Blame view
kernel/cpuset.c
75.7 KB
1da177e4c Linux-2.6.12-rc2 |
1 2 3 4 5 6 |
/* * kernel/cpuset.c * * Processor and Memory placement constraints for sets of tasks. * * Copyright (C) 2003 BULL SA. |
029190c51 cpuset sched_load... |
7 |
* Copyright (C) 2004-2007 Silicon Graphics, Inc. |
8793d854e Task Control Grou... |
8 |
* Copyright (C) 2006 Google, Inc |
1da177e4c Linux-2.6.12-rc2 |
9 10 11 |
* * Portions derived from Patrick Mochel's sysfs code. * sysfs is Copyright (c) 2001-3 Patrick Mochel |
1da177e4c Linux-2.6.12-rc2 |
12 |
* |
825a46af5 [PATCH] cpuset me... |
13 |
* 2003-10-10 Written by Simon Derr. |
1da177e4c Linux-2.6.12-rc2 |
14 |
* 2003-10-22 Updates by Stephen Hemminger. |
825a46af5 [PATCH] cpuset me... |
15 |
* 2004 May-July Rework by Paul Jackson. |
8793d854e Task Control Grou... |
16 |
* 2006 Rework by Paul Menage to use generic cgroups |
cf417141c sched, cpuset: re... |
17 18 |
* 2008 Rework of the scheduler domains and CPU hotplug handling * by Max Krasnyansky |
1da177e4c Linux-2.6.12-rc2 |
19 20 21 22 23 |
* * 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. */ |
1da177e4c Linux-2.6.12-rc2 |
24 25 26 27 28 29 30 31 32 33 34 35 |
#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> |
68860ec10 [PATCH] cpusets: ... |
36 |
#include <linux/mempolicy.h> |
1da177e4c Linux-2.6.12-rc2 |
37 |
#include <linux/mm.h> |
f481891fd cpuset: update to... |
38 |
#include <linux/memory.h> |
9984de1a5 kernel: Map most ... |
39 |
#include <linux/export.h> |
1da177e4c Linux-2.6.12-rc2 |
40 41 42 43 |
#include <linux/mount.h> #include <linux/namei.h> #include <linux/pagemap.h> #include <linux/proc_fs.h> |
6b9c2603c [PATCH] cpuset: u... |
44 |
#include <linux/rcupdate.h> |
1da177e4c Linux-2.6.12-rc2 |
45 46 |
#include <linux/sched.h> #include <linux/seq_file.h> |
22fb52dd7 [PATCH] SELinux: ... |
47 |
#include <linux/security.h> |
1da177e4c Linux-2.6.12-rc2 |
48 |
#include <linux/slab.h> |
1da177e4c Linux-2.6.12-rc2 |
49 50 51 52 53 54 55 56 |
#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> |
60063497a atomic: use <linu... |
57 |
#include <linux/atomic.h> |
3d3f26a7b [PATCH] kernel/cp... |
58 |
#include <linux/mutex.h> |
956db3ca0 hotplug cpu: move... |
59 60 |
#include <linux/workqueue.h> #include <linux/cgroup.h> |
e44193d39 cpuset: let hotpl... |
61 |
#include <linux/wait.h> |
1da177e4c Linux-2.6.12-rc2 |
62 |
|
664eeddee mm: page_alloc: u... |
63 |
struct static_key cpusets_enabled_key __read_mostly = STATIC_KEY_INIT_FALSE; |
202f72d5d [PATCH] cpuset: n... |
64 |
|
3e0d98b9f [PATCH] cpuset: m... |
65 66 67 68 69 70 71 72 |
/* 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 */ }; |
1da177e4c Linux-2.6.12-rc2 |
73 |
struct cpuset { |
8793d854e Task Control Grou... |
74 |
struct cgroup_subsys_state css; |
1da177e4c Linux-2.6.12-rc2 |
75 |
unsigned long flags; /* "unsigned long" so bitops work */ |
e2b9a3d7d cpuset: add cs->e... |
76 |
|
7e88291be cpuset: make cs->... |
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 |
/* * On default hierarchy: * * The user-configured masks can only be changed by writing to * cpuset.cpus and cpuset.mems, and won't be limited by the * parent masks. * * The effective masks is the real masks that apply to the tasks * in the cpuset. They may be changed if the configured masks are * changed or hotplug happens. * * effective_mask == configured_mask & parent's effective_mask, * and if it ends up empty, it will inherit the parent's mask. * * * On legacy hierachy: * * The user-configured masks are always the same with effective masks. */ |
e2b9a3d7d cpuset: add cs->e... |
96 97 98 99 100 101 102 |
/* user-configured CPUs and Memory Nodes allow to tasks */ cpumask_var_t cpus_allowed; nodemask_t mems_allowed; /* effective CPUs and Memory Nodes allow to tasks */ cpumask_var_t effective_cpus; nodemask_t effective_mems; |
1da177e4c Linux-2.6.12-rc2 |
103 |
|
33ad801df cpuset: record ol... |
104 105 106 107 108 109 110 111 112 113 114 |
/* * This is old Memory Nodes tasks took on. * * - top_cpuset.old_mems_allowed is initialized to mems_allowed. * - A new cpuset's old_mems_allowed is initialized when some * task is moved into it. * - old_mems_allowed is used in cpuset_migrate_mm() when we change * cpuset.mems_allowed and have tasks' nodemask updated, and * then old_mems_allowed is updated to mems_allowed. */ nodemask_t old_mems_allowed; |
3e0d98b9f [PATCH] cpuset: m... |
115 |
struct fmeter fmeter; /* memory_pressure filter */ |
029190c51 cpuset sched_load... |
116 |
|
452477fa6 cpuset: pin down ... |
117 118 119 120 121 |
/* * Tasks are being attached to this cpuset. Used to prevent * zeroing cpus/mems_allowed between ->can_attach() and ->attach(). */ int attach_in_progress; |
029190c51 cpuset sched_load... |
122 123 |
/* partition number for rebuild_sched_domains() */ int pn; |
956db3ca0 hotplug cpu: move... |
124 |
|
1d3504fcf sched, cpuset: cu... |
125 126 |
/* for custom sched domain */ int relax_domain_level; |
1da177e4c Linux-2.6.12-rc2 |
127 |
}; |
a7c6d554a cgroup: add/updat... |
128 |
static inline struct cpuset *css_cs(struct cgroup_subsys_state *css) |
8793d854e Task Control Grou... |
129 |
{ |
a7c6d554a cgroup: add/updat... |
130 |
return css ? container_of(css, struct cpuset, css) : NULL; |
8793d854e Task Control Grou... |
131 132 133 134 135 |
} /* Retrieve the cpuset for a task */ static inline struct cpuset *task_cs(struct task_struct *task) { |
073219e99 cgroup: clean up ... |
136 |
return css_cs(task_css(task, cpuset_cgrp_id)); |
8793d854e Task Control Grou... |
137 |
} |
8793d854e Task Control Grou... |
138 |
|
c9710d801 cpuset: drop "con... |
139 |
static inline struct cpuset *parent_cs(struct cpuset *cs) |
c431069fe cpuset: remove cp... |
140 |
{ |
5c9d535b8 cgroup: remove cs... |
141 |
return css_cs(cs->css.parent); |
c431069fe cpuset: remove cp... |
142 |
} |
b246272ec cpusets: stall wh... |
143 144 145 146 147 148 149 150 151 152 153 |
#ifdef CONFIG_NUMA static inline bool task_has_mempolicy(struct task_struct *task) { return task->mempolicy; } #else static inline bool task_has_mempolicy(struct task_struct *task) { return false; } #endif |
1da177e4c Linux-2.6.12-rc2 |
154 155 |
/* bits in struct cpuset flags field */ typedef enum { |
efeb77b2f cpuset: introduce... |
156 |
CS_ONLINE, |
1da177e4c Linux-2.6.12-rc2 |
157 158 |
CS_CPU_EXCLUSIVE, CS_MEM_EXCLUSIVE, |
786083667 Cpuset hardwall f... |
159 |
CS_MEM_HARDWALL, |
45b07ef31 [PATCH] cpusets: ... |
160 |
CS_MEMORY_MIGRATE, |
029190c51 cpuset sched_load... |
161 |
CS_SCHED_LOAD_BALANCE, |
825a46af5 [PATCH] cpuset me... |
162 163 |
CS_SPREAD_PAGE, CS_SPREAD_SLAB, |
1da177e4c Linux-2.6.12-rc2 |
164 165 166 |
} cpuset_flagbits_t; /* convenient tests for these bits */ |
efeb77b2f cpuset: introduce... |
167 168 169 170 |
static inline bool is_cpuset_online(const struct cpuset *cs) { return test_bit(CS_ONLINE, &cs->flags); } |
1da177e4c Linux-2.6.12-rc2 |
171 172 |
static inline int is_cpu_exclusive(const struct cpuset *cs) { |
7b5b9ef0e [PATCH] cpuset cl... |
173 |
return test_bit(CS_CPU_EXCLUSIVE, &cs->flags); |
1da177e4c Linux-2.6.12-rc2 |
174 175 176 177 |
} static inline int is_mem_exclusive(const struct cpuset *cs) { |
7b5b9ef0e [PATCH] cpuset cl... |
178 |
return test_bit(CS_MEM_EXCLUSIVE, &cs->flags); |
1da177e4c Linux-2.6.12-rc2 |
179 |
} |
786083667 Cpuset hardwall f... |
180 181 182 183 |
static inline int is_mem_hardwall(const struct cpuset *cs) { return test_bit(CS_MEM_HARDWALL, &cs->flags); } |
029190c51 cpuset sched_load... |
184 185 186 187 |
static inline int is_sched_load_balance(const struct cpuset *cs) { return test_bit(CS_SCHED_LOAD_BALANCE, &cs->flags); } |
45b07ef31 [PATCH] cpusets: ... |
188 189 |
static inline int is_memory_migrate(const struct cpuset *cs) { |
7b5b9ef0e [PATCH] cpuset cl... |
190 |
return test_bit(CS_MEMORY_MIGRATE, &cs->flags); |
45b07ef31 [PATCH] cpusets: ... |
191 |
} |
825a46af5 [PATCH] cpuset me... |
192 193 194 195 196 197 198 199 200 |
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); } |
1da177e4c Linux-2.6.12-rc2 |
201 |
static struct cpuset top_cpuset = { |
efeb77b2f cpuset: introduce... |
202 203 |
.flags = ((1 << CS_ONLINE) | (1 << CS_CPU_EXCLUSIVE) | (1 << CS_MEM_EXCLUSIVE)), |
1da177e4c Linux-2.6.12-rc2 |
204 |
}; |
ae8086ce1 cpuset: introduce... |
205 206 207 |
/** * cpuset_for_each_child - traverse online children of a cpuset * @child_cs: loop cursor pointing to the current child |
492eb21b9 cgroup: make hier... |
208 |
* @pos_css: used for iteration |
ae8086ce1 cpuset: introduce... |
209 210 211 212 213 |
* @parent_cs: target cpuset to walk children of * * Walk @child_cs through the online children of @parent_cs. Must be used * with RCU read locked. */ |
492eb21b9 cgroup: make hier... |
214 215 216 |
#define cpuset_for_each_child(child_cs, pos_css, parent_cs) \ css_for_each_child((pos_css), &(parent_cs)->css) \ if (is_cpuset_online(((child_cs) = css_cs((pos_css))))) |
ae8086ce1 cpuset: introduce... |
217 |
|
fc560a26a cpuset: replace c... |
218 219 220 |
/** * cpuset_for_each_descendant_pre - pre-order walk of a cpuset's descendants * @des_cs: loop cursor pointing to the current descendant |
492eb21b9 cgroup: make hier... |
221 |
* @pos_css: used for iteration |
fc560a26a cpuset: replace c... |
222 223 224 |
* @root_cs: target cpuset to walk ancestor of * * Walk @des_cs through the online descendants of @root_cs. Must be used |
492eb21b9 cgroup: make hier... |
225 |
* with RCU read locked. The caller may modify @pos_css by calling |
bd8815a6d cgroup: make css_... |
226 227 |
* css_rightmost_descendant() to skip subtree. @root_cs is included in the * iteration and the first node to be visited. |
fc560a26a cpuset: replace c... |
228 |
*/ |
492eb21b9 cgroup: make hier... |
229 230 231 |
#define cpuset_for_each_descendant_pre(des_cs, pos_css, root_cs) \ css_for_each_descendant_pre((pos_css), &(root_cs)->css) \ if (is_cpuset_online(((des_cs) = css_cs((pos_css))))) |
fc560a26a cpuset: replace c... |
232 |
|
1da177e4c Linux-2.6.12-rc2 |
233 |
/* |
8447a0fee cpuset: convert c... |
234 235 236 237 |
* There are two global locks guarding cpuset structures - cpuset_mutex and * callback_lock. We also require taking task_lock() when dereferencing a * task's cpuset pointer. See "The task_lock() exception", at the end of this * comment. |
5d21cc2db cpuset: replace c... |
238 |
* |
8447a0fee cpuset: convert c... |
239 |
* A task must hold both locks to modify cpusets. If a task holds |
5d21cc2db cpuset: replace c... |
240 |
* cpuset_mutex, then it blocks others wanting that mutex, ensuring that it |
8447a0fee cpuset: convert c... |
241 |
* is the only task able to also acquire callback_lock and be able to |
5d21cc2db cpuset: replace c... |
242 243 244 |
* modify cpusets. It can perform various checks on the cpuset structure * first, knowing nothing will change. It can also allocate memory while * just holding cpuset_mutex. While it is performing these checks, various |
8447a0fee cpuset: convert c... |
245 246 |
* callback routines can briefly acquire callback_lock to query cpusets. * Once it is ready to make the changes, it takes callback_lock, blocking |
5d21cc2db cpuset: replace c... |
247 |
* everyone else. |
053199edf [PATCH] cpusets: ... |
248 249 |
* * Calls to the kernel memory allocator can not be made while holding |
8447a0fee cpuset: convert c... |
250 |
* callback_lock, as that would risk double tripping on callback_lock |
053199edf [PATCH] cpusets: ... |
251 252 253 |
* from one of the callbacks into the cpuset code from within * __alloc_pages(). * |
8447a0fee cpuset: convert c... |
254 |
* If a task is only holding callback_lock, then it has read-only |
053199edf [PATCH] cpusets: ... |
255 256 |
* access to cpusets. * |
58568d2a8 cpuset,mm: update... |
257 258 259 |
* Now, the task_struct fields mems_allowed and mempolicy may be changed * by other task, we use alloc_lock in the task_struct fields to protect * them. |
053199edf [PATCH] cpusets: ... |
260 |
* |
8447a0fee cpuset: convert c... |
261 |
* The cpuset_common_file_read() handlers only hold callback_lock across |
053199edf [PATCH] cpusets: ... |
262 263 264 |
* small pieces of code, such as when reading out possibly multi-word * cpumasks and nodemasks. * |
2df167a30 cgroups: update c... |
265 266 |
* Accessing a task's cpuset should be done in accordance with the * guidelines for accessing subsystem state in kernel/cgroup.c |
1da177e4c Linux-2.6.12-rc2 |
267 |
*/ |
5d21cc2db cpuset: replace c... |
268 |
static DEFINE_MUTEX(cpuset_mutex); |
8447a0fee cpuset: convert c... |
269 |
static DEFINE_SPINLOCK(callback_lock); |
4247bdc60 [PATCH] cpuset se... |
270 |
|
cf417141c sched, cpuset: re... |
271 |
/* |
3a5a6d0c2 cpuset: don't nes... |
272 273 274 |
* CPU / memory hotplug is handled asynchronously. */ static void cpuset_hotplug_workfn(struct work_struct *work); |
3a5a6d0c2 cpuset: don't nes... |
275 |
static DECLARE_WORK(cpuset_hotplug_work, cpuset_hotplug_workfn); |
e44193d39 cpuset: let hotpl... |
276 |
static DECLARE_WAIT_QUEUE_HEAD(cpuset_attach_wq); |
3a5a6d0c2 cpuset: don't nes... |
277 |
/* |
cf417141c sched, cpuset: re... |
278 |
* This is ugly, but preserves the userspace API for existing cpuset |
8793d854e Task Control Grou... |
279 |
* users. If someone tries to mount the "cpuset" filesystem, we |
cf417141c sched, cpuset: re... |
280 281 |
* silently switch it to mount "cgroup" instead */ |
f7e835710 convert cgroup an... |
282 283 |
static struct dentry *cpuset_mount(struct file_system_type *fs_type, int flags, const char *unused_dev_name, void *data) |
1da177e4c Linux-2.6.12-rc2 |
284 |
{ |
8793d854e Task Control Grou... |
285 |
struct file_system_type *cgroup_fs = get_fs_type("cgroup"); |
f7e835710 convert cgroup an... |
286 |
struct dentry *ret = ERR_PTR(-ENODEV); |
8793d854e Task Control Grou... |
287 288 289 290 |
if (cgroup_fs) { char mountopts[] = "cpuset,noprefix," "release_agent=/sbin/cpuset_release_agent"; |
f7e835710 convert cgroup an... |
291 292 |
ret = cgroup_fs->mount(cgroup_fs, flags, unused_dev_name, mountopts); |
8793d854e Task Control Grou... |
293 294 295 |
put_filesystem(cgroup_fs); } return ret; |
1da177e4c Linux-2.6.12-rc2 |
296 297 298 299 |
} static struct file_system_type cpuset_fs_type = { .name = "cpuset", |
f7e835710 convert cgroup an... |
300 |
.mount = cpuset_mount, |
1da177e4c Linux-2.6.12-rc2 |
301 |
}; |
1da177e4c Linux-2.6.12-rc2 |
302 |
/* |
300ed6cbb cpuset: convert c... |
303 |
* Return in pmask the portion of a cpusets's cpus_allowed that |
1da177e4c Linux-2.6.12-rc2 |
304 |
* are online. If none are online, walk up the cpuset hierarchy |
40df2deb5 cpuset: cleanup g... |
305 306 |
* until we find one that does have some online cpus. The top * cpuset always has some cpus online. |
1da177e4c Linux-2.6.12-rc2 |
307 308 |
* * One way or another, we guarantee to return some non-empty subset |
5f054e31c documentation: re... |
309 |
* of cpu_online_mask. |
1da177e4c Linux-2.6.12-rc2 |
310 |
* |
8447a0fee cpuset: convert c... |
311 |
* Call with callback_lock or cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
312 |
*/ |
c9710d801 cpuset: drop "con... |
313 |
static void guarantee_online_cpus(struct cpuset *cs, struct cpumask *pmask) |
1da177e4c Linux-2.6.12-rc2 |
314 |
{ |
ae1c80238 cpuset: apply cs-... |
315 |
while (!cpumask_intersects(cs->effective_cpus, cpu_online_mask)) |
c431069fe cpuset: remove cp... |
316 |
cs = parent_cs(cs); |
ae1c80238 cpuset: apply cs-... |
317 |
cpumask_and(pmask, cs->effective_cpus, cpu_online_mask); |
1da177e4c Linux-2.6.12-rc2 |
318 319 320 321 |
} /* * Return in *pmask the portion of a cpusets's mems_allowed that |
0e1e7c7a7 Memoryless nodes:... |
322 323 |
* are online, with memory. If none are online with memory, walk * up the cpuset hierarchy until we find one that does have some |
40df2deb5 cpuset: cleanup g... |
324 |
* online mems. The top cpuset always has some mems online. |
1da177e4c Linux-2.6.12-rc2 |
325 326 |
* * One way or another, we guarantee to return some non-empty subset |
38d7bee9d cpuset: use N_MEM... |
327 |
* of node_states[N_MEMORY]. |
1da177e4c Linux-2.6.12-rc2 |
328 |
* |
8447a0fee cpuset: convert c... |
329 |
* Call with callback_lock or cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
330 |
*/ |
c9710d801 cpuset: drop "con... |
331 |
static void guarantee_online_mems(struct cpuset *cs, nodemask_t *pmask) |
1da177e4c Linux-2.6.12-rc2 |
332 |
{ |
ae1c80238 cpuset: apply cs-... |
333 |
while (!nodes_intersects(cs->effective_mems, node_states[N_MEMORY])) |
c431069fe cpuset: remove cp... |
334 |
cs = parent_cs(cs); |
ae1c80238 cpuset: apply cs-... |
335 |
nodes_and(*pmask, cs->effective_mems, node_states[N_MEMORY]); |
1da177e4c Linux-2.6.12-rc2 |
336 |
} |
f3b39d47e cpusets: restruct... |
337 338 339 |
/* * update task's spread flag if cpuset's page/slab spread flag is set * |
8447a0fee cpuset: convert c... |
340 |
* Call with callback_lock or cpuset_mutex held. |
f3b39d47e cpusets: restruct... |
341 342 343 344 345 |
*/ static void cpuset_update_task_spread_flag(struct cpuset *cs, struct task_struct *tsk) { if (is_spread_page(cs)) |
2ad654bc5 cpuset: PF_SPREAD... |
346 |
task_set_spread_page(tsk); |
f3b39d47e cpusets: restruct... |
347 |
else |
2ad654bc5 cpuset: PF_SPREAD... |
348 |
task_clear_spread_page(tsk); |
f3b39d47e cpusets: restruct... |
349 |
if (is_spread_slab(cs)) |
2ad654bc5 cpuset: PF_SPREAD... |
350 |
task_set_spread_slab(tsk); |
f3b39d47e cpusets: restruct... |
351 |
else |
2ad654bc5 cpuset: PF_SPREAD... |
352 |
task_clear_spread_slab(tsk); |
f3b39d47e cpusets: restruct... |
353 |
} |
1da177e4c Linux-2.6.12-rc2 |
354 355 356 357 358 |
/* * 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 |
5d21cc2db cpuset: replace c... |
359 |
* are only set if the other's are set. Call holding cpuset_mutex. |
1da177e4c Linux-2.6.12-rc2 |
360 361 362 363 |
*/ static int is_cpuset_subset(const struct cpuset *p, const struct cpuset *q) { |
300ed6cbb cpuset: convert c... |
364 |
return cpumask_subset(p->cpus_allowed, q->cpus_allowed) && |
1da177e4c Linux-2.6.12-rc2 |
365 366 367 368 |
nodes_subset(p->mems_allowed, q->mems_allowed) && is_cpu_exclusive(p) <= is_cpu_exclusive(q) && is_mem_exclusive(p) <= is_mem_exclusive(q); } |
645fcc9d2 cpuset: don't all... |
369 370 371 372 |
/** * alloc_trial_cpuset - allocate a trial cpuset * @cs: the cpuset that the trial cpuset duplicates */ |
c9710d801 cpuset: drop "con... |
373 |
static struct cpuset *alloc_trial_cpuset(struct cpuset *cs) |
645fcc9d2 cpuset: don't all... |
374 |
{ |
300ed6cbb cpuset: convert c... |
375 376 377 378 379 |
struct cpuset *trial; trial = kmemdup(cs, sizeof(*cs), GFP_KERNEL); if (!trial) return NULL; |
e2b9a3d7d cpuset: add cs->e... |
380 381 382 383 |
if (!alloc_cpumask_var(&trial->cpus_allowed, GFP_KERNEL)) goto free_cs; if (!alloc_cpumask_var(&trial->effective_cpus, GFP_KERNEL)) goto free_cpus; |
300ed6cbb cpuset: convert c... |
384 |
|
e2b9a3d7d cpuset: add cs->e... |
385 386 |
cpumask_copy(trial->cpus_allowed, cs->cpus_allowed); cpumask_copy(trial->effective_cpus, cs->effective_cpus); |
300ed6cbb cpuset: convert c... |
387 |
return trial; |
e2b9a3d7d cpuset: add cs->e... |
388 389 390 391 392 393 |
free_cpus: free_cpumask_var(trial->cpus_allowed); free_cs: kfree(trial); return NULL; |
645fcc9d2 cpuset: don't all... |
394 395 396 397 398 399 400 401 |
} /** * free_trial_cpuset - free the trial cpuset * @trial: the trial cpuset to be freed */ static void free_trial_cpuset(struct cpuset *trial) { |
e2b9a3d7d cpuset: add cs->e... |
402 |
free_cpumask_var(trial->effective_cpus); |
300ed6cbb cpuset: convert c... |
403 |
free_cpumask_var(trial->cpus_allowed); |
645fcc9d2 cpuset: don't all... |
404 405 |
kfree(trial); } |
1da177e4c Linux-2.6.12-rc2 |
406 407 408 409 410 411 412 |
/* * 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 |
5d21cc2db cpuset: replace c... |
413 |
* cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
414 415 416 417 418 419 420 421 422 423 424 |
* * '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. */ |
c9710d801 cpuset: drop "con... |
425 |
static int validate_change(struct cpuset *cur, struct cpuset *trial) |
1da177e4c Linux-2.6.12-rc2 |
426 |
{ |
492eb21b9 cgroup: make hier... |
427 |
struct cgroup_subsys_state *css; |
1da177e4c Linux-2.6.12-rc2 |
428 |
struct cpuset *c, *par; |
ae8086ce1 cpuset: introduce... |
429 430 431 |
int ret; rcu_read_lock(); |
1da177e4c Linux-2.6.12-rc2 |
432 433 |
/* Each of our child cpusets must be a subset of us */ |
ae8086ce1 cpuset: introduce... |
434 |
ret = -EBUSY; |
492eb21b9 cgroup: make hier... |
435 |
cpuset_for_each_child(c, css, cur) |
ae8086ce1 cpuset: introduce... |
436 437 |
if (!is_cpuset_subset(c, trial)) goto out; |
1da177e4c Linux-2.6.12-rc2 |
438 439 |
/* Remaining checks don't apply to root cpuset */ |
ae8086ce1 cpuset: introduce... |
440 |
ret = 0; |
696040670 [PATCH] cpuset: m... |
441 |
if (cur == &top_cpuset) |
ae8086ce1 cpuset: introduce... |
442 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
443 |
|
c431069fe cpuset: remove cp... |
444 |
par = parent_cs(cur); |
696040670 [PATCH] cpuset: m... |
445 |
|
7e88291be cpuset: make cs->... |
446 |
/* On legacy hiearchy, we must be a subset of our parent cpuset. */ |
ae8086ce1 cpuset: introduce... |
447 |
ret = -EACCES; |
7e88291be cpuset: make cs->... |
448 |
if (!cgroup_on_dfl(cur->css.cgroup) && !is_cpuset_subset(trial, par)) |
ae8086ce1 cpuset: introduce... |
449 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
450 |
|
2df167a30 cgroups: update c... |
451 452 453 454 |
/* * If either I or some sibling (!= me) is exclusive, we can't * overlap */ |
ae8086ce1 cpuset: introduce... |
455 |
ret = -EINVAL; |
492eb21b9 cgroup: make hier... |
456 |
cpuset_for_each_child(c, css, par) { |
1da177e4c Linux-2.6.12-rc2 |
457 458 |
if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) && c != cur && |
300ed6cbb cpuset: convert c... |
459 |
cpumask_intersects(trial->cpus_allowed, c->cpus_allowed)) |
ae8086ce1 cpuset: introduce... |
460 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
461 462 463 |
if ((is_mem_exclusive(trial) || is_mem_exclusive(c)) && c != cur && nodes_intersects(trial->mems_allowed, c->mems_allowed)) |
ae8086ce1 cpuset: introduce... |
464 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
465 |
} |
452477fa6 cpuset: pin down ... |
466 467 |
/* * Cpusets with tasks - existing or newly being attached - can't |
1c09b195d cpuset: fix a reg... |
468 |
* be changed to have empty cpus_allowed or mems_allowed. |
452477fa6 cpuset: pin down ... |
469 |
*/ |
ae8086ce1 cpuset: introduce... |
470 |
ret = -ENOSPC; |
07bc356ed cgroup: implement... |
471 |
if ((cgroup_has_tasks(cur->css.cgroup) || cur->attach_in_progress)) { |
1c09b195d cpuset: fix a reg... |
472 473 474 475 476 477 478 |
if (!cpumask_empty(cur->cpus_allowed) && cpumask_empty(trial->cpus_allowed)) goto out; if (!nodes_empty(cur->mems_allowed) && nodes_empty(trial->mems_allowed)) goto out; } |
020958b62 cpusets: decrusti... |
479 |
|
f82f80426 sched/deadline: E... |
480 481 482 483 484 485 486 487 488 |
/* * We can't shrink if we won't have enough room for SCHED_DEADLINE * tasks. */ ret = -EBUSY; if (is_cpu_exclusive(cur) && !cpuset_cpumask_can_shrink(cur->cpus_allowed, trial->cpus_allowed)) goto out; |
ae8086ce1 cpuset: introduce... |
489 490 491 492 |
ret = 0; out: rcu_read_unlock(); return ret; |
1da177e4c Linux-2.6.12-rc2 |
493 |
} |
db7f47cf4 cpusets: allow cp... |
494 |
#ifdef CONFIG_SMP |
85d7b9498 [PATCH] Dynamic s... |
495 |
/* |
cf417141c sched, cpuset: re... |
496 |
* Helper routine for generate_sched_domains(). |
8b5f1c52d cpuset: use effec... |
497 |
* Do cpusets a, b have overlapping effective cpus_allowed masks? |
029190c51 cpuset sched_load... |
498 |
*/ |
029190c51 cpuset sched_load... |
499 500 |
static int cpusets_overlap(struct cpuset *a, struct cpuset *b) { |
8b5f1c52d cpuset: use effec... |
501 |
return cpumask_intersects(a->effective_cpus, b->effective_cpus); |
029190c51 cpuset sched_load... |
502 |
} |
1d3504fcf sched, cpuset: cu... |
503 504 505 |
static void update_domain_attr(struct sched_domain_attr *dattr, struct cpuset *c) { |
1d3504fcf sched, cpuset: cu... |
506 507 508 509 |
if (dattr->relax_domain_level < c->relax_domain_level) dattr->relax_domain_level = c->relax_domain_level; return; } |
fc560a26a cpuset: replace c... |
510 511 |
static void update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *root_cs) |
f5393693e cpuset: speed up ... |
512 |
{ |
fc560a26a cpuset: replace c... |
513 |
struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
514 |
struct cgroup_subsys_state *pos_css; |
f5393693e cpuset: speed up ... |
515 |
|
fc560a26a cpuset: replace c... |
516 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
517 |
cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { |
bd8815a6d cgroup: make css_... |
518 519 |
if (cp == root_cs) continue; |
fc560a26a cpuset: replace c... |
520 521 |
/* skip the whole subtree if @cp doesn't have any CPU */ if (cpumask_empty(cp->cpus_allowed)) { |
492eb21b9 cgroup: make hier... |
522 |
pos_css = css_rightmost_descendant(pos_css); |
f5393693e cpuset: speed up ... |
523 |
continue; |
fc560a26a cpuset: replace c... |
524 |
} |
f5393693e cpuset: speed up ... |
525 526 527 |
if (is_sched_load_balance(cp)) update_domain_attr(dattr, cp); |
f5393693e cpuset: speed up ... |
528 |
} |
fc560a26a cpuset: replace c... |
529 |
rcu_read_unlock(); |
f5393693e cpuset: speed up ... |
530 |
} |
029190c51 cpuset sched_load... |
531 |
/* |
cf417141c sched, cpuset: re... |
532 533 534 535 536 |
* 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. |
0a0fca9d8 sched: Rename sch... |
537 |
* The output of this function needs to be passed to kernel/sched/core.c |
cf417141c sched, cpuset: re... |
538 539 540 |
* partition_sched_domains() routine, which will rebuild the scheduler's * load balancing domains (sched domains) as specified by that partial * partition. |
029190c51 cpuset sched_load... |
541 |
* |
45ce80fb6 cgroups: consolid... |
542 |
* See "What is sched_load_balance" in Documentation/cgroups/cpusets.txt |
029190c51 cpuset sched_load... |
543 544 545 546 547 548 549 |
* 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. * |
5d21cc2db cpuset: replace c... |
550 |
* Must be called with cpuset_mutex held. |
029190c51 cpuset sched_load... |
551 552 |
* * The three key local variables below are: |
aeed68242 cpuset: clean up ... |
553 |
* q - a linked-list queue of cpuset pointers, used to implement a |
029190c51 cpuset sched_load... |
554 555 556 557 558 559 560 561 562 563 564 565 |
* 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 |
0a0fca9d8 sched: Rename sch... |
566 |
* the kernel/sched/core.c routine partition_sched_domains() in a |
029190c51 cpuset sched_load... |
567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 |
* 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(). */ |
acc3f5d7c cpumask: Partitio... |
585 |
static int generate_sched_domains(cpumask_var_t **domains, |
cf417141c sched, cpuset: re... |
586 |
struct sched_domain_attr **attributes) |
029190c51 cpuset sched_load... |
587 |
{ |
029190c51 cpuset sched_load... |
588 589 590 591 |
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 */ |
acc3f5d7c cpumask: Partitio... |
592 |
cpumask_var_t *doms; /* resulting partition; i.e. sched domains */ |
1d3504fcf sched, cpuset: cu... |
593 |
struct sched_domain_attr *dattr; /* attributes for custom domains */ |
1583715dd sched, cpusets: f... |
594 |
int ndoms = 0; /* number of sched domains in result */ |
6af866af3 cpuset: remove re... |
595 |
int nslot; /* next empty doms[] struct cpumask slot */ |
492eb21b9 cgroup: make hier... |
596 |
struct cgroup_subsys_state *pos_css; |
029190c51 cpuset sched_load... |
597 |
|
029190c51 cpuset sched_load... |
598 |
doms = NULL; |
1d3504fcf sched, cpuset: cu... |
599 |
dattr = NULL; |
cf417141c sched, cpuset: re... |
600 |
csa = NULL; |
029190c51 cpuset sched_load... |
601 602 603 |
/* Special case for the 99% of systems with one, full, sched domain */ if (is_sched_load_balance(&top_cpuset)) { |
acc3f5d7c cpumask: Partitio... |
604 605 |
ndoms = 1; doms = alloc_sched_domains(ndoms); |
029190c51 cpuset sched_load... |
606 |
if (!doms) |
cf417141c sched, cpuset: re... |
607 |
goto done; |
1d3504fcf sched, cpuset: cu... |
608 609 610 |
dattr = kmalloc(sizeof(struct sched_domain_attr), GFP_KERNEL); if (dattr) { *dattr = SD_ATTR_INIT; |
93a655755 cpuset: fix wrong... |
611 |
update_domain_attr_tree(dattr, &top_cpuset); |
1d3504fcf sched, cpuset: cu... |
612 |
} |
8b5f1c52d cpuset: use effec... |
613 |
cpumask_copy(doms[0], top_cpuset.effective_cpus); |
cf417141c sched, cpuset: re... |
614 |
|
cf417141c sched, cpuset: re... |
615 |
goto done; |
029190c51 cpuset sched_load... |
616 |
} |
664eeddee mm: page_alloc: u... |
617 |
csa = kmalloc(nr_cpusets() * sizeof(cp), GFP_KERNEL); |
029190c51 cpuset sched_load... |
618 619 620 |
if (!csa) goto done; csn = 0; |
fc560a26a cpuset: replace c... |
621 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
622 |
cpuset_for_each_descendant_pre(cp, pos_css, &top_cpuset) { |
bd8815a6d cgroup: make css_... |
623 624 |
if (cp == &top_cpuset) continue; |
f5393693e cpuset: speed up ... |
625 |
/* |
fc560a26a cpuset: replace c... |
626 627 628 629 630 631 |
* Continue traversing beyond @cp iff @cp has some CPUs and * isn't load balancing. The former is obvious. The * latter: 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. |
f5393693e cpuset: speed up ... |
632 |
*/ |
fc560a26a cpuset: replace c... |
633 634 |
if (!cpumask_empty(cp->cpus_allowed) && !is_sched_load_balance(cp)) |
f5393693e cpuset: speed up ... |
635 |
continue; |
489a5393a cpuset: don't pas... |
636 |
|
fc560a26a cpuset: replace c... |
637 638 639 640 |
if (is_sched_load_balance(cp)) csa[csn++] = cp; /* skip @cp's subtree */ |
492eb21b9 cgroup: make hier... |
641 |
pos_css = css_rightmost_descendant(pos_css); |
fc560a26a cpuset: replace c... |
642 643 |
} rcu_read_unlock(); |
029190c51 cpuset sched_load... |
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 |
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; } } } |
cf417141c sched, cpuset: re... |
671 672 673 674 |
/* * Now we know how many domains to create. * Convert <csn, csa> to <ndoms, doms> and populate cpu masks. */ |
acc3f5d7c cpumask: Partitio... |
675 |
doms = alloc_sched_domains(ndoms); |
700018e0a cpuset: fix regre... |
676 |
if (!doms) |
cf417141c sched, cpuset: re... |
677 |
goto done; |
cf417141c sched, cpuset: re... |
678 679 680 681 682 |
/* * The rest of the code, including the scheduler, can deal with * dattr==NULL case. No need to abort if alloc fails. */ |
1d3504fcf sched, cpuset: cu... |
683 |
dattr = kmalloc(ndoms * sizeof(struct sched_domain_attr), GFP_KERNEL); |
029190c51 cpuset sched_load... |
684 685 686 |
for (nslot = 0, i = 0; i < csn; i++) { struct cpuset *a = csa[i]; |
6af866af3 cpuset: remove re... |
687 |
struct cpumask *dp; |
029190c51 cpuset sched_load... |
688 |
int apn = a->pn; |
cf417141c sched, cpuset: re... |
689 690 691 692 |
if (apn < 0) { /* Skip completed partitions */ continue; } |
acc3f5d7c cpumask: Partitio... |
693 |
dp = doms[nslot]; |
cf417141c sched, cpuset: re... |
694 695 696 697 |
if (nslot == ndoms) { static int warnings = 10; if (warnings) { |
12d3089c1 kernel/cpuset.c: ... |
698 699 700 |
pr_warn("rebuild_sched_domains confused: nslot %d, ndoms %d, csn %d, i %d, apn %d ", nslot, ndoms, csn, i, apn); |
cf417141c sched, cpuset: re... |
701 |
warnings--; |
029190c51 cpuset sched_load... |
702 |
} |
cf417141c sched, cpuset: re... |
703 704 |
continue; } |
029190c51 cpuset sched_load... |
705 |
|
6af866af3 cpuset: remove re... |
706 |
cpumask_clear(dp); |
cf417141c sched, cpuset: re... |
707 708 709 710 711 712 |
if (dattr) *(dattr + nslot) = SD_ATTR_INIT; for (j = i; j < csn; j++) { struct cpuset *b = csa[j]; if (apn == b->pn) { |
8b5f1c52d cpuset: use effec... |
713 |
cpumask_or(dp, dp, b->effective_cpus); |
cf417141c sched, cpuset: re... |
714 715 716 717 718 |
if (dattr) update_domain_attr_tree(dattr + nslot, b); /* Done with this partition */ b->pn = -1; |
029190c51 cpuset sched_load... |
719 |
} |
029190c51 cpuset sched_load... |
720 |
} |
cf417141c sched, cpuset: re... |
721 |
nslot++; |
029190c51 cpuset sched_load... |
722 723 |
} BUG_ON(nslot != ndoms); |
cf417141c sched, cpuset: re... |
724 725 |
done: kfree(csa); |
700018e0a cpuset: fix regre... |
726 727 728 729 730 731 |
/* * Fallback to the default domain if kmalloc() failed. * See comments in partition_sched_domains(). */ if (doms == NULL) ndoms = 1; |
cf417141c sched, cpuset: re... |
732 733 734 735 736 737 738 739 |
*domains = doms; *attributes = dattr; return ndoms; } /* * Rebuild scheduler domains. * |
699140ba8 cpuset: drop asyn... |
740 741 742 743 744 |
* 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. |
cf417141c sched, cpuset: re... |
745 |
* |
5d21cc2db cpuset: replace c... |
746 |
* Call with cpuset_mutex held. Takes get_online_cpus(). |
cf417141c sched, cpuset: re... |
747 |
*/ |
699140ba8 cpuset: drop asyn... |
748 |
static void rebuild_sched_domains_locked(void) |
cf417141c sched, cpuset: re... |
749 750 |
{ struct sched_domain_attr *attr; |
acc3f5d7c cpumask: Partitio... |
751 |
cpumask_var_t *doms; |
cf417141c sched, cpuset: re... |
752 |
int ndoms; |
5d21cc2db cpuset: replace c... |
753 |
lockdep_assert_held(&cpuset_mutex); |
86ef5c9a8 cpu-hotplug: repl... |
754 |
get_online_cpus(); |
cf417141c sched, cpuset: re... |
755 |
|
5b16c2a49 cpuset: fix cpu h... |
756 757 758 759 760 |
/* * We have raced with CPU hotplug. Don't do anything to avoid * passing doms with offlined cpu to partition_sched_domains(). * Anyways, hotplug work item will rebuild sched domains. */ |
8b5f1c52d cpuset: use effec... |
761 |
if (!cpumask_equal(top_cpuset.effective_cpus, cpu_active_mask)) |
5b16c2a49 cpuset: fix cpu h... |
762 |
goto out; |
cf417141c sched, cpuset: re... |
763 |
/* Generate domain masks and attrs */ |
cf417141c sched, cpuset: re... |
764 |
ndoms = generate_sched_domains(&doms, &attr); |
cf417141c sched, cpuset: re... |
765 766 767 |
/* Have scheduler rebuild the domains */ partition_sched_domains(ndoms, doms, attr); |
5b16c2a49 cpuset: fix cpu h... |
768 |
out: |
86ef5c9a8 cpu-hotplug: repl... |
769 |
put_online_cpus(); |
cf417141c sched, cpuset: re... |
770 |
} |
db7f47cf4 cpusets: allow cp... |
771 |
#else /* !CONFIG_SMP */ |
699140ba8 cpuset: drop asyn... |
772 |
static void rebuild_sched_domains_locked(void) |
db7f47cf4 cpusets: allow cp... |
773 774 |
{ } |
db7f47cf4 cpusets: allow cp... |
775 |
#endif /* CONFIG_SMP */ |
029190c51 cpuset sched_load... |
776 |
|
cf417141c sched, cpuset: re... |
777 778 |
void rebuild_sched_domains(void) { |
5d21cc2db cpuset: replace c... |
779 |
mutex_lock(&cpuset_mutex); |
699140ba8 cpuset: drop asyn... |
780 |
rebuild_sched_domains_locked(); |
5d21cc2db cpuset: replace c... |
781 |
mutex_unlock(&cpuset_mutex); |
029190c51 cpuset sched_load... |
782 |
} |
58f4790b7 cpusets: update_c... |
783 |
/** |
0b2f630a2 cpusets: restruct... |
784 785 |
* 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 |
0b2f630a2 cpusets: restruct... |
786 |
* |
d66393e54 cpuset: use css_t... |
787 788 789 |
* Iterate through each task of @cs updating its cpus_allowed to the * effective cpuset's. As this function is called with cpuset_mutex held, * cpuset membership stays stable. |
0b2f630a2 cpusets: restruct... |
790 |
*/ |
d66393e54 cpuset: use css_t... |
791 |
static void update_tasks_cpumask(struct cpuset *cs) |
0b2f630a2 cpusets: restruct... |
792 |
{ |
d66393e54 cpuset: use css_t... |
793 794 795 796 797 |
struct css_task_iter it; struct task_struct *task; css_task_iter_start(&cs->css, &it); while ((task = css_task_iter_next(&it))) |
ae1c80238 cpuset: apply cs-... |
798 |
set_cpus_allowed_ptr(task, cs->effective_cpus); |
d66393e54 cpuset: use css_t... |
799 |
css_task_iter_end(&it); |
0b2f630a2 cpusets: restruct... |
800 |
} |
5c5cc6232 cpuset: allow to ... |
801 |
/* |
734d45130 cpuset: update cs... |
802 803 804 805 806 807 |
* update_cpumasks_hier - Update effective cpumasks and tasks in the subtree * @cs: the cpuset to consider * @new_cpus: temp variable for calculating new effective_cpus * * When congifured cpumask is changed, the effective cpumasks of this cpuset * and all its descendants need to be updated. |
5c5cc6232 cpuset: allow to ... |
808 |
* |
734d45130 cpuset: update cs... |
809 |
* On legacy hierachy, effective_cpus will be the same with cpu_allowed. |
5c5cc6232 cpuset: allow to ... |
810 811 812 |
* * Called with cpuset_mutex held */ |
734d45130 cpuset: update cs... |
813 |
static void update_cpumasks_hier(struct cpuset *cs, struct cpumask *new_cpus) |
5c5cc6232 cpuset: allow to ... |
814 815 |
{ struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
816 |
struct cgroup_subsys_state *pos_css; |
8b5f1c52d cpuset: use effec... |
817 |
bool need_rebuild_sched_domains = false; |
5c5cc6232 cpuset: allow to ... |
818 819 |
rcu_read_lock(); |
734d45130 cpuset: update cs... |
820 821 822 823 |
cpuset_for_each_descendant_pre(cp, pos_css, cs) { struct cpuset *parent = parent_cs(cp); cpumask_and(new_cpus, cp->cpus_allowed, parent->effective_cpus); |
554b0d1c8 cpuset: inherit a... |
824 825 826 827 828 829 |
/* * If it becomes empty, inherit the effective mask of the * parent, which is guaranteed to have some CPUs. */ if (cpumask_empty(new_cpus)) cpumask_copy(new_cpus, parent->effective_cpus); |
734d45130 cpuset: update cs... |
830 831 832 833 |
/* Skip the whole subtree if the cpumask remains the same. */ if (cpumask_equal(new_cpus, cp->effective_cpus)) { pos_css = css_rightmost_descendant(pos_css); continue; |
5c5cc6232 cpuset: allow to ... |
834 |
} |
734d45130 cpuset: update cs... |
835 |
|
ec903c0c8 cgroup: rename cs... |
836 |
if (!css_tryget_online(&cp->css)) |
5c5cc6232 cpuset: allow to ... |
837 838 |
continue; rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
839 |
spin_lock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
840 |
cpumask_copy(cp->effective_cpus, new_cpus); |
8447a0fee cpuset: convert c... |
841 |
spin_unlock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
842 843 844 |
WARN_ON(!cgroup_on_dfl(cp->css.cgroup) && !cpumask_equal(cp->cpus_allowed, cp->effective_cpus)); |
d66393e54 cpuset: use css_t... |
845 |
update_tasks_cpumask(cp); |
5c5cc6232 cpuset: allow to ... |
846 |
|
8b5f1c52d cpuset: use effec... |
847 848 849 850 851 852 853 |
/* * If the effective cpumask of any non-empty cpuset is changed, * we need to rebuild sched domains. */ if (!cpumask_empty(cp->cpus_allowed) && is_sched_load_balance(cp)) need_rebuild_sched_domains = true; |
5c5cc6232 cpuset: allow to ... |
854 855 856 857 |
rcu_read_lock(); css_put(&cp->css); } rcu_read_unlock(); |
8b5f1c52d cpuset: use effec... |
858 859 860 |
if (need_rebuild_sched_domains) rebuild_sched_domains_locked(); |
5c5cc6232 cpuset: allow to ... |
861 |
} |
0b2f630a2 cpusets: restruct... |
862 |
/** |
58f4790b7 cpusets: update_c... |
863 864 |
* update_cpumask - update the cpus_allowed mask of a cpuset and all tasks in it * @cs: the cpuset to consider |
fc34ac1dc kernel/cpuset.c: ... |
865 |
* @trialcs: trial cpuset |
58f4790b7 cpusets: update_c... |
866 867 |
* @buf: buffer of cpu numbers written to this cpuset */ |
645fcc9d2 cpuset: don't all... |
868 869 |
static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) |
1da177e4c Linux-2.6.12-rc2 |
870 |
{ |
58f4790b7 cpusets: update_c... |
871 |
int retval; |
1da177e4c Linux-2.6.12-rc2 |
872 |
|
5f054e31c documentation: re... |
873 |
/* top_cpuset.cpus_allowed tracks cpu_online_mask; it's read-only */ |
4c4d50f7b [PATCH] cpuset: t... |
874 875 |
if (cs == &top_cpuset) return -EACCES; |
6f7f02e78 cpusets: allow em... |
876 |
/* |
c8d9c90c7 hotplug cpu: move... |
877 |
* An empty cpus_allowed is ok only if the cpuset has no tasks. |
020958b62 cpusets: decrusti... |
878 879 880 |
* Since cpulist_parse() fails on an empty mask, we special case * that parsing. The validate_change() call ensures that cpusets * with tasks have cpus. |
6f7f02e78 cpusets: allow em... |
881 |
*/ |
020958b62 cpusets: decrusti... |
882 |
if (!*buf) { |
300ed6cbb cpuset: convert c... |
883 |
cpumask_clear(trialcs->cpus_allowed); |
6f7f02e78 cpusets: allow em... |
884 |
} else { |
300ed6cbb cpuset: convert c... |
885 |
retval = cpulist_parse(buf, trialcs->cpus_allowed); |
6f7f02e78 cpusets: allow em... |
886 887 |
if (retval < 0) return retval; |
37340746a cpusets: fix bug ... |
888 |
|
5d8ba82c3 cpuset: allow wri... |
889 890 |
if (!cpumask_subset(trialcs->cpus_allowed, top_cpuset.cpus_allowed)) |
37340746a cpusets: fix bug ... |
891 |
return -EINVAL; |
6f7f02e78 cpusets: allow em... |
892 |
} |
029190c51 cpuset sched_load... |
893 |
|
8707d8b8c Fix cpusets updat... |
894 |
/* Nothing to do if the cpus didn't change */ |
300ed6cbb cpuset: convert c... |
895 |
if (cpumask_equal(cs->cpus_allowed, trialcs->cpus_allowed)) |
8707d8b8c Fix cpusets updat... |
896 |
return 0; |
58f4790b7 cpusets: update_c... |
897 |
|
a73456f37 cpuset: re-struct... |
898 899 900 |
retval = validate_change(cs, trialcs); if (retval < 0) return retval; |
8447a0fee cpuset: convert c... |
901 |
spin_lock_irq(&callback_lock); |
300ed6cbb cpuset: convert c... |
902 |
cpumask_copy(cs->cpus_allowed, trialcs->cpus_allowed); |
8447a0fee cpuset: convert c... |
903 |
spin_unlock_irq(&callback_lock); |
029190c51 cpuset sched_load... |
904 |
|
734d45130 cpuset: update cs... |
905 906 |
/* use trialcs->cpus_allowed as a temp variable */ update_cpumasks_hier(cs, trialcs->cpus_allowed); |
85d7b9498 [PATCH] Dynamic s... |
907 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
908 |
} |
053199edf [PATCH] cpusets: ... |
909 |
/* |
e4e364e86 [PATCH] cpuset: m... |
910 911 912 913 914 915 916 |
* 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. * |
e4e364e86 [PATCH] cpuset: m... |
917 918 919 920 |
* 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. |
e4e364e86 [PATCH] cpuset: m... |
921 922 923 924 925 926 |
*/ static void cpuset_migrate_mm(struct mm_struct *mm, const nodemask_t *from, const nodemask_t *to) { struct task_struct *tsk = current; |
e4e364e86 [PATCH] cpuset: m... |
927 |
tsk->mems_allowed = *to; |
e4e364e86 [PATCH] cpuset: m... |
928 929 |
do_migrate_pages(mm, from, to, MPOL_MF_MOVE_ALL); |
472958300 cpuset: fix a loc... |
930 |
rcu_read_lock(); |
ae1c80238 cpuset: apply cs-... |
931 |
guarantee_online_mems(task_cs(tsk), &tsk->mems_allowed); |
472958300 cpuset: fix a loc... |
932 |
rcu_read_unlock(); |
e4e364e86 [PATCH] cpuset: m... |
933 |
} |
3b6766fe6 cpuset: rewrite u... |
934 |
/* |
58568d2a8 cpuset,mm: update... |
935 936 937 938 939 940 941 |
* cpuset_change_task_nodemask - change task's mems_allowed and mempolicy * @tsk: the task to change * @newmems: new nodes that the task will be set * * In order to avoid seeing no nodes if the old and new nodes are disjoint, * we structure updates as setting all new allowed nodes, then clearing newly * disallowed ones. |
58568d2a8 cpuset,mm: update... |
942 943 944 945 |
*/ static void cpuset_change_task_nodemask(struct task_struct *tsk, nodemask_t *newmems) { |
b246272ec cpusets: stall wh... |
946 |
bool need_loop; |
89e8a244b cpusets: avoid lo... |
947 |
|
c0ff7453b cpuset,mm: fix no... |
948 949 950 951 952 953 954 955 956 957 |
/* * 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; if (current->flags & PF_EXITING) /* Let dying task have memory */ return; task_lock(tsk); |
b246272ec cpusets: stall wh... |
958 959 |
/* * Determine if a loop is necessary if another thread is doing |
d26914d11 mm: optimize put_... |
960 |
* read_mems_allowed_begin(). If at least one node remains unchanged and |
b246272ec cpusets: stall wh... |
961 962 963 964 965 |
* tsk does not have a mempolicy, then an empty nodemask will not be * possible when mems_allowed is larger than a word. */ need_loop = task_has_mempolicy(tsk) || !nodes_intersects(*newmems, tsk->mems_allowed); |
c0ff7453b cpuset,mm: fix no... |
966 |
|
0fc0287c9 cpuset: Fix memor... |
967 968 |
if (need_loop) { local_irq_disable(); |
cc9a6c877 cpuset: mm: reduc... |
969 |
write_seqcount_begin(&tsk->mems_allowed_seq); |
0fc0287c9 cpuset: Fix memor... |
970 |
} |
c0ff7453b cpuset,mm: fix no... |
971 |
|
cc9a6c877 cpuset: mm: reduc... |
972 973 |
nodes_or(tsk->mems_allowed, tsk->mems_allowed, *newmems); mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP1); |
c0ff7453b cpuset,mm: fix no... |
974 975 |
mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP2); |
58568d2a8 cpuset,mm: update... |
976 |
tsk->mems_allowed = *newmems; |
cc9a6c877 cpuset: mm: reduc... |
977 |
|
0fc0287c9 cpuset: Fix memor... |
978 |
if (need_loop) { |
cc9a6c877 cpuset: mm: reduc... |
979 |
write_seqcount_end(&tsk->mems_allowed_seq); |
0fc0287c9 cpuset: Fix memor... |
980 981 |
local_irq_enable(); } |
cc9a6c877 cpuset: mm: reduc... |
982 |
|
c0ff7453b cpuset,mm: fix no... |
983 |
task_unlock(tsk); |
58568d2a8 cpuset,mm: update... |
984 |
} |
8793d854e Task Control Grou... |
985 |
static void *cpuset_being_rebound; |
0b2f630a2 cpusets: restruct... |
986 987 988 |
/** * 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 |
0b2f630a2 cpusets: restruct... |
989 |
* |
d66393e54 cpuset: use css_t... |
990 991 992 |
* Iterate through each task of @cs updating its mems_allowed to the * effective cpuset's. As this function is called with cpuset_mutex held, * cpuset membership stays stable. |
0b2f630a2 cpusets: restruct... |
993 |
*/ |
d66393e54 cpuset: use css_t... |
994 |
static void update_tasks_nodemask(struct cpuset *cs) |
1da177e4c Linux-2.6.12-rc2 |
995 |
{ |
33ad801df cpuset: record ol... |
996 |
static nodemask_t newmems; /* protected by cpuset_mutex */ |
d66393e54 cpuset: use css_t... |
997 998 |
struct css_task_iter it; struct task_struct *task; |
59dac16fb [PATCH] cpuset: u... |
999 |
|
846a16bf0 mempolicy: rename... |
1000 |
cpuset_being_rebound = cs; /* causes mpol_dup() rebind */ |
4225399a6 [PATCH] cpuset: r... |
1001 |
|
ae1c80238 cpuset: apply cs-... |
1002 |
guarantee_online_mems(cs, &newmems); |
33ad801df cpuset: record ol... |
1003 |
|
4225399a6 [PATCH] cpuset: r... |
1004 |
/* |
3b6766fe6 cpuset: rewrite u... |
1005 1006 1007 1008 |
* The mpol_rebind_mm() call takes mmap_sem, which we couldn't * take while holding tasklist_lock. Forks can happen - the * mpol_dup() cpuset_being_rebound check will catch such forks, * and rebind their vma mempolicies too. Because we still hold |
5d21cc2db cpuset: replace c... |
1009 |
* the global cpuset_mutex, we know that no other rebind effort |
3b6766fe6 cpuset: rewrite u... |
1010 |
* will be contending for the global variable cpuset_being_rebound. |
4225399a6 [PATCH] cpuset: r... |
1011 |
* It's ok if we rebind the same mm twice; mpol_rebind_mm() |
04c19fa6f [PATCH] cpuset: m... |
1012 |
* is idempotent. Also migrate pages in each mm to new nodes. |
4225399a6 [PATCH] cpuset: r... |
1013 |
*/ |
d66393e54 cpuset: use css_t... |
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 |
css_task_iter_start(&cs->css, &it); while ((task = css_task_iter_next(&it))) { struct mm_struct *mm; bool migrate; cpuset_change_task_nodemask(task, &newmems); mm = get_task_mm(task); if (!mm) continue; migrate = is_memory_migrate(cs); mpol_rebind_mm(mm, &cs->mems_allowed); if (migrate) cpuset_migrate_mm(mm, &cs->old_mems_allowed, &newmems); mmput(mm); } css_task_iter_end(&it); |
4225399a6 [PATCH] cpuset: r... |
1033 |
|
33ad801df cpuset: record ol... |
1034 1035 1036 1037 1038 |
/* * All the tasks' nodemasks have been updated, update * cs->old_mems_allowed. */ cs->old_mems_allowed = newmems; |
2df167a30 cgroups: update c... |
1039 |
/* We're done rebinding vmas to this cpuset's new mems_allowed. */ |
8793d854e Task Control Grou... |
1040 |
cpuset_being_rebound = NULL; |
1da177e4c Linux-2.6.12-rc2 |
1041 |
} |
0b2f630a2 cpusets: restruct... |
1042 |
/* |
734d45130 cpuset: update cs... |
1043 1044 1045 |
* update_nodemasks_hier - Update effective nodemasks and tasks in the subtree * @cs: the cpuset to consider * @new_mems: a temp variable for calculating new effective_mems |
5c5cc6232 cpuset: allow to ... |
1046 |
* |
734d45130 cpuset: update cs... |
1047 1048 |
* When configured nodemask is changed, the effective nodemasks of this cpuset * and all its descendants need to be updated. |
5c5cc6232 cpuset: allow to ... |
1049 |
* |
734d45130 cpuset: update cs... |
1050 |
* On legacy hiearchy, effective_mems will be the same with mems_allowed. |
5c5cc6232 cpuset: allow to ... |
1051 1052 1053 |
* * Called with cpuset_mutex held */ |
734d45130 cpuset: update cs... |
1054 |
static void update_nodemasks_hier(struct cpuset *cs, nodemask_t *new_mems) |
5c5cc6232 cpuset: allow to ... |
1055 1056 |
{ struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
1057 |
struct cgroup_subsys_state *pos_css; |
5c5cc6232 cpuset: allow to ... |
1058 1059 |
rcu_read_lock(); |
734d45130 cpuset: update cs... |
1060 1061 1062 1063 |
cpuset_for_each_descendant_pre(cp, pos_css, cs) { struct cpuset *parent = parent_cs(cp); nodes_and(*new_mems, cp->mems_allowed, parent->effective_mems); |
554b0d1c8 cpuset: inherit a... |
1064 1065 1066 1067 1068 1069 |
/* * If it becomes empty, inherit the effective mask of the * parent, which is guaranteed to have some MEMs. */ if (nodes_empty(*new_mems)) *new_mems = parent->effective_mems; |
734d45130 cpuset: update cs... |
1070 1071 1072 1073 |
/* Skip the whole subtree if the nodemask remains the same. */ if (nodes_equal(*new_mems, cp->effective_mems)) { pos_css = css_rightmost_descendant(pos_css); continue; |
5c5cc6232 cpuset: allow to ... |
1074 |
} |
734d45130 cpuset: update cs... |
1075 |
|
ec903c0c8 cgroup: rename cs... |
1076 |
if (!css_tryget_online(&cp->css)) |
5c5cc6232 cpuset: allow to ... |
1077 1078 |
continue; rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
1079 |
spin_lock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
1080 |
cp->effective_mems = *new_mems; |
8447a0fee cpuset: convert c... |
1081 |
spin_unlock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
1082 1083 |
WARN_ON(!cgroup_on_dfl(cp->css.cgroup) && |
a13812683 cpuset: fix the W... |
1084 |
!nodes_equal(cp->mems_allowed, cp->effective_mems)); |
734d45130 cpuset: update cs... |
1085 |
|
d66393e54 cpuset: use css_t... |
1086 |
update_tasks_nodemask(cp); |
5c5cc6232 cpuset: allow to ... |
1087 1088 1089 1090 1091 1092 1093 1094 |
rcu_read_lock(); css_put(&cp->css); } rcu_read_unlock(); } /* |
0b2f630a2 cpusets: restruct... |
1095 1096 |
* Handle user request to change the 'mems' memory placement * of a cpuset. Needs to validate the request, update the |
58568d2a8 cpuset,mm: update... |
1097 1098 1099 1100 |
* cpusets mems_allowed, and for each task in the cpuset, * update mems_allowed and rebind task's mempolicy and any vma * mempolicies and if the cpuset is marked 'memory_migrate', * migrate the tasks pages to the new memory. |
0b2f630a2 cpusets: restruct... |
1101 |
* |
8447a0fee cpuset: convert c... |
1102 |
* Call with cpuset_mutex held. May take callback_lock during call. |
0b2f630a2 cpusets: restruct... |
1103 1104 1105 1106 |
* 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. */ |
645fcc9d2 cpuset: don't all... |
1107 1108 |
static int update_nodemask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) |
0b2f630a2 cpusets: restruct... |
1109 |
{ |
0b2f630a2 cpusets: restruct... |
1110 1111 1112 |
int retval; /* |
38d7bee9d cpuset: use N_MEM... |
1113 |
* top_cpuset.mems_allowed tracks node_stats[N_MEMORY]; |
0b2f630a2 cpusets: restruct... |
1114 1115 |
* it's read-only */ |
53feb2976 cpuset: alloc nod... |
1116 1117 1118 1119 |
if (cs == &top_cpuset) { retval = -EACCES; goto done; } |
0b2f630a2 cpusets: restruct... |
1120 |
|
0b2f630a2 cpusets: restruct... |
1121 1122 1123 1124 1125 1126 1127 |
/* * 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) { |
645fcc9d2 cpuset: don't all... |
1128 |
nodes_clear(trialcs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1129 |
} else { |
645fcc9d2 cpuset: don't all... |
1130 |
retval = nodelist_parse(buf, trialcs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1131 1132 |
if (retval < 0) goto done; |
645fcc9d2 cpuset: don't all... |
1133 |
if (!nodes_subset(trialcs->mems_allowed, |
5d8ba82c3 cpuset: allow wri... |
1134 1135 |
top_cpuset.mems_allowed)) { retval = -EINVAL; |
53feb2976 cpuset: alloc nod... |
1136 1137 |
goto done; } |
0b2f630a2 cpusets: restruct... |
1138 |
} |
33ad801df cpuset: record ol... |
1139 1140 |
if (nodes_equal(cs->mems_allowed, trialcs->mems_allowed)) { |
0b2f630a2 cpusets: restruct... |
1141 1142 1143 |
retval = 0; /* Too easy - nothing to do */ goto done; } |
645fcc9d2 cpuset: don't all... |
1144 |
retval = validate_change(cs, trialcs); |
0b2f630a2 cpusets: restruct... |
1145 1146 |
if (retval < 0) goto done; |
8447a0fee cpuset: convert c... |
1147 |
spin_lock_irq(&callback_lock); |
645fcc9d2 cpuset: don't all... |
1148 |
cs->mems_allowed = trialcs->mems_allowed; |
8447a0fee cpuset: convert c... |
1149 |
spin_unlock_irq(&callback_lock); |
0b2f630a2 cpusets: restruct... |
1150 |
|
734d45130 cpuset: update cs... |
1151 1152 |
/* use trialcs->mems_allowed as a temp variable */ update_nodemasks_hier(cs, &cs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1153 1154 1155 |
done: return retval; } |
8793d854e Task Control Grou... |
1156 1157 |
int current_cpuset_is_being_rebound(void) { |
391acf970 cpuset,mempolicy:... |
1158 1159 1160 1161 1162 1163 1164 |
int ret; rcu_read_lock(); ret = task_cs(current) == cpuset_being_rebound; rcu_read_unlock(); return ret; |
8793d854e Task Control Grou... |
1165 |
} |
5be7a4792 Fix cpuset sched_... |
1166 |
static int update_relax_domain_level(struct cpuset *cs, s64 val) |
1d3504fcf sched, cpuset: cu... |
1167 |
{ |
db7f47cf4 cpusets: allow cp... |
1168 |
#ifdef CONFIG_SMP |
60495e776 sched: Dynamic sc... |
1169 |
if (val < -1 || val >= sched_domain_level_max) |
30e0e1781 cpuset: limit the... |
1170 |
return -EINVAL; |
db7f47cf4 cpusets: allow cp... |
1171 |
#endif |
1d3504fcf sched, cpuset: cu... |
1172 1173 1174 |
if (val != cs->relax_domain_level) { cs->relax_domain_level = val; |
300ed6cbb cpuset: convert c... |
1175 1176 |
if (!cpumask_empty(cs->cpus_allowed) && is_sched_load_balance(cs)) |
699140ba8 cpuset: drop asyn... |
1177 |
rebuild_sched_domains_locked(); |
1d3504fcf sched, cpuset: cu... |
1178 1179 1180 1181 |
} return 0; } |
72ec70299 cgroup: make task... |
1182 |
/** |
950592f7b cpusets: update t... |
1183 1184 |
* update_tasks_flags - update the spread flags of tasks in the cpuset. * @cs: the cpuset in which each task's spread flags needs to be changed |
950592f7b cpusets: update t... |
1185 |
* |
d66393e54 cpuset: use css_t... |
1186 1187 1188 |
* Iterate through each task of @cs updating its spread flags. As this * function is called with cpuset_mutex held, cpuset membership stays * stable. |
950592f7b cpusets: update t... |
1189 |
*/ |
d66393e54 cpuset: use css_t... |
1190 |
static void update_tasks_flags(struct cpuset *cs) |
950592f7b cpusets: update t... |
1191 |
{ |
d66393e54 cpuset: use css_t... |
1192 1193 1194 1195 1196 1197 1198 |
struct css_task_iter it; struct task_struct *task; css_task_iter_start(&cs->css, &it); while ((task = css_task_iter_next(&it))) cpuset_update_task_spread_flag(cs, task); css_task_iter_end(&it); |
950592f7b cpusets: update t... |
1199 1200 1201 |
} /* |
1da177e4c Linux-2.6.12-rc2 |
1202 |
* update_flag - read a 0 or a 1 in a file and update associated flag |
786083667 Cpuset hardwall f... |
1203 1204 1205 |
* bit: the bit to update (see cpuset_flagbits_t) * cs: the cpuset to update * turning_on: whether the flag is being set or cleared |
053199edf [PATCH] cpusets: ... |
1206 |
* |
5d21cc2db cpuset: replace c... |
1207 |
* Call with cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
1208 |
*/ |
700fe1ab9 CGroup API files:... |
1209 1210 |
static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, int turning_on) |
1da177e4c Linux-2.6.12-rc2 |
1211 |
{ |
645fcc9d2 cpuset: don't all... |
1212 |
struct cpuset *trialcs; |
40b6a7623 cpuset.c: remove ... |
1213 |
int balance_flag_changed; |
950592f7b cpusets: update t... |
1214 |
int spread_flag_changed; |
950592f7b cpusets: update t... |
1215 |
int err; |
1da177e4c Linux-2.6.12-rc2 |
1216 |
|
645fcc9d2 cpuset: don't all... |
1217 1218 1219 |
trialcs = alloc_trial_cpuset(cs); if (!trialcs) return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
1220 |
if (turning_on) |
645fcc9d2 cpuset: don't all... |
1221 |
set_bit(bit, &trialcs->flags); |
1da177e4c Linux-2.6.12-rc2 |
1222 |
else |
645fcc9d2 cpuset: don't all... |
1223 |
clear_bit(bit, &trialcs->flags); |
1da177e4c Linux-2.6.12-rc2 |
1224 |
|
645fcc9d2 cpuset: don't all... |
1225 |
err = validate_change(cs, trialcs); |
85d7b9498 [PATCH] Dynamic s... |
1226 |
if (err < 0) |
645fcc9d2 cpuset: don't all... |
1227 |
goto out; |
029190c51 cpuset sched_load... |
1228 |
|
029190c51 cpuset sched_load... |
1229 |
balance_flag_changed = (is_sched_load_balance(cs) != |
645fcc9d2 cpuset: don't all... |
1230 |
is_sched_load_balance(trialcs)); |
029190c51 cpuset sched_load... |
1231 |
|
950592f7b cpusets: update t... |
1232 1233 |
spread_flag_changed = ((is_spread_slab(cs) != is_spread_slab(trialcs)) || (is_spread_page(cs) != is_spread_page(trialcs))); |
8447a0fee cpuset: convert c... |
1234 |
spin_lock_irq(&callback_lock); |
645fcc9d2 cpuset: don't all... |
1235 |
cs->flags = trialcs->flags; |
8447a0fee cpuset: convert c... |
1236 |
spin_unlock_irq(&callback_lock); |
85d7b9498 [PATCH] Dynamic s... |
1237 |
|
300ed6cbb cpuset: convert c... |
1238 |
if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed) |
699140ba8 cpuset: drop asyn... |
1239 |
rebuild_sched_domains_locked(); |
029190c51 cpuset sched_load... |
1240 |
|
950592f7b cpusets: update t... |
1241 |
if (spread_flag_changed) |
d66393e54 cpuset: use css_t... |
1242 |
update_tasks_flags(cs); |
645fcc9d2 cpuset: don't all... |
1243 1244 1245 |
out: free_trial_cpuset(trialcs); return err; |
1da177e4c Linux-2.6.12-rc2 |
1246 |
} |
053199edf [PATCH] cpusets: ... |
1247 |
/* |
80f7228b5 typo fixes: occur... |
1248 |
* Frequency meter - How fast is some event occurring? |
3e0d98b9f [PATCH] cpuset: m... |
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 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 |
* * 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; } |
57fce0a68 cpuset: don't use... |
1344 |
static struct cpuset *cpuset_attach_old_cs; |
5d21cc2db cpuset: replace c... |
1345 |
/* Called by cgroups to determine if a cpuset is usable; cpuset_mutex held */ |
eb95419b0 cgroup: pass arou... |
1346 1347 |
static int cpuset_can_attach(struct cgroup_subsys_state *css, struct cgroup_taskset *tset) |
f780bdb7c cgroups: add per-... |
1348 |
{ |
eb95419b0 cgroup: pass arou... |
1349 |
struct cpuset *cs = css_cs(css); |
bb9d97b6d cgroup: don't use... |
1350 1351 |
struct task_struct *task; int ret; |
1da177e4c Linux-2.6.12-rc2 |
1352 |
|
57fce0a68 cpuset: don't use... |
1353 1354 |
/* used later by cpuset_attach() */ cpuset_attach_old_cs = task_cs(cgroup_taskset_first(tset)); |
5d21cc2db cpuset: replace c... |
1355 |
mutex_lock(&cpuset_mutex); |
aa6ec29be cgroup: remove sa... |
1356 |
/* allow moving tasks into an empty cpuset if on default hierarchy */ |
5d21cc2db cpuset: replace c... |
1357 |
ret = -ENOSPC; |
aa6ec29be cgroup: remove sa... |
1358 |
if (!cgroup_on_dfl(css->cgroup) && |
88fa523bf cpuset: allow to ... |
1359 |
(cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed))) |
5d21cc2db cpuset: replace c... |
1360 |
goto out_unlock; |
9985b0bab sched: prevent bo... |
1361 |
|
924f0d9a2 cgroup: drop @ski... |
1362 |
cgroup_taskset_for_each(task, tset) { |
7f51412a4 sched/deadline: F... |
1363 1364 |
ret = task_can_attach(task, cs->cpus_allowed); if (ret) |
5d21cc2db cpuset: replace c... |
1365 1366 1367 1368 |
goto out_unlock; ret = security_task_setscheduler(task); if (ret) goto out_unlock; |
bb9d97b6d cgroup: don't use... |
1369 |
} |
f780bdb7c cgroups: add per-... |
1370 |
|
452477fa6 cpuset: pin down ... |
1371 1372 1373 1374 1375 |
/* * Mark attach is in progress. This makes validate_change() fail * changes which zero cpus/mems_allowed. */ cs->attach_in_progress++; |
5d21cc2db cpuset: replace c... |
1376 1377 1378 1379 |
ret = 0; out_unlock: mutex_unlock(&cpuset_mutex); return ret; |
8793d854e Task Control Grou... |
1380 |
} |
f780bdb7c cgroups: add per-... |
1381 |
|
eb95419b0 cgroup: pass arou... |
1382 |
static void cpuset_cancel_attach(struct cgroup_subsys_state *css, |
452477fa6 cpuset: pin down ... |
1383 1384 |
struct cgroup_taskset *tset) { |
5d21cc2db cpuset: replace c... |
1385 |
mutex_lock(&cpuset_mutex); |
eb95419b0 cgroup: pass arou... |
1386 |
css_cs(css)->attach_in_progress--; |
5d21cc2db cpuset: replace c... |
1387 |
mutex_unlock(&cpuset_mutex); |
8793d854e Task Control Grou... |
1388 |
} |
1da177e4c Linux-2.6.12-rc2 |
1389 |
|
4e4c9a140 cpuset: cleanup c... |
1390 |
/* |
5d21cc2db cpuset: replace c... |
1391 |
* Protected by cpuset_mutex. cpus_attach is used only by cpuset_attach() |
4e4c9a140 cpuset: cleanup c... |
1392 1393 1394 1395 |
* but we can't allocate it dynamically there. Define it global and * allocate from cpuset_init(). */ static cpumask_var_t cpus_attach; |
eb95419b0 cgroup: pass arou... |
1396 1397 |
static void cpuset_attach(struct cgroup_subsys_state *css, struct cgroup_taskset *tset) |
8793d854e Task Control Grou... |
1398 |
{ |
67bd2c598 cpuset: remove un... |
1399 |
/* static buf protected by cpuset_mutex */ |
4e4c9a140 cpuset: cleanup c... |
1400 |
static nodemask_t cpuset_attach_nodemask_to; |
8793d854e Task Control Grou... |
1401 |
struct mm_struct *mm; |
bb9d97b6d cgroup: don't use... |
1402 1403 |
struct task_struct *task; struct task_struct *leader = cgroup_taskset_first(tset); |
eb95419b0 cgroup: pass arou... |
1404 |
struct cpuset *cs = css_cs(css); |
57fce0a68 cpuset: don't use... |
1405 |
struct cpuset *oldcs = cpuset_attach_old_cs; |
22fb52dd7 [PATCH] SELinux: ... |
1406 |
|
5d21cc2db cpuset: replace c... |
1407 |
mutex_lock(&cpuset_mutex); |
4e4c9a140 cpuset: cleanup c... |
1408 1409 1410 1411 |
/* prepare for attach */ if (cs == &top_cpuset) cpumask_copy(cpus_attach, cpu_possible_mask); else |
ae1c80238 cpuset: apply cs-... |
1412 |
guarantee_online_cpus(cs, cpus_attach); |
4e4c9a140 cpuset: cleanup c... |
1413 |
|
ae1c80238 cpuset: apply cs-... |
1414 |
guarantee_online_mems(cs, &cpuset_attach_nodemask_to); |
4e4c9a140 cpuset: cleanup c... |
1415 |
|
924f0d9a2 cgroup: drop @ski... |
1416 |
cgroup_taskset_for_each(task, tset) { |
bb9d97b6d cgroup: don't use... |
1417 1418 1419 1420 1421 1422 1423 1424 1425 |
/* * can_attach beforehand should guarantee that this doesn't * fail. TODO: have a better way to handle failure here */ WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach)); cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to); cpuset_update_task_spread_flag(cs, task); } |
22fb52dd7 [PATCH] SELinux: ... |
1426 |
|
f780bdb7c cgroups: add per-... |
1427 1428 1429 1430 |
/* * Change mm, possibly for multiple threads in a threadgroup. This is * expensive and may sleep. */ |
ae1c80238 cpuset: apply cs-... |
1431 |
cpuset_attach_nodemask_to = cs->effective_mems; |
bb9d97b6d cgroup: don't use... |
1432 |
mm = get_task_mm(leader); |
4225399a6 [PATCH] cpuset: r... |
1433 |
if (mm) { |
f780bdb7c cgroups: add per-... |
1434 |
mpol_rebind_mm(mm, &cpuset_attach_nodemask_to); |
f047cecf2 cpuset: fix to mi... |
1435 1436 1437 1438 1439 1440 1441 1442 1443 |
/* * old_mems_allowed is the same with mems_allowed here, except * if this task is being moved automatically due to hotplug. * In that case @mems_allowed has been updated and is empty, * so @old_mems_allowed is the right nodesets that we migrate * mm from. */ if (is_memory_migrate(cs)) { |
ae1c80238 cpuset: apply cs-... |
1444 |
cpuset_migrate_mm(mm, &oldcs->old_mems_allowed, |
f780bdb7c cgroups: add per-... |
1445 |
&cpuset_attach_nodemask_to); |
f047cecf2 cpuset: fix to mi... |
1446 |
} |
4225399a6 [PATCH] cpuset: r... |
1447 1448 |
mmput(mm); } |
452477fa6 cpuset: pin down ... |
1449 |
|
33ad801df cpuset: record ol... |
1450 |
cs->old_mems_allowed = cpuset_attach_nodemask_to; |
02bb58637 cpuset: schedule ... |
1451 |
|
452477fa6 cpuset: pin down ... |
1452 |
cs->attach_in_progress--; |
e44193d39 cpuset: let hotpl... |
1453 1454 |
if (!cs->attach_in_progress) wake_up(&cpuset_attach_wq); |
5d21cc2db cpuset: replace c... |
1455 1456 |
mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1457 1458 1459 1460 1461 |
} /* The various types of files and directories in a cpuset file system */ typedef enum { |
45b07ef31 [PATCH] cpusets: ... |
1462 |
FILE_MEMORY_MIGRATE, |
1da177e4c Linux-2.6.12-rc2 |
1463 1464 |
FILE_CPULIST, FILE_MEMLIST, |
afd1a8b3e cpuset: export ef... |
1465 1466 |
FILE_EFFECTIVE_CPULIST, FILE_EFFECTIVE_MEMLIST, |
1da177e4c Linux-2.6.12-rc2 |
1467 1468 |
FILE_CPU_EXCLUSIVE, FILE_MEM_EXCLUSIVE, |
786083667 Cpuset hardwall f... |
1469 |
FILE_MEM_HARDWALL, |
029190c51 cpuset sched_load... |
1470 |
FILE_SCHED_LOAD_BALANCE, |
1d3504fcf sched, cpuset: cu... |
1471 |
FILE_SCHED_RELAX_DOMAIN_LEVEL, |
3e0d98b9f [PATCH] cpuset: m... |
1472 1473 |
FILE_MEMORY_PRESSURE_ENABLED, FILE_MEMORY_PRESSURE, |
825a46af5 [PATCH] cpuset me... |
1474 1475 |
FILE_SPREAD_PAGE, FILE_SPREAD_SLAB, |
1da177e4c Linux-2.6.12-rc2 |
1476 |
} cpuset_filetype_t; |
182446d08 cgroup: pass arou... |
1477 1478 |
static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, u64 val) |
700fe1ab9 CGroup API files:... |
1479 |
{ |
182446d08 cgroup: pass arou... |
1480 |
struct cpuset *cs = css_cs(css); |
700fe1ab9 CGroup API files:... |
1481 |
cpuset_filetype_t type = cft->private; |
a903f0865 cpuset: fix the r... |
1482 |
int retval = 0; |
700fe1ab9 CGroup API files:... |
1483 |
|
5d21cc2db cpuset: replace c... |
1484 |
mutex_lock(&cpuset_mutex); |
a903f0865 cpuset: fix the r... |
1485 1486 |
if (!is_cpuset_online(cs)) { retval = -ENODEV; |
5d21cc2db cpuset: replace c... |
1487 |
goto out_unlock; |
a903f0865 cpuset: fix the r... |
1488 |
} |
700fe1ab9 CGroup API files:... |
1489 1490 |
switch (type) { |
1da177e4c Linux-2.6.12-rc2 |
1491 |
case FILE_CPU_EXCLUSIVE: |
700fe1ab9 CGroup API files:... |
1492 |
retval = update_flag(CS_CPU_EXCLUSIVE, cs, val); |
1da177e4c Linux-2.6.12-rc2 |
1493 1494 |
break; case FILE_MEM_EXCLUSIVE: |
700fe1ab9 CGroup API files:... |
1495 |
retval = update_flag(CS_MEM_EXCLUSIVE, cs, val); |
1da177e4c Linux-2.6.12-rc2 |
1496 |
break; |
786083667 Cpuset hardwall f... |
1497 1498 1499 |
case FILE_MEM_HARDWALL: retval = update_flag(CS_MEM_HARDWALL, cs, val); break; |
029190c51 cpuset sched_load... |
1500 |
case FILE_SCHED_LOAD_BALANCE: |
700fe1ab9 CGroup API files:... |
1501 |
retval = update_flag(CS_SCHED_LOAD_BALANCE, cs, val); |
1d3504fcf sched, cpuset: cu... |
1502 |
break; |
45b07ef31 [PATCH] cpusets: ... |
1503 |
case FILE_MEMORY_MIGRATE: |
700fe1ab9 CGroup API files:... |
1504 |
retval = update_flag(CS_MEMORY_MIGRATE, cs, val); |
45b07ef31 [PATCH] cpusets: ... |
1505 |
break; |
3e0d98b9f [PATCH] cpuset: m... |
1506 |
case FILE_MEMORY_PRESSURE_ENABLED: |
700fe1ab9 CGroup API files:... |
1507 |
cpuset_memory_pressure_enabled = !!val; |
3e0d98b9f [PATCH] cpuset: m... |
1508 1509 1510 1511 |
break; case FILE_MEMORY_PRESSURE: retval = -EACCES; break; |
825a46af5 [PATCH] cpuset me... |
1512 |
case FILE_SPREAD_PAGE: |
700fe1ab9 CGroup API files:... |
1513 |
retval = update_flag(CS_SPREAD_PAGE, cs, val); |
825a46af5 [PATCH] cpuset me... |
1514 1515 |
break; case FILE_SPREAD_SLAB: |
700fe1ab9 CGroup API files:... |
1516 |
retval = update_flag(CS_SPREAD_SLAB, cs, val); |
825a46af5 [PATCH] cpuset me... |
1517 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1518 1519 |
default: retval = -EINVAL; |
700fe1ab9 CGroup API files:... |
1520 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1521 |
} |
5d21cc2db cpuset: replace c... |
1522 1523 |
out_unlock: mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1524 1525 |
return retval; } |
182446d08 cgroup: pass arou... |
1526 1527 |
static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, s64 val) |
5be7a4792 Fix cpuset sched_... |
1528 |
{ |
182446d08 cgroup: pass arou... |
1529 |
struct cpuset *cs = css_cs(css); |
5be7a4792 Fix cpuset sched_... |
1530 |
cpuset_filetype_t type = cft->private; |
5d21cc2db cpuset: replace c... |
1531 |
int retval = -ENODEV; |
5be7a4792 Fix cpuset sched_... |
1532 |
|
5d21cc2db cpuset: replace c... |
1533 1534 1535 |
mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; |
e37123953 cgroup files: rem... |
1536 |
|
5be7a4792 Fix cpuset sched_... |
1537 1538 1539 1540 1541 1542 1543 1544 |
switch (type) { case FILE_SCHED_RELAX_DOMAIN_LEVEL: retval = update_relax_domain_level(cs, val); break; default: retval = -EINVAL; break; } |
5d21cc2db cpuset: replace c... |
1545 1546 |
out_unlock: mutex_unlock(&cpuset_mutex); |
5be7a4792 Fix cpuset sched_... |
1547 1548 |
return retval; } |
1da177e4c Linux-2.6.12-rc2 |
1549 |
/* |
e37123953 cgroup files: rem... |
1550 1551 |
* Common handling for a write to a "cpus" or "mems" file. */ |
451af504d cgroup: replace c... |
1552 1553 |
static ssize_t cpuset_write_resmask(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
e37123953 cgroup files: rem... |
1554 |
{ |
451af504d cgroup: replace c... |
1555 |
struct cpuset *cs = css_cs(of_css(of)); |
645fcc9d2 cpuset: don't all... |
1556 |
struct cpuset *trialcs; |
5d21cc2db cpuset: replace c... |
1557 |
int retval = -ENODEV; |
e37123953 cgroup files: rem... |
1558 |
|
451af504d cgroup: replace c... |
1559 |
buf = strstrip(buf); |
3a5a6d0c2 cpuset: don't nes... |
1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 |
/* * CPU or memory hotunplug may leave @cs w/o any execution * resources, in which case the hotplug code asynchronously updates * configuration and transfers all tasks to the nearest ancestor * which can execute. * * As writes to "cpus" or "mems" may restore @cs's execution * resources, wait for the previously scheduled operations before * proceeding, so that we don't end up keep removing tasks added * after execution capability is restored. |
76bb5ab8f cpuset: break ker... |
1570 1571 1572 1573 1574 1575 1576 1577 |
* * cpuset_hotplug_work calls back into cgroup core via * cgroup_transfer_tasks() and waiting for it from a cgroupfs * operation like this one can lead to a deadlock through kernfs * active_ref protection. Let's break the protection. Losing the * protection is okay as we check whether @cs is online after * grabbing cpuset_mutex anyway. This only happens on the legacy * hierarchies. |
3a5a6d0c2 cpuset: don't nes... |
1578 |
*/ |
76bb5ab8f cpuset: break ker... |
1579 1580 |
css_get(&cs->css); kernfs_break_active_protection(of->kn); |
3a5a6d0c2 cpuset: don't nes... |
1581 |
flush_work(&cpuset_hotplug_work); |
5d21cc2db cpuset: replace c... |
1582 1583 1584 |
mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; |
e37123953 cgroup files: rem... |
1585 |
|
645fcc9d2 cpuset: don't all... |
1586 |
trialcs = alloc_trial_cpuset(cs); |
b75f38d65 cpuset: add a mis... |
1587 1588 |
if (!trialcs) { retval = -ENOMEM; |
5d21cc2db cpuset: replace c... |
1589 |
goto out_unlock; |
b75f38d65 cpuset: add a mis... |
1590 |
} |
645fcc9d2 cpuset: don't all... |
1591 |
|
451af504d cgroup: replace c... |
1592 |
switch (of_cft(of)->private) { |
e37123953 cgroup files: rem... |
1593 |
case FILE_CPULIST: |
645fcc9d2 cpuset: don't all... |
1594 |
retval = update_cpumask(cs, trialcs, buf); |
e37123953 cgroup files: rem... |
1595 1596 |
break; case FILE_MEMLIST: |
645fcc9d2 cpuset: don't all... |
1597 |
retval = update_nodemask(cs, trialcs, buf); |
e37123953 cgroup files: rem... |
1598 1599 1600 1601 1602 |
break; default: retval = -EINVAL; break; } |
645fcc9d2 cpuset: don't all... |
1603 1604 |
free_trial_cpuset(trialcs); |
5d21cc2db cpuset: replace c... |
1605 1606 |
out_unlock: mutex_unlock(&cpuset_mutex); |
76bb5ab8f cpuset: break ker... |
1607 1608 |
kernfs_unbreak_active_protection(of->kn); css_put(&cs->css); |
451af504d cgroup: replace c... |
1609 |
return retval ?: nbytes; |
e37123953 cgroup files: rem... |
1610 1611 1612 |
} /* |
1da177e4c Linux-2.6.12-rc2 |
1613 1614 1615 1616 1617 1618 |
* 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. |
1da177e4c Linux-2.6.12-rc2 |
1619 |
*/ |
2da8ca822 cgroup: replace c... |
1620 |
static int cpuset_common_seq_show(struct seq_file *sf, void *v) |
1da177e4c Linux-2.6.12-rc2 |
1621 |
{ |
2da8ca822 cgroup: replace c... |
1622 1623 |
struct cpuset *cs = css_cs(seq_css(sf)); cpuset_filetype_t type = seq_cft(sf)->private; |
51ffe4117 cpuset: convert a... |
1624 1625 1626 |
ssize_t count; char *buf, *s; int ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
1627 |
|
51ffe4117 cpuset: convert a... |
1628 1629 |
count = seq_get_buf(sf, &buf); s = buf; |
1da177e4c Linux-2.6.12-rc2 |
1630 |
|
8447a0fee cpuset: convert c... |
1631 |
spin_lock_irq(&callback_lock); |
1da177e4c Linux-2.6.12-rc2 |
1632 1633 1634 |
switch (type) { case FILE_CPULIST: |
51ffe4117 cpuset: convert a... |
1635 |
s += cpulist_scnprintf(s, count, cs->cpus_allowed); |
1da177e4c Linux-2.6.12-rc2 |
1636 1637 |
break; case FILE_MEMLIST: |
51ffe4117 cpuset: convert a... |
1638 |
s += nodelist_scnprintf(s, count, cs->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
1639 |
break; |
afd1a8b3e cpuset: export ef... |
1640 1641 1642 1643 1644 1645 |
case FILE_EFFECTIVE_CPULIST: s += cpulist_scnprintf(s, count, cs->effective_cpus); break; case FILE_EFFECTIVE_MEMLIST: s += nodelist_scnprintf(s, count, cs->effective_mems); break; |
1da177e4c Linux-2.6.12-rc2 |
1646 |
default: |
51ffe4117 cpuset: convert a... |
1647 1648 |
ret = -EINVAL; goto out_unlock; |
1da177e4c Linux-2.6.12-rc2 |
1649 |
} |
1da177e4c Linux-2.6.12-rc2 |
1650 |
|
51ffe4117 cpuset: convert a... |
1651 1652 1653 1654 1655 1656 1657 1658 |
if (s < buf + count - 1) { *s++ = ' '; seq_commit(sf, s - buf); } else { seq_commit(sf, -1); } out_unlock: |
8447a0fee cpuset: convert c... |
1659 |
spin_unlock_irq(&callback_lock); |
51ffe4117 cpuset: convert a... |
1660 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1661 |
} |
182446d08 cgroup: pass arou... |
1662 |
static u64 cpuset_read_u64(struct cgroup_subsys_state *css, struct cftype *cft) |
700fe1ab9 CGroup API files:... |
1663 |
{ |
182446d08 cgroup: pass arou... |
1664 |
struct cpuset *cs = css_cs(css); |
700fe1ab9 CGroup API files:... |
1665 1666 1667 1668 1669 1670 |
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); |
786083667 Cpuset hardwall f... |
1671 1672 |
case FILE_MEM_HARDWALL: return is_mem_hardwall(cs); |
700fe1ab9 CGroup API files:... |
1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 |
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(); } |
cf417141c sched, cpuset: re... |
1688 1689 1690 |
/* Unreachable but makes gcc happy */ return 0; |
700fe1ab9 CGroup API files:... |
1691 |
} |
1da177e4c Linux-2.6.12-rc2 |
1692 |
|
182446d08 cgroup: pass arou... |
1693 |
static s64 cpuset_read_s64(struct cgroup_subsys_state *css, struct cftype *cft) |
5be7a4792 Fix cpuset sched_... |
1694 |
{ |
182446d08 cgroup: pass arou... |
1695 |
struct cpuset *cs = css_cs(css); |
5be7a4792 Fix cpuset sched_... |
1696 1697 1698 1699 1700 1701 1702 |
cpuset_filetype_t type = cft->private; switch (type) { case FILE_SCHED_RELAX_DOMAIN_LEVEL: return cs->relax_domain_level; default: BUG(); } |
cf417141c sched, cpuset: re... |
1703 1704 1705 |
/* Unrechable but makes gcc happy */ return 0; |
5be7a4792 Fix cpuset sched_... |
1706 |
} |
1da177e4c Linux-2.6.12-rc2 |
1707 1708 1709 1710 |
/* * for the common functions, 'private' gives the type of file */ |
addf2c739 Cpuset hardwall f... |
1711 1712 1713 |
static struct cftype files[] = { { .name = "cpus", |
2da8ca822 cgroup: replace c... |
1714 |
.seq_show = cpuset_common_seq_show, |
451af504d cgroup: replace c... |
1715 |
.write = cpuset_write_resmask, |
e37123953 cgroup files: rem... |
1716 |
.max_write_len = (100U + 6 * NR_CPUS), |
addf2c739 Cpuset hardwall f... |
1717 1718 1719 1720 1721 |
.private = FILE_CPULIST, }, { .name = "mems", |
2da8ca822 cgroup: replace c... |
1722 |
.seq_show = cpuset_common_seq_show, |
451af504d cgroup: replace c... |
1723 |
.write = cpuset_write_resmask, |
e37123953 cgroup files: rem... |
1724 |
.max_write_len = (100U + 6 * MAX_NUMNODES), |
addf2c739 Cpuset hardwall f... |
1725 1726 1727 1728 |
.private = FILE_MEMLIST, }, { |
afd1a8b3e cpuset: export ef... |
1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 |
.name = "effective_cpus", .seq_show = cpuset_common_seq_show, .private = FILE_EFFECTIVE_CPULIST, }, { .name = "effective_mems", .seq_show = cpuset_common_seq_show, .private = FILE_EFFECTIVE_MEMLIST, }, { |
addf2c739 Cpuset hardwall f... |
1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 |
.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, }, { |
786083667 Cpuset hardwall f... |
1755 1756 1757 1758 1759 1760 1761 |
.name = "mem_hardwall", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_MEM_HARDWALL, }, { |
addf2c739 Cpuset hardwall f... |
1762 1763 1764 1765 1766 1767 1768 1769 |
.name = "sched_load_balance", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_SCHED_LOAD_BALANCE, }, { .name = "sched_relax_domain_level", |
5be7a4792 Fix cpuset sched_... |
1770 1771 |
.read_s64 = cpuset_read_s64, .write_s64 = cpuset_write_s64, |
addf2c739 Cpuset hardwall f... |
1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 |
.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, |
099fca322 cgroups: show cor... |
1787 |
.mode = S_IRUGO, |
addf2c739 Cpuset hardwall f... |
1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 |
}, { .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, }, |
3e0d98b9f [PATCH] cpuset: m... |
1803 |
|
4baf6e332 cgroup: convert a... |
1804 1805 1806 1807 1808 1809 1810 |
{ .name = "memory_pressure_enabled", .flags = CFTYPE_ONLY_ON_ROOT, .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_MEMORY_PRESSURE_ENABLED, }, |
1da177e4c Linux-2.6.12-rc2 |
1811 |
|
4baf6e332 cgroup: convert a... |
1812 1813 |
{ } /* terminate */ }; |
1da177e4c Linux-2.6.12-rc2 |
1814 1815 |
/* |
92fb97487 cgroup: rename ->... |
1816 |
* cpuset_css_alloc - allocate a cpuset css |
c9e5fe66f cpuset: rename @c... |
1817 |
* cgrp: control group that the new cpuset will be part of |
1da177e4c Linux-2.6.12-rc2 |
1818 |
*/ |
eb95419b0 cgroup: pass arou... |
1819 1820 |
static struct cgroup_subsys_state * cpuset_css_alloc(struct cgroup_subsys_state *parent_css) |
1da177e4c Linux-2.6.12-rc2 |
1821 |
{ |
c8f699bb5 cpuset: introduce... |
1822 |
struct cpuset *cs; |
1da177e4c Linux-2.6.12-rc2 |
1823 |
|
eb95419b0 cgroup: pass arou... |
1824 |
if (!parent_css) |
8793d854e Task Control Grou... |
1825 |
return &top_cpuset.css; |
033fa1c5f cgroup, cpuset: r... |
1826 |
|
c8f699bb5 cpuset: introduce... |
1827 |
cs = kzalloc(sizeof(*cs), GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
1828 |
if (!cs) |
8793d854e Task Control Grou... |
1829 |
return ERR_PTR(-ENOMEM); |
e2b9a3d7d cpuset: add cs->e... |
1830 1831 1832 1833 |
if (!alloc_cpumask_var(&cs->cpus_allowed, GFP_KERNEL)) goto free_cs; if (!alloc_cpumask_var(&cs->effective_cpus, GFP_KERNEL)) goto free_cpus; |
1da177e4c Linux-2.6.12-rc2 |
1834 |
|
029190c51 cpuset sched_load... |
1835 |
set_bit(CS_SCHED_LOAD_BALANCE, &cs->flags); |
300ed6cbb cpuset: convert c... |
1836 |
cpumask_clear(cs->cpus_allowed); |
f9a86fcbb cpuset: modify cp... |
1837 |
nodes_clear(cs->mems_allowed); |
e2b9a3d7d cpuset: add cs->e... |
1838 1839 |
cpumask_clear(cs->effective_cpus); nodes_clear(cs->effective_mems); |
3e0d98b9f [PATCH] cpuset: m... |
1840 |
fmeter_init(&cs->fmeter); |
1d3504fcf sched, cpuset: cu... |
1841 |
cs->relax_domain_level = -1; |
1da177e4c Linux-2.6.12-rc2 |
1842 |
|
c8f699bb5 cpuset: introduce... |
1843 |
return &cs->css; |
e2b9a3d7d cpuset: add cs->e... |
1844 1845 1846 1847 1848 1849 |
free_cpus: free_cpumask_var(cs->cpus_allowed); free_cs: kfree(cs); return ERR_PTR(-ENOMEM); |
c8f699bb5 cpuset: introduce... |
1850 |
} |
eb95419b0 cgroup: pass arou... |
1851 |
static int cpuset_css_online(struct cgroup_subsys_state *css) |
c8f699bb5 cpuset: introduce... |
1852 |
{ |
eb95419b0 cgroup: pass arou... |
1853 |
struct cpuset *cs = css_cs(css); |
c431069fe cpuset: remove cp... |
1854 |
struct cpuset *parent = parent_cs(cs); |
ae8086ce1 cpuset: introduce... |
1855 |
struct cpuset *tmp_cs; |
492eb21b9 cgroup: make hier... |
1856 |
struct cgroup_subsys_state *pos_css; |
c8f699bb5 cpuset: introduce... |
1857 1858 1859 |
if (!parent) return 0; |
5d21cc2db cpuset: replace c... |
1860 |
mutex_lock(&cpuset_mutex); |
efeb77b2f cpuset: introduce... |
1861 |
set_bit(CS_ONLINE, &cs->flags); |
c8f699bb5 cpuset: introduce... |
1862 1863 1864 1865 |
if (is_spread_page(parent)) set_bit(CS_SPREAD_PAGE, &cs->flags); if (is_spread_slab(parent)) set_bit(CS_SPREAD_SLAB, &cs->flags); |
1da177e4c Linux-2.6.12-rc2 |
1866 |
|
664eeddee mm: page_alloc: u... |
1867 |
cpuset_inc(); |
033fa1c5f cgroup, cpuset: r... |
1868 |
|
8447a0fee cpuset: convert c... |
1869 |
spin_lock_irq(&callback_lock); |
e2b9a3d7d cpuset: add cs->e... |
1870 1871 1872 1873 |
if (cgroup_on_dfl(cs->css.cgroup)) { cpumask_copy(cs->effective_cpus, parent->effective_cpus); cs->effective_mems = parent->effective_mems; } |
8447a0fee cpuset: convert c... |
1874 |
spin_unlock_irq(&callback_lock); |
e2b9a3d7d cpuset: add cs->e... |
1875 |
|
eb95419b0 cgroup: pass arou... |
1876 |
if (!test_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags)) |
5d21cc2db cpuset: replace c... |
1877 |
goto out_unlock; |
033fa1c5f cgroup, cpuset: r... |
1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 |
/* * Clone @parent's configuration if CGRP_CPUSET_CLONE_CHILDREN is * set. This flag handling is implemented in cgroup core for * histrical reasons - the flag may be specified during mount. * * Currently, if any sibling cpusets have exclusive cpus or mem, we * refuse to clone the configuration - thereby refusing the task to * be entered, and as a result refusing the sys_unshare() or * clone() which initiated it. If this becomes a problem for some * users who wish to allow that scenario, then this could be * changed to grant parent->cpus_allowed-sibling_cpus_exclusive * (and likewise for mems) to the new cgroup. */ |
ae8086ce1 cpuset: introduce... |
1892 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
1893 |
cpuset_for_each_child(tmp_cs, pos_css, parent) { |
ae8086ce1 cpuset: introduce... |
1894 1895 |
if (is_mem_exclusive(tmp_cs) || is_cpu_exclusive(tmp_cs)) { rcu_read_unlock(); |
5d21cc2db cpuset: replace c... |
1896 |
goto out_unlock; |
ae8086ce1 cpuset: introduce... |
1897 |
} |
033fa1c5f cgroup, cpuset: r... |
1898 |
} |
ae8086ce1 cpuset: introduce... |
1899 |
rcu_read_unlock(); |
033fa1c5f cgroup, cpuset: r... |
1900 |
|
8447a0fee cpuset: convert c... |
1901 |
spin_lock_irq(&callback_lock); |
033fa1c5f cgroup, cpuset: r... |
1902 1903 |
cs->mems_allowed = parent->mems_allowed; cpumask_copy(cs->cpus_allowed, parent->cpus_allowed); |
cea74465e cpuset: lock vs u... |
1904 |
spin_unlock_irq(&callback_lock); |
5d21cc2db cpuset: replace c... |
1905 1906 |
out_unlock: mutex_unlock(&cpuset_mutex); |
c8f699bb5 cpuset: introduce... |
1907 1908 |
return 0; } |
0b9e6965a cpuset: relocate ... |
1909 1910 1911 1912 1913 |
/* * If the cpuset being removed has its flag 'sched_load_balance' * enabled, then simulate turning sched_load_balance off, which * will call rebuild_sched_domains_locked(). */ |
eb95419b0 cgroup: pass arou... |
1914 |
static void cpuset_css_offline(struct cgroup_subsys_state *css) |
c8f699bb5 cpuset: introduce... |
1915 |
{ |
eb95419b0 cgroup: pass arou... |
1916 |
struct cpuset *cs = css_cs(css); |
c8f699bb5 cpuset: introduce... |
1917 |
|
5d21cc2db cpuset: replace c... |
1918 |
mutex_lock(&cpuset_mutex); |
c8f699bb5 cpuset: introduce... |
1919 1920 1921 |
if (is_sched_load_balance(cs)) update_flag(CS_SCHED_LOAD_BALANCE, cs, 0); |
664eeddee mm: page_alloc: u... |
1922 |
cpuset_dec(); |
efeb77b2f cpuset: introduce... |
1923 |
clear_bit(CS_ONLINE, &cs->flags); |
c8f699bb5 cpuset: introduce... |
1924 |
|
5d21cc2db cpuset: replace c... |
1925 |
mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1926 |
} |
eb95419b0 cgroup: pass arou... |
1927 |
static void cpuset_css_free(struct cgroup_subsys_state *css) |
1da177e4c Linux-2.6.12-rc2 |
1928 |
{ |
eb95419b0 cgroup: pass arou... |
1929 |
struct cpuset *cs = css_cs(css); |
1da177e4c Linux-2.6.12-rc2 |
1930 |
|
e2b9a3d7d cpuset: add cs->e... |
1931 |
free_cpumask_var(cs->effective_cpus); |
300ed6cbb cpuset: convert c... |
1932 |
free_cpumask_var(cs->cpus_allowed); |
8793d854e Task Control Grou... |
1933 |
kfree(cs); |
1da177e4c Linux-2.6.12-rc2 |
1934 |
} |
39bd0d15e cpuset: initializ... |
1935 1936 1937 |
static void cpuset_bind(struct cgroup_subsys_state *root_css) { mutex_lock(&cpuset_mutex); |
8447a0fee cpuset: convert c... |
1938 |
spin_lock_irq(&callback_lock); |
39bd0d15e cpuset: initializ... |
1939 1940 1941 1942 1943 1944 1945 1946 1947 |
if (cgroup_on_dfl(root_css->cgroup)) { cpumask_copy(top_cpuset.cpus_allowed, cpu_possible_mask); top_cpuset.mems_allowed = node_possible_map; } else { cpumask_copy(top_cpuset.cpus_allowed, top_cpuset.effective_cpus); top_cpuset.mems_allowed = top_cpuset.effective_mems; } |
8447a0fee cpuset: convert c... |
1948 |
spin_unlock_irq(&callback_lock); |
39bd0d15e cpuset: initializ... |
1949 1950 |
mutex_unlock(&cpuset_mutex); } |
073219e99 cgroup: clean up ... |
1951 |
struct cgroup_subsys cpuset_cgrp_subsys = { |
39bd0d15e cpuset: initializ... |
1952 1953 1954 1955 1956 1957 1958 1959 |
.css_alloc = cpuset_css_alloc, .css_online = cpuset_css_online, .css_offline = cpuset_css_offline, .css_free = cpuset_css_free, .can_attach = cpuset_can_attach, .cancel_attach = cpuset_cancel_attach, .attach = cpuset_attach, .bind = cpuset_bind, |
5577964e6 cgroup: rename cg... |
1960 |
.legacy_cftypes = files, |
39bd0d15e cpuset: initializ... |
1961 |
.early_init = 1, |
8793d854e Task Control Grou... |
1962 |
}; |
1da177e4c Linux-2.6.12-rc2 |
1963 1964 1965 1966 1967 1968 1969 1970 |
/** * cpuset_init - initialize cpusets at system boot * * Description: Initialize top_cpuset and the cpuset internal file system, **/ int __init cpuset_init(void) { |
8793d854e Task Control Grou... |
1971 |
int err = 0; |
1da177e4c Linux-2.6.12-rc2 |
1972 |
|
58568d2a8 cpuset,mm: update... |
1973 1974 |
if (!alloc_cpumask_var(&top_cpuset.cpus_allowed, GFP_KERNEL)) BUG(); |
e2b9a3d7d cpuset: add cs->e... |
1975 1976 |
if (!alloc_cpumask_var(&top_cpuset.effective_cpus, GFP_KERNEL)) BUG(); |
58568d2a8 cpuset,mm: update... |
1977 |
|
300ed6cbb cpuset: convert c... |
1978 |
cpumask_setall(top_cpuset.cpus_allowed); |
f9a86fcbb cpuset: modify cp... |
1979 |
nodes_setall(top_cpuset.mems_allowed); |
e2b9a3d7d cpuset: add cs->e... |
1980 1981 |
cpumask_setall(top_cpuset.effective_cpus); nodes_setall(top_cpuset.effective_mems); |
1da177e4c Linux-2.6.12-rc2 |
1982 |
|
3e0d98b9f [PATCH] cpuset: m... |
1983 |
fmeter_init(&top_cpuset.fmeter); |
029190c51 cpuset sched_load... |
1984 |
set_bit(CS_SCHED_LOAD_BALANCE, &top_cpuset.flags); |
1d3504fcf sched, cpuset: cu... |
1985 |
top_cpuset.relax_domain_level = -1; |
1da177e4c Linux-2.6.12-rc2 |
1986 |
|
1da177e4c Linux-2.6.12-rc2 |
1987 1988 |
err = register_filesystem(&cpuset_fs_type); if (err < 0) |
8793d854e Task Control Grou... |
1989 |
return err; |
2341d1b65 cpuset: convert c... |
1990 1991 |
if (!alloc_cpumask_var(&cpus_attach, GFP_KERNEL)) BUG(); |
8793d854e Task Control Grou... |
1992 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
1993 |
} |
b1aac8bb8 [PATCH] cpuset: h... |
1994 |
/* |
cf417141c sched, cpuset: re... |
1995 |
* If CPU and/or memory hotplug handlers, below, unplug any CPUs |
b1aac8bb8 [PATCH] cpuset: h... |
1996 1997 |
* or memory nodes, we need to walk over the cpuset hierarchy, * removing that CPU or node from all cpusets. If this removes the |
956db3ca0 hotplug cpu: move... |
1998 1999 |
* last CPU or node from a cpuset, then move the tasks in the empty * cpuset to its next-highest non-empty parent. |
b1aac8bb8 [PATCH] cpuset: h... |
2000 |
*/ |
956db3ca0 hotplug cpu: move... |
2001 2002 2003 |
static void remove_tasks_in_empty_cpuset(struct cpuset *cs) { struct cpuset *parent; |
c8d9c90c7 hotplug cpu: move... |
2004 |
/* |
956db3ca0 hotplug cpu: move... |
2005 2006 2007 |
* Find its next-highest non-empty parent, (top cpuset * has online cpus, so can't be empty). */ |
c431069fe cpuset: remove cp... |
2008 |
parent = parent_cs(cs); |
300ed6cbb cpuset: convert c... |
2009 |
while (cpumask_empty(parent->cpus_allowed) || |
b45012955 hotplug cpu move ... |
2010 |
nodes_empty(parent->mems_allowed)) |
c431069fe cpuset: remove cp... |
2011 |
parent = parent_cs(parent); |
956db3ca0 hotplug cpu: move... |
2012 |
|
8cc993452 cgroup, cpuset: r... |
2013 |
if (cgroup_transfer_tasks(parent->css.cgroup, cs->css.cgroup)) { |
12d3089c1 kernel/cpuset.c: ... |
2014 |
pr_err("cpuset: failed to transfer tasks out of empty cpuset "); |
e61734c55 cgroup: remove cg... |
2015 2016 2017 |
pr_cont_cgroup_name(cs->css.cgroup); pr_cont(" "); |
8cc993452 cgroup, cpuset: r... |
2018 |
} |
956db3ca0 hotplug cpu: move... |
2019 |
} |
be4c9dd7a cpuset: enable on... |
2020 2021 2022 2023 |
static void hotplug_update_tasks_legacy(struct cpuset *cs, struct cpumask *new_cpus, nodemask_t *new_mems, bool cpus_updated, bool mems_updated) |
390a36aad cpuset: refactor ... |
2024 2025 |
{ bool is_empty; |
8447a0fee cpuset: convert c... |
2026 |
spin_lock_irq(&callback_lock); |
be4c9dd7a cpuset: enable on... |
2027 2028 2029 2030 |
cpumask_copy(cs->cpus_allowed, new_cpus); cpumask_copy(cs->effective_cpus, new_cpus); cs->mems_allowed = *new_mems; cs->effective_mems = *new_mems; |
8447a0fee cpuset: convert c... |
2031 |
spin_unlock_irq(&callback_lock); |
390a36aad cpuset: refactor ... |
2032 2033 2034 2035 2036 |
/* * Don't call update_tasks_cpumask() if the cpuset becomes empty, * as the tasks will be migratecd to an ancestor. */ |
be4c9dd7a cpuset: enable on... |
2037 |
if (cpus_updated && !cpumask_empty(cs->cpus_allowed)) |
390a36aad cpuset: refactor ... |
2038 |
update_tasks_cpumask(cs); |
be4c9dd7a cpuset: enable on... |
2039 |
if (mems_updated && !nodes_empty(cs->mems_allowed)) |
390a36aad cpuset: refactor ... |
2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 |
update_tasks_nodemask(cs); is_empty = cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed); mutex_unlock(&cpuset_mutex); /* * Move tasks to the nearest ancestor with execution resources, * This is full cgroup operation which will also call back into * cpuset. Should be done outside any lock. */ if (is_empty) remove_tasks_in_empty_cpuset(cs); mutex_lock(&cpuset_mutex); } |
be4c9dd7a cpuset: enable on... |
2057 2058 2059 2060 |
static void hotplug_update_tasks(struct cpuset *cs, struct cpumask *new_cpus, nodemask_t *new_mems, bool cpus_updated, bool mems_updated) |
390a36aad cpuset: refactor ... |
2061 |
{ |
be4c9dd7a cpuset: enable on... |
2062 2063 2064 2065 |
if (cpumask_empty(new_cpus)) cpumask_copy(new_cpus, parent_cs(cs)->effective_cpus); if (nodes_empty(*new_mems)) *new_mems = parent_cs(cs)->effective_mems; |
8447a0fee cpuset: convert c... |
2066 |
spin_lock_irq(&callback_lock); |
be4c9dd7a cpuset: enable on... |
2067 2068 |
cpumask_copy(cs->effective_cpus, new_cpus); cs->effective_mems = *new_mems; |
8447a0fee cpuset: convert c... |
2069 |
spin_unlock_irq(&callback_lock); |
390a36aad cpuset: refactor ... |
2070 |
|
be4c9dd7a cpuset: enable on... |
2071 |
if (cpus_updated) |
390a36aad cpuset: refactor ... |
2072 |
update_tasks_cpumask(cs); |
be4c9dd7a cpuset: enable on... |
2073 |
if (mems_updated) |
390a36aad cpuset: refactor ... |
2074 2075 |
update_tasks_nodemask(cs); } |
deb7aa308 cpuset: reorganiz... |
2076 |
/** |
388afd854 cpuset: remove as... |
2077 |
* cpuset_hotplug_update_tasks - update tasks in a cpuset for hotunplug |
deb7aa308 cpuset: reorganiz... |
2078 |
* @cs: cpuset in interest |
956db3ca0 hotplug cpu: move... |
2079 |
* |
deb7aa308 cpuset: reorganiz... |
2080 2081 2082 |
* Compare @cs's cpu and mem masks against top_cpuset and if some have gone * offline, update @cs accordingly. If @cs ends up with no CPU or memory, * all its tasks are moved to the nearest ancestor with both resources. |
80d1fa646 cpusets, hotplug:... |
2083 |
*/ |
388afd854 cpuset: remove as... |
2084 |
static void cpuset_hotplug_update_tasks(struct cpuset *cs) |
80d1fa646 cpusets, hotplug:... |
2085 |
{ |
be4c9dd7a cpuset: enable on... |
2086 2087 2088 2089 |
static cpumask_t new_cpus; static nodemask_t new_mems; bool cpus_updated; bool mems_updated; |
e44193d39 cpuset: let hotpl... |
2090 2091 |
retry: wait_event(cpuset_attach_wq, cs->attach_in_progress == 0); |
80d1fa646 cpusets, hotplug:... |
2092 |
|
5d21cc2db cpuset: replace c... |
2093 |
mutex_lock(&cpuset_mutex); |
7ddf96b02 cpusets, hotplug:... |
2094 |
|
e44193d39 cpuset: let hotpl... |
2095 2096 2097 2098 2099 2100 2101 2102 |
/* * We have raced with task attaching. We wait until attaching * is finished, so we won't attach a task to an empty cpuset. */ if (cs->attach_in_progress) { mutex_unlock(&cpuset_mutex); goto retry; } |
be4c9dd7a cpuset: enable on... |
2103 2104 |
cpumask_and(&new_cpus, cs->cpus_allowed, parent_cs(cs)->effective_cpus); nodes_and(new_mems, cs->mems_allowed, parent_cs(cs)->effective_mems); |
80d1fa646 cpusets, hotplug:... |
2105 |
|
be4c9dd7a cpuset: enable on... |
2106 2107 |
cpus_updated = !cpumask_equal(&new_cpus, cs->effective_cpus); mems_updated = !nodes_equal(new_mems, cs->effective_mems); |
deb7aa308 cpuset: reorganiz... |
2108 |
|
390a36aad cpuset: refactor ... |
2109 |
if (cgroup_on_dfl(cs->css.cgroup)) |
be4c9dd7a cpuset: enable on... |
2110 2111 |
hotplug_update_tasks(cs, &new_cpus, &new_mems, cpus_updated, mems_updated); |
390a36aad cpuset: refactor ... |
2112 |
else |
be4c9dd7a cpuset: enable on... |
2113 2114 |
hotplug_update_tasks_legacy(cs, &new_cpus, &new_mems, cpus_updated, mems_updated); |
8d0339487 cpuset: make CPU ... |
2115 |
|
5d21cc2db cpuset: replace c... |
2116 |
mutex_unlock(&cpuset_mutex); |
b1aac8bb8 [PATCH] cpuset: h... |
2117 |
} |
deb7aa308 cpuset: reorganiz... |
2118 |
/** |
3a5a6d0c2 cpuset: don't nes... |
2119 |
* cpuset_hotplug_workfn - handle CPU/memory hotunplug for a cpuset |
956db3ca0 hotplug cpu: move... |
2120 |
* |
deb7aa308 cpuset: reorganiz... |
2121 2122 2123 2124 2125 |
* This function is called after either CPU or memory configuration has * changed and updates cpuset accordingly. The top_cpuset is always * synchronized to cpu_active_mask and N_MEMORY, which 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. |
956db3ca0 hotplug cpu: move... |
2126 |
* |
deb7aa308 cpuset: reorganiz... |
2127 |
* Non-root cpusets are only affected by offlining. If any CPUs or memory |
388afd854 cpuset: remove as... |
2128 2129 |
* nodes have been taken down, cpuset_hotplug_update_tasks() is invoked on * all descendants. |
956db3ca0 hotplug cpu: move... |
2130 |
* |
deb7aa308 cpuset: reorganiz... |
2131 2132 |
* Note that CPU offlining during suspend is ignored. We don't modify * cpusets across suspend/resume cycles at all. |
956db3ca0 hotplug cpu: move... |
2133 |
*/ |
3a5a6d0c2 cpuset: don't nes... |
2134 |
static void cpuset_hotplug_workfn(struct work_struct *work) |
b1aac8bb8 [PATCH] cpuset: h... |
2135 |
{ |
5c5cc6232 cpuset: allow to ... |
2136 2137 |
static cpumask_t new_cpus; static nodemask_t new_mems; |
deb7aa308 cpuset: reorganiz... |
2138 |
bool cpus_updated, mems_updated; |
7e88291be cpuset: make cs->... |
2139 |
bool on_dfl = cgroup_on_dfl(top_cpuset.css.cgroup); |
b1aac8bb8 [PATCH] cpuset: h... |
2140 |
|
5d21cc2db cpuset: replace c... |
2141 |
mutex_lock(&cpuset_mutex); |
956db3ca0 hotplug cpu: move... |
2142 |
|
deb7aa308 cpuset: reorganiz... |
2143 2144 2145 |
/* fetch the available cpus/mems and find out which changed how */ cpumask_copy(&new_cpus, cpu_active_mask); new_mems = node_states[N_MEMORY]; |
7ddf96b02 cpusets, hotplug:... |
2146 |
|
7e88291be cpuset: make cs->... |
2147 2148 |
cpus_updated = !cpumask_equal(top_cpuset.effective_cpus, &new_cpus); mems_updated = !nodes_equal(top_cpuset.effective_mems, new_mems); |
7ddf96b02 cpusets, hotplug:... |
2149 |
|
deb7aa308 cpuset: reorganiz... |
2150 2151 |
/* synchronize cpus_allowed to cpu_active_mask */ if (cpus_updated) { |
8447a0fee cpuset: convert c... |
2152 |
spin_lock_irq(&callback_lock); |
7e88291be cpuset: make cs->... |
2153 2154 |
if (!on_dfl) cpumask_copy(top_cpuset.cpus_allowed, &new_cpus); |
1344ab9c2 cpuset: update cp... |
2155 |
cpumask_copy(top_cpuset.effective_cpus, &new_cpus); |
8447a0fee cpuset: convert c... |
2156 |
spin_unlock_irq(&callback_lock); |
deb7aa308 cpuset: reorganiz... |
2157 2158 |
/* we don't mess with cpumasks of tasks in top_cpuset */ } |
b45012955 hotplug cpu move ... |
2159 |
|
deb7aa308 cpuset: reorganiz... |
2160 2161 |
/* synchronize mems_allowed to N_MEMORY */ if (mems_updated) { |
8447a0fee cpuset: convert c... |
2162 |
spin_lock_irq(&callback_lock); |
7e88291be cpuset: make cs->... |
2163 2164 |
if (!on_dfl) top_cpuset.mems_allowed = new_mems; |
1344ab9c2 cpuset: update cp... |
2165 |
top_cpuset.effective_mems = new_mems; |
8447a0fee cpuset: convert c... |
2166 |
spin_unlock_irq(&callback_lock); |
d66393e54 cpuset: use css_t... |
2167 |
update_tasks_nodemask(&top_cpuset); |
deb7aa308 cpuset: reorganiz... |
2168 |
} |
b45012955 hotplug cpu move ... |
2169 |
|
388afd854 cpuset: remove as... |
2170 |
mutex_unlock(&cpuset_mutex); |
5c5cc6232 cpuset: allow to ... |
2171 2172 |
/* if cpus or mems changed, we need to propagate to descendants */ if (cpus_updated || mems_updated) { |
deb7aa308 cpuset: reorganiz... |
2173 |
struct cpuset *cs; |
492eb21b9 cgroup: make hier... |
2174 |
struct cgroup_subsys_state *pos_css; |
f9b4fb8da cpusets: update t... |
2175 |
|
fc560a26a cpuset: replace c... |
2176 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
2177 |
cpuset_for_each_descendant_pre(cs, pos_css, &top_cpuset) { |
ec903c0c8 cgroup: rename cs... |
2178 |
if (cs == &top_cpuset || !css_tryget_online(&cs->css)) |
388afd854 cpuset: remove as... |
2179 2180 |
continue; rcu_read_unlock(); |
7ddf96b02 cpusets, hotplug:... |
2181 |
|
388afd854 cpuset: remove as... |
2182 |
cpuset_hotplug_update_tasks(cs); |
b45012955 hotplug cpu move ... |
2183 |
|
388afd854 cpuset: remove as... |
2184 2185 2186 2187 2188 |
rcu_read_lock(); css_put(&cs->css); } rcu_read_unlock(); } |
8d0339487 cpuset: make CPU ... |
2189 |
|
deb7aa308 cpuset: reorganiz... |
2190 |
/* rebuild sched domains if cpus_allowed has changed */ |
e0e80a02e cpuset: use rebui... |
2191 2192 |
if (cpus_updated) rebuild_sched_domains(); |
b1aac8bb8 [PATCH] cpuset: h... |
2193 |
} |
7ddf96b02 cpusets, hotplug:... |
2194 |
void cpuset_update_active_cpus(bool cpu_online) |
4c4d50f7b [PATCH] cpuset: t... |
2195 |
{ |
3a5a6d0c2 cpuset: don't nes... |
2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 |
/* * We're inside cpu hotplug critical region which usually nests * inside cgroup synchronization. Bounce actual hotplug processing * to a work item to avoid reverse locking order. * * We still need to do partition_sched_domains() synchronously; * otherwise, the scheduler will get confused and put tasks to the * dead CPU. Fall back to the default single domain. * cpuset_hotplug_workfn() will rebuild it as necessary. */ partition_sched_domains(1, NULL, NULL); schedule_work(&cpuset_hotplug_work); |
4c4d50f7b [PATCH] cpuset: t... |
2208 |
} |
4c4d50f7b [PATCH] cpuset: t... |
2209 |
|
38837fc75 [PATCH] cpuset: t... |
2210 |
/* |
38d7bee9d cpuset: use N_MEM... |
2211 2212 |
* Keep top_cpuset.mems_allowed tracking node_states[N_MEMORY]. * Call this routine anytime after node_states[N_MEMORY] changes. |
a1cd2b13f cpusets: Remove/u... |
2213 |
* See cpuset_update_active_cpus() for CPU hotplug handling. |
38837fc75 [PATCH] cpuset: t... |
2214 |
*/ |
f481891fd cpuset: update to... |
2215 2216 |
static int cpuset_track_online_nodes(struct notifier_block *self, unsigned long action, void *arg) |
38837fc75 [PATCH] cpuset: t... |
2217 |
{ |
3a5a6d0c2 cpuset: don't nes... |
2218 |
schedule_work(&cpuset_hotplug_work); |
f481891fd cpuset: update to... |
2219 |
return NOTIFY_OK; |
38837fc75 [PATCH] cpuset: t... |
2220 |
} |
d8f10cb3d kernel/cpuset.c: ... |
2221 2222 2223 2224 2225 |
static struct notifier_block cpuset_track_online_nodes_nb = { .notifier_call = cpuset_track_online_nodes, .priority = 10, /* ??! */ }; |
38837fc75 [PATCH] cpuset: t... |
2226 |
|
1da177e4c Linux-2.6.12-rc2 |
2227 2228 2229 2230 |
/** * cpuset_init_smp - initialize cpus_allowed * * Description: Finish top cpuset after cpu, node maps are initialized |
d8f10cb3d kernel/cpuset.c: ... |
2231 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
2232 2233 |
void __init cpuset_init_smp(void) { |
6ad4c1888 sched: Fix balanc... |
2234 |
cpumask_copy(top_cpuset.cpus_allowed, cpu_active_mask); |
38d7bee9d cpuset: use N_MEM... |
2235 |
top_cpuset.mems_allowed = node_states[N_MEMORY]; |
33ad801df cpuset: record ol... |
2236 |
top_cpuset.old_mems_allowed = top_cpuset.mems_allowed; |
4c4d50f7b [PATCH] cpuset: t... |
2237 |
|
e2b9a3d7d cpuset: add cs->e... |
2238 2239 |
cpumask_copy(top_cpuset.effective_cpus, cpu_active_mask); top_cpuset.effective_mems = node_states[N_MEMORY]; |
d8f10cb3d kernel/cpuset.c: ... |
2240 |
register_hotmemory_notifier(&cpuset_track_online_nodes_nb); |
1da177e4c Linux-2.6.12-rc2 |
2241 2242 2243 |
} /** |
1da177e4c Linux-2.6.12-rc2 |
2244 2245 |
* cpuset_cpus_allowed - return cpus_allowed mask from a tasks cpuset. * @tsk: pointer to task_struct from which to obtain cpuset->cpus_allowed. |
6af866af3 cpuset: remove re... |
2246 |
* @pmask: pointer to struct cpumask variable to receive cpus_allowed set. |
1da177e4c Linux-2.6.12-rc2 |
2247 |
* |
300ed6cbb cpuset: convert c... |
2248 |
* Description: Returns the cpumask_var_t cpus_allowed of the cpuset |
1da177e4c Linux-2.6.12-rc2 |
2249 |
* attached to the specified @tsk. Guaranteed to return some non-empty |
5f054e31c documentation: re... |
2250 |
* subset of cpu_online_mask, even if this means going outside the |
1da177e4c Linux-2.6.12-rc2 |
2251 2252 |
* tasks cpuset. **/ |
6af866af3 cpuset: remove re... |
2253 |
void cpuset_cpus_allowed(struct task_struct *tsk, struct cpumask *pmask) |
1da177e4c Linux-2.6.12-rc2 |
2254 |
{ |
8447a0fee cpuset: convert c... |
2255 2256 2257 |
unsigned long flags; spin_lock_irqsave(&callback_lock, flags); |
b8dadcb58 cpuset: use rcu_r... |
2258 |
rcu_read_lock(); |
ae1c80238 cpuset: apply cs-... |
2259 |
guarantee_online_cpus(task_cs(tsk), pmask); |
b8dadcb58 cpuset: use rcu_r... |
2260 |
rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
2261 |
spin_unlock_irqrestore(&callback_lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
2262 |
} |
2baab4e90 sched: Fix select... |
2263 |
void cpuset_cpus_allowed_fallback(struct task_struct *tsk) |
9084bb824 sched: Make selec... |
2264 |
{ |
9084bb824 sched: Make selec... |
2265 |
rcu_read_lock(); |
ae1c80238 cpuset: apply cs-... |
2266 |
do_set_cpus_allowed(tsk, task_cs(tsk)->effective_cpus); |
9084bb824 sched: Make selec... |
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 |
rcu_read_unlock(); /* * We own tsk->cpus_allowed, nobody can change it under us. * * But we used cs && cs->cpus_allowed lockless and thus can * race with cgroup_attach_task() or update_cpumask() and get * the wrong tsk->cpus_allowed. However, both cases imply the * subsequent cpuset_change_cpumask()->set_cpus_allowed_ptr() * which takes task_rq_lock(). * * If we are called after it dropped the lock we must see all * changes in tsk_cs()->cpus_allowed. Otherwise we can temporary * set any mask even if it is not right from task_cs() pov, * the pending set_cpus_allowed_ptr() will fix things. |
2baab4e90 sched: Fix select... |
2282 2283 2284 |
* * select_fallback_rq() will fix things ups and set cpu_possible_mask * if required. |
9084bb824 sched: Make selec... |
2285 |
*/ |
9084bb824 sched: Make selec... |
2286 |
} |
1da177e4c Linux-2.6.12-rc2 |
2287 2288 |
void cpuset_init_current_mems_allowed(void) { |
f9a86fcbb cpuset: modify cp... |
2289 |
nodes_setall(current->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
2290 |
} |
d9fd8a6d4 [PATCH] kernel/cp... |
2291 |
/** |
909d75a3b [PATCH] cpuset: i... |
2292 2293 2294 2295 2296 |
* 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 |
38d7bee9d cpuset: use N_MEM... |
2297 |
* subset of node_states[N_MEMORY], even if this means going outside the |
909d75a3b [PATCH] cpuset: i... |
2298 2299 2300 2301 2302 2303 |
* tasks cpuset. **/ nodemask_t cpuset_mems_allowed(struct task_struct *tsk) { nodemask_t mask; |
8447a0fee cpuset: convert c... |
2304 |
unsigned long flags; |
909d75a3b [PATCH] cpuset: i... |
2305 |
|
8447a0fee cpuset: convert c... |
2306 |
spin_lock_irqsave(&callback_lock, flags); |
b8dadcb58 cpuset: use rcu_r... |
2307 |
rcu_read_lock(); |
ae1c80238 cpuset: apply cs-... |
2308 |
guarantee_online_mems(task_cs(tsk), &mask); |
b8dadcb58 cpuset: use rcu_r... |
2309 |
rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
2310 |
spin_unlock_irqrestore(&callback_lock, flags); |
909d75a3b [PATCH] cpuset: i... |
2311 2312 2313 2314 2315 |
return mask; } /** |
19770b326 mm: filter based ... |
2316 2317 |
* cpuset_nodemask_valid_mems_allowed - check nodemask vs. curremt mems_allowed * @nodemask: the nodemask to be checked |
d9fd8a6d4 [PATCH] kernel/cp... |
2318 |
* |
19770b326 mm: filter based ... |
2319 |
* Are any of the nodes in the nodemask allowed in current->mems_allowed? |
1da177e4c Linux-2.6.12-rc2 |
2320 |
*/ |
19770b326 mm: filter based ... |
2321 |
int cpuset_nodemask_valid_mems_allowed(nodemask_t *nodemask) |
1da177e4c Linux-2.6.12-rc2 |
2322 |
{ |
19770b326 mm: filter based ... |
2323 |
return nodes_intersects(*nodemask, current->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
2324 |
} |
9bf2229f8 [PATCH] cpusets: ... |
2325 |
/* |
786083667 Cpuset hardwall f... |
2326 2327 |
* nearest_hardwall_ancestor() - Returns the nearest mem_exclusive or * mem_hardwall ancestor to the specified cpuset. Call holding |
8447a0fee cpuset: convert c... |
2328 |
* callback_lock. If no ancestor is mem_exclusive or mem_hardwall |
786083667 Cpuset hardwall f... |
2329 |
* (an unusual configuration), then returns the root cpuset. |
9bf2229f8 [PATCH] cpusets: ... |
2330 |
*/ |
c9710d801 cpuset: drop "con... |
2331 |
static struct cpuset *nearest_hardwall_ancestor(struct cpuset *cs) |
9bf2229f8 [PATCH] cpusets: ... |
2332 |
{ |
c431069fe cpuset: remove cp... |
2333 2334 |
while (!(is_mem_exclusive(cs) || is_mem_hardwall(cs)) && parent_cs(cs)) cs = parent_cs(cs); |
9bf2229f8 [PATCH] cpusets: ... |
2335 2336 |
return cs; } |
d9fd8a6d4 [PATCH] kernel/cp... |
2337 |
/** |
344736f29 cpuset: simplify ... |
2338 |
* cpuset_node_allowed - Can we allocate on a memory node? |
a1bc5a4ee cpusets: replace ... |
2339 |
* @node: is this an allowed node? |
02a0e53d8 [PATCH] cpuset: r... |
2340 |
* @gfp_mask: memory allocation flags |
d9fd8a6d4 [PATCH] kernel/cp... |
2341 |
* |
a1bc5a4ee cpusets: replace ... |
2342 2343 2344 2345 2346 2347 |
* If we're in interrupt, yes, we can always allocate. If __GFP_THISNODE is * set, yes, we can always allocate. If node is in our task's mems_allowed, * yes. If it's not a __GFP_HARDWALL request and this node is in the nearest * hardwalled cpuset ancestor to this task's cpuset, yes. If the task has been * OOM killed and has access to memory reserves as specified by the TIF_MEMDIE * flag, yes. |
9bf2229f8 [PATCH] cpusets: ... |
2348 2349 |
* Otherwise, no. * |
02a0e53d8 [PATCH] cpuset: r... |
2350 2351 2352 2353 2354 2355 |
* 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'. * |
9bf2229f8 [PATCH] cpusets: ... |
2356 |
* GFP_USER allocations are marked with the __GFP_HARDWALL bit, |
c596d9f32 cpusets: allow TI... |
2357 2358 |
* and do not allow allocations outside the current tasks cpuset * unless the task has been OOM killed as is marked TIF_MEMDIE. |
9bf2229f8 [PATCH] cpusets: ... |
2359 |
* GFP_KERNEL allocations are not so marked, so can escape to the |
786083667 Cpuset hardwall f... |
2360 |
* nearest enclosing hardwalled ancestor cpuset. |
9bf2229f8 [PATCH] cpusets: ... |
2361 |
* |
8447a0fee cpuset: convert c... |
2362 |
* Scanning up parent cpusets requires callback_lock. The |
02a0e53d8 [PATCH] cpuset: r... |
2363 2364 2365 2366 |
* __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 |
8447a0fee cpuset: convert c... |
2367 |
* cpuset are short of memory, might require taking the callback_lock. |
9bf2229f8 [PATCH] cpusets: ... |
2368 |
* |
36be57ffe [PATCH] cpuset: u... |
2369 |
* The first call here from mm/page_alloc:get_page_from_freelist() |
02a0e53d8 [PATCH] cpuset: r... |
2370 2371 2372 |
* 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). |
36be57ffe [PATCH] cpuset: u... |
2373 2374 2375 2376 2377 2378 |
* * 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: |
9bf2229f8 [PATCH] cpusets: ... |
2379 2380 |
* in_interrupt - any node ok (current task context irrelevant) * GFP_ATOMIC - any node ok |
c596d9f32 cpusets: allow TI... |
2381 |
* TIF_MEMDIE - any node ok |
786083667 Cpuset hardwall f... |
2382 |
* GFP_KERNEL - any node in enclosing hardwalled cpuset ok |
9bf2229f8 [PATCH] cpusets: ... |
2383 |
* GFP_USER - only nodes in current tasks mems allowed ok. |
02a0e53d8 [PATCH] cpuset: r... |
2384 |
*/ |
344736f29 cpuset: simplify ... |
2385 |
int __cpuset_node_allowed(int node, gfp_t gfp_mask) |
1da177e4c Linux-2.6.12-rc2 |
2386 |
{ |
c9710d801 cpuset: drop "con... |
2387 |
struct cpuset *cs; /* current cpuset ancestors */ |
29afd49b7 [PATCH] cpuset: r... |
2388 |
int allowed; /* is allocation in zone z allowed? */ |
8447a0fee cpuset: convert c... |
2389 |
unsigned long flags; |
9bf2229f8 [PATCH] cpusets: ... |
2390 |
|
9b819d204 [PATCH] Add __GFP... |
2391 |
if (in_interrupt() || (gfp_mask & __GFP_THISNODE)) |
9bf2229f8 [PATCH] cpusets: ... |
2392 |
return 1; |
9bf2229f8 [PATCH] cpusets: ... |
2393 2394 |
if (node_isset(node, current->mems_allowed)) return 1; |
c596d9f32 cpusets: allow TI... |
2395 2396 2397 2398 2399 2400 |
/* * 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; |
9bf2229f8 [PATCH] cpusets: ... |
2401 2402 |
if (gfp_mask & __GFP_HARDWALL) /* If hardwall request, stop here */ return 0; |
5563e7707 [PATCH] cpuset: f... |
2403 2404 |
if (current->flags & PF_EXITING) /* Let dying task have memory */ return 1; |
9bf2229f8 [PATCH] cpusets: ... |
2405 |
/* Not hardwall and node outside mems_allowed: scan up cpusets */ |
8447a0fee cpuset: convert c... |
2406 |
spin_lock_irqsave(&callback_lock, flags); |
053199edf [PATCH] cpusets: ... |
2407 |
|
b8dadcb58 cpuset: use rcu_r... |
2408 |
rcu_read_lock(); |
786083667 Cpuset hardwall f... |
2409 |
cs = nearest_hardwall_ancestor(task_cs(current)); |
99afb0fd5 cpuset: fix a rac... |
2410 |
allowed = node_isset(node, cs->mems_allowed); |
b8dadcb58 cpuset: use rcu_r... |
2411 |
rcu_read_unlock(); |
053199edf [PATCH] cpusets: ... |
2412 |
|
8447a0fee cpuset: convert c... |
2413 |
spin_unlock_irqrestore(&callback_lock, flags); |
9bf2229f8 [PATCH] cpusets: ... |
2414 |
return allowed; |
1da177e4c Linux-2.6.12-rc2 |
2415 |
} |
ef08e3b49 [PATCH] cpusets: ... |
2416 |
/** |
6adef3ebe cpusets: new roun... |
2417 2418 |
* cpuset_mem_spread_node() - On which node to begin search for a file page * cpuset_slab_spread_node() - On which node to begin search for a slab page |
825a46af5 [PATCH] cpuset me... |
2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 |
* * 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(). */ |
6adef3ebe cpusets: new roun... |
2442 |
static int cpuset_spread_node(int *rotor) |
825a46af5 [PATCH] cpuset me... |
2443 2444 |
{ int node; |
6adef3ebe cpusets: new roun... |
2445 |
node = next_node(*rotor, current->mems_allowed); |
825a46af5 [PATCH] cpuset me... |
2446 2447 |
if (node == MAX_NUMNODES) node = first_node(current->mems_allowed); |
6adef3ebe cpusets: new roun... |
2448 |
*rotor = node; |
825a46af5 [PATCH] cpuset me... |
2449 2450 |
return node; } |
6adef3ebe cpusets: new roun... |
2451 2452 2453 |
int cpuset_mem_spread_node(void) { |
778d3b0ff cpusets: randomiz... |
2454 2455 2456 |
if (current->cpuset_mem_spread_rotor == NUMA_NO_NODE) current->cpuset_mem_spread_rotor = node_random(¤t->mems_allowed); |
6adef3ebe cpusets: new roun... |
2457 2458 2459 2460 2461 |
return cpuset_spread_node(¤t->cpuset_mem_spread_rotor); } int cpuset_slab_spread_node(void) { |
778d3b0ff cpusets: randomiz... |
2462 2463 2464 |
if (current->cpuset_slab_spread_rotor == NUMA_NO_NODE) current->cpuset_slab_spread_rotor = node_random(¤t->mems_allowed); |
6adef3ebe cpusets: new roun... |
2465 2466 |
return cpuset_spread_node(¤t->cpuset_slab_spread_rotor); } |
825a46af5 [PATCH] cpuset me... |
2467 2468 2469 |
EXPORT_SYMBOL_GPL(cpuset_mem_spread_node); /** |
bbe373f2c oom: compare cpus... |
2470 2471 2472 2473 2474 2475 2476 2477 |
* 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. |
ef08e3b49 [PATCH] cpusets: ... |
2478 |
**/ |
bbe373f2c oom: compare cpus... |
2479 2480 |
int cpuset_mems_allowed_intersects(const struct task_struct *tsk1, const struct task_struct *tsk2) |
ef08e3b49 [PATCH] cpusets: ... |
2481 |
{ |
bbe373f2c oom: compare cpus... |
2482 |
return nodes_intersects(tsk1->mems_allowed, tsk2->mems_allowed); |
ef08e3b49 [PATCH] cpusets: ... |
2483 |
} |
f440d98f8 cpuset: use cgrou... |
2484 |
#define CPUSET_NODELIST_LEN (256) |
75aa19941 oom: print trigge... |
2485 2486 |
/** * cpuset_print_task_mems_allowed - prints task's cpuset and mems_allowed |
fc34ac1dc kernel/cpuset.c: ... |
2487 |
* @tsk: pointer to task_struct of some task. |
75aa19941 oom: print trigge... |
2488 2489 |
* * Description: Prints @task's name, cpuset name, and cached copy of its |
b8dadcb58 cpuset: use rcu_r... |
2490 |
* mems_allowed to the kernel log. |
75aa19941 oom: print trigge... |
2491 2492 2493 |
*/ void cpuset_print_task_mems_allowed(struct task_struct *tsk) { |
f440d98f8 cpuset: use cgrou... |
2494 2495 2496 |
/* Statically allocated to prevent using excess stack. */ static char cpuset_nodelist[CPUSET_NODELIST_LEN]; static DEFINE_SPINLOCK(cpuset_buffer_lock); |
b8dadcb58 cpuset: use rcu_r... |
2497 |
struct cgroup *cgrp; |
75aa19941 oom: print trigge... |
2498 |
|
f440d98f8 cpuset: use cgrou... |
2499 |
spin_lock(&cpuset_buffer_lock); |
b8dadcb58 cpuset: use rcu_r... |
2500 |
rcu_read_lock(); |
63f43f55c cpuset: fix cpuse... |
2501 |
|
b8dadcb58 cpuset: use rcu_r... |
2502 |
cgrp = task_cs(tsk)->css.cgroup; |
75aa19941 oom: print trigge... |
2503 2504 |
nodelist_scnprintf(cpuset_nodelist, CPUSET_NODELIST_LEN, tsk->mems_allowed); |
12d3089c1 kernel/cpuset.c: ... |
2505 |
pr_info("%s cpuset=", tsk->comm); |
e61734c55 cgroup: remove cg... |
2506 2507 2508 |
pr_cont_cgroup_name(cgrp); pr_cont(" mems_allowed=%s ", cpuset_nodelist); |
f440d98f8 cpuset: use cgrou... |
2509 |
|
cfb5966be cpuset: fix RCU l... |
2510 |
rcu_read_unlock(); |
75aa19941 oom: print trigge... |
2511 2512 |
spin_unlock(&cpuset_buffer_lock); } |
1da177e4c Linux-2.6.12-rc2 |
2513 |
/* |
3e0d98b9f [PATCH] cpuset: m... |
2514 2515 2516 2517 |
* 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. */ |
c5b2aff89 [PATCH] cpuset: m... |
2518 |
int cpuset_memory_pressure_enabled __read_mostly; |
3e0d98b9f [PATCH] cpuset: m... |
2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 |
/** * 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) { |
b8dadcb58 cpuset: use rcu_r... |
2540 |
rcu_read_lock(); |
8793d854e Task Control Grou... |
2541 |
fmeter_markevent(&task_cs(current)->fmeter); |
b8dadcb58 cpuset: use rcu_r... |
2542 |
rcu_read_unlock(); |
3e0d98b9f [PATCH] cpuset: m... |
2543 |
} |
8793d854e Task Control Grou... |
2544 |
#ifdef CONFIG_PROC_PID_CPUSET |
3e0d98b9f [PATCH] cpuset: m... |
2545 |
/* |
1da177e4c Linux-2.6.12-rc2 |
2546 2547 2548 |
* proc_cpuset_show() * - Print tasks cpuset path into seq_file. * - Used for /proc/<pid>/cpuset. |
053199edf [PATCH] cpusets: ... |
2549 2550 |
* - 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, |
5d21cc2db cpuset: replace c... |
2551 |
* and we take cpuset_mutex, keeping cpuset_attach() from changing it |
2df167a30 cgroups: update c... |
2552 |
* anyway. |
1da177e4c Linux-2.6.12-rc2 |
2553 |
*/ |
52de4779f cpuset: simplify ... |
2554 2555 |
int proc_cpuset_show(struct seq_file *m, struct pid_namespace *ns, struct pid *pid, struct task_struct *tsk) |
1da177e4c Linux-2.6.12-rc2 |
2556 |
{ |
e61734c55 cgroup: remove cg... |
2557 |
char *buf, *p; |
8793d854e Task Control Grou... |
2558 |
struct cgroup_subsys_state *css; |
99f895518 [PATCH] proc: don... |
2559 |
int retval; |
1da177e4c Linux-2.6.12-rc2 |
2560 |
|
99f895518 [PATCH] proc: don... |
2561 |
retval = -ENOMEM; |
e61734c55 cgroup: remove cg... |
2562 |
buf = kmalloc(PATH_MAX, GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
2563 |
if (!buf) |
99f895518 [PATCH] proc: don... |
2564 |
goto out; |
e61734c55 cgroup: remove cg... |
2565 |
retval = -ENAMETOOLONG; |
27e89ae5d cpuset: fix RCU l... |
2566 |
rcu_read_lock(); |
073219e99 cgroup: clean up ... |
2567 |
css = task_css(tsk, cpuset_cgrp_id); |
e61734c55 cgroup: remove cg... |
2568 |
p = cgroup_path(css->cgroup, buf, PATH_MAX); |
27e89ae5d cpuset: fix RCU l... |
2569 |
rcu_read_unlock(); |
e61734c55 cgroup: remove cg... |
2570 |
if (!p) |
52de4779f cpuset: simplify ... |
2571 |
goto out_free; |
e61734c55 cgroup: remove cg... |
2572 |
seq_puts(m, p); |
1da177e4c Linux-2.6.12-rc2 |
2573 2574 |
seq_putc(m, ' '); |
e61734c55 cgroup: remove cg... |
2575 |
retval = 0; |
99f895518 [PATCH] proc: don... |
2576 |
out_free: |
1da177e4c Linux-2.6.12-rc2 |
2577 |
kfree(buf); |
99f895518 [PATCH] proc: don... |
2578 |
out: |
1da177e4c Linux-2.6.12-rc2 |
2579 2580 |
return retval; } |
8793d854e Task Control Grou... |
2581 |
#endif /* CONFIG_PROC_PID_CPUSET */ |
1da177e4c Linux-2.6.12-rc2 |
2582 |
|
d01d48278 sched: Always sho... |
2583 |
/* Display task mems_allowed in /proc/<pid>/status file. */ |
df5f8314c proc: seqfile con... |
2584 2585 |
void cpuset_task_status_allowed(struct seq_file *m, struct task_struct *task) { |
fc34ac1dc kernel/cpuset.c: ... |
2586 |
seq_puts(m, "Mems_allowed:\t"); |
30e8e1360 cpuset: use seq_*... |
2587 |
seq_nodemask(m, &task->mems_allowed); |
fc34ac1dc kernel/cpuset.c: ... |
2588 2589 2590 |
seq_puts(m, " "); seq_puts(m, "Mems_allowed_list:\t"); |
30e8e1360 cpuset: use seq_*... |
2591 |
seq_nodemask_list(m, &task->mems_allowed); |
fc34ac1dc kernel/cpuset.c: ... |
2592 2593 |
seq_puts(m, " "); |
1da177e4c Linux-2.6.12-rc2 |
2594 |
} |