diff options
50 files changed, 3499 insertions, 83 deletions
diff --git a/Documentation/devicetree/bindings/arm/pmu.txt b/Documentation/devicetree/bindings/arm/pmu.txt index 343781b9f246..4ce82d045a6b 100644 --- a/Documentation/devicetree/bindings/arm/pmu.txt +++ b/Documentation/devicetree/bindings/arm/pmu.txt @@ -16,6 +16,9 @@ Required properties: "arm,arm1176-pmu" "arm,arm1136-pmu" - interrupts : 1 combined interrupt or 1 per core. +- cluster : a phandle to the cluster to which it belongs + If there are more than one cluster with same CPU type + then there should be separate PMU nodes per cluster. Example: diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index 2fe6e767b3d6..ba7daaa688f5 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -1240,6 +1240,15 @@ bytes respectively. Such letter suffixes can also be entirely omitted. See comment before ip2_setup() in drivers/char/ip2/ip2base.c. + irqaffinity= [SMP] Set the default irq affinity mask + Format: + <cpu number>,...,<cpu number> + or + <cpu number>-<cpu number> + (must be a positive range in ascending order) + or a mixture + <cpu number>,...,<cpu number>-<cpu number> + irqfixup [HW] When an interrupt is not handled search all handlers for it. Intended to get systems with badly broken @@ -3341,6 +3350,21 @@ bytes respectively. Such letter suffixes can also be entirely omitted. that this also can be controlled per-workqueue for workqueues visible under /sys/bus/workqueue/. + workqueue.power_efficient + Per-cpu workqueues are generally preferred because + they show better performance thanks to cache + locality; unfortunately, per-cpu workqueues tend to + be more power hungry than unbound workqueues. + + Enabling this makes the per-cpu workqueues which + were observed to contribute significantly to power + consumption unbound, leading to measurably lower + power usage at the cost of small performance + overhead. + + The default value of this parameter is determined by + the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. + x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of default x2apic cluster mode on platforms supporting x2apic. diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index c24b364bb8cb..effe59d47749 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -1494,6 +1494,90 @@ config SCHED_SMT MultiThreading at a cost of slightly increased overhead in some places. If unsure say N here. +config DISABLE_CPU_SCHED_DOMAIN_BALANCE + bool "(EXPERIMENTAL) Disable CPU level scheduler load-balancing" + help + Disables scheduler load-balancing at CPU sched domain level. + +config SCHED_HMP + bool "(EXPERIMENTAL) Heterogenous multiprocessor scheduling" + depends on DISABLE_CPU_SCHED_DOMAIN_BALANCE && SCHED_MC && FAIR_GROUP_SCHED && !SCHED_AUTOGROUP + help + Experimental scheduler optimizations for heterogeneous platforms. + Attempts to introspectively select task affinity to optimize power + and performance. Basic support for multiple (>2) cpu types is in place, + but it has only been tested with two types of cpus. + There is currently no support for migration of task groups, hence + !SCHED_AUTOGROUP. Furthermore, normal load-balancing must be disabled + between cpus of different type (DISABLE_CPU_SCHED_DOMAIN_BALANCE). + +config SCHED_HMP_PRIO_FILTER + bool "(EXPERIMENTAL) Filter HMP migrations by task priority" + depends on SCHED_HMP + help + Enables task priority based HMP migration filter. Any task with + a NICE value above the threshold will always be on low-power cpus + with less compute capacity. + +config SCHED_HMP_PRIO_FILTER_VAL + int "NICE priority threshold" + default 5 + depends on SCHED_HMP_PRIO_FILTER + +config HMP_FAST_CPU_MASK + string "HMP scheduler fast CPU mask" + depends on SCHED_HMP + help + Leave empty to use device tree information. + Specify the cpuids of the fast CPUs in the system as a list string, + e.g. cpuid 0+1 should be specified as 0-1. + +config HMP_SLOW_CPU_MASK + string "HMP scheduler slow CPU mask" + depends on SCHED_HMP + help + Leave empty to use device tree information. + Specify the cpuids of the slow CPUs in the system as a list string, + e.g. cpuid 0+1 should be specified as 0-1. + +config HMP_VARIABLE_SCALE + bool "Allows changing the load tracking scale through sysfs" + depends on SCHED_HMP + help + When turned on, this option exports the thresholds and load average + period value for the load tracking patches through sysfs. + The values can be modified to change the rate of load accumulation + and the thresholds used for HMP migration. + The load_avg_period_ms is the time in ms to reach a load average of + 0.5 for an idle task of 0 load average ratio that start a busy loop. + The up_threshold and down_threshold is the value to go to a faster + CPU or to go back to a slower cpu. + The {up,down}_threshold are devided by 1024 before being compared + to the load average. + For examples, with load_avg_period_ms = 128 and up_threshold = 512, + a running task with a load of 0 will be migrated to a bigger CPU after + 128ms, because after 128ms its load_avg_ratio is 0.5 and the real + up_threshold is 0.5. + This patch has the same behavior as changing the Y of the load + average computation to + (1002/1024)^(LOAD_AVG_PERIOD/load_avg_period_ms) + but it remove intermadiate overflows in computation. + +config HMP_FREQUENCY_INVARIANT_SCALE + bool "(EXPERIMENTAL) Frequency-Invariant Tracked Load for HMP" + depends on HMP_VARIABLE_SCALE && CPU_FREQ + help + Scales the current load contribution in line with the frequency + of the CPU that the task was executed on. + In this version, we use a simple linear scale derived from the + maximum frequency reported by CPUFreq. + Restricting tracked load to be scaled by the CPU's frequency + represents the consumption of possible compute capacity + (rather than consumption of actual instantaneous capacity as + normal) and allows the HMP migration's simple threshold + migration strategy to interact more predictably with CPUFreq's + asynchronous compute capacity changes. + config HAVE_ARM_SCU bool help @@ -1528,6 +1612,24 @@ config BIG_LITTLE help This option enables support for the big.LITTLE architecture. +config BL_SWITCHER + bool "big.LITTLE switcher support" + depends on BIG_LITTLE && MCPM && HOTPLUG_CPU + select CPU_PM + select ARM_CPU_SUSPEND + help + The big.LITTLE "switcher" provides the core functionality to + transparently handle transition between a cluster of A15's + and a cluster of A7's in a big.LITTLE system. + +config BL_SWITCHER_DUMMY_IF + tristate "Simple big.LITTLE switcher user interface" + depends on BL_SWITCHER && DEBUG_KERNEL + help + This is a simple and dummy char dev interface to control + the big.LITTLE switcher core code. It is meant for + debugging purposes only. + choice prompt "Memory split" default VMSPLIT_3G diff --git a/arch/arm/common/Makefile b/arch/arm/common/Makefile index f27d6a7af57f..462cd580fc2d 100644 --- a/arch/arm/common/Makefile +++ b/arch/arm/common/Makefile @@ -14,6 +14,9 @@ obj-$(CONFIG_SHARP_SCOOP) += scoop.o obj-$(CONFIG_PCI_HOST_ITE8152) += it8152.o obj-$(CONFIG_ARM_TIMER_SP804) += timer-sp.o obj-$(CONFIG_MCPM) += mcpm_head.o mcpm_entry.o mcpm_platsmp.o vlock.o +obj-$(CONFIG_BL_SWITCHER) += bL_switcher.o +obj-$(CONFIG_BL_SWITCHER_DUMMY_IF) += bL_switcher_dummy_if.o + AFLAGS_mcpm_head.o := -march=armv7-a AFLAGS_vlock.o := -march=armv7-a CFLAGS_REMOVE_mcpm_entry.o = -pg diff --git a/arch/arm/common/bL_switcher.c b/arch/arm/common/bL_switcher.c new file mode 100644 index 000000000000..1883c5b3e3f5 --- /dev/null +++ b/arch/arm/common/bL_switcher.c @@ -0,0 +1,860 @@ +/* + * arch/arm/common/bL_switcher.c -- big.LITTLE cluster switcher core driver + * + * Created by: Nicolas Pitre, March 2012 + * Copyright: (C) 2012 Linaro Limited + * + * 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/atomic.h> +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/sched.h> +#include <linux/interrupt.h> +#include <linux/cpu_pm.h> +#include <linux/cpu.h> +#include <linux/cpumask.h> +#include <linux/kthread.h> +#include <linux/wait.h> +#include <linux/time.h> +#include <linux/clockchips.h> +#include <linux/hrtimer.h> +#include <linux/tick.h> +#include <linux/notifier.h> +#include <linux/mm.h> +#include <linux/mutex.h> +#include <linux/smp.h> +#include <linux/spinlock.h> +#include <linux/string.h> +#include <linux/sysfs.h> +#include <linux/irqchip/arm-gic.h> +#include <linux/moduleparam.h> + +#include <asm/smp_plat.h> +#include <asm/cacheflush.h> +#include <asm/cputype.h> +#include <asm/suspend.h> +#include <asm/mcpm.h> +#include <asm/bL_switcher.h> + +#define CREATE_TRACE_POINTS +#include <trace/events/power_cpu_migrate.h> + + +/* + * Use our own MPIDR accessors as the generic ones in asm/cputype.h have + * __attribute_const__ and we don't want the compiler to assume any + * constness here as the value _does_ change along some code paths. + */ + +static int read_mpidr(void) +{ + unsigned int id; + asm volatile ("mrc\tp15, 0, %0, c0, c0, 5" : "=r" (id)); + return id & MPIDR_HWID_BITMASK; +} + +/* + * Get a global nanosecond time stamp for tracing. + */ +static s64 get_ns(void) +{ + struct timespec ts; + getnstimeofday(&ts); + return timespec_to_ns(&ts); +} + +/* + * bL switcher core code. + */ + +static void bL_do_switch(void *_arg) +{ + unsigned ib_mpidr, ib_cpu, ib_cluster; + long volatile handshake, **handshake_ptr = _arg; + + pr_debug("%s\n", __func__); + + ib_mpidr = cpu_logical_map(smp_processor_id()); + ib_cpu = MPIDR_AFFINITY_LEVEL(ib_mpidr, 0); + ib_cluster = MPIDR_AFFINITY_LEVEL(ib_mpidr, 1); + + /* Advertise our handshake location */ + if (handshake_ptr) { + handshake = 0; + *handshake_ptr = &handshake; + } else + handshake = -1; + + /* + * Our state has been saved at this point. Let's release our + * inbound CPU. + */ + mcpm_set_entry_vector(ib_cpu, ib_cluster, cpu_resume); + sev(); + + /* + * From this point, we must assume that our counterpart CPU might + * have taken over in its parallel world already, as if execution + * just returned from cpu_suspend(). It is therefore important to + * be very careful not to make any change the other guy is not + * expecting. This is why we need stack isolation. + * + * Fancy under cover tasks could be performed here. For now + * we have none. + */ + + /* + * Let's wait until our inbound is alive. + */ + while (!handshake) { + wfe(); + smp_mb(); + } + + /* Let's put ourself down. */ + mcpm_cpu_power_down(); + + /* should never get here */ + BUG(); +} + +/* + * Stack isolation. To ensure 'current' remains valid, we just use another + * piece of our thread's stack space which should be fairly lightly used. + * The selected area starts just above the thread_info structure located + * at the very bottom of the stack, aligned to a cache line, and indexed + * with the cluster number. + */ +#define STACK_SIZE 512 +extern void call_with_stack(void (*fn)(void *), void *arg, void *sp); +static int bL_switchpoint(unsigned long _arg) +{ + unsigned int mpidr = read_mpidr(); + unsigned int clusterid = MPIDR_AFFINITY_LEVEL(mpidr, 1); + void *stack = current_thread_info() + 1; + stack = PTR_ALIGN(stack, L1_CACHE_BYTES); + stack += clusterid * STACK_SIZE + STACK_SIZE; + call_with_stack(bL_do_switch, (void *)_arg, stack); + BUG(); +} + +/* + * Generic switcher interface + */ + +static unsigned int bL_gic_id[MAX_CPUS_PER_CLUSTER][MAX_NR_CLUSTERS]; +static int bL_switcher_cpu_pairing[NR_CPUS]; + +/* + * bL_switch_to - Switch to a specific cluster for the current CPU + * @new_cluster_id: the ID of the cluster to switch to. + * + * This function must be called on the CPU to be switched. + * Returns 0 on success, else a negative status code. + */ +static int bL_switch_to(unsigned int new_cluster_id) +{ + unsigned int mpidr, this_cpu, that_cpu; + unsigned int ob_mpidr, ob_cpu, ob_cluster, ib_mpidr, ib_cpu, ib_cluster; + struct completion inbound_alive; + struct tick_device *tdev; + enum clock_event_mode tdev_mode; + long volatile *handshake_ptr; + int ipi_nr, ret; + + this_cpu = smp_processor_id(); + ob_mpidr = read_mpidr(); + ob_cpu = MPIDR_AFFINITY_LEVEL(ob_mpidr, 0); + ob_cluster = MPIDR_AFFINITY_LEVEL(ob_mpidr, 1); + BUG_ON(cpu_logical_map(this_cpu) != ob_mpidr); + + if (new_cluster_id == ob_cluster) + return 0; + + that_cpu = bL_switcher_cpu_pairing[this_cpu]; + ib_mpidr = cpu_logical_map(that_cpu); + ib_cpu = MPIDR_AFFINITY_LEVEL(ib_mpidr, 0); + ib_cluster = MPIDR_AFFINITY_LEVEL(ib_mpidr, 1); + + pr_debug("before switch: CPU %d MPIDR %#x -> %#x\n", + this_cpu, ob_mpidr, ib_mpidr); + + this_cpu = smp_processor_id(); + + /* Close the gate for our entry vectors */ + mcpm_set_entry_vector(ob_cpu, ob_cluster, NULL); + mcpm_set_entry_vector(ib_cpu, ib_cluster, NULL); + + /* Install our "inbound alive" notifier. */ + init_completion(&inbound_alive); + ipi_nr = register_ipi_completion(&inbound_alive, this_cpu); + ipi_nr |= ((1 << 16) << bL_gic_id[ob_cpu][ob_cluster]); + mcpm_set_early_poke(ib_cpu, ib_cluster, gic_get_sgir_physaddr(), ipi_nr); + + /* + * Let's wake up the inbound CPU now in case it requires some delay + * to come online, but leave it gated in our entry vector code. + */ + ret = mcpm_cpu_power_up(ib_cpu, ib_cluster); + if (ret) { + pr_err("%s: mcpm_cpu_power_up() returned %d\n", __func__, ret); + return ret; + } + + /* + * Raise a SGI on the inbound CPU to make sure it doesn't stall + * in a possible WFI, such as in bL_power_down(). + */ + gic_send_sgi(bL_gic_id[ib_cpu][ib_cluster], 0); + + /* + * Wait for the inbound to come up. This allows for other + * tasks to be scheduled in the mean time. + */ + wait_for_completion(&inbound_alive); + mcpm_set_early_poke(ib_cpu, ib_cluster, 0, 0); + + /* + * From this point we are entering the switch critical zone + * and can't sleep/schedule anymore. + */ + local_irq_disable(); + local_fiq_disable(); + trace_cpu_migrate_begin(get_ns(), ob_mpidr); + + /* redirect GIC's SGIs to our counterpart */ + gic_migrate_target(bL_gic_id[ib_cpu][ib_cluster]); + + tdev = tick_get_device(this_cpu); + if (tdev && !cpumask_equal(tdev->evtdev->cpumask, cpumask_of(this_cpu))) + tdev = NULL; + if (tdev) { + tdev_mode = tdev->evtdev->mode; + clockevents_set_mode(tdev->evtdev, CLOCK_EVT_MODE_SHUTDOWN); + } + + ret = cpu_pm_enter(); + + /* we can not tolerate errors at this point */ + if (ret) + panic("%s: cpu_pm_enter() returned %d\n", __func__, ret); + + /* + * Swap the physical CPUs in the logical map for this logical CPU. + * This must be flushed to RAM as the resume code + * needs to access it while the caches are still disabled. + */ + cpu_logical_map(this_cpu) = ib_mpidr; + cpu_logical_map(that_cpu) = ob_mpidr; + sync_cache_w(&cpu_logical_map(this_cpu)); + + /* Let's do the actual CPU switch. */ + ret = cpu_suspend((unsigned long)&handshake_ptr, bL_switchpoint); + if (ret > 0) + panic("%s: cpu_suspend() returned %d\n", __func__, ret); + + /* We are executing on the inbound CPU at this point */ + mpidr = read_mpidr(); + pr_debug("after switch: CPU %d MPIDR %#x\n", this_cpu, mpidr); + BUG_ON(mpidr != ib_mpidr); + + mcpm_cpu_powered_up(); + + ret = cpu_pm_exit(); + + if (tdev) { + clockevents_set_mode(tdev->evtdev, tdev_mode); + clockevents_program_event(tdev->evtdev, + tdev->evtdev->next_event, 1); + } + + trace_cpu_migrate_finish(get_ns(), ib_mpidr); + local_fiq_enable(); + local_irq_enable(); + + *handshake_ptr = 1; + dsb_sev(); + + if (ret) + pr_err("%s exiting with error %d\n", __func__, ret); + return ret; +} + +struct bL_thread { + spinlock_t lock; + struct task_struct *task; + wait_queue_head_t wq; + int wanted_cluster; + struct completion started; + bL_switch_completion_handler completer; + void *completer_cookie; +}; + +static struct bL_thread bL_threads[NR_CPUS]; + +static int bL_switcher_thread(void *arg) +{ + struct bL_thread *t = arg; + struct sched_param param = { .sched_priority = 1 }; + int cluster; + bL_switch_completion_handler completer; + void *completer_cookie; + + sched_setscheduler_nocheck(current, SCHED_FIFO, ¶m); + complete(&t->started); + + do { + if (signal_pending(current)) + flush_signals(current); + wait_event_interruptible(t->wq, + t->wanted_cluster != -1 || + kthread_should_stop()); + + spin_lock(&t->lock); + cluster = t->wanted_cluster; + completer = t->completer; + completer_cookie = t->completer_cookie; + t->wanted_cluster = -1; + t->completer = NULL; + spin_unlock(&t->lock); + + if (cluster != -1) { + bL_switch_to(cluster); + + if (completer) + completer(completer_cookie); + } + } while (!kthread_should_stop()); + + return 0; +} + +static struct task_struct * bL_switcher_thread_create(int cpu, void *arg) +{ + struct task_struct *task; + + task = kthread_create_on_node(bL_switcher_thread, arg, + cpu_to_node(cpu), "kswitcher_%d", cpu); + if (!IS_ERR(task)) { + kthread_bind(task, cpu); + wake_up_process(task); + } else + pr_err("%s failed for CPU %d\n", __func__, cpu); + return task; +} + +/* + * bL_switch_request_cb - Switch to a specific cluster for the given CPU, + * with completion notification via a callback + * + * @cpu: the CPU to switch + * @new_cluster_id: the ID of the cluster to switch to. + * @completer: switch completion callback. if non-NULL, + * @completer(@completer_cookie) will be called on completion of + * the switch, in non-atomic context. + * @completer_cookie: opaque context argument for @completer. + * + * This function causes a cluster switch on the given CPU by waking up + * the appropriate switcher thread. This function may or may not return + * before the switch has occurred. + * + * If a @completer callback function is supplied, it will be called when + * the switch is complete. This can be used to determine asynchronously + * when the switch is complete, regardless of when bL_switch_request() + * returns. When @completer is supplied, no new switch request is permitted + * for the affected CPU until after the switch is complete, and @completer + * has returned. + */ +int bL_switch_request_cb(unsigned int cpu, unsigned int new_cluster_id, + bL_switch_completion_handler completer, + void *completer_cookie) +{ + struct bL_thread *t; + + if (cpu >= ARRAY_SIZE(bL_threads)) { + pr_err("%s: cpu %d out of bounds\n", __func__, cpu); + return -EINVAL; + } + + t = &bL_threads[cpu]; + + if (IS_ERR(t->task)) + return PTR_ERR(t->task); + if (!t->task) + return -ESRCH; + + spin_lock(&t->lock); + if (t->completer) { + spin_unlock(&t->lock); + return -EBUSY; + } + t->completer = completer; + t->completer_cookie = completer_cookie; + t->wanted_cluster = new_cluster_id; + spin_unlock(&t->lock); + wake_up(&t->wq); + return 0; +} + +EXPORT_SYMBOL_GPL(bL_switch_request_cb); + +/* + * Detach an outstanding switch request. + * + * The switcher will continue with the switch request in the background, + * but the completer function will not be called. + * + * This may be necessary if the completer is in a kernel module which is + * about to be unloaded. + */ +void bL_switch_request_detach(unsigned int cpu, + bL_switch_completion_handler completer) +{ + struct bL_thread *t; + + if (cpu >= ARRAY_SIZE(bL_threads)) { + pr_err("%s: cpu %d out of bounds\n", __func__, cpu); + return; + } + + t = &bL_threads[cpu]; + + if (IS_ERR(t->task) || !t->task) + return; + + spin_lock(&t->lock); + if (t->completer == completer) + t->completer = NULL; + spin_unlock(&t->lock); +} + +EXPORT_SYMBOL_GPL(bL_switch_request_detach); + +/* + * Activation and configuration code. + */ + +static DEFINE_MUTEX(bL_switcher_activation_lock); +static BLOCKING_NOTIFIER_HEAD(bL_activation_notifier); +static unsigned int bL_switcher_active; +static unsigned int bL_switcher_cpu_original_cluster[NR_CPUS]; +static cpumask_t bL_switcher_removed_logical_cpus; + +int bL_switcher_register_notifier(struct notifier_block *nb) +{ + return blocking_notifier_chain_register(&bL_activation_notifier, nb); +} +EXPORT_SYMBOL_GPL(bL_switcher_register_notifier); + +int bL_switcher_unregister_notifier(struct notifier_block *nb) +{ + return blocking_notifier_chain_unregister(&bL_activation_notifier, nb); +} +EXPORT_SYMBOL_GPL(bL_switcher_unregister_notifier); + +static int bL_activation_notify(unsigned long val) +{ + int ret; + + ret = blocking_notifier_call_chain(&bL_activation_notifier, val, NULL); + if (ret & NOTIFY_STOP_MASK) + pr_err("%s: notifier chain failed with status 0x%x\n", + __func__, ret); + return notifier_to_errno(ret); +} + +static void bL_switcher_restore_cpus(void) +{ + int i; + + for_each_cpu(i, &bL_switcher_removed_logical_cpus) + cpu_up(i); +} + +static int bL_switcher_halve_cpus(void) +{ + int i, j, cluster_0, gic_id, ret; + unsigned int cpu, cluster, mask; + cpumask_t available_cpus; + + /* First pass to validate what we have */ + mask = 0; + for_each_online_cpu(i) { + cpu = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 0); + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 1); + if (cluster >= 2) { + pr_err("%s: only dual cluster systems are supported\n", __func__); + return -EINVAL; + } + if (WARN_ON(cpu >= MAX_CPUS_PER_CLUSTER)) + return -EINVAL; + mask |= (1 << cluster); + } + if (mask != 3) { + pr_err("%s: no CPU pairing possible\n", __func__); + return -EINVAL; + } + + /* + * Now let's do the pairing. We match each CPU with another CPU + * from a different cluster. To get a uniform scheduling behavior + * without fiddling with CPU topology and compute capacity data, + * we'll use logical CPUs initially belonging to the same cluster. + */ + memset(bL_switcher_cpu_pairing, -1, sizeof(bL_switcher_cpu_pairing)); + cpumask_copy(&available_cpus, cpu_online_mask); + cluster_0 = -1; + for_each_cpu(i, &available_cpus) { + int match = -1; + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 1); + if (cluster_0 == -1) + cluster_0 = cluster; + if (cluster != cluster_0) + continue; + cpumask_clear_cpu(i, &available_cpus); + for_each_cpu(j, &available_cpus) { + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(j), 1); + /* + * Let's remember the last match to create "odd" + * pairing on purpose in order for other code not + * to assume any relation between physical and + * logical CPU numbers. + */ + if (cluster != cluster_0) + match = j; + } + if (match != -1) { + bL_switcher_cpu_pairing[i] = match; + cpumask_clear_cpu(match, &available_cpus); + pr_info("CPU%d paired with CPU%d\n", i, match); + } + } + + /* + * Now we disable the unwanted CPUs i.e. everything that has no + * pairing information (that includes the pairing counterparts). + */ + cpumask_clear(&bL_switcher_removed_logical_cpus); + for_each_online_cpu(i) { + cpu = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 0); + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 1); + + /* Let's take note of the GIC ID for this CPU */ + gic_id = gic_get_cpu_id(i); + if (gic_id < 0) { + pr_err("%s: bad GIC ID for CPU %d\n", __func__, i); + bL_switcher_restore_cpus(); + return -EINVAL; + } + bL_gic_id[cpu][cluster] = gic_id; + pr_info("GIC ID for CPU %u cluster %u is %u\n", + cpu, cluster, gic_id); + + if (bL_switcher_cpu_pairing[i] != -1) { + bL_switcher_cpu_original_cluster[i] = cluster; + continue; + } + + ret = cpu_down(i); + if (ret) { + bL_switcher_restore_cpus(); + return ret; + } + cpumask_set_cpu(i, &bL_switcher_removed_logical_cpus); + } + + return 0; +} + +/* Determine the logical CPU a given physical CPU is grouped on. */ +int bL_switcher_get_logical_index(u32 mpidr) +{ + int cpu; + + if (!bL_switcher_active) + return -EUNATCH; + + mpidr &= MPIDR_HWID_BITMASK; + for_each_online_cpu(cpu) { + int pairing = bL_switcher_cpu_pairing[cpu]; + if (pairing == -1) + continue; + if ((mpidr == cpu_logical_map(cpu)) || + (mpidr == cpu_logical_map(pairing))) + return cpu; + } + return -EINVAL; +} + +static void bL_switcher_trace_trigger_cpu(void *__always_unused info) +{ + trace_cpu_migrate_current(get_ns(), read_mpidr()); +} + +int bL_switcher_trace_trigger(void) +{ + int ret; + + preempt_disable(); + + bL_switcher_trace_trigger_cpu(NULL); + ret = smp_call_function(bL_switcher_trace_trigger_cpu, NULL, true); + + preempt_enable(); + + return ret; +} +EXPORT_SYMBOL_GPL(bL_switcher_trace_trigger); + +static int bL_switcher_enable(void) +{ + int cpu, ret; + + mutex_lock(&bL_switcher_activation_lock); + cpu_hotplug_driver_lock(); + if (bL_switcher_active) { + cpu_hotplug_driver_unlock(); + mutex_unlock(&bL_switcher_activation_lock); + return 0; + } + + pr_info("big.LITTLE switcher initializing\n"); + + ret = bL_activation_notify(BL_NOTIFY_PRE_ENABLE); + if (ret) + goto error; + + ret = bL_switcher_halve_cpus(); + if (ret) + goto error; + + bL_switcher_trace_trigger(); + + for_each_online_cpu(cpu) { + struct bL_thread *t = &bL_threads[cpu]; + spin_lock_init(&t->lock); + init_waitqueue_head(&t->wq); + init_completion(&t->started); + t->wanted_cluster = -1; + t->task = bL_switcher_thread_create(cpu, t); + } + + bL_switcher_active = 1; + bL_activation_notify(BL_NOTIFY_POST_ENABLE); + pr_info("big.LITTLE switcher initialized\n"); + goto out; + +error: + pr_warning("big.LITTLE switcher initialization failed\n"); + bL_activation_notify(BL_NOTIFY_POST_DISABLE); + +out: + cpu_hotplug_driver_unlock(); + mutex_unlock(&bL_switcher_activation_lock); + return ret; +} + +#ifdef CONFIG_SYSFS + +static void bL_switcher_disable(void) +{ + unsigned int cpu, cluster; + struct bL_thread *t; + struct task_struct *task; + + mutex_lock(&bL_switcher_activation_lock); + cpu_hotplug_driver_lock(); + + if (!bL_switcher_active) + goto out; + + if (bL_activation_notify(BL_NOTIFY_PRE_DISABLE) != 0) { + bL_activation_notify(BL_NOTIFY_POST_ENABLE); + goto out; + } + + bL_switcher_active = 0; + + /* + * To deactivate the switcher, we must shut down the switcher + * threads to prevent any other requests from being accepted. + * Then, if the final cluster for given logical CPU is not the + * same as the original one, we'll recreate a switcher thread + * just for the purpose of switching the CPU back without any + * possibility for interference from external requests. + */ + for_each_online_cpu(cpu) { + t = &bL_threads[cpu]; + task = t->task; + t->task = NULL; + if (!task || IS_ERR(task)) + continue; + kthread_stop(task); + /* no more switch may happen on this CPU at this point */ + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(cpu), 1); + if (cluster == bL_switcher_cpu_original_cluster[cpu]) + continue; + init_completion(&t->started); + t->wanted_cluster = bL_switcher_cpu_original_cluster[cpu]; + task = bL_switcher_thread_create(cpu, t); + if (!IS_ERR(task)) { + wait_for_completion(&t->started); + kthread_stop(task); + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(cpu), 1); + if (cluster == bL_switcher_cpu_original_cluster[cpu]) + continue; + } + /* If execution gets here, we're in trouble. */ + pr_crit("%s: unable to restore original cluster for CPU %d\n", + __func__, cpu); + pr_crit("%s: CPU %d can't be restored\n", + __func__, bL_switcher_cpu_pairing[cpu]); + cpumask_clear_cpu(bL_switcher_cpu_pairing[cpu], + &bL_switcher_removed_logical_cpus); + } + + bL_switcher_restore_cpus(); + bL_switcher_trace_trigger(); + + bL_activation_notify(BL_NOTIFY_POST_DISABLE); + +out: + cpu_hotplug_driver_unlock(); + mutex_unlock(&bL_switcher_activation_lock); +} + +static ssize_t bL_switcher_active_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + return sprintf(buf, "%u\n", bL_switcher_active); +} + +static ssize_t bL_switcher_active_store(struct kobject *kobj, + struct kobj_attribute *attr, const char *buf, size_t count) +{ + int ret; + + switch (buf[0]) { + case '0': + bL_switcher_disable(); + ret = 0; + break; + case '1': + ret = bL_switcher_enable(); + break; + default: + ret = -EINVAL; + } + + return (ret >= 0) ? count : ret; +} + +static ssize_t bL_switcher_trace_trigger_store(struct kobject *kobj, + struct kobj_attribute *attr, const char *buf, size_t count) +{ + int ret = bL_switcher_trace_trigger(); + + return ret ? ret : count; +} + +static struct kobj_attribute bL_switcher_active_attr = + __ATTR(active, 0644, bL_switcher_active_show, bL_switcher_active_store); + +static struct kobj_attribute bL_switcher_trace_trigger_attr = + __ATTR(trace_trigger, 0200, NULL, bL_switcher_trace_trigger_store); + +static struct attribute *bL_switcher_attrs[] = { + &bL_switcher_active_attr.attr, + &bL_switcher_trace_trigger_attr.attr, + NULL, +}; + +static struct attribute_group bL_switcher_attr_group = { + .attrs = bL_switcher_attrs, +}; + +static struct kobject *bL_switcher_kobj; + +static int __init bL_switcher_sysfs_init(void) +{ + int ret; + + bL_switcher_kobj = kobject_create_and_add("bL_switcher", kernel_kobj); + if (!bL_switcher_kobj) + return -ENOMEM; + ret = sysfs_create_group(bL_switcher_kobj, &bL_switcher_attr_group); + if (ret) + kobject_put(bL_switcher_kobj); + return ret; +} + +#endif /* CONFIG_SYSFS */ + +bool bL_switcher_get_enabled(void) +{ + mutex_lock(&bL_switcher_activation_lock); + + return bL_switcher_active; +} +EXPORT_SYMBOL_GPL(bL_switcher_get_enabled); + +void bL_switcher_put_enabled(void) +{ + mutex_unlock(&bL_switcher_activation_lock); +} +EXPORT_SYMBOL_GPL(bL_switcher_put_enabled); + +/* + * Veto any CPU hotplug operation while the switcher is active. + * We're just not ready to deal with that given the trickery involved. + */ +static int bL_switcher_hotplug_callback(struct notifier_block *nfb, + unsigned long action, void *hcpu) +{ + switch (action) { + case CPU_UP_PREPARE: + case CPU_DOWN_PREPARE: + if (bL_switcher_active) + return NOTIFY_BAD; + } + return NOTIFY_DONE; +} + +static struct notifier_block bL_switcher_hotplug_notifier = + { &bL_switcher_hotplug_callback, NULL, 0 }; + +static bool no_bL_switcher; +core_param(no_bL_switcher, no_bL_switcher, bool, 0644); + +static int __init bL_switcher_init(void) +{ + int ret; + + if (MAX_NR_CLUSTERS != 2) { + pr_err("%s: only dual cluster systems are supported\n", __func__); + return -EINVAL; + } + + register_cpu_notifier(&bL_switcher_hotplug_notifier); + + if (!no_bL_switcher) { + ret = bL_switcher_enable(); + if (ret) + return ret; + } + +#ifdef CONFIG_SYSFS + ret = bL_switcher_sysfs_init(); + if (ret) + pr_err("%s: unable to create sysfs entry\n", __func__); +#endif + + return 0; +} + +late_initcall(bL_switcher_init); diff --git a/arch/arm/common/bL_switcher_dummy_if.c b/arch/arm/common/bL_switcher_dummy_if.c new file mode 100644 index 000000000000..5e2dd197e728 --- /dev/null +++ b/arch/arm/common/bL_switcher_dummy_if.c @@ -0,0 +1,71 @@ +/* + * arch/arm/common/bL_switcher_dummy_if.c -- b.L switcher dummy interface + * + * Created by: Nicolas Pitre, November 2012 + * Copyright: (C) 2012 Linaro Limited + * + * Dummy interface to user space for debugging purpose only. + * + * 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/init.h> +#include <linux/module.h> +#include <linux/fs.h> +#include <linux/miscdevice.h> +#include <asm/uaccess.h> +#include <asm/bL_switcher.h> + +static ssize_t bL_switcher_write(struct file *file, const char __user *buf, + size_t len, loff_t *pos) +{ + unsigned char val[3]; + unsigned int cpu, cluster; + int ret; + + pr_debug("%s\n", __func__); + + if (len < 3) + return -EINVAL; + + if (copy_from_user(val, buf, 3)) + return -EFAULT; + + /* format: <cpu#>,<cluster#> */ + if (val[0] < '0' || val[0] > '4' || + val[1] != ',' || + val[2] < '0' || val[2] > '1') + return -EINVAL; + + cpu = val[0] - '0'; + cluster = val[2] - '0'; + ret = bL_switch_request(cpu, cluster); + + return ret ? : len; +} + +static const struct file_operations bL_switcher_fops = { + .write = bL_switcher_write, + .owner = THIS_MODULE, +}; + +static struct miscdevice bL_switcher_device = { + MISC_DYNAMIC_MINOR, + "b.L_switcher", + &bL_switcher_fops +}; + +static int __init bL_switcher_dummy_if_init(void) +{ + return misc_register(&bL_switcher_device); +} + +static void __exit bL_switcher_dummy_if_exit(void) +{ + misc_deregister(&bL_switcher_device); +} + +module_init(bL_switcher_dummy_if_init); +module_exit(bL_switcher_dummy_if_exit); diff --git a/arch/arm/common/mcpm_entry.c b/arch/arm/common/mcpm_entry.c index 370236dd1a03..4a2b32fd53a1 100644 --- a/arch/arm/common/mcpm_entry.c +++ b/arch/arm/common/mcpm_entry.c @@ -27,6 +27,18 @@ void mcpm_set_entry_vector(unsigned cpu, unsigned cluster, void *ptr) sync_cache_w(&mcpm_entry_vectors[cluster][cpu]); } +extern unsigned long mcpm_entry_early_pokes[MAX_NR_CLUSTERS][MAX_CPUS_PER_CLUSTER][2]; + +void mcpm_set_early_poke(unsigned cpu, unsigned cluster, + unsigned long poke_phys_addr, unsigned long poke_val) +{ + unsigned long *poke = &mcpm_entry_early_pokes[cluster][cpu][0]; + poke[0] = poke_phys_addr; + poke[1] = poke_val; + __cpuc_flush_dcache_area((void *)poke, 8); + outer_clean_range(__pa(poke), __pa(poke + 2)); +} + static const struct mcpm_platform_ops *platform_ops; int __init mcpm_platform_register(const struct mcpm_platform_ops *ops) diff --git a/arch/arm/common/mcpm_head.S b/arch/arm/common/mcpm_head.S index 8178705c4b24..057e9c5a9e1f 100644 --- a/arch/arm/common/mcpm_head.S +++ b/arch/arm/common/mcpm_head.S @@ -71,12 +71,19 @@ ENTRY(mcpm_entry_point) * position independent way. */ adr r5, 3f - ldmia r5, {r6, r7, r8, r11} + ldmia r5, {r0, r6, r7, r8, r11} + add r0, r5, r0 @ r0 = mcpm_entry_early_pokes add r6, r5, r6 @ r6 = mcpm_entry_vectors ldr r7, [r5, r7] @ r7 = mcpm_power_up_setup_phys add r8, r5, r8 @ r8 = mcpm_sync add r11, r5, r11 @ r11 = first_man_locks + @ Perform an early poke, if any + add r0, r0, r4, lsl #3 + ldmia r0, {r0, r1} + teq r0, #0 + strne r1, [r0] + mov r0, #MCPM_SYNC_CLUSTER_SIZE mla r8, r0, r10, r8 @ r8 = sync cluster base @@ -195,7 +202,8 @@ mcpm_entry_gated: .align 2 -3: .word mcpm_entry_vectors - . +3: .word mcpm_entry_early_pokes - . + .word mcpm_entry_vectors - 3b .word mcpm_power_up_setup_phys - 3b .word mcpm_sync - 3b .word first_man_locks - 3b @@ -214,6 +222,10 @@ first_man_locks: ENTRY(mcpm_entry_vectors) .space 4 * MAX_NR_CLUSTERS * MAX_CPUS_PER_CLUSTER + .type mcpm_entry_early_pokes, #object +ENTRY(mcpm_entry_early_pokes) + .space 8 * MAX_NR_CLUSTERS * MAX_CPUS_PER_CLUSTER + .type mcpm_power_up_setup_phys, #object ENTRY(mcpm_power_up_setup_phys) .space 4 @ set by mcpm_sync_init() diff --git a/arch/arm/include/asm/bL_switcher.h b/arch/arm/include/asm/bL_switcher.h new file mode 100644 index 000000000000..482383b45c91 --- /dev/null +++ b/arch/arm/include/asm/bL_switcher.h @@ -0,0 +1,83 @@ +/* + * arch/arm/include/asm/bL_switcher.h + * + * Created by: Nicolas Pitre, April 2012 + * Copyright: (C) 2012 Linaro Limited + * + * 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. + */ + +#ifndef ASM_BL_SWITCHER_H +#define ASM_BL_SWITCHER_H + +#include <linux/compiler.h> +#include <linux/types.h> + +typedef void (*bL_switch_completion_handler)(void *cookie); + +int bL_switch_request_cb(unsigned int cpu, unsigned int new_cluster_id, + bL_switch_completion_handler completer, + void *completer_cookie); +static inline int bL_switch_request(unsigned int cpu, unsigned int new_cluster_id) +{ + return bL_switch_request_cb(cpu, new_cluster_id, NULL, NULL); +} + +/* + * Register here to be notified about runtime enabling/disabling of + * the switcher. + * + * The notifier chain is called with the switcher activation lock held: + * the switcher will not be enabled or disabled during callbacks. + * Callbacks must not call bL_switcher_{get,put}_enabled(). + */ +#define BL_NOTIFY_PRE_ENABLE 0 +#define BL_NOTIFY_POST_ENABLE 1 +#define BL_NOTIFY_PRE_DISABLE 2 +#define BL_NOTIFY_POST_DISABLE 3 + +#ifdef CONFIG_BL_SWITCHER + +void bL_switch_request_detach(unsigned int cpu, + bL_switch_completion_handler completer); + +int bL_switcher_register_notifier(struct notifier_block *nb); +int bL_switcher_unregister_notifier(struct notifier_block *nb); + +/* + * Use these functions to temporarily prevent enabling/disabling of + * the switcher. + * bL_switcher_get_enabled() returns true if the switcher is currently + * enabled. Each call to bL_switcher_get_enabled() must be followed + * by a call to bL_switcher_put_enabled(). These functions are not + * recursive. + */ +bool bL_switcher_get_enabled(void); +void bL_switcher_put_enabled(void); + +int bL_switcher_trace_trigger(void); +int bL_switcher_get_logical_index(u32 mpidr); + +#else +static void bL_switch_request_detach(unsigned int cpu, + bL_switch_completion_handler completer) { } + +static inline int bL_switcher_register_notifier(struct notifier_block *nb) +{ + return 0; +} + +static inline int bL_switcher_unregister_notifier(struct notifier_block *nb) +{ + return 0; +} + +static inline bool bL_switcher_get_enabled(void) { return false; } +static inline void bL_switcher_put_enabled(void) { } +static inline int bL_switcher_trace_trigger(void) { return 0; } +static inline int bL_switcher_get_logical_index(u32 mpidr) { return -EUNATCH; } +#endif /* CONFIG_BL_SWITCHER */ + +#endif diff --git a/arch/arm/include/asm/hardirq.h b/arch/arm/include/asm/hardirq.h index 2740c2a2df63..3d7351c844aa 100644 --- a/arch/arm/include/asm/hardirq.h +++ b/arch/arm/include/asm/hardirq.h @@ -5,7 +5,7 @@ #include <linux/threads.h> #include <asm/irq.h> -#define NR_IPI 6 +#define NR_IPI 7 typedef struct { unsigned int __softirq_pending; diff --git a/arch/arm/include/asm/mcpm.h b/arch/arm/include/asm/mcpm.h index 0f7b7620e9a5..7626a7fd4938 100644 --- a/arch/arm/include/asm/mcpm.h +++ b/arch/arm/include/asm/mcpm.h @@ -42,6 +42,14 @@ extern void mcpm_entry_point(void); void mcpm_set_entry_vector(unsigned cpu, unsigned cluster, void *ptr); /* + * This sets an early poke i.e a value to be poked into some address + * from very early assembly code before the CPU is ungated. The + * address must be physical, and if 0 then nothing will happen. + */ +void mcpm_set_early_poke(unsigned cpu, unsigned cluster, + unsigned long poke_phys_addr, unsigned long poke_val); + +/* * CPU/cluster power operations API for higher subsystems to use. */ diff --git a/arch/arm/include/asm/pmu.h b/arch/arm/include/asm/pmu.h index f24edad26c70..0cd7824ca762 100644 --- a/arch/arm/include/asm/pmu.h +++ b/arch/arm/include/asm/pmu.h @@ -62,9 +62,19 @@ struct pmu_hw_events { raw_spinlock_t pmu_lock; }; +struct cpupmu_regs { + u32 pmc; + u32 pmcntenset; + u32 pmuseren; + u32 pmintenset; + u32 pmxevttype[8]; + u32 pmxevtcnt[8]; +}; + struct arm_pmu { struct pmu pmu; cpumask_t active_irqs; + cpumask_t valid_cpus; char *name; irqreturn_t (*handle_irq)(int irq_num, void *dev); void (*enable)(struct perf_event *event); @@ -81,6 +91,8 @@ struct arm_pmu { int (*request_irq)(struct arm_pmu *, irq_handler_t handler); void (*free_irq)(struct arm_pmu *); int (*map_event)(struct perf_event *event); + void (*save_regs)(struct arm_pmu *, struct cpupmu_regs *); + void (*restore_regs)(struct arm_pmu *, struct cpupmu_regs *); int num_events; atomic_t active_events; struct mutex reserve_mutex; diff --git a/arch/arm/include/asm/smp.h b/arch/arm/include/asm/smp.h index d3a22bebe6ce..610ccf33f5e7 100644 --- a/arch/arm/include/asm/smp.h +++ b/arch/arm/include/asm/smp.h @@ -81,6 +81,8 @@ extern void arch_send_call_function_single_ipi(int cpu); extern void arch_send_call_function_ipi_mask(const struct cpumask *mask); extern void arch_send_wakeup_ipi_mask(const struct cpumask *mask); +extern int register_ipi_completion(struct completion *completion, int cpu); + struct smp_operations { #ifdef CONFIG_SMP /* diff --git a/arch/arm/include/asm/topology.h b/arch/arm/include/asm/topology.h index 58b8b84adcd2..983fa7c153a2 100644 --- a/arch/arm/include/asm/topology.h +++ b/arch/arm/include/asm/topology.h @@ -26,11 +26,45 @@ extern struct cputopo_arm cpu_topology[NR_CPUS]; void init_cpu_topology(void); void store_cpu_topology(unsigned int cpuid); const struct cpumask *cpu_coregroup_mask(int cpu); +int cluster_to_logical_mask(unsigned int socket_id, cpumask_t *cluster_mask); + +#ifdef CONFIG_DISABLE_CPU_SCHED_DOMAIN_BALANCE +/* Common values for CPUs */ +#ifndef SD_CPU_INIT +#define SD_CPU_INIT (struct sched_domain) { \ + .min_interval = 1, \ + .max_interval = 4, \ + .busy_factor = 64, \ + .imbalance_pct = 125, \ + .cache_nice_tries = 1, \ + .busy_idx = 2, \ + .idle_idx = 1, \ + .newidle_idx = 0, \ + .wake_idx = 0, \ + .forkexec_idx = 0, \ + \ + .flags = 0*SD_LOAD_BALANCE \ + | 1*SD_BALANCE_NEWIDLE \ + | 1*SD_BALANCE_EXEC \ + | 1*SD_BALANCE_FORK \ + | 0*SD_BALANCE_WAKE \ + | 1*SD_WAKE_AFFINE \ + | 0*SD_SHARE_CPUPOWER \ + | 0*SD_SHARE_PKG_RESOURCES \ + | 0*SD_SERIALIZE \ + , \ + .last_balance = jiffies, \ + .balance_interval = 1, \ +} +#endif +#endif /* CONFIG_DISABLE_CPU_SCHED_DOMAIN_BALANCE */ #else static inline void init_cpu_topology(void) { } static inline void store_cpu_topology(unsigned int cpuid) { } +static inline int cluster_to_logical_mask(unsigned int socket_id, + cpumask_t *cluster_mask) { return -EINVAL; } #endif diff --git a/arch/arm/kernel/hw_breakpoint.c b/arch/arm/kernel/hw_breakpoint.c index 1fd749ee4a1b..1b803117ed91 100644 --- a/arch/arm/kernel/hw_breakpoint.c +++ b/arch/arm/kernel/hw_breakpoint.c @@ -1049,7 +1049,8 @@ static struct notifier_block dbg_cpu_pm_nb = { static void __init pm_init(void) { - cpu_pm_register_notifier(&dbg_cpu_pm_nb); + if (has_ossr) + cpu_pm_register_notifier(&dbg_cpu_pm_nb); } #else static inline void pm_init(void) diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c index 8c3094d0f7b7..d847c622a7b5 100644 --- a/arch/arm/kernel/perf_event.c +++ b/arch/arm/kernel/perf_event.c @@ -12,6 +12,7 @@ */ #define pr_fmt(fmt) "hw perfevents: " fmt +#include <linux/cpumask.h> #include <linux/kernel.h> #include <linux/platform_device.h> #include <linux/pm_runtime.h> @@ -81,6 +82,9 @@ armpmu_map_event(struct perf_event *event, return armpmu_map_cache_event(cache_map, config); case PERF_TYPE_RAW: return armpmu_map_raw_event(raw_event_mask, config); + default: + if (event->attr.type >= PERF_TYPE_MAX) + return armpmu_map_raw_event(raw_event_mask, config); } return -ENOENT; @@ -158,6 +162,8 @@ armpmu_stop(struct perf_event *event, int flags) struct arm_pmu *armpmu = to_arm_pmu(event->pmu); struct hw_perf_event *hwc = &event->hw; + if (!cpumask_test_cpu(smp_processor_id(), &armpmu->valid_cpus)) + return; /* * ARM pmu always has to update the counter, so ignore * PERF_EF_UPDATE, see comments in armpmu_start(). @@ -174,6 +180,8 @@ static void armpmu_start(struct perf_event *event, int flags) struct arm_pmu *armpmu = to_arm_pmu(event->pmu); struct hw_perf_event *hwc = &event->hw; + if (!cpumask_test_cpu(smp_processor_id(), &armpmu->valid_cpus)) + return; /* * ARM pmu always has to reprogram the period, so ignore * PERF_EF_RELOAD, see the comment below. @@ -201,6 +209,9 @@ armpmu_del(struct perf_event *event, int flags) struct hw_perf_event *hwc = &event->hw; int idx = hwc->idx; + if (!cpumask_test_cpu(smp_processor_id(), &armpmu->valid_cpus)) + return; + armpmu_stop(event, PERF_EF_UPDATE); hw_events->events[idx] = NULL; clear_bit(idx, hw_events->used_mask); @@ -217,6 +228,10 @@ armpmu_add(struct perf_event *event, int flags) int idx; int err = 0; + /* An event following a process won't be stopped earlier */ + if (!cpumask_test_cpu(smp_processor_id(), &armpmu->valid_cpus)) + return 0; + perf_pmu_disable(event->pmu); /* If we don't have a space for the counter then finish early. */ @@ -416,6 +431,10 @@ static int armpmu_event_init(struct perf_event *event) int err = 0; atomic_t *active_events = &armpmu->active_events; + if (event->cpu != -1 && + !cpumask_test_cpu(event->cpu, &armpmu->valid_cpus)) + return -ENOENT; + /* does not support taken branch sampling */ if (has_branch_stack(event)) return -EOPNOTSUPP; diff --git a/arch/arm/kernel/perf_event_cpu.c b/arch/arm/kernel/perf_event_cpu.c index 1f2740e3dbc0..0b48a38e3cf4 100644 --- a/arch/arm/kernel/perf_event_cpu.c +++ b/arch/arm/kernel/perf_event_cpu.c @@ -19,6 +19,7 @@ #define pr_fmt(fmt) "CPU PMU: " fmt #include <linux/bitmap.h> +#include <linux/cpu_pm.h> #include <linux/export.h> #include <linux/kernel.h> #include <linux/of.h> @@ -31,33 +32,36 @@ #include <asm/pmu.h> /* Set at runtime when we know what CPU type we are. */ -static struct arm_pmu *cpu_pmu; +static DEFINE_PER_CPU(struct arm_pmu *, cpu_pmu); static DEFINE_PER_CPU(struct perf_event * [ARMPMU_MAX_HWEVENTS], hw_events); static DEFINE_PER_CPU(unsigned long [BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)], used_mask); static DEFINE_PER_CPU(struct pmu_hw_events, cpu_hw_events); +static DEFINE_PER_CPU(struct cpupmu_regs, cpu_pmu_regs); + /* * Despite the names, these two functions are CPU-specific and are used * by the OProfile/perf code. */ const char *perf_pmu_name(void) { - if (!cpu_pmu) + struct arm_pmu *pmu = per_cpu(cpu_pmu, 0); + if (!pmu) return NULL; - return cpu_pmu->name; + return pmu->name; } EXPORT_SYMBOL_GPL(perf_pmu_name); int perf_num_counters(void) { - int max_events = 0; + struct arm_pmu *pmu = per_cpu(cpu_pmu, 0); - if (cpu_pmu != NULL) - max_events = cpu_pmu->num_events; + if (!pmu) + return 0; - return max_events; + return pmu->num_events; } EXPORT_SYMBOL_GPL(perf_num_counters); @@ -75,11 +79,13 @@ static void cpu_pmu_free_irq(struct arm_pmu *cpu_pmu) { int i, irq, irqs; struct platform_device *pmu_device = cpu_pmu->plat_device; + int cpu = -1; irqs = min(pmu_device->num_resources, num_possible_cpus()); for (i = 0; i < irqs; ++i) { - if (!cpumask_test_and_clear_cpu(i, &cpu_pmu->active_irqs)) + cpu = cpumask_next(cpu, &cpu_pmu->valid_cpus); + if (!cpumask_test_and_clear_cpu(cpu, &cpu_pmu->active_irqs)) continue; irq = platform_get_irq(pmu_device, i); if (irq >= 0) @@ -91,6 +97,7 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler) { int i, err, irq, irqs; struct platform_device *pmu_device = cpu_pmu->plat_device; + int cpu = -1; if (!pmu_device) return -ENODEV; @@ -103,6 +110,7 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler) for (i = 0; i < irqs; ++i) { err = 0; + cpu = cpumask_next(cpu, &cpu_pmu->valid_cpus); irq = platform_get_irq(pmu_device, i); if (irq < 0) continue; @@ -112,7 +120,7 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler) * assume that we're running on a uniprocessor machine and * continue. Otherwise, continue without this interrupt. */ - if (irq_set_affinity(irq, cpumask_of(i)) && irqs > 1) { + if (irq_set_affinity(irq, cpumask_of(cpu)) && irqs > 1) { pr_warning("unable to set irq affinity (irq=%d, cpu=%u)\n", irq, i); continue; @@ -126,7 +134,7 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler) return err; } - cpumask_set_cpu(i, &cpu_pmu->active_irqs); + cpumask_set_cpu(cpu, &cpu_pmu->active_irqs); } return 0; @@ -135,7 +143,7 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler) static void cpu_pmu_init(struct arm_pmu *cpu_pmu) { int cpu; - for_each_possible_cpu(cpu) { + for_each_cpu_mask(cpu, cpu_pmu->valid_cpus) { struct pmu_hw_events *events = &per_cpu(cpu_hw_events, cpu); events->events = per_cpu(hw_events, cpu); events->used_mask = per_cpu(used_mask, cpu); @@ -148,7 +156,7 @@ static void cpu_pmu_init(struct arm_pmu *cpu_pmu) /* Ensure the PMU has sane values out of reset. */ if (cpu_pmu->reset) - on_each_cpu(cpu_pmu->reset, cpu_pmu, 1); + on_each_cpu_mask(&cpu_pmu->valid_cpus, cpu_pmu->reset, cpu_pmu, 1); } /* @@ -160,21 +168,46 @@ static void cpu_pmu_init(struct arm_pmu *cpu_pmu) static int __cpuinit cpu_pmu_notify(struct notifier_block *b, unsigned long action, void *hcpu) { + struct arm_pmu *pmu = per_cpu(cpu_pmu, (long)hcpu); + if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING) return NOTIFY_DONE; - if (cpu_pmu && cpu_pmu->reset) - cpu_pmu->reset(cpu_pmu); + if (pmu && pmu->reset) + pmu->reset(pmu); else return NOTIFY_DONE; return NOTIFY_OK; } +static int cpu_pmu_pm_notify(struct notifier_block *b, + unsigned long action, void *hcpu) +{ + int cpu = smp_processor_id(); + struct arm_pmu *pmu = per_cpu(cpu_pmu, cpu); + struct cpupmu_regs *pmuregs = &per_cpu(cpu_pmu_regs, cpu); + + if (!pmu) + return NOTIFY_DONE; + + if (action == CPU_PM_ENTER && pmu->save_regs) { + pmu->save_regs(pmu, pmuregs); + } else if (action == CPU_PM_EXIT && pmu->restore_regs) { + pmu->restore_regs(pmu, pmuregs); + } + + return NOTIFY_OK; +} + static struct notifier_block __cpuinitdata cpu_pmu_hotplug_notifier = { .notifier_call = cpu_pmu_notify, }; +static struct notifier_block __cpuinitdata cpu_pmu_pm_notifier = { + .notifier_call = cpu_pmu_pm_notify, +}; + /* * PMU platform driver and devicetree bindings. */ @@ -246,6 +279,9 @@ static int probe_current_pmu(struct arm_pmu *pmu) } } + /* assume PMU support all the CPUs in this case */ + cpumask_setall(&pmu->valid_cpus); + put_cpu(); return ret; } @@ -253,15 +289,10 @@ static int probe_current_pmu(struct arm_pmu *pmu) static int cpu_pmu_device_probe(struct platform_device *pdev) { const struct of_device_id *of_id; - int (*init_fn)(struct arm_pmu *); struct device_node *node = pdev->dev.of_node; struct arm_pmu *pmu; - int ret = -ENODEV; - - if (cpu_pmu) { - pr_info("attempt to register multiple PMU devices!"); - return -ENOSPC; - } + int ret = 0; + int cpu; pmu = kzalloc(sizeof(struct arm_pmu), GFP_KERNEL); if (!pmu) { @@ -270,8 +301,28 @@ static int cpu_pmu_device_probe(struct platform_device *pdev) } if (node && (of_id = of_match_node(cpu_pmu_of_device_ids, pdev->dev.of_node))) { - init_fn = of_id->data; - ret = init_fn(pmu); + smp_call_func_t init_fn = (smp_call_func_t)of_id->data; + struct device_node *ncluster; + int cluster = -1; + cpumask_t sibling_mask; + + ncluster = of_parse_phandle(node, "cluster", 0); + if (ncluster) { + int len; + const u32 *hwid; + hwid = of_get_property(ncluster, "reg", &len); + if (hwid && len == 4) + cluster = be32_to_cpup(hwid); + } + /* set sibling mask to all cpu mask if socket is not specified */ + if (cluster == -1 || + cluster_to_logical_mask(cluster, &sibling_mask)) + cpumask_setall(&sibling_mask); + + smp_call_function_any(&sibling_mask, init_fn, pmu, 1); + + /* now set the valid_cpus after init */ + cpumask_copy(&pmu->valid_cpus, &sibling_mask); } else { ret = probe_current_pmu(pmu); } @@ -281,10 +332,12 @@ static int cpu_pmu_device_probe(struct platform_device *pdev) goto out_free; } - cpu_pmu = pmu; - cpu_pmu->plat_device = pdev; - cpu_pmu_init(cpu_pmu); - ret = armpmu_register(cpu_pmu, PERF_TYPE_RAW); + for_each_cpu_mask(cpu, pmu->valid_cpus) + per_cpu(cpu_pmu, cpu) = pmu; + + pmu->plat_device = pdev; + cpu_pmu_init(pmu); + ret = armpmu_register(pmu, -1); if (!ret) return 0; @@ -313,9 +366,17 @@ static int __init register_pmu_driver(void) if (err) return err; + err = cpu_pm_register_notifier(&cpu_pmu_pm_notifier); + if (err) { + unregister_cpu_notifier(&cpu_pmu_hotplug_notifier); + return err; + } + err = platform_driver_register(&cpu_pmu_driver); - if (err) + if (err) { + cpu_pm_unregister_notifier(&cpu_pmu_pm_notifier); unregister_cpu_notifier(&cpu_pmu_hotplug_notifier); + } return err; } diff --git a/arch/arm/kernel/perf_event_v7.c b/arch/arm/kernel/perf_event_v7.c index 039cffb053a7..654db5030c31 100644 --- a/arch/arm/kernel/perf_event_v7.c +++ b/arch/arm/kernel/perf_event_v7.c @@ -950,6 +950,51 @@ static void armv7_pmnc_dump_regs(struct arm_pmu *cpu_pmu) } #endif +static void armv7pmu_save_regs(struct arm_pmu *cpu_pmu, + struct cpupmu_regs *regs) +{ + unsigned int cnt; + asm volatile("mrc p15, 0, %0, c9, c12, 0" : "=r" (regs->pmc)); + if (!(regs->pmc & ARMV7_PMNC_E)) + return; + + asm volatile("mrc p15, 0, %0, c9, c12, 1" : "=r" (regs->pmcntenset)); + asm volatile("mrc p15, 0, %0, c9, c14, 0" : "=r" (regs->pmuseren)); + asm volatile("mrc p15, 0, %0, c9, c14, 1" : "=r" (regs->pmintenset)); + asm volatile("mrc p15, 0, %0, c9, c13, 0" : "=r" (regs->pmxevtcnt[0])); + for (cnt = ARMV7_IDX_COUNTER0; + cnt <= ARMV7_IDX_COUNTER_LAST(cpu_pmu); cnt++) { + armv7_pmnc_select_counter(cnt); + asm volatile("mrc p15, 0, %0, c9, c13, 1" + : "=r"(regs->pmxevttype[cnt])); + asm volatile("mrc p15, 0, %0, c9, c13, 2" + : "=r"(regs->pmxevtcnt[cnt])); + } + return; +} + +static void armv7pmu_restore_regs(struct arm_pmu *cpu_pmu, + struct cpupmu_regs *regs) +{ + unsigned int cnt; + if (!(regs->pmc & ARMV7_PMNC_E)) + return; + + asm volatile("mcr p15, 0, %0, c9, c12, 1" : : "r" (regs->pmcntenset)); + asm volatile("mcr p15, 0, %0, c9, c14, 0" : : "r" (regs->pmuseren)); + asm volatile("mcr p15, 0, %0, c9, c14, 1" : : "r" (regs->pmintenset)); + asm volatile("mcr p15, 0, %0, c9, c13, 0" : : "r" (regs->pmxevtcnt[0])); + for (cnt = ARMV7_IDX_COUNTER0; + cnt <= ARMV7_IDX_COUNTER_LAST(cpu_pmu); cnt++) { + armv7_pmnc_select_counter(cnt); + asm volatile("mcr p15, 0, %0, c9, c13, 1" + : : "r"(regs->pmxevttype[cnt])); + asm volatile("mcr p15, 0, %0, c9, c13, 2" + : : "r"(regs->pmxevtcnt[cnt])); + } + asm volatile("mcr p15, 0, %0, c9, c12, 0" : : "r" (regs->pmc)); +} + static void armv7pmu_enable_event(struct perf_event *event) { unsigned long flags; @@ -1223,6 +1268,8 @@ static void armv7pmu_init(struct arm_pmu *cpu_pmu) cpu_pmu->start = armv7pmu_start; cpu_pmu->stop = armv7pmu_stop; cpu_pmu->reset = armv7pmu_reset; + cpu_pmu->save_regs = armv7pmu_save_regs; + cpu_pmu->restore_regs = armv7pmu_restore_regs; cpu_pmu->max_period = (1LLU << 32) - 1; }; @@ -1240,7 +1287,7 @@ static u32 armv7_read_num_pmnc_events(void) static int armv7_a8_pmu_init(struct arm_pmu *cpu_pmu) { armv7pmu_init(cpu_pmu); - cpu_pmu->name = "ARMv7 Cortex-A8"; + cpu_pmu->name = "ARMv7_Cortex_A8"; cpu_pmu->map_event = armv7_a8_map_event; cpu_pmu->num_events = armv7_read_num_pmnc_events(); return 0; @@ -1249,7 +1296,7 @@ static int armv7_a8_pmu_init(struct arm_pmu *cpu_pmu) static int armv7_a9_pmu_init(struct arm_pmu *cpu_pmu) { armv7pmu_init(cpu_pmu); - cpu_pmu->name = "ARMv7 Cortex-A9"; + cpu_pmu->name = "ARMv7_Cortex_A9"; cpu_pmu->map_event = armv7_a9_map_event; cpu_pmu->num_events = armv7_read_num_pmnc_events(); return 0; @@ -1258,7 +1305,7 @@ static int armv7_a9_pmu_init(struct arm_pmu *cpu_pmu) static int armv7_a5_pmu_init(struct arm_pmu *cpu_pmu) { armv7pmu_init(cpu_pmu); - cpu_pmu->name = "ARMv7 Cortex-A5"; + cpu_pmu->name = "ARMv7_Cortex_A5"; cpu_pmu->map_event = armv7_a5_map_event; cpu_pmu->num_events = armv7_read_num_pmnc_events(); return 0; @@ -1267,7 +1314,7 @@ static int armv7_a5_pmu_init(struct arm_pmu *cpu_pmu) static int armv7_a15_pmu_init(struct arm_pmu *cpu_pmu) { armv7pmu_init(cpu_pmu); - cpu_pmu->name = "ARMv7 Cortex-A15"; + cpu_pmu->name = "ARMv7_Cortex_A15"; cpu_pmu->map_event = armv7_a15_map_event; cpu_pmu->num_events = armv7_read_num_pmnc_events(); cpu_pmu->set_event_filter = armv7pmu_set_event_filter; @@ -1277,7 +1324,7 @@ static int armv7_a15_pmu_init(struct arm_pmu *cpu_pmu) static int armv7_a7_pmu_init(struct arm_pmu *cpu_pmu) { armv7pmu_init(cpu_pmu); - cpu_pmu->name = "ARMv7 Cortex-A7"; + cpu_pmu->name = "ARMv7_Cortex_A7"; cpu_pmu->map_event = armv7_a7_map_event; cpu_pmu->num_events = armv7_read_num_pmnc_events(); cpu_pmu->set_event_filter = armv7pmu_set_event_filter; diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c index 5919eb451bb9..dc05ca77963b 100644 --- a/arch/arm/kernel/smp.c +++ b/arch/arm/kernel/smp.c @@ -66,6 +66,7 @@ enum ipi_msg_type { IPI_CALL_FUNC, IPI_CALL_FUNC_SINGLE, IPI_CPU_STOP, + IPI_COMPLETION, }; static DECLARE_COMPLETION(cpu_running); @@ -463,6 +464,7 @@ static const char *ipi_types[NR_IPI] = { S(IPI_CALL_FUNC, "Function call interrupts"), S(IPI_CALL_FUNC_SINGLE, "Single function call interrupts"), S(IPI_CPU_STOP, "CPU stop interrupts"), + S(IPI_COMPLETION, "completion interrupts"), }; void show_ipi_list(struct seq_file *p, int prec) @@ -588,6 +590,19 @@ static void ipi_cpu_stop(unsigned int cpu) cpu_relax(); } +static DEFINE_PER_CPU(struct completion *, cpu_completion); + +int register_ipi_completion(struct completion *completion, int cpu) +{ + per_cpu(cpu_completion, cpu) = completion; + return IPI_COMPLETION; +} + +static void ipi_complete(unsigned int cpu) +{ + complete(per_cpu(cpu_completion, cpu)); +} + /* * Main handler for inter-processor interrupts */ @@ -638,6 +653,12 @@ void handle_IPI(int ipinr, struct pt_regs *regs) irq_exit(); break; + case IPI_COMPLETION: + irq_enter(); + ipi_complete(cpu); + irq_exit(); + break; + default: printk(KERN_CRIT "CPU%u: Unknown IPI message 0x%x\n", cpu, ipinr); diff --git a/arch/arm/kernel/topology.c b/arch/arm/kernel/topology.c index c5a59546a256..677da58d9e88 100644 --- a/arch/arm/kernel/topology.c +++ b/arch/arm/kernel/topology.c @@ -23,6 +23,7 @@ #include <linux/slab.h> #include <asm/cputype.h> +#include <asm/smp_plat.h> #include <asm/topology.h> /* @@ -289,6 +290,140 @@ void store_cpu_topology(unsigned int cpuid) cpu_topology[cpuid].socket_id, mpidr); } + +#ifdef CONFIG_SCHED_HMP + +static const char * const little_cores[] = { + "arm,cortex-a7", + NULL, +}; + +static bool is_little_cpu(struct device_node *cn) +{ + const char * const *lc; + for (lc = little_cores; *lc; lc++) + if (of_device_is_compatible(cn, *lc)) + return true; + return false; +} + +void __init arch_get_fast_and_slow_cpus(struct cpumask *fast, + struct cpumask *slow) +{ + struct device_node *cn = NULL; + int cpu; + + cpumask_clear(fast); + cpumask_clear(slow); + + /* + * Use the config options if they are given. This helps testing + * HMP scheduling on systems without a big.LITTLE architecture. + */ + if (strlen(CONFIG_HMP_FAST_CPU_MASK) && strlen(CONFIG_HMP_SLOW_CPU_MASK)) { + if (cpulist_parse(CONFIG_HMP_FAST_CPU_MASK, fast)) + WARN(1, "Failed to parse HMP fast cpu mask!\n"); + if (cpulist_parse(CONFIG_HMP_SLOW_CPU_MASK, slow)) + WARN(1, "Failed to parse HMP slow cpu mask!\n"); + return; + } + + /* + * Else, parse device tree for little cores. + */ + while ((cn = of_find_node_by_type(cn, "cpu"))) { + + const u32 *mpidr; + int len; + + mpidr = of_get_property(cn, "reg", &len); + if (!mpidr || len != 4) { + pr_err("* %s missing reg property\n", cn->full_name); + continue; + } + + cpu = get_logical_index(be32_to_cpup(mpidr)); + if (cpu == -EINVAL) { + pr_err("couldn't get logical index for mpidr %x\n", + be32_to_cpup(mpidr)); + break; + } + + if (is_little_cpu(cn)) + cpumask_set_cpu(cpu, slow); + else + cpumask_set_cpu(cpu, fast); + } + + if (!cpumask_empty(fast) && !cpumask_empty(slow)) + return; + + /* + * We didn't find both big and little cores so let's call all cores + * fast as this will keep the system running, with all cores being + * treated equal. + */ + cpumask_setall(fast); + cpumask_clear(slow); +} + +struct cpumask hmp_slow_cpu_mask; + +void __init arch_get_hmp_domains(struct list_head *hmp_domains_list) +{ + struct cpumask hmp_fast_cpu_mask; + struct hmp_domain *domain; + + arch_get_fast_and_slow_cpus(&hmp_fast_cpu_mask, &hmp_slow_cpu_mask); + + /* + * Initialize hmp_domains + * Must be ordered with respect to compute capacity. + * Fastest domain at head of list. + */ + if(!cpumask_empty(&hmp_slow_cpu_mask)) { + domain = (struct hmp_domain *) + kmalloc(sizeof(struct hmp_domain), GFP_KERNEL); + cpumask_copy(&domain->possible_cpus, &hmp_slow_cpu_mask); + cpumask_and(&domain->cpus, cpu_online_mask, &domain->possible_cpus); + list_add(&domain->hmp_domains, hmp_domains_list); + } + domain = (struct hmp_domain *) + kmalloc(sizeof(struct hmp_domain), GFP_KERNEL); + cpumask_copy(&domain->possible_cpus, &hmp_fast_cpu_mask); + cpumask_and(&domain->cpus, cpu_online_mask, &domain->possible_cpus); + list_add(&domain->hmp_domains, hmp_domains_list); +} +#endif /* CONFIG_SCHED_HMP */ + + +/* + * cluster_to_logical_mask - return cpu logical mask of CPUs in a cluster + * @socket_id: cluster HW identifier + * @cluster_mask: the cpumask location to be initialized, modified by the + * function only if return value == 0 + * + * Return: + * + * 0 on success + * -EINVAL if cluster_mask is NULL or there is no record matching socket_id + */ +int cluster_to_logical_mask(unsigned int socket_id, cpumask_t *cluster_mask) +{ + int cpu; + + if (!cluster_mask) + return -EINVAL; + + for_each_online_cpu(cpu) + if (socket_id == topology_physical_package_id(cpu)) { + cpumask_copy(cluster_mask, topology_core_cpumask(cpu)); + return 0; + } + + return -EINVAL; +} + /* * init_cpu_topology is called at boot when only one cpu is running * which prevent simultaneous write access to cpu_topology array diff --git a/block/blk-core.c b/block/blk-core.c index d5745b5833c9..0852e5d43436 100644 --- a/block/blk-core.c +++ b/block/blk-core.c @@ -3180,7 +3180,8 @@ int __init blk_dev_init(void) /* used for unplugging and affects IO latency/throughput - HIGHPRI */ kblockd_workqueue = alloc_workqueue("kblockd", - WQ_MEM_RECLAIM | WQ_HIGHPRI, 0); + WQ_MEM_RECLAIM | WQ_HIGHPRI | + WQ_POWER_EFFICIENT, 0); if (!kblockd_workqueue) panic("Failed to create kblockd\n"); diff --git a/block/blk-ioc.c b/block/blk-ioc.c index 9c4bb8266bc8..4464c823cff2 100644 --- a/block/blk-ioc.c +++ b/block/blk-ioc.c @@ -144,7 +144,8 @@ void put_io_context(struct io_context *ioc) if (atomic_long_dec_and_test(&ioc->refcount)) { spin_lock_irqsave(&ioc->lock, flags); if (!hlist_empty(&ioc->icq_list)) - schedule_work(&ioc->release_work); + queue_work(system_power_efficient_wq, + &ioc->release_work); else free_ioc = true; spin_unlock_irqrestore(&ioc->lock, flags); diff --git a/block/genhd.c b/block/genhd.c index 20625eed5511..e9094b375c05 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -1489,9 +1489,11 @@ static void __disk_unblock_events(struct gendisk *disk, bool check_now) intv = disk_events_poll_jiffies(disk); set_timer_slack(&ev->dwork.timer, intv / 4); if (check_now) - queue_delayed_work(system_freezable_wq, &ev->dwork, 0); + queue_delayed_work(system_freezable_power_efficient_wq, + &ev->dwork, 0); else if (intv) - queue_delayed_work(system_freezable_wq, &ev->dwork, intv); + queue_delayed_work(system_freezable_power_efficient_wq, + &ev->dwork, intv); out_unlock: spin_unlock_irqrestore(&ev->lock, flags); } @@ -1534,7 +1536,8 @@ void disk_flush_events(struct gendisk *disk, unsigned int mask) spin_lock_irq(&ev->lock); ev->clearing |= mask; if (!ev->block) - mod_delayed_work(system_freezable_wq, &ev->dwork, 0); + mod_delayed_work(system_freezable_power_efficient_wq, + &ev->dwork, 0); spin_unlock_irq(&ev->lock); } @@ -1627,7 +1630,8 @@ static void disk_check_events(struct disk_events *ev, intv = disk_events_poll_jiffies(disk); if (!ev->block && intv) - queue_delayed_work(system_freezable_wq, &ev->dwork, intv); + queue_delayed_work(system_freezable_power_efficient_wq, + &ev->dwork, intv); spin_unlock_irq(&ev->lock); diff --git a/drivers/irqchip/irq-gic.c b/drivers/irqchip/irq-gic.c index fe44d3e2c702..6f00dfa7d4fd 100644 --- a/drivers/irqchip/irq-gic.c +++ b/drivers/irqchip/irq-gic.c @@ -253,10 +253,9 @@ static int gic_set_affinity(struct irq_data *d, const struct cpumask *mask_val, if (cpu >= NR_GIC_CPU_IF || cpu >= nr_cpu_ids) return -EINVAL; + raw_spin_lock(&irq_controller_lock); mask = 0xff << shift; bit = gic_cpu_map[cpu] << shift; - - raw_spin_lock(&irq_controller_lock); val = readl_relaxed(reg) & ~mask; writel_relaxed(val | bit, reg); raw_spin_unlock(&irq_controller_lock); @@ -652,7 +651,9 @@ static void __init gic_pm_init(struct gic_chip_data *gic) void gic_raise_softirq(const struct cpumask *mask, unsigned int irq) { int cpu; - unsigned long map = 0; + unsigned long flags, map = 0; + + raw_spin_lock_irqsave(&irq_controller_lock, flags); /* Convert our logical CPU mask into a physical one. */ for_each_cpu(cpu, mask) @@ -666,9 +667,145 @@ void gic_raise_softirq(const struct cpumask *mask, unsigned int irq) /* this always happens on GIC0 */ writel_relaxed(map << 16 | irq, gic_data_dist_base(&gic_data[0]) + GIC_DIST_SOFTINT); + + raw_spin_unlock_irqrestore(&irq_controller_lock, flags); } #endif +#ifdef CONFIG_BL_SWITCHER +/* + * gic_send_sgi - send a SGI directly to given CPU interface number + * + * cpu_id: the ID for the destination CPU interface + * irq: the IPI number to send a SGI for + */ +void gic_send_sgi(unsigned int cpu_id, unsigned int irq) +{ + BUG_ON(cpu_id >= NR_GIC_CPU_IF); + cpu_id = 1 << cpu_id; + /* this always happens on GIC0 */ + writel_relaxed((cpu_id << 16) | irq, gic_data_dist_base(&gic_data[0]) + GIC_DIST_SOFTINT); +} + +/* + * gic_get_cpu_id - get the CPU interface ID for the specified CPU + * + * @cpu: the logical CPU number to get the GIC ID for. + * + * Return the CPU interface ID for the given logical CPU number, + * or -1 if the CPU number is too large or the interface ID is + * unknown (more than one bit set). + */ +int gic_get_cpu_id(unsigned int cpu) +{ + unsigned int cpu_bit; + + if (cpu >= NR_GIC_CPU_IF) + return -1; + cpu_bit = gic_cpu_map[cpu]; + if (cpu_bit & (cpu_bit - 1)) + return -1; + return __ffs(cpu_bit); +} + +/* + * gic_migrate_target - migrate IRQs to another PU interface + * + * @new_cpu_id: the CPU target ID to migrate IRQs to + * + * Migrate all peripheral interrupts with a target matching the current CPU + * to the interface corresponding to @new_cpu_id. The CPU interface mapping + * is also updated. Targets to other CPU interfaces are unchanged. + * This must be called with IRQs locally disabled. + */ +void gic_migrate_target(unsigned int new_cpu_id) +{ + unsigned int old_cpu_id, gic_irqs, gic_nr = 0; + void __iomem *dist_base; + int i, ror_val, cpu = smp_processor_id(); + u32 val, old_mask, active_mask; + + if (gic_nr >= MAX_GIC_NR) + BUG(); + + dist_base = gic_data_dist_base(&gic_data[gic_nr]); + if (!dist_base) + return; + gic_irqs = gic_data[gic_nr].gic_irqs; + + old_cpu_id = __ffs(gic_cpu_map[cpu]); + old_mask = 0x01010101 << old_cpu_id; + ror_val = (old_cpu_id - new_cpu_id) & 31; + + raw_spin_lock(&irq_controller_lock); + + gic_cpu_map[cpu] = 1 << new_cpu_id; + + for (i = 8; i < DIV_ROUND_UP(gic_irqs, 4); i++) { + val = readl_relaxed(dist_base + GIC_DIST_TARGET + i * 4); + active_mask = val & old_mask; + if (active_mask) { + val &= ~active_mask; + val |= ror32(active_mask, ror_val); + writel_relaxed(val, dist_base + GIC_DIST_TARGET + i * 4); + } + } + + raw_spin_unlock(&irq_controller_lock); + + /* + * Now let's migrate and clear any potential SGIs that might be + * pending for us (old_cpu_id). Since GIC_DIST_SGI_PENDING_SET + * is a banked register, we can only forward the SGI using + * GIC_DIST_SOFTINT. The original SGI source is lost but Linux + * doesn't use that information anyway. + * + * For the same reason we do not adjust SGI source information + * for previously sent SGIs by us to other CPUs either. + */ + for (i = 0; i < 16; i += 4) { + int j; + val = readl_relaxed(dist_base + GIC_DIST_SGI_PENDING_SET + i); + if (!val) + continue; + writel_relaxed(val, dist_base + GIC_DIST_SGI_PENDING_CLEAR + i); + for (j = i; j < i + 4; j++) { + if (val & 0xff) + writel_relaxed((1 << (new_cpu_id + 16)) | j, + dist_base + GIC_DIST_SOFTINT); + val >>= 8; + } + } +} + +/* + * gic_get_sgir_physaddr - get the physical address for the SGI register + * + * REturn the physical address of the SGI register to be used + * by some early assembly code when the kernel is not yet available. + */ +static unsigned long gic_dist_physaddr; + +unsigned long gic_get_sgir_physaddr(void) +{ + if (!gic_dist_physaddr) + return 0; + return gic_dist_physaddr + GIC_DIST_SOFTINT; +} + +void __init gic_init_physaddr(struct device_node *node) +{ + struct resource res; + if (of_address_to_resource(node, 0, &res) == 0) { + gic_dist_physaddr = res.start; + pr_info("GIC physical location is %#lx\n", gic_dist_physaddr); + } +} + +#else +#define gic_init_physaddr(node) do { } while(0) +#endif + static int gic_irq_domain_map(struct irq_domain *d, unsigned int irq, irq_hw_number_t hw) { @@ -850,6 +987,8 @@ int __init gic_of_init(struct device_node *node, struct device_node *parent) percpu_offset = 0; gic_init_bases(gic_cnt, -1, dist_base, cpu_base, percpu_offset, node); + if (!gic_cnt) + gic_init_physaddr(node); if (parent) { irq = irq_of_parse_and_map(node, 0); diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c index 38f0b312ff85..663d2d0448b7 100644 --- a/drivers/net/phy/phy.c +++ b/drivers/net/phy/phy.c @@ -439,7 +439,7 @@ void phy_start_machine(struct phy_device *phydev, { phydev->adjust_state = handler; - schedule_delayed_work(&phydev->state_queue, HZ); + queue_delayed_work(system_power_efficient_wq, &phydev->state_queue, HZ); } /** @@ -500,7 +500,7 @@ static irqreturn_t phy_interrupt(int irq, void *phy_dat) disable_irq_nosync(irq); atomic_inc(&phydev->irq_disable); - schedule_work(&phydev->phy_queue); + queue_work(system_power_efficient_wq, &phydev->phy_queue); return IRQ_HANDLED; } @@ -655,7 +655,7 @@ static void phy_change(struct work_struct *work) /* reschedule state queue work to run as soon as possible */ cancel_delayed_work_sync(&phydev->state_queue); - schedule_delayed_work(&phydev->state_queue, 0); + queue_delayed_work(system_power_efficient_wq, &phydev->state_queue, 0); return; @@ -918,7 +918,8 @@ void phy_state_machine(struct work_struct *work) if (err < 0) phy_error(phydev); - schedule_delayed_work(&phydev->state_queue, PHY_STATE_TIME * HZ); + queue_delayed_work(system_power_efficient_wq, &phydev->state_queue, + PHY_STATE_TIME * HZ); } static inline void mmd_phy_indirect(struct mii_bus *bus, int prtad, int devad, diff --git a/drivers/video/console/fbcon.c b/drivers/video/console/fbcon.c index a92783e480e6..0d8f98c79a6c 100644 --- a/drivers/video/console/fbcon.c +++ b/drivers/video/console/fbcon.c @@ -404,7 +404,7 @@ static void cursor_timer_handler(unsigned long dev_addr) struct fb_info *info = (struct fb_info *) dev_addr; struct fbcon_ops *ops = info->fbcon_par; - schedule_work(&info->queue); + queue_work(system_power_efficient_wq, &info->queue); mod_timer(&ops->cursor_timer, jiffies + HZ/5); } diff --git a/include/linux/irqchip/arm-gic.h b/include/linux/irqchip/arm-gic.h index b5696108c06e..40643ca79cd9 100644 --- a/include/linux/irqchip/arm-gic.h +++ b/include/linux/irqchip/arm-gic.h @@ -31,6 +31,8 @@ #define GIC_DIST_TARGET 0x800 #define GIC_DIST_CONFIG 0xc00 #define GIC_DIST_SOFTINT 0xf00 +#define GIC_DIST_SGI_PENDING_CLEAR 0xf10 +#define GIC_DIST_SGI_PENDING_SET 0xf20 #define GICH_HCR 0x0 #define GICH_VTR 0x4 @@ -75,6 +77,11 @@ static inline void gic_init(unsigned int nr, int start, gic_init_bases(nr, start, dist, cpu, 0, NULL); } +void gic_send_sgi(unsigned int cpu_id, unsigned int irq); +int gic_get_cpu_id(unsigned int cpu); +void gic_migrate_target(unsigned int new_cpu_id); +unsigned long gic_get_sgir_physaddr(void); + #endif /* __ASSEMBLY */ #endif diff --git a/include/linux/sched.h b/include/linux/sched.h index 178a8d909f14..0e2a546cdade 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -885,6 +885,13 @@ void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms); bool cpus_share_cache(int this_cpu, int that_cpu); +#ifdef CONFIG_SCHED_HMP +struct hmp_domain { + struct cpumask cpus; + struct cpumask possible_cpus; + struct list_head hmp_domains; +}; +#endif /* CONFIG_SCHED_HMP */ #else /* CONFIG_SMP */ struct sched_domain_attr; @@ -931,6 +938,12 @@ struct sched_avg { u64 last_runnable_update; s64 decay_count; unsigned long load_avg_contrib; + unsigned long load_avg_ratio; +#ifdef CONFIG_SCHED_HMP + u64 hmp_last_up_migration; + u64 hmp_last_down_migration; +#endif + u32 usage_avg_sum; }; #ifdef CONFIG_SCHEDSTATS diff --git a/include/linux/vmstat.h b/include/linux/vmstat.h index c586679b6fef..a30ab7910ff4 100644 --- a/include/linux/vmstat.h +++ b/include/linux/vmstat.h @@ -198,7 +198,7 @@ extern void __inc_zone_state(struct zone *, enum zone_stat_item); extern void dec_zone_state(struct zone *, enum zone_stat_item); extern void __dec_zone_state(struct zone *, enum zone_stat_item); -void refresh_cpu_vm_stats(int); +bool refresh_cpu_vm_stats(int); void refresh_zone_stat_thresholds(void); void drain_zonestat(struct zone *zone, struct per_cpu_pageset *); diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h index 623488fdc1f5..a9f4119c7e2e 100644 --- a/include/linux/workqueue.h +++ b/include/linux/workqueue.h @@ -303,6 +303,33 @@ enum { WQ_CPU_INTENSIVE = 1 << 5, /* cpu instensive workqueue */ WQ_SYSFS = 1 << 6, /* visible in sysfs, see wq_sysfs_register() */ + /* + * Per-cpu workqueues are generally preferred because they tend to + * show better performance thanks to cache locality. Per-cpu + * workqueues exclude the scheduler from choosing the CPU to + * execute the worker threads, which has an unfortunate side effect + * of increasing power consumption. + * + * The scheduler considers a CPU idle if it doesn't have any task + * to execute and tries to keep idle cores idle to conserve power; + * however, for example, a per-cpu work item scheduled from an + * interrupt handler on an idle CPU will force the scheduler to + * excute the work item on that CPU breaking the idleness, which in + * turn may lead to more scheduling choices which are sub-optimal + * in terms of power consumption. + * + * Workqueues marked with WQ_POWER_EFFICIENT are per-cpu by default + * but become unbound if workqueue.power_efficient kernel param is + * specified. Per-cpu workqueues which are identified to + * contribute significantly to power-consumption are identified and + * marked with this flag and enabling the power_efficient mode + * leads to noticeable power saving at the cost of small + * performance disadvantage. + * + * http://thread.gmane.org/gmane.linux.kernel/1480396 + */ + WQ_POWER_EFFICIENT = 1 << 7, + __WQ_DRAINING = 1 << 16, /* internal: workqueue is draining */ __WQ_ORDERED = 1 << 17, /* internal: workqueue is ordered */ @@ -333,11 +360,19 @@ enum { * * system_freezable_wq is equivalent to system_wq except that it's * freezable. + * + * *_power_efficient_wq are inclined towards saving power and converted + * into WQ_UNBOUND variants if 'wq_power_efficient' is enabled; otherwise, + * they are same as their non-power-efficient counterparts - e.g. + * system_power_efficient_wq is identical to system_wq if + * 'wq_power_efficient' is disabled. See WQ_POWER_EFFICIENT for more info. */ extern struct workqueue_struct *system_wq; extern struct workqueue_struct *system_long_wq; extern struct workqueue_struct *system_unbound_wq; extern struct workqueue_struct *system_freezable_wq; +extern struct workqueue_struct *system_power_efficient_wq; +extern struct workqueue_struct *system_freezable_power_efficient_wq; static inline struct workqueue_struct * __deprecated __system_nrt_wq(void) { diff --git a/include/trace/events/power_cpu_migrate.h b/include/trace/events/power_cpu_migrate.h new file mode 100644 index 000000000000..f76dd4de625e --- /dev/null +++ b/include/trace/events/power_cpu_migrate.h @@ -0,0 +1,67 @@ +#undef TRACE_SYSTEM +#define TRACE_SYSTEM power + +#if !defined(_TRACE_POWER_CPU_MIGRATE_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_POWER_CPU_MIGRATE_H + +#include <linux/tracepoint.h> + +#define __cpu_migrate_proto \ + TP_PROTO(u64 timestamp, \ + u32 cpu_hwid) +#define __cpu_migrate_args \ + TP_ARGS(timestamp, \ + cpu_hwid) + +DECLARE_EVENT_CLASS(cpu_migrate, + + __cpu_migrate_proto, + __cpu_migrate_args, + + TP_STRUCT__entry( + __field(u64, timestamp ) + __field(u32, cpu_hwid ) + ), + + TP_fast_assign( + __entry->timestamp = timestamp; + __entry->cpu_hwid = cpu_hwid; + ), + + TP_printk("timestamp=%llu cpu_hwid=0x%08lX", + (unsigned long long)__entry->timestamp, + (unsigned long)__entry->cpu_hwid + ) +); + +#define __define_cpu_migrate_event(name) \ + DEFINE_EVENT(cpu_migrate, cpu_migrate_##name, \ + __cpu_migrate_proto, \ + __cpu_migrate_args \ + ) + +__define_cpu_migrate_event(begin); +__define_cpu_migrate_event(finish); +__define_cpu_migrate_event(current); + +#undef __define_cpu_migrate +#undef __cpu_migrate_proto +#undef __cpu_migrate_args + +/* This file can get included multiple times, TRACE_HEADER_MULTI_READ at top */ +#ifndef _PWR_CPU_MIGRATE_EVENT_AVOID_DOUBLE_DEFINING +#define _PWR_CPU_MIGRATE_EVENT_AVOID_DOUBLE_DEFINING + +/* + * Set from_phys_cpu and to_phys_cpu to CPU_MIGRATE_ALL_CPUS to indicate + * a whole-cluster migration: + */ +#define CPU_MIGRATE_ALL_CPUS 0x80000000U +#endif + +#endif /* _TRACE_POWER_CPU_MIGRATE_H */ + +/* This part must be outside protection */ +#undef TRACE_INCLUDE_FILE +#define TRACE_INCLUDE_FILE power_cpu_migrate +#include <trace/define_trace.h> diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h index e5586caff67a..203e8e9933ba 100644 --- a/include/trace/events/sched.h +++ b/include/trace/events/sched.h @@ -430,6 +430,159 @@ TRACE_EVENT(sched_pi_setprio, __entry->oldprio, __entry->newprio) ); +/* + * Tracepoint for showing tracked load contribution. + */ +TRACE_EVENT(sched_task_load_contrib, + + TP_PROTO(struct task_struct *tsk, unsigned long load_contrib), + + TP_ARGS(tsk, load_contrib), + + TP_STRUCT__entry( + __array(char, comm, TASK_COMM_LEN) + __field(pid_t, pid) + __field(unsigned long, load_contrib) + ), + + TP_fast_assign( + memcpy(__entry->comm, tsk->comm, TASK_COMM_LEN); + __entry->pid = tsk->pid; + __entry->load_contrib = load_contrib; + ), + + TP_printk("comm=%s pid=%d load_contrib=%lu", + __entry->comm, __entry->pid, + __entry->load_contrib) +); + +/* + * Tracepoint for showing tracked task runnable ratio [0..1023]. + */ +TRACE_EVENT(sched_task_runnable_ratio, + + TP_PROTO(struct task_struct *tsk, unsigned long ratio), + + TP_ARGS(tsk, ratio), + + TP_STRUCT__entry( + __array(char, comm, TASK_COMM_LEN) + __field(pid_t, pid) + __field(unsigned long, ratio) + ), + + TP_fast_assign( + memcpy(__entry->comm, tsk->comm, TASK_COMM_LEN); + __entry->pid = tsk->pid; + __entry->ratio = ratio; + ), + + TP_printk("comm=%s pid=%d ratio=%lu", + __entry->comm, __entry->pid, + __entry->ratio) +); + +/* + * Tracepoint for showing tracked rq runnable ratio [0..1023]. + */ +TRACE_EVENT(sched_rq_runnable_ratio, + + TP_PROTO(int cpu, unsigned long ratio), + + TP_ARGS(cpu, ratio), + + TP_STRUCT__entry( + __field(int, cpu) + __field(unsigned long, ratio) + ), + + TP_fast_assign( + __entry->cpu = cpu; + __entry->ratio = ratio; + ), + + TP_printk("cpu=%d ratio=%lu", + __entry->cpu, + __entry->ratio) +); + +/* + * Tracepoint for showing tracked rq runnable load. + */ +TRACE_EVENT(sched_rq_runnable_load, + + TP_PROTO(int cpu, u64 load), + + TP_ARGS(cpu, load), + + TP_STRUCT__entry( + __field(int, cpu) + __field(u64, load) + ), + + TP_fast_assign( + __entry->cpu = cpu; + __entry->load = load; + ), + + TP_printk("cpu=%d load=%llu", + __entry->cpu, + __entry->load) +); + +/* + * Tracepoint for showing tracked task cpu usage ratio [0..1023]. + */ +TRACE_EVENT(sched_task_usage_ratio, + + TP_PROTO(struct task_struct *tsk, unsigned long ratio), + + TP_ARGS(tsk, ratio), + + TP_STRUCT__entry( + __array(char, comm, TASK_COMM_LEN) + __field(pid_t, pid) + __field(unsigned long, ratio) + ), + + TP_fast_assign( + memcpy(__entry->comm, tsk->comm, TASK_COMM_LEN); + __entry->pid = tsk->pid; + __entry->ratio = ratio; + ), + + TP_printk("comm=%s pid=%d ratio=%lu", + __entry->comm, __entry->pid, + __entry->ratio) +); + +/* + * Tracepoint for HMP (CONFIG_SCHED_HMP) task migrations. + */ +TRACE_EVENT(sched_hmp_migrate, + + TP_PROTO(struct task_struct *tsk, int dest, int force), + + TP_ARGS(tsk, dest, force), + + TP_STRUCT__entry( + __array(char, comm, TASK_COMM_LEN) + __field(pid_t, pid) + __field(int, dest) + __field(int, force) + ), + + TP_fast_assign( + memcpy(__entry->comm, tsk->comm, TASK_COMM_LEN); + __entry->pid = tsk->pid; + __entry->dest = dest; + __entry->force = force; + ), + + TP_printk("comm=%s pid=%d dest=%d force=%d", + __entry->comm, __entry->pid, + __entry->dest, __entry->force) +); #endif /* _TRACE_SCHED_H */ /* This part must be outside protection */ diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 192a302d6cfd..473b2b6eccb5 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -23,10 +23,27 @@ static struct lock_class_key irq_desc_lock_class; #if defined(CONFIG_SMP) +static int __init irq_affinity_setup(char *str) +{ + zalloc_cpumask_var(&irq_default_affinity, GFP_NOWAIT); + cpulist_parse(str, irq_default_affinity); + /* + * Set at least the boot cpu. We don't want to end up with + * bugreports caused by random comandline masks + */ + cpumask_set_cpu(smp_processor_id(), irq_default_affinity); + return 1; +} +__setup("irqaffinity=", irq_affinity_setup); + static void __init init_irq_default_affinity(void) { - alloc_cpumask_var(&irq_default_affinity, GFP_NOWAIT); - cpumask_setall(irq_default_affinity); +#ifdef CONFIG_CPUMASK_OFFSTACK + if (!irq_default_affinity) + zalloc_cpumask_var(&irq_default_affinity, GFP_NOWAIT); +#endif + if (cpumask_empty(irq_default_affinity)) + cpumask_setall(irq_default_affinity); } #else static void __init init_irq_default_affinity(void) diff --git a/kernel/power/Kconfig b/kernel/power/Kconfig index 5dfdc9ea180b..46455961a88f 100644 --- a/kernel/power/Kconfig +++ b/kernel/power/Kconfig @@ -263,6 +263,26 @@ config PM_GENERIC_DOMAINS bool depends on PM +config WQ_POWER_EFFICIENT_DEFAULT + bool "Enable workqueue power-efficient mode by default" + depends on PM + default n + help + Per-cpu workqueues are generally preferred because they show + better performance thanks to cache locality; unfortunately, + per-cpu workqueues tend to be more power hungry than unbound + workqueues. + + Enabling workqueue.power_efficient kernel parameter makes the + per-cpu workqueues which were observed to contribute + significantly to power consumption unbound, leading to measurably + lower power usage at the cost of small performance overhead. + + This config option determines whether workqueue.power_efficient + is enabled by default. + + If in doubt, say N. + config PM_GENERIC_DOMAINS_SLEEP def_bool y depends on PM_SLEEP && PM_GENERIC_DOMAINS diff --git a/kernel/sched/core.c b/kernel/sched/core.c index e8b335016c52..50d9e9849ce7 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -1617,6 +1617,20 @@ static void __sched_fork(struct task_struct *p) #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED) p->se.avg.runnable_avg_period = 0; p->se.avg.runnable_avg_sum = 0; +#ifdef CONFIG_SCHED_HMP + /* keep LOAD_AVG_MAX in sync with fair.c if load avg series is changed */ +#define LOAD_AVG_MAX 47742 + if (p->mm) { + p->se.avg.hmp_last_up_migration = 0; + p->se.avg.hmp_last_down_migration = 0; + p->se.avg.load_avg_ratio = 1023; + p->se.avg.load_avg_contrib = + (1023 * scale_load_down(p->se.load.weight)); + p->se.avg.runnable_avg_period = LOAD_AVG_MAX; + p->se.avg.runnable_avg_sum = LOAD_AVG_MAX; + p->se.avg.usage_avg_sum = LOAD_AVG_MAX; + } +#endif #endif #ifdef CONFIG_SCHEDSTATS memset(&p->se.statistics, 0, sizeof(p->se.statistics)); @@ -3813,6 +3827,8 @@ static struct task_struct *find_process_by_pid(pid_t pid) return pid ? find_task_by_vpid(pid) : current; } +extern struct cpumask hmp_slow_cpu_mask; + /* Actually do priority change: must hold rq lock. */ static void __setscheduler(struct rq *rq, struct task_struct *p, int policy, int prio) @@ -3822,8 +3838,13 @@ __setscheduler(struct rq *rq, struct task_struct *p, int policy, int prio) p->normal_prio = normal_prio(p); /* we are holding p->pi_lock already */ p->prio = rt_mutex_getprio(p); - if (rt_prio(p->prio)) + if (rt_prio(p->prio)) { p->sched_class = &rt_sched_class; +#ifdef CONFIG_SCHED_HMP + if (cpumask_equal(&p->cpus_allowed, cpu_all_mask)) + do_set_cpus_allowed(p, &hmp_slow_cpu_mask); +#endif + } else p->sched_class = &fair_sched_class; set_load_weight(p); diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index 75024a673520..fbd8caa83ef5 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -94,6 +94,7 @@ static void print_cfs_group_stats(struct seq_file *m, int cpu, struct task_group #ifdef CONFIG_SMP P(se->avg.runnable_avg_sum); P(se->avg.runnable_avg_period); + P(se->avg.usage_avg_sum); P(se->avg.load_avg_contrib); P(se->avg.decay_count); #endif @@ -223,6 +224,8 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) cfs_rq->tg_runnable_contrib); SEQ_printf(m, " .%-30s: %d\n", "tg->runnable_avg", atomic_read(&cfs_rq->tg->runnable_avg)); + SEQ_printf(m, " .%-30s: %d\n", "tg->usage_avg", + atomic_read(&cfs_rq->tg->usage_avg)); #endif print_cfs_group_stats(m, cpu, cfs_rq->tg); diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index c61a614465c8..c849d68a9b76 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -31,9 +31,20 @@ #include <linux/task_work.h> #include <trace/events/sched.h> +#ifdef CONFIG_HMP_VARIABLE_SCALE +#include <linux/sysfs.h> +#include <linux/vmalloc.h> +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE +/* Include cpufreq header to add a notifier so that cpu frequency + * scaling can track the current CPU frequency + */ +#include <linux/cpufreq.h> +#endif /* CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ +#endif /* CONFIG_HMP_VARIABLE_SCALE */ #include "sched.h" + /* * Targeted preemption latency for CPU-bound tasks: * (default: 6ms * (1 + ilog(ncpus)), units: nanoseconds) @@ -1201,8 +1212,95 @@ static u32 __compute_runnable_contrib(u64 n) return contrib + runnable_avg_yN_sum[n]; } -/* - * We can represent the historical contribution to runnable average as the +#ifdef CONFIG_HMP_VARIABLE_SCALE + +#define HMP_VARIABLE_SCALE_SHIFT 16ULL +struct hmp_global_attr { + struct attribute attr; + ssize_t (*show)(struct kobject *kobj, + struct attribute *attr, char *buf); + ssize_t (*store)(struct kobject *a, struct attribute *b, + const char *c, size_t count); + int *value; + int (*to_sysfs)(int); + int (*from_sysfs)(int); +}; + +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE +#define HMP_DATA_SYSFS_MAX 4 +#else +#define HMP_DATA_SYSFS_MAX 3 +#endif + +struct hmp_data_struct { +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + int freqinvar_load_scale_enabled; +#endif + int multiplier; /* used to scale the time delta */ + struct attribute_group attr_group; + struct attribute *attributes[HMP_DATA_SYSFS_MAX + 1]; + struct hmp_global_attr attr[HMP_DATA_SYSFS_MAX]; +} hmp_data; + +static u64 hmp_variable_scale_convert(u64 delta); +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE +/* Frequency-Invariant Load Modification: + * Loads are calculated as in PJT's patch however we also scale the current + * contribution in line with the frequency of the CPU that the task was + * executed on. + * In this version, we use a simple linear scale derived from the maximum + * frequency reported by CPUFreq. As an example: + * + * Consider that we ran a task for 100% of the previous interval. + * + * Our CPU was under asynchronous frequency control through one of the + * CPUFreq governors. + * + * The CPUFreq governor reports that it is able to scale the CPU between + * 500MHz and 1GHz. + * + * During the period, the CPU was running at 1GHz. + * + * In this case, our load contribution for that period is calculated as + * 1 * (number_of_active_microseconds) + * + * This results in our task being able to accumulate maximum load as normal. + * + * + * Consider now that our CPU was executing at 500MHz. + * + * We now scale the load contribution such that it is calculated as + * 0.5 * (number_of_active_microseconds) + * + * Our task can only record 50% maximum load during this period. + * + * This represents the task consuming 50% of the CPU's *possible* compute + * capacity. However the task did consume 100% of the CPU's *available* + * compute capacity which is the value seen by the CPUFreq governor and + * user-side CPU Utilization tools. + * + * Restricting tracked load to be scaled by the CPU's frequency accurately + * represents the consumption of possible compute capacity and allows the + * HMP migration's simple threshold migration strategy to interact more + * predictably with CPUFreq's asynchronous compute capacity changes. + */ +#define SCHED_FREQSCALE_SHIFT 10 +struct cpufreq_extents { + u32 curr_scale; + u32 min; + u32 max; + u32 flags; +}; +/* Flag set when the governor in use only allows one frequency. + * Disables scaling. + */ +#define SCHED_LOAD_FREQINVAR_SINGLEFREQ 0x01 + +static struct cpufreq_extents freq_scale[CONFIG_NR_CPUS]; +#endif /* CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ +#endif /* CONFIG_HMP_VARIABLE_SCALE */ + +/* We can represent the historical contribution to runnable average as the * coefficients of a geometric series. To do this we sub-divide our runnable * history into segments of approximately 1ms (1024us); label the segment that * occurred N-ms ago p_N, with p_0 corresponding to the current period, e.g. @@ -1231,13 +1329,24 @@ static u32 __compute_runnable_contrib(u64 n) */ static __always_inline int __update_entity_runnable_avg(u64 now, struct sched_avg *sa, - int runnable) + int runnable, + int running, + int cpu) { u64 delta, periods; u32 runnable_contrib; int delta_w, decayed = 0; +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + u64 scaled_delta; + u32 scaled_runnable_contrib; + int scaled_delta_w; + u32 curr_scale = 1024; +#endif /* CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ delta = now - sa->last_runnable_update; +#ifdef CONFIG_HMP_VARIABLE_SCALE + delta = hmp_variable_scale_convert(delta); +#endif /* * This should only happen when time goes backwards, which it * unfortunately does during sched clock init when we swap over to TSC. @@ -1256,6 +1365,12 @@ static __always_inline int __update_entity_runnable_avg(u64 now, return 0; sa->last_runnable_update = now; +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + /* retrieve scale factor for load */ + if (hmp_data.freqinvar_load_scale_enabled) + curr_scale = freq_scale[cpu].curr_scale; +#endif /* CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ + /* delta_w is the amount already accumulated against our next period */ delta_w = sa->runnable_avg_period % 1024; if (delta + delta_w >= 1024) { @@ -1268,8 +1383,20 @@ static __always_inline int __update_entity_runnable_avg(u64 now, * period and accrue it. */ delta_w = 1024 - delta_w; + /* scale runnable time if necessary */ +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + scaled_delta_w = (delta_w * curr_scale) + >> SCHED_FREQSCALE_SHIFT; + if (runnable) + sa->runnable_avg_sum += scaled_delta_w; + if (running) + sa->usage_avg_sum += scaled_delta_w; +#else if (runnable) sa->runnable_avg_sum += delta_w; + if (running) + sa->usage_avg_sum += delta_w; +#endif /* #ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ sa->runnable_avg_period += delta_w; delta -= delta_w; @@ -1277,22 +1404,49 @@ static __always_inline int __update_entity_runnable_avg(u64 now, /* Figure out how many additional periods this update spans */ periods = delta / 1024; delta %= 1024; - + /* decay the load we have accumulated so far */ sa->runnable_avg_sum = decay_load(sa->runnable_avg_sum, periods + 1); sa->runnable_avg_period = decay_load(sa->runnable_avg_period, periods + 1); - + sa->usage_avg_sum = decay_load(sa->usage_avg_sum, periods + 1); + /* add the contribution from this period */ /* Efficiently calculate \sum (1..n_period) 1024*y^i */ runnable_contrib = __compute_runnable_contrib(periods); + /* Apply load scaling if necessary. + * Note that multiplying the whole series is same as + * multiplying all terms + */ +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + scaled_runnable_contrib = (runnable_contrib * curr_scale) + >> SCHED_FREQSCALE_SHIFT; + if (runnable) + sa->runnable_avg_sum += scaled_runnable_contrib; + if (running) + sa->usage_avg_sum += scaled_runnable_contrib; +#else if (runnable) sa->runnable_avg_sum += runnable_contrib; + if (running) + sa->usage_avg_sum += runnable_contrib; +#endif /* CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ sa->runnable_avg_period += runnable_contrib; } /* Remainder of delta accrued against u_0` */ + /* scale if necessary */ +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + scaled_delta = ((delta * curr_scale) >> SCHED_FREQSCALE_SHIFT); + if (runnable) + sa->runnable_avg_sum += scaled_delta; + if (running) + sa->usage_avg_sum += scaled_delta; +#else if (runnable) sa->runnable_avg_sum += delta; + if (running) + sa->usage_avg_sum += delta; +#endif /* CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ sa->runnable_avg_period += delta; return decayed; @@ -1338,16 +1492,28 @@ static inline void __update_tg_runnable_avg(struct sched_avg *sa, struct cfs_rq *cfs_rq) { struct task_group *tg = cfs_rq->tg; - long contrib; + long contrib, usage_contrib; /* The fraction of a cpu used by this cfs_rq */ contrib = div_u64(sa->runnable_avg_sum << NICE_0_SHIFT, sa->runnable_avg_period + 1); contrib -= cfs_rq->tg_runnable_contrib; - if (abs(contrib) > cfs_rq->tg_runnable_contrib / 64) { + usage_contrib = div_u64(sa->usage_avg_sum << NICE_0_SHIFT, + sa->runnable_avg_period + 1); + usage_contrib -= cfs_rq->tg_usage_contrib; + + /* + * contrib/usage at this point represent deltas, only update if they + * are substantive. + */ + if ((abs(contrib) > cfs_rq->tg_runnable_contrib / 64) || + (abs(usage_contrib) > cfs_rq->tg_usage_contrib / 64)) { atomic_add(contrib, &tg->runnable_avg); cfs_rq->tg_runnable_contrib += contrib; + + atomic_add(usage_contrib, &tg->usage_avg); + cfs_rq->tg_usage_contrib += usage_contrib; } } @@ -1408,6 +1574,11 @@ static inline void __update_task_entity_contrib(struct sched_entity *se) contrib = se->avg.runnable_avg_sum * scale_load_down(se->load.weight); contrib /= (se->avg.runnable_avg_period + 1); se->avg.load_avg_contrib = scale_load(contrib); + trace_sched_task_load_contrib(task_of(se), se->avg.load_avg_contrib); + contrib = se->avg.runnable_avg_sum * scale_load_down(NICE_0_LOAD); + contrib /= (se->avg.runnable_avg_period + 1); + se->avg.load_avg_ratio = scale_load(contrib); + trace_sched_task_runnable_ratio(task_of(se), se->avg.load_avg_ratio); } /* Compute the current contribution to load_avg by se, return any delta */ @@ -1443,7 +1614,11 @@ static inline void update_entity_load_avg(struct sched_entity *se, struct cfs_rq *cfs_rq = cfs_rq_of(se); long contrib_delta; u64 now; + int cpu = -1; /* not used in normal case */ +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + cpu = cfs_rq->rq->cpu; +#endif /* * For a group entity we need to use their owned cfs_rq_clock_task() in * case they are the parent of a throttled hierarchy. @@ -1453,7 +1628,8 @@ static inline void update_entity_load_avg(struct sched_entity *se, else now = cfs_rq_clock_task(group_cfs_rq(se)); - if (!__update_entity_runnable_avg(now, &se->avg, se->on_rq)) + if (!__update_entity_runnable_avg(now, &se->avg, se->on_rq, + cfs_rq->curr == se, cpu)) return; contrib_delta = __update_entity_load_avg_contrib(se); @@ -1497,8 +1673,19 @@ static void update_cfs_rq_blocked_load(struct cfs_rq *cfs_rq, int force_update) static inline void update_rq_runnable_avg(struct rq *rq, int runnable) { - __update_entity_runnable_avg(rq->clock_task, &rq->avg, runnable); + u32 contrib; + int cpu = -1; /* not used in normal case */ + +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + cpu = rq->cpu; +#endif + __update_entity_runnable_avg(rq->clock_task, &rq->avg, runnable, + runnable, cpu); __update_tg_runnable_avg(&rq->avg, &rq->cfs); + contrib = rq->avg.runnable_avg_sum * scale_load_down(1024); + contrib /= (rq->avg.runnable_avg_period + 1); + trace_sched_rq_runnable_ratio(cpu_of(rq), scale_load(contrib)); + trace_sched_rq_runnable_load(cpu_of(rq), rq->cfs.runnable_load_avg); } /* Add the load generated by se into cfs_rq's child load-average */ @@ -1886,6 +2073,7 @@ set_next_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) */ update_stats_wait_end(cfs_rq, se); __dequeue_entity(cfs_rq, se); + update_entity_load_avg(se, 1); } update_stats_curr_start(cfs_rq, se); @@ -3314,6 +3502,437 @@ done: return target; } +#ifdef CONFIG_SCHED_HMP +/* + * Heterogenous multiprocessor (HMP) optimizations + * + * The cpu types are distinguished using a list of hmp_domains + * which each represent one cpu type using a cpumask. + * The list is assumed ordered by compute capacity with the + * fastest domain first. + */ +DEFINE_PER_CPU(struct hmp_domain *, hmp_cpu_domain); + +extern void __init arch_get_hmp_domains(struct list_head *hmp_domains_list); + +/* Setup hmp_domains */ +static int __init hmp_cpu_mask_setup(void) +{ + char buf[64]; + struct hmp_domain *domain; + struct list_head *pos; + int dc, cpu; + + pr_debug("Initializing HMP scheduler:\n"); + + /* Initialize hmp_domains using platform code */ + arch_get_hmp_domains(&hmp_domains); + if (list_empty(&hmp_domains)) { + pr_debug("HMP domain list is empty!\n"); + return 0; + } + + /* Print hmp_domains */ + dc = 0; + list_for_each(pos, &hmp_domains) { + domain = list_entry(pos, struct hmp_domain, hmp_domains); + cpulist_scnprintf(buf, 64, &domain->possible_cpus); + pr_debug(" HMP domain %d: %s\n", dc, buf); + + for_each_cpu_mask(cpu, domain->possible_cpus) { + per_cpu(hmp_cpu_domain, cpu) = domain; + } + dc++; + } + + return 1; +} + +static struct hmp_domain *hmp_get_hmp_domain_for_cpu(int cpu) +{ + struct hmp_domain *domain; + struct list_head *pos; + + list_for_each(pos, &hmp_domains) { + domain = list_entry(pos, struct hmp_domain, hmp_domains); + if(cpumask_test_cpu(cpu, &domain->possible_cpus)) + return domain; + } + return NULL; +} + +static void hmp_online_cpu(int cpu) +{ + struct hmp_domain *domain = hmp_get_hmp_domain_for_cpu(cpu); + + if(domain) + cpumask_set_cpu(cpu, &domain->cpus); +} + +static void hmp_offline_cpu(int cpu) +{ + struct hmp_domain *domain = hmp_get_hmp_domain_for_cpu(cpu); + + if(domain) + cpumask_clear_cpu(cpu, &domain->cpus); +} + +/* + * Migration thresholds should be in the range [0..1023] + * hmp_up_threshold: min. load required for migrating tasks to a faster cpu + * hmp_down_threshold: max. load allowed for tasks migrating to a slower cpu + * The default values (512, 256) offer good responsiveness, but may need + * tweaking suit particular needs. + * + * hmp_up_prio: Only up migrate task with high priority (<hmp_up_prio) + * hmp_next_up_threshold: Delay before next up migration (1024 ~= 1 ms) + * hmp_next_down_threshold: Delay before next down migration (1024 ~= 1 ms) + */ +unsigned int hmp_up_threshold = 512; +unsigned int hmp_down_threshold = 256; +#ifdef CONFIG_SCHED_HMP_PRIO_FILTER +unsigned int hmp_up_prio = NICE_TO_PRIO(CONFIG_SCHED_HMP_PRIO_FILTER_VAL); +#endif +unsigned int hmp_next_up_threshold = 4096; +unsigned int hmp_next_down_threshold = 4096; + +static unsigned int hmp_up_migration(int cpu, int *target_cpu, struct sched_entity *se); +static unsigned int hmp_down_migration(int cpu, struct sched_entity *se); +static inline unsigned int hmp_domain_min_load(struct hmp_domain *hmpd, + int *min_cpu); + +/* Check if cpu is in fastest hmp_domain */ +static inline unsigned int hmp_cpu_is_fastest(int cpu) +{ + struct list_head *pos; + + pos = &hmp_cpu_domain(cpu)->hmp_domains; + return pos == hmp_domains.next; +} + +/* Check if cpu is in slowest hmp_domain */ +static inline unsigned int hmp_cpu_is_slowest(int cpu) +{ + struct list_head *pos; + + pos = &hmp_cpu_domain(cpu)->hmp_domains; + return list_is_last(pos, &hmp_domains); +} + +/* Next (slower) hmp_domain relative to cpu */ +static inline struct hmp_domain *hmp_slower_domain(int cpu) +{ + struct list_head *pos; + + pos = &hmp_cpu_domain(cpu)->hmp_domains; + return list_entry(pos->next, struct hmp_domain, hmp_domains); +} + +/* Previous (faster) hmp_domain relative to cpu */ +static inline struct hmp_domain *hmp_faster_domain(int cpu) +{ + struct list_head *pos; + + pos = &hmp_cpu_domain(cpu)->hmp_domains; + return list_entry(pos->prev, struct hmp_domain, hmp_domains); +} + +/* + * Selects a cpu in previous (faster) hmp_domain + * Note that cpumask_any_and() returns the first cpu in the cpumask + */ +static inline unsigned int hmp_select_faster_cpu(struct task_struct *tsk, + int cpu) +{ + int lowest_cpu=NR_CPUS; + __always_unused int lowest_ratio = hmp_domain_min_load(hmp_faster_domain(cpu), &lowest_cpu); + /* + * If the lowest-loaded CPU in the domain is allowed by the task affinity + * select that one, otherwise select one which is allowed + */ + if(lowest_cpu != NR_CPUS && cpumask_test_cpu(lowest_cpu,tsk_cpus_allowed(tsk))) + return lowest_cpu; + else + return cpumask_any_and(&hmp_faster_domain(cpu)->cpus, + tsk_cpus_allowed(tsk)); +} + +/* + * Selects a cpu in next (slower) hmp_domain + * Note that cpumask_any_and() returns the first cpu in the cpumask + */ +static inline unsigned int hmp_select_slower_cpu(struct task_struct *tsk, + int cpu) +{ + int lowest_cpu=NR_CPUS; + __always_unused int lowest_ratio = hmp_domain_min_load(hmp_slower_domain(cpu), &lowest_cpu); + /* + * If the lowest-loaded CPU in the domain is allowed by the task affinity + * select that one, otherwise select one which is allowed + */ + if(lowest_cpu != NR_CPUS && cpumask_test_cpu(lowest_cpu,tsk_cpus_allowed(tsk))) + return lowest_cpu; + else + return cpumask_any_and(&hmp_slower_domain(cpu)->cpus, + tsk_cpus_allowed(tsk)); +} + +static inline void hmp_next_up_delay(struct sched_entity *se, int cpu) +{ + struct cfs_rq *cfs_rq = &cpu_rq(cpu)->cfs; + + se->avg.hmp_last_up_migration = cfs_rq_clock_task(cfs_rq); + se->avg.hmp_last_down_migration = 0; +} + +static inline void hmp_next_down_delay(struct sched_entity *se, int cpu) +{ + struct cfs_rq *cfs_rq = &cpu_rq(cpu)->cfs; + + se->avg.hmp_last_down_migration = cfs_rq_clock_task(cfs_rq); + se->avg.hmp_last_up_migration = 0; +} + +#ifdef CONFIG_HMP_VARIABLE_SCALE +/* + * Heterogenous multiprocessor (HMP) optimizations + * + * These functions allow to change the growing speed of the load_avg_ratio + * by default it goes from 0 to 0.5 in LOAD_AVG_PERIOD = 32ms + * This can now be changed with /sys/kernel/hmp/load_avg_period_ms. + * + * These functions also allow to change the up and down threshold of HMP + * using /sys/kernel/hmp/{up,down}_threshold. + * Both must be between 0 and 1023. The threshold that is compared + * to the load_avg_ratio is up_threshold/1024 and down_threshold/1024. + * + * For instance, if load_avg_period = 64 and up_threshold = 512, an idle + * task with a load of 0 will reach the threshold after 64ms of busy loop. + * + * Changing load_avg_periods_ms has the same effect than changing the + * default scaling factor Y=1002/1024 in the load_avg_ratio computation to + * (1002/1024.0)^(LOAD_AVG_PERIOD/load_avg_period_ms), but the last one + * could trigger overflows. + * For instance, with Y = 1023/1024 in __update_task_entity_contrib() + * "contrib = se->avg.runnable_avg_sum * scale_load_down(se->load.weight);" + * could be overflowed for a weight > 2^12 even is the load_avg_contrib + * should still be a 32bits result. This would not happen by multiplicating + * delta time by 1/22 and setting load_avg_period_ms = 706. + */ + +/* + * By scaling the delta time it end-up increasing or decrease the + * growing speed of the per entity load_avg_ratio + * The scale factor hmp_data.multiplier is a fixed point + * number: (32-HMP_VARIABLE_SCALE_SHIFT).HMP_VARIABLE_SCALE_SHIFT + */ +static u64 hmp_variable_scale_convert(u64 delta) +{ + u64 high = delta >> 32ULL; + u64 low = delta & 0xffffffffULL; + low *= hmp_data.multiplier; + high *= hmp_data.multiplier; + return (low >> HMP_VARIABLE_SCALE_SHIFT) + + (high << (32ULL - HMP_VARIABLE_SCALE_SHIFT)); +} + +static ssize_t hmp_show(struct kobject *kobj, + struct attribute *attr, char *buf) +{ + ssize_t ret = 0; + struct hmp_global_attr *hmp_attr = + container_of(attr, struct hmp_global_attr, attr); + int temp = *(hmp_attr->value); + if (hmp_attr->to_sysfs != NULL) + temp = hmp_attr->to_sysfs(temp); + ret = sprintf(buf, "%d\n", temp); + return ret; +} + +static ssize_t hmp_store(struct kobject *a, struct attribute *attr, + const char *buf, size_t count) +{ + int temp; + ssize_t ret = count; + struct hmp_global_attr *hmp_attr = + container_of(attr, struct hmp_global_attr, attr); + char *str = vmalloc(count + 1); + if (str == NULL) + return -ENOMEM; + memcpy(str, buf, count); + str[count] = 0; + if (sscanf(str, "%d", &temp) < 1) + ret = -EINVAL; + else { + if (hmp_attr->from_sysfs != NULL) + temp = hmp_attr->from_sysfs(temp); + if (temp < 0) + ret = -EINVAL; + else + *(hmp_attr->value) = temp; + } + vfree(str); + return ret; +} + +static int hmp_period_tofrom_sysfs(int value) +{ + return (LOAD_AVG_PERIOD << HMP_VARIABLE_SCALE_SHIFT) / value; +} + +/* max value for threshold is 1024 */ +static int hmp_theshold_from_sysfs(int value) +{ + if (value > 1024) + return -1; + return value; +} +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE +/* freqinvar control is only 0,1 off/on */ +static int hmp_freqinvar_from_sysfs(int value) +{ + if (value < 0 || value > 1) + return -1; + return value; +} +#endif +static void hmp_attr_add( + const char *name, + int *value, + int (*to_sysfs)(int), + int (*from_sysfs)(int)) +{ + int i = 0; + while (hmp_data.attributes[i] != NULL) { + i++; + if (i >= HMP_DATA_SYSFS_MAX) + return; + } + hmp_data.attr[i].attr.mode = 0644; + hmp_data.attr[i].show = hmp_show; + hmp_data.attr[i].store = hmp_store; + hmp_data.attr[i].attr.name = name; + hmp_data.attr[i].value = value; + hmp_data.attr[i].to_sysfs = to_sysfs; + hmp_data.attr[i].from_sysfs = from_sysfs; + hmp_data.attributes[i] = &hmp_data.attr[i].attr; + hmp_data.attributes[i + 1] = NULL; +} + +static int hmp_attr_init(void) +{ + int ret; + memset(&hmp_data, sizeof(hmp_data), 0); + /* by default load_avg_period_ms == LOAD_AVG_PERIOD + * meaning no change + */ + hmp_data.multiplier = hmp_period_tofrom_sysfs(LOAD_AVG_PERIOD); + + hmp_attr_add("load_avg_period_ms", + &hmp_data.multiplier, + hmp_period_tofrom_sysfs, + hmp_period_tofrom_sysfs); + hmp_attr_add("up_threshold", + &hmp_up_threshold, + NULL, + hmp_theshold_from_sysfs); + hmp_attr_add("down_threshold", + &hmp_down_threshold, + NULL, + hmp_theshold_from_sysfs); +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE + /* default frequency-invariant scaling ON */ + hmp_data.freqinvar_load_scale_enabled = 1; + hmp_attr_add("frequency_invariant_load_scale", + &hmp_data.freqinvar_load_scale_enabled, + NULL, + hmp_freqinvar_from_sysfs); +#endif + hmp_data.attr_group.name = "hmp"; + hmp_data.attr_group.attrs = hmp_data.attributes; + ret = sysfs_create_group(kernel_kobj, + &hmp_data.attr_group); + return 0; +} +late_initcall(hmp_attr_init); +#endif /* CONFIG_HMP_VARIABLE_SCALE */ + +static inline unsigned int hmp_domain_min_load(struct hmp_domain *hmpd, + int *min_cpu) +{ + int cpu; + int min_cpu_runnable_temp = NR_CPUS; + unsigned long min_runnable_load = INT_MAX; + unsigned long contrib; + + for_each_cpu_mask(cpu, hmpd->cpus) { + /* don't use the divisor in the loop, just at the end */ + contrib = cpu_rq(cpu)->avg.runnable_avg_sum * scale_load_down(1024); + if (contrib < min_runnable_load) { + min_runnable_load = contrib; + min_cpu_runnable_temp = cpu; + } + } + + if (min_cpu) + *min_cpu = min_cpu_runnable_temp; + + /* domain will often have at least one empty CPU */ + return min_runnable_load ? min_runnable_load / (LOAD_AVG_MAX + 1) : 0; +} + +/* + * Calculate the task starvation + * This is the ratio of actually running time vs. runnable time. + * If the two are equal the task is getting the cpu time it needs or + * it is alone on the cpu and the cpu is fully utilized. + */ +static inline unsigned int hmp_task_starvation(struct sched_entity *se) +{ + u32 starvation; + + starvation = se->avg.usage_avg_sum * scale_load_down(NICE_0_LOAD); + starvation /= (se->avg.runnable_avg_sum + 1); + + return scale_load(starvation); +} + +static inline unsigned int hmp_offload_down(int cpu, struct sched_entity *se) +{ + int min_usage; + int dest_cpu = NR_CPUS; + + if (hmp_cpu_is_slowest(cpu)) + return NR_CPUS; + + /* Is the current domain fully loaded? */ + /* load < ~50% */ + min_usage = hmp_domain_min_load(hmp_cpu_domain(cpu), NULL); + if (min_usage < (NICE_0_LOAD>>1)) + return NR_CPUS; + + /* Is the task alone on the cpu? */ + if (cpu_rq(cpu)->cfs.nr_running < 2) + return NR_CPUS; + + /* Is the task actually starving? */ + /* >=25% ratio running/runnable = starving */ + if (hmp_task_starvation(se) > 768) + return NR_CPUS; + + /* Does the slower domain have spare cycles? */ + min_usage = hmp_domain_min_load(hmp_slower_domain(cpu), &dest_cpu); + /* load > 50% */ + if (min_usage > NICE_0_LOAD/2) + return NR_CPUS; + + if (cpumask_test_cpu(dest_cpu, &hmp_slower_domain(cpu)->cpus)) + return dest_cpu; + + return NR_CPUS; +} +#endif /* CONFIG_SCHED_HMP */ + /* * sched_balance_self: balance the current task (running on cpu) in domains * that have the 'flag' flag set. In practice, this is SD_BALANCE_FORK and @@ -3338,6 +3957,28 @@ select_task_rq_fair(struct task_struct *p, int sd_flag, int wake_flags) if (p->nr_cpus_allowed == 1) return prev_cpu; +#ifdef CONFIG_SCHED_HMP + /* always put non-kernel forking tasks on a big domain */ + if (p->mm && (sd_flag & SD_BALANCE_FORK)) { + if(hmp_cpu_is_fastest(prev_cpu)) { + struct hmp_domain *hmpdom = list_entry(&hmp_cpu_domain(prev_cpu)->hmp_domains, struct hmp_domain, hmp_domains); + __always_unused int lowest_ratio = hmp_domain_min_load(hmpdom, &new_cpu); + if(new_cpu != NR_CPUS && cpumask_test_cpu(new_cpu,tsk_cpus_allowed(p))) + return new_cpu; + else { + new_cpu = cpumask_any_and(&hmp_faster_domain(cpu)->cpus, + tsk_cpus_allowed(p)); + if(new_cpu < nr_cpu_ids) + return new_cpu; + } + } else { + new_cpu = hmp_select_faster_cpu(p, prev_cpu); + if (new_cpu != NR_CPUS) + return new_cpu; + } + } +#endif + if (sd_flag & SD_BALANCE_WAKE) { if (cpumask_test_cpu(cpu, tsk_cpus_allowed(p))) want_affine = 1; @@ -3412,6 +4053,23 @@ select_task_rq_fair(struct task_struct *p, int sd_flag, int wake_flags) unlock: rcu_read_unlock(); +#ifdef CONFIG_SCHED_HMP + if (hmp_up_migration(prev_cpu, &new_cpu, &p->se)) { + hmp_next_up_delay(&p->se, new_cpu); + trace_sched_hmp_migrate(p, new_cpu, 0); + return new_cpu; + } + if (hmp_down_migration(prev_cpu, &p->se)) { + new_cpu = hmp_select_slower_cpu(p, prev_cpu); + hmp_next_down_delay(&p->se, new_cpu); + trace_sched_hmp_migrate(p, new_cpu, 0); + return new_cpu; + } + /* Make sure that the task stays in its previous hmp domain */ + if (!cpumask_test_cpu(new_cpu, &hmp_cpu_domain(prev_cpu)->cpus)) + return prev_cpu; +#endif + return new_cpu; } @@ -3945,7 +4603,6 @@ int can_migrate_task(struct task_struct *p, struct lb_env *env) * 1) task is cache cold, or * 2) too many balance attempts have failed. */ - tsk_cache_hot = task_hot(p, env->src_rq->clock_task, env->sd); if (!tsk_cache_hot || env->sd->nr_balance_failed > env->sd->cache_nice_tries) { @@ -5371,7 +6028,11 @@ static struct { static inline int find_new_ilb(int call_cpu) { int ilb = cpumask_first(nohz.idle_cpus_mask); - +#ifdef CONFIG_SCHED_HMP + /* restrict nohz balancing to occur in the same hmp domain */ + ilb = cpumask_first_and(nohz.idle_cpus_mask, + &((struct hmp_domain *)hmp_cpu_domain(call_cpu))->cpus); +#endif if (ilb < nr_cpu_ids && idle_cpu(ilb)) return ilb; @@ -5650,6 +6311,18 @@ static inline int nohz_kick_needed(struct rq *rq, int cpu) if (time_before(now, nohz.next_balance)) return 0; +#ifdef CONFIG_SCHED_HMP + /* + * Bail out if there are no nohz CPUs in our + * HMP domain, since we will move tasks between + * domains through wakeup and force balancing + * as necessary based upon task load. + */ + if (cpumask_first_and(nohz.idle_cpus_mask, + &((struct hmp_domain *)hmp_cpu_domain(cpu))->cpus) >= nr_cpu_ids) + return 0; +#endif + if (rq->nr_running >= 2) goto need_kick; @@ -5682,6 +6355,300 @@ need_kick: static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle) { } #endif +#ifdef CONFIG_SCHED_HMP +/* Check if task should migrate to a faster cpu */ +static unsigned int hmp_up_migration(int cpu, int *target_cpu, struct sched_entity *se) +{ + struct task_struct *p = task_of(se); + struct cfs_rq *cfs_rq = &cpu_rq(cpu)->cfs; + u64 now; + + if (target_cpu) + *target_cpu = NR_CPUS; + + if (hmp_cpu_is_fastest(cpu)) + return 0; + +#ifdef CONFIG_SCHED_HMP_PRIO_FILTER + /* Filter by task priority */ + if (p->prio >= hmp_up_prio) + return 0; +#endif + if (se->avg.load_avg_ratio < hmp_up_threshold) + return 0; + + /* Let the task load settle before doing another up migration */ + now = cfs_rq_clock_task(cfs_rq); + if (((now - se->avg.hmp_last_up_migration) >> 10) + < hmp_next_up_threshold) + return 0; + + /* Target domain load < 94% */ + if (hmp_domain_min_load(hmp_faster_domain(cpu), target_cpu) + > NICE_0_LOAD-64) + return 0; + + if (cpumask_intersects(&hmp_faster_domain(cpu)->cpus, + tsk_cpus_allowed(p))) + return 1; + + return 0; +} + +/* Check if task should migrate to a slower cpu */ +static unsigned int hmp_down_migration(int cpu, struct sched_entity *se) +{ + struct task_struct *p = task_of(se); + struct cfs_rq *cfs_rq = &cpu_rq(cpu)->cfs; + u64 now; + + if (hmp_cpu_is_slowest(cpu)) + return 0; + +#ifdef CONFIG_SCHED_HMP_PRIO_FILTER + /* Filter by task priority */ + if ((p->prio >= hmp_up_prio) && + cpumask_intersects(&hmp_slower_domain(cpu)->cpus, + tsk_cpus_allowed(p))) { + return 1; + } +#endif + + /* Let the task load settle before doing another down migration */ + now = cfs_rq_clock_task(cfs_rq); + if (((now - se->avg.hmp_last_down_migration) >> 10) + < hmp_next_down_threshold) + return 0; + + if (cpumask_intersects(&hmp_slower_domain(cpu)->cpus, + tsk_cpus_allowed(p)) + && se->avg.load_avg_ratio < hmp_down_threshold) { + return 1; + } + return 0; +} + +/* + * hmp_can_migrate_task - may task p from runqueue rq be migrated to this_cpu? + * Ideally this function should be merged with can_migrate_task() to avoid + * redundant code. + */ +static int hmp_can_migrate_task(struct task_struct *p, struct lb_env *env) +{ + int tsk_cache_hot = 0; + + /* + * We do not migrate tasks that are: + * 1) running (obviously), or + * 2) cannot be migrated to this CPU due to cpus_allowed + */ + if (!cpumask_test_cpu(env->dst_cpu, tsk_cpus_allowed(p))) { + schedstat_inc(p, se.statistics.nr_failed_migrations_affine); + return 0; + } + env->flags &= ~LBF_ALL_PINNED; + + if (task_running(env->src_rq, p)) { + schedstat_inc(p, se.statistics.nr_failed_migrations_running); + return 0; + } + + /* + * Aggressive migration if: + * 1) task is cache cold, or + * 2) too many balance attempts have failed. + */ + + tsk_cache_hot = task_hot(p, env->src_rq->clock_task, env->sd); + if (!tsk_cache_hot || + env->sd->nr_balance_failed > env->sd->cache_nice_tries) { +#ifdef CONFIG_SCHEDSTATS + if (tsk_cache_hot) { + schedstat_inc(env->sd, lb_hot_gained[env->idle]); + schedstat_inc(p, se.statistics.nr_forced_migrations); + } +#endif + return 1; + } + + return 1; +} + +/* + * move_specific_task tries to move a specific task. + * Returns 1 if successful and 0 otherwise. + * Called with both runqueues locked. + */ +static int move_specific_task(struct lb_env *env, struct task_struct *pm) +{ + struct task_struct *p, *n; + + list_for_each_entry_safe(p, n, &env->src_rq->cfs_tasks, se.group_node) { + if (throttled_lb_pair(task_group(p), env->src_rq->cpu, + env->dst_cpu)) + continue; + + if (!hmp_can_migrate_task(p, env)) + continue; + /* Check if we found the right task */ + if (p != pm) + continue; + + move_task(p, env); + /* + * Right now, this is only the third place move_task() + * is called, so we can safely collect move_task() + * stats here rather than inside move_task(). + */ + schedstat_inc(env->sd, lb_gained[env->idle]); + return 1; + } + return 0; +} + +/* + * hmp_active_task_migration_cpu_stop is run by cpu stopper and used to + * migrate a specific task from one runqueue to another. + * hmp_force_up_migration uses this to push a currently running task + * off a runqueue. + * Based on active_load_balance_stop_cpu and can potentially be merged. + */ +static int hmp_active_task_migration_cpu_stop(void *data) +{ + struct rq *busiest_rq = data; + struct task_struct *p = busiest_rq->migrate_task; + int busiest_cpu = cpu_of(busiest_rq); + int target_cpu = busiest_rq->push_cpu; + struct rq *target_rq = cpu_rq(target_cpu); + struct sched_domain *sd; + + raw_spin_lock_irq(&busiest_rq->lock); + /* make sure the requested cpu hasn't gone down in the meantime */ + if (unlikely(busiest_cpu != smp_processor_id() || + !busiest_rq->active_balance)) { + goto out_unlock; + } + /* Is there any task to move? */ + if (busiest_rq->nr_running <= 1) + goto out_unlock; + /* Task has migrated meanwhile, abort forced migration */ + if (task_rq(p) != busiest_rq) + goto out_unlock; + /* + * This condition is "impossible", if it occurs + * we need to fix it. Originally reported by + * Bjorn Helgaas on a 128-cpu setup. + */ + BUG_ON(busiest_rq == target_rq); + + /* move a task from busiest_rq to target_rq */ + double_lock_balance(busiest_rq, target_rq); + + /* Search for an sd spanning us and the target CPU. */ + rcu_read_lock(); + for_each_domain(target_cpu, sd) { + if (cpumask_test_cpu(busiest_cpu, sched_domain_span(sd))) + break; + } + + if (likely(sd)) { + struct lb_env env = { + .sd = sd, + .dst_cpu = target_cpu, + .dst_rq = target_rq, + .src_cpu = busiest_rq->cpu, + .src_rq = busiest_rq, + .idle = CPU_IDLE, + }; + + schedstat_inc(sd, alb_count); + + if (move_specific_task(&env, p)) + schedstat_inc(sd, alb_pushed); + else + schedstat_inc(sd, alb_failed); + } + rcu_read_unlock(); + double_unlock_balance(busiest_rq, target_rq); +out_unlock: + busiest_rq->active_balance = 0; + raw_spin_unlock_irq(&busiest_rq->lock); + return 0; +} + +static DEFINE_SPINLOCK(hmp_force_migration); + +/* + * hmp_force_up_migration checks runqueues for tasks that need to + * be actively migrated to a faster cpu. + */ +static void hmp_force_up_migration(int this_cpu) +{ + int cpu, target_cpu; + struct sched_entity *curr; + struct rq *target; + unsigned long flags; + unsigned int force; + struct task_struct *p; + + if (!spin_trylock(&hmp_force_migration)) + return; + for_each_online_cpu(cpu) { + force = 0; + target = cpu_rq(cpu); + raw_spin_lock_irqsave(&target->lock, flags); + curr = target->cfs.curr; + if (!curr) { + raw_spin_unlock_irqrestore(&target->lock, flags); + continue; + } + if (!entity_is_task(curr)) { + struct cfs_rq *cfs_rq; + + cfs_rq = group_cfs_rq(curr); + while (cfs_rq) { + curr = cfs_rq->curr; + cfs_rq = group_cfs_rq(curr); + } + } + p = task_of(curr); + if (hmp_up_migration(cpu, &target_cpu, curr)) { + if (!target->active_balance) { + target->active_balance = 1; + target->push_cpu = target_cpu; + target->migrate_task = p; + force = 1; + trace_sched_hmp_migrate(p, target->push_cpu, 1); + hmp_next_up_delay(&p->se, target->push_cpu); + } + } + if (!force && !target->active_balance) { + /* + * For now we just check the currently running task. + * Selecting the lightest task for offloading will + * require extensive book keeping. + */ + target->push_cpu = hmp_offload_down(cpu, curr); + if (target->push_cpu < NR_CPUS) { + target->active_balance = 1; + target->migrate_task = p; + force = 1; + trace_sched_hmp_migrate(p, target->push_cpu, 2); + hmp_next_down_delay(&p->se, target->push_cpu); + } + } + raw_spin_unlock_irqrestore(&target->lock, flags); + if (force) + stop_one_cpu_nowait(cpu_of(target), + hmp_active_task_migration_cpu_stop, + target, &target->active_balance_work); + } + spin_unlock(&hmp_force_migration); +} +#else +static void hmp_force_up_migration(int this_cpu) { } +#endif /* CONFIG_SCHED_HMP */ + /* * run_rebalance_domains is triggered when needed from the scheduler tick. * Also triggered for nohz idle balancing (with nohz_balancing_kick set). @@ -5693,6 +6660,8 @@ static void run_rebalance_domains(struct softirq_action *h) enum cpu_idle_type idle = this_rq->idle_balance ? CPU_IDLE : CPU_NOT_IDLE; + hmp_force_up_migration(this_cpu); + rebalance_domains(this_cpu, idle); /* @@ -5725,11 +6694,17 @@ void trigger_load_balance(struct rq *rq, int cpu) static void rq_online_fair(struct rq *rq) { +#ifdef CONFIG_SCHED_HMP + hmp_online_cpu(rq->cpu); +#endif update_sysctl(); } static void rq_offline_fair(struct rq *rq) { +#ifdef CONFIG_SCHED_HMP + hmp_offline_cpu(rq->cpu); +#endif update_sysctl(); /* Ensure any throttled groups are reachable by pick_next_task */ @@ -6192,6 +7167,139 @@ __init void init_sched_fair_class(void) zalloc_cpumask_var(&nohz.idle_cpus_mask, GFP_NOWAIT); cpu_notifier(sched_ilb_notifier, 0); #endif + +#ifdef CONFIG_SCHED_HMP + hmp_cpu_mask_setup(); +#endif #endif /* SMP */ } + +#ifdef CONFIG_HMP_FREQUENCY_INVARIANT_SCALE +static u32 cpufreq_calc_scale(u32 min, u32 max, u32 curr) +{ + u32 result = curr / max; + return result; +} + +/* Called when the CPU Frequency is changed. + * Once for each CPU. + */ +static int cpufreq_callback(struct notifier_block *nb, + unsigned long val, void *data) +{ + struct cpufreq_freqs *freq = data; + int cpu = freq->cpu; + struct cpufreq_extents *extents; + + if (freq->flags & CPUFREQ_CONST_LOOPS) + return NOTIFY_OK; + + if (val != CPUFREQ_POSTCHANGE) + return NOTIFY_OK; + + /* if dynamic load scale is disabled, set the load scale to 1.0 */ + if (!hmp_data.freqinvar_load_scale_enabled) { + freq_scale[cpu].curr_scale = 1024; + return NOTIFY_OK; + } + + extents = &freq_scale[cpu]; + if (extents->flags & SCHED_LOAD_FREQINVAR_SINGLEFREQ) { + /* If our governor was recognised as a single-freq governor, + * use 1.0 + */ + extents->curr_scale = 1024; + } else { + extents->curr_scale = cpufreq_calc_scale(extents->min, + extents->max, freq->new); + } + + return NOTIFY_OK; +} + +/* Called when the CPUFreq governor is changed. + * Only called for the CPUs which are actually changed by the + * userspace. + */ +static int cpufreq_policy_callback(struct notifier_block *nb, + unsigned long event, void *data) +{ + struct cpufreq_policy *policy = data; + struct cpufreq_extents *extents; + int cpu, singleFreq = 0; + static const char performance_governor[] = "performance"; + static const char powersave_governor[] = "powersave"; + + if (event == CPUFREQ_START) + return 0; + + if (event != CPUFREQ_INCOMPATIBLE) + return 0; + + /* CPUFreq governors do not accurately report the range of + * CPU Frequencies they will choose from. + * We recognise performance and powersave governors as + * single-frequency only. + */ + if (!strncmp(policy->governor->name, performance_governor, + strlen(performance_governor)) || + !strncmp(policy->governor->name, powersave_governor, + strlen(powersave_governor))) + singleFreq = 1; + + /* Make sure that all CPUs impacted by this policy are + * updated since we will only get a notification when the + * user explicitly changes the policy on a CPU. + */ + for_each_cpu(cpu, policy->cpus) { + extents = &freq_scale[cpu]; + extents->max = policy->max >> SCHED_FREQSCALE_SHIFT; + extents->min = policy->min >> SCHED_FREQSCALE_SHIFT; + if (!hmp_data.freqinvar_load_scale_enabled) { + extents->curr_scale = 1024; + } else if (singleFreq) { + extents->flags |= SCHED_LOAD_FREQINVAR_SINGLEFREQ; + extents->curr_scale = 1024; + } else { + extents->flags &= ~SCHED_LOAD_FREQINVAR_SINGLEFREQ; + extents->curr_scale = cpufreq_calc_scale(extents->min, + extents->max, policy->cur); + } + } + + return 0; +} + +static struct notifier_block cpufreq_notifier = { + .notifier_call = cpufreq_callback, +}; +static struct notifier_block cpufreq_policy_notifier = { + .notifier_call = cpufreq_policy_callback, +}; + +static int __init register_sched_cpufreq_notifier(void) +{ + int ret = 0; + + /* init safe defaults since there are no policies at registration */ + for (ret = 0; ret < CONFIG_NR_CPUS; ret++) { + /* safe defaults */ + freq_scale[ret].max = 1024; + freq_scale[ret].min = 1024; + freq_scale[ret].curr_scale = 1024; + } + + pr_info("sched: registering cpufreq notifiers for scale-invariant loads\n"); + ret = cpufreq_register_notifier(&cpufreq_policy_notifier, + CPUFREQ_POLICY_NOTIFIER); + + if (ret != -EINVAL) + ret = cpufreq_register_notifier(&cpufreq_notifier, + CPUFREQ_TRANSITION_NOTIFIER); + + return ret; +} + +core_initcall(register_sched_cpufreq_notifier); +#endif /* CONFIG_HMP_FREQUENCY_INVARIANT_SCALE */ diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index ce39224d6155..27f51ac86700 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -142,7 +142,7 @@ struct task_group { atomic_t load_weight; atomic64_t load_avg; - atomic_t runnable_avg; + atomic_t runnable_avg, usage_avg; #endif #ifdef CONFIG_RT_GROUP_SCHED @@ -279,7 +279,7 @@ struct cfs_rq { #endif /* CONFIG_FAIR_GROUP_SCHED */ /* These always depend on CONFIG_FAIR_GROUP_SCHED */ #ifdef CONFIG_FAIR_GROUP_SCHED - u32 tg_runnable_contrib; + u32 tg_runnable_contrib, tg_usage_contrib; u64 tg_load_contrib; #endif /* CONFIG_FAIR_GROUP_SCHED */ @@ -464,6 +464,9 @@ struct rq { int active_balance; int push_cpu; struct cpu_stop_work active_balance_work; +#ifdef CONFIG_SCHED_HMP + struct task_struct *migrate_task; +#endif /* cpu of this runqueue: */ int cpu; int online; @@ -642,6 +645,12 @@ static inline unsigned int group_first_cpu(struct sched_group *group) extern int group_balance_cpu(struct sched_group *sg); +#ifdef CONFIG_SCHED_HMP +static LIST_HEAD(hmp_domains); +DECLARE_PER_CPU(struct hmp_domain *, hmp_cpu_domain); +#define hmp_cpu_domain(cpu) (per_cpu(hmp_cpu_domain, (cpu))) +#endif /* CONFIG_SCHED_HMP */ + #endif /* CONFIG_SMP */ #include "stats.h" diff --git a/kernel/workqueue.c b/kernel/workqueue.c index ee8e29a2320c..f02c4a4a0c3c 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -272,6 +272,15 @@ static cpumask_var_t *wq_numa_possible_cpumask; static bool wq_disable_numa; module_param_named(disable_numa, wq_disable_numa, bool, 0444); +/* see the comment above the definition of WQ_POWER_EFFICIENT */ +#ifdef CONFIG_WQ_POWER_EFFICIENT_DEFAULT +static bool wq_power_efficient = true; +#else +static bool wq_power_efficient; +#endif + +module_param_named(power_efficient, wq_power_efficient, bool, 0444); + static bool wq_numa_enabled; /* unbound NUMA affinity enabled */ /* buf for wq_update_unbound_numa_attrs(), protected by CPU hotplug exclusion */ @@ -305,6 +314,10 @@ struct workqueue_struct *system_unbound_wq __read_mostly; EXPORT_SYMBOL_GPL(system_unbound_wq); struct workqueue_struct *system_freezable_wq __read_mostly; EXPORT_SYMBOL_GPL(system_freezable_wq); +struct workqueue_struct *system_power_efficient_wq __read_mostly; +EXPORT_SYMBOL_GPL(system_power_efficient_wq); +struct workqueue_struct *system_freezable_power_efficient_wq __read_mostly; +EXPORT_SYMBOL_GPL(system_freezable_power_efficient_wq); static int worker_thread(void *__worker); static void copy_workqueue_attrs(struct workqueue_attrs *to, @@ -4086,6 +4099,10 @@ struct workqueue_struct *__alloc_workqueue_key(const char *fmt, struct workqueue_struct *wq; struct pool_workqueue *pwq; + /* see the comment above the definition of WQ_POWER_EFFICIENT */ + if ((flags & WQ_POWER_EFFICIENT) && wq_power_efficient) + flags |= WQ_UNBOUND; + /* allocate wq and format name */ if (flags & WQ_UNBOUND) tbl_size = wq_numa_tbl_len * sizeof(wq->numa_pwq_tbl[0]); @@ -4985,8 +5002,15 @@ static int __init init_workqueues(void) WQ_UNBOUND_MAX_ACTIVE); system_freezable_wq = alloc_workqueue("events_freezable", WQ_FREEZABLE, 0); + system_power_efficient_wq = alloc_workqueue("events_power_efficient", + WQ_POWER_EFFICIENT, 0); + system_freezable_power_efficient_wq = alloc_workqueue("events_freezable_power_efficient", + WQ_FREEZABLE | WQ_POWER_EFFICIENT, + 0); BUG_ON(!system_wq || !system_highpri_wq || !system_long_wq || - !system_unbound_wq || !system_freezable_wq); + !system_unbound_wq || !system_freezable_wq || + !system_power_efficient_wq || + !system_freezable_power_efficient_wq); return 0; } early_initcall(init_workqueues); diff --git a/linaro/configs/android.conf b/linaro/configs/android.conf new file mode 100644 index 000000000000..bb90ecd9e162 --- /dev/null +++ b/linaro/configs/android.conf @@ -0,0 +1,31 @@ +CONFIG_IPV6=y +# CONFIG_IPV6_SIT is not set +CONFIG_PANIC_TIMEOUT=0 +CONFIG_HAS_WAKELOCK=y +CONFIG_WAKELOCK=y +CONFIG_BLK_DEV_LOOP=y +CONFIG_DM_CRYPT=y +CONFIG_POWER_SUPPLY=y +CONFIG_ANDROID_PARANOID_NETWORK=y +CONFIG_NET_ACTIVITY_STATS=y +CONFIG_INPUT_MISC=y +CONFIG_INPUT_UINPUT=y +CONFIG_INPUT_GPIO=y +CONFIG_USB_G_ANDROID=y +CONFIG_SWITCH=y +CONFIG_STAGING=y +CONFIG_ANDROID=y +CONFIG_ANDROID_BINDER_IPC=y +CONFIG_ASHMEM=y +CONFIG_ANDROID_LOGGER=y +CONFIG_ANDROID_TIMED_OUTPUT=y +CONFIG_ANDROID_TIMED_GPIO=y +CONFIG_ANDROID_LOW_MEMORY_KILLER=y +CONFIG_ANDROID_INTF_ALARM_DEV=y +CONFIG_CRYPTO_TWOFISH=y +CONFIG_BLK_DEV_RAM=y +CONFIG_BLK_DEV_RAM_COUNT=16 +CONFIG_BLK_DEV_RAM_SIZE=16384 +CONFIG_FUSE_FS=y +CONFIG_CPU_FREQ_GOV_INTERACTIVE=y +CONFIG_CPU_FREQ_DEFAULT_GOV_INTERACTIVE=y diff --git a/linaro/configs/big-LITTLE-IKS.conf b/linaro/configs/big-LITTLE-IKS.conf new file mode 100644 index 000000000000..b067fde86eaa --- /dev/null +++ b/linaro/configs/big-LITTLE-IKS.conf @@ -0,0 +1,5 @@ +CONFIG_BIG_LITTLE=y +CONFIG_BL_SWITCHER=y +CONFIG_ARM_DT_BL_CPUFREQ=y +CONFIG_ARM_VEXPRESS_BL_CPUFREQ=y +CONFIG_CPU_FREQ_GOV_USERSPACE=y diff --git a/linaro/configs/big-LITTLE-MP.conf b/linaro/configs/big-LITTLE-MP.conf new file mode 100644 index 000000000000..0bbc603a13e5 --- /dev/null +++ b/linaro/configs/big-LITTLE-MP.conf @@ -0,0 +1,11 @@ +CONFIG_CGROUPS=y +CONFIG_CGROUP_SCHED=y +CONFIG_FAIR_GROUP_SCHED=y +CONFIG_NO_HZ=y +CONFIG_SCHED_MC=y +CONFIG_DISABLE_CPU_SCHED_DOMAIN_BALANCE=y +CONFIG_SCHED_HMP=y +CONFIG_HMP_FAST_CPU_MASK="" +CONFIG_HMP_SLOW_CPU_MASK="" +CONFIG_HMP_VARIABLE_SCALE=y +CONFIG_HMP_FREQUENCY_INVARIANT_SCALE=y diff --git a/linaro/configs/debug.conf b/linaro/configs/debug.conf new file mode 100644 index 000000000000..36980566b2d8 --- /dev/null +++ b/linaro/configs/debug.conf @@ -0,0 +1 @@ +CONFIG_PROVE_LOCKING=y diff --git a/linaro/configs/distribution.conf b/linaro/configs/distribution.conf new file mode 100644 index 000000000000..fbcfed1b6ce8 --- /dev/null +++ b/linaro/configs/distribution.conf @@ -0,0 +1,44 @@ +# CONFIG_LOCALVERSION_AUTO is not set +CONFIG_CGROUPS=y +# CONFIG_COMPAT_BRK is not set +CONFIG_DEFAULT_MMAP_MIN_ADDR=32768 +CONFIG_SECCOMP=y +CONFIG_CC_STACKPROTECTOR=y +CONFIG_SYN_COOKIES=y +CONFIG_IPV6=y +CONFIG_NETLABEL=y +CONFIG_BRIDGE_NETFILTER=y +CONFIG_NF_CONNTRACK=m +CONFIG_NETFILTER_XT_CONNMARK=m +CONFIG_NETFILTER_XT_MARK=m +CONFIG_NETFILTER_XT_TARGET_CHECKSUM=m +CONFIG_NF_CONNTRACK_IPV4=m +CONFIG_NF_NAT_IPV4=m +CONFIG_IP_NF_IPTABLES=m +CONFIG_IP_NF_FILTER=m +CONFIG_IP_NF_MANGLE=m +CONFIG_NF_CONNTRACK_IPV6=m +CONFIG_NF_NAT_IPV6=m +CONFIG_IP6_NF_IPTABLES=m +CONFIG_IP6_NF_FILTER=m +CONFIG_IP6_NF_MANGLE=m +CONFIG_BRIDGE_NF_EBTABLES=m +CONFIG_BRIDGE_EBT_MARK_T=m +CONFIG_BRIDGE=m +CONFIG_DEVTMPFS=y +CONFIG_DEVTMPFS_MOUNT=y +CONFIG_BLK_DEV_RAM=y +CONFIG_BLK_DEV_RAM_SIZE=65536 +CONFIG_INPUT_MISC=y +CONFIG_INPUT_UINPUT=y +# CONFIG_DEVKMEM is not set +CONFIG_FRAMEBUFFER_CONSOLE=y +CONFIG_AUTOFS4_FS=y +CONFIG_TMPFS_POSIX_ACL=y +CONFIG_STRICT_DEVMEM=y +CONFIG_SECURITY=y +CONFIG_LSM_MMAP_MIN_ADDR=0 +CONFIG_SECURITY_SELINUX=y +CONFIG_SECURITY_SMACK=y +CONFIG_SECURITY_APPARMOR=y +CONFIG_DEFAULT_SECURITY_APPARMOR=y diff --git a/linaro/configs/kvm-guest.conf b/linaro/configs/kvm-guest.conf new file mode 100644 index 000000000000..00e84a3ba1ec --- /dev/null +++ b/linaro/configs/kvm-guest.conf @@ -0,0 +1,11 @@ +CONFIG_BALLOON_COMPACTION=y +CONFIG_VIRTIO_BLK=y +CONFIG_VIRTIO_NET=y +CONFIG_HVC_DRIVER=y +CONFIG_VIRTIO_CONSOLE=y +CONFIG_VIRTIO=y +CONFIG_VIRTIO_BALLOON=y +CONFIG_VIRTIO_MMIO=y +CONFIG_VIRTIO_MMIO_CMDLINE_DEVICES=y +CONFIG_VIRTUALIZATION=y +# CONFIG_THUMB2_KERNEL is not set diff --git a/linaro/configs/kvm-host.conf b/linaro/configs/kvm-host.conf new file mode 100644 index 000000000000..21a40e031372 --- /dev/null +++ b/linaro/configs/kvm-host.conf @@ -0,0 +1,11 @@ +CONFIG_VIRTUALIZATION=y +CONFIG_ARM_LPAE=y +CONFIG_ARM_VIRT_EXT=y +CONFIG_HAVE_KVM_IRQCHIP=y +CONFIG_KVM_ARM_HOST=y +CONFIG_KVM_ARM_MAX_VCPUS=4 +CONFIG_KVM_ARM_TIMER=y +CONFIG_KVM_ARM_VGIC=y +CONFIG_KVM_MMIO=y +CONFIG_KVM=y +CONFIG_BLK_DEV_NBD=m diff --git a/linaro/configs/linaro-base.conf b/linaro/configs/linaro-base.conf new file mode 100644 index 000000000000..947ca1f50935 --- /dev/null +++ b/linaro/configs/linaro-base.conf @@ -0,0 +1,94 @@ +CONFIG_SYSVIPC=y +CONFIG_POSIX_MQUEUE=y +CONFIG_BSD_PROCESS_ACCT=y +CONFIG_IKCONFIG=y +CONFIG_IKCONFIG_PROC=y +CONFIG_LOG_BUF_SHIFT=16 +CONFIG_BLK_DEV_INITRD=y +CONFIG_EMBEDDED=y +CONFIG_HOTPLUG=y +CONFIG_PERF_EVENTS=y +CONFIG_SLAB=y +CONFIG_PROFILING=y +CONFIG_OPROFILE=y +CONFIG_MODULES=y +CONFIG_MODULE_UNLOAD=y +CONFIG_NO_HZ=y +CONFIG_HIGH_RES_TIMERS=y +CONFIG_SMP=y +CONFIG_SCHED_MC=y +CONFIG_SCHED_SMT=y +CONFIG_THUMB2_KERNEL=y +CONFIG_AEABI=y +# CONFIG_OABI_COMPAT is not set +CONFIG_CPU_FREQ=y +CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y +CONFIG_CPU_IDLE=y +CONFIG_BINFMT_MISC=y +CONFIG_MD=y +CONFIG_BLK_DEV_DM=y +CONFIG_NET=y +CONFIG_PACKET=y +CONFIG_UNIX=y +CONFIG_XFRM_USER=y +CONFIG_NET_KEY=y +CONFIG_NET_KEY_MIGRATE=y +CONFIG_INET=y +CONFIG_IP_MULTICAST=y +CONFIG_IP_PNP=y +CONFIG_IP_PNP_DHCP=y +CONFIG_IP_PNP_BOOTP=y +CONFIG_IP_PNP_RARP=y +# CONFIG_INET_LRO is not set +CONFIG_NETFILTER=y +CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug" +CONFIG_CONNECTOR=y +CONFIG_MTD=y +CONFIG_MTD_CMDLINE_PARTS=y +CONFIG_MTD_BLOCK=y +CONFIG_MTD_OOPS=y +CONFIG_MTD_CFI=y +CONFIG_MTD_CFI_INTELEXT=y +CONFIG_MTD_NAND=y +CONFIG_NETDEVICES=y +CONFIG_EXT2_FS=y +CONFIG_EXT3_FS=y +CONFIG_EXT4_FS=y +CONFIG_BTRFS_FS=y +CONFIG_QUOTA=y +CONFIG_QFMT_V2=y +CONFIG_MSDOS_FS=y +CONFIG_VFAT_FS=y +CONFIG_TMPFS=y +CONFIG_ECRYPT_FS=y +CONFIG_JFFS2_FS=y +CONFIG_JFFS2_SUMMARY=y +CONFIG_JFFS2_FS_XATTR=y +CONFIG_JFFS2_COMPRESSION_OPTIONS=y +CONFIG_JFFS2_LZO=y +CONFIG_JFFS2_RUBIN=y +CONFIG_CRAMFS=y +CONFIG_NETWORK_FILESYSTEMS=y +CONFIG_NFS_FS=y +# CONFIG_NFS_V2 is not set +CONFIG_NFS_V3=y +CONFIG_NFS_V3_ACL=y +CONFIG_NFS_V4=y +CONFIG_ROOT_NFS=y +CONFIG_NLS_CODEPAGE_437=y +CONFIG_NLS_ISO8859_1=y +CONFIG_PRINTK_TIME=y +CONFIG_MAGIC_SYSRQ=y +CONFIG_DEBUG_FS=y +CONFIG_SCHEDSTATS=y +CONFIG_TIMER_STATS=y +CONFIG_KEYS=y +CONFIG_CRYPTO_MICHAEL_MIC=y +CONFIG_CRC_CCITT=y +CONFIG_CRC_T10DIF=y +CONFIG_CRC_ITU_T=y +CONFIG_CRC7=y +CONFIG_HW_PERF_EVENTS=y +CONFIG_FUNCTION_TRACER=y +CONFIG_ENABLE_DEFAULT_TRACERS=y +CONFIG_PROC_DEVICETREE=y diff --git a/linaro/configs/ubuntu-minimal.conf b/linaro/configs/ubuntu-minimal.conf new file mode 120000 index 000000000000..794e82f3bc17 --- /dev/null +++ b/linaro/configs/ubuntu-minimal.conf @@ -0,0 +1 @@ +distribution.conf
\ No newline at end of file diff --git a/linaro/configs/xen.conf b/linaro/configs/xen.conf new file mode 100644 index 000000000000..d24fabbea076 --- /dev/null +++ b/linaro/configs/xen.conf @@ -0,0 +1,7 @@ +CONFIG_XEN=y +CONFIG_XEN_NETDEV_FRONTEND=y +CONFIG_XEN_NETDEV_BACKEND=y +CONFIG_XEN_BLKDEV_FRONTEND=y +CONFIG_XEN_BLKDEV_BACKEND=y +CONFIG_XENFS=y +CONFIG_XEN_COMPAT_XENFS=y diff --git a/mm/vmstat.c b/mm/vmstat.c index f42745e65780..b916a43a6b37 100644 --- a/mm/vmstat.c +++ b/mm/vmstat.c @@ -14,6 +14,7 @@ #include <linux/module.h> #include <linux/slab.h> #include <linux/cpu.h> +#include <linux/cpumask.h> #include <linux/vmstat.h> #include <linux/sched.h> #include <linux/math64.h> @@ -432,11 +433,12 @@ EXPORT_SYMBOL(dec_zone_page_state); * with the global counters. These could cause remote node cache line * bouncing and will have to be only done when necessary. */ -void refresh_cpu_vm_stats(int cpu) +bool refresh_cpu_vm_stats(int cpu) { struct zone *zone; int i; int global_diff[NR_VM_ZONE_STAT_ITEMS] = { 0, }; + bool vm_activity = false; for_each_populated_zone(zone) { struct per_cpu_pageset *p; @@ -483,14 +485,21 @@ void refresh_cpu_vm_stats(int cpu) if (p->expire) continue; - if (p->pcp.count) + if (p->pcp.count) { + vm_activity = true; drain_zone_pages(zone, &p->pcp); + } #endif } for (i = 0; i < NR_VM_ZONE_STAT_ITEMS; i++) - if (global_diff[i]) + if (global_diff[i]) { atomic_long_add(global_diff[i], &vm_stat[i]); + vm_activity = true; + } + + return vm_activity; + } /* @@ -1174,22 +1183,72 @@ static const struct file_operations proc_vmstat_file_operations = { #ifdef CONFIG_SMP static DEFINE_PER_CPU(struct delayed_work, vmstat_work); int sysctl_stat_interval __read_mostly = HZ; +static struct cpumask vmstat_off_cpus; +struct delayed_work vmstat_monitor_work; -static void vmstat_update(struct work_struct *w) +static inline bool need_vmstat(int cpu) { - refresh_cpu_vm_stats(smp_processor_id()); - schedule_delayed_work(&__get_cpu_var(vmstat_work), - round_jiffies_relative(sysctl_stat_interval)); + struct zone *zone; + int i; + + for_each_populated_zone(zone) { + struct per_cpu_pageset *p; + + p = per_cpu_ptr(zone->pageset, cpu); + + for (i = 0; i < NR_VM_ZONE_STAT_ITEMS; i++) + if (p->vm_stat_diff[i]) + return true; + + if (zone_to_nid(zone) != numa_node_id() && p->pcp.count) + return true; + } + + return false; } -static void __cpuinit start_cpu_timer(int cpu) +static void vmstat_update(struct work_struct *w); + +static void start_cpu_timer(int cpu) { struct delayed_work *work = &per_cpu(vmstat_work, cpu); - INIT_DEFERRABLE_WORK(work, vmstat_update); + cpumask_clear_cpu(cpu, &vmstat_off_cpus); schedule_delayed_work_on(cpu, work, __round_jiffies_relative(HZ, cpu)); } +static void __cpuinit setup_cpu_timer(int cpu) +{ + struct delayed_work *work = &per_cpu(vmstat_work, cpu); + + INIT_DEFERRABLE_WORK(work, vmstat_update); + start_cpu_timer(cpu); +} + +static void vmstat_update_monitor(struct work_struct *w) +{ + int cpu; + + for_each_cpu_and(cpu, &vmstat_off_cpus, cpu_online_mask) + if (need_vmstat(cpu)) + start_cpu_timer(cpu); + + queue_delayed_work(system_unbound_wq, &vmstat_monitor_work, + round_jiffies_relative(sysctl_stat_interval)); +} + + +static void vmstat_update(struct work_struct *w) +{ + int cpu = smp_processor_id(); + + if (likely(refresh_cpu_vm_stats(cpu))) + schedule_delayed_work(&__get_cpu_var(vmstat_work), + round_jiffies_relative(sysctl_stat_interval)); + else + cpumask_set_cpu(cpu, &vmstat_off_cpus); +} + /* * Use the cpu notifier to insure that the thresholds are recalculated * when necessary. @@ -1204,17 +1263,19 @@ static int __cpuinit vmstat_cpuup_callback(struct notifier_block *nfb, case CPU_ONLINE: case CPU_ONLINE_FROZEN: refresh_zone_stat_thresholds(); - start_cpu_timer(cpu); + setup_cpu_timer(cpu); node_set_state(cpu_to_node(cpu), N_CPU); break; case CPU_DOWN_PREPARE: case CPU_DOWN_PREPARE_FROZEN: - cancel_delayed_work_sync(&per_cpu(vmstat_work, cpu)); - per_cpu(vmstat_work, cpu).work.func = NULL; + if (!cpumask_test_cpu(cpu, &vmstat_off_cpus)) { + cancel_delayed_work_sync(&per_cpu(vmstat_work, cpu)); + per_cpu(vmstat_work, cpu).work.func = NULL; + } break; case CPU_DOWN_FAILED: case CPU_DOWN_FAILED_FROZEN: - start_cpu_timer(cpu); + setup_cpu_timer(cpu); break; case CPU_DEAD: case CPU_DEAD_FROZEN: @@ -1237,8 +1298,14 @@ static int __init setup_vmstat(void) register_cpu_notifier(&vmstat_notifier); + INIT_DEFERRABLE_WORK(&vmstat_monitor_work, + vmstat_update_monitor); + queue_delayed_work(system_unbound_wq, + &vmstat_monitor_work, + round_jiffies_relative(HZ)); + for_each_online_cpu(cpu) - start_cpu_timer(cpu); + setup_cpu_timer(cpu); #endif #ifdef CONFIG_PROC_FS proc_create("buddyinfo", S_IRUGO, NULL, &fragmentation_file_operations); |