From 554ec22f075d46e4363520a407d2b7eeb5dfdd43 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:03 +0200 Subject: namespacecheck: more sched.c fixes [ Stephen Rothwell : build fix ] Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 18 ------------------ 1 file changed, 18 deletions(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index ae0be3c62375..dc36c3aea018 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -134,7 +134,6 @@ extern unsigned long nr_running(void); extern unsigned long nr_uninterruptible(void); extern unsigned long nr_active(void); extern unsigned long nr_iowait(void); -extern unsigned long weighted_cpuload(const int cpu); struct seq_file; struct cfs_rq; @@ -823,23 +822,6 @@ extern int arch_reinit_sched_domains(void); #endif /* CONFIG_SMP */ -/* - * A runqueue laden with a single nice 0 task scores a weighted_cpuload of - * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a - * task of nice 0 or enough lower priority tasks to bring up the - * weighted_cpuload - */ -static inline int above_background_load(void) -{ - unsigned long cpu; - - for_each_online_cpu(cpu) { - if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE) - return 1; - } - return 0; -} - struct io_context; /* See blkdev.h */ #define NGROUPS_SMALL 32 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t))) -- cgit v1.2.3 From c7aceaba042702538b23cf4e0de1b2891ad8e671 Mon Sep 17 00:00:00 2001 From: Richard Kennedy Date: Thu, 15 May 2008 12:09:15 +0100 Subject: sched: reorder task_struct to reduce padding on 64bit builds This patch removes 24 bytes of padding and allows 1 extra object per slab on my fedora based config. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index dc36c3aea018..ea2857b99596 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1021,6 +1021,7 @@ struct task_struct { #endif int prio, static_prio, normal_prio; + unsigned int rt_priority; const struct sched_class *sched_class; struct sched_entity se; struct sched_rt_entity rt; @@ -1104,7 +1105,6 @@ struct task_struct { int __user *set_child_tid; /* CLONE_CHILD_SETTID */ int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ - unsigned int rt_priority; cputime_t utime, stime, utimescaled, stimescaled; cputime_t gtime; cputime_t prev_utime, prev_stime; @@ -1123,12 +1123,12 @@ struct task_struct { gid_t gid,egid,sgid,fsgid; struct group_info *group_info; kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset; - unsigned securebits; struct user_struct *user; + unsigned securebits; #ifdef CONFIG_KEYS + unsigned char jit_keyring; /* default keyring to attach requested keys to */ struct key *request_key_auth; /* assumed request_key authority */ struct key *thread_keyring; /* keyring private to this thread */ - unsigned char jit_keyring; /* default keyring to attach requested keys to */ #endif char comm[TASK_COMM_LEN]; /* executable name excluding path - access with [gs]et_task_comm (which lock @@ -1215,8 +1215,8 @@ struct task_struct { # define MAX_LOCK_DEPTH 48UL u64 curr_chain_key; int lockdep_depth; - struct held_lock held_locks[MAX_LOCK_DEPTH]; unsigned int lockdep_recursion; + struct held_lock held_locks[MAX_LOCK_DEPTH]; #endif /* journalling filesystem info */ @@ -1244,10 +1244,6 @@ struct task_struct { u64 acct_vm_mem1; /* accumulated virtual memory usage */ cputime_t acct_stimexpd;/* stime since last update */ #endif -#ifdef CONFIG_NUMA - struct mempolicy *mempolicy; - short il_next; -#endif #ifdef CONFIG_CPUSETS nodemask_t mems_allowed; int cpuset_mems_generation; @@ -1266,6 +1262,10 @@ struct task_struct { #endif struct list_head pi_state_list; struct futex_pi_state *pi_state_cache; +#endif +#ifdef CONFIG_NUMA + struct mempolicy *mempolicy; + short il_next; #endif atomic_t fs_excl; /* holding fs exclusive resources */ struct rcu_head rcu; -- cgit v1.2.3 From 1f11eb6a8bc92536d9e93ead48fa3ffbd1478571 Mon Sep 17 00:00:00 2001 From: Gregory Haskins Date: Wed, 4 Jun 2008 15:04:05 -0400 Subject: sched: fix cpupri hotplug support The RT folks over at RedHat found an issue w.r.t. hotplug support which was traced to problems with the cpupri infrastructure in the scheduler: https://bugzilla.redhat.com/show_bug.cgi?id=449676 This bug affects 23-rt12+, 24-rtX, 25-rtX, and sched-devel. This patch applies to 25.4-rt4, though it should trivially apply to most cpupri enabled kernels mentioned above. It turned out that the issue was that offline cpus could get inadvertently registered with cpupri so that they were erroneously selected during migration decisions. The end result would be an OOPS as the offline cpu had tasks routed to it. This patch generalizes the old join/leave domain interface into an online/offline interface, and adjusts the root-domain/hotplug code to utilize it. I was able to easily reproduce the issue prior to this patch, and am no longer able to reproduce it after this patch. I can offline cpus indefinately and everything seems to be in working order. Thanks to Arnaldo (acme), Thomas, and Peter for doing the legwork to point me in the right direction. Also thank you to Peter for reviewing the early iterations of this patch. Signed-off-by: Gregory Haskins Cc: Peter Zijlstra Cc: Steven Rostedt Cc: Arnaldo Carvalho de Melo Cc: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 4 ++-- kernel/sched.c | 54 ++++++++++++++++++++++++++++++++++++++------------- kernel/sched_rt.c | 24 +++++++++++++++++------ 3 files changed, 60 insertions(+), 22 deletions(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index ea2857b99596..d25acf600a32 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -903,8 +903,8 @@ struct sched_class { void (*set_cpus_allowed)(struct task_struct *p, const cpumask_t *newmask); - void (*join_domain)(struct rq *rq); - void (*leave_domain)(struct rq *rq); + void (*rq_online)(struct rq *rq); + void (*rq_offline)(struct rq *rq); void (*switched_from) (struct rq *this_rq, struct task_struct *task, int running); diff --git a/kernel/sched.c b/kernel/sched.c index dc0be113f41d..f0ed81b71282 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -529,6 +529,7 @@ struct rq { int push_cpu; /* cpu of this runqueue: */ int cpu; + int online; struct task_struct *migration_thread; struct list_head migration_queue; @@ -1498,6 +1499,8 @@ static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares) #endif #define sched_class_highest (&rt_sched_class) +#define for_each_class(class) \ + for (class = sched_class_highest; class; class = class->next) static inline void inc_load(struct rq *rq, const struct task_struct *p) { @@ -6065,6 +6068,36 @@ static void unregister_sched_domain_sysctl(void) } #endif +static void set_rq_online(struct rq *rq) +{ + if (!rq->online) { + const struct sched_class *class; + + cpu_set(rq->cpu, rq->rd->online); + rq->online = 1; + + for_each_class(class) { + if (class->rq_online) + class->rq_online(rq); + } + } +} + +static void set_rq_offline(struct rq *rq) +{ + if (rq->online) { + const struct sched_class *class; + + for_each_class(class) { + if (class->rq_offline) + class->rq_offline(rq); + } + + cpu_clear(rq->cpu, rq->rd->online); + rq->online = 0; + } +} + /* * migration_call - callback that gets triggered when a CPU is added. * Here we can start up the necessary migration thread for the new CPU. @@ -6102,7 +6135,8 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu) spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { BUG_ON(!cpu_isset(cpu, rq->rd->span)); - cpu_set(cpu, rq->rd->online); + + set_rq_online(rq); } spin_unlock_irqrestore(&rq->lock, flags); break; @@ -6163,7 +6197,7 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu) spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { BUG_ON(!cpu_isset(cpu, rq->rd->span)); - cpu_clear(cpu, rq->rd->online); + set_rq_offline(rq); } spin_unlock_irqrestore(&rq->lock, flags); break; @@ -6385,20 +6419,16 @@ sd_parent_degenerate(struct sched_domain *sd, struct sched_domain *parent) static void rq_attach_root(struct rq *rq, struct root_domain *rd) { unsigned long flags; - const struct sched_class *class; spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { struct root_domain *old_rd = rq->rd; - for (class = sched_class_highest; class; class = class->next) { - if (class->leave_domain) - class->leave_domain(rq); - } + if (cpu_isset(rq->cpu, old_rd->online)) + set_rq_offline(rq); cpu_clear(rq->cpu, old_rd->span); - cpu_clear(rq->cpu, old_rd->online); if (atomic_dec_and_test(&old_rd->refcount)) kfree(old_rd); @@ -6409,12 +6439,7 @@ static void rq_attach_root(struct rq *rq, struct root_domain *rd) cpu_set(rq->cpu, rd->span); if (cpu_isset(rq->cpu, cpu_online_map)) - cpu_set(rq->cpu, rd->online); - - for (class = sched_class_highest; class; class = class->next) { - if (class->join_domain) - class->join_domain(rq); - } + set_rq_online(rq); spin_unlock_irqrestore(&rq->lock, flags); } @@ -7824,6 +7849,7 @@ void __init sched_init(void) rq->next_balance = jiffies; rq->push_cpu = 0; rq->cpu = i; + rq->online = 0; rq->migration_thread = NULL; INIT_LIST_HEAD(&rq->migration_queue); rq_attach_root(rq, &def_root_domain); diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 44b06d75416e..e4821593d4de 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -12,6 +12,9 @@ static inline int rt_overloaded(struct rq *rq) static inline void rt_set_overload(struct rq *rq) { + if (!rq->online) + return; + cpu_set(rq->cpu, rq->rd->rto_mask); /* * Make sure the mask is visible before we set @@ -26,6 +29,9 @@ static inline void rt_set_overload(struct rq *rq) static inline void rt_clear_overload(struct rq *rq) { + if (!rq->online) + return; + /* the order here really doesn't matter */ atomic_dec(&rq->rd->rto_count); cpu_clear(rq->cpu, rq->rd->rto_mask); @@ -394,7 +400,10 @@ void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) if (rt_se_prio(rt_se) < rt_rq->highest_prio) { struct rq *rq = rq_of_rt_rq(rt_rq); rt_rq->highest_prio = rt_se_prio(rt_se); - cpupri_set(&rq->rd->cpupri, rq->cpu, rt_se_prio(rt_se)); + + if (rq->online) + cpupri_set(&rq->rd->cpupri, rq->cpu, + rt_se_prio(rt_se)); } #endif #ifdef CONFIG_SMP @@ -448,7 +457,10 @@ void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) if (rt_rq->highest_prio != highest_prio) { struct rq *rq = rq_of_rt_rq(rt_rq); - cpupri_set(&rq->rd->cpupri, rq->cpu, rt_rq->highest_prio); + + if (rq->online) + cpupri_set(&rq->rd->cpupri, rq->cpu, + rt_rq->highest_prio); } update_rt_migration(rq_of_rt_rq(rt_rq)); @@ -1154,7 +1166,7 @@ static void set_cpus_allowed_rt(struct task_struct *p, } /* Assumes rq->lock is held */ -static void join_domain_rt(struct rq *rq) +static void rq_online_rt(struct rq *rq) { if (rq->rt.overloaded) rt_set_overload(rq); @@ -1163,7 +1175,7 @@ static void join_domain_rt(struct rq *rq) } /* Assumes rq->lock is held */ -static void leave_domain_rt(struct rq *rq) +static void rq_offline_rt(struct rq *rq) { if (rq->rt.overloaded) rt_clear_overload(rq); @@ -1331,8 +1343,8 @@ static const struct sched_class rt_sched_class = { .load_balance = load_balance_rt, .move_one_task = move_one_task_rt, .set_cpus_allowed = set_cpus_allowed_rt, - .join_domain = join_domain_rt, - .leave_domain = leave_domain_rt, + .rq_online = rq_online_rt, + .rq_offline = rq_offline_rt, .pre_schedule = pre_schedule_rt, .post_schedule = post_schedule_rt, .task_wake_up = task_wake_up_rt, -- cgit v1.2.3 From 9985b0bab332289f14837eff3c6e0bcc658b58f7 Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Thu, 5 Jun 2008 12:57:11 -0700 Subject: sched: prevent bound kthreads from changing cpus_allowed Kthreads that have called kthread_bind() are bound to specific cpus, so other tasks should not be able to change their cpus_allowed from under them. Otherwise, it is possible to move kthreads, such as the migration or software watchdog threads, so they are not allowed access to the cpu they work on. Cc: Peter Zijlstra Cc: Paul Menage Cc: Paul Jackson Signed-off-by: David Rientjes Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 + kernel/cpuset.c | 14 +++++++++++++- kernel/kthread.c | 1 + kernel/sched.c | 6 ++++++ 4 files changed, 21 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index d25acf600a32..2db1485f865d 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1486,6 +1486,7 @@ static inline void put_task_struct(struct task_struct *t) #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ +#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ diff --git a/kernel/cpuset.c b/kernel/cpuset.c index 6090d18b58a9..b84354f4de36 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -1190,6 +1190,15 @@ static int cpuset_can_attach(struct cgroup_subsys *ss, if (cpus_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed)) return -ENOSPC; + if (tsk->flags & PF_THREAD_BOUND) { + cpumask_t mask; + + mutex_lock(&callback_mutex); + mask = cs->cpus_allowed; + mutex_unlock(&callback_mutex); + if (!cpus_equal(tsk->cpus_allowed, mask)) + return -EINVAL; + } return security_task_setscheduler(tsk, 0, NULL); } @@ -1203,11 +1212,14 @@ static void cpuset_attach(struct cgroup_subsys *ss, struct mm_struct *mm; struct cpuset *cs = cgroup_cs(cont); struct cpuset *oldcs = cgroup_cs(oldcont); + int err; mutex_lock(&callback_mutex); guarantee_online_cpus(cs, &cpus); - set_cpus_allowed_ptr(tsk, &cpus); + err = set_cpus_allowed_ptr(tsk, &cpus); mutex_unlock(&callback_mutex); + if (err) + return; from = oldcs->mems_allowed; to = cs->mems_allowed; diff --git a/kernel/kthread.c b/kernel/kthread.c index bd1b9ea024e1..97747cdd37c9 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -180,6 +180,7 @@ void kthread_bind(struct task_struct *k, unsigned int cpu) set_task_cpu(k, cpu); k->cpus_allowed = cpumask_of_cpu(cpu); k->rt.nr_cpus_allowed = 1; + k->flags |= PF_THREAD_BOUND; } EXPORT_SYMBOL(kthread_bind); diff --git a/kernel/sched.c b/kernel/sched.c index e9c24a128655..164fe7fe0d89 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5563,6 +5563,12 @@ int set_cpus_allowed_ptr(struct task_struct *p, const cpumask_t *new_mask) goto out; } + if (unlikely((p->flags & PF_THREAD_BOUND) && p != current && + !cpus_equal(p->cpus_allowed, *new_mask))) { + ret = -EINVAL; + goto out; + } + if (p->sched_class->set_cpus_allowed) p->sched_class->set_cpus_allowed(p, new_mask); else { -- cgit v1.2.3 From c09595f63bb1909c5dc4dca288f4fe818561b5f3 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 27 Jun 2008 13:41:14 +0200 Subject: sched: revert revert of: fair-group: SMP-nice for group scheduling Try again.. Initial commit: 18d95a2832c1392a2d63227a7a6d433cb9f2037e Revert: 6363ca57c76b7b83639ca8c83fc285fa26a7880e Signed-off-by: Peter Zijlstra Cc: Srivatsa Vaddagiri Cc: Mike Galbraith Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 + kernel/sched.c | 430 ++++++++++++++++++++++++++++++++++++++++++++++---- kernel/sched_debug.c | 5 + kernel/sched_fair.c | 124 +++++++++------ kernel/sched_rt.c | 4 + 5 files changed, 489 insertions(+), 75 deletions(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index eaf821072dbd..97a58b622ee1 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -765,6 +765,7 @@ struct sched_domain { struct sched_domain *child; /* bottom domain must be null terminated */ struct sched_group *groups; /* the balancing groups of the domain */ cpumask_t span; /* span of all CPUs in this domain */ + int first_cpu; /* cache of the first cpu in this domain */ unsigned long min_interval; /* Minimum balance interval ms */ unsigned long max_interval; /* Maximum balance interval ms */ unsigned int busy_factor; /* less balancing by factor if busy */ diff --git a/kernel/sched.c b/kernel/sched.c index f653af684fb3..874b6da15430 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -403,6 +403,43 @@ struct cfs_rq { */ struct list_head leaf_cfs_rq_list; struct task_group *tg; /* group that "owns" this runqueue */ + +#ifdef CONFIG_SMP + unsigned long task_weight; + unsigned long shares; + /* + * We need space to build a sched_domain wide view of the full task + * group tree, in order to avoid depending on dynamic memory allocation + * during the load balancing we place this in the per cpu task group + * hierarchy. This limits the load balancing to one instance per cpu, + * but more should not be needed anyway. + */ + struct aggregate_struct { + /* + * load = weight(cpus) * f(tg) + * + * Where f(tg) is the recursive weight fraction assigned to + * this group. + */ + unsigned long load; + + /* + * part of the group weight distributed to this span. + */ + unsigned long shares; + + /* + * The sum of all runqueue weights within this span. + */ + unsigned long rq_weight; + + /* + * Weight contributed by tasks; this is the part we can + * influence by moving tasks around. + */ + unsigned long task_weight; + } aggregate; +#endif #endif }; @@ -1484,6 +1521,326 @@ static unsigned long source_load(int cpu, int type); static unsigned long target_load(int cpu, int type); static unsigned long cpu_avg_load_per_task(int cpu); static int task_hot(struct task_struct *p, u64 now, struct sched_domain *sd); + +#ifdef CONFIG_FAIR_GROUP_SCHED + +/* + * Group load balancing. + * + * We calculate a few balance domain wide aggregate numbers; load and weight. + * Given the pictures below, and assuming each item has equal weight: + * + * root 1 - thread + * / | \ A - group + * A 1 B + * /|\ / \ + * C 2 D 3 4 + * | | + * 5 6 + * + * load: + * A and B get 1/3-rd of the total load. C and D get 1/3-rd of A's 1/3-rd, + * which equals 1/9-th of the total load. + * + * shares: + * The weight of this group on the selected cpus. + * + * rq_weight: + * Direct sum of all the cpu's their rq weight, e.g. A would get 3 while + * B would get 2. + * + * task_weight: + * Part of the rq_weight contributed by tasks; all groups except B would + * get 1, B gets 2. + */ + +static inline struct aggregate_struct * +aggregate(struct task_group *tg, struct sched_domain *sd) +{ + return &tg->cfs_rq[sd->first_cpu]->aggregate; +} + +typedef void (*aggregate_func)(struct task_group *, struct sched_domain *); + +/* + * Iterate the full tree, calling @down when first entering a node and @up when + * leaving it for the final time. + */ +static +void aggregate_walk_tree(aggregate_func down, aggregate_func up, + struct sched_domain *sd) +{ + struct task_group *parent, *child; + + rcu_read_lock(); + parent = &root_task_group; +down: + (*down)(parent, sd); + list_for_each_entry_rcu(child, &parent->children, siblings) { + parent = child; + goto down; + +up: + continue; + } + (*up)(parent, sd); + + child = parent; + parent = parent->parent; + if (parent) + goto up; + rcu_read_unlock(); +} + +/* + * Calculate the aggregate runqueue weight. + */ +static +void aggregate_group_weight(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long rq_weight = 0; + unsigned long task_weight = 0; + int i; + + for_each_cpu_mask(i, sd->span) { + rq_weight += tg->cfs_rq[i]->load.weight; + task_weight += tg->cfs_rq[i]->task_weight; + } + + aggregate(tg, sd)->rq_weight = rq_weight; + aggregate(tg, sd)->task_weight = task_weight; +} + +/* + * Compute the weight of this group on the given cpus. + */ +static +void aggregate_group_shares(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long shares = 0; + int i; + + for_each_cpu_mask(i, sd->span) + shares += tg->cfs_rq[i]->shares; + + if ((!shares && aggregate(tg, sd)->rq_weight) || shares > tg->shares) + shares = tg->shares; + + aggregate(tg, sd)->shares = shares; +} + +/* + * Compute the load fraction assigned to this group, relies on the aggregate + * weight and this group's parent's load, i.e. top-down. + */ +static +void aggregate_group_load(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long load; + + if (!tg->parent) { + int i; + + load = 0; + for_each_cpu_mask(i, sd->span) + load += cpu_rq(i)->load.weight; + + } else { + load = aggregate(tg->parent, sd)->load; + + /* + * shares is our weight in the parent's rq so + * shares/parent->rq_weight gives our fraction of the load + */ + load *= aggregate(tg, sd)->shares; + load /= aggregate(tg->parent, sd)->rq_weight + 1; + } + + aggregate(tg, sd)->load = load; +} + +static void __set_se_shares(struct sched_entity *se, unsigned long shares); + +/* + * Calculate and set the cpu's group shares. + */ +static void +__update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, + int tcpu) +{ + int boost = 0; + unsigned long shares; + unsigned long rq_weight; + + if (!tg->se[tcpu]) + return; + + rq_weight = tg->cfs_rq[tcpu]->load.weight; + + /* + * If there are currently no tasks on the cpu pretend there is one of + * average load so that when a new task gets to run here it will not + * get delayed by group starvation. + */ + if (!rq_weight) { + boost = 1; + rq_weight = NICE_0_LOAD; + } + + /* + * \Sum shares * rq_weight + * shares = ----------------------- + * \Sum rq_weight + * + */ + shares = aggregate(tg, sd)->shares * rq_weight; + shares /= aggregate(tg, sd)->rq_weight + 1; + + /* + * record the actual number of shares, not the boosted amount. + */ + tg->cfs_rq[tcpu]->shares = boost ? 0 : shares; + + if (shares < MIN_SHARES) + shares = MIN_SHARES; + else if (shares > MAX_SHARES) + shares = MAX_SHARES; + + __set_se_shares(tg->se[tcpu], shares); +} + +/* + * Re-adjust the weights on the cpu the task came from and on the cpu the + * task went to. + */ +static void +__move_group_shares(struct task_group *tg, struct sched_domain *sd, + int scpu, int dcpu) +{ + unsigned long shares; + + shares = tg->cfs_rq[scpu]->shares + tg->cfs_rq[dcpu]->shares; + + __update_group_shares_cpu(tg, sd, scpu); + __update_group_shares_cpu(tg, sd, dcpu); + + /* + * ensure we never loose shares due to rounding errors in the + * above redistribution. + */ + shares -= tg->cfs_rq[scpu]->shares + tg->cfs_rq[dcpu]->shares; + if (shares) + tg->cfs_rq[dcpu]->shares += shares; +} + +/* + * Because changing a group's shares changes the weight of the super-group + * we need to walk up the tree and change all shares until we hit the root. + */ +static void +move_group_shares(struct task_group *tg, struct sched_domain *sd, + int scpu, int dcpu) +{ + while (tg) { + __move_group_shares(tg, sd, scpu, dcpu); + tg = tg->parent; + } +} + +static +void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long shares = aggregate(tg, sd)->shares; + int i; + + for_each_cpu_mask(i, sd->span) { + struct rq *rq = cpu_rq(i); + unsigned long flags; + + spin_lock_irqsave(&rq->lock, flags); + __update_group_shares_cpu(tg, sd, i); + spin_unlock_irqrestore(&rq->lock, flags); + } + + aggregate_group_shares(tg, sd); + + /* + * ensure we never loose shares due to rounding errors in the + * above redistribution. + */ + shares -= aggregate(tg, sd)->shares; + if (shares) { + tg->cfs_rq[sd->first_cpu]->shares += shares; + aggregate(tg, sd)->shares += shares; + } +} + +/* + * Calculate the accumulative weight and recursive load of each task group + * while walking down the tree. + */ +static +void aggregate_get_down(struct task_group *tg, struct sched_domain *sd) +{ + aggregate_group_weight(tg, sd); + aggregate_group_shares(tg, sd); + aggregate_group_load(tg, sd); +} + +/* + * Rebalance the cpu shares while walking back up the tree. + */ +static +void aggregate_get_up(struct task_group *tg, struct sched_domain *sd) +{ + aggregate_group_set_shares(tg, sd); +} + +static DEFINE_PER_CPU(spinlock_t, aggregate_lock); + +static void __init init_aggregate(void) +{ + int i; + + for_each_possible_cpu(i) + spin_lock_init(&per_cpu(aggregate_lock, i)); +} + +static int get_aggregate(struct sched_domain *sd) +{ + if (!spin_trylock(&per_cpu(aggregate_lock, sd->first_cpu))) + return 0; + + aggregate_walk_tree(aggregate_get_down, aggregate_get_up, sd); + return 1; +} + +static void put_aggregate(struct sched_domain *sd) +{ + spin_unlock(&per_cpu(aggregate_lock, sd->first_cpu)); +} + +static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares) +{ + cfs_rq->shares = shares; +} + +#else + +static inline void init_aggregate(void) +{ +} + +static inline int get_aggregate(struct sched_domain *sd) +{ + return 0; +} + +static inline void put_aggregate(struct sched_domain *sd) +{ +} +#endif + #endif #include "sched_stats.h" @@ -1498,26 +1855,14 @@ static int task_hot(struct task_struct *p, u64 now, struct sched_domain *sd); #define for_each_class(class) \ for (class = sched_class_highest; class; class = class->next) -static inline void inc_load(struct rq *rq, const struct task_struct *p) -{ - update_load_add(&rq->load, p->se.load.weight); -} - -static inline void dec_load(struct rq *rq, const struct task_struct *p) -{ - update_load_sub(&rq->load, p->se.load.weight); -} - -static void inc_nr_running(struct task_struct *p, struct rq *rq) +static void inc_nr_running(struct rq *rq) { rq->nr_running++; - inc_load(rq, p); } -static void dec_nr_running(struct task_struct *p, struct rq *rq) +static void dec_nr_running(struct rq *rq) { rq->nr_running--; - dec_load(rq, p); } static void set_load_weight(struct task_struct *p) @@ -1609,7 +1954,7 @@ static void activate_task(struct rq *rq, struct task_struct *p, int wakeup) rq->nr_uninterruptible--; enqueue_task(rq, p, wakeup); - inc_nr_running(p, rq); + inc_nr_running(rq); } /* @@ -1621,7 +1966,7 @@ static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep) rq->nr_uninterruptible++; dequeue_task(rq, p, sleep); - dec_nr_running(p, rq); + dec_nr_running(rq); } /** @@ -2274,7 +2619,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) * management (if any): */ p->sched_class->task_new(rq, p); - inc_nr_running(p, rq); + inc_nr_running(rq); } check_preempt_curr(rq, p); #ifdef CONFIG_SMP @@ -3265,9 +3610,12 @@ static int load_balance(int this_cpu, struct rq *this_rq, unsigned long imbalance; struct rq *busiest; unsigned long flags; + int unlock_aggregate; cpus_setall(*cpus); + unlock_aggregate = get_aggregate(sd); + /* * When power savings policy is enabled for the parent domain, idle * sibling can pick up load irrespective of busy siblings. In this case, @@ -3383,8 +3731,9 @@ redo: if (!ld_moved && !sd_idle && sd->flags & SD_SHARE_CPUPOWER && !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE)) - return -1; - return ld_moved; + ld_moved = -1; + + goto out; out_balanced: schedstat_inc(sd, lb_balanced[idle]); @@ -3399,8 +3748,13 @@ out_one_pinned: if (!sd_idle && sd->flags & SD_SHARE_CPUPOWER && !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE)) - return -1; - return 0; + ld_moved = -1; + else + ld_moved = 0; +out: + if (unlock_aggregate) + put_aggregate(sd); + return ld_moved; } /* @@ -4588,10 +4942,8 @@ void set_user_nice(struct task_struct *p, long nice) goto out_unlock; } on_rq = p->se.on_rq; - if (on_rq) { + if (on_rq) dequeue_task(rq, p, 0); - dec_load(rq, p); - } p->static_prio = NICE_TO_PRIO(nice); set_load_weight(p); @@ -4601,7 +4953,6 @@ void set_user_nice(struct task_struct *p, long nice) if (on_rq) { enqueue_task(rq, p, 0); - inc_load(rq, p); /* * If the task increased its priority or is running and * lowered its priority, then reschedule its CPU: @@ -7016,6 +7367,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, ALLNODES); set_domain_attribute(sd, attr); sd->span = *cpu_map; + sd->first_cpu = first_cpu(sd->span); cpu_to_allnodes_group(i, cpu_map, &sd->groups, tmpmask); p = sd; sd_allnodes = 1; @@ -7026,6 +7378,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, NODE); set_domain_attribute(sd, attr); sched_domain_node_span(cpu_to_node(i), &sd->span); + sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7037,6 +7390,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, CPU); set_domain_attribute(sd, attr); sd->span = *nodemask; + sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7048,6 +7402,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, MC); set_domain_attribute(sd, attr); sd->span = cpu_coregroup_map(i); + sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; @@ -7060,6 +7415,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, SIBLING); set_domain_attribute(sd, attr); sd->span = per_cpu(cpu_sibling_map, i); + sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; @@ -7757,6 +8113,7 @@ void __init sched_init(void) } #ifdef CONFIG_SMP + init_aggregate(); init_defrootdomain(); #endif @@ -8322,14 +8679,11 @@ void sched_move_task(struct task_struct *tsk) #endif /* CONFIG_GROUP_SCHED */ #ifdef CONFIG_FAIR_GROUP_SCHED -static void set_se_shares(struct sched_entity *se, unsigned long shares) +static void __set_se_shares(struct sched_entity *se, unsigned long shares) { struct cfs_rq *cfs_rq = se->cfs_rq; - struct rq *rq = cfs_rq->rq; int on_rq; - spin_lock_irq(&rq->lock); - on_rq = se->on_rq; if (on_rq) dequeue_entity(cfs_rq, se, 0); @@ -8339,8 +8693,17 @@ static void set_se_shares(struct sched_entity *se, unsigned long shares) if (on_rq) enqueue_entity(cfs_rq, se, 0); +} - spin_unlock_irq(&rq->lock); +static void set_se_shares(struct sched_entity *se, unsigned long shares) +{ + struct cfs_rq *cfs_rq = se->cfs_rq; + struct rq *rq = cfs_rq->rq; + unsigned long flags; + + spin_lock_irqsave(&rq->lock, flags); + __set_se_shares(se, shares); + spin_unlock_irqrestore(&rq->lock, flags); } static DEFINE_MUTEX(shares_mutex); @@ -8379,8 +8742,13 @@ int sched_group_set_shares(struct task_group *tg, unsigned long shares) * w/o tripping rebalance_share or load_balance_fair. */ tg->shares = shares; - for_each_possible_cpu(i) + for_each_possible_cpu(i) { + /* + * force a rebalance + */ + cfs_rq_set_shares(tg->cfs_rq[i], 0); set_se_shares(tg->se[i], shares); + } /* * Enable load balance activity on this group, by inserting it back on diff --git a/kernel/sched_debug.c b/kernel/sched_debug.c index 8e077b9c91cb..04394ccac88d 100644 --- a/kernel/sched_debug.c +++ b/kernel/sched_debug.c @@ -167,6 +167,11 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) #endif SEQ_printf(m, " .%-30s: %ld\n", "nr_spread_over", cfs_rq->nr_spread_over); +#ifdef CONFIG_FAIR_GROUP_SCHED +#ifdef CONFIG_SMP + SEQ_printf(m, " .%-30s: %lu\n", "shares", cfs_rq->shares); +#endif +#endif } void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq) diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 2e197b8e43f1..183388c4dead 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -567,10 +567,27 @@ update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se) * Scheduling class queueing methods: */ +#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED +static void +add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight) +{ + cfs_rq->task_weight += weight; +} +#else +static inline void +add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight) +{ +} +#endif + static void account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se) { update_load_add(&cfs_rq->load, se->load.weight); + if (!parent_entity(se)) + inc_cpu_load(rq_of(cfs_rq), se->load.weight); + if (entity_is_task(se)) + add_cfs_task_weight(cfs_rq, se->load.weight); cfs_rq->nr_running++; se->on_rq = 1; list_add(&se->group_node, &cfs_rq->tasks); @@ -580,6 +597,10 @@ static void account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) { update_load_sub(&cfs_rq->load, se->load.weight); + if (!parent_entity(se)) + dec_cpu_load(rq_of(cfs_rq), se->load.weight); + if (entity_is_task(se)) + add_cfs_task_weight(cfs_rq, -se->load.weight); cfs_rq->nr_running--; se->on_rq = 0; list_del_init(&se->group_node); @@ -1372,75 +1393,90 @@ static struct task_struct *load_balance_next_fair(void *arg) return __load_balance_iterator(cfs_rq, cfs_rq->balance_iterator); } -#ifdef CONFIG_FAIR_GROUP_SCHED -static int cfs_rq_best_prio(struct cfs_rq *cfs_rq) +static unsigned long +__load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, + unsigned long max_load_move, struct sched_domain *sd, + enum cpu_idle_type idle, int *all_pinned, int *this_best_prio, + struct cfs_rq *cfs_rq) { - struct sched_entity *curr; - struct task_struct *p; - - if (!cfs_rq->nr_running || !first_fair(cfs_rq)) - return MAX_PRIO; - - curr = cfs_rq->curr; - if (!curr) - curr = __pick_next_entity(cfs_rq); + struct rq_iterator cfs_rq_iterator; - p = task_of(curr); + cfs_rq_iterator.start = load_balance_start_fair; + cfs_rq_iterator.next = load_balance_next_fair; + cfs_rq_iterator.arg = cfs_rq; - return p->prio; + return balance_tasks(this_rq, this_cpu, busiest, + max_load_move, sd, idle, all_pinned, + this_best_prio, &cfs_rq_iterator); } -#endif +#ifdef CONFIG_FAIR_GROUP_SCHED static unsigned long load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, struct sched_domain *sd, enum cpu_idle_type idle, int *all_pinned, int *this_best_prio) { - struct cfs_rq *busy_cfs_rq; long rem_load_move = max_load_move; - struct rq_iterator cfs_rq_iterator; - - cfs_rq_iterator.start = load_balance_start_fair; - cfs_rq_iterator.next = load_balance_next_fair; + int busiest_cpu = cpu_of(busiest); + struct task_group *tg; - for_each_leaf_cfs_rq(busiest, busy_cfs_rq) { -#ifdef CONFIG_FAIR_GROUP_SCHED - struct cfs_rq *this_cfs_rq; + rcu_read_lock(); + list_for_each_entry(tg, &task_groups, list) { long imbalance; - unsigned long maxload; + unsigned long this_weight, busiest_weight; + long rem_load, max_load, moved_load; + + /* + * empty group + */ + if (!aggregate(tg, sd)->task_weight) + continue; + + rem_load = rem_load_move * aggregate(tg, sd)->rq_weight; + rem_load /= aggregate(tg, sd)->load + 1; + + this_weight = tg->cfs_rq[this_cpu]->task_weight; + busiest_weight = tg->cfs_rq[busiest_cpu]->task_weight; - this_cfs_rq = cpu_cfs_rq(busy_cfs_rq, this_cpu); + imbalance = (busiest_weight - this_weight) / 2; - imbalance = busy_cfs_rq->load.weight - this_cfs_rq->load.weight; - /* Don't pull if this_cfs_rq has more load than busy_cfs_rq */ - if (imbalance <= 0) + if (imbalance < 0) + imbalance = busiest_weight; + + max_load = max(rem_load, imbalance); + moved_load = __load_balance_fair(this_rq, this_cpu, busiest, + max_load, sd, idle, all_pinned, this_best_prio, + tg->cfs_rq[busiest_cpu]); + + if (!moved_load) continue; - /* Don't pull more than imbalance/2 */ - imbalance /= 2; - maxload = min(rem_load_move, imbalance); + move_group_shares(tg, sd, busiest_cpu, this_cpu); - *this_best_prio = cfs_rq_best_prio(this_cfs_rq); -#else -# define maxload rem_load_move -#endif - /* - * pass busy_cfs_rq argument into - * load_balance_[start|next]_fair iterators - */ - cfs_rq_iterator.arg = busy_cfs_rq; - rem_load_move -= balance_tasks(this_rq, this_cpu, busiest, - maxload, sd, idle, all_pinned, - this_best_prio, - &cfs_rq_iterator); + moved_load *= aggregate(tg, sd)->load; + moved_load /= aggregate(tg, sd)->rq_weight + 1; - if (rem_load_move <= 0) + rem_load_move -= moved_load; + if (rem_load_move < 0) break; } + rcu_read_unlock(); return max_load_move - rem_load_move; } +#else +static unsigned long +load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, + unsigned long max_load_move, + struct sched_domain *sd, enum cpu_idle_type idle, + int *all_pinned, int *this_best_prio) +{ + return __load_balance_fair(this_rq, this_cpu, busiest, + max_load_move, sd, idle, all_pinned, + this_best_prio, &busiest->cfs); +} +#endif static int move_one_task_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 6b4a6b5a4167..765932d0399d 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -670,6 +670,8 @@ static void enqueue_task_rt(struct rq *rq, struct task_struct *p, int wakeup) rt_se->timeout = 0; enqueue_rt_entity(rt_se); + + inc_cpu_load(rq, p->se.load.weight); } static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int sleep) @@ -678,6 +680,8 @@ static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int sleep) update_curr_rt(rq); dequeue_rt_entity(rt_se); + + dec_cpu_load(rq, p->se.load.weight); } /* -- cgit v1.2.3 From b6a86c746f5b708012809958462234d19e9c8177 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 27 Jun 2008 13:41:18 +0200 Subject: sched: fix sched_domain aggregation Keeping the aggregate on the first cpu of the sched domain has two problems: - it could collide between different sched domains on different cpus - it could slow things down because of the remote accesses Signed-off-by: Peter Zijlstra Cc: Srivatsa Vaddagiri Cc: Mike Galbraith Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 - kernel/sched.c | 113 ++++++++++++++++++++++++-------------------------- kernel/sched_fair.c | 12 +++--- 3 files changed, 60 insertions(+), 66 deletions(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index 97a58b622ee1..eaf821072dbd 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -765,7 +765,6 @@ struct sched_domain { struct sched_domain *child; /* bottom domain must be null terminated */ struct sched_group *groups; /* the balancing groups of the domain */ cpumask_t span; /* span of all CPUs in this domain */ - int first_cpu; /* cache of the first cpu in this domain */ unsigned long min_interval; /* Minimum balance interval ms */ unsigned long max_interval; /* Maximum balance interval ms */ unsigned int busy_factor; /* less balancing by factor if busy */ diff --git a/kernel/sched.c b/kernel/sched.c index 7d282c52bd42..160d3c209b8f 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1480,12 +1480,12 @@ static int task_hot(struct task_struct *p, u64 now, struct sched_domain *sd); */ static inline struct aggregate_struct * -aggregate(struct task_group *tg, struct sched_domain *sd) +aggregate(struct task_group *tg, int cpu) { - return &tg->cfs_rq[sd->first_cpu]->aggregate; + return &tg->cfs_rq[cpu]->aggregate; } -typedef void (*aggregate_func)(struct task_group *, struct sched_domain *); +typedef void (*aggregate_func)(struct task_group *, int, struct sched_domain *); /* * Iterate the full tree, calling @down when first entering a node and @up when @@ -1493,14 +1493,14 @@ typedef void (*aggregate_func)(struct task_group *, struct sched_domain *); */ static void aggregate_walk_tree(aggregate_func down, aggregate_func up, - struct sched_domain *sd) + int cpu, struct sched_domain *sd) { struct task_group *parent, *child; rcu_read_lock(); parent = &root_task_group; down: - (*down)(parent, sd); + (*down)(parent, cpu, sd); list_for_each_entry_rcu(child, &parent->children, siblings) { parent = child; goto down; @@ -1508,7 +1508,7 @@ down: up: continue; } - (*up)(parent, sd); + (*up)(parent, cpu, sd); child = parent; parent = parent->parent; @@ -1520,8 +1520,8 @@ up: /* * Calculate the aggregate runqueue weight. */ -static -void aggregate_group_weight(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_weight(struct task_group *tg, int cpu, struct sched_domain *sd) { unsigned long rq_weight = 0; unsigned long task_weight = 0; @@ -1532,15 +1532,15 @@ void aggregate_group_weight(struct task_group *tg, struct sched_domain *sd) task_weight += tg->cfs_rq[i]->task_weight; } - aggregate(tg, sd)->rq_weight = rq_weight; - aggregate(tg, sd)->task_weight = task_weight; + aggregate(tg, cpu)->rq_weight = rq_weight; + aggregate(tg, cpu)->task_weight = task_weight; } /* * Compute the weight of this group on the given cpus. */ -static -void aggregate_group_shares(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_shares(struct task_group *tg, int cpu, struct sched_domain *sd) { unsigned long shares = 0; int i; @@ -1548,18 +1548,18 @@ void aggregate_group_shares(struct task_group *tg, struct sched_domain *sd) for_each_cpu_mask(i, sd->span) shares += tg->cfs_rq[i]->shares; - if ((!shares && aggregate(tg, sd)->rq_weight) || shares > tg->shares) + if ((!shares && aggregate(tg, cpu)->rq_weight) || shares > tg->shares) shares = tg->shares; - aggregate(tg, sd)->shares = shares; + aggregate(tg, cpu)->shares = shares; } /* * Compute the load fraction assigned to this group, relies on the aggregate * weight and this group's parent's load, i.e. top-down. */ -static -void aggregate_group_load(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_load(struct task_group *tg, int cpu, struct sched_domain *sd) { unsigned long load; @@ -1571,17 +1571,17 @@ void aggregate_group_load(struct task_group *tg, struct sched_domain *sd) load += cpu_rq(i)->load.weight; } else { - load = aggregate(tg->parent, sd)->load; + load = aggregate(tg->parent, cpu)->load; /* * shares is our weight in the parent's rq so * shares/parent->rq_weight gives our fraction of the load */ - load *= aggregate(tg, sd)->shares; - load /= aggregate(tg->parent, sd)->rq_weight + 1; + load *= aggregate(tg, cpu)->shares; + load /= aggregate(tg->parent, cpu)->rq_weight + 1; } - aggregate(tg, sd)->load = load; + aggregate(tg, cpu)->load = load; } static void __set_se_shares(struct sched_entity *se, unsigned long shares); @@ -1590,8 +1590,8 @@ static void __set_se_shares(struct sched_entity *se, unsigned long shares); * Calculate and set the cpu's group shares. */ static void -__update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, - int tcpu) +__update_group_shares_cpu(struct task_group *tg, int cpu, + struct sched_domain *sd, int tcpu) { int boost = 0; unsigned long shares; @@ -1618,8 +1618,8 @@ __update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, * \Sum rq_weight * */ - shares = aggregate(tg, sd)->shares * rq_weight; - shares /= aggregate(tg, sd)->rq_weight + 1; + shares = aggregate(tg, cpu)->shares * rq_weight; + shares /= aggregate(tg, cpu)->rq_weight + 1; /* * record the actual number of shares, not the boosted amount. @@ -1639,15 +1639,15 @@ __update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, * task went to. */ static void -__move_group_shares(struct task_group *tg, struct sched_domain *sd, +__move_group_shares(struct task_group *tg, int cpu, struct sched_domain *sd, int scpu, int dcpu) { unsigned long shares; shares = tg->cfs_rq[scpu]->shares + tg->cfs_rq[dcpu]->shares; - __update_group_shares_cpu(tg, sd, scpu); - __update_group_shares_cpu(tg, sd, dcpu); + __update_group_shares_cpu(tg, cpu, sd, scpu); + __update_group_shares_cpu(tg, cpu, sd, dcpu); /* * ensure we never loose shares due to rounding errors in the @@ -1663,19 +1663,19 @@ __move_group_shares(struct task_group *tg, struct sched_domain *sd, * we need to walk up the tree and change all shares until we hit the root. */ static void -move_group_shares(struct task_group *tg, struct sched_domain *sd, +move_group_shares(struct task_group *tg, int cpu, struct sched_domain *sd, int scpu, int dcpu) { while (tg) { - __move_group_shares(tg, sd, scpu, dcpu); + __move_group_shares(tg, cpu, sd, scpu, dcpu); tg = tg->parent; } } -static -void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_set_shares(struct task_group *tg, int cpu, struct sched_domain *sd) { - unsigned long shares = aggregate(tg, sd)->shares; + unsigned long shares = aggregate(tg, cpu)->shares; int i; for_each_cpu_mask(i, sd->span) { @@ -1683,20 +1683,20 @@ void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) unsigned long flags; spin_lock_irqsave(&rq->lock, flags); - __update_group_shares_cpu(tg, sd, i); + __update_group_shares_cpu(tg, cpu, sd, i); spin_unlock_irqrestore(&rq->lock, flags); } - aggregate_group_shares(tg, sd); + aggregate_group_shares(tg, cpu, sd); /* * ensure we never loose shares due to rounding errors in the * above redistribution. */ - shares -= aggregate(tg, sd)->shares; + shares -= aggregate(tg, cpu)->shares; if (shares) { - tg->cfs_rq[sd->first_cpu]->shares += shares; - aggregate(tg, sd)->shares += shares; + tg->cfs_rq[cpu]->shares += shares; + aggregate(tg, cpu)->shares += shares; } } @@ -1704,21 +1704,21 @@ void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) * Calculate the accumulative weight and recursive load of each task group * while walking down the tree. */ -static -void aggregate_get_down(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_get_down(struct task_group *tg, int cpu, struct sched_domain *sd) { - aggregate_group_weight(tg, sd); - aggregate_group_shares(tg, sd); - aggregate_group_load(tg, sd); + aggregate_group_weight(tg, cpu, sd); + aggregate_group_shares(tg, cpu, sd); + aggregate_group_load(tg, cpu, sd); } /* * Rebalance the cpu shares while walking back up the tree. */ -static -void aggregate_get_up(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_get_up(struct task_group *tg, int cpu, struct sched_domain *sd) { - aggregate_group_set_shares(tg, sd); + aggregate_group_set_shares(tg, cpu, sd); } static DEFINE_PER_CPU(spinlock_t, aggregate_lock); @@ -1731,18 +1731,18 @@ static void __init init_aggregate(void) spin_lock_init(&per_cpu(aggregate_lock, i)); } -static int get_aggregate(struct sched_domain *sd) +static int get_aggregate(int cpu, struct sched_domain *sd) { - if (!spin_trylock(&per_cpu(aggregate_lock, sd->first_cpu))) + if (!spin_trylock(&per_cpu(aggregate_lock, cpu))) return 0; - aggregate_walk_tree(aggregate_get_down, aggregate_get_up, sd); + aggregate_walk_tree(aggregate_get_down, aggregate_get_up, cpu, sd); return 1; } -static void put_aggregate(struct sched_domain *sd) +static void put_aggregate(int cpu, struct sched_domain *sd) { - spin_unlock(&per_cpu(aggregate_lock, sd->first_cpu)); + spin_unlock(&per_cpu(aggregate_lock, cpu)); } static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares) @@ -1756,12 +1756,12 @@ static inline void init_aggregate(void) { } -static inline int get_aggregate(struct sched_domain *sd) +static inline int get_aggregate(int cpu, struct sched_domain *sd) { return 0; } -static inline void put_aggregate(struct sched_domain *sd) +static inline void put_aggregate(int cpu, struct sched_domain *sd) { } #endif @@ -3539,7 +3539,7 @@ static int load_balance(int this_cpu, struct rq *this_rq, cpus_setall(*cpus); - unlock_aggregate = get_aggregate(sd); + unlock_aggregate = get_aggregate(this_cpu, sd); /* * When power savings policy is enabled for the parent domain, idle @@ -3678,7 +3678,7 @@ out_one_pinned: ld_moved = 0; out: if (unlock_aggregate) - put_aggregate(sd); + put_aggregate(this_cpu, sd); return ld_moved; } @@ -7292,7 +7292,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, ALLNODES); set_domain_attribute(sd, attr); sd->span = *cpu_map; - sd->first_cpu = first_cpu(sd->span); cpu_to_allnodes_group(i, cpu_map, &sd->groups, tmpmask); p = sd; sd_allnodes = 1; @@ -7303,7 +7302,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, NODE); set_domain_attribute(sd, attr); sched_domain_node_span(cpu_to_node(i), &sd->span); - sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7315,7 +7313,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, CPU); set_domain_attribute(sd, attr); sd->span = *nodemask; - sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7327,7 +7324,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, MC); set_domain_attribute(sd, attr); sd->span = cpu_coregroup_map(i); - sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; @@ -7340,7 +7336,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, SIBLING); set_domain_attribute(sd, attr); sd->span = per_cpu(cpu_sibling_map, i); - sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 509092af0330..40cf24ab4de8 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1429,11 +1429,11 @@ load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, /* * empty group */ - if (!aggregate(tg, sd)->task_weight) + if (!aggregate(tg, this_cpu)->task_weight) continue; - rem_load = rem_load_move * aggregate(tg, sd)->rq_weight; - rem_load /= aggregate(tg, sd)->load + 1; + rem_load = rem_load_move * aggregate(tg, this_cpu)->rq_weight; + rem_load /= aggregate(tg, this_cpu)->load + 1; this_weight = tg->cfs_rq[this_cpu]->task_weight; busiest_weight = tg->cfs_rq[busiest_cpu]->task_weight; @@ -1451,10 +1451,10 @@ load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, if (!moved_load) continue; - move_group_shares(tg, sd, busiest_cpu, this_cpu); + move_group_shares(tg, this_cpu, sd, busiest_cpu, this_cpu); - moved_load *= aggregate(tg, sd)->load; - moved_load /= aggregate(tg, sd)->rq_weight + 1; + moved_load *= aggregate(tg, this_cpu)->load; + moved_load /= aggregate(tg, this_cpu)->rq_weight + 1; rem_load_move -= moved_load; if (rem_load_move < 0) -- cgit v1.2.3 From 2398f2c6d34b43025f274fc42eaca34d23ec2320 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 27 Jun 2008 13:41:35 +0200 Subject: sched: update shares on wakeup We found that the affine wakeup code needs rather accurate load figures to be effective. The trouble is that updating the load figures is fairly expensive with group scheduling. Therefore ratelimit the updating. Signed-off-by: Peter Zijlstra Cc: Srivatsa Vaddagiri Cc: Mike Galbraith Signed-off-by: Ingo Molnar --- include/linux/sched.h | 3 +++ kernel/sched.c | 30 +++++++++++++++++++++++++++++- kernel/sched_features.h | 3 ++- kernel/sysctl.c | 8 ++++++++ 4 files changed, 42 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index eaf821072dbd..835b6c6fcc56 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -783,6 +783,8 @@ struct sched_domain { unsigned int balance_interval; /* initialise to 1. units in ms. */ unsigned int nr_balance_failed; /* initialise to 0 */ + u64 last_update; + #ifdef CONFIG_SCHEDSTATS /* load_balance() stats */ unsigned int lb_count[CPU_MAX_IDLE_TYPES]; @@ -1605,6 +1607,7 @@ extern unsigned int sysctl_sched_child_runs_first; extern unsigned int sysctl_sched_features; extern unsigned int sysctl_sched_migration_cost; extern unsigned int sysctl_sched_nr_migrate; +extern unsigned int sysctl_sched_shares_ratelimit; int sched_nr_latency_handler(struct ctl_table *table, int write, struct file *file, void __user *buffer, size_t *length, diff --git a/kernel/sched.c b/kernel/sched.c index 1cff969f6646..62db0891025a 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -777,6 +777,12 @@ late_initcall(sched_init_debug); */ const_debug unsigned int sysctl_sched_nr_migrate = 32; +/* + * ratelimit for updating the group shares. + * default: 0.5ms + */ +const_debug unsigned int sysctl_sched_shares_ratelimit = 500000; + /* * period over which we measure -rt task cpu usage in us. * default: 1s @@ -1590,7 +1596,13 @@ tg_nop(struct task_group *tg, int cpu, struct sched_domain *sd) static void update_shares(struct sched_domain *sd) { - walk_tg_tree(tg_nop, tg_shares_up, 0, sd); + u64 now = cpu_clock(raw_smp_processor_id()); + s64 elapsed = now - sd->last_update; + + if (elapsed >= (s64)(u64)sysctl_sched_shares_ratelimit) { + sd->last_update = now; + walk_tg_tree(tg_nop, tg_shares_up, 0, sd); + } } static void update_shares_locked(struct rq *rq, struct sched_domain *sd) @@ -2199,6 +2211,22 @@ static int try_to_wake_up(struct task_struct *p, unsigned int state, int sync) if (!sched_feat(SYNC_WAKEUPS)) sync = 0; +#ifdef CONFIG_SMP + if (sched_feat(LB_WAKEUP_UPDATE)) { + struct sched_domain *sd; + + this_cpu = raw_smp_processor_id(); + cpu = task_cpu(p); + + for_each_domain(this_cpu, sd) { + if (cpu_isset(cpu, sd->span)) { + update_shares(sd); + break; + } + } + } +#endif + smp_wmb(); rq = task_rq_lock(p, &flags); old_state = p->state; diff --git a/kernel/sched_features.h b/kernel/sched_features.h index d56e3053e746..7d616d2a2a3f 100644 --- a/kernel/sched_features.h +++ b/kernel/sched_features.h @@ -8,4 +8,5 @@ SCHED_FEAT(SYNC_WAKEUPS, 1) SCHED_FEAT(HRTICK, 1) SCHED_FEAT(DOUBLE_TICK, 0) SCHED_FEAT(ASYM_GRAN, 1) -SCHED_FEAT(LB_BIAS, 0) \ No newline at end of file +SCHED_FEAT(LB_BIAS, 0) +SCHED_FEAT(LB_WAKEUP_UPDATE, 1) diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 29116652dca8..fe8cdc80ff02 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -264,6 +264,14 @@ static struct ctl_table kern_table[] = { .extra1 = &min_wakeup_granularity_ns, .extra2 = &max_wakeup_granularity_ns, }, + { + .ctl_name = CTL_UNNUMBERED, + .procname = "sched_shares_ratelimit", + .data = &sysctl_sched_shares_ratelimit, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = &proc_dointvec, + }, { .ctl_name = CTL_UNNUMBERED, .procname = "sched_child_runs_first", -- cgit v1.2.3 From af52a90a14cdaa54ecbfb6e6982abb13466a4b56 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 7 Jul 2008 14:16:52 -0400 Subject: sched_clock: stop maximum check on NO HZ Working with ftrace I would get large jumps of 11 millisecs or more with the clock tracer. This killed the latencing timings of ftrace and also caused the irqoff self tests to fail. What was happening is with NO_HZ the idle would stop the jiffy counter and before the jiffy counter was updated the sched_clock would have a bad delta jiffies to compare with the gtod with the maximum. The jiffies would stop and the last sched_tick would record the last gtod. On wakeup, the sched clock update would compare the gtod + delta jiffies (which would be zero) and compare it to the TSC. The TSC would have correctly (with a stable TSC) moved forward several jiffies. But because the jiffies has not been updated yet the clock would be prevented from moving forward because it would appear that the TSC jumped too far ahead. The clock would then virtually stop, until the jiffies are updated. Then the next sched clock update would see that the clock was very much behind since the delta jiffies is now correct. This would then jump the clock forward by several jiffies. This caused ftrace to report several milliseconds of interrupts off latency at every resume from NO_HZ idle. This patch adds hooks into the nohz code to disable the checking of the maximum clock update when nohz is in effect. It resumes the max check when nohz has updated the jiffies again. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- include/linux/sched.h | 17 ++++++++++++++++- kernel/sched_clock.c | 39 ++++++++++++++++++++++++++++++++++++++- kernel/time/tick-sched.c | 2 ++ 3 files changed, 56 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/linux/sched.h b/include/linux/sched.h index c5d3f847ca8d..33a8f42041fa 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1573,13 +1573,28 @@ static inline void sched_clock_idle_sleep_event(void) static inline void sched_clock_idle_wakeup_event(u64 delta_ns) { } -#else + +#ifdef CONFIG_NO_HZ +static inline void sched_clock_tick_stop(int cpu) +{ +} + +static inline void sched_clock_tick_start(int cpu) +{ +} +#endif + +#else /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ extern void sched_clock_init(void); extern u64 sched_clock_cpu(int cpu); extern void sched_clock_tick(void); extern void sched_clock_idle_sleep_event(void); extern void sched_clock_idle_wakeup_event(u64 delta_ns); +#ifdef CONFIG_NO_HZ +extern void sched_clock_tick_stop(int cpu); +extern void sched_clock_tick_start(int cpu); #endif +#endif /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ /* * For kernel-internal use: high-speed (but slightly incorrect) per-cpu diff --git a/kernel/sched_clock.c b/kernel/sched_clock.c index 42b81fa38cbd..97159e225a77 100644 --- a/kernel/sched_clock.c +++ b/kernel/sched_clock.c @@ -45,6 +45,9 @@ struct sched_clock_data { u64 tick_raw; u64 tick_gtod; u64 clock; +#ifdef CONFIG_NO_HZ + int check_max; +#endif }; static DEFINE_PER_CPU_SHARED_ALIGNED(struct sched_clock_data, sched_clock_data); @@ -76,11 +79,45 @@ void sched_clock_init(void) scd->tick_raw = 0; scd->tick_gtod = ktime_now; scd->clock = ktime_now; +#ifdef CONFIG_NO_HZ + scd->check_max = 1; +#endif } sched_clock_running = 1; } +#ifdef CONFIG_NO_HZ +/* + * The dynamic ticks makes the delta jiffies inaccurate. This + * prevents us from checking the maximum time update. + * Disable the maximum check during stopped ticks. + */ +void sched_clock_tick_stop(int cpu) +{ + struct sched_clock_data *scd = cpu_sdc(cpu); + + scd->check_max = 0; +} + +void sched_clock_tick_start(int cpu) +{ + struct sched_clock_data *scd = cpu_sdc(cpu); + + scd->check_max = 1; +} + +static int check_max(struct sched_clock_data *scd) +{ + return scd->check_max; +} +#else +static int check_max(struct sched_clock_data *scd) +{ + return 1; +} +#endif /* CONFIG_NO_HZ */ + /* * update the percpu scd from the raw @now value * @@ -112,7 +149,7 @@ static void __update_sched_clock(struct sched_clock_data *scd, u64 now) */ max_clock = scd->tick_gtod + (2 + delta_jiffies) * TICK_NSEC; - if (unlikely(clock + delta > max_clock)) { + if (unlikely(clock + delta > max_clock) && check_max(scd)) { if (clock < max_clock) clock = max_clock; else diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index b854a895591e..d63008b09a4c 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -276,6 +276,7 @@ void tick_nohz_stop_sched_tick(void) ts->tick_stopped = 1; ts->idle_jiffies = last_jiffies; rcu_enter_nohz(); + sched_clock_tick_stop(cpu); } /* @@ -375,6 +376,7 @@ void tick_nohz_restart_sched_tick(void) select_nohz_load_balancer(0); now = ktime_get(); tick_do_update_jiffies64(now); + sched_clock_tick_start(cpu); cpu_clear(cpu, nohz_cpu_mask); /* -- cgit v1.2.3