From 018d6db4cb5bbdcd65424a16f2dcca692ed32ae4 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 14 Apr 2008 08:53:32 +0200 Subject: sched: re-do "sched: fix fair sleepers" re-apply: | commit e22ecef1d2658ba54ed7d3fdb5d60829fb434c23 | Author: Ingo Molnar | Date: Fri Mar 14 22:16:08 2008 +0100 | | sched: fix fair sleepers | | Fair sleepers need to scale their latency target down by runqueue | weight. Otherwise busy systems will gain ever larger sleep bonus. Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 0080968d3e4a..86a93376282c 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -510,8 +510,10 @@ place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial) if (!initial) { /* sleeps upto a single latency don't count. */ - if (sched_feat(NEW_FAIR_SLEEPERS)) - vruntime -= sysctl_sched_latency; + if (sched_feat(NEW_FAIR_SLEEPERS)) { + vruntime -= calc_delta_fair(sysctl_sched_latency, + &cfs_rq->load); + } /* ensure we never gain time by being placed backwards. */ vruntime = max_vruntime(se->vruntime, vruntime); -- cgit v1.2.3 From 27ec4407790d075c325e1f4da0a19c56953cce23 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 28 Feb 2008 21:00:21 +0100 Subject: sched: make cpu_clock() globally synchronous Alexey Zaytsev reported (and bisected) that the introduction of cpu_clock() in printk made the timestamps jump back and forth. Make cpu_clock() more reliable while still keeping it fast when it's called frequently. Signed-off-by: Ingo Molnar --- kernel/sched.c | 52 +++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 49 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 8dcdec6fe0fe..7377222ab42f 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -632,11 +632,39 @@ int sysctl_sched_rt_runtime = 950000; */ #define RUNTIME_INF ((u64)~0ULL) +static const unsigned long long time_sync_thresh = 100000; + +static DEFINE_PER_CPU(unsigned long long, time_offset); +static DEFINE_PER_CPU(unsigned long long, prev_cpu_time); + /* - * For kernel-internal use: high-speed (but slightly incorrect) per-cpu - * clock constructed from sched_clock(): + * Global lock which we take every now and then to synchronize + * the CPUs time. This method is not warp-safe, but it's good + * enough to synchronize slowly diverging time sources and thus + * it's good enough for tracing: */ -unsigned long long cpu_clock(int cpu) +static DEFINE_SPINLOCK(time_sync_lock); +static unsigned long long prev_global_time; + +static unsigned long long __sync_cpu_clock(cycles_t time, int cpu) +{ + unsigned long flags; + + spin_lock_irqsave(&time_sync_lock, flags); + + if (time < prev_global_time) { + per_cpu(time_offset, cpu) += prev_global_time - time; + time = prev_global_time; + } else { + prev_global_time = time; + } + + spin_unlock_irqrestore(&time_sync_lock, flags); + + return time; +} + +static unsigned long long __cpu_clock(int cpu) { unsigned long long now; unsigned long flags; @@ -657,6 +685,24 @@ unsigned long long cpu_clock(int cpu) return now; } + +/* + * For kernel-internal use: high-speed (but slightly incorrect) per-cpu + * clock constructed from sched_clock(): + */ +unsigned long long cpu_clock(int cpu) +{ + unsigned long long prev_cpu_time, time, delta_time; + + prev_cpu_time = per_cpu(prev_cpu_time, cpu); + time = __cpu_clock(cpu) + per_cpu(time_offset, cpu); + delta_time = time-prev_cpu_time; + + if (unlikely(delta_time > time_sync_thresh)) + time = __sync_cpu_clock(time, cpu); + + return time; +} EXPORT_SYMBOL_GPL(cpu_clock); #ifndef prepare_arch_switch -- cgit v1.2.3 From 30914a58af9d21c5f1831adabb5d7a800a378675 Mon Sep 17 00:00:00 2001 From: Reynes Philippe Date: Mon, 17 Mar 2008 16:19:05 -0700 Subject: sched: sched.c needs tick.h kernel/sched.c:506: erreur: implicit declaration of function tick_get_tick_sched kernel/sched.c:506: erreur: invalid type argument of -> kernel/sched.c:506: erreur: NOHZ_MODE_INACTIVE undeclared (first use in this function) kernel/sched.c:506: erreur: (Each undeclared identifier is reported only once kernel/sched.c:506: erreur: for each function it appears in.) Signed-off-by: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/sched.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 7377222ab42f..7fe334ead4f9 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -66,6 +66,7 @@ #include #include #include +#include #include #include -- cgit v1.2.3 From 15934a37324f32e0fda633dc7984a671ea81cd75 Mon Sep 17 00:00:00 2001 From: Guillaume Chazarain Date: Sat, 19 Apr 2008 19:44:57 +0200 Subject: sched: fix rq->clock overflows detection with CONFIG_NO_HZ When using CONFIG_NO_HZ, rq->tick_timestamp is not updated every TICK_NSEC. We check that the number of skipped ticks matches the clock jump seen in __update_rq_clock(). Signed-off-by: Guillaume Chazarain Signed-off-by: Ingo Molnar --- kernel/sched.c | 38 +++++++++++++++++++++++++++++++++++--- 1 file changed, 35 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 7fe334ead4f9..d8456a9ac9af 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -397,6 +397,7 @@ struct rq { unsigned long cpu_load[CPU_LOAD_IDX_MAX]; unsigned char idle_at_tick; #ifdef CONFIG_NO_HZ + unsigned long last_tick_seen; unsigned char in_nohz_recently; #endif /* capture load from *all* tasks on this cpu: */ @@ -500,6 +501,32 @@ static inline int cpu_of(struct rq *rq) #endif } +#ifdef CONFIG_NO_HZ +static inline bool nohz_on(int cpu) +{ + return tick_get_tick_sched(cpu)->nohz_mode != NOHZ_MODE_INACTIVE; +} + +static inline u64 max_skipped_ticks(struct rq *rq) +{ + return nohz_on(cpu_of(rq)) ? jiffies - rq->last_tick_seen + 2 : 1; +} + +static inline void update_last_tick_seen(struct rq *rq) +{ + rq->last_tick_seen = jiffies; +} +#else +static inline u64 max_skipped_ticks(struct rq *rq) +{ + return 1; +} + +static inline void update_last_tick_seen(struct rq *rq) +{ +} +#endif + /* * Update the per-runqueue clock, as finegrained as the platform can give * us, but without assuming monotonicity, etc.: @@ -524,9 +551,12 @@ static void __update_rq_clock(struct rq *rq) /* * Catch too large forward jumps too: */ - if (unlikely(clock + delta > rq->tick_timestamp + TICK_NSEC)) { - if (clock < rq->tick_timestamp + TICK_NSEC) - clock = rq->tick_timestamp + TICK_NSEC; + u64 max_jump = max_skipped_ticks(rq) * TICK_NSEC; + u64 max_time = rq->tick_timestamp + max_jump; + + if (unlikely(clock + delta > max_time)) { + if (clock < max_time) + clock = max_time; else clock++; rq->clock_overflows++; @@ -3812,6 +3842,7 @@ void scheduler_tick(void) rq->clock_underflows++; } rq->tick_timestamp = rq->clock; + update_last_tick_seen(rq); update_cpu_load(rq); curr->sched_class->task_tick(rq, curr, 0); update_sched_rt_period(rq); @@ -7261,6 +7292,7 @@ void __init sched_init(void) lockdep_set_class(&rq->lock, &rq->rq_lock_key); rq->nr_running = 0; rq->clock = 1; + update_last_tick_seen(rq); init_cfs_rq(&rq->cfs, rq); init_rt_rq(&rq->rt, rq); #ifdef CONFIG_FAIR_GROUP_SCHED -- cgit v1.2.3 From 0bbd3336eee1e712a290e0dfd1a64cbbdd63a508 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sat, 19 Apr 2008 19:44:57 +0200 Subject: sched: fix wakeup granularity for buddies The wakeup buddy logic didn't use the same wakeup granularity logic as the wakeup preemption did, this might cause the ->next buddy to be selected past the point where we would have preempted had the task been a single running instance. Signed-off-by: Peter Zijlstra Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 69 +++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 49 insertions(+), 20 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 86a93376282c..b01f8e77f2ac 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -73,13 +73,13 @@ unsigned int sysctl_sched_batch_wakeup_granularity = 10000000UL; /* * SCHED_OTHER wake-up granularity. - * (default: 5 msec * (1 + ilog(ncpus)), units: nanoseconds) + * (default: 10 msec * (1 + ilog(ncpus)), units: nanoseconds) * * This option delays the preemption effects of decoupled workloads * and reduces their over-scheduling. Synchronous workloads will still * have immediate wakeup/sleep latencies. */ -unsigned int sysctl_sched_wakeup_granularity = 5000000UL; +unsigned int sysctl_sched_wakeup_granularity = 10000000UL; const_debug unsigned int sysctl_sched_migration_cost = 500000UL; @@ -629,20 +629,16 @@ set_next_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) se->prev_sum_exec_runtime = se->sum_exec_runtime; } +static int +wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se); + static struct sched_entity * pick_next(struct cfs_rq *cfs_rq, struct sched_entity *se) { - s64 diff, gran; - if (!cfs_rq->next) return se; - diff = cfs_rq->next->vruntime - se->vruntime; - if (diff < 0) - return se; - - gran = calc_delta_fair(sysctl_sched_wakeup_granularity, &cfs_rq->load); - if (diff > gran) + if (wakeup_preempt_entity(cfs_rq->next, se) != 0) return se; return cfs_rq->next; @@ -1101,6 +1097,48 @@ out: } #endif /* CONFIG_SMP */ +static unsigned long wakeup_gran(struct sched_entity *se) +{ + unsigned long gran = sysctl_sched_wakeup_granularity; + + /* + * More easily preempt - nice tasks, while not making + * it harder for + nice tasks. + */ + if (unlikely(se->load.weight > NICE_0_LOAD)) + gran = calc_delta_fair(gran, &se->load); + + return gran; +} + +/* + * Should 'se' preempt 'curr'. + * + * |s1 + * |s2 + * |s3 + * g + * |<--->|c + * + * w(c, s1) = -1 + * w(c, s2) = 0 + * w(c, s3) = 1 + * + */ +static int +wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se) +{ + s64 gran, vdiff = curr->vruntime - se->vruntime; + + if (vdiff < 0) + return -1; + + gran = wakeup_gran(curr); + if (vdiff > gran) + return 1; + + return 0; +} /* * Preempt the current task with a newly woken task if needed: @@ -1110,7 +1148,6 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p) struct task_struct *curr = rq->curr; struct cfs_rq *cfs_rq = task_cfs_rq(curr); struct sched_entity *se = &curr->se, *pse = &p->se; - unsigned long gran; if (unlikely(rt_prio(p->prio))) { update_rq_clock(rq); @@ -1140,15 +1177,7 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p) pse = parent_entity(pse); } - gran = sysctl_sched_wakeup_granularity; - /* - * More easily preempt - nice tasks, while not making - * it harder for + nice tasks. - */ - if (unlikely(se->load.weight > NICE_0_LOAD)) - gran = calc_delta_fair(gran, &se->load); - - if (pse->vruntime + gran < se->vruntime) + if (wakeup_preempt_entity(se, pse) == 1) resched_task(curr); } -- cgit v1.2.3 From b85d0667268320072ccdeb07c27c25b300ab3724 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sun, 16 Mar 2008 20:03:22 +0100 Subject: sched: introduce SCHED_FEAT_SYNC_WAKEUPS, turn it off turn off sync wakeups by default. They are not needed anymore - the buddy logic should be smart enough to keep the system from overscheduling. Signed-off-by: Ingo Molnar --- kernel/sched.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index d8456a9ac9af..263e25e10204 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -627,6 +627,7 @@ enum { SCHED_FEAT_START_DEBIT = 4, SCHED_FEAT_HRTICK = 8, SCHED_FEAT_DOUBLE_TICK = 16, + SCHED_FEAT_SYNC_WAKEUPS = 32, }; const_debug unsigned int sysctl_sched_features = @@ -634,7 +635,8 @@ const_debug unsigned int sysctl_sched_features = SCHED_FEAT_WAKEUP_PREEMPT * 1 | SCHED_FEAT_START_DEBIT * 1 | SCHED_FEAT_HRTICK * 1 | - SCHED_FEAT_DOUBLE_TICK * 0; + SCHED_FEAT_DOUBLE_TICK * 0 | + SCHED_FEAT_SYNC_WAKEUPS * 0; #define sched_feat(x) (sysctl_sched_features & SCHED_FEAT_##x) @@ -1916,6 +1918,9 @@ static int try_to_wake_up(struct task_struct *p, unsigned int state, int sync) long old_state; struct rq *rq; + if (!sched_feat(SYNC_WAKEUPS)) + sync = 0; + smp_wmb(); rq = task_rq_lock(p, &flags); old_state = p->state; -- cgit v1.2.3 From 1fc8afa4c820fcde3658238eab5c010476ede521 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 19 Mar 2008 01:39:19 +0100 Subject: sched: feat affine wakeups Signed-off-by: Ingo Molnar --- kernel/sched.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 263e25e10204..7c5efad78c09 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -628,6 +628,7 @@ enum { SCHED_FEAT_HRTICK = 8, SCHED_FEAT_DOUBLE_TICK = 16, SCHED_FEAT_SYNC_WAKEUPS = 32, + SCHED_FEAT_AFFINE_WAKEUPS = 64, }; const_debug unsigned int sysctl_sched_features = @@ -636,7 +637,8 @@ const_debug unsigned int sysctl_sched_features = SCHED_FEAT_START_DEBIT * 1 | SCHED_FEAT_HRTICK * 1 | SCHED_FEAT_DOUBLE_TICK * 0 | - SCHED_FEAT_SYNC_WAKEUPS * 0; + SCHED_FEAT_SYNC_WAKEUPS * 0 | + SCHED_FEAT_AFFINE_WAKEUPS * 1; #define sched_feat(x) (sysctl_sched_features & SCHED_FEAT_##x) -- cgit v1.2.3 From d25ce4cd499a21aab89ff8755f8c4a2800eae25f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 17 Mar 2008 09:36:53 +0100 Subject: sched: cache hot buddy Signed-off-by: Ingo Molnar --- kernel/sched.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 7c5efad78c09..42d2f1155d30 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -625,20 +625,22 @@ enum { SCHED_FEAT_NEW_FAIR_SLEEPERS = 1, SCHED_FEAT_WAKEUP_PREEMPT = 2, SCHED_FEAT_START_DEBIT = 4, - SCHED_FEAT_HRTICK = 8, - SCHED_FEAT_DOUBLE_TICK = 16, - SCHED_FEAT_SYNC_WAKEUPS = 32, - SCHED_FEAT_AFFINE_WAKEUPS = 64, + SCHED_FEAT_AFFINE_WAKEUPS = 8, + SCHED_FEAT_CACHE_HOT_BUDDY = 16, + SCHED_FEAT_HRTICK = 32, + SCHED_FEAT_DOUBLE_TICK = 64, + SCHED_FEAT_SYNC_WAKEUPS = 128, }; const_debug unsigned int sysctl_sched_features = SCHED_FEAT_NEW_FAIR_SLEEPERS * 1 | SCHED_FEAT_WAKEUP_PREEMPT * 1 | SCHED_FEAT_START_DEBIT * 1 | + SCHED_FEAT_AFFINE_WAKEUPS * 1 | + SCHED_FEAT_CACHE_HOT_BUDDY * 1 | SCHED_FEAT_HRTICK * 1 | SCHED_FEAT_DOUBLE_TICK * 0 | - SCHED_FEAT_SYNC_WAKEUPS * 0 | - SCHED_FEAT_AFFINE_WAKEUPS * 1; + SCHED_FEAT_SYNC_WAKEUPS * 0; #define sched_feat(x) (sysctl_sched_features & SCHED_FEAT_##x) @@ -1519,7 +1521,7 @@ task_hot(struct task_struct *p, u64 now, struct sched_domain *sd) /* * Buddy candidates are cache hot: */ - if (&p->se == cfs_rq_of(&p->se)->next) + if (sched_feat(CACHE_HOT_BUDDY) && (&p->se == cfs_rq_of(&p->se)->next)) return 1; if (p->sched_class != &fair_sched_class) -- cgit v1.2.3 From 02e2b83bd25bb05ac2e69cb31458b7d1b3c70707 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 19 Mar 2008 01:37:10 +0100 Subject: sched: reenable sync wakeups Signed-off-by: Ingo Molnar --- kernel/sched.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 42d2f1155d30..770449bee6da 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -627,9 +627,9 @@ enum { SCHED_FEAT_START_DEBIT = 4, SCHED_FEAT_AFFINE_WAKEUPS = 8, SCHED_FEAT_CACHE_HOT_BUDDY = 16, - SCHED_FEAT_HRTICK = 32, - SCHED_FEAT_DOUBLE_TICK = 64, - SCHED_FEAT_SYNC_WAKEUPS = 128, + SCHED_FEAT_SYNC_WAKEUPS = 32, + SCHED_FEAT_HRTICK = 64, + SCHED_FEAT_DOUBLE_TICK = 128, }; const_debug unsigned int sysctl_sched_features = @@ -638,9 +638,9 @@ const_debug unsigned int sysctl_sched_features = SCHED_FEAT_START_DEBIT * 1 | SCHED_FEAT_AFFINE_WAKEUPS * 1 | SCHED_FEAT_CACHE_HOT_BUDDY * 1 | + SCHED_FEAT_SYNC_WAKEUPS * 1 | SCHED_FEAT_HRTICK * 1 | - SCHED_FEAT_DOUBLE_TICK * 0 | - SCHED_FEAT_SYNC_WAKEUPS * 0; + SCHED_FEAT_DOUBLE_TICK * 0; #define sched_feat(x) (sysctl_sched_features & SCHED_FEAT_##x) -- cgit v1.2.3 From 50df5d6aea6694ca481b8005900401e8c95c2603 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 14 Mar 2008 16:09:59 +0100 Subject: sched: remove sysctl_sched_batch_wakeup_granularity it's unused. Signed-off-by: Ingo Molnar --- kernel/sched.c | 1 - kernel/sched_debug.c | 1 - kernel/sched_fair.c | 10 ---------- kernel/sysctl.c | 11 ----------- 4 files changed, 23 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 770449bee6da..e813e845d9cf 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5396,7 +5396,6 @@ static inline void sched_init_granularity(void) sysctl_sched_latency = limit; sysctl_sched_wakeup_granularity *= factor; - sysctl_sched_batch_wakeup_granularity *= factor; } #ifdef CONFIG_SMP diff --git a/kernel/sched_debug.c b/kernel/sched_debug.c index ef358ba07683..3d09106990cb 100644 --- a/kernel/sched_debug.c +++ b/kernel/sched_debug.c @@ -214,7 +214,6 @@ static int sched_debug_show(struct seq_file *m, void *v) PN(sysctl_sched_latency); PN(sysctl_sched_min_granularity); PN(sysctl_sched_wakeup_granularity); - PN(sysctl_sched_batch_wakeup_granularity); PN(sysctl_sched_child_runs_first); P(sysctl_sched_features); #undef PN diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index b01f8e77f2ac..bedda18f37a5 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -61,16 +61,6 @@ const_debug unsigned int sysctl_sched_child_runs_first = 1; */ unsigned int __read_mostly sysctl_sched_compat_yield; -/* - * SCHED_BATCH wake-up granularity. - * (default: 10 msec * (1 + ilog(ncpus)), units: nanoseconds) - * - * This option delays the preemption effects of decoupled workloads - * and reduces their over-scheduling. Synchronous workloads will still - * have immediate wakeup/sleep latencies. - */ -unsigned int sysctl_sched_batch_wakeup_granularity = 10000000UL; - /* * SCHED_OTHER wake-up granularity. * (default: 10 msec * (1 + ilog(ncpus)), units: nanoseconds) diff --git a/kernel/sysctl.c b/kernel/sysctl.c index b2a2d6889bab..be332e1a0c29 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -268,17 +268,6 @@ static struct ctl_table kern_table[] = { .extra1 = &min_wakeup_granularity_ns, .extra2 = &max_wakeup_granularity_ns, }, - { - .ctl_name = CTL_UNNUMBERED, - .procname = "sched_batch_wakeup_granularity_ns", - .data = &sysctl_sched_batch_wakeup_granularity, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = &proc_dointvec_minmax, - .strategy = &sysctl_intvec, - .extra1 = &min_wakeup_granularity_ns, - .extra2 = &max_wakeup_granularity_ns, - }, { .ctl_name = CTL_UNNUMBERED, .procname = "sched_child_runs_first", -- cgit v1.2.3 From 19fb518c2a0c5d88ed22bba7083b7e7bc2a9c231 Mon Sep 17 00:00:00 2001 From: Dmitry Adamushko Date: Sun, 17 Feb 2008 22:34:07 +0100 Subject: latencytop: optimize LT_BACKTRACEDEPTH loops a bit There is no need to loop any longer when 'same == 0'. Signed-off-by: Dmitry Adamushko Signed-off-by: Ingo Molnar --- kernel/latencytop.c | 27 +++++++++++++++++---------- 1 file changed, 17 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/latencytop.c b/kernel/latencytop.c index b4e3c85abe74..7c74dab0d21b 100644 --- a/kernel/latencytop.c +++ b/kernel/latencytop.c @@ -64,8 +64,8 @@ account_global_scheduler_latency(struct task_struct *tsk, struct latency_record return; for (i = 0; i < MAXLR; i++) { - int q; - int same = 1; + int q, same = 1; + /* Nothing stored: */ if (!latency_record[i].backtrace[0]) { if (firstnonnull > i) @@ -73,12 +73,15 @@ account_global_scheduler_latency(struct task_struct *tsk, struct latency_record continue; } for (q = 0 ; q < LT_BACKTRACEDEPTH ; q++) { - if (latency_record[i].backtrace[q] != - lat->backtrace[q]) + unsigned long record = lat->backtrace[q]; + + if (latency_record[i].backtrace[q] != record) { same = 0; - if (same && lat->backtrace[q] == 0) break; - if (same && lat->backtrace[q] == ULONG_MAX) + } + + /* 0 and ULONG_MAX entries mean end of backtrace: */ + if (record == 0 || record == ULONG_MAX) break; } if (same) { @@ -143,14 +146,18 @@ account_scheduler_latency(struct task_struct *tsk, int usecs, int inter) for (i = 0; i < LT_SAVECOUNT ; i++) { struct latency_record *mylat; int same = 1; + mylat = &tsk->latency_record[i]; for (q = 0 ; q < LT_BACKTRACEDEPTH ; q++) { - if (mylat->backtrace[q] != - lat.backtrace[q]) + unsigned long record = lat.backtrace[q]; + + if (mylat->backtrace[q] != record) { same = 0; - if (same && lat.backtrace[q] == 0) break; - if (same && lat.backtrace[q] == ULONG_MAX) + } + + /* 0 and ULONG_MAX entries mean end of backtrace: */ + if (record == 0 || record == ULONG_MAX) break; } if (same) { -- cgit v1.2.3 From 79b3feffb10417f197d2ab48dd4fa3c0c9e7d788 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 18 Feb 2008 13:39:37 +0100 Subject: sched: fix regression with sched yield Balbir Singh reported: > 1:mon> t > [c0000000e7677da0] c000000000067de0 .sys_sched_yield+0x6c/0xbc > [c0000000e7677e30] c000000000008748 syscall_exit+0x0/0x40 > --- Exception: c01 (System Call) at 00000400001d09e4 > SP (4000664cb10) is in userspace > 1:mon> r > cpu 0x1: Vector: 300 (Data Access) at [c0000000e7677aa0] > pc: c000000000068e50: .yield_task_fair+0x94/0xc4 > lr: c000000000067de0: .sys_sched_yield+0x6c/0xbc the check that should have avoided that is: /* * Are we the only task in the tree? */ if (unlikely(rq->load.weight == curr->se.load.weight)) return; But I guess that overlooks rt tasks, they also increase the load. So I guess something like this ought to fix it.. Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index bedda18f37a5..290cf770b712 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -904,7 +904,7 @@ static void yield_task_fair(struct rq *rq) /* * Already in the rightmost position? */ - if (unlikely(rightmost->vruntime < se->vruntime)) + if (unlikely(!rightmost || rightmost->vruntime < se->vruntime)) return; /* -- cgit v1.2.3 From d0b27fa77854b149ad4af08b0fe47fe712a47ade Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sat, 19 Apr 2008 19:44:57 +0200 Subject: sched: rt-group: synchonised bandwidth period Various SMP balancing algorithms require that the bandwidth period run in sync. Possible improvements are moving the rt_bandwidth thing into root_domain and keeping a span per rt_bandwidth which marks throttled cpus. Signed-off-by: Peter Zijlstra Signed-off-by: Ingo Molnar --- kernel/sched.c | 260 ++++++++++++++++++++++++++++++++++++++--------- kernel/sched_rt.c | 104 +++++++++++++------ kernel/sysctl.c | 4 +- kernel/time/tick-sched.c | 5 - kernel/user.c | 28 +++++ 5 files changed, 313 insertions(+), 88 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index e813e845d9cf..bb20323f7d09 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -115,6 +115,11 @@ unsigned long long __attribute__((weak)) sched_clock(void) */ #define DEF_TIMESLICE (100 * HZ / 1000) +/* + * single value that denotes runtime == period, ie unlimited time. + */ +#define RUNTIME_INF ((u64)~0ULL) + #ifdef CONFIG_SMP /* * Divide a load by a sched group cpu_power : (load / sg->__cpu_power) @@ -156,6 +161,80 @@ struct rt_prio_array { struct list_head queue[MAX_RT_PRIO]; }; +struct rt_bandwidth { + ktime_t rt_period; + u64 rt_runtime; + struct hrtimer rt_period_timer; +}; + +static struct rt_bandwidth def_rt_bandwidth; + +static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun); + +static enum hrtimer_restart sched_rt_period_timer(struct hrtimer *timer) +{ + struct rt_bandwidth *rt_b = + container_of(timer, struct rt_bandwidth, rt_period_timer); + ktime_t now; + int overrun; + int idle = 0; + + for (;;) { + now = hrtimer_cb_get_time(timer); + overrun = hrtimer_forward(timer, now, rt_b->rt_period); + + if (!overrun) + break; + + idle = do_sched_rt_period_timer(rt_b, overrun); + } + + return idle ? HRTIMER_NORESTART : HRTIMER_RESTART; +} + +static +void init_rt_bandwidth(struct rt_bandwidth *rt_b, u64 period, u64 runtime) +{ + rt_b->rt_period = ns_to_ktime(period); + rt_b->rt_runtime = runtime; + + hrtimer_init(&rt_b->rt_period_timer, + CLOCK_MONOTONIC, HRTIMER_MODE_REL); + rt_b->rt_period_timer.function = sched_rt_period_timer; + rt_b->rt_period_timer.cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ; +} + +static void start_rt_bandwidth(struct rt_bandwidth *rt_b) +{ + ktime_t now; + + if (rt_b->rt_runtime == RUNTIME_INF) + return; + + if (hrtimer_active(&rt_b->rt_period_timer)) + return; + + spin_lock(&rt_b->rt_runtime_lock); + for (;;) { + if (hrtimer_active(&rt_b->rt_period_timer)) + break; + + now = hrtimer_cb_get_time(&rt_b->rt_period_timer); + hrtimer_forward(&rt_b->rt_period_timer, now, rt_b->rt_period); + hrtimer_start(&rt_b->rt_period_timer, + rt_b->rt_period_timer.expires, + HRTIMER_MODE_ABS); + } + spin_unlock(&rt_b->rt_runtime_lock); +} + +#ifdef CONFIG_RT_GROUP_SCHED +static void destroy_rt_bandwidth(struct rt_bandwidth *rt_b) +{ + hrtimer_cancel(&rt_b->rt_period_timer); +} +#endif + #ifdef CONFIG_GROUP_SCHED #include @@ -182,7 +261,7 @@ struct task_group { struct sched_rt_entity **rt_se; struct rt_rq **rt_rq; - u64 rt_runtime; + struct rt_bandwidth rt_bandwidth; #endif struct rcu_head rcu; @@ -407,8 +486,6 @@ struct rq { struct cfs_rq cfs; struct rt_rq rt; - u64 rt_period_expire; - int rt_throttled; #ifdef CONFIG_FAIR_GROUP_SCHED /* list of leaf cfs_rq on this cpu: */ @@ -592,23 +669,6 @@ static void update_rq_clock(struct rq *rq) #define task_rq(p) cpu_rq(task_cpu(p)) #define cpu_curr(cpu) (cpu_rq(cpu)->curr) -unsigned long rt_needs_cpu(int cpu) -{ - struct rq *rq = cpu_rq(cpu); - u64 delta; - - if (!rq->rt_throttled) - return 0; - - if (rq->clock > rq->rt_period_expire) - return 1; - - delta = rq->rt_period_expire - rq->clock; - do_div(delta, NSEC_PER_SEC / HZ); - - return (unsigned long)delta; -} - /* * Tunables that become constants when CONFIG_SCHED_DEBUG is off: */ @@ -664,10 +724,18 @@ static __read_mostly int scheduler_running; */ int sysctl_sched_rt_runtime = 950000; -/* - * single value that denotes runtime == period, ie unlimited time. - */ -#define RUNTIME_INF ((u64)~0ULL) +static inline u64 global_rt_period(void) +{ + return (u64)sysctl_sched_rt_period * NSEC_PER_USEC; +} + +static inline u64 global_rt_runtime(void) +{ + if (sysctl_sched_rt_period < 0) + return RUNTIME_INF; + + return (u64)sysctl_sched_rt_runtime * NSEC_PER_USEC; +} static const unsigned long long time_sync_thresh = 100000; @@ -3854,7 +3922,6 @@ void scheduler_tick(void) update_last_tick_seen(rq); update_cpu_load(rq); curr->sched_class->task_tick(rq, curr, 0); - update_sched_rt_period(rq); spin_unlock(&rq->lock); #ifdef CONFIG_SMP @@ -4689,7 +4756,7 @@ recheck: * Do not allow realtime tasks into groups that have no runtime * assigned. */ - if (rt_policy(policy) && task_group(p)->rt_runtime == 0) + if (rt_policy(policy) && task_group(p)->rt_bandwidth.rt_runtime == 0) return -EPERM; #endif @@ -7288,6 +7355,14 @@ void __init sched_init(void) init_defrootdomain(); #endif + init_rt_bandwidth(&def_rt_bandwidth, + global_rt_period(), global_rt_runtime()); + +#ifdef CONFIG_RT_GROUP_SCHED + init_rt_bandwidth(&init_task_group.rt_bandwidth, + global_rt_period(), global_rt_runtime()); +#endif + #ifdef CONFIG_GROUP_SCHED list_add(&init_task_group.list, &task_groups); #endif @@ -7312,15 +7387,11 @@ void __init sched_init(void) #endif #ifdef CONFIG_RT_GROUP_SCHED - init_task_group.rt_runtime = - sysctl_sched_rt_runtime * NSEC_PER_USEC; INIT_LIST_HEAD(&rq->leaf_rt_rq_list); init_tg_rt_entry(rq, &init_task_group, &per_cpu(init_rt_rq, i), &per_cpu(init_sched_rt_entity, i), i, 1); #endif - rq->rt_period_expire = 0; - rq->rt_throttled = 0; for (j = 0; j < CPU_LOAD_IDX_MAX; j++) rq->cpu_load[j] = 0; @@ -7506,8 +7577,6 @@ void set_curr_task(int cpu, struct task_struct *p) #endif -#ifdef CONFIG_GROUP_SCHED - #ifdef CONFIG_FAIR_GROUP_SCHED static void free_fair_sched_group(struct task_group *tg) { @@ -7596,6 +7665,8 @@ static void free_rt_sched_group(struct task_group *tg) { int i; + destroy_rt_bandwidth(&tg->rt_bandwidth); + for_each_possible_cpu(i) { if (tg->rt_rq) kfree(tg->rt_rq[i]); @@ -7621,7 +7692,8 @@ static int alloc_rt_sched_group(struct task_group *tg) if (!tg->rt_se) goto err; - tg->rt_runtime = 0; + init_rt_bandwidth(&tg->rt_bandwidth, + ktime_to_ns(def_rt_bandwidth.rt_period), 0); for_each_possible_cpu(i) { rq = cpu_rq(i); @@ -7674,6 +7746,7 @@ static inline void unregister_rt_sched_group(struct task_group *tg, int cpu) } #endif +#ifdef CONFIG_GROUP_SCHED static void free_sched_group(struct task_group *tg) { free_fair_sched_group(tg); @@ -7775,6 +7848,7 @@ void sched_move_task(struct task_struct *tsk) task_rq_unlock(rq, &flags); } +#endif #ifdef CONFIG_FAIR_GROUP_SCHED static void set_se_shares(struct sched_entity *se, unsigned long shares) @@ -7871,16 +7945,15 @@ static int __rt_schedulable(struct task_group *tg, u64 period, u64 runtime) struct task_group *tgi; unsigned long total = 0; unsigned long global_ratio = - to_ratio(sysctl_sched_rt_period, - sysctl_sched_rt_runtime < 0 ? - RUNTIME_INF : sysctl_sched_rt_runtime); + to_ratio(global_rt_period(), global_rt_runtime()); rcu_read_lock(); list_for_each_entry_rcu(tgi, &task_groups, list) { if (tgi == tg) continue; - total += to_ratio(period, tgi->rt_runtime); + total += to_ratio(ktime_to_ns(tgi->rt_bandwidth.rt_period), + tgi->rt_bandwidth.rt_runtime); } rcu_read_unlock(); @@ -7898,16 +7971,11 @@ static inline int tg_has_rt_tasks(struct task_group *tg) return 0; } -int sched_group_set_rt_runtime(struct task_group *tg, long rt_runtime_us) +static int tg_set_bandwidth(struct task_group *tg, + u64 rt_period, u64 rt_runtime) { - u64 rt_runtime, rt_period; int err = 0; - rt_period = (u64)sysctl_sched_rt_period * NSEC_PER_USEC; - rt_runtime = (u64)rt_runtime_us * NSEC_PER_USEC; - if (rt_runtime_us == -1) - rt_runtime = RUNTIME_INF; - mutex_lock(&rt_constraints_mutex); read_lock(&tasklist_lock); if (rt_runtime_us == 0 && tg_has_rt_tasks(tg)) { @@ -7918,7 +7986,8 @@ int sched_group_set_rt_runtime(struct task_group *tg, long rt_runtime_us) err = -EINVAL; goto unlock; } - tg->rt_runtime = rt_runtime; + tg->rt_bandwidth.rt_period = ns_to_ktime(rt_period); + tg->rt_bandwidth.rt_runtime = rt_runtime; unlock: read_unlock(&tasklist_lock); mutex_unlock(&rt_constraints_mutex); @@ -7926,19 +7995,96 @@ int sched_group_set_rt_runtime(struct task_group *tg, long rt_runtime_us) return err; } +int sched_group_set_rt_runtime(struct task_group *tg, long rt_runtime_us) +{ + u64 rt_runtime, rt_period; + + rt_period = ktime_to_ns(tg->rt_bandwidth.rt_period); + rt_runtime = (u64)rt_runtime_us * NSEC_PER_USEC; + if (rt_runtime_us < 0) + rt_runtime = RUNTIME_INF; + + return tg_set_bandwidth(tg, rt_period, rt_runtime); +} + long sched_group_rt_runtime(struct task_group *tg) { u64 rt_runtime_us; - if (tg->rt_runtime == RUNTIME_INF) + if (tg->rt_bandwidth.rt_runtime == RUNTIME_INF) return -1; - rt_runtime_us = tg->rt_runtime; + rt_runtime_us = tg->rt_bandwidth.rt_runtime; do_div(rt_runtime_us, NSEC_PER_USEC); return rt_runtime_us; } + +int sched_group_set_rt_period(struct task_group *tg, long rt_period_us) +{ + u64 rt_runtime, rt_period; + + rt_period = (u64)rt_period_us * NSEC_PER_USEC; + rt_runtime = tg->rt_bandwidth.rt_runtime; + + return tg_set_bandwidth(tg, rt_period, rt_runtime); +} + +long sched_group_rt_period(struct task_group *tg) +{ + u64 rt_period_us; + + rt_period_us = ktime_to_ns(tg->rt_bandwidth.rt_period); + do_div(rt_period_us, NSEC_PER_USEC); + return rt_period_us; +} + +static int sched_rt_global_constraints(void) +{ + int ret = 0; + + mutex_lock(&rt_constraints_mutex); + if (!__rt_schedulable(NULL, 1, 0)) + ret = -EINVAL; + mutex_unlock(&rt_constraints_mutex); + + return ret; +} +#else +static int sched_rt_global_constraints(void) +{ + return 0; +} #endif -#endif /* CONFIG_GROUP_SCHED */ + +int sched_rt_handler(struct ctl_table *table, int write, + struct file *filp, void __user *buffer, size_t *lenp, + loff_t *ppos) +{ + int ret; + int old_period, old_runtime; + static DEFINE_MUTEX(mutex); + + mutex_lock(&mutex); + old_period = sysctl_sched_rt_period; + old_runtime = sysctl_sched_rt_runtime; + + ret = proc_dointvec(table, write, filp, buffer, lenp, ppos); + + if (!ret && write) { + ret = sched_rt_global_constraints(); + if (ret) { + sysctl_sched_rt_period = old_period; + sysctl_sched_rt_runtime = old_runtime; + } else { + def_rt_bandwidth.rt_runtime = global_rt_runtime(); + def_rt_bandwidth.rt_period = + ns_to_ktime(global_rt_period()); + } + } + mutex_unlock(&mutex); + + return ret; +} #ifdef CONFIG_CGROUP_SCHED @@ -7988,7 +8134,7 @@ cpu_cgroup_can_attach(struct cgroup_subsys *ss, struct cgroup *cgrp, { #ifdef CONFIG_RT_GROUP_SCHED /* Don't accept realtime tasks when there is no way for them to run */ - if (rt_task(tsk) && cgroup_tg(cgrp)->rt_runtime == 0) + if (rt_task(tsk) && cgroup_tg(cgrp)->rt_bandwidth.rt_runtime == 0) return -EINVAL; #else /* We don't support RT-tasks being in separate groups */ @@ -8066,6 +8212,17 @@ static ssize_t cpu_rt_runtime_read(struct cgroup *cgrp, struct cftype *cft, return simple_read_from_buffer(buf, nbytes, ppos, tmp, len); } + +static int cpu_rt_period_write_uint(struct cgroup *cgrp, struct cftype *cftype, + u64 rt_period_us) +{ + return sched_group_set_rt_period(cgroup_tg(cgrp), rt_period_us); +} + +static u64 cpu_rt_period_read_uint(struct cgroup *cgrp, struct cftype *cft) +{ + return sched_group_rt_period(cgroup_tg(cgrp)); +} #endif static struct cftype cpu_files[] = { @@ -8082,6 +8239,11 @@ static struct cftype cpu_files[] = { .read = cpu_rt_runtime_read, .write = cpu_rt_runtime_write, }, + { + .name = "rt_period_us", + .read_uint = cpu_rt_period_read_uint, + .write_uint = cpu_rt_period_write_uint, + }, #endif }; diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 0a6d2e516420..8bc176136666 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -62,7 +62,7 @@ static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) if (!rt_rq->tg) return RUNTIME_INF; - return rt_rq->tg->rt_runtime; + return rt_rq->tg->rt_bandwidth.rt_runtime; } #define for_each_leaf_rt_rq(rt_rq, rq) \ @@ -127,14 +127,29 @@ static int rt_se_boosted(struct sched_rt_entity *rt_se) return p->prio != p->normal_prio; } +#ifdef CONFIG_SMP +static inline cpumask_t sched_rt_period_mask(void) +{ + return cpu_rq(smp_processor_id())->rd->span; +} #else +static inline cpumask_t sched_rt_period_mask(void) +{ + return cpu_online_map; +} +#endif -static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) +static inline +struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) { - if (sysctl_sched_rt_runtime == -1) - return RUNTIME_INF; + return container_of(rt_b, struct task_group, rt_bandwidth)->rt_rq[cpu]; +} - return (u64)sysctl_sched_rt_runtime * NSEC_PER_USEC; +#else + +static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) +{ + return def_rt_bandwidth.rt_runtime; } #define for_each_leaf_rt_rq(rt_rq, rq) \ @@ -173,8 +188,55 @@ static inline int rt_rq_throttled(struct rt_rq *rt_rq) { return rt_rq->rt_throttled; } + +static inline cpumask_t sched_rt_period_mask(void) +{ + return cpu_online_map; +} + +static inline +struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) +{ + return &cpu_rq(cpu)->rt; +} + #endif +static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) +{ + int i, idle = 1; + cpumask_t span; + + if (rt_b->rt_runtime == RUNTIME_INF) + return 1; + + span = sched_rt_period_mask(); + for_each_cpu_mask(i, span) { + int enqueue = 0; + struct rt_rq *rt_rq = sched_rt_period_rt_rq(rt_b, i); + struct rq *rq = rq_of_rt_rq(rt_rq); + + spin_lock(&rq->lock); + if (rt_rq->rt_time) { + u64 runtime = rt_b->rt_runtime; + + rt_rq->rt_time -= min(rt_rq->rt_time, overrun*runtime); + if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) { + rt_rq->rt_throttled = 0; + enqueue = 1; + } + if (rt_rq->rt_time || rt_rq->rt_nr_running) + idle = 0; + } + + if (enqueue) + sched_rt_rq_enqueue(rt_rq); + spin_unlock(&rq->lock); + } + + return idle; +} + static inline int rt_se_prio(struct sched_rt_entity *rt_se) { #ifdef CONFIG_RT_GROUP_SCHED @@ -198,11 +260,7 @@ static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) return rt_rq_throttled(rt_rq); if (rt_rq->rt_time > runtime) { - struct rq *rq = rq_of_rt_rq(rt_rq); - - rq->rt_throttled = 1; rt_rq->rt_throttled = 1; - if (rt_rq_throttled(rt_rq)) { sched_rt_rq_dequeue(rt_rq); return 1; @@ -212,29 +270,6 @@ static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) return 0; } -static void update_sched_rt_period(struct rq *rq) -{ - struct rt_rq *rt_rq; - u64 period; - - while (rq->clock > rq->rt_period_expire) { - period = (u64)sysctl_sched_rt_period * NSEC_PER_USEC; - rq->rt_period_expire += period; - - for_each_leaf_rt_rq(rt_rq, rq) { - u64 runtime = sched_rt_runtime(rt_rq); - - rt_rq->rt_time -= min(rt_rq->rt_time, runtime); - if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) { - rt_rq->rt_throttled = 0; - sched_rt_rq_enqueue(rt_rq); - } - } - - rq->rt_throttled = 0; - } -} - /* * Update the current task's runtime statistics. Skip current tasks that * are not in our scheduling class. @@ -284,6 +319,11 @@ void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) #ifdef CONFIG_RT_GROUP_SCHED if (rt_se_boosted(rt_se)) rt_rq->rt_nr_boosted++; + + if (rt_rq->tg) + start_rt_bandwidth(&rt_rq->tg->rt_bandwidth); +#else + start_rt_bandwidth(&def_rt_bandwidth); #endif } diff --git a/kernel/sysctl.c b/kernel/sysctl.c index be332e1a0c29..fd3364827ccf 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -307,7 +307,7 @@ static struct ctl_table kern_table[] = { .data = &sysctl_sched_rt_period, .maxlen = sizeof(unsigned int), .mode = 0644, - .proc_handler = &proc_dointvec, + .proc_handler = &sched_rt_handler, }, { .ctl_name = CTL_UNNUMBERED, @@ -315,7 +315,7 @@ static struct ctl_table kern_table[] = { .data = &sysctl_sched_rt_runtime, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = &proc_dointvec, + .proc_handler = &sched_rt_handler, }, { .ctl_name = CTL_UNNUMBERED, diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index 69dba0c71727..d358d4e3a958 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -191,7 +191,6 @@ u64 get_cpu_idle_time_us(int cpu, u64 *last_update_time) void tick_nohz_stop_sched_tick(void) { unsigned long seq, last_jiffies, next_jiffies, delta_jiffies, flags; - unsigned long rt_jiffies; struct tick_sched *ts; ktime_t last_update, expires, now; struct clock_event_device *dev = __get_cpu_var(tick_cpu_device).evtdev; @@ -243,10 +242,6 @@ void tick_nohz_stop_sched_tick(void) next_jiffies = get_next_timer_interrupt(last_jiffies); delta_jiffies = next_jiffies - last_jiffies; - rt_jiffies = rt_needs_cpu(cpu); - if (rt_jiffies && rt_jiffies < delta_jiffies) - delta_jiffies = rt_jiffies; - if (rcu_needs_cpu(cpu)) delta_jiffies = 1; /* diff --git a/kernel/user.c b/kernel/user.c index 7132022a040c..5925c6887c10 100644 --- a/kernel/user.c +++ b/kernel/user.c @@ -193,6 +193,33 @@ static ssize_t cpu_rt_runtime_store(struct kobject *kobj, static struct kobj_attribute cpu_rt_runtime_attr = __ATTR(cpu_rt_runtime, 0644, cpu_rt_runtime_show, cpu_rt_runtime_store); + +static ssize_t cpu_rt_period_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *buf) +{ + struct user_struct *up = container_of(kobj, struct user_struct, kobj); + + return sprintf(buf, "%lu\n", sched_group_rt_period(up->tg)); +} + +static ssize_t cpu_rt_period_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t size) +{ + struct user_struct *up = container_of(kobj, struct user_struct, kobj); + unsigned long rt_period; + int rc; + + sscanf(buf, "%lu", &rt_period); + + rc = sched_group_set_rt_period(up->tg, rt_period); + + return (rc ? rc : size); +} + +static struct kobj_attribute cpu_rt_period_attr = + __ATTR(cpu_rt_period, 0644, cpu_rt_period_show, cpu_rt_period_store); #endif /* default attributes per uid directory */ @@ -202,6 +229,7 @@ static struct attribute *uids_attributes[] = { #endif #ifdef CONFIG_RT_GROUP_SCHED &cpu_rt_runtime_attr.attr, + &cpu_rt_period_attr.attr, #endif NULL }; -- cgit v1.2.3 From ac086bc22997a2be24fc40fc8d46522fe7e03d11 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sat, 19 Apr 2008 19:44:58 +0200 Subject: sched: rt-group: smp balancing Currently the rt group scheduling does a per cpu runtime limit, however the rt load balancer makes no guarantees about an equal spread of real- time tasks, just that at any one time, the highest priority tasks run. Solve this by making the runtime limit a global property by borrowing excessive runtime from the other cpus once the local limit runs out. Signed-off-by: Peter Zijlstra Signed-off-by: Ingo Molnar --- kernel/sched.c | 40 +++++++++++++++++++++++-- kernel/sched_rt.c | 88 +++++++++++++++++++++++++++++++++++++++++++++++++++++-- 2 files changed, 122 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index bb20323f7d09..313cd4f057cf 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -164,6 +164,7 @@ struct rt_prio_array { struct rt_bandwidth { ktime_t rt_period; u64 rt_runtime; + spinlock_t rt_runtime_lock; struct hrtimer rt_period_timer; }; @@ -198,6 +199,8 @@ void init_rt_bandwidth(struct rt_bandwidth *rt_b, u64 period, u64 runtime) rt_b->rt_period = ns_to_ktime(period); rt_b->rt_runtime = runtime; + spin_lock_init(&rt_b->rt_runtime_lock); + hrtimer_init(&rt_b->rt_period_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); rt_b->rt_period_timer.function = sched_rt_period_timer; @@ -414,6 +417,8 @@ struct rt_rq { #endif int rt_throttled; u64 rt_time; + u64 rt_runtime; + spinlock_t rt_runtime_lock; #ifdef CONFIG_RT_GROUP_SCHED unsigned long rt_nr_boosted; @@ -7299,6 +7304,8 @@ static void init_rt_rq(struct rt_rq *rt_rq, struct rq *rq) rt_rq->rt_time = 0; rt_rq->rt_throttled = 0; + rt_rq->rt_runtime = 0; + spin_lock_init(&rt_rq->rt_runtime_lock); #ifdef CONFIG_RT_GROUP_SCHED rt_rq->rt_nr_boosted = 0; @@ -7335,6 +7342,7 @@ static void init_tg_rt_entry(struct rq *rq, struct task_group *tg, init_rt_rq(rt_rq, rq); rt_rq->tg = tg; rt_rq->rt_se = rt_se; + rt_rq->rt_runtime = tg->rt_bandwidth.rt_runtime; if (add) list_add(&rt_rq->leaf_rt_rq_list, &rq->leaf_rt_rq_list); @@ -7391,6 +7399,8 @@ void __init sched_init(void) init_tg_rt_entry(rq, &init_task_group, &per_cpu(init_rt_rq, i), &per_cpu(init_sched_rt_entity, i), i, 1); +#else + rq->rt.rt_runtime = def_rt_bandwidth.rt_runtime; #endif for (j = 0; j < CPU_LOAD_IDX_MAX; j++) @@ -7974,11 +7984,11 @@ static inline int tg_has_rt_tasks(struct task_group *tg) static int tg_set_bandwidth(struct task_group *tg, u64 rt_period, u64 rt_runtime) { - int err = 0; + int i, err = 0; mutex_lock(&rt_constraints_mutex); read_lock(&tasklist_lock); - if (rt_runtime_us == 0 && tg_has_rt_tasks(tg)) { + if (rt_runtime == 0 && tg_has_rt_tasks(tg)) { err = -EBUSY; goto unlock; } @@ -7986,8 +7996,19 @@ static int tg_set_bandwidth(struct task_group *tg, err = -EINVAL; goto unlock; } + + spin_lock_irq(&tg->rt_bandwidth.rt_runtime_lock); tg->rt_bandwidth.rt_period = ns_to_ktime(rt_period); tg->rt_bandwidth.rt_runtime = rt_runtime; + + for_each_possible_cpu(i) { + struct rt_rq *rt_rq = tg->rt_rq[i]; + + spin_lock(&rt_rq->rt_runtime_lock); + rt_rq->rt_runtime = rt_runtime; + spin_unlock(&rt_rq->rt_runtime_lock); + } + spin_unlock_irq(&tg->rt_bandwidth.rt_runtime_lock); unlock: read_unlock(&tasklist_lock); mutex_unlock(&rt_constraints_mutex); @@ -8052,6 +8073,19 @@ static int sched_rt_global_constraints(void) #else static int sched_rt_global_constraints(void) { + unsigned long flags; + int i; + + spin_lock_irqsave(&def_rt_bandwidth.rt_runtime_lock, flags); + for_each_possible_cpu(i) { + struct rt_rq *rt_rq = &cpu_rq(i)->rt; + + spin_lock(&rt_rq->rt_runtime_lock); + rt_rq->rt_runtime = global_rt_runtime(); + spin_unlock(&rt_rq->rt_runtime_lock); + } + spin_unlock_irqrestore(&def_rt_bandwidth.rt_runtime_lock, flags); + return 0; } #endif @@ -8168,7 +8202,7 @@ static u64 cpu_shares_read_uint(struct cgroup *cgrp, struct cftype *cft) #endif #ifdef CONFIG_RT_GROUP_SCHED -static int cpu_rt_runtime_write(struct cgroup *cgrp, struct cftype *cft, +static ssize_t cpu_rt_runtime_write(struct cgroup *cgrp, struct cftype *cft, struct file *file, const char __user *userbuf, size_t nbytes, loff_t *unused_ppos) diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 8bc176136666..6928ded24da1 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -62,7 +62,12 @@ static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) if (!rt_rq->tg) return RUNTIME_INF; - return rt_rq->tg->rt_bandwidth.rt_runtime; + return rt_rq->rt_runtime; +} + +static inline u64 sched_rt_period(struct rt_rq *rt_rq) +{ + return ktime_to_ns(rt_rq->tg->rt_bandwidth.rt_period); } #define for_each_leaf_rt_rq(rt_rq, rq) \ @@ -145,11 +150,21 @@ struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) return container_of(rt_b, struct task_group, rt_bandwidth)->rt_rq[cpu]; } +static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq) +{ + return &rt_rq->tg->rt_bandwidth; +} + #else static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) { - return def_rt_bandwidth.rt_runtime; + return rt_rq->rt_runtime; +} + +static inline u64 sched_rt_period(struct rt_rq *rt_rq) +{ + return ktime_to_ns(def_rt_bandwidth.rt_period); } #define for_each_leaf_rt_rq(rt_rq, rq) \ @@ -200,6 +215,11 @@ struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) return &cpu_rq(cpu)->rt; } +static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq) +{ + return &def_rt_bandwidth; +} + #endif static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) @@ -218,8 +238,10 @@ static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) spin_lock(&rq->lock); if (rt_rq->rt_time) { - u64 runtime = rt_b->rt_runtime; + u64 runtime; + spin_lock(&rt_rq->rt_runtime_lock); + runtime = rt_rq->rt_runtime; rt_rq->rt_time -= min(rt_rq->rt_time, overrun*runtime); if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) { rt_rq->rt_throttled = 0; @@ -227,6 +249,7 @@ static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) } if (rt_rq->rt_time || rt_rq->rt_nr_running) idle = 0; + spin_unlock(&rt_rq->rt_runtime_lock); } if (enqueue) @@ -237,6 +260,47 @@ static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) return idle; } +#ifdef CONFIG_SMP +static int balance_runtime(struct rt_rq *rt_rq) +{ + struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); + struct root_domain *rd = cpu_rq(smp_processor_id())->rd; + int i, weight, more = 0; + u64 rt_period; + + weight = cpus_weight(rd->span); + + spin_lock(&rt_b->rt_runtime_lock); + rt_period = ktime_to_ns(rt_b->rt_period); + for_each_cpu_mask(i, rd->span) { + struct rt_rq *iter = sched_rt_period_rt_rq(rt_b, i); + s64 diff; + + if (iter == rt_rq) + continue; + + spin_lock(&iter->rt_runtime_lock); + diff = iter->rt_runtime - iter->rt_time; + if (diff > 0) { + do_div(diff, weight); + if (rt_rq->rt_runtime + diff > rt_period) + diff = rt_period - rt_rq->rt_runtime; + iter->rt_runtime -= diff; + rt_rq->rt_runtime += diff; + more = 1; + if (rt_rq->rt_runtime == rt_period) { + spin_unlock(&iter->rt_runtime_lock); + break; + } + } + spin_unlock(&iter->rt_runtime_lock); + } + spin_unlock(&rt_b->rt_runtime_lock); + + return more; +} +#endif + static inline int rt_se_prio(struct sched_rt_entity *rt_se) { #ifdef CONFIG_RT_GROUP_SCHED @@ -259,6 +323,22 @@ static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) if (rt_rq->rt_throttled) return rt_rq_throttled(rt_rq); + if (sched_rt_runtime(rt_rq) >= sched_rt_period(rt_rq)) + return 0; + +#ifdef CONFIG_SMP + if (rt_rq->rt_time > runtime) { + int more; + + spin_unlock(&rt_rq->rt_runtime_lock); + more = balance_runtime(rt_rq); + spin_lock(&rt_rq->rt_runtime_lock); + + if (more) + runtime = sched_rt_runtime(rt_rq); + } +#endif + if (rt_rq->rt_time > runtime) { rt_rq->rt_throttled = 1; if (rt_rq_throttled(rt_rq)) { @@ -294,9 +374,11 @@ static void update_curr_rt(struct rq *rq) curr->se.exec_start = rq->clock; cpuacct_charge(curr, delta_exec); + spin_lock(&rt_rq->rt_runtime_lock); rt_rq->rt_time += delta_exec; if (sched_rt_runtime_exceeded(rt_rq)) resched_task(curr); + spin_unlock(&rt_rq->rt_runtime_lock); } static inline -- cgit v1.2.3 From 48f20a9a9488c432fc86df1ff4b7f4fa895d1183 Mon Sep 17 00:00:00 2001 From: Olof Johansson Date: Tue, 4 Mar 2008 15:23:25 -0800 Subject: tasklets: execute tasklets in the same order they were queued I noticed this when looking at an openswan issue. Openswan (ab?)uses the tasklet API to defer processing of packets in some situations, with one packet per tasklet_action(). I started noticing sequences of backwards-ordered sequence numbers coming over the wire, since new tasklets are always queued at the head of the list but processed sequentially. Convert it to instead append new entries to the tail of the list. As an extra bonus, the splicing code in takeover_tasklets() no longer has to iterate over the list. Signed-off-by: Olof Johansson Cc: "David S. Miller" Signed-off-by: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/softirq.c | 63 ++++++++++++++++++++++++++++++++++++-------------------- 1 file changed, 41 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/softirq.c b/kernel/softirq.c index 31e9f2a47928..3c44956ee7e2 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -356,7 +356,8 @@ void open_softirq(int nr, void (*action)(struct softirq_action*), void *data) /* Tasklets */ struct tasklet_head { - struct tasklet_struct *list; + struct tasklet_struct *head; + struct tasklet_struct **tail; }; /* Some compilers disobey section attribute on statics when not @@ -369,8 +370,9 @@ void __tasklet_schedule(struct tasklet_struct *t) unsigned long flags; local_irq_save(flags); - t->next = __get_cpu_var(tasklet_vec).list; - __get_cpu_var(tasklet_vec).list = t; + t->next = NULL; + *__get_cpu_var(tasklet_vec).tail = t; + __get_cpu_var(tasklet_vec).tail = &(t->next); raise_softirq_irqoff(TASKLET_SOFTIRQ); local_irq_restore(flags); } @@ -382,8 +384,9 @@ void __tasklet_hi_schedule(struct tasklet_struct *t) unsigned long flags; local_irq_save(flags); - t->next = __get_cpu_var(tasklet_hi_vec).list; - __get_cpu_var(tasklet_hi_vec).list = t; + t->next = NULL; + *__get_cpu_var(tasklet_hi_vec).tail = t; + __get_cpu_var(tasklet_hi_vec).tail = &(t->next); raise_softirq_irqoff(HI_SOFTIRQ); local_irq_restore(flags); } @@ -395,8 +398,9 @@ static void tasklet_action(struct softirq_action *a) struct tasklet_struct *list; local_irq_disable(); - list = __get_cpu_var(tasklet_vec).list; - __get_cpu_var(tasklet_vec).list = NULL; + list = __get_cpu_var(tasklet_vec).head; + __get_cpu_var(tasklet_vec).head = NULL; + __get_cpu_var(tasklet_vec).tail = &__get_cpu_var(tasklet_vec).head; local_irq_enable(); while (list) { @@ -416,8 +420,9 @@ static void tasklet_action(struct softirq_action *a) } local_irq_disable(); - t->next = __get_cpu_var(tasklet_vec).list; - __get_cpu_var(tasklet_vec).list = t; + t->next = NULL; + *__get_cpu_var(tasklet_vec).tail = t; + __get_cpu_var(tasklet_vec).tail = &(t->next); __raise_softirq_irqoff(TASKLET_SOFTIRQ); local_irq_enable(); } @@ -428,8 +433,9 @@ static void tasklet_hi_action(struct softirq_action *a) struct tasklet_struct *list; local_irq_disable(); - list = __get_cpu_var(tasklet_hi_vec).list; - __get_cpu_var(tasklet_hi_vec).list = NULL; + list = __get_cpu_var(tasklet_hi_vec).head; + __get_cpu_var(tasklet_hi_vec).head = NULL; + __get_cpu_var(tasklet_hi_vec).tail = &__get_cpu_var(tasklet_hi_vec).head; local_irq_enable(); while (list) { @@ -449,8 +455,9 @@ static void tasklet_hi_action(struct softirq_action *a) } local_irq_disable(); - t->next = __get_cpu_var(tasklet_hi_vec).list; - __get_cpu_var(tasklet_hi_vec).list = t; + t->next = NULL; + *__get_cpu_var(tasklet_hi_vec).tail = t; + __get_cpu_var(tasklet_hi_vec).tail = &(t->next); __raise_softirq_irqoff(HI_SOFTIRQ); local_irq_enable(); } @@ -487,6 +494,15 @@ EXPORT_SYMBOL(tasklet_kill); void __init softirq_init(void) { + int cpu; + + for_each_possible_cpu(cpu) { + per_cpu(tasklet_vec, cpu).tail = + &per_cpu(tasklet_vec, cpu).head; + per_cpu(tasklet_hi_vec, cpu).tail = + &per_cpu(tasklet_hi_vec, cpu).head; + } + open_softirq(TASKLET_SOFTIRQ, tasklet_action, NULL); open_softirq(HI_SOFTIRQ, tasklet_hi_action, NULL); } @@ -555,9 +571,12 @@ void tasklet_kill_immediate(struct tasklet_struct *t, unsigned int cpu) return; /* CPU is dead, so no lock needed. */ - for (i = &per_cpu(tasklet_vec, cpu).list; *i; i = &(*i)->next) { + for (i = &per_cpu(tasklet_vec, cpu).head; *i; i = &(*i)->next) { if (*i == t) { *i = t->next; + /* If this was the tail element, move the tail ptr */ + if (*i == NULL) + per_cpu(tasklet_vec, cpu).tail = i; return; } } @@ -566,20 +585,20 @@ void tasklet_kill_immediate(struct tasklet_struct *t, unsigned int cpu) static void takeover_tasklets(unsigned int cpu) { - struct tasklet_struct **i; - /* CPU is dead, so no lock needed. */ local_irq_disable(); /* Find end, append list for that CPU. */ - for (i = &__get_cpu_var(tasklet_vec).list; *i; i = &(*i)->next); - *i = per_cpu(tasklet_vec, cpu).list; - per_cpu(tasklet_vec, cpu).list = NULL; + *__get_cpu_var(tasklet_vec).tail = per_cpu(tasklet_vec, cpu).head; + __get_cpu_var(tasklet_vec).tail = per_cpu(tasklet_vec, cpu).tail; + per_cpu(tasklet_vec, cpu).head = NULL; + per_cpu(tasklet_vec, cpu).tail = &per_cpu(tasklet_vec, cpu).head; raise_softirq_irqoff(TASKLET_SOFTIRQ); - for (i = &__get_cpu_var(tasklet_hi_vec).list; *i; i = &(*i)->next); - *i = per_cpu(tasklet_hi_vec, cpu).list; - per_cpu(tasklet_hi_vec, cpu).list = NULL; + *__get_cpu_var(tasklet_hi_vec).tail = per_cpu(tasklet_hi_vec, cpu).head; + __get_cpu_var(tasklet_hi_vec).tail = per_cpu(tasklet_hi_vec, cpu).tail; + per_cpu(tasklet_hi_vec, cpu).head = NULL; + per_cpu(tasklet_hi_vec, cpu).tail = &per_cpu(tasklet_hi_vec, cpu).head; raise_softirq_irqoff(HI_SOFTIRQ); local_irq_enable(); -- cgit v1.2.3 From 32cd756a80aaef657ac09c76e6eff3ba65567790 Mon Sep 17 00:00:00 2001 From: Dhaval Giani Date: Fri, 29 Feb 2008 10:02:43 +0530 Subject: sched: cleanup cpuacct variable names Change the variable names to the common convention for the cpuacct subsystem. Signed-off-by: Dhaval Giani Acked-by: Balbir Singh Signed-off-by: Ingo Molnar --- kernel/sched.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 313cd4f057cf..e2f85c7a7476 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -8318,9 +8318,9 @@ struct cpuacct { struct cgroup_subsys cpuacct_subsys; /* return cpu accounting group corresponding to this container */ -static inline struct cpuacct *cgroup_ca(struct cgroup *cont) +static inline struct cpuacct *cgroup_ca(struct cgroup *cgrp) { - return container_of(cgroup_subsys_state(cont, cpuacct_subsys_id), + return container_of(cgroup_subsys_state(cgrp, cpuacct_subsys_id), struct cpuacct, css); } @@ -8333,7 +8333,7 @@ static inline struct cpuacct *task_ca(struct task_struct *tsk) /* create a new cpu accounting group */ static struct cgroup_subsys_state *cpuacct_create( - struct cgroup_subsys *ss, struct cgroup *cont) + struct cgroup_subsys *ss, struct cgroup *cgrp) { struct cpuacct *ca = kzalloc(sizeof(*ca), GFP_KERNEL); @@ -8351,18 +8351,18 @@ static struct cgroup_subsys_state *cpuacct_create( /* destroy an existing cpu accounting group */ static void -cpuacct_destroy(struct cgroup_subsys *ss, struct cgroup *cont) +cpuacct_destroy(struct cgroup_subsys *ss, struct cgroup *cgrp) { - struct cpuacct *ca = cgroup_ca(cont); + struct cpuacct *ca = cgroup_ca(cgrp); free_percpu(ca->cpuusage); kfree(ca); } /* return total cpu usage (in nanoseconds) of a group */ -static u64 cpuusage_read(struct cgroup *cont, struct cftype *cft) +static u64 cpuusage_read(struct cgroup *cgrp, struct cftype *cft) { - struct cpuacct *ca = cgroup_ca(cont); + struct cpuacct *ca = cgroup_ca(cgrp); u64 totalcpuusage = 0; int i; @@ -8388,9 +8388,9 @@ static struct cftype files[] = { }, }; -static int cpuacct_populate(struct cgroup_subsys *ss, struct cgroup *cont) +static int cpuacct_populate(struct cgroup_subsys *ss, struct cgroup *cgrp) { - return cgroup_add_files(cont, ss, files, ARRAY_SIZE(files)); + return cgroup_add_files(cgrp, ss, files, ARRAY_SIZE(files)); } /* -- cgit v1.2.3 From 0297b80339d545045490716fa8591b215fdd9458 Mon Sep 17 00:00:00 2001 From: Dhaval Giani Date: Fri, 29 Feb 2008 10:02:44 +0530 Subject: sched: allow cpuacct stats to be reset Currently the schedstats implementation does not allow the statistics to be reset. This patch aims to allow that. echo 0 > cpuacct.usage resets the usage. Any other value is not allowed and returns -EINVAL. Signed-off-by: Dhaval Giani Signed-off-by: Ingo Molnar --- kernel/sched.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index e2f85c7a7476..e4bf4477aee5 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -8381,10 +8381,34 @@ static u64 cpuusage_read(struct cgroup *cgrp, struct cftype *cft) return totalcpuusage; } +static int cpuusage_write(struct cgroup *cgrp, struct cftype *cftype, + u64 reset) +{ + struct cpuacct *ca = cgroup_ca(cgrp); + int err = 0; + int i; + + if (reset) { + err = -EINVAL; + goto out; + } + + for_each_possible_cpu(i) { + u64 *cpuusage = percpu_ptr(ca->cpuusage, i); + + spin_lock_irq(&cpu_rq(i)->lock); + *cpuusage = 0; + spin_unlock_irq(&cpu_rq(i)->lock); + } +out: + return err; +} + static struct cftype files[] = { { .name = "usage", .read_uint = cpuusage_read, + .write_uint = cpuusage_write, }, }; -- cgit v1.2.3 From 9f0e738f492522a2f70ad9a2a0287e4e966c633a Mon Sep 17 00:00:00 2001 From: Gregory Haskins Date: Tue, 12 Feb 2008 13:30:05 -0500 Subject: sched: fix cpus_allowed settings Signed-off-by: Gregory Haskins Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/kthread.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/kthread.c b/kernel/kthread.c index 0ac887882f90..25241d6ec8cd 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -180,6 +180,7 @@ void kthread_bind(struct task_struct *k, unsigned int cpu) wait_task_inactive(k); set_task_cpu(k, cpu); k->cpus_allowed = cpumask_of_cpu(cpu); + k->rt.nr_cpus_allowed = 1; } EXPORT_SYMBOL(kthread_bind); -- cgit v1.2.3 From d366f8cbc16882e93538d9a52423c2f50dad7c06 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Fri, 4 Apr 2008 18:11:12 -0700 Subject: cpumask: Cleanup more uses of CPU_MASK and NODE_MASK * Replace usages of CPU_MASK_NONE, CPU_MASK_ALL, NODE_MASK_NONE, NODE_MASK_ALL to reduce stack requirements for large NR_CPUS and MAXNODES counts. * In some cases, the cpumask variable was initialized but then overwritten with another value. This is the case for changes like this: - cpumask_t oldmask = CPU_MASK_ALL; + cpumask_t oldmask; Signed-off-by: Mike Travis Signed-off-by: Ingo Molnar --- kernel/irq/chip.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index fdb3fbe2b0c4..964964baefa2 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -47,7 +47,7 @@ void dynamic_irq_init(unsigned int irq) desc->irq_count = 0; desc->irqs_unhandled = 0; #ifdef CONFIG_SMP - desc->affinity = CPU_MASK_ALL; + cpus_setall(desc->affinity); #endif spin_unlock_irqrestore(&desc->lock, flags); } -- cgit v1.2.3 From 434d53b00d6bb7be0a1d3dcc0d0d5df6c042e164 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Fri, 4 Apr 2008 18:11:04 -0700 Subject: sched: remove fixed NR_CPUS sized arrays in kernel_sched_c * Change fixed size arrays to per_cpu variables or dynamically allocated arrays in sched_init() and sched_init_smp(). (1) static struct sched_entity *init_sched_entity_p[NR_CPUS]; (1) static struct cfs_rq *init_cfs_rq_p[NR_CPUS]; (1) static struct sched_rt_entity *init_sched_rt_entity_p[NR_CPUS]; (1) static struct rt_rq *init_rt_rq_p[NR_CPUS]; static struct sched_group **sched_group_nodes_bycpu[NR_CPUS]; (1) - these arrays are allocated via alloc_bootmem_low() * Change sched_domain_debug_one() to use cpulist_scnprintf instead of cpumask_scnprintf. This reduces the output buffer required and improves readability when large NR_CPU count machines arrive. * In sched_create_group() we allocate new arrays based on nr_cpu_ids. Signed-off-by: Mike Travis Signed-off-by: Ingo Molnar --- kernel/sched.c | 80 ++++++++++++++++++++++++++++++++++++++-------------------- 1 file changed, 52 insertions(+), 28 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index e4bf4477aee5..ef3f28b334ea 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -67,6 +67,7 @@ #include #include #include +#include #include #include @@ -276,17 +277,11 @@ struct task_group { static DEFINE_PER_CPU(struct sched_entity, init_sched_entity); /* Default task group's cfs_rq on each cpu */ static DEFINE_PER_CPU(struct cfs_rq, init_cfs_rq) ____cacheline_aligned_in_smp; - -static struct sched_entity *init_sched_entity_p[NR_CPUS]; -static struct cfs_rq *init_cfs_rq_p[NR_CPUS]; #endif #ifdef CONFIG_RT_GROUP_SCHED static DEFINE_PER_CPU(struct sched_rt_entity, init_sched_rt_entity); static DEFINE_PER_CPU(struct rt_rq, init_rt_rq) ____cacheline_aligned_in_smp; - -static struct sched_rt_entity *init_sched_rt_entity_p[NR_CPUS]; -static struct rt_rq *init_rt_rq_p[NR_CPUS]; #endif /* task_group_lock serializes add/remove of task groups and also changes to @@ -310,17 +305,7 @@ static int in