Normalize the CPU priority system between the two search algorithms, and modularlize the search function within push_rt_tasks. Signed-off-by: Gregory Haskins <ghaskins@xxxxxxxxxx> --- kernel/sched.c | 91 ++++++++++++++++++++++++++++++++++++++------------------ 1 files changed, 61 insertions(+), 30 deletions(-) diff --git a/kernel/sched.c b/kernel/sched.c index 0065551..e8942c5 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -24,6 +24,7 @@ * by Peter Williams * 2007-05-06 Interactivity improvements to CFS by Mike Galbraith * 2007-07-01 Group scheduling enhancements by Srivatsa Vaddagiri + * 2007-10-11 RT overload enhancements by Steven Rostedt, Greg Haskins */ #include <linux/mm.h> @@ -305,8 +306,10 @@ struct rq { #ifdef CONFIG_PREEMPT_RT unsigned long rt_nr_running; unsigned long rt_nr_uninterruptible; +#ifndef CONFIG_CPU_PRIORITIES int curr_prio; #endif +#endif unsigned long switch_timestamp; unsigned long slice_avg; @@ -501,9 +504,48 @@ static int get_cpupri(struct rq *rq) return cpupri_get(rq->cpu); } +static int find_lowest_cpu(int defcpu, int pri, struct task_struct *p) +{ + return cpupri_find_best(defcpu, pri, p); +} #else -#define set_cpupri(rq, task) do { } while (0) -#define get_cpupri(rq) CPUPRI_INVALID +static void set_cpupri(struct rq *rq, struct task_struct *p) +{ + rq->curr_prio = calc_task_cpupri(rq, p); +} + +static int get_cpupri(struct rq *rq) +{ + return rq->curr_prio; +} + +static int find_lowest_cpu(int defcpu, int pri, struct task_struct *p) +{ + cpumask_t mask; + int cpu; + struct rq *lowest_rq = NULL; + + cpus_and(mask, cpu_online_map, p->cpus_allowed); + + /* + * Scan each rq for the lowest prio. + */ + for_each_cpu_mask(cpu, mask) { + struct rq *rq = &per_cpu(runqueues, cpu); + + if (cpu == smp_processor_id()) + continue; + + /* no locking for now */ + if (rq->curr_prio < pri && + (!lowest_rq || rq->curr_prio < lowest_rq->curr_prio)) { + lowest_rq = rq; + } + } + + return lowest_rq ? lowest_rq->cpu : defcpu; +} + #endif /* @@ -1531,11 +1573,9 @@ static int double_lock_balance(struct rq *this_rq, struct rq *busiest); static int push_rt_task(struct rq *this_rq) { struct task_struct *next_task; - struct rq *lowest_rq = NULL; int tries; - int cpu; - int dst_cpu = -1; int ret = 0; + int pri; assert_spin_locked(&this_rq->lock); @@ -1546,32 +1586,19 @@ static int push_rt_task(struct rq *this_rq) /* We might release this_rq lock */ get_task_struct(next_task); + pri = calc_task_cpupri(this_rq, next_task); + /* Only try this algorithm three times */ for (tries = 0; tries < 3; tries++) { - cpumask_t mask; - - cpus_and(mask, cpu_online_map, next_task->cpus_allowed); - - /* - * Scan each rq for the lowest prio. - */ - for_each_cpu_mask(cpu, mask) { - struct rq *rq = &per_cpu(runqueues, cpu); - - if (cpu == smp_processor_id()) - continue; - - /* no locking for now */ - if (rq->curr_prio > next_task->prio && - (!lowest_rq || rq->curr_prio < lowest_rq->curr_prio)) { - dst_cpu = cpu; - lowest_rq = rq; - } - } + struct rq *lowest_rq; + int cpu; - if (!lowest_rq) + cpu = find_lowest_cpu(this_rq->cpu, pri, next_task); + if (cpu == this_rq->cpu) break; + lowest_rq = cpu_rq(cpu); + if (double_lock_balance(this_rq, lowest_rq)) { /* * We had to unlock the run queue. In @@ -1579,20 +1606,20 @@ static int push_rt_task(struct rq *this_rq) * migrated already or had its affinity changed. */ if (unlikely(task_rq(next_task) != this_rq || - !cpu_isset(dst_cpu, next_task->cpus_allowed))) { + !cpu_isset(cpu, next_task->cpus_allowed))) { spin_unlock(&lowest_rq->lock); break; } } /* if the prio of this runqueue changed, try again */ - if (lowest_rq->curr_prio <= next_task->prio) { + if (get_cpupri(lowest_rq) > pri) { spin_unlock(&lowest_rq->lock); continue; } deactivate_task(this_rq, next_task, 0); - set_task_cpu(next_task, dst_cpu); + set_task_cpu(next_task, lowest_rq->cpu); activate_task(lowest_rq, next_task, 0); resched_task(lowest_rq->curr); @@ -2333,7 +2360,7 @@ static inline void finish_task_switch(struct rq *rq, struct task_struct *prev) * If we pushed an RT task off the runqueue, * then kick other CPUs, they might run it: */ - rq->curr_prio = current->prio; + if (unlikely(rq->rt_nr_running > 1)) push_rt_task(rq); @@ -7371,6 +7398,10 @@ void __init sched_init(void) highest_cpu = i; /* delimiter for bitsearch: */ __set_bit(MAX_RT_PRIO, array->bitmap); + +#if defined(CONFIG_PREEMPT_RT) && !defined(CONFIG_CPU_PRIORITIES) + rq->curr_prio = CPUPRI_INVALID; +#endif } set_load_weight(&init_task); - To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html