Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 77ecebbc authored by Vikram Mulukutla's avatar Vikram Mulukutla Committed by Pavankumar Kondeti
Browse files

sched: walt: Optimize cycle counter reads 



The cycle counter read is a bit of an expensive operation and requires
locking across all CPUs in a frequency domain. Optimize this by
returning the same value if the delta between two reads is zero i.e
two reads are done in the same sched context for the same CPU.

Change-Id: I99da5a704d3652f53c8564ba7532783d3288f227
Signed-off-by: default avatarVikram Mulukutla <markivx@codeaurora.org>
[pkondeti@codeaurora.org: limit the optimization to the
same CPU for the sched context]
Signed-off-by: default avatarPavankumar Kondeti <pkondeti@codeaurora.org>
parent f5c48058

kernel/sched/sched.h

+2 −0
Original line number Diff line number Diff line
@@ -823,6 +823,8 @@ struct rq { 
	int prev_top;

	int curr_top;

	bool notif_pending;

	u64 last_cc_update;

	u64 cycles;

#endif



#ifdef CONFIG_IRQ_TIME_ACCOUNTING

kernel/sched/walt.c

+26 −7
Original line number Diff line number Diff line
@@ -301,10 +301,27 @@ int register_cpu_cycle_counter_cb(struct cpu_cycle_counter_cb *cb) 
	return 0;

}



static void update_task_cpu_cycles(struct task_struct *p, int cpu)

/*

 * Assumes rq_lock is held and wallclock was recorded in the same critical

 * section as this function's invocation.

 */

static inline u64 read_cycle_counter(int cpu, u64 wallclock)

{

	struct rq *rq = cpu_rq(cpu);



	if (rq->last_cc_update != wallclock) {

		rq->cycles = cpu_cycle_counter_cb.get_cpu_cycle_counter(cpu);

		rq->last_cc_update = wallclock;

	}



	return rq->cycles;

}



static void update_task_cpu_cycles(struct task_struct *p, int cpu,

				   u64 wallclock)

{

	if (use_cycle_counter)

		p->cpu_cycles = cpu_cycle_counter_cb.get_cpu_cycle_counter(cpu);

		p->cpu_cycles = read_cycle_counter(cpu, wallclock);

}



void clear_ed_task(struct task_struct *p, struct rq *rq)
@@ -348,7 +365,7 @@ void sched_account_irqstart(int cpu, struct task_struct *curr, u64 wallclock) 
	if (is_idle_task(curr)) {

		/* We're here without rq->lock held, IRQ disabled */

		raw_spin_lock(&rq->lock);

		update_task_cpu_cycles(curr, cpu);

		update_task_cpu_cycles(curr, cpu, ktime_get_ns());

		raw_spin_unlock(&rq->lock);

	}

}
@@ -757,7 +774,7 @@ void fixup_busy_time(struct task_struct *p, int new_cpu) 
	update_task_ravg(p, task_rq(p), TASK_MIGRATE,

			 wallclock, 0);



	update_task_cpu_cycles(p, new_cpu);

	update_task_cpu_cycles(p, new_cpu, wallclock);



	/*

	 * When a task is migrating during the wakeup, adjust
@@ -1839,7 +1856,7 @@ update_task_rq_cpu_cycles(struct task_struct *p, struct rq *rq, int event, 
		return;

	}



	cur_cycles = cpu_cycle_counter_cb.get_cpu_cycle_counter(cpu);

	cur_cycles = read_cycle_counter(cpu, wallclock);



	/*

	 * If current task is idle task and irqtime == 0 CPU was
@@ -1904,7 +1921,7 @@ void update_task_ravg(struct task_struct *p, struct rq *rq, int event, 
	old_window_start = update_window_start(rq, wallclock, event);



	if (!p->ravg.mark_start) {

		update_task_cpu_cycles(p, cpu_of(rq));

		update_task_cpu_cycles(p, cpu_of(rq), wallclock);

		goto done;

	}
@@ -2035,7 +2052,7 @@ void mark_task_starting(struct task_struct *p) 
	p->ravg.mark_start = p->last_wake_ts = wallclock;

	p->last_enqueued_ts = wallclock;

	p->last_switch_out_ts = 0;

	update_task_cpu_cycles(p, cpu_of(rq));

	update_task_cpu_cycles(p, cpu_of(rq), wallclock);

}



static cpumask_t all_cluster_cpus = CPU_MASK_NONE;
@@ -3255,6 +3272,8 @@ void walt_sched_init(struct rq *rq) 
	rq->curr_table = 0;

	rq->prev_top = 0;

	rq->curr_top = 0;

	rq->last_cc_update = 0;

	rq->cycles = 0;

	for (j = 0; j < NUM_TRACKED_WINDOWS; j++) {

		memset(&rq->load_subs[j], 0,

				sizeof(struct load_subtractions));