From 21aa9af03d06cb1d19a3738e5cf12acff984e69b Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 8 Jun 2010 21:40:37 +0200 Subject: sched: add hooks for workqueue Concurrency managed workqueue needs to know when workers are going to sleep and waking up. Using these two hooks, cmwq keeps track of the current concurrency level and throttles execution of new works if it's too high and wakes up another worker from the sleep hook if it becomes too low. This patch introduces PF_WQ_WORKER to identify workqueue workers and adds the following two hooks. * wq_worker_waking_up(): called when a worker is woken up. * wq_worker_sleeping(): called when a worker is going to sleep and may return a pointer to a local task which should be woken up. The returned task is woken up using try_to_wake_up_local() which is simplified ttwu which is called under rq lock and can only wake up local tasks. Both hooks are currently defined as noop in kernel/workqueue_sched.h. Later cmwq implementation will replace them with proper implementation. These hooks are hard coded as they'll always be enabled. Signed-off-by: Tejun Heo Acked-by: Peter Zijlstra Cc: Mike Galbraith Cc: Ingo Molnar --- include/linux/sched.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index f118809c953f..edc3dd168d87 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1696,6 +1696,7 @@ extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t * #define PF_EXITING 0x00000004 /* getting shut down */ #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ +#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */ #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ -- cgit v1.2.3 From c676329abb2b8359d9a5d734dec0c81779823fd6 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 25 May 2010 10:48:51 +0200 Subject: sched_clock: Add local_clock() API and improve documentation For people who otherwise get to write: cpu_clock(smp_processor_id()), there is now: local_clock(). Also, as per suggestion from Andrew, provide some documentation on the various clock interfaces, and minimize the unsigned long long vs u64 mess. Signed-off-by: Peter Zijlstra Cc: Andrew Morton Cc: Linus Torvalds Cc: Jens Axboe LKML-Reference: <1275052414.1645.52.camel@laptop> Signed-off-by: Ingo Molnar --- arch/parisc/kernel/ftrace.c | 4 +- include/linux/sched.h | 37 ++++++++++-------- kernel/lockdep.c | 2 +- kernel/perf_event.c | 2 +- kernel/rcutorture.c | 3 +- kernel/sched.c | 2 +- kernel/sched_clock.c | 95 ++++++++++++++++++++++++++++++++++++++++----- kernel/trace/trace_clock.c | 2 +- 8 files changed, 113 insertions(+), 34 deletions(-) (limited to 'include/linux/sched.h') diff --git a/arch/parisc/kernel/ftrace.c b/arch/parisc/kernel/ftrace.c index 9877372ffdba..5beb97bafbb1 100644 --- a/arch/parisc/kernel/ftrace.c +++ b/arch/parisc/kernel/ftrace.c @@ -82,7 +82,7 @@ unsigned long ftrace_return_to_handler(unsigned long retval0, unsigned long ret; pop_return_trace(&trace, &ret); - trace.rettime = cpu_clock(raw_smp_processor_id()); + trace.rettime = local_clock(); ftrace_graph_return(&trace); if (unlikely(!ret)) { @@ -126,7 +126,7 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) return; } - calltime = cpu_clock(raw_smp_processor_id()); + calltime = local_clock(); if (push_return_trace(old, calltime, self_addr, &trace.depth) == -EBUSY) { diff --git a/include/linux/sched.h b/include/linux/sched.h index edc3dd168d87..c2d4316a04bb 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1791,20 +1791,23 @@ static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) #endif /* - * Architectures can set this to 1 if they have specified - * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, - * but then during bootup it turns out that sched_clock() - * is reliable after all: + * Do not use outside of architecture code which knows its limitations. + * + * sched_clock() has no promise of monotonicity or bounded drift between + * CPUs, use (which you should not) requires disabling IRQs. + * + * Please use one of the three interfaces below. */ -#ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK -extern int sched_clock_stable; -#endif - -/* ftrace calls sched_clock() directly */ extern unsigned long long notrace sched_clock(void); +/* + * See the comment in kernel/sched_clock.c + */ +extern u64 cpu_clock(int cpu); +extern u64 local_clock(void); +extern u64 sched_clock_cpu(int cpu); + extern void sched_clock_init(void); -extern u64 sched_clock_cpu(int cpu); #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK static inline void sched_clock_tick(void) @@ -1819,17 +1822,19 @@ static inline void sched_clock_idle_wakeup_event(u64 delta_ns) { } #else +/* + * Architectures can set this to 1 if they have specified + * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, + * but then during bootup it turns out that sched_clock() + * is reliable after all: + */ +extern int sched_clock_stable; + extern void sched_clock_tick(void); extern void sched_clock_idle_sleep_event(void); extern void sched_clock_idle_wakeup_event(u64 delta_ns); #endif -/* - * For kernel-internal use: high-speed (but slightly incorrect) per-cpu - * clock constructed from sched_clock(): - */ -extern unsigned long long cpu_clock(int cpu); - extern unsigned long long task_sched_runtime(struct task_struct *task); extern unsigned long long thread_group_sched_runtime(struct task_struct *task); diff --git a/kernel/lockdep.c b/kernel/lockdep.c index 54286798c37b..f2852a510232 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -146,7 +146,7 @@ static DEFINE_PER_CPU(struct lock_class_stats[MAX_LOCKDEP_KEYS], static inline u64 lockstat_clock(void) { - return cpu_clock(smp_processor_id()); + return local_clock(); } static int lock_point(unsigned long points[], unsigned long ip) diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 31d6afe92594..109c5ec88933 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -214,7 +214,7 @@ static void perf_unpin_context(struct perf_event_context *ctx) static inline u64 perf_clock(void) { - return cpu_clock(raw_smp_processor_id()); + return local_clock(); } /* diff --git a/kernel/rcutorture.c b/kernel/rcutorture.c index 6535ac8bc6a5..2e2726d790b9 100644 --- a/kernel/rcutorture.c +++ b/kernel/rcutorture.c @@ -239,8 +239,7 @@ static unsigned long rcu_random(struct rcu_random_state *rrsp) { if (--rrsp->rrs_count < 0) { - rrsp->rrs_state += - (unsigned long)cpu_clock(raw_smp_processor_id()); + rrsp->rrs_state += (unsigned long)local_clock(); rrsp->rrs_count = RCU_RANDOM_REFRESH; } rrsp->rrs_state = rrsp->rrs_state * RCU_RANDOM_MULT + RCU_RANDOM_ADD; diff --git a/kernel/sched.c b/kernel/sched.c index 8f351c56567f..3abd8f780dae 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1647,7 +1647,7 @@ static void update_shares(struct sched_domain *sd) if (root_task_group_empty()) return; - now = cpu_clock(raw_smp_processor_id()); + now = local_clock(); elapsed = now - sd->last_update; if (elapsed >= (s64)(u64)sysctl_sched_shares_ratelimit) { diff --git a/kernel/sched_clock.c b/kernel/sched_clock.c index 906a0f718cb3..52f1a149bfb1 100644 --- a/kernel/sched_clock.c +++ b/kernel/sched_clock.c @@ -10,19 +10,55 @@ * Ingo Molnar * Guillaume Chazarain * - * Create a semi stable clock from a mixture of other events, including: - * - gtod + * + * What: + * + * cpu_clock(i) provides a fast (execution time) high resolution + * clock with bounded drift between CPUs. The value of cpu_clock(i) + * is monotonic for constant i. The timestamp returned is in nanoseconds. + * + * ######################### BIG FAT WARNING ########################## + * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can # + * # go backwards !! # + * #################################################################### + * + * There is no strict promise about the base, although it tends to start + * at 0 on boot (but people really shouldn't rely on that). + * + * cpu_clock(i) -- can be used from any context, including NMI. + * sched_clock_cpu(i) -- must be used with local IRQs disabled (implied by NMI) + * local_clock() -- is cpu_clock() on the current cpu. + * + * How: + * + * The implementation either uses sched_clock() when + * !CONFIG_HAVE_UNSTABLE_SCHED_CLOCK, which means in that case the + * sched_clock() is assumed to provide these properties (mostly it means + * the architecture provides a globally synchronized highres time source). + * + * Otherwise it tries to create a semi stable clock from a mixture of other + * clocks, including: + * + * - GTOD (clock monotomic) * - sched_clock() * - explicit idle events * - * We use gtod as base and the unstable clock deltas. The deltas are filtered, - * making it monotonic and keeping it within an expected window. + * We use GTOD as base and use sched_clock() deltas to improve resolution. The + * deltas are filtered to provide monotonicity and keeping it within an + * expected window. * * Furthermore, explicit sleep and wakeup hooks allow us to account for time * that is otherwise invisible (TSC gets stopped). * - * The clock: sched_clock_cpu() is monotonic per cpu, and should be somewhat - * consistent between cpus (never more than 2 jiffies difference). + * + * Notes: + * + * The !IRQ-safetly of sched_clock() and sched_clock_cpu() comes from things + * like cpufreq interrupts that can change the base clock (TSC) multiplier + * and cause funny jumps in time -- although the filtering provided by + * sched_clock_cpu() should mitigate serious artifacts we cannot rely on it + * in general since for !CONFIG_HAVE_UNSTABLE_SCHED_CLOCK we fully rely on + * sched_clock(). */ #include #include @@ -170,6 +206,11 @@ again: return val; } +/* + * Similar to cpu_clock(), but requires local IRQs to be disabled. + * + * See cpu_clock(). + */ u64 sched_clock_cpu(int cpu) { struct sched_clock_data *scd; @@ -237,9 +278,19 @@ void sched_clock_idle_wakeup_event(u64 delta_ns) } EXPORT_SYMBOL_GPL(sched_clock_idle_wakeup_event); -unsigned long long cpu_clock(int cpu) +/* + * As outlined at the top, provides a fast, high resolution, nanosecond + * time source that is monotonic per cpu argument and has bounded drift + * between cpus. + * + * ######################### BIG FAT WARNING ########################## + * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can # + * # go backwards !! # + * #################################################################### + */ +u64 cpu_clock(int cpu) { - unsigned long long clock; + u64 clock; unsigned long flags; local_irq_save(flags); @@ -249,6 +300,25 @@ unsigned long long cpu_clock(int cpu) return clock; } +/* + * Similar to cpu_clock() for the current cpu. Time will only be observed + * to be monotonic if care is taken to only compare timestampt taken on the + * same CPU. + * + * See cpu_clock(). + */ +u64 local_clock(void) +{ + u64 clock; + unsigned long flags; + + local_irq_save(flags); + clock = sched_clock_cpu(smp_processor_id()); + local_irq_restore(flags); + + return clock; +} + #else /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ void sched_clock_init(void) @@ -264,12 +334,17 @@ u64 sched_clock_cpu(int cpu) return sched_clock(); } - -unsigned long long cpu_clock(int cpu) +u64 cpu_clock(int cpu) { return sched_clock_cpu(cpu); } +u64 local_clock(void) +{ + return sched_clock_cpu(0); +} + #endif /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ EXPORT_SYMBOL_GPL(cpu_clock); +EXPORT_SYMBOL_GPL(local_clock); diff --git a/kernel/trace/trace_clock.c b/kernel/trace/trace_clock.c index 9d589d8dcd1a..1723e2b8c589 100644 --- a/kernel/trace/trace_clock.c +++ b/kernel/trace/trace_clock.c @@ -56,7 +56,7 @@ u64 notrace trace_clock_local(void) */ u64 notrace trace_clock(void) { - return cpu_clock(raw_smp_processor_id()); + return local_clock(); } -- cgit v1.2.3 From 83cd4fe27ad8446619b2e030b171b858501de87d Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Fri, 21 May 2010 17:09:41 -0700 Subject: sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Suresh Siddha Signed-off-by: Peter Zijlstra Cc: Thomas Gleixner LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar --- include/linux/sched.h | 9 +- kernel/hrtimer.c | 8 +- kernel/sched.c | 34 ++++- kernel/sched_fair.c | 329 ++++++++++++++++++++++++++++------------------- kernel/time/tick-sched.c | 8 +- kernel/timer.c | 8 +- 6 files changed, 237 insertions(+), 159 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index c2d4316a04bb..a3e5b1cd0438 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -271,14 +271,11 @@ extern int runqueue_is_locked(int cpu); extern cpumask_var_t nohz_cpu_mask; #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) -extern int select_nohz_load_balancer(int cpu); -extern int get_nohz_load_balancer(void); +extern void select_nohz_load_balancer(int stop_tick); +extern int get_nohz_timer_target(void); extern int nohz_ratelimit(int cpu); #else -static inline int select_nohz_load_balancer(int cpu) -{ - return 0; -} +static inline void select_nohz_load_balancer(int stop_tick) { } static inline int nohz_ratelimit(int cpu) { diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c index 5c69e996bd0f..e934339fbbef 100644 --- a/kernel/hrtimer.c +++ b/kernel/hrtimer.c @@ -144,12 +144,8 @@ struct hrtimer_clock_base *lock_hrtimer_base(const struct hrtimer *timer, static int hrtimer_get_target(int this_cpu, int pinned) { #ifdef CONFIG_NO_HZ - if (!pinned && get_sysctl_timer_migration() && idle_cpu(this_cpu)) { - int preferred_cpu = get_nohz_load_balancer(); - - if (preferred_cpu >= 0) - return preferred_cpu; - } + if (!pinned && get_sysctl_timer_migration() && idle_cpu(this_cpu)) + return get_nohz_timer_target(); #endif return this_cpu; } diff --git a/kernel/sched.c b/kernel/sched.c index a757f6b11cbd..132950b33dde 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -460,7 +460,7 @@ struct rq { unsigned long last_load_update_tick; #ifdef CONFIG_NO_HZ u64 nohz_stamp; - unsigned char in_nohz_recently; + unsigned char nohz_balance_kick; #endif unsigned int skip_clock_update; @@ -1194,6 +1194,27 @@ static void resched_cpu(int cpu) } #ifdef CONFIG_NO_HZ +/* + * In the semi idle case, use the nearest busy cpu for migrating timers + * from an idle cpu. This is good for power-savings. + * + * We don't do similar optimization for completely idle system, as + * selecting an idle cpu will add more delays to the timers than intended + * (as that cpu's timer base may not be uptodate wrt jiffies etc). + */ +int get_nohz_timer_target(void) +{ + int cpu = smp_processor_id(); + int i; + struct sched_domain *sd; + + for_each_domain(cpu, sd) { + for_each_cpu(i, sched_domain_span(sd)) + if (!idle_cpu(i)) + return i; + } + return cpu; +} /* * When add_timer_on() enqueues a timer into the timer wheel of an * idle CPU then this timer might expire before the next timer event @@ -7791,6 +7812,10 @@ void __init sched_init(void) rq->idle_stamp = 0; rq->avg_idle = 2*sysctl_sched_migration_cost; rq_attach_root(rq, &def_root_domain); +#ifdef CONFIG_NO_HZ + rq->nohz_balance_kick = 0; + init_sched_softirq_csd(&per_cpu(remote_sched_softirq_cb, i)); +#endif #endif init_rq_hrtick(rq); atomic_set(&rq->nr_iowait, 0); @@ -7835,8 +7860,11 @@ void __init sched_init(void) zalloc_cpumask_var(&nohz_cpu_mask, GFP_NOWAIT); #ifdef CONFIG_SMP #ifdef CONFIG_NO_HZ - zalloc_cpumask_var(&nohz.cpu_mask, GFP_NOWAIT); - alloc_cpumask_var(&nohz.ilb_grp_nohz_mask, GFP_NOWAIT); + zalloc_cpumask_var(&nohz.idle_cpus_mask, GFP_NOWAIT); + alloc_cpumask_var(&nohz.grp_idle_mask, GFP_NOWAIT); + atomic_set(&nohz.load_balancer, nr_cpu_ids); + atomic_set(&nohz.first_pick_cpu, nr_cpu_ids); + atomic_set(&nohz.second_pick_cpu, nr_cpu_ids); #endif /* May be allocated at isolcpus cmdline parse time */ if (cpu_isolated_map == NULL) diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 22b8b4f2b616..6ee2e0af665b 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -3091,13 +3091,40 @@ out_unlock: } #ifdef CONFIG_NO_HZ + +static DEFINE_PER_CPU(struct call_single_data, remote_sched_softirq_cb); + +static void trigger_sched_softirq(void *data) +{ + raise_softirq_irqoff(SCHED_SOFTIRQ); +} + +static inline void init_sched_softirq_csd(struct call_single_data *csd) +{ + csd->func = trigger_sched_softirq; + csd->info = NULL; + csd->flags = 0; + csd->priv = 0; +} + +/* + * idle load balancing details + * - One of the idle CPUs nominates itself as idle load_balancer, while + * entering idle. + * - This idle load balancer CPU will also go into tickless mode when + * it is idle, just like all other idle CPUs + * - When one of the busy CPUs notice that there may be an idle rebalancing + * needed, they will kick the idle load balancer, which then does idle + * load balancing for all the idle CPUs. + */ static struct { atomic_t load_balancer; - cpumask_var_t cpu_mask; - cpumask_var_t ilb_grp_nohz_mask; -} nohz ____cacheline_aligned = { - .load_balancer = ATOMIC_INIT(-1), -}; + atomic_t first_pick_cpu; + atomic_t second_pick_cpu; + cpumask_var_t idle_cpus_mask; + cpumask_var_t grp_idle_mask; + unsigned long next_balance; /* in jiffy units */ +} nohz ____cacheline_aligned; int get_nohz_load_balancer(void) { @@ -3151,17 +3178,17 @@ static inline struct sched_domain *lowest_flag_domain(int cpu, int flag) */ static inline int is_semi_idle_group(struct sched_group *ilb_group) { - cpumask_and(nohz.ilb_grp_nohz_mask, nohz.cpu_mask, + cpumask_and(nohz.grp_idle_mask, nohz.idle_cpus_mask, sched_group_cpus(ilb_group)); /* * A sched_group is semi-idle when it has atleast one busy cpu * and atleast one idle cpu. */ - if (cpumask_empty(nohz.ilb_grp_nohz_mask)) + if (cpumask_empty(nohz.grp_idle_mask)) return 0; - if (cpumask_equal(nohz.ilb_grp_nohz_mask, sched_group_cpus(ilb_group))) + if (cpumask_equal(nohz.grp_idle_mask, sched_group_cpus(ilb_group))) return 0; return 1; @@ -3194,7 +3221,7 @@ static int find_new_ilb(int cpu) * Optimize for the case when we have no idle CPUs or only one * idle CPU. Don't walk the sched_domain hierarchy in such cases */ - if (cpumask_weight(nohz.cpu_mask) < 2) + if (cpumask_weight(nohz.idle_cpus_mask) < 2) goto out_done; for_each_flag_domain(cpu, sd, SD_POWERSAVINGS_BALANCE) { @@ -3202,7 +3229,7 @@ static int find_new_ilb(int cpu) do { if (is_semi_idle_group(ilb_group)) - return cpumask_first(nohz.ilb_grp_nohz_mask); + return cpumask_first(nohz.grp_idle_mask); ilb_group = ilb_group->next; @@ -3210,98 +3237,116 @@ static int find_new_ilb(int cpu) } out_done: - return cpumask_first(nohz.cpu_mask); + return nr_cpu_ids; } #else /* (CONFIG_SCHED_MC || CONFIG_SCHED_SMT) */ static inline int find_new_ilb(int call_cpu) { - return cpumask_first(nohz.cpu_mask); + return nr_cpu_ids; } #endif +/* + * Kick a CPU to do the nohz balancing, if it is time for it. We pick the + * nohz_load_balancer CPU (if there is one) otherwise fallback to any idle + * CPU (if there is one). + */ +static void nohz_balancer_kick(int cpu) +{ + int ilb_cpu; + + nohz.next_balance++; + + ilb_cpu = get_nohz_load_balancer(); + + if (ilb_cpu >= nr_cpu_ids) { + ilb_cpu = cpumask_first(nohz.idle_cpus_mask); + if (ilb_cpu >= nr_cpu_ids) + return; + } + + if (!cpu_rq(ilb_cpu)->nohz_balance_kick) { + struct call_single_data *cp; + + cpu_rq(ilb_cpu)->nohz_balance_kick = 1; + cp = &per_cpu(remote_sched_softirq_cb, cpu); + __smp_call_function_single(ilb_cpu, cp, 0); + } + return; +} + /* * This routine will try to nominate the ilb (idle load balancing) * owner among the cpus whose ticks are stopped. ilb owner will do the idle - * load balancing on behalf of all those cpus. If all the cpus in the system - * go into this tickless mode, then there will be no ilb owner (as there is - * no need for one) and all the cpus will sleep till the next wakeup event - * arrives... - * - * For the ilb owner, tick is not stopped. And this tick will be used - * for idle load balancing. ilb owner will still be part of - * nohz.cpu_mask.. + * load balancing on behalf of all those cpus. * - * While stopping the tick, this cpu will become the ilb owner if there - * is no other owner. And will be the owner till that cpu becomes busy - * or if all cpus in the system stop their ticks at which point - * there is no need for ilb owner. + * When the ilb owner becomes busy, we will not have new ilb owner until some + * idle CPU wakes up and goes back to idle or some busy CPU tries to kick + * idle load balancing by kicking one of the idle CPUs. * - * When the ilb owner becomes busy, it nominates another owner, during the - * next busy scheduler_tick() + * Ticks are stopped for the ilb owner as well, with busy CPU kicking this + * ilb owner CPU in future (when there is a need for idle load balancing on + * behalf of all idle CPUs). */ -int select_nohz_load_balancer(int stop_tick) +void select_nohz_load_balancer(int stop_tick) { int cpu = smp_processor_id(); if (stop_tick) { - cpu_rq(cpu)->in_nohz_recently = 1; - if (!cpu_active(cpu)) { if (atomic_read(&nohz.load_balancer) != cpu) - return 0; + return; /* * If we are going offline and still the leader, * give up! */ - if (atomic_cmpxchg(&nohz.load_balancer, cpu, -1) != cpu) + if (atomic_cmpxchg(&nohz.load_balancer, cpu, + nr_cpu_ids) != cpu) BUG(); - return 0; + return; } - cpumask_set_cpu(cpu, nohz.cpu_mask); + cpumask_set_cpu(cpu, nohz.idle_cpus_mask); - /* time for ilb owner also to sleep */ - if (cpumask_weight(nohz.cpu_mask) == num_active_cpus()) { - if (atomic_read(&nohz.load_balancer) == cpu) - atomic_set(&nohz.load_balancer, -1); - return 0; - } + if (atomic_read(&nohz.first_pick_cpu) == cpu) + atomic_cmpxchg(&nohz.first_pick_cpu, cpu, nr_cpu_ids); + if (atomic_read(&nohz.second_pick_cpu) == cpu) + atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids); - if (atomic_read(&nohz.load_balancer) == -1) { - /* make me the ilb owner */ - if (atomic_cmpxchg(&nohz.load_balancer, -1, cpu) == -1) - return 1; - } else if (atomic_read(&nohz.load_balancer) == cpu) { + if (atomic_read(&nohz.load_balancer) >= nr_cpu_ids) { int new_ilb; - if (!(sched_smt_power_savings || - sched_mc_power_savings)) - return 1; + /* make me the ilb owner */ + if (atomic_cmpxchg(&nohz.load_balancer, nr_cpu_ids, + cpu) != nr_cpu_ids) + return; + /* * Check to see if there is a more power-efficient * ilb. */ new_ilb = find_new_ilb(cpu); if (new_ilb < nr_cpu_ids && new_ilb != cpu) { - atomic_set(&nohz.load_balancer, -1); + atomic_set(&nohz.load_balancer, nr_cpu_ids); resched_cpu(new_ilb); - return 0; + return; } - return 1; + return; } } else { - if (!cpumask_test_cpu(cpu, nohz.cpu_mask)) - return 0; + if (!cpumask_test_cpu(cpu, nohz.idle_cpus_mask)) + return; - cpumask_clear_cpu(cpu, nohz.cpu_mask); + cpumask_clear_cpu(cpu, nohz.idle_cpus_mask); if (atomic_read(&nohz.load_balancer) == cpu) - if (atomic_cmpxchg(&nohz.load_balancer, cpu, -1) != cpu) + if (atomic_cmpxchg(&nohz.load_balancer, cpu, + nr_cpu_ids) != cpu) BUG(); } - return 0; + return; } #endif @@ -3383,11 +3428,101 @@ out: rq->next_balance = next_balance; } +#ifdef CONFIG_NO_HZ /* - * run_rebalance_domains is triggered when needed from the scheduler tick. - * In CONFIG_NO_HZ case, the idle load balance owner will do the + * In CONFIG_NO_HZ case, the idle balance kickee will do the * rebalancing for all the cpus for whom scheduler ticks are stopped. */ +static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle) +{ + struct rq *this_rq = cpu_rq(this_cpu); + struct rq *rq; + int balance_cpu; + + if (idle != CPU_IDLE || !this_rq->nohz_balance_kick) + return; + + for_each_cpu(balance_cpu, nohz.idle_cpus_mask) { + if (balance_cpu == this_cpu) + continue; + + /* + * If this cpu gets work to do, stop the load balancing + * work being done for other cpus. Next load + * balancing owner will pick it up. + */ + if (need_resched()) { + this_rq->nohz_balance_kick = 0; + break; + } + + raw_spin_lock_irq(&this_rq->lock); + update_cpu_load(this_rq); + raw_spin_unlock_irq(&this_rq->lock); + + rebalance_domains(balance_cpu, CPU_IDLE); + + rq = cpu_rq(balance_cpu); + if (time_after(this_rq->next_balance, rq->next_balance)) + this_rq->next_balance = rq->next_balance; + } + nohz.next_balance = this_rq->next_balance; + this_rq->nohz_balance_kick = 0; +} + +/* + * Current heuristic for kicking the idle load balancer + * - first_pick_cpu is the one of the busy CPUs. It will kick + * idle load balancer when it has more than one process active. This + * eliminates the need for idle load balancing altogether when we have + * only one running process in the system (common case). + * - If there are more than one busy CPU, idle load balancer may have + * to run for active_load_balance to happen (i.e., two busy CPUs are + * SMT or core siblings and can run better if they move to different + * physical CPUs). So, second_pick_cpu is the second of the busy CPUs + * which will kick idle load balancer as soon as it has any load. + */ +static inline int nohz_kick_needed(struct rq *rq, int cpu) +{ + unsigned long now = jiffies; + int ret; + int first_pick_cpu, second_pick_cpu; + + if (time_before(now, nohz.next_balance)) + return 0; + + if (!rq->nr_running) + return 0; + + first_pick_cpu = atomic_read(&nohz.first_pick_cpu); + second_pick_cpu = atomic_read(&nohz.second_pick_cpu); + + if (first_pick_cpu < nr_cpu_ids && first_pick_cpu != cpu && + second_pick_cpu < nr_cpu_ids && second_pick_cpu != cpu) + return 0; + + ret = atomic_cmpxchg(&nohz.first_pick_cpu, nr_cpu_ids, cpu); + if (ret == nr_cpu_ids || ret == cpu) { + atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids); + if (rq->nr_running > 1) + return 1; + } else { + ret = atomic_cmpxchg(&nohz.second_pick_cpu, nr_cpu_ids, cpu); + if (ret == nr_cpu_ids || ret == cpu) { + if (rq->nr_running) + return 1; + } + } + return 0; +} +#else +static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle) { } +#endif + +/* + * run_rebalance_domains is triggered when needed from the scheduler tick. + * Also triggered for nohz idle balancing (with nohz_balancing_kick set). + */ static void run_rebalance_domains(struct softirq_action *h) { int this_cpu = smp_processor_id(); @@ -3397,40 +3532,12 @@ static void run_rebalance_domains(struct softirq_action *h) rebalance_domains(this_cpu, idle); -#ifdef CONFIG_NO_HZ /* - * If this cpu is the owner for idle load balancing, then do the + * If this cpu has a pending nohz_balance_kick, then do the * balancing on behalf of the other idle cpus whose ticks are * stopped. */ - if (this_rq->idle_at_tick && - atomic_read(&nohz.load_balancer) == this_cpu) { - struct rq *rq; - int balance_cpu; - - for_each_cpu(balance_cpu, nohz.cpu_mask) { - if (balance_cpu == this_cpu) - continue; - - /* - * If this cpu gets work to do, stop the load balancing - * work being done for other cpus. Next load - * balancing owner will pick it up. - */ - if (need_resched()) - break; - - rq = cpu_rq(balance_cpu); - raw_spin_lock_irq(&rq->lock); - update_cpu_load(rq); - raw_spin_unlock_irq(&rq->lock); - rebalance_domains(balance_cpu, CPU_IDLE); - - if (time_after(this_rq->next_balance, rq->next_balance)) - this_rq->next_balance = rq->next_balance; - } - } -#endif + nohz_idle_balance(this_cpu, idle); } static inline int on_null_domain(int cpu) @@ -3440,57 +3547,17 @@ static inline int on_null_domain(int cpu) /* * Trigger the SCHED_SOFTIRQ if it is time to do periodic load balancing. - * - * In case of CONFIG_NO_HZ, this is the place where we nominate a new - * idle load balancing owner or decide to stop the periodic load balancing, - * if the whole system is idle. */ static inline void trigger_load_balance(struct rq *rq, int cpu) { -#ifdef CONFIG_NO_HZ - /* - * If we were in the nohz mode recently and busy at the current - * scheduler tick, then check if we need to nominate new idle - * load balancer. - */ - if (rq->in_nohz_recently && !rq->idle_at_tick) { - rq->in_nohz_recently = 0; - - if (atomic_read(&nohz.load_balancer) == cpu) { - cpumask_clear_cpu(cpu, nohz.cpu_mask); - atomic_set(&nohz.load_balancer, -1); - } - - if (atomic_read(&nohz.load_balancer) == -1) { - int ilb = find_new_ilb(cpu); - - if (ilb < nr_cpu_ids) - resched_cpu(ilb); - } - } - - /* - * If this cpu is idle and doing idle load balancing for all the - * cpus with ticks stopped, is it time for that to stop? - */ - if (rq->idle_at_tick && atomic_read(&nohz.load_balancer) == cpu && - cpumask_weight(nohz.cpu_mask) == num_online_cpus()) { - resched_cpu(cpu); - return; - } - - /* - * If this cpu is idle and the idle load balancing is done by - * someone else, then no need raise the SCHED_SOFTIRQ - */ - if (rq->idle_at_tick && atomic_read(&nohz.load_balancer) != cpu && - cpumask_test_cpu(cpu, nohz.cpu_mask)) - return; -#endif /* Don't need to rebalance while attached to NULL domain */ if (time_after_eq(jiffies, rq->next_balance) && likely(!on_null_domain(cpu))) raise_softirq(SCHED_SOFTIRQ); +#ifdef CONFIG_NO_HZ + else if (nohz_kick_needed(rq, cpu) && likely(!on_null_domain(cpu))) + nohz_balancer_kick(cpu); +#endif } static void rq_online_fair(struct rq *rq) diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index 1d7b9bc1c034..5f171f04ab00 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -408,13 +408,7 @@ void tick_nohz_stop_sched_tick(int inidle) * the scheduler tick in nohz_restart_sched_tick. */ if (!ts->tick_stopped) { - if (select_nohz_load_balancer(1)) { - /* - * sched tick not stopped! - */ - cpumask_clear_cpu(cpu, nohz_cpu_mask); - goto out; - } + select_nohz_load_balancer(1); ts->idle_tick = hrtimer_get_expires(&ts->sched_timer); ts->tick_stopped = 1; diff --git a/kernel/timer.c b/kernel/timer.c index ee305c8d4e18..48d6aec0789c 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -679,12 +679,8 @@ __mod_timer(struct timer_list *timer, unsigned long expires, cpu = smp_processor_id(); #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) - if (!pinned && get_sysctl_timer_migration() && idle_cpu(cpu)) { - int preferred_cpu = get_nohz_load_balancer(); - - if (preferred_cpu >= 0) - cpu = preferred_cpu; - } + if (!pinned && get_sysctl_timer_migration() && idle_cpu(cpu)) + cpu = get_nohz_timer_target(); #endif new_base = per_cpu(tvec_bases, cpu); -- cgit v1.2.3 From 9d5efe05eb0c904545a28b19c18b949f23334de0 Mon Sep 17 00:00:00 2001 From: Srivatsa Vaddagiri Date: Tue, 8 Jun 2010 14:57:02 +1000 Subject: sched: Fix capacity calculations for SMT4 Handle cpu capacity being reported as 0 on cores with more number of hardware threads. For example on a Power7 core with 4 hardware threads, core power is 1177 and thus power of each hardware thread is 1177/4 = 294. This low power can lead to capacity for each hardware thread being calculated as 0, which leads to tasks bouncing within the core madly! Fix this by reporting capacity for hardware threads as 1, provided their power is not scaled down significantly because of frequency scaling or real-time tasks usage of cpu. Signed-off-by: Srivatsa Vaddagiri Signed-off-by: Michael Neuling Signed-off-by: Peter Zijlstra Cc: Arjan van de Ven LKML-Reference: <20100608045702.21D03CC895@localhost.localdomain> Signed-off-by: Ingo Molnar --- include/linux/sched.h | 2 +- kernel/sched_fair.c | 53 +++++++++++++++++++++++++++++++++++++++++---------- 2 files changed, 44 insertions(+), 11 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index a3e5b1cd0438..c731296e5e93 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -857,7 +857,7 @@ struct sched_group { * CPU power of this group, SCHED_LOAD_SCALE being max power for a * single CPU. */ - unsigned int cpu_power; + unsigned int cpu_power, cpu_power_orig; /* * The CPUs this group covers. diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 6ee2e0af665b..b9b3462483b7 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2285,13 +2285,6 @@ static void update_cpu_power(struct sched_domain *sd, int cpu) unsigned long power = SCHED_LOAD_SCALE; struct sched_group *sdg = sd->groups; - if (sched_feat(ARCH_POWER)) - power *= arch_scale_freq_power(sd, cpu); - else - power *= default_scale_freq_power(sd, cpu); - - power >>= SCHED_LOAD_SHIFT; - if ((sd->flags & SD_SHARE_CPUPOWER) && weight > 1) { if (sched_feat(ARCH_POWER)) power *= arch_scale_smt_power(sd, cpu); @@ -2301,6 +2294,15 @@ static void update_cpu_power(struct sched_domain *sd, int cpu) power >>= SCHED_LOAD_SHIFT; } + sdg->cpu_power_orig = power; + + if (sched_feat(ARCH_POWER)) + power *= arch_scale_freq_power(sd, cpu); + else + power *= default_scale_freq_power(sd, cpu); + + power >>= SCHED_LOAD_SHIFT; + power *= scale_rt_power(cpu); power >>= SCHED_LOAD_SHIFT; @@ -2333,6 +2335,31 @@ static void update_group_power(struct sched_domain *sd, int cpu) sdg->cpu_power = power; } +/* + * Try and fix up capacity for tiny siblings, this is needed when + * things like SD_ASYM_PACKING need f_b_g to select another sibling + * which on its own isn't powerful enough. + * + * See update_sd_pick_busiest() and check_asym_packing(). + */ +static inline int +fix_small_capacity(struct sched_domain *sd, struct sched_group *group) +{ + /* + * Only siblings can have significantly less than SCHED_LOAD_SCALE + */ + if (sd->level != SD_LV_SIBLING) + return 0; + + /* + * If ~90% of the cpu_power is still there, we're good. + */ + if (group->cpu_power * 32 < group->cpu_power_orig * 29) + return 1; + + return 0; +} + /** * update_sg_lb_stats - Update sched_group's statistics for load balancing. * @sd: The sched_domain whose statistics are to be updated. @@ -2426,6 +2453,8 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, sgs->group_capacity = DIV_ROUND_CLOSEST(group->cpu_power, SCHED_LOAD_SCALE); + if (!sgs->group_capacity) + sgs->group_capacity = fix_small_capacity(sd, group); } /** @@ -2724,8 +2753,9 @@ ret: * find_busiest_queue - find the busiest runqueue among the cpus in group. */ static struct rq * -find_busiest_queue(struct sched_group *group, enum cpu_idle_type idle, - unsigned long imbalance, const struct cpumask *cpus) +find_busiest_queue(struct sched_domain *sd, struct sched_group *group, + enum cpu_idle_type idle, unsigned long imbalance, + const struct cpumask *cpus) { struct rq *busiest = NULL, *rq; unsigned long max_load = 0; @@ -2736,6 +2766,9 @@ find_busiest_queue(struct sched_group *group, enum cpu_idle_type idle, unsigned long capacity = DIV_ROUND_CLOSEST(power, SCHED_LOAD_SCALE); unsigned long wl; + if (!capacity) + capacity = fix_small_capacity(sd, group); + if (!cpumask_test_cpu(i, cpus)) continue; @@ -2852,7 +2885,7 @@ redo: goto out_balanced; } - busiest = find_busiest_queue(group, idle, imbalance, cpus); + busiest = find_busiest_queue(sd, group, idle, imbalance, cpus); if (!busiest) { schedstat_inc(sd, lb_nobusyq[idle]); goto out_balanced; -- cgit v1.2.3 From 532cb4c401e225b084c14d6bd6a2f8ee561de2f1 Mon Sep 17 00:00:00 2001 From: Michael Neuling Date: Tue, 8 Jun 2010 14:57:02 +1000 Subject: sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling Signed-off-by: Srivatsa Vaddagiri Signed-off-by: Peter Zijlstra Cc: Arjan van de Ven Cc: "H. Peter Anvin" Cc: Thomas Gleixner LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar --- include/linux/sched.h | 4 +- include/linux/topology.h | 1 + kernel/sched_fair.c | 139 +++++++++++++++++++++++++++++++++++++++++------ 3 files changed, 126 insertions(+), 18 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index c731296e5e93..ff154e10752b 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -801,7 +801,7 @@ enum cpu_idle_type { #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */ #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */ #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */ - +#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */ #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */ enum powersavings_balance_level { @@ -836,6 +836,8 @@ static inline int sd_balance_for_package_power(void) return SD_PREFER_SIBLING; } +extern int __weak arch_sd_sibiling_asym_packing(void); + /* * Optimise SD flags for power savings: * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings. diff --git a/include/linux/topology.h b/include/linux/topology.h index c44df50a05ab..cf57f30d0dcb 100644 --- a/include/linux/topology.h +++ b/include/linux/topology.h @@ -103,6 +103,7 @@ int arch_update_cpu_topology(void); | 1*SD_SHARE_PKG_RESOURCES \ | 0*SD_SERIALIZE \ | 0*SD_PREFER_SIBLING \ + | arch_sd_sibiling_asym_packing() \ , \ .last_balance = jiffies, \ .balance_interval = 1, \ diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index b9b3462483b7..593424f91a8a 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2457,12 +2457,54 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, sgs->group_capacity = fix_small_capacity(sd, group); } +/** + * update_sd_pick_busiest - return 1 on busiest group + * @sd: sched_domain whose statistics are to be checked + * @sds: sched_domain statistics + * @sg: sched_group candidate to be checked for being the busiest + * @sds: sched_group statistics + * + * Determine if @sg is a busier group than the previously selected + * busiest group. + */ +static bool update_sd_pick_busiest(struct sched_domain *sd, + struct sd_lb_stats *sds, + struct sched_group *sg, + struct sg_lb_stats *sgs, + int this_cpu) +{ + if (sgs->avg_load <= sds->max_load) + return false; + + if (sgs->sum_nr_running > sgs->group_capacity) + return true; + + if (sgs->group_imb) + return true; + + /* + * ASYM_PACKING needs to move all the work to the lowest + * numbered CPUs in the group, therefore mark all groups + * higher than ourself as busy. + */ + if ((sd->flags & SD_ASYM_PACKING) && sgs->sum_nr_running && + this_cpu < group_first_cpu(sg)) { + if (!sds->busiest) + return true; + + if (group_first_cpu(sds->busiest) > group_first_cpu(sg)) + return true; + } + + return false; +} + /** * update_sd_lb_stats - Update sched_group's statistics for load balancing. * @sd: sched_domain whose statistics are to be updated. * @this_cpu: Cpu for which load balance is currently performed. * @idle: Idle status of this_cpu - * @sd_idle: Idle status of the sched_domain containing group. + * @sd_idle: Idle status of the sched_domain containing sg. * @cpus: Set of cpus considered for load balancing. * @balance: Should we balance. * @sds: variable to hold the statistics for this sched_domain. @@ -2473,7 +2515,7 @@ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, struct sd_lb_stats *sds) { struct sched_domain *child = sd->child; - struct sched_group *group = sd->groups; + struct sched_group *sg = sd->groups; struct sg_lb_stats sgs; int load_idx, prefer_sibling = 0; @@ -2486,21 +2528,20 @@ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, do { int local_group; - local_group = cpumask_test_cpu(this_cpu, - sched_group_cpus(group)); + local_group = cpumask_test_cpu(this_cpu, sched_group_cpus(sg)); memset(&sgs, 0, sizeof(sgs)); - update_sg_lb_stats(sd, group, this_cpu, idle, load_idx, sd_idle, + update_sg_lb_stats(sd, sg, this_cpu, idle, load_idx, sd_idle, local_group, cpus, balance, &sgs); if (local_group && !(*balance)) return; sds->total_load += sgs.group_load; - sds->total_pwr += group->cpu_power; + sds->total_pwr += sg->cpu_power; /* * In case the child domain prefers tasks go to siblings - * first, lower the group capacity to one so that we'll try + * first, lower the sg capacity to one so that we'll try * and move all the excess tasks away. */ if (prefer_sibling) @@ -2508,23 +2549,72 @@ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, if (local_group) { sds->this_load = sgs.avg_load; - sds->this = group; + sds->this = sg; sds->this_nr_running = sgs.sum_nr_running; sds->this_load_per_task = sgs.sum_weighted_load; - } else if (sgs.avg_load > sds->max_load && - (sgs.sum_nr_running > sgs.group_capacity || - sgs.group_imb)) { + } else if (update_sd_pick_busiest(sd, sds, sg, &sgs, this_cpu)) { sds->max_load = sgs.avg_load; - sds->busiest = group; + sds->busiest = sg; sds->busiest_nr_running = sgs.sum_nr_running; sds->busiest_group_capacity = sgs.group_capacity; sds->busiest_load_per_task = sgs.sum_weighted_load; sds->group_imb = sgs.group_imb; } - update_sd_power_savings_stats(group, sds, local_group, &sgs); - group = group->next; - } while (group != sd->groups); + update_sd_power_savings_stats(sg, sds, local_group, &sgs); + sg = sg->next; + } while (sg != sd->groups); +} + +int __weak arch_sd_sibiling_asym_packing(void) +{ + return 0*SD_ASYM_PACKING; +} + +/** + * check_asym_packing - Check to see if the group is packed into the + * sched doman. + * + * This is primarily intended to used at the sibling level. Some + * cores like POWER7 prefer to use lower numbered SMT threads. In the + * case of POWER7, it can move to lower SMT modes only when higher + * threads are idle. When in lower SMT modes, the threads will + * perform better since they share less core resources. Hence when we + * have idle threads, we want them to be the higher ones. + * + * This packing function is run on idle threads. It checks to see if + * the busiest CPU in this domain (core in the P7 case) has a higher + * CPU number than the packing function is being run on. Here we are + * assuming lower CPU number will be equivalent to lower a SMT thread + * number. + * + * @sd: The sched_domain whose packing is to be checked. + * @sds: Statistics of the sched_domain which is to be packed + * @this_cpu: The cpu at whose sched_domain we're performing load-balance. + * @imbalance: returns amount of imbalanced due to packing. + * + * Returns 1 when packing is required and a task should be moved to + * this CPU. The amount of the imbalance is returned in *imbalance. + */ +static int check_asym_packing(struct sched_domain *sd, + struct sd_lb_stats *sds, + int this_cpu, unsigned long *imbalance) +{ + int busiest_cpu; + + if (!(sd->flags & SD_ASYM_PACKING)) + return 0; + + if (!sds->busiest) + return 0; + + busiest_cpu = group_first_cpu(sds->busiest); + if (this_cpu > busiest_cpu) + return 0; + + *imbalance = DIV_ROUND_CLOSEST(sds->max_load * sds->busiest->cpu_power, + SCHED_LOAD_SCALE); + return 1; } /** @@ -2719,6 +2809,10 @@ find_busiest_group(struct sched_domain *sd, int this_cpu, if (!(*balance)) goto ret; + if ((idle == CPU_IDLE || idle == CPU_NEWLY_IDLE) && + check_asym_packing(sd, &sds, this_cpu, imbalance)) + return sds.busiest; + if (!sds.busiest || sds.busiest_nr_running == 0) goto out_balanced; @@ -2808,9 +2902,19 @@ find_busiest_queue(struct sched_domain *sd, struct sched_group *group, /* Working cpumask for load_balance and load_balance_newidle. */ static DEFINE_PER_CPU(cpumask_var_t, load_balance_tmpmask); -static int need_active_balance(struct sched_domain *sd, int sd_idle, int idle) +static int need_active_balance(struct sched_domain *sd, int sd_idle, int idle, + int busiest_cpu, int this_cpu) { if (idle == CPU_NEWLY_IDLE) { + + /* + * ASYM_PACKING needs to force migrate tasks from busy but + * higher numbered CPUs in order to pack all tasks in the + * lowest numbered CPUs. + */ + if ((sd->flags & SD_ASYM_PACKING) && busiest_cpu > this_cpu) + return 1; + /* * The only task running in a non-idle cpu can be moved to this * cpu in an attempt to completely freeup the other CPU @@ -2929,7 +3033,8 @@ redo: schedstat_inc(sd, lb_failed[idle]); sd->nr_balance_failed++; - if (need_active_balance(sd, sd_idle, idle)) { + if (need_active_balance(sd, sd_idle, idle, cpu_of(busiest), + this_cpu)) { raw_spin_lock_irqsave(&busiest->lock, flags); /* don't kick the active_load_balance_cpu_stop, -- cgit v1.2.3 From 396e894d289d69bacf5acd983c97cd6e21a14c08 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 9 Jul 2010 15:12:27 +0200 Subject: sched: Revert nohz_ratelimit() for now Norbert reported that nohz_ratelimit() causes his laptop to burn about 4W (40%) extra. For now back out the change and see if we can adjust the power management code to make better decisions. Reported-by: Norbert Preining Signed-off-by: Peter Zijlstra Acked-by: Mike Galbraith Cc: Arjan van de Ven LKML-Reference: Signed-off-by: Ingo Molnar --- include/linux/sched.h | 6 ------ kernel/sched.c | 10 ---------- kernel/time/tick-sched.c | 2 +- 3 files changed, 1 insertion(+), 17 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 747fcaedddb7..6e0bb86de990 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -273,17 +273,11 @@ extern cpumask_var_t nohz_cpu_mask; #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) extern int select_nohz_load_balancer(int cpu); extern int get_nohz_load_balancer(void); -extern int nohz_ratelimit(int cpu); #else static inline int select_nohz_load_balancer(int cpu) { return 0; } - -static inline int nohz_ratelimit(int cpu) -{ - return 0; -} #endif /* diff --git a/kernel/sched.c b/kernel/sched.c index f52a8801b7a2..63b4a14682fa 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1232,16 +1232,6 @@ void wake_up_idle_cpu(int cpu) smp_send_reschedule(cpu); } -int nohz_ratelimit(int cpu) -{ - struct rq *rq = cpu_rq(cpu); - u64 diff = rq->clock - rq->nohz_stamp; - - rq->nohz_stamp = rq->clock; - - return diff < (NSEC_PER_SEC / HZ) >> 1; -} - #endif /* CONFIG_NO_HZ */ static u64 sched_avg_period(void) diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index 813993b5fb61..f898af608171 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -325,7 +325,7 @@ void tick_nohz_stop_sched_tick(int inidle) } while (read_seqretry(&xtime_lock, seq)); if (rcu_needs_cpu(cpu) || printk_needs_cpu(cpu) || - arch_needs_cpu(cpu) || nohz_ratelimit(cpu)) { + arch_needs_cpu(cpu)) { next_jiffies = last_jiffies + 1; delta_jiffies = 1; } else { -- cgit v1.2.3