Change the rt_ratio interface to rt_runtime_us, to match rt_period_us. This avoids picking a granularity for the ratio. Extend the /sys/kernel/uids// interface to allow setting the group's rt_runtime. Signed-off-by: Peter Zijlstra --- Documentation/ABI/testing/sysfs-kernel-uids | 6 + Documentation/sched-rt-group.txt | 59 +++++++++++ include/linux/sched.h | 7 - kernel/sched.c | 145 +++++++++++++++++++++------- kernel/sched_rt.c | 53 ++++------ kernel/sysctl.c | 32 +++--- kernel/user.c | 28 +++++ 7 files changed, 250 insertions(+), 80 deletions(-) Index: linux-2.6/include/linux/sched.h =================================================================== --- linux-2.6.orig/include/linux/sched.h +++ linux-2.6/include/linux/sched.h @@ -1507,8 +1507,6 @@ extern unsigned int sysctl_sched_child_r extern unsigned int sysctl_sched_features; extern unsigned int sysctl_sched_migration_cost; extern unsigned int sysctl_sched_nr_migrate; -extern unsigned int sysctl_sched_rt_period; -extern unsigned int sysctl_sched_rt_ratio; #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP) extern unsigned int sysctl_sched_min_bal_int_shares; extern unsigned int sysctl_sched_max_bal_int_shares; @@ -1518,6 +1516,8 @@ int sched_nr_latency_handler(struct ctl_ struct file *file, void __user *buffer, size_t *length, loff_t *ppos); #endif +extern unsigned int sysctl_sched_rt_period; +extern int sysctl_sched_rt_runtime; extern unsigned int sysctl_sched_compat_yield; @@ -1997,6 +1997,9 @@ extern void sched_destroy_group(struct t extern void sched_move_task(struct task_struct *tsk); extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); extern unsigned long sched_group_shares(struct task_group *tg); +extern int sched_group_set_rt_runtime(struct task_group *tg, + long rt_runtime_us); +extern long sched_group_rt_runtime(struct task_group *tg); #endif Index: linux-2.6/kernel/sched.c =================================================================== --- linux-2.6.orig/kernel/sched.c +++ linux-2.6/kernel/sched.c @@ -176,7 +176,7 @@ struct task_group { struct sched_rt_entity **rt_se; struct rt_rq **rt_rq; - unsigned int rt_ratio; + u64 rt_runtime; /* * shares assigned to a task group governs how much of cpu bandwidth @@ -654,19 +654,21 @@ const_debug unsigned int sysctl_sched_fe const_debug unsigned int sysctl_sched_nr_migrate = 32; /* - * period over which we measure -rt task cpu usage in ms. + * period over which we measure -rt task cpu usage in us. * default: 1s */ -const_debug unsigned int sysctl_sched_rt_period = 1000; +unsigned int sysctl_sched_rt_period = 1000000; -#define SCHED_RT_FRAC_SHIFT 16 -#define SCHED_RT_FRAC (1UL << SCHED_RT_FRAC_SHIFT) +/* + * part of the period that we allow rt tasks to run in us. + * default: 0.95s + */ +int sysctl_sched_rt_runtime = 950000; /* - * ratio of time -rt tasks may consume. - * default: 95% + * single value that denotes runtime == period, ie unlimited time. */ -const_debug unsigned int sysctl_sched_rt_ratio = 62259; +#define RUNTIME_INF ((u64)~0ULL) /* * For kernel-internal use: high-speed (but slightly incorrect) per-cpu @@ -7191,7 +7193,8 @@ void __init sched_init(void) &per_cpu(init_cfs_rq, i), &per_cpu(init_sched_entity, i), i, 1); - init_task_group.rt_ratio = sysctl_sched_rt_ratio; /* XXX */ + 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), @@ -7586,7 +7589,7 @@ struct task_group *sched_create_group(vo goto err; tg->shares = NICE_0_LOAD; - tg->rt_ratio = 0; /* XXX */ + tg->rt_runtime = 0; for_each_possible_cpu(i) { rq = cpu_rq(i); @@ -7780,30 +7783,76 @@ unsigned long sched_group_shares(struct } /* - * Ensure the total rt_ratio <= sysctl_sched_rt_ratio + * Ensure that the real time constraints are schedulable. */ -int sched_group_set_rt_ratio(struct task_group *tg, unsigned long rt_ratio) +static DEFINE_MUTEX(rt_constraints_mutex); + +static unsigned long to_ratio(u64 period, u64 runtime) +{ + if (runtime == RUNTIME_INF) + return 1ULL << 16; + + runtime *= (1ULL << 16); + do_div(runtime, period); + return runtime; +} + +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); rcu_read_lock(); - list_for_each_entry_rcu(tgi, &task_groups, list) - total += tgi->rt_ratio; - rcu_read_unlock(); + list_for_each_entry_rcu(tgi, &task_groups, list) { + if (tgi == tg) + continue; - if (total + rt_ratio - tg->rt_ratio > sysctl_sched_rt_ratio) - return -EINVAL; + total += to_ratio(period, tgi->rt_runtime); + } + rcu_read_unlock(); - tg->rt_ratio = rt_ratio; - return 0; + return total + to_ratio(period, runtime) < global_ratio; } -unsigned long sched_group_rt_ratio(struct task_group *tg) +int sched_group_set_rt_runtime(struct task_group *tg, long rt_runtime_us) { - return tg->rt_ratio; + u64 rt_runtime, rt_period; + int err = 0; + + rt_period = sysctl_sched_rt_period * NSEC_PER_USEC; + rt_runtime = (u64)rt_runtime_us * NSEC_PER_USEC; + if (rt_runtime_us == -1) + rt_runtime = rt_period; + + mutex_lock(&rt_constraints_mutex); + if (!__rt_schedulable(tg, rt_period, rt_runtime)) { + err = -EINVAL; + goto unlock; + } + if (rt_runtime_us == -1) + rt_runtime = RUNTIME_INF; + tg->rt_runtime = rt_runtime; + unlock: + mutex_unlock(&rt_constraints_mutex); + + return err; } +long sched_group_rt_runtime(struct task_group *tg) +{ + u64 rt_runtime_us; + + if (tg->rt_runtime == RUNTIME_INF) + return -1; + + rt_runtime_us = tg->rt_runtime; + do_div(rt_runtime_us, NSEC_PER_USEC); + return rt_runtime_us; +} #endif /* CONFIG_FAIR_GROUP_SCHED */ #ifdef CONFIG_FAIR_CGROUP_SCHED @@ -7879,17 +7928,49 @@ static u64 cpu_shares_read_uint(struct c return (u64) tg->shares; } -static int cpu_rt_ratio_write_uint(struct cgroup *cgrp, struct cftype *cftype, - u64 rt_ratio_val) -{ - return sched_group_set_rt_ratio(cgroup_tg(cgrp), rt_ratio_val); +static int cpu_rt_runtime_write(struct cgroup *cgrp, struct cftype *cft, + struct file *file, + const char __user *userbuf, + size_t nbytes, loff_t *unused_ppos) +{ + char buffer[64]; + int retval = 0; + s64 val; + char *end; + + if (!nbytes) + return -EINVAL; + if (nbytes >= sizeof(buffer)) + return -E2BIG; + if (copy_from_user(buffer, userbuf, nbytes)) + return -EFAULT; + + buffer[nbytes] = 0; /* nul-terminate */ + + /* strip newline if necessary */ + if (nbytes && (buffer[nbytes-1] == '\n')) + buffer[nbytes-1] = 0; + val = simple_strtoll(buffer, &end, 0); + if (*end) + return -EINVAL; + + /* Pass to subsystem */ + retval = sched_group_set_rt_runtime(cgroup_tg(cgrp), val); + if (!retval) + retval = nbytes; + return retval; } -static u64 cpu_rt_ratio_read_uint(struct cgroup *cgrp, struct cftype *cft) -{ - struct task_group *tg = cgroup_tg(cgrp); +static ssize_t cpu_rt_runtime_read(struct cgroup *cgrp, struct cftype *cft, + struct file *file, + char __user *buf, size_t nbytes, + loff_t *ppos) +{ + char tmp[64]; + long val = sched_group_rt_runtime(cgroup_tg(cgrp)); + int len = sprintf(tmp, "%ld\n", val); - return (u64) tg->rt_ratio; + return simple_read_from_buffer(buf, nbytes, ppos, tmp, len); } static struct cftype cpu_files[] = { @@ -7899,9 +7980,9 @@ static struct cftype cpu_files[] = { .write_uint = cpu_shares_write_uint, }, { - .name = "rt_ratio", - .read_uint = cpu_rt_ratio_read_uint, - .write_uint = cpu_rt_ratio_write_uint, + .name = "rt_runtime_us", + .read = cpu_rt_runtime_read, + .write = cpu_rt_runtime_write, }, }; Index: linux-2.6/kernel/sched_rt.c =================================================================== --- linux-2.6.orig/kernel/sched_rt.c +++ linux-2.6/kernel/sched_rt.c @@ -57,12 +57,12 @@ static inline int on_rt_rq(struct sched_ #ifdef CONFIG_FAIR_GROUP_SCHED -static inline unsigned int sched_rt_ratio(struct rt_rq *rt_rq) +static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) { if (!rt_rq->tg) - return SCHED_RT_FRAC; + return RUNTIME_INF; - return rt_rq->tg->rt_ratio; + return rt_rq->tg->rt_runtime; } #define for_each_leaf_rt_rq(rt_rq, rq) \ @@ -89,7 +89,7 @@ static inline struct rt_rq *group_rt_rq( static void enqueue_rt_entity(struct sched_rt_entity *rt_se); static void dequeue_rt_entity(struct sched_rt_entity *rt_se); -static void sched_rt_ratio_enqueue(struct rt_rq *rt_rq) +static void sched_rt_rq_enqueue(struct rt_rq *rt_rq) { struct sched_rt_entity *rt_se = rt_rq->rt_se; @@ -102,7 +102,7 @@ static void sched_rt_ratio_enqueue(struc } } -static void sched_rt_ratio_dequeue(struct rt_rq *rt_rq) +static void sched_rt_rq_dequeue(struct rt_rq *rt_rq) { struct sched_rt_entity *rt_se = rt_rq->rt_se; @@ -129,9 +129,12 @@ static int rt_se_boosted(struct sched_rt #else -static inline unsigned int sched_rt_ratio(struct rt_rq *rt_rq) +static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) { - return sysctl_sched_rt_ratio; + if (sysctl_sched_rt_runtime == -1) + return RUNTIME_INF; + + return (u64)sysctl_sched_rt_runtime * NSEC_PER_USEC; } #define for_each_leaf_rt_rq(rt_rq, rq) \ @@ -158,11 +161,11 @@ static inline struct rt_rq *group_rt_rq( return NULL; } -static inline void sched_rt_ratio_enqueue(struct rt_rq *rt_rq) +static inline void sched_rt_rq_enqueue(struct rt_rq *rt_rq) { } -static inline void sched_rt_ratio_dequeue(struct rt_rq *rt_rq) +static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq) { } @@ -184,28 +187,24 @@ static inline int rt_se_prio(struct sche return rt_task_of(rt_se)->prio; } -static int sched_rt_ratio_exceeded(struct rt_rq *rt_rq) +static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) { - unsigned int rt_ratio = sched_rt_ratio(rt_rq); - u64 period, ratio; + u64 runtime = sched_rt_runtime(rt_rq); - if (rt_ratio == SCHED_RT_FRAC) + if (runtime == RUNTIME_INF) return 0; if (rt_rq->rt_throttled) return rt_rq_throttled(rt_rq); - period = (u64)sysctl_sched_rt_period * NSEC_PER_MSEC; - ratio = (period * rt_ratio) >> SCHED_RT_FRAC_SHIFT; - - if (rt_rq->rt_time > ratio) { + 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_ratio_dequeue(rt_rq); + sched_rt_rq_dequeue(rt_rq); return 1; } } @@ -219,17 +218,16 @@ static void update_sched_rt_period(struc u64 period; while (rq->clock > rq->rt_period_expire) { - period = (u64)sysctl_sched_rt_period * NSEC_PER_MSEC; + period = (u64)sysctl_sched_rt_period * NSEC_PER_USEC; rq->rt_period_expire += period; for_each_leaf_rt_rq(rt_rq, rq) { - unsigned long rt_ratio = sched_rt_ratio(rt_rq); - u64 ratio = (period * rt_ratio) >> SCHED_RT_FRAC_SHIFT; + u64 runtime = sched_rt_runtime(rt_rq); - rt_rq->rt_time -= min(rt_rq->rt_time, ratio); - if (rt_rq->rt_throttled) { + 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_ratio_enqueue(rt_rq); + sched_rt_rq_enqueue(rt_rq); } } @@ -262,12 +260,7 @@ static void update_curr_rt(struct rq *rq cpuacct_charge(curr, delta_exec); rt_rq->rt_time += delta_exec; - /* - * might make it a tad more accurate: - * - * update_sched_rt_period(rq); - */ - if (sched_rt_ratio_exceeded(rt_rq)) + if (sched_rt_runtime_exceeded(rt_rq)) resched_task(curr); } Index: linux-2.6/kernel/sysctl.c =================================================================== --- linux-2.6.orig/kernel/sysctl.c +++ linux-2.6/kernel/sysctl.c @@ -311,22 +311,6 @@ static struct ctl_table kern_table[] = { .mode = 0644, .proc_handler = &proc_dointvec, }, - { - .ctl_name = CTL_UNNUMBERED, - .procname = "sched_rt_period_ms", - .data = &sysctl_sched_rt_period, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = &proc_dointvec, - }, - { - .ctl_name = CTL_UNNUMBERED, - .procname = "sched_rt_ratio", - .data = &sysctl_sched_rt_ratio, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = &proc_dointvec, - }, #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP) { .ctl_name = CTL_UNNUMBERED, @@ -348,6 +332,22 @@ static struct ctl_table kern_table[] = { #endif { .ctl_name = CTL_UNNUMBERED, + .procname = "sched_rt_period_us", + .data = &sysctl_sched_rt_period, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = &proc_dointvec, + }, + { + .ctl_name = CTL_UNNUMBERED, + .procname = "sched_rt_runtime_us", + .data = &sysctl_sched_rt_runtime, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = &proc_dointvec, + }, + { + .ctl_name = CTL_UNNUMBERED, .procname = "sched_compat_yield", .data = &sysctl_sched_compat_yield, .maxlen = sizeof(unsigned int), Index: linux-2.6/Documentation/ABI/testing/sysfs-kernel-uids =================================================================== --- linux-2.6.orig/Documentation/ABI/testing/sysfs-kernel-uids +++ linux-2.6/Documentation/ABI/testing/sysfs-kernel-uids @@ -12,3 +12,9 @@ Description: B has shares = 2048, User B will get twice the CPU bandwidth user A will. For more details refer Documentation/sched-design-CFS.txt + +What: /sys/kernel/uids//cpu_rt_runtime_us +Date: January 2008 +Contact: Peter Zijlstra +Description: See Documentation/sched-rt-group.txt + Index: linux-2.6/Documentation/sched-rt-group.txt =================================================================== --- /dev/null +++ linux-2.6/Documentation/sched-rt-group.txt @@ -0,0 +1,59 @@ + + +Real-Time group scheduling. + +The problem space: + +In order to schedule multiple groups of realtime tasks each group must +be assigned a fixed portion of the cpu time available. Without a minimum +guarantee a realtime group can obviously fall short. A fuzzy upper limit +is of no use since it cannot be relied upon. Which leaves us with just +the single fixed portion. + +CPU time is divided by means of specifying how much time can be spend +running in a given period. Say a frame fixed realtime renderer must +deliver a 25 frames a second, which yields a period of 0.04s. Now say +it will also have to play some music and respond to input, leaving it +with around 80% for the graphics. We can then give this group a runtime +of 0.8 * 0.04s = 0.032s. + +This way the graphics group will have a 0.04s period with a 0.032s runtime +limit. + +Now if the audio thread needs to refill the dma buffer every 0.005s, but +needs only about 3% cpu time to do so, it will can do with a 0.03 * 0.005s += 0.00015s. + + +The Interface: + +system wide: + +/proc/sys/kernel/sched_rt_period_ms +/proc/sys/kernel/sched_rt_runtime_us + +CONFIG_FAIR_USER_SCHED + +/sys/kernel/uids//cpu_rt_runtime_us + +or + +CONFIG_FAIR_CGROUP_SCHED + +/cgroup//cpu.rt_runtime_us + +[ time is specified in us because the interface is s32, this gives an + operating range of ~35m to 1us ] + +The period takes values in [ 1, INT_MAX ], runtime in [ -1, INT_MAX - 1 ]. + +A runtime of -1 specifies runtime == period, ie. no limit. + +New groups get the period from /proc/sys/kernel/sched_rt_period_us and +a runtime of 0. + +Settings are constrainted to: + + \Sum_{i} runtime_{i} / global_period <= global_runtime / global_period + +in order to keep the configuration schedulable. Index: linux-2.6/kernel/user.c =================================================================== --- linux-2.6.orig/kernel/user.c +++ linux-2.6/kernel/user.c @@ -156,9 +156,37 @@ static ssize_t cpu_shares_store(struct k static struct kobj_attribute cpu_share_attr = __ATTR(cpu_share, 0644, cpu_shares_show, cpu_shares_store); +static ssize_t cpu_rt_runtime_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_runtime(up->tg)); +} + +static ssize_t cpu_rt_runtime_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_runtime; + int rc; + + sscanf(buf, "%lu", &rt_runtime); + + rc = sched_group_set_rt_runtime(up->tg, rt_runtime); + + return (rc ? rc : size); +} + +static struct kobj_attribute cpu_rt_runtime_attr = + __ATTR(cpu_rt_runtime, 0644, cpu_rt_runtime_show, cpu_rt_runtime_store); + /* default attributes per uid directory */ static struct attribute *uids_attributes[] = { &cpu_share_attr.attr, + &cpu_rt_runtime_attr.attr, NULL }; -- -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/