Blame view
kernel/cpuset.c
74.5 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 53 54 55 56 |
#include <linux/spinlock.h> #include <linux/stat.h> #include <linux/string.h> #include <linux/time.h> #include <linux/backing-dev.h> #include <linux/sort.h> #include <asm/uaccess.h> |
60063497a atomic: use <linu... |
57 |
#include <linux/atomic.h> |
3d3f26a7b [PATCH] kernel/cp... |
58 |
#include <linux/mutex.h> |
956db3ca0 hotplug cpu: move... |
59 60 |
#include <linux/workqueue.h> #include <linux/cgroup.h> |
e44193d39 cpuset: let hotpl... |
61 |
#include <linux/wait.h> |
1da177e4c Linux-2.6.12-rc2 |
62 |
|
202f72d5d [PATCH] cpuset: n... |
63 64 65 66 67 |
/* * Tracks how many cpusets are currently defined in system. * When there is only one cpuset (the root cpuset) we can * short circuit some hooks. */ |
7edc59628 [PATCH] cpuset: m... |
68 |
int number_of_cpusets __read_mostly; |
202f72d5d [PATCH] cpuset: n... |
69 |
|
3e0d98b9f [PATCH] cpuset: m... |
70 71 72 73 74 75 76 77 |
/* See "Frequency meter" comments, below. */ struct fmeter { int cnt; /* unprocessed events count */ int val; /* most recent output value */ time_t time; /* clock (secs) when val computed */ spinlock_t lock; /* guards read or write of above */ }; |
1da177e4c Linux-2.6.12-rc2 |
78 |
struct cpuset { |
8793d854e Task Control Grou... |
79 |
struct cgroup_subsys_state css; |
1da177e4c Linux-2.6.12-rc2 |
80 |
unsigned long flags; /* "unsigned long" so bitops work */ |
300ed6cbb cpuset: convert c... |
81 |
cpumask_var_t cpus_allowed; /* CPUs allowed to tasks in cpuset */ |
1da177e4c Linux-2.6.12-rc2 |
82 |
nodemask_t mems_allowed; /* Memory Nodes allowed to tasks */ |
33ad801df cpuset: record ol... |
83 84 85 86 87 88 89 90 91 92 93 |
/* * 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... |
94 |
struct fmeter fmeter; /* memory_pressure filter */ |
029190c51 cpuset sched_load... |
95 |
|
452477fa6 cpuset: pin down ... |
96 97 98 99 100 |
/* * 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... |
101 102 |
/* partition number for rebuild_sched_domains() */ int pn; |
956db3ca0 hotplug cpu: move... |
103 |
|
1d3504fcf sched, cpuset: cu... |
104 105 |
/* for custom sched domain */ int relax_domain_level; |
1da177e4c Linux-2.6.12-rc2 |
106 |
}; |
a7c6d554a cgroup: add/updat... |
107 |
static inline struct cpuset *css_cs(struct cgroup_subsys_state *css) |
8793d854e Task Control Grou... |
108 |
{ |
a7c6d554a cgroup: add/updat... |
109 |
return css ? container_of(css, struct cpuset, css) : NULL; |
8793d854e Task Control Grou... |
110 111 112 113 114 |
} /* Retrieve the cpuset for a task */ static inline struct cpuset *task_cs(struct task_struct *task) { |
073219e99 cgroup: clean up ... |
115 |
return css_cs(task_css(task, cpuset_cgrp_id)); |
8793d854e Task Control Grou... |
116 |
} |
8793d854e Task Control Grou... |
117 |
|
c9710d801 cpuset: drop "con... |
118 |
static inline struct cpuset *parent_cs(struct cpuset *cs) |
c431069fe cpuset: remove cp... |
119 |
{ |
638769869 cgroup: add css_p... |
120 |
return css_cs(css_parent(&cs->css)); |
c431069fe cpuset: remove cp... |
121 |
} |
b246272ec cpusets: stall wh... |
122 123 124 125 126 127 128 129 130 131 132 |
#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 |
133 134 |
/* bits in struct cpuset flags field */ typedef enum { |
efeb77b2f cpuset: introduce... |
135 |
CS_ONLINE, |
1da177e4c Linux-2.6.12-rc2 |
136 137 |
CS_CPU_EXCLUSIVE, CS_MEM_EXCLUSIVE, |
786083667 Cpuset hardwall f... |
138 |
CS_MEM_HARDWALL, |
45b07ef31 [PATCH] cpusets: ... |
139 |
CS_MEMORY_MIGRATE, |
029190c51 cpuset sched_load... |
140 |
CS_SCHED_LOAD_BALANCE, |
825a46af5 [PATCH] cpuset me... |
141 142 |
CS_SPREAD_PAGE, CS_SPREAD_SLAB, |
1da177e4c Linux-2.6.12-rc2 |
143 144 145 |
} cpuset_flagbits_t; /* convenient tests for these bits */ |
efeb77b2f cpuset: introduce... |
146 147 148 149 |
static inline bool is_cpuset_online(const struct cpuset *cs) { return test_bit(CS_ONLINE, &cs->flags); } |
1da177e4c Linux-2.6.12-rc2 |
150 151 |
static inline int is_cpu_exclusive(const struct cpuset *cs) { |
7b5b9ef0e [PATCH] cpuset cl... |
152 |
return test_bit(CS_CPU_EXCLUSIVE, &cs->flags); |
1da177e4c Linux-2.6.12-rc2 |
153 154 155 156 |
} static inline int is_mem_exclusive(const struct cpuset *cs) { |
7b5b9ef0e [PATCH] cpuset cl... |
157 |
return test_bit(CS_MEM_EXCLUSIVE, &cs->flags); |
1da177e4c Linux-2.6.12-rc2 |
158 |
} |
786083667 Cpuset hardwall f... |
159 160 161 162 |
static inline int is_mem_hardwall(const struct cpuset *cs) { return test_bit(CS_MEM_HARDWALL, &cs->flags); } |
029190c51 cpuset sched_load... |
163 164 165 166 |
static inline int is_sched_load_balance(const struct cpuset *cs) { return test_bit(CS_SCHED_LOAD_BALANCE, &cs->flags); } |
45b07ef31 [PATCH] cpusets: ... |
167 168 |
static inline int is_memory_migrate(const struct cpuset *cs) { |
7b5b9ef0e [PATCH] cpuset cl... |
169 |
return test_bit(CS_MEMORY_MIGRATE, &cs->flags); |
45b07ef31 [PATCH] cpusets: ... |
170 |
} |
825a46af5 [PATCH] cpuset me... |
171 172 173 174 175 176 177 178 179 |
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 |
180 |
static struct cpuset top_cpuset = { |
efeb77b2f cpuset: introduce... |
181 182 |
.flags = ((1 << CS_ONLINE) | (1 << CS_CPU_EXCLUSIVE) | (1 << CS_MEM_EXCLUSIVE)), |
1da177e4c Linux-2.6.12-rc2 |
183 |
}; |
ae8086ce1 cpuset: introduce... |
184 185 186 |
/** * cpuset_for_each_child - traverse online children of a cpuset * @child_cs: loop cursor pointing to the current child |
492eb21b9 cgroup: make hier... |
187 |
* @pos_css: used for iteration |
ae8086ce1 cpuset: introduce... |
188 189 190 191 192 |
* @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... |
193 194 195 |
#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... |
196 |
|
fc560a26a cpuset: replace c... |
197 198 199 |
/** * 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... |
200 |
* @pos_css: used for iteration |
fc560a26a cpuset: replace c... |
201 202 203 |
* @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... |
204 |
* with RCU read locked. The caller may modify @pos_css by calling |
bd8815a6d cgroup: make css_... |
205 206 |
* css_rightmost_descendant() to skip subtree. @root_cs is included in the * iteration and the first node to be visited. |
fc560a26a cpuset: replace c... |
207 |
*/ |
492eb21b9 cgroup: make hier... |
208 209 210 |
#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... |
211 |
|
1da177e4c Linux-2.6.12-rc2 |
212 |
/* |
5d21cc2db cpuset: replace c... |
213 214 215 216 217 218 219 220 221 222 223 224 225 226 |
* There are two global mutexes guarding cpuset structures - cpuset_mutex * and callback_mutex. The latter may nest inside the former. We also * require taking task_lock() when dereferencing a task's cpuset pointer. * See "The task_lock() exception", at the end of this comment. * * A task must hold both mutexes to modify cpusets. If a task holds * cpuset_mutex, then it blocks others wanting that mutex, ensuring that it * is the only task able to also acquire callback_mutex and be able to * 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 * callback routines can briefly acquire callback_mutex to query cpusets. * Once it is ready to make the changes, it takes callback_mutex, blocking * everyone else. |
053199edf [PATCH] cpusets: ... |
227 228 |
* * Calls to the kernel memory allocator can not be made while holding |
3d3f26a7b [PATCH] kernel/cp... |
229 |
* callback_mutex, as that would risk double tripping on callback_mutex |
053199edf [PATCH] cpusets: ... |
230 231 232 |
* from one of the callbacks into the cpuset code from within * __alloc_pages(). * |
3d3f26a7b [PATCH] kernel/cp... |
233 |
* If a task is only holding callback_mutex, then it has read-only |
053199edf [PATCH] cpusets: ... |
234 235 |
* access to cpusets. * |
58568d2a8 cpuset,mm: update... |
236 237 238 |
* 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: ... |
239 |
* |
3d3f26a7b [PATCH] kernel/cp... |
240 |
* The cpuset_common_file_read() handlers only hold callback_mutex across |
053199edf [PATCH] cpusets: ... |
241 242 243 |
* small pieces of code, such as when reading out possibly multi-word * cpumasks and nodemasks. * |
2df167a30 cgroups: update c... |
244 245 |
* 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 |
246 |
*/ |
5d21cc2db cpuset: replace c... |
247 |
static DEFINE_MUTEX(cpuset_mutex); |
3d3f26a7b [PATCH] kernel/cp... |
248 |
static DEFINE_MUTEX(callback_mutex); |
4247bdc60 [PATCH] cpuset se... |
249 |
|
cf417141c sched, cpuset: re... |
250 |
/* |
3a5a6d0c2 cpuset: don't nes... |
251 252 253 |
* CPU / memory hotplug is handled asynchronously. */ static void cpuset_hotplug_workfn(struct work_struct *work); |
3a5a6d0c2 cpuset: don't nes... |
254 |
static DECLARE_WORK(cpuset_hotplug_work, cpuset_hotplug_workfn); |
e44193d39 cpuset: let hotpl... |
255 |
static DECLARE_WAIT_QUEUE_HEAD(cpuset_attach_wq); |
3a5a6d0c2 cpuset: don't nes... |
256 |
/* |
cf417141c sched, cpuset: re... |
257 |
* This is ugly, but preserves the userspace API for existing cpuset |
8793d854e Task Control Grou... |
258 |
* users. If someone tries to mount the "cpuset" filesystem, we |
cf417141c sched, cpuset: re... |
259 260 |
* silently switch it to mount "cgroup" instead */ |
f7e835710 convert cgroup an... |
261 262 |
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 |
263 |
{ |
8793d854e Task Control Grou... |
264 |
struct file_system_type *cgroup_fs = get_fs_type("cgroup"); |
f7e835710 convert cgroup an... |
265 |
struct dentry *ret = ERR_PTR(-ENODEV); |
8793d854e Task Control Grou... |
266 267 268 269 |
if (cgroup_fs) { char mountopts[] = "cpuset,noprefix," "release_agent=/sbin/cpuset_release_agent"; |
f7e835710 convert cgroup an... |
270 271 |
ret = cgroup_fs->mount(cgroup_fs, flags, unused_dev_name, mountopts); |
8793d854e Task Control Grou... |
272 273 274 |
put_filesystem(cgroup_fs); } return ret; |
1da177e4c Linux-2.6.12-rc2 |
275 276 277 278 |
} static struct file_system_type cpuset_fs_type = { .name = "cpuset", |
f7e835710 convert cgroup an... |
279 |
.mount = cpuset_mount, |
1da177e4c Linux-2.6.12-rc2 |
280 |
}; |
1da177e4c Linux-2.6.12-rc2 |
281 |
/* |
300ed6cbb cpuset: convert c... |
282 |
* Return in pmask the portion of a cpusets's cpus_allowed that |
1da177e4c Linux-2.6.12-rc2 |
283 |
* are online. If none are online, walk up the cpuset hierarchy |
40df2deb5 cpuset: cleanup g... |
284 285 |
* until we find one that does have some online cpus. The top * cpuset always has some cpus online. |
1da177e4c Linux-2.6.12-rc2 |
286 287 |
* * One way or another, we guarantee to return some non-empty subset |
5f054e31c documentation: re... |
288 |
* of cpu_online_mask. |
1da177e4c Linux-2.6.12-rc2 |
289 |
* |
3d3f26a7b [PATCH] kernel/cp... |
290 |
* Call with callback_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
291 |
*/ |
c9710d801 cpuset: drop "con... |
292 |
static void guarantee_online_cpus(struct cpuset *cs, struct cpumask *pmask) |
1da177e4c Linux-2.6.12-rc2 |
293 |
{ |
40df2deb5 cpuset: cleanup g... |
294 |
while (!cpumask_intersects(cs->cpus_allowed, cpu_online_mask)) |
c431069fe cpuset: remove cp... |
295 |
cs = parent_cs(cs); |
40df2deb5 cpuset: cleanup g... |
296 |
cpumask_and(pmask, cs->cpus_allowed, cpu_online_mask); |
1da177e4c Linux-2.6.12-rc2 |
297 298 299 300 |
} /* * Return in *pmask the portion of a cpusets's mems_allowed that |
0e1e7c7a7 Memoryless nodes:... |
301 302 |
* 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... |
303 |
* online mems. The top cpuset always has some mems online. |
1da177e4c Linux-2.6.12-rc2 |
304 305 |
* * One way or another, we guarantee to return some non-empty subset |
38d7bee9d cpuset: use N_MEM... |
306 |
* of node_states[N_MEMORY]. |
1da177e4c Linux-2.6.12-rc2 |
307 |
* |
3d3f26a7b [PATCH] kernel/cp... |
308 |
* Call with callback_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
309 |
*/ |
c9710d801 cpuset: drop "con... |
310 |
static void guarantee_online_mems(struct cpuset *cs, nodemask_t *pmask) |
1da177e4c Linux-2.6.12-rc2 |
311 |
{ |
40df2deb5 cpuset: cleanup g... |
312 |
while (!nodes_intersects(cs->mems_allowed, node_states[N_MEMORY])) |
c431069fe cpuset: remove cp... |
313 |
cs = parent_cs(cs); |
40df2deb5 cpuset: cleanup g... |
314 |
nodes_and(*pmask, cs->mems_allowed, node_states[N_MEMORY]); |
1da177e4c Linux-2.6.12-rc2 |
315 |
} |
f3b39d47e cpusets: restruct... |
316 317 318 |
/* * update task's spread flag if cpuset's page/slab spread flag is set * |
5d21cc2db cpuset: replace c... |
319 |
* Called with callback_mutex/cpuset_mutex held |
f3b39d47e cpusets: restruct... |
320 321 322 323 324 325 326 327 328 329 330 331 332 |
*/ static void cpuset_update_task_spread_flag(struct cpuset *cs, struct task_struct *tsk) { if (is_spread_page(cs)) tsk->flags |= PF_SPREAD_PAGE; else tsk->flags &= ~PF_SPREAD_PAGE; if (is_spread_slab(cs)) tsk->flags |= PF_SPREAD_SLAB; else tsk->flags &= ~PF_SPREAD_SLAB; } |
1da177e4c Linux-2.6.12-rc2 |
333 334 335 336 337 |
/* * 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... |
338 |
* are only set if the other's are set. Call holding cpuset_mutex. |
1da177e4c Linux-2.6.12-rc2 |
339 340 341 342 |
*/ static int is_cpuset_subset(const struct cpuset *p, const struct cpuset *q) { |
300ed6cbb cpuset: convert c... |
343 |
return cpumask_subset(p->cpus_allowed, q->cpus_allowed) && |
1da177e4c Linux-2.6.12-rc2 |
344 345 346 347 |
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... |
348 349 350 351 |
/** * alloc_trial_cpuset - allocate a trial cpuset * @cs: the cpuset that the trial cpuset duplicates */ |
c9710d801 cpuset: drop "con... |
352 |
static struct cpuset *alloc_trial_cpuset(struct cpuset *cs) |
645fcc9d2 cpuset: don't all... |
353 |
{ |
300ed6cbb cpuset: convert c... |
354 355 356 357 358 359 360 361 362 363 364 365 366 |
struct cpuset *trial; trial = kmemdup(cs, sizeof(*cs), GFP_KERNEL); if (!trial) return NULL; if (!alloc_cpumask_var(&trial->cpus_allowed, GFP_KERNEL)) { kfree(trial); return NULL; } cpumask_copy(trial->cpus_allowed, cs->cpus_allowed); return trial; |
645fcc9d2 cpuset: don't all... |
367 368 369 370 371 372 373 374 |
} /** * free_trial_cpuset - free the trial cpuset * @trial: the trial cpuset to be freed */ static void free_trial_cpuset(struct cpuset *trial) { |
300ed6cbb cpuset: convert c... |
375 |
free_cpumask_var(trial->cpus_allowed); |
645fcc9d2 cpuset: don't all... |
376 377 |
kfree(trial); } |
1da177e4c Linux-2.6.12-rc2 |
378 379 380 381 382 383 384 |
/* * 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... |
385 |
* cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
386 387 388 389 390 391 392 393 394 395 396 |
* * '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... |
397 |
static int validate_change(struct cpuset *cur, struct cpuset *trial) |
1da177e4c Linux-2.6.12-rc2 |
398 |
{ |
492eb21b9 cgroup: make hier... |
399 |
struct cgroup_subsys_state *css; |
1da177e4c Linux-2.6.12-rc2 |
400 |
struct cpuset *c, *par; |
ae8086ce1 cpuset: introduce... |
401 402 403 |
int ret; rcu_read_lock(); |
1da177e4c Linux-2.6.12-rc2 |
404 405 |
/* Each of our child cpusets must be a subset of us */ |
ae8086ce1 cpuset: introduce... |
406 |
ret = -EBUSY; |
492eb21b9 cgroup: make hier... |
407 |
cpuset_for_each_child(c, css, cur) |
ae8086ce1 cpuset: introduce... |
408 409 |
if (!is_cpuset_subset(c, trial)) goto out; |
1da177e4c Linux-2.6.12-rc2 |
410 411 |
/* Remaining checks don't apply to root cpuset */ |
ae8086ce1 cpuset: introduce... |
412 |
ret = 0; |
696040670 [PATCH] cpuset: m... |
413 |
if (cur == &top_cpuset) |
ae8086ce1 cpuset: introduce... |
414 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
415 |
|
c431069fe cpuset: remove cp... |
416 |
par = parent_cs(cur); |
696040670 [PATCH] cpuset: m... |
417 |
|
1da177e4c Linux-2.6.12-rc2 |
418 |
/* We must be a subset of our parent cpuset */ |
ae8086ce1 cpuset: introduce... |
419 |
ret = -EACCES; |
1da177e4c Linux-2.6.12-rc2 |
420 |
if (!is_cpuset_subset(trial, par)) |
ae8086ce1 cpuset: introduce... |
421 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
422 |
|
2df167a30 cgroups: update c... |
423 424 425 426 |
/* * If either I or some sibling (!= me) is exclusive, we can't * overlap */ |
ae8086ce1 cpuset: introduce... |
427 |
ret = -EINVAL; |
492eb21b9 cgroup: make hier... |
428 |
cpuset_for_each_child(c, css, par) { |
1da177e4c Linux-2.6.12-rc2 |
429 430 |
if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) && c != cur && |
300ed6cbb cpuset: convert c... |
431 |
cpumask_intersects(trial->cpus_allowed, c->cpus_allowed)) |
ae8086ce1 cpuset: introduce... |
432 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
433 434 435 |
if ((is_mem_exclusive(trial) || is_mem_exclusive(c)) && c != cur && nodes_intersects(trial->mems_allowed, c->mems_allowed)) |
ae8086ce1 cpuset: introduce... |
436 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
437 |
} |
452477fa6 cpuset: pin down ... |
438 439 |
/* * Cpusets with tasks - existing or newly being attached - can't |
1c09b195d cpuset: fix a reg... |
440 |
* be changed to have empty cpus_allowed or mems_allowed. |
452477fa6 cpuset: pin down ... |
441 |
*/ |
ae8086ce1 cpuset: introduce... |
442 |
ret = -ENOSPC; |
07bc356ed cgroup: implement... |
443 |
if ((cgroup_has_tasks(cur->css.cgroup) || cur->attach_in_progress)) { |
1c09b195d cpuset: fix a reg... |
444 445 446 447 448 449 450 |
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... |
451 |
|
ae8086ce1 cpuset: introduce... |
452 453 454 455 |
ret = 0; out: rcu_read_unlock(); return ret; |
1da177e4c Linux-2.6.12-rc2 |
456 |
} |
db7f47cf4 cpusets: allow cp... |
457 |
#ifdef CONFIG_SMP |
85d7b9498 [PATCH] Dynamic s... |
458 |
/* |
cf417141c sched, cpuset: re... |
459 |
* Helper routine for generate_sched_domains(). |
029190c51 cpuset sched_load... |
460 461 |
* Do cpusets a, b have overlapping cpus_allowed masks? */ |
029190c51 cpuset sched_load... |
462 463 |
static int cpusets_overlap(struct cpuset *a, struct cpuset *b) { |
300ed6cbb cpuset: convert c... |
464 |
return cpumask_intersects(a->cpus_allowed, b->cpus_allowed); |
029190c51 cpuset sched_load... |
465 |
} |
1d3504fcf sched, cpuset: cu... |
466 467 468 |
static void update_domain_attr(struct sched_domain_attr *dattr, struct cpuset *c) { |
1d3504fcf sched, cpuset: cu... |
469 470 471 472 |
if (dattr->relax_domain_level < c->relax_domain_level) dattr->relax_domain_level = c->relax_domain_level; return; } |
fc560a26a cpuset: replace c... |
473 474 |
static void update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *root_cs) |
f5393693e cpuset: speed up ... |
475 |
{ |
fc560a26a cpuset: replace c... |
476 |
struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
477 |
struct cgroup_subsys_state *pos_css; |
f5393693e cpuset: speed up ... |
478 |
|
fc560a26a cpuset: replace c... |
479 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
480 |
cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { |
bd8815a6d cgroup: make css_... |
481 482 |
if (cp == root_cs) continue; |
fc560a26a cpuset: replace c... |
483 484 |
/* skip the whole subtree if @cp doesn't have any CPU */ if (cpumask_empty(cp->cpus_allowed)) { |
492eb21b9 cgroup: make hier... |
485 |
pos_css = css_rightmost_descendant(pos_css); |
f5393693e cpuset: speed up ... |
486 |
continue; |
fc560a26a cpuset: replace c... |
487 |
} |
f5393693e cpuset: speed up ... |
488 489 490 |
if (is_sched_load_balance(cp)) update_domain_attr(dattr, cp); |
f5393693e cpuset: speed up ... |
491 |
} |
fc560a26a cpuset: replace c... |
492 |
rcu_read_unlock(); |
f5393693e cpuset: speed up ... |
493 |
} |
029190c51 cpuset sched_load... |
494 |
/* |
cf417141c sched, cpuset: re... |
495 496 497 498 499 |
* 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... |
500 |
* The output of this function needs to be passed to kernel/sched/core.c |
cf417141c sched, cpuset: re... |
501 502 503 |
* 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... |
504 |
* |
45ce80fb6 cgroups: consolid... |
505 |
* See "What is sched_load_balance" in Documentation/cgroups/cpusets.txt |
029190c51 cpuset sched_load... |
506 507 508 509 510 511 512 |
* 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... |
513 |
* Must be called with cpuset_mutex held. |
029190c51 cpuset sched_load... |
514 515 |
* * The three key local variables below are: |
aeed68242 cpuset: clean up ... |
516 |
* q - a linked-list queue of cpuset pointers, used to implement a |
029190c51 cpuset sched_load... |
517 518 519 520 521 522 523 524 525 526 527 528 |
* 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... |
529 |
* the kernel/sched/core.c routine partition_sched_domains() in a |
029190c51 cpuset sched_load... |
530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 |
* 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... |
548 |
static int generate_sched_domains(cpumask_var_t **domains, |
cf417141c sched, cpuset: re... |
549 |
struct sched_domain_attr **attributes) |
029190c51 cpuset sched_load... |
550 |
{ |
029190c51 cpuset sched_load... |
551 552 553 554 |
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... |
555 |
cpumask_var_t *doms; /* resulting partition; i.e. sched domains */ |
1d3504fcf sched, cpuset: cu... |
556 |
struct sched_domain_attr *dattr; /* attributes for custom domains */ |
1583715dd sched, cpusets: f... |
557 |
int ndoms = 0; /* number of sched domains in result */ |
6af866af3 cpuset: remove re... |
558 |
int nslot; /* next empty doms[] struct cpumask slot */ |
492eb21b9 cgroup: make hier... |
559 |
struct cgroup_subsys_state *pos_css; |
029190c51 cpuset sched_load... |
560 |
|
029190c51 cpuset sched_load... |
561 |
doms = NULL; |
1d3504fcf sched, cpuset: cu... |
562 |
dattr = NULL; |
cf417141c sched, cpuset: re... |
563 |
csa = NULL; |
029190c51 cpuset sched_load... |
564 565 566 |
/* Special case for the 99% of systems with one, full, sched domain */ if (is_sched_load_balance(&top_cpuset)) { |
acc3f5d7c cpumask: Partitio... |
567 568 |
ndoms = 1; doms = alloc_sched_domains(ndoms); |
029190c51 cpuset sched_load... |
569 |
if (!doms) |
cf417141c sched, cpuset: re... |
570 |
goto done; |
1d3504fcf sched, cpuset: cu... |
571 572 573 |
dattr = kmalloc(sizeof(struct sched_domain_attr), GFP_KERNEL); if (dattr) { *dattr = SD_ATTR_INIT; |
93a655755 cpuset: fix wrong... |
574 |
update_domain_attr_tree(dattr, &top_cpuset); |
1d3504fcf sched, cpuset: cu... |
575 |
} |
acc3f5d7c cpumask: Partitio... |
576 |
cpumask_copy(doms[0], top_cpuset.cpus_allowed); |
cf417141c sched, cpuset: re... |
577 |
|
cf417141c sched, cpuset: re... |
578 |
goto done; |
029190c51 cpuset sched_load... |
579 |
} |
029190c51 cpuset sched_load... |
580 581 582 583 |
csa = kmalloc(number_of_cpusets * sizeof(cp), GFP_KERNEL); if (!csa) goto done; csn = 0; |
fc560a26a cpuset: replace c... |
584 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
585 |
cpuset_for_each_descendant_pre(cp, pos_css, &top_cpuset) { |
bd8815a6d cgroup: make css_... |
586 587 |
if (cp == &top_cpuset) continue; |
f5393693e cpuset: speed up ... |
588 |
/* |
fc560a26a cpuset: replace c... |
589 590 591 592 593 594 |
* 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 ... |
595 |
*/ |
fc560a26a cpuset: replace c... |
596 597 |
if (!cpumask_empty(cp->cpus_allowed) && !is_sched_load_balance(cp)) |
f5393693e cpuset: speed up ... |
598 |
continue; |
489a5393a cpuset: don't pas... |
599 |
|
fc560a26a cpuset: replace c... |
600 601 602 603 |
if (is_sched_load_balance(cp)) csa[csn++] = cp; /* skip @cp's subtree */ |
492eb21b9 cgroup: make hier... |
604 |
pos_css = css_rightmost_descendant(pos_css); |
fc560a26a cpuset: replace c... |
605 606 |
} rcu_read_unlock(); |
029190c51 cpuset sched_load... |
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 |
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... |
634 635 636 637 |
/* * Now we know how many domains to create. * Convert <csn, csa> to <ndoms, doms> and populate cpu masks. */ |
acc3f5d7c cpumask: Partitio... |
638 |
doms = alloc_sched_domains(ndoms); |
700018e0a cpuset: fix regre... |
639 |
if (!doms) |
cf417141c sched, cpuset: re... |
640 |
goto done; |
cf417141c sched, cpuset: re... |
641 642 643 644 645 |
/* * 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... |
646 |
dattr = kmalloc(ndoms * sizeof(struct sched_domain_attr), GFP_KERNEL); |
029190c51 cpuset sched_load... |
647 648 649 |
for (nslot = 0, i = 0; i < csn; i++) { struct cpuset *a = csa[i]; |
6af866af3 cpuset: remove re... |
650 |
struct cpumask *dp; |
029190c51 cpuset sched_load... |
651 |
int apn = a->pn; |
cf417141c sched, cpuset: re... |
652 653 654 655 |
if (apn < 0) { /* Skip completed partitions */ continue; } |
acc3f5d7c cpumask: Partitio... |
656 |
dp = doms[nslot]; |
cf417141c sched, cpuset: re... |
657 658 659 660 661 662 663 664 665 666 667 |
if (nslot == ndoms) { static int warnings = 10; if (warnings) { printk(KERN_WARNING "rebuild_sched_domains confused:" " nslot %d, ndoms %d, csn %d, i %d," " apn %d ", nslot, ndoms, csn, i, apn); warnings--; |
029190c51 cpuset sched_load... |
668 |
} |
cf417141c sched, cpuset: re... |
669 670 |
continue; } |
029190c51 cpuset sched_load... |
671 |
|
6af866af3 cpuset: remove re... |
672 |
cpumask_clear(dp); |
cf417141c sched, cpuset: re... |
673 674 675 676 677 678 |
if (dattr) *(dattr + nslot) = SD_ATTR_INIT; for (j = i; j < csn; j++) { struct cpuset *b = csa[j]; if (apn == b->pn) { |
300ed6cbb cpuset: convert c... |
679 |
cpumask_or(dp, dp, b->cpus_allowed); |
cf417141c sched, cpuset: re... |
680 681 682 683 684 |
if (dattr) update_domain_attr_tree(dattr + nslot, b); /* Done with this partition */ b->pn = -1; |
029190c51 cpuset sched_load... |
685 |
} |
029190c51 cpuset sched_load... |
686 |
} |
cf417141c sched, cpuset: re... |
687 |
nslot++; |
029190c51 cpuset sched_load... |
688 689 |
} BUG_ON(nslot != ndoms); |
cf417141c sched, cpuset: re... |
690 691 |
done: kfree(csa); |
700018e0a cpuset: fix regre... |
692 693 694 695 696 697 |
/* * Fallback to the default domain if kmalloc() failed. * See comments in partition_sched_domains(). */ if (doms == NULL) ndoms = 1; |
cf417141c sched, cpuset: re... |
698 699 700 701 702 703 704 705 |
*domains = doms; *attributes = dattr; return ndoms; } /* * Rebuild scheduler domains. * |
699140ba8 cpuset: drop asyn... |
706 707 708 709 710 |
* 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... |
711 |
* |
5d21cc2db cpuset: replace c... |
712 |
* Call with cpuset_mutex held. Takes get_online_cpus(). |
cf417141c sched, cpuset: re... |
713 |
*/ |
699140ba8 cpuset: drop asyn... |
714 |
static void rebuild_sched_domains_locked(void) |
cf417141c sched, cpuset: re... |
715 716 |
{ struct sched_domain_attr *attr; |
acc3f5d7c cpumask: Partitio... |
717 |
cpumask_var_t *doms; |
cf417141c sched, cpuset: re... |
718 |
int ndoms; |
5d21cc2db cpuset: replace c... |
719 |
lockdep_assert_held(&cpuset_mutex); |
86ef5c9a8 cpu-hotplug: repl... |
720 |
get_online_cpus(); |
cf417141c sched, cpuset: re... |
721 |
|
5b16c2a49 cpuset: fix cpu h... |
722 723 724 725 726 727 728 |
/* * 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. */ if (!cpumask_equal(top_cpuset.cpus_allowed, cpu_active_mask)) goto out; |
cf417141c sched, cpuset: re... |
729 |
/* Generate domain masks and attrs */ |
cf417141c sched, cpuset: re... |
730 |
ndoms = generate_sched_domains(&doms, &attr); |
cf417141c sched, cpuset: re... |
731 732 733 |
/* Have scheduler rebuild the domains */ partition_sched_domains(ndoms, doms, attr); |
5b16c2a49 cpuset: fix cpu h... |
734 |
out: |
86ef5c9a8 cpu-hotplug: repl... |
735 |
put_online_cpus(); |
cf417141c sched, cpuset: re... |
736 |
} |
db7f47cf4 cpusets: allow cp... |
737 |
#else /* !CONFIG_SMP */ |
699140ba8 cpuset: drop asyn... |
738 |
static void rebuild_sched_domains_locked(void) |
db7f47cf4 cpusets: allow cp... |
739 740 |
{ } |
db7f47cf4 cpusets: allow cp... |
741 |
#endif /* CONFIG_SMP */ |
029190c51 cpuset sched_load... |
742 |
|
cf417141c sched, cpuset: re... |
743 744 |
void rebuild_sched_domains(void) { |
5d21cc2db cpuset: replace c... |
745 |
mutex_lock(&cpuset_mutex); |
699140ba8 cpuset: drop asyn... |
746 |
rebuild_sched_domains_locked(); |
5d21cc2db cpuset: replace c... |
747 |
mutex_unlock(&cpuset_mutex); |
029190c51 cpuset sched_load... |
748 |
} |
070b57fca cpuset: introduce... |
749 750 751 |
/* * effective_cpumask_cpuset - return nearest ancestor with non-empty cpus * @cs: the cpuset in interest |
58f4790b7 cpusets: update_c... |
752 |
* |
070b57fca cpuset: introduce... |
753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 |
* A cpuset's effective cpumask is the cpumask of the nearest ancestor * with non-empty cpus. We use effective cpumask whenever: * - we update tasks' cpus_allowed. (they take on the ancestor's cpumask * if the cpuset they reside in has no cpus) * - we want to retrieve task_cs(tsk)'s cpus_allowed. * * Called with cpuset_mutex held. cpuset_cpus_allowed_fallback() is an * exception. See comments there. */ static struct cpuset *effective_cpumask_cpuset(struct cpuset *cs) { while (cpumask_empty(cs->cpus_allowed)) cs = parent_cs(cs); return cs; } /* * effective_nodemask_cpuset - return nearest ancestor with non-empty mems * @cs: the cpuset in interest * * A cpuset's effective nodemask is the nodemask of the nearest ancestor * with non-empty memss. We use effective nodemask whenever: * - we update tasks' mems_allowed. (they take on the ancestor's nodemask * if the cpuset they reside in has no mems) * - we want to retrieve task_cs(tsk)'s mems_allowed. * * Called with cpuset_mutex held. |
053199edf [PATCH] cpusets: ... |
780 |
*/ |
070b57fca cpuset: introduce... |
781 |
static struct cpuset *effective_nodemask_cpuset(struct cpuset *cs) |
58f4790b7 cpusets: update_c... |
782 |
{ |
070b57fca cpuset: introduce... |
783 784 785 |
while (nodes_empty(cs->mems_allowed)) cs = parent_cs(cs); return cs; |
58f4790b7 cpusets: update_c... |
786 |
} |
053199edf [PATCH] cpusets: ... |
787 |
|
58f4790b7 cpusets: update_c... |
788 |
/** |
0b2f630a2 cpusets: restruct... |
789 790 |
* 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... |
791 |
* |
d66393e54 cpuset: use css_t... |
792 793 794 |
* 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... |
795 |
*/ |
d66393e54 cpuset: use css_t... |
796 |
static void update_tasks_cpumask(struct cpuset *cs) |
0b2f630a2 cpusets: restruct... |
797 |
{ |
d66393e54 cpuset: use css_t... |
798 799 800 801 802 803 804 805 |
struct cpuset *cpus_cs = effective_cpumask_cpuset(cs); struct css_task_iter it; struct task_struct *task; css_task_iter_start(&cs->css, &it); while ((task = css_task_iter_next(&it))) set_cpus_allowed_ptr(task, cpus_cs->cpus_allowed); css_task_iter_end(&it); |
0b2f630a2 cpusets: restruct... |
806 |
} |
5c5cc6232 cpuset: allow to ... |
807 808 809 810 |
/* * update_tasks_cpumask_hier - Update the cpumasks of tasks in the hierarchy. * @root_cs: the root cpuset of the hierarchy * @update_root: update root cpuset or not? |
5c5cc6232 cpuset: allow to ... |
811 812 813 814 815 816 |
* * This will update cpumasks of tasks in @root_cs and all other empty cpusets * which take on cpumask of @root_cs. * * Called with cpuset_mutex held */ |
d66393e54 cpuset: use css_t... |
817 |
static void update_tasks_cpumask_hier(struct cpuset *root_cs, bool update_root) |
5c5cc6232 cpuset: allow to ... |
818 819 |
{ struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
820 |
struct cgroup_subsys_state *pos_css; |
5c5cc6232 cpuset: allow to ... |
821 822 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
823 |
cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { |
bd8815a6d cgroup: make css_... |
824 825 826 827 828 829 830 831 832 |
if (cp == root_cs) { if (!update_root) continue; } else { /* skip the whole subtree if @cp have some CPU */ if (!cpumask_empty(cp->cpus_allowed)) { pos_css = css_rightmost_descendant(pos_css); continue; } |
5c5cc6232 cpuset: allow to ... |
833 834 835 836 |
} if (!css_tryget(&cp->css)) continue; rcu_read_unlock(); |
d66393e54 cpuset: use css_t... |
837 |
update_tasks_cpumask(cp); |
5c5cc6232 cpuset: allow to ... |
838 839 840 841 842 843 |
rcu_read_lock(); css_put(&cp->css); } rcu_read_unlock(); } |
0b2f630a2 cpusets: restruct... |
844 |
/** |
58f4790b7 cpusets: update_c... |
845 846 847 848 |
* update_cpumask - update the cpus_allowed mask of a cpuset and all tasks in it * @cs: the cpuset to consider * @buf: buffer of cpu numbers written to this cpuset */ |
645fcc9d2 cpuset: don't all... |
849 850 |
static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) |
1da177e4c Linux-2.6.12-rc2 |
851 |
{ |
58f4790b7 cpusets: update_c... |
852 853 |
int retval; int is_load_balanced; |
1da177e4c Linux-2.6.12-rc2 |
854 |
|
5f054e31c documentation: re... |
855 |
/* top_cpuset.cpus_allowed tracks cpu_online_mask; it's read-only */ |
4c4d50f7b [PATCH] cpuset: t... |
856 857 |
if (cs == &top_cpuset) return -EACCES; |
6f7f02e78 cpusets: allow em... |
858 |
/* |
c8d9c90c7 hotplug cpu: move... |
859 |
* An empty cpus_allowed is ok only if the cpuset has no tasks. |
020958b62 cpusets: decrusti... |
860 861 862 |
* 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... |
863 |
*/ |
020958b62 cpusets: decrusti... |
864 |
if (!*buf) { |
300ed6cbb cpuset: convert c... |
865 |
cpumask_clear(trialcs->cpus_allowed); |
6f7f02e78 cpusets: allow em... |
866 |
} else { |
300ed6cbb cpuset: convert c... |
867 |
retval = cpulist_parse(buf, trialcs->cpus_allowed); |
6f7f02e78 cpusets: allow em... |
868 869 |
if (retval < 0) return retval; |
37340746a cpusets: fix bug ... |
870 |
|
6ad4c1888 sched: Fix balanc... |
871 |
if (!cpumask_subset(trialcs->cpus_allowed, cpu_active_mask)) |
37340746a cpusets: fix bug ... |
872 |
return -EINVAL; |
6f7f02e78 cpusets: allow em... |
873 |
} |
029190c51 cpuset sched_load... |
874 |
|
8707d8b8c Fix cpusets updat... |
875 |
/* Nothing to do if the cpus didn't change */ |
300ed6cbb cpuset: convert c... |
876 |
if (cpumask_equal(cs->cpus_allowed, trialcs->cpus_allowed)) |
8707d8b8c Fix cpusets updat... |
877 |
return 0; |
58f4790b7 cpusets: update_c... |
878 |
|
a73456f37 cpuset: re-struct... |
879 880 881 |
retval = validate_change(cs, trialcs); if (retval < 0) return retval; |
645fcc9d2 cpuset: don't all... |
882 |
is_load_balanced = is_sched_load_balance(trialcs); |
029190c51 cpuset sched_load... |
883 |
|
3d3f26a7b [PATCH] kernel/cp... |
884 |
mutex_lock(&callback_mutex); |
300ed6cbb cpuset: convert c... |
885 |
cpumask_copy(cs->cpus_allowed, trialcs->cpus_allowed); |
3d3f26a7b [PATCH] kernel/cp... |
886 |
mutex_unlock(&callback_mutex); |
029190c51 cpuset sched_load... |
887 |
|
d66393e54 cpuset: use css_t... |
888 |
update_tasks_cpumask_hier(cs, true); |
58f4790b7 cpusets: update_c... |
889 |
|
8707d8b8c Fix cpusets updat... |
890 |
if (is_load_balanced) |
699140ba8 cpuset: drop asyn... |
891 |
rebuild_sched_domains_locked(); |
85d7b9498 [PATCH] Dynamic s... |
892 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
893 |
} |
053199edf [PATCH] cpusets: ... |
894 |
/* |
e4e364e86 [PATCH] cpuset: m... |
895 896 897 898 899 900 901 |
* cpuset_migrate_mm * * Migrate memory region from one set of nodes to another. * * Temporarilly set tasks mems_allowed to target nodes of migration, * so that the migration code can allocate pages on these nodes. * |
e4e364e86 [PATCH] cpuset: m... |
902 903 904 905 |
* While the mm_struct we are migrating is typically from some * other task, the task_struct mems_allowed that we are hacking * is for our current task, which must allocate new pages for that * migrating memory region. |
e4e364e86 [PATCH] cpuset: m... |
906 907 908 909 910 911 |
*/ static void cpuset_migrate_mm(struct mm_struct *mm, const nodemask_t *from, const nodemask_t *to) { struct task_struct *tsk = current; |
070b57fca cpuset: introduce... |
912 |
struct cpuset *mems_cs; |
e4e364e86 [PATCH] cpuset: m... |
913 |
|
e4e364e86 [PATCH] cpuset: m... |
914 |
tsk->mems_allowed = *to; |
e4e364e86 [PATCH] cpuset: m... |
915 916 |
do_migrate_pages(mm, from, to, MPOL_MF_MOVE_ALL); |
472958300 cpuset: fix a loc... |
917 |
rcu_read_lock(); |
070b57fca cpuset: introduce... |
918 919 |
mems_cs = effective_nodemask_cpuset(task_cs(tsk)); guarantee_online_mems(mems_cs, &tsk->mems_allowed); |
472958300 cpuset: fix a loc... |
920 |
rcu_read_unlock(); |
e4e364e86 [PATCH] cpuset: m... |
921 |
} |
3b6766fe6 cpuset: rewrite u... |
922 |
/* |
58568d2a8 cpuset,mm: update... |
923 924 925 926 927 928 929 |
* 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... |
930 931 932 933 |
*/ static void cpuset_change_task_nodemask(struct task_struct *tsk, nodemask_t *newmems) { |
b246272ec cpusets: stall wh... |
934 |
bool need_loop; |
89e8a244b cpusets: avoid lo... |
935 |
|
c0ff7453b cpuset,mm: fix no... |
936 937 938 939 940 941 942 943 944 945 |
/* * Allow tasks that have access to memory reserves because they have * been OOM killed to get memory anywhere. */ if (unlikely(test_thread_flag(TIF_MEMDIE))) return; if (current->flags & PF_EXITING) /* Let dying task have memory */ return; task_lock(tsk); |
b246272ec cpusets: stall wh... |
946 947 |
/* * Determine if a loop is necessary if another thread is doing |
d26914d11 mm: optimize put_... |
948 |
* read_mems_allowed_begin(). If at least one node remains unchanged and |
b246272ec cpusets: stall wh... |
949 950 951 952 953 |
* 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... |
954 |
|
0fc0287c9 cpuset: Fix memor... |
955 956 |
if (need_loop) { local_irq_disable(); |
cc9a6c877 cpuset: mm: reduc... |
957 |
write_seqcount_begin(&tsk->mems_allowed_seq); |
0fc0287c9 cpuset: Fix memor... |
958 |
} |
c0ff7453b cpuset,mm: fix no... |
959 |
|
cc9a6c877 cpuset: mm: reduc... |
960 961 |
nodes_or(tsk->mems_allowed, tsk->mems_allowed, *newmems); mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP1); |
c0ff7453b cpuset,mm: fix no... |
962 963 |
mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP2); |
58568d2a8 cpuset,mm: update... |
964 |
tsk->mems_allowed = *newmems; |
cc9a6c877 cpuset: mm: reduc... |
965 |
|
0fc0287c9 cpuset: Fix memor... |
966 |
if (need_loop) { |
cc9a6c877 cpuset: mm: reduc... |
967 |
write_seqcount_end(&tsk->mems_allowed_seq); |
0fc0287c9 cpuset: Fix memor... |
968 969 |
local_irq_enable(); } |
cc9a6c877 cpuset: mm: reduc... |
970 |
|
c0ff7453b cpuset,mm: fix no... |
971 |
task_unlock(tsk); |
58568d2a8 cpuset,mm: update... |
972 |
} |
8793d854e Task Control Grou... |
973 |
static void *cpuset_being_rebound; |
0b2f630a2 cpusets: restruct... |
974 975 976 |
/** * 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... |
977 |
* |
d66393e54 cpuset: use css_t... |
978 979 980 |
* 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... |
981 |
*/ |
d66393e54 cpuset: use css_t... |
982 |
static void update_tasks_nodemask(struct cpuset *cs) |
1da177e4c Linux-2.6.12-rc2 |
983 |
{ |
33ad801df cpuset: record ol... |
984 |
static nodemask_t newmems; /* protected by cpuset_mutex */ |
070b57fca cpuset: introduce... |
985 |
struct cpuset *mems_cs = effective_nodemask_cpuset(cs); |
d66393e54 cpuset: use css_t... |
986 987 |
struct css_task_iter it; struct task_struct *task; |
59dac16fb [PATCH] cpuset: u... |
988 |
|
846a16bf0 mempolicy: rename... |
989 |
cpuset_being_rebound = cs; /* causes mpol_dup() rebind */ |
4225399a6 [PATCH] cpuset: r... |
990 |
|
070b57fca cpuset: introduce... |
991 |
guarantee_online_mems(mems_cs, &newmems); |
33ad801df cpuset: record ol... |
992 |
|
4225399a6 [PATCH] cpuset: r... |
993 |
/* |
3b6766fe6 cpuset: rewrite u... |
994 995 996 997 |
* 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... |
998 |
* the global cpuset_mutex, we know that no other rebind effort |
3b6766fe6 cpuset: rewrite u... |
999 |
* will be contending for the global variable cpuset_being_rebound. |
4225399a6 [PATCH] cpuset: r... |
1000 |
* It's ok if we rebind the same mm twice; mpol_rebind_mm() |
04c19fa6f [PATCH] cpuset: m... |
1001 |
* is idempotent. Also migrate pages in each mm to new nodes. |
4225399a6 [PATCH] cpuset: r... |
1002 |
*/ |
d66393e54 cpuset: use css_t... |
1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 |
css_task_iter_start(&cs->css, &it); while ((task = css_task_iter_next(&it))) { struct mm_struct *mm; bool migrate; cpuset_change_task_nodemask(task, &newmems); mm = get_task_mm(task); if (!mm) continue; migrate = is_memory_migrate(cs); mpol_rebind_mm(mm, &cs->mems_allowed); if (migrate) cpuset_migrate_mm(mm, &cs->old_mems_allowed, &newmems); mmput(mm); } css_task_iter_end(&it); |
4225399a6 [PATCH] cpuset: r... |
1022 |
|
33ad801df cpuset: record ol... |
1023 1024 1025 1026 1027 |
/* * All the tasks' nodemasks have been updated, update * cs->old_mems_allowed. */ cs->old_mems_allowed = newmems; |
2df167a30 cgroups: update c... |
1028 |
/* We're done rebinding vmas to this cpuset's new mems_allowed. */ |
8793d854e Task Control Grou... |
1029 |
cpuset_being_rebound = NULL; |
1da177e4c Linux-2.6.12-rc2 |
1030 |
} |
0b2f630a2 cpusets: restruct... |
1031 |
/* |
5c5cc6232 cpuset: allow to ... |
1032 1033 1034 |
* update_tasks_nodemask_hier - Update the nodemasks of tasks in the hierarchy. * @cs: the root cpuset of the hierarchy * @update_root: update the root cpuset or not? |
5c5cc6232 cpuset: allow to ... |
1035 1036 1037 1038 1039 1040 |
* * This will update nodemasks of tasks in @root_cs and all other empty cpusets * which take on nodemask of @root_cs. * * Called with cpuset_mutex held */ |
d66393e54 cpuset: use css_t... |
1041 |
static void update_tasks_nodemask_hier(struct cpuset *root_cs, bool update_root) |
5c5cc6232 cpuset: allow to ... |
1042 1043 |
{ struct cpuset *cp; |
492eb21b9 cgroup: make hier... |
1044 |
struct cgroup_subsys_state *pos_css; |
5c5cc6232 cpuset: allow to ... |
1045 1046 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
1047 |
cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { |
bd8815a6d cgroup: make css_... |
1048 1049 1050 1051 1052 1053 1054 1055 1056 |
if (cp == root_cs) { if (!update_root) continue; } else { /* skip the whole subtree if @cp have some CPU */ if (!nodes_empty(cp->mems_allowed)) { pos_css = css_rightmost_descendant(pos_css); continue; } |
5c5cc6232 cpuset: allow to ... |
1057 1058 1059 1060 |
} if (!css_tryget(&cp->css)) continue; rcu_read_unlock(); |
d66393e54 cpuset: use css_t... |
1061 |
update_tasks_nodemask(cp); |
5c5cc6232 cpuset: allow to ... |
1062 1063 1064 1065 1066 1067 1068 1069 |
rcu_read_lock(); css_put(&cp->css); } rcu_read_unlock(); } /* |
0b2f630a2 cpusets: restruct... |
1070 1071 |
* Handle user request to change the 'mems' memory placement * of a cpuset. Needs to validate the request, update the |
58568d2a8 cpuset,mm: update... |
1072 1073 1074 1075 |
* 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... |
1076 |
* |
5d21cc2db cpuset: replace c... |
1077 |
* Call with cpuset_mutex held. May take callback_mutex during call. |
0b2f630a2 cpusets: restruct... |
1078 1079 1080 1081 |
* 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... |
1082 1083 |
static int update_nodemask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) |
0b2f630a2 cpusets: restruct... |
1084 |
{ |
0b2f630a2 cpusets: restruct... |
1085 1086 1087 |
int retval; /* |
38d7bee9d cpuset: use N_MEM... |
1088 |
* top_cpuset.mems_allowed tracks node_stats[N_MEMORY]; |
0b2f630a2 cpusets: restruct... |
1089 1090 |
* it's read-only */ |
53feb2976 cpuset: alloc nod... |
1091 1092 1093 1094 |
if (cs == &top_cpuset) { retval = -EACCES; goto done; } |
0b2f630a2 cpusets: restruct... |
1095 |
|
0b2f630a2 cpusets: restruct... |
1096 1097 1098 1099 1100 1101 1102 |
/* * 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... |
1103 |
nodes_clear(trialcs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1104 |
} else { |
645fcc9d2 cpuset: don't all... |
1105 |
retval = nodelist_parse(buf, trialcs->mems_allowed); |
0b2f630a2 cpusets: restruct... |
1106 1107 |
if (retval < 0) goto done; |
645fcc9d2 cpuset: don't all... |
1108 |
if (!nodes_subset(trialcs->mems_allowed, |
38d7bee9d cpuset: use N_MEM... |
1109 |
node_states[N_MEMORY])) { |
53feb2976 cpuset: alloc nod... |
1110 1111 1112 |
retval = -EINVAL; goto done; } |
0b2f630a2 cpusets: restruct... |
1113 |
} |
33ad801df cpuset: record ol... |
1114 1115 |
if (nodes_equal(cs->mems_allowed, trialcs->mems_allowed)) { |
0b2f630a2 cpusets: restruct... |
1116 1117 1118 |
retval = 0; /* Too easy - nothing to do */ goto done; } |
645fcc9d2 cpuset: don't all... |
1119 |
retval = validate_change(cs, trialcs); |
0b2f630a2 cpusets: restruct... |
1120 1121 1122 1123 |
if (retval < 0) goto done; mutex_lock(&callback_mutex); |
645fcc9d2 cpuset: don't all... |
1124 |
cs->mems_allowed = trialcs->mems_allowed; |
0b2f630a2 cpusets: restruct... |
1125 |
mutex_unlock(&callback_mutex); |
d66393e54 cpuset: use css_t... |
1126 |
update_tasks_nodemask_hier(cs, true); |
0b2f630a2 cpusets: restruct... |
1127 1128 1129 |
done: return retval; } |
8793d854e Task Control Grou... |
1130 1131 1132 1133 |
int current_cpuset_is_being_rebound(void) { return task_cs(current) == cpuset_being_rebound; } |
5be7a4792 Fix cpuset sched_... |
1134 |
static int update_relax_domain_level(struct cpuset *cs, s64 val) |
1d3504fcf sched, cpuset: cu... |
1135 |
{ |
db7f47cf4 cpusets: allow cp... |
1136 |
#ifdef CONFIG_SMP |
60495e776 sched: Dynamic sc... |
1137 |
if (val < -1 || val >= sched_domain_level_max) |
30e0e1781 cpuset: limit the... |
1138 |
return -EINVAL; |
db7f47cf4 cpusets: allow cp... |
1139 |
#endif |
1d3504fcf sched, cpuset: cu... |
1140 1141 1142 |
if (val != cs->relax_domain_level) { cs->relax_domain_level = val; |
300ed6cbb cpuset: convert c... |
1143 1144 |
if (!cpumask_empty(cs->cpus_allowed) && is_sched_load_balance(cs)) |
699140ba8 cpuset: drop asyn... |
1145 |
rebuild_sched_domains_locked(); |
1d3504fcf sched, cpuset: cu... |
1146 1147 1148 1149 |
} return 0; } |
72ec70299 cgroup: make task... |
1150 |
/** |
950592f7b cpusets: update t... |
1151 1152 |
* 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... |
1153 |
* |
d66393e54 cpuset: use css_t... |
1154 1155 1156 |
* 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... |
1157 |
*/ |
d66393e54 cpuset: use css_t... |
1158 |
static void update_tasks_flags(struct cpuset *cs) |
950592f7b cpusets: update t... |
1159 |
{ |
d66393e54 cpuset: use css_t... |
1160 1161 1162 1163 1164 1165 1166 |
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... |
1167 1168 1169 |
} /* |
1da177e4c Linux-2.6.12-rc2 |
1170 |
* update_flag - read a 0 or a 1 in a file and update associated flag |
786083667 Cpuset hardwall f... |
1171 1172 1173 |
* 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: ... |
1174 |
* |
5d21cc2db cpuset: replace c... |
1175 |
* Call with cpuset_mutex held. |
1da177e4c Linux-2.6.12-rc2 |
1176 |
*/ |
700fe1ab9 CGroup API files:... |
1177 1178 |
static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, int turning_on) |
1da177e4c Linux-2.6.12-rc2 |
1179 |
{ |
645fcc9d2 cpuset: don't all... |
1180 |
struct cpuset *trialcs; |
40b6a7623 cpuset.c: remove ... |
1181 |
int balance_flag_changed; |
950592f7b cpusets: update t... |
1182 |
int spread_flag_changed; |
950592f7b cpusets: update t... |
1183 |
int err; |
1da177e4c Linux-2.6.12-rc2 |
1184 |
|
645fcc9d2 cpuset: don't all... |
1185 1186 1187 |
trialcs = alloc_trial_cpuset(cs); if (!trialcs) return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
1188 |
if (turning_on) |
645fcc9d2 cpuset: don't all... |
1189 |
set_bit(bit, &trialcs->flags); |
1da177e4c Linux-2.6.12-rc2 |
1190 |
else |
645fcc9d2 cpuset: don't all... |
1191 |
clear_bit(bit, &trialcs->flags); |
1da177e4c Linux-2.6.12-rc2 |
1192 |
|
645fcc9d2 cpuset: don't all... |
1193 |
err = validate_change(cs, trialcs); |
85d7b9498 [PATCH] Dynamic s... |
1194 |
if (err < 0) |
645fcc9d2 cpuset: don't all... |
1195 |
goto out; |
029190c51 cpuset sched_load... |
1196 |
|
029190c51 cpuset sched_load... |
1197 |
balance_flag_changed = (is_sched_load_balance(cs) != |
645fcc9d2 cpuset: don't all... |
1198 |
is_sched_load_balance(trialcs)); |
029190c51 cpuset sched_load... |
1199 |
|
950592f7b cpusets: update t... |
1200 1201 |
spread_flag_changed = ((is_spread_slab(cs) != is_spread_slab(trialcs)) || (is_spread_page(cs) != is_spread_page(trialcs))); |
3d3f26a7b [PATCH] kernel/cp... |
1202 |
mutex_lock(&callback_mutex); |
645fcc9d2 cpuset: don't all... |
1203 |
cs->flags = trialcs->flags; |
3d3f26a7b [PATCH] kernel/cp... |
1204 |
mutex_unlock(&callback_mutex); |
85d7b9498 [PATCH] Dynamic s... |
1205 |
|
300ed6cbb cpuset: convert c... |
1206 |
if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed) |
699140ba8 cpuset: drop asyn... |
1207 |
rebuild_sched_domains_locked(); |
029190c51 cpuset sched_load... |
1208 |
|
950592f7b cpusets: update t... |
1209 |
if (spread_flag_changed) |
d66393e54 cpuset: use css_t... |
1210 |
update_tasks_flags(cs); |
645fcc9d2 cpuset: don't all... |
1211 1212 1213 |
out: free_trial_cpuset(trialcs); return err; |
1da177e4c Linux-2.6.12-rc2 |
1214 |
} |
053199edf [PATCH] cpusets: ... |
1215 |
/* |
80f7228b5 typo fixes: occur... |
1216 |
* Frequency meter - How fast is some event occurring? |
3e0d98b9f [PATCH] cpuset: m... |
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 |
* * These routines manage a digitally filtered, constant time based, * event frequency meter. There are four routines: * fmeter_init() - initialize a frequency meter. * fmeter_markevent() - called each time the event happens. * fmeter_getrate() - returns the recent rate of such events. * fmeter_update() - internal routine used to update fmeter. * * A common data structure is passed to each of these routines, * which is used to keep track of the state required to manage the * frequency meter and its digital filter. * * The filter works on the number of events marked per unit time. * The filter is single-pole low-pass recursive (IIR). The time unit * is 1 second. Arithmetic is done using 32-bit integers scaled to * simulate 3 decimal digits of precision (multiplied by 1000). * * With an FM_COEF of 933, and a time base of 1 second, the filter * has a half-life of 10 seconds, meaning that if the events quit * happening, then the rate returned from the fmeter_getrate() * will be cut in half each 10 seconds, until it converges to zero. * * It is not worth doing a real infinitely recursive filter. If more * than FM_MAXTICKS ticks have elapsed since the last filter event, * just compute FM_MAXTICKS ticks worth, by which point the level * will be stable. * * Limit the count of unprocessed events to FM_MAXCNT, so as to avoid * arithmetic overflow in the fmeter_update() routine. * * Given the simple 32 bit integer arithmetic used, this meter works * best for reporting rates between one per millisecond (msec) and * one per 32 (approx) seconds. At constant rates faster than one * per msec it maxes out at values just under 1,000,000. At constant * rates between one per msec, and one per second it will stabilize * to a value N*1000, where N is the rate of events per second. * At constant rates between one per second and one per 32 seconds, * it will be choppy, moving up on the seconds that have an event, * and then decaying until the next event. At rates slower than * about one in 32 seconds, it decays all the way back to zero between * each event. */ #define FM_COEF 933 /* coefficient for half-life of 10 secs */ #define FM_MAXTICKS ((time_t)99) /* useless computing more ticks than this */ #define FM_MAXCNT 1000000 /* limit cnt to avoid overflow */ #define FM_SCALE 1000 /* faux fixed point scale */ /* Initialize a frequency meter */ static void fmeter_init(struct fmeter *fmp) { fmp->cnt = 0; fmp->val = 0; fmp->time = 0; spin_lock_init(&fmp->lock); } /* Internal meter update - process cnt events and update value */ static void fmeter_update(struct fmeter *fmp) { time_t now = get_seconds(); time_t ticks = now - fmp->time; if (ticks == 0) return; ticks = min(FM_MAXTICKS, ticks); while (ticks-- > 0) fmp->val = (FM_COEF * fmp->val) / FM_SCALE; fmp->time = now; fmp->val += ((FM_SCALE - FM_COEF) * fmp->cnt) / FM_SCALE; fmp->cnt = 0; } /* Process any previous ticks, then bump cnt by one (times scale). */ static void fmeter_markevent(struct fmeter *fmp) { spin_lock(&fmp->lock); fmeter_update(fmp); fmp->cnt = min(FM_MAXCNT, fmp->cnt + FM_SCALE); spin_unlock(&fmp->lock); } /* Process any previous ticks, then return current value. */ static int fmeter_getrate(struct fmeter *fmp) { int val; spin_lock(&fmp->lock); fmeter_update(fmp); val = fmp->val; spin_unlock(&fmp->lock); return val; } |
57fce0a68 cpuset: don't use... |
1312 |
static struct cpuset *cpuset_attach_old_cs; |
5d21cc2db cpuset: replace c... |
1313 |
/* Called by cgroups to determine if a cpuset is usable; cpuset_mutex held */ |
eb95419b0 cgroup: pass arou... |
1314 1315 |
static int cpuset_can_attach(struct cgroup_subsys_state *css, struct cgroup_taskset *tset) |
f780bdb7c cgroups: add per-... |
1316 |
{ |
eb95419b0 cgroup: pass arou... |
1317 |
struct cpuset *cs = css_cs(css); |
bb9d97b6d cgroup: don't use... |
1318 1319 |
struct task_struct *task; int ret; |
1da177e4c Linux-2.6.12-rc2 |
1320 |
|
57fce0a68 cpuset: don't use... |
1321 1322 |
/* used later by cpuset_attach() */ cpuset_attach_old_cs = task_cs(cgroup_taskset_first(tset)); |
5d21cc2db cpuset: replace c... |
1323 |
mutex_lock(&cpuset_mutex); |
88fa523bf cpuset: allow to ... |
1324 1325 1326 1327 |
/* * We allow to move tasks into an empty cpuset if sane_behavior * flag is set. */ |
5d21cc2db cpuset: replace c... |
1328 |
ret = -ENOSPC; |
eb95419b0 cgroup: pass arou... |
1329 |
if (!cgroup_sane_behavior(css->cgroup) && |
88fa523bf cpuset: allow to ... |
1330 |
(cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed))) |
5d21cc2db cpuset: replace c... |
1331 |
goto out_unlock; |
9985b0bab sched: prevent bo... |
1332 |
|
924f0d9a2 cgroup: drop @ski... |
1333 |
cgroup_taskset_for_each(task, tset) { |
bb9d97b6d cgroup: don't use... |
1334 |
/* |
14a40ffcc sched: replace PF... |
1335 1336 1337 1338 1339 1340 1341 |
* Kthreads which disallow setaffinity shouldn't be moved * to a new cpuset; we don't want to change their cpu * affinity and isolating such threads by their set of * allowed nodes is unnecessary. Thus, cpusets are not * applicable for such threads. This prevents checking for * success of set_cpus_allowed_ptr() on all attached tasks * before cpus_allowed may be changed. |
bb9d97b6d cgroup: don't use... |
1342 |
*/ |
5d21cc2db cpuset: replace c... |
1343 |
ret = -EINVAL; |
14a40ffcc sched: replace PF... |
1344 |
if (task->flags & PF_NO_SETAFFINITY) |
5d21cc2db cpuset: replace c... |
1345 1346 1347 1348 |
goto out_unlock; ret = security_task_setscheduler(task); if (ret) goto out_unlock; |
bb9d97b6d cgroup: don't use... |
1349 |
} |
f780bdb7c cgroups: add per-... |
1350 |
|
452477fa6 cpuset: pin down ... |
1351 1352 1353 1354 1355 |
/* * Mark attach is in progress. This makes validate_change() fail * changes which zero cpus/mems_allowed. */ cs->attach_in_progress++; |
5d21cc2db cpuset: replace c... |
1356 1357 1358 1359 |
ret = 0; out_unlock: mutex_unlock(&cpuset_mutex); return ret; |
8793d854e Task Control Grou... |
1360 |
} |
f780bdb7c cgroups: add per-... |
1361 |
|
eb95419b0 cgroup: pass arou... |
1362 |
static void cpuset_cancel_attach(struct cgroup_subsys_state *css, |
452477fa6 cpuset: pin down ... |
1363 1364 |
struct cgroup_taskset *tset) { |
5d21cc2db cpuset: replace c... |
1365 |
mutex_lock(&cpuset_mutex); |
eb95419b0 cgroup: pass arou... |
1366 |
css_cs(css)->attach_in_progress--; |
5d21cc2db cpuset: replace c... |
1367 |
mutex_unlock(&cpuset_mutex); |
8793d854e Task Control Grou... |
1368 |
} |
1da177e4c Linux-2.6.12-rc2 |
1369 |
|
4e4c9a140 cpuset: cleanup c... |
1370 |
/* |
5d21cc2db cpuset: replace c... |
1371 |
* Protected by cpuset_mutex. cpus_attach is used only by cpuset_attach() |
4e4c9a140 cpuset: cleanup c... |
1372 1373 1374 1375 |
* but we can't allocate it dynamically there. Define it global and * allocate from cpuset_init(). */ static cpumask_var_t cpus_attach; |
eb95419b0 cgroup: pass arou... |
1376 1377 |
static void cpuset_attach(struct cgroup_subsys_state *css, struct cgroup_taskset *tset) |
8793d854e Task Control Grou... |
1378 |
{ |
67bd2c598 cpuset: remove un... |
1379 |
/* static buf protected by cpuset_mutex */ |
4e4c9a140 cpuset: cleanup c... |
1380 |
static nodemask_t cpuset_attach_nodemask_to; |
8793d854e Task Control Grou... |
1381 |
struct mm_struct *mm; |
bb9d97b6d cgroup: don't use... |
1382 1383 |
struct task_struct *task; struct task_struct *leader = cgroup_taskset_first(tset); |
eb95419b0 cgroup: pass arou... |
1384 |
struct cpuset *cs = css_cs(css); |
57fce0a68 cpuset: don't use... |
1385 |
struct cpuset *oldcs = cpuset_attach_old_cs; |
070b57fca cpuset: introduce... |
1386 1387 |
struct cpuset *cpus_cs = effective_cpumask_cpuset(cs); struct cpuset *mems_cs = effective_nodemask_cpuset(cs); |
22fb52dd7 [PATCH] SELinux: ... |
1388 |
|
5d21cc2db cpuset: replace c... |
1389 |
mutex_lock(&cpuset_mutex); |
4e4c9a140 cpuset: cleanup c... |
1390 1391 1392 1393 |
/* prepare for attach */ if (cs == &top_cpuset) cpumask_copy(cpus_attach, cpu_possible_mask); else |
070b57fca cpuset: introduce... |
1394 |
guarantee_online_cpus(cpus_cs, cpus_attach); |
4e4c9a140 cpuset: cleanup c... |
1395 |
|
070b57fca cpuset: introduce... |
1396 |
guarantee_online_mems(mems_cs, &cpuset_attach_nodemask_to); |
4e4c9a140 cpuset: cleanup c... |
1397 |
|
924f0d9a2 cgroup: drop @ski... |
1398 |
cgroup_taskset_for_each(task, tset) { |
bb9d97b6d cgroup: don't use... |
1399 1400 1401 1402 1403 1404 1405 1406 1407 |
/* * 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: ... |
1408 |
|
f780bdb7c cgroups: add per-... |
1409 1410 1411 1412 |
/* * Change mm, possibly for multiple threads in a threadgroup. This is * expensive and may sleep. */ |
f780bdb7c cgroups: add per-... |
1413 |
cpuset_attach_nodemask_to = cs->mems_allowed; |
bb9d97b6d cgroup: don't use... |
1414 |
mm = get_task_mm(leader); |
4225399a6 [PATCH] cpuset: r... |
1415 |
if (mm) { |
070b57fca cpuset: introduce... |
1416 |
struct cpuset *mems_oldcs = effective_nodemask_cpuset(oldcs); |
f780bdb7c cgroups: add per-... |
1417 |
mpol_rebind_mm(mm, &cpuset_attach_nodemask_to); |
f047cecf2 cpuset: fix to mi... |
1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 |
/* * old_mems_allowed is the same with mems_allowed here, except * if this task is being moved automatically due to hotplug. * In that case @mems_allowed has been updated and is empty, * so @old_mems_allowed is the right nodesets that we migrate * mm from. */ if (is_memory_migrate(cs)) { cpuset_migrate_mm(mm, &mems_oldcs->old_mems_allowed, |
f780bdb7c cgroups: add per-... |
1428 |
&cpuset_attach_nodemask_to); |
f047cecf2 cpuset: fix to mi... |
1429 |
} |
4225399a6 [PATCH] cpuset: r... |
1430 1431 |
mmput(mm); } |
452477fa6 cpuset: pin down ... |
1432 |
|
33ad801df cpuset: record ol... |
1433 |
cs->old_mems_allowed = cpuset_attach_nodemask_to; |
02bb58637 cpuset: schedule ... |
1434 |
|
452477fa6 cpuset: pin down ... |
1435 |
cs->attach_in_progress--; |
e44193d39 cpuset: let hotpl... |
1436 1437 |
if (!cs->attach_in_progress) wake_up(&cpuset_attach_wq); |
5d21cc2db cpuset: replace c... |
1438 1439 |
mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1440 1441 1442 1443 1444 |
} /* The various types of files and directories in a cpuset file system */ typedef enum { |
45b07ef31 [PATCH] cpusets: ... |
1445 |
FILE_MEMORY_MIGRATE, |
1da177e4c Linux-2.6.12-rc2 |
1446 1447 1448 1449 |
FILE_CPULIST, FILE_MEMLIST, FILE_CPU_EXCLUSIVE, FILE_MEM_EXCLUSIVE, |
786083667 Cpuset hardwall f... |
1450 |
FILE_MEM_HARDWALL, |
029190c51 cpuset sched_load... |
1451 |
FILE_SCHED_LOAD_BALANCE, |
1d3504fcf sched, cpuset: cu... |
1452 |
FILE_SCHED_RELAX_DOMAIN_LEVEL, |
3e0d98b9f [PATCH] cpuset: m... |
1453 1454 |
FILE_MEMORY_PRESSURE_ENABLED, FILE_MEMORY_PRESSURE, |
825a46af5 [PATCH] cpuset me... |
1455 1456 |
FILE_SPREAD_PAGE, FILE_SPREAD_SLAB, |
1da177e4c Linux-2.6.12-rc2 |
1457 |
} cpuset_filetype_t; |
182446d08 cgroup: pass arou... |
1458 1459 |
static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, u64 val) |
700fe1ab9 CGroup API files:... |
1460 |
{ |
182446d08 cgroup: pass arou... |
1461 |
struct cpuset *cs = css_cs(css); |
700fe1ab9 CGroup API files:... |
1462 |
cpuset_filetype_t type = cft->private; |
a903f0865 cpuset: fix the r... |
1463 |
int retval = 0; |
700fe1ab9 CGroup API files:... |
1464 |
|
5d21cc2db cpuset: replace c... |
1465 |
mutex_lock(&cpuset_mutex); |
a903f0865 cpuset: fix the r... |
1466 1467 |
if (!is_cpuset_online(cs)) { retval = -ENODEV; |
5d21cc2db cpuset: replace c... |
1468 |
goto out_unlock; |
a903f0865 cpuset: fix the r... |
1469 |
} |
700fe1ab9 CGroup API files:... |
1470 1471 |
switch (type) { |
1da177e4c Linux-2.6.12-rc2 |
1472 |
case FILE_CPU_EXCLUSIVE: |
700fe1ab9 CGroup API files:... |
1473 |
retval = update_flag(CS_CPU_EXCLUSIVE, cs, val); |
1da177e4c Linux-2.6.12-rc2 |
1474 1475 |
break; case FILE_MEM_EXCLUSIVE: |
700fe1ab9 CGroup API files:... |
1476 |
retval = update_flag(CS_MEM_EXCLUSIVE, cs, val); |
1da177e4c Linux-2.6.12-rc2 |
1477 |
break; |
786083667 Cpuset hardwall f... |
1478 1479 1480 |
case FILE_MEM_HARDWALL: retval = update_flag(CS_MEM_HARDWALL, cs, val); break; |
029190c51 cpuset sched_load... |
1481 |
case FILE_SCHED_LOAD_BALANCE: |
700fe1ab9 CGroup API files:... |
1482 |
retval = update_flag(CS_SCHED_LOAD_BALANCE, cs, val); |
1d3504fcf sched, cpuset: cu... |
1483 |
break; |
45b07ef31 [PATCH] cpusets: ... |
1484 |
case FILE_MEMORY_MIGRATE: |
700fe1ab9 CGroup API files:... |
1485 |
retval = update_flag(CS_MEMORY_MIGRATE, cs, val); |
45b07ef31 [PATCH] cpusets: ... |
1486 |
break; |
3e0d98b9f [PATCH] cpuset: m... |
1487 |
case FILE_MEMORY_PRESSURE_ENABLED: |
700fe1ab9 CGroup API files:... |
1488 |
cpuset_memory_pressure_enabled = !!val; |
3e0d98b9f [PATCH] cpuset: m... |
1489 1490 1491 1492 |
break; case FILE_MEMORY_PRESSURE: retval = -EACCES; break; |
825a46af5 [PATCH] cpuset me... |
1493 |
case FILE_SPREAD_PAGE: |
700fe1ab9 CGroup API files:... |
1494 |
retval = update_flag(CS_SPREAD_PAGE, cs, val); |
825a46af5 [PATCH] cpuset me... |
1495 1496 |
break; case FILE_SPREAD_SLAB: |
700fe1ab9 CGroup API files:... |
1497 |
retval = update_flag(CS_SPREAD_SLAB, cs, val); |
825a46af5 [PATCH] cpuset me... |
1498 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1499 1500 |
default: retval = -EINVAL; |
700fe1ab9 CGroup API files:... |
1501 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1502 |
} |
5d21cc2db cpuset: replace c... |
1503 1504 |
out_unlock: mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1505 1506 |
return retval; } |
182446d08 cgroup: pass arou... |
1507 1508 |
static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, s64 val) |
5be7a4792 Fix cpuset sched_... |
1509 |
{ |
182446d08 cgroup: pass arou... |
1510 |
struct cpuset *cs = css_cs(css); |
5be7a4792 Fix cpuset sched_... |
1511 |
cpuset_filetype_t type = cft->private; |
5d21cc2db cpuset: replace c... |
1512 |
int retval = -ENODEV; |
5be7a4792 Fix cpuset sched_... |
1513 |
|
5d21cc2db cpuset: replace c... |
1514 1515 1516 |
mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; |
e37123953 cgroup files: rem... |
1517 |
|
5be7a4792 Fix cpuset sched_... |
1518 1519 1520 1521 1522 1523 1524 1525 |
switch (type) { case FILE_SCHED_RELAX_DOMAIN_LEVEL: retval = update_relax_domain_level(cs, val); break; default: retval = -EINVAL; break; } |
5d21cc2db cpuset: replace c... |
1526 1527 |
out_unlock: mutex_unlock(&cpuset_mutex); |
5be7a4792 Fix cpuset sched_... |
1528 1529 |
return retval; } |
1da177e4c Linux-2.6.12-rc2 |
1530 |
/* |
e37123953 cgroup files: rem... |
1531 1532 |
* Common handling for a write to a "cpus" or "mems" file. */ |
182446d08 cgroup: pass arou... |
1533 |
static int cpuset_write_resmask(struct cgroup_subsys_state *css, |
4d3bb511b cgroup: drop cons... |
1534 |
struct cftype *cft, char *buf) |
e37123953 cgroup files: rem... |
1535 |
{ |
182446d08 cgroup: pass arou... |
1536 |
struct cpuset *cs = css_cs(css); |
645fcc9d2 cpuset: don't all... |
1537 |
struct cpuset *trialcs; |
5d21cc2db cpuset: replace c... |
1538 |
int retval = -ENODEV; |
e37123953 cgroup files: rem... |
1539 |
|
3a5a6d0c2 cpuset: don't nes... |
1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 |
/* * 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. */ flush_work(&cpuset_hotplug_work); |
5d21cc2db cpuset: replace c... |
1552 1553 1554 |
mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; |
e37123953 cgroup files: rem... |
1555 |
|
645fcc9d2 cpuset: don't all... |
1556 |
trialcs = alloc_trial_cpuset(cs); |
b75f38d65 cpuset: add a mis... |
1557 1558 |
if (!trialcs) { retval = -ENOMEM; |
5d21cc2db cpuset: replace c... |
1559 |
goto out_unlock; |
b75f38d65 cpuset: add a mis... |
1560 |
} |
645fcc9d2 cpuset: don't all... |
1561 |
|
e37123953 cgroup files: rem... |
1562 1563 |
switch (cft->private) { case FILE_CPULIST: |
645fcc9d2 cpuset: don't all... |
1564 |
retval = update_cpumask(cs, trialcs, buf); |
e37123953 cgroup files: rem... |
1565 1566 |
break; case FILE_MEMLIST: |
645fcc9d2 cpuset: don't all... |
1567 |
retval = update_nodemask(cs, trialcs, buf); |
e37123953 cgroup files: rem... |
1568 1569 1570 1571 1572 |
break; default: retval = -EINVAL; break; } |
645fcc9d2 cpuset: don't all... |
1573 1574 |
free_trial_cpuset(trialcs); |
5d21cc2db cpuset: replace c... |
1575 1576 |
out_unlock: mutex_unlock(&cpuset_mutex); |
e37123953 cgroup files: rem... |
1577 1578 1579 1580 |
return retval; } /* |
1da177e4c Linux-2.6.12-rc2 |
1581 1582 1583 1584 1585 1586 |
* 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 |
1587 |
*/ |
2da8ca822 cgroup: replace c... |
1588 |
static int cpuset_common_seq_show(struct seq_file *sf, void *v) |
1da177e4c Linux-2.6.12-rc2 |
1589 |
{ |
2da8ca822 cgroup: replace c... |
1590 1591 |
struct cpuset *cs = css_cs(seq_css(sf)); cpuset_filetype_t type = seq_cft(sf)->private; |
51ffe4117 cpuset: convert a... |
1592 1593 1594 |
ssize_t count; char *buf, *s; int ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
1595 |
|
51ffe4117 cpuset: convert a... |
1596 1597 |
count = seq_get_buf(sf, &buf); s = buf; |
1da177e4c Linux-2.6.12-rc2 |
1598 |
|
51ffe4117 cpuset: convert a... |
1599 |
mutex_lock(&callback_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1600 1601 1602 |
switch (type) { case FILE_CPULIST: |
51ffe4117 cpuset: convert a... |
1603 |
s += cpulist_scnprintf(s, count, cs->cpus_allowed); |
1da177e4c Linux-2.6.12-rc2 |
1604 1605 |
break; case FILE_MEMLIST: |
51ffe4117 cpuset: convert a... |
1606 |
s += nodelist_scnprintf(s, count, cs->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
1607 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1608 |
default: |
51ffe4117 cpuset: convert a... |
1609 1610 |
ret = -EINVAL; goto out_unlock; |
1da177e4c Linux-2.6.12-rc2 |
1611 |
} |
1da177e4c Linux-2.6.12-rc2 |
1612 |
|
51ffe4117 cpuset: convert a... |
1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 |
if (s < buf + count - 1) { *s++ = ' '; seq_commit(sf, s - buf); } else { seq_commit(sf, -1); } out_unlock: mutex_unlock(&callback_mutex); return ret; |
1da177e4c Linux-2.6.12-rc2 |
1623 |
} |
182446d08 cgroup: pass arou... |
1624 |
static u64 cpuset_read_u64(struct cgroup_subsys_state *css, struct cftype *cft) |
700fe1ab9 CGroup API files:... |
1625 |
{ |
182446d08 cgroup: pass arou... |
1626 |
struct cpuset *cs = css_cs(css); |
700fe1ab9 CGroup API files:... |
1627 1628 1629 1630 1631 1632 |
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... |
1633 1634 |
case FILE_MEM_HARDWALL: return is_mem_hardwall(cs); |
700fe1ab9 CGroup API files:... |
1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 |
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... |
1650 1651 1652 |
/* Unreachable but makes gcc happy */ return 0; |
700fe1ab9 CGroup API files:... |
1653 |
} |
1da177e4c Linux-2.6.12-rc2 |
1654 |
|
182446d08 cgroup: pass arou... |
1655 |
static s64 cpuset_read_s64(struct cgroup_subsys_state *css, struct cftype *cft) |
5be7a4792 Fix cpuset sched_... |
1656 |
{ |
182446d08 cgroup: pass arou... |
1657 |
struct cpuset *cs = css_cs(css); |
5be7a4792 Fix cpuset sched_... |
1658 1659 1660 1661 1662 1663 1664 |
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... |
1665 1666 1667 |
/* Unrechable but makes gcc happy */ return 0; |
5be7a4792 Fix cpuset sched_... |
1668 |
} |
1da177e4c Linux-2.6.12-rc2 |
1669 1670 1671 1672 |
/* * for the common functions, 'private' gives the type of file */ |
addf2c739 Cpuset hardwall f... |
1673 1674 1675 |
static struct cftype files[] = { { .name = "cpus", |
2da8ca822 cgroup: replace c... |
1676 |
.seq_show = cpuset_common_seq_show, |
e37123953 cgroup files: rem... |
1677 1678 |
.write_string = cpuset_write_resmask, .max_write_len = (100U + 6 * NR_CPUS), |
addf2c739 Cpuset hardwall f... |
1679 1680 1681 1682 1683 |
.private = FILE_CPULIST, }, { .name = "mems", |
2da8ca822 cgroup: replace c... |
1684 |
.seq_show = cpuset_common_seq_show, |
e37123953 cgroup files: rem... |
1685 1686 |
.write_string = cpuset_write_resmask, .max_write_len = (100U + 6 * MAX_NUMNODES), |
addf2c739 Cpuset hardwall f... |
1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 |
.private = FILE_MEMLIST, }, { .name = "cpu_exclusive", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_CPU_EXCLUSIVE, }, { .name = "mem_exclusive", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_MEM_EXCLUSIVE, }, { |
786083667 Cpuset hardwall f... |
1705 1706 1707 1708 1709 1710 1711 |
.name = "mem_hardwall", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_MEM_HARDWALL, }, { |
addf2c739 Cpuset hardwall f... |
1712 1713 1714 1715 1716 1717 1718 1719 |
.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_... |
1720 1721 |
.read_s64 = cpuset_read_s64, .write_s64 = cpuset_write_s64, |
addf2c739 Cpuset hardwall f... |
1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 |
.private = FILE_SCHED_RELAX_DOMAIN_LEVEL, }, { .name = "memory_migrate", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_MEMORY_MIGRATE, }, { .name = "memory_pressure", .read_u64 = cpuset_read_u64, .write_u64 = cpuset_write_u64, .private = FILE_MEMORY_PRESSURE, |
099fca322 cgroups: show cor... |
1737 |
.mode = S_IRUGO, |
addf2c739 Cpuset hardwall f... |
1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 |
}, { .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... |
1753 |
|
4baf6e332 cgroup: convert a... |
1754 1755 1756 1757 1758 1759 1760 |
{ .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 |
1761 |
|
4baf6e332 cgroup: convert a... |
1762 1763 |
{ } /* terminate */ }; |
1da177e4c Linux-2.6.12-rc2 |
1764 1765 |
/* |
92fb97487 cgroup: rename ->... |
1766 |
* cpuset_css_alloc - allocate a cpuset css |
c9e5fe66f cpuset: rename @c... |
1767 |
* cgrp: control group that the new cpuset will be part of |
1da177e4c Linux-2.6.12-rc2 |
1768 |
*/ |
eb95419b0 cgroup: pass arou... |
1769 1770 |
static struct cgroup_subsys_state * cpuset_css_alloc(struct cgroup_subsys_state *parent_css) |
1da177e4c Linux-2.6.12-rc2 |
1771 |
{ |
c8f699bb5 cpuset: introduce... |
1772 |
struct cpuset *cs; |
1da177e4c Linux-2.6.12-rc2 |
1773 |
|
eb95419b0 cgroup: pass arou... |
1774 |
if (!parent_css) |
8793d854e Task Control Grou... |
1775 |
return &top_cpuset.css; |
033fa1c5f cgroup, cpuset: r... |
1776 |
|
c8f699bb5 cpuset: introduce... |
1777 |
cs = kzalloc(sizeof(*cs), GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
1778 |
if (!cs) |
8793d854e Task Control Grou... |
1779 |
return ERR_PTR(-ENOMEM); |
300ed6cbb cpuset: convert c... |
1780 1781 1782 1783 |
if (!alloc_cpumask_var(&cs->cpus_allowed, GFP_KERNEL)) { kfree(cs); return ERR_PTR(-ENOMEM); } |
1da177e4c Linux-2.6.12-rc2 |
1784 |
|
029190c51 cpuset sched_load... |
1785 |
set_bit(CS_SCHED_LOAD_BALANCE, &cs->flags); |
300ed6cbb cpuset: convert c... |
1786 |
cpumask_clear(cs->cpus_allowed); |
f9a86fcbb cpuset: modify cp... |
1787 |
nodes_clear(cs->mems_allowed); |
3e0d98b9f [PATCH] cpuset: m... |
1788 |
fmeter_init(&cs->fmeter); |
1d3504fcf sched, cpuset: cu... |
1789 |
cs->relax_domain_level = -1; |
1da177e4c Linux-2.6.12-rc2 |
1790 |
|
c8f699bb5 cpuset: introduce... |
1791 1792 |
return &cs->css; } |
eb95419b0 cgroup: pass arou... |
1793 |
static int cpuset_css_online(struct cgroup_subsys_state *css) |
c8f699bb5 cpuset: introduce... |
1794 |
{ |
eb95419b0 cgroup: pass arou... |
1795 |
struct cpuset *cs = css_cs(css); |
c431069fe cpuset: remove cp... |
1796 |
struct cpuset *parent = parent_cs(cs); |
ae8086ce1 cpuset: introduce... |
1797 |
struct cpuset *tmp_cs; |
492eb21b9 cgroup: make hier... |
1798 |
struct cgroup_subsys_state *pos_css; |
c8f699bb5 cpuset: introduce... |
1799 1800 1801 |
if (!parent) return 0; |
5d21cc2db cpuset: replace c... |
1802 |
mutex_lock(&cpuset_mutex); |
efeb77b2f cpuset: introduce... |
1803 |
set_bit(CS_ONLINE, &cs->flags); |
c8f699bb5 cpuset: introduce... |
1804 1805 1806 1807 |
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 |
1808 |
|
202f72d5d [PATCH] cpuset: n... |
1809 |
number_of_cpusets++; |
033fa1c5f cgroup, cpuset: r... |
1810 |
|
eb95419b0 cgroup: pass arou... |
1811 |
if (!test_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags)) |
5d21cc2db cpuset: replace c... |
1812 |
goto out_unlock; |
033fa1c5f cgroup, cpuset: r... |
1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 |
/* * 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... |
1827 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
1828 |
cpuset_for_each_child(tmp_cs, pos_css, parent) { |
ae8086ce1 cpuset: introduce... |
1829 1830 |
if (is_mem_exclusive(tmp_cs) || is_cpu_exclusive(tmp_cs)) { rcu_read_unlock(); |
5d21cc2db cpuset: replace c... |
1831 |
goto out_unlock; |
ae8086ce1 cpuset: introduce... |
1832 |
} |
033fa1c5f cgroup, cpuset: r... |
1833 |
} |
ae8086ce1 cpuset: introduce... |
1834 |
rcu_read_unlock(); |
033fa1c5f cgroup, cpuset: r... |
1835 1836 1837 1838 1839 |
mutex_lock(&callback_mutex); cs->mems_allowed = parent->mems_allowed; cpumask_copy(cs->cpus_allowed, parent->cpus_allowed); mutex_unlock(&callback_mutex); |
5d21cc2db cpuset: replace c... |
1840 1841 |
out_unlock: mutex_unlock(&cpuset_mutex); |
c8f699bb5 cpuset: introduce... |
1842 1843 |
return 0; } |
0b9e6965a cpuset: relocate ... |
1844 1845 1846 1847 1848 |
/* * 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... |
1849 |
static void cpuset_css_offline(struct cgroup_subsys_state *css) |
c8f699bb5 cpuset: introduce... |
1850 |
{ |
eb95419b0 cgroup: pass arou... |
1851 |
struct cpuset *cs = css_cs(css); |
c8f699bb5 cpuset: introduce... |
1852 |
|
5d21cc2db cpuset: replace c... |
1853 |
mutex_lock(&cpuset_mutex); |
c8f699bb5 cpuset: introduce... |
1854 1855 1856 1857 1858 |
if (is_sched_load_balance(cs)) update_flag(CS_SCHED_LOAD_BALANCE, cs, 0); number_of_cpusets--; |
efeb77b2f cpuset: introduce... |
1859 |
clear_bit(CS_ONLINE, &cs->flags); |
c8f699bb5 cpuset: introduce... |
1860 |
|
5d21cc2db cpuset: replace c... |
1861 |
mutex_unlock(&cpuset_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1862 |
} |
eb95419b0 cgroup: pass arou... |
1863 |
static void cpuset_css_free(struct cgroup_subsys_state *css) |
1da177e4c Linux-2.6.12-rc2 |
1864 |
{ |
eb95419b0 cgroup: pass arou... |
1865 |
struct cpuset *cs = css_cs(css); |
1da177e4c Linux-2.6.12-rc2 |
1866 |
|
300ed6cbb cpuset: convert c... |
1867 |
free_cpumask_var(cs->cpus_allowed); |
8793d854e Task Control Grou... |
1868 |
kfree(cs); |
1da177e4c Linux-2.6.12-rc2 |
1869 |
} |
073219e99 cgroup: clean up ... |
1870 |
struct cgroup_subsys cpuset_cgrp_subsys = { |
92fb97487 cgroup: rename ->... |
1871 |
.css_alloc = cpuset_css_alloc, |
c8f699bb5 cpuset: introduce... |
1872 1873 |
.css_online = cpuset_css_online, .css_offline = cpuset_css_offline, |
92fb97487 cgroup: rename ->... |
1874 |
.css_free = cpuset_css_free, |
8793d854e Task Control Grou... |
1875 |
.can_attach = cpuset_can_attach, |
452477fa6 cpuset: pin down ... |
1876 |
.cancel_attach = cpuset_cancel_attach, |
8793d854e Task Control Grou... |
1877 |
.attach = cpuset_attach, |
4baf6e332 cgroup: convert a... |
1878 |
.base_cftypes = files, |
8793d854e Task Control Grou... |
1879 1880 |
.early_init = 1, }; |
1da177e4c Linux-2.6.12-rc2 |
1881 1882 1883 1884 1885 1886 1887 1888 |
/** * 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... |
1889 |
int err = 0; |
1da177e4c Linux-2.6.12-rc2 |
1890 |
|
58568d2a8 cpuset,mm: update... |
1891 1892 |
if (!alloc_cpumask_var(&top_cpuset.cpus_allowed, GFP_KERNEL)) BUG(); |
300ed6cbb cpuset: convert c... |
1893 |
cpumask_setall(top_cpuset.cpus_allowed); |
f9a86fcbb cpuset: modify cp... |
1894 |
nodes_setall(top_cpuset.mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
1895 |
|
3e0d98b9f [PATCH] cpuset: m... |
1896 |
fmeter_init(&top_cpuset.fmeter); |
029190c51 cpuset sched_load... |
1897 |
set_bit(CS_SCHED_LOAD_BALANCE, &top_cpuset.flags); |
1d3504fcf sched, cpuset: cu... |
1898 |
top_cpuset.relax_domain_level = -1; |
1da177e4c Linux-2.6.12-rc2 |
1899 |
|
1da177e4c Linux-2.6.12-rc2 |
1900 1901 |
err = register_filesystem(&cpuset_fs_type); if (err < 0) |
8793d854e Task Control Grou... |
1902 |
return err; |
2341d1b65 cpuset: convert c... |
1903 1904 |
if (!alloc_cpumask_var(&cpus_attach, GFP_KERNEL)) BUG(); |
202f72d5d [PATCH] cpuset: n... |
1905 |
number_of_cpusets = 1; |
8793d854e Task Control Grou... |
1906 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
1907 |
} |
b1aac8bb8 [PATCH] cpuset: h... |
1908 |
/* |
cf417141c sched, cpuset: re... |
1909 |
* If CPU and/or memory hotplug handlers, below, unplug any CPUs |
b1aac8bb8 [PATCH] cpuset: h... |
1910 1911 |
* 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... |
1912 1913 |
* 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... |
1914 |
*/ |
956db3ca0 hotplug cpu: move... |
1915 1916 1917 |
static void remove_tasks_in_empty_cpuset(struct cpuset *cs) { struct cpuset *parent; |
c8d9c90c7 hotplug cpu: move... |
1918 |
/* |
956db3ca0 hotplug cpu: move... |
1919 1920 1921 |
* Find its next-highest non-empty parent, (top cpuset * has online cpus, so can't be empty). */ |
c431069fe cpuset: remove cp... |
1922 |
parent = parent_cs(cs); |
300ed6cbb cpuset: convert c... |
1923 |
while (cpumask_empty(parent->cpus_allowed) || |
b45012955 hotplug cpu move ... |
1924 |
nodes_empty(parent->mems_allowed)) |
c431069fe cpuset: remove cp... |
1925 |
parent = parent_cs(parent); |
956db3ca0 hotplug cpu: move... |
1926 |
|
8cc993452 cgroup, cpuset: r... |
1927 |
if (cgroup_transfer_tasks(parent->css.cgroup, cs->css.cgroup)) { |
e61734c55 cgroup: remove cg... |
1928 1929 1930 1931 |
printk(KERN_ERR "cpuset: failed to transfer tasks out of empty cpuset "); pr_cont_cgroup_name(cs->css.cgroup); pr_cont(" "); |
8cc993452 cgroup, cpuset: r... |
1932 |
} |
956db3ca0 hotplug cpu: move... |
1933 |
} |
deb7aa308 cpuset: reorganiz... |
1934 |
/** |
388afd854 cpuset: remove as... |
1935 |
* cpuset_hotplug_update_tasks - update tasks in a cpuset for hotunplug |
deb7aa308 cpuset: reorganiz... |
1936 |
* @cs: cpuset in interest |
956db3ca0 hotplug cpu: move... |
1937 |
* |
deb7aa308 cpuset: reorganiz... |
1938 1939 1940 |
* 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:... |
1941 |
*/ |
388afd854 cpuset: remove as... |
1942 |
static void cpuset_hotplug_update_tasks(struct cpuset *cs) |
80d1fa646 cpusets, hotplug:... |
1943 |
{ |
deb7aa308 cpuset: reorganiz... |
1944 |
static cpumask_t off_cpus; |
33ad801df cpuset: record ol... |
1945 |
static nodemask_t off_mems; |
5d21cc2db cpuset: replace c... |
1946 |
bool is_empty; |
5c5cc6232 cpuset: allow to ... |
1947 |
bool sane = cgroup_sane_behavior(cs->css.cgroup); |
80d1fa646 cpusets, hotplug:... |
1948 |
|
e44193d39 cpuset: let hotpl... |
1949 1950 |
retry: wait_event(cpuset_attach_wq, cs->attach_in_progress == 0); |
80d1fa646 cpusets, hotplug:... |
1951 |
|
5d21cc2db cpuset: replace c... |
1952 |
mutex_lock(&cpuset_mutex); |
7ddf96b02 cpusets, hotplug:... |
1953 |
|
e44193d39 cpuset: let hotpl... |
1954 1955 1956 1957 1958 1959 1960 1961 |
/* * 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; } |
deb7aa308 cpuset: reorganiz... |
1962 1963 |
cpumask_andnot(&off_cpus, cs->cpus_allowed, top_cpuset.cpus_allowed); nodes_andnot(off_mems, cs->mems_allowed, top_cpuset.mems_allowed); |
80d1fa646 cpusets, hotplug:... |
1964 |
|
5c5cc6232 cpuset: allow to ... |
1965 1966 1967 1968 1969 1970 |
mutex_lock(&callback_mutex); cpumask_andnot(cs->cpus_allowed, cs->cpus_allowed, &off_cpus); mutex_unlock(&callback_mutex); /* * If sane_behavior flag is set, we need to update tasks' cpumask |
f047cecf2 cpuset: fix to mi... |
1971 1972 1973 |
* for empty cpuset to take on ancestor's cpumask. Otherwise, don't * call update_tasks_cpumask() if the cpuset becomes empty, as * the tasks in it will be migrated to an ancestor. |
5c5cc6232 cpuset: allow to ... |
1974 1975 |
*/ if ((sane && cpumask_empty(cs->cpus_allowed)) || |
f047cecf2 cpuset: fix to mi... |
1976 |
(!cpumask_empty(&off_cpus) && !cpumask_empty(cs->cpus_allowed))) |
d66393e54 cpuset: use css_t... |
1977 |
update_tasks_cpumask(cs); |
80d1fa646 cpusets, hotplug:... |
1978 |
|
5c5cc6232 cpuset: allow to ... |
1979 1980 1981 1982 1983 1984 |
mutex_lock(&callback_mutex); nodes_andnot(cs->mems_allowed, cs->mems_allowed, off_mems); mutex_unlock(&callback_mutex); /* * If sane_behavior flag is set, we need to update tasks' nodemask |
f047cecf2 cpuset: fix to mi... |
1985 1986 1987 |
* for empty cpuset to take on ancestor's nodemask. Otherwise, don't * call update_tasks_nodemask() if the cpuset becomes empty, as * the tasks in it will be migratd to an ancestor. |
5c5cc6232 cpuset: allow to ... |
1988 1989 |
*/ if ((sane && nodes_empty(cs->mems_allowed)) || |
f047cecf2 cpuset: fix to mi... |
1990 |
(!nodes_empty(off_mems) && !nodes_empty(cs->mems_allowed))) |
d66393e54 cpuset: use css_t... |
1991 |
update_tasks_nodemask(cs); |
deb7aa308 cpuset: reorganiz... |
1992 |
|
5d21cc2db cpuset: replace c... |
1993 1994 |
is_empty = cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed); |
8d0339487 cpuset: make CPU ... |
1995 |
|
5d21cc2db cpuset: replace c... |
1996 1997 1998 |
mutex_unlock(&cpuset_mutex); /* |
5c5cc6232 cpuset: allow to ... |
1999 2000 2001 2002 |
* If sane_behavior flag is set, we'll keep tasks in empty cpusets. * * Otherwise move tasks to the nearest ancestor with execution * resources. This is full cgroup operation which will |
5d21cc2db cpuset: replace c... |
2003 2004 |
* also call back into cpuset. Should be done outside any lock. */ |
5c5cc6232 cpuset: allow to ... |
2005 |
if (!sane && is_empty) |
5d21cc2db cpuset: replace c... |
2006 |
remove_tasks_in_empty_cpuset(cs); |
b1aac8bb8 [PATCH] cpuset: h... |
2007 |
} |
deb7aa308 cpuset: reorganiz... |
2008 |
/** |
3a5a6d0c2 cpuset: don't nes... |
2009 |
* cpuset_hotplug_workfn - handle CPU/memory hotunplug for a cpuset |
956db3ca0 hotplug cpu: move... |
2010 |
* |
deb7aa308 cpuset: reorganiz... |
2011 2012 2013 2014 2015 |
* 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... |
2016 |
* |
deb7aa308 cpuset: reorganiz... |
2017 |
* Non-root cpusets are only affected by offlining. If any CPUs or memory |
388afd854 cpuset: remove as... |
2018 2019 |
* nodes have been taken down, cpuset_hotplug_update_tasks() is invoked on * all descendants. |
956db3ca0 hotplug cpu: move... |
2020 |
* |
deb7aa308 cpuset: reorganiz... |
2021 2022 |
* Note that CPU offlining during suspend is ignored. We don't modify * cpusets across suspend/resume cycles at all. |
956db3ca0 hotplug cpu: move... |
2023 |
*/ |
3a5a6d0c2 cpuset: don't nes... |
2024 |
static void cpuset_hotplug_workfn(struct work_struct *work) |
b1aac8bb8 [PATCH] cpuset: h... |
2025 |
{ |
5c5cc6232 cpuset: allow to ... |
2026 2027 |
static cpumask_t new_cpus; static nodemask_t new_mems; |
deb7aa308 cpuset: reorganiz... |
2028 |
bool cpus_updated, mems_updated; |
b1aac8bb8 [PATCH] cpuset: h... |
2029 |
|
5d21cc2db cpuset: replace c... |
2030 |
mutex_lock(&cpuset_mutex); |
956db3ca0 hotplug cpu: move... |
2031 |
|
deb7aa308 cpuset: reorganiz... |
2032 2033 2034 |
/* 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:... |
2035 |
|
deb7aa308 cpuset: reorganiz... |
2036 |
cpus_updated = !cpumask_equal(top_cpuset.cpus_allowed, &new_cpus); |
deb7aa308 cpuset: reorganiz... |
2037 |
mems_updated = !nodes_equal(top_cpuset.mems_allowed, new_mems); |
7ddf96b02 cpusets, hotplug:... |
2038 |
|
deb7aa308 cpuset: reorganiz... |
2039 2040 2041 2042 2043 2044 2045 |
/* synchronize cpus_allowed to cpu_active_mask */ if (cpus_updated) { mutex_lock(&callback_mutex); cpumask_copy(top_cpuset.cpus_allowed, &new_cpus); mutex_unlock(&callback_mutex); /* we don't mess with cpumasks of tasks in top_cpuset */ } |
b45012955 hotplug cpu move ... |
2046 |
|
deb7aa308 cpuset: reorganiz... |
2047 2048 |
/* synchronize mems_allowed to N_MEMORY */ if (mems_updated) { |
deb7aa308 cpuset: reorganiz... |
2049 2050 2051 |
mutex_lock(&callback_mutex); top_cpuset.mems_allowed = new_mems; mutex_unlock(&callback_mutex); |
d66393e54 cpuset: use css_t... |
2052 |
update_tasks_nodemask(&top_cpuset); |
deb7aa308 cpuset: reorganiz... |
2053 |
} |
b45012955 hotplug cpu move ... |
2054 |
|
388afd854 cpuset: remove as... |
2055 |
mutex_unlock(&cpuset_mutex); |
5c5cc6232 cpuset: allow to ... |
2056 2057 |
/* if cpus or mems changed, we need to propagate to descendants */ if (cpus_updated || mems_updated) { |
deb7aa308 cpuset: reorganiz... |
2058 |
struct cpuset *cs; |
492eb21b9 cgroup: make hier... |
2059 |
struct cgroup_subsys_state *pos_css; |
f9b4fb8da cpusets: update t... |
2060 |
|
fc560a26a cpuset: replace c... |
2061 |
rcu_read_lock(); |
492eb21b9 cgroup: make hier... |
2062 |
cpuset_for_each_descendant_pre(cs, pos_css, &top_cpuset) { |
bd8815a6d cgroup: make css_... |
2063 |
if (cs == &top_cpuset || !css_tryget(&cs->css)) |
388afd854 cpuset: remove as... |
2064 2065 |
continue; rcu_read_unlock(); |
7ddf96b02 cpusets, hotplug:... |
2066 |
|
388afd854 cpuset: remove as... |
2067 |
cpuset_hotplug_update_tasks(cs); |
b45012955 hotplug cpu move ... |
2068 |
|
388afd854 cpuset: remove as... |
2069 2070 2071 2072 2073 |
rcu_read_lock(); css_put(&cs->css); } rcu_read_unlock(); } |
8d0339487 cpuset: make CPU ... |
2074 |
|
deb7aa308 cpuset: reorganiz... |
2075 |
/* rebuild sched domains if cpus_allowed has changed */ |
e0e80a02e cpuset: use rebui... |
2076 2077 |
if (cpus_updated) rebuild_sched_domains(); |
b1aac8bb8 [PATCH] cpuset: h... |
2078 |
} |
7ddf96b02 cpusets, hotplug:... |
2079 |
void cpuset_update_active_cpus(bool cpu_online) |
4c4d50f7b [PATCH] cpuset: t... |
2080 |
{ |
3a5a6d0c2 cpuset: don't nes... |
2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 |
/* * 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... |
2093 |
} |
4c4d50f7b [PATCH] cpuset: t... |
2094 |
|
38837fc75 [PATCH] cpuset: t... |
2095 |
/* |
38d7bee9d cpuset: use N_MEM... |
2096 2097 |
* Keep top_cpuset.mems_allowed tracking node_states[N_MEMORY]. * Call this routine anytime after node_states[N_MEMORY] changes. |
a1cd2b13f cpusets: Remove/u... |
2098 |
* See cpuset_update_active_cpus() for CPU hotplug handling. |
38837fc75 [PATCH] cpuset: t... |
2099 |
*/ |
f481891fd cpuset: update to... |
2100 2101 |
static int cpuset_track_online_nodes(struct notifier_block *self, unsigned long action, void *arg) |
38837fc75 [PATCH] cpuset: t... |
2102 |
{ |
3a5a6d0c2 cpuset: don't nes... |
2103 |
schedule_work(&cpuset_hotplug_work); |
f481891fd cpuset: update to... |
2104 |
return NOTIFY_OK; |
38837fc75 [PATCH] cpuset: t... |
2105 |
} |
d8f10cb3d kernel/cpuset.c: ... |
2106 2107 2108 2109 2110 |
static struct notifier_block cpuset_track_online_nodes_nb = { .notifier_call = cpuset_track_online_nodes, .priority = 10, /* ??! */ }; |
38837fc75 [PATCH] cpuset: t... |
2111 |
|
1da177e4c Linux-2.6.12-rc2 |
2112 2113 2114 2115 |
/** * cpuset_init_smp - initialize cpus_allowed * * Description: Finish top cpuset after cpu, node maps are initialized |
d8f10cb3d kernel/cpuset.c: ... |
2116 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
2117 2118 |
void __init cpuset_init_smp(void) { |
6ad4c1888 sched: Fix balanc... |
2119 |
cpumask_copy(top_cpuset.cpus_allowed, cpu_active_mask); |
38d7bee9d cpuset: use N_MEM... |
2120 |
top_cpuset.mems_allowed = node_states[N_MEMORY]; |
33ad801df cpuset: record ol... |
2121 |
top_cpuset.old_mems_allowed = top_cpuset.mems_allowed; |
4c4d50f7b [PATCH] cpuset: t... |
2122 |
|
d8f10cb3d kernel/cpuset.c: ... |
2123 |
register_hotmemory_notifier(&cpuset_track_online_nodes_nb); |
1da177e4c Linux-2.6.12-rc2 |
2124 2125 2126 |
} /** |
1da177e4c Linux-2.6.12-rc2 |
2127 2128 |
* 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... |
2129 |
* @pmask: pointer to struct cpumask variable to receive cpus_allowed set. |
1da177e4c Linux-2.6.12-rc2 |
2130 |
* |
300ed6cbb cpuset: convert c... |
2131 |
* Description: Returns the cpumask_var_t cpus_allowed of the cpuset |
1da177e4c Linux-2.6.12-rc2 |
2132 |
* attached to the specified @tsk. Guaranteed to return some non-empty |
5f054e31c documentation: re... |
2133 |
* subset of cpu_online_mask, even if this means going outside the |
1da177e4c Linux-2.6.12-rc2 |
2134 2135 |
* tasks cpuset. **/ |
6af866af3 cpuset: remove re... |
2136 |
void cpuset_cpus_allowed(struct task_struct *tsk, struct cpumask *pmask) |
1da177e4c Linux-2.6.12-rc2 |
2137 |
{ |
070b57fca cpuset: introduce... |
2138 |
struct cpuset *cpus_cs; |
3d3f26a7b [PATCH] kernel/cp... |
2139 |
mutex_lock(&callback_mutex); |
b8dadcb58 cpuset: use rcu_r... |
2140 |
rcu_read_lock(); |
070b57fca cpuset: introduce... |
2141 2142 |
cpus_cs = effective_cpumask_cpuset(task_cs(tsk)); guarantee_online_cpus(cpus_cs, pmask); |
b8dadcb58 cpuset: use rcu_r... |
2143 |
rcu_read_unlock(); |
897f0b3c3 sched: Kill the b... |
2144 |
mutex_unlock(&callback_mutex); |
1da177e4c Linux-2.6.12-rc2 |
2145 |
} |
2baab4e90 sched: Fix select... |
2146 |
void cpuset_cpus_allowed_fallback(struct task_struct *tsk) |
9084bb824 sched: Make selec... |
2147 |
{ |
c9710d801 cpuset: drop "con... |
2148 |
struct cpuset *cpus_cs; |
9084bb824 sched: Make selec... |
2149 2150 |
rcu_read_lock(); |
070b57fca cpuset: introduce... |
2151 2152 |
cpus_cs = effective_cpumask_cpuset(task_cs(tsk)); do_set_cpus_allowed(tsk, cpus_cs->cpus_allowed); |
9084bb824 sched: Make selec... |
2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 |
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... |
2168 2169 2170 |
* * select_fallback_rq() will fix things ups and set cpu_possible_mask * if required. |
9084bb824 sched: Make selec... |
2171 |
*/ |
9084bb824 sched: Make selec... |
2172 |
} |
1da177e4c Linux-2.6.12-rc2 |
2173 2174 |
void cpuset_init_current_mems_allowed(void) { |
f9a86fcbb cpuset: modify cp... |
2175 |
nodes_setall(current->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
2176 |
} |
d9fd8a6d4 [PATCH] kernel/cp... |
2177 |
/** |
909d75a3b [PATCH] cpuset: i... |
2178 2179 2180 2181 2182 |
* 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... |
2183 |
* subset of node_states[N_MEMORY], even if this means going outside the |
909d75a3b [PATCH] cpuset: i... |
2184 2185 2186 2187 2188 |
* tasks cpuset. **/ nodemask_t cpuset_mems_allowed(struct task_struct *tsk) { |
070b57fca cpuset: introduce... |
2189 |
struct cpuset *mems_cs; |
909d75a3b [PATCH] cpuset: i... |
2190 |
nodemask_t mask; |
3d3f26a7b [PATCH] kernel/cp... |
2191 |
mutex_lock(&callback_mutex); |
b8dadcb58 cpuset: use rcu_r... |
2192 |
rcu_read_lock(); |
070b57fca cpuset: introduce... |
2193 2194 |
mems_cs = effective_nodemask_cpuset(task_cs(tsk)); guarantee_online_mems(mems_cs, &mask); |
b8dadcb58 cpuset: use rcu_r... |
2195 |
rcu_read_unlock(); |
3d3f26a7b [PATCH] kernel/cp... |
2196 |
mutex_unlock(&callback_mutex); |
909d75a3b [PATCH] cpuset: i... |
2197 2198 2199 2200 2201 |
return mask; } /** |
19770b326 mm: filter based ... |
2202 2203 |
* cpuset_nodemask_valid_mems_allowed - check nodemask vs. curremt mems_allowed * @nodemask: the nodemask to be checked |
d9fd8a6d4 [PATCH] kernel/cp... |
2204 |
* |
19770b326 mm: filter based ... |
2205 |
* Are any of the nodes in the nodemask allowed in current->mems_allowed? |
1da177e4c Linux-2.6.12-rc2 |
2206 |
*/ |
19770b326 mm: filter based ... |
2207 |
int cpuset_nodemask_valid_mems_allowed(nodemask_t *nodemask) |
1da177e4c Linux-2.6.12-rc2 |
2208 |
{ |
19770b326 mm: filter based ... |
2209 |
return nodes_intersects(*nodemask, current->mems_allowed); |
1da177e4c Linux-2.6.12-rc2 |
2210 |
} |
9bf2229f8 [PATCH] cpusets: ... |
2211 |
/* |
786083667 Cpuset hardwall f... |
2212 2213 2214 2215 |
* nearest_hardwall_ancestor() - Returns the nearest mem_exclusive or * mem_hardwall ancestor to the specified cpuset. Call holding * callback_mutex. If no ancestor is mem_exclusive or mem_hardwall * (an unusual configuration), then returns the root cpuset. |
9bf2229f8 [PATCH] cpusets: ... |
2216 |
*/ |
c9710d801 cpuset: drop "con... |
2217 |
static struct cpuset *nearest_hardwall_ancestor(struct cpuset *cs) |
9bf2229f8 [PATCH] cpusets: ... |
2218 |
{ |
c431069fe cpuset: remove cp... |
2219 2220 |
while (!(is_mem_exclusive(cs) || is_mem_hardwall(cs)) && parent_cs(cs)) cs = parent_cs(cs); |
9bf2229f8 [PATCH] cpusets: ... |
2221 2222 |
return cs; } |
d9fd8a6d4 [PATCH] kernel/cp... |
2223 |
/** |
a1bc5a4ee cpusets: replace ... |
2224 2225 |
* cpuset_node_allowed_softwall - Can we allocate on a memory node? * @node: is this an allowed node? |
02a0e53d8 [PATCH] cpuset: r... |
2226 |
* @gfp_mask: memory allocation flags |
d9fd8a6d4 [PATCH] kernel/cp... |
2227 |
* |
a1bc5a4ee cpusets: replace ... |
2228 2229 2230 2231 2232 2233 |
* If we're in interrupt, yes, we can always allocate. If __GFP_THISNODE is * set, yes, we can always allocate. If node is in our task's mems_allowed, * yes. If it's not a __GFP_HARDWALL request and this node is in the nearest * hardwalled cpuset ancestor to this task's cpuset, yes. If the task has been * OOM killed and has access to memory reserves as specified by the TIF_MEMDIE * flag, yes. |
9bf2229f8 [PATCH] cpusets: ... |
2234 2235 |
* Otherwise, no. * |
a1bc5a4ee cpusets: replace ... |
2236 2237 2238 |
* If __GFP_HARDWALL is set, cpuset_node_allowed_softwall() reduces to * cpuset_node_allowed_hardwall(). Otherwise, cpuset_node_allowed_softwall() * might sleep, and might allow a node from an enclosing cpuset. |
02a0e53d8 [PATCH] cpuset: r... |
2239 |
* |
a1bc5a4ee cpusets: replace ... |
2240 2241 |
* cpuset_node_allowed_hardwall() only handles the simpler case of hardwall * cpusets, and never sleeps. |
02a0e53d8 [PATCH] cpuset: r... |
2242 2243 2244 2245 2246 2247 2248 |
* * The __GFP_THISNODE placement logic is really handled elsewhere, * by forcibly using a zonelist starting at a specified node, and by * (in get_page_from_freelist()) refusing to consider the zones for * any node on the zonelist except the first. By the time any such * calls get to this routine, we should just shut up and say 'yes'. * |
9bf2229f8 [PATCH] cpusets: ... |
2249 |
* GFP_USER allocations are marked with the __GFP_HARDWALL bit, |
c596d9f32 cpusets: allow TI... |
2250 2251 |
* 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: ... |
2252 |
* GFP_KERNEL allocations are not so marked, so can escape to the |
786083667 Cpuset hardwall f... |
2253 |
* nearest enclosing hardwalled ancestor cpuset. |
9bf2229f8 [PATCH] cpusets: ... |
2254 |
* |
02a0e53d8 [PATCH] cpuset: r... |
2255 2256 2257 2258 2259 2260 2261 |
* Scanning up parent cpusets requires callback_mutex. The * __alloc_pages() routine only calls here with __GFP_HARDWALL bit * _not_ set if it's a GFP_KERNEL allocation, and all nodes in the * current tasks mems_allowed came up empty on the first pass over * the zonelist. So only GFP_KERNEL allocations, if all nodes in the * cpuset are short of memory, might require taking the callback_mutex * mutex. |
9bf2229f8 [PATCH] cpusets: ... |
2262 |
* |
36be57ffe [PATCH] cpuset: u... |
2263 |
* The first call here from mm/page_alloc:get_page_from_freelist() |
02a0e53d8 [PATCH] cpuset: r... |
2264 2265 2266 |
* 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... |
2267 2268 2269 2270 2271 2272 |
* * 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: ... |
2273 2274 |
* in_interrupt - any node ok (current task context irrelevant) * GFP_ATOMIC - any node ok |
c596d9f32 cpusets: allow TI... |
2275 |
* TIF_MEMDIE - any node ok |
786083667 Cpuset hardwall f... |
2276 |
* GFP_KERNEL - any node in enclosing hardwalled cpuset ok |
9bf2229f8 [PATCH] cpusets: ... |
2277 |
* GFP_USER - only nodes in current tasks mems allowed ok. |
36be57ffe [PATCH] cpuset: u... |
2278 2279 |
* * Rule: |
a1bc5a4ee cpusets: replace ... |
2280 |
* Don't call cpuset_node_allowed_softwall if you can't sleep, unless you |
36be57ffe [PATCH] cpuset: u... |
2281 2282 |
* pass in the __GFP_HARDWALL flag set in gfp_flag, which disables * the code that might scan up ancestor cpusets and sleep. |
02a0e53d8 [PATCH] cpuset: r... |
2283 |
*/ |
a1bc5a4ee cpusets: replace ... |
2284 |
int __cpuset_node_allowed_softwall(int node, gfp_t gfp_mask) |
1da177e4c Linux-2.6.12-rc2 |
2285 |
{ |
c9710d801 cpuset: drop "con... |
2286 |
struct cpuset *cs; /* current cpuset ancestors */ |
29afd49b7 [PATCH] cpuset: r... |
2287 |
int allowed; /* is allocation in zone z allowed? */ |
9bf2229f8 [PATCH] cpusets: ... |
2288 |
|
9b819d204 [PATCH] Add __GFP... |
2289 |
if (in_interrupt() || (gfp_mask & __GFP_THISNODE)) |
9bf2229f8 [PATCH] cpusets: ... |
2290 |
return 1; |
92d1dbd27 [PATCH] cpuset: m... |
2291 |
might_sleep_if(!(gfp_mask & __GFP_HARDWALL)); |
9bf2229f8 [PATCH] cpusets: ... |
2292 2293 |
if (node_isset(node, current->mems_allowed)) return 1; |
c596d9f32 cpusets: allow TI... |
2294 2295 2296 2297 2298 2299 |
/* * Allow tasks that have access to memory reserves because they have * been OOM killed to get memory anywhere. */ if (unlikely(test_thread_flag(TIF_MEMDIE))) return 1; |
9bf2229f8 [PATCH] cpusets: ... |
2300 2301 |
if (gfp_mask & __GFP_HARDWALL) /* If hardwall request, stop here */ return 0; |
5563e7707 [PATCH] cpuset: f... |
2302 2303 |
if (current->flags & PF_EXITING) /* Let dying task have memory */ return 1; |
9bf2229f8 [PATCH] cpusets: ... |
2304 |
/* Not hardwall and node outside mems_allowed: scan up cpusets */ |
3d3f26a7b [PATCH] kernel/cp... |
2305 |
mutex_lock(&callback_mutex); |
053199edf [PATCH] cpusets: ... |
2306 |
|
b8dadcb58 cpuset: use rcu_r... |
2307 |
rcu_read_lock(); |
786083667 Cpuset hardwall f... |
2308 |
cs = nearest_hardwall_ancestor(task_cs(current)); |
99afb0fd5 cpuset: fix a rac... |
2309 |
allowed = node_isset(node, cs->mems_allowed); |
b8dadcb58 cpuset: use rcu_r... |
2310 |
rcu_read_unlock(); |
053199edf [PATCH] cpusets: ... |
2311 |
|
3d3f26a7b [PATCH] kernel/cp... |
2312 |
mutex_unlock(&callback_mutex); |
9bf2229f8 [PATCH] cpusets: ... |
2313 |
return allowed; |
1da177e4c Linux-2.6.12-rc2 |
2314 |
} |
02a0e53d8 [PATCH] cpuset: r... |
2315 |
/* |
a1bc5a4ee cpusets: replace ... |
2316 2317 |
* cpuset_node_allowed_hardwall - Can we allocate on a memory node? * @node: is this an allowed node? |
02a0e53d8 [PATCH] cpuset: r... |
2318 2319 |
* @gfp_mask: memory allocation flags * |
a1bc5a4ee cpusets: replace ... |
2320 2321 2322 2323 2324 |
* If we're in interrupt, yes, we can always allocate. If __GFP_THISNODE is * set, yes, we can always allocate. If node is in our task's mems_allowed, * yes. If the task has been OOM killed and has access to memory reserves as * specified by the TIF_MEMDIE flag, yes. * Otherwise, no. |
02a0e53d8 [PATCH] cpuset: r... |
2325 2326 2327 2328 2329 2330 2331 |
* * The __GFP_THISNODE placement logic is really handled elsewhere, * by forcibly using a zonelist starting at a specified node, and by * (in get_page_from_freelist()) refusing to consider the zones for * any node on the zonelist except the first. By the time any such * calls get to this routine, we should just shut up and say 'yes'. * |
a1bc5a4ee cpusets: replace ... |
2332 2333 |
* Unlike the cpuset_node_allowed_softwall() variant, above, * this variant requires that the node be in the current task's |
02a0e53d8 [PATCH] cpuset: r... |
2334 2335 2336 2337 |
* mems_allowed or that we're in interrupt. It does not scan up the * cpuset hierarchy for the nearest enclosing mem_exclusive cpuset. * It never sleeps. */ |
a1bc5a4ee cpusets: replace ... |
2338 |
int __cpuset_node_allowed_hardwall(int node, gfp_t gfp_mask) |
02a0e53d8 [PATCH] cpuset: r... |
2339 |
{ |
02a0e53d8 [PATCH] cpuset: r... |
2340 2341 |
if (in_interrupt() || (gfp_mask & __GFP_THISNODE)) return 1; |
02a0e53d8 [PATCH] cpuset: r... |
2342 2343 |
if (node_isset(node, current->mems_allowed)) return 1; |
dedf8b79e whitespace fixes:... |
2344 2345 2346 2347 2348 2349 |
/* * Allow tasks that have access to memory reserves because they have * been OOM killed to get memory anywhere. */ if (unlikely(test_thread_flag(TIF_MEMDIE))) return 1; |
02a0e53d8 [PATCH] cpuset: r... |
2350 2351 |
return 0; } |
ef08e3b49 [PATCH] cpusets: ... |
2352 |
/** |
6adef3ebe cpusets: new roun... |
2353 2354 |
* 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... |
2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 |
* * 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... |
2378 |
static int cpuset_spread_node(int *rotor) |
825a46af5 [PATCH] cpuset me... |
2379 2380 |
{ int node; |
6adef3ebe cpusets: new roun... |
2381 |
node = next_node(*rotor, current->mems_allowed); |
825a46af5 [PATCH] cpuset me... |
2382 2383 |
if (node == MAX_NUMNODES) node = first_node(current->mems_allowed); |
6adef3ebe cpusets: new roun... |
2384 |
*rotor = node; |
825a46af5 [PATCH] cpuset me... |
2385 2386 |
return node; } |
6adef3ebe cpusets: new roun... |
2387 2388 2389 |
int cpuset_mem_spread_node(void) { |
778d3b0ff cpusets: randomiz... |
2390 2391 2392 |
if (current->cpuset_mem_spread_rotor == NUMA_NO_NODE) current->cpuset_mem_spread_rotor = node_random(¤t->mems_allowed); |
6adef3ebe cpusets: new roun... |
2393 2394 2395 2396 2397 |
return cpuset_spread_node(¤t->cpuset_mem_spread_rotor); } int cpuset_slab_spread_node(void) { |
778d3b0ff cpusets: randomiz... |
2398 2399 2400 |
if (current->cpuset_slab_spread_rotor == NUMA_NO_NODE) current->cpuset_slab_spread_rotor = node_random(¤t->mems_allowed); |
6adef3ebe cpusets: new roun... |
2401 2402 |
return cpuset_spread_node(¤t->cpuset_slab_spread_rotor); } |
825a46af5 [PATCH] cpuset me... |
2403 2404 2405 |
EXPORT_SYMBOL_GPL(cpuset_mem_spread_node); /** |
bbe373f2c oom: compare cpus... |
2406 2407 2408 2409 2410 2411 2412 2413 |
* 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: ... |
2414 |
**/ |
bbe373f2c oom: compare cpus... |
2415 2416 |
int cpuset_mems_allowed_intersects(const struct task_struct *tsk1, const struct task_struct *tsk2) |
ef08e3b49 [PATCH] cpusets: ... |
2417 |
{ |
bbe373f2c oom: compare cpus... |
2418 |
return nodes_intersects(tsk1->mems_allowed, tsk2->mems_allowed); |
ef08e3b49 [PATCH] cpusets: ... |
2419 |
} |
f440d98f8 cpuset: use cgrou... |
2420 |
#define CPUSET_NODELIST_LEN (256) |
75aa19941 oom: print trigge... |
2421 2422 2423 2424 2425 |
/** * cpuset_print_task_mems_allowed - prints task's cpuset and mems_allowed * @task: pointer to task_struct of some task. * * Description: Prints @task's name, cpuset name, and cached copy of its |
b8dadcb58 cpuset: use rcu_r... |
2426 |
* mems_allowed to the kernel log. |
75aa19941 oom: print trigge... |
2427 2428 2429 |
*/ void cpuset_print_task_mems_allowed(struct task_struct *tsk) { |
f440d98f8 cpuset: use cgrou... |
2430 2431 2432 |
/* Statically allocated to prevent using excess stack. */ static char cpuset_nodelist[CPUSET_NODELIST_LEN]; static DEFINE_SPINLOCK(cpuset_buffer_lock); |
b8dadcb58 cpuset: use rcu_r... |
2433 |
struct cgroup *cgrp; |
75aa19941 oom: print trigge... |
2434 |
|
f440d98f8 cpuset: use cgrou... |
2435 |
spin_lock(&cpuset_buffer_lock); |
b8dadcb58 cpuset: use rcu_r... |
2436 |
rcu_read_lock(); |
63f43f55c cpuset: fix cpuse... |
2437 |
|
b8dadcb58 cpuset: use rcu_r... |
2438 |
cgrp = task_cs(tsk)->css.cgroup; |
75aa19941 oom: print trigge... |
2439 2440 |
nodelist_scnprintf(cpuset_nodelist, CPUSET_NODELIST_LEN, tsk->mems_allowed); |
e61734c55 cgroup: remove cg... |
2441 2442 2443 2444 |
printk(KERN_INFO "%s cpuset=", tsk->comm); pr_cont_cgroup_name(cgrp); pr_cont(" mems_allowed=%s ", cpuset_nodelist); |
f440d98f8 cpuset: use cgrou... |
2445 |
|
cfb5966be cpuset: fix RCU l... |
2446 |
rcu_read_unlock(); |
75aa19941 oom: print trigge... |
2447 2448 |
spin_unlock(&cpuset_buffer_lock); } |
1da177e4c Linux-2.6.12-rc2 |
2449 |
/* |
3e0d98b9f [PATCH] cpuset: m... |
2450 2451 2452 2453 |
* 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... |
2454 |
int cpuset_memory_pressure_enabled __read_mostly; |
3e0d98b9f [PATCH] cpuset: m... |
2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 |
/** * 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... |
2476 |
rcu_read_lock(); |
8793d854e Task Control Grou... |
2477 |
fmeter_markevent(&task_cs(current)->fmeter); |
b8dadcb58 cpuset: use rcu_r... |
2478 |
rcu_read_unlock(); |
3e0d98b9f [PATCH] cpuset: m... |
2479 |
} |
8793d854e Task Control Grou... |
2480 |
#ifdef CONFIG_PROC_PID_CPUSET |
3e0d98b9f [PATCH] cpuset: m... |
2481 |
/* |
1da177e4c Linux-2.6.12-rc2 |
2482 2483 2484 |
* proc_cpuset_show() * - Print tasks cpuset path into seq_file. * - Used for /proc/<pid>/cpuset. |
053199edf [PATCH] cpusets: ... |
2485 2486 |
* - 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... |
2487 |
* and we take cpuset_mutex, keeping cpuset_attach() from changing it |
2df167a30 cgroups: update c... |
2488 |
* anyway. |
1da177e4c Linux-2.6.12-rc2 |
2489 |
*/ |
8d8b97ba4 take cgroup_open(... |
2490 |
int proc_cpuset_show(struct seq_file *m, void *unused_v) |
1da177e4c Linux-2.6.12-rc2 |
2491 |
{ |
13b41b094 [PATCH] proc: Use... |
2492 |
struct pid *pid; |
1da177e4c Linux-2.6.12-rc2 |
2493 |
struct task_struct *tsk; |
e61734c55 cgroup: remove cg... |
2494 |
char *buf, *p; |
8793d854e Task Control Grou... |
2495 |
struct cgroup_subsys_state *css; |
99f895518 [PATCH] proc: don... |
2496 |
int retval; |
1da177e4c Linux-2.6.12-rc2 |
2497 |
|
99f895518 [PATCH] proc: don... |
2498 |
retval = -ENOMEM; |
e61734c55 cgroup: remove cg... |
2499 |
buf = kmalloc(PATH_MAX, GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
2500 |
if (!buf) |
99f895518 [PATCH] proc: don... |
2501 2502 2503 |
goto out; retval = -ESRCH; |
13b41b094 [PATCH] proc: Use... |
2504 2505 |
pid = m->private; tsk = get_pid_task(pid, PIDTYPE_PID); |
99f895518 [PATCH] proc: don... |
2506 2507 |
if (!tsk) goto out_free; |
1da177e4c Linux-2.6.12-rc2 |
2508 |
|
e61734c55 cgroup: remove cg... |
2509 |
retval = -ENAMETOOLONG; |
27e89ae5d cpuset: fix RCU l... |
2510 |
rcu_read_lock(); |
073219e99 cgroup: clean up ... |
2511 |
css = task_css(tsk, cpuset_cgrp_id); |
e61734c55 cgroup: remove cg... |
2512 |
p = cgroup_path(css->cgroup, buf, PATH_MAX); |
27e89ae5d cpuset: fix RCU l... |
2513 |
rcu_read_unlock(); |
e61734c55 cgroup: remove cg... |
2514 |
if (!p) |
27e89ae5d cpuset: fix RCU l... |
2515 |
goto out_put_task; |
e61734c55 cgroup: remove cg... |
2516 |
seq_puts(m, p); |
1da177e4c Linux-2.6.12-rc2 |
2517 2518 |
seq_putc(m, ' '); |
e61734c55 cgroup: remove cg... |
2519 |
retval = 0; |
27e89ae5d cpuset: fix RCU l... |
2520 |
out_put_task: |
99f895518 [PATCH] proc: don... |
2521 2522 |
put_task_struct(tsk); out_free: |
1da177e4c Linux-2.6.12-rc2 |
2523 |
kfree(buf); |
99f895518 [PATCH] proc: don... |
2524 |
out: |
1da177e4c Linux-2.6.12-rc2 |
2525 2526 |
return retval; } |
8793d854e Task Control Grou... |
2527 |
#endif /* CONFIG_PROC_PID_CPUSET */ |
1da177e4c Linux-2.6.12-rc2 |
2528 |
|
d01d48278 sched: Always sho... |
2529 |
/* Display task mems_allowed in /proc/<pid>/status file. */ |
df5f8314c proc: seqfile con... |
2530 2531 |
void cpuset_task_status_allowed(struct seq_file *m, struct task_struct *task) { |
df5f8314c proc: seqfile con... |
2532 |
seq_printf(m, "Mems_allowed:\t"); |
30e8e1360 cpuset: use seq_*... |
2533 |
seq_nodemask(m, &task->mems_allowed); |
df5f8314c proc: seqfile con... |
2534 2535 |
seq_printf(m, " "); |
39106dcf8 cpumask: use new ... |
2536 |
seq_printf(m, "Mems_allowed_list:\t"); |
30e8e1360 cpuset: use seq_*... |
2537 |
seq_nodemask_list(m, &task->mems_allowed); |
39106dcf8 cpumask: use new ... |
2538 2539 |
seq_printf(m, " "); |
1da177e4c Linux-2.6.12-rc2 |
2540 |
} |