt mems_allowed
 * @nodemask: the nodemask to be checked
2097
 *
2098
 * Are any of the nodes in the nodemask allowed in current->mems_allowed?
L
Linus Torvalds 已提交
2099
 */
2100
int cpuset_nodemask_valid_mems_allowed(nodemask_t *nodemask)
L
Linus Torvalds 已提交
2101
{
2102
	return nodes_intersects(*nodemask, current->mems_allowed);
L
Linus Torvalds 已提交
2103 2104
}

2105
/*
2106 2107 2108 2109
 * nearest_hardwall_ancestor() - Returns the nearest mem_exclusive or
 * mem_hardwall ancestor to the specified cpuset.  Call holding
 * callback_mutex.  If no ancestor is mem_exclusive or mem_hardwall
 * (an unusual configuration), then returns the root cpuset.
2110
 */
2111
static const struct cpuset *nearest_hardwall_ancestor(const struct cpuset *cs)
2112
{
2113
	while (!(is_mem_exclusive(cs) || is_mem_hardwall(cs)) && cs->parent)
2114 2115 2116 2117
		cs = cs->parent;
	return cs;
}

2118
/**
2119
 * cpuset_zone_allowed_softwall - Can we allocate on zone z's memory node?
2120
 * @z: is this zone on an allowed node?
2121
 * @gfp_mask: memory allocation flags
2122
 *
2123 2124
 * If we're in interrupt, yes, we can always allocate.  If
 * __GFP_THISNODE is set, yes, we can always allocate.  If zone
2125 2126
 * z's node is in our tasks mems_allowed, yes.  If it's not a
 * __GFP_HARDWALL request and this zone's nodes is in the nearest
2127
 * hardwalled cpuset ancestor to this tasks cpuset, yes.
2128 2129
 * If the task has been OOM killed and has access to memory reserves
 * as specified by the TIF_MEMDIE flag, yes.
2130 2131
 * Otherwise, no.
 *
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145
 * If __GFP_HARDWALL is set, cpuset_zone_allowed_softwall()
 * reduces to cpuset_zone_allowed_hardwall().  Otherwise,
 * cpuset_zone_allowed_softwall() might sleep, and might allow a zone
 * from an enclosing cpuset.
 *
 * cpuset_zone_allowed_hardwall() only handles the simpler case of
 * hardwall cpusets, and never sleeps.
 *
 * The __GFP_THISNODE placement logic is really handled elsewhere,
 * by forcibly using a zonelist starting at a specified node, and by
 * (in get_page_from_freelist()) refusing to consider the zones for
 * any node on the zonelist except the first.  By the time any such
 * calls get to this routine, we should just shut up and say 'yes'.
 *
2146
 * GFP_USER allocations are marked with the __GFP_HARDWALL bit,
2147 2148
 * and do not allow allocations outside the current tasks cpuset
 * unless the task has been OOM killed as is marked TIF_MEMDIE.
2149
 * GFP_KERNEL allocations are not so marked, so can escape to the
2150
 * nearest enclosing hardwalled ancestor cpuset.
2151
 *
2152 2153 2154 2155 2156 2157 2158
 * Scanning up parent cpusets requires callback_mutex.  The
 * __alloc_pages() routine only calls here with __GFP_HARDWALL bit
 * _not_ set if it's a GFP_KERNEL allocation, and all nodes in the
 * current tasks mems_allowed came up empty on the first pass over
 * the zonelist.  So only GFP_KERNEL allocations, if all nodes in the
 * cpuset are short of memory, might require taking the callback_mutex
 * mutex.
2159
 *
2160
 * The first call here from mm/page_alloc:get_page_from_freelist()
2161 2162 2163
 * has __GFP_HARDWALL set in gfp_mask, enforcing hardwall cpusets,
 * so no allocation on a node outside the cpuset is allowed (unless
 * in interrupt, of course).
2164 2165 2166 2167 2168 2169
 *
 * The second pass through get_page_from_freelist() doesn't even call
 * here for GFP_ATOMIC calls.  For those calls, the __alloc_pages()
 * variable 'wait' is not set, and the bit ALLOC_CPUSET is not set
 * in alloc_flags.  That logic and the checks below have the combined
 * affect that:
2170 2171
 *	in_interrupt - any node ok (current task context irrelevant)
 *	GFP_ATOMIC   - any node ok
2172
 *	TIF_MEMDIE   - any node ok
2173
 *	GFP_KERNEL   - any node in enclosing hardwalled cpuset ok
2174
 *	GFP_USER     - only nodes in current tasks mems allowed ok.
2175 2176
 *
 * Rule:
2177
 *    Don't call cpuset_zone_allowed_softwall if you can't sleep, unless you
2178 2179
 *    pass in the __GFP_HARDWALL flag set in gfp_flag, which disables
 *    the code that might scan up ancestor cpusets and sleep.
2180
 */
2181

2182
int __cpuset_zone_allowed_softwall(struct zone *z, gfp_t gfp_mask)
L
Linus Torvalds 已提交
2183
{
2184 2185
	int node;			/* node that zone z is on */
	const struct cpuset *cs;	/* current cpuset ancestors */
2186
	int allowed;			/* is allocation in zone z allowed? */
2187

2188
	if (in_interrupt() || (gfp_mask & __GFP_THISNODE))
2189
		return 1;
2190
	node = zone_to_nid(z);
2191
	might_sleep_if(!(gfp_mask & __GFP_HARDWALL));
2192 2193
	if (node_isset(node, current->mems_allowed))
		return 1;
2194 2195 2196 2197 2198 2199
	/*
	 * Allow tasks that have access to memory reserves because they have
	 * been OOM killed to get memory anywhere.
	 */
	if (unlikely(test_thread_flag(TIF_MEMDIE)))
		return 1;
2200 2201 2202
	if (gfp_mask & __GFP_HARDWALL)	/* If hardwall request, stop here */
		return 0;

2203 2204 2205
	if (current->flags & PF_EXITING) /* Let dying task have memory */
		return 1;

2206
	/* Not hardwall and node outside mems_allowed: scan up cpusets */
2207
	mutex_lock(&callback_mutex);
2208 2209

	task_lock(current);
2210
	cs = nearest_hardwall_ancestor(task_cs(current));
2211 2212
	task_unlock(current);

2213
	allowed = node_isset(node, cs->mems_allowed);
2214
	mutex_unlock(&callback_mutex);
2215
	return allowed;
L
Linus Torvalds 已提交
2216 2217
}

2218 2219 2220 2221 2222 2223 2224
/*
 * cpuset_zone_allowed_hardwall - Can we allocate on zone z's memory node?
 * @z: is this zone on an allowed node?
 * @gfp_mask: memory allocation flags
 *
 * If we're in interrupt, yes, we can always allocate.
 * If __GFP_THISNODE is set, yes, we can always allocate.  If zone
2225 2226 2227
 * z's node is in our tasks mems_allowed, yes.   If the task has been
 * OOM killed and has access to memory reserves as specified by the
 * TIF_MEMDIE flag, yes.  Otherwise, no.
2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250
 *
 * The __GFP_THISNODE placement logic is really handled elsewhere,
 * by forcibly using a zonelist starting at a specified node, and by
 * (in get_page_from_freelist()) refusing to consider the zones for
 * any node on the zonelist except the first.  By the time any such
 * calls get to this routine, we should just shut up and say 'yes'.
 *
 * Unlike the cpuset_zone_allowed_softwall() variant, above,
 * this variant requires that the zone be in the current tasks
 * mems_allowed or that we're in interrupt.  It does not scan up the
 * cpuset hierarchy for the nearest enclosing mem_exclusive cpuset.
 * It never sleeps.
 */

int __cpuset_zone_allowed_hardwall(struct zone *z, gfp_t gfp_mask)
{
	int node;			/* node that zone z is on */

	if (in_interrupt() || (gfp_mask & __GFP_THISNODE))
		return 1;
	node = zone_to_nid(z);
	if (node_isset(node, current->mems_allowed))
		return 1;
D
Daniel Walker 已提交
2251 2252 2253 2254 2255 2256
	/*
	 * Allow tasks that have access to memory reserves because they have
	 * been OOM killed to get memory anywhere.
	 */
	if (unlikely(test_thread_flag(TIF_MEMDIE)))
		return 1;
2257 2258 2259
	return 0;
}

P
Paul Jackson 已提交
2260 2261 2262
/**
 * cpuset_lock - lock out any changes to cpuset structures
 *
2263
 * The out of memory (oom) code needs to mutex_lock cpusets
P
Paul Jackson 已提交
2264
 * from being changed while it scans the tasklist looking for a
2265
 * task in an overlapping cpuset.  Expose callback_mutex via this
P
Paul Jackson 已提交
2266 2267
 * cpuset_lock() routine, so the oom code can lock it, before
 * locking the task list.  The tasklist_lock is a spinlock, so
2268
 * must be taken inside callback_mutex.
P
Paul Jackson 已提交
2269 2270 2271 2272
 */

void cpuset_lock(void)
{
2273
	mutex_lock(&callback_mutex);
P
Paul Jackson 已提交
2274 2275 2276 2277 2278 2279 2280 2281 2282 2283
}

/**
 * cpuset_unlock - release lock on cpuset changes
 *
 * Undo the lock taken in a previous cpuset_lock() call.
 */

void cpuset_unlock(void)
{
2284
	mutex_unlock(&callback_mutex);
P
Paul Jackson 已提交
2285 2286
}

2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324
/**
 * cpuset_mem_spread_node() - On which node to begin search for a page
 *
 * If a task is marked PF_SPREAD_PAGE or PF_SPREAD_SLAB (as for
 * tasks in a cpuset with is_spread_page or is_spread_slab set),
 * and if the memory allocation used cpuset_mem_spread_node()
 * to determine on which node to start looking, as it will for
 * certain page cache or slab cache pages such as used for file
 * system buffers and inode caches, then instead of starting on the
 * local node to look for a free page, rather spread the starting
 * node around the tasks mems_allowed nodes.
 *
 * We don't have to worry about the returned node being offline
 * because "it can't happen", and even if it did, it would be ok.
 *
 * The routines calling guarantee_online_mems() are careful to
 * only set nodes in task->mems_allowed that are online.  So it
 * should not be possible for the following code to return an
 * offline node.  But if it did, that would be ok, as this routine
 * is not returning the node where the allocation must be, only
 * the node where the search should start.  The zonelist passed to
 * __alloc_pages() will include all nodes.  If the slab allocator
 * is passed an offline node, it will fall back to the local node.
 * See kmem_cache_alloc_node().
 */

int cpuset_mem_spread_node(void)
{
	int node;

	node = next_node(current->cpuset_mem_spread_rotor, current->mems_allowed);
	if (node == MAX_NUMNODES)
		node = first_node(current->mems_allowed);
	current->cpuset_mem_spread_rotor = node;
	return node;
}
EXPORT_SYMBOL_GPL(cpuset_mem_spread_node);

2325
/**
2326 2327 2328 2329 2330 2331 2332 2333
 * cpuset_mems_allowed_intersects - Does @tsk1's mems_allowed intersect @tsk2's?
 * @tsk1: pointer to task_struct of some task.
 * @tsk2: pointer to task_struct of some other task.
 *
 * Description: Return true if @tsk1's mems_allowed intersects the
 * mems_allowed of @tsk2.  Used by the OOM killer to determine if
 * one of the task's memory usage might impact the memory available
 * to the other.
2334 2335
 **/

2336 2337
int cpuset_mems_allowed_intersects(const struct task_struct *tsk1,
				   const struct task_struct *tsk2)
2338
{
2339
	return nodes_intersects(tsk1->mems_allowed, tsk2->mems_allowed);
2340 2341
}

2342 2343 2344 2345 2346 2347
/*
 * Collection of memory_pressure is suppressed unless
 * this flag is enabled by writing "1" to the special
 * cpuset file 'memory_pressure_enabled' in the root cpuset.
 */

2348
int cpuset_memory_pressure_enabled __read_mostly;
2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370

/**
 * cpuset_memory_pressure_bump - keep stats of per-cpuset reclaims.
 *
 * Keep a running average of the rate of synchronous (direct)
 * page reclaim efforts initiated by tasks in each cpuset.
 *
 * This represents the rate at which some task in the cpuset
 * ran low on memory on all nodes it was allowed to use, and
 * had to enter the kernels page reclaim code in an effort to
 * create more free memory by tossing clean pages or swapping
 * or writing dirty pages.
 *
 * Display to user space in the per-cpuset read-only file
 * "memory_pressure".  Value displayed is an integer
 * representing the recent rate of entry into the synchronous
 * (direct) page reclaim by any task attached to the cpuset.
 **/

void __cpuset_memory_pressure_bump(void)
{
	task_lock(current);
2371
	fmeter_markevent(&task_cs(current)->fmeter);
2372 2373 2374
	task_unlock(current);
}

2375
#ifdef CONFIG_PROC_PID_CPUSET
L
Linus Torvalds 已提交
2376 2377 2378 2379
/*
 * proc_cpuset_show()
 *  - Print tasks cpuset path into seq_file.
 *  - Used for /proc/<pid>/cpuset.
2380 2381
 *  - No need to task_lock(tsk) on this tsk->cpuset reference, as it
 *    doesn't really matter if tsk->cpuset changes after we read it,
2382
 *    and we take cgroup_mutex, keeping cpuset_attach() from changing it
2383
 *    anyway.
L
Linus Torvalds 已提交
2384
 */
P
Paul Jackson 已提交
2385
static int proc_cpuset_show(struct seq_file *m, void *unused_v)
L
Linus Torvalds 已提交
2386
{
2387
	struct pid *pid;
L
Linus Torvalds 已提交
2388 2389
	struct task_struct *tsk;
	char *buf;
2390
	struct cgroup_subsys_state *css;
2391
	int retval;
L
Linus Torvalds 已提交
2392

2393
	retval = -ENOMEM;
L
Linus Torvalds 已提交
2394 2395
	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
	if (!buf)
2396 2397 2398
		goto out;

	retval = -ESRCH;
2399 2400
	pid = m->private;
	tsk = get_pid_task(pid, PIDTYPE_PID);
2401 2402
	if (!tsk)
		goto out_free;
L
Linus Torvalds 已提交
2403

2404
	retval = -EINVAL;
2405 2406 2407
	cgroup_lock();
	css = task_subsys_state(tsk, cpuset_subsys_id);
	retval = cgroup_path(css->cgroup, buf, PAGE_SIZE);
L
Linus Torvalds 已提交
2408
	if (retval < 0)
2409
		goto out_unlock;
L
Linus Torvalds 已提交
2410 2411
	seq_puts(m, buf);
	seq_putc(m, '\n');
2412
out_unlock:
2413
	cgroup_unlock();
2414 2415
	put_task_struct(tsk);
out_free:
L
Linus Torvalds 已提交
2416
	kfree(buf);
2417
out:
L
Linus Torvalds 已提交
2418 2419 2420 2421 2422
	return retval;
}

static int cpuset_open(struct inode *inode, struct file *file)
{
2423 2424
	struct pid *pid = PROC_I(inode)->pid;
	return single_open(file, proc_cpuset_show, pid);
L
Linus Torvalds 已提交
2425 2426
}

2427
const struct file_operations proc_cpuset_operations = {
L
Linus Torvalds 已提交
2428 2429 2430 2431 2432
	.open		= cpuset_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= single_release,
};
2433
#endif /* CONFIG_PROC_PID_CPUSET */
L
Linus Torvalds 已提交
2434 2435

/* Display task cpus_allowed, mems_allowed in /proc/<pid>/status file. */
2436 2437 2438
void cpuset_task_status_allowed(struct seq_file *m, struct task_struct *task)
{
	seq_printf(m, "Cpus_allowed:\t");
2439
	seq_cpumask(m, &task->cpus_allowed);
2440
	seq_printf(m, "\n");
2441
	seq_printf(m, "Cpus_allowed_list:\t");
2442
	seq_cpumask_list(m, &task->cpus_allowed);
2443
	seq_printf(m, "\n");
2444
	seq_printf(m, "Mems_allowed:\t");
2445
	seq_nodemask(m, &task->mems_allowed);
2446
	seq_printf(m, "\n");
2447
	seq_printf(m, "Mems_allowed_list:\t");
2448
	seq_nodemask_list(m, &task->mems_allowed);
2449
	seq_printf(m, "\n");
L
Linus Torvalds 已提交
2450
}