From 2df8406a439bdeacbb5b74bbf91c376447d447dc Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Wed, 15 Apr 2015 18:51:18 +0200 Subject: sched/autogroup: Remove unnecessary #ifdef guards Since commit: 029632fbb7b7 ("sched: Make separate sched*.c translation units") autogroup is a separate translation unit built from the Makefile and thus no longer needs its content wrapped with #ifdef CONFIG_SCHED_AUTOGROUP. Signed-off-by: Tobias Klauser Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1429116678-17000-1-git-send-email-tklauser@distanz.ch Signed-off-by: Ingo Molnar --- kernel/sched/auto_group.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/auto_group.c b/kernel/sched/auto_group.c index eae160dd669d..1a3b58d531b2 100644 --- a/kernel/sched/auto_group.c +++ b/kernel/sched/auto_group.c @@ -1,5 +1,3 @@ -#ifdef CONFIG_SCHED_AUTOGROUP - #include "sched.h" #include @@ -249,5 +247,3 @@ int autogroup_path(struct task_group *tg, char *buf, int buflen) return snprintf(buf, buflen, "%s-%ld", "/autogroup", tg->autogroup->id); } #endif /* CONFIG_SCHED_DEBUG */ - -#endif /* CONFIG_SCHED_AUTOGROUP */ -- cgit v1.2.3 From 6a82b60da26bc404a8fca242521d988c437d0611 Mon Sep 17 00:00:00 2001 From: Paul Gortmaker Date: Mon, 27 Apr 2015 18:47:50 -0400 Subject: sched/core: Remove __cpuinit section tag that crept back in We removed __cpuinit support (leaving no-op stubs) quite some time ago. However this one crept back in as of commit a803f0261bb2bb57aab ("sched: Initialize rq->age_stamp on processor start") Since we want to clobber the stubs too, get this removed now. Signed-off-by: Paul Gortmaker Signed-off-by: Peter Zijlstra (Intel) Cc: Borislav Petkov Cc: Corey Minyard Cc: H. Peter Anvin Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1430174880-27958-2-git-send-email-paul.gortmaker@windriver.com Signed-off-by: Ingo Molnar --- kernel/sched/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index fe22f7510bce..43ba7651d620 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -5315,7 +5315,7 @@ static struct notifier_block migration_notifier = { .priority = CPU_PRI_MIGRATION, }; -static void __cpuinit set_cpu_rq_start_time(void) +static void set_cpu_rq_start_time(void) { int cpu = smp_processor_id(); struct rq *rq = cpu_rq(cpu); -- cgit v1.2.3 From 3289bdb429884c0279bf9ab72dff7b934f19dfc6 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 14 Apr 2015 13:19:42 +0200 Subject: sched: Move the loadavg code to a more obvious location I could not find the loadavg code.. turns out it was hidden in a file called proc.c. It further got mingled up with the cruft per rq load indexes (which we really want to get rid of). Move the per rq load indexes into the fair.c load-balance code (that's the only thing that uses them) and rename proc.c to loadavg.c so we can find it again. Signed-off-by: Peter Zijlstra (Intel) Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Paul Gortmaker Cc: Thomas Gleixner [ Did minor cleanups to the code. ] Signed-off-by: Ingo Molnar --- include/linux/sched.h | 5 + kernel/sched/Makefile | 2 +- kernel/sched/core.c | 7 +- kernel/sched/fair.c | 183 ++++++++++++++++ kernel/sched/loadavg.c | 394 +++++++++++++++++++++++++++++++++ kernel/sched/proc.c | 584 ------------------------------------------------- kernel/sched/sched.h | 8 +- 7 files changed, 593 insertions(+), 590 deletions(-) create mode 100644 kernel/sched/loadavg.c delete mode 100644 kernel/sched/proc.c (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 26a2e6122734..85cf253bc366 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -173,7 +173,12 @@ extern unsigned long nr_iowait_cpu(int cpu); extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load); extern void calc_global_load(unsigned long ticks); + +#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) extern void update_cpu_load_nohz(void); +#else +static inline void update_cpu_load_nohz(void) { } +#endif extern unsigned long get_parent_ip(unsigned long addr); diff --git a/kernel/sched/Makefile b/kernel/sched/Makefile index 46be87024875..67687973ce80 100644 --- a/kernel/sched/Makefile +++ b/kernel/sched/Makefile @@ -11,7 +11,7 @@ ifneq ($(CONFIG_SCHED_OMIT_FRAME_POINTER),y) CFLAGS_core.o := $(PROFILING) -fno-omit-frame-pointer endif -obj-y += core.o proc.o clock.o cputime.o +obj-y += core.o loadavg.o clock.o cputime.o obj-y += idle_task.o fair.o rt.o deadline.o stop_task.o obj-y += wait.o completion.o idle.o obj-$(CONFIG_SMP) += cpupri.o cpudeadline.o diff --git a/kernel/sched/core.c b/kernel/sched/core.c index fdf972d56f65..527fc28a737a 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2397,9 +2397,9 @@ unsigned long nr_iowait_cpu(int cpu) void get_iowait_load(unsigned long *nr_waiters, unsigned long *load) { - struct rq *this = this_rq(); - *nr_waiters = atomic_read(&this->nr_iowait); - *load = this->cpu_load[0]; + struct rq *rq = this_rq(); + *nr_waiters = atomic_read(&rq->nr_iowait); + *load = rq->load.weight; } #ifdef CONFIG_SMP @@ -2497,6 +2497,7 @@ void scheduler_tick(void) update_rq_clock(rq); curr->sched_class->task_tick(rq, curr, 0); update_cpu_load_active(rq); + calc_global_load_tick(rq); raw_spin_unlock(&rq->lock); perf_event_task_tick(); diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index ffeaa4105e48..4bc6013886ec 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -4323,6 +4323,189 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags) } #ifdef CONFIG_SMP + +/* + * per rq 'load' arrray crap; XXX kill this. + */ + +/* + * The exact cpuload at various idx values, calculated at every tick would be + * load = (2^idx - 1) / 2^idx * load + 1 / 2^idx * cur_load + * + * If a cpu misses updates for n-1 ticks (as it was idle) and update gets called + * on nth tick when cpu may be busy, then we have: + * load = ((2^idx - 1) / 2^idx)^(n-1) * load + * load = (2^idx - 1) / 2^idx) * load + 1 / 2^idx * cur_load + * + * decay_load_missed() below does efficient calculation of + * load = ((2^idx - 1) / 2^idx)^(n-1) * load + * avoiding 0..n-1 loop doing load = ((2^idx - 1) / 2^idx) * load + * + * The calculation is approximated on a 128 point scale. + * degrade_zero_ticks is the number of ticks after which load at any + * particular idx is approximated to be zero. + * degrade_factor is a precomputed table, a row for each load idx. + * Each column corresponds to degradation factor for a power of two ticks, + * based on 128 point scale. + * Example: + * row 2, col 3 (=12) says that the degradation at load idx 2 after + * 8 ticks is 12/128 (which is an approximation of exact factor 3^8/4^8). + * + * With this power of 2 load factors, we can degrade the load n times + * by looking at 1 bits in n and doing as many mult/shift instead of + * n mult/shifts needed by the exact degradation. + */ +#define DEGRADE_SHIFT 7 +static const unsigned char + degrade_zero_ticks[CPU_LOAD_IDX_MAX] = {0, 8, 32, 64, 128}; +static const unsigned char + degrade_factor[CPU_LOAD_IDX_MAX][DEGRADE_SHIFT + 1] = { + {0, 0, 0, 0, 0, 0, 0, 0}, + {64, 32, 8, 0, 0, 0, 0, 0}, + {96, 72, 40, 12, 1, 0, 0}, + {112, 98, 75, 43, 15, 1, 0}, + {120, 112, 98, 76, 45, 16, 2} }; + +/* + * Update cpu_load for any missed ticks, due to tickless idle. The backlog + * would be when CPU is idle and so we just decay the old load without + * adding any new load. + */ +static unsigned long +decay_load_missed(unsigned long load, unsigned long missed_updates, int idx) +{ + int j = 0; + + if (!missed_updates) + return load; + + if (missed_updates >= degrade_zero_ticks[idx]) + return 0; + + if (idx == 1) + return load >> missed_updates; + + while (missed_updates) { + if (missed_updates % 2) + load = (load * degrade_factor[idx][j]) >> DEGRADE_SHIFT; + + missed_updates >>= 1; + j++; + } + return load; +} + +/* + * Update rq->cpu_load[] statistics. This function is usually called every + * scheduler tick (TICK_NSEC). With tickless idle this will not be called + * every tick. We fix it up based on jiffies. + */ +static void __update_cpu_load(struct rq *this_rq, unsigned long this_load, + unsigned long pending_updates) +{ + int i, scale; + + this_rq->nr_load_updates++; + + /* Update our load: */ + this_rq->cpu_load[0] = this_load; /* Fasttrack for idx 0 */ + for (i = 1, scale = 2; i < CPU_LOAD_IDX_MAX; i++, scale += scale) { + unsigned long old_load, new_load; + + /* scale is effectively 1 << i now, and >> i divides by scale */ + + old_load = this_rq->cpu_load[i]; + old_load = decay_load_missed(old_load, pending_updates - 1, i); + new_load = this_load; + /* + * Round up the averaging division if load is increasing. This + * prevents us from getting stuck on 9 if the load is 10, for + * example. + */ + if (new_load > old_load) + new_load += scale - 1; + + this_rq->cpu_load[i] = (old_load * (scale - 1) + new_load) >> i; + } + + sched_avg_update(this_rq); +} + +#ifdef CONFIG_NO_HZ_COMMON +/* + * There is no sane way to deal with nohz on smp when using jiffies because the + * cpu doing the jiffies update might drift wrt the cpu doing the jiffy reading + * causing off-by-one errors in observed deltas; {0,2} instead of {1,1}. + * + * Therefore we cannot use the delta approach from the regular tick since that + * would seriously skew the load calculation. However we'll make do for those + * updates happening while idle (nohz_idle_balance) or coming out of idle + * (tick_nohz_idle_exit). + * + * This means we might still be one tick off for nohz periods. + */ + +/* + * Called from nohz_idle_balance() to update the load ratings before doing the + * idle balance. + */ +static void update_idle_cpu_load(struct rq *this_rq) +{ + unsigned long curr_jiffies = ACCESS_ONCE(jiffies); + unsigned long load = this_rq->cfs.runnable_load_avg; + unsigned long pending_updates; + + /* + * bail if there's load or we're actually up-to-date. + */ + if (load || curr_jiffies == this_rq->last_load_update_tick) + return; + + pending_updates = curr_jiffies - this_rq->last_load_update_tick; + this_rq->last_load_update_tick = curr_jiffies; + + __update_cpu_load(this_rq, load, pending_updates); +} + +/* + * Called from tick_nohz_idle_exit() -- try and fix up the ticks we missed. + */ +void update_cpu_load_nohz(void) +{ + struct rq *this_rq = this_rq(); + unsigned long curr_jiffies = ACCESS_ONCE(jiffies); + unsigned long pending_updates; + + if (curr_jiffies == this_rq->last_load_update_tick) + return; + + raw_spin_lock(&this_rq->lock); + pending_updates = curr_jiffies - this_rq->last_load_update_tick; + if (pending_updates) { + this_rq->last_load_update_tick = curr_jiffies; + /* + * We were idle, this means load 0, the current load might be + * !0 due to remote wakeups and the sort. + */ + __update_cpu_load(this_rq, 0, pending_updates); + } + raw_spin_unlock(&this_rq->lock); +} +#endif /* CONFIG_NO_HZ */ + +/* + * Called from scheduler_tick() + */ +void update_cpu_load_active(struct rq *this_rq) +{ + unsigned long load = this_rq->cfs.runnable_load_avg; + /* + * See the mess around update_idle_cpu_load() / update_cpu_load_nohz(). + */ + this_rq->last_load_update_tick = jiffies; + __update_cpu_load(this_rq, load, 1); +} + /* Used instead of source_load when we know the type == 0 */ static unsigned long weighted_cpuload(const int cpu) { diff --git a/kernel/sched/loadavg.c b/kernel/sched/loadavg.c new file mode 100644 index 000000000000..ef7159012cf3 --- /dev/null +++ b/kernel/sched/loadavg.c @@ -0,0 +1,394 @@ +/* + * kernel/sched/loadavg.c + * + * This file contains the magic bits required to compute the global loadavg + * figure. Its a silly number but people think its important. We go through + * great pains to make it work on big machines and tickless kernels. + */ + +#include + +#include "sched.h" + +/* + * Global load-average calculations + * + * We take a distributed and async approach to calculating the global load-avg + * in order to minimize overhead. + * + * The global load average is an exponentially decaying average of nr_running + + * nr_uninterruptible. + * + * Once every LOAD_FREQ: + * + * nr_active = 0; + * for_each_possible_cpu(cpu) + * nr_active += cpu_of(cpu)->nr_running + cpu_of(cpu)->nr_uninterruptible; + * + * avenrun[n] = avenrun[0] * exp_n + nr_active * (1 - exp_n) + * + * Due to a number of reasons the above turns in the mess below: + * + * - for_each_possible_cpu() is prohibitively expensive on machines with + * serious number of cpus, therefore we need to take a distributed approach + * to calculating nr_active. + * + * \Sum_i x_i(t) = \Sum_i x_i(t) - x_i(t_0) | x_i(t_0) := 0 + * = \Sum_i { \Sum_j=1 x_i(t_j) - x_i(t_j-1) } + * + * So assuming nr_active := 0 when we start out -- true per definition, we + * can simply take per-cpu deltas and fold those into a global accumulate + * to obtain the same result. See calc_load_fold_active(). + * + * Furthermore, in order to avoid synchronizing all per-cpu delta folding + * across the machine, we assume 10 ticks is sufficient time for every + * cpu to have completed this task. + * + * This places an upper-bound on the IRQ-off latency of the machine. Then + * again, being late doesn't loose the delta, just wrecks the sample. + * + * - cpu_rq()->nr_uninterruptible isn't accurately tracked per-cpu because + * this would add another cross-cpu cacheline miss and atomic operation + * to the wakeup path. Instead we increment on whatever cpu the task ran + * when it went into uninterruptible state and decrement on whatever cpu + * did the wakeup. This means that only the sum of nr_uninterruptible over + * all cpus yields the correct result. + * + * This covers the NO_HZ=n code, for extra head-aches, see the comment below. + */ + +/* Variables and functions for calc_load */ +atomic_long_t calc_load_tasks; +unsigned long calc_load_update; +unsigned long avenrun[3]; +EXPORT_SYMBOL(avenrun); /* should be removed */ + +/** + * get_avenrun - get the load average array + * @loads: pointer to dest load array + * @offset: offset to add + * @shift: shift count to shift the result left + * + * These values are estimates at best, so no need for locking. + */ +void get_avenrun(unsigned long *loads, unsigned long offset, int shift) +{ + loads[0] = (avenrun[0] + offset) << shift; + loads[1] = (avenrun[1] + offset) << shift; + loads[2] = (avenrun[2] + offset) << shift; +} + +long calc_load_fold_active(struct rq *this_rq) +{ + long nr_active, delta = 0; + + nr_active = this_rq->nr_running; + nr_active += (long)this_rq->nr_uninterruptible; + + if (nr_active != this_rq->calc_load_active) { + delta = nr_active - this_rq->calc_load_active; + this_rq->calc_load_active = nr_active; + } + + return delta; +} + +/* + * a1 = a0 * e + a * (1 - e) + */ +static unsigned long +calc_load(unsigned long load, unsigned long exp, unsigned long active) +{ + load *= exp; + load += active * (FIXED_1 - exp); + load += 1UL << (FSHIFT - 1); + return load >> FSHIFT; +} + +#ifdef CONFIG_NO_HZ_COMMON +/* + * Handle NO_HZ for the global load-average. + * + * Since the above described distributed algorithm to compute the global + * load-average relies on per-cpu sampling from the tick, it is affected by + * NO_HZ. + * + * The basic idea is to fold the nr_active delta into a global idle-delta upon + * entering NO_HZ state such that we can include this as an 'extra' cpu delta + * when we read the global state. + * + * Obviously reality has to ruin such a delightfully simple scheme: + * + * - When we go NO_HZ idle during the window, we can negate our sample + * contribution, causing under-accounting. + * + * We avoid this by keeping two idle-delta counters and flipping them + * when the window starts, thus separating old and new NO_HZ load. + * + * The only trick is the slight shift in index flip for read vs write. + * + * 0s 5s 10s 15s + * +10 +10 +10 +10 + * |-|-----------|-|-----------|-|-----------|-| + * r:0 0 1 1 0 0 1 1 0 + * w:0 1 1 0 0 1 1 0 0 + * + * This ensures we'll fold the old idle contribution in this window while + * accumlating the new one. + * + * - When we wake up from NO_HZ idle during the window, we push up our + * contribution, since we effectively move our sample point to a known + * busy state. + * + * This is solved by pushing the window forward, and thus skipping the + * sample, for this cpu (effectively using the idle-delta for this cpu which + * was in effect at the time the window opened). This also solves the issue + * of having to deal with a cpu having been in NOHZ idle for multiple + * LOAD_FREQ intervals. + * + * When making the ILB scale, we should try to pull this in as well. + */ +static atomic_long_t calc_load_idle[2]; +static int calc_load_idx; + +static inline int calc_load_write_idx(void) +{ + int idx = calc_load_idx; + + /* + * See calc_global_nohz(), if we observe the new index, we also + * need to observe the new update time. + */ + smp_rmb(); + + /* + * If the folding window started, make sure we start writing in the + * next idle-delta. + */ + if (!time_before(jiffies, calc_load_update)) + idx++; + + return idx & 1; +} + +static inline int calc_load_read_idx(void) +{ + return calc_load_idx & 1; +} + +void calc_load_enter_idle(void) +{ + struct rq *this_rq = this_rq(); + long delta; + + /* + * We're going into NOHZ mode, if there's any pending delta, fold it + * into the pending idle delta. + */ + delta = calc_load_fold_active(this_rq); + if (delta) { + int idx = calc_load_write_idx(); + + atomic_long_add(delta, &calc_load_idle[idx]); + } +} + +void calc_load_exit_idle(void) +{ + struct rq *this_rq = this_rq(); + + /* + * If we're still before the sample window, we're done. + */ + if (time_before(jiffies, this_rq->calc_load_update)) + return; + + /* + * We woke inside or after the sample window, this means we're already + * accounted through the nohz accounting, so skip the entire deal and + * sync up for the next window. + */ + this_rq->calc_load_update = calc_load_update; + if (time_before(jiffies, this_rq->calc_load_update + 10)) + this_rq->calc_load_update += LOAD_FREQ; +} + +static long calc_load_fold_idle(void) +{ + int idx = calc_load_read_idx(); + long delta = 0; + + if (atomic_long_read(&calc_load_idle[idx])) + delta = atomic_long_xchg(&calc_load_idle[idx], 0); + + return delta; +} + +/** + * fixed_power_int - compute: x^n, in O(log n) time + * + * @x: base of the power + * @frac_bits: fractional bits of @x + * @n: power to raise @x to. + * + * By exploiting the relation between the definition of the natural power + * function: x^n := x*x*...*x (x multiplied by itself for n times), and + * the binary encoding of numbers used by computers: n := \Sum n_i * 2^i, + * (where: n_i \elem {0, 1}, the binary vector representing n), + * we find: x^n := x^(\Sum n_i * 2^i) := \Prod x^(n_i * 2^i), which is + * of course trivially computable in O(log_2 n), the length of our binary + * vector. + */ +static unsigned long +fixed_power_int(unsigned long x, unsigned int frac_bits, unsigned int n) +{ + unsigned long result = 1UL << frac_bits; + + if (n) { + for (;;) { + if (n & 1) { + result *= x; + result += 1UL << (frac_bits - 1); + result >>= frac_bits; + } + n >>= 1; + if (!n) + break; + x *= x; + x += 1UL << (frac_bits - 1); + x >>= frac_bits; + } + } + + return result; +} + +/* + * a1 = a0 * e + a * (1 - e) + * + * a2 = a1 * e + a * (1 - e) + * = (a0 * e + a * (1 - e)) * e + a * (1 - e) + * = a0 * e^2 + a * (1 - e) * (1 + e) + * + * a3 = a2 * e + a * (1 - e) + * = (a0 * e^2 + a * (1 - e) * (1 + e)) * e + a * (1 - e) + * = a0 * e^3 + a * (1 - e) * (1 + e + e^2) + * + * ... + * + * an = a0 * e^n + a * (1 - e) * (1 + e + ... + e^n-1) [1] + * = a0 * e^n + a * (1 - e) * (1 - e^n)/(1 - e) + * = a0 * e^n + a * (1 - e^n) + * + * [1] application of the geometric series: + * + * n 1 - x^(n+1) + * S_n := \Sum x^i = ------------- + * i=0 1 - x + */ +static unsigned long +calc_load_n(unsigned long load, unsigned long exp, + unsigned long active, unsigned int n) +{ + return calc_load(load, fixed_power_int(exp, FSHIFT, n), active); +} + +/* + * NO_HZ can leave us missing all per-cpu ticks calling + * calc_load_account_active(), but since an idle CPU folds its delta into + * calc_load_tasks_idle per calc_load_account_idle(), all we need to do is fold + * in the pending idle delta if our idle period crossed a load cycle boundary. + * + * Once we've updated the global active value, we need to apply the exponential + * weights adjusted to the number of cycles missed. + */ +static void calc_global_nohz(void) +{ + long delta, active, n; + + if (!time_before(jiffies, calc_load_update + 10)) { + /* + * Catch-up, fold however many we are behind still + */ + delta = jiffies - calc_load_update - 10; + n = 1 + (delta / LOAD_FREQ); + + active = atomic_long_read(&calc_load_tasks); + active = active > 0 ? active * FIXED_1 : 0; + + avenrun[0] = calc_load_n(avenrun[0], EXP_1, active, n); + avenrun[1] = calc_load_n(avenrun[1], EXP_5, active, n); + avenrun[2] = calc_load_n(avenrun[2], EXP_15, active, n); + + calc_load_update += n * LOAD_FREQ; + } + + /* + * Flip the idle index... + * + * Make sure we first write the new time then flip the index, so that + * calc_load_write_idx() will see the new time when it reads the new + * index, this avoids a double flip messing things up. + */ + smp_wmb(); + calc_load_idx++; +} +#else /* !CONFIG_NO_HZ_COMMON */ + +static inline long calc_load_fold_idle(void) { return 0; } +static inline void calc_global_nohz(void) { } + +#endif /* CONFIG_NO_HZ_COMMON */ + +/* + * calc_load - update the avenrun load estimates 10 ticks after the + * CPUs have updated calc_load_tasks. + * + * Called from the global timer code. + */ +void calc_global_load(unsigned long ticks) +{ + long active, delta; + + if (time_before(jiffies, calc_load_update + 10)) + return; + + /* + * Fold the 'old' idle-delta to include all NO_HZ cpus. + */ + delta = calc_load_fold_idle(); + if (delta) + atomic_long_add(delta, &calc_load_tasks); + + active = atomic_long_read(&calc_load_tasks); + active = active > 0 ? active * FIXED_1 : 0; + + avenrun[0] = calc_load(avenrun[0], EXP_1, active); + avenrun[1] = calc_load(avenrun[1], EXP_5, active); + avenrun[2] = calc_load(avenrun[2], EXP_15, active); + + calc_load_update += LOAD_FREQ; + + /* + * In case we idled for multiple LOAD_FREQ intervals, catch up in bulk. + */ + calc_global_nohz(); +} + +/* + * Called from scheduler_tick() to periodically update this CPU's + * active count. + */ +void calc_global_load_tick(struct rq *this_rq) +{ + long delta; + + if (time_before(jiffies, this_rq->calc_load_update)) + return; + + delta = calc_load_fold_active(this_rq); + if (delta) + atomic_long_add(delta, &calc_load_tasks); + + this_rq->calc_load_update += LOAD_FREQ; +} diff --git a/kernel/sched/proc.c b/kernel/sched/proc.c deleted file mode 100644 index 8ecd552fe4f2..000000000000 --- a/kernel/sched/proc.c +++ /dev/null @@ -1,584 +0,0 @@ -/* - * kernel/sched/proc.c - * - * Kernel load calculations, forked from sched/core.c - */ - -#include - -#include "sched.h" - -/* - * Global load-average calculations - * - * We take a distributed and async approach to calculating the global load-avg - * in order to minimize overhead. - * - * The global load average is an exponentially decaying average of nr_running + - * nr_uninterruptible. - * - * Once every LOAD_FREQ: - * - * nr_active = 0; - * for_each_possible_cpu(cpu) - * nr_active += cpu_of(cpu)->nr_running + cpu_of(cpu)->nr_uninterruptible; - * - * avenrun[n] = avenrun[0] * exp_n + nr_active * (1 - exp_n) - * - * Due to a number of reasons the above turns in the mess below: - * - * - for_each_possible_cpu() is prohibitively expensive on machines with - * serious number of cpus, therefore we need to take a distributed approach - * to calculating nr_active. - * - * \Sum_i x_i(t) = \Sum_i x_i(t) - x_i(t_0) | x_i(t_0) := 0 - * = \Sum_i { \Sum_j=1 x_i(t_j) - x_i(t_j-1) } - * - * So assuming nr_active := 0 when we start out -- true per definition, we - * can simply take per-cpu deltas and fold those into a global accumulate - * to obtain the same result. See calc_load_fold_active(). - * - * Furthermore, in order to avoid synchronizing all per-cpu delta folding - * across the machine, we assume 10 ticks is sufficient time for every - * cpu to have completed this task. - * - * This places an upper-bound on the IRQ-off latency of the machine. Then - * again, being late doesn't loose the delta, just wrecks the sample. - * - * - cpu_rq()->nr_uninterruptible isn't accurately tracked per-cpu because - * this would add another cross-cpu cacheline miss and atomic operation - * to the wakeup path. Instead we increment on whatever cpu the task ran - * when it went into uninterruptible state and decrement on whatever cpu - * did the wakeup. This means that only the sum of nr_uninterruptible over - * all cpus yields the correct result. - * - * This covers the NO_HZ=n code, for extra head-aches, see the comment below. - */ - -/* Variables and functions for calc_load */ -atomic_long_t calc_load_tasks; -unsigned long calc_load_update; -unsigned long avenrun[3]; -EXPORT_SYMBOL(avenrun); /* should be removed */ - -/** - * get_avenrun - get the load average array - * @loads: pointer to dest load array - * @offset: offset to add - * @shift: shift count to shift the result left - * - * These values are estimates at best, so no need for locking. - */ -void get_avenrun(unsigned long *loads, unsigned long offset, int shift) -{ - loads[0] = (avenrun[0] + offset) << shift; - loads[1] = (avenrun[1] + offset) << shift; - loads[2] = (avenrun[2] + offset) << shift; -} - -long calc_load_fold_active(struct rq *this_rq) -{ - long nr_active, delta = 0; - - nr_active = this_rq->nr_running; - nr_active += (long) this_rq->nr_uninterruptible; - - if (nr_active != this_rq->calc_load_active) { - delta = nr_active - this_rq->calc_load_active; - this_rq->calc_load_active = nr_active; - } - - return delta; -} - -/* - * a1 = a0 * e + a * (1 - e) - */ -static unsigned long -calc_load(unsigned long load, unsigned long exp, unsigned long active) -{ - load *= exp; - load += active * (FIXED_1 - exp); - load += 1UL << (FSHIFT - 1); - return load >> FSHIFT; -} - -#ifdef CONFIG_NO_HZ_COMMON -/* - * Handle NO_HZ for the global load-average. - * - * Since the above described distributed algorithm to compute the global - * load-average relies on per-cpu sampling from the tick, it is affected by - * NO_HZ. - * - * The basic idea is to fold the nr_active delta into a global idle-delta upon - * entering NO_HZ state such that we can include this as an 'extra' cpu delta - * when we read the global state. - * - * Obviously reality has to ruin such a delightfully simple scheme: - * - * - When we go NO_HZ idle during the window, we can negate our sample - * contribution, causing under-accounting. - * - * We avoid this by keeping two idle-delta counters and flipping them - * when the window starts, thus separating old and new NO_HZ load. - * - * The only trick is the slight shift in index flip for read vs write. - * - * 0s 5s 10s 15s - * +10 +10 +10 +10 - * |-|-----------|-|-----------|-|-----------|-| - * r:0 0 1 1 0 0 1 1 0 - * w:0 1 1 0 0 1 1 0 0 - * - * This ensures we'll fold the old idle contribution in this window while - * accumlating the new one. - * - * - When we wake up from NO_HZ idle during the window, we push up our - * contribution, since we effectively move our sample point to a known - * busy state. - * - * This is solved by pushing the window forward, and thus skipping the - * sample, for this cpu (effectively using the idle-delta for this cpu which - * was in effect at the time the window opened). This also solves the issue - * of having to deal with a cpu having been in NOHZ idle for multiple - * LOAD_FREQ intervals. - * - * When making the ILB scale, we should try to pull this in as well. - */ -static atomic_long_t calc_load_idle[2]; -static int calc_load_idx; - -static inline int calc_load_write_idx(void) -{ - int idx = calc_load_idx; - - /* - * See calc_global_nohz(), if we observe the new index, we also - * need to observe the new update time. - */ - smp_rmb(); - - /* - * If the folding window started, make sure we start writing in the - * next idle-delta. - */ - if (!time_before(jiffies, calc_load_update)) - idx++; - - return idx & 1; -} - -static inline int calc_load_read_idx(void) -{ - return calc_load_idx & 1; -} - -void calc_load_enter_idle(void) -{ - struct rq *this_rq = this_rq(); - long delta; - - /* - * We're going into NOHZ mode, if there's any pending delta, fold it - * into the pending idle delta. - */ - delta = calc_load_fold_active(this_rq); - if (delta) { - int idx = calc_load_write_idx(); - atomic_long_add(delta, &calc_load_idle[idx]); - } -} - -void calc_load_exit_idle(void) -{ - struct rq *this_rq = this_rq(); - - /* - * If we're still before the sample window, we're done. - */ - if (time_before(jiffies, this_rq->calc_load_update)) - return; - - /* - * We woke inside or after the sample window, this means we're already - * accounted through the nohz accounting, so skip the entire deal and - * sync up for the next window. - */ - this_rq->calc_load_update = calc_load_update; - if (time_before(jiffies, this_rq->calc_load_update + 10)) - this_rq->calc_load_update += LOAD_FREQ; -} - -static long calc_load_fold_idle(void) -{ - int idx = calc_load_read_idx(); - long delta = 0; - - if (atomic_long_read(&calc_load_idle[idx])) - delta = atomic_long_xchg(&calc_load_idle[idx], 0); - - return delta; -} - -/** - * fixed_power_int - compute: x^n, in O(log n) time - * - * @x: base of the power - * @frac_bits: fractional bits of @x - * @n: power to raise @x to. - * - * By exploiting the relation between the definition of the natural power - * function: x^n := x*x*...*x (x multiplied by itself for n times), and - * the binary encoding of numbers used by computers: n := \Sum n_i * 2^i, - * (where: n_i \elem {0, 1}, the binary vector representing n), - * we find: x^n := x^(\Sum n_i * 2^i) := \Prod x^(n_i * 2^i), which is - * of course trivially computable in O(log_2 n), the length of our binary - * vector. - */ -static unsigned long -fixed_power_int(unsigned long x, unsigned int frac_bits, unsigned int n) -{ - unsigned long result = 1UL << frac_bits; - - if (n) for (;;) { - if (n & 1) { - result *= x; - result += 1UL << (frac_bits - 1); - result >>= frac_bits; - } - n >>= 1; - if (!n) - break; - x *= x; - x += 1UL << (frac_bits - 1); - x >>= frac_bits; - } - - return result; -} - -/* - * a1 = a0 * e + a * (1 - e) - * - * a2 = a1 * e + a * (1 - e) - * = (a0 * e + a * (1 - e)) * e + a * (1 - e) - * = a0 * e^2 + a * (1 - e) * (1 + e) - * - * a3 = a2 * e + a * (1 - e) - * = (a0 * e^2 + a * (1 - e) * (1 + e)) * e + a * (1 - e) - * = a0 * e^3 + a * (1 - e) * (1 + e + e^2) - * - * ... - * - * an = a0 * e^n + a * (1 - e) * (1 + e + ... + e^n-1) [1] - * = a0 * e^n + a * (1 - e) * (1 - e^n)/(1 - e) - * = a0 * e^n + a * (1 - e^n) - * - * [1] application of the geometric series: - * - * n 1 - x^(n+1) - * S_n := \Sum x^i = ------------- - * i=0 1 - x - */ -static unsigned long -calc_load_n(unsigned long load, unsigned long exp, - unsigned long active, unsigned int n) -{ - - return calc_load(load, fixed_power_int(exp, FSHIFT, n), active); -} - -/* - * NO_HZ can leave us missing all per-cpu ticks calling - * calc_load_account_active(), but since an idle CPU folds its delta into - * calc_load_tasks_idle per calc_load_account_idle(), all we need to do is fold - * in the pending idle delta if our idle period crossed a load cycle boundary. - * - * Once we've updated the global active value, we need to apply the exponential - * weights adjusted to the number of cycles missed. - */ -static void calc_global_nohz(void) -{ - long delta, active, n; - - if (!time_before(jiffies, calc_load_update + 10)) { - /* - * Catch-up, fold however many we are behind still - */ - delta = jiffies - calc_load_update - 10; - n = 1 + (delta / LOAD_FREQ); - - active = atomic_long_read(&calc_load_tasks); - active = active > 0 ? active * FIXED_1 : 0; - - avenrun[0] = calc_load_n(avenrun[0], EXP_1, active, n); - avenrun[1] = calc_load_n(avenrun[1], EXP_5, active, n); - avenrun[2] = calc_load_n(avenrun[2], EXP_15, active, n); - - calc_load_update += n * LOAD_FREQ; - } - - /* - * Flip the idle index... - * - * Make sure we first write the new time then flip the index, so that - * calc_load_write_idx() will see the new time when it reads the new - * index, this avoids a double flip messing things up. - */ - smp_wmb(); - calc_load_idx++; -} -#else /* !CONFIG_NO_HZ_COMMON */ - -static inline long calc_load_fold_idle(void) { return 0; } -static inline void calc_global_nohz(void) { } - -#endif /* CONFIG_NO_HZ_COMMON */ - -/* - * calc_load - update the avenrun load estimates 10 ticks after the - * CPUs have updated calc_load_tasks. - */ -void calc_global_load(unsigned long ticks) -{ - long active, delta; - - if (time_before(jiffies, calc_load_update + 10)) - return; - - /* - * Fold the 'old' idle-delta to include all NO_HZ cpus. - */ - delta = calc_load_fold_idle(); - if (delta) - atomic_long_add(delta, &calc_load_tasks); - - active = atomic_long_read(&calc_load_tasks); - active = active > 0 ? active * FIXED_1 : 0; - - avenrun[0] = calc_load(avenrun[0], EXP_1, active); - avenrun[1] = calc_load(avenrun[1], EXP_5, active); - avenrun[2] = calc_load(avenrun[2], EXP_15, active); - - calc_load_update += LOAD_FREQ; - - /* - * In case we idled for multiple LOAD_FREQ intervals, catch up in bulk. - */ - calc_global_nohz(); -} - -/* - * Called from update_cpu_load() to periodically update this CPU's - * active count. - */ -static void calc_load_account_active(struct rq *this_rq) -{ - long delta; - - if (time_before(jiffies, this_rq->calc_load_update)) - return; - - delta = calc_load_fold_active(this_rq); - if (delta) - atomic_long_add(delta, &calc_load_tasks); - - this_rq->calc_load_update += LOAD_FREQ; -} - -/* - * End of global load-average stuff - */ - -/* - * The exact cpuload at various idx values, calculated at every tick would be - * load = (2^idx - 1) / 2^idx * load + 1 / 2^idx * cur_load - * - * If a cpu misses updates for n-1 ticks (as it was idle) and update gets called - * on nth tick when cpu may be busy, then we have: - * load = ((2^idx - 1) / 2^idx)^(n-1) * load - * load = (2^idx - 1) / 2^idx) * load + 1 / 2^idx * cur_load - * - * decay_load_missed() below does efficient calculation of - * load = ((2^idx - 1) / 2^idx)^(n-1) * load - * avoiding 0..n-1 loop doing load = ((2^idx - 1) / 2^idx) * load - * - * The calculation is approximated on a 128 point scale. - * degrade_zero_ticks is the number of ticks after which load at any - * particular idx is approximated to be zero. - * degrade_factor is a precomputed table, a row for each load idx. - * Each column corresponds to degradation factor for a power of two ticks, - * based on 128 point scale. - * Example: - * row 2, col 3 (=12) says that the degradation at load idx 2 after - * 8 ticks is 12/128 (which is an approximation of exact factor 3^8/4^8). - * - * With this power of 2 load factors, we can degrade the load n times - * by looking at 1 bits in n and doing as many mult/shift instead of - * n mult/shifts needed by the exact degradation. - */ -#define DEGRADE_SHIFT 7 -static const unsigned char - degrade_zero_ticks[CPU_LOAD_IDX_MAX] = {0, 8, 32, 64, 128}; -static const unsigned char - degrade_factor[CPU_LOAD_IDX_MAX][DEGRADE_SHIFT + 1] = { - {0, 0, 0, 0, 0, 0, 0, 0}, - {64, 32, 8, 0, 0, 0, 0, 0}, - {96, 72, 40, 12, 1, 0, 0}, - {112, 98, 75, 43, 15, 1, 0}, - {120, 112, 98, 76, 45, 16, 2} }; - -/* - * Update cpu_load for any missed ticks, due to tickless idle. The backlog - * would be when CPU is idle and so we just decay the old load without - * adding any new load. - */ -static unsigned long -decay_load_missed(unsigned long load, unsigned long missed_updates, int idx) -{ - int j = 0; - - if (!missed_updates) - return load; - - if (missed_updates >= degrade_zero_ticks[idx]) - return 0; - - if (idx == 1) - return load >> missed_updates; - - while (missed_updates) { - if (missed_updates % 2) - load = (load * degrade_factor[idx][j]) >> DEGRADE_SHIFT; - - missed_updates >>= 1; - j++; - } - return load; -} - -/* - * Update rq->cpu_load[] statistics. This function is usually called every - * scheduler tick (TICK_NSEC). With tickless idle this will not be called - * every tick. We fix it up based on jiffies. - */ -static void __update_cpu_load(struct rq *this_rq, unsigned long this_load, - unsigned long pending_updates) -{ - int i, scale; - - this_rq->nr_load_updates++; - - /* Update our load: */ - this_rq->cpu_load[0] = this_load; /* Fasttrack for idx 0 */ - for (i = 1, scale = 2; i < CPU_LOAD_IDX_MAX; i++, scale += scale) { - unsigned long old_load, new_load; - - /* scale is effectively 1 << i now, and >> i divides by scale */ - - old_load = this_rq->cpu_load[i]; - old_load = decay_load_missed(old_load, pending_updates - 1, i); - new_load = this_load; - /* - * Round up the averaging division if load is increasing. This - * prevents us from getting stuck on 9 if the load is 10, for - * example. - */ - if (new_load > old_load) - new_load += scale - 1; - - this_rq->cpu_load[i] = (old_load * (scale - 1) + new_load) >> i; - } - - sched_avg_update(this_rq); -} - -#ifdef CONFIG_SMP -static inline unsigned long get_rq_runnable_load(struct rq *rq) -{ - return rq->cfs.runnable_load_avg; -} -#else -static inline unsigned long get_rq_runnable_load(struct rq *rq) -{ - return rq->load.weight; -} -#endif - -#ifdef CONFIG_NO_HZ_COMMON -/* - * There is no sane way to deal with nohz on smp when using jiffies because the - * cpu doing the jiffies update might drift wrt the cpu doing the jiffy reading - * causing off-by-one errors in observed deltas; {0,2} instead of {1,1}. - * - * Therefore we cannot use the delta approach from the regular tick since that - * would seriously skew the load calculation. However we'll make do for those - * updates happening while idle (nohz_idle_balance) or coming out of idle - * (tick_nohz_idle_exit). - * - * This means we might still be one tick off for nohz periods. - */ - -/* - * Called from nohz_idle_balance() to update the load ratings before doing the - * idle balance. - */ -void update_idle_cpu_load(struct rq *this_rq) -{ - unsigned long curr_jiffies = ACCESS_ONCE(jiffies); - unsigned long load = get_rq_runnable_load(this_rq); - unsigned long pending_updates; - - /* - * bail if there's load or we're actually up-to-date. - */ - if (load || curr_jiffies == this_rq->last_load_update_tick) - return; - - pending_updates = curr_jiffies - this_rq->last_load_update_tick; - this_rq->last_load_update_tick = curr_jiffies; - - __update_cpu_load(this_rq, load, pending_updates); -} - -/* - * Called from tick_nohz_idle_exit() -- try and fix up the ticks we missed. - */ -void update_cpu_load_nohz(void) -{ - struct rq *this_rq = this_rq(); - unsigned long curr_jiffies = ACCESS_ONCE(jiffies); - unsigned long pending_updates; - - if (curr_jiffies == this_rq->last_load_update_tick) - return; - - raw_spin_lock(&this_rq->lock); - pending_updates = curr_jiffies - this_rq->last_load_update_tick; - if (pending_updates) { - this_rq->last_load_update_tick = curr_jiffies; - /* - * We were idle, this means load 0, the current load might be - * !0 due to remote wakeups and the sort. - */ - __update_cpu_load(this_rq, 0, pending_updates); - } - raw_spin_unlock(&this_rq->lock); -} -#endif /* CONFIG_NO_HZ */ - -/* - * Called from scheduler_tick() - */ -void update_cpu_load_active(struct rq *this_rq) -{ - unsigned long load = get_rq_runnable_load(this_rq); - /* - * See the mess around update_idle_cpu_load() / update_cpu_load_nohz(). - */ - this_rq->last_load_update_tick = jiffies; - __update_cpu_load(this_rq, load, 1); - - calc_load_account_active(this_rq); -} diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index e0e129993958..09ed26a89f31 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -26,8 +26,14 @@ extern __read_mostly int scheduler_running; extern unsigned long calc_load_update; extern atomic_long_t calc_load_tasks; +extern void calc_global_load_tick(struct rq *this_rq); extern long calc_load_fold_active(struct rq *this_rq); + +#ifdef CONFIG_SMP extern void update_cpu_load_active(struct rq *this_rq); +#else +static inline void update_cpu_load_active(struct rq *this_rq) { } +#endif /* * Helpers for converting nanosecond timing to jiffy resolution @@ -1298,8 +1304,6 @@ extern void init_dl_task_timer(struct sched_dl_entity *dl_se); unsigned long to_ratio(u64 period, u64 runtime); -extern void update_idle_cpu_load(struct rq *this_rq); - extern void init_task_runnable_average(struct task_struct *p); static inline void add_nr_running(struct rq *rq, unsigned count) -- cgit v1.2.3 From b76808e6808e34e7e78131d2b8cb0535622b8e9f Mon Sep 17 00:00:00 2001 From: Palmer Dabbelt Date: Thu, 30 Apr 2015 21:19:57 -0700 Subject: signals, sched: Change all uses of JOBCTL_* from 'int' to 'long' c56fb6564dcd ("Fix a misaligned load inside ptrace_attach()") makes jobctl an "unsigned long". It makes sense to have the masks applied to it match that type. This is currently just a cosmetic change, but it will prevent the mask from being unexpectedly truncated if we ever end up with masks with more bits. One instance of "signr" is an int, but I left this alone because the mask ensures that it will never overflow. Signed-off-by: Palmer Dabbelt Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Chris Metcalf Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Thomas Gleixner Cc: bobby.prani@gmail.com Cc: oleg@redhat.com Cc: paulmck@linux.vnet.ibm.com Cc: richard@nod.at Cc: vdavydov@parallels.com Link: http://lkml.kernel.org/r/1430453997-32459-4-git-send-email-palmer@dabbelt.com Signed-off-by: Ingo Molnar --- include/linux/sched.h | 18 +++++++++--------- kernel/signal.c | 6 +++--- 2 files changed, 12 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 85cf253bc366..4f066cb625ad 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2082,22 +2082,22 @@ TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */ #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */ -#define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT) -#define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT) -#define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT) -#define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT) -#define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT) -#define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT) -#define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT) +#define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT) +#define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT) +#define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT) +#define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT) +#define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT) +#define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT) +#define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT) #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY) #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK) extern bool task_set_jobctl_pending(struct task_struct *task, - unsigned int mask); + unsigned long mask); extern void task_clear_jobctl_trapping(struct task_struct *task); extern void task_clear_jobctl_pending(struct task_struct *task, - unsigned int mask); + unsigned long mask); static inline void rcu_copy_process(struct task_struct *p) { diff --git a/kernel/signal.c b/kernel/signal.c index d51c5ddd855c..f19833b5db3c 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -245,7 +245,7 @@ static inline void print_dropped_signal(int sig) * RETURNS: * %true if @mask is set, %false if made noop because @task was dying. */ -bool task_set_jobctl_pending(struct task_struct *task, unsigned int mask) +bool task_set_jobctl_pending(struct task_struct *task, unsigned long mask) { BUG_ON(mask & ~(JOBCTL_PENDING_MASK | JOBCTL_STOP_CONSUME | JOBCTL_STOP_SIGMASK | JOBCTL_TRAPPING)); @@ -297,7 +297,7 @@ void task_clear_jobctl_trapping(struct task_struct *task) * CONTEXT: * Must be called with @task->sighand->siglock held. */ -void task_clear_jobctl_pending(struct task_struct *task, unsigned int mask) +void task_clear_jobctl_pending(struct task_struct *task, unsigned long mask) { BUG_ON(mask & ~JOBCTL_PENDING_MASK); @@ -2000,7 +2000,7 @@ static bool do_signal_stop(int signr) struct signal_struct *sig = current->signal; if (!(current->jobctl & JOBCTL_STOP_PENDING)) { - unsigned int gstop = JOBCTL_STOP_PENDING | JOBCTL_STOP_CONSUME; + unsigned long gstop = JOBCTL_STOP_PENDING | JOBCTL_STOP_CONSUME; struct task_struct *t; /* signr will be recorded in task->jobctl for retries */ -- cgit v1.2.3 From ce2f5fe46303d1e1a2ba453753a7e8200d32182c Mon Sep 17 00:00:00 2001 From: Nicholas Mc Guire Date: Sun, 3 May 2015 10:51:56 +0200 Subject: sched/core: Remove unnecessary down/up conversion 'rt_period_us' is automatically type converted from u64 to long and then cast back to u64 - this down/up conversion is unnecessary and can be removed to improve readability. This will also help us not truncate 'rt_period_us' to 32 bits on 32-bit kernels, should we ever have so large values. (unlikely, not the least due to procfs.) Signed-off-by: Nicholas Mc Guire Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1430643116-24049-1-git-send-email-hofrat@osadl.org Signed-off-by: Ingo Molnar --- kernel/sched/core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 527fc28a737a..46a5d6f05208 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -7738,11 +7738,11 @@ static long sched_group_rt_runtime(struct task_group *tg) return rt_runtime_us; } -static int sched_group_set_rt_period(struct task_group *tg, long rt_period_us) +static int sched_group_set_rt_period(struct task_group *tg, u64 rt_period_us) { u64 rt_runtime, rt_period; - rt_period = (u64)rt_period_us * NSEC_PER_USEC; + rt_period = rt_period_us * NSEC_PER_USEC; rt_runtime = tg->rt_bandwidth.rt_runtime; return tg_set_rt_bandwidth(tg, rt_period, rt_runtime); -- cgit v1.2.3 From 316c1608d15c736439d4065ed12f306db554b3da Mon Sep 17 00:00:00 2001 From: Jason Low Date: Tue, 28 Apr 2015 13:00:20 -0700 Subject: sched, timer: Convert usages of ACCESS_ONCE() in the scheduler to READ_ONCE()/WRITE_ONCE() ACCESS_ONCE doesn't work reliably on non-scalar types. This patch removes the rest of the existing usages of ACCESS_ONCE() in the scheduler, and use the new READ_ONCE() and WRITE_ONCE() APIs as appropriate. Signed-off-by: Jason Low Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Acked-by: Rik van Riel Acked-by: Waiman Long Cc: Andrew Morton Cc: Aswin Chandramouleeswaran Cc: Borislav Petkov Cc: Davidlohr Bueso Cc: Frederic Weisbecker Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Mel Gorman Cc: Mike Galbraith Cc: Oleg Nesterov Cc: Paul E. McKenney Cc: Preeti U Murthy Cc: Scott J Norton Cc: Steven Rostedt Link: http://lkml.kernel.org/r/1430251224-5764-2-git-send-email-jason.low2@hp.com Signed-off-by: Ingo Molnar --- include/linux/sched.h | 4 ++-- kernel/fork.c | 2 +- kernel/sched/auto_group.c | 2 +- kernel/sched/auto_group.h | 2 +- kernel/sched/core.c | 4 ++-- kernel/sched/cputime.c | 2 +- kernel/sched/deadline.c | 2 +- kernel/sched/fair.c | 18 +++++++++--------- kernel/sched/rt.c | 2 +- kernel/sched/sched.h | 2 +- kernel/sched/wait.c | 4 ++-- kernel/time/posix-cpu-timers.c | 8 ++++---- 12 files changed, 26 insertions(+), 26 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index fb650a2f4a73..d70910355b20 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -3085,13 +3085,13 @@ static inline void mm_update_next_owner(struct mm_struct *mm) static inline unsigned long task_rlimit(const struct task_struct *tsk, unsigned int limit) { - return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur); + return READ_ONCE(tsk->signal->rlim[limit].rlim_cur); } static inline unsigned long task_rlimit_max(const struct task_struct *tsk, unsigned int limit) { - return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max); + return READ_ONCE(tsk->signal->rlim[limit].rlim_max); } static inline unsigned long rlimit(unsigned int limit) diff --git a/kernel/fork.c b/kernel/fork.c index 03c1eaaa6ef5..47c37a411a62 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1094,7 +1094,7 @@ static void posix_cpu_timers_init_group(struct signal_struct *sig) /* Thread group counters. */ thread_group_cputime_init(sig); - cpu_limit = ACCESS_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur); + cpu_limit = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur); if (cpu_limit != RLIM_INFINITY) { sig->cputime_expires.prof_exp = secs_to_cputime(cpu_limit); sig->cputimer.running = 1; diff --git a/kernel/sched/auto_group.c b/kernel/sched/auto_group.c index 1a3b58d531b2..750ed601ddf7 100644 --- a/kernel/sched/auto_group.c +++ b/kernel/sched/auto_group.c @@ -139,7 +139,7 @@ autogroup_move_group(struct task_struct *p, struct autogroup *ag) p->signal->autogroup = autogroup_kref_get(ag); - if (!ACCESS_ONCE(sysctl_sched_autogroup_enabled)) + if (!READ_ONCE(sysctl_sched_autogroup_enabled)) goto out; for_each_thread(p, t) diff --git a/kernel/sched/auto_group.h b/kernel/sched/auto_group.h index 8bd047142816..890c95f2587a 100644 --- a/kernel/sched/auto_group.h +++ b/kernel/sched/auto_group.h @@ -29,7 +29,7 @@ extern bool task_wants_autogroup(struct task_struct *p, struct task_group *tg); static inline struct task_group * autogroup_task_group(struct task_struct *p, struct task_group *tg) { - int enabled = ACCESS_ONCE(sysctl_sched_autogroup_enabled); + int enabled = READ_ONCE(sysctl_sched_autogroup_enabled); if (enabled && task_wants_autogroup(p, tg)) return p->signal->autogroup->tg; diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 46a5d6f05208..22b53c863ef3 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -511,7 +511,7 @@ static bool set_nr_and_not_polling(struct task_struct *p) static bool set_nr_if_polling(struct task_struct *p) { struct thread_info *ti = task_thread_info(p); - typeof(ti->flags) old, val = ACCESS_ONCE(ti->flags); + typeof(ti->flags) old, val = READ_ONCE(ti->flags); for (;;) { if (!(val & _TIF_POLLING_NRFLAG)) @@ -2526,7 +2526,7 @@ void scheduler_tick(void) u64 scheduler_tick_max_deferment(void) { struct rq *rq = this_rq(); - unsigned long next, now = ACCESS_ONCE(jiffies); + unsigned long next, now = READ_ONCE(jiffies); next = rq->last_sched_tick + HZ; diff --git a/kernel/sched/cputime.c b/kernel/sched/cputime.c index 8394b1ee600c..f5a64ffad176 100644 --- a/kernel/sched/cputime.c +++ b/kernel/sched/cputime.c @@ -567,7 +567,7 @@ static void cputime_advance(cputime_t *counter, cputime_t new) { cputime_t old; - while (new > (old = ACCESS_ONCE(*counter))) + while (new > (old = READ_ONCE(*counter))) cmpxchg_cputime(counter, old, new); } diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c index 5e95145088fd..890ce951c717 100644 --- a/kernel/sched/deadline.c +++ b/kernel/sched/deadline.c @@ -995,7 +995,7 @@ select_task_rq_dl(struct task_struct *p, int cpu, int sd_flag, int flags) rq = cpu_rq(cpu); rcu_read_lock(); - curr = ACCESS_ONCE(rq->curr); /* unlocked access */ + curr = READ_ONCE(rq->curr); /* unlocked access */ /* * If we are dealing with a -deadline task, we must diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 4bc6013886ec..d6915a038d8a 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -834,7 +834,7 @@ static unsigned int task_nr_scan_windows(struct task_struct *p) static unsigned int task_scan_min(struct task_struct *p) { - unsigned int scan_size = ACCESS_ONCE(sysctl_numa_balancing_scan_size); + unsigned int scan_size = READ_ONCE(sysctl_numa_balancing_scan_size); unsigned int scan, floor; unsigned int windows = 1; @@ -1794,7 +1794,7 @@ static void task_numa_placement(struct task_struct *p) u64 runtime, period; spinlock_t *group_lock = NULL; - seq = ACCESS_ONCE(p->mm->numa_scan_seq); + seq = READ_ONCE(p->mm->numa_scan_seq); if (p->numa_scan_seq == seq) return; p->numa_scan_seq = seq; @@ -1938,7 +1938,7 @@ static void task_numa_group(struct task_struct *p, int cpupid, int flags, } rcu_read_lock(); - tsk = ACCESS_ONCE(cpu_rq(cpu)->curr); + tsk = READ_ONCE(cpu_rq(cpu)->curr); if (!cpupid_match_pid(tsk, cpupid)) goto no_join; @@ -2107,7 +2107,7 @@ void task_numa_fault(int last_cpupid, int mem_node, int pages, int flags) static void reset_ptenuma_scan(struct task_struct *p) { - ACCESS_ONCE(p->mm->numa_scan_seq)++; + WRITE_ONCE(p->mm->numa_scan_seq, READ_ONCE(p->mm->numa_scan_seq) + 1); p->mm->numa_scan_offset = 0; } @@ -4451,7 +4451,7 @@ static void __update_cpu_load(struct rq *this_rq, unsigned long this_load, */ static void update_idle_cpu_load(struct rq *this_rq) { - unsigned long curr_jiffies = ACCESS_ONCE(jiffies); + unsigned long curr_jiffies = READ_ONCE(jiffies); unsigned long load = this_rq->cfs.runnable_load_avg; unsigned long pending_updates; @@ -4473,7 +4473,7 @@ static void update_idle_cpu_load(struct rq *this_rq) void update_cpu_load_nohz(void) { struct rq *this_rq = this_rq(); - unsigned long curr_jiffies = ACCESS_ONCE(jiffies); + unsigned long curr_jiffies = READ_ONCE(jiffies); unsigned long pending_updates; if (curr_jiffies == this_rq->last_load_update_tick) @@ -4558,7 +4558,7 @@ static unsigned long capacity_orig_of(int cpu) static unsigned long cpu_avg_load_per_task(int cpu) { struct rq *rq = cpu_rq(cpu); - unsigned long nr_running = ACCESS_ONCE(rq->cfs.h_nr_running); + unsigned long nr_running = READ_ONCE(rq->cfs.h_nr_running); unsigned long load_avg = rq->cfs.runnable_load_avg; if (nr_running) @@ -6220,8 +6220,8 @@ static unsigned long scale_rt_capacity(int cpu) * Since we're reading these variables without serialization make sure * we read them once before doing sanity checks on them. */ - age_stamp = ACCESS_ONCE(rq->age_stamp); - avg = ACCESS_ONCE(rq->rt_avg); + age_stamp = READ_ONCE(rq->age_stamp); + avg = READ_ONCE(rq->rt_avg); delta = __rq_clock_broken(rq) - age_stamp; if (unlikely(delta < 0)) diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index 575da76a3874..560d2fa623c3 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -1323,7 +1323,7 @@ select_task_rq_rt(struct task_struct *p, int cpu, int sd_flag, int flags) rq = cpu_rq(cpu); rcu_read_lock(); - curr = ACCESS_ONCE(rq->curr); /* unlocked access */ + curr = READ_ONCE(rq->curr); /* unlocked access */ /* * If the current task on @p's runqueue is an RT task, then diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 09ed26a89f31..d85455539d5c 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -713,7 +713,7 @@ DECLARE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues); static inline u64 __rq_clock_broken(struct rq *rq) { - return ACCESS_ONCE(rq->clock); + return READ_ONCE(rq->clock); } static inline u64 rq_clock(struct rq *rq) diff --git a/kernel/sched/wait.c b/kernel/sched/wait.c index 852143a79f36..2ccec988d6b7 100644 --- a/kernel/sched/wait.c +++ b/kernel/sched/wait.c @@ -601,7 +601,7 @@ EXPORT_SYMBOL(bit_wait_io); __sched int bit_wait_timeout(struct wait_bit_key *word) { - unsigned long now = ACCESS_ONCE(jiffies); + unsigned long now = READ_ONCE(jiffies); if (signal_pending_state(current->state, current)) return 1; if (time_after_eq(now, word->timeout)) @@ -613,7 +613,7 @@ EXPORT_SYMBOL_GPL(bit_wait_timeout); __sched int bit_wait_io_timeout(struct wait_bit_key *word) { - unsigned long now = ACCESS_ONCE(jiffies); + unsigned long now = READ_ONCE(jiffies); if (signal_pending_state(current->state, current)) return 1; if (time_after_eq(now, word->timeout)) diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c index 0075da74abf0..e072d982f64c 100644 --- a/kernel/time/posix-cpu-timers.c +++ b/kernel/time/posix-cpu-timers.c @@ -852,10 +852,10 @@ static void check_thread_timers(struct task_struct *tsk, /* * Check for the special case thread timers. */ - soft = ACCESS_ONCE(sig->rlim[RLIMIT_RTTIME].rlim_cur); + soft = READ_ONCE(sig->rlim[RLIMIT_RTTIME].rlim_cur); if (soft != RLIM_INFINITY) { unsigned long hard = - ACCESS_ONCE(sig->rlim[RLIMIT_RTTIME].rlim_max); + READ_ONCE(sig->rlim[RLIMIT_RTTIME].rlim_max); if (hard != RLIM_INFINITY && tsk->rt.timeout > DIV_ROUND_UP(hard, USEC_PER_SEC/HZ)) { @@ -958,11 +958,11 @@ static void check_process_timers(struct task_struct *tsk, SIGPROF); check_cpu_itimer(tsk, &sig->it[CPUCLOCK_VIRT], &virt_expires, utime, SIGVTALRM); - soft = ACCESS_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur); + soft = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur); if (soft != RLIM_INFINITY) { unsigned long psecs = cputime_to_secs(ptime); unsigned long hard = - ACCESS_ONCE(sig->rlim[RLIMIT_CPU].rlim_max); + READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_max); cputime_t x; if (psecs >= hard) { /* -- cgit v1.2.3 From 7e5a2c1729f1612618ed236249a15bf15f309325 Mon Sep 17 00:00:00 2001 From: Jason Low Date: Thu, 30 Apr 2015 17:28:14 -0700 Subject: sched/numa: Document usages of mm->numa_scan_seq The p->mm->numa_scan_seq is accessed using READ_ONCE/WRITE_ONCE and modified without exclusive access. It is not clear why it is accessed this way. This patch provides some documentation on that. Suggested-by: Ingo Molnar Signed-off-by: Jason Low Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Acked-by: Rik van Riel Cc: Andrew Morton Cc: Aswin Chandramouleeswaran Cc: Borislav Petkov Cc: Davidlohr Bueso Cc: Frederic Weisbecker Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Mel Gorman Cc: Mike Galbraith Cc: Oleg Nesterov Cc: Paul E. McKenney Cc: Peter Zijlstra Cc: Preeti U Murthy Cc: Scott J Norton Cc: Steven Rostedt Cc: Waiman Long Link: http://lkml.kernel.org/r/1430440094.2475.61.camel@j-VirtualBox Signed-off-by: Ingo Molnar --- kernel/sched/fair.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index d6915a038d8a..f18ddb72fe88 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1794,6 +1794,11 @@ static void task_numa_placement(struct task_struct *p) u64 runtime, period; spinlock_t *group_lock = NULL; + /* + * The p->mm->numa_scan_seq field gets updated without + * exclusive access. Use READ_ONCE() here to ensure + * that the field is read in a single access: + */ seq = READ_ONCE(p->mm->numa_scan_seq); if (p->numa_scan_seq == seq) return; @@ -2107,6 +2112,14 @@ void task_numa_fault(int last_cpupid, int mem_node, int pages, int flags) static void reset_ptenuma_scan(struct task_struct *p) { + /* + * We only did a read acquisition of the mmap sem, so + * p->mm->numa_scan_seq is written to without exclusive access + * and the update is not guaranteed to be atomic. That's not + * much of an issue though, since this is just used for + * statistical sampling. Use READ_ONCE/WRITE_ONCE, which are not + * expensive, to avoid any form of compiler optimizations: + */ WRITE_ONCE(p->mm->numa_scan_seq, READ_ONCE(p->mm->numa_scan_seq) + 1); p->mm->numa_scan_offset = 0; } -- cgit v1.2.3 From 1018016c706f7ff9f56fde3a649789c47085a293 Mon Sep 17 00:00:00 2001 From: Jason Low Date: Tue, 28 Apr 2015 13:00:22 -0700 Subject: sched, timer: Replace spinlocks with atomics in thread_group_cputimer(), to improve scalability While running a database workload, we found a scalability issue with itimers. Much of the problem was caused by the thread_group_cputimer spinlock. Each time we account for group system/user time, we need to obtain a thread_group_cputimer's spinlock to update the timers. On larger systems (such as a 16 socket machine), this caused more than 30% of total time spent trying to obtain this kernel lock to update these group timer stats. This patch converts the timers to 64-bit atomic variables and use atomic add to update them without a lock. With this patch, the percent of total time spent updating thread group cputimer timers was reduced from 30% down to less than 1%. Note: On 32-bit systems using the generic 64-bit atomics, this causes sample_group_cputimer() to take locks 3 times instead of just 1 time. However, we tested this patch on a 32-bit system ARM system using the generic atomics and did not find the overhead to be much of an issue. An explanation for why this isn't an issue is that 32-bit systems usually have small numbers of CPUs, and cacheline contention from extra spinlocks called periodically is not really apparent on smaller systems. Signed-off-by: Jason Low Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Acked-by: Rik van Riel Cc: Andrew Morton Cc: Aswin Chandramouleeswaran Cc: Borislav Petkov Cc: Davidlohr Bueso Cc: Frederic Weisbecker Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Mel Gorman Cc: Mike Galbraith Cc: Oleg Nesterov Cc: Paul E. McKenney Cc: Peter Zijlstra Cc: Preeti U Murthy Cc: Scott J Norton Cc: Steven Rostedt Cc: Waiman Long Link: http://lkml.kernel.org/r/1430251224-5764-4-git-send-email-jason.low2@hp.com Signed-off-by: Ingo Molnar --- include/linux/init_task.h | 7 ++-- include/linux/sched.h | 10 ++---- kernel/fork.c | 3 -- kernel/sched/stats.h | 15 +++----- kernel/time/posix-cpu-timers.c | 79 ++++++++++++++++++++++++++---------------- 5 files changed, 62 insertions(+), 52 deletions(-) (limited to 'kernel') diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 696d22312b31..7b9d8b59e7bf 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -50,9 +50,10 @@ extern struct fs_struct init_fs; .cpu_timers = INIT_CPU_TIMERS(sig.cpu_timers), \ .rlim = INIT_RLIMITS, \ .cputimer = { \ - .cputime = INIT_CPUTIME, \ - .running = 0, \ - .lock = __RAW_SPIN_LOCK_UNLOCKED(sig.cputimer.lock), \ + .utime = ATOMIC64_INIT(0), \ + .stime = ATOMIC64_INIT(0), \ + .sum_exec_runtime = ATOMIC64_INIT(0), \ + .running = 0 \ }, \ .cred_guard_mutex = \ __MUTEX_INITIALIZER(sig.cred_guard_mutex), \ diff --git a/include/linux/sched.h b/include/linux/sched.h index d70910355b20..a45874c3fab6 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -598,9 +598,10 @@ struct task_cputime { * used for thread group CPU timer calculations. */ struct thread_group_cputimer { - struct task_cputime cputime; + atomic64_t utime; + atomic64_t stime; + atomic64_t sum_exec_runtime; int running; - raw_spinlock_t lock; }; #include @@ -2967,11 +2968,6 @@ static __always_inline bool need_resched(void) void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); -static inline void thread_group_cputime_init(struct signal_struct *sig) -{ - raw_spin_lock_init(&sig->cputimer.lock); -} - /* * Reevaluate whether the task has signals pending delivery. * Wake the task if so. diff --git a/kernel/fork.c b/kernel/fork.c index 47c37a411a62..2e670864174f 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1091,9 +1091,6 @@ static void posix_cpu_timers_init_group(struct signal_struct *sig) { unsigned long cpu_limit; - /* Thread group counters. */ - thread_group_cputime_init(sig); - cpu_limit = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur); if (cpu_limit != RLIM_INFINITY) { sig->cputime_expires.prof_exp = secs_to_cputime(cpu_limit); diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h index 4ab704339656..c6d1c7da3ea5 100644 --- a/kernel/sched/stats.h +++ b/kernel/sched/stats.h @@ -174,7 +174,8 @@ static inline bool cputimer_running(struct task_struct *tsk) { struct thread_group_cputimer *cputimer = &tsk->signal->cputimer; - if (!cputimer->running) + /* Check if cputimer isn't running. This is accessed without locking. */ + if (!READ_ONCE(cputimer->running)) return false; /* @@ -215,9 +216,7 @@ static inline void account_group_user_time(struct task_struct *tsk, if (!cputimer_running(tsk)) return; - raw_spin_lock(&cputimer->lock); - cputimer->cputime.utime += cputime; - raw_spin_unlock(&cputimer->lock); + atomic64_add(cputime, &cputimer->utime); } /** @@ -238,9 +237,7 @@ static inline void account_group_system_time(struct task_struct *tsk, if (!cputimer_running(tsk)) return; - raw_spin_lock(&cputimer->lock); - cputimer->cputime.stime += cputime; - raw_spin_unlock(&cputimer->lock); + atomic64_add(cputime, &cputimer->stime); } /** @@ -261,7 +258,5 @@ static inline void account_group_exec_runtime(struct task_struct *tsk, if (!cputimer_running(tsk)) return; - raw_spin_lock(&cputimer->lock); - cputimer->cputime.sum_exec_runtime += ns; - raw_spin_unlock(&cputimer->lock); + atomic64_add(ns, &cputimer->sum_exec_runtime); } diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c index e072d982f64c..d85730669410 100644 --- a/kernel/time/posix-cpu-timers.c +++ b/kernel/time/posix-cpu-timers.c @@ -196,39 +196,62 @@ static int cpu_clock_sample(const clockid_t which_clock, struct task_struct *p, return 0; } -static void update_gt_cputime(struct task_cputime *a, struct task_cputime *b) +/* + * Set cputime to sum_cputime if sum_cputime > cputime. Use cmpxchg + * to avoid race conditions with concurrent updates to cputime. + */ +static inline void __update_gt_cputime(atomic64_t *cputime, u64 sum_cputime) { - if (b->utime > a->utime) - a->utime = b->utime; + u64 curr_cputime; +retry: + curr_cputime = atomic64_read(cputime); + if (sum_cputime > curr_cputime) { + if (atomic64_cmpxchg(cputime, curr_cputime, sum_cputime) != curr_cputime) + goto retry; + } +} - if (b->stime > a->stime) - a->stime = b->stime; +static void update_gt_cputime(struct thread_group_cputimer *cputimer, struct task_cputime *sum) +{ + __update_gt_cputime(&cputimer->utime, sum->utime); + __update_gt_cputime(&cputimer->stime, sum->stime); + __update_gt_cputime(&cputimer->sum_exec_runtime, sum->sum_exec_runtime); +} - if (b->sum_exec_runtime > a->sum_exec_runtime) - a->sum_exec_runtime = b->sum_exec_runtime; +/* Sample thread_group_cputimer values in "cputimer", store results in "times". */ +static inline void sample_group_cputimer(struct task_cputime *times, + struct thread_group_cputimer *cputimer) +{ + times->utime = atomic64_read(&cputimer->utime); + times->stime = atomic64_read(&cputimer->stime); + times->sum_exec_runtime = atomic64_read(&cputimer->sum_exec_runtime); } void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times) { struct thread_group_cputimer *cputimer = &tsk->signal->cputimer; struct task_cputime sum; - unsigned long flags; - if (!cputimer->running) { + /* Check if cputimer isn't running. This is accessed without locking. */ + if (!READ_ONCE(cputimer->running)) { /* * The POSIX timer interface allows for absolute time expiry * values through the TIMER_ABSTIME flag, therefore we have - * to synchronize the timer to the clock every time we start - * it. + * to synchronize the timer to the clock every time we start it. */ thread_group_cputime(tsk, &sum); - raw_spin_lock_irqsave(&cputimer->lock, flags); - cputimer->running = 1; - update_gt_cputime(&cputimer->cputime, &sum); - } else - raw_spin_lock_irqsave(&cputimer->lock, flags); - *times = cputimer->cputime; - raw_spin_unlock_irqrestore(&cputimer->lock, flags); + update_gt_cputime(cputimer, &sum); + + /* + * We're setting cputimer->running without a lock. Ensure + * this only gets written to in one operation. We set + * running after update_gt_cputime() as a small optimization, + * but barriers are not required because update_gt_cputime() + * can handle concurrent updates. + */ + WRITE_ONCE(cputimer->running, 1); + } + sample_group_cputimer(times, cputimer); } /* @@ -582,7 +605,8 @@ bool posix_cpu_timers_can_stop_tick(struct task_struct *tsk) if (!task_cputime_zero(&tsk->cputime_expires)) return false; - if (tsk->signal->cputimer.running) + /* Check if cputimer is running. This is accessed without locking. */ + if (READ_ONCE(tsk->signal->cputimer.running)) return false; return true; @@ -882,14 +906,12 @@ static void check_thread_timers(struct task_struct *tsk, } } -static void stop_process_timers(struct signal_struct *sig) +static inline void stop_process_timers(struct signal_struct *sig) { struct thread_group_cputimer *cputimer = &sig->cputimer; - unsigned long flags; - raw_spin_lock_irqsave(&cputimer->lock, flags); - cputimer->running = 0; - raw_spin_unlock_irqrestore(&cputimer->lock, flags); + /* Turn off cputimer->running. This is done without locking. */ + WRITE_ONCE(cputimer->running, 0); } static u32 onecputick; @@ -1111,12 +1133,11 @@ static inline int fastpath_timer_check(struct task_struct *tsk) } sig = tsk->signal; - if (sig->cputimer.running) { + /* Check if cputimer is running. This is accessed without locking. */ + if (READ_ONCE(sig->cputimer.running)) { struct task_cputime group_sample; - raw_spin_lock(&sig->cputimer.lock); - group_sample = sig->cputimer.cputime; - raw_spin_unlock(&sig->cputimer.lock); + sample_group_cputimer(&group_sample, &sig->cputimer); if (task_cputime_expired(&group_sample, &sig->cputime_expires)) return 1; @@ -1157,7 +1178,7 @@ void run_posix_cpu_timers(struct task_struct *tsk) * If there are any active process wide timers (POSIX 1.b, itimers, * RLIMIT_CPU) cputimer must be running. */ - if (tsk->signal->cputimer.running) + if (READ_ONCE(tsk->signal->cputimer.running)) check_process_timers(tsk, &firing); /* -- cgit v1.2.3 From 7110744516276e906f9197e2857d026eb2343393 Mon Sep 17 00:00:00 2001 From: Jason Low Date: Tue, 28 Apr 2015 13:00:24 -0700 Subject: sched, timer: Use the atomic task_cputime in thread_group_cputimer Recent optimizations were made to thread_group_cputimer to improve its scalability by keeping track of cputime stats without a lock. However, the values were open coded to the structure, causing them to be at a different abstraction level from the regular task_cputime structure. Furthermore, any subsequent similar optimizations would not be able to share the new code, since they are specific to thread_group_cputimer. This patch adds the new task_cputime_atomic data structure (introduced in the previous patch in the series) to thread_group_cputimer for keeping track of the cputime atomically, which also helps generalize the code. Suggested-by: Ingo Molnar Signed-off-by: Jason Low Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Acked-by: Rik van Riel Cc: Andrew Morton Cc: Aswin Chandramouleeswaran Cc: Borislav Petkov Cc: Davidlohr Bueso Cc: Frederic Weisbecker Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Mel Gorman Cc: Mike Galbraith Cc: Oleg Nesterov Cc: Paul E. McKenney Cc: Peter Zijlstra Cc: Preeti U Murthy Cc: Scott J Norton Cc: Steven Rostedt Cc: Waiman Long Link: http://lkml.kernel.org/r/1430251224-5764-6-git-send-email-jason.low2@hp.com Signed-off-by: Ingo Molnar --- include/linux/init_task.h | 6 ++---- include/linux/sched.h | 4 +--- kernel/sched/stats.h | 6 +++--- kernel/time/posix-cpu-timers.c | 26 +++++++++++++------------- 4 files changed, 19 insertions(+), 23 deletions(-) (limited to 'kernel') diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 7b9d8b59e7bf..bb9b075f0eb0 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -50,10 +50,8 @@ extern struct fs_struct init_fs; .cpu_timers = INIT_CPU_TIMERS(sig.cpu_timers), \ .rlim = INIT_RLIMITS, \ .cputimer = { \ - .utime = ATOMIC64_INIT(0), \ - .stime = ATOMIC64_INIT(0), \ - .sum_exec_runtime = ATOMIC64_INIT(0), \ - .running = 0 \ + .cputime_atomic = INIT_CPUTIME_ATOMIC, \ + .running = 0, \ }, \ .cred_guard_mutex = \ __MUTEX_INITIALIZER(sig.cred_guard_mutex), \ diff --git a/include/linux/sched.h b/include/linux/sched.h index 6eb78cd45da7..4adc536a3b03 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -615,9 +615,7 @@ struct task_cputime_atomic { * used for thread group CPU timer calculations. */ struct thread_group_cputimer { - atomic64_t utime; - atomic64_t stime; - atomic64_t sum_exec_runtime; + struct task_cputime_atomic cputime_atomic; int running; }; diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h index c6d1c7da3ea5..077ebbd5e10f 100644 --- a/kernel/sched/stats.h +++ b/kernel/sched/stats.h @@ -216,7 +216,7 @@ static inline void account_group_user_time(struct task_struct *tsk, if (!cputimer_running(tsk)) return; - atomic64_add(cputime, &cputimer->utime); + atomic64_add(cputime, &cputimer->cputime_atomic.utime); } /** @@ -237,7 +237,7 @@ static inline void account_group_system_time(struct task_struct *tsk, if (!cputimer_running(tsk)) return; - atomic64_add(cputime, &cputimer->stime); + atomic64_add(cputime, &cputimer->cputime_atomic.stime); } /** @@ -258,5 +258,5 @@ static inline void account_group_exec_runtime(struct task_struct *tsk, if (!cputimer_running(tsk)) return; - atomic64_add(ns, &cputimer->sum_exec_runtime); + atomic64_add(ns, &cputimer->cputime_atomic.sum_exec_runtime); } diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c index d85730669410..892e3dae0aac 100644 --- a/kernel/time/posix-cpu-timers.c +++ b/kernel/time/posix-cpu-timers.c @@ -211,20 +211,20 @@ retry: } } -static void update_gt_cputime(struct thread_group_cputimer *cputimer, struct task_cputime *sum) +static void update_gt_cputime(struct task_cputime_atomic *cputime_atomic, struct task_cputime *sum) { - __update_gt_cputime(&cputimer->utime, sum->utime); - __update_gt_cputime(&cputimer->stime, sum->stime); - __update_gt_cputime(&cputimer->sum_exec_runtime, sum->sum_exec_runtime); + __update_gt_cputime(&cputime_atomic->utime, sum->utime); + __update_gt_cputime(&cputime_atomic->stime, sum->stime); + __update_gt_cputime(&cputime_atomic->sum_exec_runtime, sum->sum_exec_runtime); } -/* Sample thread_group_cputimer values in "cputimer", store results in "times". */ -static inline void sample_group_cputimer(struct task_cputime *times, - struct thread_group_cputimer *cputimer) +/* Sample task_cputime_atomic values in "atomic_timers", store results in "times". */ +static inline void sample_cputime_atomic(struct task_cputime *times, + struct task_cputime_atomic *atomic_times) { - times->utime = atomic64_read(&cputimer->utime); - times->stime = atomic64_read(&cputimer->stime); - times->sum_exec_runtime = atomic64_read(&cputimer->sum_exec_runtime); + times->utime = atomic64_read(&atomic_times->utime); + times->stime = atomic64_read(&atomic_times->stime); + times->sum_exec_runtime = atomic64_read(&atomic_times->sum_exec_runtime); } void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times) @@ -240,7 +240,7 @@ void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times) * to synchronize the timer to the clock every time we start it. */ thread_group_cputime(tsk, &sum); - update_gt_cputime(cputimer, &sum); + update_gt_cputime(&cputimer->cputime_atomic, &sum); /* * We're setting cputimer->running without a lock. Ensure @@ -251,7 +251,7 @@ void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times) */ WRITE_ONCE(cputimer->running, 1); } - sample_group_cputimer(times, cputimer); + sample_cputime_atomic(times, &cputimer->cputime_atomic); } /* @@ -1137,7 +1137,7 @@ static inline int fastpath_timer_check(struct task_struct *tsk) if (READ_ONCE(sig->cputimer.running)) { struct task_cputime group_sample; - sample_group_cputimer(&group_sample, &sig->cputimer); + sample_cputime_atomic(&group_sample, &sig->cputimer.cputime_atomic); if (task_cputime_expired(&group_sample, &sig->cputime_expires)) return 1; -- cgit v1.2.3 From 7675104990ed255b9315a82ae827ff312a2a88a2 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 1 May 2015 08:27:50 -0700 Subject: sched: Implement lockless wake-queues This is useful for locking primitives that can effect multiple wakeups per operation and want to avoid lock internal lock contention by delaying the wakeups until we've released the lock internal locks. Alternatively it can be used to avoid issuing multiple wakeups, and thus save a few cycles, in packet processing. Queue all target tasks and wakeup once you've processed all packets. That way you avoid waking the target task multiple times if there were multiple packets for the same task. Properties of a wake_q are: - Lockless, as queue head must reside on the stack. - Being a queue, maintains wakeup order passed by the callers. This can be important for otherwise, in scenarios where highly contended locks could affect any reliance on lock fairness. - A queued task cannot be added again until it is woken up. This patch adds the needed infrastructure into the scheduler code and uses the new wake_list to delay the futex wakeups until after we've released the hash bucket locks. Signed-off-by: Peter Zijlstra (Intel) [tweaks, adjustments, comments, etc.] Signed-off-by: Davidlohr Bueso Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Cc: Borislav Petkov Cc: Chris Mason Cc: Davidlohr Bueso Cc: George Spelvin Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Manfred Spraul Cc: Sebastian Andrzej Siewior Cc: Steven Rostedt Link: http://lkml.kernel.org/r/1430494072-30283-2-git-send-email-dave@stgolabs.net Signed-off-by: Ingo Molnar --- include/linux/sched.h | 46 ++++++++++++++++++++++++++++++++++++++++++++++ kernel/sched/core.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 92 insertions(+) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 4adc536a3b03..254d88e80f65 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -920,6 +920,50 @@ enum cpu_idle_type { #define SCHED_CAPACITY_SHIFT 10 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT) +/* + * Wake-queues are lists of tasks with a pending wakeup, whose + * callers have already marked the task as woken internally, + * and can thus carry on. A common use case is being able to + * do the wakeups once the corresponding user lock as been + * released. + * + * We hold reference to each task in the list across the wakeup, + * thus guaranteeing that the memory is still valid by the time + * the actual wakeups are performed in wake_up_q(). + * + * One per task suffices, because there's never a need for a task to be + * in two wake queues simultaneously; it is forbidden to abandon a task + * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is + * already in a wake queue, the wakeup will happen soon and the second + * waker can just skip it. + * + * The WAKE_Q macro declares and initializes the list head. + * wake_up_q() does NOT reinitialize the list; it's expected to be + * called near the end of a function, where the fact that the queue is + * not used again will be easy to see by inspection. + * + * Note that this can cause spurious wakeups. schedule() callers + * must ensure the call is done inside a loop, confirming that the + * wakeup condition has in fact occurred. + */ +struct wake_q_node { + struct wake_q_node *next; +}; + +struct wake_q_head { + struct wake_q_node *first; + struct wake_q_node **lastp; +}; + +#define WAKE_Q_TAIL ((struct wake_q_node *) 0x01) + +#define WAKE_Q(name) \ + struct wake_q_head name = { WAKE_Q_TAIL, &name.first } + +extern void wake_q_add(struct wake_q_head *head, + struct task_struct *task); +extern void wake_up_q(struct wake_q_head *head); + /* * sched-domains (multiprocessor balancing) declarations: */ @@ -1532,6 +1576,8 @@ struct task_struct { /* Protection of the PI data structures: */ raw_spinlock_t pi_lock; + struct wake_q_node wake_q; + #ifdef CONFIG_RT_MUTEXES /* PI waiters blocked on a rt_mutex held by this task */ struct rb_root pi_waiters; diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 22b53c863ef3..355f9538ca33 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -541,6 +541,52 @@ static bool set_nr_if_polling(struct task_struct *p) #endif #endif +void wake_q_add(struct wake_q_head *head, struct task_struct *task) +{ + struct wake_q_node *node = &task->wake_q; + + /* + * Atomically grab the task, if ->wake_q is !nil already it means + * its already queued (either by us or someone else) and will get the + * wakeup due to that. + * + * This cmpxchg() implies a full barrier, which pairs with the write + * barrier implied by the wakeup in wake_up_list(). + */ + if (cmpxchg(&node->next, NULL, WAKE_Q_TAIL)) + return; + + get_task_struct(task); + + /* + * The head is context local, there can be no concurrency. + */ + *head->lastp = node; + head->lastp = &node->next; +} + +void wake_up_q(struct wake_q_head *head) +{ + struct wake_q_node *node = head->first; + + while (node != WAKE_Q_TAIL) { + struct task_struct *task; + + task = container_of(node, struct task_struct, wake_q); + BUG_ON(!task); + /* task can safely be re-inserted now */ + node = node->next; + task->wake_q.next = NULL; + + /* + * wake_up_process() implies a wmb() to pair with the queueing + * in wake_q_add() so as not to miss wakeups. + */ + wake_up_process(task); + put_task_struct(task); + } +} + /* * resched_curr - mark rq's current task 'to be rescheduled now'. * -- cgit v1.2.3 From 1d0dcb3ad9d336e6d6ee020a750a7f8d907e28de Mon Sep 17 00:00:00 2001 From: Davidlohr Bueso Date: Fri, 1 May 2015 08:27:51 -0700 Subject: futex: Implement lockless wakeups Given the overall futex architecture, any chance of reducing hb->lock contention is welcome. In this particular case, using wake-queues to enable lockless wakeups addresses very much real world performance concerns, even cases of soft-lockups in cases of large amounts of blocked tasks (which is not hard to find in large boxes, using but just a handful of futex). At the lowest level, this patch can reduce latency of a single thread attempting to acquire hb->lock in highly contended scenarios by a up to 2x. At lower counts of nr_wake there are no regressions, confirming, of course, that the wake_q handling overhead is practically non existent. For instance, while a fair amount of variation, the extended pef-bench wakeup benchmark shows for a 20 core machine the following avg per-thread time to wakeup its share of tasks: nr_thr ms-before ms-after 16 0.0590 0.0215 32 0.0396 0.0220 48 0.0417 0.0182 64 0.0536 0.0236 80 0.0414 0.0097 96 0.0672 0.0152 Naturally, this can cause spurious wakeups. However there is no core code that cannot handle them afaict, and furthermore tglx does have the point that other events can already trigger them anyway. Signed-off-by: Davidlohr Bueso Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Cc: Andrew Morton Cc: Borislav Petkov Cc: Chris Mason Cc: Davidlohr Bueso Cc: George Spelvin Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Manfred Spraul Cc: Peter Zijlstra Cc: Sebastian Andrzej Siewior Cc: Steven Rostedt Link: http://lkml.kernel.org/r/1430494072-30283-3-git-send-email-dave@stgolabs.net Signed-off-by: Ingo Molnar --- kernel/futex.c | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 2579e407ff67..f9984c363e9a 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -1090,9 +1090,11 @@ static void __unqueue_futex(struct futex_q *q) /* * The hash bucket lock must be held when this is called. - * Afterwards, the futex_q must not be accessed. + * Afterwards, the futex_q must not be accessed. Callers + * must ensure to later call wake_up_q() for the actual + * wakeups to occur. */ -static void wake_futex(struct futex_q *q) +static void mark_wake_futex(struct wake_q_head *wake_q, struct futex_q *q) { struct task_struct *p = q->task; @@ -1100,14 +1102,10 @@ static void wake_futex(struct futex_q *q) return; /* - * We set q->lock_ptr = NULL _before_ we wake up the task. If - * a non-futex wake up happens on another CPU then the task - * might exit and p would dereference a non-existing task - * struct. Prevent this by holding a reference on p across the - * wake up. + * Queue the task for later wakeup for after we've released + * the hb->lock. wake_q_add() grabs reference to p. */ - get_task_struct(p); - + wake_q_add(wake_q, p); __unqueue_futex(q); /* * The waiting task can free the futex_q as soon as @@ -1117,9 +1115,6 @@ static void wake_futex(struct futex_q *q) */ smp_wmb(); q->lock_ptr = NULL; - - wake_up_state(p, TASK_NORMAL); - put_task_struct(p); } static int wake_futex_pi(u32 __user *uaddr, u32 uval, struct futex_q *this) @@ -1217,6 +1212,7 @@ futex_wake(u32 __user *uaddr, unsigned int flags, int nr_wake, u32 bitset) struct futex_q *this, *next; union futex_key key = FUTEX_KEY_INIT; int ret; + WAKE_Q(wake_q); if (!bitset) return -EINVAL; @@ -1244,13 +1240,14 @@ futex_wake(u32 __user *uaddr, unsigned int flags, int nr_wake, u32 bitset) if (!(this->bitset & bitset)) continue; - wake_futex(this); + mark_wake_futex(&wake_q, this); if (++ret >= nr_wake) break; } } spin_unlock(&hb->lock); + wake_up_q(&wake_q); out_put_key: put_futex_key(&key); out: @@ -1269,6 +1266,7 @@ futex_wake_op(u32 __user *uaddr1, unsigned int flags, u32 __user *uaddr2, struct futex_hash_bucket *hb1, *hb2; struct futex_q *this, *next; int ret, op_ret; + WAKE_Q(wake_q); retry: ret = get_futex_key(uaddr1, flags & FLAGS_SHARED, &key1, VERIFY_READ); @@ -1320,7 +1318,7 @@ retry_private: ret = -EINVAL; goto out_unlock; } - wake_futex(this); + mark_wake_futex(&wake_q, this); if (++ret >= nr_wake) break; } @@ -1334,7 +1332,7 @@ retry_private: ret = -EINVAL; goto out_unlock; } - wake_futex(this); + mark_wake_futex(&wake_q, this); if (++op_ret >= nr_wake2) break; } @@ -1344,6 +1342,7 @@ retry_private: out_unlock: double_unlock_hb(hb1, hb2); + wake_up_q(&wake_q); out_put_keys: put_futex_key(&key2); out_put_key1: @@ -1503,6 +1502,7 @@ static int futex_requeue(u32 __user *uaddr1, unsigned int flags, struct futex_pi_state *pi_state = NULL; struct futex_hash_bucket *hb1, *hb2; struct futex_q *this, *next; + WAKE_Q(wake_q); if (requeue_pi) { /* @@ -1679,7 +1679,7 @@ retry_private: * woken by futex_unlock_pi(). */ if (++task_count <= nr_wake && !requeue_pi) { - wake_futex(this); + mark_wake_futex(&wake_q, this); continue; } @@ -1719,6 +1719,7 @@ retry_private: out_unlock: free_pi_state(pi_state); double_unlock_hb(hb1, hb2); + wake_up_q(&wake_q); hb_waiters_dec(hb2); /* -- cgit v1.2.3 From 58ac93e4f2e4b15beffdf0e3749b7fea3208ef66 Mon Sep 17 00:00:00 2001 From: Nicholas Mc Guire Date: Fri, 15 May 2015 21:05:42 +0200 Subject: sched: Fix function declaration return type mismatch static code checking was unhappy with: ./kernel/sched/fair.c:162 WARNING: return of wrong type int != unsigned int get_update_sysctl_factor() is declared to return int but is currently returning an unsigned int. The first few preprocessed lines are: static int get_update_sysctl_factor(void) { unsigned int cpus = ({ int __min1 = (cpumask_weight(cpu_online_mask)); int __min2 = (8); __min1 < __min2 ? __min1: __min2; }); unsigned int factor; The type used by min_t() should be 'unsigned int' and the return type of get_update_sysctl_factor() should also be 'unsigned int' as its call-site update_sysctl() is expecting 'unsigned int' and the values utilizing: 'factor' 'sysctl_sched_min_granularity' 'sched_nr_latency' 'sysctl_sched_wakeup_granularity' ... are also all 'unsigned int', plus cpumask_weight() is also returning 'unsigned int'. So the natural type to use around here is 'unsigned int'. ( Patch was compile tested with x86_64_defconfig + CONFIG_SCHED_DEBUG=y and the changed sections in kernel/sched/fair.i were reviewed. ) Signed-off-by: Nicholas Mc Guire [ Improved the changelog a bit. ] Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431716742-11077-1-git-send-email-hofrat@osadl.org Signed-off-by: Ingo Molnar --- kernel/sched/fair.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index f18ddb72fe88..a27d9883f8ba 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -141,9 +141,9 @@ static inline void update_load_set(struct load_weight *lw, unsigned long w) * * This idea comes from the SD scheduler of Con Kolivas: */ -static int get_update_sysctl_factor(void) +static unsigned int get_update_sysctl_factor(void) { - unsigned int cpus = min_t(int, num_online_cpus(), 8); + unsigned int cpus = min_t(unsigned int, num_online_cpus(), 8); unsigned int factor; switch (sysctl_sched_tunable_scaling) { @@ -576,7 +576,7 @@ int sched_proc_update_handler(struct ctl_table *table, int write, loff_t *ppos) { int ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); - int factor = get_update_sysctl_factor(); + unsigned int factor = get_update_sysctl_factor(); if (ret || !write) return ret; -- cgit v1.2.3 From b30f0e3ffedfa52b1d67a302ae5860c49998e5e2 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 12 May 2015 16:41:49 +0200 Subject: sched/preempt: Optimize preemption operations on __schedule() callers __schedule() disables preemption and some of its callers (the preempt_schedule*() family) also set PREEMPT_ACTIVE. So we have two preempt_count() modifications that could be performed at once. Lets remove the preemption disablement from __schedule() and pull this responsibility to its callers in order to optimize preempt_count() operations in a single place. Suggested-by: Linus Torvalds Signed-off-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra (Intel) Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431441711-29753-5-git-send-email-fweisbec@gmail.com Signed-off-by: Ingo Molnar --- include/linux/preempt.h | 12 ++++++++++++ kernel/sched/core.c | 29 +++++++++-------------------- 2 files changed, 21 insertions(+), 20 deletions(-) (limited to 'kernel') diff --git a/include/linux/preempt.h b/include/linux/preempt.h index 4689ef210a13..45da394f2779 100644 --- a/include/linux/preempt.h +++ b/include/linux/preempt.h @@ -137,6 +137,18 @@ extern void preempt_count_sub(int val); #define preempt_count_inc() preempt_count_add(1) #define preempt_count_dec() preempt_count_sub(1) +#define preempt_active_enter() \ +do { \ + preempt_count_add(PREEMPT_ACTIVE + PREEMPT_DISABLE_OFFSET); \ + barrier(); \ +} while (0) + +#define preempt_active_exit() \ +do { \ + barrier(); \ + preempt_count_sub(PREEMPT_ACTIVE + PREEMPT_DISABLE_OFFSET); \ +} while (0) + #ifdef CONFIG_PREEMPT_COUNT #define preempt_disable() \ diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 355f9538ca33..5140db62c621 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2773,9 +2773,7 @@ again: * - return from syscall or exception to user-space * - return from interrupt-handler to user-space * - * WARNING: all callers must re-check need_resched() afterward and reschedule - * accordingly in case an event triggered the need for rescheduling (such as - * an interrupt waking up a task) while preemption was disabled in __schedule(). + * WARNING: must be called with preemption disabled! */ static void __sched __schedule(void) { @@ -2784,7 +2782,6 @@ static void __sched __schedule(void) struct rq *rq; int cpu; - preempt_disable(); cpu = smp_processor_id(); rq = cpu_rq(cpu); rcu_note_context_switch(); @@ -2848,8 +2845,6 @@ static void __sched __schedule(void) raw_spin_unlock_irq(&rq->lock); post_schedule(rq); - - sched_preempt_enable_no_resched(); } static inline void sched_submit_work(struct task_struct *tsk) @@ -2870,7 +2865,9 @@ asmlinkage __visible void __sched schedule(void) sched_submit_work(tsk); do { + preempt_disable(); __schedule(); + sched_preempt_enable_no_resched(); } while (need_resched()); } EXPORT_SYMBOL(schedule); @@ -2909,15 +2906,14 @@ void __sched schedule_preempt_disabled(void) static void __sched notrace preempt_schedule_common(void) { do { - __preempt_count_add(PREEMPT_ACTIVE); + preempt_active_enter(); __schedule(); - __preempt_count_sub(PREEMPT_ACTIVE); + preempt_active_exit(); /* * Check again in case we missed a preemption opportunity * between schedule and now. */ - barrier(); } while (need_resched()); } @@ -2964,7 +2960,7 @@ asmlinkage __visible void __sched notrace preempt_schedule_context(void) return; do { - __preempt_count_add(PREEMPT_ACTIVE); + preempt_active_enter(); /* * Needs preempt disabled in case user_exit() is traced * and the tracer calls preempt_enable_notrace() causing @@ -2974,8 +2970,7 @@ asmlinkage __visible void __sched notrace preempt_schedule_context(void) __schedule(); exception_exit(prev_ctx); - __preempt_count_sub(PREEMPT_ACTIVE); - barrier(); + preempt_active_exit(); } while (need_resched()); } EXPORT_SYMBOL_GPL(preempt_schedule_context); @@ -2999,17 +2994,11 @@ asmlinkage __visible void __sched preempt_schedule_irq(void) prev_state = exception_enter(); do { - __preempt_count_add(PREEMPT_ACTIVE); + preempt_active_enter(); local_irq_enable(); __schedule(); local_irq_disable(); - __preempt_count_sub(PREEMPT_ACTIVE); - - /* - * Check again in case we missed a preemption opportunity - * between schedule and now. - */ - barrier(); + preempt_active_exit(); } while (need_resched()); exception_exit(prev_state); -- cgit v1.2.3 From 8bcbde5480f9777f8b74d71493722c663e22c21b Mon Sep 17 00:00:00 2001 From: David Hildenbrand Date: Mon, 11 May 2015 17:52:06 +0200 Subject: sched/preempt, mm/fault: Count pagefault_disable() levels in pagefault_disabled Until now, pagefault_disable()/pagefault_enabled() used the preempt count to track whether in an environment with pagefaults disabled (can be queried via in_atomic()). This patch introduces a separate counter in task_struct to count the level of pagefault_disable() calls. We'll keep manipulating the preempt count to retain compatibility to existing pagefault handlers. It is now possible to verify whether in a pagefault_disable() envionment by calling pagefault_disabled(). In contrast to in_atomic() it will not be influenced by preempt_enable()/preempt_disable(). This patch is based on a patch from Ingo Molnar. Reviewed-and-tested-by: Thomas Gleixner Signed-off-by: David Hildenbrand Signed-off-by: Peter Zijlstra (Intel) Cc: David.Laight@ACULAB.COM Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: airlied@linux.ie Cc: akpm@linux-foundation.org Cc: benh@kernel.crashing.org Cc: bigeasy@linutronix.de Cc: borntraeger@de.ibm.com Cc: daniel.vetter@intel.com Cc: heiko.carstens@de.ibm.com Cc: herbert@gondor.apana.org.au Cc: hocko@suse.cz Cc: hughd@google.com Cc: mst@redhat.com Cc: paulus@samba.org Cc: ralf@linux-mips.org Cc: schwidefsky@de.ibm.com Cc: yang.shi@windriver.com Link: http://lkml.kernel.org/r/1431359540-32227-2-git-send-email-dahi@linux.vnet.ibm.com Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 + include/linux/uaccess.h | 36 +++++++++++++++++++++++++++++------- kernel/fork.c | 3 +++ 3 files changed, 33 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index c53a1784d7a9..dd07ac03f82a 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1788,6 +1788,7 @@ struct task_struct { #ifdef CONFIG_DEBUG_ATOMIC_SLEEP unsigned long task_state_change; #endif + int pagefault_disabled; }; /* Future-safe accessor for struct task_struct's cpus_allowed. */ diff --git a/include/linux/uaccess.h b/include/linux/uaccess.h index ecd3319dac33..23290cc93a24 100644 --- a/include/linux/uaccess.h +++ b/include/linux/uaccess.h @@ -2,20 +2,36 @@ #define __LINUX_UACCESS_H__ #include +#include #include +static __always_inline void pagefault_disabled_inc(void) +{ + current->pagefault_disabled++; +} + +static __always_inline void pagefault_disabled_dec(void) +{ + current->pagefault_disabled--; + WARN_ON(current->pagefault_disabled < 0); +} + /* - * These routines enable/disable the pagefault handler in that - * it will not take any locks and go straight to the fixup table. + * These routines enable/disable the pagefault handler. If disabled, it will + * not take any locks and go straight to the fixup table. + * + * We increase the preempt and the pagefault count, to be able to distinguish + * whether we run in simple atomic context or in a real pagefault_disable() + * context. + * + * For now, after pagefault_disabled() has been called, we run in atomic + * context. User access methods will not sleep. * - * They have great resemblance to the preempt_disable/enable calls - * and in fact they are identical; this is because currently there is - * no other way to make the pagefault handlers do this. So we do - * disable preemption but we don't necessarily care about that. */ static inline void pagefault_disable(void) { preempt_count_inc(); + pagefault_disabled_inc(); /* * make sure to have issued the store before a pagefault * can hit. @@ -25,18 +41,24 @@ static inline void pagefault_disable(void) static inline void pagefault_enable(void) { -#ifndef CONFIG_PREEMPT /* * make sure to issue those last loads/stores before enabling * the pagefault handler again. */ barrier(); + pagefault_disabled_dec(); +#ifndef CONFIG_PREEMPT preempt_count_dec(); #else preempt_enable(); #endif } +/* + * Is the pagefault handler disabled? If so, user access methods will not sleep. + */ +#define pagefault_disabled() (current->pagefault_disabled != 0) + #ifndef ARCH_HAS_NOCACHE_UACCESS static inline unsigned long __copy_from_user_inatomic_nocache(void *to, diff --git a/kernel/fork.c b/kernel/fork.c index 2e670864174f..0bb88b555550 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1393,6 +1393,9 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->hardirq_context = 0; p->softirq_context = 0; #endif + + p->pagefault_disabled = 0; + #ifdef CONFIG_LOCKDEP p->lockdep_depth = 0; /* no locks held yet */ p->curr_chain_key = 0; -- cgit v1.2.3 From c1ceac6276e4ee12e4129afd380db10fae0db7df Mon Sep 17 00:00:00 2001 From: Rik van Riel Date: Thu, 14 May 2015 22:59:36 -0400 Subject: sched/numa: Reduce conflict between fbq_classify_rq() and migration It is possible for fbq_classify_rq() to indicate that a CPU has tasks that should be moved to another NUMA node, but for migrate_improves_locality and migrate_degrades_locality to not identify those tasks. This patch always gives preference to preferred node evaluations, and only checks the number of faults when evaluating moves between two non-preferred nodes on a larger NUMA system. On a two node system, the number of faults is never evaluated. Either a task is about to be pulled off its preferred node, or migrated onto it. Signed-off-by: Rik van Riel Signed-off-by: Peter Zijlstra (Intel) Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: mgorman@suse.de Link: http://lkml.kernel.org/r/20150514225936.35b91717@annuminas.surriel.com Signed-off-by: Ingo Molnar --- kernel/sched/fair.c | 60 +++++++++++++++++++++++++++++------------------------ 1 file changed, 33 insertions(+), 27 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index a27d9883f8ba..0d4632f7799b 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -5663,10 +5663,15 @@ static int task_hot(struct task_struct *p, struct lb_env *env) } #ifdef CONFIG_NUMA_BALANCING -/* Returns true if the destination node has incurred more faults */ +/* + * Returns true if the destination node is the preferred node. + * Needs to match fbq_classify_rq(): if there is a runnable task + * that is not on its preferred node, we should identify it. + */ static bool migrate_improves_locality(struct task_struct *p, struct lb_env *env) { struct numa_group *numa_group = rcu_dereference(p->numa_group); + unsigned long src_faults, dst_faults; int src_nid, dst_nid; if (!sched_feat(NUMA_FAVOUR_HIGHER) || !p->numa_faults || @@ -5680,29 +5685,30 @@ static bool migrate_improves_locality(struct task_struct *p, struct lb_env *env) if (src_nid == dst_nid) return false; - if (numa_group) { - /* Task is already in the group's interleave set. */ - if (node_isset(src_nid, numa_group->active_nodes)) - return false; - - /* Task is moving into the group's interleave set. */ - if (node_isset(dst_nid, numa_group->active_nodes)) - return true; - - return group_faults(p, dst_nid) > group_faults(p, src_nid); - } - /* Encourage migration to the preferred node. */ if (dst_nid == p->numa_preferred_nid) return true; - return task_faults(p, dst_nid) > task_faults(p, src_nid); + /* Migrating away from the preferred node is bad. */ + if (src_nid == p->numa_preferred_nid) + return false; + + if (numa_group) { + src_faults = group_faults(p, src_nid); + dst_faults = group_faults(p, dst_nid); + } else { + src_faults = task_faults(p, src_nid); + dst_faults = task_faults(p, dst_nid); + } + + return dst_faults > src_faults; } static bool migrate_degrades_locality(struct task_struct *p, struct lb_env *env) { struct numa_group *numa_group = rcu_dereference(p->numa_group); + unsigned long src_faults, dst_faults; int src_nid, dst_nid; if (!sched_feat(NUMA) || !sched_feat(NUMA_RESIST_LOWER)) @@ -5717,23 +5723,23 @@ static bool migrate_degrades_locality(struct task_struct *p, struct lb_env *env) if (src_nid == dst_nid) return false; - if (numa_group) { - /* Task is moving within/into the group's interleave set. */ - if (node_isset(dst_nid, numa_group->active_nodes)) - return false; + /* Migrating away from the preferred node is bad. */ + if (src_nid == p->numa_preferred_nid) + return true; - /* Task is moving out of the group's interleave set. */ - if (node_isset(src_nid, numa_group->active_nodes)) - return true; + /* Encourage migration to the preferred node. */ + if (dst_nid == p->numa_preferred_nid) + return false; - return group_faults(p, dst_nid) < group_faults(p, src_nid); + if (numa_group) { + src_faults = group_faults(p, src_nid); + dst_faults = group_faults(p, dst_nid); + } else { + src_faults = task_faults(p, src_nid); + dst_faults = task_faults(p, dst_nid); } - /* Migrating away from the preferred node is always bad. */ - if (src_nid == p->numa_preferred_nid) - return true; - - return task_faults(p, dst_nid) < task_faults(p, src_nid); + return dst_faults < src_faults; } #else -- cgit v1.2.3 From be690035df893385ceaac2323b29be1fb7f2a67f Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 4 Jun 2015 17:39:07 +0200 Subject: sched: Make preempt_schedule_context() function-tracing safe Since function tracing disables preemption, it needs a safe preemption point to use when preemption is re-enabled without worrying about tracing recursion. Ie: to avoid tracing recursion, that preemption point can't be traced (use of notrace qualifier) and it can't call any traceable function before that preemption point disables preemption itself, which disarms the recursion. preempt_schedule() was fine until commit: b30f0e3ffedf ("sched/preempt: Optimize preemption operations on __schedule() callers") because PREEMPT_ACTIVE (which has the property to disable preemption and this disarm tracing preemption recursion) was set before calling any further function. But that commit introduced the use of preempt_count_add/sub() functions to set PREEMPT_ACTIVE and because these functions are called before preemption gets a chance to be disabled, we have a tracing recursion. preempt_schedule_context() is one of the possible preemption functions used by tracing. Its special purpose is to avoid tracing recursion against context tracking. Lets enhance this function to become more generally tracing safe by disabling preemption with raw accessors, such that no function is called before preemption gets disabled and disarm the tracing recursion. This function is going to become the specific tracing-safe preemption point in further commit. Reported-by: Fengguang Wu Signed-off-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Steven Rostedt Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1433432349-1021-2-git-send-email-fweisbec@gmail.com Signed-off-by: Ingo Molnar --- kernel/sched/core.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 20b858f2db22..4e925ea10c0c 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2960,7 +2960,13 @@ asmlinkage __visible void __sched notrace preempt_schedule_context(void) return; do { - preempt_active_enter(); + /* + * Use raw __prempt_count() ops that don't call function. + * We can't call functions before disabling preemption which + * disarm preemption tracing recursions. + */ + __preempt_count_add(PREEMPT_ACTIVE + PREEMPT_DISABLE_OFFSET); + barrier(); /* * Needs preempt disabled in case user_exit() is traced * and the tracer calls preempt_enable_notrace() causing @@ -2970,7 +2976,8 @@ asmlinkage __visible void __sched notrace preempt_schedule_context(void) __schedule(); exception_exit(prev_ctx); - preempt_active_exit(); + barrier(); + __preempt_count_sub(PREEMPT_ACTIVE + PREEMPT_DISABLE_OFFSET); } while (need_resched()); } EXPORT_SYMBOL_GPL(preempt_schedule_context); -- cgit v1.2.3 From 4eaca0a887eaee04fc7a3866d0f5b51b34030dfa Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 4 Jun 2015 17:39:08 +0200 Subject: preempt: Use preempt_schedule_context() as the official tracing preemption point preempt_schedule_context() is a tracing safe preemption point but it's only used when CONFIG_CONTEXT_TRACKING=y. Other configs have tracing recursion issues since commit: b30f0e3ffedf ("sched/preempt: Optimize preemption operations on __schedule() callers") introduced function based preemp_count_*() ops. Lets make it available on all configs and give it a more appropriate name for its new position. Reported-by: Fengguang Wu Signed-off-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Steven Rostedt Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1433432349-1021-3-git-send-email-fweisbec@gmail.com Signed-off-by: Ingo Molnar --- arch/x86/include/asm/preempt.h | 8 +++----- arch/x86/kernel/i386_ksyms_32.c | 4 +--- arch/x86/kernel/x8664_ksyms_64.c | 4 +--- arch/x86/lib/thunk_32.S | 4 +--- arch/x86/lib/thunk_64.S | 4 +--- include/asm-generic/preempt.h | 7 ++----- include/linux/preempt.h | 6 +----- kernel/sched/core.c | 8 +++----- 8 files changed, 13 insertions(+), 32 deletions(-) (limited to 'kernel') diff --git a/arch/x86/include/asm/preempt.h b/arch/x86/include/asm/preempt.h index 8f3271842533..dca71714f860 100644 --- a/arch/x86/include/asm/preempt.h +++ b/arch/x86/include/asm/preempt.h @@ -99,11 +99,9 @@ static __always_inline bool should_resched(void) extern asmlinkage void ___preempt_schedule(void); # define __preempt_schedule() asm ("call ___preempt_schedule") extern asmlinkage void preempt_schedule(void); -# ifdef CONFIG_CONTEXT_TRACKING - extern asmlinkage void ___preempt_schedule_context(void); -# define __preempt_schedule_context() asm ("call ___preempt_schedule_context") - extern asmlinkage void preempt_schedule_context(void); -# endif + extern asmlinkage void ___preempt_schedule_notrace(void); +# define __preempt_schedule_notrace() asm ("call ___preempt_schedule_notrace") + extern asmlinkage void preempt_schedule_notrace(void); #endif #endif /* __ASM_PREEMPT_H */ diff --git a/arch/x86/kernel/i386_ksyms_32.c b/arch/x86/kernel/i386_ksyms_32.c index 05fd74f537d6..64341aa485ae 100644 --- a/arch/x86/kernel/i386_ksyms_32.c +++ b/arch/x86/kernel/i386_ksyms_32.c @@ -40,7 +40,5 @@ EXPORT_SYMBOL(empty_zero_page); #ifdef CONFIG_PREEMPT EXPORT_SYMBOL(___preempt_schedule); -#ifdef CONFIG_CONTEXT_TRACKING -EXPORT_SYMBOL(___preempt_schedule_context); -#endif +EXPORT_SYMBOL(___preempt_schedule_notrace); #endif diff --git a/arch/x86/kernel/x8664_ksyms_64.c b/arch/x86/kernel/x8664_ksyms_64.c index 37d8fa4438f0..a0695be19864 100644 --- a/arch/x86/kernel/x8664_ksyms_64.c +++ b/arch/x86/kernel/x8664_ksyms_64.c @@ -75,7 +75,5 @@ EXPORT_SYMBOL(native_load_gs_index); #ifdef CONFIG_PREEMPT EXPORT_SYMBOL(___preempt_schedule); -#ifdef CONFIG_CONTEXT_TRACKING -EXPORT_SYMBOL(___preempt_schedule_context); -#endif +EXPORT_SYMBOL(___preempt_schedule_notrace); #endif diff --git a/arch/x86/lib/thunk_32.S b/arch/x86/lib/thunk_32.S index 5eb715087b80..e407941d0488 100644 --- a/arch/x86/lib/thunk_32.S +++ b/arch/x86/lib/thunk_32.S @@ -38,8 +38,6 @@ #ifdef CONFIG_PREEMPT THUNK ___preempt_schedule, preempt_schedule -#ifdef CONFIG_CONTEXT_TRACKING - THUNK ___preempt_schedule_context, preempt_schedule_context -#endif + THUNK ___preempt_schedule_notrace, preempt_schedule_notrace #endif diff --git a/arch/x86/lib/thunk_64.S b/arch/x86/lib/thunk_64.S index f89ba4e93025..2198902329b5 100644 --- a/arch/x86/lib/thunk_64.S +++ b/arch/x86/lib/thunk_64.S @@ -49,9 +49,7 @@ #ifdef CONFIG_PREEMPT THUNK ___preempt_schedule, preempt_schedule -#ifdef CONFIG_CONTEXT_TRACKING - THUNK ___preempt_schedule_context, preempt_schedule_context -#endif + THUNK ___preempt_schedule_notrace, preempt_schedule_notrace #endif #if defined(CONFIG_TRACE_IRQFLAGS) \ diff --git a/include/asm-generic/preempt.h b/include/asm-generic/preempt.h index eb6f9e6c3075..d0a7a4753db2 100644 --- a/include/asm-generic/preempt.h +++ b/include/asm-generic/preempt.h @@ -79,11 +79,8 @@ static __always_inline bool should_resched(void) #ifdef CONFIG_PREEMPT extern asmlinkage void preempt_schedule(void); #define __preempt_schedule() preempt_schedule() - -#ifdef CONFIG_CONTEXT_TRACKING -extern asmlinkage void preempt_schedule_context(void); -#define __preempt_schedule_context() preempt_schedule_context() -#endif +extern asmlinkage void preempt_schedule_notrace(void); +#define __preempt_schedule_notrace() preempt_schedule_notrace() #endif /* CONFIG_PREEMPT */ #endif /* __ASM_PREEMPT_H */ diff --git a/include/linux/preempt.h b/include/linux/preempt.h index a1a00e14c14f..7686dd63bc35 100644 --- a/include/linux/preempt.h +++ b/include/linux/preempt.h @@ -204,15 +204,11 @@ do { \ #ifdef CONFIG_PREEMPT -#ifndef CONFIG_CONTEXT_TRACKING -#define __preempt_schedule_context() __preempt_schedule() -#endif - #define preempt_enable_notrace() \ do { \ barrier(); \ if (unlikely(__preempt_count_dec_and_test())) \ - __preempt_schedule_context(); \ + __preempt_schedule_notrace(); \ } while (0) #else #define preempt_enable_notrace() \ diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 4e925ea10c0c..af0a5a6cee98 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2937,9 +2937,8 @@ asmlinkage __visible void __sched notrace preempt_schedule(void) NOKPROBE_SYMBOL(preempt_schedule); EXPORT_SYMBOL(preempt_schedule); -#ifdef CONFIG_CONTEXT_TRACKING /** - * preempt_schedule_context - preempt_schedule called by tracing + * preempt_schedule_notrace - preempt_schedule called by tracing * * The tracing infrastructure uses preempt_enable_notrace to prevent * recursion and tracing preempt enabling caused by the tracing @@ -2952,7 +2951,7 @@ EXPORT_SYMBOL(preempt_schedule); * instead of preempt_schedule() to exit user context if needed before * calling the scheduler. */ -asmlinkage __visible void __sched notrace preempt_schedule_context(void) +asmlinkage __visible void __sched notrace preempt_schedule_notrace(void) { enum ctx_state prev_ctx; @@ -2980,8 +2979,7 @@ asmlinkage __visible void __sched notrace preempt_schedule_context(void) __preempt_count_sub(PREEMPT_ACTIVE + PREEMPT_DISABLE_OFFSET); } while (need_resched()); } -EXPORT_SYMBOL_GPL(preempt_schedule_context); -#endif /* CONFIG_CONTEXT_TRACKING */ +EXPORT_SYMBOL_GPL(preempt_schedule_notrace); #endif /* CONFIG_PREEMPT */ -- cgit v1.2.3 From 54d27365cae88fbcc853b391dcd561e71acb81fa Mon Sep 17 00:00:00 2001 From: Ben Segall Date: Mon, 6 Apr 2015 15:28:10 -0700 Subject: sched/fair: Prevent throttling in early pick_next_task_fair() The optimized task selection logic optimistically selects a new task to run without first doing a full put_prev_task(). This is so that we can avoid a put/set on the common ancestors of the old and new task. Similarly, we should only call check_cfs_rq_runtime() to throttle eligible groups if they're part of the common ancestry, otherwise it is possible to end up with no eligible task in the simple task selection. Imagine: /root /prev /next /A /B If our optimistic selection ends up throttling /next, we goto simple and our put_prev_task() ends up throttling /prev, after which we're going to bug out in set_next_entity() because there aren't any tasks left. Avoid this scenario by only throttling common ancestors. Reported-by: Mohammed Naser Reported-by: Konstantin Khlebnikov Signed-off-by: Ben Segall [ munged Changelog ] Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Roman Gushchin Cc: Thomas Gleixner Cc: pjt@google.com Fixes: 678d5718d8d0 ("sched/fair: Optimize cgroup pick_next_task_fair()") Link: http://lkml.kernel.org/r/xm26wq1oswoq.fsf@sword-of-the-dawn.mtv.corp.google.com Signed-off-by: Ingo Molnar --- kernel/sched/fair.c | 25 ++++++++++++++----------- 1 file changed, 14 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 0d4632f7799b..84ada054c6a8 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -5322,18 +5322,21 @@ again: * entity, update_curr() will update its vruntime, otherwise * forget we've ever seen it. */ - if (curr && curr->on_rq) - update_curr(cfs_rq); - else - curr = NULL; + if (curr) { + if (curr->on_rq) + update_curr(cfs_rq); + else + curr = NULL; - /* - * This call to check_cfs_rq_runtime() will do the throttle and - * dequeue its entity in the parent(s). Therefore the 'simple' - * nr_running test will indeed be correct. - */ - if (unlikely(check_cfs_rq_runtime(cfs_rq))) - goto simple; + /* + * This call to check_cfs_rq_runtime() will do the + * throttle and dequeue its entity in the parent(s). + * Therefore the 'simple' nr_running test will indeed + * be correct. + */ + if (unlikely(check_cfs_rq_runtime(cfs_rq))) + goto simple; + } se = pick_next_entity(cfs_rq, curr); cfs_rq = group_cfs_rq(se); -- cgit v1.2.3 From e4991b240c622f0441c21f4869e13209abc08c5e Mon Sep 17 00:00:00 2001 From: Rik van Riel Date: Wed, 27 May 2015 15:04:27 -0400 Subject: Revert 095bebf61a46 ("sched/numa: Do not move past the balance point if unbalanced") Commit 095bebf61a46 ("sched/numa: Do not move past the balance point if unbalanced") broke convergence of workloads with just one runnable thread, by making it impossible for the one runnable thread on the system to move from one NUMA node to another. Instead, the thread would remain where it was, and pull all the memory across to its location, which is much slower than just migrating the thread to where the memory is. The next patch has a better fix for the issue that 095bebf61a46 tried to address. Reported-by: Jirka Hladky Signed-off-by: Rik van Riel Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: dedekind1@gmail.com Cc: mgorman@suse.de Link: http://lkml.kernel.org/r/1432753468-7785-2-git-send-email-riel@redhat.com Signed-off-by: Ingo Molnar --- kernel/sched/fair.c | 41 +++++++++++++++-------------------------- 1 file changed, 15 insertions(+), 26 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 84ada054c6a8..723d69e241be 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1198,11 +1198,9 @@ static void task_numa_assign(struct task_numa_env *env, static bool load_too_imbalanced(long src_load, long dst_load, struct task_numa_env *env) { + long imb, old_imb; + long orig_src_load, orig_dst_load; long src_capacity, dst_capacity; - long orig_src_load; - long load_a, load_b; - long moved_load; - long imb; /* * The load is corrected for the CPU capacity available on each node. @@ -1215,39 +1213,30 @@ static bool load_too_imbalanced(long src_load, long dst_load, dst_capacity = env->dst_stats.compute_capacity; /* We care about the slope of the imbalance, not the direction. */ - load_a = dst_load; - load_b = src_load; - if (load_a < load_b) - swap(load_a, load_b); + if (dst_load < src_load) + swap(dst_load, src_load); /* Is the difference below the threshold? */ - imb = load_a * src_capacity * 100 - - load_b * dst_capacity * env->imbalance_pct; + imb = dst_load * src_capacity * 100 - + src_load * dst_capacity * env->imbalance_pct; if (imb <= 0) return false; /* * The imbalance is above the allowed threshold. - * Allow a move that brings us closer to a balanced situation, - * without moving things past the point of balance. + * Compare it with the old imbalance. */ orig_src_load = env->src_stats.load; + orig_dst_load = env->dst_stats.load; - /* - * In a task swap, there will be one load moving from src to dst, - * and another moving back. This is the net sum of both moves. - * A simple task move will always have a positive value. - * Allow the move if it brings the system closer to a balanced - * situation, without crossing over the balance point. - */ - moved_load = orig_src_load - src_load; + if (orig_dst_load < orig_src_load) + swap(orig_dst_load, orig_src_load); - if (moved_load > 0) - /* Moving src -> dst. Did we overshoot balance? */ - return src_load * dst_capacity < dst_load * src_capacity; - else - /* Moving dst -> src. Did we overshoot balance? */ - return dst_load * src_capacity < src_load * dst_capacity; + old_imb = orig_dst_load * src_capacity * 100 - + orig_src_load * dst_capacity * env->imbalance_pct; + + /* Would this change make things worse? */ + return (imb > old_imb); } /* -- cgit v1.2.3 From 6f9aad0bc37286c0441b57f0ba8cffee50715426 Mon Sep 17 00:00:00 2001 From: Rik van Riel Date: Thu, 28 May 2015 09:52:49 -0400 Subject: sched/numa: Only consider less busy nodes as numa balancing destinations Changeset a43455a1d572 ("sched/numa: Ensure task_numa_migrate() checks the preferred node") fixes an issue where workloads would never converge on a fully loaded (or overloaded) system. However, it introduces a regression on less than fully loaded systems, where workloads converge on a few NUMA nodes, instead of properly staying spread out across the whole system. This leads to a reduction in available memory bandwidth, and usable CPU cache, with predictable performance problems. The root cause appears to be an interaction between the load balancer and NUMA balancing, where the short term load represented by the load balancer differs from the long term load the NUMA balancing code would like to base its decisions on. Simply reverting a43455a1d572 would re-introduce the non-convergence of workloads on fully loaded systems, so that is not a good option. As an aside, the check done before a43455a1d572 only applied to a task's preferred node, not to other candidate nodes in the system, so the converge-on-too-few-nodes problem still happens, just to a lesser degree. Instead, try to compensate for the impedance mismatch between the load balancer and NUMA balancing by only ever considering a lesser loaded node as a destination for NUMA balancing, regardless of whether the task is trying to move to the preferred node, or to another node. This patch also addresses the issue that a system with a single runnable thread would never migrate that thread to near its memory, introduced by 095bebf61a46 ("sched/numa: Do not move past the balance point if unbalanced"). A test where the main thread creates a large memory area, and spawns a worker thread to iterate over the memory (placed on another node by select_task_rq_fair), after which the main thread goes to sleep and waits for the worker thread to loop over all the memory now sees the worker thread migrated to where the memory is, instead of having all the memory migrated over like before. Jirka has run a number of performance tests on several systems: single instance SpecJBB 2005 performance is 7-15% higher on a 4 node system, with higher gains on systems with more cores per socket. Multi-instance SpecJBB 2005 (one per node), linpack, and stream see little or no changes with the revert of 095bebf61a46 and this patch. Reported-by: Artem Bityutski Reported-by: Jirka Hladky Tested-by: Jirka Hladky Tested-by: Artem Bityutskiy Signed-off-by: Rik van Riel Signed-off-by: Peter Zijlstra (Intel) Acked-by: Mel Gorman Cc: Andrew Morton Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/20150528095249.3083ade0@annuminas.surriel.com Signed-off-by: Ingo Molnar --- kernel/sched/fair.c | 30 ++++++++++++++++++++++++++++-- 1 file changed, 28 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 723d69e241be..4b6e5f63d9af 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1398,6 +1398,30 @@ static void task_numa_find_cpu(struct task_numa_env *env, } } +/* Only move tasks to a NUMA node less busy than the current node. */ +static bool numa_has_capacity(struct task_numa_env *env) +{ + struct numa_stats *src = &env->src_stats; + struct numa_stats *dst = &env->dst_stats; + + if (src->has_free_capacity && !dst->has_free_capacity) + return false; + + /* + * Only consider a task move if the source has a higher load + * than the destination, corrected for CPU capacity on each node. + * + * src->load dst->load + * --------------------- vs --------------------- + * src->compute_capacity dst->compute_capacity + */ + if (src->load * dst->compute_capacity > + dst->load * src->compute_capacity) + return true; + + return false; +} + static int task_numa_migrate(struct task_struct *p) { struct task_numa_env env = { @@ -1452,7 +1476,8 @@ static int task_numa_migrate(struct task_struct *p) update_numa_stats(&env.dst_stats, env.dst_nid); /* Try to find a spot on the preferred nid. */ - task_numa_find_cpu(&env, taskimp, groupimp); + if (numa_has_capacity(&env)) + task_numa_find_cpu(&env, taskimp, groupimp); /* * Look at other nodes in these cases: @@ -1483,7 +1508,8 @@ static int task_numa_migrate(struct task_struct *p) env.dist = dist; env.dst_nid = nid; update_numa_stats(&env.dst_stats, env.dst_nid); - task_numa_find_cpu(&env, taskimp, groupimp); + if (numa_has_capacity(&env)) + task_numa_find_cpu(&env, taskimp, groupimp); } } -- cgit v1.2.3 From 33d6176eb12d1b0ae6d2f672b47367fd90726b91 Mon Sep 17 00:00:00 2001 From: Srikar Dronamraju Date: Mon, 8 Jun 2015 13:40:39 +0530 Subject: sched/debug: Properly format runnable tasks in /proc/sched_debug With !CONFIG_SCHEDSTATS, runnable tasks in /proc/sched_debug has too many columns than required. Fix this by printing appropriate columns. While at this, print sum_exec_runtime, since this information is available even in !CONFIG_SCHEDSTATS case. Signed-off-by: Srikar Dronamraju Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1433751041-11724-2-git-send-email-srikar@linux.vnet.ibm.com Signed-off-by: Ingo Molnar --- kernel/sched/debug.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index a245c1fc6f0a..59cb603081a3 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -136,8 +136,10 @@ print_task(struct seq_file *m, struct rq *rq, struct task_struct *p) SPLIT_NS(p->se.sum_exec_runtime), SPLIT_NS(p->se.statistics.sum_sleep_runtime)); #else - SEQ_printf(m, "%15Ld %15Ld %15Ld.%06ld %15Ld.%06ld %15Ld.%06ld", - 0LL, 0LL, 0LL, 0L, 0LL, 0L, 0LL, 0L); + SEQ_printf(m, "%9Ld.%06ld %9Ld.%06ld %9Ld.%06ld", + 0LL, 0L, + SPLIT_NS(p->se.sum_exec_runtime), + 0LL, 0L); #endif #ifdef CONFIG_NUMA_BALANCING SEQ_printf(m, " %d", task_node(p)); -- cgit v1.2.3 From c5f3ab1c3b2e277cca6462415038dab02b4ad396 Mon Sep 17 00:00:00 2001 From: Srikar Dronamraju Date: Mon, 8 Jun 2015 13:40:40 +0530 Subject: sched/debug: Replace vruntime with wait_sum in /proc/sched_debug Within runnable tasks in /proc/sched_debug, vruntime is printed twice, once as tree-key and again as exec-runtime. Since exec-runtime isnt populated in !CONFIG_SCHEDSTATS, use this field to print wait_sum. Signed-off-by: Srikar Dronamraju Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1433751041-11724-3-git-send-email-srikar@linux.vnet.ibm.com Signed-off-by: Ingo Molnar --- kernel/sched/debug.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index 59cb603081a3..7dc547eb56d7 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -132,7 +132,7 @@ print_task(struct seq_file *m, struct rq *rq, struct task_struct *p) p->prio); #ifdef CONFIG_SCHEDSTATS SEQ_printf(m, "%9Ld.%06ld %9Ld.%06ld %9Ld.%06ld", - SPLIT_NS(p->se.vruntime), + SPLIT_NS(p->se.statistics.wait_sum), SPLIT_NS(p->se.sum_exec_runtime), SPLIT_NS(p->se.statistics.sum_sleep_runtime)); #else @@ -158,7 +158,7 @@ static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu) SEQ_printf(m, "\nrunnable tasks:\n" " task PID tree-key switches prio" - " exec-runtime sum-exec sum-sleep\n" + " wait-time sum-exec sum-sleep\n" "------------------------------------------------------" "----------------------------------------------------\n"); -- cgit v1.2.3 From 82a0d2762699b95d6ce4114d00dc1865df9b0df3 Mon Sep 17 00:00:00 2001 From: Srikar Dronamraju Date: Mon, 8 Jun 2015 13:40:41 +0530 Subject: sched/debug: Add sum_sleep_runtime to /proc//sched When CONFIG_SCHEDSTATS is enabled, /proc//sched prints almost all sched statistics except sum_sleep_runtime. Since sum_sleep_runtime is a good info to collect, add this it to /proc//sched. Signed-off-by: Srikar Dronamraju Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1433751041-11724-4-git-send-email-srikar@linux.vnet.ibm.com Signed-off-by: Ingo Molnar --- kernel/sched/debug.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index 7dc547eb56d7..704683cc9042 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -584,6 +584,7 @@ void proc_sched_show_task(struct task_struct *p, struct seq_file *m) nr_switches = p->nvcsw + p->nivcsw; #ifdef CONFIG_SCHEDSTATS + PN(se.statistics.sum_sleep_runtime); PN(se.statistics.wait_start); PN(se.statistics.sleep_start); PN(se.statistics.block_start); -- cgit v1.2.3 From b17718d02f54b90978d0e0146368b512b11c3e84 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 5 Jun 2015 17:30:23 +0200 Subject: sched/stop_machine: Fix deadlock between multiple stop_two_cpus() Jiri reported a machine stuck in multi_cpu_stop() with migrate_swap_stop() as function and with the following src,dst cpu pairs: {11, 4} {13, 11} { 4, 13} 4 11 13 cpuM: queue(4 ,13) *Ma cpuN: queue(13,11) *N Na *M Mb cpuO: queue(11, 4) *O Oa *Nb *Ob Where *X denotes the cpu running the queueing of cpu-X and X[ab] denotes the first/second queued work. You'll observe the top of the workqueue for each cpu: 4,11,13 to be work from cpus: M, O, N resp. IOW. deadlock. Do away with the queueing trickery and introduce lg_double_lock() to lock both CPUs and fully serialize the stop_two_cpus() callers instead of the partial (and buggy) serialization we have now. Reported-by: Jiri Olsa Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Oleg Nesterov Cc: Peter Zijlstra Cc: Rik van Riel Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/20150605153023.GH19282@twins.programming.kicks-ass.net Signed-off-by: Ingo Molnar --- include/linux/lglock.h | 5 +++++ kernel/locking/lglock.c | 22 ++++++++++++++++++++++ kernel/stop_machine.c | 42 +++++------------------------------------- 3 files changed, 32 insertions(+), 37 deletions(-) (limited to 'kernel') diff --git a/include/linux/lglock.h b/include/linux/lglock.h index 0081f000e34b..c92ebd100d9b 100644 --- a/include/linux/lglock.h +++ b/include/linux/lglock.h @@ -52,10 +52,15 @@ struct lglock { static struct lglock name = { .lock = &name ## _lock } void lg_lock_init(struct lglock *lg, char *name); + void lg_local_lock(struct lglock *lg); void lg_local_unlock(struct lglock *lg); void lg_local_lock_cpu(struct lglock *lg, int cpu); void lg_local_unlock_cpu(struct lglock *lg, int cpu); + +void lg_double_lock(struct lglock *lg, int cpu1, int cpu2); +void lg_double_unlock(struct lglock *lg, int cpu1, int cpu2); + void lg_global_lock(struct lglock *lg); void lg_global_unlock(struct lglock *lg); diff --git a/kernel/locking/lglock.c b/kernel/locking/lglock.c index 86ae2aebf004..951cfcd10b4a 100644 --- a/kernel/locking/lglock.c +++ b/kernel/locking/lglock.c @@ -60,6 +60,28 @@ void lg_local_unlock_cpu(struct lglock *lg, int cpu) } EXPORT_SYMBOL(lg_local_unlock_cpu); +void lg_double_lock(struct lglock *lg, int cpu1, int cpu2) +{ + BUG_ON(cpu1 == cpu2); + + /* lock in cpu order, just like lg_global_lock */ + if (cpu2 < cpu1) + swap(cpu1, cpu2); + + preempt_disable(); + lock_acquire_shared(&lg->lock_dep_map, 0, 0, NULL, _RET_IP_); + arch_spin_lock(per_cpu_ptr(lg->lock, cpu1)); + arch_spin_lock(per_cpu_ptr(lg->lock, cpu2)); +} + +void lg_double_unlock(struct lglock *lg, int cpu1, int cpu2) +{ + lock_release(&lg->lock_dep_map, 1, _RET_IP_); + arch_spin_unlock(per_cpu_ptr(lg->lock, cpu1)); + arch_spin_unlock(per_cpu_ptr(lg->lock, cpu2)); + preempt_enable(); +} + void lg_global_lock(struct lglock *lg) { int i; diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index 695f0c6cd169..fd643d8c4b42 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -211,25 +211,6 @@ static int multi_cpu_stop(void *data) return err; } -struct irq_cpu_stop_queue_work_info { - int cpu1; - int cpu2; - struct cpu_stop_work *work1; - struct cpu_stop_work *work2; -}; - -/* - * This function is always run with irqs and preemption disabled. - * This guarantees that both work1 and work2 get queued, before - * our local migrate thread gets the chance to preempt us. - */ -static void irq_cpu_stop_queue_work(void *arg) -{ - struct irq_cpu_stop_queue_work_info *info = arg; - cpu_stop_queue_work(info->cpu1, info->work1); - cpu_stop_queue_work(info->cpu2, info->work2); -} - /** * stop_two_cpus - stops two cpus * @cpu1: the cpu to stop @@ -245,7 +226,6 @@ int stop_two_cpus(unsigned int cpu1, unsigned int cpu2, cpu_stop_fn_t fn, void * { struct cpu_stop_done done; struct cpu_stop_work work1, work2; - struct irq_cpu_stop_queue_work_info call_args; struct multi_stop_data msdata; preempt_disable(); @@ -262,13 +242,6 @@ int stop_two_cpus(unsigned int cpu1, unsigned int cpu2, cpu_stop_fn_t fn, void * .done = &done }; - call_args = (struct irq_cpu_stop_queue_work_info){ - .cpu1 = cpu1, - .cpu2 = cpu2, - .work1 = &work1, - .work2 = &work2, - }; - cpu_stop_init_done(&done, 2); set_state(&msdata, MULTI_STOP_PREPARE); @@ -285,16 +258,11 @@ int stop_two_cpus(unsigned int cpu1, unsigned int cpu2, cpu_stop_fn_t fn, void * return -ENOENT; } - lg_local_lock(&stop_cpus_lock); - /* - * Queuing needs to be done by the lowest numbered CPU, to ensure - * that works are always queued in the same order on every CPU. - * This prevents deadlocks. - */ - smp_call_function_single(min(cpu1, cpu2), - &irq_cpu_stop_queue_work, - &call_args, 1); - lg_local_unlock(&stop_cpus_lock); + lg_double_lock(&stop_cpus_lock, cpu1, cpu2); + cpu_stop_queue_work(cpu1, &work1); + cpu_stop_queue_work(cpu2, &work2); + lg_double_unlock(&stop_cpus_lock, cpu1, cpu2); + preempt_enable(); wait_for_completion(&done.completion); -- cgit v1.2.3 From d84525a845cc2617d638349f8756a9fec9ac8113 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Sun, 17 May 2015 12:53:10 -0400 Subject: sched/preempt: Fix preempt notifiers documentation about hlist_del() within unsafe iteration preempt_notifier_unregister() documents: "This is safe to call from within a preemption notifier." However, both fire_sched_in_preempt_notifiers() and fire_sched_out_preempt_notifiers() are using hlist_for_each_entry(), which is not safe against entry removal during iteration. Inspection of the KVM code does not reveal any use of preempt_notifier_unregister() within the preempt notifiers. Therefore, fix the comment. Signed-off-by: Mathieu Desnoyers Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431881590-1456-1-git-send-email-mathieu.desnoyers@efficios.com Signed-off-by: Ingo Molnar --- kernel/sched/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index af0a5a6cee98..bdb7aa67baef 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2165,7 +2165,7 @@ EXPORT_SYMBOL_GPL(preempt_notifier_register); * preempt_notifier_unregister - no longer interested in preemption notifications * @notifier: notifier struct to unregister * - * This is safe to call from within a preemption notifier. + * This is *not* safe to call from within a preemption notifier. */ void preempt_notifier_unregister(struct preempt_notifier *notifier) { -- cgit v1.2.3 From 1cde2930e15473cb4dd7e5a07d83e605a969bd6e Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 8 Jun 2015 16:00:30 +0200 Subject: sched/preempt: Add static_key() to preempt_notifiers Avoid touching the curr->preempt_notifier cacheline when not needed. Provides a small improvement on pipe-bench: taskset 01 perf stat --repeat 10 -- perf bench sched pipe before: Performance counter stats for 'perf bench sched pipe' (10 runs): 12385.016204 task-clock (msec) # 1.001 CPUs utilized ( +- 0.34% ) 2,000,023 context-switches # 0.161 M/sec ( +- 0.00% ) 0 cpu-migrations # 0.000 K/sec 175 page-faults # 0.014 K/sec ( +- 0.26% ) 41,376,162,250 cycles # 3.341 GHz ( +- 0.11% ) 17,389,139,321 stalled-cycles-frontend # 42.03% frontend cycles idle ( +- 0.25% ) stalled-cycles-backend 68,788,588,003 instructions # 1.66 insns per cycle # 0.25 stalled cycles per insn ( +- 0.02% ) 13,449,387,620 branches # 1085.940 M/sec ( +- 0.02% ) 20,880,690 branch-misses # 0.16% of all branches ( +- 0.98% ) 12.372646094 seconds time elapsed ( +- 0.34% ) after: Performance counter stats for 'perf bench sched pipe' (10 runs): 12180.936528 task-clock (msec) # 1.001 CPUs utilized ( +- 0.33% ) 2,000,077 context-switches # 0.164 M/sec ( +- 0.00% ) 0 cpu-migrations # 0.000 K/sec 174 page-faults # 0.014 K/sec ( +- 0.27% ) 40,691,545,577 cycles # 3.341 GHz ( +- 0.06% ) 16,446,333,371 stalled-cycles-frontend # 40.42% frontend cycles idle ( +- 0.18% ) stalled-cycles-backend 68,570,100,387 instructions # 1.69 insns per cycle # 0.24 stalled cycles per insn ( +- 0.01% ) 13,389,740,014 branches # 1099.237 M/sec ( +- 0.01% ) 20,175,440 branch-misses # 0.15% of all branches ( +- 0.52% ) 12.169253010 seconds time elapsed ( +- 0.33% ) Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar --- kernel/sched/core.c | 28 +++++++++++++++++++++++----- 1 file changed, 23 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index bdb7aa67baef..1428c7cebe2f 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2151,12 +2151,15 @@ void wake_up_new_task(struct task_struct *p) #ifdef CONFIG_PREEMPT_NOTIFIERS +static struct static_key preempt_notifier_key = STATIC_KEY_INIT_FALSE; + /** * preempt_notifier_register - tell me when current is being preempted & rescheduled * @notifier: notifier struct to register */ void preempt_notifier_register(struct preempt_notifier *notifier) { + static_key_slow_inc(&preempt_notifier_key); hlist_add_head(¬ifier->link, ¤t->preempt_notifiers); } EXPORT_SYMBOL_GPL(preempt_notifier_register); @@ -2170,10 +2173,11 @@ EXPORT_SYMBOL_GPL(preempt_notifier_register); void preempt_notifier_unregister(struct preempt_notifier *notifier) { hlist_del(¬ifier->link); + static_key_slow_dec(&preempt_notifier_key); } EXPORT_SYMBOL_GPL(preempt_notifier_unregister); -static void fire_sched_in_preempt_notifiers(struct task_struct *curr) +static void __fire_sched_in_preempt_notifiers(struct task_struct *curr) { struct preempt_notifier *notifier; @@ -2181,9 +2185,15 @@ static void fire_sched_in_preempt_notifiers(struct task_struct *curr) notifier->ops->sched_in(notifier, raw_smp_processor_id()); } +static __always_inline void fire_sched_in_preempt_notifiers(struct task_struct *curr) +{ + if (static_key_false(&preempt_notifier_key)) + __fire_sched_in_preempt_notifiers(curr); +} + static void -fire_sched_out_preempt_notifiers(struct task_struct *curr, - struct task_struct *next) +__fire_sched_out_preempt_notifiers(struct task_struct *curr, + struct task_struct *next) { struct preempt_notifier *notifier; @@ -2191,13 +2201,21 @@ fire_sched_out_preempt_notifiers(struct task_struct *curr, notifier->ops->sched_out(notifier, next); } +static __always_inline void +fire_sched_out_preempt_notifiers(struct task_struct *curr, + struct task_struct *next) +{ + if (static_key_false(&preempt_notifier_key)) + __fire_sched_out_preempt_notifiers(curr, next); +} + #else /* !CONFIG_PREEMPT_NOTIFIERS */ -static void fire_sched_in_preempt_notifiers(struct task_struct *curr) +static inline void fire_sched_in_preempt_notifiers(struct task_struct *curr) { } -static void +static inline void fire_sched_out_preempt_notifiers(struct task_struct *curr, struct task_struct *next) { -- cgit v1.2.3 From 8b5e770ed7c05a65ffd2d33a83c14572696236dc Mon Sep 17 00:00:00 2001 From: Wanpeng Li Date: Wed, 13 May 2015 14:01:01 +0800 Subject: sched/deadline: Optimize pull_dl_task() pull_dl_task() uses pick_next_earliest_dl_task() to select a migration candidate; this is sub-optimal since the next earliest task -- as per the regular runqueue -- might not be migratable at all. This could result in iterating the entire runqueue looking for a task. Instead iterate the pushable queue -- this queue only contains tasks that have at least 2 cpus set in their cpus_allowed mask. Signed-off-by: Wanpeng Li [ Improved the changelog. ] Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Juri Lelli Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431496867-4194-1-git-send-email-wanpeng.li@linux.intel.com Signed-off-by: Ingo Molnar --- kernel/sched/deadline.c | 28 +++++++++++++++++++++++++++- 1 file changed, 27 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c index 890ce951c717..9cbe1c7fd36e 100644 --- a/kernel/sched/deadline.c +++ b/kernel/sched/deadline.c @@ -1230,6 +1230,32 @@ next_node: return NULL; } +/* + * Return the earliest pushable rq's task, which is suitable to be executed + * on the CPU, NULL otherwise: + */ +static struct task_struct *pick_earliest_pushable_dl_task(struct rq *rq, int cpu) +{ + struct rb_node *next_node = rq->dl.pushable_dl_tasks_leftmost; + struct task_struct *p = NULL; + + if (!has_pushable_dl_tasks(rq)) + return NULL; + +next_node: + if (next_node) { + p = rb_entry(next_node, struct task_struct, pushable_dl_tasks); + + if (pick_dl_task(rq, p, cpu)) + return p; + + next_node = rb_next(next_node); + goto next_node; + } + + return NULL; +} + static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask_dl); static int find_later_rq(struct task_struct *task) @@ -1514,7 +1540,7 @@ static int pull_dl_task(struct rq *this_rq) if (src_rq->dl.dl_nr_running <= 1) goto skip; - p = pick_next_earliest_dl_task(src_rq, this_cpu); + p = pick_earliest_pushable_dl_task(src_rq, this_cpu); /* * We found a task to be pulled if: -- cgit v1.2.3 From a6c0e746fb8f4ea6508f274314378325a6e1ec9b Mon Sep 17 00:00:00 2001 From: Wanpeng Li Date: Wed, 13 May 2015 14:01:02 +0800 Subject: sched/deadline: Make init_sched_dl_class() __init It's a bootstrap function, make init_sched_dl_class() __init. Signed-off-by: Wanpeng Li Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Juri Lelli Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431496867-4194-2-git-send-email-wanpeng.li@linux.intel.com Signed-off-by: Ingo Molnar --- kernel/sched/deadline.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c index 9cbe1c7fd36e..1c4bc31eb0f5 100644 --- a/kernel/sched/deadline.c +++ b/kernel/sched/deadline.c @@ -1685,7 +1685,7 @@ static void rq_offline_dl(struct rq *rq) cpudl_clear_freecpu(&rq->rd->cpudl, rq->cpu); } -void init_sched_dl_class(void) +void __init init_sched_dl_class(void) { unsigned int i; -- cgit v1.2.3 From 9d514262425691dddf942edea8bc9919e66fe140 Mon Sep 17 00:00:00 2001 From: Wanpeng Li Date: Wed, 13 May 2015 14:01:03 +0800 Subject: sched/deadline: Reduce rq lock contention by eliminating locking of non-feasible target This patch adds a check that prevents futile attempts to move DL tasks to a CPU with active tasks of equal or earlier deadline. The same behavior as commit 80e3d87b2c55 ("sched/rt: Reduce rq lock contention by eliminating locking of non-feasible target") for rt class. Signed-off-by: Wanpeng Li Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Juri Lelli Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431496867-4194-3-git-send-email-wanpeng.li@linux.intel.com Signed-off-by: Ingo Molnar --- kernel/sched/deadline.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c index 1c4bc31eb0f5..98f787143db4 100644 --- a/kernel/sched/deadline.c +++ b/kernel/sched/deadline.c @@ -1012,7 +1012,9 @@ select_task_rq_dl(struct task_struct *p, int cpu, int sd_flag, int flags) (p->nr_cpus_allowed > 1)) { int target = find_later_rq(p); - if (target != -1) + if (target != -1 && + dl_time_before(p->dl.deadline, + cpu_rq(target)->dl.earliest_dl.curr)) cpu = target; } rcu_read_unlock(); @@ -1359,6 +1361,17 @@ static struct rq *find_lock_later_rq(struct task_struct *task, struct rq *rq) later_rq = cpu_rq(cpu); + if (!dl_time_before(task->dl.deadline, + later_rq->dl.earliest_dl.curr)) { + /* + * Target rq has tasks of equal or earlier deadline, + * retrying does not release any lock and is unlikely + * to yield a different result. + */ + later_rq = NULL; + break; + } + /* Retry if something changed. */ if (double_lock_balance(rq, later_rq)) { if (unlikely(task_rq(task) != rq || -- cgit v1.2.3 From 178a4d23e4e6a0a90b086dad86697676b49db60a Mon Sep 17 00:00:00 2001 From: Wanpeng Li Date: Wed, 13 May 2015 14:01:05 +0800 Subject: sched/deadline: Drop duplicate init_sched_dl_class() declaration There are two init_sched_dl_class() declarations, this patch drops the duplicate. Signed-off-by: Wanpeng Li Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Juri Lelli Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431496867-4194-5-git-send-email-wanpeng.li@linux.intel.com Signed-off-by: Ingo Molnar --- kernel/sched/sched.h | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index d85455539d5c..d62b2882232b 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -1290,7 +1290,6 @@ extern void update_max_interval(void); extern void init_sched_dl_class(void); extern void init_sched_rt_class(void); extern void init_sched_fair_class(void); -extern void init_sched_dl_class(void); extern void resched_curr(struct rq *rq); extern void resched_cpu(int cpu); -- cgit v1.2.3 From 6713c3aa7f63626c0cecf9c509fb48d885b2dd12 Mon Sep 17 00:00:00 2001 From: Wanpeng Li Date: Wed, 13 May 2015 14:01:06 +0800 Subject: sched: Remove superfluous resetting of the p->dl_throttled flag Resetting the p->dl_throttled flag in rt_mutex_setprio() (for a task that is going to be boosted) is superfluous, as the natural place to do so is in replenish_dl_entity(). If the task was on the runqueue and it is boosted by a DL task, it will be enqueued back with ENQUEUE_REPLENISH flag set, which can guarantee that dl_throttled is reset in replenish_dl_entity(). This patch drops the resetting of throttled status in function rt_mutex_setprio(). Signed-off-by: Wanpeng Li Signed-off-by: Peter Zijlstra (Intel) Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Juri Lelli Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1431496867-4194-6-git-send-email-wanpeng.li@linux.intel.com Signed-off-by: Ingo Molnar --- kernel/sched/core.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 1428c7cebe2f..10338ce78be4 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3099,7 +3099,6 @@ void rt_mutex_setprio(struct task_struct *p, int prio) if (!dl_prio(p->normal_prio) || (pi_task && dl_entity_preempt(&pi_task->dl, &p->dl))) { p->dl.dl_boosted = 1; - p->dl.dl_throttled = 0; enqueue_flag = ENQUEUE_REPLENISH; } else p->dl.dl_boosted = 0; -- cgit v1.2.3 From 6fab54101923044712baee429ff573f03b99fc47 Mon Sep 17 00:00:00 2001 From: Zhiqiang Zhang Date: Mon, 15 Jun 2015 11:15:20 +0800 Subject: sched/deadline: Remove needless parameter in dl_runtime_exceeded() Sine commit 269ad8015a6b ("sched/deadline: Avoid double-accounting in case of missed deadlines), parameter 'rq' is no longer used, so remove it. Signed-off-by: Zhiqiang Zhang Signed-off-by: Peter Zijlstra (Intel) Cc: Cc: Cc: Andrew Morton Cc: Borislav Petkov Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1434338120-43773-1-git-send-email-zhangzhiqiang.zhang@huawei.com Signed-off-by: Ingo Molnar --- kernel/sched/deadline.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c index 98f787143db4..392e8fb94db3 100644 --- a/kernel/sched/deadline.c +++ b/kernel/sched/deadline.c @@ -640,7 +640,7 @@ void init_dl_task_timer(struct sched_dl_entity *dl_se) } static -int dl_runtime_exceeded(struct rq *rq, struct sched_dl_entity *dl_se) +int dl_runtime_exceeded(struct sched_dl_entity *dl_se) { return (dl_se->runtime <= 0); } @@ -684,7 +684,7 @@ static void update_curr_dl(struct rq *rq) sched_rt_avg_update(rq, delta_exec); dl_se->runtime -= dl_se->dl_yielded ? 0 : delta_exec; - if (dl_runtime_exceeded(rq, dl_se)) { + if (dl_runtime_exceeded(dl_se)) { dl_se->dl_throttled = 1; __dequeue_task_dl(rq, curr, 0); if (unlikely(!start_dl_timer(dl_se, curr->dl.dl_boosted))) -- cgit v1.2.3