cpuset.c 70.3 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6
/*
 *  kernel/cpuset.c
 *
 *  Processor and Memory placement constraints for sets of tasks.
 *
 *  Copyright (C) 2003 BULL SA.
Paul Jackson's avatar
Paul Jackson committed
7
 *  Copyright (C) 2004-2007 Silicon Graphics, Inc.
8
 *  Copyright (C) 2006 Google, Inc
Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
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
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
45 46
#include <linux/sched.h>
#include <linux/seq_file.h>
47
#include <linux/security.h>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
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 */
};

Linus Torvalds's avatar
Linus Torvalds committed
82
struct cpuset {
83 84
	struct cgroup_subsys_state css;

Linus Torvalds's avatar
Linus Torvalds committed
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 */
Paul Jackson's avatar
Paul Jackson committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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

Linus Torvalds's avatar
Linus Torvalds committed
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,
Paul Jackson's avatar
Paul Jackson committed
133
	CS_SCHED_LOAD_BALANCE,
134 135
	CS_SPREAD_PAGE,
	CS_SPREAD_SLAB,
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
142 143 144 145
}

static inline int is_mem_exclusive(const struct cpuset *cs)
{
146
	return test_bit(CS_MEM_EXCLUSIVE, &cs->flags);
Linus Torvalds's avatar
Linus Torvalds committed
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);
}

Paul Jackson's avatar
Paul Jackson committed
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);
}

Linus Torvalds's avatar
Linus Torvalds committed
174
/*
175
 * Increment this integer everytime any cpuset changes its
Linus Torvalds's avatar
Linus Torvalds committed
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
Linus Torvalds's avatar
Linus Torvalds committed
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,
Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
192
 */
193
static int cpuset_mems_generation;
Linus Torvalds's avatar
Linus Torvalds committed
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
Linus Torvalds's avatar
Linus Torvalds committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
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].
Linus Torvalds's avatar
Linus Torvalds committed
299 300
 *
 * One way or another, we guarantee to return some non-empty subset
301
 * of node_states[N_HIGH_MEMORY].
Linus Torvalds's avatar
Linus Torvalds committed
302
 *
303
 * Call with callback_mutex held.
Linus Torvalds's avatar
Linus Torvalds committed
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]))
Linus Torvalds's avatar
Linus Torvalds committed
310 311
		cs = cs->parent;
	if (cs)
312 313
		nodes_and(*pmask, cs->mems_allowed,
					node_states[N_HIGH_MEMORY]);
Linus Torvalds's avatar
Linus Torvalds committed
314
	else
315 316
		*pmask = node_states[N_HIGH_MEMORY];
	BUG_ON(!nodes_intersects(*pmask, node_states[N_HIGH_MEMORY]));
Linus Torvalds's avatar
Linus Torvalds committed
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
David Rientjes's avatar
David Rientjes committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
358 359
 */

360
void cpuset_update_task_memory_state(void)
Linus Torvalds's avatar
Linus Torvalds committed
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();
371
		my_cpusets_mem_gen = task_cs(tsk)->mems_generation;
372 373
		rcu_read_unlock();
	}
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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))
Linus Torvalds's avatar
Linus Torvalds committed
439 440 441 442
			return -EBUSY;
	}

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

446 447
	par = cur->parent;

Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
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;
		}
	}

Linus Torvalds's avatar
Linus Torvalds committed
476 477 478
	return 0;
}

Paul Jackson's avatar
Paul Jackson committed
479
/*
480
 * Helper routine for generate_sched_domains().
Paul Jackson's avatar
Paul Jackson committed
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);
		}
	}
}

Paul Jackson's avatar
Paul Jackson committed
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.
Paul Jackson's avatar
Paul Jackson committed
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.
Paul Jackson's avatar
Paul Jackson committed
543 544
 *
 * The three key local variables below are:
545
 *    q  - a linked-list queue of cpuset pointers, used to implement a
Paul Jackson's avatar
Paul Jackson committed
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)
Paul Jackson's avatar
Paul Jackson committed
579
{
580
	LIST_HEAD(q);		/* queue of cpusets to be scanned */
Paul Jackson's avatar
Paul Jackson committed
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 */
Paul Jackson's avatar
Paul Jackson committed
587 588 589
	int ndoms;		/* number of sched domains in result */
	int nslot;		/* next empty doms[] cpumask_t slot */

590
	ndoms = 0;
Paul Jackson's avatar
Paul Jackson committed
591
	doms = NULL;
592
	dattr = NULL;
593
	csa = NULL;
Paul Jackson's avatar
Paul Jackson committed
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
		}
Paul Jackson's avatar
Paul Jackson committed
606
		*doms = top_cpuset.cpus_allowed;
607 608 609

		ndoms = 1;
		goto done;
Paul Jackson's avatar
Paul Jackson committed
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)) {
Paul Jackson's avatar
Paul Jackson committed
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)) {
Paul Jackson's avatar
Paul Jackson committed
635
			csa[csn++] = cp;
636 637
			continue;
		}
638

Paul Jackson's avatar
Paul Jackson committed
639 640
		list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {
			child = cgroup_cs(cont);
641
			list_add_tail(&child->stack_list, &q);
Paul Jackson's avatar
Paul Jackson committed
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.
	 */
Paul Jackson's avatar
Paul Jackson committed
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);
Paul Jackson's avatar
Paul Jackson committed
687 688 689

	for (nslot = 0, i = 0; i < csn; i++) {
		struct cpuset *a = csa[i];
690
		cpumask_t *dp;
Paul Jackson's avatar
Paul Jackson committed
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--;
Paul Jackson's avatar
Paul Jackson committed
709
			}
710 711
			continue;
		}
Paul Jackson's avatar
Paul Jackson committed
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;
Paul Jackson's avatar
Paul Jackson committed
726 727
			}
		}
728
		nslot++;
Paul Jackson's avatar
Paul Jackson committed
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
}
Paul Jackson's avatar
Paul Jackson committed
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);
Paul Jackson's avatar
Paul Jackson committed
807 808
}

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.
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)
821 822 823 824
{
	return !cpus_equal(tsk->cpus_allowed,
			(cgroup_cs(scan->cg))->cpus_allowed);
}
825

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)
839
{
840
	set_cpus_allowed_ptr(tsk, &((cgroup_cs(scan->cg))->cpus_allowed));
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
}

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)
Linus Torvalds's avatar
Linus Torvalds committed
873
{
874
	struct ptr_heap heap;
Linus Torvalds's avatar
Linus Torvalds committed
875
	struct cpuset trialcs;
876 877
	int retval;
	int is_load_balanced;
Linus Torvalds's avatar
Linus Torvalds committed
878

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

Linus Torvalds's avatar
Linus Torvalds committed
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
	}
Linus Torvalds's avatar
Linus Torvalds committed
901
	retval = validate_change(cs, &trialcs);
902 903
	if (retval < 0)
		return retval;
Paul Jackson's avatar
Paul Jackson committed
904

Paul Menage's avatar
Paul Menage committed
905 906 907
	/* Nothing to do if the cpus didn't change */
	if (cpus_equal(cs->cpus_allowed, trialcs.cpus_allowed))
		return 0;
908

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

Paul Jackson's avatar
Paul Jackson committed
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);
Paul Jackson's avatar
Paul Jackson committed
918

Paul Menage's avatar
Paul Menage committed
919 920
	/*
	 * Scan tasks in the cpuset, and update the cpumasks of any
921
	 * that need an update.
Paul Menage's avatar
Paul Menage committed
922
	 */
923 924 925
	update_tasks_cpumask(cs, &heap);

	heap_free(&heap);
926

Paul Menage's avatar
Paul Menage committed
927
	if (is_load_balanced)
928
		async_rebuild_sched_domains();
929
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
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:
Linus Torvalds's avatar
Linus Torvalds committed
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;
}

Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
1168 1169
 */

1170 1171
static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs,
		       int turning_on)
Linus Torvalds's avatar
Linus Torvalds committed
1172 1173
{
	struct cpuset trialcs;
1174
	int err;
1175
	int balance_flag_changed;
Linus Torvalds's avatar
Linus Torvalds committed
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;
Paul Jackson's avatar
Paul Jackson committed
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

1194
	if (!cpus_empty(trialcs.cpus_allowed) && balance_flag_changed)
1195
		async_rebuild_sched_domains();
Paul Jackson's avatar
Paul Jackson committed
1196

1197
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
1198 1199
}

1200
/*
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)