Blame view
kernel/cpuset.c
76.5 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 |
#include <linux/spinlock.h> #include <linux/stat.h> #include <linux/string.h> #include <linux/time.h> |
d2b436580 cpuset: Replace a... |
53 |
#include <linux/time64.h> |
1da177e4c Linux-2.6.12-rc2 |
54 55 56 57 |
#include <linux/backing-dev.h> #include <linux/sort.h> #include <asm/uaccess.h> |
60063497a atomic: use <linu... |
58 |
#include <linux/atomic.h> |
3d3f26a7b [PATCH] kernel/cp... |
59 |
#include <linux/mutex.h> |
956db3ca0 hotplug cpu: move... |
60 |
#include <linux/cgroup.h> |
e44193d39 cpuset: let hotpl... |
61 |
#include <linux/wait.h> |
1da177e4c Linux-2.6.12-rc2 |
62 |
|
002f29062 cpuset: use stati... |
63 |
DEFINE_STATIC_KEY_FALSE(cpusets_enabled_key); |
202f72d5d [PATCH] cpuset: n... |
64 |
|
3e0d98b9f [PATCH] cpuset: m... |
65 66 67 68 69 |
/* See "Frequency meter" comments, below. */ struct fmeter { int cnt; /* unprocessed events count */ int val; /* most recent output value */ |
d2b436580 cpuset: Replace a... |
70 |
time64_t time; /* clock (secs) when val computed */ |
3e0d98b9f [PATCH] cpuset: m... |
71 72 |
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 |
|
e93ad19d0 cpuset: make mm m... |
271 |
static struct workqueue_struct *cpuset_migrate_mm_wq; |
cf417141c sched, cpuset: re... |
272 |
/* |
3a5a6d0c2 cpuset: don't nes... |
273 274 275 |
* CPU / memory hotplug is handled asynchronously. */ static void cpuset_hotplug_workfn(struct work_struct *work); |
3a5a6d0c2 cpuset: don't nes... |
276 |
static DECLARE_WORK(cpuset_hotplug_work, cpuset_hotplug_workfn); |
e44193d39 cpuset: let hotpl... |
277 |
static DECLARE_WAIT_QUEUE_HEAD(cpuset_attach_wq); |
3a5a6d0c2 cpuset: don't nes... |
278 |
/* |
cf417141c sched, cpuset: re... |
279 |
* This is ugly, but preserves the userspace API for existing cpuset |
8793d854e Task Control Grou... |
280 |
* users. If someone tries to mount the "cpuset" filesystem, we |
cf417141c sched, cpuset: re... |
281 282 |
* silently switch it to mount "cgroup" instead */ |
f7e835710 convert cgroup an... |
283 284 |
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 |
285 |
{ |
8793d854e Task Control Grou... |
286 |
struct file_system_type *cgroup_fs = get_fs_type("cgroup"); |
f7e835710 convert cgroup an... |
287 |
struct dentry *ret = ERR_PTR(-ENODEV); |
8793d854e Task Control Grou... |
288 289 290 291 |
if (cgroup_fs) { char mountopts[] = "cpuset,noprefix," "release_agent=/sbin/cpuset_release_agent"; |
f7e835710 convert cgroup an... |
292 293 |
ret = cgroup_fs->mount(cgroup_fs, flags, unused_dev_name, mountopts); |
8793d854e Task Control Grou... |
294 295 296 |
put_filesystem(cgroup_fs); } return ret; |
1da177e4c Linux-2.6.12-rc2 |
297 298 299 300 |
} static struct file_system_type cpuset_fs_type = { .name = "cpuset", |
f7e835710 convert cgroup an... |
301 |
.mount = cpuset_mount, |
1da177e4c Linux-2.6.12-rc2 |
302 |
}; |
1da177e4c Linux-2.6.12-rc2 |
303 |
/* |
300ed6cbb cpuset: convert c... |
304 |
* Return in pmask the portion of a cpusets's cpus_allowed that |
1da177e4c Linux-2.6.12-rc2 |
305 |
* are online. If none are online, walk up the cpuset hierarchy |
28b89b9e6 cpuset: handle ra... |
306 |
* until we find one that does have some online cpus. |
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 |
{ |
28b89b9e6 cpuset: handle ra... |
315 |
while (!cpumask_intersects(cs->effective_cpus, cpu_online_mask)) { |
c431069fe cpuset: remove cp... |
316 |
cs = parent_cs(cs); |
28b89b9e6 cpuset: handle ra... |
317 318 319 320 321 322 323 324 325 326 327 328 |
if (unlikely(!cs)) { /* * The top cpuset doesn't have any online cpu as a * consequence of a race between cpuset_hotplug_work * and cpu hotplug notifier. But we know the top * cpuset's effective_cpus is on its way to to be * identical to cpu_online_mask. */ cpumask_copy(pmask, cpu_online_mask); return; } } |
ae1c80238 cpuset: apply cs-... |
329 |
cpumask_and(pmask, cs->effective_cpus, cpu_online_mask); |
1da177e4c Linux-2.6.12-rc2 |
330 331 332 333 |
} /* * Return in *pmask the portion of a cpusets's mems_allowed that |
0e1e7c7a7 Memoryless nodes:... |
334 335 |
* 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... |
336 |
* online mems. The top cpuset always has some mems online. |
1da177e4c Linux-2.6.12-rc2 |
337 338 |
* * One way or another, we guarantee to return some non-empty subset |
38d7bee9d cpuset: use N_MEM... |
339 |
* of node_states[N_MEMORY]. |
1da177e4c Linux-2.6.12-rc2 |
340 |
* |
8447a0fee cpuset: convert c... |
341 |
* Call with callback_lock or cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
342 |
*/ |
c9710d801 cpuset: drop "con... |
343 |
static void guarantee_online_mems(struct cpuset *cs, nodemask_t *pmask) |
1da177e4c Linux-2.6.12-rc2 |
344 |
{ |
ae1c80238 cpuset: apply cs-... |
345 |
while (!nodes_intersects(cs->effective_mems, node_states[N_MEMORY])) |
c431069fe cpuset: remove cp... |
346 |
cs = parent_cs(cs); |
ae1c80238 cpuset: apply cs-... |
347 |
nodes_and(*pmask, cs->effective_mems, node_states[N_MEMORY]); |
1da177e4c Linux-2.6.12-rc2 |
348 |
} |
f3b39d47e cpusets: restruct... |
349 350 351 |
/* * update task's spread flag if cpuset's page/slab spread flag is set * |
8447a0fee cpuset: convert c... |
352 |
* Call with callback_lock or cpuset_mutex held. |
f3b39d47e cpusets: restruct... |
353 354 355 356 357 |
*/ static void cpuset_update_task_spread_flag(struct cpuset *cs, struct task_struct *tsk) { if (is_spread_page(cs)) |
2ad654bc5 cpuset: PF_SPREAD... |
358 |
task_set_spread_page(tsk); |
f3b39d47e cpusets: restruct... |
359 |
else |
2ad654bc5 cpuset: PF_SPREAD... |
360 |
task_clear_spread_page(tsk); |
f3b39d47e cpusets: restruct... |
361 |
if (is_spread_slab(cs)) |
2ad654bc5 cpuset: PF_SPREAD... |
362 |
task_set_spread_slab(tsk); |
f3b39d47e cpusets: restruct... |
363 |
else |
2ad654bc5 cpuset: PF_SPREAD... |
364 |
task_clear_spread_slab(tsk); |
f3b39d47e cpusets: restruct... |
365 |
} |
1da177e4c Linux-2.6.12-rc2 |
366 367 368 369 370 |
/* * 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... |
371 |
* are only set if the other's are set. Call holding cpuset_mutex. |
1da177e4c Linux-2.6.12-rc2 |
372 373 374 375 |
*/ static int is_cpuset_subset(const struct cpuset *p, const struct cpuset *q) { |
300ed6cbb cpuset: convert c... |
376 |
return cpumask_subset(p->cpus_allowed, q->cpus_allowed) && |
1da177e4c Linux-2.6.12-rc2 |
377 378 379 380 |
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... |
381 382 383 384 |
/** * alloc_trial_cpuset - allocate a trial cpuset * @cs: the cpuset that the trial cpuset duplicates */ |
c9710d801 cpuset: drop "con... |
385 |
static struct cpuset *alloc_trial_cpuset(struct cpuset *cs) |
645fcc9d2 cpuset: don't all... |
386 |
{ |
300ed6cbb cpuset: convert c... |
387 388 389 390 391 |
struct cpuset *trial; trial = kmemdup(cs, sizeof(*cs), GFP_KERNEL); if (!trial) return NULL; |
e2b9a3d7d cpuset: add cs->e... |
392 393 394 395 |
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... |
396 |
|
e2b9a3d7d cpuset: add cs->e... |
397 398 |
cpumask_copy(trial->cpus_allowed, cs->cpus_allowed); cpumask_copy(trial->effective_cpus, cs->effective_cpus); |
300ed6cbb cpuset: convert c... |
399 |
return trial; |
e2b9a3d7d cpuset: add cs->e... |
400 401 402 403 404 405 |
free_cpus: free_cpumask_var(trial->cpus_allowed); free_cs: kfree(trial); return NULL; |
645fcc9d2 cpuset: don't all... |
406 407 408 409 410 411 412 413 |
} /** * 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... |
414 |
free_cpumask_var(trial->effective_cpus); |
300ed6cbb cpuset: convert c... |
415 |
free_cpumask_var(trial->cpus_allowed); |
645fcc9d2 cpuset: don't all... |
416 417 |
kfree(trial); } |
1da177e4c Linux-2.6.12-rc2 |
418 419 420 421 422 423 424 |
/* * 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... |
425 |
* cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
426 427 428 429 430 431 432 433 434 435 436 |
* * '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... |
437 |
static int validate_change(struct cpuset *cur, struct cpuset *trial) |
1da177e4c Linux-2.6.12-rc2 |
438 |
{ |
492eb21b9 cgroup: make hier... |
439 |
struct cgroup_subsys_state *css; |
1da177e4c Linux-2.6.12-rc2 |
440 |
struct cpuset *c, *par; |
ae8086ce1 cpuset: introduce... |
441 442 443 |
int ret; rcu_read_lock(); |
1da177e4c Linux-2.6.12-rc2 |
444 445 |
/* Each of our child cpusets must be a subset of us */ |
ae8086ce1 cpuset: introduce... |
446 |
ret = -EBUSY; |
492eb21b9 cgroup: make hier... |
447 |
cpuset_for_each_child(c, css, cur) |
ae8086ce1 cpuset: introduce... |
448 449 |
if (!is_cpuset_subset(c, trial)) goto out; |
1da177e4c Linux-2.6.12-rc2 |
450 451 |
/* Remaining checks don't apply to root cpuset */ |
ae8086ce1 cpuset: introduce... |
452 |
ret = 0; |
696040670 [PATCH] cpuset: m... |
453 |
if (cur == &top_cpuset) |
ae8086ce1 cpuset: introduce... |
454 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
455 |
|
c431069fe cpuset: remove cp... |
456 |
par = parent_cs(cur); |
696040670 [PATCH] cpuset: m... |
457 |
|
7e88291be cpuset: make cs->... |
458 |
/* On legacy hiearchy, we must be a subset of our parent cpuset. */ |
ae8086ce1 cpuset: introduce... |
459 |
ret = -EACCES; |
9e10a130d cgroup: replace c... |
460 461 |
if (!cgroup_subsys_on_dfl(cpuset_cgrp_subsys) && !is_cpuset_subset(trial, par)) |
ae8086ce1 cpuset: introduce... |
462 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
463 |
|
2df167a30 cgroups: update c... |
464 465 466 467 |
/* * If either I or some sibling (!= me) is exclusive, we can't * overlap */ |
ae8086ce1 cpuset: introduce... |
468 |
ret = -EINVAL; |
492eb21b9 cgroup: make hier... |
469 |
cpuset_for_each_child(c, css, par) { |
1da177e4c Linux-2.6.12-rc2 |
470 471 |
if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) && c != cur && |
300ed6cbb cpuset: convert c... |
472 |
cpumask_intersects(trial->cpus_allowed, c->cpus_allowed)) |
ae8086ce1 cpuset: introduce... |
473 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
474 475 476 |
if ((is_mem_exclusive(trial) || is_mem_exclusive(c)) && c != cur && nodes_intersects(trial->mems_allowed, c->mems_allowed)) |
ae8086ce1 cpuset: introduce... |
477 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
478 |
} |
452477fa6 cpuset: pin down ... |
479 480 |
/* * Cpusets with tasks - existing or newly being attached - can't |
1c09b195d cpuset: fix a reg... |
481 |
* be changed to have empty cpus_allowed or mems_allowed. |
452477fa6 cpuset: pin down ... |
482 |
*/ |
ae8086ce1 cpuset: introduce... |
483 |
ret = -ENOSPC; |
27bd4dbb8 cgroup: replace c... |
484 |
if ((cgroup_is_populated(cur->css.cgroup) || cur->attach_in_progress)) { |
1c09b195d cpuset: fix a reg... |
485 486 487 488 489 490 491 |
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... |
492 |
|
f82f80426 sched/deadline: E... |
493 494 495 496 497 498 499 500 501 |
/* * 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... |
502 503 504 505 |
ret = 0; out: rcu_read_unlock(); return ret; |
1da177e4c Linux-2.6.12-rc2 |
506 |
} |
db7f47cf4 cpusets: allow cp... |
507 |
#ifdef CONFIG_SMP |
85d7b9498 [PATCH] Dynamic s... |
508 |
/* |
cf417141c sched, cpuset: re... |
509 |
* Helper routine for generate_sched_domains(). |
8b5f1c52d cpuset: use effec... |
510 |
* Do cpusets a, b have overlapping effective cpus_allowed masks? |
029190c51 cpuset sched_load... |
511 |
*/ |
029190c51 cpuset sched_load... |
512 513 |
static int cpusets_overlap(struct cpuset *a, struct cpuset *b) { |
8b5f1c52d cpuset: use effec... |
514 |
return cpumask_intersects(a->effective_cpus, b->effective_cpus); |
029190c51 cpuset sched_load... |
515 |
} |
1d3504fcf sched, cpuset: cu... |
516 517 518 |
static void update_domain_attr(struct sched_domain_attr *dattr, struct cpuset *c) { |
1d3504fcf sched, cpuset: cu... |
519 520 521 522 |
if (dattr->relax_domain_level < c->relax_domain_level) dattr->relax_domain_level = c->relax_domain_level; return; } |
fc560a26a cpuset: replace c... |
523 524 |
static void update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *root_cs) |
f5393693e cpuset: speed up ... |
525 |
{ |
fc560a26a cpuset: replace c... |
526 |
struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
527 |
struct cgroup_subsys_state *pos_css; |
f5393693e cpuset: speed up ... |
528 |
|
fc560a26a cpuset: replace c... |
529 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
530 |
cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { |
fc560a26a cpuset: replace c... |
531 532 |
/* skip the whole subtree if @cp doesn't have any CPU */ if (cpumask_empty(cp->cpus_allowed)) { |
492eb21b9 cgroup: make hier... |
533 |
pos_css = css_rightmost_descendant(pos_css); |
f5393693e cpuset: speed up ... |
534 |
continue; |
fc560a26a cpuset: replace c... |
535 |
} |
f5393693e cpuset: speed up ... |
536 537 538 |
if (is_sched_load_balance(cp)) update_domain_attr(dattr, cp); |
f5393693e cpuset: speed up ... |
539 |
} |
fc560a26a cpuset: replace c... |
540 |
rcu_read_unlock(); |
f5393693e cpuset: speed up ... |
541 |
} |
029190c51 cpuset sched_load... |
542 |
/* |
cf417141c sched, cpuset: re... |
543 544 545 546 547 |
* 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... |
548 |
* The output of this function needs to be passed to kernel/sched/core.c |
cf417141c sched, cpuset: re... |
549 550 551 |
* 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... |
552 |
* |
45ce80fb6 cgroups: consolid... |
553 |
* See "What is sched_load_balance" in Documentation/cgroups/cpusets.txt |
029190c51 cpuset sched_load... |
554 555 556 557 558 559 560 |
* 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... |
561 |
* Must be called with cpuset_mutex held. |
029190c51 cpuset sched_load... |
562 563 |
* * The three key local variables below are: |
aeed68242 cpuset: clean up ... |
564 |
* q - a linked-list queue of cpuset pointers, used to implement a |
029190c51 cpuset sched_load... |
565 566 567 568 569 570 571 572 573 574 575 576 |
* top-down scan of all cpusets. This scan loads a pointer * to each cpuset marked is_sched_load_balance into the * array 'csa'. For our purposes, rebuilding the schedulers * sched domains, we can ignore !is_sched_load_balance cpusets. * csa - (for CpuSet Array) Array of pointers to all the cpusets * that need to be load balanced, for convenient iterative * access by the subsequent code that finds the best partition, * i.e the set of domains (subsets) of CPUs such that the * cpus_allowed of every cpuset marked is_sched_load_balance * is a subset of one of these domains, while there are as * many such domains as possible, each as small as possible. * doms - Conversion of 'csa' to an array of cpumasks, for passing to |
0a0fca9d8 sched: Rename sch... |
577 |
* the kernel/sched/core.c routine partition_sched_domains() in a |
029190c51 cpuset sched_load... |
578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 |
* 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... |
596 |
static int generate_sched_domains(cpumask_var_t **domains, |
cf417141c sched, cpuset: re... |
597 |
struct sched_domain_attr **attributes) |
029190c51 cpuset sched_load... |
598 |
{ |
029190c51 cpuset sched_load... |
599 600 601 602 |
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... |
603 |
cpumask_var_t *doms; /* resulting partition; i.e. sched domains */ |
47b8ea718 cpusets, isolcpus... |
604 |
cpumask_var_t non_isolated_cpus; /* load balanced CPUs */ |
1d3504fcf sched, cpuset: cu... |
605 |
struct sched_domain_attr *dattr; /* attributes for custom domains */ |
1583715dd sched, cpusets: f... |
606 |
int ndoms = 0; /* number of sched domains in result */ |
6af866af3 cpuset: remove re... |
607 |
int nslot; /* next empty doms[] struct cpumask slot */ |
492eb21b9 cgroup: make hier... |
608 |
struct cgroup_subsys_state *pos_css; |
029190c51 cpuset sched_load... |
609 |
|
029190c51 cpuset sched_load... |
610 |
doms = NULL; |
1d3504fcf sched, cpuset: cu... |
611 |
dattr = NULL; |
cf417141c sched, cpuset: re... |
612 |
csa = NULL; |
029190c51 cpuset sched_load... |
613 |
|
47b8ea718 cpusets, isolcpus... |
614 615 616 |
if (!alloc_cpumask_var(&non_isolated_cpus, GFP_KERNEL)) goto done; cpumask_andnot(non_isolated_cpus, cpu_possible_mask, cpu_isolated_map); |
029190c51 cpuset sched_load... |
617 618 |
/* Special case for the 99% of systems with one, full, sched domain */ if (is_sched_load_balance(&top_cpuset)) { |
acc3f5d7c cpumask: Partitio... |
619 620 |
ndoms = 1; doms = alloc_sched_domains(ndoms); |
029190c51 cpuset sched_load... |
621 |
if (!doms) |
cf417141c sched, cpuset: re... |
622 |
goto done; |
1d3504fcf sched, cpuset: cu... |
623 624 625 |
dattr = kmalloc(sizeof(struct sched_domain_attr), GFP_KERNEL); if (dattr) { *dattr = SD_ATTR_INIT; |
93a655755 cpuset: fix wrong... |
626 |
update_domain_attr_tree(dattr, &top_cpuset); |
1d3504fcf sched, cpuset: cu... |
627 |
} |
47b8ea718 cpusets, isolcpus... |
628 629 |
cpumask_and(doms[0], top_cpuset.effective_cpus, non_isolated_cpus); |
cf417141c sched, cpuset: re... |
630 |
|
cf417141c sched, cpuset: re... |
631 |
goto done; |
029190c51 cpuset sched_load... |
632 |
} |
664eeddee mm: page_alloc: u... |
633 |
csa = kmalloc(nr_cpusets() * sizeof(cp), GFP_KERNEL); |
029190c51 cpuset sched_load... |
634 635 636 |
if (!csa) goto done; csn = 0; |
fc560a26a cpuset: replace c... |
637 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
638 |
cpuset_for_each_descendant_pre(cp, pos_css, &top_cpuset) { |
bd8815a6d cgroup: make css_... |
639 640 |
if (cp == &top_cpuset) continue; |
f5393693e cpuset: speed up ... |
641 |
/* |
fc560a26a cpuset: replace c... |
642 643 644 645 646 647 |
* 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 ... |
648 |
*/ |
fc560a26a cpuset: replace c... |
649 |
if (!cpumask_empty(cp->cpus_allowed) && |
47b8ea718 cpusets, isolcpus... |
650 651 |
!(is_sched_load_balance(cp) && cpumask_intersects(cp->cpus_allowed, non_isolated_cpus))) |
f5393693e cpuset: speed up ... |
652 |
continue; |
489a5393a cpuset: don't pas... |
653 |
|
fc560a26a cpuset: replace c... |
654 655 656 657 |
if (is_sched_load_balance(cp)) csa[csn++] = cp; /* skip @cp's subtree */ |
492eb21b9 cgroup: make hier... |
658 |
pos_css = css_rightmost_descendant(pos_css); |
fc560a26a cpuset: replace c... |
659 660 |
} rcu_read_unlock(); |
029190c51 cpuset sched_load... |
661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 |
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... |
688 689 690 691 |
/* * Now we know how many domains to create. * Convert <csn, csa> to <ndoms, doms> and populate cpu masks. */ |
acc3f5d7c cpumask: Partitio... |
692 |
doms = alloc_sched_domains(ndoms); |
700018e0a cpuset: fix regre... |
693 |
if (!doms) |
cf417141c sched, cpuset: re... |
694 |
goto done; |
cf417141c sched, cpuset: re... |
695 696 697 698 699 |
/* * 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... |
700 |
dattr = kmalloc(ndoms * sizeof(struct sched_domain_attr), GFP_KERNEL); |
029190c51 cpuset sched_load... |
701 702 703 |
for (nslot = 0, i = 0; i < csn; i++) { struct cpuset *a = csa[i]; |
6af866af3 cpuset: remove re... |
704 |
struct cpumask *dp; |
029190c51 cpuset sched_load... |
705 |
int apn = a->pn; |
cf417141c sched, cpuset: re... |
706 707 708 709 |
if (apn < 0) { /* Skip completed partitions */ continue; } |
acc3f5d7c cpumask: Partitio... |
710 |
dp = doms[nslot]; |
cf417141c sched, cpuset: re... |
711 712 713 714 |
if (nslot == ndoms) { static int warnings = 10; if (warnings) { |
12d3089c1 kernel/cpuset.c: ... |
715 716 717 |
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... |
718 |
warnings--; |
029190c51 cpuset sched_load... |
719 |
} |
cf417141c sched, cpuset: re... |
720 721 |
continue; } |
029190c51 cpuset sched_load... |
722 |
|
6af866af3 cpuset: remove re... |
723 |
cpumask_clear(dp); |
cf417141c sched, cpuset: re... |
724 725 726 727 728 729 |
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... |
730 |
cpumask_or(dp, dp, b->effective_cpus); |
47b8ea718 cpusets, isolcpus... |
731 |
cpumask_and(dp, dp, non_isolated_cpus); |
cf417141c sched, cpuset: re... |
732 733 734 735 736 |
if (dattr) update_domain_attr_tree(dattr + nslot, b); /* Done with this partition */ b->pn = -1; |
029190c51 cpuset sched_load... |
737 |
} |
029190c51 cpuset sched_load... |
738 |
} |
cf417141c sched, cpuset: re... |
739 |
nslot++; |
029190c51 cpuset sched_load... |
740 741 |
} BUG_ON(nslot != ndoms); |
cf417141c sched, cpuset: re... |
742 |
done: |
47b8ea718 cpusets, isolcpus... |
743 |
free_cpumask_var(non_isolated_cpus); |
cf417141c sched, cpuset: re... |
744 |
kfree(csa); |
700018e0a cpuset: fix regre... |
745 746 747 748 749 750 |
/* * Fallback to the default domain if kmalloc() failed. * See comments in partition_sched_domains(). */ if (doms == NULL) ndoms = 1; |
cf417141c sched, cpuset: re... |
751 752 753 754 755 756 757 758 |
*domains = doms; *attributes = dattr; return ndoms; } /* * Rebuild scheduler domains. * |
699140ba8 cpuset: drop asyn... |
759 760 761 762 763 |
* 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... |
764 |
* |
5d21cc2db cpuset: replace c... |
765 |
* Call with cpuset_mutex held. Takes get_online_cpus(). |
cf417141c sched, cpuset: re... |
766 |
*/ |
699140ba8 cpuset: drop asyn... |
767 |
static void rebuild_sched_domains_locked(void) |
cf417141c sched, cpuset: re... |
768 769 |
{ struct sched_domain_attr *attr; |
acc3f5d7c cpumask: Partitio... |
770 |
cpumask_var_t *doms; |
cf417141c sched, cpuset: re... |
771 |
int ndoms; |
5d21cc2db cpuset: replace c... |
772 |
lockdep_assert_held(&cpuset_mutex); |
86ef5c9a8 cpu-hotplug: repl... |
773 |
get_online_cpus(); |
cf417141c sched, cpuset: re... |
774 |
|
5b16c2a49 cpuset: fix cpu h... |
775 776 777 778 779 |
/* * 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... |
780 |
if (!cpumask_equal(top_cpuset.effective_cpus, cpu_active_mask)) |
5b16c2a49 cpuset: fix cpu h... |
781 |
goto out; |
cf417141c sched, cpuset: re... |
782 |
/* Generate domain masks and attrs */ |
cf417141c sched, cpuset: re... |
783 |
ndoms = generate_sched_domains(&doms, &attr); |
cf417141c sched, cpuset: re... |
784 785 786 |
/* Have scheduler rebuild the domains */ partition_sched_domains(ndoms, doms, attr); |
5b16c2a49 cpuset: fix cpu h... |
787 |
out: |
86ef5c9a8 cpu-hotplug: repl... |
788 |
put_online_cpus(); |
cf417141c sched, cpuset: re... |
789 |
} |
db7f47cf4 cpusets: allow cp... |
790 |
#else /* !CONFIG_SMP */ |
699140ba8 cpuset: drop asyn... |
791 |
static void rebuild_sched_domains_locked(void) |
db7f47cf4 cpusets: allow cp... |
792 793 |
{ } |
db7f47cf4 cpusets: allow cp... |
794 |
#endif /* CONFIG_SMP */ |
029190c51 cpuset sched_load... |
795 |
|
cf417141c sched, cpuset: re... |
796 797 |
void rebuild_sched_domains(void) { |
5d21cc2db cpuset: replace c... |
798 |
mutex_lock(&cpuset_mutex); |
699140ba8 cpuset: drop asyn... |
799 |
rebuild_sched_domains_locked(); |
5d21cc2db cpuset: replace c... |
800 |
mutex_unlock(&cpuset_mutex); |
029190c51 cpuset sched_load... |
801 |
} |
58f4790b7 cpusets: update_c... |
802 |
/** |
0b2f630a2 cpusets: restruct... |
803 804 |
* 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... |
805 |
* |
d66393e54 cpuset: use css_t... |
806 807 808 |
* 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... |
809 |
*/ |
d66393e54 cpuset: use css_t... |
810 |
static void update_tasks_cpumask(struct cpuset *cs) |
0b2f630a2 cpusets: restruct... |
811 |
{ |
d66393e54 cpuset: use css_t... |
812 813 814 815 816 |
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-... |
817 |
set_cpus_allowed_ptr(task, cs->effective_cpus); |
d66393e54 cpuset: use css_t... |
818 |
css_task_iter_end(&it); |
0b2f630a2 cpusets: restruct... |
819 |
} |
5c5cc6232 cpuset: allow to ... |
820 |
/* |
734d45130 cpuset: update cs... |
821 822 823 824 825 826 |
* 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 ... |
827 |
* |
734d45130 cpuset: update cs... |
828 |
* On legacy hierachy, effective_cpus will be the same with cpu_allowed. |
5c5cc6232 cpuset: allow to ... |
829 830 831 |
* * Called with cpuset_mutex held */ |
734d45130 cpuset: update cs... |
832 |
static void update_cpumasks_hier(struct cpuset *cs, struct cpumask *new_cpus) |
5c5cc6232 cpuset: allow to ... |
833 834 |
{ struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
835 |
struct cgroup_subsys_state *pos_css; |
8b5f1c52d cpuset: use effec... |
836 |
bool need_rebuild_sched_domains = false; |
5c5cc6232 cpuset: allow to ... |
837 838 |
rcu_read_lock(); |
734d45130 cpuset: update cs... |
839 840 841 842 |
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... |
843 844 845 846 |
/* * If it becomes empty, inherit the effective mask of the * parent, which is guaranteed to have some CPUs. */ |
9e10a130d cgroup: replace c... |
847 848 |
if (cgroup_subsys_on_dfl(cpuset_cgrp_subsys) && cpumask_empty(new_cpus)) |
554b0d1c8 cpuset: inherit a... |
849 |
cpumask_copy(new_cpus, parent->effective_cpus); |
734d45130 cpuset: update cs... |
850 851 852 853 |
/* 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 ... |
854 |
} |
734d45130 cpuset: update cs... |
855 |
|
ec903c0c8 cgroup: rename cs... |
856 |
if (!css_tryget_online(&cp->css)) |
5c5cc6232 cpuset: allow to ... |
857 858 |
continue; rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
859 |
spin_lock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
860 |
cpumask_copy(cp->effective_cpus, new_cpus); |
8447a0fee cpuset: convert c... |
861 |
spin_unlock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
862 |
|
9e10a130d cgroup: replace c... |
863 |
WARN_ON(!cgroup_subsys_on_dfl(cpuset_cgrp_subsys) && |
734d45130 cpuset: update cs... |
864 |
!cpumask_equal(cp->cpus_allowed, cp->effective_cpus)); |
d66393e54 cpuset: use css_t... |
865 |
update_tasks_cpumask(cp); |
5c5cc6232 cpuset: allow to ... |
866 |
|
8b5f1c52d cpuset: use effec... |
867 868 869 870 871 872 873 |
/* * 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 ... |
874 875 876 877 |
rcu_read_lock(); css_put(&cp->css); } rcu_read_unlock(); |
8b5f1c52d cpuset: use effec... |
878 879 880 |
if (need_rebuild_sched_domains) rebuild_sched_domains_locked(); |
5c5cc6232 cpuset: allow to ... |
881 |
} |
0b2f630a2 cpusets: restruct... |
882 |
/** |
58f4790b7 cpusets: update_c... |
883 884 |
* update_cpumask - update the cpus_allowed mask of a cpuset and all tasks in it * @cs: the cpuset to consider |
fc34ac1dc kernel/cpuset.c: ... |
885 |
* @trialcs: trial cpuset |
58f4790b7 cpusets: update_c... |
886 887 |
* @buf: buffer of cpu numbers written to this cpuset */ |
645fcc9d2 cpuset: don't all... |
888 889 |
static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) |
1da177e4c Linux-2.6.12-rc2 |
890 |
{ |
58f4790b7 cpusets: update_c... |
891 |
int retval; |
1da177e4c Linux-2.6.12-rc2 |
892 |
|
5f054e31c documentation: re... |
893 |
/* top_cpuset.cpus_allowed tracks cpu_online_mask; it's read-only */ |
4c4d50f7b [PATCH] cpuset: t... |
894 895 |
if (cs == &top_cpuset) return -EACCES; |
6f7f02e78 cpusets: allow em... |
896 |
/* |
c8d9c90c7 hotplug cpu: move... |
897 |
* An empty cpus_allowed is ok only if the cpuset has no tasks. |
020958b62 cpusets: decrusti... |
898 899 900 |
* 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... |
901 |
*/ |
020958b62 cpusets: decrusti... |
902 |
if (!*buf) { |
300ed6cbb cpuset: convert c... |
903 |
cpumask_clear(trialcs->cpus_allowed); |
6f7f02e78 cpusets: allow em... |
904 |
} else { |
300ed6cbb cpuset: convert c... |
905 |
retval = cpulist_parse(buf, trialcs->cpus_allowed); |
6f7f02e78 cpusets: allow em... |
906 907 |
if (retval < 0) return retval; |
37340746a cpusets: fix bug ... |
908 |
|
5d8ba82c3 cpuset: allow wri... |
909 910 |
if (!cpumask_subset(trialcs->cpus_allowed, top_cpuset.cpus_allowed)) |
37340746a cpusets: fix bug ... |
911 |
return -EINVAL; |
6f7f02e78 cpusets: allow em... |
912 |
} |
029190c51 cpuset sched_load... |
913 |
|
8707d8b8c Fix cpusets updat... |
914 |
/* Nothing to do if the cpus didn't change */ |
300ed6cbb cpuset: convert c... |
915 |
if (cpumask_equal(cs->cpus_allowed, trialcs->cpus_allowed)) |
8707d8b8c Fix cpusets updat... |
916 |
return 0; |
58f4790b7 cpusets: update_c... |
917 |
|
a73456f37 cpuset: re-struct... |
918 919 920 |
retval = validate_change(cs, trialcs); if (retval < 0) return retval; |
8447a0fee cpuset: convert c... |
921 |
spin_lock_irq(&callback_lock); |
300ed6cbb cpuset: convert c... |
922 |
cpumask_copy(cs->cpus_allowed, trialcs->cpus_allowed); |
8447a0fee cpuset: convert c... |
923 |
spin_unlock_irq(&callback_lock); |
029190c51 cpuset sched_load... |
924 |
|
734d45130 cpuset: update cs... |
925 926 |
/* use trialcs->cpus_allowed as a temp variable */ update_cpumasks_hier(cs, trialcs->cpus_allowed); |
85d7b9498 [PATCH] Dynamic s... |
927 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
928 |
} |
053199edf [PATCH] cpusets: ... |
929 |
/* |
e93ad19d0 cpuset: make mm m... |
930 931 932 933 934 |
* Migrate memory region from one set of nodes to another. This is * performed asynchronously as it can be called from process migration path * holding locks involved in process management. All mm migrations are * performed in the queued order and can be waited for by flushing * cpuset_migrate_mm_wq. |
e4e364e86 [PATCH] cpuset: m... |
935 |
*/ |
e93ad19d0 cpuset: make mm m... |
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 |
struct cpuset_migrate_mm_work { struct work_struct work; struct mm_struct *mm; nodemask_t from; nodemask_t to; }; static void cpuset_migrate_mm_workfn(struct work_struct *work) { struct cpuset_migrate_mm_work *mwork = container_of(work, struct cpuset_migrate_mm_work, work); /* on a wq worker, no need to worry about %current's mems_allowed */ do_migrate_pages(mwork->mm, &mwork->from, &mwork->to, MPOL_MF_MOVE_ALL); mmput(mwork->mm); kfree(mwork); } |
e4e364e86 [PATCH] cpuset: m... |
953 954 955 |
static void cpuset_migrate_mm(struct mm_struct *mm, const nodemask_t *from, const nodemask_t *to) { |
e93ad19d0 cpuset: make mm m... |
956 |
struct cpuset_migrate_mm_work *mwork; |
e4e364e86 [PATCH] cpuset: m... |
957 |
|
e93ad19d0 cpuset: make mm m... |
958 959 960 961 962 963 964 965 966 967 968 |
mwork = kzalloc(sizeof(*mwork), GFP_KERNEL); if (mwork) { mwork->mm = mm; mwork->from = *from; mwork->to = *to; INIT_WORK(&mwork->work, cpuset_migrate_mm_workfn); queue_work(cpuset_migrate_mm_wq, &mwork->work); } else { mmput(mm); } } |
e4e364e86 [PATCH] cpuset: m... |
969 |
|
5cf1cacb4 cgroup, cpuset: r... |
970 |
static void cpuset_post_attach(void) |
e93ad19d0 cpuset: make mm m... |
971 972 |
{ flush_workqueue(cpuset_migrate_mm_wq); |
e4e364e86 [PATCH] cpuset: m... |
973 |
} |
3b6766fe6 cpuset: rewrite u... |
974 |
/* |
58568d2a8 cpuset,mm: update... |
975 976 977 978 979 980 981 |
* 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... |
982 983 984 985 |
*/ static void cpuset_change_task_nodemask(struct task_struct *tsk, nodemask_t *newmems) { |
b246272ec cpusets: stall wh... |
986 |
bool need_loop; |
89e8a244b cpusets: avoid lo... |
987 |
|
c0ff7453b cpuset,mm: fix no... |
988 |
task_lock(tsk); |
b246272ec cpusets: stall wh... |
989 990 |
/* * Determine if a loop is necessary if another thread is doing |
d26914d11 mm: optimize put_... |
991 |
* read_mems_allowed_begin(). If at least one node remains unchanged and |
b246272ec cpusets: stall wh... |
992 993 994 995 996 |
* 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... |
997 |
|
0fc0287c9 cpuset: Fix memor... |
998 999 |
if (need_loop) { local_irq_disable(); |
cc9a6c877 cpuset: mm: reduc... |
1000 |
write_seqcount_begin(&tsk->mems_allowed_seq); |
0fc0287c9 cpuset: Fix memor... |
1001 |
} |
c0ff7453b cpuset,mm: fix no... |
1002 |
|
cc9a6c877 cpuset: mm: reduc... |
1003 1004 |
nodes_or(tsk->mems_allowed, tsk->mems_allowed, *newmems); mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP1); |
c0ff7453b cpuset,mm: fix no... |
1005 1006 |
mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP2); |
58568d2a8 cpuset,mm: update... |
1007 |
tsk->mems_allowed = *newmems; |
cc9a6c877 cpuset: mm: reduc... |
1008 |
|
0fc0287c9 cpuset: Fix memor... |
1009 |
if (need_loop) { |
cc9a6c877 cpuset: mm: reduc... |
1010 |
write_seqcount_end(&tsk->mems_allowed_seq); |
0fc0287c9 cpuset: Fix memor... |
1011 1012 |
local_irq_enable(); } |
cc9a6c877 cpuset: mm: reduc... |
1013 |
|
c0ff7453b cpuset,mm: fix no... |
1014 |
task_unlock(tsk); |
58568d2a8 cpuset,mm: update... |
1015 |
} |
8793d854e Task Control Grou... |
1016 |
static void *cpuset_being_rebound; |
0b2f630a2 cpusets: restruct... |
1017 1018 1019 |
/** * 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... |
1020 |
* |
d66393e54 cpuset: use css_t... |
1021 1022 1023 |
* 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... |
1024 |
*/ |
d66393e54 cpuset: use css_t... |
1025 |
static void update_tasks_nodemask(struct cpuset *cs) |
1da177e4c Linux-2.6.12-rc2 |
1026 |
{ |
33ad801df cpuset: record ol... |
1027 |
static nodemask_t newmems; /* protected by cpuset_mutex */ |
d66393e54 cpuset: use css_t... |
1028 1029 |
struct css_task_iter it; struct task_struct *task; |
59dac16fb [PATCH] cpuset: u... |
1030 |
|
846a16bf0 mempolicy: rename... |
1031 |
cpuset_being_rebound = cs; /* causes mpol_dup() rebind */ |
4225399a6 [PATCH] cpuset: r... |
1032 |
|
ae1c80238 cpuset: apply cs-... |
1033 |
guarantee_online_mems(cs, &newmems); |
33ad801df cpuset: record ol... |
1034 |
|
4225399a6 [PATCH] cpuset: r... |
1035 |
/* |
3b6766fe6 cpuset: rewrite u... |
1036 1037 1038 1039 |
* 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... |
1040 |
* the global cpuset_mutex, we know that no other rebind effort |
3b6766fe6 cpuset: rewrite u... |
1041 |
* will be contending for the global variable cpuset_being_rebound. |
4225399a6 [PATCH] cpuset: r... |
1042 |
* It's ok if we rebind the same mm twice; mpol_rebind_mm() |
04c19fa6f [PATCH] cpuset: m... |
1043 |
* is idempotent. Also migrate pages in each mm to new nodes. |
4225399a6 [PATCH] cpuset: r... |
1044 |
*/ |
d66393e54 cpuset: use css_t... |
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 |
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); |
e93ad19d0 cpuset: make mm m... |
1061 1062 |
else mmput(mm); |
d66393e54 cpuset: use css_t... |
1063 1064 |
} css_task_iter_end(&it); |
4225399a6 [PATCH] cpuset: r... |
1065 |
|
33ad801df cpuset: record ol... |
1066 1067 1068 1069 1070 |
/* * All the tasks' nodemasks have been updated, update * cs->old_mems_allowed. */ cs->old_mems_allowed = newmems; |
2df167a30 cgroups: update c... |
1071 |
/* We're done rebinding vmas to this cpuset's new mems_allowed. */ |
8793d854e Task Control Grou... |
1072 |
cpuset_being_rebound = NULL; |
1da177e4c Linux-2.6.12-rc2 |
1073 |
} |
0b2f630a2 cpusets: restruct... |
1074 |
/* |
734d45130 cpuset: update cs... |
1075 1076 1077 |
* 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 ... |
1078 |
* |
734d45130 cpuset: update cs... |
1079 1080 |
* When configured nodemask is changed, the effective nodemasks of this cpuset * and all its descendants need to be updated. |
5c5cc6232 cpuset: allow to ... |
1081 |
* |
734d45130 cpuset: update cs... |
1082 |
* On legacy hiearchy, effective_mems will be the same with mems_allowed. |
5c5cc6232 cpuset: allow to ... |
1083 1084 1085 |
* * Called with cpuset_mutex held */ |
734d45130 cpuset: update cs... |
1086 |
static void update_nodemasks_hier(struct cpuset *cs, nodemask_t *new_mems) |
5c5cc6232 cpuset: allow to ... |
1087 1088 |
{ struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
1089 |
struct cgroup_subsys_state *pos_css; |
5c5cc6232 cpuset: allow to ... |
1090 1091 |
rcu_read_lock(); |
734d45130 cpuset: update cs... |
1092 1093 1094 1095 |
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... |
1096 1097 1098 1099 |
/* * If it becomes empty, inherit the effective mask of the * parent, which is guaranteed to have some MEMs. */ |
9e10a130d cgroup: replace c... |
1100 1101 |
if (cgroup_subsys_on_dfl(cpuset_cgrp_subsys) && nodes_empty(*new_mems)) |
554b0d1c8 cpuset: inherit a... |
1102 |
*new_mems = parent->effective_mems; |
734d45130 cpuset: update cs... |
1103 1104 1105 1106 |
/* 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 ... |
1107 |
} |
734d45130 cpuset: update cs... |
1108 |
|
ec903c0c8 cgroup: rename cs... |
1109 |
if (!css_tryget_online(&cp->css)) |
5c5cc6232 cpuset: allow to ... |
1110 1111 |
continue; rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
1112 |
spin_lock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
1113 |
cp->effective_mems = *new_mems; |
8447a0fee cpuset: convert c... |
1114 |
spin_unlock_irq(&callback_lock); |
734d45130 cpuset: update cs... |
1115 |
|
9e10a130d cgroup: replace c... |
1116 |
WARN_ON(!cgroup_subsys_on_dfl(cpuset_cgrp_subsys) && |
a13812683 cpuset: fix the W... |
1117 |
!nodes_equal(cp->mems_allowed, cp->effective_mems)); |
734d45130 cpuset: update cs... |
1118 |
|
d66393e54 cpuset: use css_t... |
1119 |
update_tasks_nodemask(cp); |
5c5cc6232 cpuset: allow to ... |
1120 1121 1122 1123 1124 1125 1126 1127 |
rcu_read_lock(); css_put(&cp->css); } rcu_read_unlock(); } /* |
0b2f630a2 cpusets: restruct... |
1128 1129 |
* Handle user request to change the 'mems' memory placement * of a cpuset. Needs to validate the request, update the |
58568d2a8 cpuset,mm: update... |
1130 1131 1132 1133 |
* 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... |
1134 |
* |
8447a0fee cpuset: convert c... |
1135 |
* Call with cpuset_mutex held. May take callback_lock during call. |
0b2f630a2 cpusets: restruct... |
1136 1137 1138 1139 |
* 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... |
1140 1141 |
static int update_nodemask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) |
0b2f630a2 cpusets: restruct... |
1142 |
{ |
0b2f630a2 cpusets: restruct... |
1143 1144 1145 |
int retval; /* |
38d7bee9d cpuset: use N_MEM... |
1146 |
* top_cpuset.mems_allowed tracks node_stats[N_MEMORY]; |
0b2f630a2 cpusets: restruct... |
1147 1148 |
* it's read-only */ |
53feb2976 cpuset: alloc nod... |
1149 1150 1151 1152 |
if (cs == &top_cpuset) { retval = -EACCES; goto done; } |
0b2f630a2 cpusets: restruct... |
1153 |
|
0b2f630a2 cpusets: restruct... |
1154 1155 1156 1157 1158 1159 1160 |
/* * 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... |
1161 |
nodes_clear(trialcs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1162 |
} else { |
645fcc9d2 cpuset: don't all... |
1163 |
retval = nodelist_parse(buf, trialcs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1164 1165 |
if (retval < 0) goto done; |
645fcc9d2 cpuset: don't all... |
1166 |
if (!nodes_subset(trialcs->mems_allowed, |
5d8ba82c3 cpuset: allow wri... |
1167 1168 |
top_cpuset.mems_allowed)) { retval = -EINVAL; |
53feb2976 cpuset: alloc nod... |
1169 1170 |
goto done; } |
0b2f630a2 cpusets: restruct... |
1171 |
} |
33ad801df cpuset: record ol... |
1172 1173 |
if (nodes_equal(cs->mems_allowed, trialcs->mems_allowed)) { |
0b2f630a2 cpusets: restruct... |
1174 1175 1176 |
retval = 0; /* Too easy - nothing to do */ goto done; } |
645fcc9d2 cpuset: don't all... |
1177 |
retval = validate_change(cs, trialcs); |
0b2f630a2 cpusets: restruct... |
1178 1179 |
if (retval < 0) goto done; |
8447a0fee cpuset: convert c... |
1180 |
spin_lock_irq(&callback_lock); |
645fcc9d2 cpuset: don't all... |
1181 |
cs->mems_allowed = trialcs->mems_allowed; |
8447a0fee cpuset: convert c... |
1182 |
spin_unlock_irq(&callback_lock); |
0b2f630a2 cpusets: restruct... |
1183 |
|
734d45130 cpuset: update cs... |
1184 |
/* use trialcs->mems_allowed as a temp variable */ |
24ee3cf89 cpuset: use trial... |
1185 |
update_nodemasks_hier(cs, &trialcs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1186 1187 1188 |
done: return retval; } |
8793d854e Task Control Grou... |
1189 1190 |
int current_cpuset_is_being_rebound(void) { |
391acf970 cpuset,mempolicy:... |
1191 1192 1193 1194 1195 1196 1197 |
int ret; rcu_read_lock(); ret = task_cs(current) == cpuset_being_rebound; rcu_read_unlock(); return ret; |
8793d854e Task Control Grou... |
1198 |
} |
5be7a4792 Fix cpuset sched_... |
1199 |
static int update_relax_domain_level(struct cpuset *cs, s64 val) |
1d3504fcf sched, cpuset: cu... |
1200 |
{ |
db7f47cf4 cpusets: allow cp... |
1201 |
#ifdef CONFIG_SMP |
60495e776 sched: Dynamic sc... |
1202 |
if (val < -1 || val >= sched_domain_level_max) |
30e0e1781 cpuset: limit the... |
1203 |
return -EINVAL; |
db7f47cf4 cpusets: allow cp... |
1204 |
#endif |
1d3504fcf sched, cpuset: cu... |
1205 1206 1207 |
if (val != cs->relax_domain_level) { cs->relax_domain_level = val; |
300ed6cbb cpuset: convert c... |
1208 1209 |
if (!cpumask_empty(cs->cpus_allowed) && is_sched_load_balance(cs)) |
699140ba8 cpuset: drop asyn... |
1210 |
rebuild_sched_domains_locked(); |
1d3504fcf sched, cpuset: cu... |
1211 1212 1213 1214 |
} return 0; } |
72ec70299 cgroup: make task... |
1215 |
/** |
950592f7b cpusets: update t... |
1216 1217 |
* 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... |
1218 |
* |
d66393e54 cpuset: use css_t... |
1219 1220 1221 |
* 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... |
1222 |
*/ |
d66393e54 cpuset: use css_t... |
1223 |
static void update_tasks_flags(struct cpuset *cs) |
950592f7b cpusets: update t... |
1224 |
{ |
d66393e54 cpuset: use css_t... |
1225 1226 1227 1228 1229 1230 1231 |
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... |
1232 1233 1234 |
} /* |
1da177e4c Linux-2.6.12-rc2 |
1235 |
* update_flag - read a 0 or a 1 in a file and update associated flag |
786083667 Cpuset hardwall f... |
1236 1237 1238 |
* 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: ... |
1239 |
* |
5d21cc2db cpuset: replace c... |
1240 |
* Call with cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
1241 |
*/ |
700fe1ab9 CGroup API files:... |
1242 1243 |
static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, int turning_on) |
1da177e4c Linux-2.6.12-rc2 |
1244 |
{ |
645fcc9d2 cpuset: don't all... |
1245 |
struct cpuset *trialcs; |
40b6a7623 cpuset.c: remove ... |
1246 |
int balance_flag_changed; |
950592f7b cpusets: update t... |
1247 |
int spread_flag_changed; |
950592f7b cpusets: update t... |
1248 |
int err; |
1da177e4c Linux-2.6.12-rc2 |
1249 |
|
645fcc9d2 cpuset: don't all... |
1250 1251 1252 |
trialcs = alloc_trial_cpuset(cs); if (!trialcs) return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
1253 |
if (turning_on) |
645fcc9d2 cpuset: don't all... |
1254 |
set_bit(bit, &trialcs->flags); |
1da177e4c Linux-2.6.12-rc2 |
1255 |
else |
645fcc9d2 cpuset: don't all... |
1256 |
clear_bit(bit, &trialcs->flags); |
1da177e4c Linux-2.6.12-rc2 |
1257 |
|
645fcc9d2 cpuset: don't all... |
1258 |
err = validate_change(cs, trialcs); |
85d7b9498 [PATCH] Dynamic s... |
1259 |
if (err < 0) |
645fcc9d2 cpuset: don't all... |
1260 |
goto out; |
029190c51 cpuset sched_load... |
1261 |
|
029190c51 cpuset sched_load... |
1262 |
balance_flag_changed = (is_sched_load_balance(cs) != |
645fcc9d2 cpuset: don't all... |
1263 |
is_sched_load_balance(trialcs)); |
029190c51 cpuset sched_load... |
1264 |
|
950592f7b cpusets: update t... |
1265 1266 |
spread_flag_changed = ((is_spread_slab(cs) != is_spread_slab(trialcs)) || (is_spread_page(cs) != is_spread_page(trialcs))); |
8447a0fee cpuset: convert c... |
1267 |
spin_lock_irq(&callback_lock); |
645fcc9d2 cpuset: don't all... |
1268 |
cs->flags = trialcs->flags; |
8447a0fee cpuset: convert c... |
1269 |
spin_unlock_irq(&callback_lock); |
85d7b9498 [PATCH] Dynamic s... |
1270 |
|
300ed6cbb cpuset: convert c... |
1271 |
if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed) |
699140ba8 cpuset: drop asyn... |
1272 |
rebuild_sched_domains_locked(); |
029190c51 cpuset sched_load... |
1273 |
|
950592f7b cpusets: update t... |
1274 |
if (spread_flag_changed) |
d66393e54 cpuset: use css_t... |
1275 |
update_tasks_flags(cs); |
645fcc9d2 cpuset: don't all... |
1276 1277 1278 |
out: free_trial_cpuset(trialcs); return err; |
1da177e4c Linux-2.6.12-rc2 |
1279 |
} |
053199edf [PATCH] cpusets: ... |
1280 |
/* |
80f7228b5 typo fixes: occur... |
1281 |
* Frequency meter - How fast is some event occurring? |
3e0d98b9f [PATCH] cpuset: m... |
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 |
* * 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 */ |
d2b436580 cpuset: Replace a... |
1326 |
#define FM_MAXTICKS ((u32)99) /* useless computing more ticks than this */ |
3e0d98b9f [PATCH] cpuset: m... |
1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 |
#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) { |
d2b436580 cpuset: Replace a... |
1342 1343 1344 1345 1346 |
time64_t now; u32 ticks; now = ktime_get_seconds(); ticks = now - fmp->time; |
3e0d98b9f [PATCH] cpuset: m... |
1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 |
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... |
1380 |
static struct cpuset *cpuset_attach_old_cs; |
5d21cc2db cpuset: replace c... |
1381 |
/* Called by cgroups to determine if a cpuset is usable; cpuset_mutex held */ |
1f7dd3e5a cgroup: fix handl... |
1382 |
static int cpuset_can_attach(struct cgroup_taskset *tset) |
f780bdb7c cgroups: add per-... |
1383 |
{ |
1f7dd3e5a cgroup: fix handl... |
1384 1385 |
struct cgroup_subsys_state *css; struct cpuset *cs; |
bb9d97b6d cgroup: don't use... |
1386 1387 |
struct task_struct *task; int ret; |
1da177e4c Linux-2.6.12-rc2 |
1388 |
|
57fce0a68 cpuset: don't use... |
1389 |
/* used later by cpuset_attach() */ |
1f7dd3e5a cgroup: fix handl... |
1390 1391 |
cpuset_attach_old_cs = task_cs(cgroup_taskset_first(tset, &css)); cs = css_cs(css); |
57fce0a68 cpuset: don't use... |
1392 |
|
5d21cc2db cpuset: replace c... |
1393 |
mutex_lock(&cpuset_mutex); |
aa6ec29be cgroup: remove sa... |
1394 |
/* allow moving tasks into an empty cpuset if on default hierarchy */ |
5d21cc2db cpuset: replace c... |
1395 |
ret = -ENOSPC; |
9e10a130d cgroup: replace c... |
1396 |
if (!cgroup_subsys_on_dfl(cpuset_cgrp_subsys) && |
88fa523bf cpuset: allow to ... |
1397 |
(cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed))) |
5d21cc2db cpuset: replace c... |
1398 |
goto out_unlock; |
9985b0bab sched: prevent bo... |
1399 |
|
1f7dd3e5a cgroup: fix handl... |
1400 |
cgroup_taskset_for_each(task, css, tset) { |
7f51412a4 sched/deadline: F... |
1401 1402 |
ret = task_can_attach(task, cs->cpus_allowed); if (ret) |
5d21cc2db cpuset: replace c... |
1403 1404 1405 1406 |
goto out_unlock; ret = security_task_setscheduler(task); if (ret) goto out_unlock; |
bb9d97b6d cgroup: don't use... |
1407 |
} |
f780bdb7c cgroups: add per-... |
1408 |
|
452477fa6 cpuset: pin down ... |
1409 1410 1411 1412 1413 |
/* * Mark attach is in progress. This makes validate_change() fail * changes which zero cpus/mems_allowed. */ cs->attach_in_progress++; |
5d21cc2db cpuset: replace c... |
1414 1415 1416 1417 |
ret = 0; out_unlock: mutex_unlock(&cpuset_mutex); return ret; |
8793d854e Task Control Grou... |
1418 |
} |
f780bdb7c cgroups: add per-... |
1419 |
|
1f7dd3e5a cgroup: fix handl... |
1420 |
static void cpuset_cancel_attach(struct cgroup_taskset *tset) |
452477fa6 cpuset: pin down ... |
1421 |
{ |
1f7dd3e5a cgroup: fix handl... |
1422 1423 1424 1425 1426 |
struct cgroup_subsys_state *css; struct cpuset *cs; cgroup_taskset_first(tset, &css); cs = css_cs(css); |
5d21cc2db cpuset: replace c... |
1427 |
mutex_lock(&cpuset_mutex); |
eb95419b0 cgroup: pass arou... |
1428 |
css_cs(css)->attach_in_progress--; |
5d21cc2db cpuset: replace c... |
1429 |
mutex_unlock(&cpuset_mutex); |
8793d854e Task Control Grou... |
1430 |
} |
1da177e4c Linux-2.6.12-rc2 |
1431 |
|
4e4c9a140 cpuset: cleanup c... |
1432 |
/* |
5d21cc2db cpuset: replace c... |
1433 |
* Protected by cpuset_mutex. cpus_attach is used only by cpuset_attach() |
4e4c9a140 cpuset: cleanup c... |
1434 1435 1436 1437 |
* but we can't allocate it dynamically there. Define it global and * allocate from cpuset_init(). */ static cpumask_var_t cpus_attach; |
1f7dd3e5a cgroup: fix handl... |
1438 |
static void cpuset_attach(struct cgroup_taskset *tset) |
8793d854e Task Control Grou... |
1439 |
{ |
67bd2c598 cpuset: remove un... |
1440 |
/* static buf protected by cpuset_mutex */ |
4e4c9a140 cpuset: cleanup c... |
1441 |
static nodemask_t cpuset_attach_nodemask_to; |
bb9d97b6d cgroup: don't use... |
1442 |
struct task_struct *task; |
4530eddb5 cgroup, memcg, cp... |
1443 |
struct task_struct *leader; |
1f7dd3e5a cgroup: fix handl... |
1444 1445 |
struct cgroup_subsys_state *css; struct cpuset *cs; |
57fce0a68 cpuset: don't use... |
1446 |
struct cpuset *oldcs = cpuset_attach_old_cs; |
22fb52dd7 [PATCH] SELinux: ... |
1447 |
|
1f7dd3e5a cgroup: fix handl... |
1448 1449 |
cgroup_taskset_first(tset, &css); cs = css_cs(css); |
5d21cc2db cpuset: replace c... |
1450 |
mutex_lock(&cpuset_mutex); |
4e4c9a140 cpuset: cleanup c... |
1451 1452 1453 1454 |
/* prepare for attach */ if (cs == &top_cpuset) cpumask_copy(cpus_attach, cpu_possible_mask); else |
ae1c80238 cpuset: apply cs-... |
1455 |
guarantee_online_cpus(cs, cpus_attach); |
4e4c9a140 cpuset: cleanup c... |
1456 |
|
ae1c80238 cpuset: apply cs-... |
1457 |
guarantee_online_mems(cs, &cpuset_attach_nodemask_to); |
4e4c9a140 cpuset: cleanup c... |
1458 |
|
1f7dd3e5a cgroup: fix handl... |
1459 |
cgroup_taskset_for_each(task, css, tset) { |
bb9d97b6d cgroup: don't use... |
1460 1461 1462 1463 1464 1465 1466 1467 1468 |
/* * 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: ... |
1469 |
|
f780bdb7c cgroups: add per-... |
1470 |
/* |
4530eddb5 cgroup, memcg, cp... |
1471 1472 |
* Change mm for all threadgroup leaders. This is expensive and may * sleep and should be moved outside migration path proper. |
f780bdb7c cgroups: add per-... |
1473 |
*/ |
ae1c80238 cpuset: apply cs-... |
1474 |
cpuset_attach_nodemask_to = cs->effective_mems; |
1f7dd3e5a cgroup: fix handl... |
1475 |
cgroup_taskset_for_each_leader(leader, css, tset) { |
3df9ca0a2 cpuset: migrate m... |
1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 |
struct mm_struct *mm = get_task_mm(leader); if (mm) { mpol_rebind_mm(mm, &cpuset_attach_nodemask_to); /* * 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. */ |
e93ad19d0 cpuset: make mm m... |
1489 |
if (is_memory_migrate(cs)) |
3df9ca0a2 cpuset: migrate m... |
1490 1491 |
cpuset_migrate_mm(mm, &oldcs->old_mems_allowed, &cpuset_attach_nodemask_to); |
e93ad19d0 cpuset: make mm m... |
1492 1493 |
else mmput(mm); |
f047cecf2 cpuset: fix to mi... |
1494 |
} |
4225399a6 [PATCH] cpuset: r... |
1495 |
} |
452477fa6 cpuset: pin down ... |
1496 |
|
33ad801df cpuset: record ol... |
1497 |
cs->old_mems_allowed = cpuset_attach_nodemask_to; |
02bb58637 cpuset: schedule ... |
1498 |
|
452477fa6 cpuset: pin down ... |
1499 |
cs->attach_in_progress--; |
e44193d39 cpuset: let hotpl... |
1500 1501 |
if (!cs->attach_in_progress) wake_up(&cpuset_attach_wq); |
5d21cc2db cpuset: replace c... |
1502 1503 |
mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1504 1505 1506 1507 1508 |
} /* The various types of files and directories in a cpuset file system */ typedef enum { |
45b07ef31 [PATCH] cpusets: ... |
1509 |
FILE_MEMORY_MIGRATE, |
1da177e4c Linux-2.6.12-rc2 |
1510 1511 |
FILE_CPULIST, FILE_MEMLIST, |
afd1a8b3e cpuset: export ef... |
1512 1513 |
FILE_EFFECTIVE_CPULIST, FILE_EFFECTIVE_MEMLIST, |
1da177e4c Linux-2.6.12-rc2 |
1514 1515 |
FILE_CPU_EXCLUSIVE, FILE_MEM_EXCLUSIVE, |
786083667 Cpuset hardwall f... |
1516 |
FILE_MEM_HARDWALL, |
029190c51 cpuset sched_load... |
1517 |
FILE_SCHED_LOAD_BALANCE, |
1d3504fcf sched, cpuset: cu... |
1518 |
FILE_SCHED_RELAX_DOMAIN_LEVEL, |
3e0d98b9f [PATCH] cpuset: m... |
1519 1520 |
FILE_MEMORY_PRESSURE_ENABLED, FILE_MEMORY_PRESSURE, |
825a46af5 [PATCH] cpuset me... |
1521 1522 |
FILE_SPREAD_PAGE, FILE_SPREAD_SLAB, |
1da177e4c Linux-2.6.12-rc2 |
1523 |
} cpuset_filetype_t; |
182446d08 cgroup: pass arou... |
1524 1525 |
static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, u64 val) |
700fe1ab9 CGroup API files:... |
1526 |
{ |
182446d08 cgroup: pass arou... |
1527 |
struct cpuset *cs = css_cs(css); |
700fe1ab9 CGroup API files:... |
1528 |
cpuset_filetype_t type = cft->private; |
a903f0865 cpuset: fix the r... |
1529 |
int retval = 0; |
700fe1ab9 CGroup API files:... |
1530 |
|
5d21cc2db cpuset: replace c... |
1531 |
mutex_lock(&cpuset_mutex); |
a903f0865 cpuset: fix the r... |
1532 1533 |
if (!is_cpuset_online(cs)) { retval = -ENODEV; |
5d21cc2db cpuset: replace c... |
1534 |
goto out_unlock; |
a903f0865 cpuset: fix the r... |
1535 |
} |
700fe1ab9 CGroup API files:... |
1536 1537 |
switch (type) { |
1da177e4c Linux-2.6.12-rc2 |
1538 |
case FILE_CPU_EXCLUSIVE: |
700fe1ab9 CGroup API files:... |
1539 |
retval = update_flag(CS_CPU_EXCLUSIVE, cs, val); |
1da177e4c Linux-2.6.12-rc2 |
1540 1541 |
break; case FILE_MEM_EXCLUSIVE: |
700fe1ab9 CGroup API files:... |
1542 |
retval = update_flag(CS_MEM_EXCLUSIVE, cs, val); |
1da177e4c Linux-2.6.12-rc2 |
1543 |
break; |
786083667 Cpuset hardwall f... |
1544 1545 1546 |
case FILE_MEM_HARDWALL: retval = update_flag(CS_MEM_HARDWALL, cs, val); break; |
029190c51 cpuset sched_load... |
1547 |
case FILE_SCHED_LOAD_BALANCE: |
700fe1ab9 CGroup API files:... |
1548 |
retval = update_flag(CS_SCHED_LOAD_BALANCE, cs, val); |
1d3504fcf sched, cpuset: cu... |
1549 |
break; |
45b07ef31 [PATCH] cpusets: ... |
1550 |
case FILE_MEMORY_MIGRATE: |
700fe1ab9 CGroup API files:... |
1551 |
retval = update_flag(CS_MEMORY_MIGRATE, cs, val); |
45b07ef31 [PATCH] cpusets: ... |
1552 |
break; |
3e0d98b9f [PATCH] cpuset: m... |
1553 |
case FILE_MEMORY_PRESSURE_ENABLED: |
700fe1ab9 CGroup API files:... |
1554 |
cpuset_memory_pressure_enabled = !!val; |
3e0d98b9f [PATCH] cpuset: m... |
1555 |
break; |
825a46af5 [PATCH] cpuset me... |
1556 |
case FILE_SPREAD_PAGE: |
700fe1ab9 CGroup API files:... |
1557 |
retval = update_flag(CS_SPREAD_PAGE, cs, val); |
825a46af5 [PATCH] cpuset me... |
1558 1559 |
break; case FILE_SPREAD_SLAB: |
700fe1ab9 CGroup API files:... |
1560 |
retval = update_flag(CS_SPREAD_SLAB, cs, val); |
825a46af5 [PATCH] cpuset me... |
1561 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1562 1563 |
default: retval = -EINVAL; |
700fe1ab9 CGroup API files:... |
1564 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1565 |
} |
5d21cc2db cpuset: replace c... |
1566 1567 |
out_unlock: mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1568 1569 |
return retval; } |
182446d08 cgroup: pass arou... |
1570 1571 |
static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, s64 val) |
5be7a4792 Fix cpuset sched_... |
1572 |
{ |
182446d08 cgroup: pass arou... |
1573 |
struct cpuset *cs = css_cs(css); |
5be7a4792 Fix cpuset sched_... |
1574 |
cpuset_filetype_t type = cft->private; |
5d21cc2db cpuset: replace c... |
1575 |
int retval = -ENODEV; |
5be7a4792 Fix cpuset sched_... |
1576 |
|
5d21cc2db cpuset: replace c... |
1577 1578 1579 |
mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; |
e37123953 cgroup files: rem... |
1580 |
|
5be7a4792 Fix cpuset sched_... |
1581 1582 1583 1584 1585 1586 1587 1588 |
switch (type) { case FILE_SCHED_RELAX_DOMAIN_LEVEL: retval = update_relax_domain_level(cs, val); break; default: retval = -EINVAL; break; } |
5d21cc2db cpuset: replace c... |
1589 1590 |
out_unlock: mutex_unlock(&cpuset_mutex); |
5be7a4792 Fix cpuset sched_... |
1591 1592 |
return retval; } |
1da177e4c Linux-2.6.12-rc2 |
1593 |
/* |
e37123953 cgroup files: rem... |
1594 1595 |
* Common handling for a write to a "cpus" or "mems" file. */ |
451af504d cgroup: replace c... |
1596 1597 |
static ssize_t cpuset_write_resmask(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
e37123953 cgroup files: rem... |
1598 |
{ |
451af504d cgroup: replace c... |
1599 |
struct cpuset *cs = css_cs(of_css(of)); |
645fcc9d2 cpuset: don't all... |
1600 |
struct cpuset *trialcs; |
5d21cc2db cpuset: replace c... |
1601 |
int retval = -ENODEV; |
e37123953 cgroup files: rem... |
1602 |
|
451af504d cgroup: replace c... |
1603 |
buf = strstrip(buf); |
3a5a6d0c2 cpuset: don't nes... |
1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 |
/* * 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... |
1614 1615 1616 1617 1618 1619 1620 1621 |
* * 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... |
1622 |
*/ |
76bb5ab8f cpuset: break ker... |
1623 1624 |
css_get(&cs->css); kernfs_break_active_protection(of->kn); |
3a5a6d0c2 cpuset: don't nes... |
1625 |
flush_work(&cpuset_hotplug_work); |
5d21cc2db cpuset: replace c... |
1626 1627 1628 |
mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; |
e37123953 cgroup files: rem... |
1629 |
|
645fcc9d2 cpuset: don't all... |
1630 |
trialcs = alloc_trial_cpuset(cs); |
b75f38d65 cpuset: add a mis... |
1631 1632 |
if (!trialcs) { retval = -ENOMEM; |
5d21cc2db cpuset: replace c... |
1633 |
goto out_unlock; |
b75f38d65 cpuset: add a mis... |
1634 |
} |
645fcc9d2 cpuset: don't all... |
1635 |
|
451af504d cgroup: replace c... |
1636 |
switch (of_cft(of)->private) { |
e37123953 cgroup files: rem... |
1637 |
case FILE_CPULIST: |
645fcc9d2 cpuset: don't all... |
1638 |
retval = update_cpumask(cs, trialcs, buf); |
e37123953 cgroup files: rem... |
1639 1640 |
break; case FILE_MEMLIST: |
645fcc9d2 cpuset: don't all... |
1641 |
retval = update_nodemask(cs, trialcs, buf); |
e37123953 cgroup files: rem... |
1642 1643 1644 1645 1646 |
break; default: retval = -EINVAL; break; } |
645fcc9d2 cpuset: don't all... |
1647 1648 |
free_trial_cpuset(trialcs); |
5d21cc2db cpuset: replace c... |
1649 1650 |
out_unlock: mutex_unlock(&cpuset_mutex); |
76bb5ab8f cpuset: break ker... |
1651 1652 |
kernfs_unbreak_active_protection(of->kn); css_put(&cs->css); |
e93ad19d0 cpuset: make mm m... |
1653 |
flush_workqueue(cpuset_migrate_mm_wq); |
451af504d cgroup: replace c... |
1654 |
return retval ?: nbytes; |
e37123953 cgroup files: rem... |
1655 1656 1657 |
} /* |
1da177e4c Linux-2.6.12-rc2 |
1658 1659 1660 1661 1662 1663 |
* 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 |
1664 |
*/ |
2da8ca822 cgroup: replace c... |
1665 |
static int cpuset_common_seq_show(struct seq_file *sf, void *v) |
1da177e4c Linux-2.6.12-rc2 |
1666 |
{ |
2da8ca822 cgroup: replace c... |
1667 1668 |
struct cpuset *cs = css_cs(seq_css(sf)); cpuset_filetype_t type = seq_cft(sf)->private; |
51ffe4117 cpuset: convert a... |
1669 |
int ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
1670 |
|
8447a0fee cpuset: convert c... |
1671 |
spin_lock_irq(&callback_lock); |
1da177e4c Linux-2.6.12-rc2 |
1672 1673 1674 |
switch (type) { case FILE_CPULIST: |
e8e6d97c9 cpuset: use %*pb[... |
1675 1676 |
seq_printf(sf, "%*pbl ", cpumask_pr_args(cs->cpus_allowed)); |
1da177e4c Linux-2.6.12-rc2 |
1677 1678 |
break; case FILE_MEMLIST: |
e8e6d97c9 cpuset: use %*pb[... |
1679 1680 |
seq_printf(sf, "%*pbl ", nodemask_pr_args(&cs->mems_allowed)); |
1da177e4c Linux-2.6.12-rc2 |
1681 |
break; |
afd1a8b3e cpuset: export ef... |
1682 |
case FILE_EFFECTIVE_CPULIST: |
e8e6d97c9 cpuset: use %*pb[... |
1683 1684 |
seq_printf(sf, "%*pbl ", cpumask_pr_args(cs->effective_cpus)); |
afd1a8b3e cpuset: export ef... |
1685 1686 |
break; case FILE_EFFECTIVE_MEMLIST: |
e8e6d97c9 cpuset: use %*pb[... |
1687 1688 |
seq_printf(sf, "%*pbl ", nodemask_pr_args(&cs->effective_mems)); |
afd1a8b3e cpuset: export ef... |
1689 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1690 |
default: |
51ffe4117 cpuset: convert a... |
1691 |
ret = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1692 |
} |
1da177e4c Linux-2.6.12-rc2 |
1693 |
|
8447a0fee cpuset: convert c... |
1694 |
spin_unlock_irq(&callback_lock); |
51ffe4117 cpuset: convert a... |
1695 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1696 |
} |
182446d08 cgroup: pass arou... |
1697 |
static u64 cpuset_read_u64(struct cgroup_subsys_state *css, struct cftype *cft) |
700fe1ab9 CGroup API files:... |
1698 |
{ |
182446d08 cgroup: pass arou... |
1699 |
struct cpuset *cs = css_cs(css); |
700fe1ab9 CGroup API files:... |
1700 1701 1702 1703 1704 1705 |
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... |
1706 1707 |
case FILE_MEM_HARDWALL: return is_mem_hardwall(cs); |
700fe1ab9 CGroup API files:... |
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 |
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... |
1723 1724 1725 |
/* Unreachable but makes gcc happy */ return 0; |
700fe1ab9 CGroup API files:... |
1726 |
} |
1da177e4c Linux-2.6.12-rc2 |
1727 |
|
182446d08 cgroup: pass arou... |
1728 |
static s64 cpuset_read_s64(struct cgroup_subsys_state *css, struct cftype *cft) |
5be7a4792 Fix cpuset sched_... |
1729 |
{ |
182446d08 cgroup: pass arou... |
1730 |
struct cpuset *cs = css_cs(css); |
5be7a4792 Fix cpuset sched_... |
1731 1732 1733 1734 1735 1736 1737 |
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... |
1738 1739 1740 |
/* Unrechable but makes gcc happy */ return 0; |
5be7a4792 Fix cpuset sched_... |
1741 |
} |
1da177e4c Linux-2.6.12-rc2 |
1742 1743 1744 1745 |
/* * for the common functions, 'private' gives the type of file */ |
addf2c739 Cpuset hardwall f... |
1746 1747 1748 |
static struct cftype files[] = { { .name = "cpus", |
2da8ca822 cgroup: replace c... |
1749 |
.seq_show = cpuset_common_seq_show, |
451af504d cgroup: replace c... |
1750 |
.write = cpuset_write_resmask, |
e37123953 cgroup files: rem... |
1751 |
.max_write_len = (100U + 6 * NR_CPUS), |
addf2c739 Cpuset hardwall f... |
1752 1753 1754 1755 1756 |
.private = FILE_CPULIST, }, { .name = "mems", |
2da8ca822 cgroup: replace c... |
1757 |
.seq_show = cpuset_common_seq_show, |
451af504d cgroup: replace c... |
1758 |
.write = cpuset_write_resmask, |
e37123953 cgroup files: rem... |
1759 |
.max_write_len = (100U + 6 * MAX_NUMNODES), |
addf2c739 Cpuset hardwall f... |
1760 1761 1762 1763 |
.private = FILE_MEMLIST, }, { |
afd1a8b3e cpuset: export ef... |
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 |
.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... |
1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 |
.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... |
1790 1791 1792 1793 1794 1795 1796 |
.name = "mem_hardwall", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_MEM_HARDWALL, }, { |
addf2c739 Cpuset hardwall f... |
1797 1798 1799 1800 1801 1802 1803 1804 |
.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_... |
1805 1806 |
.read_s64 = cpuset_read_s64, .write_s64 = cpuset_write_s64, |
addf2c739 Cpuset hardwall f... |
1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 |
.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, |
addf2c739 Cpuset hardwall f... |
1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 |
}, { .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... |
1835 |
|
4baf6e332 cgroup: convert a... |
1836 1837 1838 1839 1840 1841 1842 |
{ .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 |
1843 |
|
4baf6e332 cgroup: convert a... |
1844 1845 |
{ } /* terminate */ }; |
1da177e4c Linux-2.6.12-rc2 |
1846 1847 |
/* |
92fb97487 cgroup: rename ->... |
1848 |
* cpuset_css_alloc - allocate a cpuset css |
c9e5fe66f cpuset: rename @c... |
1849 |
* cgrp: control group that the new cpuset will be part of |
1da177e4c Linux-2.6.12-rc2 |
1850 |
*/ |
eb95419b0 cgroup: pass arou... |
1851 1852 |
static struct cgroup_subsys_state * cpuset_css_alloc(struct cgroup_subsys_state *parent_css) |
1da177e4c Linux-2.6.12-rc2 |
1853 |
{ |
c8f699bb5 cpuset: introduce... |
1854 |
struct cpuset *cs; |
1da177e4c Linux-2.6.12-rc2 |
1855 |
|
eb95419b0 cgroup: pass arou... |
1856 |
if (!parent_css) |
8793d854e Task Control Grou... |
1857 |
return &top_cpuset.css; |
033fa1c5f cgroup, cpuset: r... |
1858 |
|
c8f699bb5 cpuset: introduce... |
1859 |
cs = kzalloc(sizeof(*cs), GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
1860 |
if (!cs) |
8793d854e Task Control Grou... |
1861 |
return ERR_PTR(-ENOMEM); |
e2b9a3d7d cpuset: add cs->e... |
1862 1863 1864 1865 |
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 |
1866 |
|
029190c51 cpuset sched_load... |
1867 |
set_bit(CS_SCHED_LOAD_BALANCE, &cs->flags); |
300ed6cbb cpuset: convert c... |
1868 |
cpumask_clear(cs->cpus_allowed); |
f9a86fcbb cpuset: modify cp... |
1869 |
nodes_clear(cs->mems_allowed); |
e2b9a3d7d cpuset: add cs->e... |
1870 1871 |
cpumask_clear(cs->effective_cpus); nodes_clear(cs->effective_mems); |
3e0d98b9f [PATCH] cpuset: m... |
1872 |
fmeter_init(&cs->fmeter); |
1d3504fcf sched, cpuset: cu... |
1873 |
cs->relax_domain_level = -1; |
1da177e4c Linux-2.6.12-rc2 |
1874 |
|
c8f699bb5 cpuset: introduce... |
1875 |
return &cs->css; |
e2b9a3d7d cpuset: add cs->e... |
1876 1877 1878 1879 1880 1881 |
free_cpus: free_cpumask_var(cs->cpus_allowed); free_cs: kfree(cs); return ERR_PTR(-ENOMEM); |
c8f699bb5 cpuset: introduce... |
1882 |
} |
eb95419b0 cgroup: pass arou... |
1883 |
static int cpuset_css_online(struct cgroup_subsys_state *css) |
c8f699bb5 cpuset: introduce... |
1884 |
{ |
eb95419b0 cgroup: pass arou... |
1885 |
struct cpuset *cs = css_cs(css); |
c431069fe cpuset: remove cp... |
1886 |
struct cpuset *parent = parent_cs(cs); |
ae8086ce1 cpuset: introduce... |
1887 |
struct cpuset *tmp_cs; |
492eb21b9 cgroup: make hier... |
1888 |
struct cgroup_subsys_state *pos_css; |
c8f699bb5 cpuset: introduce... |
1889 1890 1891 |
if (!parent) return 0; |
5d21cc2db cpuset: replace c... |
1892 |
mutex_lock(&cpuset_mutex); |
efeb77b2f cpuset: introduce... |
1893 |
set_bit(CS_ONLINE, &cs->flags); |
c8f699bb5 cpuset: introduce... |
1894 1895 1896 1897 |
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 |
1898 |
|
664eeddee mm: page_alloc: u... |
1899 |
cpuset_inc(); |
033fa1c5f cgroup, cpuset: r... |
1900 |
|
8447a0fee cpuset: convert c... |
1901 |
spin_lock_irq(&callback_lock); |
9e10a130d cgroup: replace c... |
1902 |
if (cgroup_subsys_on_dfl(cpuset_cgrp_subsys)) { |
e2b9a3d7d cpuset: add cs->e... |
1903 1904 1905 |
cpumask_copy(cs->effective_cpus, parent->effective_cpus); cs->effective_mems = parent->effective_mems; } |
8447a0fee cpuset: convert c... |
1906 |
spin_unlock_irq(&callback_lock); |
e2b9a3d7d cpuset: add cs->e... |
1907 |
|
eb95419b0 cgroup: pass arou... |
1908 |
if (!test_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags)) |
5d21cc2db cpuset: replace c... |
1909 |
goto out_unlock; |
033fa1c5f cgroup, cpuset: r... |
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 |
/* * 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... |
1924 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
1925 |
cpuset_for_each_child(tmp_cs, pos_css, parent) { |
ae8086ce1 cpuset: introduce... |
1926 1927 |
if (is_mem_exclusive(tmp_cs) || is_cpu_exclusive(tmp_cs)) { rcu_read_unlock(); |
5d21cc2db cpuset: replace c... |
1928 |
goto out_unlock; |
ae8086ce1 cpuset: introduce... |
1929 |
} |
033fa1c5f cgroup, cpuset: r... |
1930 |
} |
ae8086ce1 cpuset: introduce... |
1931 |
rcu_read_unlock(); |
033fa1c5f cgroup, cpuset: r... |
1932 |
|
8447a0fee cpuset: convert c... |
1933 |
spin_lock_irq(&callback_lock); |
033fa1c5f cgroup, cpuset: r... |
1934 |
cs->mems_allowed = parent->mems_allowed; |
790317e1b cpuset: initializ... |
1935 |
cs->effective_mems = parent->mems_allowed; |
033fa1c5f cgroup, cpuset: r... |
1936 |
cpumask_copy(cs->cpus_allowed, parent->cpus_allowed); |
790317e1b cpuset: initializ... |
1937 |
cpumask_copy(cs->effective_cpus, parent->cpus_allowed); |
cea74465e cpuset: lock vs u... |
1938 |
spin_unlock_irq(&callback_lock); |
5d21cc2db cpuset: replace c... |
1939 1940 |
out_unlock: mutex_unlock(&cpuset_mutex); |
c8f699bb5 cpuset: introduce... |
1941 1942 |
return 0; } |
0b9e6965a cpuset: relocate ... |
1943 1944 1945 1946 1947 |
/* * 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... |
1948 |
static void cpuset_css_offline(struct cgroup_subsys_state *css) |
c8f699bb5 cpuset: introduce... |
1949 |
{ |
eb95419b0 cgroup: pass arou... |
1950 |
struct cpuset *cs = css_cs(css); |
c8f699bb5 cpuset: introduce... |
1951 |
|
5d21cc2db cpuset: replace c... |
1952 |
mutex_lock(&cpuset_mutex); |
c8f699bb5 cpuset: introduce... |
1953 1954 1955 |
if (is_sched_load_balance(cs)) update_flag(CS_SCHED_LOAD_BALANCE, cs, 0); |
664eeddee mm: page_alloc: u... |
1956 |
cpuset_dec(); |
efeb77b2f cpuset: introduce... |
1957 |
clear_bit(CS_ONLINE, &cs->flags); |
c8f699bb5 cpuset: introduce... |
1958 |
|
5d21cc2db cpuset: replace c... |
1959 |
mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1960 |
} |
eb95419b0 cgroup: pass arou... |
1961 |
static void cpuset_css_free(struct cgroup_subsys_state *css) |
1da177e4c Linux-2.6.12-rc2 |
1962 |
{ |
eb95419b0 cgroup: pass arou... |
1963 |
struct cpuset *cs = css_cs(css); |
1da177e4c Linux-2.6.12-rc2 |
1964 |
|
e2b9a3d7d cpuset: add cs->e... |
1965 |
free_cpumask_var(cs->effective_cpus); |
300ed6cbb cpuset: convert c... |
1966 |
free_cpumask_var(cs->cpus_allowed); |
8793d854e Task Control Grou... |
1967 |
kfree(cs); |
1da177e4c Linux-2.6.12-rc2 |
1968 |
} |
39bd0d15e cpuset: initializ... |
1969 1970 1971 |
static void cpuset_bind(struct cgroup_subsys_state *root_css) { mutex_lock(&cpuset_mutex); |
8447a0fee cpuset: convert c... |
1972 |
spin_lock_irq(&callback_lock); |
39bd0d15e cpuset: initializ... |
1973 |
|
9e10a130d cgroup: replace c... |
1974 |
if (cgroup_subsys_on_dfl(cpuset_cgrp_subsys)) { |
39bd0d15e cpuset: initializ... |
1975 1976 1977 1978 1979 1980 1981 |
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... |
1982 |
spin_unlock_irq(&callback_lock); |
39bd0d15e cpuset: initializ... |
1983 1984 |
mutex_unlock(&cpuset_mutex); } |
06f4e9489 cpuset: make sure... |
1985 1986 1987 1988 1989 |
/* * Make sure the new task conform to the current state of its parent, * which could have been changed by cpuset just after it inherits the * state from the parent and before it sits on the cgroup's task list. */ |
8a15b8174 cpuset: fix non s... |
1990 |
static void cpuset_fork(struct task_struct *task) |
06f4e9489 cpuset: make sure... |
1991 1992 1993 1994 1995 1996 1997 |
{ if (task_css_is_root(task, cpuset_cgrp_id)) return; set_cpus_allowed_ptr(task, ¤t->cpus_allowed); task->mems_allowed = current->mems_allowed; } |
073219e99 cgroup: clean up ... |
1998 |
struct cgroup_subsys cpuset_cgrp_subsys = { |
39bd0d15e cpuset: initializ... |
1999 2000 2001 2002 2003 2004 2005 |
.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, |
5cf1cacb4 cgroup, cpuset: r... |
2006 |
.post_attach = cpuset_post_attach, |
39bd0d15e cpuset: initializ... |
2007 |
.bind = cpuset_bind, |
06f4e9489 cpuset: make sure... |
2008 |
.fork = cpuset_fork, |
5577964e6 cgroup: rename cg... |
2009 |
.legacy_cftypes = files, |
b38e42e96 cgroup: convert c... |
2010 |
.early_init = true, |
8793d854e Task Control Grou... |
2011 |
}; |
1da177e4c Linux-2.6.12-rc2 |
2012 2013 2014 2015 2016 2017 2018 2019 |
/** * 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... |
2020 |
int err = 0; |
1da177e4c Linux-2.6.12-rc2 |
2021 |
|
58568d2a8 cpuset,mm: update... |
2022 2023 |
if (!alloc_cpumask_var(&top_cpuset.cpus_allowed, GFP_KERNEL)) BUG(); |
e2b9a3d7d cpuset: add cs->e... |
2024 2025 |
if (!alloc_cpumask_var(&top_cpuset.effective_cpus, GFP_KERNEL)) BUG(); |
58568d2a8 cpuset,mm: update... |
2026 |
|
300ed6cbb cpuset: convert c... |
2027 |
cpumask_setall(top_cpuset.cpus_allowed); |
f9a86fcbb cpuset: modify cp... |
2028 |
nodes_setall(top_cpuset.mems_allowed); |
e2b9a3d7d cpuset: add cs->e... |
2029 2030 |
cpumask_setall(top_cpuset.effective_cpus); nodes_setall(top_cpuset.effective_mems); |
1da177e4c Linux-2.6.12-rc2 |
2031 |
|
3e0d98b9f [PATCH] cpuset: m... |
2032 |
fmeter_init(&top_cpuset.fmeter); |
029190c51 cpuset sched_load... |
2033 |
set_bit(CS_SCHED_LOAD_BALANCE, &top_cpuset.flags); |
1d3504fcf sched, cpuset: cu... |
2034 |
top_cpuset.relax_domain_level = -1; |
1da177e4c Linux-2.6.12-rc2 |
2035 |
|
1da177e4c Linux-2.6.12-rc2 |
2036 2037 |
err = register_filesystem(&cpuset_fs_type); if (err < 0) |
8793d854e Task Control Grou... |
2038 |
return err; |
2341d1b65 cpuset: convert c... |
2039 2040 |
if (!alloc_cpumask_var(&cpus_attach, GFP_KERNEL)) BUG(); |
8793d854e Task Control Grou... |
2041 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
2042 |
} |
b1aac8bb8 [PATCH] cpuset: h... |
2043 |
/* |
cf417141c sched, cpuset: re... |
2044 |
* If CPU and/or memory hotplug handlers, below, unplug any CPUs |
b1aac8bb8 [PATCH] cpuset: h... |
2045 2046 |
* 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... |
2047 2048 |
* 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... |
2049 |
*/ |
956db3ca0 hotplug cpu: move... |
2050 2051 2052 |
static void remove_tasks_in_empty_cpuset(struct cpuset *cs) { struct cpuset *parent; |
c8d9c90c7 hotplug cpu: move... |
2053 |
/* |
956db3ca0 hotplug cpu: move... |
2054 2055 2056 |
* Find its next-highest non-empty parent, (top cpuset * has online cpus, so can't be empty). */ |
c431069fe cpuset: remove cp... |
2057 |
parent = parent_cs(cs); |
300ed6cbb cpuset: convert c... |
2058 |
while (cpumask_empty(parent->cpus_allowed) || |
b45012955 hotplug cpu move ... |
2059 |
nodes_empty(parent->mems_allowed)) |
c431069fe cpuset: remove cp... |
2060 |
parent = parent_cs(parent); |
956db3ca0 hotplug cpu: move... |
2061 |
|
8cc993452 cgroup, cpuset: r... |
2062 |
if (cgroup_transfer_tasks(parent->css.cgroup, cs->css.cgroup)) { |
12d3089c1 kernel/cpuset.c: ... |
2063 |
pr_err("cpuset: failed to transfer tasks out of empty cpuset "); |
e61734c55 cgroup: remove cg... |
2064 2065 2066 |
pr_cont_cgroup_name(cs->css.cgroup); pr_cont(" "); |
8cc993452 cgroup, cpuset: r... |
2067 |
} |
956db3ca0 hotplug cpu: move... |
2068 |
} |
be4c9dd7a cpuset: enable on... |
2069 2070 2071 2072 |
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 ... |
2073 2074 |
{ bool is_empty; |
8447a0fee cpuset: convert c... |
2075 |
spin_lock_irq(&callback_lock); |
be4c9dd7a cpuset: enable on... |
2076 2077 2078 2079 |
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... |
2080 |
spin_unlock_irq(&callback_lock); |
390a36aad cpuset: refactor ... |
2081 2082 2083 2084 2085 |
/* * Don't call update_tasks_cpumask() if the cpuset becomes empty, * as the tasks will be migratecd to an ancestor. */ |
be4c9dd7a cpuset: enable on... |
2086 |
if (cpus_updated && !cpumask_empty(cs->cpus_allowed)) |
390a36aad cpuset: refactor ... |
2087 |
update_tasks_cpumask(cs); |
be4c9dd7a cpuset: enable on... |
2088 |
if (mems_updated && !nodes_empty(cs->mems_allowed)) |
390a36aad cpuset: refactor ... |
2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 |
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... |
2106 2107 2108 2109 |
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 ... |
2110 |
{ |
be4c9dd7a cpuset: enable on... |
2111 2112 2113 2114 |
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... |
2115 |
spin_lock_irq(&callback_lock); |
be4c9dd7a cpuset: enable on... |
2116 2117 |
cpumask_copy(cs->effective_cpus, new_cpus); cs->effective_mems = *new_mems; |
8447a0fee cpuset: convert c... |
2118 |
spin_unlock_irq(&callback_lock); |
390a36aad cpuset: refactor ... |
2119 |
|
be4c9dd7a cpuset: enable on... |
2120 |
if (cpus_updated) |
390a36aad cpuset: refactor ... |
2121 |
update_tasks_cpumask(cs); |
be4c9dd7a cpuset: enable on... |
2122 |
if (mems_updated) |
390a36aad cpuset: refactor ... |
2123 2124 |
update_tasks_nodemask(cs); } |
deb7aa308 cpuset: reorganiz... |
2125 |
/** |
388afd854 cpuset: remove as... |
2126 |
* cpuset_hotplug_update_tasks - update tasks in a cpuset for hotunplug |
deb7aa308 cpuset: reorganiz... |
2127 |
* @cs: cpuset in interest |
956db3ca0 hotplug cpu: move... |
2128 |
* |
deb7aa308 cpuset: reorganiz... |
2129 2130 2131 |
* 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:... |
2132 |
*/ |
388afd854 cpuset: remove as... |
2133 |
static void cpuset_hotplug_update_tasks(struct cpuset *cs) |
80d1fa646 cpusets, hotplug:... |
2134 |
{ |
be4c9dd7a cpuset: enable on... |
2135 2136 2137 2138 |
static cpumask_t new_cpus; static nodemask_t new_mems; bool cpus_updated; bool mems_updated; |
e44193d39 cpuset: let hotpl... |
2139 2140 |
retry: wait_event(cpuset_attach_wq, cs->attach_in_progress == 0); |
80d1fa646 cpusets, hotplug:... |
2141 |
|
5d21cc2db cpuset: replace c... |
2142 |
mutex_lock(&cpuset_mutex); |
7ddf96b02 cpusets, hotplug:... |
2143 |
|
e44193d39 cpuset: let hotpl... |
2144 2145 2146 2147 2148 2149 2150 2151 |
/* * 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... |
2152 2153 |
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:... |
2154 |
|
be4c9dd7a cpuset: enable on... |
2155 2156 |
cpus_updated = !cpumask_equal(&new_cpus, cs->effective_cpus); mems_updated = !nodes_equal(new_mems, cs->effective_mems); |
deb7aa308 cpuset: reorganiz... |
2157 |
|
9e10a130d cgroup: replace c... |
2158 |
if (cgroup_subsys_on_dfl(cpuset_cgrp_subsys)) |
be4c9dd7a cpuset: enable on... |
2159 2160 |
hotplug_update_tasks(cs, &new_cpus, &new_mems, cpus_updated, mems_updated); |
390a36aad cpuset: refactor ... |
2161 |
else |
be4c9dd7a cpuset: enable on... |
2162 2163 |
hotplug_update_tasks_legacy(cs, &new_cpus, &new_mems, cpus_updated, mems_updated); |
8d0339487 cpuset: make CPU ... |
2164 |
|
5d21cc2db cpuset: replace c... |
2165 |
mutex_unlock(&cpuset_mutex); |
b1aac8bb8 [PATCH] cpuset: h... |
2166 |
} |
deb7aa308 cpuset: reorganiz... |
2167 |
/** |
3a5a6d0c2 cpuset: don't nes... |
2168 |
* cpuset_hotplug_workfn - handle CPU/memory hotunplug for a cpuset |
956db3ca0 hotplug cpu: move... |
2169 |
* |
deb7aa308 cpuset: reorganiz... |
2170 2171 2172 2173 2174 |
* 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... |
2175 |
* |
deb7aa308 cpuset: reorganiz... |
2176 |
* Non-root cpusets are only affected by offlining. If any CPUs or memory |
388afd854 cpuset: remove as... |
2177 2178 |
* nodes have been taken down, cpuset_hotplug_update_tasks() is invoked on * all descendants. |
956db3ca0 hotplug cpu: move... |
2179 |
* |
deb7aa308 cpuset: reorganiz... |
2180 2181 |
* Note that CPU offlining during suspend is ignored. We don't modify * cpusets across suspend/resume cycles at all. |
956db3ca0 hotplug cpu: move... |
2182 |
*/ |
3a5a6d0c2 cpuset: don't nes... |
2183 |
static void cpuset_hotplug_workfn(struct work_struct *work) |
b1aac8bb8 [PATCH] cpuset: h... |
2184 |
{ |
5c5cc6232 cpuset: allow to ... |
2185 2186 |
static cpumask_t new_cpus; static nodemask_t new_mems; |
deb7aa308 cpuset: reorganiz... |
2187 |
bool cpus_updated, mems_updated; |
9e10a130d cgroup: replace c... |
2188 |
bool on_dfl = cgroup_subsys_on_dfl(cpuset_cgrp_subsys); |
b1aac8bb8 [PATCH] cpuset: h... |
2189 |
|
5d21cc2db cpuset: replace c... |
2190 |
mutex_lock(&cpuset_mutex); |
956db3ca0 hotplug cpu: move... |
2191 |
|
deb7aa308 cpuset: reorganiz... |
2192 2193 2194 |
/* 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:... |
2195 |
|
7e88291be cpuset: make cs->... |
2196 2197 |
cpus_updated = !cpumask_equal(top_cpuset.effective_cpus, &new_cpus); mems_updated = !nodes_equal(top_cpuset.effective_mems, new_mems); |
7ddf96b02 cpusets, hotplug:... |
2198 |
|
deb7aa308 cpuset: reorganiz... |
2199 2200 |
/* synchronize cpus_allowed to cpu_active_mask */ if (cpus_updated) { |
8447a0fee cpuset: convert c... |
2201 |
spin_lock_irq(&callback_lock); |
7e88291be cpuset: make cs->... |
2202 2203 |
if (!on_dfl) cpumask_copy(top_cpuset.cpus_allowed, &new_cpus); |
1344ab9c2 cpuset: update cp... |
2204 |
cpumask_copy(top_cpuset.effective_cpus, &new_cpus); |
8447a0fee cpuset: convert c... |
2205 |
spin_unlock_irq(&callback_lock); |
deb7aa308 cpuset: reorganiz... |
2206 2207 |
/* we don't mess with cpumasks of tasks in top_cpuset */ } |
b45012955 hotplug cpu move ... |
2208 |
|
deb7aa308 cpuset: reorganiz... |
2209 2210 |
/* synchronize mems_allowed to N_MEMORY */ if (mems_updated) { |
8447a0fee cpuset: convert c... |
2211 |
spin_lock_irq(&callback_lock); |
7e88291be cpuset: make cs->... |
2212 2213 |
if (!on_dfl) top_cpuset.mems_allowed = new_mems; |
1344ab9c2 cpuset: update cp... |
2214 |
top_cpuset.effective_mems = new_mems; |
8447a0fee cpuset: convert c... |
2215 |
spin_unlock_irq(&callback_lock); |
d66393e54 cpuset: use css_t... |
2216 |
update_tasks_nodemask(&top_cpuset); |
deb7aa308 cpuset: reorganiz... |
2217 |
} |
b45012955 hotplug cpu move ... |
2218 |
|
388afd854 cpuset: remove as... |
2219 |
mutex_unlock(&cpuset_mutex); |
5c5cc6232 cpuset: allow to ... |
2220 2221 |
/* if cpus or mems changed, we need to propagate to descendants */ if (cpus_updated || mems_updated) { |
deb7aa308 cpuset: reorganiz... |
2222 |
struct cpuset *cs; |
492eb21b9 cgroup: make hier... |
2223 |
struct cgroup_subsys_state *pos_css; |
f9b4fb8da cpusets: update t... |
2224 |
|
fc560a26a cpuset: replace c... |
2225 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
2226 |
cpuset_for_each_descendant_pre(cs, pos_css, &top_cpuset) { |
ec903c0c8 cgroup: rename cs... |
2227 |
if (cs == &top_cpuset || !css_tryget_online(&cs->css)) |
388afd854 cpuset: remove as... |
2228 2229 |
continue; rcu_read_unlock(); |
7ddf96b02 cpusets, hotplug:... |
2230 |
|
388afd854 cpuset: remove as... |
2231 |
cpuset_hotplug_update_tasks(cs); |
b45012955 hotplug cpu move ... |
2232 |
|
388afd854 cpuset: remove as... |
2233 2234 2235 2236 2237 |
rcu_read_lock(); css_put(&cs->css); } rcu_read_unlock(); } |
8d0339487 cpuset: make CPU ... |
2238 |
|
deb7aa308 cpuset: reorganiz... |
2239 |
/* rebuild sched domains if cpus_allowed has changed */ |
e0e80a02e cpuset: use rebui... |
2240 2241 |
if (cpus_updated) rebuild_sched_domains(); |
b1aac8bb8 [PATCH] cpuset: h... |
2242 |
} |
7ddf96b02 cpusets, hotplug:... |
2243 |
void cpuset_update_active_cpus(bool cpu_online) |
4c4d50f7b [PATCH] cpuset: t... |
2244 |
{ |
3a5a6d0c2 cpuset: don't nes... |
2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 |
/* * 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... |
2257 |
} |
4c4d50f7b [PATCH] cpuset: t... |
2258 |
|
38837fc75 [PATCH] cpuset: t... |
2259 |
/* |
38d7bee9d cpuset: use N_MEM... |
2260 2261 |
* Keep top_cpuset.mems_allowed tracking node_states[N_MEMORY]. * Call this routine anytime after node_states[N_MEMORY] changes. |
a1cd2b13f cpusets: Remove/u... |
2262 |
* See cpuset_update_active_cpus() for CPU hotplug handling. |
38837fc75 [PATCH] cpuset: t... |
2263 |
*/ |
f481891fd cpuset: update to... |
2264 2265 |
static int cpuset_track_online_nodes(struct notifier_block *self, unsigned long action, void *arg) |
38837fc75 [PATCH] cpuset: t... |
2266 |
{ |
3a5a6d0c2 cpuset: don't nes... |
2267 |
schedule_work(&cpuset_hotplug_work); |
f481891fd cpuset: update to... |
2268 |
return NOTIFY_OK; |
38837fc75 [PATCH] cpuset: t... |
2269 |
} |
d8f10cb3d kernel/cpuset.c: ... |
2270 2271 2272 2273 2274 |
static struct notifier_block cpuset_track_online_nodes_nb = { .notifier_call = cpuset_track_online_nodes, .priority = 10, /* ??! */ }; |
38837fc75 [PATCH] cpuset: t... |
2275 |
|
1da177e4c Linux-2.6.12-rc2 |
2276 2277 2278 2279 |
/** * cpuset_init_smp - initialize cpus_allowed * * Description: Finish top cpuset after cpu, node maps are initialized |
d8f10cb3d kernel/cpuset.c: ... |
2280 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
2281 2282 |
void __init cpuset_init_smp(void) { |
6ad4c1888 sched: Fix balanc... |
2283 |
cpumask_copy(top_cpuset.cpus_allowed, cpu_active_mask); |
38d7bee9d cpuset: use N_MEM... |
2284 |
top_cpuset.mems_allowed = node_states[N_MEMORY]; |
33ad801df cpuset: record ol... |
2285 |
top_cpuset.old_mems_allowed = top_cpuset.mems_allowed; |
4c4d50f7b [PATCH] cpuset: t... |
2286 |
|
e2b9a3d7d cpuset: add cs->e... |
2287 2288 |
cpumask_copy(top_cpuset.effective_cpus, cpu_active_mask); top_cpuset.effective_mems = node_states[N_MEMORY]; |
d8f10cb3d kernel/cpuset.c: ... |
2289 |
register_hotmemory_notifier(&cpuset_track_online_nodes_nb); |
e93ad19d0 cpuset: make mm m... |
2290 2291 2292 |
cpuset_migrate_mm_wq = alloc_ordered_workqueue("cpuset_migrate_mm", 0); BUG_ON(!cpuset_migrate_mm_wq); |
1da177e4c Linux-2.6.12-rc2 |
2293 2294 2295 |
} /** |
1da177e4c Linux-2.6.12-rc2 |
2296 2297 |
* 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... |
2298 |
* @pmask: pointer to struct cpumask variable to receive cpus_allowed set. |
1da177e4c Linux-2.6.12-rc2 |
2299 |
* |
300ed6cbb cpuset: convert c... |
2300 |
* Description: Returns the cpumask_var_t cpus_allowed of the cpuset |
1da177e4c Linux-2.6.12-rc2 |
2301 |
* attached to the specified @tsk. Guaranteed to return some non-empty |
5f054e31c documentation: re... |
2302 |
* subset of cpu_online_mask, even if this means going outside the |
1da177e4c Linux-2.6.12-rc2 |
2303 2304 |
* tasks cpuset. **/ |
6af866af3 cpuset: remove re... |
2305 |
void cpuset_cpus_allowed(struct task_struct *tsk, struct cpumask *pmask) |
1da177e4c Linux-2.6.12-rc2 |
2306 |
{ |
8447a0fee cpuset: convert c... |
2307 2308 2309 |
unsigned long flags; spin_lock_irqsave(&callback_lock, flags); |
b8dadcb58 cpuset: use rcu_r... |
2310 |
rcu_read_lock(); |
ae1c80238 cpuset: apply cs-... |
2311 |
guarantee_online_cpus(task_cs(tsk), pmask); |
b8dadcb58 cpuset: use rcu_r... |
2312 |
rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
2313 |
spin_unlock_irqrestore(&callback_lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
2314 |
} |
2baab4e90 sched: Fix select... |
2315 |
void cpuset_cpus_allowed_fallback(struct task_struct *tsk) |
9084bb824 sched: Make selec... |
2316 |
{ |
9084bb824 sched: Make selec... |
2317 |
rcu_read_lock(); |
ae1c80238 cpuset: apply cs-... |
2318 |
do_set_cpus_allowed(tsk, task_cs(tsk)->effective_cpus); |
9084bb824 sched: Make selec... |
2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 |
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... |
2334 2335 2336 |
* * select_fallback_rq() will fix things ups and set cpu_possible_mask * if required. |
9084bb824 sched: Make selec... |
2337 |
*/ |
9084bb824 sched: Make selec... |
2338 |
} |
8f4ab07f4 kernel/cpuset.c: ... |
2339 |
void __init cpuset_init_current_mems_allowed(void) |
1da177e4c Linux-2.6.12-rc2 |
2340 |
{ |
f9a86fcbb cpuset: modify cp... |
2341 |
nodes_setall(current->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
2342 |
} |
d9fd8a6d4 [PATCH] kernel/cp... |
2343 |
/** |
909d75a3b [PATCH] cpuset: i... |
2344 2345 2346 2347 2348 |
* 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... |
2349 |
* subset of node_states[N_MEMORY], even if this means going outside the |
909d75a3b [PATCH] cpuset: i... |
2350 2351 2352 2353 2354 2355 |
* tasks cpuset. **/ nodemask_t cpuset_mems_allowed(struct task_struct *tsk) { nodemask_t mask; |
8447a0fee cpuset: convert c... |
2356 |
unsigned long flags; |
909d75a3b [PATCH] cpuset: i... |
2357 |
|
8447a0fee cpuset: convert c... |
2358 |
spin_lock_irqsave(&callback_lock, flags); |
b8dadcb58 cpuset: use rcu_r... |
2359 |
rcu_read_lock(); |
ae1c80238 cpuset: apply cs-... |
2360 |
guarantee_online_mems(task_cs(tsk), &mask); |
b8dadcb58 cpuset: use rcu_r... |
2361 |
rcu_read_unlock(); |
8447a0fee cpuset: convert c... |
2362 |
spin_unlock_irqrestore(&callback_lock, flags); |
909d75a3b [PATCH] cpuset: i... |
2363 2364 2365 2366 2367 |
return mask; } /** |
19770b326 mm: filter based ... |
2368 2369 |
* cpuset_nodemask_valid_mems_allowed - check nodemask vs. curremt mems_allowed * @nodemask: the nodemask to be checked |
d9fd8a6d4 [PATCH] kernel/cp... |
2370 |
* |
19770b326 mm: filter based ... |
2371 |
* Are any of the nodes in the nodemask allowed in current->mems_allowed? |
1da177e4c Linux-2.6.12-rc2 |
2372 |
*/ |
19770b326 mm: filter based ... |
2373 |
int cpuset_nodemask_valid_mems_allowed(nodemask_t *nodemask) |
1da177e4c Linux-2.6.12-rc2 |
2374 |
{ |
19770b326 mm: filter based ... |
2375 |
return nodes_intersects(*nodemask, current->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
2376 |
} |
9bf2229f8 [PATCH] cpusets: ... |
2377 |
/* |
786083667 Cpuset hardwall f... |
2378 2379 |
* nearest_hardwall_ancestor() - Returns the nearest mem_exclusive or * mem_hardwall ancestor to the specified cpuset. Call holding |
8447a0fee cpuset: convert c... |
2380 |
* callback_lock. If no ancestor is mem_exclusive or mem_hardwall |
786083667 Cpuset hardwall f... |
2381 |
* (an unusual configuration), then returns the root cpuset. |
9bf2229f8 [PATCH] cpusets: ... |
2382 |
*/ |
c9710d801 cpuset: drop "con... |
2383 |
static struct cpuset *nearest_hardwall_ancestor(struct cpuset *cs) |
9bf2229f8 [PATCH] cpusets: ... |
2384 |
{ |
c431069fe cpuset: remove cp... |
2385 2386 |
while (!(is_mem_exclusive(cs) || is_mem_hardwall(cs)) && parent_cs(cs)) cs = parent_cs(cs); |
9bf2229f8 [PATCH] cpusets: ... |
2387 2388 |
return cs; } |
d9fd8a6d4 [PATCH] kernel/cp... |
2389 |
/** |
344736f29 cpuset: simplify ... |
2390 |
* cpuset_node_allowed - Can we allocate on a memory node? |
a1bc5a4ee cpusets: replace ... |
2391 |
* @node: is this an allowed node? |
02a0e53d8 [PATCH] cpuset: r... |
2392 |
* @gfp_mask: memory allocation flags |
d9fd8a6d4 [PATCH] kernel/cp... |
2393 |
* |
6e276d2a5 kernel, cpuset: r... |
2394 2395 2396 2397 |
* If we're in interrupt, yes, we can always allocate. If @node is set in * current's mems_allowed, yes. If it's not a __GFP_HARDWALL request and this * node is set in the nearest hardwalled cpuset ancestor to current's cpuset, * yes. If current has access to memory reserves due to TIF_MEMDIE, yes. |
9bf2229f8 [PATCH] cpusets: ... |
2398 2399 2400 |
* Otherwise, no. * * GFP_USER allocations are marked with the __GFP_HARDWALL bit, |
c596d9f32 cpusets: allow TI... |
2401 2402 |
* 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: ... |
2403 |
* GFP_KERNEL allocations are not so marked, so can escape to the |
786083667 Cpuset hardwall f... |
2404 |
* nearest enclosing hardwalled ancestor cpuset. |
9bf2229f8 [PATCH] cpusets: ... |
2405 |
* |
8447a0fee cpuset: convert c... |
2406 |
* Scanning up parent cpusets requires callback_lock. The |
02a0e53d8 [PATCH] cpuset: r... |
2407 2408 2409 2410 |
* __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... |
2411 |
* cpuset are short of memory, might require taking the callback_lock. |
9bf2229f8 [PATCH] cpusets: ... |
2412 |
* |
36be57ffe [PATCH] cpuset: u... |
2413 |
* The first call here from mm/page_alloc:get_page_from_freelist() |
02a0e53d8 [PATCH] cpuset: r... |
2414 2415 2416 |
* 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... |
2417 2418 2419 2420 2421 2422 |
* * 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: ... |
2423 2424 |
* in_interrupt - any node ok (current task context irrelevant) * GFP_ATOMIC - any node ok |
c596d9f32 cpusets: allow TI... |
2425 |
* TIF_MEMDIE - any node ok |
786083667 Cpuset hardwall f... |
2426 |
* GFP_KERNEL - any node in enclosing hardwalled cpuset ok |
9bf2229f8 [PATCH] cpusets: ... |
2427 |
* GFP_USER - only nodes in current tasks mems allowed ok. |
02a0e53d8 [PATCH] cpuset: r... |
2428 |
*/ |
002f29062 cpuset: use stati... |
2429 |
bool __cpuset_node_allowed(int node, gfp_t gfp_mask) |
1da177e4c Linux-2.6.12-rc2 |
2430 |
{ |
c9710d801 cpuset: drop "con... |
2431 |
struct cpuset *cs; /* current cpuset ancestors */ |
29afd49b7 [PATCH] cpuset: r... |
2432 |
int allowed; /* is allocation in zone z allowed? */ |
8447a0fee cpuset: convert c... |
2433 |
unsigned long flags; |
9bf2229f8 [PATCH] cpusets: ... |
2434 |
|
6e276d2a5 kernel, cpuset: r... |
2435 |
if (in_interrupt()) |
002f29062 cpuset: use stati... |
2436 |
return true; |
9bf2229f8 [PATCH] cpusets: ... |
2437 |
if (node_isset(node, current->mems_allowed)) |
002f29062 cpuset: use stati... |
2438 |
return true; |
c596d9f32 cpusets: allow TI... |
2439 2440 2441 2442 2443 |
/* * 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))) |
002f29062 cpuset: use stati... |
2444 |
return true; |
9bf2229f8 [PATCH] cpusets: ... |
2445 |
if (gfp_mask & __GFP_HARDWALL) /* If hardwall request, stop here */ |
002f29062 cpuset: use stati... |
2446 |
return false; |
9bf2229f8 [PATCH] cpusets: ... |
2447 |
|
5563e7707 [PATCH] cpuset: f... |
2448 |
if (current->flags & PF_EXITING) /* Let dying task have memory */ |
002f29062 cpuset: use stati... |
2449 |
return true; |
5563e7707 [PATCH] cpuset: f... |
2450 |
|
9bf2229f8 [PATCH] cpusets: ... |
2451 |
/* Not hardwall and node outside mems_allowed: scan up cpusets */ |
8447a0fee cpuset: convert c... |
2452 |
spin_lock_irqsave(&callback_lock, flags); |
053199edf [PATCH] cpusets: ... |
2453 |
|
b8dadcb58 cpuset: use rcu_r... |
2454 |
rcu_read_lock(); |
786083667 Cpuset hardwall f... |
2455 |
cs = nearest_hardwall_ancestor(task_cs(current)); |
99afb0fd5 cpuset: fix a rac... |
2456 |
allowed = node_isset(node, cs->mems_allowed); |
b8dadcb58 cpuset: use rcu_r... |
2457 |
rcu_read_unlock(); |
053199edf [PATCH] cpusets: ... |
2458 |
|
8447a0fee cpuset: convert c... |
2459 |
spin_unlock_irqrestore(&callback_lock, flags); |
9bf2229f8 [PATCH] cpusets: ... |
2460 |
return allowed; |
1da177e4c Linux-2.6.12-rc2 |
2461 |
} |
ef08e3b49 [PATCH] cpusets: ... |
2462 |
/** |
6adef3ebe cpusets: new roun... |
2463 2464 |
* 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... |
2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 |
* * 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... |
2488 |
static int cpuset_spread_node(int *rotor) |
825a46af5 [PATCH] cpuset me... |
2489 |
{ |
0edaf86cf include/linux/nod... |
2490 |
return *rotor = next_node_in(*rotor, current->mems_allowed); |
825a46af5 [PATCH] cpuset me... |
2491 |
} |
6adef3ebe cpusets: new roun... |
2492 2493 2494 |
int cpuset_mem_spread_node(void) { |
778d3b0ff cpusets: randomiz... |
2495 2496 2497 |
if (current->cpuset_mem_spread_rotor == NUMA_NO_NODE) current->cpuset_mem_spread_rotor = node_random(¤t->mems_allowed); |
6adef3ebe cpusets: new roun... |
2498 2499 2500 2501 2502 |
return cpuset_spread_node(¤t->cpuset_mem_spread_rotor); } int cpuset_slab_spread_node(void) { |
778d3b0ff cpusets: randomiz... |
2503 2504 2505 |
if (current->cpuset_slab_spread_rotor == NUMA_NO_NODE) current->cpuset_slab_spread_rotor = node_random(¤t->mems_allowed); |
6adef3ebe cpusets: new roun... |
2506 2507 |
return cpuset_spread_node(¤t->cpuset_slab_spread_rotor); } |
825a46af5 [PATCH] cpuset me... |
2508 2509 2510 |
EXPORT_SYMBOL_GPL(cpuset_mem_spread_node); /** |
bbe373f2c oom: compare cpus... |
2511 2512 2513 2514 2515 2516 2517 2518 |
* 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: ... |
2519 |
**/ |
bbe373f2c oom: compare cpus... |
2520 2521 |
int cpuset_mems_allowed_intersects(const struct task_struct *tsk1, const struct task_struct *tsk2) |
ef08e3b49 [PATCH] cpusets: ... |
2522 |
{ |
bbe373f2c oom: compare cpus... |
2523 |
return nodes_intersects(tsk1->mems_allowed, tsk2->mems_allowed); |
ef08e3b49 [PATCH] cpusets: ... |
2524 |
} |
75aa19941 oom: print trigge... |
2525 |
/** |
da39da3a5 mm, oom: remove t... |
2526 |
* cpuset_print_current_mems_allowed - prints current's cpuset and mems_allowed |
75aa19941 oom: print trigge... |
2527 |
* |
da39da3a5 mm, oom: remove t... |
2528 |
* Description: Prints current's name, cpuset name, and cached copy of its |
b8dadcb58 cpuset: use rcu_r... |
2529 |
* mems_allowed to the kernel log. |
75aa19941 oom: print trigge... |
2530 |
*/ |
da39da3a5 mm, oom: remove t... |
2531 |
void cpuset_print_current_mems_allowed(void) |
75aa19941 oom: print trigge... |
2532 |
{ |
b8dadcb58 cpuset: use rcu_r... |
2533 |
struct cgroup *cgrp; |
75aa19941 oom: print trigge... |
2534 |
|
b8dadcb58 cpuset: use rcu_r... |
2535 |
rcu_read_lock(); |
63f43f55c cpuset: fix cpuse... |
2536 |
|
da39da3a5 mm, oom: remove t... |
2537 2538 |
cgrp = task_cs(current)->css.cgroup; pr_info("%s cpuset=", current->comm); |
e61734c55 cgroup: remove cg... |
2539 |
pr_cont_cgroup_name(cgrp); |
da39da3a5 mm, oom: remove t... |
2540 2541 2542 |
pr_cont(" mems_allowed=%*pbl ", nodemask_pr_args(¤t->mems_allowed)); |
f440d98f8 cpuset: use cgrou... |
2543 |
|
cfb5966be cpuset: fix RCU l... |
2544 |
rcu_read_unlock(); |
75aa19941 oom: print trigge... |
2545 |
} |
1da177e4c Linux-2.6.12-rc2 |
2546 |
/* |
3e0d98b9f [PATCH] cpuset: m... |
2547 2548 2549 2550 |
* 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... |
2551 |
int cpuset_memory_pressure_enabled __read_mostly; |
3e0d98b9f [PATCH] cpuset: m... |
2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 |
/** * 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... |
2573 |
rcu_read_lock(); |
8793d854e Task Control Grou... |
2574 |
fmeter_markevent(&task_cs(current)->fmeter); |
b8dadcb58 cpuset: use rcu_r... |
2575 |
rcu_read_unlock(); |
3e0d98b9f [PATCH] cpuset: m... |
2576 |
} |
8793d854e Task Control Grou... |
2577 |
#ifdef CONFIG_PROC_PID_CPUSET |
3e0d98b9f [PATCH] cpuset: m... |
2578 |
/* |
1da177e4c Linux-2.6.12-rc2 |
2579 2580 2581 |
* proc_cpuset_show() * - Print tasks cpuset path into seq_file. * - Used for /proc/<pid>/cpuset. |
053199edf [PATCH] cpusets: ... |
2582 2583 |
* - 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... |
2584 |
* and we take cpuset_mutex, keeping cpuset_attach() from changing it |
2df167a30 cgroups: update c... |
2585 |
* anyway. |
1da177e4c Linux-2.6.12-rc2 |
2586 |
*/ |
52de4779f cpuset: simplify ... |
2587 2588 |
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 |
2589 |
{ |
4c737b41d cgroup: make cgro... |
2590 |
char *buf; |
8793d854e Task Control Grou... |
2591 |
struct cgroup_subsys_state *css; |
99f895518 [PATCH] proc: don... |
2592 |
int retval; |
1da177e4c Linux-2.6.12-rc2 |
2593 |
|
99f895518 [PATCH] proc: don... |
2594 |
retval = -ENOMEM; |
e61734c55 cgroup: remove cg... |
2595 |
buf = kmalloc(PATH_MAX, GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
2596 |
if (!buf) |
99f895518 [PATCH] proc: don... |
2597 |
goto out; |
a79a908fd cgroup: introduce... |
2598 |
css = task_get_css(tsk, cpuset_cgrp_id); |
4c737b41d cgroup: make cgro... |
2599 2600 |
retval = cgroup_path_ns(css->cgroup, buf, PATH_MAX, current->nsproxy->cgroup_ns); |
a79a908fd cgroup: introduce... |
2601 |
css_put(css); |
4c737b41d cgroup: make cgro... |
2602 |
if (retval >= PATH_MAX) |
679a5e3f1 cpuset: fix error... |
2603 2604 |
retval = -ENAMETOOLONG; if (retval < 0) |
52de4779f cpuset: simplify ... |
2605 |
goto out_free; |
4c737b41d cgroup: make cgro... |
2606 |
seq_puts(m, buf); |
1da177e4c Linux-2.6.12-rc2 |
2607 2608 |
seq_putc(m, ' '); |
e61734c55 cgroup: remove cg... |
2609 |
retval = 0; |
99f895518 [PATCH] proc: don... |
2610 |
out_free: |
1da177e4c Linux-2.6.12-rc2 |
2611 |
kfree(buf); |
99f895518 [PATCH] proc: don... |
2612 |
out: |
1da177e4c Linux-2.6.12-rc2 |
2613 2614 |
return retval; } |
8793d854e Task Control Grou... |
2615 |
#endif /* CONFIG_PROC_PID_CPUSET */ |
1da177e4c Linux-2.6.12-rc2 |
2616 |
|
d01d48278 sched: Always sho... |
2617 |
/* Display task mems_allowed in /proc/<pid>/status file. */ |
df5f8314c proc: seqfile con... |
2618 2619 |
void cpuset_task_status_allowed(struct seq_file *m, struct task_struct *task) { |
e8e6d97c9 cpuset: use %*pb[... |
2620 2621 2622 2623 2624 2625 |
seq_printf(m, "Mems_allowed:\t%*pb ", nodemask_pr_args(&task->mems_allowed)); seq_printf(m, "Mems_allowed_list:\t%*pbl ", nodemask_pr_args(&task->mems_allowed)); |
1da177e4c Linux-2.6.12-rc2 |
2626 |
} |