linux/kernel/sched/fair.c

5604 lines
142 KiB
C
Raw Normal View History

/*
* Completely Fair Scheduling (CFS) Class (SCHED_NORMAL/SCHED_BATCH)
*
* Copyright (C) 2007 Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
*
* Interactivity improvements by Mike Galbraith
* (C) 2007 Mike Galbraith <efault@gmx.de>
*
* Various enhancements by Dmitry Adamushko.
* (C) 2007 Dmitry Adamushko <dmitry.adamushko@gmail.com>
*
* Group scheduling enhancements by Srivatsa Vaddagiri
* Copyright IBM Corporation, 2007
* Author: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
*
* Scaled math optimizations by Thomas Gleixner
* Copyright (C) 2007, Thomas Gleixner <tglx@linutronix.de>
*
* Adaptive scheduling granularity, math enhancements by Peter Zijlstra
* Copyright (C) 2007 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
*/
#include <linux/latencytop.h>
#include <linux/sched.h>
#include <linux/cpumask.h>
#include <linux/slab.h>
#include <linux/profile.h>
#include <linux/interrupt.h>
#include <trace/events/sched.h>
#include "sched.h"
/*
* Targeted preemption latency for CPU-bound tasks:
* (default: 6ms * (1 + ilog(ncpus)), units: nanoseconds)
*
* NOTE: this latency value is not the same as the concept of
* 'timeslice length' - timeslices in CFS are of variable length
* and have no persistent notion like in traditional, time-slice
* based scheduling concepts.
*
* (to see the precise effective timeslice length of your workload,
* run vmstat and monitor the context-switches (cs) field)
*/
unsigned int sysctl_sched_latency = 6000000ULL;
unsigned int normalized_sysctl_sched_latency = 6000000ULL;
/*
* The initial- and re-scaling of tunables is configurable
* (default SCHED_TUNABLESCALING_LOG = *(1+ilog(ncpus))
*
* Options are:
* SCHED_TUNABLESCALING_NONE - unscaled, always *1
* SCHED_TUNABLESCALING_LOG - scaled logarithmical, *1+ilog(ncpus)
* SCHED_TUNABLESCALING_LINEAR - scaled linear, *ncpus
*/
enum sched_tunable_scaling sysctl_sched_tunable_scaling
= SCHED_TUNABLESCALING_LOG;
/*
* Minimal preemption granularity for CPU-bound tasks:
* (default: 0.75 msec * (1 + ilog(ncpus)), units: nanoseconds)
*/
unsigned int sysctl_sched_min_granularity = 750000ULL;
unsigned int normalized_sysctl_sched_min_granularity = 750000ULL;
/*
* is kept at sysctl_sched_latency / sysctl_sched_min_granularity
*/
static unsigned int sched_nr_latency = 8;
/*
* After fork, child runs first. If set to 0 (default) then
* parent will (try to) run first.
*/
unsigned int sysctl_sched_child_runs_first __read_mostly;
/*
* SCHED_OTHER wake-up granularity.
* (default: 1 msec * (1 + ilog(ncpus)), units: nanoseconds)
*
* This option delays the preemption effects of decoupled workloads
* and reduces their over-scheduling. Synchronous workloads will still
* have immediate wakeup/sleep latencies.
*/
unsigned int sysctl_sched_wakeup_granularity = 1000000UL;
unsigned int normalized_sysctl_sched_wakeup_granularity = 1000000UL;
const_debug unsigned int sysctl_sched_migration_cost = 500000UL;
/*
* The exponential sliding window over which load is averaged for shares
* distribution.
* (default: 10msec)
*/
unsigned int __read_mostly sysctl_sched_shares_window = 10000000UL;
#ifdef CONFIG_CFS_BANDWIDTH
/*
* Amount of runtime to allocate from global (tg) to local (per-cfs_rq) pool
* each time a cfs_rq requests quota.
*
* Note: in the case that the slice exceeds the runtime remaining (either due
* to consumption or the quota being specified to be smaller than the slice)
* we will always only issue the remaining available time.
*
* default: 5 msec, units: microseconds
*/
unsigned int sysctl_sched_cfs_bandwidth_slice = 5000UL;
#endif
/*
* Increase the granularity value when there are more CPUs,
* because with more CPUs the 'effective latency' as visible
* to users decreases. But the relationship is not linear,
* so pick a second-best guess by going with the log2 of the
* number of CPUs.
*
* This idea comes from the SD scheduler of Con Kolivas:
*/
static int get_update_sysctl_factor(void)
{
unsigned int cpus = min_t(int, num_online_cpus(), 8);
unsigned int factor;
switch (sysctl_sched_tunable_scaling) {
case SCHED_TUNABLESCALING_NONE:
factor = 1;
break;
case SCHED_TUNABLESCALING_LINEAR:
factor = cpus;
break;
case SCHED_TUNABLESCALING_LOG:
default:
factor = 1 + ilog2(cpus);
break;
}
return factor;
}
static void update_sysctl(void)
{
unsigned int factor = get_update_sysctl_factor();
#define SET_SYSCTL(name) \
(sysctl_##name = (factor) * normalized_sysctl_##name)
SET_SYSCTL(sched_min_granularity);
SET_SYSCTL(sched_latency);
SET_SYSCTL(sched_wakeup_granularity);
#undef SET_SYSCTL
}
void sched_init_granularity(void)
{
update_sysctl();
}
#if BITS_PER_LONG == 32
# define WMULT_CONST (~0UL)
#else
# define WMULT_CONST (1UL << 32)
#endif
#define WMULT_SHIFT 32
/*
* Shift right and round:
*/
#define SRR(x, y) (((x) + (1UL << ((y) - 1))) >> (y))
/*
* delta *= weight / lw
*/
static unsigned long
calc_delta_mine(unsigned long delta_exec, unsigned long weight,
struct load_weight *lw)
{
u64 tmp;
/*
* weight can be less than 2^SCHED_LOAD_RESOLUTION for task group sched
* entities since MIN_SHARES = 2. Treat weight as 1 if less than
* 2^SCHED_LOAD_RESOLUTION.
*/
if (likely(weight > (1UL << SCHED_LOAD_RESOLUTION)))
tmp = (u64)delta_exec * scale_load_down(weight);
else
tmp = (u64)delta_exec;
if (!lw->inv_weight) {
unsigned long w = scale_load_down(lw->weight);
if (BITS_PER_LONG > 32 && unlikely(w >= WMULT_CONST))
lw->inv_weight = 1;
else if (unlikely(!w))
lw->inv_weight = WMULT_CONST;
else
lw->inv_weight = WMULT_CONST / w;
}
/*
* Check whether we'd overflow the 64-bit multiplication:
*/
if (unlikely(tmp > WMULT_CONST))
tmp = SRR(SRR(tmp, WMULT_SHIFT/2) * lw->inv_weight,
WMULT_SHIFT/2);
else
tmp = SRR(tmp * lw->inv_weight, WMULT_SHIFT);
return (unsigned long)min(tmp, (u64)(unsigned long)LONG_MAX);
}
const struct sched_class fair_sched_class;
/**************************************************************
* CFS operations on generic schedulable entities:
*/
#ifdef CONFIG_FAIR_GROUP_SCHED
/* cpu runqueue to which this cfs_rq is attached */
static inline struct rq *rq_of(struct cfs_rq *cfs_rq)
{
return cfs_rq->rq;
}
/* An entity is a task if it doesn't "own" a runqueue */
#define entity_is_task(se) (!se->my_q)
static inline struct task_struct *task_of(struct sched_entity *se)
{
#ifdef CONFIG_SCHED_DEBUG
WARN_ON_ONCE(!entity_is_task(se));
#endif
return container_of(se, struct task_struct, se);
}
/* Walk up scheduling entities hierarchy */
#define for_each_sched_entity(se) \
for (; se; se = se->parent)
static inline struct cfs_rq *task_cfs_rq(struct task_struct *p)
{
return p->se.cfs_rq;
}
/* runqueue on which this entity is (to be) queued */
static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se)
{
return se->cfs_rq;
}
/* runqueue "owned" by this group */
static inline struct cfs_rq *group_cfs_rq(struct sched_entity *grp)
{
return grp->my_q;
}
static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
if (!cfs_rq->on_list) {
/*
* Ensure we either appear before our parent (if already
* enqueued) or force our parent to appear after us when it is
* enqueued. The fact that we always enqueue bottom-up
* reduces this to two cases.
*/
if (cfs_rq->tg->parent &&
cfs_rq->tg->parent->cfs_rq[cpu_of(rq_of(cfs_rq))]->on_list) {
list_add_rcu(&cfs_rq->leaf_cfs_rq_list,
&rq_of(cfs_rq)->leaf_cfs_rq_list);
} else {
list_add_tail_rcu(&cfs_rq->leaf_cfs_rq_list,
&rq_of(cfs_rq)->leaf_cfs_rq_list);
}
cfs_rq->on_list = 1;
}
}
static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
if (cfs_rq->on_list) {
list_del_rcu(&cfs_rq->leaf_cfs_rq_list);
cfs_rq->on_list = 0;
}
}
/* Iterate thr' all leaf cfs_rq's on a runqueue */
#define for_each_leaf_cfs_rq(rq, cfs_rq) \
list_for_each_entry_rcu(cfs_rq, &rq->leaf_cfs_rq_list, leaf_cfs_rq_list)
/* Do the two (enqueued) entities belong to the same group ? */
static inline int
is_same_group(struct sched_entity *se, struct sched_entity *pse)
{
if (se->cfs_rq == pse->cfs_rq)
return 1;
return 0;
}
static inline struct sched_entity *parent_entity(struct sched_entity *se)
{
return se->parent;
}
/* return depth at which a sched entity is present in the hierarchy */
static inline int depth_se(struct sched_entity *se)
{
int depth = 0;
for_each_sched_entity(se)
depth++;
return depth;
}
static void
find_matching_se(struct sched_entity **se, struct sched_entity **pse)
{
int se_depth, pse_depth;
/*
* preemption test can be made between sibling entities who are in the
* same cfs_rq i.e who have a common parent. Walk up the hierarchy of
* both tasks until we find their ancestors who are siblings of common
* parent.
*/
/* First walk up until both entities are at same depth */
se_depth = depth_se(*se);
pse_depth = depth_se(*pse);
while (se_depth > pse_depth) {
se_depth--;
*se = parent_entity(*se);
}
while (pse_depth > se_depth) {
pse_depth--;
*pse = parent_entity(*pse);
}
while (!is_same_group(*se, *pse)) {
*se = parent_entity(*se);
*pse = parent_entity(*pse);
}
}
#else /* !CONFIG_FAIR_GROUP_SCHED */
static inline struct task_struct *task_of(struct sched_entity *se)
{
return container_of(se, struct task_struct, se);
}
static inline struct rq *rq_of(struct cfs_rq *cfs_rq)
{
return container_of(cfs_rq, struct rq, cfs);
}
#define entity_is_task(se) 1
#define for_each_sched_entity(se) \
for (; se; se = NULL)
static inline struct cfs_rq *task_cfs_rq(struct task_struct *p)
{
return &task_rq(p)->cfs;
}
static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se)
{
struct task_struct *p = task_of(se);
struct rq *rq = task_rq(p);
return &rq->cfs;
}
/* runqueue "owned" by this group */
static inline struct cfs_rq *group_cfs_rq(struct sched_entity *grp)
{
return NULL;
}
static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
}
static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
}
#define for_each_leaf_cfs_rq(rq, cfs_rq) \
for (cfs_rq = &rq->cfs; cfs_rq; cfs_rq = NULL)
static inline int
is_same_group(struct sched_entity *se, struct sched_entity *pse)
{
return 1;
}
static inline struct sched_entity *parent_entity(struct sched_entity *se)
{
return NULL;
}
static inline void
find_matching_se(struct sched_entity **se, struct sched_entity **pse)
{
}
#endif /* CONFIG_FAIR_GROUP_SCHED */
static void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
unsigned long delta_exec);
/**************************************************************
* Scheduling class tree data structure manipulation methods:
*/
static inline u64 max_vruntime(u64 min_vruntime, u64 vruntime)
{
s64 delta = (s64)(vruntime - min_vruntime);
if (delta > 0)
min_vruntime = vruntime;
return min_vruntime;
}
static inline u64 min_vruntime(u64 min_vruntime, u64 vruntime)
{
s64 delta = (s64)(vruntime - min_vruntime);
if (delta < 0)
min_vruntime = vruntime;
return min_vruntime;
}
static inline int entity_before(struct sched_entity *a,
struct sched_entity *b)
{
return (s64)(a->vruntime - b->vruntime) < 0;
}
static void update_min_vruntime(struct cfs_rq *cfs_rq)
{
u64 vruntime = cfs_rq->min_vruntime;
if (cfs_rq->curr)
vruntime = cfs_rq->curr->vruntime;
if (cfs_rq->rb_leftmost) {
struct sched_entity *se = rb_entry(cfs_rq->rb_leftmost,
struct sched_entity,
run_node);
if (!cfs_rq->curr)
vruntime = se->vruntime;
else
vruntime = min_vruntime(vruntime, se->vruntime);
}
cfs_rq->min_vruntime = max_vruntime(cfs_rq->min_vruntime, vruntime);
#ifndef CONFIG_64BIT
smp_wmb();
cfs_rq->min_vruntime_copy = cfs_rq->min_vruntime;
#endif
}
/*
* Enqueue an entity into the rb-tree:
*/
static void __enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
struct rb_node **link = &cfs_rq->tasks_timeline.rb_node;
struct rb_node *parent = NULL;
struct sched_entity *entry;
int leftmost = 1;
/*
* Find the right place in the rbtree:
*/
while (*link) {
parent = *link;
entry = rb_entry(parent, struct sched_entity, run_node);
/*
* We dont care about collisions. Nodes with
* the same key stay together.
*/
if (entity_before(se, entry)) {
link = &parent->rb_left;
} else {
link = &parent->rb_right;
leftmost = 0;
}
}
/*
* Maintain a cache of leftmost tree entries (it is frequently
* used):
*/
if (leftmost)
cfs_rq->rb_leftmost = &se->run_node;
rb_link_node(&se->run_node, parent, link);
rb_insert_color(&se->run_node, &cfs_rq->tasks_timeline);
}
static void __dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
if (cfs_rq->rb_leftmost == &se->run_node) {
struct rb_node *next_node;
next_node = rb_next(&se->run_node);
cfs_rq->rb_leftmost = next_node;
}
rb_erase(&se->run_node, &cfs_rq->tasks_timeline);
}
struct sched_entity *__pick_first_entity(struct cfs_rq *cfs_rq)
{
struct rb_node *left = cfs_rq->rb_leftmost;
if (!left)
return NULL;
return rb_entry(left, struct sched_entity, run_node);
}
static struct sched_entity *__pick_next_entity(struct sched_entity *se)
{
struct rb_node *next = rb_next(&se->run_node);
if (!next)
return NULL;
return rb_entry(next, struct sched_entity, run_node);
}
#ifdef CONFIG_SCHED_DEBUG
struct sched_entity *__pick_last_entity(struct cfs_rq *cfs_rq)
{
struct rb_node *last = rb_last(&cfs_rq->tasks_timeline);
if (!last)
return NULL;
return rb_entry(last, struct sched_entity, run_node);
}
/**************************************************************
* Scheduling class statistics methods:
*/
int sched_proc_update_handler(struct ctl_table *table, int write,
void __user *buffer, size_t *lenp,
loff_t *ppos)
{
int ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
int factor = get_update_sysctl_factor();
if (ret || !write)
return ret;
sched_nr_latency = DIV_ROUND_UP(sysctl_sched_latency,
sysctl_sched_min_granularity);
#define WRT_SYSCTL(name) \
(normalized_sysctl_##name = sysctl_##name / (factor))
WRT_SYSCTL(sched_min_granularity);
WRT_SYSCTL(sched_latency);
WRT_SYSCTL(sched_wakeup_granularity);
#undef WRT_SYSCTL
return 0;
}
#endif
/*
* delta /= w
*/
static inline unsigned long
calc_delta_fair(unsigned long delta, struct sched_entity *se)
{
if (unlikely(se->load.weight != NICE_0_LOAD))
delta = calc_delta_mine(delta, NICE_0_LOAD, &se->load);
return delta;
}
/*
* The idea is to set a period in which each task runs once.
*
* When there are too many tasks (sysctl_sched_nr_latency) we have to stretch
* this period because otherwise the slices get too small.
*
* p = (nr <= nl) ? l : l*nr/nl
*/
static u64 __sched_period(unsigned long nr_running)
{
u64 period = sysctl_sched_latency;
unsigned long nr_latency = sched_nr_latency;
if (unlikely(nr_running > nr_latency)) {
period = sysctl_sched_min_granularity;
period *= nr_running;
}
return period;
}
/*
* We calculate the wall-time slice from the period by taking a part
* proportional to the weight.
*
* s = p*P[w/rw]
*/
static u64 sched_slice(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
u64 slice = __sched_period(cfs_rq->nr_running + !se->on_rq);
for_each_sched_entity(se) {
struct load_weight *load;
struct load_weight lw;
cfs_rq = cfs_rq_of(se);
load = &cfs_rq->load;
if (unlikely(!se->on_rq)) {
lw = cfs_rq->load;
update_load_add(&lw, se->load.weight);
load = &lw;
}
slice = calc_delta_mine(slice, se->load.weight, load);
}
return slice;
}
/*
* We calculate the vruntime slice of a to be inserted task
*
* vs = s/w
*/
static u64 sched_vslice(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
return calc_delta_fair(sched_slice(cfs_rq, se), se);
}
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update);
static void update_cfs_shares(struct cfs_rq *cfs_rq);
/*
* Update the current task's runtime statistics. Skip current tasks that
* are not in our scheduling class.
*/
static inline void
__update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr,
unsigned long delta_exec)
{
unsigned long delta_exec_weighted;
schedstat_set(curr->statistics.exec_max,
max((u64)delta_exec, curr->statistics.exec_max));
curr->sum_exec_runtime += delta_exec;
schedstat_add(cfs_rq, exec_clock, delta_exec);
delta_exec_weighted = calc_delta_fair(delta_exec, curr);
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
curr->vruntime += delta_exec_weighted;
update_min_vruntime(cfs_rq);
#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED
cfs_rq->load_unacc_exec_time += delta_exec;
#endif
}
static void update_curr(struct cfs_rq *cfs_rq)
{
struct sched_entity *curr = cfs_rq->curr;
u64 now = rq_of(cfs_rq)->clock_task;
unsigned long delta_exec;
if (unlikely(!curr))
return;
/*
* Get the amount of time the current task was running
* since the last time we changed load (this cannot
* overflow on 32 bits):
*/
delta_exec = (unsigned long)(now - curr->exec_start);
if (!delta_exec)
return;
__update_curr(cfs_rq, curr, delta_exec);
curr->exec_start = now;
sched: cpu accounting controller (V2) Commit cfb5285660aad4931b2ebbfa902ea48a37dfffa1 removed a useful feature for us, which provided a cpu accounting resource controller. This feature would be useful if someone wants to group tasks only for accounting purpose and doesnt really want to exercise any control over their cpu consumption. The patch below reintroduces the feature. It is based on Paul Menage's original patch (Commit 62d0df64065e7c135d0002f069444fbdfc64768f), with these differences: - Removed load average information. I felt it needs more thought (esp to deal with SMP and virtualized platforms) and can be added for 2.6.25 after more discussions. - Convert group cpu usage to be nanosecond accurate (as rest of the cfs stats are) and invoke cpuacct_charge() from the respective scheduler classes - Make accounting scalable on SMP systems by splitting the usage counter to be per-cpu - Move the code from kernel/cpu_acct.c to kernel/sched.c (since the code is not big enough to warrant a new file and also this rightly needs to live inside the scheduler. Also things like accessing rq->lock while reading cpu usage becomes easier if the code lived in kernel/sched.c) The patch also modifies the cpu controller not to provide the same accounting information. Tested-by: Balbir Singh <balbir@linux.vnet.ibm.com> Tested the patches on top of 2.6.24-rc3. The patches work fine. Ran some simple tests like cpuspin (spin on the cpu), ran several tasks in the same group and timed them. Compared their time stamps with cpuacct.usage. Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2007-12-02 19:04:49 +00:00
if (entity_is_task(curr)) {
struct task_struct *curtask = task_of(curr);
trace_sched_stat_runtime(curtask, delta_exec, curr->vruntime);
sched: cpu accounting controller (V2) Commit cfb5285660aad4931b2ebbfa902ea48a37dfffa1 removed a useful feature for us, which provided a cpu accounting resource controller. This feature would be useful if someone wants to group tasks only for accounting purpose and doesnt really want to exercise any control over their cpu consumption. The patch below reintroduces the feature. It is based on Paul Menage's original patch (Commit 62d0df64065e7c135d0002f069444fbdfc64768f), with these differences: - Removed load average information. I felt it needs more thought (esp to deal with SMP and virtualized platforms) and can be added for 2.6.25 after more discussions. - Convert group cpu usage to be nanosecond accurate (as rest of the cfs stats are) and invoke cpuacct_charge() from the respective scheduler classes - Make accounting scalable on SMP systems by splitting the usage counter to be per-cpu - Move the code from kernel/cpu_acct.c to kernel/sched.c (since the code is not big enough to warrant a new file and also this rightly needs to live inside the scheduler. Also things like accessing rq->lock while reading cpu usage becomes easier if the code lived in kernel/sched.c) The patch also modifies the cpu controller not to provide the same accounting information. Tested-by: Balbir Singh <balbir@linux.vnet.ibm.com> Tested the patches on top of 2.6.24-rc3. The patches work fine. Ran some simple tests like cpuspin (spin on the cpu), ran several tasks in the same group and timed them. Compared their time stamps with cpuacct.usage. Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2007-12-02 19:04:49 +00:00
cpuacct_charge(curtask, delta_exec);
timers: fix itimer/many thread hang Overview This patch reworks the handling of POSIX CPU timers, including the ITIMER_PROF, ITIMER_VIRT timers and rlimit handling. It was put together with the help of Roland McGrath, the owner and original writer of this code. The problem we ran into, and the reason for this rework, has to do with using a profiling timer in a process with a large number of threads. It appears that the performance of the old implementation of run_posix_cpu_timers() was at least O(n*3) (where "n" is the number of threads in a process) or worse. Everything is fine with an increasing number of threads until the time taken for that routine to run becomes the same as or greater than the tick time, at which point things degrade rather quickly. This patch fixes bug 9906, "Weird hang with NPTL and SIGPROF." Code Changes This rework corrects the implementation of run_posix_cpu_timers() to make it run in constant time for a particular machine. (Performance may vary between one machine and another depending upon whether the kernel is built as single- or multiprocessor and, in the latter case, depending upon the number of running processors.) To do this, at each tick we now update fields in signal_struct as well as task_struct. The run_posix_cpu_timers() function uses those fields to make its decisions. We define a new structure, "task_cputime," to contain user, system and scheduler times and use these in appropriate places: struct task_cputime { cputime_t utime; cputime_t stime; unsigned long long sum_exec_runtime; }; This is included in the structure "thread_group_cputime," which is a new substructure of signal_struct and which varies for uniprocessor versus multiprocessor kernels. For uniprocessor kernels, it uses "task_cputime" as a simple substructure, while for multiprocessor kernels it is a pointer: struct thread_group_cputime { struct task_cputime totals; }; struct thread_group_cputime { struct task_cputime *totals; }; We also add a new task_cputime substructure directly to signal_struct, to cache the earliest expiration of process-wide timers, and task_cputime also replaces the it_*_expires fields of task_struct (used for earliest expiration of thread timers). The "thread_group_cputime" structure contains process-wide timers that are updated via account_user_time() and friends. In the non-SMP case the structure is a simple aggregator; unfortunately in the SMP case that simplicity was not achievable due to cache-line contention between CPUs (in one measured case performance was actually _worse_ on a 16-cpu system than the same test on a 4-cpu system, due to this contention). For SMP, the thread_group_cputime counters are maintained as a per-cpu structure allocated using alloc_percpu(). The timer functions update only the timer field in the structure corresponding to the running CPU, obtained using per_cpu_ptr(). We define a set of inline functions in sched.h that we use to maintain the thread_group_cputime structure and hide the differences between UP and SMP implementations from the rest of the kernel. The thread_group_cputime_init() function initializes the thread_group_cputime structure for the given task. The thread_group_cputime_alloc() is a no-op for UP; for SMP it calls the out-of-line function thread_group_cputime_alloc_smp() to allocate and fill in the per-cpu structures and fields. The thread_group_cputime_free() function, also a no-op for UP, in SMP frees the per-cpu structures. The thread_group_cputime_clone_thread() function (also a UP no-op) for SMP calls thread_group_cputime_alloc() if the per-cpu structures haven't yet been allocated. The thread_group_cputime() function fills the task_cputime structure it is passed with the contents of the thread_group_cputime fields; in UP it's that simple but in SMP it must also safely check that tsk->signal is non-NULL (if it is it just uses the appropriate fields of task_struct) and, if so, sums the per-cpu values for each online CPU. Finally, the three functions account_group_user_time(), account_group_system_time() and account_group_exec_runtime() are used by timer functions to update the respective fields of the thread_group_cputime structure. Non-SMP operation is trivial and will not be mentioned further. The per-cpu structure is always allocated when a task creates its first new thread, via a call to thread_group_cputime_clone_thread() from copy_signal(). It is freed at process exit via a call to thread_group_cputime_free() from cleanup_signal(). All functions that formerly summed utime/stime/sum_sched_runtime values from from all threads in the thread group now use thread_group_cputime() to snapshot the values in the thread_group_cputime structure or the values in the task structure itself if the per-cpu structure hasn't been allocated. Finally, the code in kernel/posix-cpu-timers.c has changed quite a bit. The run_posix_cpu_timers() function has been split into a fast path and a slow path; the former safely checks whether there are any expired thread timers and, if not, just returns, while the slow path does the heavy lifting. With the dedicated thread group fields, timers are no longer "rebalanced" and the process_timer_rebalance() function and related code has gone away. All summing loops are gone and all code that used them now uses the thread_group_cputime() inline. When process-wide timers are set, the new task_cputime structure in signal_struct is used to cache the earliest expiration; this is checked in the fast path. Performance The fix appears not to add significant overhead to existing operations. It generally performs the same as the current code except in two cases, one in which it performs slightly worse (Case 5 below) and one in which it performs very significantly better (Case 2 below). Overall it's a wash except in those two cases. I've since done somewhat more involved testing on a dual-core Opteron system. Case 1: With no itimer running, for a test with 100,000 threads, the fixed kernel took 1428.5 seconds, 513 seconds more than the unfixed system, all of which was spent in the system. There were twice as many voluntary context switches with the fix as without it. Case 2: With an itimer running at .01 second ticks and 4000 threads (the most an unmodified kernel can handle), the fixed kernel ran the test in eight percent of the time (5.8 seconds as opposed to 70 seconds) and had better tick accuracy (.012 seconds per tick as opposed to .023 seconds per tick). Case 3: A 4000-thread test with an initial timer tick of .01 second and an interval of 10,000 seconds (i.e. a timer that ticks only once) had very nearly the same performance in both cases: 6.3 seconds elapsed for the fixed kernel versus 5.5 seconds for the unfixed kernel. With fewer threads (eight in these tests), the Case 1 test ran in essentially the same time on both the modified and unmodified kernels (5.2 seconds versus 5.8 seconds). The Case 2 test ran in about the same time as well, 5.9 seconds versus 5.4 seconds but again with much better tick accuracy, .013 seconds per tick versus .025 seconds per tick for the unmodified kernel. Since the fix affected the rlimit code, I also tested soft and hard CPU limits. Case 4: With a hard CPU limit of 20 seconds and eight threads (and an itimer running), the modified kernel was very slightly favored in that while it killed the process in 19.997 seconds of CPU time (5.002 seconds of wall time), only .003 seconds of that was system time, the rest was user time. The unmodified kernel killed the process in 20.001 seconds of CPU (5.014 seconds of wall time) of which .016 seconds was system time. Really, though, the results were too close to call. The results were essentially the same with no itimer running. Case 5: With a soft limit of 20 seconds and a hard limit of 2000 seconds (where the hard limit would never be reached) and an itimer running, the modified kernel exhibited worse tick accuracy than the unmodified kernel: .050 seconds/tick versus .028 seconds/tick. Otherwise, performance was almost indistinguishable. With no itimer running this test exhibited virtually identical behavior and times in both cases. In times past I did some limited performance testing. those results are below. On a four-cpu Opteron system without this fix, a sixteen-thread test executed in 3569.991 seconds, of which user was 3568.435s and system was 1.556s. On the same system with the fix, user and elapsed time were about the same, but system time dropped to 0.007 seconds. Performance with eight, four and one thread were comparable. Interestingly, the timer ticks with the fix seemed more accurate: The sixteen-thread test with the fix received 149543 ticks for 0.024 seconds per tick, while the same test without the fix received 58720 for 0.061 seconds per tick. Both cases were configured for an interval of 0.01 seconds. Again, the other tests were comparable. Each thread in this test computed the primes up to 25,000,000. I also did a test with a large number of threads, 100,000 threads, which is impossible without the fix. In this case each thread computed the primes only up to 10,000 (to make the runtime manageable). System time dominated, at 1546.968 seconds out of a total 2176.906 seconds (giving a user time of 629.938s). It received 147651 ticks for 0.015 seconds per tick, still quite accurate. There is obviously no comparable test without the fix. Signed-off-by: Frank Mayhar <fmayhar@google.com> Cc: Roland McGrath <roland@redhat.com> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-09-12 16:54:39 +00:00
account_group_exec_runtime(curtask, delta_exec);
sched: cpu accounting controller (V2) Commit cfb5285660aad4931b2ebbfa902ea48a37dfffa1 removed a useful feature for us, which provided a cpu accounting resource controller. This feature would be useful if someone wants to group tasks only for accounting purpose and doesnt really want to exercise any control over their cpu consumption. The patch below reintroduces the feature. It is based on Paul Menage's original patch (Commit 62d0df64065e7c135d0002f069444fbdfc64768f), with these differences: - Removed load average information. I felt it needs more thought (esp to deal with SMP and virtualized platforms) and can be added for 2.6.25 after more discussions. - Convert group cpu usage to be nanosecond accurate (as rest of the cfs stats are) and invoke cpuacct_charge() from the respective scheduler classes - Make accounting scalable on SMP systems by splitting the usage counter to be per-cpu - Move the code from kernel/cpu_acct.c to kernel/sched.c (since the code is not big enough to warrant a new file and also this rightly needs to live inside the scheduler. Also things like accessing rq->lock while reading cpu usage becomes easier if the code lived in kernel/sched.c) The patch also modifies the cpu controller not to provide the same accounting information. Tested-by: Balbir Singh <balbir@linux.vnet.ibm.com> Tested the patches on top of 2.6.24-rc3. The patches work fine. Ran some simple tests like cpuspin (spin on the cpu), ran several tasks in the same group and timed them. Compared their time stamps with cpuacct.usage. Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2007-12-02 19:04:49 +00:00
}
account_cfs_rq_runtime(cfs_rq, delta_exec);
}
static inline void
update_stats_wait_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
schedstat_set(se->statistics.wait_start, rq_of(cfs_rq)->clock);
}
/*
* Task is being enqueued - update stats:
*/
static void update_stats_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
/*
* Are we enqueueing a waiting task? (for current tasks
* a dequeue/enqueue event is a NOP)
*/
if (se != cfs_rq->curr)
update_stats_wait_start(cfs_rq, se);
}
static void
update_stats_wait_end(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
schedstat_set(se->statistics.wait_max, max(se->statistics.wait_max,
rq_of(cfs_rq)->clock - se->statistics.wait_start));
schedstat_set(se->statistics.wait_count, se->statistics.wait_count + 1);
schedstat_set(se->statistics.wait_sum, se->statistics.wait_sum +
rq_of(cfs_rq)->clock - se->statistics.wait_start);
#ifdef CONFIG_SCHEDSTATS
if (entity_is_task(se)) {
trace_sched_stat_wait(task_of(se),
rq_of(cfs_rq)->clock - se->statistics.wait_start);
}
#endif
schedstat_set(se->statistics.wait_start, 0);
}
static inline void
update_stats_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
/*
* Mark the end of the wait period if dequeueing a
* waiting task:
*/
if (se != cfs_rq->curr)
update_stats_wait_end(cfs_rq, se);
}
/*
* We are picking a new current task - update its stats:
*/
static inline void
update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
/*
* We are starting a new run period:
*/
se->exec_start = rq_of(cfs_rq)->clock_task;
}
/**************************************************
* Scheduling class queueing methods:
*/
#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED
static void
add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight)
{
cfs_rq->task_weight += weight;
}
#else
static inline void
add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight)
{
}
#endif
static void
account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
update_load_add(&cfs_rq->load, se->load.weight);
if (!parent_entity(se))
update_load_add(&rq_of(cfs_rq)->load, se->load.weight);
if (entity_is_task(se)) {
add_cfs_task_weight(cfs_rq, se->load.weight);
list_add(&se->group_node, &cfs_rq->tasks);
}
cfs_rq->nr_running++;
}
static void
account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
update_load_sub(&cfs_rq->load, se->load.weight);
if (!parent_entity(se))
update_load_sub(&rq_of(cfs_rq)->load, se->load.weight);
if (entity_is_task(se)) {
add_cfs_task_weight(cfs_rq, -se->load.weight);
list_del_init(&se->group_node);
}
cfs_rq->nr_running--;
}
#ifdef CONFIG_FAIR_GROUP_SCHED
/* we need this in update_cfs_load and load-balance functions below */
static inline int throttled_hierarchy(struct cfs_rq *cfs_rq);
# ifdef CONFIG_SMP
static void update_cfs_rq_load_contribution(struct cfs_rq *cfs_rq,
int global_update)
{
struct task_group *tg = cfs_rq->tg;
long load_avg;
load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1);
load_avg -= cfs_rq->load_contribution;
if (global_update || abs(load_avg) > cfs_rq->load_contribution / 8) {
atomic_add(load_avg, &tg->load_weight);
cfs_rq->load_contribution += load_avg;
}
}
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
{
u64 period = sysctl_sched_shares_window;
u64 now, delta;
unsigned long load = cfs_rq->load.weight;
if (cfs_rq->tg == &root_task_group || throttled_hierarchy(cfs_rq))
return;
now = rq_of(cfs_rq)->clock_task;
delta = now - cfs_rq->load_stamp;
/* truncate load history at 4 idle periods */
if (cfs_rq->load_stamp > cfs_rq->load_last &&
now - cfs_rq->load_last > 4 * period) {
cfs_rq->load_period = 0;
cfs_rq->load_avg = 0;
delta = period - 1;
}
cfs_rq->load_stamp = now;
cfs_rq->load_unacc_exec_time = 0;
cfs_rq->load_period += delta;
if (load) {
cfs_rq->load_last = now;
cfs_rq->load_avg += delta * load;
}
/* consider updating load contribution on each fold or truncate */
if (global_update || cfs_rq->load_period > period
|| !cfs_rq->load_period)
update_cfs_rq_load_contribution(cfs_rq, global_update);
while (cfs_rq->load_period > period) {
/*
* Inline assembly required to prevent the compiler
* optimising this loop into a divmod call.
* See __iter_div_u64_rem() for another example of this.
*/
asm("" : "+rm" (cfs_rq->load_period));
cfs_rq->load_period /= 2;
cfs_rq->load_avg /= 2;
}
if (!cfs_rq->curr && !cfs_rq->nr_running && !cfs_rq->load_avg)
list_del_leaf_cfs_rq(cfs_rq);
}
static inline long calc_tg_weight(struct task_group *tg, struct cfs_rq *cfs_rq)
{
long tg_weight;
/*
* Use this CPU's actual weight instead of the last load_contribution
* to gain a more accurate current total weight. See
* update_cfs_rq_load_contribution().
*/
tg_weight = atomic_read(&tg->load_weight);
tg_weight -= cfs_rq->load_contribution;
tg_weight += cfs_rq->load.weight;
return tg_weight;
}
static long calc_cfs_shares(struct cfs_rq *cfs_rq, struct task_group *tg)
{
long tg_weight, load, shares;
tg_weight = calc_tg_weight(tg, cfs_rq);
load = cfs_rq->load.weight;
shares = (tg->shares * load);
if (tg_weight)
shares /= tg_weight;
if (shares < MIN_SHARES)
shares = MIN_SHARES;
if (shares > tg->shares)
shares = tg->shares;
return shares;
}
static void update_entity_shares_tick(struct cfs_rq *cfs_rq)
{
if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) {
update_cfs_load(cfs_rq, 0);
update_cfs_shares(cfs_rq);
}
}
# else /* CONFIG_SMP */
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
{
}
static inline long calc_cfs_shares(struct cfs_rq *cfs_rq, struct task_group *tg)
{
return tg->shares;
}
static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq)
{
}
# endif /* CONFIG_SMP */
static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se,
unsigned long weight)
{
if (se->on_rq) {
/* commit outstanding execution time */
if (cfs_rq->curr == se)
update_curr(cfs_rq);
account_entity_dequeue(cfs_rq, se);
}
update_load_set(&se->load, weight);
if (se->on_rq)
account_entity_enqueue(cfs_rq, se);
}
static void update_cfs_shares(struct cfs_rq *cfs_rq)
{
struct task_group *tg;
struct sched_entity *se;
long shares;
tg = cfs_rq->tg;
se = tg->se[cpu_of(rq_of(cfs_rq))];
if (!se || throttled_hierarchy(cfs_rq))
return;
#ifndef CONFIG_SMP
if (likely(se->load.weight == tg->shares))
return;
#endif
shares = calc_cfs_shares(cfs_rq, tg);
reweight_entity(cfs_rq_of(se), se, shares);
}
#else /* CONFIG_FAIR_GROUP_SCHED */
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
{
}
static inline void update_cfs_shares(struct cfs_rq *cfs_rq)
{
}
static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq)
{
}
#endif /* CONFIG_FAIR_GROUP_SCHED */
static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
#ifdef CONFIG_SCHEDSTATS
struct task_struct *tsk = NULL;
if (entity_is_task(se))
tsk = task_of(se);
if (se->statistics.sleep_start) {
u64 delta = rq_of(cfs_rq)->clock - se->statistics.sleep_start;
if ((s64)delta < 0)
delta = 0;
if (unlikely(delta > se->statistics.sleep_max))
se->statistics.sleep_max = delta;
se->statistics.sleep_start = 0;
se->statistics.sum_sleep_runtime += delta;
if (tsk) {
account_scheduler_latency(tsk, delta >> 10, 1);
trace_sched_stat_sleep(tsk, delta);
}
}
if (se->statistics.block_start) {
u64 delta = rq_of(cfs_rq)->clock - se->statistics.block_start;
if ((s64)delta < 0)
delta = 0;
if (unlikely(delta > se->statistics.block_max))
se->statistics.block_max = delta;
se->statistics.block_start = 0;
se->statistics.sum_sleep_runtime += delta;
if (tsk) {
if (tsk->in_iowait) {
se->statistics.iowait_sum += delta;
se->statistics.iowait_count++;
trace_sched_stat_iowait(tsk, delta);
}
trace_sched_stat_blocked(tsk, delta);
/*
* Blocking time is in units of nanosecs, so shift by
* 20 to get a milliseconds-range estimation of the
* amount of time that the task spent sleeping:
*/
if (unlikely(prof_on == SLEEP_PROFILING)) {
profile_hits(SLEEP_PROFILING,
(void *)get_wchan(tsk),
delta >> 20);
}
account_scheduler_latency(tsk, delta >> 10, 0);
}
}
#endif
}
static void check_spread(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
#ifdef CONFIG_SCHED_DEBUG
s64 d = se->vruntime - cfs_rq->min_vruntime;
if (d < 0)
d = -d;
if (d > 3*sysctl_sched_latency)
schedstat_inc(cfs_rq, nr_spread_over);
#endif
}
static void
place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial)
{
u64 vruntime = cfs_rq->min_vruntime;
/*
* The 'current' period is already promised to the current tasks,
* however the extra weight of the new task will slow them down a
* little, place the new task so that it fits in the slot that
* stays open at the end.
*/
if (initial && sched_feat(START_DEBIT))
vruntime += sched_vslice(cfs_rq, se);
/* sleeps up to a single latency don't count. */
if (!initial) {
unsigned long thresh = sysctl_sched_latency;
/*
* Halve their sleep time's effect, to allow
* for a gentler effect of sleepers:
*/
if (sched_feat(GENTLE_FAIR_SLEEPERS))
thresh >>= 1;
vruntime -= thresh;
}
/* ensure we never gain time by being placed backwards. */
vruntime = max_vruntime(se->vruntime, vruntime);
se->vruntime = vruntime;
}
static void check_enqueue_throttle(struct cfs_rq *cfs_rq);
static void
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
{
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
/*
* Update the normalized vruntime before updating min_vruntime
* through callig update_curr().
*/
if (!(flags & ENQUEUE_WAKEUP) || (flags & ENQUEUE_WAKING))
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
se->vruntime += cfs_rq->min_vruntime;
/*
* Update run-time statistics of the 'current'.
*/
update_curr(cfs_rq);
update_cfs_load(cfs_rq, 0);
account_entity_enqueue(cfs_rq, se);
update_cfs_shares(cfs_rq);
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
if (flags & ENQUEUE_WAKEUP) {
place_entity(cfs_rq, se, 0);
enqueue_sleeper(cfs_rq, se);
}
update_stats_enqueue(cfs_rq, se);
check_spread(cfs_rq, se);
if (se != cfs_rq->curr)
__enqueue_entity(cfs_rq, se);
se->on_rq = 1;
if (cfs_rq->nr_running == 1) {
list_add_leaf_cfs_rq(cfs_rq);
check_enqueue_throttle(cfs_rq);
}
}
static void __clear_buddies_last(struct sched_entity *se)
{
for_each_sched_entity(se) {
struct cfs_rq *cfs_rq = cfs_rq_of(se);
if (cfs_rq->last == se)
cfs_rq->last = NULL;
else
break;
}
}
static void __clear_buddies_next(struct sched_entity *se)
{
for_each_sched_entity(se) {
struct cfs_rq *cfs_rq = cfs_rq_of(se);
if (cfs_rq->next == se)
cfs_rq->next = NULL;
else
break;
}
}
static void __clear_buddies_skip(struct sched_entity *se)
{
for_each_sched_entity(se) {
struct cfs_rq *cfs_rq = cfs_rq_of(se);
if (cfs_rq->skip == se)
cfs_rq->skip = NULL;
else
break;
}
}
static void clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
if (cfs_rq->last == se)
__clear_buddies_last(se);
if (cfs_rq->next == se)
__clear_buddies_next(se);
if (cfs_rq->skip == se)
__clear_buddies_skip(se);
}
sched: Return unused runtime on group dequeue When a local cfs_rq blocks we return the majority of its remaining quota to the global bandwidth pool for use by other runqueues. We do this only when the quota is current and there is more than min_cfs_rq_quota [1ms by default] of runtime remaining on the rq. In the case where there are throttled runqueues and we have sufficient bandwidth to meter out a slice, a second timer is kicked off to handle this delivery, unthrottling where appropriate. Using a 'worst case' antagonist which executes on each cpu for 1ms before moving onto the next on a fairly large machine: no quota generations: 197.47 ms /cgroup/a/cpuacct.usage 199.46 ms /cgroup/a/cpuacct.usage 205.46 ms /cgroup/a/cpuacct.usage 198.46 ms /cgroup/a/cpuacct.usage 208.39 ms /cgroup/a/cpuacct.usage Since we are allowed to use "stale" quota our usage is effectively bounded by the rate of input into the global pool and performance is relatively stable. with quota generations [1s increments]: 119.58 ms /cgroup/a/cpuacct.usage 119.65 ms /cgroup/a/cpuacct.usage 119.64 ms /cgroup/a/cpuacct.usage 119.63 ms /cgroup/a/cpuacct.usage 119.60 ms /cgroup/a/cpuacct.usage The large deficit here is due to quota generations (/intentionally/) preventing us from now using previously stranded slack quota. The cost is that this quota becomes unavailable. with quota generations and quota return: 200.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 198.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 200.06 ms /cgroup/a/cpuacct.usage By returning unused quota we're able to both stably consume our desired quota and prevent unintentional overages due to the abuse of slack quota from previous quota periods (especially on a large machine). Signed-off-by: Paul Turner <pjt@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20110721184758.306848658@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-07-21 16:43:41 +00:00
static void return_cfs_rq_runtime(struct cfs_rq *cfs_rq);
static void
dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
{
/*
* Update run-time statistics of the 'current'.
*/
update_curr(cfs_rq);
update_stats_dequeue(cfs_rq, se);
if (flags & DEQUEUE_SLEEP) {
#ifdef CONFIG_SCHEDSTATS
if (entity_is_task(se)) {
struct task_struct *tsk = task_of(se);
if (tsk->state & TASK_INTERRUPTIBLE)
se->statistics.sleep_start = rq_of(cfs_rq)->clock;
if (tsk->state & TASK_UNINTERRUPTIBLE)
se->statistics.block_start = rq_of(cfs_rq)->clock;
}
#endif
}
clear_buddies(cfs_rq, se);
if (se != cfs_rq->curr)
__dequeue_entity(cfs_rq, se);
se->on_rq = 0;
update_cfs_load(cfs_rq, 0);
account_entity_dequeue(cfs_rq, se);
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
/*
* Normalize the entity after updating the min_vruntime because the
* update can refer to the ->curr item and we need to reflect this
* movement in our normalized position.
*/
if (!(flags & DEQUEUE_SLEEP))
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
se->vruntime -= cfs_rq->min_vruntime;
sched: Return unused runtime on group dequeue When a local cfs_rq blocks we return the majority of its remaining quota to the global bandwidth pool for use by other runqueues. We do this only when the quota is current and there is more than min_cfs_rq_quota [1ms by default] of runtime remaining on the rq. In the case where there are throttled runqueues and we have sufficient bandwidth to meter out a slice, a second timer is kicked off to handle this delivery, unthrottling where appropriate. Using a 'worst case' antagonist which executes on each cpu for 1ms before moving onto the next on a fairly large machine: no quota generations: 197.47 ms /cgroup/a/cpuacct.usage 199.46 ms /cgroup/a/cpuacct.usage 205.46 ms /cgroup/a/cpuacct.usage 198.46 ms /cgroup/a/cpuacct.usage 208.39 ms /cgroup/a/cpuacct.usage Since we are allowed to use "stale" quota our usage is effectively bounded by the rate of input into the global pool and performance is relatively stable. with quota generations [1s increments]: 119.58 ms /cgroup/a/cpuacct.usage 119.65 ms /cgroup/a/cpuacct.usage 119.64 ms /cgroup/a/cpuacct.usage 119.63 ms /cgroup/a/cpuacct.usage 119.60 ms /cgroup/a/cpuacct.usage The large deficit here is due to quota generations (/intentionally/) preventing us from now using previously stranded slack quota. The cost is that this quota becomes unavailable. with quota generations and quota return: 200.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 198.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 200.06 ms /cgroup/a/cpuacct.usage By returning unused quota we're able to both stably consume our desired quota and prevent unintentional overages due to the abuse of slack quota from previous quota periods (especially on a large machine). Signed-off-by: Paul Turner <pjt@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20110721184758.306848658@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-07-21 16:43:41 +00:00
/* return excess runtime on last dequeue */
return_cfs_rq_runtime(cfs_rq);
update_min_vruntime(cfs_rq);
update_cfs_shares(cfs_rq);
}
/*
* Preempt the current task with a newly woken task if needed:
*/
static void
check_preempt_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr)
{
unsigned long ideal_runtime, delta_exec;
struct sched_entity *se;
s64 delta;
ideal_runtime = sched_slice(cfs_rq, curr);
delta_exec = curr->sum_exec_runtime - curr->prev_sum_exec_runtime;
if (delta_exec > ideal_runtime) {
resched_task(rq_of(cfs_rq)->curr);
/*
* The current task ran long enough, ensure it doesn't get
* re-elected due to buddy favours.
*/
clear_buddies(cfs_rq, curr);
sched: Strengthen buddies and mitigate buddy induced latencies This patch restores the effectiveness of LAST_BUDDY in preventing pgsql+oltp from collapsing due to wakeup preemption. It also switches LAST_BUDDY to exclusively do what it does best, namely mitigate the effects of aggressive wakeup preemption, which improves vmark throughput markedly, and restores mysql+oltp scalability. Since buddies are about scalability, enable them beginning at the point where we begin expanding sched_latency, namely sched_nr_latency. Previously, buddies were cleared aggressively, which seriously reduced their effectiveness. Not clearing aggressively however, produces a small drop in mysql+oltp throughput immediately after peak, indicating that LAST_BUDDY is actually doing some harm. This is right at the point where X on the desktop in competition with another load wants low latency service. Ergo, do not enable until we need to scale. To mitigate latency induced by buddies, or by a task just missing wakeup preemption, check latency at tick time. Last hunk prevents buddies from stymieing BALANCE_NEWIDLE via CACHE_HOT_BUDDY. Supporting performance tests: tip = v2.6.32-rc5-1497-ga525b32 tipx = NO_GENTLE_FAIR_SLEEPERS NEXT_BUDDY granularity knobs = 31 knobs + 31 buddies tip+x = NO_GENTLE_FAIR_SLEEPERS granularity knobs = 31 knobs (Three run averages except where noted.) vmark: ------ tip 108466 messages per second tip+ 125307 messages per second tip+x 125335 messages per second tipx 117781 messages per second 2.6.31.3 122729 messages per second mysql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 9949.89 18690.20 34801.24 34460.04 32682.88 30765.97 28305.27 25059.64 19548.08 tip+ 10013.90 18526.84 34900.38 34420.14 33069.83 32083.40 30578.30 28010.71 25605.47 tipx 9698.71 18002.70 34477.56 33420.01 32634.30 31657.27 29932.67 26827.52 21487.18 2.6.31.3 8243.11 18784.20 34404.83 33148.38 31900.32 31161.90 29663.81 25995.94 18058.86 pgsql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 13686.37 26609.25 51934.28 51347.81 49479.51 45312.65 36691.91 26851.57 24145.35 tip+ (1x) 13907.85 27135.87 52951.98 52514.04 51742.52 50705.43 49947.97 48374.19 46227.94 tip+x 13906.78 27065.81 52951.19 52542.59 52176.11 51815.94 50838.90 49439.46 46891.00 tipx 13742.46 26769.81 52351.99 51891.73 51320.79 50938.98 50248.65 48908.70 46553.84 2.6.31.3 13815.35 26906.46 52683.34 52061.31 51937.10 51376.80 50474.28 49394.47 47003.25 Signed-off-by: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-23 21:09:22 +00:00
return;
}
/*
* Ensure that a task that missed wakeup preemption by a
* narrow margin doesn't have to wait for a full slice.
* This also mitigates buddy induced latencies under load.
*/
if (delta_exec < sysctl_sched_min_granularity)
return;
se = __pick_first_entity(cfs_rq);
delta = curr->vruntime - se->vruntime;
sched: Strengthen buddies and mitigate buddy induced latencies This patch restores the effectiveness of LAST_BUDDY in preventing pgsql+oltp from collapsing due to wakeup preemption. It also switches LAST_BUDDY to exclusively do what it does best, namely mitigate the effects of aggressive wakeup preemption, which improves vmark throughput markedly, and restores mysql+oltp scalability. Since buddies are about scalability, enable them beginning at the point where we begin expanding sched_latency, namely sched_nr_latency. Previously, buddies were cleared aggressively, which seriously reduced their effectiveness. Not clearing aggressively however, produces a small drop in mysql+oltp throughput immediately after peak, indicating that LAST_BUDDY is actually doing some harm. This is right at the point where X on the desktop in competition with another load wants low latency service. Ergo, do not enable until we need to scale. To mitigate latency induced by buddies, or by a task just missing wakeup preemption, check latency at tick time. Last hunk prevents buddies from stymieing BALANCE_NEWIDLE via CACHE_HOT_BUDDY. Supporting performance tests: tip = v2.6.32-rc5-1497-ga525b32 tipx = NO_GENTLE_FAIR_SLEEPERS NEXT_BUDDY granularity knobs = 31 knobs + 31 buddies tip+x = NO_GENTLE_FAIR_SLEEPERS granularity knobs = 31 knobs (Three run averages except where noted.) vmark: ------ tip 108466 messages per second tip+ 125307 messages per second tip+x 125335 messages per second tipx 117781 messages per second 2.6.31.3 122729 messages per second mysql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 9949.89 18690.20 34801.24 34460.04 32682.88 30765.97 28305.27 25059.64 19548.08 tip+ 10013.90 18526.84 34900.38 34420.14 33069.83 32083.40 30578.30 28010.71 25605.47 tipx 9698.71 18002.70 34477.56 33420.01 32634.30 31657.27 29932.67 26827.52 21487.18 2.6.31.3 8243.11 18784.20 34404.83 33148.38 31900.32 31161.90 29663.81 25995.94 18058.86 pgsql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 13686.37 26609.25 51934.28 51347.81 49479.51 45312.65 36691.91 26851.57 24145.35 tip+ (1x) 13907.85 27135.87 52951.98 52514.04 51742.52 50705.43 49947.97 48374.19 46227.94 tip+x 13906.78 27065.81 52951.19 52542.59 52176.11 51815.94 50838.90 49439.46 46891.00 tipx 13742.46 26769.81 52351.99 51891.73 51320.79 50938.98 50248.65 48908.70 46553.84 2.6.31.3 13815.35 26906.46 52683.34 52061.31 51937.10 51376.80 50474.28 49394.47 47003.25 Signed-off-by: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-23 21:09:22 +00:00
if (delta < 0)
return;
if (delta > ideal_runtime)
resched_task(rq_of(cfs_rq)->curr);
}
static void
set_next_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
/* 'current' is not kept within the tree. */
if (se->on_rq) {
/*
* Any task has to be enqueued before it get to execute on
* a CPU. So account for the time it spent waiting on the
* runqueue.
*/
update_stats_wait_end(cfs_rq, se);
__dequeue_entity(cfs_rq, se);
}
update_stats_curr_start(cfs_rq, se);
cfs_rq->curr = se;
#ifdef CONFIG_SCHEDSTATS
/*
* Track our maximum slice length, if the CPU's load is at
* least twice that of our own weight (i.e. dont track it
* when there are only lesser-weight tasks around):
*/
if (rq_of(cfs_rq)->load.weight >= 2*se->load.weight) {
se->statistics.slice_max = max(se->statistics.slice_max,
se->sum_exec_runtime - se->prev_sum_exec_runtime);
}
#endif
se->prev_sum_exec_runtime = se->sum_exec_runtime;
}
static int
wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se);
/*
* Pick the next process, keeping these things in mind, in this order:
* 1) keep things fair between processes/task groups
* 2) pick the "next" process, since someone really wants that to run
* 3) pick the "last" process, for cache locality
* 4) do not run the "skip" process, if something else is available
*/
static struct sched_entity *pick_next_entity(struct cfs_rq *cfs_rq)
{
struct sched_entity *se = __pick_first_entity(cfs_rq);
sched: Strengthen buddies and mitigate buddy induced latencies This patch restores the effectiveness of LAST_BUDDY in preventing pgsql+oltp from collapsing due to wakeup preemption. It also switches LAST_BUDDY to exclusively do what it does best, namely mitigate the effects of aggressive wakeup preemption, which improves vmark throughput markedly, and restores mysql+oltp scalability. Since buddies are about scalability, enable them beginning at the point where we begin expanding sched_latency, namely sched_nr_latency. Previously, buddies were cleared aggressively, which seriously reduced their effectiveness. Not clearing aggressively however, produces a small drop in mysql+oltp throughput immediately after peak, indicating that LAST_BUDDY is actually doing some harm. This is right at the point where X on the desktop in competition with another load wants low latency service. Ergo, do not enable until we need to scale. To mitigate latency induced by buddies, or by a task just missing wakeup preemption, check latency at tick time. Last hunk prevents buddies from stymieing BALANCE_NEWIDLE via CACHE_HOT_BUDDY. Supporting performance tests: tip = v2.6.32-rc5-1497-ga525b32 tipx = NO_GENTLE_FAIR_SLEEPERS NEXT_BUDDY granularity knobs = 31 knobs + 31 buddies tip+x = NO_GENTLE_FAIR_SLEEPERS granularity knobs = 31 knobs (Three run averages except where noted.) vmark: ------ tip 108466 messages per second tip+ 125307 messages per second tip+x 125335 messages per second tipx 117781 messages per second 2.6.31.3 122729 messages per second mysql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 9949.89 18690.20 34801.24 34460.04 32682.88 30765.97 28305.27 25059.64 19548.08 tip+ 10013.90 18526.84 34900.38 34420.14 33069.83 32083.40 30578.30 28010.71 25605.47 tipx 9698.71 18002.70 34477.56 33420.01 32634.30 31657.27 29932.67 26827.52 21487.18 2.6.31.3 8243.11 18784.20 34404.83 33148.38 31900.32 31161.90 29663.81 25995.94 18058.86 pgsql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 13686.37 26609.25 51934.28 51347.81 49479.51 45312.65 36691.91 26851.57 24145.35 tip+ (1x) 13907.85 27135.87 52951.98 52514.04 51742.52 50705.43 49947.97 48374.19 46227.94 tip+x 13906.78 27065.81 52951.19 52542.59 52176.11 51815.94 50838.90 49439.46 46891.00 tipx 13742.46 26769.81 52351.99 51891.73 51320.79 50938.98 50248.65 48908.70 46553.84 2.6.31.3 13815.35 26906.46 52683.34 52061.31 51937.10 51376.80 50474.28 49394.47 47003.25 Signed-off-by: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-23 21:09:22 +00:00
struct sched_entity *left = se;
/*
* Avoid running the skip buddy, if running something else can
* be done without getting too unfair.
*/
if (cfs_rq->skip == se) {
struct sched_entity *second = __pick_next_entity(se);
if (second && wakeup_preempt_entity(second, left) < 1)
se = second;
}
sched: Strengthen buddies and mitigate buddy induced latencies This patch restores the effectiveness of LAST_BUDDY in preventing pgsql+oltp from collapsing due to wakeup preemption. It also switches LAST_BUDDY to exclusively do what it does best, namely mitigate the effects of aggressive wakeup preemption, which improves vmark throughput markedly, and restores mysql+oltp scalability. Since buddies are about scalability, enable them beginning at the point where we begin expanding sched_latency, namely sched_nr_latency. Previously, buddies were cleared aggressively, which seriously reduced their effectiveness. Not clearing aggressively however, produces a small drop in mysql+oltp throughput immediately after peak, indicating that LAST_BUDDY is actually doing some harm. This is right at the point where X on the desktop in competition with another load wants low latency service. Ergo, do not enable until we need to scale. To mitigate latency induced by buddies, or by a task just missing wakeup preemption, check latency at tick time. Last hunk prevents buddies from stymieing BALANCE_NEWIDLE via CACHE_HOT_BUDDY. Supporting performance tests: tip = v2.6.32-rc5-1497-ga525b32 tipx = NO_GENTLE_FAIR_SLEEPERS NEXT_BUDDY granularity knobs = 31 knobs + 31 buddies tip+x = NO_GENTLE_FAIR_SLEEPERS granularity knobs = 31 knobs (Three run averages except where noted.) vmark: ------ tip 108466 messages per second tip+ 125307 messages per second tip+x 125335 messages per second tipx 117781 messages per second 2.6.31.3 122729 messages per second mysql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 9949.89 18690.20 34801.24 34460.04 32682.88 30765.97 28305.27 25059.64 19548.08 tip+ 10013.90 18526.84 34900.38 34420.14 33069.83 32083.40 30578.30 28010.71 25605.47 tipx 9698.71 18002.70 34477.56 33420.01 32634.30 31657.27 29932.67 26827.52 21487.18 2.6.31.3 8243.11 18784.20 34404.83 33148.38 31900.32 31161.90 29663.81 25995.94 18058.86 pgsql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 13686.37 26609.25 51934.28 51347.81 49479.51 45312.65 36691.91 26851.57 24145.35 tip+ (1x) 13907.85 27135.87 52951.98 52514.04 51742.52 50705.43 49947.97 48374.19 46227.94 tip+x 13906.78 27065.81 52951.19 52542.59 52176.11 51815.94 50838.90 49439.46 46891.00 tipx 13742.46 26769.81 52351.99 51891.73 51320.79 50938.98 50248.65 48908.70 46553.84 2.6.31.3 13815.35 26906.46 52683.34 52061.31 51937.10 51376.80 50474.28 49394.47 47003.25 Signed-off-by: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-23 21:09:22 +00:00
/*
* Prefer last buddy, try to return the CPU to a preempted task.
*/
if (cfs_rq->last && wakeup_preempt_entity(cfs_rq->last, left) < 1)
se = cfs_rq->last;
/*
* Someone really wants this to run. If it's not unfair, run it.
*/
if (cfs_rq->next && wakeup_preempt_entity(cfs_rq->next, left) < 1)
se = cfs_rq->next;
sched: Strengthen buddies and mitigate buddy induced latencies This patch restores the effectiveness of LAST_BUDDY in preventing pgsql+oltp from collapsing due to wakeup preemption. It also switches LAST_BUDDY to exclusively do what it does best, namely mitigate the effects of aggressive wakeup preemption, which improves vmark throughput markedly, and restores mysql+oltp scalability. Since buddies are about scalability, enable them beginning at the point where we begin expanding sched_latency, namely sched_nr_latency. Previously, buddies were cleared aggressively, which seriously reduced their effectiveness. Not clearing aggressively however, produces a small drop in mysql+oltp throughput immediately after peak, indicating that LAST_BUDDY is actually doing some harm. This is right at the point where X on the desktop in competition with another load wants low latency service. Ergo, do not enable until we need to scale. To mitigate latency induced by buddies, or by a task just missing wakeup preemption, check latency at tick time. Last hunk prevents buddies from stymieing BALANCE_NEWIDLE via CACHE_HOT_BUDDY. Supporting performance tests: tip = v2.6.32-rc5-1497-ga525b32 tipx = NO_GENTLE_FAIR_SLEEPERS NEXT_BUDDY granularity knobs = 31 knobs + 31 buddies tip+x = NO_GENTLE_FAIR_SLEEPERS granularity knobs = 31 knobs (Three run averages except where noted.) vmark: ------ tip 108466 messages per second tip+ 125307 messages per second tip+x 125335 messages per second tipx 117781 messages per second 2.6.31.3 122729 messages per second mysql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 9949.89 18690.20 34801.24 34460.04 32682.88 30765.97 28305.27 25059.64 19548.08 tip+ 10013.90 18526.84 34900.38 34420.14 33069.83 32083.40 30578.30 28010.71 25605.47 tipx 9698.71 18002.70 34477.56 33420.01 32634.30 31657.27 29932.67 26827.52 21487.18 2.6.31.3 8243.11 18784.20 34404.83 33148.38 31900.32 31161.90 29663.81 25995.94 18058.86 pgsql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 13686.37 26609.25 51934.28 51347.81 49479.51 45312.65 36691.91 26851.57 24145.35 tip+ (1x) 13907.85 27135.87 52951.98 52514.04 51742.52 50705.43 49947.97 48374.19 46227.94 tip+x 13906.78 27065.81 52951.19 52542.59 52176.11 51815.94 50838.90 49439.46 46891.00 tipx 13742.46 26769.81 52351.99 51891.73 51320.79 50938.98 50248.65 48908.70 46553.84 2.6.31.3 13815.35 26906.46 52683.34 52061.31 51937.10 51376.80 50474.28 49394.47 47003.25 Signed-off-by: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-23 21:09:22 +00:00
clear_buddies(cfs_rq, se);
return se;
}
static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq);
static void put_prev_entity(struct cfs_rq *cfs_rq, struct sched_entity *prev)
{
/*
* If still on the runqueue then deactivate_task()
* was not called and update_curr() has to be done:
*/
if (prev->on_rq)
update_curr(cfs_rq);
/* throttle cfs_rqs exceeding runtime */
check_cfs_rq_runtime(cfs_rq);
check_spread(cfs_rq, prev);
if (prev->on_rq) {
update_stats_wait_start(cfs_rq, prev);
/* Put 'current' back into the tree. */
__enqueue_entity(cfs_rq, prev);
}
cfs_rq->curr = NULL;
}
static void
entity_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr, int queued)
{
/*
* Update run-time statistics of the 'current'.
*/
update_curr(cfs_rq);
/*
* Update share accounting for long-running entities.
*/
update_entity_shares_tick(cfs_rq);
#ifdef CONFIG_SCHED_HRTICK
/*
* queued ticks are scheduled to match the slice, so don't bother
* validating it and just reschedule.
*/
if (queued) {
resched_task(rq_of(cfs_rq)->curr);
return;
}
/*
* don't let the period tick interfere with the hrtick preemption
*/
if (!sched_feat(DOUBLE_TICK) &&
hrtimer_active(&rq_of(cfs_rq)->hrtick_timer))
return;
#endif
if (cfs_rq->nr_running > 1)
check_preempt_tick(cfs_rq, curr);
}
/**************************************************
* CFS bandwidth control machinery
*/
#ifdef CONFIG_CFS_BANDWIDTH
#ifdef HAVE_JUMP_LABEL
static struct jump_label_key __cfs_bandwidth_used;
static inline bool cfs_bandwidth_used(void)
{
return static_branch(&__cfs_bandwidth_used);
}
void account_cfs_bandwidth_used(int enabled, int was_enabled)
{
/* only need to count groups transitioning between enabled/!enabled */
if (enabled && !was_enabled)
jump_label_inc(&__cfs_bandwidth_used);
else if (!enabled && was_enabled)
jump_label_dec(&__cfs_bandwidth_used);
}
#else /* HAVE_JUMP_LABEL */
static bool cfs_bandwidth_used(void)
{
return true;
}
void account_cfs_bandwidth_used(int enabled, int was_enabled) {}
#endif /* HAVE_JUMP_LABEL */
/*
* default period for cfs group bandwidth.
* default: 0.1s, units: nanoseconds
*/
static inline u64 default_cfs_period(void)
{
return 100000000ULL;
}
static inline u64 sched_cfs_bandwidth_slice(void)
{
return (u64)sysctl_sched_cfs_bandwidth_slice * NSEC_PER_USEC;
}
/*
* Replenish runtime according to assigned quota and update expiration time.
* We use sched_clock_cpu directly instead of rq->clock to avoid adding
* additional synchronization around rq->lock.
*
* requires cfs_b->lock
*/
void __refill_cfs_bandwidth_runtime(struct cfs_bandwidth *cfs_b)
{
u64 now;
if (cfs_b->quota == RUNTIME_INF)
return;
now = sched_clock_cpu(smp_processor_id());
cfs_b->runtime = cfs_b->quota;
cfs_b->runtime_expires = now + ktime_to_ns(cfs_b->period);
}
static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
{
return &tg->cfs_bandwidth;
}
/* returns 0 on failure to allocate runtime */
static int assign_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
struct task_group *tg = cfs_rq->tg;
struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(tg);
u64 amount = 0, min_amount, expires;
/* note: this is a positive sum as runtime_remaining <= 0 */
min_amount = sched_cfs_bandwidth_slice() - cfs_rq->runtime_remaining;
raw_spin_lock(&cfs_b->lock);
if (cfs_b->quota == RUNTIME_INF)
amount = min_amount;
else {
/*
* If the bandwidth pool has become inactive, then at least one
* period must have elapsed since the last consumption.
* Refresh the global state and ensure bandwidth timer becomes
* active.
*/
if (!cfs_b->timer_active) {
__refill_cfs_bandwidth_runtime(cfs_b);
__start_cfs_bandwidth(cfs_b);
}
if (cfs_b->runtime > 0) {
amount = min(cfs_b->runtime, min_amount);
cfs_b->runtime -= amount;
cfs_b->idle = 0;
}
}
expires = cfs_b->runtime_expires;
raw_spin_unlock(&cfs_b->lock);
cfs_rq->runtime_remaining += amount;
/*
* we may have advanced our local expiration to account for allowed
* spread between our sched_clock and the one on which runtime was
* issued.
*/
if ((s64)(expires - cfs_rq->runtime_expires) > 0)
cfs_rq->runtime_expires = expires;
return cfs_rq->runtime_remaining > 0;
}
/*
* Note: This depends on the synchronization provided by sched_clock and the
* fact that rq->clock snapshots this value.
*/
static void expire_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
struct rq *rq = rq_of(cfs_rq);
/* if the deadline is ahead of our clock, nothing to do */
if (likely((s64)(rq->clock - cfs_rq->runtime_expires) < 0))
return;
if (cfs_rq->runtime_remaining < 0)
return;
/*
* If the local deadline has passed we have to consider the
* possibility that our sched_clock is 'fast' and the global deadline
* has not truly expired.
*
* Fortunately we can check determine whether this the case by checking
* whether the global deadline has advanced.
*/
if ((s64)(cfs_rq->runtime_expires - cfs_b->runtime_expires) >= 0) {
/* extend local deadline, drift is bounded above by 2 ticks */
cfs_rq->runtime_expires += TICK_NSEC;
} else {
/* global deadline is ahead, expiration has passed */
cfs_rq->runtime_remaining = 0;
}
}
static void __account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
unsigned long delta_exec)
{
/* dock delta_exec before expiring quota (as it could span periods) */
cfs_rq->runtime_remaining -= delta_exec;
expire_cfs_rq_runtime(cfs_rq);
if (likely(cfs_rq->runtime_remaining > 0))
return;
/*
* if we're unable to extend our runtime we resched so that the active
* hierarchy can be throttled
*/
if (!assign_cfs_rq_runtime(cfs_rq) && likely(cfs_rq->curr))
resched_task(rq_of(cfs_rq)->curr);
}
static __always_inline void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
unsigned long delta_exec)
{
if (!cfs_bandwidth_used() || !cfs_rq->runtime_enabled)
return;
__account_cfs_rq_runtime(cfs_rq, delta_exec);
}
static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
{
return cfs_bandwidth_used() && cfs_rq->throttled;
}
/* check whether cfs_rq, or any parent, is throttled */
static inline int throttled_hierarchy(struct cfs_rq *cfs_rq)
{
return cfs_bandwidth_used() && cfs_rq->throttle_count;
}
/*
* Ensure that neither of the group entities corresponding to src_cpu or
* dest_cpu are members of a throttled hierarchy when performing group
* load-balance operations.
*/
static inline int throttled_lb_pair(struct task_group *tg,
int src_cpu, int dest_cpu)
{
struct cfs_rq *src_cfs_rq, *dest_cfs_rq;
src_cfs_rq = tg->cfs_rq[src_cpu];
dest_cfs_rq = tg->cfs_rq[dest_cpu];
return throttled_hierarchy(src_cfs_rq) ||
throttled_hierarchy(dest_cfs_rq);
}
/* updated child weight may affect parent so we have to do this bottom up */
static int tg_unthrottle_up(struct task_group *tg, void *data)
{
struct rq *rq = data;
struct cfs_rq *cfs_rq = tg->cfs_rq[cpu_of(rq)];
cfs_rq->throttle_count--;
#ifdef CONFIG_SMP
if (!cfs_rq->throttle_count) {
u64 delta = rq->clock_task - cfs_rq->load_stamp;
/* leaving throttled state, advance shares averaging windows */
cfs_rq->load_stamp += delta;
cfs_rq->load_last += delta;
/* update entity weight now that we are on_rq again */
update_cfs_shares(cfs_rq);
}
#endif
return 0;
}
static int tg_throttle_down(struct task_group *tg, void *data)
{
struct rq *rq = data;
struct cfs_rq *cfs_rq = tg->cfs_rq[cpu_of(rq)];
/* group is entering throttled state, record last load */
if (!cfs_rq->throttle_count)
update_cfs_load(cfs_rq, 0);
cfs_rq->throttle_count++;
return 0;
}
static void throttle_cfs_rq(struct cfs_rq *cfs_rq)
{
struct rq *rq = rq_of(cfs_rq);
struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
struct sched_entity *se;
long task_delta, dequeue = 1;
se = cfs_rq->tg->se[cpu_of(rq_of(cfs_rq))];
/* account load preceding throttle */
rcu_read_lock();
walk_tg_tree_from(cfs_rq->tg, tg_throttle_down, tg_nop, (void *)rq);
rcu_read_unlock();
task_delta = cfs_rq->h_nr_running;
for_each_sched_entity(se) {
struct cfs_rq *qcfs_rq = cfs_rq_of(se);
/* throttled entity or throttle-on-deactivate */
if (!se->on_rq)
break;
if (dequeue)
dequeue_entity(qcfs_rq, se, DEQUEUE_SLEEP);
qcfs_rq->h_nr_running -= task_delta;
if (qcfs_rq->load.weight)
dequeue = 0;
}
if (!se)
rq->nr_running -= task_delta;
cfs_rq->throttled = 1;
cfs_rq->throttled_timestamp = rq->clock;
raw_spin_lock(&cfs_b->lock);
list_add_tail_rcu(&cfs_rq->throttled_list, &cfs_b->throttled_cfs_rq);
raw_spin_unlock(&cfs_b->lock);
}
void unthrottle_cfs_rq(struct cfs_rq *cfs_rq)
{
struct rq *rq = rq_of(cfs_rq);
struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
struct sched_entity *se;
int enqueue = 1;
long task_delta;
se = cfs_rq->tg->se[cpu_of(rq_of(cfs_rq))];
cfs_rq->throttled = 0;
raw_spin_lock(&cfs_b->lock);
cfs_b->throttled_time += rq->clock - cfs_rq->throttled_timestamp;
list_del_rcu(&cfs_rq->throttled_list);
raw_spin_unlock(&cfs_b->lock);
cfs_rq->throttled_timestamp = 0;
update_rq_clock(rq);
/* update hierarchical throttle state */
walk_tg_tree_from(cfs_rq->tg, tg_nop, tg_unthrottle_up, (void *)rq);
if (!cfs_rq->load.weight)
return;
task_delta = cfs_rq->h_nr_running;
for_each_sched_entity(se) {
if (se->on_rq)
enqueue = 0;
cfs_rq = cfs_rq_of(se);
if (enqueue)
enqueue_entity(cfs_rq, se, ENQUEUE_WAKEUP);
cfs_rq->h_nr_running += task_delta;
if (cfs_rq_throttled(cfs_rq))
break;
}
if (!se)
rq->nr_running += task_delta;
/* determine whether we need to wake up potentially idle cpu */
if (rq->curr == rq->idle && rq->cfs.nr_running)
resched_task(rq->curr);
}
static u64 distribute_cfs_runtime(struct cfs_bandwidth *cfs_b,
u64 remaining, u64 expires)
{
struct cfs_rq *cfs_rq;
u64 runtime = remaining;
rcu_read_lock();
list_for_each_entry_rcu(cfs_rq, &cfs_b->throttled_cfs_rq,
throttled_list) {
struct rq *rq = rq_of(cfs_rq);
raw_spin_lock(&rq->lock);
if (!cfs_rq_throttled(cfs_rq))
goto next;
runtime = -cfs_rq->runtime_remaining + 1;
if (runtime > remaining)
runtime = remaining;
remaining -= runtime;
cfs_rq->runtime_remaining += runtime;
cfs_rq->runtime_expires = expires;
/* we check whether we're throttled above */
if (cfs_rq->runtime_remaining > 0)
unthrottle_cfs_rq(cfs_rq);
next:
raw_spin_unlock(&rq->lock);
if (!remaining)
break;
}
rcu_read_unlock();
return remaining;
}
/*
* Responsible for refilling a task_group's bandwidth and unthrottling its
* cfs_rqs as appropriate. If there has been no activity within the last
* period the timer is deactivated until scheduling resumes; cfs_b->idle is
* used to track this state.
*/
static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun)
{
u64 runtime, runtime_expires;
int idle = 1, throttled;
raw_spin_lock(&cfs_b->lock);
/* no need to continue the timer with no bandwidth constraint */
if (cfs_b->quota == RUNTIME_INF)
goto out_unlock;
throttled = !list_empty(&cfs_b->throttled_cfs_rq);
/* idle depends on !throttled (for the case of a large deficit) */
idle = cfs_b->idle && !throttled;
cfs_b->nr_periods += overrun;
/* if we're going inactive then everything else can be deferred */
if (idle)
goto out_unlock;
__refill_cfs_bandwidth_runtime(cfs_b);
if (!throttled) {
/* mark as potentially idle for the upcoming period */
cfs_b->idle = 1;
goto out_unlock;
}
/* account preceding periods in which throttling occurred */
cfs_b->nr_throttled += overrun;
/*
* There are throttled entities so we must first use the new bandwidth
* to unthrottle them before making it generally available. This
* ensures that all existing debts will be paid before a new cfs_rq is
* allowed to run.
*/
runtime = cfs_b->runtime;
runtime_expires = cfs_b->runtime_expires;
cfs_b->runtime = 0;
/*
* This check is repeated as we are holding onto the new bandwidth
* while we unthrottle. This can potentially race with an unthrottled
* group trying to acquire new bandwidth from the global pool.
*/
while (throttled && runtime > 0) {
raw_spin_unlock(&cfs_b->lock);
/* we can't nest cfs_b->lock while distributing bandwidth */
runtime = distribute_cfs_runtime(cfs_b, runtime,
runtime_expires);
raw_spin_lock(&cfs_b->lock);
throttled = !list_empty(&cfs_b->throttled_cfs_rq);
}
/* return (any) remaining runtime */
cfs_b->runtime = runtime;
/*
* While we are ensured activity in the period following an
* unthrottle, this also covers the case in which the new bandwidth is
* insufficient to cover the existing bandwidth deficit. (Forcing the
* timer to remain active while there are any throttled entities.)
*/
cfs_b->idle = 0;
out_unlock:
if (idle)
cfs_b->timer_active = 0;
raw_spin_unlock(&cfs_b->lock);
return idle;
}
sched: Return unused runtime on group dequeue When a local cfs_rq blocks we return the majority of its remaining quota to the global bandwidth pool for use by other runqueues. We do this only when the quota is current and there is more than min_cfs_rq_quota [1ms by default] of runtime remaining on the rq. In the case where there are throttled runqueues and we have sufficient bandwidth to meter out a slice, a second timer is kicked off to handle this delivery, unthrottling where appropriate. Using a 'worst case' antagonist which executes on each cpu for 1ms before moving onto the next on a fairly large machine: no quota generations: 197.47 ms /cgroup/a/cpuacct.usage 199.46 ms /cgroup/a/cpuacct.usage 205.46 ms /cgroup/a/cpuacct.usage 198.46 ms /cgroup/a/cpuacct.usage 208.39 ms /cgroup/a/cpuacct.usage Since we are allowed to use "stale" quota our usage is effectively bounded by the rate of input into the global pool and performance is relatively stable. with quota generations [1s increments]: 119.58 ms /cgroup/a/cpuacct.usage 119.65 ms /cgroup/a/cpuacct.usage 119.64 ms /cgroup/a/cpuacct.usage 119.63 ms /cgroup/a/cpuacct.usage 119.60 ms /cgroup/a/cpuacct.usage The large deficit here is due to quota generations (/intentionally/) preventing us from now using previously stranded slack quota. The cost is that this quota becomes unavailable. with quota generations and quota return: 200.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 198.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 200.06 ms /cgroup/a/cpuacct.usage By returning unused quota we're able to both stably consume our desired quota and prevent unintentional overages due to the abuse of slack quota from previous quota periods (especially on a large machine). Signed-off-by: Paul Turner <pjt@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20110721184758.306848658@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-07-21 16:43:41 +00:00
/* a cfs_rq won't donate quota below this amount */
static const u64 min_cfs_rq_runtime = 1 * NSEC_PER_MSEC;
/* minimum remaining period time to redistribute slack quota */
static const u64 min_bandwidth_expiration = 2 * NSEC_PER_MSEC;
/* how long we wait to gather additional slack before distributing */
static const u64 cfs_bandwidth_slack_period = 5 * NSEC_PER_MSEC;
/* are we near the end of the current quota period? */
static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
{
struct hrtimer *refresh_timer = &cfs_b->period_timer;
u64 remaining;
/* if the call-back is running a quota refresh is already occurring */
if (hrtimer_callback_running(refresh_timer))
return 1;
/* is a quota refresh about to occur? */
remaining = ktime_to_ns(hrtimer_expires_remaining(refresh_timer));
if (remaining < min_expire)
return 1;
return 0;
}
static void start_cfs_slack_bandwidth(struct cfs_bandwidth *cfs_b)
{
u64 min_left = cfs_bandwidth_slack_period + min_bandwidth_expiration;
/* if there's a quota refresh soon don't bother with slack */
if (runtime_refresh_within(cfs_b, min_left))
return;
start_bandwidth_timer(&cfs_b->slack_timer,
ns_to_ktime(cfs_bandwidth_slack_period));
}
/* we know any runtime found here is valid as update_curr() precedes return */
static void __return_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
s64 slack_runtime = cfs_rq->runtime_remaining - min_cfs_rq_runtime;
if (slack_runtime <= 0)
return;
raw_spin_lock(&cfs_b->lock);
if (cfs_b->quota != RUNTIME_INF &&
cfs_rq->runtime_expires == cfs_b->runtime_expires) {
cfs_b->runtime += slack_runtime;
/* we are under rq->lock, defer unthrottling using a timer */
if (cfs_b->runtime > sched_cfs_bandwidth_slice() &&
!list_empty(&cfs_b->throttled_cfs_rq))
start_cfs_slack_bandwidth(cfs_b);
}
raw_spin_unlock(&cfs_b->lock);
/* even if it's not valid for return we don't want to try again */
cfs_rq->runtime_remaining -= slack_runtime;
}
static __always_inline void return_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
if (!cfs_bandwidth_used())
return;
if (!cfs_rq->runtime_enabled || cfs_rq->nr_running)
sched: Return unused runtime on group dequeue When a local cfs_rq blocks we return the majority of its remaining quota to the global bandwidth pool for use by other runqueues. We do this only when the quota is current and there is more than min_cfs_rq_quota [1ms by default] of runtime remaining on the rq. In the case where there are throttled runqueues and we have sufficient bandwidth to meter out a slice, a second timer is kicked off to handle this delivery, unthrottling where appropriate. Using a 'worst case' antagonist which executes on each cpu for 1ms before moving onto the next on a fairly large machine: no quota generations: 197.47 ms /cgroup/a/cpuacct.usage 199.46 ms /cgroup/a/cpuacct.usage 205.46 ms /cgroup/a/cpuacct.usage 198.46 ms /cgroup/a/cpuacct.usage 208.39 ms /cgroup/a/cpuacct.usage Since we are allowed to use "stale" quota our usage is effectively bounded by the rate of input into the global pool and performance is relatively stable. with quota generations [1s increments]: 119.58 ms /cgroup/a/cpuacct.usage 119.65 ms /cgroup/a/cpuacct.usage 119.64 ms /cgroup/a/cpuacct.usage 119.63 ms /cgroup/a/cpuacct.usage 119.60 ms /cgroup/a/cpuacct.usage The large deficit here is due to quota generations (/intentionally/) preventing us from now using previously stranded slack quota. The cost is that this quota becomes unavailable. with quota generations and quota return: 200.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 198.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 200.06 ms /cgroup/a/cpuacct.usage By returning unused quota we're able to both stably consume our desired quota and prevent unintentional overages due to the abuse of slack quota from previous quota periods (especially on a large machine). Signed-off-by: Paul Turner <pjt@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20110721184758.306848658@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-07-21 16:43:41 +00:00
return;
__return_cfs_rq_runtime(cfs_rq);
}
/*
* This is done with a timer (instead of inline with bandwidth return) since
* it's necessary to juggle rq->locks to unthrottle their respective cfs_rqs.
*/
static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b)
{
u64 runtime = 0, slice = sched_cfs_bandwidth_slice();
u64 expires;
/* confirm we're still not at a refresh boundary */
if (runtime_refresh_within(cfs_b, min_bandwidth_expiration))
return;
raw_spin_lock(&cfs_b->lock);
if (cfs_b->quota != RUNTIME_INF && cfs_b->runtime > slice) {
runtime = cfs_b->runtime;
cfs_b->runtime = 0;
}
expires = cfs_b->runtime_expires;
raw_spin_unlock(&cfs_b->lock);
if (!runtime)
return;
runtime = distribute_cfs_runtime(cfs_b, runtime, expires);
raw_spin_lock(&cfs_b->lock);
if (expires == cfs_b->runtime_expires)
cfs_b->runtime = runtime;
raw_spin_unlock(&cfs_b->lock);
}
/*
* When a group wakes up we want to make sure that its quota is not already
* expired/exceeded, otherwise it may be allowed to steal additional ticks of
* runtime as update_curr() throttling can not not trigger until it's on-rq.
*/
static void check_enqueue_throttle(struct cfs_rq *cfs_rq)
{
if (!cfs_bandwidth_used())
return;
/* an active group must be handled by the update_curr()->put() path */
if (!cfs_rq->runtime_enabled || cfs_rq->curr)
return;
/* ensure the group is not already throttled */
if (cfs_rq_throttled(cfs_rq))
return;
/* update runtime allocation */
account_cfs_rq_runtime(cfs_rq, 0);
if (cfs_rq->runtime_remaining <= 0)
throttle_cfs_rq(cfs_rq);
}
/* conditionally throttle active cfs_rq's from put_prev_entity() */
static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
if (!cfs_bandwidth_used())
return;
if (likely(!cfs_rq->runtime_enabled || cfs_rq->runtime_remaining > 0))
return;
/*
* it's possible for a throttled entity to be forced into a running
* state (e.g. set_curr_task), in this case we're finished.
*/
if (cfs_rq_throttled(cfs_rq))
return;
throttle_cfs_rq(cfs_rq);
}
static inline u64 default_cfs_period(void);
static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun);
static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b);
static enum hrtimer_restart sched_cfs_slack_timer(struct hrtimer *timer)
{
struct cfs_bandwidth *cfs_b =
container_of(timer, struct cfs_bandwidth, slack_timer);
do_sched_cfs_slack_timer(cfs_b);
return HRTIMER_NORESTART;
}
static enum hrtimer_restart sched_cfs_period_timer(struct hrtimer *timer)
{
struct cfs_bandwidth *cfs_b =
container_of(timer, struct cfs_bandwidth, period_timer);
ktime_t now;
int overrun;
int idle = 0;
for (;;) {
now = hrtimer_cb_get_time(timer);
overrun = hrtimer_forward(timer, now, cfs_b->period);
if (!overrun)
break;
idle = do_sched_cfs_period_timer(cfs_b, overrun);
}
return idle ? HRTIMER_NORESTART : HRTIMER_RESTART;
}
void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b)
{
raw_spin_lock_init(&cfs_b->lock);
cfs_b->runtime = 0;
cfs_b->quota = RUNTIME_INF;
cfs_b->period = ns_to_ktime(default_cfs_period());
INIT_LIST_HEAD(&cfs_b->throttled_cfs_rq);
hrtimer_init(&cfs_b->period_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
cfs_b->period_timer.function = sched_cfs_period_timer;
hrtimer_init(&cfs_b->slack_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
cfs_b->slack_timer.function = sched_cfs_slack_timer;
}
static void init_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
cfs_rq->runtime_enabled = 0;
INIT_LIST_HEAD(&cfs_rq->throttled_list);
}
/* requires cfs_b->lock, may release to reprogram timer */
void __start_cfs_bandwidth(struct cfs_bandwidth *cfs_b)
{
/*
* The timer may be active because we're trying to set a new bandwidth
* period or because we're racing with the tear-down path
* (timer_active==0 becomes visible before the hrtimer call-back
* terminates). In either case we ensure that it's re-programmed
*/
while (unlikely(hrtimer_active(&cfs_b->period_timer))) {
raw_spin_unlock(&cfs_b->lock);
/* ensure cfs_b->lock is available while we wait */
hrtimer_cancel(&cfs_b->period_timer);
raw_spin_lock(&cfs_b->lock);
/* if someone else restarted the timer then we're done */
if (cfs_b->timer_active)
return;
}
cfs_b->timer_active = 1;
start_bandwidth_timer(&cfs_b->period_timer, cfs_b->period);
}
static void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b)
{
hrtimer_cancel(&cfs_b->period_timer);
hrtimer_cancel(&cfs_b->slack_timer);
}
void unthrottle_offline_cfs_rqs(struct rq *rq)
{
struct cfs_rq *cfs_rq;
for_each_leaf_cfs_rq(rq, cfs_rq) {
struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
if (!cfs_rq->runtime_enabled)
continue;
/*
* clock_task is not advancing so we just need to make sure
* there's some valid quota amount
*/
cfs_rq->runtime_remaining = cfs_b->quota;
if (cfs_rq_throttled(cfs_rq))
unthrottle_cfs_rq(cfs_rq);
}
}
#else /* CONFIG_CFS_BANDWIDTH */
static void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
unsigned long delta_exec) {}
static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq) {}
static void check_enqueue_throttle(struct cfs_rq *cfs_rq) {}
sched: Return unused runtime on group dequeue When a local cfs_rq blocks we return the majority of its remaining quota to the global bandwidth pool for use by other runqueues. We do this only when the quota is current and there is more than min_cfs_rq_quota [1ms by default] of runtime remaining on the rq. In the case where there are throttled runqueues and we have sufficient bandwidth to meter out a slice, a second timer is kicked off to handle this delivery, unthrottling where appropriate. Using a 'worst case' antagonist which executes on each cpu for 1ms before moving onto the next on a fairly large machine: no quota generations: 197.47 ms /cgroup/a/cpuacct.usage 199.46 ms /cgroup/a/cpuacct.usage 205.46 ms /cgroup/a/cpuacct.usage 198.46 ms /cgroup/a/cpuacct.usage 208.39 ms /cgroup/a/cpuacct.usage Since we are allowed to use "stale" quota our usage is effectively bounded by the rate of input into the global pool and performance is relatively stable. with quota generations [1s increments]: 119.58 ms /cgroup/a/cpuacct.usage 119.65 ms /cgroup/a/cpuacct.usage 119.64 ms /cgroup/a/cpuacct.usage 119.63 ms /cgroup/a/cpuacct.usage 119.60 ms /cgroup/a/cpuacct.usage The large deficit here is due to quota generations (/intentionally/) preventing us from now using previously stranded slack quota. The cost is that this quota becomes unavailable. with quota generations and quota return: 200.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 198.09 ms /cgroup/a/cpuacct.usage 200.09 ms /cgroup/a/cpuacct.usage 200.06 ms /cgroup/a/cpuacct.usage By returning unused quota we're able to both stably consume our desired quota and prevent unintentional overages due to the abuse of slack quota from previous quota periods (especially on a large machine). Signed-off-by: Paul Turner <pjt@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20110721184758.306848658@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-07-21 16:43:41 +00:00
static void return_cfs_rq_runtime(struct cfs_rq *cfs_rq) {}
static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
{
return 0;
}
static inline int throttled_hierarchy(struct cfs_rq *cfs_rq)
{
return 0;
}
static inline int throttled_lb_pair(struct task_group *tg,
int src_cpu, int dest_cpu)
{
return 0;
}
void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
#ifdef CONFIG_FAIR_GROUP_SCHED
static void init_cfs_rq_runtime(struct cfs_rq *cfs_rq) {}
#endif
static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
{
return NULL;
}
static inline void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
void unthrottle_offline_cfs_rqs(struct rq *rq) {}
#endif /* CONFIG_CFS_BANDWIDTH */
/**************************************************
* CFS operations on tasks:
*/
#ifdef CONFIG_SCHED_HRTICK
static void hrtick_start_fair(struct rq *rq, struct task_struct *p)
{
struct sched_entity *se = &p->se;
struct cfs_rq *cfs_rq = cfs_rq_of(se);
WARN_ON(task_rq(p) != rq);
if (hrtick_enabled(rq) && cfs_rq->nr_running > 1) {
u64 slice = sched_slice(cfs_rq, se);
u64 ran = se->sum_exec_runtime - se->prev_sum_exec_runtime;
s64 delta = slice - ran;
if (delta < 0) {
if (rq->curr == p)
resched_task(p);
return;
}
/*
* Don't schedule slices shorter than 10000ns, that just
* doesn't make sense. Rely on vruntime for fairness.
*/
sched, x86: clean up hrtick implementation random uvesafb failures were reported against Gentoo: http://bugs.gentoo.org/show_bug.cgi?id=222799 and Mihai Moldovan bisected it back to: > 8f4d37ec073c17e2d4aa8851df5837d798606d6f is first bad commit > commit 8f4d37ec073c17e2d4aa8851df5837d798606d6f > Author: Peter Zijlstra <a.p.zijlstra@chello.nl> > Date: Fri Jan 25 21:08:29 2008 +0100 > > sched: high-res preemption tick Linus suspected it to be hrtick + vm86 interaction and observed: > Btw, Peter, Ingo: I think that commit is doing bad things. They aren't > _incorrect_ per se, but they are definitely bad. > > Why? > > Using random _TIF_WORK_MASK flags is really impolite for doing > "scheduling" work. There's a reason that arch/x86/kernel/entry_32.S > special-cases the _TIF_NEED_RESCHED flag: we don't want to exit out of > vm86 mode unnecessarily. > > See the "work_notifysig_v86" label, and how it does that > "save_v86_state()" thing etc etc. Right, I never liked having to fiddle with those TIF flags. Initially I needed it because the hrtimer base lock could not nest in the rq lock. That however is fixed these days. Currently the only reason left to fiddle with the TIF flags is remote wakeups. We cannot program a remote cpu's hrtimer. I've been thinking about using the new and improved IPI function call stuff to implement hrtimer_start_on(). However that does require that smp_call_function_single(.wait=0) works from interrupt context - /me looks at the latest series from Jens - Yes that does seem to be supported, good. Here's a stab at cleaning this stuff up ... Mihai reported test success as well. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Tested-by: Mihai Moldovan <ionic@ionic.de> Cc: Michal Januszewski <spock@gentoo.org> Cc: Antonino Daplas <adaplas@gmail.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-18 16:01:23 +00:00
if (rq->curr != p)
delta = max_t(s64, 10000LL, delta);
sched, x86: clean up hrtick implementation random uvesafb failures were reported against Gentoo: http://bugs.gentoo.org/show_bug.cgi?id=222799 and Mihai Moldovan bisected it back to: > 8f4d37ec073c17e2d4aa8851df5837d798606d6f is first bad commit > commit 8f4d37ec073c17e2d4aa8851df5837d798606d6f > Author: Peter Zijlstra <a.p.zijlstra@chello.nl> > Date: Fri Jan 25 21:08:29 2008 +0100 > > sched: high-res preemption tick Linus suspected it to be hrtick + vm86 interaction and observed: > Btw, Peter, Ingo: I think that commit is doing bad things. They aren't > _incorrect_ per se, but they are definitely bad. > > Why? > > Using random _TIF_WORK_MASK flags is really impolite for doing > "scheduling" work. There's a reason that arch/x86/kernel/entry_32.S > special-cases the _TIF_NEED_RESCHED flag: we don't want to exit out of > vm86 mode unnecessarily. > > See the "work_notifysig_v86" label, and how it does that > "save_v86_state()" thing etc etc. Right, I never liked having to fiddle with those TIF flags. Initially I needed it because the hrtimer base lock could not nest in the rq lock. That however is fixed these days. Currently the only reason left to fiddle with the TIF flags is remote wakeups. We cannot program a remote cpu's hrtimer. I've been thinking about using the new and improved IPI function call stuff to implement hrtimer_start_on(). However that does require that smp_call_function_single(.wait=0) works from interrupt context - /me looks at the latest series from Jens - Yes that does seem to be supported, good. Here's a stab at cleaning this stuff up ... Mihai reported test success as well. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Tested-by: Mihai Moldovan <ionic@ionic.de> Cc: Michal Januszewski <spock@gentoo.org> Cc: Antonino Daplas <adaplas@gmail.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-18 16:01:23 +00:00
hrtick_start(rq, delta);
}
}
/*
* called from enqueue/dequeue and updates the hrtick when the
* current task is from our class and nr_running is low enough
* to matter.
*/
static void hrtick_update(struct rq *rq)
{
struct task_struct *curr = rq->curr;
if (curr->sched_class != &fair_sched_class)
return;
if (cfs_rq_of(&curr->se)->nr_running < sched_nr_latency)
hrtick_start_fair(rq, curr);
}
#else /* !CONFIG_SCHED_HRTICK */
static inline void
hrtick_start_fair(struct rq *rq, struct task_struct *p)
{
}
static inline void hrtick_update(struct rq *rq)
{
}
#endif
/*
* The enqueue_task method is called before nr_running is
* increased. Here we update the fair scheduling stats and
* then put the task into the rbtree:
*/
static void
enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
{
struct cfs_rq *cfs_rq;
struct sched_entity *se = &p->se;
for_each_sched_entity(se) {
if (se->on_rq)
break;
cfs_rq = cfs_rq_of(se);
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
enqueue_entity(cfs_rq, se, flags);
/*
* end evaluation on encountering a throttled cfs_rq
*
* note: in the case of encountering a throttled cfs_rq we will
* post the final h_nr_running increment below.
*/
if (cfs_rq_throttled(cfs_rq))
break;
cfs_rq->h_nr_running++;
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
flags = ENQUEUE_WAKEUP;
}
for_each_sched_entity(se) {
cfs_rq = cfs_rq_of(se);
cfs_rq->h_nr_running++;
if (cfs_rq_throttled(cfs_rq))
break;
update_cfs_load(cfs_rq, 0);
update_cfs_shares(cfs_rq);
}
if (!se)
inc_nr_running(rq);
hrtick_update(rq);
}
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
static void set_next_buddy(struct sched_entity *se);
/*
* The dequeue_task method is called before nr_running is
* decreased. We remove the task from the rbtree and
* update the fair scheduling stats:
*/
static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags)
{
struct cfs_rq *cfs_rq;
struct sched_entity *se = &p->se;
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
int task_sleep = flags & DEQUEUE_SLEEP;
for_each_sched_entity(se) {
cfs_rq = cfs_rq_of(se);
dequeue_entity(cfs_rq, se, flags);
/*
* end evaluation on encountering a throttled cfs_rq
*
* note: in the case of encountering a throttled cfs_rq we will
* post the final h_nr_running decrement below.
*/
if (cfs_rq_throttled(cfs_rq))
break;
cfs_rq->h_nr_running--;
/* Don't dequeue parent if it has other entities besides us */
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
if (cfs_rq->load.weight) {
/*
* Bias pick_next to pick a task from this cfs_rq, as
* p is sleeping when it is within its sched_slice.
*/
if (task_sleep && parent_entity(se))
set_next_buddy(parent_entity(se));
/* avoid re-evaluating load for this entity */
se = parent_entity(se);
break;
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
}
flags |= DEQUEUE_SLEEP;
}
for_each_sched_entity(se) {
cfs_rq = cfs_rq_of(se);
cfs_rq->h_nr_running--;
if (cfs_rq_throttled(cfs_rq))
break;
update_cfs_load(cfs_rq, 0);
update_cfs_shares(cfs_rq);
}
if (!se)
dec_nr_running(rq);
hrtick_update(rq);
}
#ifdef CONFIG_SMP
/* Used instead of source_load when we know the type == 0 */
static unsigned long weighted_cpuload(const int cpu)
{
return cpu_rq(cpu)->load.weight;
}
/*
* Return a low guess at the load of a migration-source cpu weighted
* according to the scheduling class and "nice" value.
*
* We want to under-estimate the load of migration sources, to
* balance conservatively.
*/
static unsigned long source_load(int cpu, int type)
{
struct rq *rq = cpu_rq(cpu);
unsigned long total = weighted_cpuload(cpu);
if (type == 0 || !sched_feat(LB_BIAS))
return total;
return min(rq->cpu_load[type-1], total);
}
/*
* Return a high guess at the load of a migration-target cpu weighted
* according to the scheduling class and "nice" value.
*/
static unsigned long target_load(int cpu, int type)
{
struct rq *rq = cpu_rq(cpu);
unsigned long total = weighted_cpuload(cpu);
if (type == 0 || !sched_feat(LB_BIAS))
return total;
return max(rq->cpu_load[type-1], total);
}
static unsigned long power_of(int cpu)
{
return cpu_rq(cpu)->cpu_power;
}
static unsigned long cpu_avg_load_per_task(int cpu)
{
struct rq *rq = cpu_rq(cpu);
unsigned long nr_running = ACCESS_ONCE(rq->nr_running);
if (nr_running)
return rq->load.weight / nr_running;
return 0;
}
static void task_waking_fair(struct task_struct *p)
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
{
struct sched_entity *se = &p->se;
struct cfs_rq *cfs_rq = cfs_rq_of(se);
u64 min_vruntime;
#ifndef CONFIG_64BIT
u64 min_vruntime_copy;
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
do {
min_vruntime_copy = cfs_rq->min_vruntime_copy;
smp_rmb();
min_vruntime = cfs_rq->min_vruntime;
} while (min_vruntime != min_vruntime_copy);
#else
min_vruntime = cfs_rq->min_vruntime;
#endif
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
se->vruntime -= min_vruntime;
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
}
#ifdef CONFIG_FAIR_GROUP_SCHED
/*
* effective_load() calculates the load change as seen from the root_task_group
*
* Adding load to a group doesn't make a group heavier, but can cause movement
* of group shares between cpus. Assuming the shares were perfectly aligned one
* can calculate the shift in shares.
*
* Calculate the effective load difference if @wl is added (subtracted) to @tg
* on this @cpu and results in a total addition (subtraction) of @wg to the
* total group weight.
*
* Given a runqueue weight distribution (rw_i) we can compute a shares
* distribution (s_i) using:
*
* s_i = rw_i / \Sum rw_j (1)
*
* Suppose we have 4 CPUs and our @tg is a direct child of the root group and
* has 7 equal weight tasks, distributed as below (rw_i), with the resulting
* shares distribution (s_i):
*
* rw_i = { 2, 4, 1, 0 }
* s_i = { 2/7, 4/7, 1/7, 0 }
*
* As per wake_affine() we're interested in the load of two CPUs (the CPU the
* task used to run on and the CPU the waker is running on), we need to
* compute the effect of waking a task on either CPU and, in case of a sync
* wakeup, compute the effect of the current task going to sleep.
*
* So for a change of @wl to the local @cpu with an overall group weight change
* of @wl we can compute the new shares distribution (s'_i) using:
*
* s'_i = (rw_i + @wl) / (@wg + \Sum rw_j) (2)
*
* Suppose we're interested in CPUs 0 and 1, and want to compute the load
* differences in waking a task to CPU 0. The additional task changes the
* weight and shares distributions like:
*
* rw'_i = { 3, 4, 1, 0 }
* s'_i = { 3/8, 4/8, 1/8, 0 }
*
* We can then compute the difference in effective weight by using:
*
* dw_i = S * (s'_i - s_i) (3)
*
* Where 'S' is the group weight as seen by its parent.
*
* Therefore the effective change in loads on CPU 0 would be 5/56 (3/8 - 2/7)
* times the weight of the group. The effect on CPU 1 would be -4/56 (4/8 -
* 4/7) times the weight of the group.
*/
static long effective_load(struct task_group *tg, int cpu, long wl, long wg)
{
struct sched_entity *se = tg->se[cpu];
if (!tg->parent) /* the trivial, non-cgroup case */
return wl;
for_each_sched_entity(se) {
long w, W;
tg = se->my_q->tg;
/*
* W = @wg + \Sum rw_j
*/
W = wg + calc_tg_weight(tg, se->my_q);
/*
* w = rw_i + @wl
*/
w = se->my_q->load.weight + wl;
/*
* wl = S * s'_i; see (2)
*/
if (W > 0 && w < W)
wl = (w * tg->shares) / W;
else
wl = tg->shares;
/*
* Per the above, wl is the new se->load.weight value; since
* those are clipped to [MIN_SHARES, ...) do so now. See
* calc_cfs_shares().
*/
if (wl < MIN_SHARES)
wl = MIN_SHARES;
/*
* wl = dw_i = S * (s'_i - s_i); see (3)
*/
wl -= se->load.weight;
/*
* Recursively apply this logic to all parent groups to compute
* the final effective load change on the root group. Since
* only the @tg group gets extra weight, all parent groups can
* only redistribute existing shares. @wl is the shift in shares
* resulting from this level per the above.
*/
wg = 0;
}
return wl;
}
#else
static inline unsigned long effective_load(struct task_group *tg, int cpu,
unsigned long wl, unsigned long wg)
{
return wl;
}
#endif
static int wake_affine(struct sched_domain *sd, struct task_struct *p, int sync)
{
s64 this_load, load;
int idx, this_cpu, prev_cpu;
unsigned long tl_per_task;
struct task_group *tg;
unsigned long weight;
int balanced;
idx = sd->wake_idx;
this_cpu = smp_processor_id();
prev_cpu = task_cpu(p);
load = source_load(prev_cpu, idx);
this_load = target_load(this_cpu, idx);
/*
* If sync wakeup then subtract the (maximum possible)
* effect of the currently running task from the load
* of the current CPU:
*/
if (sync) {
tg = task_group(current);
weight = current->se.load.weight;
this_load += effective_load(tg, this_cpu, -weight, -weight);
load += effective_load(tg, prev_cpu, 0, -weight);
}
tg = task_group(p);
weight = p->se.load.weight;
/*
* In low-load situations, where prev_cpu is idle and this_cpu is idle
* due to the sync cause above having dropped this_load to 0, we'll
* always have an imbalance, but there's really nothing you can do
* about that, so that's good too.
*
* Otherwise check if either cpus are near enough in load to allow this
* task to be woken on this_cpu.
*/
if (this_load > 0) {
s64 this_eff_load, prev_eff_load;
this_eff_load = 100;
this_eff_load *= power_of(prev_cpu);
this_eff_load *= this_load +
effective_load(tg, this_cpu, weight, weight);
prev_eff_load = 100 + (sd->imbalance_pct - 100) / 2;
prev_eff_load *= power_of(this_cpu);
prev_eff_load *= load + effective_load(tg, prev_cpu, 0, weight);
balanced = this_eff_load <= prev_eff_load;
} else
balanced = true;
/*
* If the currently running task will sleep within
* a reasonable amount of time then attract this newly
* woken task:
*/
if (sync && balanced)
return 1;
schedstat_inc(p, se.statistics.nr_wakeups_affine_attempts);
tl_per_task = cpu_avg_load_per_task(this_cpu);
if (balanced ||
(this_load <= load &&
this_load + target_load(prev_cpu, idx) <= tl_per_task)) {
/*
* This domain has SD_WAKE_AFFINE and
* p is cache cold in this domain, and
* there is no bad imbalance.
*/
schedstat_inc(sd, ttwu_move_affine);
schedstat_inc(p, se.statistics.nr_wakeups_affine);
return 1;
}
return 0;
}
/*
* find_idlest_group finds and returns the least busy CPU group within the
* domain.
*/
static struct sched_group *
find_idlest_group(struct sched_domain *sd, struct task_struct *p,
int this_cpu, int load_idx)
{
struct sched_group *idlest = NULL, *group = sd->groups;
unsigned long min_load = ULONG_MAX, this_load = 0;
int imbalance = 100 + (sd->imbalance_pct-100)/2;
do {
unsigned long load, avg_load;
int local_group;
int i;
/* Skip over this group if it has no CPUs allowed */
if (!cpumask_intersects(sched_group_cpus(group),
tsk_cpus_allowed(p)))
continue;
local_group = cpumask_test_cpu(this_cpu,
sched_group_cpus(group));
/* Tally up the load of all CPUs in the group */
avg_load = 0;
for_each_cpu(i, sched_group_cpus(group)) {
/* Bias balancing toward cpus of our domain */
if (local_group)
load = source_load(i, load_idx);
else
load = target_load(i, load_idx);
avg_load += load;
}
/* Adjust by relative CPU power of the group */
avg_load = (avg_load * SCHED_POWER_SCALE) / group->sgp->power;
if (local_group) {
this_load = avg_load;
} else if (avg_load < min_load) {
min_load = avg_load;
idlest = group;
}
} while (group = group->next, group != sd->groups);
if (!idlest || 100*this_load < imbalance*min_load)
return NULL;
return idlest;
}
/*
* find_idlest_cpu - find the idlest cpu among the cpus in group.
*/
static int
find_idlest_cpu(struct sched_group *group, struct task_struct *p, int this_cpu)
{
unsigned long load, min_load = ULONG_MAX;
int idlest = -1;
int i;
/* Traverse only the allowed CPUs */
for_each_cpu_and(i, sched_group_cpus(group), tsk_cpus_allowed(p)) {
load = weighted_cpuload(i);
if (load < min_load || (load == min_load && i == this_cpu)) {
min_load = load;
idlest = i;
}
}
return idlest;
}
/*
* Try and locate an idle CPU in the sched_domain.
*/
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
static int select_idle_sibling(struct task_struct *p, int target)
{
int cpu = smp_processor_id();
int prev_cpu = task_cpu(p);
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
struct sched_domain *sd;
struct sched_group *sg;
int i, smt = 0;
/*
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
* If the task is going to be woken-up on this cpu and if it is
* already idle, then it is the right target.
*/
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
if (target == cpu && idle_cpu(cpu))
return cpu;
/*
* If the task is going to be woken-up on the cpu where it previously
* ran and if it is currently idle, then it the right target.
*/
if (target == prev_cpu && idle_cpu(prev_cpu))
return prev_cpu;
/*
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
* Otherwise, iterate the domains and find an elegible idle cpu.
*/
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_lock();
again:
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
for_each_domain(target, sd) {
if (!smt && (sd->flags & SD_SHARE_CPUPOWER))
continue;
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
if (!(sd->flags & SD_SHARE_PKG_RESOURCES)) {
if (!smt) {
smt = 1;
goto again;
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
}
break;
}
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
sg = sd->groups;
do {
if (!cpumask_intersects(sched_group_cpus(sg),
tsk_cpus_allowed(p)))
goto next;
for_each_cpu(i, sched_group_cpus(sg)) {
if (!idle_cpu(i))
goto next;
}
target = cpumask_first_and(sched_group_cpus(sg),
tsk_cpus_allowed(p));
goto done;
next:
sg = sg->next;
} while (sg != sd->groups);
}
done:
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_unlock();
return target;
}
/*
* 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
* SD_BALANCE_EXEC.
*
* Balance, ie. select the least loaded group.
*
* Returns the target CPU number, or the same CPU if no balancing is needed.
*
* preempt must be disabled.
*/
static int
select_task_rq_fair(struct task_struct *p, int sd_flag, int wake_flags)
{
struct sched_domain *tmp, *affine_sd = NULL, *sd = NULL;
int cpu = smp_processor_id();
int prev_cpu = task_cpu(p);
int new_cpu = cpu;
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
int want_affine = 0;
int want_sd = 1;
int sync = wake_flags & WF_SYNC;
if (sd_flag & SD_BALANCE_WAKE) {
if (cpumask_test_cpu(cpu, tsk_cpus_allowed(p)))
want_affine = 1;
new_cpu = prev_cpu;
}
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_lock();
for_each_domain(cpu, tmp) {
if (!(tmp->flags & SD_LOAD_BALANCE))
continue;
/*
* If power savings logic is enabled for a domain, see if we
* are not overloaded, if so, don't balance wider.
*/
if (tmp->flags & (SD_POWERSAVINGS_BALANCE|SD_PREFER_LOCAL)) {
unsigned long power = 0;
unsigned long nr_running = 0;
unsigned long capacity;
int i;
for_each_cpu(i, sched_domain_span(tmp)) {
power += power_of(i);
nr_running += cpu_rq(i)->cfs.nr_running;
}
capacity = DIV_ROUND_CLOSEST(power, SCHED_POWER_SCALE);
if (tmp->flags & SD_POWERSAVINGS_BALANCE)
nr_running /= 2;
if (nr_running < capacity)
want_sd = 0;
}
/*
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
* If both cpu and prev_cpu are part of this domain,
* cpu is a valid SD_WAKE_AFFINE target.
*/
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
if (want_affine && (tmp->flags & SD_WAKE_AFFINE) &&
cpumask_test_cpu(prev_cpu, sched_domain_span(tmp))) {
affine_sd = tmp;
want_affine = 0;
}
if (!want_sd && !want_affine)
break;
if (!(tmp->flags & sd_flag))
continue;
if (want_sd)
sd = tmp;
}
if (affine_sd) {
sched: Fix select_idle_sibling() logic in select_task_rq_fair() Issues in the current select_idle_sibling() logic in select_task_rq_fair() in the context of a task wake-up: a) Once we select the idle sibling, we use that domain (spanning the cpu that the task is currently woken-up and the idle sibling that we found) in our wake_affine() decisions. This domain is completely different from the domain(we are supposed to use) that spans the cpu that the task currently woken-up and the cpu where the task previously ran. b) We do select_idle_sibling() check only for the cpu that the task is currently woken-up on. If select_task_rq_fair() selects the previously run cpu for waking the task, doing a select_idle_sibling() check for that cpu also helps and we don't do this currently. c) In the scenarios where the cpu that the task is woken-up is busy but with its HT siblings are idle, we are selecting the task be woken-up on the idle HT sibling instead of a core that it previously ran and currently completely idle. i.e., we are not taking decisions based on wake_affine() but directly selecting an idle sibling that can cause an imbalance at the SMT/MC level which will be later corrected by the periodic load balancer. Fix this by first going through the load imbalance calculations using wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu, then choose a possible idle sibling for waking up the task on. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-31 23:47:45 +00:00
if (cpu == prev_cpu || wake_affine(affine_sd, p, sync))
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
prev_cpu = cpu;
new_cpu = select_idle_sibling(p, prev_cpu);
goto unlock;
}
while (sd) {
int load_idx = sd->forkexec_idx;
struct sched_group *group;
int weight;
if (!(sd->flags & sd_flag)) {
sd = sd->child;
continue;
}
if (sd_flag & SD_BALANCE_WAKE)
load_idx = sd->wake_idx;
group = find_idlest_group(sd, p, cpu, load_idx);
if (!group) {
sd = sd->child;
continue;
}
new_cpu = find_idlest_cpu(group, p, cpu);
if (new_cpu == -1 || new_cpu == cpu) {
/* Now try balancing at a lower domain level of cpu */
sd = sd->child;
continue;
}
/* Now try balancing at a lower domain level of new_cpu */
cpu = new_cpu;
weight = sd->span_weight;
sd = NULL;
for_each_domain(cpu, tmp) {
if (weight <= tmp->span_weight)
break;
if (tmp->flags & sd_flag)
sd = tmp;
}
/* while loop will break here if sd == NULL */
}
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
unlock:
rcu_read_unlock();
return new_cpu;
}
#endif /* CONFIG_SMP */
static unsigned long
wakeup_gran(struct sched_entity *curr, struct sched_entity *se)
{
unsigned long gran = sysctl_sched_wakeup_granularity;
/*
* Since its curr running now, convert the gran from real-time
* to virtual-time in his units.
*
* By using 'se' instead of 'curr' we penalize light tasks, so
* they get preempted easier. That is, if 'se' < 'curr' then
* the resulting gran will be larger, therefore penalizing the
* lighter, if otoh 'se' > 'curr' then the resulting gran will
* be smaller, again penalizing the lighter task.
*
* This is especially important for buddies when the leftmost
* task is higher priority than the buddy.
*/
return calc_delta_fair(gran, se);
}
/*
* Should 'se' preempt 'curr'.
*
* |s1
* |s2
* |s3
* g
* |<--->|c
*
* w(c, s1) = -1
* w(c, s2) = 0
* w(c, s3) = 1
*
*/
static int
wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se)
{
s64 gran, vdiff = curr->vruntime - se->vruntime;
if (vdiff <= 0)
return -1;
gran = wakeup_gran(curr, se);
if (vdiff > gran)
return 1;
return 0;
}
static void set_last_buddy(struct sched_entity *se)
{
if (entity_is_task(se) && unlikely(task_of(se)->policy == SCHED_IDLE))
return;
for_each_sched_entity(se)
cfs_rq_of(se)->last = se;
}
static void set_next_buddy(struct sched_entity *se)
{
if (entity_is_task(se) && unlikely(task_of(se)->policy == SCHED_IDLE))
return;
for_each_sched_entity(se)
cfs_rq_of(se)->next = se;
}
static void set_skip_buddy(struct sched_entity *se)
{
for_each_sched_entity(se)
cfs_rq_of(se)->skip = se;
}
/*
* Preempt the current task with a newly woken task if needed:
*/
static void check_preempt_wakeup(struct rq *rq, struct task_struct *p, int wake_flags)
{
struct task_struct *curr = rq->curr;
struct sched_entity *se = &curr->se, *pse = &p->se;
struct cfs_rq *cfs_rq = task_cfs_rq(curr);
sched: Strengthen buddies and mitigate buddy induced latencies This patch restores the effectiveness of LAST_BUDDY in preventing pgsql+oltp from collapsing due to wakeup preemption. It also switches LAST_BUDDY to exclusively do what it does best, namely mitigate the effects of aggressive wakeup preemption, which improves vmark throughput markedly, and restores mysql+oltp scalability. Since buddies are about scalability, enable them beginning at the point where we begin expanding sched_latency, namely sched_nr_latency. Previously, buddies were cleared aggressively, which seriously reduced their effectiveness. Not clearing aggressively however, produces a small drop in mysql+oltp throughput immediately after peak, indicating that LAST_BUDDY is actually doing some harm. This is right at the point where X on the desktop in competition with another load wants low latency service. Ergo, do not enable until we need to scale. To mitigate latency induced by buddies, or by a task just missing wakeup preemption, check latency at tick time. Last hunk prevents buddies from stymieing BALANCE_NEWIDLE via CACHE_HOT_BUDDY. Supporting performance tests: tip = v2.6.32-rc5-1497-ga525b32 tipx = NO_GENTLE_FAIR_SLEEPERS NEXT_BUDDY granularity knobs = 31 knobs + 31 buddies tip+x = NO_GENTLE_FAIR_SLEEPERS granularity knobs = 31 knobs (Three run averages except where noted.) vmark: ------ tip 108466 messages per second tip+ 125307 messages per second tip+x 125335 messages per second tipx 117781 messages per second 2.6.31.3 122729 messages per second mysql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 9949.89 18690.20 34801.24 34460.04 32682.88 30765.97 28305.27 25059.64 19548.08 tip+ 10013.90 18526.84 34900.38 34420.14 33069.83 32083.40 30578.30 28010.71 25605.47 tipx 9698.71 18002.70 34477.56 33420.01 32634.30 31657.27 29932.67 26827.52 21487.18 2.6.31.3 8243.11 18784.20 34404.83 33148.38 31900.32 31161.90 29663.81 25995.94 18058.86 pgsql+oltp: ----------- clients 1 2 4 8 16 32 64 128 256 .......................................................................................... tip 13686.37 26609.25 51934.28 51347.81 49479.51 45312.65 36691.91 26851.57 24145.35 tip+ (1x) 13907.85 27135.87 52951.98 52514.04 51742.52 50705.43 49947.97 48374.19 46227.94 tip+x 13906.78 27065.81 52951.19 52542.59 52176.11 51815.94 50838.90 49439.46 46891.00 tipx 13742.46 26769.81 52351.99 51891.73 51320.79 50938.98 50248.65 48908.70 46553.84 2.6.31.3 13815.35 26906.46 52683.34 52061.31 51937.10 51376.80 50474.28 49394.47 47003.25 Signed-off-by: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-23 21:09:22 +00:00
int scale = cfs_rq->nr_running >= sched_nr_latency;
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
int next_buddy_marked = 0;
if (unlikely(se == pse))
return;
/*
* This is possible from callers such as pull_task(), in which we
* unconditionally check_prempt_curr() after an enqueue (which may have
* lead to a throttle). This both saves work and prevents false
* next-buddy nomination below.
*/
if (unlikely(throttled_hierarchy(cfs_rq_of(pse))))
return;
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
if (sched_feat(NEXT_BUDDY) && scale && !(wake_flags & WF_FORK)) {
set_next_buddy(pse);
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
next_buddy_marked = 1;
}
/*
* We can come here with TIF_NEED_RESCHED already set from new task
* wake up path.
*
* Note: this also catches the edge-case of curr being in a throttled
* group (e.g. via set_curr_task), since update_curr() (in the
* enqueue of curr) will have resulted in resched being set. This
* prevents us from potentially nominating it as a false LAST_BUDDY
* below.
*/
if (test_tsk_need_resched(curr))
return;
/* Idle tasks are by definition preempted by non-idle tasks. */
if (unlikely(curr->policy == SCHED_IDLE) &&
likely(p->policy != SCHED_IDLE))
goto preempt;
/*
* Batch and idle tasks do not preempt non-idle tasks (their preemption
* is driven by the tick):
*/
if (unlikely(p->policy != SCHED_NORMAL))
return;
find_matching_se(&se, &pse);
update_curr(cfs_rq_of(se));
BUG_ON(!pse);
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
if (wakeup_preempt_entity(se, pse) == 1) {
/*
* Bias pick_next to pick the sched entity that is
* triggering this preemption.
*/
if (!next_buddy_marked)
set_next_buddy(pse);
goto preempt;
sched: Next buddy hint on sleep and preempt path When a task in a taskgroup sleeps, pick_next_task starts all the way back at the root and picks the task/taskgroup with the min vruntime across all runnable tasks. But when there are many frequently sleeping tasks across different taskgroups, it makes better sense to stay with same taskgroup for its slice period (or until all tasks in the taskgroup sleeps) instead of switching cross taskgroup on each sleep after a short runtime. This helps specifically where taskgroups corresponds to a process with multiple threads. The change reduces the number of CR3 switches in this case. Example: Two taskgroups with 2 threads each which are running for 2ms and sleeping for 1ms. Looking at sched:sched_switch shows: BEFORE: taskgroup_1 threads [5004, 5005], taskgroup_2 threads [5016, 5017] cpu-soaker-5004 [003] 3683.391089 cpu-soaker-5016 [003] 3683.393106 cpu-soaker-5005 [003] 3683.395119 cpu-soaker-5017 [003] 3683.397130 cpu-soaker-5004 [003] 3683.399143 cpu-soaker-5016 [003] 3683.401155 cpu-soaker-5005 [003] 3683.403168 cpu-soaker-5017 [003] 3683.405170 AFTER: taskgroup_1 threads [21890, 21891], taskgroup_2 threads [21934, 21935] cpu-soaker-21890 [003] 865.895494 cpu-soaker-21935 [003] 865.897506 cpu-soaker-21934 [003] 865.899520 cpu-soaker-21935 [003] 865.901532 cpu-soaker-21934 [003] 865.903543 cpu-soaker-21935 [003] 865.905546 cpu-soaker-21891 [003] 865.907548 cpu-soaker-21890 [003] 865.909560 cpu-soaker-21891 [003] 865.911571 cpu-soaker-21890 [003] 865.913582 cpu-soaker-21891 [003] 865.915594 cpu-soaker-21934 [003] 865.917606 Similar problem is there when there are multiple taskgroups and say a task A preempts currently running task B of taskgroup_1. On schedule, pick_next_task can pick an unrelated task on taskgroup_2. Here it would be better to give some preference to task B on pick_next_task. A simple (may be extreme case) benchmark I tried was tbench with 2 tbench client processes with 2 threads each running on a single CPU. Avg throughput across 5 50 sec runs was: BEFORE: 105.84 MB/sec AFTER: 112.42 MB/sec Signed-off-by: Venkatesh Pallipadi <venki@google.com> Acked-by: Rik van Riel <riel@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1302802253-25760-1-git-send-email-venki@google.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-14 17:30:53 +00:00
}
return;
preempt:
resched_task(curr);
/*
* Only set the backward buddy when the current task is still
* on the rq. This can happen when a wakeup gets interleaved
* with schedule on the ->pre_schedule() or idle_balance()
* point, either of which can * drop the rq lock.
*
* Also, during early boot the idle thread is in the fair class,
* for obvious reasons its a bad idea to schedule back to it.
*/
if (unlikely(!se->on_rq || curr == rq->idle))
return;
if (sched_feat(LAST_BUDDY) && scale && entity_is_task(se))
set_last_buddy(se);
}
static struct task_struct *pick_next_task_fair(struct rq *rq)
{
struct task_struct *p;
struct cfs_rq *cfs_rq = &rq->cfs;
struct sched_entity *se;
if (!cfs_rq->nr_running)
return NULL;
do {
se = pick_next_entity(cfs_rq);
set_next_entity(cfs_rq, se);
cfs_rq = group_cfs_rq(se);
} while (cfs_rq);
p = task_of(se);
hrtick_start_fair(rq, p);
return p;
}
/*
* Account for a descheduled task:
*/
static void put_prev_task_fair(struct rq *rq, struct task_struct *prev)
{
struct sched_entity *se = &prev->se;
struct cfs_rq *cfs_rq;
for_each_sched_entity(se) {
cfs_rq = cfs_rq_of(se);
put_prev_entity(cfs_rq, se);
}
}
/*
* sched_yield() is very simple
*
* The magic of dealing with the ->skip buddy is in pick_next_entity.
*/
static void yield_task_fair(struct rq *rq)
{
struct task_struct *curr = rq->curr;
struct cfs_rq *cfs_rq = task_cfs_rq(curr);
struct sched_entity *se = &curr->se;
/*
* Are we the only task in the tree?
*/
if (unlikely(rq->nr_running == 1))
return;
clear_buddies(cfs_rq, se);
if (curr->policy != SCHED_BATCH) {
update_rq_clock(rq);
/*
* Update run-time statistics of the 'current'.
*/
update_curr(cfs_rq);
}
set_skip_buddy(se);
}
static bool yield_to_task_fair(struct rq *rq, struct task_struct *p, bool preempt)
{
struct sched_entity *se = &p->se;
/* throttled hierarchies are not runnable */
if (!se->on_rq || throttled_hierarchy(cfs_rq_of(se)))
return false;
/* Tell the scheduler that we'd really like pse to run next. */
set_next_buddy(se);
yield_task_fair(rq);
return true;
}
#ifdef CONFIG_SMP
/**************************************************
* Fair scheduling class load-balancing methods:
*/
/*
* pull_task - move a task from a remote runqueue to the local runqueue.
* Both runqueues must be locked.
*/
static void pull_task(struct rq *src_rq, struct task_struct *p,
struct rq *this_rq, int this_cpu)
{
deactivate_task(src_rq, p, 0);
set_task_cpu(p, this_cpu);
activate_task(this_rq, p, 0);
check_preempt_curr(this_rq, p, 0);
}
/*
* Is this task likely cache-hot:
*/
static int
task_hot(struct task_struct *p, u64 now, struct sched_domain *sd)
{
s64 delta;
if (p->sched_class != &fair_sched_class)
return 0;
if (unlikely(p->policy == SCHED_IDLE))
return 0;
/*
* Buddy candidates are cache hot:
*/
if (sched_feat(CACHE_HOT_BUDDY) && this_rq()->nr_running &&
(&p->se == cfs_rq_of(&p->se)->next ||
&p->se == cfs_rq_of(&p->se)->last))
return 1;
if (sysctl_sched_migration_cost == -1)
return 1;
if (sysctl_sched_migration_cost == 0)
return 0;
delta = now - p->se.exec_start;
return delta < (s64)sysctl_sched_migration_cost;
}
/*
* can_migrate_task - may task p from runqueue rq be migrated to this_cpu?
*/
static
int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu,
struct sched_domain *sd, enum cpu_idle_type idle,
int *all_pinned)
{
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, or
* 3) are cache-hot on their current CPU.
*/
if (!cpumask_test_cpu(this_cpu, tsk_cpus_allowed(p))) {
schedstat_inc(p, se.statistics.nr_failed_migrations_affine);
return 0;
}
*all_pinned = 0;
if (task_running(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, rq->clock_task, sd);
if (!tsk_cache_hot ||
sd->nr_balance_failed > sd->cache_nice_tries) {
#ifdef CONFIG_SCHEDSTATS
if (tsk_cache_hot) {
schedstat_inc(sd, lb_hot_gained[idle]);
schedstat_inc(p, se.statistics.nr_forced_migrations);
}
#endif
return 1;
}
if (tsk_cache_hot) {
schedstat_inc(p, se.statistics.nr_failed_migrations_hot);
return 0;
}
return 1;
}
/*
* move_one_task tries to move exactly one task from busiest to this_rq, as
* part of active balancing operations within "domain".
* Returns 1 if successful and 0 otherwise.
*
* Called with both runqueues locked.
*/
static int
move_one_task(struct rq *this_rq, int this_cpu, struct rq *busiest,
struct sched_domain *sd, enum cpu_idle_type idle)
{
struct task_struct *p, *n;
struct cfs_rq *cfs_rq;
int pinned = 0;
for_each_leaf_cfs_rq(busiest, cfs_rq) {
list_for_each_entry_safe(p, n, &cfs_rq->tasks, se.group_node) {
if (throttled_lb_pair(task_group(p),
busiest->cpu, this_cpu))
break;
if (!can_migrate_task(p, busiest, this_cpu,
sd, idle, &pinned))
continue;
pull_task(busiest, p, this_rq, this_cpu);
/*
* Right now, this is only the second place pull_task()
* is called, so we can safely collect pull_task()
* stats here rather than inside pull_task().
*/
schedstat_inc(sd, lb_gained[idle]);
return 1;
}
}
return 0;
}
static unsigned long
balance_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest,
unsigned long max_load_move, struct sched_domain *sd,
enum cpu_idle_type idle, int *all_pinned,
struct cfs_rq *busiest_cfs_rq)
{
sched: Fix erroneous all_pinned logic The scheduler load balancer has specific code to deal with cases of unbalanced system due to lots of unmovable tasks (for example because of hard CPU affinity). In those situation, it excludes the busiest CPU that has pinned tasks for load balance consideration such that it can perform second 2nd load balance pass on the rest of the system. This all works as designed if there is only one cgroup in the system. However, when we have multiple cgroups, this logic has false positives and triggers multiple load balance passes despite there are actually no pinned tasks at all. The reason it has false positives is that the all pinned logic is deep in the lowest function of can_migrate_task() and is too low level: load_balance_fair() iterates each task group and calls balance_tasks() to migrate target load. Along the way, balance_tasks() will also set a all_pinned variable. Given that task-groups are iterated, this all_pinned variable is essentially the status of last group in the scanning process. Task group can have number of reasons that no load being migrated, none due to cpu affinity. However, this status bit is being propagated back up to the higher level load_balance(), which incorrectly think that no tasks were moved. It kick off the all pinned logic and start multiple passes attempt to move load onto puller CPU. To fix this, move the all_pinned aggregation up at the iterator level. This ensures that the status is aggregated over all task-groups, not just last one in the list. Signed-off-by: Ken Chen <kenchen@google.com> Cc: stable@kernel.org Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/BANLkTi=ernzNawaR5tJZEsV_QVnfxqXmsQ@mail.gmail.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-08 19:20:16 +00:00
int loops = 0, pulled = 0;
long rem_load_move = max_load_move;
struct task_struct *p, *n;
if (max_load_move == 0)
goto out;
list_for_each_entry_safe(p, n, &busiest_cfs_rq->tasks, se.group_node) {
if (loops++ > sysctl_sched_nr_migrate)
break;
if ((p->se.load.weight >> 1) > rem_load_move ||
sched: Fix erroneous all_pinned logic The scheduler load balancer has specific code to deal with cases of unbalanced system due to lots of unmovable tasks (for example because of hard CPU affinity). In those situation, it excludes the busiest CPU that has pinned tasks for load balance consideration such that it can perform second 2nd load balance pass on the rest of the system. This all works as designed if there is only one cgroup in the system. However, when we have multiple cgroups, this logic has false positives and triggers multiple load balance passes despite there are actually no pinned tasks at all. The reason it has false positives is that the all pinned logic is deep in the lowest function of can_migrate_task() and is too low level: load_balance_fair() iterates each task group and calls balance_tasks() to migrate target load. Along the way, balance_tasks() will also set a all_pinned variable. Given that task-groups are iterated, this all_pinned variable is essentially the status of last group in the scanning process. Task group can have number of reasons that no load being migrated, none due to cpu affinity. However, this status bit is being propagated back up to the higher level load_balance(), which incorrectly think that no tasks were moved. It kick off the all pinned logic and start multiple passes attempt to move load onto puller CPU. To fix this, move the all_pinned aggregation up at the iterator level. This ensures that the status is aggregated over all task-groups, not just last one in the list. Signed-off-by: Ken Chen <kenchen@google.com> Cc: stable@kernel.org Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/BANLkTi=ernzNawaR5tJZEsV_QVnfxqXmsQ@mail.gmail.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-08 19:20:16 +00:00
!can_migrate_task(p, busiest, this_cpu, sd, idle,
all_pinned))
continue;
pull_task(busiest, p, this_rq, this_cpu);
pulled++;
rem_load_move -= p->se.load.weight;
#ifdef CONFIG_PREEMPT
/*
* NEWIDLE balancing is a source of latency, so preemptible
* kernels will stop after the first task is pulled to minimize
* the critical section.
*/
if (idle == CPU_NEWLY_IDLE)
break;
#endif
/*
* We only want to steal up to the prescribed amount of
* weighted load.
*/
if (rem_load_move <= 0)
break;
}
out:
/*
* Right now, this is one of only two places pull_task() is called,
* so we can safely collect pull_task() stats here rather than
* inside pull_task().
*/
schedstat_add(sd, lb_gained[idle], pulled);
return max_load_move - rem_load_move;
}
#ifdef CONFIG_FAIR_GROUP_SCHED
/*
* update tg->load_weight by folding this cpu's load_avg
*/
static int update_shares_cpu(struct task_group *tg, int cpu)
{
struct cfs_rq *cfs_rq;
unsigned long flags;
struct rq *rq;
if (!tg->se[cpu])
return 0;
rq = cpu_rq(cpu);
cfs_rq = tg->cfs_rq[cpu];
raw_spin_lock_irqsave(&rq->lock, flags);
update_rq_clock(rq);
update_cfs_load(cfs_rq, 1);
/*
* We need to update shares after updating tg->load_weight in
* order to adjust the weight of groups with long running tasks.
*/
update_cfs_shares(cfs_rq);
raw_spin_unlock_irqrestore(&rq->lock, flags);
return 0;
}
static void update_shares(int cpu)
{
struct cfs_rq *cfs_rq;
struct rq *rq = cpu_rq(cpu);
rcu_read_lock();
/*
* Iterates the task_group tree in a bottom up fashion, see
* list_add_leaf_cfs_rq() for details.
*/
for_each_leaf_cfs_rq(rq, cfs_rq) {
/* throttled entities do not contribute to load */
if (throttled_hierarchy(cfs_rq))
continue;
update_shares_cpu(cfs_rq->tg, cpu);
}
rcu_read_unlock();
}
/*
* Compute the cpu's hierarchical load factor for each task group.
* This needs to be done in a top-down fashion because the load of a child
* group is a fraction of its parents load.
*/
static int tg_load_down(struct task_group *tg, void *data)
{
unsigned long load;
long cpu = (long)data;
if (!tg->parent) {
load = cpu_rq(cpu)->load.weight;
} else {
load = tg->parent->cfs_rq[cpu]->h_load;
load *= tg->se[cpu]->load.weight;
load /= tg->parent->cfs_rq[cpu]->load.weight + 1;
}
tg->cfs_rq[cpu]->h_load = load;
return 0;
}
static void update_h_load(long cpu)
{
walk_tg_tree(tg_load_down, tg_nop, (void *)cpu);
}
static unsigned long
load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest,
unsigned long max_load_move,
struct sched_domain *sd, enum cpu_idle_type idle,
int *all_pinned)
{
long rem_load_move = max_load_move;
struct cfs_rq *busiest_cfs_rq;
rcu_read_lock();
update_h_load(cpu_of(busiest));
for_each_leaf_cfs_rq(busiest, busiest_cfs_rq) {
unsigned long busiest_h_load = busiest_cfs_rq->h_load;
unsigned long busiest_weight = busiest_cfs_rq->load.weight;
u64 rem_load, moved_load;
/*
* empty group or part of a throttled hierarchy
*/
if (!busiest_cfs_rq->task_weight ||
throttled_lb_pair(busiest_cfs_rq->tg, cpu_of(busiest), this_cpu))
continue;
rem_load = (u64)rem_load_move * busiest_weight;
rem_load = div_u64(rem_load, busiest_h_load + 1);
moved_load = balance_tasks(this_rq, this_cpu, busiest,
rem_load, sd, idle, all_pinned,
busiest_cfs_rq);
if (!moved_load)
continue;
moved_load *= busiest_h_load;
moved_load = div_u64(moved_load, busiest_weight + 1);
rem_load_move -= moved_load;
if (rem_load_move < 0)
break;
}
rcu_read_unlock();
return max_load_move - rem_load_move;
}
#else
static inline void update_shares(int cpu)
{
}
static unsigned long
load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest,
unsigned long max_load_move,
struct sched_domain *sd, enum cpu_idle_type idle,
int *all_pinned)
{
return balance_tasks(this_rq, this_cpu, busiest,
max_load_move, sd, idle, all_pinned,
&busiest->cfs);
}
#endif
/*
* move_tasks tries to move up to max_load_move weighted load from busiest to
* this_rq, as part of a balancing operation within domain "sd".
* Returns 1 if successful and 0 otherwise.
*
* Called with both runqueues locked.
*/
static int move_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest,
unsigned long max_load_move,
struct sched_domain *sd, enum cpu_idle_type idle,
int *all_pinned)
{
unsigned long total_load_moved = 0, load_moved;
do {
load_moved = load_balance_fair(this_rq, this_cpu, busiest,
max_load_move - total_load_moved,
sd, idle, all_pinned);
total_load_moved += load_moved;
#ifdef CONFIG_PREEMPT
/*
* NEWIDLE balancing is a source of latency, so preemptible
* kernels will stop after the first task is pulled to minimize
* the critical section.
*/
if (idle == CPU_NEWLY_IDLE && this_rq->nr_running)
break;
if (raw_spin_is_contended(&this_rq->lock) ||
raw_spin_is_contended(&busiest->lock))
break;
#endif
} while (load_moved && max_load_move > total_load_moved);
return total_load_moved > 0;
}
/********** Helpers for find_busiest_group ************************/
/*
* sd_lb_stats - Structure to store the statistics of a sched_domain
* during load balancing.
*/
struct sd_lb_stats {
struct sched_group *busiest; /* Busiest group in this sd */
struct sched_group *this; /* Local group in this sd */
unsigned long total_load; /* Total load of all groups in sd */
unsigned long total_pwr; /* Total power of all groups in sd */
unsigned long avg_load; /* Average load across all groups in sd */
/** Statistics of this group */
unsigned long this_load;
unsigned long this_load_per_task;
unsigned long this_nr_running;
unsigned long this_has_capacity;
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
unsigned int this_idle_cpus;
/* Statistics of the busiest group */
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
unsigned int busiest_idle_cpus;
unsigned long max_load;
unsigned long busiest_load_per_task;
unsigned long busiest_nr_running;
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
unsigned long busiest_group_capacity;
unsigned long busiest_has_capacity;
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
unsigned int busiest_group_weight;
int group_imb; /* Is there imbalance in this sd */
#if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
int power_savings_balance; /* Is powersave balance needed for this sd */
struct sched_group *group_min; /* Least loaded group in sd */
struct sched_group *group_leader; /* Group which relieves group_min */
unsigned long min_load_per_task; /* load_per_task in group_min */
unsigned long leader_nr_running; /* Nr running of group_leader */
unsigned long min_nr_running; /* Nr running of group_min */
#endif
};
/*
* sg_lb_stats - stats of a sched_group required for load_balancing
*/
struct sg_lb_stats {
unsigned long avg_load; /*Avg load across the CPUs of the group */
unsigned long group_load; /* Total load over the CPUs of the group */
unsigned long sum_nr_running; /* Nr tasks running in the group */
unsigned long sum_weighted_load; /* Weighted load of group's tasks */
unsigned long group_capacity;
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
unsigned long idle_cpus;
unsigned long group_weight;
int group_imb; /* Is there an imbalance in the group ? */
int group_has_capacity; /* Is there extra capacity in the group? */
};
/**
* get_sd_load_idx - Obtain the load index for a given sched domain.
* @sd: The sched_domain whose load_idx is to be obtained.
* @idle: The Idle status of the CPU for whose sd load_icx is obtained.
*/
static inline int get_sd_load_idx(struct sched_domain *sd,
enum cpu_idle_type idle)
{
int load_idx;
switch (idle) {
case CPU_NOT_IDLE:
load_idx = sd->busy_idx;
break;
case CPU_NEWLY_IDLE:
load_idx = sd->newidle_idx;
break;
default:
load_idx = sd->idle_idx;
break;
}
return load_idx;
}
#if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
/**
* init_sd_power_savings_stats - Initialize power savings statistics for
* the given sched_domain, during load balancing.
*
* @sd: Sched domain whose power-savings statistics are to be initialized.
* @sds: Variable containing the statistics for sd.
* @idle: Idle status of the CPU at which we're performing load-balancing.
*/
static inline void init_sd_power_savings_stats(struct sched_domain *sd,
struct sd_lb_stats *sds, enum cpu_idle_type idle)
{
/*
* Busy processors will not participate in power savings
* balance.
*/
if (idle == CPU_NOT_IDLE || !(sd->flags & SD_POWERSAVINGS_BALANCE))
sds->power_savings_balance = 0;
else {
sds->power_savings_balance = 1;
sds->min_nr_running = ULONG_MAX;
sds->leader_nr_running = 0;
}
}
/**
* update_sd_power_savings_stats - Update the power saving stats for a
* sched_domain while performing load balancing.
*
* @group: sched_group belonging to the sched_domain under consideration.
* @sds: Variable containing the statistics of the sched_domain
* @local_group: Does group contain the CPU for which we're performing
* load balancing ?
* @sgs: Variable containing the statistics of the group.
*/
static inline void update_sd_power_savings_stats(struct sched_group *group,
struct sd_lb_stats *sds, int local_group, struct sg_lb_stats *sgs)
{
if (!sds->power_savings_balance)
return;
/*
* If the local group is idle or completely loaded
* no need to do power savings balance at this domain
*/
if (local_group && (sds->this_nr_running >= sgs->group_capacity ||
!sds->this_nr_running))
sds->power_savings_balance = 0;
/*
* If a group is already running at full capacity or idle,
* don't include that group in power savings calculations
*/
if (!sds->power_savings_balance ||
sgs->sum_nr_running >= sgs->group_capacity ||
!sgs->sum_nr_running)
return;
/*
* Calculate the group which has the least non-idle load.
* This is the group from where we need to pick up the load
* for saving power
*/
if ((sgs->sum_nr_running < sds->min_nr_running) ||
(sgs->sum_nr_running == sds->min_nr_running &&
group_first_cpu(group) > group_first_cpu(sds->group_min))) {
sds->group_min = group;
sds->min_nr_running = sgs->sum_nr_running;
sds->min_load_per_task = sgs->sum_weighted_load /
sgs->sum_nr_running;
}
/*
* Calculate the group which is almost near its
* capacity but still has some space to pick up some load
* from other group and save more power
*/
if (sgs->sum_nr_running + 1 > sgs->group_capacity)
return;
if (sgs->sum_nr_running > sds->leader_nr_running ||
(sgs->sum_nr_running == sds->leader_nr_running &&
group_first_cpu(group) < group_first_cpu(sds->group_leader))) {
sds->group_leader = group;
sds->leader_nr_running = sgs->sum_nr_running;
}
}
/**
* check_power_save_busiest_group - see if there is potential for some power-savings balance
* @sds: Variable containing the statistics of the sched_domain
* under consideration.
* @this_cpu: Cpu at which we're currently performing load-balancing.
* @imbalance: Variable to store the imbalance.
*
* Description:
* Check if we have potential to perform some power-savings balance.
* If yes, set the busiest group to be the least loaded group in the
* sched_domain, so that it's CPUs can be put to idle.
*
* Returns 1 if there is potential to perform power-savings balance.
* Else returns 0.
*/
static inline int check_power_save_busiest_group(struct sd_lb_stats *sds,
int this_cpu, unsigned long *imbalance)
{
if (!sds->power_savings_balance)
return 0;
if (sds->this != sds->group_leader ||
sds->group_leader == sds->group_min)
return 0;
*imbalance = sds->min_load_per_task;
sds->busiest = sds->group_min;
return 1;
}
#else /* CONFIG_SCHED_MC || CONFIG_SCHED_SMT */
static inline void init_sd_power_savings_stats(struct sched_domain *sd,
struct sd_lb_stats *sds, enum cpu_idle_type idle)
{
return;
}
static inline void update_sd_power_savings_stats(struct sched_group *group,
struct sd_lb_stats *sds, int local_group, struct sg_lb_stats *sgs)
{
return;
}
static inline int check_power_save_busiest_group(struct sd_lb_stats *sds,
int this_cpu, unsigned long *imbalance)
{
return 0;
}
#endif /* CONFIG_SCHED_MC || CONFIG_SCHED_SMT */
unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu)
{
return SCHED_POWER_SCALE;
}
unsigned long __weak arch_scale_freq_power(struct sched_domain *sd, int cpu)
{
return default_scale_freq_power(sd, cpu);
}
unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu)
{
unsigned long weight = sd->span_weight;
unsigned long smt_gain = sd->smt_gain;
smt_gain /= weight;
return smt_gain;
}
unsigned long __weak arch_scale_smt_power(struct sched_domain *sd, int cpu)
{
return default_scale_smt_power(sd, cpu);
}
unsigned long scale_rt_power(int cpu)
{
struct rq *rq = cpu_rq(cpu);
u64 total, available;
total = sched_avg_period() + (rq->clock - rq->age_stamp);
if (unlikely(total < rq->rt_avg)) {
/* Ensures that power won't end up being negative */
available = 0;
} else {
available = total - rq->rt_avg;
}
if (unlikely((s64)total < SCHED_POWER_SCALE))
total = SCHED_POWER_SCALE;
total >>= SCHED_POWER_SHIFT;
return div_u64(available, total);
}
static void update_cpu_power(struct sched_domain *sd, int cpu)
{
unsigned long weight = sd->span_weight;
unsigned long power = SCHED_POWER_SCALE;
struct sched_group *sdg = sd->groups;
if ((sd->flags & SD_SHARE_CPUPOWER) && weight > 1) {
if (sched_feat(ARCH_POWER))
power *= arch_scale_smt_power(sd, cpu);
else
power *= default_scale_smt_power(sd, cpu);
power >>= SCHED_POWER_SHIFT;
}
sdg->sgp->power_orig = power;
if (sched_feat(ARCH_POWER))
power *= arch_scale_freq_power(sd, cpu);
else
power *= default_scale_freq_power(sd, cpu);
power >>= SCHED_POWER_SHIFT;
power *= scale_rt_power(cpu);
power >>= SCHED_POWER_SHIFT;
if (!power)
power = 1;
cpu_rq(cpu)->cpu_power = power;
sdg->sgp->power = power;
}
void update_group_power(struct sched_domain *sd, int cpu)
{
struct sched_domain *child = sd->child;
struct sched_group *group, *sdg = sd->groups;
unsigned long power;
if (!child) {
update_cpu_power(sd, cpu);
return;
}
power = 0;
group = child->groups;
do {
power += group->sgp->power;
group = group->next;
} while (group != child->groups);
sdg->sgp->power = power;
}
/*
* Try and fix up capacity for tiny siblings, this is needed when
* things like SD_ASYM_PACKING need f_b_g to select another sibling
* which on its own isn't powerful enough.
*
* See update_sd_pick_busiest() and check_asym_packing().
*/
static inline int
fix_small_capacity(struct sched_domain *sd, struct sched_group *group)
{
/*
* Only siblings can have significantly less than SCHED_POWER_SCALE
*/
if (!(sd->flags & SD_SHARE_CPUPOWER))
return 0;
/*
* If ~90% of the cpu_power is still there, we're good.
*/
if (group->sgp->power * 32 > group->sgp->power_orig * 29)
return 1;
return 0;
}
/**
* update_sg_lb_stats - Update sched_group's statistics for load balancing.
* @sd: The sched_domain whose statistics are to be updated.
* @group: sched_group whose statistics are to be updated.
* @this_cpu: Cpu for which load balance is currently performed.
* @idle: Idle status of this_cpu
* @load_idx: Load index of sched_domain of this_cpu for load calc.
* @local_group: Does group contain this_cpu.
* @cpus: Set of cpus considered for load balancing.
* @balance: Should we balance.
* @sgs: variable to hold the statistics for this group.
*/
static inline void update_sg_lb_stats(struct sched_domain *sd,
struct sched_group *group, int this_cpu,
enum cpu_idle_type idle, int load_idx,
int local_group, const struct cpumask *cpus,
int *balance, struct sg_lb_stats *sgs)
{
unsigned long load, max_cpu_load, min_cpu_load, max_nr_running;
int i;
unsigned int balance_cpu = -1, first_idle_cpu = 0;
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
unsigned long avg_load_per_task = 0;
if (local_group)
balance_cpu = group_first_cpu(group);
/* Tally up the load of all CPUs in the group */
max_cpu_load = 0;
min_cpu_load = ~0UL;
max_nr_running = 0;
for_each_cpu_and(i, sched_group_cpus(group), cpus) {
struct rq *rq = cpu_rq(i);
/* Bias balancing toward cpus of our domain */
if (local_group) {
if (idle_cpu(i) && !first_idle_cpu) {
first_idle_cpu = 1;
balance_cpu = i;
}
load = target_load(i, load_idx);
} else {
load = source_load(i, load_idx);
if (load > max_cpu_load) {
max_cpu_load = load;
max_nr_running = rq->nr_running;
}
if (min_cpu_load > load)
min_cpu_load = load;
}
sgs->group_load += load;
sgs->sum_nr_running += rq->nr_running;
sgs->sum_weighted_load += weighted_cpuload(i);
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
if (idle_cpu(i))
sgs->idle_cpus++;
}
/*
* First idle cpu or the first cpu(busiest) in this sched group
* is eligible for doing load balancing at this and above
* domains. In the newly idle case, we will allow all the cpu's
* to do the newly idle load balance.
*/
if (idle != CPU_NEWLY_IDLE && local_group) {
if (balance_cpu != this_cpu) {
*balance = 0;
return;
}
update_group_power(sd, this_cpu);
}
/* Adjust by relative CPU power of the group */
sgs->avg_load = (sgs->group_load*SCHED_POWER_SCALE) / group->sgp->power;
/*
* Consider the group unbalanced when the imbalance is larger
* than the average weight of a task.
*
* APZ: with cgroup the avg task weight can vary wildly and
* might not be a suitable number - should we keep a
* normalized nr_running number somewhere that negates
* the hierarchy?
*/
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
if (sgs->sum_nr_running)
avg_load_per_task = sgs->sum_weighted_load / sgs->sum_nr_running;
if ((max_cpu_load - min_cpu_load) >= avg_load_per_task && max_nr_running > 1)
sgs->group_imb = 1;
sgs->group_capacity = DIV_ROUND_CLOSEST(group->sgp->power,
SCHED_POWER_SCALE);
if (!sgs->group_capacity)
sgs->group_capacity = fix_small_capacity(sd, group);
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
sgs->group_weight = group->group_weight;
if (sgs->group_capacity > sgs->sum_nr_running)
sgs->group_has_capacity = 1;
}
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
/**
* update_sd_pick_busiest - return 1 on busiest group
* @sd: sched_domain whose statistics are to be checked
* @sds: sched_domain statistics
* @sg: sched_group candidate to be checked for being the busiest
* @sgs: sched_group statistics
* @this_cpu: the current cpu
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
*
* Determine if @sg is a busier group than the previously selected
* busiest group.
*/
static bool update_sd_pick_busiest(struct sched_domain *sd,
struct sd_lb_stats *sds,
struct sched_group *sg,
struct sg_lb_stats *sgs,
int this_cpu)
{
if (sgs->avg_load <= sds->max_load)
return false;
if (sgs->sum_nr_running > sgs->group_capacity)
return true;
if (sgs->group_imb)
return true;
/*
* ASYM_PACKING needs to move all the work to the lowest
* numbered CPUs in the group, therefore mark all groups
* higher than ourself as busy.
*/
if ((sd->flags & SD_ASYM_PACKING) && sgs->sum_nr_running &&
this_cpu < group_first_cpu(sg)) {
if (!sds->busiest)
return true;
if (group_first_cpu(sds->busiest) > group_first_cpu(sg))
return true;
}
return false;
}
/**
* update_sd_lb_stats - Update sched_domain's statistics for load balancing.
* @sd: sched_domain whose statistics are to be updated.
* @this_cpu: Cpu for which load balance is currently performed.
* @idle: Idle status of this_cpu
* @cpus: Set of cpus considered for load balancing.
* @balance: Should we balance.
* @sds: variable to hold the statistics for this sched_domain.
*/
static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu,
enum cpu_idle_type idle, const struct cpumask *cpus,
int *balance, struct sd_lb_stats *sds)
{
struct sched_domain *child = sd->child;
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
struct sched_group *sg = sd->groups;
struct sg_lb_stats sgs;
int load_idx, prefer_sibling = 0;
if (child && child->flags & SD_PREFER_SIBLING)
prefer_sibling = 1;
init_sd_power_savings_stats(sd, sds, idle);
load_idx = get_sd_load_idx(sd, idle);
do {
int local_group;
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
local_group = cpumask_test_cpu(this_cpu, sched_group_cpus(sg));
memset(&sgs, 0, sizeof(sgs));
update_sg_lb_stats(sd, sg, this_cpu, idle, load_idx,
local_group, cpus, balance, &sgs);
if (local_group && !(*balance))
return;
sds->total_load += sgs.group_load;
sds->total_pwr += sg->sgp->power;
/*
* In case the child domain prefers tasks go to siblings
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
* first, lower the sg capacity to one so that we'll try
sched: Drop group_capacity to 1 only if local group has extra capacity When SD_PREFER_SIBLING is set on a sched domain, drop group_capacity to 1 only if the local group has extra capacity. The extra check prevents the case where you always pull from the heaviest group when it is already under-utilized (possible with a large weight task outweighs the tasks on the system). For example, consider a 16-cpu quad-core quad-socket machine with MC and NUMA scheduling domains. Let's say we spawn 15 nice0 tasks and one nice-15 task, and each task is running on one core. In this case, we observe the following events when balancing at the NUMA domain: - find_busiest_group() will always pick the sched group containing the niced task to be the busiest group. - find_busiest_queue() will then always pick one of the cpus running the nice0 task (never picks the cpu with the nice -15 task since weighted_cpuload > imbalance). - The load balancer fails to migrate the task since it is the running task and increments sd->nr_balance_failed. - It repeats the above steps a few more times until sd->nr_balance_failed > 5, at which point it kicks off the active load balancer, wakes up the migration thread and kicks the nice 0 task off the cpu. The load balancer doesn't stop until we kick out all nice 0 tasks from the sched group, leaving you with 3 idle cpus and one cpu running the nice -15 task. When balancing at the NUMA domain, we drop sgs.group_capacity to 1 if the child domain (in this case MC) has SD_PREFER_SIBLING set. Subsequent load checks are not relevant because the niced task has a very large weight. In this patch, we add an extra condition to the "if(prefer_sibling)" check in update_sd_lb_stats(). We drop the capacity of a group only if the local group has extra capacity, ie. nr_running < group_capacity. This patch preserves the original intent of the prefer_siblings check (to spread tasks across the system in low utilization scenarios) and fixes the case above. It helps in the following ways: - In low utilization cases (where nr_tasks << nr_cpus), we still drop group_capacity down to 1 if we prefer siblings. - On very busy systems (where nr_tasks >> nr_cpus), sgs.nr_running will most likely be > sgs.group_capacity. - When balancing large weight tasks, if the local group does not have extra capacity, we do not pick the group with the niced task as the busiest group. This prevents failed balances, active migration and the under-utilization described above. Signed-off-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1287173550-30365-5-git-send-email-ncrao@google.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-10-15 20:12:30 +00:00
* and move all the excess tasks away. We lower the capacity
* of a group only if the local group has the capacity to fit
* these excess tasks, i.e. nr_running < group_capacity. The
* extra check prevents the case where you always pull from the
* heaviest group when it is already under-utilized (possible
* with a large weight task outweighs the tasks on the system).
*/
sched: Drop group_capacity to 1 only if local group has extra capacity When SD_PREFER_SIBLING is set on a sched domain, drop group_capacity to 1 only if the local group has extra capacity. The extra check prevents the case where you always pull from the heaviest group when it is already under-utilized (possible with a large weight task outweighs the tasks on the system). For example, consider a 16-cpu quad-core quad-socket machine with MC and NUMA scheduling domains. Let's say we spawn 15 nice0 tasks and one nice-15 task, and each task is running on one core. In this case, we observe the following events when balancing at the NUMA domain: - find_busiest_group() will always pick the sched group containing the niced task to be the busiest group. - find_busiest_queue() will then always pick one of the cpus running the nice0 task (never picks the cpu with the nice -15 task since weighted_cpuload > imbalance). - The load balancer fails to migrate the task since it is the running task and increments sd->nr_balance_failed. - It repeats the above steps a few more times until sd->nr_balance_failed > 5, at which point it kicks off the active load balancer, wakes up the migration thread and kicks the nice 0 task off the cpu. The load balancer doesn't stop until we kick out all nice 0 tasks from the sched group, leaving you with 3 idle cpus and one cpu running the nice -15 task. When balancing at the NUMA domain, we drop sgs.group_capacity to 1 if the child domain (in this case MC) has SD_PREFER_SIBLING set. Subsequent load checks are not relevant because the niced task has a very large weight. In this patch, we add an extra condition to the "if(prefer_sibling)" check in update_sd_lb_stats(). We drop the capacity of a group only if the local group has extra capacity, ie. nr_running < group_capacity. This patch preserves the original intent of the prefer_siblings check (to spread tasks across the system in low utilization scenarios) and fixes the case above. It helps in the following ways: - In low utilization cases (where nr_tasks << nr_cpus), we still drop group_capacity down to 1 if we prefer siblings. - On very busy systems (where nr_tasks >> nr_cpus), sgs.nr_running will most likely be > sgs.group_capacity. - When balancing large weight tasks, if the local group does not have extra capacity, we do not pick the group with the niced task as the busiest group. This prevents failed balances, active migration and the under-utilization described above. Signed-off-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1287173550-30365-5-git-send-email-ncrao@google.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-10-15 20:12:30 +00:00
if (prefer_sibling && !local_group && sds->this_has_capacity)
sgs.group_capacity = min(sgs.group_capacity, 1UL);
if (local_group) {
sds->this_load = sgs.avg_load;
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
sds->this = sg;
sds->this_nr_running = sgs.sum_nr_running;
sds->this_load_per_task = sgs.sum_weighted_load;
sds->this_has_capacity = sgs.group_has_capacity;
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
sds->this_idle_cpus = sgs.idle_cpus;
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
} else if (update_sd_pick_busiest(sd, sds, sg, &sgs, this_cpu)) {
sds->max_load = sgs.avg_load;
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
sds->busiest = sg;
sds->busiest_nr_running = sgs.sum_nr_running;
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
sds->busiest_idle_cpus = sgs.idle_cpus;
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
sds->busiest_group_capacity = sgs.group_capacity;
sds->busiest_load_per_task = sgs.sum_weighted_load;
sds->busiest_has_capacity = sgs.group_has_capacity;
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
sds->busiest_group_weight = sgs.group_weight;
sds->group_imb = sgs.group_imb;
}
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
update_sd_power_savings_stats(sg, sds, local_group, &sgs);
sg = sg->next;
} while (sg != sd->groups);
}
/**
* check_asym_packing - Check to see if the group is packed into the
* sched doman.
*
* This is primarily intended to used at the sibling level. Some
* cores like POWER7 prefer to use lower numbered SMT threads. In the
* case of POWER7, it can move to lower SMT modes only when higher
* threads are idle. When in lower SMT modes, the threads will
* perform better since they share less core resources. Hence when we
* have idle threads, we want them to be the higher ones.
*
* This packing function is run on idle threads. It checks to see if
* the busiest CPU in this domain (core in the P7 case) has a higher
* CPU number than the packing function is being run on. Here we are
* assuming lower CPU number will be equivalent to lower a SMT thread
* number.
*
* Returns 1 when packing is required and a task should be moved to
* this CPU. The amount of the imbalance is returned in *imbalance.
*
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
* @sd: The sched_domain whose packing is to be checked.
* @sds: Statistics of the sched_domain which is to be packed
* @this_cpu: The cpu at whose sched_domain we're performing load-balance.
* @imbalance: returns amount of imbalanced due to packing.
*/
static int check_asym_packing(struct sched_domain *sd,
struct sd_lb_stats *sds,
int this_cpu, unsigned long *imbalance)
{
int busiest_cpu;
if (!(sd->flags & SD_ASYM_PACKING))
return 0;
if (!sds->busiest)
return 0;
busiest_cpu = group_first_cpu(sds->busiest);
if (this_cpu > busiest_cpu)
return 0;
*imbalance = DIV_ROUND_CLOSEST(sds->max_load * sds->busiest->sgp->power,
SCHED_POWER_SCALE);
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
return 1;
}
/**
* fix_small_imbalance - Calculate the minor imbalance that exists
* amongst the groups of a sched_domain, during
* load balancing.
* @sds: Statistics of the sched_domain whose imbalance is to be calculated.
* @this_cpu: The cpu at whose sched_domain we're performing load-balance.
* @imbalance: Variable to store the imbalance.
*/
static inline void fix_small_imbalance(struct sd_lb_stats *sds,
int this_cpu, unsigned long *imbalance)
{
unsigned long tmp, pwr_now = 0, pwr_move = 0;
unsigned int imbn = 2;
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
unsigned long scaled_busy_load_per_task;
if (sds->this_nr_running) {
sds->this_load_per_task /= sds->this_nr_running;
if (sds->busiest_load_per_task >
sds->this_load_per_task)
imbn = 1;
} else
sds->this_load_per_task =
cpu_avg_load_per_task(this_cpu);
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
scaled_busy_load_per_task = sds->busiest_load_per_task
* SCHED_POWER_SCALE;
scaled_busy_load_per_task /= sds->busiest->sgp->power;
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
if (sds->max_load - sds->this_load + scaled_busy_load_per_task >=
(scaled_busy_load_per_task * imbn)) {
*imbalance = sds->busiest_load_per_task;
return;
}
/*
* OK, we don't have enough imbalance to justify moving tasks,
* however we may be able to increase total CPU power used by
* moving them.
*/
pwr_now += sds->busiest->sgp->power *
min(sds->busiest_load_per_task, sds->max_load);
pwr_now += sds->this->sgp->power *
min(sds->this_load_per_task, sds->this_load);
pwr_now /= SCHED_POWER_SCALE;
/* Amount of load we'd subtract */
tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) /
sds->busiest->sgp->power;
if (sds->max_load > tmp)
pwr_move += sds->busiest->sgp->power *
min(sds->busiest_load_per_task, sds->max_load - tmp);
/* Amount of load we'd add */
if (sds->max_load * sds->busiest->sgp->power <
sds->busiest_load_per_task * SCHED_POWER_SCALE)
tmp = (sds->max_load * sds->busiest->sgp->power) /
sds->this->sgp->power;
else
tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) /
sds->this->sgp->power;
pwr_move += sds->this->sgp->power *
min(sds->this_load_per_task, sds->this_load + tmp);
pwr_move /= SCHED_POWER_SCALE;
/* Move if we gain throughput */
if (pwr_move > pwr_now)
*imbalance = sds->busiest_load_per_task;
}
/**
* calculate_imbalance - Calculate the amount of imbalance present within the
* groups of a given sched_domain during load balance.
* @sds: statistics of the sched_domain whose imbalance is to be calculated.
* @this_cpu: Cpu for which currently load balance is being performed.
* @imbalance: The variable to store the imbalance.
*/
static inline void calculate_imbalance(struct sd_lb_stats *sds, int this_cpu,
unsigned long *imbalance)
{
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
unsigned long max_pull, load_above_capacity = ~0UL;
sds->busiest_load_per_task /= sds->busiest_nr_running;
if (sds->group_imb) {
sds->busiest_load_per_task =
min(sds->busiest_load_per_task, sds->avg_load);
}
/*
* In the presence of smp nice balancing, certain scenarios can have
* max load less than avg load(as we skip the groups at or below
* its cpu_power, while calculating max_load..)
*/
if (sds->max_load < sds->avg_load) {
*imbalance = 0;
return fix_small_imbalance(sds, this_cpu, imbalance);
}
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
if (!sds->group_imb) {
/*
* Don't want to pull so many tasks that a group would go idle.
*/
load_above_capacity = (sds->busiest_nr_running -
sds->busiest_group_capacity);
load_above_capacity *= (SCHED_LOAD_SCALE * SCHED_POWER_SCALE);
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
load_above_capacity /= sds->busiest->sgp->power;
sched: Fix SCHED_MC regression caused by change in sched cpu_power On platforms like dual socket quad-core platform, the scheduler load balancer is not detecting the load imbalances in certain scenarios. This is leading to scenarios like where one socket is completely busy (with all the 4 cores running with 4 tasks) and leaving another socket completely idle. This causes performance issues as those 4 tasks share the memory controller, last-level cache bandwidth etc. Also we won't be taking advantage of turbo-mode as much as we would like, etc. Some of the comparisons in the scheduler load balancing code are comparing the "weighted cpu load that is scaled wrt sched_group's cpu_power" with the "weighted average load per task that is not scaled wrt sched_group's cpu_power". While this has probably been broken for a longer time (for multi socket numa nodes etc), the problem got aggrevated via this recent change: | | commit f93e65c186ab3c05ce2068733ca10e34fd00125e | Author: Peter Zijlstra <a.p.zijlstra@chello.nl> | Date: Tue Sep 1 10:34:32 2009 +0200 | | sched: Restore __cpu_power to a straight sum of power | Also with this change, the sched group cpu power alone no longer reflects the group capacity that is needed to implement MC, MT performance (default) and power-savings (user-selectable) policies. We need to use the computed group capacity (sgs.group_capacity, that is computed using the SD_PREFER_SIBLING logic in update_sd_lb_stats()) to find out if the group with the max load is above its capacity and how much load to move etc. Reported-by: Ma Ling <ling.ma@intel.com> Initial-Analysis-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> [ -v2: build fix ] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: <stable@kernel.org> # [2.6.32.x, 2.6.33.x] LKML-Reference: <1266970432.11588.22.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-24 00:13:52 +00:00
}
/*
* We're trying to get all the cpus to the average_load, so we don't
* want to push ourselves above the average load, nor do we wish to
* reduce the max loaded cpu below the average load. At the same time,
* we also don't want to reduce the group load below the group capacity
* (so that we can implement power-savings policies etc). Thus we look
* for the minimum possible imbalance.
* Be careful of negative numbers as they'll appear as very large values
* with unsigned longs.
*/
max_pull = min(sds->max_load - sds->avg_load, load_above_capacity);
/* How much load to actually move to equalise the imbalance */
*imbalance = min(max_pull * sds->busiest->sgp->power,
(sds->avg_load - sds->this_load) * sds->this->sgp->power)
/ SCHED_POWER_SCALE;
/*
* if *imbalance is less than the average load per runnable task
* there is no guarantee that any tasks will be moved so we'll have
* a think about bumping its value to force at least one task to be
* moved
*/
if (*imbalance < sds->busiest_load_per_task)
return fix_small_imbalance(sds, this_cpu, imbalance);
}
/******* find_busiest_group() helpers end here *********************/
/**
* find_busiest_group - Returns the busiest group within the sched_domain
* if there is an imbalance. If there isn't an imbalance, and
* the user has opted for power-savings, it returns a group whose
* CPUs can be put to idle by rebalancing those tasks elsewhere, if
* such a group exists.
*
* Also calculates the amount of weighted load which should be moved
* to restore balance.
*
* @sd: The sched_domain whose busiest group is to be returned.
* @this_cpu: The cpu for which load balancing is currently being performed.
* @imbalance: Variable which stores amount of weighted load which should
* be moved to restore balance/put a group to idle.
* @idle: The idle status of this_cpu.
* @cpus: The set of CPUs under consideration for load-balancing.
* @balance: Pointer to a variable indicating if this_cpu
* is the appropriate cpu to perform load balancing at this_level.
*
* Returns: - the busiest group if imbalance exists.
* - If no imbalance and user has opted for power-savings balance,
* return the least loaded group whose CPUs can be
* put to idle by rebalancing its tasks onto our group.
*/
static struct sched_group *
find_busiest_group(struct sched_domain *sd, int this_cpu,
unsigned long *imbalance, enum cpu_idle_type idle,
const struct cpumask *cpus, int *balance)
{
struct sd_lb_stats sds;
memset(&sds, 0, sizeof(sds));
/*
* Compute the various statistics relavent for load balancing at
* this level.
*/
update_sd_lb_stats(sd, this_cpu, idle, cpus, balance, &sds);
/*
* this_cpu is not the appropriate cpu to perform load balancing at
* this level.
*/
if (!(*balance))
goto ret;
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
if ((idle == CPU_IDLE || idle == CPU_NEWLY_IDLE) &&
check_asym_packing(sd, &sds, this_cpu, imbalance))
return sds.busiest;
/* There is no busy sibling group to pull tasks from */
if (!sds.busiest || sds.busiest_nr_running == 0)
goto out_balanced;
sds.avg_load = (SCHED_POWER_SCALE * sds.total_load) / sds.total_pwr;
/*
* If the busiest group is imbalanced the below checks don't
* work because they assumes all things are equal, which typically
* isn't true due to cpus_allowed constraints and the like.
*/
if (sds.group_imb)
goto force_balance;
/* SD_BALANCE_NEWIDLE trumps SMP nice when underutilized */
if (idle == CPU_NEWLY_IDLE && sds.this_has_capacity &&
!sds.busiest_has_capacity)
goto force_balance;
/*
* If the local group is more busy than the selected busiest group
* don't try and pull any tasks.
*/
if (sds.this_load >= sds.max_load)
goto out_balanced;
/*
* Don't pull any tasks if this group is already above the domain
* average load.
*/
if (sds.this_load >= sds.avg_load)
goto out_balanced;
if (idle == CPU_IDLE) {
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
/*
* This cpu is idle. If the busiest group load doesn't
* have more tasks than the number of available cpu's and
* there is no imbalance between this and busiest group
* wrt to idle cpu's, it is balanced.
*/
if ((sds.this_idle_cpus <= sds.busiest_idle_cpus + 1) &&
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
sds.busiest_nr_running <= sds.busiest_group_weight)
goto out_balanced;
} else {
/*
* In the CPU_NEWLY_IDLE, CPU_NOT_IDLE cases, use
* imbalance_pct to be conservative.
*/
if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load)
goto out_balanced;
sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao <ncrao@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-17 22:02:32 +00:00
}
force_balance:
/* Looks like there is an imbalance. Compute it */
calculate_imbalance(&sds, this_cpu, imbalance);
return sds.busiest;
out_balanced:
/*
* There is no obvious imbalance. But check if we can do some balancing
* to save power.
*/
if (check_power_save_busiest_group(&sds, this_cpu, imbalance))
return sds.busiest;
ret:
*imbalance = 0;
return NULL;
}
/*
* find_busiest_queue - find the busiest runqueue among the cpus in group.
*/
static struct rq *
find_busiest_queue(struct sched_domain *sd, struct sched_group *group,
enum cpu_idle_type idle, unsigned long imbalance,
const struct cpumask *cpus)
{
struct rq *busiest = NULL, *rq;
unsigned long max_load = 0;
int i;
for_each_cpu(i, sched_group_cpus(group)) {
unsigned long power = power_of(i);
unsigned long capacity = DIV_ROUND_CLOSEST(power,
SCHED_POWER_SCALE);
unsigned long wl;
if (!capacity)
capacity = fix_small_capacity(sd, group);
if (!cpumask_test_cpu(i, cpus))
continue;
rq = cpu_rq(i);
wl = weighted_cpuload(i);
/*
* When comparing with imbalance, use weighted_cpuload()
* which is not scaled with the cpu power.
*/
if (capacity && rq->nr_running == 1 && wl > imbalance)
continue;
/*
* For the load comparisons with the other cpu's, consider
* the weighted_cpuload() scaled with the cpu power, so that
* the load can be moved away from the cpu that is potentially
* running at a lower capacity.
*/
wl = (wl * SCHED_POWER_SCALE) / power;
if (wl > max_load) {
max_load = wl;
busiest = rq;
}
}
return busiest;
}
/*
* Max backoff if we encounter pinned tasks. Pretty arbitrary value, but
* so long as it is large enough.
*/
#define MAX_PINNED_INTERVAL 512
/* Working cpumask for load_balance and load_balance_newidle. */
DEFINE_PER_CPU(cpumask_var_t, load_balance_tmpmask);
static int need_active_balance(struct sched_domain *sd, int idle,
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
int busiest_cpu, int this_cpu)
{
if (idle == CPU_NEWLY_IDLE) {
sched: Add asymmetric group packing option for sibling domain Check to see if the group is packed in a sched doman. This is primarily intended to used at the sibling level. Some cores like POWER7 prefer to use lower numbered SMT threads. In the case of POWER7, it can move to lower SMT modes only when higher threads are idle. When in lower SMT modes, the threads will perform better since they share less core resources. Hence when we have idle threads, we want them to be the higher ones. This adds a hook into f_b_g() called check_asym_packing() to check the packing. This packing function is run on idle threads. It checks to see if the busiest CPU in this domain (core in the P7 case) has a higher CPU number than what where the packing function is being run on. If it is, calculate the imbalance and return the higher busier thread as the busiest group to f_b_g(). Here we are assuming a lower CPU number will be equivalent to a lower SMT thread number. It also creates a new SD_ASYM_PACKING flag to enable this feature at any scheduler domain level. It also creates an arch hook to enable this feature at the sibling level. The default function doesn't enable this feature. Based heavily on patch from Peter Zijlstra. Fixes from Srivatsa Vaddagiri. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <20100608045702.2936CCC897@localhost.localdomain> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-06-08 04:57:02 +00:00
/*
* ASYM_PACKING needs to force migrate tasks from busy but
* higher numbered CPUs in order to pack all tasks in the
* lowest numbered CPUs.
*/
if ((sd->flags & SD_ASYM_PACKING) && busiest_cpu > this_cpu)
return 1;
/*
* The only task running in a non-idle cpu can be moved to this
* cpu in an attempt to completely freeup the other CPU
* package.
*
* The package power saving logic comes from
* find_busiest_group(). If there are no imbalance, then
* f_b_g() will return NULL. However when sched_mc={1,2} then
* f_b_g() will select a group from which a running task may be
* pulled to this cpu in order to make the other package idle.
* If there is no opportunity to make a package idle and if
* there are no imbalance, then f_b_g() will return NULL and no
* action will be taken in load_balance_newidle().
*
* Under normal task pull operation due to imbalance, there
* will be more than one task in the source run queue and
* move_tasks() will succeed. ld_moved will be true and this
* active balance code will not be triggered.
*/
if (sched_mc_power_savings < POWERSAVINGS_BALANCE_WAKEUP)
return 0;
}
return unlikely(sd->nr_balance_failed > sd->cache_nice_tries+2);
}
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
static int active_load_balance_cpu_stop(void *data);
/*
* Check this_cpu to ensure it is balanced within domain. Attempt to move
* tasks if there is an imbalance.
*/
static int load_balance(int this_cpu, struct rq *this_rq,
struct sched_domain *sd, enum cpu_idle_type idle,
int *balance)
{
int ld_moved, all_pinned = 0, active_balance = 0;
struct sched_group *group;
unsigned long imbalance;
struct rq *busiest;
unsigned long flags;
struct cpumask *cpus = __get_cpu_var(load_balance_tmpmask);
cpumask_copy(cpus, cpu_active_mask);
schedstat_inc(sd, lb_count[idle]);
redo:
group = find_busiest_group(sd, this_cpu, &imbalance, idle,
cpus, balance);
if (*balance == 0)
goto out_balanced;
if (!group) {
schedstat_inc(sd, lb_nobusyg[idle]);
goto out_balanced;
}
busiest = find_busiest_queue(sd, group, idle, imbalance, cpus);
if (!busiest) {
schedstat_inc(sd, lb_nobusyq[idle]);
goto out_balanced;
}
BUG_ON(busiest == this_rq);
schedstat_add(sd, lb_imbalance[idle], imbalance);
ld_moved = 0;
if (busiest->nr_running > 1) {
/*
* Attempt to move tasks. If find_busiest_group has found
* an imbalance but busiest->nr_running <= 1, the group is
* still unbalanced. ld_moved simply stays zero, so it is
* correctly treated as an imbalance.
*/
sched: Fix erroneous all_pinned logic The scheduler load balancer has specific code to deal with cases of unbalanced system due to lots of unmovable tasks (for example because of hard CPU affinity). In those situation, it excludes the busiest CPU that has pinned tasks for load balance consideration such that it can perform second 2nd load balance pass on the rest of the system. This all works as designed if there is only one cgroup in the system. However, when we have multiple cgroups, this logic has false positives and triggers multiple load balance passes despite there are actually no pinned tasks at all. The reason it has false positives is that the all pinned logic is deep in the lowest function of can_migrate_task() and is too low level: load_balance_fair() iterates each task group and calls balance_tasks() to migrate target load. Along the way, balance_tasks() will also set a all_pinned variable. Given that task-groups are iterated, this all_pinned variable is essentially the status of last group in the scanning process. Task group can have number of reasons that no load being migrated, none due to cpu affinity. However, this status bit is being propagated back up to the higher level load_balance(), which incorrectly think that no tasks were moved. It kick off the all pinned logic and start multiple passes attempt to move load onto puller CPU. To fix this, move the all_pinned aggregation up at the iterator level. This ensures that the status is aggregated over all task-groups, not just last one in the list. Signed-off-by: Ken Chen <kenchen@google.com> Cc: stable@kernel.org Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/BANLkTi=ernzNawaR5tJZEsV_QVnfxqXmsQ@mail.gmail.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-08 19:20:16 +00:00
all_pinned = 1;
local_irq_save(flags);
double_rq_lock(this_rq, busiest);
ld_moved = move_tasks(this_rq, this_cpu, busiest,
imbalance, sd, idle, &all_pinned);
double_rq_unlock(this_rq, busiest);
local_irq_restore(flags);
/*
* some other cpu did the load balance for us.
*/
if (ld_moved && this_cpu != smp_processor_id())
resched_cpu(this_cpu);
/* All tasks on this runqueue were pinned by CPU affinity */
if (unlikely(all_pinned)) {
cpumask_clear_cpu(cpu_of(busiest), cpus);
if (!cpumask_empty(cpus))
goto redo;
goto out_balanced;
}
}
if (!ld_moved) {
schedstat_inc(sd, lb_failed[idle]);
sched: Increment cache_nice_tries only on periodic lb scheduler uses cache_nice_tries as an indicator to do cache_hot and active load balance, when normal load balance fails. Currently, this value is changed on any failed load balance attempt. That ends up being not so nice to workloads that enter/exit idle often, as they do more frequent new_idle balance and that pretty soon results in cache hot tasks being pulled in. Making the cache_nice_tries ignore failed new_idle balance seems to make better sense. With that only the failed load balance in periodic load balance gets accounted and the rate of accumulation of cache_nice_tries will not depend on idle entry/exit (short running sleep-wakeup kind of tasks). This reduces movement of cache_hot tasks. schedstat diff (after-before) excerpt from a workload that has frequent and short wakeup-idle pattern (:2 in cpu col below refers to NEWIDLE idx) This snapshot was across ~400 seconds. Without this change: domainstats: domain0 cpu cnt bln fld imb gain hgain nobusyq nobusyg 0:2 306487 219575 73167 110069413 44583 19070 1172 218403 1:2 292139 194853 81421 120893383 50745 21902 1259 193594 2:2 283166 174607 91359 129699642 54931 23688 1287 173320 3:2 273998 161788 93991 132757146 57122 24351 1366 160422 4:2 289851 215692 62190 83398383 36377 13680 851 214841 5:2 316312 222146 77605 117582154 49948 20281 988 221158 6:2 297172 195596 83623 122133390 52801 21301 929 194667 7:2 283391 178078 86378 126622761 55122 22239 928 177150 8:2 297655 210359 72995 110246694 45798 19777 1125 209234 9:2 297357 202011 79363 119753474 50953 22088 1089 200922 10:2 278797 178703 83180 122514385 52969 22726 1128 177575 11:2 272661 167669 86978 127342327 55857 24342 1195 166474 12:2 293039 204031 73211 110282059 47285 19651 948 203083 13:2 289502 196762 76803 114712942 49339 20547 1016 195746 14:2 264446 169609 78292 115715605 50459 21017 982 168627 15:2 260968 163660 80142 116811793 51483 21281 1064 162596 With this change: domainstats: domain0 cpu cnt bln fld imb gain hgain nobusyq nobusyg 0:2 272347 187380 77455 105420270 24975 1 953 186427 1:2 267276 172360 86234 116242264 28087 6 1028 171332 2:2 259769 156777 93281 123243134 30555 1 1043 155734 3:2 250870 143129 97627 127370868 32026 6 1188 141941 4:2 248422 177116 64096 78261112 22202 2 757 176359 5:2 275595 180683 84950 116075022 29400 6 778 179905 6:2 262418 162609 88944 119256898 31056 4 817 161792 7:2 252204 147946 92646 122388300 32879 4 824 147122 8:2 262335 172239 81631 110477214 26599 4 864 171375 9:2 261563 164775 88016 117203621 28331 3 849 163926 10:2 243389 140949 93379 121353071 29585 2 909 140040 11:2 242795 134651 98310 124768957 30895 2 1016 133635 12:2 255234 166622 79843 104696912 26483 4 746 165876 13:2 244944 151595 83855 109808099 27787 3 801 150794 14:2 241301 140982 89935 116954383 30403 6 845 140137 15:2 232271 128564 92821 119185207 31207 4 1416 127148 Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> LKML-Reference: <1284167957-3675-1-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-09-11 01:19:17 +00:00
/*
* Increment the failure counter only on periodic balance.
* We do not want newidle balance, which can be very
* frequent, pollute the failure counter causing
* excessive cache_hot migrations and active balances.
*/
if (idle != CPU_NEWLY_IDLE)
sd->nr_balance_failed++;
if (need_active_balance(sd, idle, cpu_of(busiest), this_cpu)) {
raw_spin_lock_irqsave(&busiest->lock, flags);
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
/* don't kick the active_load_balance_cpu_stop,
* if the curr task on busiest cpu can't be
* moved to this_cpu
*/
if (!cpumask_test_cpu(this_cpu,
tsk_cpus_allowed(busiest->curr))) {
raw_spin_unlock_irqrestore(&busiest->lock,
flags);
all_pinned = 1;
goto out_one_pinned;
}
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
/*
* ->active_balance synchronizes accesses to
* ->active_balance_work. Once set, it's cleared
* only after active load balance is finished.
*/
if (!busiest->active_balance) {
busiest->active_balance = 1;
busiest->push_cpu = this_cpu;
active_balance = 1;
}
raw_spin_unlock_irqrestore(&busiest->lock, flags);
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
if (active_balance)
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
stop_one_cpu_nowait(cpu_of(busiest),
active_load_balance_cpu_stop, busiest,
&busiest->active_balance_work);
/*
* We've kicked active balancing, reset the failure
* counter.
*/
sd->nr_balance_failed = sd->cache_nice_tries+1;
}
} else
sd->nr_balance_failed = 0;
if (likely(!active_balance)) {
/* We were unbalanced, so reset the balancing interval */
sd->balance_interval = sd->min_interval;
} else {
/*
* If we've begun active balancing, start to back off. This
* case may not be covered by the all_pinned logic if there
* is only 1 task on the busy runqueue (because we don't call
* move_tasks).
*/
if (sd->balance_interval < sd->max_interval)
sd->balance_interval *= 2;
}
goto out;
out_balanced:
schedstat_inc(sd, lb_balanced[idle]);
sd->nr_balance_failed = 0;
out_one_pinned:
/* tune up the balancing interval */
if ((all_pinned && sd->balance_interval < MAX_PINNED_INTERVAL) ||
(sd->balance_interval < sd->max_interval))
sd->balance_interval *= 2;
ld_moved = 0;
out:
return ld_moved;
}
/*
* idle_balance is called by schedule() if this_cpu is about to become
* idle. Attempts to pull tasks from other CPUs.
*/
void idle_balance(int this_cpu, struct rq *this_rq)
{
struct sched_domain *sd;
int pulled_task = 0;
unsigned long next_balance = jiffies + HZ;
this_rq->idle_stamp = this_rq->clock;
if (this_rq->avg_idle < sysctl_sched_migration_cost)
return;
/*
* Drop the rq->lock, but keep IRQ/preempt disabled.
*/
raw_spin_unlock(&this_rq->lock);
update_shares(this_cpu);
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_lock();
for_each_domain(this_cpu, sd) {
unsigned long interval;
int balance = 1;
if (!(sd->flags & SD_LOAD_BALANCE))
continue;
if (sd->flags & SD_BALANCE_NEWIDLE) {
/* If we've pulled tasks over stop searching: */
pulled_task = load_balance(this_cpu, this_rq,
sd, CPU_NEWLY_IDLE, &balance);
}
interval = msecs_to_jiffies(sd->balance_interval);
if (time_after(next_balance, sd->last_balance + interval))
next_balance = sd->last_balance + interval;
if (pulled_task) {
this_rq->idle_stamp = 0;
break;
}
}
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_unlock();
raw_spin_lock(&this_rq->lock);
if (pulled_task || time_after(jiffies, this_rq->next_balance)) {
/*
* We are going idle. next_balance may be set based on
* a busy processor. So reset next_balance.
*/
this_rq->next_balance = next_balance;
}
}
/*
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
* active_load_balance_cpu_stop is run by cpu stopper. It pushes
* running tasks off the busiest CPU onto idle CPUs. It requires at
* least 1 task to be running on each physical CPU where possible, and
* avoids physical / logical imbalances.
*/
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
static int active_load_balance_cpu_stop(void *data)
{
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
struct rq *busiest_rq = data;
int busiest_cpu = cpu_of(busiest_rq);
int target_cpu = busiest_rq->push_cpu;
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
struct rq *target_rq = cpu_rq(target_cpu);
struct sched_domain *sd;
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
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)
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
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. */
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_lock();
for_each_domain(target_cpu, sd) {
if ((sd->flags & SD_LOAD_BALANCE) &&
cpumask_test_cpu(busiest_cpu, sched_domain_span(sd)))
break;
}
if (likely(sd)) {
schedstat_inc(sd, alb_count);
if (move_one_task(target_rq, target_cpu, busiest_rq,
sd, CPU_IDLE))
schedstat_inc(sd, alb_pushed);
else
schedstat_inc(sd, alb_failed);
}
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_unlock();
double_unlock_balance(busiest_rq, target_rq);
sched: replace migration_thread with cpu_stop Currently migration_thread is serving three purposes - migration pusher, context to execute active_load_balance() and forced context switcher for expedited RCU synchronize_sched. All three roles are hardcoded into migration_thread() and determining which job is scheduled is slightly messy. This patch kills migration_thread and replaces all three uses with cpu_stop. The three different roles of migration_thread() are splitted into three separate cpu_stop callbacks - migration_cpu_stop(), active_load_balance_cpu_stop() and synchronize_sched_expedited_cpu_stop() - and each use case now simply asks cpu_stop to execute the callback as necessary. synchronize_sched_expedited() was implemented with private preallocated resources and custom multi-cpu queueing and waiting logic, both of which are provided by cpu_stop. synchronize_sched_expedited_count is made atomic and all other shared resources along with the mutex are dropped. synchronize_sched_expedited() also implemented a check to detect cases where not all the callback got executed on their assigned cpus and fall back to synchronize_sched(). If called with cpu hotplug blocked, cpu_stop already guarantees that and the condition cannot happen; otherwise, stop_machine() would break. However, this patch preserves the paranoid check using a cpumask to record on which cpus the stopper ran so that it can serve as a bisection point if something actually goes wrong theree. Because the internal execution state is no longer visible, rcu_expedited_torture_stats() is removed. This patch also renames cpu_stop threads to from "stopper/%d" to "migration/%d". The names of these threads ultimately don't matter and there's no reason to make unnecessary userland visible changes. With this patch applied, stop_machine() and sched now share the same resources. stop_machine() is faster without wasting any resources and sched migration users are much cleaner. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Dipankar Sarma <dipankar@in.ibm.com> Cc: Josh Triplett <josh@freedesktop.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Dimitri Sivanich <sivanich@sgi.com>
2010-05-06 16:49:21 +00:00
out_unlock:
busiest_rq->active_balance = 0;
raw_spin_unlock_irq(&busiest_rq->lock);
return 0;
}
#ifdef CONFIG_NO_HZ
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
/*
* idle load balancing details
* - One of the idle CPUs nominates itself as idle load_balancer, while
* entering idle.
* - This idle load balancer CPU will also go into tickless mode when
* it is idle, just like all other idle CPUs
* - When one of the busy CPUs notice that there may be an idle rebalancing
* needed, they will kick the idle load balancer, which then does idle
* load balancing for all the idle CPUs.
*/
static struct {
atomic_t load_balancer;
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
atomic_t first_pick_cpu;
atomic_t second_pick_cpu;
cpumask_var_t idle_cpus_mask;
cpumask_var_t grp_idle_mask;
unsigned long next_balance; /* in jiffy units */
} nohz ____cacheline_aligned;
int get_nohz_load_balancer(void)
{
return atomic_read(&nohz.load_balancer);
}
#if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
/**
* lowest_flag_domain - Return lowest sched_domain containing flag.
* @cpu: The cpu whose lowest level of sched domain is to
* be returned.
* @flag: The flag to check for the lowest sched_domain
* for the given cpu.
*
* Returns the lowest sched_domain of a cpu which contains the given flag.
*/
static inline struct sched_domain *lowest_flag_domain(int cpu, int flag)
{
struct sched_domain *sd;
for_each_domain(cpu, sd)
if (sd->flags & flag)
break;
return sd;
}
/**
* for_each_flag_domain - Iterates over sched_domains containing the flag.
* @cpu: The cpu whose domains we're iterating over.
* @sd: variable holding the value of the power_savings_sd
* for cpu.
* @flag: The flag to filter the sched_domains to be iterated.
*
* Iterates over all the scheduler domains for a given cpu that has the 'flag'
* set, starting from the lowest sched_domain to the highest.
*/
#define for_each_flag_domain(cpu, sd, flag) \
for (sd = lowest_flag_domain(cpu, flag); \
(sd && (sd->flags & flag)); sd = sd->parent)
/**
* is_semi_idle_group - Checks if the given sched_group is semi-idle.
* @ilb_group: group to be checked for semi-idleness
*
* Returns: 1 if the group is semi-idle. 0 otherwise.
*
* We define a sched_group to be semi idle if it has atleast one idle-CPU
* and atleast one non-idle CPU. This helper function checks if the given
* sched_group is semi-idle or not.
*/
static inline int is_semi_idle_group(struct sched_group *ilb_group)
{
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
cpumask_and(nohz.grp_idle_mask, nohz.idle_cpus_mask,
sched_group_cpus(ilb_group));
/*
* A sched_group is semi-idle when it has atleast one busy cpu
* and atleast one idle cpu.
*/
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (cpumask_empty(nohz.grp_idle_mask))
return 0;
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (cpumask_equal(nohz.grp_idle_mask, sched_group_cpus(ilb_group)))
return 0;
return 1;
}
/**
* find_new_ilb - Finds the optimum idle load balancer for nomination.
* @cpu: The cpu which is nominating a new idle_load_balancer.
*
* Returns: Returns the id of the idle load balancer if it exists,
* Else, returns >= nr_cpu_ids.
*
* This algorithm picks the idle load balancer such that it belongs to a
* semi-idle powersavings sched_domain. The idea is to try and avoid
* completely idle packages/cores just for the purpose of idle load balancing
* when there are other idle cpu's which are better suited for that job.
*/
static int find_new_ilb(int cpu)
{
struct sched_domain *sd;
struct sched_group *ilb_group;
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
int ilb = nr_cpu_ids;
/*
* Have idle load balancer selection from semi-idle packages only
* when power-aware load balancing is enabled
*/
if (!(sched_smt_power_savings || sched_mc_power_savings))
goto out_done;
/*
* Optimize for the case when we have no idle CPUs or only one
* idle CPU. Don't walk the sched_domain hierarchy in such cases
*/
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (cpumask_weight(nohz.idle_cpus_mask) < 2)
goto out_done;
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_lock();
for_each_flag_domain(cpu, sd, SD_POWERSAVINGS_BALANCE) {
ilb_group = sd->groups;
do {
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
if (is_semi_idle_group(ilb_group)) {
ilb = cpumask_first(nohz.grp_idle_mask);
goto unlock;
}
ilb_group = ilb_group->next;
} while (ilb_group != sd->groups);
}
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
unlock:
rcu_read_unlock();
out_done:
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
return ilb;
}
#else /* (CONFIG_SCHED_MC || CONFIG_SCHED_SMT) */
static inline int find_new_ilb(int call_cpu)
{
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
return nr_cpu_ids;
}
#endif
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
/*
* Kick a CPU to do the nohz balancing, if it is time for it. We pick the
* nohz_load_balancer CPU (if there is one) otherwise fallback to any idle
* CPU (if there is one).
*/
static void nohz_balancer_kick(int cpu)
{
int ilb_cpu;
nohz.next_balance++;
ilb_cpu = get_nohz_load_balancer();
if (ilb_cpu >= nr_cpu_ids) {
ilb_cpu = cpumask_first(nohz.idle_cpus_mask);
if (ilb_cpu >= nr_cpu_ids)
return;
}
if (!cpu_rq(ilb_cpu)->nohz_balance_kick) {
cpu_rq(ilb_cpu)->nohz_balance_kick = 1;
sched: Use resched IPI to kick off the nohz idle balance Current use of smp call function to kick the nohz idle balance can deadlock in this scenario. 1. cpu-A did a generic_exec_single() to cpu-B and after queuing its call single data (csd) to the call single queue, cpu-A took a timer interrupt. Actual IPI to cpu-B to process the call single queue is not yet sent. 2. As part of the timer interrupt handler, cpu-A decided to kick cpu-B for the idle load balancing (sets cpu-B's rq->nohz_balance_kick to 1) and __smp_call_function_single() with nowait will queue the csd to the cpu-B's queue. But the generic_exec_single() won't send an IPI to cpu-B as the call single queue was not empty. 3. cpu-A is busy with lot of interrupts 4. Meanwhile cpu-B is entering and exiting idle and noticed that it has it's rq->nohz_balance_kick set to '1'. So it will go ahead and do the idle load balancer and clear its rq->nohz_balance_kick. 5. At this point, csd queued as part of the step-2 above is still locked and waiting to be serviced on cpu-B. 6. cpu-A is still busy with interrupt load and now it got another timer interrupt and as part of it decided to kick cpu-B for another idle load balancing (as it finds cpu-B's rq->nohz_balance_kick cleared in step-4 above) and does __smp_call_function_single() with the same csd that is still locked. 7. And we get a deadlock waiting for the csd_lock() in the __smp_call_function_single(). Main issue here is that cpu-B can service the idle load balancer kick request from cpu-A even with out receiving the IPI and this lead to doing multiple __smp_call_function_single() on the same csd leading to deadlock. To kick a cpu, scheduler already has the reschedule vector reserved. Use that mechanism (kick_process()) instead of using the generic smp call function mechanism to kick off the nohz idle load balancing and avoid the deadlock. [ This issue is present from 2.6.35+ kernels, but marking it -stable only from v3.0+ as the proposed fix depends on the scheduler_ipi() that is introduced recently. ] Reported-by: Prarit Bhargava <prarit@redhat.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: stable@kernel.org # v3.0+ Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20111003220934.834943260@sbsiddha-desk.sc.intel.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-10-03 22:09:00 +00:00
smp_mb();
/*
* Use smp_send_reschedule() instead of resched_cpu().
* This way we generate a sched IPI on the target cpu which
* is idle. And the softirq performing nohz idle load balance
* will be run before returning from the IPI.
*/
smp_send_reschedule(ilb_cpu);
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
}
return;
}
/*
* This routine will try to nominate the ilb (idle load balancing)
* owner among the cpus whose ticks are stopped. ilb owner will do the idle
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
* load balancing on behalf of all those cpus.
*
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
* When the ilb owner becomes busy, we will not have new ilb owner until some
* idle CPU wakes up and goes back to idle or some busy CPU tries to kick
* idle load balancing by kicking one of the idle CPUs.
*
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
* Ticks are stopped for the ilb owner as well, with busy CPU kicking this
* ilb owner CPU in future (when there is a need for idle load balancing on
* behalf of all idle CPUs).
*/
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
void select_nohz_load_balancer(int stop_tick)
{
int cpu = smp_processor_id();
if (stop_tick) {
if (!cpu_active(cpu)) {
if (atomic_read(&nohz.load_balancer) != cpu)
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
return;
/*
* If we are going offline and still the leader,
* give up!
*/
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (atomic_cmpxchg(&nohz.load_balancer, cpu,
nr_cpu_ids) != cpu)
BUG();
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
return;
}
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
cpumask_set_cpu(cpu, nohz.idle_cpus_mask);
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (atomic_read(&nohz.first_pick_cpu) == cpu)
atomic_cmpxchg(&nohz.first_pick_cpu, cpu, nr_cpu_ids);
if (atomic_read(&nohz.second_pick_cpu) == cpu)
atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids);
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (atomic_read(&nohz.load_balancer) >= nr_cpu_ids) {
int new_ilb;
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
/* make me the ilb owner */
if (atomic_cmpxchg(&nohz.load_balancer, nr_cpu_ids,
cpu) != nr_cpu_ids)
return;
/*
* Check to see if there is a more power-efficient
* ilb.
*/
new_ilb = find_new_ilb(cpu);
if (new_ilb < nr_cpu_ids && new_ilb != cpu) {
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
atomic_set(&nohz.load_balancer, nr_cpu_ids);
resched_cpu(new_ilb);
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
return;
}
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
return;
}
} else {
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (!cpumask_test_cpu(cpu, nohz.idle_cpus_mask))
return;
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
cpumask_clear_cpu(cpu, nohz.idle_cpus_mask);
if (atomic_read(&nohz.load_balancer) == cpu)
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
if (atomic_cmpxchg(&nohz.load_balancer, cpu,
nr_cpu_ids) != cpu)
BUG();
}
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
return;
}
#endif
static DEFINE_SPINLOCK(balancing);
static unsigned long __read_mostly max_load_balance_interval = HZ/10;
/*
* Scale the max load_balance interval with the number of CPUs in the system.
* This trades load-balance latency on larger machines for less cross talk.
*/
void update_max_interval(void)
{
max_load_balance_interval = HZ*num_online_cpus()/10;
}
/*
* It checks each scheduling domain to see if it is due to be balanced,
* and initiates a balancing operation if so.
*
* Balancing parameters are set up in arch_init_sched_domains.
*/
static void rebalance_domains(int cpu, enum cpu_idle_type idle)
{
int balance = 1;
struct rq *rq = cpu_rq(cpu);
unsigned long interval;
struct sched_domain *sd;
/* Earliest time when we have to do rebalance again */
unsigned long next_balance = jiffies + 60*HZ;
int update_next_balance = 0;
int need_serialize;
update_shares(cpu);
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_lock();
for_each_domain(cpu, sd) {
if (!(sd->flags & SD_LOAD_BALANCE))
continue;
interval = sd->balance_interval;
if (idle != CPU_IDLE)
interval *= sd->busy_factor;
/* scale ms to jiffies */
interval = msecs_to_jiffies(interval);
interval = clamp(interval, 1UL, max_load_balance_interval);
need_serialize = sd->flags & SD_SERIALIZE;
if (need_serialize) {
if (!spin_trylock(&balancing))
goto out;
}
if (time_after_eq(jiffies, sd->last_balance + interval)) {
if (load_balance(cpu, rq, sd, idle, &balance)) {
/*
* We've pulled tasks over so either we're no
* longer idle.
*/
idle = CPU_NOT_IDLE;
}
sd->last_balance = jiffies;
}
if (need_serialize)
spin_unlock(&balancing);
out:
if (time_after(next_balance, sd->last_balance + interval)) {
next_balance = sd->last_balance + interval;
update_next_balance = 1;
}
/*
* Stop the load balance at this level. There is another
* CPU in our sched group which is doing load balancing more
* actively.
*/
if (!balance)
break;
}
sched: Dynamically allocate sched_domain/sched_group data-structures Instead of relying on static allocations for the sched_domain and sched_group trees, dynamically allocate and RCU free them. Allocating this dynamically also allows for some build_sched_groups() simplification since we can now (like with other simplifications) rely on the sched_domain tree instead of hard-coded knowledge. One tricky to note is that detach_destroy_domains() needs to hold rcu_read_lock() over the entire tear-down, per-cpu is not sufficient since that can lead to partial sched_group existance (could possibly be solved by doing the tear-down backwards but this is much more robust). A concequence of the above is that we can no longer print the sched_domain debug stuff from cpu_attach_domain() since that might now run with preemption disabled (due to classic RCU etc.) and sched_domain_debug() does some GFP_KERNEL allocations. Another thing to note is that we now fully rely on normal RCU and not RCU-sched, this is because with the new and exiting RCU flavours we grew over the years BH doesn't necessarily hold off RCU-sched grace periods (-rt is known to break this). This would in fact already cause us grief since we do sched_domain/sched_group iterations from softirq context. This patch is somewhat larger than I would like it to be, but I didn't find any means of shrinking/splitting this. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-04-07 12:09:50 +00:00
rcu_read_unlock();
/*
* next_balance will be updated only when there is a need.
* When the cpu is attached to null domain for ex, it will not be
* updated.
*/
if (likely(update_next_balance))
rq->next_balance = next_balance;
}
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
#ifdef CONFIG_NO_HZ
/*
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
* In CONFIG_NO_HZ case, the idle balance kickee will do the
* rebalancing for all the cpus for whom scheduler ticks are stopped.
*/
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle)
{
struct rq *this_rq = cpu_rq(this_cpu);
struct rq *rq;
int balance_cpu;
if (idle != CPU_IDLE || !this_rq->nohz_balance_kick)
return;
for_each_cpu(balance_cpu, nohz.idle_cpus_mask) {
if (balance_cpu == this_cpu)
continue;
/*
* If this cpu gets work to do, stop the load balancing
* work being done for other cpus. Next load
* balancing owner will pick it up.
*/
if (need_resched()) {
this_rq->nohz_balance_kick = 0;
break;
}
raw_spin_lock_irq(&this_rq->lock);
update_rq_clock(this_rq);
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
update_cpu_load(this_rq);
raw_spin_unlock_irq(&this_rq->lock);
rebalance_domains(balance_cpu, CPU_IDLE);
rq = cpu_rq(balance_cpu);
if (time_after(this_rq->next_balance, rq->next_balance))
this_rq->next_balance = rq->next_balance;
}
nohz.next_balance = this_rq->next_balance;
this_rq->nohz_balance_kick = 0;
}
/*
* Current heuristic for kicking the idle load balancer
* - first_pick_cpu is the one of the busy CPUs. It will kick
* idle load balancer when it has more than one process active. This
* eliminates the need for idle load balancing altogether when we have
* only one running process in the system (common case).
* - If there are more than one busy CPU, idle load balancer may have
* to run for active_load_balance to happen (i.e., two busy CPUs are
* SMT or core siblings and can run better if they move to different
* physical CPUs). So, second_pick_cpu is the second of the busy CPUs
* which will kick idle load balancer as soon as it has any load.
*/
static inline int nohz_kick_needed(struct rq *rq, int cpu)
{
unsigned long now = jiffies;
int ret;
int first_pick_cpu, second_pick_cpu;
if (time_before(now, nohz.next_balance))
return 0;
if (idle_cpu(cpu))
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
return 0;
first_pick_cpu = atomic_read(&nohz.first_pick_cpu);
second_pick_cpu = atomic_read(&nohz.second_pick_cpu);
if (first_pick_cpu < nr_cpu_ids && first_pick_cpu != cpu &&
second_pick_cpu < nr_cpu_ids && second_pick_cpu != cpu)
return 0;
ret = atomic_cmpxchg(&nohz.first_pick_cpu, nr_cpu_ids, cpu);
if (ret == nr_cpu_ids || ret == cpu) {
atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids);
if (rq->nr_running > 1)
return 1;
} else {
ret = atomic_cmpxchg(&nohz.second_pick_cpu, nr_cpu_ids, cpu);
if (ret == nr_cpu_ids || ret == cpu) {
if (rq->nr_running)
return 1;
}
}
return 0;
}
#else
static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle) { }
#endif
/*
* run_rebalance_domains is triggered when needed from the scheduler tick.
* Also triggered for nohz idle balancing (with nohz_balancing_kick set).
*/
static void run_rebalance_domains(struct softirq_action *h)
{
int this_cpu = smp_processor_id();
struct rq *this_rq = cpu_rq(this_cpu);
enum cpu_idle_type idle = this_rq->idle_balance ?
CPU_IDLE : CPU_NOT_IDLE;
rebalance_domains(this_cpu, idle);
/*
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
* If this cpu has a pending nohz_balance_kick, then do the
* balancing on behalf of the other idle cpus whose ticks are
* stopped.
*/
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
nohz_idle_balance(this_cpu, idle);
}
static inline int on_null_domain(int cpu)
{
return !rcu_dereference_sched(cpu_rq(cpu)->sd);
}
/*
* Trigger the SCHED_SOFTIRQ if it is time to do periodic load balancing.
*/
void trigger_load_balance(struct rq *rq, int cpu)
{
/* Don't need to rebalance while attached to NULL domain */
if (time_after_eq(jiffies, rq->next_balance) &&
likely(!on_null_domain(cpu)))
raise_softirq(SCHED_SOFTIRQ);
sched: Change nohz idle load balancing logic to push model In the new push model, all idle CPUs indeed go into nohz mode. There is still the concept of idle load balancer (performing the load balancing on behalf of all the idle cpu's in the system). Busy CPU kicks the nohz balancer when any of the nohz CPUs need idle load balancing. The kickee CPU does the idle load balancing on behalf of all idle CPUs instead of the normal idle balance. This addresses the below two problems with the current nohz ilb logic: * the idle load balancer continued to have periodic ticks during idle and wokeup frequently, even though it did not have any rebalancing to do on behalf of any of the idle CPUs. * On x86 and CPUs that have APIC timer stoppage on idle CPUs, this periodic wakeup can result in a periodic additional interrupt on a CPU doing the timer broadcast. Also currently we are migrating the unpinned timers from an idle to the cpu doing idle load balancing (when all the cpus in the system are idle, there is no idle load balancing cpu and timers get added to the same idle cpu where the request was made. So the existing optimization works only on semi idle system). And In semi idle system, we no longer have periodic ticks on the idle load balancer CPU. Using that cpu will add more delays to the timers than intended (as that cpu's timer base may not be uptodate wrt jiffies etc). This was causing mysterious slowdowns during boot etc. For now, in the semi idle case, use the nearest busy cpu for migrating timers from an idle cpu. This is good for power-savings anyway. Signed-off-by: Venkatesh Pallipadi <venki@google.com> Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> LKML-Reference: <1274486981.2840.46.camel@sbs-t61.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-05-22 00:09:41 +00:00
#ifdef CONFIG_NO_HZ
else if (nohz_kick_needed(rq, cpu) && likely(!on_null_domain(cpu)))
nohz_balancer_kick(cpu);
#endif
}
static void rq_online_fair(struct rq *rq)
{
update_sysctl();
}
static void rq_offline_fair(struct rq *rq)
{
update_sysctl();
}
#endif /* CONFIG_SMP */
/*
* scheduler tick hitting a task of our scheduling class:
*/
static void task_tick_fair(struct rq *rq, struct task_struct *curr, int queued)
{
struct cfs_rq *cfs_rq;
struct sched_entity *se = &curr->se;
for_each_sched_entity(se) {
cfs_rq = cfs_rq_of(se);
entity_tick(cfs_rq, se, queued);
}
}
/*
* called on fork with the child task as argument from the parent's context
* - child not yet on the tasklist
* - preemption disabled
*/
static void task_fork_fair(struct task_struct *p)
{
struct cfs_rq *cfs_rq = task_cfs_rq(current);
struct sched_entity *se = &p->se, *curr = cfs_rq->curr;
int this_cpu = smp_processor_id();
struct rq *rq = this_rq();
unsigned long flags;
raw_spin_lock_irqsave(&rq->lock, flags);
update_rq_clock(rq);
if (unlikely(task_cpu(p) != this_cpu)) {
rcu_read_lock();
__set_task_cpu(p, this_cpu);
rcu_read_unlock();
}
update_curr(cfs_rq);
if (curr)
se->vruntime = curr->vruntime;
place_entity(cfs_rq, se, 1);
if (sysctl_sched_child_runs_first && curr && entity_before(curr, se)) {
/*
* Upon rescheduling, sched_class::put_prev_task() will place
* 'current' within the tree based on its new key value.
*/
swap(curr->vruntime, se->vruntime);
resched_task(rq->curr);
}
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
se->vruntime -= cfs_rq->min_vruntime;
raw_spin_unlock_irqrestore(&rq->lock, flags);
}
/*
* Priority of the task has changed. Check to see if we preempt
* the current task.
*/
static void
prio_changed_fair(struct rq *rq, struct task_struct *p, int oldprio)
{
if (!p->se.on_rq)
return;
/*
* Reschedule if we are currently running on this runqueue and
* our priority decreased, or if we are not currently running on
* this runqueue and our priority is higher than the current's
*/
if (rq->curr == p) {
if (p->prio > oldprio)
resched_task(rq->curr);
} else
check_preempt_curr(rq, p, 0);
}
static void switched_from_fair(struct rq *rq, struct task_struct *p)
{
struct sched_entity *se = &p->se;
struct cfs_rq *cfs_rq = cfs_rq_of(se);
/*
* Ensure the task's vruntime is normalized, so that when its
* switched back to the fair class the enqueue_entity(.flags=0) will
* do the right thing.
*
* If it was on_rq, then the dequeue_entity(.flags=0) will already
* have normalized the vruntime, if it was !on_rq, then only when
* the task is sleeping will it still have non-normalized vruntime.
*/
if (!se->on_rq && p->state != TASK_RUNNING) {
/*
* Fix up our vruntime so that the current sleep doesn't
* cause 'unlimited' sleep bonus.
*/
place_entity(cfs_rq, se, 0);
se->vruntime -= cfs_rq->min_vruntime;
}
}
/*
* We switched to the sched_fair class.
*/
static void switched_to_fair(struct rq *rq, struct task_struct *p)
{
if (!p->se.on_rq)
return;
/*
* We were most likely switched from sched_rt, so
* kick off the schedule if running, otherwise just see
* if we can still preempt the current task.
*/
if (rq->curr == p)
resched_task(rq->curr);
else
check_preempt_curr(rq, p, 0);
}
/* Account for a task changing its policy or group.
*
* This routine is mostly called to set cfs_rq->curr field when a task
* migrates between groups/classes.
*/
static void set_curr_task_fair(struct rq *rq)
{
struct sched_entity *se = &rq->curr->se;
for_each_sched_entity(se) {
struct cfs_rq *cfs_rq = cfs_rq_of(se);
set_next_entity(cfs_rq, se);
/* ensure bandwidth has been allocated on our new cfs_rq */
account_cfs_rq_runtime(cfs_rq, 0);
}
}
void init_cfs_rq(struct cfs_rq *cfs_rq)
{
cfs_rq->tasks_timeline = RB_ROOT;
INIT_LIST_HEAD(&cfs_rq->tasks);
cfs_rq->min_vruntime = (u64)(-(1LL << 20));
#ifndef CONFIG_64BIT
cfs_rq->min_vruntime_copy = cfs_rq->min_vruntime;
#endif
}
#ifdef CONFIG_FAIR_GROUP_SCHED
static void task_move_group_fair(struct task_struct *p, int on_rq)
{
/*
* If the task was not on the rq at the time of this cgroup movement
* it must have been asleep, sleeping tasks keep their ->vruntime
* absolute on their old rq until wakeup (needed for the fair sleeper
* bonus in place_entity()).
*
* If it was on the rq, we've just 'preempted' it, which does convert
* ->vruntime to a relative base.
*
* Make sure both cases convert their relative position when migrating
* to another cgroup's rq. This does somewhat interfere with the
* fair sleeper stuff for the first placement, but who cares.
*/
if (!on_rq)
p->se.vruntime -= cfs_rq_of(&p->se)->min_vruntime;
set_task_rq(p, task_cpu(p));
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
if (!on_rq)
p->se.vruntime += cfs_rq_of(&p->se)->min_vruntime;
}
void free_fair_sched_group(struct task_group *tg)
{
int i;
destroy_cfs_bandwidth(tg_cfs_bandwidth(tg));
for_each_possible_cpu(i) {
if (tg->cfs_rq)
kfree(tg->cfs_rq[i]);
if (tg->se)
kfree(tg->se[i]);
}
kfree(tg->cfs_rq);
kfree(tg->se);
}
int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent)
{
struct cfs_rq *cfs_rq;
struct sched_entity *se;
int i;
tg->cfs_rq = kzalloc(sizeof(cfs_rq) * nr_cpu_ids, GFP_KERNEL);
if (!tg->cfs_rq)
goto err;
tg->se = kzalloc(sizeof(se) * nr_cpu_ids, GFP_KERNEL);
if (!tg->se)
goto err;
tg->shares = NICE_0_LOAD;
init_cfs_bandwidth(tg_cfs_bandwidth(tg));
for_each_possible_cpu(i) {
cfs_rq = kzalloc_node(sizeof(struct cfs_rq),
GFP_KERNEL, cpu_to_node(i));
if (!cfs_rq)
goto err;
se = kzalloc_node(sizeof(struct sched_entity),
GFP_KERNEL, cpu_to_node(i));
if (!se)
goto err_free_rq;
init_cfs_rq(cfs_rq);
init_tg_cfs_entry(tg, cfs_rq, se, i, parent->se[i]);
}
return 1;
err_free_rq:
kfree(cfs_rq);
err:
return 0;
}
void unregister_fair_sched_group(struct task_group *tg, int cpu)
{
struct rq *rq = cpu_rq(cpu);
unsigned long flags;
/*
* Only empty task groups can be destroyed; so we can speculatively
* check on_list without danger of it being re-added.
*/
if (!tg->cfs_rq[cpu]->on_list)
return;
raw_spin_lock_irqsave(&rq->lock, flags);
list_del_leaf_cfs_rq(tg->cfs_rq[cpu]);
raw_spin_unlock_irqrestore(&rq->lock, flags);
}
void init_tg_cfs_entry(struct task_group *tg, struct cfs_rq *cfs_rq,
struct sched_entity *se, int cpu,
struct sched_entity *parent)
{
struct rq *rq = cpu_rq(cpu);
cfs_rq->tg = tg;
cfs_rq->rq = rq;
#ifdef CONFIG_SMP
/* allow initial update_cfs_load() to truncate */
cfs_rq->load_stamp = 1;
#endif
init_cfs_rq_runtime(cfs_rq);
tg->cfs_rq[cpu] = cfs_rq;
tg->se[cpu] = se;
/* se could be NULL for root_task_group */
if (!se)
return;
if (!parent)
se->cfs_rq = &rq->cfs;
else
se->cfs_rq = parent->my_q;
se->my_q = cfs_rq;
update_load_set(&se->load, 0);
se->parent = parent;
}
static DEFINE_MUTEX(shares_mutex);
int sched_group_set_shares(struct task_group *tg, unsigned long shares)
{
int i;
unsigned long flags;
/*
* We can't change the weight of the root cgroup.
*/
if (!tg->se[0])
return -EINVAL;
shares = clamp(shares, scale_load(MIN_SHARES), scale_load(MAX_SHARES));
mutex_lock(&shares_mutex);
if (tg->shares == shares)
goto done;
tg->shares = shares;
for_each_possible_cpu(i) {
struct rq *rq = cpu_rq(i);
struct sched_entity *se;
se = tg->se[i];
/* Propagate contribution to hierarchy */
raw_spin_lock_irqsave(&rq->lock, flags);
for_each_sched_entity(se)
update_cfs_shares(group_cfs_rq(se));
raw_spin_unlock_irqrestore(&rq->lock, flags);
}
done:
mutex_unlock(&shares_mutex);
return 0;
}
#else /* CONFIG_FAIR_GROUP_SCHED */
void free_fair_sched_group(struct task_group *tg) { }
int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent)
{
return 1;
}
void unregister_fair_sched_group(struct task_group *tg, int cpu) { }
#endif /* CONFIG_FAIR_GROUP_SCHED */
static unsigned int get_rr_interval_fair(struct rq *rq, struct task_struct *task)
{
struct sched_entity *se = &task->se;
unsigned int rr_interval = 0;
/*
* Time slice is 0 for SCHED_OTHER tasks that are on an otherwise
* idle runqueue:
*/
if (rq->cfs.load.weight)
rr_interval = NS_TO_JIFFIES(sched_slice(&rq->cfs, se));
return rr_interval;
}
/*
* All the scheduling class methods:
*/
const struct sched_class fair_sched_class = {
.next = &idle_sched_class,
.enqueue_task = enqueue_task_fair,
.dequeue_task = dequeue_task_fair,
.yield_task = yield_task_fair,
.yield_to_task = yield_to_task_fair,
.check_preempt_curr = check_preempt_wakeup,
.pick_next_task = pick_next_task_fair,
.put_prev_task = put_prev_task_fair,
#ifdef CONFIG_SMP
.select_task_rq = select_task_rq_fair,
.rq_online = rq_online_fair,
.rq_offline = rq_offline_fair,
sched: Remove the cfs_rq dependency from set_task_cpu() In order to remove the cfs_rq dependency from set_task_cpu() we need to ensure the task is cfs_rq invariant for all callsites. The simple approach is to substract cfs_rq->min_vruntime from se->vruntime on dequeue, and add cfs_rq->min_vruntime on enqueue. However, this has the downside of breaking FAIR_SLEEPERS since we loose the old vruntime as we only maintain the relative position. To solve this, we observe that we only migrate runnable tasks, we do this using deactivate_task(.sleep=0) and activate_task(.wakeup=0), therefore we can restrain the min_vruntime invariance to that state. The only other case is wakeup balancing, since we want to maintain the old vruntime we cannot make it relative on dequeue, but since we don't migrate inactive tasks, we can do so right before we activate it again. This is where we need the new pre-wakeup hook, we need to call this while still holding the old rq->lock. We could fold it into ->select_task_rq(), but since that has multiple callsites and would obfuscate the locking requirements, that seems like a fudge. This leaves the fork() case, simply make sure that ->task_fork() leaves the ->vruntime in a relative state. This covers all cases where set_task_cpu() gets called, and ensures it sees a relative vruntime. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <20091216170518.191697025@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 17:04:41 +00:00
.task_waking = task_waking_fair,
#endif
.set_curr_task = set_curr_task_fair,
.task_tick = task_tick_fair,
.task_fork = task_fork_fair,
.prio_changed = prio_changed_fair,
.switched_from = switched_from_fair,
.switched_to = switched_to_fair,
.get_rr_interval = get_rr_interval_fair,
#ifdef CONFIG_FAIR_GROUP_SCHED
.task_move_group = task_move_group_fair,
#endif
};
#ifdef CONFIG_SCHED_DEBUG
void print_cfs_stats(struct seq_file *m, int cpu)
{
struct cfs_rq *cfs_rq;
rcu_read_lock();
for_each_leaf_cfs_rq(cpu_rq(cpu), cfs_rq)
print_cfs_rq(m, cpu, cfs_rq);
rcu_read_unlock();
}
#endif
__init void init_sched_fair_class(void)
{
#ifdef CONFIG_SMP
open_softirq(SCHED_SOFTIRQ, run_rebalance_domains);
#ifdef CONFIG_NO_HZ
zalloc_cpumask_var(&nohz.idle_cpus_mask, GFP_NOWAIT);
alloc_cpumask_var(&nohz.grp_idle_mask, GFP_NOWAIT);
atomic_set(&nohz.load_balancer, nr_cpu_ids);
atomic_set(&nohz.first_pick_cpu, nr_cpu_ids);
atomic_set(&nohz.second_pick_cpu, nr_cpu_ids);
#endif
#endif /* SMP */
}