From 2468c7234b366eeb799ee0648cb58f9cba394a54 Mon Sep 17 00:00:00 2001 From: Daisuke Nishimura Date: Wed, 10 Mar 2010 15:22:03 -0800 Subject: cgroup: introduce cancel_attach() Add cancel_attach() operation to struct cgroup_subsys. cancel_attach() can be used when can_attach() operation prepares something for the subsys, but we should rollback what can_attach() operation has prepared if attach task fails after we've succeeded in can_attach(). Signed-off-by: Daisuke Nishimura Acked-by: Li Zefan Reviewed-by: Paul Menage Cc: Balbir Singh Acked-by: KAMEZAWA Hiroyuki Cc: Daisuke Nishimura Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/cgroup.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux/cgroup.h') diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index c9bbcb2a75ae..d08cfe7e12e5 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -428,6 +428,8 @@ struct cgroup_subsys { void (*destroy)(struct cgroup_subsys *ss, struct cgroup *cgrp); int (*can_attach)(struct cgroup_subsys *ss, struct cgroup *cgrp, struct task_struct *tsk, bool threadgroup); + void (*cancel_attach)(struct cgroup_subsys *ss, struct cgroup *cgrp, + struct task_struct *tsk, bool threadgroup); void (*attach)(struct cgroup_subsys *ss, struct cgroup *cgrp, struct cgroup *old_cgrp, struct task_struct *tsk, bool threadgroup); -- cgit v1.2.3-71-gd317 From d7b9fff711d5e8db8c844161c684017e556c38a0 Mon Sep 17 00:00:00 2001 From: Daisuke Nishimura Date: Wed, 10 Mar 2010 15:22:05 -0800 Subject: cgroup: introduce coalesce css_get() and css_put() Current css_get() and css_put() increment/decrement css->refcnt one by one. This patch add a new function __css_get(), which takes "count" as a arg and increment the css->refcnt by "count". And this patch also add a new arg("count") to __css_put() and change the function to decrement the css->refcnt by "count". These coalesce version of __css_get()/__css_put() will be used to improve performance of memcg's moving charge feature later, where instead of calling css_get()/css_put() repeatedly, these new functions will be used. No change is needed for current users of css_get()/css_put(). Signed-off-by: Daisuke Nishimura Acked-by: Paul Menage Cc: Balbir Singh Acked-by: KAMEZAWA Hiroyuki Cc: Li Zefan Cc: Daisuke Nishimura Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/cgroup.h | 12 +++++++++--- kernel/cgroup.c | 5 +++-- 2 files changed, 12 insertions(+), 5 deletions(-) (limited to 'include/linux/cgroup.h') diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index d08cfe7e12e5..14160b5b693f 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -76,6 +76,12 @@ enum { CSS_REMOVED, /* This CSS is dead */ }; +/* Caller must verify that the css is not for root cgroup */ +static inline void __css_get(struct cgroup_subsys_state *css, int count) +{ + atomic_add(count, &css->refcnt); +} + /* * Call css_get() to hold a reference on the css; it can be used * for a reference obtained via: @@ -87,7 +93,7 @@ static inline void css_get(struct cgroup_subsys_state *css) { /* We don't need to reference count the root state */ if (!test_bit(CSS_ROOT, &css->flags)) - atomic_inc(&css->refcnt); + __css_get(css, 1); } static inline bool css_is_removed(struct cgroup_subsys_state *css) @@ -118,11 +124,11 @@ static inline bool css_tryget(struct cgroup_subsys_state *css) * css_get() or css_tryget() */ -extern void __css_put(struct cgroup_subsys_state *css); +extern void __css_put(struct cgroup_subsys_state *css, int count); static inline void css_put(struct cgroup_subsys_state *css) { if (!test_bit(CSS_ROOT, &css->flags)) - __css_put(css); + __css_put(css, 1); } /* bits in struct cgroup flags field */ diff --git a/kernel/cgroup.c b/kernel/cgroup.c index be45d2f6008a..cace83ddbcdc 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -3746,12 +3746,13 @@ static void check_for_release(struct cgroup *cgrp) } } -void __css_put(struct cgroup_subsys_state *css) +/* Caller must verify that the css is not for root cgroup */ +void __css_put(struct cgroup_subsys_state *css, int count) { struct cgroup *cgrp = css->cgroup; int val; rcu_read_lock(); - val = atomic_dec_return(&css->refcnt); + val = atomic_sub_return(count, &css->refcnt); if (val == 1) { if (notify_on_release(cgrp)) { set_bit(CGRP_RELEASABLE, &cgrp->flags); -- cgit v1.2.3-71-gd317 From aae8aab40367036931608fdaf9e2dc568b516f19 Mon Sep 17 00:00:00 2001 From: Ben Blum Date: Wed, 10 Mar 2010 15:22:07 -0800 Subject: cgroups: revamp subsys array This patch series provides the ability for cgroup subsystems to be compiled as modules both within and outside the kernel tree. This is mainly useful for classifiers and subsystems that hook into components that are already modules. cls_cgroup and blkio-cgroup serve as the example use cases for this feature. It provides an interface cgroup_load_subsys() and cgroup_unload_subsys() which modular subsystems can use to register and depart during runtime. The net_cls classifier subsystem serves as the example for a subsystem which can be converted into a module using these changes. Patch #1 sets up the subsys[] array so its contents can be dynamic as modules appear and (eventually) disappear. Iterations over the array are modified to handle when subsystems are absent, and the dynamic section of the array is protected by cgroup_mutex. Patch #2 implements an interface for modules to load subsystems, called cgroup_load_subsys, similar to cgroup_init_subsys, and adds a module pointer in struct cgroup_subsys. Patch #3 adds a mechanism for unloading modular subsystems, which includes a more advanced rework of the rudimentary reference counting introduced in patch 2. Patch #4 modifies the net_cls subsystem, which already had some module declarations, to be configurable as a module, which also serves as a simple proof-of-concept. Part of implementing patches 2 and 4 involved updating css pointers in each css_set when the module appears or leaves. In doing this, it was discovered that css_sets always remain linked to the dummy cgroup, regardless of whether or not any subsystems are actually bound to it (i.e., not mounted on an actual hierarchy). The subsystem loading and unloading code therefore should keep in mind the special cases where the added subsystem is the only one in the dummy cgroup (and therefore all css_sets need to be linked back into it) and where the removed subsys was the only one in the dummy cgroup (and therefore all css_sets should be unlinked from it) - however, as all css_sets always stay attached to the dummy cgroup anyway, these cases are ignored. Any fix that addresses this issue should also make sure these cases are addressed in the subsystem loading and unloading code. This patch: Make subsys[] able to be dynamically populated to support modular subsystems This patch reworks the way the subsys[] array is used so that subsystems can register themselves after boot time, and enables the internals of cgroups to be able to handle when subsystems are not present or may appear/disappear. Signed-off-by: Ben Blum Acked-by: Li Zefan Cc: Paul Menage Cc: "David S. Miller" Cc: KAMEZAWA Hiroyuki Cc: Lai Jiangshan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/cgroup.h | 10 ++++-- kernel/cgroup.c | 96 +++++++++++++++++++++++++++++++++++++++++--------- 2 files changed, 88 insertions(+), 18 deletions(-) (limited to 'include/linux/cgroup.h') diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index 14160b5b693f..28319a9fe569 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -40,13 +40,19 @@ extern int cgroupstats_build(struct cgroupstats *stats, extern const struct file_operations proc_cgroup_operations; -/* Define the enumeration of all cgroup subsystems */ +/* Define the enumeration of all builtin cgroup subsystems */ #define SUBSYS(_x) _x ## _subsys_id, enum cgroup_subsys_id { #include - CGROUP_SUBSYS_COUNT + CGROUP_BUILTIN_SUBSYS_COUNT }; #undef SUBSYS +/* + * This define indicates the maximum number of subsystems that can be loaded + * at once. We limit to this many since cgroupfs_root has subsys_bits to keep + * track of all of them. + */ +#define CGROUP_SUBSYS_COUNT (BITS_PER_BYTE*sizeof(unsigned long)) /* Per-subsystem/per-cgroup state maintained by the system. */ struct cgroup_subsys_state { diff --git a/kernel/cgroup.c b/kernel/cgroup.c index cace83ddbcdc..c92fb9549358 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -57,10 +57,14 @@ static DEFINE_MUTEX(cgroup_mutex); -/* Generate an array of cgroup subsystem pointers */ +/* + * Generate an array of cgroup subsystem pointers. At boot time, this is + * populated up to CGROUP_BUILTIN_SUBSYS_COUNT, and modular subsystems are + * registered after that. The mutable section of this array is protected by + * cgroup_mutex. + */ #define SUBSYS(_x) &_x ## _subsys, - -static struct cgroup_subsys *subsys[] = { +static struct cgroup_subsys *subsys[CGROUP_SUBSYS_COUNT] = { #include }; @@ -448,8 +452,11 @@ static struct css_set *find_existing_css_set( struct hlist_node *node; struct css_set *cg; - /* Built the set of subsystem state objects that we want to - * see in the new css_set */ + /* + * Build the set of subsystem state objects that we want to see in the + * new css_set. while subsystems can change globally, the entries here + * won't change, so no need for locking. + */ for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { if (root->subsys_bits & (1UL << i)) { /* Subsystem is in this hierarchy. So we want @@ -884,7 +891,9 @@ void cgroup_release_and_wakeup_rmdir(struct cgroup_subsys_state *css) css_put(css); } - +/* + * Call with cgroup_mutex held. + */ static int rebind_subsystems(struct cgroupfs_root *root, unsigned long final_bits) { @@ -892,6 +901,8 @@ static int rebind_subsystems(struct cgroupfs_root *root, struct cgroup *cgrp = &root->top_cgroup; int i; + BUG_ON(!mutex_is_locked(&cgroup_mutex)); + removed_bits = root->actual_subsys_bits & ~final_bits; added_bits = final_bits & ~root->actual_subsys_bits; /* Check that any added subsystems are currently free */ @@ -900,6 +911,12 @@ static int rebind_subsystems(struct cgroupfs_root *root, struct cgroup_subsys *ss = subsys[i]; if (!(bit & added_bits)) continue; + /* + * Nobody should tell us to do a subsys that doesn't exist: + * parse_cgroupfs_options should catch that case and refcounts + * ensure that subsystems won't disappear once selected. + */ + BUG_ON(ss == NULL); if (ss->root != &rootnode) { /* Subsystem isn't free */ return -EBUSY; @@ -919,6 +936,7 @@ static int rebind_subsystems(struct cgroupfs_root *root, unsigned long bit = 1UL << i; if (bit & added_bits) { /* We're binding this subsystem to this hierarchy */ + BUG_ON(ss == NULL); BUG_ON(cgrp->subsys[i]); BUG_ON(!dummytop->subsys[i]); BUG_ON(dummytop->subsys[i]->cgroup != dummytop); @@ -932,6 +950,7 @@ static int rebind_subsystems(struct cgroupfs_root *root, mutex_unlock(&ss->hierarchy_mutex); } else if (bit & removed_bits) { /* We're removing this subsystem */ + BUG_ON(ss == NULL); BUG_ON(cgrp->subsys[i] != dummytop->subsys[i]); BUG_ON(cgrp->subsys[i]->cgroup != cgrp); mutex_lock(&ss->hierarchy_mutex); @@ -944,6 +963,7 @@ static int rebind_subsystems(struct cgroupfs_root *root, mutex_unlock(&ss->hierarchy_mutex); } else if (bit & final_bits) { /* Subsystem state should already exist */ + BUG_ON(ss == NULL); BUG_ON(!cgrp->subsys[i]); } else { /* Subsystem state shouldn't exist */ @@ -986,14 +1006,18 @@ struct cgroup_sb_opts { }; -/* Convert a hierarchy specifier into a bitmask of subsystems and - * flags. */ +/* + * Convert a hierarchy specifier into a bitmask of subsystems and flags. Call + * with cgroup_mutex held to protect the subsys[] array. + */ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) { char *token, *o = data ?: "all"; unsigned long mask = (unsigned long)-1; + BUG_ON(!mutex_is_locked(&cgroup_mutex)); + #ifdef CONFIG_CPUSETS mask = ~(1UL << cpuset_subsys_id); #endif @@ -1009,6 +1033,8 @@ static int parse_cgroupfs_options(char *data, opts->subsys_bits = 0; for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; + if (ss == NULL) + continue; if (!ss->disabled) opts->subsys_bits |= 1ul << i; } @@ -1053,6 +1079,8 @@ static int parse_cgroupfs_options(char *data, int i; for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { ss = subsys[i]; + if (ss == NULL) + continue; if (!strcmp(token, ss->name)) { if (!ss->disabled) set_bit(i, &opts->subsys_bits); @@ -1306,7 +1334,9 @@ static int cgroup_get_sb(struct file_system_type *fs_type, struct cgroupfs_root *new_root; /* First find the desired set of subsystems */ + mutex_lock(&cgroup_mutex); ret = parse_cgroupfs_options(data, &opts); + mutex_unlock(&cgroup_mutex); if (ret) goto out_err; @@ -2918,8 +2948,14 @@ static void cgroup_lock_hierarchy(struct cgroupfs_root *root) /* We need to take each hierarchy_mutex in a consistent order */ int i; + /* + * No worry about a race with rebind_subsystems that might mess up the + * locking order, since both parties are under cgroup_mutex. + */ for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; + if (ss == NULL) + continue; if (ss->root == root) mutex_lock(&ss->hierarchy_mutex); } @@ -2931,6 +2967,8 @@ static void cgroup_unlock_hierarchy(struct cgroupfs_root *root) for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; + if (ss == NULL) + continue; if (ss->root == root) mutex_unlock(&ss->hierarchy_mutex); } @@ -3054,11 +3092,16 @@ static int cgroup_has_css_refs(struct cgroup *cgrp) * synchronization other than RCU, and the subsystem linked * list isn't RCU-safe */ int i; + /* + * We won't need to lock the subsys array, because the subsystems + * we're concerned about aren't going anywhere since our cgroup root + * has a reference on them. + */ for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; struct cgroup_subsys_state *css; - /* Skip subsystems not in this hierarchy */ - if (ss->root != cgrp->root) + /* Skip subsystems not present or not in this hierarchy */ + if (ss == NULL || ss->root != cgrp->root) continue; css = cgrp->subsys[ss->subsys_id]; /* When called from check_for_release() it's possible @@ -3279,7 +3322,8 @@ int __init cgroup_init_early(void) for (i = 0; i < CSS_SET_TABLE_SIZE; i++) INIT_HLIST_HEAD(&css_set_table[i]); - for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + /* at bootup time, we don't worry about modular subsystems */ + for (i = 0; i < CGROUP_BUILTIN_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; BUG_ON(!ss->name); @@ -3314,7 +3358,8 @@ int __init cgroup_init(void) if (err) return err; - for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + /* at bootup time, we don't worry about modular subsystems */ + for (i = 0; i < CGROUP_BUILTIN_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; if (!ss->early_init) cgroup_init_subsys(ss); @@ -3423,9 +3468,16 @@ static int proc_cgroupstats_show(struct seq_file *m, void *v) int i; seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n"); + /* + * ideally we don't want subsystems moving around while we do this. + * cgroup_mutex is also necessary to guarantee an atomic snapshot of + * subsys/hierarchy state. + */ mutex_lock(&cgroup_mutex); for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; + if (ss == NULL) + continue; seq_printf(m, "%s\t%d\t%d\t%d\n", ss->name, ss->root->hierarchy_id, ss->root->number_of_cgroups, !ss->disabled); @@ -3483,7 +3535,12 @@ void cgroup_fork_callbacks(struct task_struct *child) { if (need_forkexit_callback) { int i; - for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + /* + * forkexit callbacks are only supported for builtin + * subsystems, and the builtin section of the subsys array is + * immutable, so we don't need to lock the subsys array here. + */ + for (i = 0; i < CGROUP_BUILTIN_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; if (ss->fork) ss->fork(ss, child); @@ -3552,7 +3609,11 @@ void cgroup_exit(struct task_struct *tsk, int run_callbacks) struct css_set *cg; if (run_callbacks && need_forkexit_callback) { - for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + /* + * modular subsystems can't use callbacks, so no need to lock + * the subsys array + */ + for (i = 0; i < CGROUP_BUILTIN_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; if (ss->exit) ss->exit(ss, tsk); @@ -3844,8 +3905,11 @@ static int __init cgroup_disable(char *str) while ((token = strsep(&str, ",")) != NULL) { if (!*token) continue; - - for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + /* + * cgroup_disable, being at boot time, can't know about module + * subsystems, so we don't worry about them. + */ + for (i = 0; i < CGROUP_BUILTIN_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; if (!strcmp(token, ss->name)) { -- cgit v1.2.3-71-gd317 From e6a1105ba08b265023dd71a4174fb4a29ebc7083 Mon Sep 17 00:00:00 2001 From: Ben Blum Date: Wed, 10 Mar 2010 15:22:09 -0800 Subject: cgroups: subsystem module loading interface Add interface between cgroups subsystem management and module loading This patch implements rudimentary module-loading support for cgroups - namely, a cgroup_load_subsys (similar to cgroup_init_subsys) for use as a module initcall, and a struct module pointer in struct cgroup_subsys. Several functions that might be wanted by modules have had EXPORT_SYMBOL added to them, but it's unclear exactly which functions want it and which won't. Signed-off-by: Ben Blum Acked-by: Li Zefan Cc: Paul Menage Cc: "David S. Miller" Cc: KAMEZAWA Hiroyuki Cc: Lai Jiangshan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- Documentation/cgroups/cgroups.txt | 4 + include/linux/cgroup.h | 4 + kernel/cgroup.c | 150 ++++++++++++++++++++++++++++++++++++-- 3 files changed, 153 insertions(+), 5 deletions(-) (limited to 'include/linux/cgroup.h') diff --git a/Documentation/cgroups/cgroups.txt b/Documentation/cgroups/cgroups.txt index d45082653e3d..ae8a037a761e 100644 --- a/Documentation/cgroups/cgroups.txt +++ b/Documentation/cgroups/cgroups.txt @@ -488,6 +488,10 @@ Each subsystem should: - add an entry in linux/cgroup_subsys.h - define a cgroup_subsys object called _subsys +If a subsystem can be compiled as a module, it should also have in its +module initcall a call to cgroup_load_subsys(&its_subsys_struct). It +should also set its_subsys.module = THIS_MODULE in its .c file. + Each subsystem may export the following methods. The only mandatory methods are create/destroy. Any others that are null are presumed to be successful no-ops. diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index 28319a9fe569..402ce477c47e 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -37,6 +37,7 @@ extern void cgroup_post_fork(struct task_struct *p); extern void cgroup_exit(struct task_struct *p, int run_callbacks); extern int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry); +extern int cgroup_load_subsys(struct cgroup_subsys *ss); extern const struct file_operations proc_cgroup_operations; @@ -486,6 +487,9 @@ struct cgroup_subsys { /* used when use_id == true */ struct idr idr; spinlock_t id_lock; + + /* should be defined only by modular subsystems */ + struct module *module; }; #define SUBSYS(_x) extern struct cgroup_subsys _x ## _subsys; diff --git a/kernel/cgroup.c b/kernel/cgroup.c index c92fb9549358..2cae38e64c59 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -44,6 +44,7 @@ #include #include #include +#include #include #include #include @@ -254,7 +255,8 @@ struct cg_cgroup_link { static struct css_set init_css_set; static struct cg_cgroup_link init_css_set_link; -static int cgroup_subsys_init_idr(struct cgroup_subsys *ss); +static int cgroup_init_idr(struct cgroup_subsys *ss, + struct cgroup_subsys_state *css); /* css_set_lock protects the list of css_set objects, and the * chain of tasks off each css_set. Nests outside task->alloc_lock @@ -2125,6 +2127,7 @@ int cgroup_add_file(struct cgroup *cgrp, error = PTR_ERR(dentry); return error; } +EXPORT_SYMBOL_GPL(cgroup_add_file); int cgroup_add_files(struct cgroup *cgrp, struct cgroup_subsys *subsys, @@ -2139,6 +2142,7 @@ int cgroup_add_files(struct cgroup *cgrp, } return 0; } +EXPORT_SYMBOL_GPL(cgroup_add_files); /** * cgroup_task_count - count the number of tasks in a cgroup. @@ -3292,7 +3296,144 @@ static void __init cgroup_init_subsys(struct cgroup_subsys *ss) mutex_init(&ss->hierarchy_mutex); lockdep_set_class(&ss->hierarchy_mutex, &ss->subsys_key); ss->active = 1; + + /* this function shouldn't be used with modular subsystems, since they + * need to register a subsys_id, among other things */ + BUG_ON(ss->module); +} + +/** + * cgroup_load_subsys: load and register a modular subsystem at runtime + * @ss: the subsystem to load + * + * This function should be called in a modular subsystem's initcall. If the + * subsytem is built as a module, it will be assigned a new subsys_id and set + * up for use. If the subsystem is built-in anyway, work is delegated to the + * simpler cgroup_init_subsys. + */ +int __init_or_module cgroup_load_subsys(struct cgroup_subsys *ss) +{ + int i; + struct cgroup_subsys_state *css; + + /* check name and function validity */ + if (ss->name == NULL || strlen(ss->name) > MAX_CGROUP_TYPE_NAMELEN || + ss->create == NULL || ss->destroy == NULL) + return -EINVAL; + + /* + * we don't support callbacks in modular subsystems. this check is + * before the ss->module check for consistency; a subsystem that could + * be a module should still have no callbacks even if the user isn't + * compiling it as one. + */ + if (ss->fork || ss->exit) + return -EINVAL; + + /* + * an optionally modular subsystem is built-in: we want to do nothing, + * since cgroup_init_subsys will have already taken care of it. + */ + if (ss->module == NULL) { + /* a few sanity checks */ + BUG_ON(ss->subsys_id >= CGROUP_BUILTIN_SUBSYS_COUNT); + BUG_ON(subsys[ss->subsys_id] != ss); + return 0; + } + + /* + * need to register a subsys id before anything else - for example, + * init_cgroup_css needs it. + */ + mutex_lock(&cgroup_mutex); + /* find the first empty slot in the array */ + for (i = CGROUP_BUILTIN_SUBSYS_COUNT; i < CGROUP_SUBSYS_COUNT; i++) { + if (subsys[i] == NULL) + break; + } + if (i == CGROUP_SUBSYS_COUNT) { + /* maximum number of subsystems already registered! */ + mutex_unlock(&cgroup_mutex); + return -EBUSY; + } + /* assign ourselves the subsys_id */ + ss->subsys_id = i; + subsys[i] = ss; + + /* + * no ss->create seems to need anything important in the ss struct, so + * this can happen first (i.e. before the rootnode attachment). + */ + css = ss->create(ss, dummytop); + if (IS_ERR(css)) { + /* failure case - need to deassign the subsys[] slot. */ + subsys[i] = NULL; + mutex_unlock(&cgroup_mutex); + return PTR_ERR(css); + } + + list_add(&ss->sibling, &rootnode.subsys_list); + ss->root = &rootnode; + + /* our new subsystem will be attached to the dummy hierarchy. */ + init_cgroup_css(css, ss, dummytop); + /* init_idr must be after init_cgroup_css because it sets css->id. */ + if (ss->use_id) { + int ret = cgroup_init_idr(ss, css); + if (ret) { + dummytop->subsys[ss->subsys_id] = NULL; + ss->destroy(ss, dummytop); + subsys[i] = NULL; + mutex_unlock(&cgroup_mutex); + return ret; + } + } + + /* + * Now we need to entangle the css into the existing css_sets. unlike + * in cgroup_init_subsys, there are now multiple css_sets, so each one + * will need a new pointer to it; done by iterating the css_set_table. + * furthermore, modifying the existing css_sets will corrupt the hash + * table state, so each changed css_set will need its hash recomputed. + * this is all done under the css_set_lock. + */ + write_lock(&css_set_lock); + for (i = 0; i < CSS_SET_TABLE_SIZE; i++) { + struct css_set *cg; + struct hlist_node *node, *tmp; + struct hlist_head *bucket = &css_set_table[i], *new_bucket; + + hlist_for_each_entry_safe(cg, node, tmp, bucket, hlist) { + /* skip entries that we already rehashed */ + if (cg->subsys[ss->subsys_id]) + continue; + /* remove existing entry */ + hlist_del(&cg->hlist); + /* set new value */ + cg->subsys[ss->subsys_id] = css; + /* recompute hash and restore entry */ + new_bucket = css_set_hash(cg->subsys); + hlist_add_head(&cg->hlist, new_bucket); + } + } + write_unlock(&css_set_lock); + + mutex_init(&ss->hierarchy_mutex); + lockdep_set_class(&ss->hierarchy_mutex, &ss->subsys_key); + ss->active = 1; + + /* + * pin the subsystem's module so it doesn't go away. this shouldn't + * fail, since the module's initcall calls us. + * TODO: with module unloading, move this elsewhere + */ + BUG_ON(!try_module_get(ss->module)); + + /* success! */ + mutex_unlock(&cgroup_mutex); + return 0; } +EXPORT_SYMBOL_GPL(cgroup_load_subsys); /** * cgroup_init_early - cgroup initialization at system boot @@ -3364,7 +3505,7 @@ int __init cgroup_init(void) if (!ss->early_init) cgroup_init_subsys(ss); if (ss->use_id) - cgroup_subsys_init_idr(ss); + cgroup_init_idr(ss, init_css_set.subsys[ss->subsys_id]); } /* Add init_css_set to the hash table */ @@ -4033,15 +4174,14 @@ err_out: } -static int __init cgroup_subsys_init_idr(struct cgroup_subsys *ss) +static int __init_or_module cgroup_init_idr(struct cgroup_subsys *ss, + struct cgroup_subsys_state *rootcss) { struct css_id *newid; - struct cgroup_subsys_state *rootcss; spin_lock_init(&ss->id_lock); idr_init(&ss->idr); - rootcss = init_css_set.subsys[ss->subsys_id]; newid = get_new_cssid(ss, 0); if (IS_ERR(newid)) return PTR_ERR(newid); -- cgit v1.2.3-71-gd317 From cf5d5941fda647fe3d2f2d00cf9e0245236a5f08 Mon Sep 17 00:00:00 2001 From: Ben Blum Date: Wed, 10 Mar 2010 15:22:09 -0800 Subject: cgroups: subsystem module unloading Provides support for unloading modular subsystems. This patch adds a new function cgroup_unload_subsys which is to be used for removing a loaded subsystem during module deletion. Reference counting of the subsystems' modules is moved from once (at load time) to once per attached hierarchy (in parse_cgroupfs_options and rebind_subsystems) (i.e., 0 or 1). Signed-off-by: Ben Blum Acked-by: Li Zefan Cc: Paul Menage Cc: "David S. Miller" Cc: KAMEZAWA Hiroyuki Cc: Lai Jiangshan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- Documentation/cgroups/cgroups.txt | 5 +- include/linux/cgroup.h | 4 +- kernel/cgroup.c | 167 ++++++++++++++++++++++++++++++++------ 3 files changed, 148 insertions(+), 28 deletions(-) (limited to 'include/linux/cgroup.h') diff --git a/Documentation/cgroups/cgroups.txt b/Documentation/cgroups/cgroups.txt index ae8a037a761e..764007b63921 100644 --- a/Documentation/cgroups/cgroups.txt +++ b/Documentation/cgroups/cgroups.txt @@ -489,8 +489,9 @@ Each subsystem should: - define a cgroup_subsys object called _subsys If a subsystem can be compiled as a module, it should also have in its -module initcall a call to cgroup_load_subsys(&its_subsys_struct). It -should also set its_subsys.module = THIS_MODULE in its .c file. +module initcall a call to cgroup_load_subsys(), and in its exitcall a +call to cgroup_unload_subsys(). It should also set its_subsys.module = +THIS_MODULE in its .c file. Each subsystem may export the following methods. The only mandatory methods are create/destroy. Any others that are null are presumed to diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index 402ce477c47e..2a59d3101e5d 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -38,6 +38,7 @@ extern void cgroup_exit(struct task_struct *p, int run_callbacks); extern int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry); extern int cgroup_load_subsys(struct cgroup_subsys *ss); +extern void cgroup_unload_subsys(struct cgroup_subsys *ss); extern const struct file_operations proc_cgroup_operations; @@ -271,7 +272,8 @@ struct css_set { /* * Set of subsystem states, one for each subsystem. This array * is immutable after creation apart from the init_css_set - * during subsystem registration (at boot time). + * during subsystem registration (at boot time) and modular subsystem + * loading/unloading. */ struct cgroup_subsys_state *subsys[CGROUP_SUBSYS_COUNT]; diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 2cae38e64c59..aa889c96cc74 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -894,7 +894,9 @@ void cgroup_release_and_wakeup_rmdir(struct cgroup_subsys_state *css) } /* - * Call with cgroup_mutex held. + * Call with cgroup_mutex held. Drops reference counts on modules, including + * any duplicate ones that parse_cgroupfs_options took. If this function + * returns an error, no reference counts are touched. */ static int rebind_subsystems(struct cgroupfs_root *root, unsigned long final_bits) @@ -950,6 +952,7 @@ static int rebind_subsystems(struct cgroupfs_root *root, if (ss->bind) ss->bind(ss, cgrp); mutex_unlock(&ss->hierarchy_mutex); + /* refcount was already taken, and we're keeping it */ } else if (bit & removed_bits) { /* We're removing this subsystem */ BUG_ON(ss == NULL); @@ -963,10 +966,20 @@ static int rebind_subsystems(struct cgroupfs_root *root, subsys[i]->root = &rootnode; list_move(&ss->sibling, &rootnode.subsys_list); mutex_unlock(&ss->hierarchy_mutex); + /* subsystem is now free - drop reference on module */ + module_put(ss->module); } else if (bit & final_bits) { /* Subsystem state should already exist */ BUG_ON(ss == NULL); BUG_ON(!cgrp->subsys[i]); + /* + * a refcount was taken, but we already had one, so + * drop the extra reference. + */ + module_put(ss->module); +#ifdef CONFIG_MODULE_UNLOAD + BUG_ON(ss->module && !module_refcount(ss->module)); +#endif } else { /* Subsystem state shouldn't exist */ BUG_ON(cgrp->subsys[i]); @@ -1010,13 +1023,16 @@ struct cgroup_sb_opts { /* * Convert a hierarchy specifier into a bitmask of subsystems and flags. Call - * with cgroup_mutex held to protect the subsys[] array. + * with cgroup_mutex held to protect the subsys[] array. This function takes + * refcounts on subsystems to be used, unless it returns error, in which case + * no refcounts are taken. */ -static int parse_cgroupfs_options(char *data, - struct cgroup_sb_opts *opts) +static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) { char *token, *o = data ?: "all"; unsigned long mask = (unsigned long)-1; + int i; + bool module_pin_failed = false; BUG_ON(!mutex_is_locked(&cgroup_mutex)); @@ -1031,7 +1047,6 @@ static int parse_cgroupfs_options(char *data, return -EINVAL; if (!strcmp(token, "all")) { /* Add all non-disabled subsystems */ - int i; opts->subsys_bits = 0; for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; @@ -1054,7 +1069,6 @@ static int parse_cgroupfs_options(char *data, if (!opts->release_agent) return -ENOMEM; } else if (!strncmp(token, "name=", 5)) { - int i; const char *name = token + 5; /* Can't specify an empty name */ if (!strlen(name)) @@ -1078,7 +1092,6 @@ static int parse_cgroupfs_options(char *data, return -ENOMEM; } else { struct cgroup_subsys *ss; - int i; for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { ss = subsys[i]; if (ss == NULL) @@ -1117,9 +1130,54 @@ static int parse_cgroupfs_options(char *data, if (!opts->subsys_bits && !opts->name) return -EINVAL; + /* + * Grab references on all the modules we'll need, so the subsystems + * don't dance around before rebind_subsystems attaches them. This may + * take duplicate reference counts on a subsystem that's already used, + * but rebind_subsystems handles this case. + */ + for (i = CGROUP_BUILTIN_SUBSYS_COUNT; i < CGROUP_SUBSYS_COUNT; i++) { + unsigned long bit = 1UL << i; + + if (!(bit & opts->subsys_bits)) + continue; + if (!try_module_get(subsys[i]->module)) { + module_pin_failed = true; + break; + } + } + if (module_pin_failed) { + /* + * oops, one of the modules was going away. this means that we + * raced with a module_delete call, and to the user this is + * essentially a "subsystem doesn't exist" case. + */ + for (i--; i >= CGROUP_BUILTIN_SUBSYS_COUNT; i--) { + /* drop refcounts only on the ones we took */ + unsigned long bit = 1UL << i; + + if (!(bit & opts->subsys_bits)) + continue; + module_put(subsys[i]->module); + } + return -ENOENT; + } + return 0; } +static void drop_parsed_module_refcounts(unsigned long subsys_bits) +{ + int i; + for (i = CGROUP_BUILTIN_SUBSYS_COUNT; i < CGROUP_SUBSYS_COUNT; i++) { + unsigned long bit = 1UL << i; + + if (!(bit & subsys_bits)) + continue; + module_put(subsys[i]->module); + } +} + static int cgroup_remount(struct super_block *sb, int *flags, char *data) { int ret = 0; @@ -1136,21 +1194,19 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data) if (ret) goto out_unlock; - /* Don't allow flags to change at remount */ - if (opts.flags != root->flags) { - ret = -EINVAL; - goto out_unlock; - } - - /* Don't allow name to change at remount */ - if (opts.name && strcmp(opts.name, root->name)) { + /* Don't allow flags or name to change at remount */ + if (opts.flags != root->flags || + (opts.name && strcmp(opts.name, root->name))) { ret = -EINVAL; + drop_parsed_module_refcounts(opts.subsys_bits); goto out_unlock; } ret = rebind_subsystems(root, opts.subsys_bits); - if (ret) + if (ret) { + drop_parsed_module_refcounts(opts.subsys_bits); goto out_unlock; + } /* (re)populate subsystem files */ cgroup_populate_dir(cgrp); @@ -1349,7 +1405,7 @@ static int cgroup_get_sb(struct file_system_type *fs_type, new_root = cgroup_root_from_opts(&opts); if (IS_ERR(new_root)) { ret = PTR_ERR(new_root); - goto out_err; + goto drop_modules; } opts.new_root = new_root; @@ -1358,7 +1414,7 @@ static int cgroup_get_sb(struct file_system_type *fs_type, if (IS_ERR(sb)) { ret = PTR_ERR(sb); cgroup_drop_root(opts.new_root); - goto out_err; + goto drop_modules; } root = sb->s_fs_info; @@ -1414,6 +1470,11 @@ static int cgroup_get_sb(struct file_system_type *fs_type, free_cg_links(&tmp_cg_links); goto drop_new_super; } + /* + * There must be no failure case after here, since rebinding + * takes care of subsystems' refcounts, which are explicitly + * dropped in the failure exit path. + */ /* EBUSY should be the only error here */ BUG_ON(ret); @@ -1452,6 +1513,8 @@ static int cgroup_get_sb(struct file_system_type *fs_type, * any) is not needed */ cgroup_drop_root(opts.new_root); + /* no subsys rebinding, so refcounts don't change */ + drop_parsed_module_refcounts(opts.subsys_bits); } simple_set_mnt(mnt, sb); @@ -1461,6 +1524,8 @@ static int cgroup_get_sb(struct file_system_type *fs_type, drop_new_super: deactivate_locked_super(sb); + drop_modules: + drop_parsed_module_refcounts(opts.subsys_bits); out_err: kfree(opts.release_agent); kfree(opts.name); @@ -3422,19 +3487,71 @@ int __init_or_module cgroup_load_subsys(struct cgroup_subsys *ss) lockdep_set_class(&ss->hierarchy_mutex, &ss->subsys_key); ss->active = 1; - /* - * pin the subsystem's module so it doesn't go away. this shouldn't - * fail, since the module's initcall calls us. - * TODO: with module unloading, move this elsewhere - */ - BUG_ON(!try_module_get(ss->module)); - /* success! */ mutex_unlock(&cgroup_mutex); return 0; } EXPORT_SYMBOL_GPL(cgroup_load_subsys); +/** + * cgroup_unload_subsys: unload a modular subsystem + * @ss: the subsystem to unload + * + * This function should be called in a modular subsystem's exitcall. When this + * function is invoked, the refcount on the subsystem's module will be 0, so + * the subsystem will not be attached to any hierarchy. + */ +void cgroup_unload_subsys(struct cgroup_subsys *ss) +{ + struct cg_cgroup_link *link; + struct hlist_head *hhead; + + BUG_ON(ss->module == NULL); + + /* + * we shouldn't be called if the subsystem is in use, and the use of + * try_module_get in parse_cgroupfs_options should ensure that it + * doesn't start being used while we're killing it off. + */ + BUG_ON(ss->root != &rootnode); + + mutex_lock(&cgroup_mutex); + /* deassign the subsys_id */ + BUG_ON(ss->subsys_id < CGROUP_BUILTIN_SUBSYS_COUNT); + subsys[ss->subsys_id] = NULL; + + /* remove subsystem from rootnode's list of subsystems */ + list_del(&ss->sibling); + + /* + * disentangle the css from all css_sets attached to the dummytop. as + * in loading, we need to pay our respects to the hashtable gods. + */ + write_lock(&css_set_lock); + list_for_each_entry(link, &dummytop->css_sets, cgrp_link_list) { + struct css_set *cg = link->cg; + + hlist_del(&cg->hlist); + BUG_ON(!cg->subsys[ss->subsys_id]); + cg->subsys[ss->subsys_id] = NULL; + hhead = css_set_hash(cg->subsys); + hlist_add_head(&cg->hlist, hhead); + } + write_unlock(&css_set_lock); + + /* + * remove subsystem's css from the dummytop and free it - need to free + * before marking as null because ss->destroy needs the cgrp->subsys + * pointer to find their state. note that this also takes care of + * freeing the css_id. + */ + ss->destroy(ss, dummytop); + dummytop->subsys[ss->subsys_id] = NULL; + + mutex_unlock(&cgroup_mutex); +} +EXPORT_SYMBOL_GPL(cgroup_unload_subsys); + /** * cgroup_init_early - cgroup initialization at system boot * -- cgit v1.2.3-71-gd317 From 0dea116876eefc9c7ca9c5d74fe665481e499fa3 Mon Sep 17 00:00:00 2001 From: "Kirill A. Shutemov" Date: Wed, 10 Mar 2010 15:22:20 -0800 Subject: cgroup: implement eventfd-based generic API for notifications This patchset introduces eventfd-based API for notifications in cgroups and implements memory notifications on top of it. It uses statistics in memory controler to track memory usage. Output of time(1) on building kernel on tmpfs: Root cgroup before changes: make -j2 506.37 user 60.93s system 193% cpu 4:52.77 total Non-root cgroup before changes: make -j2 507.14 user 62.66s system 193% cpu 4:54.74 total Root cgroup after changes (0 thresholds): make -j2 507.13 user 62.20s system 193% cpu 4:53.55 total Non-root cgroup after changes (0 thresholds): make -j2 507.70 user 64.20s system 193% cpu 4:55.70 total Root cgroup after changes (1 thresholds, never crossed): make -j2 506.97 user 62.20s system 193% cpu 4:53.90 total Non-root cgroup after changes (1 thresholds, never crossed): make -j2 507.55 user 64.08s system 193% cpu 4:55.63 total This patch: Introduce the write-only file "cgroup.event_control" in every cgroup. To register new notification handler you need: - create an eventfd; - open a control file to be monitored. Callbacks register_event() and unregister_event() must be defined for the control file; - write " " to cgroup.event_control. Interpretation of args is defined by control file implementation; eventfd will be woken up by control file implementation or when the cgroup is removed. To unregister notification handler just close eventfd. If you need notification functionality for a control file you have to implement callbacks register_event() and unregister_event() in the struct cftype. [kamezawa.hiroyu@jp.fujitsu.com: Kconfig fix] Signed-off-by: Kirill A. Shutemov Reviewed-by: KAMEZAWA Hiroyuki Paul Menage Cc: Li Zefan Cc: Balbir Singh Cc: Pavel Emelyanov Cc: Dan Malek Cc: Vladislav Buzov Cc: Daisuke Nishimura Cc: Alexander Shishkin Cc: Davide Libenzi Signed-off-by: KAMEZAWA Hiroyuki Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- Documentation/cgroups/cgroups.txt | 20 ++++ include/linux/cgroup.h | 24 ++++ init/Kconfig | 1 + kernel/cgroup.c | 228 +++++++++++++++++++++++++++++++++++++- 4 files changed, 272 insertions(+), 1 deletion(-) (limited to 'include/linux/cgroup.h') diff --git a/Documentation/cgroups/cgroups.txt b/Documentation/cgroups/cgroups.txt index c0358c30c64f..fd588ff0e296 100644 --- a/Documentation/cgroups/cgroups.txt +++ b/Documentation/cgroups/cgroups.txt @@ -23,6 +23,7 @@ CONTENTS: 2.1 Basic Usage 2.2 Attaching processes 2.3 Mounting hierarchies by name + 2.4 Notification API 3. Kernel API 3.1 Overview 3.2 Synchronization @@ -435,6 +436,25 @@ you give a subsystem a name. The name of the subsystem appears as part of the hierarchy description in /proc/mounts and /proc//cgroups. +2.4 Notification API +-------------------- + +There is mechanism which allows to get notifications about changing +status of a cgroup. + +To register new notification handler you need: + - create a file descriptor for event notification using eventfd(2); + - open a control file to be monitored (e.g. memory.usage_in_bytes); + - write " " to cgroup.event_control. + Interpretation of args is defined by control file implementation; + +eventfd will be woken up by control file implementation or when the +cgroup is removed. + +To unregister notification handler just close eventfd. + +NOTE: Support of notifications should be implemented for the control +file. See documentation for the subsystem. 3. Kernel API ============= diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index 2a59d3101e5d..b4f2201321cd 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -235,6 +235,10 @@ struct cgroup { /* For RCU-protected deletion */ struct rcu_head rcu_head; + + /* List of events which userspace want to recieve */ + struct list_head event_list; + spinlock_t event_list_lock; }; /* @@ -378,6 +382,26 @@ struct cftype { int (*trigger)(struct cgroup *cgrp, unsigned int event); int (*release)(struct inode *inode, struct file *file); + + /* + * register_event() callback will be used to add new userspace + * waiter for changes related to the cftype. Implement it if + * you want to provide this functionality. Use eventfd_signal() + * on eventfd to send notification to userspace. + */ + int (*register_event)(struct cgroup *cgrp, struct cftype *cft, + struct eventfd_ctx *eventfd, const char *args); + /* + * unregister_event() callback will be called when userspace + * closes the eventfd or on cgroup removing. + * This callback must be implemented, if you want provide + * notification functionality. + * + * Be careful. It can be called after destroy(), so you have + * to keep all nesessary data, until all events are removed. + */ + int (*unregister_event)(struct cgroup *cgrp, struct cftype *cft, + struct eventfd_ctx *eventfd); }; struct cgroup_scanner { diff --git a/init/Kconfig b/init/Kconfig index 089a230e5652..eb77e8ccde1c 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -463,6 +463,7 @@ config HAVE_UNSTABLE_SCHED_CLOCK menuconfig CGROUPS boolean "Control Group support" + depends on EVENTFD help This option adds support for grouping sets of processes together, for use with process control subsystems such as Cpusets, CFS, memory diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 1bf4d6db54ab..ea94984a3895 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -4,6 +4,10 @@ * Based originally on the cpuset system, extracted by Paul Menage * Copyright (C) 2006 Google, Inc * + * Notifications support + * Copyright (C) 2009 Nokia Corporation + * Author: Kirill A. Shutemov + * * Copyright notices from the original cpuset code: * -------------------------------------------------- * Copyright (C) 2003 BULL SA. @@ -53,6 +57,8 @@ #include #include #include /* TODO: replace with more sophisticated array */ +#include +#include #include @@ -152,6 +158,35 @@ struct css_id { unsigned short stack[0]; /* Array of Length (depth+1) */ }; +/* + * cgroup_event represents events which userspace want to recieve. + */ +struct cgroup_event { + /* + * Cgroup which the event belongs to. + */ + struct cgroup *cgrp; + /* + * Control file which the event associated. + */ + struct cftype *cft; + /* + * eventfd to signal userspace about the event. + */ + struct eventfd_ctx *eventfd; + /* + * Each of these stored in a list by the cgroup. + */ + struct list_head list; + /* + * All fields below needed to unregister event when + * userspace closes eventfd. + */ + poll_table pt; + wait_queue_head_t *wqh; + wait_queue_t wait; + struct work_struct remove; +}; /* The list of hierarchy roots */ @@ -760,14 +795,28 @@ static struct inode *cgroup_new_inode(mode_t mode, struct super_block *sb) static int cgroup_call_pre_destroy(struct cgroup *cgrp) { struct cgroup_subsys *ss; + struct cgroup_event *event, *tmp; int ret = 0; for_each_subsys(cgrp->root, ss) if (ss->pre_destroy) { ret = ss->pre_destroy(ss, cgrp); if (ret) - break; + goto out; } + + /* + * Unregister events and notify userspace. + */ + spin_lock(&cgrp->event_list_lock); + list_for_each_entry_safe(event, tmp, &cgrp->event_list, list) { + list_del(&event->list); + eventfd_signal(event->eventfd, 1); + schedule_work(&event->remove); + } + spin_unlock(&cgrp->event_list_lock); + +out: return ret; } @@ -1239,6 +1288,8 @@ static void init_cgroup_housekeeping(struct cgroup *cgrp) INIT_LIST_HEAD(&cgrp->release_list); INIT_LIST_HEAD(&cgrp->pidlists); mutex_init(&cgrp->pidlist_mutex); + INIT_LIST_HEAD(&cgrp->event_list); + spin_lock_init(&cgrp->event_list_lock); } static void init_cgroup_root(struct cgroupfs_root *root) @@ -2077,6 +2128,16 @@ static const struct inode_operations cgroup_dir_inode_operations = { .rename = cgroup_rename, }; +/* + * Check if a file is a control file + */ +static inline struct cftype *__file_cft(struct file *file) +{ + if (file->f_dentry->d_inode->i_fop != &cgroup_file_operations) + return ERR_PTR(-EINVAL); + return __d_cft(file->f_dentry); +} + static int cgroup_create_file(struct dentry *dentry, mode_t mode, struct super_block *sb) { @@ -2930,6 +2991,166 @@ static int cgroup_write_notify_on_release(struct cgroup *cgrp, return 0; } +/* + * Unregister event and free resources. + * + * Gets called from workqueue. + */ +static void cgroup_event_remove(struct work_struct *work) +{ + struct cgroup_event *event = container_of(work, struct cgroup_event, + remove); + struct cgroup *cgrp = event->cgrp; + + /* TODO: check return code */ + event->cft->unregister_event(cgrp, event->cft, event->eventfd); + + eventfd_ctx_put(event->eventfd); + remove_wait_queue(event->wqh, &event->wait); + kfree(event); +} + +/* + * Gets called on POLLHUP on eventfd when user closes it. + * + * Called with wqh->lock held and interrupts disabled. + */ +static int cgroup_event_wake(wait_queue_t *wait, unsigned mode, + int sync, void *key) +{ + struct cgroup_event *event = container_of(wait, + struct cgroup_event, wait); + struct cgroup *cgrp = event->cgrp; + unsigned long flags = (unsigned long)key; + + if (flags & POLLHUP) { + spin_lock(&cgrp->event_list_lock); + list_del(&event->list); + spin_unlock(&cgrp->event_list_lock); + /* + * We are in atomic context, but cgroup_event_remove() may + * sleep, so we have to call it in workqueue. + */ + schedule_work(&event->remove); + } + + return 0; +} + +static void cgroup_event_ptable_queue_proc(struct file *file, + wait_queue_head_t *wqh, poll_table *pt) +{ + struct cgroup_event *event = container_of(pt, + struct cgroup_event, pt); + + event->wqh = wqh; + add_wait_queue(wqh, &event->wait); +} + +/* + * Parse input and register new cgroup event handler. + * + * Input must be in format ' '. + * Interpretation of args is defined by control file implementation. + */ +static int cgroup_write_event_control(struct cgroup *cgrp, struct cftype *cft, + const char *buffer) +{ + struct cgroup_event *event = NULL; + unsigned int efd, cfd; + struct file *efile = NULL; + struct file *cfile = NULL; + char *endp; + int ret; + + efd = simple_strtoul(buffer, &endp, 10); + if (*endp != ' ') + return -EINVAL; + buffer = endp + 1; + + cfd = simple_strtoul(buffer, &endp, 10); + if ((*endp != ' ') && (*endp != '\0')) + return -EINVAL; + buffer = endp + 1; + + event = kzalloc(sizeof(*event), GFP_KERNEL); + if (!event) + return -ENOMEM; + event->cgrp = cgrp; + INIT_LIST_HEAD(&event->list); + init_poll_funcptr(&event->pt, cgroup_event_ptable_queue_proc); + init_waitqueue_func_entry(&event->wait, cgroup_event_wake); + INIT_WORK(&event->remove, cgroup_event_remove); + + efile = eventfd_fget(efd); + if (IS_ERR(efile)) { + ret = PTR_ERR(efile); + goto fail; + } + + event->eventfd = eventfd_ctx_fileget(efile); + if (IS_ERR(event->eventfd)) { + ret = PTR_ERR(event->eventfd); + goto fail; + } + + cfile = fget(cfd); + if (!cfile) { + ret = -EBADF; + goto fail; + } + + /* the process need read permission on control file */ + ret = file_permission(cfile, MAY_READ); + if (ret < 0) + goto fail; + + event->cft = __file_cft(cfile); + if (IS_ERR(event->cft)) { + ret = PTR_ERR(event->cft); + goto fail; + } + + if (!event->cft->register_event || !event->cft->unregister_event) { + ret = -EINVAL; + goto fail; + } + + ret = event->cft->register_event(cgrp, event->cft, + event->eventfd, buffer); + if (ret) + goto fail; + + if (efile->f_op->poll(efile, &event->pt) & POLLHUP) { + event->cft->unregister_event(cgrp, event->cft, event->eventfd); + ret = 0; + goto fail; + } + + spin_lock(&cgrp->event_list_lock); + list_add(&event->list, &cgrp->event_list); + spin_unlock(&cgrp->event_list_lock); + + fput(cfile); + fput(efile); + + return 0; + +fail: + if (cfile) + fput(cfile); + + if (event && event->eventfd && !IS_ERR(event->eventfd)) + eventfd_ctx_put(event->eventfd); + + if (!IS_ERR_OR_NULL(efile)) + fput(efile); + + kfree(event); + + return ret; +} + /* * for the common functions, 'private' gives the type of file */ @@ -2955,6 +3176,11 @@ static struct cftype files[] = { .read_u64 = cgroup_read_notify_on_release, .write_u64 = cgroup_write_notify_on_release, }, + { + .name = CGROUP_FILE_GENERIC_PREFIX "event_control", + .write_string = cgroup_write_event_control, + .mode = S_IWUGO, + }, }; static struct cftype cft_release_agent = { -- cgit v1.2.3-71-gd317 From a0a4db548edcce067c1201ef25cf2bc29f32dca4 Mon Sep 17 00:00:00 2001 From: "Kirill A. Shutemov" Date: Wed, 10 Mar 2010 15:22:34 -0800 Subject: cgroups: remove events before destroying subsystem state objects Events should be removed after rmdir of cgroup directory, but before destroying subsystem state objects. Let's take reference to cgroup directory dentry to do that. Signed-off-by: Kirill A. Shutemov Acked-by: KAMEZAWA Hiroyuki Cc: Paul Menage Acked-by: Li Zefan Cc: Balbir Singh Cc: Pavel Emelyanov Cc: Dan Malek Cc: Daisuke Nishimura Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/cgroup.h | 3 --- kernel/cgroup.c | 8 ++++++++ mm/memcontrol.c | 9 --------- 3 files changed, 8 insertions(+), 12 deletions(-) (limited to 'include/linux/cgroup.h') diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index b4f2201321cd..b8ad1ea99586 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -396,9 +396,6 @@ struct cftype { * closes the eventfd or on cgroup removing. * This callback must be implemented, if you want provide * notification functionality. - * - * Be careful. It can be called after destroy(), so you have - * to keep all nesessary data, until all events are removed. */ int (*unregister_event)(struct cgroup *cgrp, struct cftype *cft, struct eventfd_ctx *eventfd); diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 87441fc75663..ef909a329750 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -2994,6 +2994,7 @@ static void cgroup_event_remove(struct work_struct *work) eventfd_ctx_put(event->eventfd); kfree(event); + dput(cgrp->dentry); } /* @@ -3114,6 +3115,13 @@ static int cgroup_write_event_control(struct cgroup *cgrp, struct cftype *cft, goto fail; } + /* + * Events should be removed after rmdir of cgroup directory, but before + * destroying subsystem state objects. Let's take reference to cgroup + * directory dentry to do that. + */ + dget(cgrp->dentry); + spin_lock(&cgrp->event_list_lock); list_add(&event->list, &cgrp->event_list); spin_unlock(&cgrp->event_list_lock); diff --git a/mm/memcontrol.c b/mm/memcontrol.c index f9ae4b4c36eb..f7b910fc14fb 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -3361,12 +3361,6 @@ static int mem_cgroup_register_event(struct cgroup *cgrp, struct cftype *cft, } } - /* - * We need to increment refcnt to be sure that all thresholds - * will be unregistered before calling __mem_cgroup_free() - */ - mem_cgroup_get(memcg); - if (type == _MEM) rcu_assign_pointer(memcg->thresholds, thresholds_new); else @@ -3460,9 +3454,6 @@ assign: /* To be sure that nobody uses thresholds before freeing it */ synchronize_rcu(); - for (i = 0; i < thresholds->size - size; i++) - mem_cgroup_put(memcg); - kfree(thresholds); unlock: mutex_unlock(&memcg->thresholds_lock); -- cgit v1.2.3-71-gd317