Re: [RFC V2 2/2] sched: idle: IRQ based next prediction for idle period

[Nicolas Pitre]

On Wed, 20 Jan 2016, Daniel Lezcano wrote:

> Many IRQs are quiet most of the time, or they tend to come in bursts of
> fairly equal time intervals within each burst. It is therefore possible
> to detect those IRQs with stable intervals and guestimate when the next
> IRQ event is most likely to happen.
> 
> Examples of such IRQs may include audio related IRQs where the FIFO size
> and/or DMA descriptor size with the sample rate create stable intervals,
> block devices during large data transfers, etc.  Even network streaming
> of multimedia content creates patterns of periodic network interface IRQs
> in some cases.
> 
> This patch adds code to track the mean interval and variance for each IRQ
> over a window of time intervals between IRQ events. Those statistics can
> be used to assist cpuidle in selecting the most appropriate sleep state
> by predicting the most likely time for the next interrupt.
> 
> Because the stats are gathered in interrupt context, the core computation
> is as light as possible.
> 
> Signed-off-by: Daniel Lezcano <daniel.lezcano@xxxxxxxxxx>
> ---
>  drivers/cpuidle/Kconfig   |   9 +
>  kernel/sched/Makefile     |   1 +
>  kernel/sched/idle-sched.c | 529 ++++++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 539 insertions(+)
>  create mode 100644 kernel/sched/idle-sched.c
> 
> diff --git a/drivers/cpuidle/Kconfig b/drivers/cpuidle/Kconfig
> index 8c7930b..a606106 100644
> --- a/drivers/cpuidle/Kconfig
> +++ b/drivers/cpuidle/Kconfig
> @@ -25,6 +25,15 @@ config CPU_IDLE_GOV_MENU
>  	bool "Menu governor (for tickless system)"
>  	default y
>  
> +config CPU_IDLE_GOV_SCHED
> +	bool "Sched idle governor"
> +	select IRQ_TIMINGS
> +	help
> +	  Enables an irq timings tracking mechanism to track the wakeup sources
> +	  of the platform.
> +
> +	  If you are unsure, it is safe to say N.
> +
>  config DT_IDLE_STATES
>  	bool
>  
> diff --git a/kernel/sched/Makefile b/kernel/sched/Makefile
> index 6768797..f7d5a35 100644
> --- a/kernel/sched/Makefile
> +++ b/kernel/sched/Makefile
> @@ -19,3 +19,4 @@ obj-$(CONFIG_SCHED_AUTOGROUP) += auto_group.o
>  obj-$(CONFIG_SCHEDSTATS) += stats.o
>  obj-$(CONFIG_SCHED_DEBUG) += debug.o
>  obj-$(CONFIG_CGROUP_CPUACCT) += cpuacct.o
> +obj-$(CONFIG_CPU_IDLE_GOV_SCHED) += idle-sched.o
> diff --git a/kernel/sched/idle-sched.c b/kernel/sched/idle-sched.c
> new file mode 100644
> index 0000000..c2b8568
> --- /dev/null
> +++ b/kernel/sched/idle-sched.c
> @@ -0,0 +1,529 @@
> +/*
> + *  Copyright (C) 2016 Linaro Ltd, Daniel Lezcano <daniel.lezcano@xxxxxxxxxx>
> + *                                 Nicolas Pitre <nicolas.pitre@xxxxxxxxxx>
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + */
> +#include <linux/cpuidle.h>
> +#include <linux/interrupt.h>
> +#include <linux/irqdesc.h>
> +#include <linux/ktime.h>
> +#include <linux/slab.h>
> +#include <linux/tick.h>
> +#include <linux/time64.h>
> +
> +/*
> + * Define the number of samples over which the average and variance
> + * are computed. A power of 2 is preferred so to let the compiler
> + * optimize divisions by that number with simple arithmetic shifts.
> + */
> +#define STATS_NR_VALUES 4
> +
> +/**
> + * struct stats - internal structure to encapsulate stats informations
> + *
> + * @sum: sum of the values
> + * @values: array of values to do stats on
> + * @w_ptr: current buffer pointer
> + */
> +struct stats {
> +	u64           sum;                     /* sum of values */
> +	u32           values[STATS_NR_VALUES]; /* array of values */
> +	unsigned char w_ptr;                   /* current window pointer */

Why did you change this from an unsigned int?

This won't provide any memory space saving given that the structure has 
to be padded up to the next 64-bit boundary.

> +};
> +
> +/**
> + * struct wakeup - internal structure describing a source of wakeup
> + *
> + * @stats: the stats structure on the different event intervals
> + * @timestamp: latest update timestamp
> + */
> +struct wakeup {
> +	struct stats stats;
> +	ktime_t timestamp;
> +};
> +
> +/*
> + * Per cpu and irq statistics. Each cpu receives interrupts and those
> + * ones can be distributed following an irq chip specific
> + * algorithm. Random irq distribution is the worst case to predict
> + * interruption behavior but usually that does not happen or could be
> + * fixed from userspace by setting the irq affinity.
> + */
> +static DEFINE_PER_CPU(struct wakeup, *wakeups[NR_IRQS]);
> +
> +static DECLARE_BITMAP(enabled_irq, NR_IRQS);
> +
> +/**
> + * stats_add - add a new value in the statistic structure
> + *
> + * @s: the statistic structure
> + * @value: the new value to be added
> + *
> + * Adds the value to the array, if the array is full, the oldest value
> + * is replaced.
> + */
> +static void stats_add(struct stats *s, u32 value)
> +{
> +	/*
> +	 * This is a circular buffer, so the oldest value is the next
> +	 * one in the buffer. Let's compute the next pointer to
> +	 * retrieve the oldest value and re-use it to update the w_ptr
> +	 * after adding the new value.
> +	 */
> +	s->w_ptr = (s->w_ptr + 1) % STATS_NR_VALUES;
> +
> +	/*
> +	 * Remove the oldest value from the summing. If this is the
> +	 * first time we go through this array slot, the previous
> +	 * value will be zero and we won't substract anything from the
> +	 * current sum. Hence this code relies on a zero-ed stat
> +	 * structure at init time via memset or kzalloc.
> +	 */
> +	s->sum -= s->values[s->w_ptr];
> +	s->values[s->w_ptr] = value;
> +
> +	/*
> +	 * In order to reduce the overhead and to prevent value
> +	 * derivation due to the integer computation, we just sum the
> +	 * value and do the division when the average and the variance
> +	 * are requested.
> +	 */
> +	s->sum += value;
> +}
> +
> +/**
> + * stats_reset - reset the stats
> + *
> + * @s: the statistic structure
> + *
> + * Reset the statistics and reset the values
> + */
> +static inline void stats_reset(struct stats *s)
> +{
> +	memset(s, 0, sizeof(*s));
> +}
> +
> +/**
> + * stats_mean - compute the average
> + *
> + * @s: the statistics structure
> + *
> + * Returns an u32 corresponding to the mean value, or zero if there is
> + * no data
> + */
> +static inline u32 stats_mean(struct stats *s)
> +{
> +	/*
> +	 * gcc is smart enough to convert to a bits shift when the
> +	 * divisor is constant and multiple of 2^x.
> +	 *
> +	 * The number of values could have not reached STATS_NR_VALUES
> +	 * yet, but we can consider it acceptable as the situation is
> +	 * only at the beginning of the burst of irqs.
> +	 */
> +	return s->sum / STATS_NR_VALUES;
> +}
> +
> +/**
> + * stats_variance - compute the variance
> + *
> + * @s: the statistic structure
> + *
> + * Returns an u64 corresponding to the variance, or zero if there is
> + * no data
> + */
> +static u64 stats_variance(struct stats *s, u32 mean)
> +{
> +	int i;
> +	u64 variance = 0;
> +
> +	/*
> +	 * The variance is the sum of the squared difference to the
> +	 * average divided by the number of elements.
> +	 */
> +	for (i = 0; i < STATS_NR_VALUES; i++) {
> +		s64 diff = s->values[i] - mean;
> +		variance += (u64)diff * diff;
> +	}

This is completely wrong.  Even more wrong than it used to be.  I must
have expressed myself badly about this last time.

To avoid any confusion, here's what the code should be:

	int i;
	u64 variance = 0;

	for (i = 0; i < STATS_NR_VALUES; i++) {
		s32 diff = s->values[i] - mean;
		variance += (s64)diff * diff;
	}

	[...]

> +
> +	return variance / STATS_NR_VALUES;
> +}
> +
> +/**
> + * sched_idle_irq - irq timestamp callback
> + *
> + * @irq: the irq number
> + * @timestamp: when the interrupt occured
> + * @dev_id: device id for shared interrupt (not yet used)
> + *
> + * Interrupt callback called when an interrupt happens. This function
> + * is critical as it is called under an interrupt section: minimum
> + * operations as possible are done here:
> + */
> +static void sched_irq_timing_handler(unsigned int irq, ktime_t timestamp, void *dev_id)
> +{
> +	u32 diff;
> +	unsigned int cpu = raw_smp_processor_id();
> +	struct wakeup *w = per_cpu(wakeups[irq], cpu);
> +
> +	/*
> +	 * It is the first time the interrupt occurs of the series, we
> +	 * can't do any stats as we don't have an interval, just store
> +	 * the timestamp and exit.
> +	 */
> +	if (ktime_equal(w->timestamp, ktime_set(0, 0))) {
> +		w->timestamp = timestamp;
> +		return;
> +	}
> +
> +	/*
> +	 * Microsec resolution is enough for our purpose.
> +	 */
> +	diff = ktime_us_delta(timestamp, w->timestamp);
> +	w->timestamp = timestamp;
> +
> +	/*
> +	 * There is no point attempting predictions on interrupts more
> +	 * than ~1 second apart. This has no benefit for sleep state
> +	 * selection and increases the risk of overflowing our variance
> +	 * computation. Reset all stats in that case.
> +	 */
> +	if (diff > (1 << 20)) {

You could use the USEC_PER_SEC constant here. It is already widely used 
and would make the code even more obvious.

> +		stats_reset(&w->stats);
> +		return;
> +	}
> +
> +	stats_add(&w->stats, diff);
> +}
> +
> +static ktime_t next_irq_event(void)
> +{
> +	unsigned int irq, cpu = raw_smp_processor_id();
> +	ktime_t diff, next, min = ktime_set(KTIME_SEC_MAX, 0);
> +	ktime_t now = ktime_get();
> +	struct wakeup *w;
> +	u32 interval, mean;
> +	u64 variance;
> +
> +	/*
> +	 * Lookup the interrupt array for this cpu and search for the
> +	 * earlier expected interruption.
> +	 */
> +	for (irq = 0; irq < NR_IRQS; irq = find_next_bit(enabled_irq, NR_IRQS, irq)) {
> +
> +		w = per_cpu(wakeups[irq], cpu);
> +
> +		/*
> +		 * The interrupt was not setup as a source of a wakeup
> +		 * or the wakeup source is not considered at this
> +		 * moment stable enough to do a prediction.
> +		 */
> +		if (!w)
> +			continue;
> +
> +		/*
> +		 * No statistics available yet.
> +		 */
> +		if (ktime_equal(w->timestamp, ktime_set(0, 0)))
> +			continue;
> +
> +		diff = ktime_sub(now, w->timestamp);
> +
> +		/*
> +		 * There is no point attempting predictions on interrupts more
> +		 * than 1 second apart. This has no benefit for sleep state
> +		 * selection and increases the risk of overflowing our variance
> +		 * computation. Reset all stats in that case.
> +		 */

This comment is wrong.  It is relevant in sched_irq_timing_handler() but 
not here.  Instead this should be something like:

		/*
		 * This interrupt last triggered more than a second ago.
		 * It is definitely not predictable for our purpose anymore.
		 */

> +		if (unlikely(ktime_after(diff, ktime_set(1, 0)))) {
> +			stats_reset(&w->stats);
> +			continue;
> +		}
> +
> +		/*
> +		 * If the mean value is null, just ignore this wakeup
> +		 * source.
> +		 */
> +		mean = stats_mean(&w->stats);
> +		if (!mean)
> +			continue;
> +
> +		variance = stats_variance(&w->stats, mean);
> +		/*
> +		 * We want to check the last interval is:
> +		 *
> +		 *  mean - stddev < interval < mean + stddev
> +		 *
> +		 * That simplifies to:
> +		 *
> +		 * -stddev < interval - mean < stddev
> +		 *
> +		 * abs(interval - mean) < stddev
> +		 *
> +		 * The standard deviation is the sqrt of the variance:
> +		 *
> +		 * abs(interval - mean) < sqrt(variance)
> +		 *
> +		 * and we want to prevent to do an sqrt, so we square
> +		 * the equation:
> +		 *
> +		 * (interval - mean)^2 < variance
> +		 *
> +		 * So if the latest value of the stats complies with
> +		 * this condition, then the wakeup source is
> +		 * considered predictable and can be used to predict
> +		 * the next event.
> +		 */
> +		interval = w->stats.values[w->stats.w_ptr];
> +		if ((u64)((interval - mean) * (interval - mean)) > variance)

s/u64/s64/ please.

> +			continue;
> +
> +		/*
> +		 * Let's compute the next event: the wakeup source is
> +		 * considered predictable, we add the average interval
> +		 * time added to the latest interruption event time.
> +		 */
> +		next = ktime_add_us(w->timestamp, stats_mean(&w->stats));
> +
> +		/*
> +		 * If the interrupt is supposed to happen before the
> +		 * minimum time, then it becomes the minimum.
> +		 */
> +		if (ktime_before(next, min))
> +			min = next;
> +	}
> +
> +	/*
> +	 * At this point, we have our prediction but the caller is
> +	 * expecting the remaining time before the next event, so
> +	 * compute the expected sleep length.
> +	 */
> +	diff = ktime_sub(min, now);
> +
> +	/*
> +	 * The result could be negative for different reasons:
> +	 *  - the prediction is incorrect
> +	 *  - the prediction was too near now and expired while we were
> +	 *    in this function
> +	 *
> +	 * In both cases, we return KTIME_MAX as a failure to do a
> +	 * prediction
> +	 */
> +	if (ktime_compare(diff, ktime_set(0, 0)) <= 0)
> +		return ktime_set(KTIME_SEC_MAX, 0);
> +
> +	return diff;
> +}
> +
> +/**
> + * sched_idle_next_wakeup - Predict the next wakeup on the current cpu
> + *
> + * The next event on the cpu is based on a statistic approach of the
> + * interrupt events and the timer deterministic value. From the timer
> + * or the irqs, we return the one expected to occur first.
> + *
> + * Returns the expected remaining idle time before being woken up by
> + * an interruption.
> + */
> +s64 sched_idle_next_wakeup(void)
> +{
> +	s64 next_timer = ktime_to_us(tick_nohz_get_sleep_length());
> +	s64 next_irq = ktime_to_us(next_irq_event());
> +
> +	return min(next_irq, next_timer);
> +}
> +
> +/**
> + * sched_idle - go to idle for a specified amount of time
> + *
> + * @duration: the idle duration time
> + * @latency: the latency constraint
> + *
> + * Returns 0 on success, < 0 otherwise.
> + */
> +int sched_idle(s64 duration, unsigned int latency)
> +{
> +	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
> +	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);
> +	struct cpuidle_state_usage *su;
> +	struct cpuidle_state *s;
> +	int i, ret = 0, index = -1;
> +
> +	rcu_idle_enter();
> +
> +	/*
> +	 * No cpuidle driver is available, let's use the default arch
> +	 * idle function.
> +	 */
> +	if (cpuidle_not_available(drv, dev))
> +		goto default_idle;
> +
> +	/*
> +	 * Find the idle state with the lowest power while satisfying
> +	 * our constraints. We will save energy if the duration of the
> +	 * idle time is bigger than the target residency which is the
> +	 * break even point. The choice will be modulated by the
> +	 * latency.
> +	 */
> +	for (i = 0; i < drv->state_count; i++) {
> +
> +		s = &drv->states[i];
> +
> +		su = &dev->states_usage[i];
> +
> +		if (s->disabled || su->disable)
> +			continue;
> +		if (s->target_residency > duration)
> +			continue;
> +		if (s->exit_latency > latency)
> +			continue;
> +
> +		index = i;
> +	}
> +
> +	/*
> +	 * The idle task must be scheduled, it is pointless to go to
> +	 * idle, just re-enable the interrupt and return.
> +	 */
> +	if (current_clr_polling_and_test()) {
> +		local_irq_enable();
> +		goto out;
> +	}
> +
> +	if (index < 0) {
> +		/*
> +		 * No idle callbacks fulfilled the constraints, jump
> +		 * to the default function like there wasn't any
> +		 * cpuidle driver.
> +		 */
> +		goto default_idle;
> +	} else {
> +		/*
> +		 * Enter the idle state previously returned by the
> +		 * governor decision.  This function will block until
> +		 * an interrupt occurs and will take care of
> +		 * re-enabling the local interrupts
> +		 */
> +		return cpuidle_enter(drv, dev, index);
> +	}
> +
> +default_idle:
> +	default_idle_call();
> +out:
> +	rcu_idle_exit();
> +	return ret;
> +}
> +
> +/**
> + * sched_irq_timing_remove - disable the tracking of the specified irq
> + *
> + * Clear the irq table slot to stop tracking the interrupt.
> + *
> + * @irq: the irq number to stop tracking
> + * @dev_id: the device id for shared irq
> + *
> + * This function will remove from the wakeup source prediction table.
> + */
> +static void sched_irq_timing_remove(unsigned int irq, void *dev_id)
> +{
> +	clear_bit(irq, enabled_irq);
> +}
> +
> +/**
> + * sched_irq_timing_setup - enable the tracking of the specified irq
> + *
> + * Function is called with the corresponding irqdesc lock taken. It is
> + * not allowed to do any memory allocation or blocking call. Flag the
> + * irq table slot to be tracked in order to predict the next event.
> + *
> + * @irq: the interrupt numbe to be tracked
> + * @act: the new irq action to be set to this interrupt
> + *
> + * Returns zero on success, < 0 otherwise.
> + */
> +static int sched_irq_timing_setup(unsigned int irq, struct irqaction *act)
> +{
> +	/*
> +	 * No interrupt set for this descriptor or related to a timer.
> +	 * Timers are deterministic, so no need to try to do any
> +	 * prediction on them. No error for both cases, we are just not
> +	 * interested.
> +	 */
> +	if (!(act->flags & __IRQF_TIMER))
> +		return 0;
> +
> +	set_bit(irq, enabled_irq);
> +
> +	return 0;
> +}
> +
> +/**
> + * sched_irq_timing_free - free memory previously allocated
> + *
> + * @irq: the interrupt number
> + */
> +static void sched_irq_timing_free(unsigned int irq)
> +{
> +	struct wakeup *w;
> +	int cpu;
> +
> +	for_each_possible_cpu(cpu) {
> +
> +		w = per_cpu(wakeups[irq], cpu);
> +		if (!w)
> +			continue;
> +
> +		per_cpu(wakeups[irq], cpu) = NULL;
> +		kfree(w);
> +	}
> +}
> +
> +/**
> + * sched_irq_timing_alloc - allocates memory for irq tracking
> + *
> + * Allocates the memory to track the specified irq.
> + *
> + * @irq: the interrupt number
> + *
> + * Returns 0 on success, -ENOMEM on error.
> + */
> +static int sched_irq_timing_alloc(unsigned int irq)
> +{
> +	struct wakeup *w;
> +	int cpu, ret = -ENOMEM;
> +
> +	/*
> +	 * Allocates the wakeup structure and the stats structure. As
> +	 * the interrupt can occur on any cpu, allocate the wakeup
> +	 * structure per cpu basis.
> +	 */
> +	for_each_possible_cpu(cpu) {
> +
> +		w = kzalloc(sizeof(*w), GFP_KERNEL);
> +		if (!w)
> +			goto out;
> +
> +		per_cpu(wakeups[irq], cpu) = w;
> +	}
> +
> +	ret = 0;
> +out:
> +	if (ret)
> +		sched_irq_timing_free(irq);
> +
> +	return ret;
> +}
> +
> +static struct irqtimings_ops irqt_ops = {
> +	.alloc   = sched_irq_timing_alloc,
> +	.free    = sched_irq_timing_free,
> +	.setup   = sched_irq_timing_setup,
> +	.remove  = sched_irq_timing_remove,
> +	.handler = sched_irq_timing_handler,
> +};
> +
> +DECLARE_IRQ_TIMINGS(&irqt_ops);
> -- 
> 1.9.1
> 
>