diff options
Diffstat (limited to 'kernel')
-rw-r--r-- | kernel/cgroup.c | 142 |
1 files changed, 79 insertions, 63 deletions
diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 4766bb65e4d9..36066d8a4911 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c | |||
@@ -113,9 +113,9 @@ static int root_count; | |||
113 | #define dummytop (&rootnode.top_cgroup) | 113 | #define dummytop (&rootnode.top_cgroup) |
114 | 114 | ||
115 | /* This flag indicates whether tasks in the fork and exit paths should | 115 | /* This flag indicates whether tasks in the fork and exit paths should |
116 | * take callback_mutex and check for fork/exit handlers to call. This | 116 | * check for fork/exit handlers to call. This avoids us having to do |
117 | * avoids us having to do extra work in the fork/exit path if none of the | 117 | * extra work in the fork/exit path if none of the subsystems need to |
118 | * subsystems need to be called. | 118 | * be called. |
119 | */ | 119 | */ |
120 | static int need_forkexit_callback; | 120 | static int need_forkexit_callback; |
121 | 121 | ||
@@ -307,7 +307,6 @@ static inline void put_css_set_taskexit(struct css_set *cg) | |||
307 | * template: location in which to build the desired set of subsystem | 307 | * template: location in which to build the desired set of subsystem |
308 | * state objects for the new cgroup group | 308 | * state objects for the new cgroup group |
309 | */ | 309 | */ |
310 | |||
311 | static struct css_set *find_existing_css_set( | 310 | static struct css_set *find_existing_css_set( |
312 | struct css_set *oldcg, | 311 | struct css_set *oldcg, |
313 | struct cgroup *cgrp, | 312 | struct cgroup *cgrp, |
@@ -354,7 +353,6 @@ static struct css_set *find_existing_css_set( | |||
354 | * and chains them on tmp through their cgrp_link_list fields. Returns 0 on | 353 | * and chains them on tmp through their cgrp_link_list fields. Returns 0 on |
355 | * success or a negative error | 354 | * success or a negative error |
356 | */ | 355 | */ |
357 | |||
358 | static int allocate_cg_links(int count, struct list_head *tmp) | 356 | static int allocate_cg_links(int count, struct list_head *tmp) |
359 | { | 357 | { |
360 | struct cg_cgroup_link *link; | 358 | struct cg_cgroup_link *link; |
@@ -396,7 +394,6 @@ static void free_cg_links(struct list_head *tmp) | |||
396 | * substituted into the appropriate hierarchy. Must be called with | 394 | * substituted into the appropriate hierarchy. Must be called with |
397 | * cgroup_mutex held | 395 | * cgroup_mutex held |
398 | */ | 396 | */ |
399 | |||
400 | static struct css_set *find_css_set( | 397 | static struct css_set *find_css_set( |
401 | struct css_set *oldcg, struct cgroup *cgrp) | 398 | struct css_set *oldcg, struct cgroup *cgrp) |
402 | { | 399 | { |
@@ -507,8 +504,8 @@ static struct css_set *find_css_set( | |||
507 | * critical pieces of code here. The exception occurs on cgroup_exit(), | 504 | * critical pieces of code here. The exception occurs on cgroup_exit(), |
508 | * when a task in a notify_on_release cgroup exits. Then cgroup_mutex | 505 | * when a task in a notify_on_release cgroup exits. Then cgroup_mutex |
509 | * is taken, and if the cgroup count is zero, a usermode call made | 506 | * is taken, and if the cgroup count is zero, a usermode call made |
510 | * to /sbin/cgroup_release_agent with the name of the cgroup (path | 507 | * to the release agent with the name of the cgroup (path relative to |
511 | * relative to the root of cgroup file system) as the argument. | 508 | * the root of cgroup file system) as the argument. |
512 | * | 509 | * |
513 | * A cgroup can only be deleted if both its 'count' of using tasks | 510 | * A cgroup can only be deleted if both its 'count' of using tasks |
514 | * is zero, and its list of 'children' cgroups is empty. Since all | 511 | * is zero, and its list of 'children' cgroups is empty. Since all |
@@ -521,7 +518,7 @@ static struct css_set *find_css_set( | |||
521 | * | 518 | * |
522 | * The need for this exception arises from the action of | 519 | * The need for this exception arises from the action of |
523 | * cgroup_attach_task(), which overwrites one tasks cgroup pointer with | 520 | * cgroup_attach_task(), which overwrites one tasks cgroup pointer with |
524 | * another. It does so using cgroup_mutexe, however there are | 521 | * another. It does so using cgroup_mutex, however there are |
525 | * several performance critical places that need to reference | 522 | * several performance critical places that need to reference |
526 | * task->cgroup without the expense of grabbing a system global | 523 | * task->cgroup without the expense of grabbing a system global |
527 | * mutex. Therefore except as noted below, when dereferencing or, as | 524 | * mutex. Therefore except as noted below, when dereferencing or, as |
@@ -537,7 +534,6 @@ static struct css_set *find_css_set( | |||
537 | * cgroup_lock - lock out any changes to cgroup structures | 534 | * cgroup_lock - lock out any changes to cgroup structures |
538 | * | 535 | * |
539 | */ | 536 | */ |
540 | |||
541 | void cgroup_lock(void) | 537 | void cgroup_lock(void) |
542 | { | 538 | { |
543 | mutex_lock(&cgroup_mutex); | 539 | mutex_lock(&cgroup_mutex); |
@@ -548,7 +544,6 @@ void cgroup_lock(void) | |||
548 | * | 544 | * |
549 | * Undo the lock taken in a previous cgroup_lock() call. | 545 | * Undo the lock taken in a previous cgroup_lock() call. |
550 | */ | 546 | */ |
551 | |||
552 | void cgroup_unlock(void) | 547 | void cgroup_unlock(void) |
553 | { | 548 | { |
554 | mutex_unlock(&cgroup_mutex); | 549 | mutex_unlock(&cgroup_mutex); |
@@ -590,7 +585,6 @@ static struct inode *cgroup_new_inode(mode_t mode, struct super_block *sb) | |||
590 | * Call subsys's pre_destroy handler. | 585 | * Call subsys's pre_destroy handler. |
591 | * This is called before css refcnt check. | 586 | * This is called before css refcnt check. |
592 | */ | 587 | */ |
593 | |||
594 | static void cgroup_call_pre_destroy(struct cgroup *cgrp) | 588 | static void cgroup_call_pre_destroy(struct cgroup *cgrp) |
595 | { | 589 | { |
596 | struct cgroup_subsys *ss; | 590 | struct cgroup_subsys *ss; |
@@ -600,7 +594,6 @@ static void cgroup_call_pre_destroy(struct cgroup *cgrp) | |||
600 | return; | 594 | return; |
601 | } | 595 | } |
602 | 596 | ||
603 | |||
604 | static void cgroup_diput(struct dentry *dentry, struct inode *inode) | 597 | static void cgroup_diput(struct dentry *dentry, struct inode *inode) |
605 | { | 598 | { |
606 | /* is dentry a directory ? if so, kfree() associated cgroup */ | 599 | /* is dentry a directory ? if so, kfree() associated cgroup */ |
@@ -1129,8 +1122,13 @@ static inline struct cftype *__d_cft(struct dentry *dentry) | |||
1129 | return dentry->d_fsdata; | 1122 | return dentry->d_fsdata; |
1130 | } | 1123 | } |
1131 | 1124 | ||
1132 | /* | 1125 | /** |
1133 | * Called with cgroup_mutex held. Writes path of cgroup into buf. | 1126 | * cgroup_path - generate the path of a cgroup |
1127 | * @cgrp: the cgroup in question | ||
1128 | * @buf: the buffer to write the path into | ||
1129 | * @buflen: the length of the buffer | ||
1130 | * | ||
1131 | * Called with cgroup_mutex held. Writes path of cgroup into buf. | ||
1134 | * Returns 0 on success, -errno on error. | 1132 | * Returns 0 on success, -errno on error. |
1135 | */ | 1133 | */ |
1136 | int cgroup_path(const struct cgroup *cgrp, char *buf, int buflen) | 1134 | int cgroup_path(const struct cgroup *cgrp, char *buf, int buflen) |
@@ -1188,11 +1186,13 @@ static void get_first_subsys(const struct cgroup *cgrp, | |||
1188 | *subsys_id = test_ss->subsys_id; | 1186 | *subsys_id = test_ss->subsys_id; |
1189 | } | 1187 | } |
1190 | 1188 | ||
1191 | /* | 1189 | /** |
1192 | * Attach task 'tsk' to cgroup 'cgrp' | 1190 | * cgroup_attach_task - attach task 'tsk' to cgroup 'cgrp' |
1191 | * @cgrp: the cgroup the task is attaching to | ||
1192 | * @tsk: the task to be attached | ||
1193 | * | 1193 | * |
1194 | * Call holding cgroup_mutex. May take task_lock of | 1194 | * Call holding cgroup_mutex. May take task_lock of |
1195 | * the task 'pid' during call. | 1195 | * the task 'tsk' during call. |
1196 | */ | 1196 | */ |
1197 | int cgroup_attach_task(struct cgroup *cgrp, struct task_struct *tsk) | 1197 | int cgroup_attach_task(struct cgroup *cgrp, struct task_struct *tsk) |
1198 | { | 1198 | { |
@@ -1293,7 +1293,6 @@ static int attach_task_by_pid(struct cgroup *cgrp, char *pidbuf) | |||
1293 | } | 1293 | } |
1294 | 1294 | ||
1295 | /* The various types of files and directories in a cgroup file system */ | 1295 | /* The various types of files and directories in a cgroup file system */ |
1296 | |||
1297 | enum cgroup_filetype { | 1296 | enum cgroup_filetype { |
1298 | FILE_ROOT, | 1297 | FILE_ROOT, |
1299 | FILE_DIR, | 1298 | FILE_DIR, |
@@ -1584,12 +1583,11 @@ static int cgroup_create_file(struct dentry *dentry, int mode, | |||
1584 | } | 1583 | } |
1585 | 1584 | ||
1586 | /* | 1585 | /* |
1587 | * cgroup_create_dir - create a directory for an object. | 1586 | * cgroup_create_dir - create a directory for an object. |
1588 | * cgrp: the cgroup we create the directory for. | 1587 | * @cgrp: the cgroup we create the directory for. It must have a valid |
1589 | * It must have a valid ->parent field | 1588 | * ->parent field. And we are going to fill its ->dentry field. |
1590 | * And we are going to fill its ->dentry field. | 1589 | * @dentry: dentry of the new cgroup |
1591 | * dentry: dentry of the new cgroup | 1590 | * @mode: mode to set on new directory. |
1592 | * mode: mode to set on new directory. | ||
1593 | */ | 1591 | */ |
1594 | static int cgroup_create_dir(struct cgroup *cgrp, struct dentry *dentry, | 1592 | static int cgroup_create_dir(struct cgroup *cgrp, struct dentry *dentry, |
1595 | int mode) | 1593 | int mode) |
@@ -1651,8 +1649,12 @@ int cgroup_add_files(struct cgroup *cgrp, | |||
1651 | return 0; | 1649 | return 0; |
1652 | } | 1650 | } |
1653 | 1651 | ||
1654 | /* Count the number of tasks in a cgroup. */ | 1652 | /** |
1655 | 1653 | * cgroup_task_count - count the number of tasks in a cgroup. | |
1654 | * @cgrp: the cgroup in question | ||
1655 | * | ||
1656 | * Return the number of tasks in the cgroup. | ||
1657 | */ | ||
1656 | int cgroup_task_count(const struct cgroup *cgrp) | 1658 | int cgroup_task_count(const struct cgroup *cgrp) |
1657 | { | 1659 | { |
1658 | int count = 0; | 1660 | int count = 0; |
@@ -1962,12 +1964,13 @@ static int pid_array_load(pid_t *pidarray, int npids, struct cgroup *cgrp) | |||
1962 | } | 1964 | } |
1963 | 1965 | ||
1964 | /** | 1966 | /** |
1965 | * Build and fill cgroupstats so that taskstats can export it to user | 1967 | * cgroupstats_build - build and fill cgroupstats |
1966 | * space. | ||
1967 | * | ||
1968 | * @stats: cgroupstats to fill information into | 1968 | * @stats: cgroupstats to fill information into |
1969 | * @dentry: A dentry entry belonging to the cgroup for which stats have | 1969 | * @dentry: A dentry entry belonging to the cgroup for which stats have |
1970 | * been requested. | 1970 | * been requested. |
1971 | * | ||
1972 | * Build and fill cgroupstats so that taskstats can export it to user | ||
1973 | * space. | ||
1971 | */ | 1974 | */ |
1972 | int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry) | 1975 | int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry) |
1973 | { | 1976 | { |
@@ -2199,14 +2202,13 @@ static void init_cgroup_css(struct cgroup_subsys_state *css, | |||
2199 | } | 2202 | } |
2200 | 2203 | ||
2201 | /* | 2204 | /* |
2202 | * cgroup_create - create a cgroup | 2205 | * cgroup_create - create a cgroup |
2203 | * parent: cgroup that will be parent of the new cgroup. | 2206 | * @parent: cgroup that will be parent of the new cgroup |
2204 | * name: name of the new cgroup. Will be strcpy'ed. | 2207 | * @dentry: dentry of the new cgroup |
2205 | * mode: mode to set on new inode | 2208 | * @mode: mode to set on new inode |
2206 | * | 2209 | * |
2207 | * Must be called with the mutex on the parent inode held | 2210 | * Must be called with the mutex on the parent inode held |
2208 | */ | 2211 | */ |
2209 | |||
2210 | static long cgroup_create(struct cgroup *parent, struct dentry *dentry, | 2212 | static long cgroup_create(struct cgroup *parent, struct dentry *dentry, |
2211 | int mode) | 2213 | int mode) |
2212 | { | 2214 | { |
@@ -2349,13 +2351,12 @@ static int cgroup_rmdir(struct inode *unused_dir, struct dentry *dentry) | |||
2349 | parent = cgrp->parent; | 2351 | parent = cgrp->parent; |
2350 | root = cgrp->root; | 2352 | root = cgrp->root; |
2351 | sb = root->sb; | 2353 | sb = root->sb; |
2354 | |||
2352 | /* | 2355 | /* |
2353 | * Call pre_destroy handlers of subsys | 2356 | * Call pre_destroy handlers of subsys. Notify subsystems |
2357 | * that rmdir() request comes. | ||
2354 | */ | 2358 | */ |
2355 | cgroup_call_pre_destroy(cgrp); | 2359 | cgroup_call_pre_destroy(cgrp); |
2356 | /* | ||
2357 | * Notify subsyses that rmdir() request comes. | ||
2358 | */ | ||
2359 | 2360 | ||
2360 | if (cgroup_has_css_refs(cgrp)) { | 2361 | if (cgroup_has_css_refs(cgrp)) { |
2361 | mutex_unlock(&cgroup_mutex); | 2362 | mutex_unlock(&cgroup_mutex); |
@@ -2431,8 +2432,10 @@ static void cgroup_init_subsys(struct cgroup_subsys *ss) | |||
2431 | } | 2432 | } |
2432 | 2433 | ||
2433 | /** | 2434 | /** |
2434 | * cgroup_init_early - initialize cgroups at system boot, and | 2435 | * cgroup_init_early - cgroup initialization at system boot |
2435 | * initialize any subsystems that request early init. | 2436 | * |
2437 | * Initialize cgroups at system boot, and initialize any | ||
2438 | * subsystems that request early init. | ||
2436 | */ | 2439 | */ |
2437 | int __init cgroup_init_early(void) | 2440 | int __init cgroup_init_early(void) |
2438 | { | 2441 | { |
@@ -2474,8 +2477,10 @@ int __init cgroup_init_early(void) | |||
2474 | } | 2477 | } |
2475 | 2478 | ||
2476 | /** | 2479 | /** |
2477 | * cgroup_init - register cgroup filesystem and /proc file, and | 2480 | * cgroup_init - cgroup initialization |
2478 | * initialize any subsystems that didn't request early init. | 2481 | * |
2482 | * Register cgroup filesystem and /proc file, and initialize | ||
2483 | * any subsystems that didn't request early init. | ||
2479 | */ | 2484 | */ |
2480 | int __init cgroup_init(void) | 2485 | int __init cgroup_init(void) |
2481 | { | 2486 | { |
@@ -2618,7 +2623,7 @@ static struct file_operations proc_cgroupstats_operations = { | |||
2618 | 2623 | ||
2619 | /** | 2624 | /** |
2620 | * cgroup_fork - attach newly forked task to its parents cgroup. | 2625 | * cgroup_fork - attach newly forked task to its parents cgroup. |
2621 | * @tsk: pointer to task_struct of forking parent process. | 2626 | * @child: pointer to task_struct of forking parent process. |
2622 | * | 2627 | * |
2623 | * Description: A task inherits its parent's cgroup at fork(). | 2628 | * Description: A task inherits its parent's cgroup at fork(). |
2624 | * | 2629 | * |
@@ -2642,9 +2647,12 @@ void cgroup_fork(struct task_struct *child) | |||
2642 | } | 2647 | } |
2643 | 2648 | ||
2644 | /** | 2649 | /** |
2645 | * cgroup_fork_callbacks - called on a new task very soon before | 2650 | * cgroup_fork_callbacks - run fork callbacks |
2646 | * adding it to the tasklist. No need to take any locks since no-one | 2651 | * @child: the new task |
2647 | * can be operating on this task | 2652 | * |
2653 | * Called on a new task very soon before adding it to the | ||
2654 | * tasklist. No need to take any locks since no-one can | ||
2655 | * be operating on this task. | ||
2648 | */ | 2656 | */ |
2649 | void cgroup_fork_callbacks(struct task_struct *child) | 2657 | void cgroup_fork_callbacks(struct task_struct *child) |
2650 | { | 2658 | { |
@@ -2659,11 +2667,14 @@ void cgroup_fork_callbacks(struct task_struct *child) | |||
2659 | } | 2667 | } |
2660 | 2668 | ||
2661 | /** | 2669 | /** |
2662 | * cgroup_post_fork - called on a new task after adding it to the | 2670 | * cgroup_post_fork - called on a new task after adding it to the task list |
2663 | * task list. Adds the task to the list running through its css_set | 2671 | * @child: the task in question |
2664 | * if necessary. Has to be after the task is visible on the task list | 2672 | * |
2665 | * in case we race with the first call to cgroup_iter_start() - to | 2673 | * Adds the task to the list running through its css_set if necessary. |
2666 | * guarantee that the new task ends up on its list. */ | 2674 | * Has to be after the task is visible on the task list in case we race |
2675 | * with the first call to cgroup_iter_start() - to guarantee that the | ||
2676 | * new task ends up on its list. | ||
2677 | */ | ||
2667 | void cgroup_post_fork(struct task_struct *child) | 2678 | void cgroup_post_fork(struct task_struct *child) |
2668 | { | 2679 | { |
2669 | if (use_task_css_set_links) { | 2680 | if (use_task_css_set_links) { |
@@ -2676,6 +2687,7 @@ void cgroup_post_fork(struct task_struct *child) | |||
2676 | /** | 2687 | /** |
2677 | * cgroup_exit - detach cgroup from exiting task | 2688 | * cgroup_exit - detach cgroup from exiting task |
2678 | * @tsk: pointer to task_struct of exiting process | 2689 | * @tsk: pointer to task_struct of exiting process |
2690 | * @run_callback: run exit callbacks? | ||
2679 | * | 2691 | * |
2680 | * Description: Detach cgroup from @tsk and release it. | 2692 | * Description: Detach cgroup from @tsk and release it. |
2681 | * | 2693 | * |
@@ -2706,7 +2718,6 @@ void cgroup_post_fork(struct task_struct *child) | |||
2706 | * top_cgroup isn't going away, and either task has PF_EXITING set, | 2718 | * top_cgroup isn't going away, and either task has PF_EXITING set, |
2707 | * which wards off any cgroup_attach_task() attempts, or task is a failed | 2719 | * which wards off any cgroup_attach_task() attempts, or task is a failed |
2708 | * fork, never visible to cgroup_attach_task. | 2720 | * fork, never visible to cgroup_attach_task. |
2709 | * | ||
2710 | */ | 2721 | */ |
2711 | void cgroup_exit(struct task_struct *tsk, int run_callbacks) | 2722 | void cgroup_exit(struct task_struct *tsk, int run_callbacks) |
2712 | { | 2723 | { |
@@ -2743,9 +2754,13 @@ void cgroup_exit(struct task_struct *tsk, int run_callbacks) | |||
2743 | } | 2754 | } |
2744 | 2755 | ||
2745 | /** | 2756 | /** |
2746 | * cgroup_clone - duplicate the current cgroup in the hierarchy | 2757 | * cgroup_clone - clone the cgroup the given subsystem is attached to |
2747 | * that the given subsystem is attached to, and move this task into | 2758 | * @tsk: the task to be moved |
2748 | * the new child | 2759 | * @subsys: the given subsystem |
2760 | * | ||
2761 | * Duplicate the current cgroup in the hierarchy that the given | ||
2762 | * subsystem is attached to, and move this task into the new | ||
2763 | * child. | ||
2749 | */ | 2764 | */ |
2750 | int cgroup_clone(struct task_struct *tsk, struct cgroup_subsys *subsys) | 2765 | int cgroup_clone(struct task_struct *tsk, struct cgroup_subsys *subsys) |
2751 | { | 2766 | { |
@@ -2858,9 +2873,12 @@ int cgroup_clone(struct task_struct *tsk, struct cgroup_subsys *subsys) | |||
2858 | return ret; | 2873 | return ret; |
2859 | } | 2874 | } |
2860 | 2875 | ||
2861 | /* | 2876 | /** |
2862 | * See if "cgrp" is a descendant of the current task's cgroup in | 2877 | * cgroup_is_descendant - see if @cgrp is a descendant of current task's cgrp |
2863 | * the appropriate hierarchy | 2878 | * @cgrp: the cgroup in question |
2879 | * | ||
2880 | * See if @cgrp is a descendant of the current task's cgroup in | ||
2881 | * the appropriate hierarchy. | ||
2864 | * | 2882 | * |
2865 | * If we are sending in dummytop, then presumably we are creating | 2883 | * If we are sending in dummytop, then presumably we are creating |
2866 | * the top cgroup in the subsystem. | 2884 | * the top cgroup in the subsystem. |
@@ -2939,9 +2957,7 @@ void __css_put(struct cgroup_subsys_state *css) | |||
2939 | * release agent task. We don't bother to wait because the caller of | 2957 | * release agent task. We don't bother to wait because the caller of |
2940 | * this routine has no use for the exit status of the release agent | 2958 | * this routine has no use for the exit status of the release agent |
2941 | * task, so no sense holding our caller up for that. | 2959 | * task, so no sense holding our caller up for that. |
2942 | * | ||
2943 | */ | 2960 | */ |
2944 | |||
2945 | static void cgroup_release_agent(struct work_struct *work) | 2961 | static void cgroup_release_agent(struct work_struct *work) |
2946 | { | 2962 | { |
2947 | BUG_ON(work != &release_agent_work); | 2963 | BUG_ON(work != &release_agent_work); |