From 39be350127ec60a078edffe5b4915dafba4ba514 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 26 Jan 2012 12:44:34 +0100 Subject: sched, block: Unify cache detection The block layer has some code trying to determine if two CPUs share a cache, the scheduler has a similar function. Expose the function used by the scheduler and make the block layer use it, thereby removing the block layers usage of CONFIG_SCHED* and topology bits. Signed-off-by: Peter Zijlstra Acked-by: Jens Axboe Link: http://lkml.kernel.org/r/1327579450.2446.95.camel@twins diff --git a/block/blk-softirq.c b/block/blk-softirq.c index 1366a89..467c8de 100644 --- a/block/blk-softirq.c +++ b/block/blk-softirq.c @@ -8,6 +8,7 @@ #include #include #include +#include #include "blk.h" @@ -103,9 +104,10 @@ static struct notifier_block __cpuinitdata blk_cpu_notifier = { void __blk_complete_request(struct request *req) { - int ccpu, cpu, group_cpu = NR_CPUS; + int ccpu, cpu; struct request_queue *q = req->q; unsigned long flags; + bool shared = false; BUG_ON(!q->softirq_done_fn); @@ -117,22 +119,20 @@ void __blk_complete_request(struct request *req) */ if (req->cpu != -1) { ccpu = req->cpu; - if (!test_bit(QUEUE_FLAG_SAME_FORCE, &q->queue_flags)) { - ccpu = blk_cpu_to_group(ccpu); - group_cpu = blk_cpu_to_group(cpu); - } + if (!test_bit(QUEUE_FLAG_SAME_FORCE, &q->queue_flags)) + shared = cpus_share_cache(cpu, ccpu); } else ccpu = cpu; /* - * If current CPU and requested CPU are in the same group, running - * softirq in current CPU. One might concern this is just like + * If current CPU and requested CPU share a cache, run the softirq on + * the current CPU. One might concern this is just like * QUEUE_FLAG_SAME_FORCE, but actually not. blk_complete_request() is * running in interrupt handler, and currently I/O controller doesn't * support multiple interrupts, so current CPU is unique actually. This * avoids IPI sending from current CPU to the first CPU of a group. */ - if (ccpu == cpu || ccpu == group_cpu) { + if (ccpu == cpu || shared) { struct list_head *list; do_local: list = &__get_cpu_var(blk_cpu_done); diff --git a/block/blk.h b/block/blk.h index 7efd772..df5b59a 100644 --- a/block/blk.h +++ b/block/blk.h @@ -164,22 +164,6 @@ static inline int queue_congestion_off_threshold(struct request_queue *q) return q->nr_congestion_off; } -static inline int blk_cpu_to_group(int cpu) -{ - int group = NR_CPUS; -#ifdef CONFIG_SCHED_MC - const struct cpumask *mask = cpu_coregroup_mask(cpu); - group = cpumask_first(mask); -#elif defined(CONFIG_SCHED_SMT) - group = cpumask_first(topology_thread_cpumask(cpu)); -#else - return cpu; -#endif - if (likely(group < NR_CPUS)) - return group; - return cpu; -} - /* * Contribute to IO statistics IFF: * diff --git a/include/linux/sched.h b/include/linux/sched.h index 513f524..0e19595 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1052,6 +1052,8 @@ static inline int test_sd_parent(struct sched_domain *sd, int flag) unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu); unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu); +bool cpus_share_cache(int this_cpu, int that_cpu); + #else /* CONFIG_SMP */ struct sched_domain_attr; @@ -1061,6 +1063,12 @@ partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], struct sched_domain_attr *dattr_new) { } + +static inline bool cpus_share_cache(int this_cpu, int that_cpu) +{ + return true; +} + #endif /* !CONFIG_SMP */ diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 5255c9d..d7c4322 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -1507,7 +1507,7 @@ static int ttwu_activate_remote(struct task_struct *p, int wake_flags) } #endif /* __ARCH_WANT_INTERRUPTS_ON_CTXSW */ -static inline int ttwu_share_cache(int this_cpu, int that_cpu) +bool cpus_share_cache(int this_cpu, int that_cpu) { return per_cpu(sd_llc_id, this_cpu) == per_cpu(sd_llc_id, that_cpu); } @@ -1518,7 +1518,7 @@ static void ttwu_queue(struct task_struct *p, int cpu) struct rq *rq = cpu_rq(cpu); #if defined(CONFIG_SMP) - if (sched_feat(TTWU_QUEUE) && !ttwu_share_cache(smp_processor_id(), cpu)) { + if (sched_feat(TTWU_QUEUE) && !cpus_share_cache(smp_processor_id(), cpu)) { sched_clock_cpu(cpu); /* sync clocks x-cpu */ ttwu_queue_remote(p, cpu); return; @@ -5754,7 +5754,7 @@ static void destroy_sched_domains(struct sched_domain *sd, int cpu) * * Also keep a unique ID per domain (we use the first cpu number in * the cpumask of the domain), this allows us to quickly tell if - * two cpus are in the same cache domain, see ttwu_share_cache(). + * two cpus are in the same cache domain, see cpus_share_cache(). */ DEFINE_PER_CPU(struct sched_domain *, sd_llc); DEFINE_PER_CPU(int, sd_llc_id); -- cgit v0.10.2 From 4ec4412e1e91f44a3dcb97b6c9172a13fc78bac9 Mon Sep 17 00:00:00 2001 From: Vincent Guittot Date: Mon, 12 Dec 2011 20:21:08 +0100 Subject: sched: Ensure cpu_power periodic update With a lot of small tasks, the softirq sched is nearly never called when no_hz is enabled. In this case load_balance() is mainly called with the newly_idle mode which doesn't update the cpu_power. Add a next_update field which ensure a maximum update period when there is short activity. Having stale cpu_power information can skew the load-balancing decisions, this is cured by the guaranteed update. Signed-off-by: Vincent Guittot Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1323717668-2143-1-git-send-email-vincent.guittot@linaro.org diff --git a/include/linux/sched.h b/include/linux/sched.h index 0e19595..92313a3f 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -905,6 +905,7 @@ struct sched_group_power { * single CPU. */ unsigned int power, power_orig; + unsigned long next_update; /* * Number of busy cpus in this group. */ diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 7c6414f..8e77a6b 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -215,6 +215,8 @@ calc_delta_mine(unsigned long delta_exec, unsigned long weight, const struct sched_class fair_sched_class; +static unsigned long __read_mostly max_load_balance_interval = HZ/10; + /************************************************************** * CFS operations on generic schedulable entities: */ @@ -3776,6 +3778,11 @@ void update_group_power(struct sched_domain *sd, int cpu) struct sched_domain *child = sd->child; struct sched_group *group, *sdg = sd->groups; unsigned long power; + unsigned long interval; + + interval = msecs_to_jiffies(sd->balance_interval); + interval = clamp(interval, 1UL, max_load_balance_interval); + sdg->sgp->next_update = jiffies + interval; if (!child) { update_cpu_power(sd, cpu); @@ -3883,12 +3890,15 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, * domains. In the newly idle case, we will allow all the cpu's * to do the newly idle load balance. */ - if (idle != CPU_NEWLY_IDLE && local_group) { - if (balance_cpu != this_cpu) { - *balance = 0; - return; - } - update_group_power(sd, this_cpu); + if (local_group) { + if (idle != CPU_NEWLY_IDLE) { + if (balance_cpu != this_cpu) { + *balance = 0; + return; + } + update_group_power(sd, this_cpu); + } else if (time_after_eq(jiffies, group->sgp->next_update)) + update_group_power(sd, this_cpu); } /* Adjust by relative CPU power of the group */ @@ -4945,8 +4955,6 @@ static int __cpuinit sched_ilb_notifier(struct notifier_block *nfb, static DEFINE_SPINLOCK(balancing); -static unsigned long __read_mostly max_load_balance_interval = HZ/10; - /* * Scale the max load_balance interval with the number of CPUs in the system. * This trades load-balance latency on larger machines for less cross talk. -- cgit v0.10.2 From 30fd049afcfed50e022704036e8629d6bdfe84e6 Mon Sep 17 00:00:00 2001 From: Rakib Mullick Date: Tue, 24 Jan 2012 22:33:56 +0600 Subject: sched: Remove sched_switch Currently we don't utilize the sched_switch field anymore. But, simply removing sched_switch field from the middle of the sched_stat output will break tools. So, to stay compatible we hardcode it to zero and remove the field from the scheduler data structures. Update the schedstat documentation accordingly. Signed-off-by: Rakib Mullick Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1327422836.27181.5.camel@localhost.localdomain Signed-off-by: Ingo Molnar diff --git a/Documentation/scheduler/sched-stats.txt b/Documentation/scheduler/sched-stats.txt index 1cd5d51..8259b34 100644 --- a/Documentation/scheduler/sched-stats.txt +++ b/Documentation/scheduler/sched-stats.txt @@ -38,7 +38,8 @@ First field is a sched_yield() statistic: 1) # of times sched_yield() was called Next three are schedule() statistics: - 2) # of times we switched to the expired queue and reused it + 2) This field is a legacy array expiration count field used in the O(1) + scheduler. We kept it for ABI compatibility, but it is always set to zero. 3) # of times schedule() was called 4) # of times schedule() left the processor idle diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index 2a075e1..09acaa1 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -288,7 +288,6 @@ static void print_cpu(struct seq_file *m, int cpu) P(yld_count); - P(sched_switch); P(sched_count); P(sched_goidle); #ifdef CONFIG_SMP diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 98c0c26..8a2c768 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -462,7 +462,6 @@ struct rq { unsigned int yld_count; /* schedule() stats */ - unsigned int sched_switch; unsigned int sched_count; unsigned int sched_goidle; diff --git a/kernel/sched/stats.c b/kernel/sched/stats.c index 2a581ba..903ffa9e 100644 --- a/kernel/sched/stats.c +++ b/kernel/sched/stats.c @@ -32,9 +32,9 @@ static int show_schedstat(struct seq_file *seq, void *v) /* runqueue-specific stats */ seq_printf(seq, - "cpu%d %u %u %u %u %u %u %llu %llu %lu", + "cpu%d %u 0 %u %u %u %u %llu %llu %lu", cpu, rq->yld_count, - rq->sched_switch, rq->sched_count, rq->sched_goidle, + rq->sched_count, rq->sched_goidle, rq->ttwu_count, rq->ttwu_local, rq->rq_cpu_time, rq->rq_sched_info.run_delay, rq->rq_sched_info.pcount); -- cgit v0.10.2 From ed387b781ea6e14b78f449aa2ee4f270b60b01ac Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Tue, 31 Jan 2012 11:40:32 +0900 Subject: sched: Move SMP-only variable into the SMP section This also fixes the following compilation warning on !SMP: CC kernel/sched/fair.o kernel/sched/fair.c:218:36: warning: 'max_load_balance_interval' defined but not used [-Wunused-variable] Signed-off-by: Hiroshi Shimamoto Reviewed-by: Vincent Guittot Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/4F2754A0.9090306@ct.jp.nec.com Signed-off-by: Ingo Molnar diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 8e77a6b..4ab60a2 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -215,8 +215,6 @@ calc_delta_mine(unsigned long delta_exec, unsigned long weight, const struct sched_class fair_sched_class; -static unsigned long __read_mostly max_load_balance_interval = HZ/10; - /************************************************************** * CFS operations on generic schedulable entities: */ @@ -3086,6 +3084,8 @@ static bool yield_to_task_fair(struct rq *rq, struct task_struct *p, bool preemp * Fair scheduling class load-balancing methods: */ +static unsigned long __read_mostly max_load_balance_interval = HZ/10; + /* * pull_task - move a task from a remote runqueue to the local runqueue. * Both runqueues must be locked. -- cgit v0.10.2 From 9388dc3047a88bedfd867e9ba3e1980c815ac524 Mon Sep 17 00:00:00 2001 From: Anton Vorontsov Date: Thu, 9 Feb 2012 20:45:19 +0400 Subject: sched: Turn lock_task_sighand() into a static inline It appears that sparse tool understands static inline functions for context balance checking, so let's turn the macros into an inline func. This makes the code a little bit more robust. Suggested-by: Oleg Nesterov Signed-off-by: Anton Vorontsov Cc: KOSAKI Motohiro Cc: Greg KH Cc: Arve Cc: San Mehat Cc: Colin Cross Cc: Eric W. Biederman Cc: Paul E. McKenney Cc: kernel-team@android.com Cc: linaro-kernel@lists.linaro.org Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20120209164519.GA10266@oksana.dev.rtsoft.ru Signed-off-by: Ingo Molnar diff --git a/include/linux/sched.h b/include/linux/sched.h index 92313a3f..5dba2ad 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2393,12 +2393,15 @@ static inline void task_unlock(struct task_struct *p) extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk, unsigned long *flags); -#define lock_task_sighand(tsk, flags) \ -({ struct sighand_struct *__ss; \ - __cond_lock(&(tsk)->sighand->siglock, \ - (__ss = __lock_task_sighand(tsk, flags))); \ - __ss; \ -}) \ +static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk, + unsigned long *flags) +{ + struct sighand_struct *ret; + + ret = __lock_task_sighand(tsk, flags); + (void)__cond_lock(&tsk->sighand->siglock, ret); + return ret; +} static inline void unlock_task_sighand(struct task_struct *tsk, unsigned long *flags) -- cgit v0.10.2 From 62f6536a630affe3176deb48554d27ee58b65077 Mon Sep 17 00:00:00 2001 From: "Nikunj A. Dadhania" Date: Fri, 17 Feb 2012 08:34:30 +0530 Subject: sched: Remove rcu_read_lock/unlock() from select_idle_sibling() select_idle_sibling() is called from select_task_rq_fair(), which already has the RCU read lock held. Signed-off-by: Nikunj A. Dadhania Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/20120217030409.11748.12491.stgit@abhimanyu Signed-off-by: Ingo Molnar diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 4ab60a2..6ce9992 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -2670,8 +2670,6 @@ static int select_idle_sibling(struct task_struct *p, int target) /* * Otherwise, iterate the domains and find an elegible idle cpu. */ - rcu_read_lock(); - sd = rcu_dereference(per_cpu(sd_llc, target)); for_each_lower_domain(sd) { sg = sd->groups; @@ -2693,8 +2691,6 @@ next: } while (sg != sd->groups); } done: - rcu_read_unlock(); - return target; } -- cgit v0.10.2 From de5bdff7a72acc281219be2b8edeeca1fd81c542 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Thu, 16 Feb 2012 14:52:21 +0900 Subject: sched: Make initial SCHED_RR timeslace DEF_TIMESLICE Current the initial SCHED_RR timeslice of init_task is HZ, which means 1s, and is not same as the default SCHED_RR timeslice DEF_TIMESLICE. Change that initial timeslice to the DEF_TIMESLICE. Signed-off-by: Hiroshi Shimamoto [ s/DEF_TIMESLICE/RR_TIMESLICE/g ] Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/4F3C9995.3010800@ct.jp.nec.com Signed-off-by: Ingo Molnar diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 9c66b1a..f994d51 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -149,7 +149,7 @@ extern struct cred init_cred; }, \ .rt = { \ .run_list = LIST_HEAD_INIT(tsk.rt.run_list), \ - .time_slice = HZ, \ + .time_slice = RR_TIMESLICE, \ .nr_cpus_allowed = NR_CPUS, \ }, \ .tasks = LIST_HEAD_INIT(tsk.tasks), \ diff --git a/include/linux/sched.h b/include/linux/sched.h index 5dba2ad..eb5de46 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1234,6 +1234,12 @@ struct sched_rt_entity { #endif }; +/* + * default timeslice is 100 msecs (used only for SCHED_RR tasks). + * Timeslices get refilled after they expire. + */ +#define RR_TIMESLICE (100 * HZ / 1000) + struct rcu_node; enum perf_event_task_context { diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index f42ae7f..f70206c 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -1972,7 +1972,7 @@ static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued) if (--p->rt.time_slice) return; - p->rt.time_slice = DEF_TIMESLICE; + p->rt.time_slice = RR_TIMESLICE; /* * Requeue to the end of queue if we are not the only element @@ -2000,7 +2000,7 @@ static unsigned int get_rr_interval_rt(struct rq *rq, struct task_struct *task) * Time slice is 0 for SCHED_FIFO tasks */ if (task->policy == SCHED_RR) - return DEF_TIMESLICE; + return RR_TIMESLICE; else return 0; } diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 8a2c768..c0660a1 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -36,11 +36,7 @@ extern __read_mostly int scheduler_running; /* * These are the 'tuning knobs' of the scheduler: - * - * default timeslice is 100 msecs (used only for SCHED_RR tasks). - * Timeslices get refilled after they expire. */ -#define DEF_TIMESLICE (100 * HZ / 1000) /* * single value that denotes runtime == period, ie unlimited time. -- cgit v0.10.2 From 42c62a589f1ccbf38a02cb732231f9c2fccc5ab0 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Oct 2011 22:03:48 +0200 Subject: sched/rt: Keep period timer ticking when rt throttling is active When a runqueue is throttled we cannot disable the period timer because that timer is the only way to undo the throttling. We got stale throttling entries when a rq was throttled and then the global sysctl was disabled, which stopped the timer. Signed-off-by: Peter Zijlstra [ Added changelog ] Signed-off-by: Thomas Gleixner Link: http://lkml.kernel.org/n/tip-nuj34q52p6ro7szapuz84i0v@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index f70206c..6d1eb0b 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -778,12 +778,9 @@ static inline int balance_runtime(struct rt_rq *rt_rq) static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) { - int i, idle = 1; + int i, idle = 1, throttled = 0; const struct cpumask *span; - if (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF) - return 1; - span = sched_rt_period_mask(); for_each_cpu(i, span) { int enqueue = 0; @@ -818,12 +815,17 @@ static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) if (!rt_rq_throttled(rt_rq)) enqueue = 1; } + if (rt_rq->rt_throttled) + throttled = 1; if (enqueue) sched_rt_rq_enqueue(rt_rq); raw_spin_unlock(&rq->lock); } + if (!throttled && (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF)) + return 1; + return idle; } @@ -884,7 +886,8 @@ static void update_curr_rt(struct rq *rq) if (unlikely((s64)delta_exec < 0)) delta_exec = 0; - schedstat_set(curr->se.statistics.exec_max, max(curr->se.statistics.exec_max, delta_exec)); + schedstat_set(curr->se.statistics.exec_max, + max(curr->se.statistics.exec_max, delta_exec)); curr->se.sum_exec_runtime += delta_exec; account_group_exec_runtime(curr, delta_exec); -- cgit v0.10.2 From 7abc63b1bd412f7655b62ef3e35c3c11c5134636 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Oct 2011 22:03:48 +0200 Subject: sched/rt: Do not throttle when PI boosting When a runqueue has rt_runtime_us = 0 then the only way it can accumulate rt_time is via PI boosting. That causes the runqueue to be throttled and replenishing does not change anything due to rt_runtime_us = 0. So avoid that situation by clearing rt_time and skip the throttling alltogether. Signed-off-by: Peter Zijlstra [ Changelog ] Signed-off-by: Thomas Gleixner Link: http://lkml.kernel.org/n/tip-7x70cypsotjb4jvcor3edctk@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index 6d1eb0b..7f7e7cd 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -857,8 +857,24 @@ static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) return 0; if (rt_rq->rt_time > runtime) { - rt_rq->rt_throttled = 1; - printk_once(KERN_WARNING "sched: RT throttling activated\n"); + struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); + + /* + * Don't actually throttle groups that have no runtime assigned + * but accrue some time due to boosting. + */ + if (likely(rt_b->rt_runtime)) { + rt_rq->rt_throttled = 1; + printk_once(KERN_WARNING "sched: RT throttling activated\n"); + } else { + /* + * In case we did anyway, make it go away, + * replenishment is a joke, since it will replenish us + * with exactly 0 ns. + */ + rt_rq->rt_time = 0; + } + if (rt_rq_throttled(rt_rq)) { sched_rt_rq_dequeue(rt_rq); return 1; -- cgit v0.10.2 From c5491ea779793f977d282754db478157cc409d82 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 21 Mar 2011 12:09:35 +0100 Subject: sched/rt: Add schedule_preempt_disabled() Add helper to get rid of the ever repeating: preempt_enable_no_resched(); schedule(); preempt_disable(); patterns. Signed-off-by: Thomas Gleixner Acked-by: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-wxx7btox7coby6ifv5vzhzgp@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/include/linux/sched.h b/include/linux/sched.h index 1a63984..03dd224 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -361,6 +361,7 @@ extern signed long schedule_timeout_interruptible(signed long timeout); extern signed long schedule_timeout_killable(signed long timeout); extern signed long schedule_timeout_uninterruptible(signed long timeout); asmlinkage void schedule(void); +extern void schedule_preempt_disabled(void); extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner); struct nsproxy; diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 91fe6a0..7302239 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3246,6 +3246,18 @@ asmlinkage void __sched schedule(void) } EXPORT_SYMBOL(schedule); +/** + * schedule_preempt_disabled - called with preemption disabled + * + * Returns with preemption disabled. Note: preempt_count must be 1 + */ +void __sched schedule_preempt_disabled(void) +{ + preempt_enable_no_resched(); + schedule(); + preempt_disable(); +} + #ifdef CONFIG_MUTEX_SPIN_ON_OWNER static inline bool owner_running(struct mutex *lock, struct task_struct *owner) -- cgit v0.10.2 From bd2f55361f18347e890d52ff9cfd8895455ec11b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 21 Mar 2011 12:33:18 +0100 Subject: sched/rt: Use schedule_preempt_disabled() Coccinelle based conversion. Signed-off-by: Thomas Gleixner Acked-by: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-24swm5zut3h9c4a6s46x8rws@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c index 971d65c..c2ae3cd 100644 --- a/arch/arm/kernel/process.c +++ b/arch/arm/kernel/process.c @@ -239,9 +239,7 @@ void cpu_idle(void) leds_event(led_idle_end); rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/avr32/kernel/process.c b/arch/avr32/kernel/process.c index ea33957..92c5af9 100644 --- a/arch/avr32/kernel/process.c +++ b/arch/avr32/kernel/process.c @@ -40,9 +40,7 @@ void cpu_idle(void) cpu_idle_sleep(); rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/blackfin/kernel/process.c b/arch/blackfin/kernel/process.c index 8dd0416..a80a643 100644 --- a/arch/blackfin/kernel/process.c +++ b/arch/blackfin/kernel/process.c @@ -94,9 +94,7 @@ void cpu_idle(void) idle(); rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/cris/kernel/process.c b/arch/cris/kernel/process.c index aa585e4..d8f50ff 100644 --- a/arch/cris/kernel/process.c +++ b/arch/cris/kernel/process.c @@ -115,9 +115,7 @@ void cpu_idle (void) idle = default_idle; idle(); } - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/frv/kernel/process.c b/arch/frv/kernel/process.c index 3901df1..29cc497 100644 --- a/arch/frv/kernel/process.c +++ b/arch/frv/kernel/process.c @@ -92,9 +92,7 @@ void cpu_idle(void) idle(); } - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/h8300/kernel/process.c b/arch/h8300/kernel/process.c index 933bd38..1a173b3 100644 --- a/arch/h8300/kernel/process.c +++ b/arch/h8300/kernel/process.c @@ -81,9 +81,7 @@ void cpu_idle(void) while (1) { while (!need_resched()) idle(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c index 6d33c5c..9dc52b6 100644 --- a/arch/ia64/kernel/process.c +++ b/arch/ia64/kernel/process.c @@ -330,9 +330,7 @@ cpu_idle (void) normal_xtp(); #endif } - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); check_pgt_cache(); if (cpu_is_offline(cpu)) play_dead(); diff --git a/arch/m32r/kernel/process.c b/arch/m32r/kernel/process.c index 422bea9..3a4a32b 100644 --- a/arch/m32r/kernel/process.c +++ b/arch/m32r/kernel/process.c @@ -90,9 +90,7 @@ void cpu_idle (void) idle(); } - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/m68k/kernel/process_mm.c b/arch/m68k/kernel/process_mm.c index 099283e..fe4186b 100644 --- a/arch/m68k/kernel/process_mm.c +++ b/arch/m68k/kernel/process_mm.c @@ -78,9 +78,7 @@ void cpu_idle(void) while (1) { while (!need_resched()) idle(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/m68k/kernel/process_no.c b/arch/m68k/kernel/process_no.c index 5e1078c..f7fe6c3 100644 --- a/arch/m68k/kernel/process_no.c +++ b/arch/m68k/kernel/process_no.c @@ -73,9 +73,7 @@ void cpu_idle(void) /* endless idle loop with no priority at all */ while (1) { idle(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/microblaze/kernel/process.c b/arch/microblaze/kernel/process.c index 7dcb5bf..9155f7d 100644 --- a/arch/microblaze/kernel/process.c +++ b/arch/microblaze/kernel/process.c @@ -110,9 +110,7 @@ void cpu_idle(void) rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); check_pgt_cache(); } } diff --git a/arch/mips/kernel/process.c b/arch/mips/kernel/process.c index 7955409..61f1cb4 100644 --- a/arch/mips/kernel/process.c +++ b/arch/mips/kernel/process.c @@ -80,9 +80,7 @@ void __noreturn cpu_idle(void) #endif rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/mn10300/kernel/process.c b/arch/mn10300/kernel/process.c index 28eec31..cac401d 100644 --- a/arch/mn10300/kernel/process.c +++ b/arch/mn10300/kernel/process.c @@ -123,9 +123,7 @@ void cpu_idle(void) idle(); } - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/parisc/kernel/process.c b/arch/parisc/kernel/process.c index 62c60b8..d4b94b3 100644 --- a/arch/parisc/kernel/process.c +++ b/arch/parisc/kernel/process.c @@ -71,9 +71,7 @@ void cpu_idle(void) while (1) { while (!need_resched()) barrier(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); check_pgt_cache(); } } diff --git a/arch/powerpc/kernel/idle.c b/arch/powerpc/kernel/idle.c index 0a48bf5..6503514 100644 --- a/arch/powerpc/kernel/idle.c +++ b/arch/powerpc/kernel/idle.c @@ -101,11 +101,11 @@ void cpu_idle(void) ppc64_runlatch_on(); rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - if (cpu_should_die()) + if (cpu_should_die()) { + preempt_enable_no_resched(); cpu_die(); - schedule(); - preempt_disable(); + } + schedule_preempt_disabled(); } } diff --git a/arch/powerpc/platforms/iseries/setup.c b/arch/powerpc/platforms/iseries/setup.c index 8fc6258..a5fbf4c 100644 --- a/arch/powerpc/platforms/iseries/setup.c +++ b/arch/powerpc/platforms/iseries/setup.c @@ -584,9 +584,7 @@ static void iseries_shared_idle(void) if (hvlpevent_is_pending()) process_iSeries_events(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } @@ -615,9 +613,7 @@ static void iseries_dedicated_idle(void) ppc64_runlatch_on(); rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c index e795933..7618085 100644 --- a/arch/s390/kernel/process.c +++ b/arch/s390/kernel/process.c @@ -97,9 +97,7 @@ void cpu_idle(void) tick_nohz_idle_exit(); if (test_thread_flag(TIF_MCCK_PENDING)) s390_handle_mcck(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/score/kernel/process.c b/arch/score/kernel/process.c index 25d0803..2707023 100644 --- a/arch/score/kernel/process.c +++ b/arch/score/kernel/process.c @@ -53,9 +53,7 @@ void __noreturn cpu_idle(void) while (!need_resched()) barrier(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/sh/kernel/idle.c b/arch/sh/kernel/idle.c index 406508d..7e48928 100644 --- a/arch/sh/kernel/idle.c +++ b/arch/sh/kernel/idle.c @@ -114,9 +114,7 @@ void cpu_idle(void) rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/sparc/kernel/process_32.c b/arch/sparc/kernel/process_32.c index f793742..935fdbc 100644 --- a/arch/sparc/kernel/process_32.c +++ b/arch/sparc/kernel/process_32.c @@ -113,9 +113,7 @@ void cpu_idle(void) while (!need_resched()) cpu_relax(); } - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); check_pgt_cache(); } } @@ -138,9 +136,7 @@ void cpu_idle(void) while (!need_resched()) cpu_relax(); } - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); check_pgt_cache(); } } diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c index 39d8b05..ab9a292 100644 --- a/arch/sparc/kernel/process_64.c +++ b/arch/sparc/kernel/process_64.c @@ -104,15 +104,13 @@ void cpu_idle(void) rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - #ifdef CONFIG_HOTPLUG_CPU - if (cpu_is_offline(cpu)) + if (cpu_is_offline(cpu)) { + preempt_enable_no_resched(); cpu_play_dead(); + } #endif - - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/tile/kernel/process.c b/arch/tile/kernel/process.c index 4c1ac6e..6ae495e 100644 --- a/arch/tile/kernel/process.c +++ b/arch/tile/kernel/process.c @@ -108,9 +108,7 @@ void cpu_idle(void) } rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index c08d1ff..49888fe 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -119,9 +119,7 @@ void cpu_idle(void) } rcu_idle_exit(); tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index cfa5c90..e34257c 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -156,9 +156,7 @@ void cpu_idle(void) } tick_nohz_idle_exit(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c index 47041e7..2c90047 100644 --- a/arch/xtensa/kernel/process.c +++ b/arch/xtensa/kernel/process.c @@ -113,9 +113,7 @@ void cpu_idle(void) while (1) { while (!need_resched()) platform_idle(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/init/main.c b/init/main.c index ff49a6d..4990f7e 100644 --- a/init/main.c +++ b/init/main.c @@ -374,11 +374,8 @@ static noinline void __init_refok rest_init(void) * at least once to get things moving: */ init_idle_bootup_task(current); - preempt_enable_no_resched(); - schedule(); - + schedule_preempt_disabled(); /* Call into cpu_idle with preempt disabled */ - preempt_disable(); cpu_idle(); } diff --git a/kernel/mutex.c b/kernel/mutex.c index 89096dd..a307cc9 100644 --- a/kernel/mutex.c +++ b/kernel/mutex.c @@ -240,9 +240,7 @@ __mutex_lock_common(struct mutex *lock, long state, unsigned int subclass, /* didn't get the lock, go to sleep: */ spin_unlock_mutex(&lock->wait_lock, flags); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); spin_lock_mutex(&lock->wait_lock, flags); } diff --git a/kernel/softirq.c b/kernel/softirq.c index 4eb3a0f..79b5247 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -744,9 +744,7 @@ static int run_ksoftirqd(void * __bind_cpu) while (!kthread_should_stop()) { preempt_disable(); if (!local_softirq_pending()) { - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } __set_current_state(TASK_RUNNING); -- cgit v0.10.2 From ba74c1448f127649046615ec017bded7b2a76f29 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 21 Mar 2011 13:32:17 +0100 Subject: sched/rt: Document scheduler related skip-resched-check sites Create a distinction between scheduler related preempt_enable_no_resched() calls and the nearly one hundred other places in the kernel that do not want to reschedule, for one reason or another. This distinction matters for -rt, where the scheduler and the non-scheduler preempt models (and checks) are different. For upstream it's purely documentational. Signed-off-by: Thomas Gleixner Link: http://lkml.kernel.org/n/tip-gs88fvx2mdv5psnzxnv575ke@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/arch/powerpc/kernel/idle.c b/arch/powerpc/kernel/idle.c index 6503514..c97fc60 100644 --- a/arch/powerpc/kernel/idle.c +++ b/arch/powerpc/kernel/idle.c @@ -102,7 +102,7 @@ void cpu_idle(void) rcu_idle_exit(); tick_nohz_idle_exit(); if (cpu_should_die()) { - preempt_enable_no_resched(); + sched_preempt_enable_no_resched(); cpu_die(); } schedule_preempt_disabled(); diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c index ab9a292..06b5b5f 100644 --- a/arch/sparc/kernel/process_64.c +++ b/arch/sparc/kernel/process_64.c @@ -106,7 +106,7 @@ void cpu_idle(void) #ifdef CONFIG_HOTPLUG_CPU if (cpu_is_offline(cpu)) { - preempt_enable_no_resched(); + sched_preempt_enable_no_resched(); cpu_play_dead(); } #endif diff --git a/include/linux/preempt.h b/include/linux/preempt.h index 58969b2..5a710b9 100644 --- a/include/linux/preempt.h +++ b/include/linux/preempt.h @@ -48,12 +48,14 @@ do { \ barrier(); \ } while (0) -#define preempt_enable_no_resched() \ +#define sched_preempt_enable_no_resched() \ do { \ barrier(); \ dec_preempt_count(); \ } while (0) +#define preempt_enable_no_resched() sched_preempt_enable_no_resched() + #define preempt_enable() \ do { \ preempt_enable_no_resched(); \ @@ -92,6 +94,7 @@ do { \ #else /* !CONFIG_PREEMPT_COUNT */ #define preempt_disable() do { } while (0) +#define sched_preempt_enable_no_resched() do { } while (0) #define preempt_enable_no_resched() do { } while (0) #define preempt_enable() do { } while (0) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 7302239..643cc37 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3220,7 +3220,7 @@ need_resched: post_schedule(rq); - preempt_enable_no_resched(); + sched_preempt_enable_no_resched(); if (need_resched()) goto need_resched; } @@ -3253,7 +3253,7 @@ EXPORT_SYMBOL(schedule); */ void __sched schedule_preempt_disabled(void) { - preempt_enable_no_resched(); + sched_preempt_enable_no_resched(); schedule(); preempt_disable(); } @@ -4486,7 +4486,7 @@ SYSCALL_DEFINE0(sched_yield) __release(rq->lock); spin_release(&rq->lock.dep_map, 1, _THIS_IP_); do_raw_spin_unlock(&rq->lock); - preempt_enable_no_resched(); + sched_preempt_enable_no_resched(); schedule(); diff --git a/kernel/softirq.c b/kernel/softirq.c index 79b5247..f268369 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -353,7 +353,7 @@ void irq_exit(void) tick_nohz_irq_exit(); #endif rcu_irq_exit(); - preempt_enable_no_resched(); + sched_preempt_enable_no_resched(); } /* @@ -759,7 +759,7 @@ static int run_ksoftirqd(void * __bind_cpu) if (local_softirq_pending()) __do_softirq(); local_irq_enable(); - preempt_enable_no_resched(); + sched_preempt_enable_no_resched(); cond_resched(); preempt_disable(); rcu_note_context_switch((long)__bind_cpu); -- cgit v0.10.2 From 63b2001169e75cd71e917ec953fdab572e3f944a Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 1 Dec 2011 00:04:00 +0100 Subject: sched/wait: Add __wake_up_all_locked() API For code which protects the waitqueue itself with another lock it makes no sense to acquire the waitqueue lock for wakeup all. Provide __wake_up_all_locked(). This is an optimization on the vanilla kernel (to be used by the PCI code) and an important semantic distinction on -rt. Signed-off-by: Thomas Gleixner Acked-by: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-ux6m4b8jonb9inx8xafh77ds@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/include/linux/wait.h b/include/linux/wait.h index a9ce45e..7d9a9e9 100644 --- a/include/linux/wait.h +++ b/include/linux/wait.h @@ -157,7 +157,7 @@ void __wake_up(wait_queue_head_t *q, unsigned int mode, int nr, void *key); void __wake_up_locked_key(wait_queue_head_t *q, unsigned int mode, void *key); void __wake_up_sync_key(wait_queue_head_t *q, unsigned int mode, int nr, void *key); -void __wake_up_locked(wait_queue_head_t *q, unsigned int mode); +void __wake_up_locked(wait_queue_head_t *q, unsigned int mode, int nr); void __wake_up_sync(wait_queue_head_t *q, unsigned int mode, int nr); void __wake_up_bit(wait_queue_head_t *, void *, int); int __wait_on_bit(wait_queue_head_t *, struct wait_bit_queue *, int (*)(void *), unsigned); @@ -170,7 +170,8 @@ wait_queue_head_t *bit_waitqueue(void *, int); #define wake_up(x) __wake_up(x, TASK_NORMAL, 1, NULL) #define wake_up_nr(x, nr) __wake_up(x, TASK_NORMAL, nr, NULL) #define wake_up_all(x) __wake_up(x, TASK_NORMAL, 0, NULL) -#define wake_up_locked(x) __wake_up_locked((x), TASK_NORMAL) +#define wake_up_locked(x) __wake_up_locked((x), TASK_NORMAL, 1) +#define wake_up_all_locked(x) __wake_up_locked((x), TASK_NORMAL, 0) #define wake_up_interruptible(x) __wake_up(x, TASK_INTERRUPTIBLE, 1, NULL) #define wake_up_interruptible_nr(x, nr) __wake_up(x, TASK_INTERRUPTIBLE, nr, NULL) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 643cc37..820f745 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3418,9 +3418,9 @@ EXPORT_SYMBOL(__wake_up); /* * Same as __wake_up but called with the spinlock in wait_queue_head_t held. */ -void __wake_up_locked(wait_queue_head_t *q, unsigned int mode) +void __wake_up_locked(wait_queue_head_t *q, unsigned int mode, int nr) { - __wake_up_common(q, mode, 1, 0, NULL); + __wake_up_common(q, mode, nr, 0, NULL); } EXPORT_SYMBOL_GPL(__wake_up_locked); -- cgit v0.10.2 From 1c4dd99bed5f6f70932bf8dacdd54d04a2619778 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 6 Jun 2011 20:07:38 +0200 Subject: sched/rt: Prevent idle task boosting Idle task boosting is a nono in general. There is one exception, when PREEMPT_RT and NOHZ is active: The idle task calls get_next_timer_interrupt() and holds the timer wheel base->lock on the CPU and another CPU wants to access the timer (probably to cancel it). We can safely ignore the boosting request, as the idle CPU runs this code with interrupts disabled and will complete the lock protected section without being interrupted. So there is no real need to boost. Signed-off-by: Thomas Gleixner Acked-by: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-755rvsosz7sdzot12a3gbha6@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 820f745..c2bbdec 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3779,6 +3779,24 @@ void rt_mutex_setprio(struct task_struct *p, int prio) rq = __task_rq_lock(p); + /* + * Idle task boosting is a nono in general. There is one + * exception, when PREEMPT_RT and NOHZ is active: + * + * The idle task calls get_next_timer_interrupt() and holds + * the timer wheel base->lock on the CPU and another CPU wants + * to access the timer (probably to cancel it). We can safely + * ignore the boosting request, as the idle CPU runs this code + * with interrupts disabled and will complete the lock + * protected section without being interrupted. So there is no + * real need to boost. + */ + if (unlikely(p == rq->idle)) { + WARN_ON(p != rq->curr); + WARN_ON(p->pi_blocked_on); + goto out_unlock; + } + trace_sched_pi_setprio(p, prio); oldprio = p->prio; prev_class = p->sched_class; @@ -3802,11 +3820,10 @@ void rt_mutex_setprio(struct task_struct *p, int prio) enqueue_task(rq, p, oldprio < prio ? ENQUEUE_HEAD : 0); check_class_changed(rq, p, prev_class, oldprio); +out_unlock: __task_rq_unlock(rq); } - #endif - void set_user_nice(struct task_struct *p, long nice) { int old_prio, delta, on_rq; -- cgit v0.10.2 From 3c7d51843b03a6839e9ec7cda724e54d2319a63a Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sun, 17 Jul 2011 20:46:52 +0200 Subject: sched/rt: Do not submit new work when PI-blocked When we are PI-blocked then we want to get things done ASAP. Signed-off-by: Thomas Gleixner Acked-by: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-vw8et3445km5b8mpihf4trae@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/include/linux/sched.h b/include/linux/sched.h index 03dd224..c628a91 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2082,12 +2082,20 @@ extern unsigned int sysctl_sched_cfs_bandwidth_slice; extern int rt_mutex_getprio(struct task_struct *p); extern void rt_mutex_setprio(struct task_struct *p, int prio); extern void rt_mutex_adjust_pi(struct task_struct *p); +static inline bool tsk_is_pi_blocked(struct task_struct *tsk) +{ + return tsk->pi_blocked_on != NULL; +} #else static inline int rt_mutex_getprio(struct task_struct *p) { return p->normal_prio; } # define rt_mutex_adjust_pi(p) do { } while (0) +static inline bool tsk_is_pi_blocked(struct task_struct *tsk) +{ + return false; +} #endif extern bool yield_to(struct task_struct *p, bool preempt); diff --git a/kernel/sched/core.c b/kernel/sched/core.c index c2bbdec..25e06a5 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3227,7 +3227,7 @@ need_resched: static inline void sched_submit_work(struct task_struct *tsk) { - if (!tsk->state) + if (!tsk->state || tsk_is_pi_blocked(tsk)) return; /* * If we are going to sleep and we have plugged IO queued, -- cgit v0.10.2 From 8e45cb545d98bc58e75b7de89ec8d3e5c8459ee6 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 22 Feb 2012 12:47:19 +0100 Subject: sched: Move load-balancing arguments into helper struct Passing large sets of similar arguments all around the load-balancer gets tiresom when you want to modify something. Stick them all in a helper structure and pass the structure around. Signed-off-by: Peter Zijlstra Cc: pjt@google.com Link: http://lkml.kernel.org/n/tip-5slqz0vhsdzewrfk9eza1aon@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 79e9e13..55b1f11 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -3135,13 +3135,25 @@ task_hot(struct task_struct *p, u64 now, struct sched_domain *sd) #define LBF_HAD_BREAKS 0x0C /* count HAD_BREAKs overflows into ABORT */ #define LBF_ABORT 0x10 +struct lb_env { + struct sched_domain *sd; + + int this_cpu; + struct rq *this_rq; + + struct rq *busiest_rq; + struct cfs_rq *busiest_cfs_rq; + + enum cpu_idle_type idle; + unsigned long max_load_move; + unsigned int flags; +}; + /* * can_migrate_task - may task p from runqueue rq be migrated to this_cpu? */ static -int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu, - struct sched_domain *sd, enum cpu_idle_type idle, - int *lb_flags) +int can_migrate_task(struct task_struct *p, struct lb_env *env) { int tsk_cache_hot = 0; /* @@ -3150,13 +3162,13 @@ int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu, * 2) cannot be migrated to this CPU due to cpus_allowed, or * 3) are cache-hot on their current CPU. */ - if (!cpumask_test_cpu(this_cpu, tsk_cpus_allowed(p))) { + if (!cpumask_test_cpu(env->this_cpu, tsk_cpus_allowed(p))) { schedstat_inc(p, se.statistics.nr_failed_migrations_affine); return 0; } - *lb_flags &= ~LBF_ALL_PINNED; + env->flags &= ~LBF_ALL_PINNED; - if (task_running(rq, p)) { + if (task_running(env->busiest_rq, p)) { schedstat_inc(p, se.statistics.nr_failed_migrations_running); return 0; } @@ -3167,12 +3179,12 @@ int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu, * 2) too many balance attempts have failed. */ - tsk_cache_hot = task_hot(p, rq->clock_task, sd); + tsk_cache_hot = task_hot(p, env->busiest_rq->clock_task, env->sd); if (!tsk_cache_hot || - sd->nr_balance_failed > sd->cache_nice_tries) { + env->sd->nr_balance_failed > env->sd->cache_nice_tries) { #ifdef CONFIG_SCHEDSTATS if (tsk_cache_hot) { - schedstat_inc(sd, lb_hot_gained[idle]); + schedstat_inc(env->sd, lb_hot_gained[env->idle]); schedstat_inc(p, se.statistics.nr_forced_migrations); } #endif @@ -3193,31 +3205,27 @@ int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu, * * Called with both runqueues locked. */ -static int -move_one_task(struct rq *this_rq, int this_cpu, struct rq *busiest, - struct sched_domain *sd, enum cpu_idle_type idle) +static int move_one_task(struct lb_env *env) { struct task_struct *p, *n; struct cfs_rq *cfs_rq; - int pinned = 0; - for_each_leaf_cfs_rq(busiest, cfs_rq) { + for_each_leaf_cfs_rq(env->busiest_rq, cfs_rq) { list_for_each_entry_safe(p, n, &cfs_rq->tasks, se.group_node) { if (throttled_lb_pair(task_group(p), - busiest->cpu, this_cpu)) + env->busiest_rq->cpu, env->this_cpu)) break; - if (!can_migrate_task(p, busiest, this_cpu, - sd, idle, &pinned)) + if (!can_migrate_task(p, env)) continue; - pull_task(busiest, p, this_rq, this_cpu); + pull_task(env->busiest_rq, p, env->this_rq, env->this_cpu); /* * Right now, this is only the second place pull_task() * is called, so we can safely collect pull_task() * stats here rather than inside pull_task(). */ - schedstat_inc(sd, lb_gained[idle]); + schedstat_inc(env->sd, lb_gained[env->idle]); return 1; } } @@ -3225,31 +3233,26 @@ move_one_task(struct rq *this_rq, int this_cpu, struct rq *busiest, return 0; } -static unsigned long -balance_tasks(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 *lb_flags, - struct cfs_rq *busiest_cfs_rq) +static unsigned long balance_tasks(struct lb_env *env) { int loops = 0, pulled = 0; - long rem_load_move = max_load_move; + long rem_load_move = env->max_load_move; struct task_struct *p, *n; - if (max_load_move == 0) + if (env->max_load_move == 0) goto out; - list_for_each_entry_safe(p, n, &busiest_cfs_rq->tasks, se.group_node) { + list_for_each_entry_safe(p, n, &env->busiest_cfs_rq->tasks, se.group_node) { if (loops++ > sysctl_sched_nr_migrate) { - *lb_flags |= LBF_NEED_BREAK; + env->flags |= LBF_NEED_BREAK; break; } if ((p->se.load.weight >> 1) > rem_load_move || - !can_migrate_task(p, busiest, this_cpu, sd, idle, - lb_flags)) + !can_migrate_task(p, env)) continue; - pull_task(busiest, p, this_rq, this_cpu); + pull_task(env->busiest_rq, p, env->this_rq, env->this_cpu); pulled++; rem_load_move -= p->se.load.weight; @@ -3259,8 +3262,8 @@ balance_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest, * kernels will stop after the first task is pulled to minimize * the critical section. */ - if (idle == CPU_NEWLY_IDLE) { - *lb_flags |= LBF_ABORT; + if (env->idle == CPU_NEWLY_IDLE) { + env->flags |= LBF_ABORT; break; } #endif @@ -3278,9 +3281,9 @@ out: * so we can safely collect pull_task() stats here rather than * inside pull_task(). */ - schedstat_add(sd, lb_gained[idle], pulled); + schedstat_add(env->sd, lb_gained[env->idle], pulled); - return max_load_move - rem_load_move; + return env->max_load_move - rem_load_move; } #ifdef CONFIG_FAIR_GROUP_SCHED @@ -3363,40 +3366,39 @@ static void update_h_load(long cpu) walk_tg_tree(tg_load_down, tg_nop, (void *)cpu); } -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 *lb_flags) +static unsigned long load_balance_fair(struct lb_env *env) { - long rem_load_move = max_load_move; - struct cfs_rq *busiest_cfs_rq; + unsigned long max_load_move = env->max_load_move; + long rem_load_move = env->max_load_move; rcu_read_lock(); - update_h_load(cpu_of(busiest)); + update_h_load(cpu_of(env->busiest_rq)); - for_each_leaf_cfs_rq(busiest, busiest_cfs_rq) { - unsigned long busiest_h_load = busiest_cfs_rq->h_load; - unsigned long busiest_weight = busiest_cfs_rq->load.weight; + for_each_leaf_cfs_rq(env->busiest_rq, env->busiest_cfs_rq) { + unsigned long busiest_h_load = env->busiest_cfs_rq->h_load; + unsigned long busiest_weight = env->busiest_cfs_rq->load.weight; u64 rem_load, moved_load; - if (*lb_flags & (LBF_NEED_BREAK|LBF_ABORT)) + if (env->flags & (LBF_NEED_BREAK|LBF_ABORT)) break; /* * empty group or part of a throttled hierarchy */ - if (!busiest_cfs_rq->task_weight || - throttled_lb_pair(busiest_cfs_rq->tg, cpu_of(busiest), this_cpu)) + if (!env->busiest_cfs_rq->task_weight) + continue; + + if (throttled_lb_pair(env->busiest_cfs_rq->tg, + cpu_of(env->busiest_rq), + env->this_cpu)) continue; rem_load = (u64)rem_load_move * busiest_weight; rem_load = div_u64(rem_load, busiest_h_load + 1); - moved_load = balance_tasks(this_rq, this_cpu, busiest, - rem_load, sd, idle, lb_flags, - busiest_cfs_rq); + env->max_load_move = rem_load; + moved_load = balance_tasks(env); if (!moved_load) continue; @@ -3416,15 +3418,10 @@ static inline void update_shares(int cpu) { } -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 *lb_flags) +static unsigned long load_balance_fair(struct lb_env *env) { - return balance_tasks(this_rq, this_cpu, busiest, - max_load_move, sd, idle, lb_flags, - &busiest->cfs); + env->busiest_cfs_rq = &env->busiest_rq->cfs; + return balance_tasks(env); } #endif @@ -3435,21 +3432,17 @@ load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, * * Called with both runqueues locked. */ -static int move_tasks(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 *lb_flags) +static int move_tasks(struct lb_env *env) { + unsigned long max_load_move = env->max_load_move; unsigned long total_load_moved = 0, load_moved; do { - load_moved = load_balance_fair(this_rq, this_cpu, busiest, - max_load_move - total_load_moved, - sd, idle, lb_flags); - + env->max_load_move = max_load_move - total_load_moved; + load_moved = load_balance_fair(env); total_load_moved += load_moved; - if (*lb_flags & (LBF_NEED_BREAK|LBF_ABORT)) + if (env->flags & (LBF_NEED_BREAK|LBF_ABORT)) break; #ifdef CONFIG_PREEMPT @@ -3458,8 +3451,8 @@ static int move_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest, * kernels will stop after the first task is pulled to minimize * the critical section. */ - if (idle == CPU_NEWLY_IDLE && this_rq->nr_running) { - *lb_flags |= LBF_ABORT; + if (env->idle == CPU_NEWLY_IDLE && env->this_rq->nr_running) { + env->flags |= LBF_ABORT; break; } #endif @@ -4459,13 +4452,20 @@ static int load_balance(int this_cpu, struct rq *this_rq, struct sched_domain *sd, enum cpu_idle_type idle, int *balance) { - int ld_moved, lb_flags = 0, active_balance = 0; + int ld_moved, active_balance = 0; struct sched_group *group; unsigned long imbalance; struct rq *busiest; unsigned long flags; struct cpumask *cpus = __get_cpu_var(load_balance_tmpmask); + struct lb_env env = { + .sd = sd, + .this_cpu = this_cpu, + .this_rq = this_rq, + .idle = idle, + }; + cpumask_copy(cpus, cpu_active_mask); schedstat_inc(sd, lb_count[idle]); @@ -4500,11 +4500,13 @@ redo: * still unbalanced. ld_moved simply stays zero, so it is * correctly treated as an imbalance. */ - lb_flags |= LBF_ALL_PINNED; + env.flags |= LBF_ALL_PINNED; + env.max_load_move = imbalance; + env.busiest_rq = busiest; + local_irq_save(flags); double_rq_lock(this_rq, busiest); - ld_moved = move_tasks(this_rq, this_cpu, busiest, - imbalance, sd, idle, &lb_flags); + ld_moved = move_tasks(&env); double_rq_unlock(this_rq, busiest); local_irq_restore(flags); @@ -4514,18 +4516,18 @@ redo: if (ld_moved && this_cpu != smp_processor_id()) resched_cpu(this_cpu); - if (lb_flags & LBF_ABORT) + if (env.flags & LBF_ABORT) goto out_balanced; - if (lb_flags & LBF_NEED_BREAK) { - lb_flags += LBF_HAD_BREAK - LBF_NEED_BREAK; - if (lb_flags & LBF_ABORT) + if (env.flags & LBF_NEED_BREAK) { + env.flags += LBF_HAD_BREAK - LBF_NEED_BREAK; + if (env.flags & LBF_ABORT) goto out_balanced; goto redo; } /* All tasks on this runqueue were pinned by CPU affinity */ - if (unlikely(lb_flags & LBF_ALL_PINNED)) { + if (unlikely(env.flags & LBF_ALL_PINNED)) { cpumask_clear_cpu(cpu_of(busiest), cpus); if (!cpumask_empty(cpus)) goto redo; @@ -4555,7 +4557,7 @@ redo: tsk_cpus_allowed(busiest->curr))) { raw_spin_unlock_irqrestore(&busiest->lock, flags); - lb_flags |= LBF_ALL_PINNED; + env.flags |= LBF_ALL_PINNED; goto out_one_pinned; } @@ -4608,7 +4610,7 @@ out_balanced: out_one_pinned: /* tune up the balancing interval */ - if (((lb_flags & LBF_ALL_PINNED) && + if (((env.flags & LBF_ALL_PINNED) && sd->balance_interval < MAX_PINNED_INTERVAL) || (sd->balance_interval < sd->max_interval)) sd->balance_interval *= 2; @@ -4718,10 +4720,17 @@ static int active_load_balance_cpu_stop(void *data) } if (likely(sd)) { + struct lb_env env = { + .sd = sd, + .this_cpu = target_cpu, + .this_rq = target_rq, + .busiest_rq = busiest_rq, + .idle = CPU_IDLE, + }; + schedstat_inc(sd, alb_count); - if (move_one_task(target_rq, target_cpu, busiest_rq, - sd, CPU_IDLE)) + if (move_one_task(&env)) schedstat_inc(sd, alb_pushed); else schedstat_inc(sd, alb_failed); -- cgit v0.10.2 From ddcdf6e7d9919d139031fa2a6addd9544a9a833e Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 22 Feb 2012 19:27:40 +0100 Subject: sched: Rename load-balancing fields s/env->this_/env->dst_/g s/env->busiest_/env->src_/g s/pull_task/move_task/g Makes everything clearer. Signed-off-by: Peter Zijlstra Cc: pjt@google.com Link: http://lkml.kernel.org/n/tip-0yvgms8t8x962drpvl0fu0kk@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 55b1f11..233d051 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -2918,7 +2918,7 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p, int wake_ return; /* - * This is possible from callers such as pull_task(), in which we + * This is possible from callers such as move_task(), in which we * unconditionally check_prempt_curr() after an enqueue (which may have * lead to a throttle). This both saves work and prevents false * next-buddy nomination below. @@ -3084,17 +3084,37 @@ static bool yield_to_task_fair(struct rq *rq, struct task_struct *p, bool preemp static unsigned long __read_mostly max_load_balance_interval = HZ/10; +#define LBF_ALL_PINNED 0x01 +#define LBF_NEED_BREAK 0x02 /* clears into HAD_BREAK */ +#define LBF_HAD_BREAK 0x04 +#define LBF_HAD_BREAKS 0x0C /* count HAD_BREAKs overflows into ABORT */ +#define LBF_ABORT 0x10 + +struct lb_env { + struct sched_domain *sd; + + int src_cpu; + struct rq *src_rq; + struct cfs_rq *src_cfs_rq; + + int dst_cpu; + struct rq *dst_rq; + + enum cpu_idle_type idle; + unsigned long max_load_move; + unsigned int flags; +}; + /* - * pull_task - move a task from a remote runqueue to the local runqueue. + * move_task - move a task from one runqueue to another runqueue. * Both runqueues must be locked. */ -static void pull_task(struct rq *src_rq, struct task_struct *p, - struct rq *this_rq, int this_cpu) +static void move_task(struct task_struct *p, struct lb_env *env) { - deactivate_task(src_rq, p, 0); - set_task_cpu(p, this_cpu); - activate_task(this_rq, p, 0); - check_preempt_curr(this_rq, p, 0); + deactivate_task(env->src_rq, p, 0); + set_task_cpu(p, env->dst_cpu); + activate_task(env->dst_rq, p, 0); + check_preempt_curr(env->dst_rq, p, 0); } /* @@ -3129,26 +3149,6 @@ task_hot(struct task_struct *p, u64 now, struct sched_domain *sd) return delta < (s64)sysctl_sched_migration_cost; } -#define LBF_ALL_PINNED 0x01 -#define LBF_NEED_BREAK 0x02 /* clears into HAD_BREAK */ -#define LBF_HAD_BREAK 0x04 -#define LBF_HAD_BREAKS 0x0C /* count HAD_BREAKs overflows into ABORT */ -#define LBF_ABORT 0x10 - -struct lb_env { - struct sched_domain *sd; - - int this_cpu; - struct rq *this_rq; - - struct rq *busiest_rq; - struct cfs_rq *busiest_cfs_rq; - - enum cpu_idle_type idle; - unsigned long max_load_move; - unsigned int flags; -}; - /* * can_migrate_task - may task p from runqueue rq be migrated to this_cpu? */ @@ -3162,13 +3162,13 @@ int can_migrate_task(struct task_struct *p, struct lb_env *env) * 2) cannot be migrated to this CPU due to cpus_allowed, or * 3) are cache-hot on their current CPU. */ - if (!cpumask_test_cpu(env->this_cpu, tsk_cpus_allowed(p))) { + if (!cpumask_test_cpu(env->dst_cpu, tsk_cpus_allowed(p))) { schedstat_inc(p, se.statistics.nr_failed_migrations_affine); return 0; } env->flags &= ~LBF_ALL_PINNED; - if (task_running(env->busiest_rq, p)) { + if (task_running(env->src_rq, p)) { schedstat_inc(p, se.statistics.nr_failed_migrations_running); return 0; } @@ -3179,7 +3179,7 @@ int can_migrate_task(struct task_struct *p, struct lb_env *env) * 2) too many balance attempts have failed. */ - tsk_cache_hot = task_hot(p, env->busiest_rq->clock_task, env->sd); + tsk_cache_hot = task_hot(p, env->src_rq->clock_task, env->sd); if (!tsk_cache_hot || env->sd->nr_balance_failed > env->sd->cache_nice_tries) { #ifdef CONFIG_SCHEDSTATS @@ -3210,20 +3210,20 @@ static int move_one_task(struct lb_env *env) struct task_struct *p, *n; struct cfs_rq *cfs_rq; - for_each_leaf_cfs_rq(env->busiest_rq, cfs_rq) { + for_each_leaf_cfs_rq(env->src_rq, cfs_rq) { list_for_each_entry_safe(p, n, &cfs_rq->tasks, se.group_node) { if (throttled_lb_pair(task_group(p), - env->busiest_rq->cpu, env->this_cpu)) + env->src_cpu, env->dst_cpu)) break; if (!can_migrate_task(p, env)) continue; - pull_task(env->busiest_rq, p, env->this_rq, env->this_cpu); + move_task(p, env); /* - * Right now, this is only the second place pull_task() - * is called, so we can safely collect pull_task() - * stats here rather than inside pull_task(). + * Right now, this is only the second place move_task() + * is called, so we can safely collect move_task() + * stats here rather than inside move_task(). */ schedstat_inc(env->sd, lb_gained[env->idle]); return 1; @@ -3242,7 +3242,7 @@ static unsigned long balance_tasks(struct lb_env *env) if (env->max_load_move == 0) goto out; - list_for_each_entry_safe(p, n, &env->busiest_cfs_rq->tasks, se.group_node) { + list_for_each_entry_safe(p, n, &env->src_cfs_rq->tasks, se.group_node) { if (loops++ > sysctl_sched_nr_migrate) { env->flags |= LBF_NEED_BREAK; break; @@ -3252,7 +3252,7 @@ static unsigned long balance_tasks(struct lb_env *env) !can_migrate_task(p, env)) continue; - pull_task(env->busiest_rq, p, env->this_rq, env->this_cpu); + move_task(p, env); pulled++; rem_load_move -= p->se.load.weight; @@ -3277,9 +3277,9 @@ static unsigned long balance_tasks(struct lb_env *env) } out: /* - * Right now, this is one of only two places pull_task() is called, - * so we can safely collect pull_task() stats here rather than - * inside pull_task(). + * Right now, this is one of only two places move_task() is called, + * so we can safely collect move_task() stats here rather than + * inside move_task(). */ schedstat_add(env->sd, lb_gained[env->idle], pulled); @@ -3372,11 +3372,11 @@ static unsigned long load_balance_fair(struct lb_env *env) long rem_load_move = env->max_load_move; rcu_read_lock(); - update_h_load(cpu_of(env->busiest_rq)); + update_h_load(cpu_of(env->src_rq)); - for_each_leaf_cfs_rq(env->busiest_rq, env->busiest_cfs_rq) { - unsigned long busiest_h_load = env->busiest_cfs_rq->h_load; - unsigned long busiest_weight = env->busiest_cfs_rq->load.weight; + for_each_leaf_cfs_rq(env->src_rq, env->src_cfs_rq) { + unsigned long busiest_h_load = env->src_cfs_rq->h_load; + unsigned long busiest_weight = env->src_cfs_rq->load.weight; u64 rem_load, moved_load; if (env->flags & (LBF_NEED_BREAK|LBF_ABORT)) @@ -3385,12 +3385,12 @@ static unsigned long load_balance_fair(struct lb_env *env) /* * empty group or part of a throttled hierarchy */ - if (!env->busiest_cfs_rq->task_weight) + if (!env->src_cfs_rq->task_weight) continue; - if (throttled_lb_pair(env->busiest_cfs_rq->tg, - cpu_of(env->busiest_rq), - env->this_cpu)) + if (throttled_lb_pair(env->src_cfs_rq->tg, + cpu_of(env->src_rq), + env->dst_cpu)) continue; rem_load = (u64)rem_load_move * busiest_weight; @@ -3420,7 +3420,7 @@ static inline void update_shares(int cpu) static unsigned long load_balance_fair(struct lb_env *env) { - env->busiest_cfs_rq = &env->busiest_rq->cfs; + env->src_cfs_rq = &env->src_rq->cfs; return balance_tasks(env); } #endif @@ -3451,7 +3451,7 @@ static int move_tasks(struct lb_env *env) * kernels will stop after the first task is pulled to minimize * the critical section. */ - if (env->idle == CPU_NEWLY_IDLE && env->this_rq->nr_running) { + if (env->idle == CPU_NEWLY_IDLE && env->dst_rq->nr_running) { env->flags |= LBF_ABORT; break; } @@ -4461,8 +4461,8 @@ static int load_balance(int this_cpu, struct rq *this_rq, struct lb_env env = { .sd = sd, - .this_cpu = this_cpu, - .this_rq = this_rq, + .dst_cpu = this_cpu, + .dst_rq = this_rq, .idle = idle, }; @@ -4502,7 +4502,8 @@ redo: */ env.flags |= LBF_ALL_PINNED; env.max_load_move = imbalance; - env.busiest_rq = busiest; + env.src_cpu = busiest->cpu; + env.src_rq = busiest; local_irq_save(flags); double_rq_lock(this_rq, busiest); @@ -4722,9 +4723,10 @@ static int active_load_balance_cpu_stop(void *data) if (likely(sd)) { struct lb_env env = { .sd = sd, - .this_cpu = target_cpu, - .this_rq = target_rq, - .busiest_rq = busiest_rq, + .dst_cpu = target_cpu, + .dst_rq = target_rq, + .src_cpu = busiest_rq->cpu, + .src_rq = busiest_rq, .idle = CPU_IDLE, }; -- cgit v0.10.2 From 367456c756a6b84f493ca9cc5b17b1f5d38ef466 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 20 Feb 2012 21:49:09 +0100 Subject: sched: Ditch per cgroup task lists for load-balancing Per cgroup load-balance has numerous problems, chief amongst them that there is no real sane order in them. So stop pretending it makes sense and enqueue all tasks on a single list. This also allows us to more easily fix the fwd progress issue uncovered by the lock-break stuff. Rotate the list on failure to migreate and limit the total iterations to nr_running (which with releasing the lock isn't strictly accurate but close enough). Also add a filter that skips very light tasks on the first attempt around the list, this attempts to avoid shooting whole cgroups around without affecting over balance. Signed-off-by: Peter Zijlstra Cc: pjt@google.com Link: http://lkml.kernel.org/n/tip-tx8yqydc7eimgq7i4rkc3a4g@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 25e06a5..9554512 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -6959,6 +6959,9 @@ void __init sched_init(void) rq->online = 0; rq->idle_stamp = 0; rq->avg_idle = 2*sysctl_sched_migration_cost; + + INIT_LIST_HEAD(&rq->cfs_tasks); + rq_attach_root(rq, &def_root_domain); #ifdef CONFIG_NO_HZ rq->nohz_flags = 0; diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 233d051..a0424fc 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -776,29 +776,16 @@ 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)) update_load_add(&rq_of(cfs_rq)->load, se->load.weight); - if (entity_is_task(se)) { - add_cfs_task_weight(cfs_rq, se->load.weight); - list_add(&se->group_node, &cfs_rq->tasks); - } +#ifdef CONFIG_SMP + if (entity_is_task(se)) + list_add(&se->group_node, &rq_of(cfs_rq)->cfs_tasks); +#endif cfs_rq->nr_running++; } @@ -808,10 +795,8 @@ 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)) update_load_sub(&rq_of(cfs_rq)->load, se->load.weight); - if (entity_is_task(se)) { - add_cfs_task_weight(cfs_rq, -se->load.weight); + if (entity_is_task(se)) list_del_init(&se->group_node); - } cfs_rq->nr_running--; } @@ -3085,17 +3070,14 @@ static bool yield_to_task_fair(struct rq *rq, struct task_struct *p, bool preemp static unsigned long __read_mostly max_load_balance_interval = HZ/10; #define LBF_ALL_PINNED 0x01 -#define LBF_NEED_BREAK 0x02 /* clears into HAD_BREAK */ -#define LBF_HAD_BREAK 0x04 -#define LBF_HAD_BREAKS 0x0C /* count HAD_BREAKs overflows into ABORT */ -#define LBF_ABORT 0x10 +#define LBF_NEED_BREAK 0x02 +#define LBF_ABORT 0x04 struct lb_env { struct sched_domain *sd; int src_cpu; struct rq *src_rq; - struct cfs_rq *src_cfs_rq; int dst_cpu; struct rq *dst_rq; @@ -3103,6 +3085,10 @@ struct lb_env { enum cpu_idle_type idle; unsigned long max_load_move; unsigned int flags; + + unsigned int loop; + unsigned int loop_break; + unsigned int loop_max; }; /* @@ -3208,53 +3194,69 @@ int can_migrate_task(struct task_struct *p, struct lb_env *env) static int move_one_task(struct lb_env *env) { struct task_struct *p, *n; - struct cfs_rq *cfs_rq; - for_each_leaf_cfs_rq(env->src_rq, cfs_rq) { - list_for_each_entry_safe(p, n, &cfs_rq->tasks, se.group_node) { - if (throttled_lb_pair(task_group(p), - env->src_cpu, env->dst_cpu)) - break; + list_for_each_entry_safe(p, n, &env->src_rq->cfs_tasks, se.group_node) { + if (throttled_lb_pair(task_group(p), env->src_rq->cpu, env->dst_cpu)) + continue; - if (!can_migrate_task(p, env)) - continue; + if (!can_migrate_task(p, env)) + continue; - move_task(p, env); - /* - * Right now, this is only the second place move_task() - * is called, so we can safely collect move_task() - * stats here rather than inside move_task(). - */ - schedstat_inc(env->sd, lb_gained[env->idle]); - return 1; - } + move_task(p, env); + /* + * Right now, this is only the second place move_task() + * is called, so we can safely collect move_task() + * stats here rather than inside move_task(). + */ + schedstat_inc(env->sd, lb_gained[env->idle]); + return 1; } - return 0; } +static unsigned long task_h_load(struct task_struct *p); + static unsigned long balance_tasks(struct lb_env *env) { - int loops = 0, pulled = 0; long rem_load_move = env->max_load_move; struct task_struct *p, *n; + unsigned long load; + int pulled = 0; if (env->max_load_move == 0) goto out; - list_for_each_entry_safe(p, n, &env->src_cfs_rq->tasks, se.group_node) { - if (loops++ > sysctl_sched_nr_migrate) { + list_for_each_entry_safe(p, n, &env->src_rq->cfs_tasks, se.group_node) { + env->loop++; + /* We've more or less seen every task there is, call it quits */ + if (env->loop > env->loop_max) { + env->flags |= LBF_ABORT; + break; + } + /* take a beather every nr_migrate tasks */ + if (env->loop > env->loop_break) { + env->loop_break += sysctl_sched_nr_migrate; env->flags |= LBF_NEED_BREAK; break; } - if ((p->se.load.weight >> 1) > rem_load_move || - !can_migrate_task(p, env)) - continue; + if (throttled_lb_pair(task_group(p), env->src_rq->cpu, + env->dst_cpu)) + goto next; + + load = task_h_load(p); + if (load < 16 && !env->sd->nr_balance_failed) + goto next; + + if ((load * 2) > rem_load_move) + goto next; + + if (!can_migrate_task(p, env)) + goto next; move_task(p, env); pulled++; - rem_load_move -= p->se.load.weight; + rem_load_move -= load; #ifdef CONFIG_PREEMPT /* @@ -3274,6 +3276,10 @@ static unsigned long balance_tasks(struct lb_env *env) */ if (rem_load_move <= 0) break; + + continue; +next: + list_move_tail(&p->se.group_node, &env->src_rq->cfs_tasks); } out: /* @@ -3363,65 +3369,33 @@ static int tg_load_down(struct task_group *tg, void *data) static void update_h_load(long cpu) { + rcu_read_lock(); walk_tg_tree(tg_load_down, tg_nop, (void *)cpu); + rcu_read_unlock(); } -static unsigned long load_balance_fair(struct lb_env *env) +static unsigned long task_h_load(struct task_struct *p) { - unsigned long max_load_move = env->max_load_move; - long rem_load_move = env->max_load_move; - - rcu_read_lock(); - update_h_load(cpu_of(env->src_rq)); - - for_each_leaf_cfs_rq(env->src_rq, env->src_cfs_rq) { - unsigned long busiest_h_load = env->src_cfs_rq->h_load; - unsigned long busiest_weight = env->src_cfs_rq->load.weight; - u64 rem_load, moved_load; - - if (env->flags & (LBF_NEED_BREAK|LBF_ABORT)) - break; - - /* - * empty group or part of a throttled hierarchy - */ - if (!env->src_cfs_rq->task_weight) - continue; - - if (throttled_lb_pair(env->src_cfs_rq->tg, - cpu_of(env->src_rq), - env->dst_cpu)) - continue; - - rem_load = (u64)rem_load_move * busiest_weight; - rem_load = div_u64(rem_load, busiest_h_load + 1); - - env->max_load_move = rem_load; - - moved_load = balance_tasks(env); - if (!moved_load) - continue; - - moved_load *= busiest_h_load; - moved_load = div_u64(moved_load, busiest_weight + 1); + struct cfs_rq *cfs_rq = task_cfs_rq(p); + unsigned long load; - rem_load_move -= moved_load; - if (rem_load_move < 0) - break; - } - rcu_read_unlock(); + load = p->se.load.weight; + load = div_u64(load * cfs_rq->h_load, cfs_rq->load.weight + 1); - return max_load_move - rem_load_move; + return load; } #else static inline void update_shares(int cpu) { } -static unsigned long load_balance_fair(struct lb_env *env) +static inline void update_h_load(long cpu) { - env->src_cfs_rq = &env->src_rq->cfs; - return balance_tasks(env); +} + +static unsigned long task_h_load(struct task_struct *p) +{ + return p->se.load.weight; } #endif @@ -3437,9 +3411,10 @@ static int move_tasks(struct lb_env *env) unsigned long max_load_move = env->max_load_move; unsigned long total_load_moved = 0, load_moved; + update_h_load(cpu_of(env->src_rq)); do { env->max_load_move = max_load_move - total_load_moved; - load_moved = load_balance_fair(env); + load_moved = balance_tasks(env); total_load_moved += load_moved; if (env->flags & (LBF_NEED_BREAK|LBF_ABORT)) @@ -4464,6 +4439,7 @@ static int load_balance(int this_cpu, struct rq *this_rq, .dst_cpu = this_cpu, .dst_rq = this_rq, .idle = idle, + .loop_break = sysctl_sched_nr_migrate, }; cpumask_copy(cpus, cpu_active_mask); @@ -4504,6 +4480,7 @@ redo: env.max_load_move = imbalance; env.src_cpu = busiest->cpu; env.src_rq = busiest; + env.loop_max = busiest->nr_running; local_irq_save(flags); double_rq_lock(this_rq, busiest); @@ -4521,9 +4498,7 @@ redo: goto out_balanced; if (env.flags & LBF_NEED_BREAK) { - env.flags += LBF_HAD_BREAK - LBF_NEED_BREAK; - if (env.flags & LBF_ABORT) - goto out_balanced; + env.flags &= ~LBF_NEED_BREAK; goto redo; } @@ -5357,7 +5332,6 @@ static void set_curr_task_fair(struct rq *rq) void init_cfs_rq(struct cfs_rq *cfs_rq) { cfs_rq->tasks_timeline = RB_ROOT; - INIT_LIST_HEAD(&cfs_rq->tasks); cfs_rq->min_vruntime = (u64)(-(1LL << 20)); #ifndef CONFIG_64BIT cfs_rq->min_vruntime_copy = cfs_rq->min_vruntime; diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index c0660a1..753bdd5 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -212,9 +212,6 @@ struct cfs_rq { struct rb_root tasks_timeline; struct rb_node *rb_leftmost; - struct list_head tasks; - struct list_head *balance_iterator; - /* * 'curr' points to currently running entity on this cfs_rq. * It is set to NULL otherwise (i.e when none are currently running). @@ -242,11 +239,6 @@ struct cfs_rq { #ifdef CONFIG_SMP /* - * the part of load.weight contributed by tasks - */ - unsigned long task_weight; - - /* * h_load = weight * f(tg) * * Where f(tg) is the recursive weight fraction assigned to @@ -420,6 +412,8 @@ struct rq { int cpu; int online; + struct list_head cfs_tasks; + u64 rt_avg; u64 age_stamp; u64 idle_stamp; -- cgit v0.10.2 From 2e5b5b3a1b7768c89fbfeca18e75f8ee377e924c Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Thu, 23 Feb 2012 17:41:27 +0900 Subject: sched: Clean up parameter passing of proc_sched_autogroup_set_nice() Pass nice as a value to proc_sched_autogroup_set_nice(). No side effect is expected, and the variable err will be overwritten with the return value. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/4F45FBB7.5090607@ct.jp.nec.com Signed-off-by: Ingo Molnar diff --git a/fs/proc/base.c b/fs/proc/base.c index d4548dd..965d4bd 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -1310,8 +1310,7 @@ sched_autogroup_write(struct file *file, const char __user *buf, if (!p) return -ESRCH; - err = nice; - err = proc_sched_autogroup_set_nice(p, &err); + err = proc_sched_autogroup_set_nice(p, nice); if (err) count = err; diff --git a/include/linux/sched.h b/include/linux/sched.h index c628a91..c298fb9 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2065,7 +2065,7 @@ extern void sched_autogroup_fork(struct signal_struct *sig); extern void sched_autogroup_exit(struct signal_struct *sig); #ifdef CONFIG_PROC_FS extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m); -extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice); +extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice); #endif #else static inline void sched_autogroup_create_attach(struct task_struct *p) { } diff --git a/kernel/sched/auto_group.c b/kernel/sched/auto_group.c index e8a1f83..0984a21 100644 --- a/kernel/sched/auto_group.c +++ b/kernel/sched/auto_group.c @@ -195,20 +195,20 @@ __setup("noautogroup", setup_autogroup); #ifdef CONFIG_PROC_FS -int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice) +int proc_sched_autogroup_set_nice(struct task_struct *p, int nice) { static unsigned long next = INITIAL_JIFFIES; struct autogroup *ag; int err; - if (*nice < -20 || *nice > 19) + if (nice < -20 || nice > 19) return -EINVAL; - err = security_task_setnice(current, *nice); + err = security_task_setnice(current, nice); if (err) return err; - if (*nice < 0 && !can_nice(current, *nice)) + if (nice < 0 && !can_nice(current, nice)) return -EPERM; /* this is a heavy operation taking global locks.. */ @@ -219,9 +219,9 @@ int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice) ag = autogroup_task_get(p); down_write(&ag->lock); - err = sched_group_set_shares(ag->tg, prio_to_weight[*nice + 20]); + err = sched_group_set_shares(ag->tg, prio_to_weight[nice + 20]); if (!err) - ag->nice = *nice; + ag->nice = nice; up_write(&ag->lock); autogroup_kref_put(ag); -- cgit v0.10.2 From 5d6523ebd2f67de9d23285aad7f3910e7b0aee83 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sat, 10 Mar 2012 00:07:36 +0100 Subject: sched: Fix load-balance wreckage Commit 367456c ("sched: Ditch per cgroup task lists for load-balancing") completely wrecked load-balancing due to a few silly mistakes. Correct those and remove more pointless code. Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-zk04ihygwxn7qqrlpaf73b0r@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index a0424fc..def17aa 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -784,7 +784,7 @@ account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se) update_load_add(&rq_of(cfs_rq)->load, se->load.weight); #ifdef CONFIG_SMP if (entity_is_task(se)) - list_add(&se->group_node, &rq_of(cfs_rq)->cfs_tasks); + list_add_tail(&se->group_node, &rq_of(cfs_rq)->cfs_tasks); #endif cfs_rq->nr_running++; } @@ -3071,7 +3071,6 @@ static unsigned long __read_mostly max_load_balance_interval = HZ/10; #define LBF_ALL_PINNED 0x01 #define LBF_NEED_BREAK 0x02 -#define LBF_ABORT 0x04 struct lb_env { struct sched_domain *sd; @@ -3083,7 +3082,7 @@ struct lb_env { struct rq *dst_rq; enum cpu_idle_type idle; - unsigned long max_load_move; + long load_move; unsigned int flags; unsigned int loop; @@ -3216,39 +3215,47 @@ static int move_one_task(struct lb_env *env) static unsigned long task_h_load(struct task_struct *p); -static unsigned long balance_tasks(struct lb_env *env) +/* + * move_tasks tries to move up to load_move weighted load from busiest to + * this_rq, as part of a balancing operation within domain "sd". + * Returns 1 if successful and 0 otherwise. + * + * Called with both runqueues locked. + */ +static int move_tasks(struct lb_env *env) { - long rem_load_move = env->max_load_move; - struct task_struct *p, *n; + struct list_head *tasks = &env->src_rq->cfs_tasks; + struct task_struct *p; unsigned long load; int pulled = 0; - if (env->max_load_move == 0) - goto out; + if (env->load_move <= 0) + return 0; + + while (!list_empty(tasks)) { + p = list_first_entry(tasks, struct task_struct, se.group_node); - list_for_each_entry_safe(p, n, &env->src_rq->cfs_tasks, se.group_node) { env->loop++; /* We've more or less seen every task there is, call it quits */ - if (env->loop > env->loop_max) { - env->flags |= LBF_ABORT; + if (env->loop > env->loop_max) break; - } - /* take a beather every nr_migrate tasks */ + + /* take a breather every nr_migrate tasks */ if (env->loop > env->loop_break) { env->loop_break += sysctl_sched_nr_migrate; env->flags |= LBF_NEED_BREAK; break; } - if (throttled_lb_pair(task_group(p), env->src_rq->cpu, - env->dst_cpu)) + if (throttled_lb_pair(task_group(p), env->src_cpu, env->dst_cpu)) goto next; load = task_h_load(p); + if (load < 16 && !env->sd->nr_balance_failed) goto next; - if ((load * 2) > rem_load_move) + if ((load / 2) > env->load_move) goto next; if (!can_migrate_task(p, env)) @@ -3256,7 +3263,7 @@ static unsigned long balance_tasks(struct lb_env *env) move_task(p, env); pulled++; - rem_load_move -= load; + env->load_move -= load; #ifdef CONFIG_PREEMPT /* @@ -3264,24 +3271,22 @@ static unsigned long balance_tasks(struct lb_env *env) * kernels will stop after the first task is pulled to minimize * the critical section. */ - if (env->idle == CPU_NEWLY_IDLE) { - env->flags |= LBF_ABORT; + if (env->idle == CPU_NEWLY_IDLE) break; - } #endif /* * We only want to steal up to the prescribed amount of * weighted load. */ - if (rem_load_move <= 0) + if (env->load_move <= 0) break; continue; next: - list_move_tail(&p->se.group_node, &env->src_rq->cfs_tasks); + list_move_tail(&p->se.group_node, tasks); } -out: + /* * Right now, this is one of only two places move_task() is called, * so we can safely collect move_task() stats here rather than @@ -3289,7 +3294,7 @@ out: */ schedstat_add(env->sd, lb_gained[env->idle], pulled); - return env->max_load_move - rem_load_move; + return pulled; } #ifdef CONFIG_FAIR_GROUP_SCHED @@ -3399,43 +3404,6 @@ static unsigned long task_h_load(struct task_struct *p) } #endif -/* - * move_tasks tries to move up to max_load_move weighted load from busiest to - * this_rq, as part of a balancing operation within domain "sd". - * Returns 1 if successful and 0 otherwise. - * - * Called with both runqueues locked. - */ -static int move_tasks(struct lb_env *env) -{ - unsigned long max_load_move = env->max_load_move; - unsigned long total_load_moved = 0, load_moved; - - update_h_load(cpu_of(env->src_rq)); - do { - env->max_load_move = max_load_move - total_load_moved; - load_moved = balance_tasks(env); - total_load_moved += load_moved; - - if (env->flags & (LBF_NEED_BREAK|LBF_ABORT)) - break; - -#ifdef CONFIG_PREEMPT - /* - * NEWIDLE balancing is a source of latency, so preemptible - * kernels will stop after the first task is pulled to minimize - * the critical section. - */ - if (env->idle == CPU_NEWLY_IDLE && env->dst_rq->nr_running) { - env->flags |= LBF_ABORT; - break; - } -#endif - } while (load_moved && max_load_move > total_load_moved); - - return total_load_moved > 0; -} - /********** Helpers for find_busiest_group ************************/ /* * sd_lb_stats - Structure to store the statistics of a sched_domain @@ -4477,31 +4445,31 @@ redo: * correctly treated as an imbalance. */ env.flags |= LBF_ALL_PINNED; - env.max_load_move = imbalance; + env.load_move = imbalance; env.src_cpu = busiest->cpu; env.src_rq = busiest; env.loop_max = busiest->nr_running; +more_balance: local_irq_save(flags); double_rq_lock(this_rq, busiest); - ld_moved = move_tasks(&env); + if (!env.loop) + update_h_load(env.src_cpu); + ld_moved += move_tasks(&env); double_rq_unlock(this_rq, busiest); local_irq_restore(flags); + if (env.flags & LBF_NEED_BREAK) { + env.flags &= ~LBF_NEED_BREAK; + goto more_balance; + } + /* * some other cpu did the load balance for us. */ if (ld_moved && this_cpu != smp_processor_id()) resched_cpu(this_cpu); - if (env.flags & LBF_ABORT) - goto out_balanced; - - if (env.flags & LBF_NEED_BREAK) { - env.flags &= ~LBF_NEED_BREAK; - goto redo; - } - /* All tasks on this runqueue were pinned by CPU affinity */ if (unlikely(env.flags & LBF_ALL_PINNED)) { cpumask_clear_cpu(cpu_of(busiest), cpus); -- cgit v0.10.2 From 5fbd036b552f633abb394a319f7c62a5c86a9cd7 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 15 Dec 2011 17:09:22 +0100 Subject: sched: Cleanup cpu_active madness Stepan found: CPU0 CPUn _cpu_up() __cpu_up() boostrap() notify_cpu_starting() set_cpu_online() while (!cpu_active()) cpu_relax() smp_call_function(.wait=1) /* we find cpu_online() is true */ arch_send_call_function_ipi_mask() /* wait-forever-more */ local_irq_enable() cpu_notify(CPU_ONLINE) sched_cpu_active() set_cpu_active() Now the purpose of cpu_active is mostly with bringing down a cpu, where we mark it !active to avoid the load-balancer from moving tasks to it while we tear down the cpu. This is required because we only update the sched_domain tree after we brought the cpu-down. And this is needed so that some tasks can still run while we bring it down, we just don't want new tasks to appear. On cpu-up however the sched_domain tree doesn't yet include the new cpu, so its invisible to the load-balancer, regardless of the active state. So instead of setting the active state after we boot the new cpu (and consequently having to wait for it before enabling interrupts) set the cpu active before we set it online and avoid the whole mess. Reported-by: Stepan Moskovchenko Signed-off-by: Peter Zijlstra Acked-by: Thomas Gleixner Link: http://lkml.kernel.org/r/1323965362.18942.71.camel@twins Signed-off-by: Ingo Molnar diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c index cdeb727..d616ed5 100644 --- a/arch/arm/kernel/smp.c +++ b/arch/arm/kernel/smp.c @@ -295,13 +295,6 @@ asmlinkage void __cpuinit secondary_start_kernel(void) */ percpu_timer_setup(); - while (!cpu_active(cpu)) - cpu_relax(); - - /* - * cpu_active bit is set, so it's safe to enalbe interrupts - * now. - */ local_irq_enable(); local_fiq_enable(); diff --git a/arch/hexagon/kernel/smp.c b/arch/hexagon/kernel/smp.c index c871a2c..0123c63 100644 --- a/arch/hexagon/kernel/smp.c +++ b/arch/hexagon/kernel/smp.c @@ -179,8 +179,6 @@ void __cpuinit start_secondary(void) printk(KERN_INFO "%s cpu %d\n", __func__, current_thread_info()->cpu); set_cpu_online(cpu, true); - while (!cpumask_test_cpu(cpu, cpu_active_mask)) - cpu_relax(); local_irq_enable(); cpu_idle(); diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c index 2398ce6..b0e28c4 100644 --- a/arch/s390/kernel/smp.c +++ b/arch/s390/kernel/smp.c @@ -550,12 +550,6 @@ int __cpuinit start_secondary(void *cpuvoid) S390_lowcore.restart_psw.addr = PSW_ADDR_AMODE | (unsigned long) psw_restart_int_handler; __ctl_set_bit(0, 28); /* Enable lowcore protection */ - /* - * Wait until the cpu which brought this one up marked it - * active before enabling interrupts. - */ - while (!cpumask_test_cpu(smp_processor_id(), cpu_active_mask)) - cpu_relax(); local_irq_enable(); /* cpu_idle will call schedule for us */ cpu_idle(); diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 66d250c..58f7816 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -291,19 +291,6 @@ notrace static void __cpuinit start_secondary(void *unused) per_cpu(cpu_state, smp_processor_id()) = CPU_ONLINE; x86_platform.nmi_init(); - /* - * Wait until the cpu which brought this one up marked it - * online before enabling interrupts. If we don't do that then - * we can end up waking up the softirq thread before this cpu - * reached the active state, which makes the scheduler unhappy - * and schedule the softirq thread on the wrong cpu. This is - * only observable with forced threaded interrupts, but in - * theory it could also happen w/o them. It's just way harder - * to achieve. - */ - while (!cpumask_test_cpu(smp_processor_id(), cpu_active_mask)) - cpu_relax(); - /* enable local interrupts */ local_irq_enable(); diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 9554512..b1ccce8 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -5410,7 +5410,7 @@ static int __cpuinit sched_cpu_active(struct notifier_block *nfb, unsigned long action, void *hcpu) { switch (action & ~CPU_TASKS_FROZEN) { - case CPU_ONLINE: + case CPU_STARTING: case CPU_DOWN_FAILED: set_cpu_active((long)hcpu, true); return NOTIFY_OK; -- cgit v0.10.2 From 554cecaf733623b327eef9652b65965eb1081b81 Mon Sep 17 00:00:00 2001 From: Diwakar Tundlam Date: Wed, 7 Mar 2012 14:44:26 -0800 Subject: sched/nohz: Correctly initialize 'next_balance' in 'nohz' idle balancer The 'next_balance' field of 'nohz' idle balancer must be initialized to jiffies. Since jiffies is initialized to negative 300 seconds the 'nohz' idle balancer does not run for the first 300s (5mins) after bootup. If no new processes are spawed or no idle cycles happen, the load on the cpus will remain unbalanced for that duration. Signed-off-by: Diwakar Tundlam Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1DD7BFEDD3147247B1355BEFEFE4665237994F30EF@HQMAIL04.nvidia.com Signed-off-by: Ingo Molnar diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index def17aa..11f3979 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -5571,6 +5571,7 @@ __init void init_sched_fair_class(void) open_softirq(SCHED_SOFTIRQ, run_rebalance_domains); #ifdef CONFIG_NO_HZ + nohz.next_balance = jiffies; zalloc_cpumask_var(&nohz.idle_cpus_mask, GFP_NOWAIT); cpu_notifier(sched_ilb_notifier, 0); #endif -- cgit v0.10.2 From 3ccf3e8306156a28213adc720aba807e9a901ad5 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 27 Feb 2012 10:47:00 +0100 Subject: printk/sched: Introduce special printk_sched() for those awkward moments There's a few awkward printk()s inside of scheduler guts that people prefer to keep but really are rather deadlock prone. Fudge around it by storing the text in a per-cpu buffer and poll it using the existing printk_tick() handler. This will drop output when its more frequent than once a tick, however only the affinity thing could possible go that fast and for that just one should suffice to notify the admin he's done something silly.. Signed-off-by: Peter Zijlstra Cc: Linus Torvalds Cc: Andrew Morton Link: http://lkml.kernel.org/n/tip-wua3lmkt3dg8nfts66o6brne@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/include/linux/printk.h b/include/linux/printk.h index f0e22f7..1f77a41 100644 --- a/include/linux/printk.h +++ b/include/linux/printk.h @@ -101,6 +101,11 @@ asmlinkage __printf(1, 2) __cold int printk(const char *fmt, ...); /* + * Special printk facility for scheduler use only, _DO_NOT_USE_ ! + */ +__printf(1, 2) __cold int printk_sched(const char *fmt, ...); + +/* * Please don't use printk_ratelimit(), because it shares ratelimiting state * with all other unrelated printk_ratelimit() callsites. Instead use * printk_ratelimited() or plain old __ratelimit(). @@ -127,6 +132,11 @@ int printk(const char *s, ...) { return 0; } +static inline __printf(1, 2) __cold +int printk_sched(const char *s, ...) +{ + return 0; +} static inline int printk_ratelimit(void) { return 0; diff --git a/kernel/printk.c b/kernel/printk.c index 13c0a11..7ca7ba5 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1208,13 +1208,47 @@ int is_console_locked(void) return console_locked; } +/* + * Delayed printk facility, for scheduler-internal messages: + */ +#define PRINTK_BUF_SIZE 512 + +#define PRINTK_PENDING_WAKEUP 0x01 +#define PRINTK_PENDING_SCHED 0x02 + static DEFINE_PER_CPU(int, printk_pending); +static DEFINE_PER_CPU(char [PRINTK_BUF_SIZE], printk_sched_buf); + +int printk_sched(const char *fmt, ...) +{ + unsigned long flags; + va_list args; + char *buf; + int r; + + local_irq_save(flags); + buf = __get_cpu_var(printk_sched_buf); + + va_start(args, fmt); + r = vsnprintf(buf, PRINTK_BUF_SIZE, fmt, args); + va_end(args); + + __this_cpu_or(printk_pending, PRINTK_PENDING_SCHED); + local_irq_restore(flags); + + return r; +} void printk_tick(void) { if (__this_cpu_read(printk_pending)) { - __this_cpu_write(printk_pending, 0); - wake_up_interruptible(&log_wait); + int pending = __this_cpu_xchg(printk_pending, 0); + if (pending & PRINTK_PENDING_SCHED) { + char *buf = __get_cpu_var(printk_sched_buf); + printk(KERN_WARNING "[sched_delayed] %s", buf); + } + if (pending & PRINTK_PENDING_WAKEUP) + wake_up_interruptible(&log_wait); } } @@ -1228,7 +1262,7 @@ int printk_needs_cpu(int cpu) void wake_up_klogd(void) { if (waitqueue_active(&log_wait)) - this_cpu_write(printk_pending, 1); + this_cpu_or(printk_pending, PRINTK_PENDING_WAKEUP); } /** diff --git a/kernel/sched/core.c b/kernel/sched/core.c index b1ccce8..8781cec 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -1284,7 +1284,7 @@ static int select_fallback_rq(int cpu, struct task_struct *p) * leave kernel. */ if (p->mm && printk_ratelimit()) { - printk(KERN_INFO "process %d (%s) no longer affine to cpu%d\n", + printk_sched("process %d (%s) no longer affine to cpu%d\n", task_pid_nr(p), p->comm, cpu); } diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index 7f7e7cd..b60dad7 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -864,8 +864,14 @@ static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) * but accrue some time due to boosting. */ if (likely(rt_b->rt_runtime)) { + static bool once = false; + rt_rq->rt_throttled = 1; - printk_once(KERN_WARNING "sched: RT throttling activated\n"); + + if (!once) { + once = true; + printk_sched("sched: RT throttling activated\n"); + } } else { /* * In case we did anyway, make it go away, -- cgit v0.10.2 From 8e3fabfde445a872c8aec2296846badf24d7c8b4 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 6 Mar 2012 18:54:26 +0100 Subject: sched: Update yield() docs Suggested-by: Joe Perches Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1331056466.11248.327.camel@twins Signed-off-by: Ingo Molnar diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 8781cec..47614a5 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -4577,8 +4577,24 @@ EXPORT_SYMBOL(__cond_resched_softirq); /** * yield - yield the current processor to other threads. * - * This is a shortcut for kernel-space yielding - it marks the - * thread runnable and calls sys_sched_yield(). + * Do not ever use this function, there's a 99% chance you're doing it wrong. + * + * The scheduler is at all times free to pick the calling task as the most + * eligible task to run, if removing the yield() call from your code breaks + * it, its already broken. + * + * Typical broken usage is: + * + * while (!event) + * yield(); + * + * where one assumes that yield() will let 'the other' process run that will + * make event true. If the current task is a SCHED_FIFO task that will never + * happen. Never use yield() as a progress guarantee!! + * + * If you want to use yield() to wait for something, use wait_event(). + * If you want to use yield() to be 'nice' for others, use cond_resched(). + * If you still want to use yield(), do not! */ void __sched yield(void) { -- cgit v0.10.2 From c308b56b5398779cd3da0f62ab26b0453494c3d4 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 1 Mar 2012 15:04:46 +0100 Subject: sched: Fix nohz load accounting -- again! MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Various people reported nohz load tracking still being wrecked, but Doug spotted the actual problem. We fold the nohz remainder in too soon, causing us to loose samples and under-account. So instead of playing catch-up up-front, always do a single load-fold with whatever state we encounter and only then fold the nohz remainder and play catch-up. Reported-by: Doug Smythies Reported-by: LesÅ=82aw Kope=C4=87 Reported-by: Aman Gupta Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-4v31etnhgg9kwd6ocgx3rxl8@git.kernel.org Signed-off-by: Ingo Molnar diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 47614a5..e3ccc13 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2266,13 +2266,10 @@ calc_load_n(unsigned long load, unsigned long exp, * Once we've updated the global active value, we need to apply the exponential * weights adjusted to the number of cycles missed. */ -static void calc_global_nohz(unsigned long ticks) +static void calc_global_nohz(void) { long delta, active, n; - if (time_before(jiffies, calc_load_update)) - return; - /* * If we crossed a calc_load_update boundary, make sure to fold * any pending idle changes, the respective CPUs might have @@ -2284,31 +2281,25 @@ static void calc_global_nohz(unsigned long ticks) atomic_long_add(delta, &calc_load_tasks); /* - * If we were idle for multiple load cycles, apply them. + * It could be the one fold was all it took, we done! */ - if (ticks >= LOAD_FREQ) { - n = ticks / LOAD_FREQ; + if (time_before(jiffies, calc_load_update + 10)) + return; - active = atomic_long_read(&calc_load_tasks); - active = active > 0 ? active * FIXED_1 : 0; + /* + * Catch-up, fold however many we are behind still + */ + delta = jiffies - calc_load_update - 10; + n = 1 + (delta / LOAD_FREQ); - avenrun[0] = calc_load_n(avenrun[0], EXP_1, active, n); - avenrun[1] = calc_load_n(avenrun[1], EXP_5, active, n); - avenrun[2] = calc_load_n(avenrun[2], EXP_15, active, n); + active = atomic_long_read(&calc_load_tasks); + active = active > 0 ? active * FIXED_1 : 0; - calc_load_update += n * LOAD_FREQ; - } + avenrun[0] = calc_load_n(avenrun[0], EXP_1, active, n); + avenrun[1] = calc_load_n(avenrun[1], EXP_5, active, n); + avenrun[2] = calc_load_n(avenrun[2], EXP_15, active, n); - /* - * Its possible the remainder of the above division also crosses - * a LOAD_FREQ period, the regular check in calc_global_load() - * which comes after this will take care of that. - * - * Consider us being 11 ticks before a cycle completion, and us - * sleeping for 4*LOAD_FREQ + 22 ticks, then the above code will - * age us 4 cycles, and the test in calc_global_load() will - * pick up the final one. - */ + calc_load_update += n * LOAD_FREQ; } #else void calc_load_account_idle(struct rq *this_rq) @@ -2320,7 +2311,7 @@ static inline long calc_load_fold_idle(void) return 0; } -static void calc_global_nohz(unsigned long ticks) +static void calc_global_nohz(void) { } #endif @@ -2348,8 +2339,6 @@ void calc_global_load(unsigned long ticks) { long active; - calc_global_nohz(ticks); - if (time_before(jiffies, calc_load_update + 10)) return; @@ -2361,6 +2350,16 @@ void calc_global_load(unsigned long ticks) avenrun[2] = calc_load(avenrun[2], EXP_15, active); calc_load_update += LOAD_FREQ; + + /* + * Account one period with whatever state we found before + * folding in the nohz state and ageing the entire idle period. + * + * This avoids loosing a sample when we go idle between + * calc_load_account_active() (10 ticks ago) and now and thus + * under-accounting. + */ + calc_global_nohz(); } /* -- cgit v0.10.2 From 9993bc635d01a6ee7f6b833b4ee65ce7c06350b1 Mon Sep 17 00:00:00 2001 From: Salman Qazi Date: Fri, 9 Mar 2012 16:41:01 -0800 Subject: sched/x86: Fix overflow in cyc2ns_offset When a machine boots up, the TSC generally gets reset. However, when kexec is used to boot into a kernel, the TSC value would be carried over from the previous kernel. The computation of cycns_offset in set_cyc2ns_scale is prone to an overflow, if the machine has been up more than 208 days prior to the kexec. The overflow happens when we multiply *scale, even though there is enough room to store the final answer. We fix this issue by decomposing tsc_now into the quotient and remainder of division by CYC2NS_SCALE_FACTOR and then performing the multiplication separately on the two components. Refactor code to share the calculation with the previous fix in __cycles_2_ns(). Signed-off-by: Salman Qazi Acked-by: John Stultz Acked-by: Peter Zijlstra Cc: Paul Turner Cc: john stultz Link: http://lkml.kernel.org/r/20120310004027.19291.88460.stgit@dungbeetle.mtv.corp.google.com Signed-off-by: Ingo Molnar diff --git a/arch/x86/include/asm/timer.h b/arch/x86/include/asm/timer.h index 431793e..34baa0e 100644 --- a/arch/x86/include/asm/timer.h +++ b/arch/x86/include/asm/timer.h @@ -57,14 +57,10 @@ DECLARE_PER_CPU(unsigned long long, cyc2ns_offset); static inline unsigned long long __cycles_2_ns(unsigned long long cyc) { - unsigned long long quot; - unsigned long long rem; int cpu = smp_processor_id(); unsigned long long ns = per_cpu(cyc2ns_offset, cpu); - quot = (cyc >> CYC2NS_SCALE_FACTOR); - rem = cyc & ((1ULL << CYC2NS_SCALE_FACTOR) - 1); - ns += quot * per_cpu(cyc2ns, cpu) + - ((rem * per_cpu(cyc2ns, cpu)) >> CYC2NS_SCALE_FACTOR); + ns += mult_frac(cyc, per_cpu(cyc2ns, cpu), + (1UL << CYC2NS_SCALE_FACTOR)); return ns; } diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index a62c201..183c592 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -620,7 +620,8 @@ static void set_cyc2ns_scale(unsigned long cpu_khz, int cpu) if (cpu_khz) { *scale = (NSEC_PER_MSEC << CYC2NS_SCALE_FACTOR)/cpu_khz; - *offset = ns_now - (tsc_now * *scale >> CYC2NS_SCALE_FACTOR); + *offset = ns_now - mult_frac(tsc_now, *scale, + (1UL << CYC2NS_SCALE_FACTOR)); } sched_clock_idle_wakeup_event(0); diff --git a/include/linux/kernel.h b/include/linux/kernel.h index e834342..d801acb 100644 --- a/include/linux/kernel.h +++ b/include/linux/kernel.h @@ -85,6 +85,19 @@ } \ ) +/* + * Multiplies an integer by a fraction, while avoiding unnecessary + * overflow or loss of precision. + */ +#define mult_frac(x, numer, denom)( \ +{ \ + typeof(x) quot = (x) / (denom); \ + typeof(x) rem = (x) % (denom); \ + (quot * (numer)) + ((rem * (numer)) / (denom)); \ +} \ +) + + #define _RET_IP_ (unsigned long)__builtin_return_address(0) #define _THIS_IP_ ({ __label__ __here; __here: (unsigned long)&&__here; }) -- cgit v0.10.2 From 600e145882802d6ccbfe2c4aea243d97caeb91a9 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 15 Mar 2012 12:35:37 +0100 Subject: printk: Make it compile with !CONFIG_PRINTK Commit 3ccf3e830615 ("printk/sched: Introduce special printk_sched() for those awkward moments") overlooked an #ifdef, so move code around to respect these directives. Reported-by: Stephen Rothwell Signed-off-by: Peter Zijlstra Cc: Randy Dunlap Link: http://lkml.kernel.org/r/1331811337.18960.179.camel@twins Signed-off-by: Ingo Molnar diff --git a/kernel/printk.c b/kernel/printk.c index 49a2ae4..b64ce71 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1222,26 +1222,6 @@ int is_console_locked(void) static DEFINE_PER_CPU(int, printk_pending); static DEFINE_PER_CPU(char [PRINTK_BUF_SIZE], printk_sched_buf); -int printk_sched(const char *fmt, ...) -{ - unsigned long flags; - va_list args; - char *buf; - int r; - - local_irq_save(flags); - buf = __get_cpu_var(printk_sched_buf); - - va_start(args, fmt); - r = vsnprintf(buf, PRINTK_BUF_SIZE, fmt, args); - va_end(args); - - __this_cpu_or(printk_pending, PRINTK_PENDING_SCHED); - local_irq_restore(flags); - - return r; -} - void printk_tick(void) { if (__this_cpu_read(printk_pending)) { @@ -1658,6 +1638,26 @@ late_initcall(printk_late_init); #if defined CONFIG_PRINTK +int printk_sched(const char *fmt, ...) +{ + unsigned long flags; + va_list args; + char *buf; + int r; + + local_irq_save(flags); + buf = __get_cpu_var(printk_sched_buf); + + va_start(args, fmt); + r = vsnprintf(buf, PRINTK_BUF_SIZE, fmt, args); + va_end(args); + + __this_cpu_or(printk_pending, PRINTK_PENDING_SCHED); + local_irq_restore(flags); + + return r; +} + /* * printk rate limiting, lifted from the networking subsystem. * -- cgit v0.10.2