diff options
Diffstat (limited to 'kernel/cpuset.c')
| -rw-r--r-- | kernel/cpuset.c | 1893 |
1 files changed, 1005 insertions, 888 deletions
diff --git a/kernel/cpuset.c b/kernel/cpuset.c index ba401fab459..116a4164720 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -37,7 +37,7 @@ #include <linux/mempolicy.h> #include <linux/mm.h> #include <linux/memory.h> -#include <linux/module.h> +#include <linux/export.h> #include <linux/mount.h> #include <linux/namei.h> #include <linux/pagemap.h> @@ -55,29 +55,13 @@ #include <linux/sort.h> #include <asm/uaccess.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include <linux/mutex.h> #include <linux/workqueue.h> #include <linux/cgroup.h> +#include <linux/wait.h> -/* - * Workqueue for cpuset related tasks. - * - * Using kevent workqueue may cause deadlock when memory_migrate - * is set. So we create a separate workqueue thread for cpuset. - */ -static struct workqueue_struct *cpuset_wq; - -/* - * Tracks how many cpusets are currently defined in system. - * When there is only one cpuset (the root cpuset) we can - * short circuit some hooks. - */ -int number_of_cpusets __read_mostly; - -/* Forward declare cgroup structures */ -struct cgroup_subsys cpuset_subsys; -struct cpuset; +struct static_key cpusets_enabled_key __read_mostly = STATIC_KEY_INIT_FALSE; /* See "Frequency meter" comments, below. */ @@ -95,36 +79,65 @@ struct cpuset { cpumask_var_t cpus_allowed; /* CPUs allowed to tasks in cpuset */ nodemask_t mems_allowed; /* Memory Nodes allowed to tasks */ - struct cpuset *parent; /* my parent */ + /* + * 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; struct fmeter fmeter; /* memory_pressure filter */ + /* + * Tasks are being attached to this cpuset. Used to prevent + * zeroing cpus/mems_allowed between ->can_attach() and ->attach(). + */ + int attach_in_progress; + /* partition number for rebuild_sched_domains() */ int pn; /* for custom sched domain */ int relax_domain_level; - - /* used for walking a cpuset heirarchy */ - struct list_head stack_list; }; -/* Retrieve the cpuset for a cgroup */ -static inline struct cpuset *cgroup_cs(struct cgroup *cont) +static inline struct cpuset *css_cs(struct cgroup_subsys_state *css) { - return container_of(cgroup_subsys_state(cont, cpuset_subsys_id), - struct cpuset, css); + return css ? container_of(css, struct cpuset, css) : NULL; } /* Retrieve the cpuset for a task */ static inline struct cpuset *task_cs(struct task_struct *task) { - return container_of(task_subsys_state(task, cpuset_subsys_id), - struct cpuset, css); + return css_cs(task_css(task, cpuset_cgrp_id)); } +static inline struct cpuset *parent_cs(struct cpuset *cs) +{ + return css_cs(cs->css.parent); +} + +#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 + + /* bits in struct cpuset flags field */ typedef enum { + CS_ONLINE, CS_CPU_EXCLUSIVE, CS_MEM_EXCLUSIVE, CS_MEM_HARDWALL, @@ -135,6 +148,11 @@ typedef enum { } cpuset_flagbits_t; /* convenient tests for these bits */ +static inline bool is_cpuset_online(const struct cpuset *cs) +{ + return test_bit(CS_ONLINE, &cs->flags); +} + static inline int is_cpu_exclusive(const struct cpuset *cs) { return test_bit(CS_CPU_EXCLUSIVE, &cs->flags); @@ -171,27 +189,53 @@ static inline int is_spread_slab(const struct cpuset *cs) } static struct cpuset top_cpuset = { - .flags = ((1 << CS_CPU_EXCLUSIVE) | (1 << CS_MEM_EXCLUSIVE)), + .flags = ((1 << CS_ONLINE) | (1 << CS_CPU_EXCLUSIVE) | + (1 << CS_MEM_EXCLUSIVE)), }; +/** + * cpuset_for_each_child - traverse online children of a cpuset + * @child_cs: loop cursor pointing to the current child + * @pos_css: used for iteration + * @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. + */ +#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))))) + +/** + * cpuset_for_each_descendant_pre - pre-order walk of a cpuset's descendants + * @des_cs: loop cursor pointing to the current descendant + * @pos_css: used for iteration + * @root_cs: target cpuset to walk ancestor of + * + * Walk @des_cs through the online descendants of @root_cs. Must be used + * with RCU read locked. The caller may modify @pos_css by calling + * css_rightmost_descendant() to skip subtree. @root_cs is included in the + * iteration and the first node to be visited. + */ +#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))))) + /* - * There are two global mutexes guarding cpuset structures. The first - * is the main control groups cgroup_mutex, accessed via - * cgroup_lock()/cgroup_unlock(). The second is the cpuset-specific - * callback_mutex, below. They can nest. It is ok to first take - * cgroup_mutex, then nest callback_mutex. 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 cgroup_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 cgroup_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. + * 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. * * Calls to the kernel memory allocator can not be made while holding * callback_mutex, as that would risk double tripping on callback_mutex @@ -213,36 +257,33 @@ static struct cpuset top_cpuset = { * guidelines for accessing subsystem state in kernel/cgroup.c */ +static DEFINE_MUTEX(cpuset_mutex); static DEFINE_MUTEX(callback_mutex); /* - * cpuset_buffer_lock protects both the cpuset_name and cpuset_nodelist - * buffers. They are statically allocated to prevent using excess stack - * when calling cpuset_print_task_mems_allowed(). + * CPU / memory hotplug is handled asynchronously. */ -#define CPUSET_NAME_LEN (128) -#define CPUSET_NODELIST_LEN (256) -static char cpuset_name[CPUSET_NAME_LEN]; -static char cpuset_nodelist[CPUSET_NODELIST_LEN]; -static DEFINE_SPINLOCK(cpuset_buffer_lock); +static void cpuset_hotplug_workfn(struct work_struct *work); +static DECLARE_WORK(cpuset_hotplug_work, cpuset_hotplug_workfn); + +static DECLARE_WAIT_QUEUE_HEAD(cpuset_attach_wq); /* * This is ugly, but preserves the userspace API for existing cpuset * users. If someone tries to mount the "cpuset" filesystem, we * silently switch it to mount "cgroup" instead */ -static int cpuset_get_sb(struct file_system_type *fs_type, - int flags, const char *unused_dev_name, - void *data, struct vfsmount *mnt) +static struct dentry *cpuset_mount(struct file_system_type *fs_type, + int flags, const char *unused_dev_name, void *data) { struct file_system_type *cgroup_fs = get_fs_type("cgroup"); - int ret = -ENODEV; + struct dentry *ret = ERR_PTR(-ENODEV); if (cgroup_fs) { char mountopts[] = "cpuset,noprefix," "release_agent=/sbin/cpuset_release_agent"; - ret = cgroup_fs->get_sb(cgroup_fs, flags, - unused_dev_name, mountopts, mnt); + ret = cgroup_fs->mount(cgroup_fs, flags, + unused_dev_name, mountopts); put_filesystem(cgroup_fs); } return ret; @@ -250,65 +291,49 @@ static int cpuset_get_sb(struct file_system_type *fs_type, static struct file_system_type cpuset_fs_type = { .name = "cpuset", - .get_sb = cpuset_get_sb, + .mount = cpuset_mount, }; /* * Return in pmask the portion of a cpusets's cpus_allowed that * are online. If none are online, walk up the cpuset hierarchy - * until we find one that does have some online cpus. If we get - * all the way to the top and still haven't found any online cpus, - * return cpu_online_map. Or if passed a NULL cs from an exit'ing - * task, return cpu_online_map. + * until we find one that does have some online cpus. The top + * cpuset always has some cpus online. * * One way or another, we guarantee to return some non-empty subset - * of cpu_online_map. + * of cpu_online_mask. * * Call with callback_mutex held. */ - -static void guarantee_online_cpus(const struct cpuset *cs, - struct cpumask *pmask) +static void guarantee_online_cpus(struct cpuset *cs, struct cpumask *pmask) { - while (cs && !cpumask_intersects(cs->cpus_allowed, cpu_online_mask)) - cs = cs->parent; - if (cs) - cpumask_and(pmask, cs->cpus_allowed, cpu_online_mask); - else - cpumask_copy(pmask, cpu_online_mask); - BUG_ON(!cpumask_intersects(pmask, cpu_online_mask)); + while (!cpumask_intersects(cs->cpus_allowed, cpu_online_mask)) + cs = parent_cs(cs); + cpumask_and(pmask, cs->cpus_allowed, cpu_online_mask); } /* * Return in *pmask the portion of a cpusets's mems_allowed that * are online, with memory. If none are online with memory, walk * up the cpuset hierarchy until we find one that does have some - * online mems. If we get all the way to the top and still haven't - * found any online mems, return node_states[N_HIGH_MEMORY]. + * online mems. The top cpuset always has some mems online. * * One way or another, we guarantee to return some non-empty subset - * of node_states[N_HIGH_MEMORY]. + * of node_states[N_MEMORY]. * * Call with callback_mutex held. */ - -static void guarantee_online_mems(const struct cpuset *cs, nodemask_t *pmask) +static void guarantee_online_mems(struct cpuset *cs, nodemask_t *pmask) { - while (cs && !nodes_intersects(cs->mems_allowed, - node_states[N_HIGH_MEMORY])) - cs = cs->parent; - if (cs) - nodes_and(*pmask, cs->mems_allowed, - node_states[N_HIGH_MEMORY]); - else - *pmask = node_states[N_HIGH_MEMORY]; - BUG_ON(!nodes_intersects(*pmask, node_states[N_HIGH_MEMORY])); + while (!nodes_intersects(cs->mems_allowed, node_states[N_MEMORY])) + cs = parent_cs(cs); + nodes_and(*pmask, cs->mems_allowed, node_states[N_MEMORY]); } /* * update task's spread flag if cpuset's page/slab spread flag is set * - * Called with callback_mutex/cgroup_mutex held + * Called with callback_mutex/cpuset_mutex held */ static void cpuset_update_task_spread_flag(struct cpuset *cs, struct task_struct *tsk) @@ -328,7 +353,7 @@ static void cpuset_update_task_spread_flag(struct cpuset *cs, * * 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 - * are only set if the other's are set. Call holding cgroup_mutex. + * are only set if the other's are set. Call holding cpuset_mutex. */ static int is_cpuset_subset(const struct cpuset *p, const struct cpuset *q) @@ -343,7 +368,7 @@ static int is_cpuset_subset(const struct cpuset *p, const struct cpuset *q) * alloc_trial_cpuset - allocate a trial cpuset * @cs: the cpuset that the trial cpuset duplicates */ -static struct cpuset *alloc_trial_cpuset(const struct cpuset *cs) +static struct cpuset *alloc_trial_cpuset(struct cpuset *cs) { struct cpuset *trial; @@ -377,7 +402,7 @@ static void free_trial_cpuset(struct cpuset *trial) * 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 - * cgroup_mutex held. + * cpuset_mutex held. * * 'cur' is the address of an actual, in-use cpuset. Operations * such as list traversal that depend on the actual address of the @@ -390,52 +415,66 @@ static void free_trial_cpuset(struct cpuset *trial) * Return 0 if valid, -errno if not. */ -static int validate_change(const struct cpuset *cur, const struct cpuset *trial) +static int validate_change(struct cpuset *cur, struct cpuset *trial) { - struct cgroup *cont; + struct cgroup_subsys_state *css; struct cpuset *c, *par; + int ret; + + rcu_read_lock(); /* Each of our child cpusets must be a subset of us */ - list_for_each_entry(cont, &cur->css.cgroup->children, sibling) { - if (!is_cpuset_subset(cgroup_cs(cont), trial)) - return -EBUSY; - } + ret = -EBUSY; + cpuset_for_each_child(c, css, cur) + if (!is_cpuset_subset(c, trial)) + goto out; /* Remaining checks don't apply to root cpuset */ + ret = 0; if (cur == &top_cpuset) - return 0; + goto out; - par = cur->parent; + par = parent_cs(cur); /* We must be a subset of our parent cpuset */ + ret = -EACCES; if (!is_cpuset_subset(trial, par)) - return -EACCES; + goto out; /* * If either I or some sibling (!= me) is exclusive, we can't * overlap */ - list_for_each_entry(cont, &par->css.cgroup->children, sibling) { - c = cgroup_cs(cont); + ret = -EINVAL; + cpuset_for_each_child(c, css, par) { if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) && c != cur && cpumask_intersects(trial->cpus_allowed, c->cpus_allowed)) - return -EINVAL; + goto out; if ((is_mem_exclusive(trial) || is_mem_exclusive(c)) && c != cur && nodes_intersects(trial->mems_allowed, c->mems_allowed)) - return -EINVAL; + goto out; } - /* Cpusets with tasks can't have empty cpus_allowed or mems_allowed */ - if (cgroup_task_count(cur->css.cgroup)) { - if (cpumask_empty(trial->cpus_allowed) || - nodes_empty(trial->mems_allowed)) { - return -ENOSPC; - } + /* + * Cpusets with tasks - existing or newly being attached - can't + * be changed to have empty cpus_allowed or mems_allowed. + */ + ret = -ENOSPC; + if ((cgroup_has_tasks(cur->css.cgroup) || cur->attach_in_progress)) { + 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; } - return 0; + ret = 0; +out: + rcu_read_unlock(); + return ret; } #ifdef CONFIG_SMP @@ -456,31 +495,27 @@ update_domain_attr(struct sched_domain_attr *dattr, struct cpuset *c) return; } -static void -update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *c) +static void update_domain_attr_tree(struct sched_domain_attr *dattr, + struct cpuset *root_cs) { - LIST_HEAD(q); - - list_add(&c->stack_list, &q); - while (!list_empty(&q)) { - struct cpuset *cp; - struct cgroup *cont; - struct cpuset *child; + struct cpuset *cp; + struct cgroup_subsys_state *pos_css; - cp = list_first_entry(&q, struct cpuset, stack_list); - list_del(q.next); + rcu_read_lock(); + cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { + if (cp == root_cs) + continue; - if (cpumask_empty(cp->cpus_allowed)) + /* skip the whole subtree if @cp doesn't have any CPU */ + if (cpumask_empty(cp->cpus_allowed)) { + pos_css = css_rightmost_descendant(pos_css); continue; + } if (is_sched_load_balance(cp)) update_domain_attr(dattr, cp); - - list_for_each_entry(cont, &cp->css.cgroup->children, sibling) { - child = cgroup_cs(cont); - list_add_tail(&child->stack_list, &q); - } } + rcu_read_unlock(); } /* @@ -489,7 +524,7 @@ update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *c) * 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. - * The output of this function needs to be passed to kernel/sched.c + * The output of this function needs to be passed to kernel/sched/core.c * partition_sched_domains() routine, which will rebuild the scheduler's * load balancing domains (sched domains) as specified by that partial * partition. @@ -502,7 +537,7 @@ update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *c) * domains when operating in the severe memory shortage situations * that could cause allocation failures below. * - * Must be called with cgroup_lock held. + * Must be called with cpuset_mutex held. * * The three key local variables below are: * q - a linked-list queue of cpuset pointers, used to implement a @@ -518,7 +553,7 @@ update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *c) * 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 - * the kernel/sched.c routine partition_sched_domains() in a + * the kernel/sched/core.c routine partition_sched_domains() in a * convenient format, that can be easily compared to the prior * value to determine what partition elements (sched domains) * were changed (added or removed.) @@ -540,7 +575,6 @@ update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *c) static int generate_sched_domains(cpumask_var_t **domains, struct sched_domain_attr **attributes) { - LIST_HEAD(q); /* queue of cpusets to be scanned */ struct cpuset *cp; /* scans q */ struct cpuset **csa; /* array of all cpuset ptrs */ int csn; /* how many cpuset ptrs in csa so far */ @@ -549,6 +583,7 @@ static int generate_sched_domains(cpumask_var_t **domains, struct sched_domain_attr *dattr; /* attributes for custom domains */ int ndoms = 0; /* number of sched domains in result */ int nslot; /* next empty doms[] struct cpumask slot */ + struct cgroup_subsys_state *pos_css; doms = NULL; dattr = NULL; @@ -571,38 +606,34 @@ static int generate_sched_domains(cpumask_var_t **domains, goto done; } - csa = kmalloc(number_of_cpusets * sizeof(cp), GFP_KERNEL); + csa = kmalloc(nr_cpusets() * sizeof(cp), GFP_KERNEL); if (!csa) goto done; csn = 0; - list_add(&top_cpuset.stack_list, &q); - while (!list_empty(&q)) { - struct cgroup *cont; - struct cpuset *child; /* scans child cpusets of cp */ - - cp = list_first_entry(&q, struct cpuset, stack_list); - list_del(q.next); - - if (cpumask_empty(cp->cpus_allowed)) + rcu_read_lock(); + cpuset_for_each_descendant_pre(cp, pos_css, &top_cpuset) { + if (cp == &top_cpuset) continue; - /* - * 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. + * 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. */ - if (is_sched_load_balance(cp)) { - csa[csn++] = cp; + if (!cpumask_empty(cp->cpus_allowed) && + !is_sched_load_balance(cp)) continue; - } - list_for_each_entry(cont, &cp->css.cgroup->children, sibling) { - child = cgroup_cs(cont); - list_add_tail(&child->stack_list, &q); - } - } + if (is_sched_load_balance(cp)) + csa[csn++] = cp; + + /* skip @cp's subtree */ + pos_css = css_rightmost_descendant(pos_css); + } + rcu_read_unlock(); for (i = 0; i < csn; i++) csa[i]->pn = i; @@ -660,11 +691,8 @@ restart: 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\n", - nslot, ndoms, csn, i, apn); + pr_warn("rebuild_sched_domains confused: nslot %d, ndoms %d, csn %d, i %d, apn %d\n", + nslot, ndoms, csn, i, apn); warnings--; } continue; @@ -707,155 +735,163 @@ done: /* * Rebuild scheduler domains. * - * Call with neither cgroup_mutex held nor within get_online_cpus(). - * Takes both cgroup_mutex and get_online_cpus(). + * 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. * - * Cannot be directly called from cpuset code handling changes - * to the cpuset pseudo-filesystem, because it cannot be called - * from code that already holds cgroup_mutex. + * Call with cpuset_mutex held. Takes get_online_cpus(). */ -static void do_rebuild_sched_domains(struct work_struct *unused) +static void rebuild_sched_domains_locked(void) { struct sched_domain_attr *attr; cpumask_var_t *doms; int ndoms; + lockdep_assert_held(&cpuset_mutex); get_online_cpus(); + /* + * 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; + /* Generate domain masks and attrs */ - cgroup_lock(); ndoms = generate_sched_domains(&doms, &attr); - cgroup_unlock(); /* Have scheduler rebuild the domains */ partition_sched_domains(ndoms, doms, attr); - +out: put_online_cpus(); } #else /* !CONFIG_SMP */ -static void do_rebuild_sched_domains(struct work_struct *unused) +static void rebuild_sched_domains_locked(void) { } +#endif /* CONFIG_SMP */ -static int generate_sched_domains(cpumask_var_t **domains, - struct sched_domain_attr **attributes) +void rebuild_sched_domains(void) { - *domains = NULL; - return 1; + mutex_lock(&cpuset_mutex); + rebuild_sched_domains_locked(); + mutex_unlock(&cpuset_mutex); } -#endif /* CONFIG_SMP */ - -static DECLARE_WORK(rebuild_sched_domains_work, do_rebuild_sched_domains); /* - * Rebuild scheduler domains, asynchronously via workqueue. - * - * 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. + * effective_cpumask_cpuset - return nearest ancestor with non-empty cpus + * @cs: the cpuset in interest * - * The rebuild_sched_domains() and partition_sched_domains() - * routines must nest cgroup_lock() inside get_online_cpus(), - * but such cpuset changes as these must nest that locking the - * other way, holding cgroup_lock() for much of the code. + * 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. * - * So in order to avoid an ABBA deadlock, the cpuset code handling - * these user changes delegates the actual sched domain rebuilding - * to a separate workqueue thread, which ends up processing the - * above do_rebuild_sched_domains() function. + * Called with cpuset_mutex held. cpuset_cpus_allowed_fallback() is an + * exception. See comments there. */ -static void async_rebuild_sched_domains(void) +static struct cpuset *effective_cpumask_cpuset(struct cpuset *cs) { - queue_work(cpuset_wq, &rebuild_sched_domains_work); + while (cpumask_empty(cs->cpus_allowed)) + cs = parent_cs(cs); + return cs; } /* - * Accomplishes the same scheduler domain rebuild as the above - * async_rebuild_sched_domains(), however it directly calls the - * rebuild routine synchronously rather than calling it via an - * asynchronous work thread. + * effective_nodemask_cpuset - return nearest ancestor with non-empty mems + * @cs: the cpuset in interest * - * This can only be called from code that is not holding - * cgroup_mutex (not nested in a cgroup_lock() call.) - */ -void rebuild_sched_domains(void) -{ - do_rebuild_sched_domains(NULL); -} - -/** - * cpuset_test_cpumask - test a task's cpus_allowed versus its cpuset's - * @tsk: task to test - * @scan: struct cgroup_scanner contained in its struct cpuset_hotplug_scanner - * - * Call with cgroup_mutex held. May take callback_mutex during call. - * Called for each task in a cgroup by cgroup_scan_tasks(). - * Return nonzero if this tasks's cpus_allowed mask should be changed (in other - * words, if its mask is not equal to its cpuset's mask). + * 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. */ -static int cpuset_test_cpumask(struct task_struct *tsk, - struct cgroup_scanner *scan) +static struct cpuset *effective_nodemask_cpuset(struct cpuset *cs) { - return !cpumask_equal(&tsk->cpus_allowed, - (cgroup_cs(scan->cg))->cpus_allowed); + while (nodes_empty(cs->mems_allowed)) + cs = parent_cs(cs); + return cs; } /** - * cpuset_change_cpumask - make a task's cpus_allowed the same as its cpuset's - * @tsk: task to test - * @scan: struct cgroup_scanner containing the cgroup of the task - * - * Called by cgroup_scan_tasks() for each task in a cgroup whose - * cpus_allowed mask needs to be changed. + * 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 * - * We don't need to re-check for the cgroup/cpuset membership, since we're - * holding cgroup_lock() at this point. + * 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. */ -static void cpuset_change_cpumask(struct task_struct *tsk, - struct cgroup_scanner *scan) +static void update_tasks_cpumask(struct cpuset *cs) { - set_cpus_allowed_ptr(tsk, ((cgroup_cs(scan->cg))->cpus_allowed)); + 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); } -/** - * 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 - * @heap: if NULL, defer allocating heap memory to cgroup_scan_tasks() - * - * Called with cgroup_mutex held +/* + * 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? * - * The cgroup_scan_tasks() function will scan all the tasks in a cgroup, - * calling callback functions for each. + * This will update cpumasks of tasks in @root_cs and all other empty cpusets + * which take on cpumask of @root_cs. * - * No return value. It's guaranteed that cgroup_scan_tasks() always returns 0 - * if @heap != NULL. + * Called with cpuset_mutex held */ -static void update_tasks_cpumask(struct cpuset *cs, struct ptr_heap *heap) -{ - struct cgroup_scanner scan; +static void update_tasks_cpumask_hier(struct cpuset *root_cs, bool update_root) +{ + struct cpuset *cp; + struct cgroup_subsys_state *pos_css; + + rcu_read_lock(); + cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { + 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; + } + } + if (!css_tryget_online(&cp->css)) + continue; + rcu_read_unlock(); + + update_tasks_cpumask(cp); - scan.cg = cs->css.cgroup; - scan.test_task = cpuset_test_cpumask; - scan.process_task = cpuset_change_cpumask; - scan.heap = heap; - cgroup_scan_tasks(&scan); + rcu_read_lock(); + css_put(&cp->css); + } + rcu_read_unlock(); } /** * update_cpumask - update the cpus_allowed mask of a cpuset and all tasks in it * @cs: the cpuset to consider + * @trialcs: trial cpuset * @buf: buffer of cpu numbers written to this cpuset */ static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) { - struct ptr_heap heap; int retval; int is_load_balanced; - /* top_cpuset.cpus_allowed tracks cpu_online_map; it's read-only */ + /* top_cpuset.cpus_allowed tracks cpu_online_mask; it's read-only */ if (cs == &top_cpuset) return -EACCES; @@ -875,16 +911,13 @@ static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs, if (!cpumask_subset(trialcs->cpus_allowed, cpu_active_mask)) return -EINVAL; } - retval = validate_change(cs, trialcs); - if (retval < 0) - return retval; /* Nothing to do if the cpus didn't change */ if (cpumask_equal(cs->cpus_allowed, trialcs->cpus_allowed)) return 0; - retval = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL); - if (retval) + retval = validate_change(cs, trialcs); + if (retval < 0) return retval; is_load_balanced = is_sched_load_balance(trialcs); @@ -893,16 +926,10 @@ static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs, cpumask_copy(cs->cpus_allowed, trialcs->cpus_allowed); mutex_unlock(&callback_mutex); - /* - * Scan tasks in the cpuset, and update the cpumasks of any - * that need an update. - */ - update_tasks_cpumask(cs, &heap); - - heap_free(&heap); + update_tasks_cpumask_hier(cs, true); if (is_load_balanced) - async_rebuild_sched_domains(); + rebuild_sched_domains_locked(); return 0; } @@ -914,15 +941,6 @@ static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs, * Temporarilly set tasks mems_allowed to target nodes of migration, * so that the migration code can allocate pages on these nodes. * - * Call holding cgroup_mutex, so current's cpuset won't change - * during this call, as manage_mutex holds off any cpuset_attach() - * calls. Therefore we don't need to take task_lock around the - * call to guarantee_online_mems(), as we know no one is changing - * our task's cpuset. - * - * Hold callback_mutex around the two modifications of our tasks - * mems_allowed to synchronize with cpuset_mems_allowed(). - * * 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 @@ -933,12 +951,16 @@ static void cpuset_migrate_mm(struct mm_struct *mm, const nodemask_t *from, const nodemask_t *to) { struct task_struct *tsk = current; + struct cpuset *mems_cs; tsk->mems_allowed = *to; do_migrate_pages(mm, from, to, MPOL_MF_MOVE_ALL); - guarantee_online_mems(task_cs(tsk),&tsk->mems_allowed); + rcu_read_lock(); + mems_cs = effective_nodemask_cpuset(task_cs(tsk)); + guarantee_online_mems(mems_cs, &tsk->mems_allowed); + rcu_read_unlock(); } /* @@ -949,49 +971,48 @@ static void cpuset_migrate_mm(struct mm_struct *mm, const nodemask_t *from, * 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. - * - * Called with task's alloc_lock held */ static void cpuset_change_task_nodemask(struct task_struct *tsk, nodemask_t *newmems) { - nodes_or(tsk->mems_allowed, tsk->mems_allowed, *newmems); - mpol_rebind_task(tsk, &tsk->mems_allowed); - mpol_rebind_task(tsk, newmems); - tsk->mems_allowed = *newmems; -} + bool need_loop; -/* - * Update task's mems_allowed and rebind its mempolicy and vmas' mempolicy - * of it to cpuset's new mems_allowed, and migrate pages to new nodes if - * memory_migrate flag is set. Called with cgroup_mutex held. - */ -static void cpuset_change_nodemask(struct task_struct *p, - struct cgroup_scanner *scan) -{ - struct mm_struct *mm; - struct cpuset *cs; - int migrate; - const nodemask_t *oldmem = scan->data; - nodemask_t newmems; + /* + * 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; - cs = cgroup_cs(scan->cg); - guarantee_online_mems(cs, &newmems); + task_lock(tsk); + /* + * Determine if a loop is necessary if another thread is doing + * read_mems_allowed_begin(). If at least one node remains unchanged and + * 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); - task_lock(p); - cpuset_change_task_nodemask(p, &newmems); - task_unlock(p); + if (need_loop) { + local_irq_disable(); + write_seqcount_begin(&tsk->mems_allowed_seq); + } - mm = get_task_mm(p); - if (!mm) - return; + nodes_or(tsk->mems_allowed, tsk->mems_allowed, *newmems); + mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP1); - migrate = is_memory_migrate(cs); + mpol_rebind_task(tsk, newmems, MPOL_REBIND_STEP2); + tsk->mems_allowed = *newmems; + + if (need_loop) { + write_seqcount_end(&tsk->mems_allowed_seq); + local_irq_enable(); + } - mpol_rebind_mm(mm, &cs->mems_allowed); - if (migrate) - cpuset_migrate_mm(mm, oldmem, &cs->mems_allowed); - mmput(mm); + task_unlock(tsk); } static void *cpuset_being_rebound; @@ -999,43 +1020,102 @@ static void *cpuset_being_rebound; /** * 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 - * @oldmem: old mems_allowed of cpuset cs - * @heap: if NULL, defer allocating heap memory to cgroup_scan_tasks() * - * Called with cgroup_mutex held - * No return value. It's guaranteed that cgroup_scan_tasks() always returns 0 - * if @heap != NULL. + * 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. */ -static void update_tasks_nodemask(struct cpuset *cs, const nodemask_t *oldmem, - struct ptr_heap *heap) +static void update_tasks_nodemask(struct cpuset *cs) { - struct cgroup_scanner scan; + static nodemask_t newmems; /* protected by cpuset_mutex */ + struct cpuset *mems_cs = effective_nodemask_cpuset(cs); + struct css_task_iter it; + struct task_struct *task; cpuset_being_rebound = cs; /* causes mpol_dup() rebind */ - scan.cg = cs->css.cgroup; - scan.test_task = NULL; - scan.process_task = cpuset_change_nodemask; - scan.heap = heap; - scan.data = (nodemask_t *)oldmem; + guarantee_online_mems(mems_cs, &newmems); /* * 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 - * the global cgroup_mutex, we know that no other rebind effort + * the global cpuset_mutex, we know that no other rebind effort * will be contending for the global variable cpuset_being_rebound. * It's ok if we rebind the same mm twice; mpol_rebind_mm() * is idempotent. Also migrate pages in each mm to new nodes. */ - cgroup_scan_tasks(&scan); + 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); + + /* + * All the tasks' nodemasks have been updated, update + * cs->old_mems_allowed. + */ + cs->old_mems_allowed = newmems; /* We're done rebinding vmas to this cpuset's new mems_allowed. */ cpuset_being_rebound = NULL; } /* + * 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? + * + * 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 + */ +static void update_tasks_nodemask_hier(struct cpuset *root_cs, bool update_root) +{ + struct cpuset *cp; + struct cgroup_subsys_state *pos_css; + + rcu_read_lock(); + cpuset_for_each_descendant_pre(cp, pos_css, root_cs) { + 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; + } + } + if (!css_tryget_online(&cp->css)) + continue; + rcu_read_unlock(); + + update_tasks_nodemask(cp); + + rcu_read_lock(); + css_put(&cp->css); + } + rcu_read_unlock(); +} + +/* * Handle user request to change the 'mems' memory placement * of a cpuset. Needs to validate the request, update the * cpusets mems_allowed, and for each task in the cpuset, @@ -1043,7 +1123,7 @@ static void update_tasks_nodemask(struct cpuset *cs, const nodemask_t *oldmem, * mempolicies and if the cpuset is marked 'memory_migrate', * migrate the tasks pages to the new memory. * - * Call with cgroup_mutex held. May take callback_mutex during call. + * Call with cpuset_mutex held. May take callback_mutex during call. * 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. @@ -1051,16 +1131,16 @@ static void update_tasks_nodemask(struct cpuset *cs, const nodemask_t *oldmem, static int update_nodemask(struct cpuset *cs, struct cpuset *trialcs, const char *buf) { - nodemask_t oldmem; int retval; - struct ptr_heap heap; /* - * top_cpuset.mems_allowed tracks node_stats[N_HIGH_MEMORY]; + * top_cpuset.mems_allowed tracks node_stats[N_MEMORY]; * it's read-only */ - if (cs == &top_cpuset) - return -EACCES; + if (cs == &top_cpuset) { + retval = -EACCES; + goto done; + } /* * An empty mems_allowed is ok iff there are no tasks in the cpuset. @@ -1076,11 +1156,13 @@ static int update_nodemask(struct cpuset *cs, struct cpuset *trialcs, goto done; if (!nodes_subset(trialcs->mems_allowed, - node_states[N_HIGH_MEMORY])) - return -EINVAL; + node_states[N_MEMORY])) { + retval = -EINVAL; + goto done; + } } - oldmem = cs->mems_allowed; - if (nodes_equal(oldmem, trialcs->mems_allowed)) { + + if (nodes_equal(cs->mems_allowed, trialcs->mems_allowed)) { retval = 0; /* Too easy - nothing to do */ goto done; } @@ -1088,30 +1170,30 @@ static int update_nodemask(struct cpuset *cs, struct cpuset *trialcs, if (retval < 0) goto done; - retval = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL); - if (retval < 0) - goto done; - mutex_lock(&callback_mutex); cs->mems_allowed = trialcs->mems_allowed; mutex_unlock(&callback_mutex); - update_tasks_nodemask(cs, &oldmem, &heap); - - heap_free(&heap); + update_tasks_nodemask_hier(cs, true); done: return retval; } int current_cpuset_is_being_rebound(void) { - return task_cs(current) == cpuset_being_rebound; + int ret; + + rcu_read_lock(); + ret = task_cs(current) == cpuset_being_rebound; + rcu_read_unlock(); + + return ret; } static int update_relax_domain_level(struct cpuset *cs, s64 val) { #ifdef CONFIG_SMP - if (val < -1 || val >= SD_LV_MAX) + if (val < -1 || val >= sched_domain_level_max) return -EINVAL; #endif @@ -1119,50 +1201,29 @@ static int update_relax_domain_level(struct cpuset *cs, s64 val) cs->relax_domain_level = val; if (!cpumask_empty(cs->cpus_allowed) && is_sched_load_balance(cs)) - async_rebuild_sched_domains(); + rebuild_sched_domains_locked(); } return 0; } -/* - * cpuset_change_flag - make a task's spread flags the same as its cpuset's - * @tsk: task to be updated - * @scan: struct cgroup_scanner containing the cgroup of the task - * - * Called by cgroup_scan_tasks() for each task in a cgroup. - * - * We don't need to re-check for the cgroup/cpuset membership, since we're - * holding cgroup_lock() at this point. - */ -static void cpuset_change_flag(struct task_struct *tsk, - struct cgroup_scanner *scan) -{ - cpuset_update_task_spread_flag(cgroup_cs(scan->cg), tsk); -} - -/* +/** * 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 - * @heap: if NULL, defer allocating heap memory to cgroup_scan_tasks() - * - * Called with cgroup_mutex held - * - * The cgroup_scan_tasks() function will scan all the tasks in a cgroup, - * calling callback functions for each. * - * No return value. It's guaranteed that cgroup_scan_tasks() always returns 0 - * if @heap != NULL. + * Iterate through each task of @cs updating its spread flags. As this + * function is called with cpuset_mutex held, cpuset membership stays + * stable. */ -static void update_tasks_flags(struct cpuset *cs, struct ptr_heap *heap) +static void update_tasks_flags(struct cpuset *cs) { - struct cgroup_scanner scan; + struct css_task_iter it; + struct task_struct *task; - scan.cg = cs->css.cgroup; - scan.test_task = NULL; - scan.process_task = cpuset_change_flag; - scan.heap = heap; - cgroup_scan_tasks(&scan); + 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); } /* @@ -1171,7 +1232,7 @@ static void update_tasks_flags(struct cpuset *cs, struct ptr_heap *heap) * cs: the cpuset to update * turning_on: whether the flag is being set or cleared * - * Call with cgroup_mutex held. + * Call with cpuset_mutex held. */ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, @@ -1180,7 +1241,6 @@ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, struct cpuset *trialcs; int balance_flag_changed; int spread_flag_changed; - struct ptr_heap heap; int err; trialcs = alloc_trial_cpuset(cs); @@ -1196,10 +1256,6 @@ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, if (err < 0) goto out; - err = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL); - if (err < 0) - goto out; - balance_flag_changed = (is_sched_load_balance(cs) != is_sched_load_balance(trialcs)); @@ -1211,11 +1267,10 @@ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, mutex_unlock(&callback_mutex); if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed) - async_rebuild_sched_domains(); + rebuild_sched_domains_locked(); if (spread_flag_changed) - update_tasks_flags(cs, &heap); - heap_free(&heap); + update_tasks_flags(cs); out: free_trial_cpuset(trialcs); return err; @@ -1319,105 +1374,140 @@ static int fmeter_getrate(struct fmeter *fmp) return val; } -/* Protected by cgroup_lock */ -static cpumask_var_t cpus_attach; +static struct cpuset *cpuset_attach_old_cs; -/* Called by cgroups to determine if a cpuset is usable; cgroup_mutex held */ -static int cpuset_can_attach(struct cgroup_subsys *ss, struct cgroup *cont, - struct task_struct *tsk, bool threadgroup) +/* Called by cgroups to determine if a cpuset is usable; cpuset_mutex held */ +static int cpuset_can_attach(struct cgroup_subsys_state *css, + struct cgroup_taskset *tset) { + struct cpuset *cs = css_cs(css); + struct task_struct *task; int ret; - struct cpuset *cs = cgroup_cs(cont); - if (cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed)) - return -ENOSPC; + /* used later by cpuset_attach() */ + cpuset_attach_old_cs = task_cs(cgroup_taskset_first(tset)); + + mutex_lock(&cpuset_mutex); /* - * Kthreads bound to specific cpus cannot be moved to a new cpuset; we - * cannot 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. + * We allow to move tasks into an empty cpuset if sane_behavior + * flag is set. */ - if (tsk->flags & PF_THREAD_BOUND) - return -EINVAL; - - ret = security_task_setscheduler(tsk, 0, NULL); - if (ret) - return ret; - if (threadgroup) { - struct task_struct *c; + ret = -ENOSPC; + if (!cgroup_sane_behavior(css->cgroup) && + (cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed))) + goto out_unlock; - rcu_read_lock(); - list_for_each_entry_rcu(c, &tsk->thread_group, thread_group) { - ret = security_task_setscheduler(c, 0, NULL); - if (ret) { - rcu_read_unlock(); - return ret; - } - } - rcu_read_unlock(); + cgroup_taskset_for_each(task, tset) { + /* + * 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. + */ + ret = -EINVAL; + if (task->flags & PF_NO_SETAFFINITY) + goto out_unlock; + ret = security_task_setscheduler(task); + if (ret) + goto out_unlock; } - return 0; -} -static void cpuset_attach_task(struct task_struct *tsk, nodemask_t *to, - struct cpuset *cs) -{ - int err; /* - * can_attach beforehand should guarantee that this doesn't fail. - * TODO: have a better way to handle failure here + * Mark attach is in progress. This makes validate_change() fail + * changes which zero cpus/mems_allowed. */ - err = set_cpus_allowed_ptr(tsk, cpus_attach); - WARN_ON_ONCE(err); - - task_lock(tsk); - cpuset_change_task_nodemask(tsk, to); - task_unlock(tsk); - cpuset_update_task_spread_flag(cs, tsk); + cs->attach_in_progress++; + ret = 0; +out_unlock: + mutex_unlock(&cpuset_mutex); + return ret; +} +static void cpuset_cancel_attach(struct cgroup_subsys_state *css, + struct cgroup_taskset *tset) +{ + mutex_lock(&cpuset_mutex); + css_cs(css)->attach_in_progress--; + mutex_unlock(&cpuset_mutex); } -static void cpuset_attach(struct cgroup_subsys *ss, struct cgroup *cont, - struct cgroup *oldcont, struct task_struct *tsk, - bool threadgroup) +/* + * Protected by cpuset_mutex. cpus_attach is used only by cpuset_attach() + * but we can't allocate it dynamically there. Define it global and + * allocate from cpuset_init(). + */ +static cpumask_var_t cpus_attach; + +static void cpuset_attach(struct cgroup_subsys_state *css, + struct cgroup_taskset *tset) { - nodemask_t from, to; + /* static buf protected by cpuset_mutex */ + static nodemask_t cpuset_attach_nodemask_to; struct mm_struct *mm; - struct cpuset *cs = cgroup_cs(cont); - struct cpuset *oldcs = cgroup_cs(oldcont); + struct task_struct *task; + struct task_struct *leader = cgroup_taskset_first(tset); + struct cpuset *cs = css_cs(css); + struct cpuset *oldcs = cpuset_attach_old_cs; + struct cpuset *cpus_cs = effective_cpumask_cpuset(cs); + struct cpuset *mems_cs = effective_nodemask_cpuset(cs); - if (cs == &top_cpuset) { + mutex_lock(&cpuset_mutex); + + /* prepare for attach */ + if (cs == &top_cpuset) cpumask_copy(cpus_attach, cpu_possible_mask); - to = node_possible_map; - } else { - guarantee_online_cpus(cs, cpus_attach); - guarantee_online_mems(cs, &to); - } + else + guarantee_online_cpus(cpus_cs, cpus_attach); - /* do per-task migration stuff possibly for each in the threadgroup */ - cpuset_attach_task(tsk, &to, cs); - if (threadgroup) { - struct task_struct *c; - rcu_read_lock(); - list_for_each_entry_rcu(c, &tsk->thread_group, thread_group) { - cpuset_attach_task(c, &to, cs); - } - rcu_read_unlock(); + guarantee_online_mems(mems_cs, &cpuset_attach_nodemask_to); + + cgroup_taskset_for_each(task, tset) { + /* + * 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); } - /* change mm; only needs to be done once even if threadgroup */ - from = oldcs->mems_allowed; - to = cs->mems_allowed; - mm = get_task_mm(tsk); + /* + * Change mm, possibly for multiple threads in a threadgroup. This is + * expensive and may sleep. + */ + cpuset_attach_nodemask_to = cs->mems_allowed; + mm = get_task_mm(leader); if (mm) { - mpol_rebind_mm(mm, &to); - if (is_memory_migrate(cs)) - cpuset_migrate_mm(mm, &from, &to); + struct cpuset *mems_oldcs = effective_nodemask_cpuset(oldcs); + + mpol_rebind_mm(mm, &cpuset_attach_nodemask_to); + + /* + * old_mems_allowed is the same with mems_allowed here, except + * if this task is being moved automatically due to hotplug. + * In that case @mems_allowed has been updated and is empty, + * so @old_mems_allowed is the right nodesets that we migrate + * mm from. + */ + if (is_memory_migrate(cs)) { + cpuset_migrate_mm(mm, &mems_oldcs->old_mems_allowed, + &cpuset_attach_nodemask_to); + } mmput(mm); } + + cs->old_mems_allowed = cpuset_attach_nodemask_to; + + cs->attach_in_progress--; + if (!cs->attach_in_progress) + wake_up(&cpuset_attach_wq); + + mutex_unlock(&cpuset_mutex); } /* The various types of files and directories in a cpuset file system */ @@ -1437,14 +1527,18 @@ typedef enum { FILE_SPREAD_SLAB, } cpuset_filetype_t; -static int cpuset_write_u64(struct cgroup *cgrp, struct cftype *cft, u64 val) +static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, + u64 val) { - int retval = 0; - struct cpuset *cs = cgroup_cs(cgrp); + struct cpuset *cs = css_cs(css); cpuset_filetype_t type = cft->private; + int retval = 0; - if (!cgroup_lock_live_group(cgrp)) - return -ENODEV; + mutex_lock(&cpuset_mutex); + if (!is_cpuset_online(cs)) { + retval = -ENODEV; + goto out_unlock; + } switch (type) { case FILE_CPU_EXCLUSIVE: @@ -1478,18 +1572,21 @@ static int cpuset_write_u64(struct cgroup *cgrp, struct cftype *cft, u64 val) retval = -EINVAL; break; } - cgroup_unlock(); +out_unlock: + mutex_unlock(&cpuset_mutex); return retval; } -static int cpuset_write_s64(struct cgroup *cgrp, struct cftype *cft, s64 val) +static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, + s64 val) { - int retval = 0; - struct cpuset *cs = cgroup_cs(cgrp); + struct cpuset *cs = css_cs(css); cpuset_filetype_t type = cft->private; + int retval = -ENODEV; - if (!cgroup_lock_live_group(cgrp)) - return -ENODEV; + mutex_lock(&cpuset_mutex); + if (!is_cpuset_online(cs)) + goto out_unlock; switch (type) { case FILE_SCHED_RELAX_DOMAIN_LEVEL: @@ -1499,28 +1596,57 @@ static int cpuset_write_s64(struct cgroup *cgrp, struct cftype *cft, s64 val) retval = -EINVAL; break; } - cgroup_unlock(); +out_unlock: + mutex_unlock(&cpuset_mutex); return retval; } /* * Common handling for a write to a "cpus" or "mems" file. */ -static int cpuset_write_resmask(struct cgroup *cgrp, struct cftype *cft, - const char *buf) +static ssize_t cpuset_write_resmask(struct kernfs_open_file *of, + char *buf, size_t nbytes, loff_t off) { - int retval = 0; - struct cpuset *cs = cgroup_cs(cgrp); + struct cpuset *cs = css_cs(of_css(of)); struct cpuset *trialcs; + int retval = -ENODEV; - if (!cgroup_lock_live_group(cgrp)) - return -ENODEV; + buf = strstrip(buf); + + /* + * 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. + * + * cpuset_hotplug_work calls back into cgroup core via + * cgroup_transfer_tasks() and waiting for it from a cgroupfs + * operation like this one can lead to a deadlock through kernfs + * active_ref protection. Let's break the protection. Losing the + * protection is okay as we check whether @cs is online after + * grabbing cpuset_mutex anyway. This only happens on the legacy + * hierarchies. + */ + css_get(&cs->css); + kernfs_break_active_protection(of->kn); + flush_work(&cpuset_hotplug_work); + + mutex_lock(&cpuset_mutex); + if (!is_cpuset_online(cs)) + goto out_unlock; trialcs = alloc_trial_cpuset(cs); - if (!trialcs) - return -ENOMEM; + if (!trialcs) { + retval = -ENOMEM; + goto out_unlock; + } - switch (cft->private) { + switch (of_cft(of)->private) { case FILE_CPULIST: retval = update_cpumask(cs, trialcs, buf); break; @@ -1533,8 +1659,11 @@ static int cpuset_write_resmask(struct cgroup *cgrp, struct cftype *cft, } free_trial_cpuset(trialcs); - cgroup_unlock(); - return retval; +out_unlock: + mutex_unlock(&cpuset_mutex); + kernfs_unbreak_active_protection(of->kn); + css_put(&cs->css); + return retval ?: nbytes; } /* @@ -1544,72 +1673,46 @@ static int cpuset_write_resmask(struct cgroup *cgrp, struct cftype *cft, * 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. - * A single large read to a buffer that crosses a page boundary is - * ok, because the result being copied to user land is not recomputed - * across a page fault. */ - -static int cpuset_sprintf_cpulist(char *page, struct cpuset *cs) +static int cpuset_common_seq_show(struct seq_file *sf, void *v) { - int ret; - - mutex_lock(&callback_mutex); - ret = cpulist_scnprintf(page, PAGE_SIZE, cs->cpus_allowed); - mutex_unlock(&callback_mutex); + struct cpuset *cs = css_cs(seq_css(sf)); + cpuset_filetype_t type = seq_cft(sf)->private; + ssize_t count; + char *buf, *s; + int ret = 0; - return ret; -} - -static int cpuset_sprintf_memlist(char *page, struct cpuset *cs) -{ - nodemask_t mask; + count = seq_get_buf(sf, &buf); + s = buf; mutex_lock(&callback_mutex); - mask = cs->mems_allowed; - mutex_unlock(&callback_mutex); - - return nodelist_scnprintf(page, PAGE_SIZE, mask); -} - -static ssize_t cpuset_common_file_read(struct cgroup *cont, - struct cftype *cft, - struct file *file, - char __user *buf, - size_t nbytes, loff_t *ppos) -{ - struct cpuset *cs = cgroup_cs(cont); - cpuset_filetype_t type = cft->private; - char *page; - ssize_t retval = 0; - char *s; - - if (!(page = (char *)__get_free_page(GFP_TEMPORARY))) - return -ENOMEM; - - s = page; switch (type) { case FILE_CPULIST: - s += cpuset_sprintf_cpulist(s, cs); + s += cpulist_scnprintf(s, count, cs->cpus_allowed); break; case FILE_MEMLIST: - s += cpuset_sprintf_memlist(s, cs); + s += nodelist_scnprintf(s, count, cs->mems_allowed); break; default: - retval = -EINVAL; - goto out; + ret = -EINVAL; + goto out_unlock; } - *s++ = '\n'; - retval = simple_read_from_buffer(buf, nbytes, ppos, page, s - page); -out: - free_page((unsigned long)page); - return retval; + if (s < buf + count - 1) { + *s++ = '\n'; + seq_commit(sf, s - buf); + } else { + seq_commit(sf, -1); + } +out_unlock: + mutex_unlock(&callback_mutex); + return ret; } -static u64 cpuset_read_u64(struct cgroup *cont, struct cftype *cft) +static u64 cpuset_read_u64(struct cgroup_subsys_state *css, struct cftype *cft) { - struct cpuset *cs = cgroup_cs(cont); + struct cpuset *cs = css_cs(css); cpuset_filetype_t type = cft->private; switch (type) { case FILE_CPU_EXCLUSIVE: @@ -1638,9 +1741,9 @@ static u64 cpuset_read_u64(struct cgroup *cont, struct cftype *cft) return 0; } -static s64 cpuset_read_s64(struct cgroup *cont, struct cftype *cft) +static s64 cpuset_read_s64(struct cgroup_subsys_state *css, struct cftype *cft) { - struct cpuset *cs = cgroup_cs(cont); + struct cpuset *cs = css_cs(css); cpuset_filetype_t type = cft->private; switch (type) { case FILE_SCHED_RELAX_DOMAIN_LEVEL: @@ -1661,16 +1764,16 @@ static s64 cpuset_read_s64(struct cgroup *cont, struct cftype *cft) static struct cftype files[] = { { .name = "cpus", - .read = cpuset_common_file_read, - .write_string = cpuset_write_resmask, + .seq_show = cpuset_common_seq_show, + .write = cpuset_write_resmask, .max_write_len = (100U + 6 * NR_CPUS), .private = FILE_CPULIST, }, { .name = "mems", - .read = cpuset_common_file_read, - .write_string = cpuset_write_resmask, + .seq_show = cpuset_common_seq_show, + .write = cpuset_write_resmask, .max_write_len = (100U + 6 * MAX_NUMNODES), .private = FILE_MEMLIST, }, @@ -1738,84 +1841,32 @@ static struct cftype files[] = { .write_u64 = cpuset_write_u64, .private = FILE_SPREAD_SLAB, }, -}; - -static struct cftype cft_memory_pressure_enabled = { - .name = "memory_pressure_enabled", - .read_u64 = cpuset_read_u64, - .write_u64 = cpuset_write_u64, - .private = FILE_MEMORY_PRESSURE_ENABLED, -}; - -static int cpuset_populate(struct cgroup_subsys *ss, struct cgroup *cont) -{ - int err; - - err = cgroup_add_files(cont, ss, files, ARRAY_SIZE(files)); - if (err) - return err; - /* memory_pressure_enabled is in root cpuset only */ - if (!cont->parent) - err = cgroup_add_file(cont, ss, - &cft_memory_pressure_enabled); - return err; -} - -/* - * post_clone() is called at the end of cgroup_clone(). - * 'cgroup' was just created automatically as a result of - * a cgroup_clone(), and the current task is about to - * be moved into 'cgroup'. - * - * Currently we refuse to set up the cgroup - thereby - * refusing the task to be entered, and as a result refusing - * the sys_unshare() or clone() which initiated it - if any - * sibling cpusets have exclusive cpus or mem. - * - * If this becomes a problem for some users who wish to - * allow that scenario, then cpuset_post_clone() could be - * changed to grant parent->cpus_allowed-sibling_cpus_exclusive - * (and likewise for mems) to the new cgroup. Called with cgroup_mutex - * held. - */ -static void cpuset_post_clone(struct cgroup_subsys *ss, - struct cgroup *cgroup) -{ - struct cgroup *parent, *child; - struct cpuset *cs, *parent_cs; - parent = cgroup->parent; - list_for_each_entry(child, &parent->children, sibling) { - cs = cgroup_cs(child); - if (is_mem_exclusive(cs) || is_cpu_exclusive(cs)) - return; - } - cs = cgroup_cs(cgroup); - parent_cs = cgroup_cs(parent); + { + .name = "memory_pressure_enabled", + .flags = CFTYPE_ONLY_ON_ROOT, + .read_u64 = cpuset_read_u64, + .write_u64 = cpuset_write_u64, + .private = FILE_MEMORY_PRESSURE_ENABLED, + }, - cs->mems_allowed = parent_cs->mems_allowed; - cpumask_copy(cs->cpus_allowed, parent_cs->cpus_allowed); - return; -} + { } /* terminate */ +}; /* - * cpuset_create - create a cpuset - * ss: cpuset cgroup subsystem - * cont: control group that the new cpuset will be part of + * cpuset_css_alloc - allocate a cpuset css + * cgrp: control group that the new cpuset will be part of */ -static struct cgroup_subsys_state *cpuset_create( - struct cgroup_subsys *ss, - struct cgroup *cont) +static struct cgroup_subsys_state * +cpuset_css_alloc(struct cgroup_subsys_state *parent_css) { struct cpuset *cs; - struct cpuset *parent; - if (!cont->parent) { + if (!parent_css) return &top_cpuset.css; - } - parent = cgroup_cs(cont->parent); - cs = kmalloc(sizeof(*cs), GFP_KERNEL); + + cs = kzalloc(sizeof(*cs), GFP_KERNEL); if (!cs) return ERR_PTR(-ENOMEM); if (!alloc_cpumask_var(&cs->cpus_allowed, GFP_KERNEL)) { @@ -1823,49 +1874,107 @@ static struct cgroup_subsys_state *cpuset_create( return ERR_PTR(-ENOMEM); } - cs->flags = 0; - if (is_spread_page(parent)) - set_bit(CS_SPREAD_PAGE, &cs->flags); - if (is_spread_slab(parent)) - set_bit(CS_SPREAD_SLAB, &cs->flags); set_bit(CS_SCHED_LOAD_BALANCE, &cs->flags); cpumask_clear(cs->cpus_allowed); nodes_clear(cs->mems_allowed); fmeter_init(&cs->fmeter); cs->relax_domain_level = -1; - cs->parent = parent; - number_of_cpusets++; - return &cs->css ; + return &cs->css; +} + +static int cpuset_css_online(struct cgroup_subsys_state *css) +{ + struct cpuset *cs = css_cs(css); + struct cpuset *parent = parent_cs(cs); + struct cpuset *tmp_cs; + struct cgroup_subsys_state *pos_css; + + if (!parent) + return 0; + + mutex_lock(&cpuset_mutex); + + set_bit(CS_ONLINE, &cs->flags); + if (is_spread_page(parent)) + set_bit(CS_SPREAD_PAGE, &cs->flags); + if (is_spread_slab(parent)) + set_bit(CS_SPREAD_SLAB, &cs->flags); + + cpuset_inc(); + + if (!test_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags)) + goto out_unlock; + + /* + * 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. + */ + rcu_read_lock(); + cpuset_for_each_child(tmp_cs, pos_css, parent) { + if (is_mem_exclusive(tmp_cs) || is_cpu_exclusive(tmp_cs)) { + rcu_read_unlock(); + goto out_unlock; + } + } + rcu_read_unlock(); + + mutex_lock(&callback_mutex); + cs->mems_allowed = parent->mems_allowed; + cpumask_copy(cs->cpus_allowed, parent->cpus_allowed); + mutex_unlock(&callback_mutex); +out_unlock: + mutex_unlock(&cpuset_mutex); + return 0; } /* * If the cpuset being removed has its flag 'sched_load_balance' * enabled, then simulate turning sched_load_balance off, which - * will call async_rebuild_sched_domains(). + * will call rebuild_sched_domains_locked(). */ -static void cpuset_destroy(struct cgroup_subsys *ss, struct cgroup *cont) +static void cpuset_css_offline(struct cgroup_subsys_state *css) { - struct cpuset *cs = cgroup_cs(cont); + struct cpuset *cs = css_cs(css); + + mutex_lock(&cpuset_mutex); if (is_sched_load_balance(cs)) update_flag(CS_SCHED_LOAD_BALANCE, cs, 0); - number_of_cpusets--; + cpuset_dec(); + clear_bit(CS_ONLINE, &cs->flags); + + mutex_unlock(&cpuset_mutex); +} + +static void cpuset_css_free(struct cgroup_subsys_state *css) +{ + struct cpuset *cs = css_cs(css); + free_cpumask_var(cs->cpus_allowed); kfree(cs); } -struct cgroup_subsys cpuset_subsys = { - .name = "cpuset", - .create = cpuset_create, - .destroy = cpuset_destroy, +struct cgroup_subsys cpuset_cgrp_subsys = { + .css_alloc = cpuset_css_alloc, + .css_online = cpuset_css_online, + .css_offline = cpuset_css_offline, + .css_free = cpuset_css_free, .can_attach = cpuset_can_attach, + .cancel_attach = cpuset_cancel_attach, .attach = cpuset_attach, - .populate = cpuset_populate, - .post_clone = cpuset_post_clone, - .subsys_id = cpuset_subsys_id, + .base_cftypes = files, .early_init = 1, }; @@ -1896,234 +2005,230 @@ int __init cpuset_init(void) if (!alloc_cpumask_var(&cpus_attach, GFP_KERNEL)) BUG(); - number_of_cpusets = 1; return 0; } -/** - * cpuset_do_move_task - move a given task to another cpuset - * @tsk: pointer to task_struct the task to move - * @scan: struct cgroup_scanner contained in its struct cpuset_hotplug_scanner - * - * Called by cgroup_scan_tasks() for each task in a cgroup. - * Return nonzero to stop the walk through the tasks. - */ -static void cpuset_do_move_task(struct task_struct *tsk, - struct cgroup_scanner *scan) -{ - struct cgroup *new_cgroup = scan->data; - - cgroup_attach_task(new_cgroup, tsk); -} - -/** - * move_member_tasks_to_cpuset - move tasks from one cpuset to another - * @from: cpuset in which the tasks currently reside - * @to: cpuset to which the tasks will be moved - * - * Called with cgroup_mutex held - * callback_mutex must not be held, as cpuset_attach() will take it. - * - * The cgroup_scan_tasks() function will scan all the tasks in a cgroup, - * calling callback functions for each. - */ -static void move_member_tasks_to_cpuset(struct cpuset *from, struct cpuset *to) -{ - struct cgroup_scanner scan; - - scan.cg = from->css.cgroup; - scan.test_task = NULL; /* select all tasks in cgroup */ - scan.process_task = cpuset_do_move_task; - scan.heap = NULL; - scan.data = to->css.cgroup; - - if (cgroup_scan_tasks(&scan)) - printk(KERN_ERR "move_member_tasks_to_cpuset: " - "cgroup_scan_tasks failed\n"); -} - /* * If CPU and/or memory hotplug handlers, below, unplug any CPUs * or memory nodes, we need to walk over the cpuset hierarchy, * removing that CPU or node from all cpusets. If this removes the * last CPU or node from a cpuset, then move the tasks in the empty * cpuset to its next-highest non-empty parent. - * - * Called with cgroup_mutex held - * callback_mutex must not be held, as cpuset_attach() will take it. */ static void remove_tasks_in_empty_cpuset(struct cpuset *cs) { struct cpuset *parent; /* - * The cgroup's css_sets list is in use if there are tasks - * in the cpuset; the list is empty if there are none; - * the cs->css.refcnt seems always 0. - */ - if (list_empty(&cs->css.cgroup->css_sets)) - return; - - /* * Find its next-highest non-empty parent, (top cpuset * has online cpus, so can't be empty). */ - parent = cs->parent; + parent = parent_cs(cs); while (cpumask_empty(parent->cpus_allowed) || nodes_empty(parent->mems_allowed)) - parent = parent->parent; + parent = parent_cs(parent); - move_member_tasks_to_cpuset(cs, parent); + if (cgroup_transfer_tasks(parent->css.cgroup, cs->css.cgroup)) { + pr_err("cpuset: failed to transfer tasks out of empty cpuset "); + pr_cont_cgroup_name(cs->css.cgroup); + pr_cont("\n"); + } } -/* - * Walk the specified cpuset subtree and look for empty cpusets. - * The tasks of such cpuset must be moved to a parent cpuset. - * - * Called with cgroup_mutex held. We take callback_mutex to modify - * cpus_allowed and mems_allowed. - * - * This walk processes the tree from top to bottom, completing one layer - * before dropping down to the next. It always processes a node before - * any of its children. +/** + * cpuset_hotplug_update_tasks - update tasks in a cpuset for hotunplug + * @cs: cpuset in interest * - * For now, since we lack memory hot unplug, we'll never see a cpuset - * that has tasks along with an empty 'mems'. But if we did see such - * a cpuset, we'd handle it just like we do if its 'cpus' was empty. + * 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. */ -static void scan_for_empty_cpusets(struct cpuset *root) -{ - LIST_HEAD(queue); - struct cpuset *cp; /* scans cpusets being updated */ - struct cpuset *child; /* scans child cpusets of cp */ - struct cgroup *cont; - nodemask_t oldmems; - - list_add_tail((struct list_head *)&root->stack_list, &queue); - - while (!list_empty(&queue)) { - cp = list_first_entry(&queue, struct cpuset, stack_list); - list_del(queue.next); - list_for_each_entry(cont, &cp->css.cgroup->children, sibling) { - child = cgroup_cs(cont); - list_add_tail(&child->stack_list, &queue); - } +static void cpuset_hotplug_update_tasks(struct cpuset *cs) +{ + static cpumask_t off_cpus; + static nodemask_t off_mems; + bool is_empty; + bool sane = cgroup_sane_behavior(cs->css.cgroup); - /* Continue past cpusets with all cpus, mems online */ - if (cpumask_subset(cp->cpus_allowed, cpu_active_mask) && - nodes_subset(cp->mems_allowed, node_states[N_HIGH_MEMORY])) - continue; +retry: + wait_event(cpuset_attach_wq, cs->attach_in_progress == 0); - oldmems = cp->mems_allowed; + mutex_lock(&cpuset_mutex); - /* Remove offline cpus and mems from this cpuset. */ - mutex_lock(&callback_mutex); - cpumask_and(cp->cpus_allowed, cp->cpus_allowed, - cpu_active_mask); - nodes_and(cp->mems_allowed, cp->mems_allowed, - node_states[N_HIGH_MEMORY]); - mutex_unlock(&callback_mutex); - - /* Move tasks from the empty cpuset to a parent */ - if (cpumask_empty(cp->cpus_allowed) || - nodes_empty(cp->mems_allowed)) - remove_tasks_in_empty_cpuset(cp); - else { - update_tasks_cpumask(cp, NULL); - update_tasks_nodemask(cp, &oldmems, NULL); - } + /* + * 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; } + + cpumask_andnot(&off_cpus, cs->cpus_allowed, top_cpuset.cpus_allowed); + nodes_andnot(off_mems, cs->mems_allowed, top_cpuset.mems_allowed); + + 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 + * 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. + */ + if ((sane && cpumask_empty(cs->cpus_allowed)) || + (!cpumask_empty(&off_cpus) && !cpumask_empty(cs->cpus_allowed))) + update_tasks_cpumask(cs); + + 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 + * 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. + */ + if ((sane && nodes_empty(cs->mems_allowed)) || + (!nodes_empty(off_mems) && !nodes_empty(cs->mems_allowed))) + update_tasks_nodemask(cs); + + is_empty = cpumask_empty(cs->cpus_allowed) || + nodes_empty(cs->mems_allowed); + + mutex_unlock(&cpuset_mutex); + + /* + * 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 + * also call back into cpuset. Should be done outside any lock. + */ + if (!sane && is_empty) + remove_tasks_in_empty_cpuset(cs); } -/* - * The top_cpuset tracks what CPUs and Memory Nodes are online, - * period. This 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. +/** + * cpuset_hotplug_workfn - handle CPU/memory hotunplug for a cpuset + * + * 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. * - * This routine ensures that top_cpuset.cpus_allowed tracks - * cpu_online_map on each CPU hotplug (cpuhp) event. + * Non-root cpusets are only affected by offlining. If any CPUs or memory + * nodes have been taken down, cpuset_hotplug_update_tasks() is invoked on + * all descendants. * - * Called within get_online_cpus(). Needs to call cgroup_lock() - * before calling generate_sched_domains(). + * Note that CPU offlining during suspend is ignored. We don't modify + * cpusets across suspend/resume cycles at all. */ -static int cpuset_track_online_cpus(struct notifier_block *unused_nb, - unsigned long phase, void *unused_cpu) +static void cpuset_hotplug_workfn(struct work_struct *work) { - struct sched_domain_attr *attr; - cpumask_var_t *doms; - int ndoms; + static cpumask_t new_cpus; + static nodemask_t new_mems; + bool cpus_updated, mems_updated; - switch (phase) { - case CPU_ONLINE: - case CPU_ONLINE_FROZEN: - case CPU_DOWN_PREPARE: - case CPU_DOWN_PREPARE_FROZEN: - case CPU_DOWN_FAILED: - case CPU_DOWN_FAILED_FROZEN: - break; + mutex_lock(&cpuset_mutex); - default: - return NOTIFY_DONE; + /* fetch the available cpus/mems and find out which changed how */ + cpumask_copy(&new_cpus, cpu_active_mask); + new_mems = node_states[N_MEMORY]; + + cpus_updated = !cpumask_equal(top_cpuset.cpus_allowed, &new_cpus); + mems_updated = !nodes_equal(top_cpuset.mems_allowed, new_mems); + + /* 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 */ } - cgroup_lock(); - mutex_lock(&callback_mutex); - cpumask_copy(top_cpuset.cpus_allowed, cpu_active_mask); - mutex_unlock(&callback_mutex); - scan_for_empty_cpusets(&top_cpuset); - ndoms = generate_sched_domains(&doms, &attr); - cgroup_unlock(); + /* synchronize mems_allowed to N_MEMORY */ + if (mems_updated) { + mutex_lock(&callback_mutex); + top_cpuset.mems_allowed = new_mems; + mutex_unlock(&callback_mutex); + update_tasks_nodemask(&top_cpuset); + } - /* Have scheduler rebuild the domains */ - partition_sched_domains(ndoms, doms, attr); + mutex_unlock(&cpuset_mutex); - return NOTIFY_OK; + /* if cpus or mems changed, we need to propagate to descendants */ + if (cpus_updated || mems_updated) { + struct cpuset *cs; + struct cgroup_subsys_state *pos_css; + + rcu_read_lock(); + cpuset_for_each_descendant_pre(cs, pos_css, &top_cpuset) { + if (cs == &top_cpuset || !css_tryget_online(&cs->css)) + continue; + rcu_read_unlock(); + + cpuset_hotplug_update_tasks(cs); + + rcu_read_lock(); + css_put(&cs->css); + } + rcu_read_unlock(); + } + + /* rebuild sched domains if cpus_allowed has changed */ + if (cpus_updated) + rebuild_sched_domains(); +} + +void cpuset_update_active_cpus(bool cpu_online) +{ + /* + * 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); } -#ifdef CONFIG_MEMORY_HOTPLUG /* - * Keep top_cpuset.mems_allowed tracking node_states[N_HIGH_MEMORY]. - * Call this routine anytime after node_states[N_HIGH_MEMORY] changes. - * See also the previous routine cpuset_track_online_cpus(). + * Keep top_cpuset.mems_allowed tracking node_states[N_MEMORY]. + * Call this routine anytime after node_states[N_MEMORY] changes. + * See cpuset_update_active_cpus() for CPU hotplug handling. */ static int cpuset_track_online_nodes(struct notifier_block *self, unsigned long action, void *arg) { - cgroup_lock(); - switch (action) { - case MEM_ONLINE: - case MEM_OFFLINE: - mutex_lock(&callback_mutex); - top_cpuset.mems_allowed = node_states[N_HIGH_MEMORY]; - mutex_unlock(&callback_mutex); - if (action == MEM_OFFLINE) - scan_for_empty_cpusets(&top_cpuset); - break; - default: - break; - } - cgroup_unlock(); + schedule_work(&cpuset_hotplug_work); return NOTIFY_OK; } -#endif + +static struct notifier_block cpuset_track_online_nodes_nb = { + .notifier_call = cpuset_track_online_nodes, + .priority = 10, /* ??! */ +}; /** * cpuset_init_smp - initialize cpus_allowed * * Description: Finish top cpuset after cpu, node maps are initialized - **/ - + */ void __init cpuset_init_smp(void) { cpumask_copy(top_cpuset.cpus_allowed, cpu_active_mask); - top_cpuset.mems_allowed = node_states[N_HIGH_MEMORY]; - - hotcpu_notifier(cpuset_track_online_cpus, 0); - hotplug_memory_notifier(cpuset_track_online_nodes, 10); + top_cpuset.mems_allowed = node_states[N_MEMORY]; + top_cpuset.old_mems_allowed = top_cpuset.mems_allowed; - cpuset_wq = create_singlethread_workqueue("cpuset"); - BUG_ON(!cpuset_wq); + register_hotmemory_notifier(&cpuset_track_online_nodes_nb); } /** @@ -2133,26 +2238,48 @@ void __init cpuset_init_smp(void) * * Description: Returns the cpumask_var_t cpus_allowed of the cpuset * attached to the specified @tsk. Guaranteed to return some non-empty - * subset of cpu_online_map, even if this means going outside the + * subset of cpu_online_mask, even if this means going outside the * tasks cpuset. **/ void cpuset_cpus_allowed(struct task_struct *tsk, struct cpumask *pmask) { + struct cpuset *cpus_cs; + mutex_lock(&callback_mutex); - cpuset_cpus_allowed_locked(tsk, pmask); + rcu_read_lock(); + cpus_cs = effective_cpumask_cpuset(task_cs(tsk)); + guarantee_online_cpus(cpus_cs, pmask); + rcu_read_unlock(); mutex_unlock(&callback_mutex); } -/** - * cpuset_cpus_allowed_locked - return cpus_allowed mask from a tasks cpuset. - * Must be called with callback_mutex held. - **/ -void cpuset_cpus_allowed_locked(struct task_struct *tsk, struct cpumask *pmask) +void cpuset_cpus_allowed_fallback(struct task_struct *tsk) { - task_lock(tsk); - guarantee_online_cpus(task_cs(tsk), pmask); - task_unlock(tsk); + struct cpuset *cpus_cs; + + rcu_read_lock(); + cpus_cs = effective_cpumask_cpuset(task_cs(tsk)); + do_set_cpus_allowed(tsk, cpus_cs->cpus_allowed); + 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. + * + * select_fallback_rq() will fix things ups and set cpu_possible_mask + * if required. + */ } void cpuset_init_current_mems_allowed(void) @@ -2166,18 +2293,20 @@ void cpuset_init_current_mems_allowed(void) * * Description: Returns the nodemask_t mems_allowed of the cpuset * attached to the specified @tsk. Guaranteed to return some non-empty - * subset of node_states[N_HIGH_MEMORY], even if this means going outside the + * subset of node_states[N_MEMORY], even if this means going outside the * tasks cpuset. **/ nodemask_t cpuset_mems_allowed(struct task_struct *tsk) { + struct cpuset *mems_cs; nodemask_t mask; mutex_lock(&callback_mutex); - task_lock(tsk); - guarantee_online_mems(task_cs(tsk), &mask); - task_unlock(tsk); + rcu_read_lock(); + mems_cs = effective_nodemask_cpuset(task_cs(tsk)); + guarantee_online_mems(mems_cs, &mask); + rcu_read_unlock(); mutex_unlock(&callback_mutex); return mask; @@ -2200,10 +2329,10 @@ int cpuset_nodemask_valid_mems_allowed(nodemask_t *nodemask) * callback_mutex. If no ancestor is mem_exclusive or mem_hardwall * (an unusual configuration), then returns the root cpuset. */ -static const struct cpuset *nearest_hardwall_ancestor(const struct cpuset *cs) +static struct cpuset *nearest_hardwall_ancestor(struct cpuset *cs) { - while (!(is_mem_exclusive(cs) || is_mem_hardwall(cs)) && cs->parent) - cs = cs->parent; + while (!(is_mem_exclusive(cs) || is_mem_hardwall(cs)) && parent_cs(cs)) + cs = parent_cs(cs); return cs; } @@ -2270,7 +2399,7 @@ static const struct cpuset *nearest_hardwall_ancestor(const struct cpuset *cs) */ int __cpuset_node_allowed_softwall(int node, gfp_t gfp_mask) { - const struct cpuset *cs; /* current cpuset ancestors */ + struct cpuset *cs; /* current cpuset ancestors */ int allowed; /* is allocation in zone z allowed? */ if (in_interrupt() || (gfp_mask & __GFP_THISNODE)) @@ -2293,11 +2422,11 @@ int __cpuset_node_allowed_softwall(int node, gfp_t gfp_mask) /* Not hardwall and node outside mems_allowed: scan up cpusets */ mutex_lock(&callback_mutex); - task_lock(current); + rcu_read_lock(); cs = nearest_hardwall_ancestor(task_cs(current)); - task_unlock(current); - allowed = node_isset(node, cs->mems_allowed); + rcu_read_unlock(); + mutex_unlock(&callback_mutex); return allowed; } @@ -2341,34 +2470,8 @@ int __cpuset_node_allowed_hardwall(int node, gfp_t gfp_mask) } /** - * cpuset_lock - lock out any changes to cpuset structures - * - * The out of memory (oom) code needs to mutex_lock cpusets - * from being changed while it scans the tasklist looking for a - * task in an overlapping cpuset. Expose callback_mutex via this - * cpuset_lock() routine, so the oom code can lock it, before - * locking the task list. The tasklist_lock is a spinlock, so - * must be taken inside callback_mutex. - */ - -void cpuset_lock(void) -{ - mutex_lock(&callback_mutex); -} - -/** - * cpuset_unlock - release lock on cpuset changes - * - * Undo the lock taken in a previous cpuset_lock() call. - */ - -void cpuset_unlock(void) -{ - mutex_unlock(&callback_mutex); -} - -/** - * cpuset_mem_spread_node() - On which node to begin search for a page + * 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 * * 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), @@ -2393,16 +2496,35 @@ void cpuset_unlock(void) * See kmem_cache_alloc_node(). */ -int cpuset_mem_spread_node(void) +static int cpuset_spread_node(int *rotor) { int node; - node = next_node(current->cpuset_mem_spread_rotor, current->mems_allowed); + node = next_node(*rotor, current->mems_allowed); if (node == MAX_NUMNODES) node = first_node(current->mems_allowed); - current->cpuset_mem_spread_rotor = node; + *rotor = node; return node; } + +int cpuset_mem_spread_node(void) +{ + if (current->cpuset_mem_spread_rotor == NUMA_NO_NODE) + current->cpuset_mem_spread_rotor = + node_random(¤t->mems_allowed); + + return cpuset_spread_node(¤t->cpuset_mem_spread_rotor); +} + +int cpuset_slab_spread_node(void) +{ + if (current->cpuset_slab_spread_rotor == NUMA_NO_NODE) + current->cpuset_slab_spread_rotor = + node_random(¤t->mems_allowed); + + return cpuset_spread_node(¤t->cpuset_slab_spread_rotor); +} + EXPORT_SYMBOL_GPL(cpuset_mem_spread_node); /** @@ -2422,26 +2544,33 @@ int cpuset_mems_allowed_intersects(const struct task_struct *tsk1, return nodes_intersects(tsk1->mems_allowed, tsk2->mems_allowed); } +#define CPUSET_NODELIST_LEN (256) + /** * cpuset_print_task_mems_allowed - prints task's cpuset and mems_allowed - * @task: pointer to task_struct of some task. + * @tsk: pointer to task_struct of some task. * * Description: Prints @task's name, cpuset name, and cached copy of its - * mems_allowed to the kernel log. Must hold task_lock(task) to allow - * dereferencing task_cs(task). + * mems_allowed to the kernel log. */ void cpuset_print_task_mems_allowed(struct task_struct *tsk) { - struct dentry *dentry; + /* Statically allocated to prevent using excess stack. */ + static char cpuset_nodelist[CPUSET_NODELIST_LEN]; + static DEFINE_SPINLOCK(cpuset_buffer_lock); + struct cgroup *cgrp; - dentry = task_cs(tsk)->css.cgroup->dentry; spin_lock(&cpuset_buffer_lock); - snprintf(cpuset_name, CPUSET_NAME_LEN, - dentry ? (const char *)dentry->d_name.name : "/"); + rcu_read_lock(); + + cgrp = task_cs(tsk)->css.cgroup; nodelist_scnprintf(cpuset_nodelist, CPUSET_NODELIST_LEN, tsk->mems_allowed); - printk(KERN_INFO "%s cpuset=%s mems_allowed=%s\n", - tsk->comm, cpuset_name, cpuset_nodelist); + pr_info("%s cpuset=", tsk->comm); + pr_cont_cgroup_name(cgrp); + pr_cont(" mems_allowed=%s\n", cpuset_nodelist); + + rcu_read_unlock(); spin_unlock(&cpuset_buffer_lock); } @@ -2473,9 +2602,9 @@ int cpuset_memory_pressure_enabled __read_mostly; void __cpuset_memory_pressure_bump(void) { - task_lock(current); + rcu_read_lock(); fmeter_markevent(&task_cs(current)->fmeter); - task_unlock(current); + rcu_read_unlock(); } #ifdef CONFIG_PROC_PID_CPUSET @@ -2485,19 +2614,19 @@ void __cpuset_memory_pressure_bump(void) * - Used for /proc/<pid>/cpuset. * - 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, - * and we take cgroup_mutex, keeping cpuset_attach() from changing it + * and we take cpuset_mutex, keeping cpuset_attach() from changing it * anyway. */ -static int proc_cpuset_show(struct seq_file *m, void *unused_v) +int proc_cpuset_show(struct seq_file *m, void *unused_v) { struct pid *pid; struct task_struct *tsk; - char *buf; + char *buf, *p; struct cgroup_subsys_state *css; int retval; retval = -ENOMEM; - buf = kmalloc(PAGE_SIZE, GFP_KERNEL); + buf = kmalloc(PATH_MAX, GFP_KERNEL); if (!buf) goto out; @@ -2507,44 +2636,32 @@ static int proc_cpuset_show(struct seq_file *m, void *unused_v) if (!tsk) goto out_free; - retval = -EINVAL; - cgroup_lock(); - css = task_subsys_state(tsk, cpuset_subsys_id); - retval = cgroup_path(css->cgroup, buf, PAGE_SIZE); - if (retval < 0) - goto out_unlock; - seq_puts(m, buf); + retval = -ENAMETOOLONG; + rcu_read_lock(); + css = task_css(tsk, cpuset_cgrp_id); + p = cgroup_path(css->cgroup, buf, PATH_MAX); + rcu_read_unlock(); + if (!p) + goto out_put_task; + seq_puts(m, p); seq_putc(m, '\n'); -out_unlock: - cgroup_unlock(); + retval = 0; +out_put_task: put_task_struct(tsk); out_free: kfree(buf); out: return retval; } - -static int cpuset_open(struct inode *inode, struct file *file) -{ - struct pid *pid = PROC_I(inode)->pid; - return single_open(file, proc_cpuset_show, pid); -} - -const struct file_operations proc_cpuset_operations = { - .open = cpuset_open, - .read = seq_read, - .llseek = seq_lseek, - .release = single_release, -}; #endif /* CONFIG_PROC_PID_CPUSET */ /* Display task mems_allowed in /proc/<pid>/status file. */ void cpuset_task_status_allowed(struct seq_file *m, struct task_struct *task) { - seq_printf(m, "Mems_allowed:\t"); + seq_puts(m, "Mems_allowed:\t"); seq_nodemask(m, &task->mems_allowed); - seq_printf(m, "\n"); - seq_printf(m, "Mems_allowed_list:\t"); + seq_puts(m, "\n"); + seq_puts(m, "Mems_allowed_list:\t"); seq_nodemask_list(m, &task->mems_allowed); - seq_printf(m, "\n"); + seq_puts(m, "\n"); } |