From 8e8be45e8e55daa381028aec339829929ddb53a5 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Thu, 2 Sep 2010 16:16:14 -0700 Subject: rcu: add priority-inversion testing to rcutorture Add an optional test to force long-term preemption of RCU read-side critical sections, controlled by new test_boost, test_boost_interval, and test_boost_duration module parameters. This is to be used to test RCU priority boosting. Signed-off-by: Paul E. McKenney --- kernel/rcutorture.c | 270 +++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 259 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutorture.c b/kernel/rcutorture.c index 9d8e8fb2515f..89613f97ff26 100644 --- a/kernel/rcutorture.c +++ b/kernel/rcutorture.c @@ -47,6 +47,7 @@ #include #include #include +#include MODULE_LICENSE("GPL"); MODULE_AUTHOR("Paul E. McKenney and " @@ -64,6 +65,9 @@ static int irqreader = 1; /* RCU readers from irq (timers). */ static int fqs_duration = 0; /* Duration of bursts (us), 0 to disable. */ static int fqs_holdoff = 0; /* Hold time within burst (us). */ static int fqs_stutter = 3; /* Wait time between bursts (s). */ +static int test_boost = 1; /* Test RCU prio boost: 0=no, 1=maybe, 2=yes. */ +static int test_boost_interval = 7; /* Interval between boost tests, seconds. */ +static int test_boost_duration = 4; /* Duration of each boost test, seconds. */ static char *torture_type = "rcu"; /* What RCU implementation to torture. */ module_param(nreaders, int, 0444); @@ -88,6 +92,12 @@ module_param(fqs_holdoff, int, 0444); MODULE_PARM_DESC(fqs_holdoff, "Holdoff time within fqs bursts (us)"); module_param(fqs_stutter, int, 0444); MODULE_PARM_DESC(fqs_stutter, "Wait time between fqs bursts (s)"); +module_param(test_boost, int, 0444); +MODULE_PARM_DESC(test_boost, "Test RCU prio boost: 0=no, 1=maybe, 2=yes."); +module_param(test_boost_interval, int, 0444); +MODULE_PARM_DESC(test_boost_interval, "Interval between boost tests, seconds."); +module_param(test_boost_duration, int, 0444); +MODULE_PARM_DESC(test_boost_duration, "Duration of each boost test, seconds."); module_param(torture_type, charp, 0444); MODULE_PARM_DESC(torture_type, "Type of RCU to torture (rcu, rcu_bh, srcu)"); @@ -109,6 +119,7 @@ static struct task_struct *stats_task; static struct task_struct *shuffler_task; static struct task_struct *stutter_task; static struct task_struct *fqs_task; +static struct task_struct *boost_tasks[NR_CPUS]; #define RCU_TORTURE_PIPE_LEN 10 @@ -134,6 +145,12 @@ static atomic_t n_rcu_torture_alloc_fail; static atomic_t n_rcu_torture_free; static atomic_t n_rcu_torture_mberror; static atomic_t n_rcu_torture_error; +static long n_rcu_torture_boost_ktrerror; +static long n_rcu_torture_boost_rterror; +static long n_rcu_torture_boost_allocerror; +static long n_rcu_torture_boost_afferror; +static long n_rcu_torture_boost_failure; +static long n_rcu_torture_boosts; static long n_rcu_torture_timers; static struct list_head rcu_torture_removed; static cpumask_var_t shuffle_tmp_mask; @@ -147,6 +164,16 @@ static int stutter_pause_test; #endif int rcutorture_runnable = RCUTORTURE_RUNNABLE_INIT; +#ifdef CONFIG_RCU_BOOST +#define rcu_can_boost() 1 +#else /* #ifdef CONFIG_RCU_BOOST */ +#define rcu_can_boost() 0 +#endif /* #else #ifdef CONFIG_RCU_BOOST */ + +static unsigned long boost_starttime; /* jiffies of next boost test start. */ +DEFINE_MUTEX(boost_mutex); /* protect setting boost_starttime */ + /* and boost task create/destroy. */ + /* Mediate rmmod and system shutdown. Concurrent rmmod & shutdown illegal! */ #define FULLSTOP_DONTSTOP 0 /* Normal operation. */ @@ -277,6 +304,7 @@ struct rcu_torture_ops { void (*fqs)(void); int (*stats)(char *page); int irq_capable; + int can_boost; char *name; }; @@ -366,6 +394,7 @@ static struct rcu_torture_ops rcu_ops = { .fqs = rcu_force_quiescent_state, .stats = NULL, .irq_capable = 1, + .can_boost = rcu_can_boost(), .name = "rcu" }; @@ -408,6 +437,7 @@ static struct rcu_torture_ops rcu_sync_ops = { .fqs = rcu_force_quiescent_state, .stats = NULL, .irq_capable = 1, + .can_boost = rcu_can_boost(), .name = "rcu_sync" }; @@ -424,6 +454,7 @@ static struct rcu_torture_ops rcu_expedited_ops = { .fqs = rcu_force_quiescent_state, .stats = NULL, .irq_capable = 1, + .can_boost = rcu_can_boost(), .name = "rcu_expedited" }; @@ -683,6 +714,110 @@ static struct rcu_torture_ops sched_expedited_ops = { .name = "sched_expedited" }; +/* + * RCU torture priority-boost testing. Runs one real-time thread per + * CPU for moderate bursts, repeatedly registering RCU callbacks and + * spinning waiting for them to be invoked. If a given callback takes + * too long to be invoked, we assume that priority inversion has occurred. + */ + +struct rcu_boost_inflight { + struct rcu_head rcu; + int inflight; +}; + +static void rcu_torture_boost_cb(struct rcu_head *head) +{ + struct rcu_boost_inflight *rbip = + container_of(head, struct rcu_boost_inflight, rcu); + + smp_mb(); /* Ensure RCU-core accesses precede clearing ->inflight */ + rbip->inflight = 0; +} + +static int rcu_torture_boost(void *arg) +{ + unsigned long call_rcu_time; + unsigned long endtime; + unsigned long oldstarttime; + struct rcu_boost_inflight rbi = { .inflight = 0 }; + struct sched_param sp; + + VERBOSE_PRINTK_STRING("rcu_torture_boost started"); + + /* Set real-time priority. */ + sp.sched_priority = 1; + if (sched_setscheduler(current, SCHED_FIFO, &sp) < 0) { + VERBOSE_PRINTK_STRING("rcu_torture_boost RT prio failed!"); + n_rcu_torture_boost_rterror++; + } + + /* Each pass through the following loop does one boost-test cycle. */ + do { + /* Wait for the next test interval. */ + oldstarttime = boost_starttime; + while (jiffies - oldstarttime > ULONG_MAX / 2) { + schedule_timeout_uninterruptible(1); + rcu_stutter_wait("rcu_torture_boost"); + if (kthread_should_stop() || + fullstop != FULLSTOP_DONTSTOP) + goto checkwait; + } + + /* Do one boost-test interval. */ + endtime = oldstarttime + test_boost_duration * HZ; + call_rcu_time = jiffies; + while (jiffies - endtime > ULONG_MAX / 2) { + /* If we don't have a callback in flight, post one. */ + if (!rbi.inflight) { + smp_mb(); /* RCU core before ->inflight = 1. */ + rbi.inflight = 1; + call_rcu(&rbi.rcu, rcu_torture_boost_cb); + if (jiffies - call_rcu_time > + test_boost_duration * HZ - HZ / 2) { + VERBOSE_PRINTK_STRING("rcu_torture_boost boosting failed"); + n_rcu_torture_boost_failure++; + } + call_rcu_time = jiffies; + } + cond_resched(); + rcu_stutter_wait("rcu_torture_boost"); + if (kthread_should_stop() || + fullstop != FULLSTOP_DONTSTOP) + goto checkwait; + } + + /* + * Set the start time of the next test interval. + * Yes, this is vulnerable to long delays, but such + * delays simply cause a false negative for the next + * interval. Besides, we are running at RT priority, + * so delays should be relatively rare. + */ + while (oldstarttime == boost_starttime) { + if (mutex_trylock(&boost_mutex)) { + boost_starttime = jiffies + + test_boost_interval * HZ; + n_rcu_torture_boosts++; + mutex_unlock(&boost_mutex); + break; + } + schedule_timeout_uninterruptible(1); + } + + /* Go do the stutter. */ +checkwait: rcu_stutter_wait("rcu_torture_boost"); + } while (!kthread_should_stop() && fullstop == FULLSTOP_DONTSTOP); + + /* Clean up and exit. */ + VERBOSE_PRINTK_STRING("rcu_torture_boost task stopping"); + rcutorture_shutdown_absorb("rcu_torture_boost"); + while (!kthread_should_stop() || rbi.inflight) + schedule_timeout_uninterruptible(1); + smp_mb(); /* order accesses to ->inflight before stack-frame death. */ + return 0; +} + /* * RCU torture force-quiescent-state kthread. Repeatedly induces * bursts of calls to force_quiescent_state(), increasing the probability @@ -933,7 +1068,8 @@ rcu_torture_printk(char *page) cnt += sprintf(&page[cnt], "%s%s ", torture_type, TORTURE_FLAG); cnt += sprintf(&page[cnt], "rtc: %p ver: %ld tfle: %d rta: %d rtaf: %d rtf: %d " - "rtmbe: %d nt: %ld", + "rtmbe: %d rtbke: %ld rtbre: %ld rtbae: %ld rtbafe: %ld " + "rtbf: %ld rtb: %ld nt: %ld", rcu_torture_current, rcu_torture_current_version, list_empty(&rcu_torture_freelist), @@ -941,8 +1077,19 @@ rcu_torture_printk(char *page) atomic_read(&n_rcu_torture_alloc_fail), atomic_read(&n_rcu_torture_free), atomic_read(&n_rcu_torture_mberror), + n_rcu_torture_boost_ktrerror, + n_rcu_torture_boost_rterror, + n_rcu_torture_boost_allocerror, + n_rcu_torture_boost_afferror, + n_rcu_torture_boost_failure, + n_rcu_torture_boosts, n_rcu_torture_timers); - if (atomic_read(&n_rcu_torture_mberror) != 0) + if (atomic_read(&n_rcu_torture_mberror) != 0 || + n_rcu_torture_boost_ktrerror != 0 || + n_rcu_torture_boost_rterror != 0 || + n_rcu_torture_boost_allocerror != 0 || + n_rcu_torture_boost_afferror != 0 || + n_rcu_torture_boost_failure != 0) cnt += sprintf(&page[cnt], " !!!"); cnt += sprintf(&page[cnt], "\n%s%s ", torture_type, TORTURE_FLAG); if (i > 1) { @@ -1094,22 +1241,91 @@ rcu_torture_stutter(void *arg) } static inline void -rcu_torture_print_module_parms(char *tag) +rcu_torture_print_module_parms(struct rcu_torture_ops *cur_ops, char *tag) { printk(KERN_ALERT "%s" TORTURE_FLAG "--- %s: nreaders=%d nfakewriters=%d " "stat_interval=%d verbose=%d test_no_idle_hz=%d " "shuffle_interval=%d stutter=%d irqreader=%d " - "fqs_duration=%d fqs_holdoff=%d fqs_stutter=%d\n", + "fqs_duration=%d fqs_holdoff=%d fqs_stutter=%d " + "test_boost=%d/%d test_boost_interval=%d " + "test_boost_duration=%d\n", torture_type, tag, nrealreaders, nfakewriters, stat_interval, verbose, test_no_idle_hz, shuffle_interval, - stutter, irqreader, fqs_duration, fqs_holdoff, fqs_stutter); + stutter, irqreader, fqs_duration, fqs_holdoff, fqs_stutter, + test_boost, cur_ops->can_boost, + test_boost_interval, test_boost_duration); } -static struct notifier_block rcutorture_nb = { +static struct notifier_block rcutorture_shutdown_nb = { .notifier_call = rcutorture_shutdown_notify, }; +static void rcutorture_booster_cleanup(int cpu) +{ + struct task_struct *t; + + if (boost_tasks[cpu] == NULL) + return; + mutex_lock(&boost_mutex); + VERBOSE_PRINTK_STRING("Stopping rcu_torture_boost task"); + t = boost_tasks[cpu]; + boost_tasks[cpu] = NULL; + mutex_unlock(&boost_mutex); + + /* This must be outside of the mutex, otherwise deadlock! */ + kthread_stop(t); +} + +static int rcutorture_booster_init(int cpu) +{ + int retval; + + if (boost_tasks[cpu] != NULL) + return 0; /* Already created, nothing more to do. */ + + /* Don't allow time recalculation while creating a new task. */ + mutex_lock(&boost_mutex); + VERBOSE_PRINTK_STRING("Creating rcu_torture_boost task"); + boost_tasks[cpu] = kthread_create(rcu_torture_boost, NULL, + "rcu_torture_boost"); + if (IS_ERR(boost_tasks[cpu])) { + retval = PTR_ERR(boost_tasks[cpu]); + VERBOSE_PRINTK_STRING("rcu_torture_boost task create failed"); + n_rcu_torture_boost_ktrerror++; + boost_tasks[cpu] = NULL; + mutex_unlock(&boost_mutex); + return retval; + } + kthread_bind(boost_tasks[cpu], cpu); + wake_up_process(boost_tasks[cpu]); + mutex_unlock(&boost_mutex); + return 0; +} + +static int rcutorture_cpu_notify(struct notifier_block *self, + unsigned long action, void *hcpu) +{ + long cpu = (long)hcpu; + + switch (action) { + case CPU_ONLINE: + case CPU_DOWN_FAILED: + (void)rcutorture_booster_init(cpu); + break; + case CPU_DOWN_PREPARE: + rcutorture_booster_cleanup(cpu); + break; + default: + break; + } + return NOTIFY_OK; +} + +static struct notifier_block rcutorture_cpu_nb = { + .notifier_call = rcutorture_cpu_notify, +}; + static void rcu_torture_cleanup(void) { @@ -1127,7 +1343,7 @@ rcu_torture_cleanup(void) } fullstop = FULLSTOP_RMMOD; mutex_unlock(&fullstop_mutex); - unregister_reboot_notifier(&rcutorture_nb); + unregister_reboot_notifier(&rcutorture_shutdown_nb); if (stutter_task) { VERBOSE_PRINTK_STRING("Stopping rcu_torture_stutter task"); kthread_stop(stutter_task); @@ -1184,6 +1400,12 @@ rcu_torture_cleanup(void) kthread_stop(fqs_task); } fqs_task = NULL; + if ((test_boost == 1 && cur_ops->can_boost) || + test_boost == 2) { + unregister_cpu_notifier(&rcutorture_cpu_nb); + for_each_possible_cpu(i) + rcutorture_booster_cleanup(i); + } /* Wait for all RCU callbacks to fire. */ @@ -1195,9 +1417,9 @@ rcu_torture_cleanup(void) if (cur_ops->cleanup) cur_ops->cleanup(); if (atomic_read(&n_rcu_torture_error)) - rcu_torture_print_module_parms("End of test: FAILURE"); + rcu_torture_print_module_parms(cur_ops, "End of test: FAILURE"); else - rcu_torture_print_module_parms("End of test: SUCCESS"); + rcu_torture_print_module_parms(cur_ops, "End of test: SUCCESS"); } static int __init @@ -1242,7 +1464,7 @@ rcu_torture_init(void) nrealreaders = nreaders; else nrealreaders = 2 * num_online_cpus(); - rcu_torture_print_module_parms("Start of test"); + rcu_torture_print_module_parms(cur_ops, "Start of test"); fullstop = FULLSTOP_DONTSTOP; /* Set up the freelist. */ @@ -1263,6 +1485,12 @@ rcu_torture_init(void) atomic_set(&n_rcu_torture_free, 0); atomic_set(&n_rcu_torture_mberror, 0); atomic_set(&n_rcu_torture_error, 0); + n_rcu_torture_boost_ktrerror = 0; + n_rcu_torture_boost_rterror = 0; + n_rcu_torture_boost_allocerror = 0; + n_rcu_torture_boost_afferror = 0; + n_rcu_torture_boost_failure = 0; + n_rcu_torture_boosts = 0; for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++) atomic_set(&rcu_torture_wcount[i], 0); for_each_possible_cpu(cpu) { @@ -1376,7 +1604,27 @@ rcu_torture_init(void) goto unwind; } } - register_reboot_notifier(&rcutorture_nb); + if (test_boost_interval < 1) + test_boost_interval = 1; + if (test_boost_duration < 2) + test_boost_duration = 2; + if ((test_boost == 1 && cur_ops->can_boost) || + test_boost == 2) { + int retval; + + boost_starttime = jiffies + test_boost_interval * HZ; + register_cpu_notifier(&rcutorture_cpu_nb); + for_each_possible_cpu(i) { + if (cpu_is_offline(i)) + continue; /* Heuristic: CPU can go offline. */ + retval = rcutorture_booster_init(i); + if (retval < 0) { + firsterr = retval; + goto unwind; + } + } + } + register_reboot_notifier(&rcutorture_shutdown_nb); mutex_unlock(&fullstop_mutex); return 0; -- cgit v1.2.3 From a386b5af8edda1c742ce9f77891e112eefffc005 Mon Sep 17 00:00:00 2001 From: Kasper Pedersen Date: Wed, 20 Oct 2010 15:55:15 -0700 Subject: time: Compensate for rounding on odd-frequency clocksources When the clocksource is not a multiple of HZ, the clock will be off. For acpi_pm, HZ=1000 the error is 127.111 ppm: The rounding of cycle_interval ends up generating a false error term in ntp_error accumulation since xtime_interval is not exactly 1/HZ. So, we subtract out the error caused by the rounding. This has been visible since 2.6.32-rc2 commit a092ff0f90cae22b2ac8028ecd2c6f6c1a9e4601 time: Implement logarithmic time accumulation That commit raised NTP_INTERVAL_FREQ and exposed the rounding error. testing tool: http://n1.taur.dk/permanent/testpmt.c Also tested with ntpd and a frequency counter. Signed-off-by: Kasper Pedersen Acked-by: john stultz Cc: John Kacur Cc: Clark Williams Cc: Martin Schwidefsky Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- kernel/time/timekeeping.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c index 49010d822f72..5bb86da82003 100644 --- a/kernel/time/timekeeping.c +++ b/kernel/time/timekeeping.c @@ -32,6 +32,8 @@ struct timekeeper { cycle_t cycle_interval; /* Number of clock shifted nano seconds in one NTP interval. */ u64 xtime_interval; + /* shifted nano seconds left over when rounding cycle_interval */ + s64 xtime_remainder; /* Raw nano seconds accumulated per NTP interval. */ u32 raw_interval; @@ -62,7 +64,7 @@ struct timekeeper timekeeper; static void timekeeper_setup_internals(struct clocksource *clock) { cycle_t interval; - u64 tmp; + u64 tmp, ntpinterval; timekeeper.clock = clock; clock->cycle_last = clock->read(clock); @@ -70,6 +72,7 @@ static void timekeeper_setup_internals(struct clocksource *clock) /* Do the ns -> cycle conversion first, using original mult */ tmp = NTP_INTERVAL_LENGTH; tmp <<= clock->shift; + ntpinterval = tmp; tmp += clock->mult/2; do_div(tmp, clock->mult); if (tmp == 0) @@ -80,6 +83,7 @@ static void timekeeper_setup_internals(struct clocksource *clock) /* Go back from cycles -> shifted ns */ timekeeper.xtime_interval = (u64) interval * clock->mult; + timekeeper.xtime_remainder = ntpinterval - timekeeper.xtime_interval; timekeeper.raw_interval = ((u64) interval * clock->mult) >> clock->shift; @@ -719,7 +723,8 @@ static cycle_t logarithmic_accumulation(cycle_t offset, int shift) /* Accumulate error between NTP and clock interval */ timekeeper.ntp_error += tick_length << shift; - timekeeper.ntp_error -= timekeeper.xtime_interval << + timekeeper.ntp_error -= + (timekeeper.xtime_interval + timekeeper.xtime_remainder) << (timekeeper.ntp_error_shift + shift); return offset; -- cgit v1.2.3 From 2bf1c05e3c406925e498d06da66b4828f0209ea6 Mon Sep 17 00:00:00 2001 From: Nikitas Angelinas Date: Wed, 20 Oct 2010 15:57:31 -0700 Subject: time: Use ARRAY_SIZE macro in timecompare.c Replace sizeof(buffer)/sizeof(buffer[0]) with ARRAY_SIZE(buffer) in kernel/time/timecompare.c Signed-off-by: Nikitas Angelinas Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- kernel/time/timecompare.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/time/timecompare.c b/kernel/time/timecompare.c index ac38fbb176cc..a9ae369925ce 100644 --- a/kernel/time/timecompare.c +++ b/kernel/time/timecompare.c @@ -21,6 +21,7 @@ #include #include #include +#include /* * fixed point arithmetic scale factor for skew @@ -57,11 +58,11 @@ int timecompare_offset(struct timecompare *sync, int index; int num_samples = sync->num_samples; - if (num_samples > sizeof(buffer)/sizeof(buffer[0])) { + if (num_samples > ARRAY_SIZE(buffer)) { samples = kmalloc(sizeof(*samples) * num_samples, GFP_ATOMIC); if (!samples) { samples = buffer; - num_samples = sizeof(buffer)/sizeof(buffer[0]); + num_samples = ARRAY_SIZE(buffer); } } else { samples = buffer; -- cgit v1.2.3 From dd6414b50fa2b1cd247a8aa8f8bd42414b7453e1 Mon Sep 17 00:00:00 2001 From: Phil Carmody Date: Wed, 20 Oct 2010 15:57:33 -0700 Subject: timer: Permit statically-declared work with deferrable timers Currently, you have to just define a delayed_work uninitialised, and then initialise it before first use. That's a tad clumsy. At risk of playing mind-games with the compiler, fooling it into doing pointer arithmetic with compile-time-constants, this lets clients properly initialise delayed work with deferrable timers statically. This patch was inspired by the issues which lead Artem Bityutskiy to commit 8eab945c5616fc984 ("sunrpc: make the cache cleaner workqueue deferrable"). Signed-off-by: Phil Carmody Acked-by: Artem Bityutskiy Cc: Arjan van de Ven Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- include/linux/timer.h | 25 +++++++++++++++++++++++++ include/linux/workqueue.h | 8 ++++++++ kernel/timer.c | 15 +-------------- 3 files changed, 34 insertions(+), 14 deletions(-) (limited to 'kernel') diff --git a/include/linux/timer.h b/include/linux/timer.h index 1794674c1a52..cbfb7a355d30 100644 --- a/include/linux/timer.h +++ b/include/linux/timer.h @@ -48,6 +48,18 @@ extern struct tvec_base boot_tvec_bases; #define __TIMER_LOCKDEP_MAP_INITIALIZER(_kn) #endif +/* + * Note that all tvec_bases are 2 byte aligned and lower bit of + * base in timer_list is guaranteed to be zero. Use the LSB to + * indicate whether the timer is deferrable. + * + * A deferrable timer will work normally when the system is busy, but + * will not cause a CPU to come out of idle just to service it; instead, + * the timer will be serviced when the CPU eventually wakes up with a + * subsequent non-deferrable timer. + */ +#define TBASE_DEFERRABLE_FLAG (0x1) + #define TIMER_INITIALIZER(_function, _expires, _data) { \ .entry = { .prev = TIMER_ENTRY_STATIC }, \ .function = (_function), \ @@ -59,6 +71,19 @@ extern struct tvec_base boot_tvec_bases; __FILE__ ":" __stringify(__LINE__)) \ } +#define TBASE_MAKE_DEFERRED(ptr) ((struct tvec_base *) \ + ((unsigned char *)(ptr) + TBASE_DEFERRABLE_FLAG)) + +#define TIMER_DEFERRED_INITIALIZER(_function, _expires, _data) {\ + .entry = { .prev = TIMER_ENTRY_STATIC }, \ + .function = (_function), \ + .expires = (_expires), \ + .data = (_data), \ + .base = TBASE_MAKE_DEFERRED(&boot_tvec_bases), \ + __TIMER_LOCKDEP_MAP_INITIALIZER( \ + __FILE__ ":" __stringify(__LINE__)) \ + } + #define DEFINE_TIMER(_name, _function, _expires, _data) \ struct timer_list _name = \ TIMER_INITIALIZER(_function, _expires, _data) diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h index f11100f96482..88238c15ec3e 100644 --- a/include/linux/workqueue.h +++ b/include/linux/workqueue.h @@ -127,12 +127,20 @@ struct execute_work { .timer = TIMER_INITIALIZER(NULL, 0, 0), \ } +#define __DEFERRED_WORK_INITIALIZER(n, f) { \ + .work = __WORK_INITIALIZER((n).work, (f)), \ + .timer = TIMER_DEFERRED_INITIALIZER(NULL, 0, 0), \ + } + #define DECLARE_WORK(n, f) \ struct work_struct n = __WORK_INITIALIZER(n, f) #define DECLARE_DELAYED_WORK(n, f) \ struct delayed_work n = __DELAYED_WORK_INITIALIZER(n, f) +#define DECLARE_DEFERRED_WORK(n, f) \ + struct delayed_work n = __DEFERRED_WORK_INITIALIZER(n, f) + /* * initialize a work item's function pointer */ diff --git a/kernel/timer.c b/kernel/timer.c index 97bf05baade7..72853b256ff2 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -88,18 +88,6 @@ struct tvec_base boot_tvec_bases; EXPORT_SYMBOL(boot_tvec_bases); static DEFINE_PER_CPU(struct tvec_base *, tvec_bases) = &boot_tvec_bases; -/* - * Note that all tvec_bases are 2 byte aligned and lower bit of - * base in timer_list is guaranteed to be zero. Use the LSB to - * indicate whether the timer is deferrable. - * - * A deferrable timer will work normally when the system is busy, but - * will not cause a CPU to come out of idle just to service it; instead, - * the timer will be serviced when the CPU eventually wakes up with a - * subsequent non-deferrable timer. - */ -#define TBASE_DEFERRABLE_FLAG (0x1) - /* Functions below help us manage 'deferrable' flag */ static inline unsigned int tbase_get_deferrable(struct tvec_base *base) { @@ -113,8 +101,7 @@ static inline struct tvec_base *tbase_get_base(struct tvec_base *base) static inline void timer_set_deferrable(struct timer_list *timer) { - timer->base = ((struct tvec_base *)((unsigned long)(timer->base) | - TBASE_DEFERRABLE_FLAG)); + timer->base = TBASE_MAKE_DEFERRED(timer->base); } static inline void -- cgit v1.2.3 From 20f33a03f0cf87e51165f7084f697acfb68e865b Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 20 Oct 2010 15:57:34 -0700 Subject: posix-timers: Annotate lock_timer() lock_timer() conditionally grabs it_lock in case of returning non-NULL but unlock_timer() releases it unconditionally. This leads sparse to complain about the lock context imbalance. Rename and wrap lock_timer using __cond_lock() macro to make sparse happy. Signed-off-by: Namhyung Kim Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- kernel/posix-timers.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/posix-timers.c b/kernel/posix-timers.c index 9ca4973f736d..93bd2eb2bc53 100644 --- a/kernel/posix-timers.c +++ b/kernel/posix-timers.c @@ -145,7 +145,13 @@ static int common_timer_del(struct k_itimer *timer); static enum hrtimer_restart posix_timer_fn(struct hrtimer *data); -static struct k_itimer *lock_timer(timer_t timer_id, unsigned long *flags); +static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags); + +#define lock_timer(tid, flags) \ +({ struct k_itimer *__timr; \ + __cond_lock(&__timr->it_lock, __timr = __lock_timer(tid, flags)); \ + __timr; \ +}) static inline void unlock_timer(struct k_itimer *timr, unsigned long flags) { @@ -619,7 +625,7 @@ out: * the find to the timer lock. To avoid a dead lock, the timer id MUST * be release with out holding the timer lock. */ -static struct k_itimer *lock_timer(timer_t timer_id, unsigned long *flags) +static struct k_itimer *__lock_timer(timer_t timer_id, unsigned long *flags) { struct k_itimer *timr; /* -- cgit v1.2.3 From 6f1bc451e6a79470b122a37ee1fc6bbca450f444 Mon Sep 17 00:00:00 2001 From: Yong Zhang Date: Wed, 20 Oct 2010 15:57:31 -0700 Subject: timer: Make try_to_del_timer_sync() the same on SMP and UP On UP try_to_del_timer_sync() is mapped to del_timer() which does not take the running timer callback into account, so it has different semantics. Remove the SMP dependency of try_to_del_timer_sync() by using base->running_timer in the UP case as well. [ tglx: Removed set_running_timer() inline and tweaked the changelog ] Signed-off-by: Yong Zhang Cc: Ingo Molnar Cc: Peter Zijlstra Acked-by: Oleg Nesterov Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- include/linux/timer.h | 4 ++-- kernel/timer.c | 17 +++-------------- 2 files changed, 5 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/include/linux/timer.h b/include/linux/timer.h index cbfb7a355d30..6abd9138beda 100644 --- a/include/linux/timer.h +++ b/include/linux/timer.h @@ -274,11 +274,11 @@ static inline void timer_stats_timer_clear_start_info(struct timer_list *timer) extern void add_timer(struct timer_list *timer); +extern int try_to_del_timer_sync(struct timer_list *timer); + #ifdef CONFIG_SMP - extern int try_to_del_timer_sync(struct timer_list *timer); extern int del_timer_sync(struct timer_list *timer); #else -# define try_to_del_timer_sync(t) del_timer(t) # define del_timer_sync(t) del_timer(t) #endif diff --git a/kernel/timer.c b/kernel/timer.c index 72853b256ff2..47b86c1e3226 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -330,15 +330,6 @@ void set_timer_slack(struct timer_list *timer, int slack_hz) } EXPORT_SYMBOL_GPL(set_timer_slack); - -static inline void set_running_timer(struct tvec_base *base, - struct timer_list *timer) -{ -#ifdef CONFIG_SMP - base->running_timer = timer; -#endif -} - static void internal_add_timer(struct tvec_base *base, struct timer_list *timer) { unsigned long expires = timer->expires; @@ -923,15 +914,12 @@ int del_timer(struct timer_list *timer) } EXPORT_SYMBOL(del_timer); -#ifdef CONFIG_SMP /** * try_to_del_timer_sync - Try to deactivate a timer * @timer: timer do del * * This function tries to deactivate a timer. Upon successful (ret >= 0) * exit the timer is not queued and the handler is not running on any CPU. - * - * It must not be called from interrupt contexts. */ int try_to_del_timer_sync(struct timer_list *timer) { @@ -960,6 +948,7 @@ out: } EXPORT_SYMBOL(try_to_del_timer_sync); +#ifdef CONFIG_SMP /** * del_timer_sync - deactivate a timer and wait for the handler to finish. * @timer: the timer to be deactivated @@ -1098,7 +1087,7 @@ static inline void __run_timers(struct tvec_base *base) timer_stats_account_timer(timer); - set_running_timer(base, timer); + base->running_timer = timer; detach_timer(timer, 1); spin_unlock_irq(&base->lock); @@ -1106,7 +1095,7 @@ static inline void __run_timers(struct tvec_base *base) spin_lock_irq(&base->lock); } } - set_running_timer(base, NULL); + base->running_timer = NULL; spin_unlock_irq(&base->lock); } -- cgit v1.2.3 From 1118e2cd33d47254854e1ba3ba8e32802ff14fdf Mon Sep 17 00:00:00 2001 From: Yong Zhang Date: Wed, 20 Oct 2010 15:57:32 -0700 Subject: timer: Del_timer_sync() can be used in softirq context Actually we have used del_timer_sync() in softirq context for a long time, e.g. in __dst_free()::cancel_delayed_work(). So change the comments of it to warn on hardirq context only, and make lockdep know about this change. Signed-off-by: Yong Zhang Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Oleg Nesterov Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- kernel/timer.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/timer.c b/kernel/timer.c index 47b86c1e3226..612de0306e79 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -959,7 +959,7 @@ EXPORT_SYMBOL(try_to_del_timer_sync); * * Synchronization rules: Callers must prevent restarting of the timer, * otherwise this function is meaningless. It must not be called from - * interrupt contexts. The caller must not hold locks which would prevent + * hardirq contexts. The caller must not hold locks which would prevent * completion of the timer's handler. The timer's handler must not call * add_timer_on(). Upon exit the timer is not queued and the handler is * not running on any CPU. @@ -969,12 +969,10 @@ EXPORT_SYMBOL(try_to_del_timer_sync); int del_timer_sync(struct timer_list *timer) { #ifdef CONFIG_LOCKDEP - unsigned long flags; - - local_irq_save(flags); + local_bh_disable(); lock_map_acquire(&timer->lockdep_map); lock_map_release(&timer->lockdep_map); - local_irq_restore(flags); + local_bh_enable(); #endif for (;;) { -- cgit v1.2.3 From 466bd3030973910118ca601da8072be97a1e2209 Mon Sep 17 00:00:00 2001 From: Yong Zhang Date: Wed, 20 Oct 2010 15:57:33 -0700 Subject: timer: Warn when del_timer_sync() is called in hardirq context Add explict warning when del_timer_sync() is called in hardirq context. Signed-off-by: Yong Zhang Cc: Ingo Molnar Cc: Peter Zijlstra Acked-by: Oleg Nesterov Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- kernel/timer.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/timer.c b/kernel/timer.c index 612de0306e79..483e54ba5c93 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -974,7 +974,11 @@ int del_timer_sync(struct timer_list *timer) lock_map_release(&timer->lockdep_map); local_bh_enable(); #endif - + /* + * don't use it in hardirq context, because it + * could lead to deadlock. + */ + WARN_ON(in_irq()); for (;;) { int ret = try_to_del_timer_sync(timer); if (ret >= 0) -- cgit v1.2.3 From fe7de49f9d4e53f24ec9ef762a503f70b562341c Mon Sep 17 00:00:00 2001 From: KOSAKI Motohiro Date: Wed, 20 Oct 2010 16:01:12 -0700 Subject: sched: Make sched_param argument static in sched_setscheduler() callers Andrew Morton pointed out almost all sched_setscheduler() callers are using fixed parameters and can be converted to static. It reduces runtime memory use a little. Signed-off-by: KOSAKI Motohiro Reported-by: Andrew Morton Acked-by: James Morris Cc: Ingo Molnar Cc: Steven Rostedt Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- include/linux/sched.h | 5 +++-- kernel/irq/manage.c | 4 +++- kernel/kthread.c | 2 +- kernel/sched.c | 6 +++--- kernel/softirq.c | 4 +++- kernel/trace/trace_selftest.c | 2 +- kernel/watchdog.c | 2 +- 7 files changed, 15 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 0383601a927c..849c8670583d 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1942,9 +1942,10 @@ extern int task_nice(const struct task_struct *p); extern int can_nice(const struct task_struct *p, const int nice); extern int task_curr(const struct task_struct *p); extern int idle_cpu(int cpu); -extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); +extern int sched_setscheduler(struct task_struct *, int, + const struct sched_param *); extern int sched_setscheduler_nocheck(struct task_struct *, int, - struct sched_param *); + const struct sched_param *); extern struct task_struct *idle_task(int cpu); extern struct task_struct *curr_task(int cpu); extern void set_curr_task(int cpu, struct task_struct *p); diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 644e8d5fa367..850f030fa0c2 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -573,7 +573,9 @@ irq_thread_check_affinity(struct irq_desc *desc, struct irqaction *action) { } */ static int irq_thread(void *data) { - struct sched_param param = { .sched_priority = MAX_USER_RT_PRIO/2, }; + static struct sched_param param = { + .sched_priority = MAX_USER_RT_PRIO/2, + }; struct irqaction *action = data; struct irq_desc *desc = irq_to_desc(action->irq); int wake, oneshot = desc->status & IRQ_ONESHOT; diff --git a/kernel/kthread.c b/kernel/kthread.c index 2dc3786349d1..74cf6f5e7ade 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -148,7 +148,7 @@ struct task_struct *kthread_create(int (*threadfn)(void *data), wait_for_completion(&create.done); if (!IS_ERR(create.result)) { - struct sched_param param = { .sched_priority = 0 }; + static struct sched_param param = { .sched_priority = 0 }; va_list args; va_start(args, namefmt); diff --git a/kernel/sched.c b/kernel/sched.c index d42992bccdfa..51944e8c38a8 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4701,7 +4701,7 @@ static bool check_same_owner(struct task_struct *p) } static int __sched_setscheduler(struct task_struct *p, int policy, - struct sched_param *param, bool user) + const struct sched_param *param, bool user) { int retval, oldprio, oldpolicy = -1, on_rq, running; unsigned long flags; @@ -4856,7 +4856,7 @@ recheck: * NOTE that the task may be already dead. */ int sched_setscheduler(struct task_struct *p, int policy, - struct sched_param *param) + const struct sched_param *param) { return __sched_setscheduler(p, policy, param, true); } @@ -4874,7 +4874,7 @@ EXPORT_SYMBOL_GPL(sched_setscheduler); * but our caller might not have that capability. */ int sched_setscheduler_nocheck(struct task_struct *p, int policy, - struct sched_param *param) + const struct sched_param *param) { return __sched_setscheduler(p, policy, param, false); } diff --git a/kernel/softirq.c b/kernel/softirq.c index fc978889b194..081869ed3a9f 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -851,7 +851,9 @@ static int __cpuinit cpu_callback(struct notifier_block *nfb, cpumask_any(cpu_online_mask)); case CPU_DEAD: case CPU_DEAD_FROZEN: { - struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 }; + static struct sched_param param = { + .sched_priority = MAX_RT_PRIO-1 + }; p = per_cpu(ksoftirqd, hotcpu); per_cpu(ksoftirqd, hotcpu) = NULL; diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 155a415b3209..562c56e048fd 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -558,7 +558,7 @@ trace_selftest_startup_nop(struct tracer *trace, struct trace_array *tr) static int trace_wakeup_test_thread(void *data) { /* Make this a RT thread, doesn't need to be too high */ - struct sched_param param = { .sched_priority = 5 }; + static struct sched_param param = { .sched_priority = 5 }; struct completion *x = data; sched_setscheduler(current, SCHED_FIFO, ¶m); diff --git a/kernel/watchdog.c b/kernel/watchdog.c index bafba687a6d8..94ca779aa9c2 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -307,7 +307,7 @@ static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer) */ static int watchdog(void *unused) { - struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 }; + static struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 }; struct hrtimer *hrtimer = &__raw_get_cpu_var(watchdog_hrtimer); sched_setscheduler(current, SCHED_FIFO, ¶m); -- cgit v1.2.3 From 3cf9b85b474e656a0856b88290c7a289ac5ea247 Mon Sep 17 00:00:00 2001 From: Joe Perches Date: Fri, 5 Nov 2010 16:12:38 -0700 Subject: locking, lockdep: Convert sprintf_symbol to %pS Signed-off-by: Joe Perches Cc: Peter Zijlstra Cc: Jiri Kosina LKML-Reference: <1288998760-11775-6-git-send-email-joe@perches.com> Signed-off-by: Ingo Molnar --- kernel/lockdep_proc.c | 16 ++++++---------- 1 file changed, 6 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/lockdep_proc.c b/kernel/lockdep_proc.c index 59b76c8ce9d7..1969d2fc4b36 100644 --- a/kernel/lockdep_proc.c +++ b/kernel/lockdep_proc.c @@ -494,7 +494,6 @@ static void seq_stats(struct seq_file *m, struct lock_stat_data *data) namelen += 2; for (i = 0; i < LOCKSTAT_POINTS; i++) { - char sym[KSYM_SYMBOL_LEN]; char ip[32]; if (class->contention_point[i] == 0) @@ -503,15 +502,13 @@ static void seq_stats(struct seq_file *m, struct lock_stat_data *data) if (!i) seq_line(m, '-', 40-namelen, namelen); - sprint_symbol(sym, class->contention_point[i]); snprintf(ip, sizeof(ip), "[<%p>]", (void *)class->contention_point[i]); - seq_printf(m, "%40s %14lu %29s %s\n", name, - stats->contention_point[i], - ip, sym); + seq_printf(m, "%40s %14lu %29s %pS\n", + name, stats->contention_point[i], + ip, (void *)class->contention_point[i]); } for (i = 0; i < LOCKSTAT_POINTS; i++) { - char sym[KSYM_SYMBOL_LEN]; char ip[32]; if (class->contending_point[i] == 0) @@ -520,12 +517,11 @@ static void seq_stats(struct seq_file *m, struct lock_stat_data *data) if (!i) seq_line(m, '-', 40-namelen, namelen); - sprint_symbol(sym, class->contending_point[i]); snprintf(ip, sizeof(ip), "[<%p>]", (void *)class->contending_point[i]); - seq_printf(m, "%40s %14lu %29s %s\n", name, - stats->contending_point[i], - ip, sym); + seq_printf(m, "%40s %14lu %29s %pS\n", + name, stats->contending_point[i], + ip, (void *)class->contending_point[i]); } if (i) { seq_puts(m, "\n"); -- cgit v1.2.3 From ae791a2d2e382adc69990a144a7f1a6c4bc24f1e Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 10 Nov 2010 13:30:36 +0100 Subject: futex: Cleanup stale fshared flag interfaces The fast GUP changes stopped using the fshared flag in put_futex_keys(), but we kept the interface the same. Cleanup all stale users. This patch is split out from Darren Harts combo patch which also combines various flags. This way the changes are clearly separated. Signed-off-by: Thomas Gleixner Cc: Darren Hart LKML-Reference: <1289250609-16304-1-git-send-email-dvhart@linux.intel.com> --- kernel/futex.c | 64 +++++++++++++++++++++++++++------------------------------- 1 file changed, 30 insertions(+), 34 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 40a8777a27d0..38cf606a2d7d 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -283,8 +283,7 @@ again: return 0; } -static inline -void put_futex_key(int fshared, union futex_key *key) +static inline void put_futex_key(union futex_key *key) { drop_futex_key_refs(key); } @@ -907,7 +906,7 @@ static int futex_wake(u32 __user *uaddr, int fshared, int nr_wake, u32 bitset) } spin_unlock(&hb->lock); - put_futex_key(fshared, &key); + put_futex_key(&key); out: return ret; } @@ -965,8 +964,8 @@ retry_private: if (!fshared) goto retry_private; - put_futex_key(fshared, &key2); - put_futex_key(fshared, &key1); + put_futex_key(&key2); + put_futex_key(&key1); goto retry; } @@ -996,9 +995,9 @@ retry_private: double_unlock_hb(hb1, hb2); out_put_keys: - put_futex_key(fshared, &key2); + put_futex_key(&key2); out_put_key1: - put_futex_key(fshared, &key1); + put_futex_key(&key1); out: return ret; } @@ -1219,8 +1218,8 @@ retry_private: if (!fshared) goto retry_private; - put_futex_key(fshared, &key2); - put_futex_key(fshared, &key1); + put_futex_key(&key2); + put_futex_key(&key1); goto retry; } if (curval != *cmpval) { @@ -1260,8 +1259,8 @@ retry_private: break; case -EFAULT: double_unlock_hb(hb1, hb2); - put_futex_key(fshared, &key2); - put_futex_key(fshared, &key1); + put_futex_key(&key2); + put_futex_key(&key1); ret = fault_in_user_writeable(uaddr2); if (!ret) goto retry; @@ -1269,8 +1268,8 @@ retry_private: case -EAGAIN: /* The owner was exiting, try again. */ double_unlock_hb(hb1, hb2); - put_futex_key(fshared, &key2); - put_futex_key(fshared, &key1); + put_futex_key(&key2); + put_futex_key(&key1); cond_resched(); goto retry; default: @@ -1352,9 +1351,9 @@ out_unlock: drop_futex_key_refs(&key1); out_put_keys: - put_futex_key(fshared, &key2); + put_futex_key(&key2); out_put_key1: - put_futex_key(fshared, &key1); + put_futex_key(&key1); out: if (pi_state != NULL) free_pi_state(pi_state); @@ -1494,7 +1493,7 @@ static void unqueue_me_pi(struct futex_q *q) * private futexes. */ static int fixup_pi_state_owner(u32 __user *uaddr, struct futex_q *q, - struct task_struct *newowner, int fshared) + struct task_struct *newowner) { u32 newtid = task_pid_vnr(newowner) | FUTEX_WAITERS; struct futex_pi_state *pi_state = q->pi_state; @@ -1600,7 +1599,6 @@ static long futex_wait_restart(struct restart_block *restart); /** * fixup_owner() - Post lock pi_state and corner case management * @uaddr: user address of the futex - * @fshared: whether the futex is shared (1) or not (0) * @q: futex_q (contains pi_state and access to the rt_mutex) * @locked: if the attempt to take the rt_mutex succeeded (1) or not (0) * @@ -1613,8 +1611,7 @@ static long futex_wait_restart(struct restart_block *restart); * 0 - success, lock not taken * <0 - on error (-EFAULT) */ -static int fixup_owner(u32 __user *uaddr, int fshared, struct futex_q *q, - int locked) +static int fixup_owner(u32 __user *uaddr, struct futex_q *q, int locked) { struct task_struct *owner; int ret = 0; @@ -1625,7 +1622,7 @@ static int fixup_owner(u32 __user *uaddr, int fshared, struct futex_q *q, * did a lock-steal - fix up the PI-state in that case: */ if (q->pi_state->owner != current) - ret = fixup_pi_state_owner(uaddr, q, current, fshared); + ret = fixup_pi_state_owner(uaddr, q, current); goto out; } @@ -1652,7 +1649,7 @@ static int fixup_owner(u32 __user *uaddr, int fshared, struct futex_q *q, * lock. Fix the state up. */ owner = rt_mutex_owner(&q->pi_state->pi_mutex); - ret = fixup_pi_state_owner(uaddr, q, owner, fshared); + ret = fixup_pi_state_owner(uaddr, q, owner); goto out; } @@ -1772,7 +1769,7 @@ retry_private: if (!fshared) goto retry_private; - put_futex_key(fshared, &q->key); + put_futex_key(&q->key); goto retry; } @@ -1783,7 +1780,7 @@ retry_private: out: if (ret) - put_futex_key(fshared, &q->key); + put_futex_key(&q->key); return ret; } @@ -1941,7 +1938,7 @@ retry_private: * exit to complete. */ queue_unlock(&q, hb); - put_futex_key(fshared, &q.key); + put_futex_key(&q.key); cond_resched(); goto retry; default: @@ -1971,7 +1968,7 @@ retry_private: * Fixup the pi_state owner and possibly acquire the lock if we * haven't already. */ - res = fixup_owner(uaddr, fshared, &q, !ret); + res = fixup_owner(uaddr, &q, !ret); /* * If fixup_owner() returned an error, proprogate that. If it acquired * the lock, clear our -ETIMEDOUT or -EINTR. @@ -1995,7 +1992,7 @@ out_unlock_put_key: queue_unlock(&q, hb); out_put_key: - put_futex_key(fshared, &q.key); + put_futex_key(&q.key); out: if (to) destroy_hrtimer_on_stack(&to->timer); @@ -2011,7 +2008,7 @@ uaddr_faulted: if (!fshared) goto retry_private; - put_futex_key(fshared, &q.key); + put_futex_key(&q.key); goto retry; } @@ -2093,14 +2090,14 @@ retry: out_unlock: spin_unlock(&hb->lock); - put_futex_key(fshared, &key); + put_futex_key(&key); out: return ret; pi_faulted: spin_unlock(&hb->lock); - put_futex_key(fshared, &key); + put_futex_key(&key); ret = fault_in_user_writeable(uaddr); if (!ret) @@ -2273,8 +2270,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, */ if (q.pi_state && (q.pi_state->owner != current)) { spin_lock(q.lock_ptr); - ret = fixup_pi_state_owner(uaddr2, &q, current, - fshared); + ret = fixup_pi_state_owner(uaddr2, &q, current); spin_unlock(q.lock_ptr); } } else { @@ -2293,7 +2289,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, * Fixup the pi_state owner and possibly acquire the lock if we * haven't already. */ - res = fixup_owner(uaddr2, fshared, &q, !ret); + res = fixup_owner(uaddr2, &q, !ret); /* * If fixup_owner() returned an error, proprogate that. If it * acquired the lock, clear -ETIMEDOUT or -EINTR. @@ -2324,9 +2320,9 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, } out_put_keys: - put_futex_key(fshared, &q.key); + put_futex_key(&q.key); out_key2: - put_futex_key(fshared, &key2); + put_futex_key(&key2); out: if (to) { -- cgit v1.2.3 From b41277dc7a18ee332d9e8078e978bacdf6e76157 Mon Sep 17 00:00:00 2001 From: Darren Hart Date: Mon, 8 Nov 2010 13:10:09 -0800 Subject: futex: Replace fshared and clockrt with combined flags In the early days we passed the mmap sem around. That became the "int fshared" with the fast gup improvements. Then we added "int clockrt" in places. This patch unifies these options as "flags". [ tglx: Split out the stale fshared cleanup ] Signed-off-by: Darren Hart Cc: Peter Zijlstra Cc: Eric Dumazet Cc: John Kacur Cc: Ingo Molnar LKML-Reference: <1289250609-16304-1-git-send-email-dvhart@linux.intel.com> Signed-off-by: Thomas Gleixner --- kernel/futex.c | 146 +++++++++++++++++++++++++++------------------------------ 1 file changed, 70 insertions(+), 76 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 38cf606a2d7d..87ad28746e17 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -68,6 +68,14 @@ int __read_mostly futex_cmpxchg_enabled; #define FUTEX_HASHBITS (CONFIG_BASE_SMALL ? 4 : 8) +/* + * Futex flags used to encode options to functions and preserve them across + * restarts. + */ +#define FLAGS_SHARED 0x01 +#define FLAGS_CLOCKRT 0x02 +#define FLAGS_HAS_TIMEOUT 0x04 + /* * Priority Inheritance state: */ @@ -869,7 +877,8 @@ double_unlock_hb(struct futex_hash_bucket *hb1, struct futex_hash_bucket *hb2) /* * Wake up waiters matching bitset queued on this futex (uaddr). */ -static int futex_wake(u32 __user *uaddr, int fshared, int nr_wake, u32 bitset) +static int +futex_wake(u32 __user *uaddr, unsigned int flags, int nr_wake, u32 bitset) { struct futex_hash_bucket *hb; struct futex_q *this, *next; @@ -880,7 +889,7 @@ static int futex_wake(u32 __user *uaddr, int fshared, int nr_wake, u32 bitset) if (!bitset) return -EINVAL; - ret = get_futex_key(uaddr, fshared, &key); + ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &key); if (unlikely(ret != 0)) goto out; @@ -916,7 +925,7 @@ out: * to this virtual address: */ static int -futex_wake_op(u32 __user *uaddr1, int fshared, u32 __user *uaddr2, +futex_wake_op(u32 __user *uaddr1, unsigned int flags, u32 __user *uaddr2, int nr_wake, int nr_wake2, int op) { union futex_key key1 = FUTEX_KEY_INIT, key2 = FUTEX_KEY_INIT; @@ -926,10 +935,10 @@ futex_wake_op(u32 __user *uaddr1, int fshared, u32 __user *uaddr2, int ret, op_ret; retry: - ret = get_futex_key(uaddr1, fshared, &key1); + ret = get_futex_key(uaddr1, flags & FLAGS_SHARED, &key1); if (unlikely(ret != 0)) goto out; - ret = get_futex_key(uaddr2, fshared, &key2); + ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2); if (unlikely(ret != 0)) goto out_put_key1; @@ -961,7 +970,7 @@ retry_private: if (ret) goto out_put_keys; - if (!fshared) + if (!(flags & FLAGS_SHARED)) goto retry_private; put_futex_key(&key2); @@ -1132,13 +1141,13 @@ static int futex_proxy_trylock_atomic(u32 __user *pifutex, /** * futex_requeue() - Requeue waiters from uaddr1 to uaddr2 * @uaddr1: source futex user address - * @fshared: 0 for a PROCESS_PRIVATE futex, 1 for PROCESS_SHARED + * @flags: futex flags (FLAGS_SHARED, etc.) * @uaddr2: target futex user address * @nr_wake: number of waiters to wake (must be 1 for requeue_pi) * @nr_requeue: number of waiters to requeue (0-INT_MAX) * @cmpval: @uaddr1 expected value (or %NULL) * @requeue_pi: if we are attempting to requeue from a non-pi futex to a - * pi futex (pi to pi requeue is not supported) + * pi futex (pi to pi requeue is not supported) * * Requeue waiters on uaddr1 to uaddr2. In the requeue_pi case, try to acquire * uaddr2 atomically on behalf of the top waiter. @@ -1147,9 +1156,9 @@ static int futex_proxy_trylock_atomic(u32 __user *pifutex, * >=0 - on success, the number of tasks requeued or woken * <0 - on error */ -static int futex_requeue(u32 __user *uaddr1, int fshared, u32 __user *uaddr2, - int nr_wake, int nr_requeue, u32 *cmpval, - int requeue_pi) +static int futex_requeue(u32 __user *uaddr1, unsigned int flags, + u32 __user *uaddr2, int nr_wake, int nr_requeue, + u32 *cmpval, int requeue_pi) { union futex_key key1 = FUTEX_KEY_INIT, key2 = FUTEX_KEY_INIT; int drop_count = 0, task_count = 0, ret; @@ -1190,10 +1199,10 @@ retry: pi_state = NULL; } - ret = get_futex_key(uaddr1, fshared, &key1); + ret = get_futex_key(uaddr1, flags & FLAGS_SHARED, &key1); if (unlikely(ret != 0)) goto out; - ret = get_futex_key(uaddr2, fshared, &key2); + ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2); if (unlikely(ret != 0)) goto out_put_key1; @@ -1215,7 +1224,7 @@ retry_private: if (ret) goto out_put_keys; - if (!fshared) + if (!(flags & FLAGS_SHARED)) goto retry_private; put_futex_key(&key2); @@ -1586,14 +1595,6 @@ handle_fault: goto retry; } -/* - * In case we must use restart_block to restart a futex_wait, - * we encode in the 'flags' shared capability - */ -#define FLAGS_SHARED 0x01 -#define FLAGS_CLOCKRT 0x02 -#define FLAGS_HAS_TIMEOUT 0x04 - static long futex_wait_restart(struct restart_block *restart); /** @@ -1712,7 +1713,7 @@ static void futex_wait_queue_me(struct futex_hash_bucket *hb, struct futex_q *q, * futex_wait_setup() - Prepare to wait on a futex * @uaddr: the futex userspace address * @val: the expected value - * @fshared: whether the futex is shared (1) or not (0) + * @flags: futex flags (FLAGS_SHARED, etc.) * @q: the associated futex_q * @hb: storage for hash_bucket pointer to be returned to caller * @@ -1725,7 +1726,7 @@ static void futex_wait_queue_me(struct futex_hash_bucket *hb, struct futex_q *q, * 0 - uaddr contains val and hb has been locked * <1 - -EFAULT or -EWOULDBLOCK (uaddr does not contain val) and hb is unlcoked */ -static int futex_wait_setup(u32 __user *uaddr, u32 val, int fshared, +static int futex_wait_setup(u32 __user *uaddr, u32 val, unsigned int flags, struct futex_q *q, struct futex_hash_bucket **hb) { u32 uval; @@ -1750,7 +1751,7 @@ static int futex_wait_setup(u32 __user *uaddr, u32 val, int fshared, */ retry: q->key = FUTEX_KEY_INIT; - ret = get_futex_key(uaddr, fshared, &q->key); + ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q->key); if (unlikely(ret != 0)) return ret; @@ -1766,7 +1767,7 @@ retry_private: if (ret) goto out; - if (!fshared) + if (!(flags & FLAGS_SHARED)) goto retry_private; put_futex_key(&q->key); @@ -1784,8 +1785,8 @@ out: return ret; } -static int futex_wait(u32 __user *uaddr, int fshared, - u32 val, ktime_t *abs_time, u32 bitset, int clockrt) +static int futex_wait(u32 __user *uaddr, unsigned int flags, u32 val, + ktime_t *abs_time, u32 bitset) { struct hrtimer_sleeper timeout, *to = NULL; struct restart_block *restart; @@ -1804,8 +1805,9 @@ static int futex_wait(u32 __user *uaddr, int fshared, if (abs_time) { to = &timeout; - hrtimer_init_on_stack(&to->timer, clockrt ? CLOCK_REALTIME : - CLOCK_MONOTONIC, HRTIMER_MODE_ABS); + hrtimer_init_on_stack(&to->timer, (flags & FLAGS_CLOCKRT) ? + CLOCK_REALTIME : CLOCK_MONOTONIC, + HRTIMER_MODE_ABS); hrtimer_init_sleeper(to, current); hrtimer_set_expires_range_ns(&to->timer, *abs_time, current->timer_slack_ns); @@ -1816,7 +1818,7 @@ retry: * Prepare to wait on uaddr. On success, holds hb lock and increments * q.key refs. */ - ret = futex_wait_setup(uaddr, val, fshared, &q, &hb); + ret = futex_wait_setup(uaddr, val, flags, &q, &hb); if (ret) goto out; @@ -1849,12 +1851,7 @@ retry: restart->futex.val = val; restart->futex.time = abs_time->tv64; restart->futex.bitset = bitset; - restart->futex.flags = FLAGS_HAS_TIMEOUT; - - if (fshared) - restart->futex.flags |= FLAGS_SHARED; - if (clockrt) - restart->futex.flags |= FLAGS_CLOCKRT; + restart->futex.flags = flags; ret = -ERESTART_RESTARTBLOCK; @@ -1870,7 +1867,6 @@ out: static long futex_wait_restart(struct restart_block *restart) { u32 __user *uaddr = restart->futex.uaddr; - int fshared = 0; ktime_t t, *tp = NULL; if (restart->futex.flags & FLAGS_HAS_TIMEOUT) { @@ -1878,11 +1874,9 @@ static long futex_wait_restart(struct restart_block *restart) tp = &t; } restart->fn = do_no_restart_syscall; - if (restart->futex.flags & FLAGS_SHARED) - fshared = 1; - return (long)futex_wait(uaddr, fshared, restart->futex.val, tp, - restart->futex.bitset, - restart->futex.flags & FLAGS_CLOCKRT); + + return (long)futex_wait(uaddr, restart->futex.flags, + restart->futex.val, tp, restart->futex.bitset); } @@ -1892,8 +1886,8 @@ static long futex_wait_restart(struct restart_block *restart) * if there are waiters then it will block, it does PI, etc. (Due to * races the kernel might see a 0 value of the futex too.) */ -static int futex_lock_pi(u32 __user *uaddr, int fshared, - int detect, ktime_t *time, int trylock) +static int futex_lock_pi(u32 __user *uaddr, unsigned int flags, int detect, + ktime_t *time, int trylock) { struct hrtimer_sleeper timeout, *to = NULL; struct futex_hash_bucket *hb; @@ -1916,7 +1910,7 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared, q.requeue_pi_key = NULL; retry: q.key = FUTEX_KEY_INIT; - ret = get_futex_key(uaddr, fshared, &q.key); + ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q.key); if (unlikely(ret != 0)) goto out; @@ -2005,7 +1999,7 @@ uaddr_faulted: if (ret) goto out_put_key; - if (!fshared) + if (!(flags & FLAGS_SHARED)) goto retry_private; put_futex_key(&q.key); @@ -2017,7 +2011,7 @@ uaddr_faulted: * This is the in-kernel slowpath: we look up the PI state (if any), * and do the rt-mutex unlock. */ -static int futex_unlock_pi(u32 __user *uaddr, int fshared) +static int futex_unlock_pi(u32 __user *uaddr, unsigned int flags) { struct futex_hash_bucket *hb; struct futex_q *this, *next; @@ -2035,7 +2029,7 @@ retry: if ((uval & FUTEX_TID_MASK) != task_pid_vnr(current)) return -EPERM; - ret = get_futex_key(uaddr, fshared, &key); + ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &key); if (unlikely(ret != 0)) goto out; @@ -2157,7 +2151,7 @@ int handle_early_requeue_pi_wakeup(struct futex_hash_bucket *hb, /** * futex_wait_requeue_pi() - Wait on uaddr and take uaddr2 * @uaddr: the futex we initially wait on (non-pi) - * @fshared: whether the futexes are shared (1) or not (0). They must be + * @flags: futex flags (FLAGS_SHARED, FLAGS_CLOCKRT, etc.), they must be * the same type, no requeueing from private to shared, etc. * @val: the expected value of uaddr * @abs_time: absolute timeout @@ -2195,9 +2189,9 @@ int handle_early_requeue_pi_wakeup(struct futex_hash_bucket *hb, * 0 - On success * <0 - On error */ -static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, +static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags, u32 val, ktime_t *abs_time, u32 bitset, - int clockrt, u32 __user *uaddr2) + u32 __user *uaddr2) { struct hrtimer_sleeper timeout, *to = NULL; struct rt_mutex_waiter rt_waiter; @@ -2212,8 +2206,9 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, if (abs_time) { to = &timeout; - hrtimer_init_on_stack(&to->timer, clockrt ? CLOCK_REALTIME : - CLOCK_MONOTONIC, HRTIMER_MODE_ABS); + hrtimer_init_on_stack(&to->timer, (flags & FLAGS_CLOCKRT) ? + CLOCK_REALTIME : CLOCK_MONOTONIC, + HRTIMER_MODE_ABS); hrtimer_init_sleeper(to, current); hrtimer_set_expires_range_ns(&to->timer, *abs_time, current->timer_slack_ns); @@ -2227,7 +2222,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, rt_waiter.task = NULL; key2 = FUTEX_KEY_INIT; - ret = get_futex_key(uaddr2, fshared, &key2); + ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2); if (unlikely(ret != 0)) goto out; @@ -2240,7 +2235,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, * Prepare to wait on uaddr. On success, increments q.key (key1) ref * count. */ - ret = futex_wait_setup(uaddr, val, fshared, &q, &hb); + ret = futex_wait_setup(uaddr, val, flags, &q, &hb); if (ret) goto out_key2; @@ -2547,58 +2542,57 @@ void exit_robust_list(struct task_struct *curr) long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout, u32 __user *uaddr2, u32 val2, u32 val3) { - int clockrt, ret = -ENOSYS; - int cmd = op & FUTEX_CMD_MASK; - int fshared = 0; + int ret = -ENOSYS, cmd = op & FUTEX_CMD_MASK; + unsigned int flags = 0; if (!(op & FUTEX_PRIVATE_FLAG)) - fshared = 1; + flags |= FLAGS_SHARED; - clockrt = op & FUTEX_CLOCK_REALTIME; - if (clockrt && cmd != FUTEX_WAIT_BITSET && cmd != FUTEX_WAIT_REQUEUE_PI) - return -ENOSYS; + if (op & FUTEX_CLOCK_REALTIME) { + flags |= FLAGS_CLOCKRT; + if (cmd != FUTEX_WAIT_BITSET && cmd != FUTEX_WAIT_REQUEUE_PI) + return -ENOSYS; + } switch (cmd) { case FUTEX_WAIT: val3 = FUTEX_BITSET_MATCH_ANY; case FUTEX_WAIT_BITSET: - ret = futex_wait(uaddr, fshared, val, timeout, val3, clockrt); + ret = futex_wait(uaddr, flags, val, timeout, val3); break; case FUTEX_WAKE: val3 = FUTEX_BITSET_MATCH_ANY; case FUTEX_WAKE_BITSET: - ret = futex_wake(uaddr, fshared, val, val3); + ret = futex_wake(uaddr, flags, val, val3); break; case FUTEX_REQUEUE: - ret = futex_requeue(uaddr, fshared, uaddr2, val, val2, NULL, 0); + ret = futex_requeue(uaddr, flags, uaddr2, val, val2, NULL, 0); break; case FUTEX_CMP_REQUEUE: - ret = futex_requeue(uaddr, fshared, uaddr2, val, val2, &val3, - 0); + ret = futex_requeue(uaddr, flags, uaddr2, val, val2, &val3, 0); break; case FUTEX_WAKE_OP: - ret = futex_wake_op(uaddr, fshared, uaddr2, val, val2, val3); + ret = futex_wake_op(uaddr, flags, uaddr2, val, val2, val3); break; case FUTEX_LOCK_PI: if (futex_cmpxchg_enabled) - ret = futex_lock_pi(uaddr, fshared, val, timeout, 0); + ret = futex_lock_pi(uaddr, flags, val, timeout, 0); break; case FUTEX_UNLOCK_PI: if (futex_cmpxchg_enabled) - ret = futex_unlock_pi(uaddr, fshared); + ret = futex_unlock_pi(uaddr, flags); break; case FUTEX_TRYLOCK_PI: if (futex_cmpxchg_enabled) - ret = futex_lock_pi(uaddr, fshared, 0, timeout, 1); + ret = futex_lock_pi(uaddr, flags, 0, timeout, 1); break; case FUTEX_WAIT_REQUEUE_PI: val3 = FUTEX_BITSET_MATCH_ANY; - ret = futex_wait_requeue_pi(uaddr, fshared, val, timeout, val3, - clockrt, uaddr2); + ret = futex_wait_requeue_pi(uaddr, flags, val, timeout, val3, + uaddr2); break; case FUTEX_CMP_REQUEUE_PI: - ret = futex_requeue(uaddr, fshared, uaddr2, val, val2, &val3, - 1); + ret = futex_requeue(uaddr, flags, uaddr2, val, val2, &val3, 1); break; default: ret = -ENOSYS; -- cgit v1.2.3 From 5bdb05f91b27b9361c4f348a4e05999f597df72e Mon Sep 17 00:00:00 2001 From: Darren Hart Date: Mon, 8 Nov 2010 13:40:28 -0800 Subject: futex: Add futex_q static initializer The futex_q struct has grown considerably over the last couple years. I believe it now merits a static initializer to avoid uninitialized data errors (having spent more time than I care to admit debugging an uninitialized q.bitset in an experimental new op code). With the key initializer built in, several of the FUTEX_KEY_INIT calls can be removed. V2: use a static variable instead of an init macro. use a C99 initializer and don't rely on variable ordering in the struct. V3: make futex_q_init const Signed-off-by: Darren Hart Cc: Peter Zijlstra Cc: Eric Dumazet Cc: John Kacur Cc: Ingo Molnar LKML-Reference: <1289252428-18383-1-git-send-email-dvhart@linux.intel.com> Signed-off-by: Thomas Gleixner --- kernel/futex.c | 25 ++++++++++--------------- 1 file changed, 10 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 87ad28746e17..3019b92e6917 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -131,6 +131,12 @@ struct futex_q { u32 bitset; }; +static const struct futex_q futex_q_init = { + /* list gets initialized in queue_me()*/ + .key = FUTEX_KEY_INIT, + .bitset = FUTEX_BITSET_MATCH_ANY +}; + /* * Hash buckets are shared by all the futex_keys that hash to the same * location. Each key may have multiple futex_q structures, one for each task @@ -1750,7 +1756,6 @@ static int futex_wait_setup(u32 __user *uaddr, u32 val, unsigned int flags, * rare, but normal. */ retry: - q->key = FUTEX_KEY_INIT; ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q->key); if (unlikely(ret != 0)) return ret; @@ -1791,16 +1796,12 @@ static int futex_wait(u32 __user *uaddr, unsigned int flags, u32 val, struct hrtimer_sleeper timeout, *to = NULL; struct restart_block *restart; struct futex_hash_bucket *hb; - struct futex_q q; + struct futex_q q = futex_q_init; int ret; if (!bitset) return -EINVAL; - - q.pi_state = NULL; q.bitset = bitset; - q.rt_waiter = NULL; - q.requeue_pi_key = NULL; if (abs_time) { to = &timeout; @@ -1891,7 +1892,7 @@ static int futex_lock_pi(u32 __user *uaddr, unsigned int flags, int detect, { struct hrtimer_sleeper timeout, *to = NULL; struct futex_hash_bucket *hb; - struct futex_q q; + struct futex_q q = futex_q_init; int res, ret; if (refill_pi_state_cache()) @@ -1905,11 +1906,7 @@ static int futex_lock_pi(u32 __user *uaddr, unsigned int flags, int detect, hrtimer_set_expires(&to->timer, *time); } - q.pi_state = NULL; - q.rt_waiter = NULL; - q.requeue_pi_key = NULL; retry: - q.key = FUTEX_KEY_INIT; ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q.key); if (unlikely(ret != 0)) goto out; @@ -2197,8 +2194,8 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags, struct rt_mutex_waiter rt_waiter; struct rt_mutex *pi_mutex = NULL; struct futex_hash_bucket *hb; - union futex_key key2; - struct futex_q q; + union futex_key key2 = FUTEX_KEY_INIT; + struct futex_q q = futex_q_init; int res, ret; if (!bitset) @@ -2221,12 +2218,10 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags, debug_rt_mutex_init_waiter(&rt_waiter); rt_waiter.task = NULL; - key2 = FUTEX_KEY_INIT; ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2); if (unlikely(ret != 0)) goto out; - q.pi_state = NULL; q.bitset = bitset; q.rt_waiter = &rt_waiter; q.requeue_pi_key = &key2; -- cgit v1.2.3 From b2c0710c464ede15e1fc52fb1e7ee9ba54cea186 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Thu, 9 Sep 2010 13:40:39 -0700 Subject: rcu: move TINY_RCU from softirq to kthread If RCU priority boosting is to be meaningful, callback invocation must be boosted in addition to preempted RCU readers. Otherwise, in presence of CPU real-time threads, the grace period ends, but the callbacks don't get invoked. If the callbacks don't get invoked, the associated memory doesn't get freed, so the system is still subject to OOM. But it is not reasonable to priority-boost RCU_SOFTIRQ, so this commit moves the callback invocations to a kthread, which can be boosted easily. Signed-off-by: Paul E. McKenney Signed-off-by: Paul E. McKenney --- include/linux/rcupdate.h | 1 - include/linux/rcutiny.h | 8 ++---- include/linux/rcutree.h | 1 + kernel/rcutiny.c | 71 +++++++++++++++++++++++++++++++++++++++--------- kernel/rcutiny_plugin.h | 15 +++++----- 5 files changed, 70 insertions(+), 26 deletions(-) (limited to 'kernel') diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h index 03cda7bed985..7142ee3304ab 100644 --- a/include/linux/rcupdate.h +++ b/include/linux/rcupdate.h @@ -118,7 +118,6 @@ static inline int rcu_preempt_depth(void) #endif /* #else #ifdef CONFIG_PREEMPT_RCU */ /* Internal to kernel */ -extern void rcu_init(void); extern void rcu_sched_qs(int cpu); extern void rcu_bh_qs(int cpu); extern void rcu_check_callbacks(int cpu, int user); diff --git a/include/linux/rcutiny.h b/include/linux/rcutiny.h index 13877cb93a60..ea025a611fcc 100644 --- a/include/linux/rcutiny.h +++ b/include/linux/rcutiny.h @@ -27,7 +27,9 @@ #include -#define rcu_init_sched() do { } while (0) +static inline void rcu_init(void) +{ +} #ifdef CONFIG_TINY_RCU @@ -125,16 +127,12 @@ static inline void rcu_cpu_stall_reset(void) } #ifdef CONFIG_DEBUG_LOCK_ALLOC - extern int rcu_scheduler_active __read_mostly; extern void rcu_scheduler_starting(void); - #else /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ - static inline void rcu_scheduler_starting(void) { } - #endif /* #else #ifdef CONFIG_DEBUG_LOCK_ALLOC */ #endif /* __LINUX_RCUTINY_H */ diff --git a/include/linux/rcutree.h b/include/linux/rcutree.h index 95518e628794..c0e96833aa73 100644 --- a/include/linux/rcutree.h +++ b/include/linux/rcutree.h @@ -30,6 +30,7 @@ #ifndef __LINUX_RCUTREE_H #define __LINUX_RCUTREE_H +extern void rcu_init(void); extern void rcu_note_context_switch(int cpu); extern int rcu_needs_cpu(int cpu); extern void rcu_cpu_stall_reset(void); diff --git a/kernel/rcutiny.c b/kernel/rcutiny.c index d806735342ac..86eef29cdfb2 100644 --- a/kernel/rcutiny.c +++ b/kernel/rcutiny.c @@ -59,8 +59,15 @@ int rcu_scheduler_active __read_mostly; EXPORT_SYMBOL_GPL(rcu_scheduler_active); #endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ +/* Controls for rcu_cbs() kthread, replacing RCU_SOFTIRQ used previously. */ +static struct task_struct *rcu_cbs_task; +static DECLARE_WAIT_QUEUE_HEAD(rcu_cbs_wq); +static unsigned long have_rcu_cbs; +static void invoke_rcu_cbs(void); + /* Forward declarations for rcutiny_plugin.h. */ -static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp); +static void rcu_process_callbacks(struct rcu_ctrlblk *rcp); +static int rcu_cbs(void *arg); static void __call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu), struct rcu_ctrlblk *rcp); @@ -123,7 +130,7 @@ void rcu_sched_qs(int cpu) { if (rcu_qsctr_help(&rcu_sched_ctrlblk) + rcu_qsctr_help(&rcu_bh_ctrlblk)) - raise_softirq(RCU_SOFTIRQ); + invoke_rcu_cbs(); } /* @@ -132,7 +139,7 @@ void rcu_sched_qs(int cpu) void rcu_bh_qs(int cpu) { if (rcu_qsctr_help(&rcu_bh_ctrlblk)) - raise_softirq(RCU_SOFTIRQ); + invoke_rcu_cbs(); } /* @@ -152,10 +159,10 @@ void rcu_check_callbacks(int cpu, int user) } /* - * Helper function for rcu_process_callbacks() that operates on the - * specified rcu_ctrlkblk structure. + * Invoke the RCU callbacks on the specified rcu_ctrlkblk structure + * whose grace period has elapsed. */ -static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp) +static void rcu_process_callbacks(struct rcu_ctrlblk *rcp) { struct rcu_head *next, *list; unsigned long flags; @@ -180,19 +187,52 @@ static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp) next = list->next; prefetch(next); debug_rcu_head_unqueue(list); + local_bh_disable(); list->func(list); + local_bh_enable(); list = next; } } /* - * Invoke any callbacks whose grace period has completed. + * This kthread invokes RCU callbacks whose grace periods have + * elapsed. It is awakened as needed, and takes the place of the + * RCU_SOFTIRQ that was used previously for this purpose. + * This is a kthread, but it is never stopped, at least not until + * the system goes down. + */ +static int rcu_cbs(void *arg) +{ + unsigned long work; + unsigned long flags; + + for (;;) { + wait_event(rcu_cbs_wq, have_rcu_cbs != 0); + local_irq_save(flags); + work = have_rcu_cbs; + have_rcu_cbs = 0; + local_irq_restore(flags); + if (work) { + rcu_process_callbacks(&rcu_sched_ctrlblk); + rcu_process_callbacks(&rcu_bh_ctrlblk); + rcu_preempt_process_callbacks(); + } + } + + return 0; /* Not reached, but needed to shut gcc up. */ +} + +/* + * Wake up rcu_cbs() to process callbacks now eligible for invocation. */ -static void rcu_process_callbacks(struct softirq_action *unused) +static void invoke_rcu_cbs(void) { - __rcu_process_callbacks(&rcu_sched_ctrlblk); - __rcu_process_callbacks(&rcu_bh_ctrlblk); - rcu_preempt_process_callbacks(); + unsigned long flags; + + local_irq_save(flags); + have_rcu_cbs = 1; + wake_up(&rcu_cbs_wq); + local_irq_restore(flags); } /* @@ -282,7 +322,12 @@ void rcu_barrier_sched(void) } EXPORT_SYMBOL_GPL(rcu_barrier_sched); -void __init rcu_init(void) +/* + * Spawn the kthread that invokes RCU callbacks. + */ +static int __init rcu_spawn_kthreads(void) { - open_softirq(RCU_SOFTIRQ, rcu_process_callbacks); + rcu_cbs_task = kthread_run(rcu_cbs, NULL, "rcu_cbs"); + return 0; } +early_initcall(rcu_spawn_kthreads); diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h index 6ceca4f745ff..95f9239df512 100644 --- a/kernel/rcutiny_plugin.h +++ b/kernel/rcutiny_plugin.h @@ -22,6 +22,8 @@ * Author: Paul E. McKenney */ +#include + #ifdef CONFIG_TINY_PREEMPT_RCU #include @@ -164,9 +166,9 @@ static void rcu_preempt_cpu_qs(void) if (!rcu_preempt_blocked_readers_any()) rcu_preempt_ctrlblk.rcb.donetail = rcu_preempt_ctrlblk.nexttail; - /* If there are done callbacks, make RCU_SOFTIRQ process them. */ + /* If there are done callbacks, cause them to be invoked. */ if (*rcu_preempt_ctrlblk.rcb.donetail != NULL) - raise_softirq(RCU_SOFTIRQ); + invoke_rcu_cbs(); } /* @@ -374,7 +376,7 @@ static void rcu_preempt_check_callbacks(void) rcu_preempt_cpu_qs(); if (&rcu_preempt_ctrlblk.rcb.rcucblist != rcu_preempt_ctrlblk.rcb.donetail) - raise_softirq(RCU_SOFTIRQ); + invoke_rcu_cbs(); if (rcu_preempt_gp_in_progress() && rcu_cpu_blocking_cur_gp() && rcu_preempt_running_reader()) @@ -383,7 +385,7 @@ static void rcu_preempt_check_callbacks(void) /* * TINY_PREEMPT_RCU has an extra callback-list tail pointer to - * update, so this is invoked from __rcu_process_callbacks() to + * update, so this is invoked from rcu_process_callbacks() to * handle that case. Of course, it is invoked for all flavors of * RCU, but RCU callbacks can appear only on one of the lists, and * neither ->nexttail nor ->donetail can possibly be NULL, so there @@ -400,7 +402,7 @@ static void rcu_preempt_remove_callbacks(struct rcu_ctrlblk *rcp) */ static void rcu_preempt_process_callbacks(void) { - __rcu_process_callbacks(&rcu_preempt_ctrlblk.rcb); + rcu_process_callbacks(&rcu_preempt_ctrlblk.rcb); } /* @@ -599,14 +601,13 @@ static void rcu_preempt_process_callbacks(void) #endif /* #else #ifdef CONFIG_TINY_PREEMPT_RCU */ #ifdef CONFIG_DEBUG_LOCK_ALLOC - #include /* * During boot, we forgive RCU lockdep issues. After this function is * invoked, we start taking RCU lockdep issues seriously. */ -void rcu_scheduler_starting(void) +void __init rcu_scheduler_starting(void) { WARN_ON(nr_context_switches() > 0); rcu_scheduler_active = 1; -- cgit v1.2.3 From 48c5ccae88dcd989d9de507e8510313c6cbd352b Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sat, 13 Nov 2010 19:32:29 +0100 Subject: sched: Simplify cpu-hot-unplug task migration While discussing the need for sched_idle_next(), Oleg remarked that since try_to_wake_up() ensures sleeping tasks will end up running on a sane cpu, we can do away with migrate_live_tasks(). If we then extend the existing hack of migrating current from CPU_DYING to migrating the full rq worth of tasks from CPU_DYING, the need for the sched_idle_next() abomination disappears as well, since idle will be the only possible thread left after the migration thread stops. This greatly simplifies the hot-unplug task migration path, as can be seen from the resulting code reduction (and about half the new lines are comments). Suggested-by: Oleg Nesterov Signed-off-by: Peter Zijlstra LKML-Reference: <1289851597.2109.547.camel@laptop> Signed-off-by: Ingo Molnar --- include/linux/sched.h | 3 - kernel/cpu.c | 16 ++-- kernel/sched.c | 206 +++++++++++++++----------------------------------- 3 files changed, 67 insertions(+), 158 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 3cd70cf91fde..29d953abb5ad 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1871,14 +1871,11 @@ extern void sched_clock_idle_sleep_event(void); extern void sched_clock_idle_wakeup_event(u64 delta_ns); #ifdef CONFIG_HOTPLUG_CPU -extern void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p); extern void idle_task_exit(void); #else static inline void idle_task_exit(void) {} #endif -extern void sched_idle_next(void); - #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) extern void wake_up_idle_cpu(int cpu); #else diff --git a/kernel/cpu.c b/kernel/cpu.c index f6e726f18491..8615aa65d927 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -189,7 +189,6 @@ static inline void check_for_tasks(int cpu) } struct take_cpu_down_param { - struct task_struct *caller; unsigned long mod; void *hcpu; }; @@ -208,11 +207,6 @@ static int __ref take_cpu_down(void *_param) cpu_notify(CPU_DYING | param->mod, param->hcpu); - if (task_cpu(param->caller) == cpu) - move_task_off_dead_cpu(cpu, param->caller); - /* Force idle task to run as soon as we yield: it should - immediately notice cpu is offline and die quickly. */ - sched_idle_next(); return 0; } @@ -223,7 +217,6 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen) void *hcpu = (void *)(long)cpu; unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0; struct take_cpu_down_param tcd_param = { - .caller = current, .mod = mod, .hcpu = hcpu, }; @@ -253,9 +246,12 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen) } BUG_ON(cpu_online(cpu)); - /* Wait for it to sleep (leaving idle task). */ - while (!idle_cpu(cpu)) - yield(); + /* + * The migration_call() CPU_DYING callback will have removed all + * runnable tasks from the cpu, there's only the idle task left now + * that the migration thread is done doing the stop_machine thing. + */ + BUG_ON(!idle_cpu(cpu)); /* This actually kills the CPU. */ __cpu_die(cpu); diff --git a/kernel/sched.c b/kernel/sched.c index 41f18695b730..b0d5f1b24a39 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2366,18 +2366,15 @@ static int select_fallback_rq(int cpu, struct task_struct *p) return dest_cpu; /* No more Mr. Nice Guy. */ - if (unlikely(dest_cpu >= nr_cpu_ids)) { - dest_cpu = cpuset_cpus_allowed_fallback(p); - /* - * Don't tell them about moving exiting tasks or - * kernel threads (both mm NULL), since they never - * leave kernel. - */ - if (p->mm && printk_ratelimit()) { - printk(KERN_INFO "process %d (%s) no " - "longer affine to cpu%d\n", - task_pid_nr(p), p->comm, cpu); - } + dest_cpu = cpuset_cpus_allowed_fallback(p); + /* + * Don't tell them about moving exiting tasks or + * kernel threads (both mm NULL), since they never + * leave kernel. + */ + if (p->mm && printk_ratelimit()) { + printk(KERN_INFO "process %d (%s) no longer affine to cpu%d\n", + task_pid_nr(p), p->comm, cpu); } return dest_cpu; @@ -5712,29 +5709,20 @@ static int migration_cpu_stop(void *data) } #ifdef CONFIG_HOTPLUG_CPU + /* - * Figure out where task on dead CPU should go, use force if necessary. + * Ensures that the idle task is using init_mm right before its cpu goes + * offline. */ -void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p) +void idle_task_exit(void) { - struct rq *rq = cpu_rq(dead_cpu); - int needs_cpu, uninitialized_var(dest_cpu); - unsigned long flags; + struct mm_struct *mm = current->active_mm; - local_irq_save(flags); + BUG_ON(cpu_online(smp_processor_id())); - raw_spin_lock(&rq->lock); - needs_cpu = (task_cpu(p) == dead_cpu) && (p->state != TASK_WAKING); - if (needs_cpu) - dest_cpu = select_fallback_rq(dead_cpu, p); - raw_spin_unlock(&rq->lock); - /* - * It can only fail if we race with set_cpus_allowed(), - * in the racer should migrate the task anyway. - */ - if (needs_cpu) - __migrate_task(p, dead_cpu, dest_cpu); - local_irq_restore(flags); + if (mm != &init_mm) + switch_mm(mm, &init_mm, current); + mmdrop(mm); } /* @@ -5747,128 +5735,69 @@ void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p) static void migrate_nr_uninterruptible(struct rq *rq_src) { struct rq *rq_dest = cpu_rq(cpumask_any(cpu_active_mask)); - unsigned long flags; - local_irq_save(flags); - double_rq_lock(rq_src, rq_dest); rq_dest->nr_uninterruptible += rq_src->nr_uninterruptible; rq_src->nr_uninterruptible = 0; - double_rq_unlock(rq_src, rq_dest); - local_irq_restore(flags); -} - -/* Run through task list and migrate tasks from the dead cpu. */ -static void migrate_live_tasks(int src_cpu) -{ - struct task_struct *p, *t; - - read_lock(&tasklist_lock); - - do_each_thread(t, p) { - if (p == current) - continue; - - if (task_cpu(p) == src_cpu) - move_task_off_dead_cpu(src_cpu, p); - } while_each_thread(t, p); - - read_unlock(&tasklist_lock); } /* - * Schedules idle task to be the next runnable task on current CPU. - * It does so by boosting its priority to highest possible. - * Used by CPU offline code. + * remove the tasks which were accounted by rq from calc_load_tasks. */ -void sched_idle_next(void) +static void calc_global_load_remove(struct rq *rq) { - int this_cpu = smp_processor_id(); - struct rq *rq = cpu_rq(this_cpu); - struct task_struct *p = rq->idle; - unsigned long flags; - - /* cpu has to be offline */ - BUG_ON(cpu_online(this_cpu)); - - /* - * Strictly not necessary since rest of the CPUs are stopped by now - * and interrupts disabled on the current cpu. - */ - raw_spin_lock_irqsave(&rq->lock, flags); - - __setscheduler(rq, p, SCHED_FIFO, MAX_RT_PRIO-1); - - activate_task(rq, p, 0); - - raw_spin_unlock_irqrestore(&rq->lock, flags); + atomic_long_sub(rq->calc_load_active, &calc_load_tasks); + rq->calc_load_active = 0; } /* - * Ensures that the idle task is using init_mm right before its cpu goes - * offline. + * Migrate all tasks from the rq, sleeping tasks will be migrated by + * try_to_wake_up()->select_task_rq(). + * + * Called with rq->lock held even though we'er in stop_machine() and + * there's no concurrency possible, we hold the required locks anyway + * because of lock validation efforts. */ -void idle_task_exit(void) -{ - struct mm_struct *mm = current->active_mm; - - BUG_ON(cpu_online(smp_processor_id())); - - if (mm != &init_mm) - switch_mm(mm, &init_mm, current); - mmdrop(mm); -} - -/* called under rq->lock with disabled interrupts */ -static void migrate_dead(unsigned int dead_cpu, struct task_struct *p) +static void migrate_tasks(unsigned int dead_cpu) { struct rq *rq = cpu_rq(dead_cpu); - - /* Must be exiting, otherwise would be on tasklist. */ - BUG_ON(!p->exit_state); - - /* Cannot have done final schedule yet: would have vanished. */ - BUG_ON(p->state == TASK_DEAD); - - get_task_struct(p); + struct task_struct *next, *stop = rq->stop; + int dest_cpu; /* - * Drop lock around migration; if someone else moves it, - * that's OK. No task can be added to this CPU, so iteration is - * fine. + * Fudge the rq selection such that the below task selection loop + * doesn't get stuck on the currently eligible stop task. + * + * We're currently inside stop_machine() and the rq is either stuck + * in the stop_machine_cpu_stop() loop, or we're executing this code, + * either way we should never end up calling schedule() until we're + * done here. */ - raw_spin_unlock_irq(&rq->lock); - move_task_off_dead_cpu(dead_cpu, p); - raw_spin_lock_irq(&rq->lock); - - put_task_struct(p); -} - -/* release_task() removes task from tasklist, so we won't find dead tasks. */ -static void migrate_dead_tasks(unsigned int dead_cpu) -{ - struct rq *rq = cpu_rq(dead_cpu); - struct task_struct *next; + rq->stop = NULL; for ( ; ; ) { - if (!rq->nr_running) + /* + * There's this thread running, bail when that's the only + * remaining thread. + */ + if (rq->nr_running == 1) break; + next = pick_next_task(rq); - if (!next) - break; + BUG_ON(!next); next->sched_class->put_prev_task(rq, next); - migrate_dead(dead_cpu, next); + /* Find suitable destination for @next, with force if needed. */ + dest_cpu = select_fallback_rq(dead_cpu, next); + raw_spin_unlock(&rq->lock); + + __migrate_task(next, dead_cpu, dest_cpu); + + raw_spin_lock(&rq->lock); } -} -/* - * remove the tasks which were accounted by rq from calc_load_tasks. - */ -static void calc_global_load_remove(struct rq *rq) -{ - atomic_long_sub(rq->calc_load_active, &calc_load_tasks); - rq->calc_load_active = 0; + rq->stop = stop; } + #endif /* CONFIG_HOTPLUG_CPU */ #if defined(CONFIG_SCHED_DEBUG) && defined(CONFIG_SYSCTL) @@ -6078,15 +6007,13 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu) unsigned long flags; struct rq *rq = cpu_rq(cpu); - switch (action) { + switch (action & ~CPU_TASKS_FROZEN) { case CPU_UP_PREPARE: - case CPU_UP_PREPARE_FROZEN: rq->calc_load_update = calc_load_update; break; case CPU_ONLINE: - case CPU_ONLINE_FROZEN: /* Update our root-domain */ raw_spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { @@ -6098,30 +6025,19 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu) break; #ifdef CONFIG_HOTPLUG_CPU - case CPU_DEAD: - case CPU_DEAD_FROZEN: - migrate_live_tasks(cpu); - /* Idle task back to normal (off runqueue, low prio) */ - raw_spin_lock_irq(&rq->lock); - deactivate_task(rq, rq->idle, 0); - __setscheduler(rq, rq->idle, SCHED_NORMAL, 0); - rq->idle->sched_class = &idle_sched_class; - migrate_dead_tasks(cpu); - raw_spin_unlock_irq(&rq->lock); - migrate_nr_uninterruptible(rq); - BUG_ON(rq->nr_running != 0); - calc_global_load_remove(rq); - break; - case CPU_DYING: - case CPU_DYING_FROZEN: /* Update our root-domain */ raw_spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { BUG_ON(!cpumask_test_cpu(cpu, rq->rd->span)); set_rq_offline(rq); } + migrate_tasks(cpu); + BUG_ON(rq->nr_running != 1); /* the migration thread */ raw_spin_unlock_irqrestore(&rq->lock, flags); + + migrate_nr_uninterruptible(rq); + calc_global_load_remove(rq); break; #endif } -- cgit v1.2.3 From 2069dd75c7d0f49355939e5586daf5a9ab216db7 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 15 Nov 2010 15:47:00 -0800 Subject: sched: Rewrite tg_shares_up) By tracking a per-cpu load-avg for each cfs_rq and folding it into a global task_group load on each tick we can rework tg_shares_up to be strictly per-cpu. This should improve cpu-cgroup performance for smp systems significantly. [ Paul: changed to use queueing cfs_rq + bug fixes ] Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234937.580480400@google.com> Signed-off-by: Ingo Molnar --- include/linux/sched.h | 2 - kernel/sched.c | 173 ++++++++++++------------------------------------ kernel/sched_debug.c | 15 +++-- kernel/sched_fair.c | 164 +++++++++++++++++++++++++++++---------------- kernel/sched_features.h | 2 - kernel/sysctl.c | 19 ------ 6 files changed, 162 insertions(+), 213 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 29d953abb5ad..8abb8aa59664 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1885,8 +1885,6 @@ static inline void wake_up_idle_cpu(int cpu) { } extern unsigned int sysctl_sched_latency; extern unsigned int sysctl_sched_min_granularity; extern unsigned int sysctl_sched_wakeup_granularity; -extern unsigned int sysctl_sched_shares_ratelimit; -extern unsigned int sysctl_sched_shares_thresh; extern unsigned int sysctl_sched_child_runs_first; enum sched_tunable_scaling { diff --git a/kernel/sched.c b/kernel/sched.c index b0d5f1b24a39..e2f1a3024a99 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -253,6 +253,8 @@ struct task_group { /* runqueue "owned" by this group on each cpu */ struct cfs_rq **cfs_rq; unsigned long shares; + + atomic_t load_weight; #endif #ifdef CONFIG_RT_GROUP_SCHED @@ -359,15 +361,11 @@ struct cfs_rq { */ unsigned long h_load; - /* - * this cpu's part of tg->shares - */ - unsigned long shares; + u64 load_avg; + u64 load_period; + u64 load_stamp; - /* - * load.weight at the time we set shares - */ - unsigned long rq_weight; + unsigned long load_contribution; #endif #endif }; @@ -806,20 +804,6 @@ late_initcall(sched_init_debug); */ const_debug unsigned int sysctl_sched_nr_migrate = 32; -/* - * ratelimit for updating the group shares. - * default: 0.25ms - */ -unsigned int sysctl_sched_shares_ratelimit = 250000; -unsigned int normalized_sysctl_sched_shares_ratelimit = 250000; - -/* - * Inject some fuzzyness into changing the per-cpu group shares - * this avoids remote rq-locks at the expense of fairness. - * default: 4 - */ -unsigned int sysctl_sched_shares_thresh = 4; - /* * period over which we average the RT time consumption, measured * in ms. @@ -1369,6 +1353,12 @@ static inline void update_load_sub(struct load_weight *lw, unsigned long dec) lw->inv_weight = 0; } +static inline void update_load_set(struct load_weight *lw, unsigned long w) +{ + lw->weight = w; + lw->inv_weight = 0; +} + /* * To aid in avoiding the subversion of "niceness" due to uneven distribution * of tasks with abnormal "nice" values across CPUs the contribution that @@ -1557,97 +1547,44 @@ static unsigned long cpu_avg_load_per_task(int cpu) #ifdef CONFIG_FAIR_GROUP_SCHED -static __read_mostly unsigned long __percpu *update_shares_data; - -static void __set_se_shares(struct sched_entity *se, unsigned long shares); - -/* - * Calculate and set the cpu's group shares. - */ -static void update_group_shares_cpu(struct task_group *tg, int cpu, - unsigned long sd_shares, - unsigned long sd_rq_weight, - unsigned long *usd_rq_weight) -{ - unsigned long shares, rq_weight; - int boost = 0; - - rq_weight = usd_rq_weight[cpu]; - if (!rq_weight) { - boost = 1; - rq_weight = NICE_0_LOAD; - } - - /* - * \Sum_j shares_j * rq_weight_i - * shares_i = ----------------------------- - * \Sum_j rq_weight_j - */ - shares = (sd_shares * rq_weight) / sd_rq_weight; - shares = clamp_t(unsigned long, shares, MIN_SHARES, MAX_SHARES); - - if (abs(shares - tg->se[cpu]->load.weight) > - sysctl_sched_shares_thresh) { - struct rq *rq = cpu_rq(cpu); - unsigned long flags; - - raw_spin_lock_irqsave(&rq->lock, flags); - tg->cfs_rq[cpu]->rq_weight = boost ? 0 : rq_weight; - tg->cfs_rq[cpu]->shares = boost ? 0 : shares; - __set_se_shares(tg->se[cpu], shares); - raw_spin_unlock_irqrestore(&rq->lock, flags); - } -} +static void update_cfs_load(struct cfs_rq *cfs_rq); +static void update_cfs_shares(struct cfs_rq *cfs_rq); /* - * Re-compute the task group their per cpu shares over the given domain. - * This needs to be done in a bottom-up fashion because the rq weight of a - * parent group depends on the shares of its child groups. + * update tg->load_weight by folding this cpu's load_avg */ static int tg_shares_up(struct task_group *tg, void *data) { - unsigned long weight, rq_weight = 0, sum_weight = 0, shares = 0; - unsigned long *usd_rq_weight; - struct sched_domain *sd = data; + long load_avg; + struct cfs_rq *cfs_rq; unsigned long flags; - int i; + int cpu = (long)data; + struct rq *rq; - if (!tg->se[0]) + if (!tg->se[cpu]) return 0; - local_irq_save(flags); - usd_rq_weight = per_cpu_ptr(update_shares_data, smp_processor_id()); - - for_each_cpu(i, sched_domain_span(sd)) { - weight = tg->cfs_rq[i]->load.weight; - usd_rq_weight[i] = weight; - - rq_weight += weight; - /* - * If there are currently no tasks on the cpu pretend there - * is one of average load so that when a new task gets to - * run here it will not get delayed by group starvation. - */ - if (!weight) - weight = NICE_0_LOAD; + rq = cpu_rq(cpu); + cfs_rq = tg->cfs_rq[cpu]; - sum_weight += weight; - shares += tg->cfs_rq[i]->shares; - } + raw_spin_lock_irqsave(&rq->lock, flags); - if (!rq_weight) - rq_weight = sum_weight; + update_rq_clock(rq); + update_cfs_load(cfs_rq); - if ((!shares && rq_weight) || shares > tg->shares) - shares = tg->shares; + load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); + load_avg -= cfs_rq->load_contribution; - if (!sd->parent || !(sd->parent->flags & SD_LOAD_BALANCE)) - shares = tg->shares; + atomic_add(load_avg, &tg->load_weight); + cfs_rq->load_contribution += load_avg; - for_each_cpu(i, sched_domain_span(sd)) - update_group_shares_cpu(tg, i, shares, rq_weight, usd_rq_weight); + /* + * We need to update shares after updating tg->load_weight in + * order to adjust the weight of groups with long running tasks. + */ + update_cfs_shares(cfs_rq); - local_irq_restore(flags); + raw_spin_unlock_irqrestore(&rq->lock, flags); return 0; } @@ -1666,7 +1603,7 @@ static int tg_load_down(struct task_group *tg, void *data) load = cpu_rq(cpu)->load.weight; } else { load = tg->parent->cfs_rq[cpu]->h_load; - load *= tg->cfs_rq[cpu]->shares; + load *= tg->se[cpu]->load.weight; load /= tg->parent->cfs_rq[cpu]->load.weight + 1; } @@ -1675,21 +1612,16 @@ static int tg_load_down(struct task_group *tg, void *data) return 0; } -static void update_shares(struct sched_domain *sd) +static void update_shares(long cpu) { - s64 elapsed; - u64 now; - if (root_task_group_empty()) return; - now = local_clock(); - elapsed = now - sd->last_update; + /* + * XXX: replace with an on-demand list + */ - if (elapsed >= (s64)(u64)sysctl_sched_shares_ratelimit) { - sd->last_update = now; - walk_tg_tree(tg_nop, tg_shares_up, sd); - } + walk_tg_tree(tg_nop, tg_shares_up, (void *)cpu); } static void update_h_load(long cpu) @@ -1699,7 +1631,7 @@ static void update_h_load(long cpu) #else -static inline void update_shares(struct sched_domain *sd) +static inline void update_shares(int cpu) { } @@ -1824,15 +1756,6 @@ static void double_rq_unlock(struct rq *rq1, struct rq *rq2) #endif -#ifdef CONFIG_FAIR_GROUP_SCHED -static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares) -{ -#ifdef CONFIG_SMP - cfs_rq->shares = shares; -#endif -} -#endif - static void calc_load_account_idle(struct rq *this_rq); static void update_sysctl(void); static int get_update_sysctl_factor(void); @@ -5551,7 +5474,6 @@ static void update_sysctl(void) SET_SYSCTL(sched_min_granularity); SET_SYSCTL(sched_latency); SET_SYSCTL(sched_wakeup_granularity); - SET_SYSCTL(sched_shares_ratelimit); #undef SET_SYSCTL } @@ -7787,8 +7709,7 @@ static void init_tg_cfs_entry(struct task_group *tg, struct cfs_rq *cfs_rq, se->cfs_rq = parent->my_q; se->my_q = cfs_rq; - se->load.weight = tg->shares; - se->load.inv_weight = 0; + update_load_set(&se->load, tg->shares); se->parent = parent; } #endif @@ -7881,10 +7802,6 @@ void __init sched_init(void) #endif /* CONFIG_CGROUP_SCHED */ -#if defined CONFIG_FAIR_GROUP_SCHED && defined CONFIG_SMP - update_shares_data = __alloc_percpu(nr_cpu_ids * sizeof(unsigned long), - __alignof__(unsigned long)); -#endif for_each_possible_cpu(i) { struct rq *rq; @@ -8452,8 +8369,7 @@ static void __set_se_shares(struct sched_entity *se, unsigned long shares) if (on_rq) dequeue_entity(cfs_rq, se, 0); - se->load.weight = shares; - se->load.inv_weight = 0; + update_load_set(&se->load, shares); if (on_rq) enqueue_entity(cfs_rq, se, 0); @@ -8510,7 +8426,6 @@ int sched_group_set_shares(struct task_group *tg, unsigned long shares) /* * force a rebalance */ - cfs_rq_set_shares(tg->cfs_rq[i], 0); set_se_shares(tg->se[i], shares); } diff --git a/kernel/sched_debug.c b/kernel/sched_debug.c index 2e1b0d17dd9b..e6590e7312e8 100644 --- a/kernel/sched_debug.c +++ b/kernel/sched_debug.c @@ -202,15 +202,22 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) spread0 = min_vruntime - rq0_min_vruntime; SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "spread0", SPLIT_NS(spread0)); - SEQ_printf(m, " .%-30s: %ld\n", "nr_running", cfs_rq->nr_running); - SEQ_printf(m, " .%-30s: %ld\n", "load", cfs_rq->load.weight); - SEQ_printf(m, " .%-30s: %d\n", "nr_spread_over", cfs_rq->nr_spread_over); + SEQ_printf(m, " .%-30s: %ld\n", "nr_running", cfs_rq->nr_running); + SEQ_printf(m, " .%-30s: %ld\n", "load", cfs_rq->load.weight); #ifdef CONFIG_FAIR_GROUP_SCHED #ifdef CONFIG_SMP - SEQ_printf(m, " .%-30s: %lu\n", "shares", cfs_rq->shares); + SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "load_avg", + SPLIT_NS(cfs_rq->load_avg)); + SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "load_period", + SPLIT_NS(cfs_rq->load_period)); + SEQ_printf(m, " .%-30s: %ld\n", "load_contrib", + cfs_rq->load_contribution); + SEQ_printf(m, " .%-30s: %d\n", "load_tg", + atomic_read(&tg->load_weight)); #endif + print_cfs_group_stats(m, cpu, cfs_rq->tg); #endif } diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index f4f6a8326dd0..d86544b4151c 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -417,7 +417,6 @@ int sched_proc_update_handler(struct ctl_table *table, int write, WRT_SYSCTL(sched_min_granularity); WRT_SYSCTL(sched_latency); WRT_SYSCTL(sched_wakeup_granularity); - WRT_SYSCTL(sched_shares_ratelimit); #undef WRT_SYSCTL return 0; @@ -633,7 +632,6 @@ account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se) list_add(&se->group_node, &cfs_rq->tasks); } cfs_rq->nr_running++; - se->on_rq = 1; } static void @@ -647,9 +645,89 @@ account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) list_del_init(&se->group_node); } cfs_rq->nr_running--; - se->on_rq = 0; } +#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED +static void update_cfs_load(struct cfs_rq *cfs_rq) +{ + u64 period = sched_avg_period(); + u64 now, delta; + + if (!cfs_rq) + return; + + now = rq_of(cfs_rq)->clock; + delta = now - cfs_rq->load_stamp; + + cfs_rq->load_stamp = now; + cfs_rq->load_period += delta; + cfs_rq->load_avg += delta * cfs_rq->load.weight; + + while (cfs_rq->load_period > period) { + /* + * Inline assembly required to prevent the compiler + * optimising this loop into a divmod call. + * See __iter_div_u64_rem() for another example of this. + */ + asm("" : "+rm" (cfs_rq->load_period)); + cfs_rq->load_period /= 2; + cfs_rq->load_avg /= 2; + } +} + +static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, + unsigned long weight) +{ + if (se->on_rq) + account_entity_dequeue(cfs_rq, se); + + update_load_set(&se->load, weight); + + if (se->on_rq) + account_entity_enqueue(cfs_rq, se); +} + +static void update_cfs_shares(struct cfs_rq *cfs_rq) +{ + struct task_group *tg; + struct sched_entity *se; + long load_weight, load, shares; + + if (!cfs_rq) + return; + + tg = cfs_rq->tg; + se = tg->se[cpu_of(rq_of(cfs_rq))]; + if (!se) + return; + + load = cfs_rq->load.weight; + + load_weight = atomic_read(&tg->load_weight); + load_weight -= cfs_rq->load_contribution; + load_weight += load; + + shares = (tg->shares * load); + if (load_weight) + shares /= load_weight; + + if (shares < MIN_SHARES) + shares = MIN_SHARES; + if (shares > tg->shares) + shares = tg->shares; + + reweight_entity(cfs_rq_of(se), se, shares); +} +#else /* CONFIG_FAIR_GROUP_SCHED */ +static inline void update_cfs_load(struct cfs_rq *cfs_rq) +{ +} + +static inline void update_cfs_shares(struct cfs_rq *cfs_rq) +{ +} +#endif /* CONFIG_FAIR_GROUP_SCHED */ + static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se) { #ifdef CONFIG_SCHEDSTATS @@ -771,7 +849,9 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) * Update run-time statistics of the 'current'. */ update_curr(cfs_rq); + update_cfs_load(cfs_rq); account_entity_enqueue(cfs_rq, se); + update_cfs_shares(cfs_rq); if (flags & ENQUEUE_WAKEUP) { place_entity(cfs_rq, se, 0); @@ -782,6 +862,7 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) check_spread(cfs_rq, se); if (se != cfs_rq->curr) __enqueue_entity(cfs_rq, se); + se->on_rq = 1; } static void __clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se) @@ -825,8 +906,11 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) if (se != cfs_rq->curr) __dequeue_entity(cfs_rq, se); + se->on_rq = 0; + update_cfs_load(cfs_rq); account_entity_dequeue(cfs_rq, se); update_min_vruntime(cfs_rq); + update_cfs_shares(cfs_rq); /* * Normalize the entity after updating the min_vruntime because the @@ -1055,6 +1139,13 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags) flags = ENQUEUE_WAKEUP; } + for_each_sched_entity(se) { + struct cfs_rq *cfs_rq = cfs_rq_of(se); + + update_cfs_load(cfs_rq); + update_cfs_shares(cfs_rq); + } + hrtick_update(rq); } @@ -1071,12 +1162,20 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags) for_each_sched_entity(se) { cfs_rq = cfs_rq_of(se); dequeue_entity(cfs_rq, se, flags); + /* Don't dequeue parent if it has other entities besides us */ if (cfs_rq->load.weight) break; flags |= DEQUEUE_SLEEP; } + for_each_sched_entity(se) { + struct cfs_rq *cfs_rq = cfs_rq_of(se); + + update_cfs_load(cfs_rq); + update_cfs_shares(cfs_rq); + } + hrtick_update(rq); } @@ -1143,51 +1242,20 @@ static void task_waking_fair(struct rq *rq, struct task_struct *p) * Adding load to a group doesn't make a group heavier, but can cause movement * of group shares between cpus. Assuming the shares were perfectly aligned one * can calculate the shift in shares. - * - * The problem is that perfectly aligning the shares is rather expensive, hence - * we try to avoid doing that too often - see update_shares(), which ratelimits - * this change. - * - * We compensate this by not only taking the current delta into account, but - * also considering the delta between when the shares were last adjusted and - * now. - * - * We still saw a performance dip, some tracing learned us that between - * cgroup:/ and cgroup:/foo balancing the number of affine wakeups increased - * significantly. Therefore try to bias the error in direction of failing - * the affine wakeup. - * */ -static long effective_load(struct task_group *tg, int cpu, - long wl, long wg) +static long effective_load(struct task_group *tg, int cpu, long wl, long wg) { struct sched_entity *se = tg->se[cpu]; if (!tg->parent) return wl; - /* - * By not taking the decrease of shares on the other cpu into - * account our error leans towards reducing the affine wakeups. - */ - if (!wl && sched_feat(ASYM_EFF_LOAD)) - return wl; - for_each_sched_entity(se) { long S, rw, s, a, b; - long more_w; - - /* - * Instead of using this increment, also add the difference - * between when the shares were last updated and now. - */ - more_w = se->my_q->load.weight - se->my_q->rq_weight; - wl += more_w; - wg += more_w; S = se->my_q->tg->shares; - s = se->my_q->shares; - rw = se->my_q->rq_weight; + s = se->load.weight; + rw = se->my_q->load.weight; a = S*(rw + wl); b = S*rw + s*wg; @@ -1508,23 +1576,6 @@ select_task_rq_fair(struct rq *rq, struct task_struct *p, int sd_flag, int wake_ sd = tmp; } -#ifdef CONFIG_FAIR_GROUP_SCHED - if (sched_feat(LB_SHARES_UPDATE)) { - /* - * Pick the largest domain to update shares over - */ - tmp = sd; - if (affine_sd && (!tmp || affine_sd->span_weight > sd->span_weight)) - tmp = affine_sd; - - if (tmp) { - raw_spin_unlock(&rq->lock); - update_shares(tmp); - raw_spin_lock(&rq->lock); - } - } -#endif - if (affine_sd) { if (cpu == prev_cpu || wake_affine(affine_sd, p, sync)) return select_idle_sibling(p, cpu); @@ -3014,7 +3065,6 @@ static int load_balance(int this_cpu, struct rq *this_rq, schedstat_inc(sd, lb_count[idle]); redo: - update_shares(sd); group = find_busiest_group(sd, this_cpu, &imbalance, idle, &sd_idle, cpus, balance); @@ -3156,8 +3206,6 @@ out_one_pinned: else ld_moved = 0; out: - if (ld_moved) - update_shares(sd); return ld_moved; } @@ -3549,6 +3597,8 @@ static void rebalance_domains(int cpu, enum cpu_idle_type idle) int update_next_balance = 0; int need_serialize; + update_shares(cpu); + for_each_domain(cpu, sd) { if (!(sd->flags & SD_LOAD_BALANCE)) continue; diff --git a/kernel/sched_features.h b/kernel/sched_features.h index 185f920ec1a2..68e69acc29b9 100644 --- a/kernel/sched_features.h +++ b/kernel/sched_features.h @@ -52,8 +52,6 @@ SCHED_FEAT(ARCH_POWER, 0) SCHED_FEAT(HRTICK, 0) SCHED_FEAT(DOUBLE_TICK, 0) SCHED_FEAT(LB_BIAS, 1) -SCHED_FEAT(LB_SHARES_UPDATE, 1) -SCHED_FEAT(ASYM_EFF_LOAD, 1) /* * Spin-wait on mutex acquisition when the mutex owner is running on diff --git a/kernel/sysctl.c b/kernel/sysctl.c index b65bf634035e..3132b25193db 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -259,8 +259,6 @@ static int min_wakeup_granularity_ns; /* 0 usecs */ static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; -static int min_sched_shares_ratelimit = 100000; /* 100 usec */ -static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */ #endif #ifdef CONFIG_COMPACTION @@ -304,15 +302,6 @@ static struct ctl_table kern_table[] = { .extra1 = &min_wakeup_granularity_ns, .extra2 = &max_wakeup_granularity_ns, }, - { - .procname = "sched_shares_ratelimit", - .data = &sysctl_sched_shares_ratelimit, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = sched_proc_update_handler, - .extra1 = &min_sched_shares_ratelimit, - .extra2 = &max_sched_shares_ratelimit, - }, { .procname = "sched_tunable_scaling", .data = &sysctl_sched_tunable_scaling, @@ -322,14 +311,6 @@ static struct ctl_table kern_table[] = { .extra1 = &min_sched_tunable_scaling, .extra2 = &max_sched_tunable_scaling, }, - { - .procname = "sched_shares_thresh", - .data = &sysctl_sched_shares_thresh, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = proc_dointvec_minmax, - .extra1 = &zero, - }, { .procname = "sched_migration_cost", .data = &sysctl_sched_migration_cost, -- cgit v1.2.3 From 3d4b47b4b040c9d77dd68104cfc1055d89a55afd Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 15 Nov 2010 15:47:01 -0800 Subject: sched: Implement on-demand (active) cfs_rq list Make certain load-balance actions scale per number of active cgroups instead of the number of existing cgroups. This makes wakeup/sleep paths more expensive, but is a win for systems where the vast majority of existing cgroups are idle. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234937.666535048@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 105 ++++++++++++++-------------------------------------- kernel/sched_fair.c | 46 ++++++++++++++++++++--- kernel/sched_rt.c | 24 ++++++++++++ 3 files changed, 92 insertions(+), 83 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index e2f1a3024a99..22436dd2e19f 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -274,9 +274,7 @@ struct task_group { #define root_task_group init_task_group -/* task_group_lock serializes add/remove of task groups and also changes to - * a task group's cpu shares. - */ +/* task_group_lock serializes the addition/removal of task groups */ static DEFINE_SPINLOCK(task_group_lock); #ifdef CONFIG_FAIR_GROUP_SCHED @@ -344,6 +342,7 @@ struct cfs_rq { * leaf_cfs_rq_list ties together list of leaf cfs_rq's in a cpu. This * list is used during load balance. */ + int on_list; struct list_head leaf_cfs_rq_list; struct task_group *tg; /* group that "owns" this runqueue */ @@ -1547,7 +1546,7 @@ static unsigned long cpu_avg_load_per_task(int cpu) #ifdef CONFIG_FAIR_GROUP_SCHED -static void update_cfs_load(struct cfs_rq *cfs_rq); +static void update_cfs_load(struct cfs_rq *cfs_rq, int lb); static void update_cfs_shares(struct cfs_rq *cfs_rq); /* @@ -1570,7 +1569,7 @@ static int tg_shares_up(struct task_group *tg, void *data) raw_spin_lock_irqsave(&rq->lock, flags); update_rq_clock(rq); - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 1); load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); load_avg -= cfs_rq->load_contribution; @@ -7688,15 +7687,13 @@ static void init_rt_rq(struct rt_rq *rt_rq, struct rq *rq) #ifdef CONFIG_FAIR_GROUP_SCHED static void init_tg_cfs_entry(struct task_group *tg, struct cfs_rq *cfs_rq, - struct sched_entity *se, int cpu, int add, + struct sched_entity *se, int cpu, struct sched_entity *parent) { struct rq *rq = cpu_rq(cpu); tg->cfs_rq[cpu] = cfs_rq; init_cfs_rq(cfs_rq, rq); cfs_rq->tg = tg; - if (add) - list_add(&cfs_rq->leaf_cfs_rq_list, &rq->leaf_cfs_rq_list); tg->se[cpu] = se; /* se could be NULL for init_task_group */ @@ -7716,7 +7713,7 @@ static void init_tg_cfs_entry(struct task_group *tg, struct cfs_rq *cfs_rq, #ifdef CONFIG_RT_GROUP_SCHED static void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq, - struct sched_rt_entity *rt_se, int cpu, int add, + struct sched_rt_entity *rt_se, int cpu, struct sched_rt_entity *parent) { struct rq *rq = cpu_rq(cpu); @@ -7725,8 +7722,6 @@ static void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq, init_rt_rq(rt_rq, rq); rt_rq->tg = tg; rt_rq->rt_runtime = tg->rt_bandwidth.rt_runtime; - if (add) - list_add(&rt_rq->leaf_rt_rq_list, &rq->leaf_rt_rq_list); tg->rt_se[cpu] = rt_se; if (!rt_se) @@ -7835,7 +7830,7 @@ void __init sched_init(void) * We achieve this by letting init_task_group's tasks sit * directly in rq->cfs (i.e init_task_group->se[] = NULL). */ - init_tg_cfs_entry(&init_task_group, &rq->cfs, NULL, i, 1, NULL); + init_tg_cfs_entry(&init_task_group, &rq->cfs, NULL, i, NULL); #endif #endif /* CONFIG_FAIR_GROUP_SCHED */ @@ -7843,7 +7838,7 @@ void __init sched_init(void) #ifdef CONFIG_RT_GROUP_SCHED INIT_LIST_HEAD(&rq->leaf_rt_rq_list); #ifdef CONFIG_CGROUP_SCHED - init_tg_rt_entry(&init_task_group, &rq->rt, NULL, i, 1, NULL); + init_tg_rt_entry(&init_task_group, &rq->rt, NULL, i, NULL); #endif #endif @@ -8119,7 +8114,7 @@ int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent) if (!se) goto err_free_rq; - init_tg_cfs_entry(tg, cfs_rq, se, i, 0, parent->se[i]); + init_tg_cfs_entry(tg, cfs_rq, se, i, parent->se[i]); } return 1; @@ -8130,15 +8125,22 @@ err: return 0; } -static inline void register_fair_sched_group(struct task_group *tg, int cpu) -{ - list_add_rcu(&tg->cfs_rq[cpu]->leaf_cfs_rq_list, - &cpu_rq(cpu)->leaf_cfs_rq_list); -} - static inline void unregister_fair_sched_group(struct task_group *tg, int cpu) { - list_del_rcu(&tg->cfs_rq[cpu]->leaf_cfs_rq_list); + struct rq *rq = cpu_rq(cpu); + unsigned long flags; + int i; + + /* + * Only empty task groups can be destroyed; so we can speculatively + * check on_list without danger of it being re-added. + */ + if (!tg->cfs_rq[cpu]->on_list) + return; + + raw_spin_lock_irqsave(&rq->lock, flags); + list_del_leaf_cfs_rq(tg->cfs_rq[i]); + raw_spin_unlock_irqrestore(&rq->lock, flags); } #else /* !CONFG_FAIR_GROUP_SCHED */ static inline void free_fair_sched_group(struct task_group *tg) @@ -8151,10 +8153,6 @@ int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent) return 1; } -static inline void register_fair_sched_group(struct task_group *tg, int cpu) -{ -} - static inline void unregister_fair_sched_group(struct task_group *tg, int cpu) { } @@ -8209,7 +8207,7 @@ int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent) if (!rt_se) goto err_free_rq; - init_tg_rt_entry(tg, rt_rq, rt_se, i, 0, parent->rt_se[i]); + init_tg_rt_entry(tg, rt_rq, rt_se, i, parent->rt_se[i]); } return 1; @@ -8219,17 +8217,6 @@ err_free_rq: err: return 0; } - -static inline void register_rt_sched_group(struct task_group *tg, int cpu) -{ - list_add_rcu(&tg->rt_rq[cpu]->leaf_rt_rq_list, - &cpu_rq(cpu)->leaf_rt_rq_list); -} - -static inline void unregister_rt_sched_group(struct task_group *tg, int cpu) -{ - list_del_rcu(&tg->rt_rq[cpu]->leaf_rt_rq_list); -} #else /* !CONFIG_RT_GROUP_SCHED */ static inline void free_rt_sched_group(struct task_group *tg) { @@ -8240,14 +8227,6 @@ int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent) { return 1; } - -static inline void register_rt_sched_group(struct task_group *tg, int cpu) -{ -} - -static inline void unregister_rt_sched_group(struct task_group *tg, int cpu) -{ -} #endif /* CONFIG_RT_GROUP_SCHED */ #ifdef CONFIG_CGROUP_SCHED @@ -8263,7 +8242,6 @@ struct task_group *sched_create_group(struct task_group *parent) { struct task_group *tg; unsigned long flags; - int i; tg = kzalloc(sizeof(*tg), GFP_KERNEL); if (!tg) @@ -8276,10 +8254,6 @@ struct task_group *sched_create_group(struct task_group *parent) goto err; spin_lock_irqsave(&task_group_lock, flags); - for_each_possible_cpu(i) { - register_fair_sched_group(tg, i); - register_rt_sched_group(tg, i); - } list_add_rcu(&tg->list, &task_groups); WARN_ON(!parent); /* root should already exist */ @@ -8309,11 +8283,11 @@ void sched_destroy_group(struct task_group *tg) unsigned long flags; int i; - spin_lock_irqsave(&task_group_lock, flags); - for_each_possible_cpu(i) { + /* end participation in shares distribution */ + for_each_possible_cpu(i) unregister_fair_sched_group(tg, i); - unregister_rt_sched_group(tg, i); - } + + spin_lock_irqsave(&task_group_lock, flags); list_del_rcu(&tg->list); list_del_rcu(&tg->siblings); spin_unlock_irqrestore(&task_group_lock, flags); @@ -8391,7 +8365,6 @@ static DEFINE_MUTEX(shares_mutex); int sched_group_set_shares(struct task_group *tg, unsigned long shares) { int i; - unsigned long flags; /* * We can't change the weight of the root cgroup. @@ -8408,19 +8381,6 @@ int sched_group_set_shares(struct task_group *tg, unsigned long shares) if (tg->shares == shares) goto done; - spin_lock_irqsave(&task_group_lock, flags); - for_each_possible_cpu(i) - unregister_fair_sched_group(tg, i); - list_del_rcu(&tg->siblings); - spin_unlock_irqrestore(&task_group_lock, flags); - - /* wait for any ongoing reference to this group to finish */ - synchronize_sched(); - - /* - * Now we are free to modify the group's share on each cpu - * w/o tripping rebalance_share or load_balance_fair. - */ tg->shares = shares; for_each_possible_cpu(i) { /* @@ -8429,15 +8389,6 @@ int sched_group_set_shares(struct task_group *tg, unsigned long shares) set_se_shares(tg->se[i], shares); } - /* - * Enable load balance activity on this group, by inserting it back on - * each cpu's rq->leaf_cfs_rq_list. - */ - spin_lock_irqsave(&task_group_lock, flags); - for_each_possible_cpu(i) - register_fair_sched_group(tg, i); - list_add_rcu(&tg->siblings, &tg->parent->children); - spin_unlock_irqrestore(&task_group_lock, flags); done: mutex_unlock(&shares_mutex); return 0; diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index d86544b4151c..0560e72bd732 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -143,6 +143,24 @@ static inline struct cfs_rq *cpu_cfs_rq(struct cfs_rq *cfs_rq, int this_cpu) return cfs_rq->tg->cfs_rq[this_cpu]; } +static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq) +{ + if (!cfs_rq->on_list) { + list_add_rcu(&cfs_rq->leaf_cfs_rq_list, + &rq_of(cfs_rq)->leaf_cfs_rq_list); + + cfs_rq->on_list = 1; + } +} + +static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq) +{ + if (cfs_rq->on_list) { + list_del_rcu(&cfs_rq->leaf_cfs_rq_list); + cfs_rq->on_list = 0; + } +} + /* Iterate thr' all leaf cfs_rq's on a runqueue */ #define for_each_leaf_cfs_rq(rq, cfs_rq) \ list_for_each_entry_rcu(cfs_rq, &rq->leaf_cfs_rq_list, leaf_cfs_rq_list) @@ -246,6 +264,14 @@ static inline struct cfs_rq *cpu_cfs_rq(struct cfs_rq *cfs_rq, int this_cpu) return &cpu_rq(this_cpu)->cfs; } +static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq) +{ +} + +static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq) +{ +} + #define for_each_leaf_cfs_rq(rq, cfs_rq) \ for (cfs_rq = &rq->cfs; cfs_rq; cfs_rq = NULL) @@ -648,7 +674,7 @@ account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) } #if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED -static void update_cfs_load(struct cfs_rq *cfs_rq) +static void update_cfs_load(struct cfs_rq *cfs_rq, int lb) { u64 period = sched_avg_period(); u64 now, delta; @@ -673,6 +699,11 @@ static void update_cfs_load(struct cfs_rq *cfs_rq) cfs_rq->load_period /= 2; cfs_rq->load_avg /= 2; } + + if (lb && !cfs_rq->nr_running) { + if (cfs_rq->load_avg < (period / 8)) + list_del_leaf_cfs_rq(cfs_rq); + } } static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, @@ -719,7 +750,7 @@ static void update_cfs_shares(struct cfs_rq *cfs_rq) reweight_entity(cfs_rq_of(se), se, shares); } #else /* CONFIG_FAIR_GROUP_SCHED */ -static inline void update_cfs_load(struct cfs_rq *cfs_rq) +static inline void update_cfs_load(struct cfs_rq *cfs_rq, int lb) { } @@ -849,7 +880,7 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) * Update run-time statistics of the 'current'. */ update_curr(cfs_rq); - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); account_entity_enqueue(cfs_rq, se); update_cfs_shares(cfs_rq); @@ -863,6 +894,9 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) if (se != cfs_rq->curr) __enqueue_entity(cfs_rq, se); se->on_rq = 1; + + if (cfs_rq->nr_running == 1) + list_add_leaf_cfs_rq(cfs_rq); } static void __clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se) @@ -907,7 +941,7 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) if (se != cfs_rq->curr) __dequeue_entity(cfs_rq, se); se->on_rq = 0; - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); account_entity_dequeue(cfs_rq, se); update_min_vruntime(cfs_rq); update_cfs_shares(cfs_rq); @@ -1142,7 +1176,7 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags) for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); update_cfs_shares(cfs_rq); } @@ -1172,7 +1206,7 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags) for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); update_cfs_shares(cfs_rq); } diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index bea7d79f7e9c..c914ec747ca6 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -183,6 +183,17 @@ static inline u64 sched_rt_period(struct rt_rq *rt_rq) return ktime_to_ns(rt_rq->tg->rt_bandwidth.rt_period); } +static inline void list_add_leaf_rt_rq(struct rt_rq *rt_rq) +{ + list_add_rcu(&rt_rq->leaf_rt_rq_list, + &rq_of_rt_rq(rt_rq)->leaf_rt_rq_list); +} + +static inline void list_del_leaf_rt_rq(struct rt_rq *rt_rq) +{ + list_del_rcu(&rt_rq->leaf_rt_rq_list); +} + #define for_each_leaf_rt_rq(rt_rq, rq) \ list_for_each_entry_rcu(rt_rq, &rq->leaf_rt_rq_list, leaf_rt_rq_list) @@ -276,6 +287,14 @@ static inline u64 sched_rt_period(struct rt_rq *rt_rq) return ktime_to_ns(def_rt_bandwidth.rt_period); } +static inline void list_add_leaf_rt_rq(struct rt_rq *rt_rq) +{ +} + +static inline void list_del_leaf_rt_rq(struct rt_rq *rt_rq) +{ +} + #define for_each_leaf_rt_rq(rt_rq, rq) \ for (rt_rq = &rq->rt; rt_rq; rt_rq = NULL) @@ -825,6 +844,9 @@ static void __enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head) if (group_rq && (rt_rq_throttled(group_rq) || !group_rq->rt_nr_running)) return; + if (!rt_rq->rt_nr_running) + list_add_leaf_rt_rq(rt_rq); + if (head) list_add(&rt_se->run_list, queue); else @@ -844,6 +866,8 @@ static void __dequeue_rt_entity(struct sched_rt_entity *rt_se) __clear_bit(rt_se_prio(rt_se), array->bitmap); dec_rt_tasks(rt_se, rt_rq); + if (!rt_rq->rt_nr_running) + list_del_leaf_rt_rq(rt_rq); } /* -- cgit v1.2.3 From 9e3081ca61147b29f52fddb4f7c6b6b82ea5eb7a Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 15 Nov 2010 15:47:02 -0800 Subject: sched: Make tg_shares_up() walk on-demand Make tg_shares_up() use the active cgroup list, this means we cannot do a strict bottom-up walk of the hierarchy, but assuming its a very wide tree with a small number of active groups it should be a win. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234937.754159484@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 67 ----------------------------------------------------- kernel/sched_fair.c | 58 ++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 58 insertions(+), 67 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 22436dd2e19f..6268d2dc9a91 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -279,13 +279,6 @@ static DEFINE_SPINLOCK(task_group_lock); #ifdef CONFIG_FAIR_GROUP_SCHED -#ifdef CONFIG_SMP -static int root_task_group_empty(void) -{ - return list_empty(&root_task_group.children); -} -#endif - # define INIT_TASK_GROUP_LOAD NICE_0_LOAD /* @@ -1546,48 +1539,6 @@ static unsigned long cpu_avg_load_per_task(int cpu) #ifdef CONFIG_FAIR_GROUP_SCHED -static void update_cfs_load(struct cfs_rq *cfs_rq, int lb); -static void update_cfs_shares(struct cfs_rq *cfs_rq); - -/* - * update tg->load_weight by folding this cpu's load_avg - */ -static int tg_shares_up(struct task_group *tg, void *data) -{ - long load_avg; - struct cfs_rq *cfs_rq; - unsigned long flags; - int cpu = (long)data; - struct rq *rq; - - if (!tg->se[cpu]) - return 0; - - rq = cpu_rq(cpu); - cfs_rq = tg->cfs_rq[cpu]; - - raw_spin_lock_irqsave(&rq->lock, flags); - - update_rq_clock(rq); - update_cfs_load(cfs_rq, 1); - - load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); - load_avg -= cfs_rq->load_contribution; - - atomic_add(load_avg, &tg->load_weight); - cfs_rq->load_contribution += load_avg; - - /* - * We need to update shares after updating tg->load_weight in - * order to adjust the weight of groups with long running tasks. - */ - update_cfs_shares(cfs_rq); - - raw_spin_unlock_irqrestore(&rq->lock, flags); - - return 0; -} - /* * Compute the cpu's hierarchical load factor for each task group. * This needs to be done in a top-down fashion because the load of a child @@ -1611,29 +1562,11 @@ static int tg_load_down(struct task_group *tg, void *data) return 0; } -static void update_shares(long cpu) -{ - if (root_task_group_empty()) - return; - - /* - * XXX: replace with an on-demand list - */ - - walk_tg_tree(tg_nop, tg_shares_up, (void *)cpu); -} - static void update_h_load(long cpu) { walk_tg_tree(tg_load_down, tg_nop, (void *)cpu); } -#else - -static inline void update_shares(int cpu) -{ -} - #endif #ifdef CONFIG_PREEMPT diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 0560e72bd732..46ff6587dc16 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2004,6 +2004,60 @@ out: } #ifdef CONFIG_FAIR_GROUP_SCHED +/* + * update tg->load_weight by folding this cpu's load_avg + */ +static int tg_shares_up(struct task_group *tg, int cpu) +{ + struct cfs_rq *cfs_rq; + unsigned long flags; + struct rq *rq; + long load_avg; + + if (!tg->se[cpu]) + return 0; + + rq = cpu_rq(cpu); + cfs_rq = tg->cfs_rq[cpu]; + + raw_spin_lock_irqsave(&rq->lock, flags); + + update_rq_clock(rq); + update_cfs_load(cfs_rq, 1); + + load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); + load_avg -= cfs_rq->load_contribution; + atomic_add(load_avg, &tg->load_weight); + cfs_rq->load_contribution += load_avg; + + /* + * We need to update shares after updating tg->load_weight in + * order to adjust the weight of groups with long running tasks. + */ + update_cfs_shares(cfs_rq); + + raw_spin_unlock_irqrestore(&rq->lock, flags); + + return 0; +} + +static void update_shares(int cpu) +{ + struct cfs_rq *cfs_rq; + struct rq *rq = cpu_rq(cpu); + + rcu_read_lock(); + for_each_leaf_cfs_rq(rq, cfs_rq) { + struct task_group *tg = cfs_rq->tg; + + do { + tg_shares_up(tg, cpu); + tg = tg->parent; + } while (tg); + } + rcu_read_unlock(); +} + static unsigned long load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, @@ -2051,6 +2105,10 @@ load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, return max_load_move - rem_load_move; } #else +static inline void update_shares(int cpu) +{ +} + static unsigned long load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, -- cgit v1.2.3 From f0d7442a5924a802b66eef79b3708f77297bfb35 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:03 -0800 Subject: sched: Fix load corruption from update_cfs_shares() As part of enqueue_entity both a new entity weight and its contribution to the queuing cfs_rq / rq are updated. Since update_cfs_shares will only update the queueing weights when the entity is on_rq (which in this case it is not yet), there's a dependency loop here: update_cfs_shares needs account_entity_enqueue to update cfs_rq->load.weight account_entity_enqueue needs the updated weight for the queuing cfs_rq load[*] Fix this and avoid spurious dequeue/enqueues by issuing update_cfs_shares as if we had accounted the enqueue already. This was also resulting in rq->load corruption previously. [*]: this dependency also exists when using the group cfs_rq w/ update_cfs_shares as the weight of the enqueued entity changes without the load being updated. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234937.844900206@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 46ff6587dc16..d52b97a04e7a 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -718,7 +718,7 @@ static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, account_entity_enqueue(cfs_rq, se); } -static void update_cfs_shares(struct cfs_rq *cfs_rq) +static void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta) { struct task_group *tg; struct sched_entity *se; @@ -732,7 +732,7 @@ static void update_cfs_shares(struct cfs_rq *cfs_rq) if (!se) return; - load = cfs_rq->load.weight; + load = cfs_rq->load.weight + weight_delta; load_weight = atomic_read(&tg->load_weight); load_weight -= cfs_rq->load_contribution; @@ -754,7 +754,7 @@ static inline void update_cfs_load(struct cfs_rq *cfs_rq, int lb) { } -static inline void update_cfs_shares(struct cfs_rq *cfs_rq) +static inline void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta) { } #endif /* CONFIG_FAIR_GROUP_SCHED */ @@ -881,8 +881,8 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) */ update_curr(cfs_rq); update_cfs_load(cfs_rq, 0); + update_cfs_shares(cfs_rq, se->load.weight); account_entity_enqueue(cfs_rq, se); - update_cfs_shares(cfs_rq); if (flags & ENQUEUE_WAKEUP) { place_entity(cfs_rq, se, 0); @@ -944,7 +944,7 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) update_cfs_load(cfs_rq, 0); account_entity_dequeue(cfs_rq, se); update_min_vruntime(cfs_rq); - update_cfs_shares(cfs_rq); + update_cfs_shares(cfs_rq, 0); /* * Normalize the entity after updating the min_vruntime because the @@ -1177,7 +1177,7 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags) struct cfs_rq *cfs_rq = cfs_rq_of(se); update_cfs_load(cfs_rq, 0); - update_cfs_shares(cfs_rq); + update_cfs_shares(cfs_rq, 0); } hrtick_update(rq); @@ -1207,7 +1207,7 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags) struct cfs_rq *cfs_rq = cfs_rq_of(se); update_cfs_load(cfs_rq, 0); - update_cfs_shares(cfs_rq); + update_cfs_shares(cfs_rq, 0); } hrtick_update(rq); @@ -2034,7 +2034,7 @@ static int tg_shares_up(struct task_group *tg, int cpu) * We need to update shares after updating tg->load_weight in * order to adjust the weight of groups with long running tasks. */ - update_cfs_shares(cfs_rq); + update_cfs_shares(cfs_rq, 0); raw_spin_unlock_irqrestore(&rq->lock, flags); -- cgit v1.2.3 From e33078baa4d30ad1d0e46d1f62b9e5a63a3e6ee3 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:04 -0800 Subject: sched: Fix update_cfs_load() synchronization Using cfs_rq->nr_running is not sufficient to synchronize update_cfs_load with the put path since nr_running accounting occurs at deactivation. It's also not safe to make the removal decision based on load_avg as this fails with both high periods and low shares. Resolve this by clipping history after 4 periods without activity. Note: the above will always occur from update_shares() since in the last-task-sleep-case that task will still be cfs_rq->curr when update_cfs_load is called. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234937.933428187@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 2 +- kernel/sched_fair.c | 33 +++++++++++++++++++++------------ 2 files changed, 22 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 6268d2dc9a91..dadab4d13875 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -355,7 +355,7 @@ struct cfs_rq { u64 load_avg; u64 load_period; - u64 load_stamp; + u64 load_stamp, load_last; unsigned long load_contribution; #endif diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index d52b97a04e7a..a543a5b202a4 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -674,10 +674,11 @@ account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) } #if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED -static void update_cfs_load(struct cfs_rq *cfs_rq, int lb) +static void update_cfs_load(struct cfs_rq *cfs_rq) { u64 period = sched_avg_period(); u64 now, delta; + unsigned long load = cfs_rq->load.weight; if (!cfs_rq) return; @@ -685,9 +686,19 @@ static void update_cfs_load(struct cfs_rq *cfs_rq, int lb) now = rq_of(cfs_rq)->clock; delta = now - cfs_rq->load_stamp; + /* truncate load history at 4 idle periods */ + if (cfs_rq->load_stamp > cfs_rq->load_last && + now - cfs_rq->load_last > 4 * period) { + cfs_rq->load_period = 0; + cfs_rq->load_avg = 0; + } + cfs_rq->load_stamp = now; cfs_rq->load_period += delta; - cfs_rq->load_avg += delta * cfs_rq->load.weight; + if (load) { + cfs_rq->load_last = now; + cfs_rq->load_avg += delta * load; + } while (cfs_rq->load_period > period) { /* @@ -700,10 +711,8 @@ static void update_cfs_load(struct cfs_rq *cfs_rq, int lb) cfs_rq->load_avg /= 2; } - if (lb && !cfs_rq->nr_running) { - if (cfs_rq->load_avg < (period / 8)) - list_del_leaf_cfs_rq(cfs_rq); - } + if (!cfs_rq->curr && !cfs_rq->nr_running && !cfs_rq->load_avg) + list_del_leaf_cfs_rq(cfs_rq); } static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, @@ -750,7 +759,7 @@ static void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta) reweight_entity(cfs_rq_of(se), se, shares); } #else /* CONFIG_FAIR_GROUP_SCHED */ -static inline void update_cfs_load(struct cfs_rq *cfs_rq, int lb) +static inline void update_cfs_load(struct cfs_rq *cfs_rq) { } @@ -880,7 +889,7 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) * Update run-time statistics of the 'current'. */ update_curr(cfs_rq); - update_cfs_load(cfs_rq, 0); + update_cfs_load(cfs_rq); update_cfs_shares(cfs_rq, se->load.weight); account_entity_enqueue(cfs_rq, se); @@ -941,7 +950,7 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) if (se != cfs_rq->curr) __dequeue_entity(cfs_rq, se); se->on_rq = 0; - update_cfs_load(cfs_rq, 0); + update_cfs_load(cfs_rq); account_entity_dequeue(cfs_rq, se); update_min_vruntime(cfs_rq); update_cfs_shares(cfs_rq, 0); @@ -1176,7 +1185,7 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags) for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); - update_cfs_load(cfs_rq, 0); + update_cfs_load(cfs_rq); update_cfs_shares(cfs_rq, 0); } @@ -1206,7 +1215,7 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags) for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); - update_cfs_load(cfs_rq, 0); + update_cfs_load(cfs_rq); update_cfs_shares(cfs_rq, 0); } @@ -2023,7 +2032,7 @@ static int tg_shares_up(struct task_group *tg, int cpu) raw_spin_lock_irqsave(&rq->lock, flags); update_rq_clock(rq); - update_cfs_load(cfs_rq, 1); + update_cfs_load(cfs_rq); load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); load_avg -= cfs_rq->load_contribution; -- cgit v1.2.3 From 67e86250f8ea7b8f7da53ac25ea73c6bd71f5cd9 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:05 -0800 Subject: sched: Introduce hierarchal order on shares update list Avoid duplicate shares update calls by ensuring children always appear before parents in rq->leaf_cfs_rq_list. This allows us to do a single in-order traversal for update_shares(). Since we always enqueue in bottom-up order this reduces to 2 cases: 1) Our parent is already in the list, e.g. root \ b /\ c d* (root->b->c already enqueued) Since d's parent is enqueued we push it to the head of the list, implicitly ahead of b. 2) Our parent does not appear in the list (or we have no parent) In this case we enqueue to the tail of the list, if our parent is subsequently enqueued (bottom-up) it will appear to our right by the same rule. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234938.022488865@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 26 ++++++++++++++++---------- 1 file changed, 16 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index a543a5b202a4..b320753aa6c9 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -146,8 +146,20 @@ static inline struct cfs_rq *cpu_cfs_rq(struct cfs_rq *cfs_rq, int this_cpu) static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq) { if (!cfs_rq->on_list) { - list_add_rcu(&cfs_rq->leaf_cfs_rq_list, + /* + * Ensure we either appear before our parent (if already + * enqueued) or force our parent to appear after us when it is + * enqueued. The fact that we always enqueue bottom-up + * reduces this to two cases. + */ + if (cfs_rq->tg->parent && + cfs_rq->tg->parent->cfs_rq[cpu_of(rq_of(cfs_rq))]->on_list) { + list_add_rcu(&cfs_rq->leaf_cfs_rq_list, + &rq_of(cfs_rq)->leaf_cfs_rq_list); + } else { + list_add_tail_rcu(&cfs_rq->leaf_cfs_rq_list, &rq_of(cfs_rq)->leaf_cfs_rq_list); + } cfs_rq->on_list = 1; } @@ -2016,7 +2028,7 @@ out: /* * update tg->load_weight by folding this cpu's load_avg */ -static int tg_shares_up(struct task_group *tg, int cpu) +static int update_shares_cpu(struct task_group *tg, int cpu) { struct cfs_rq *cfs_rq; unsigned long flags; @@ -2056,14 +2068,8 @@ static void update_shares(int cpu) struct rq *rq = cpu_rq(cpu); rcu_read_lock(); - for_each_leaf_cfs_rq(rq, cfs_rq) { - struct task_group *tg = cfs_rq->tg; - - do { - tg_shares_up(tg, cpu); - tg = tg->parent; - } while (tg); - } + for_each_leaf_cfs_rq(rq, cfs_rq) + update_shares_cpu(cfs_rq->tg, cpu); rcu_read_unlock(); } -- cgit v1.2.3 From a7a4f8a752ec734b2eab904fc863d5dc873de338 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:06 -0800 Subject: sched: Add sysctl_sched_shares_window Introduce a new sysctl for the shares window and disambiguate it from sched_time_avg. A 10ms window appears to be a good compromise between accuracy and performance. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234938.112173964@google.com> Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 + kernel/sched_fair.c | 9 ++++++++- kernel/sysctl.c | 7 +++++++ 3 files changed, 16 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 8abb8aa59664..840f1277492f 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1900,6 +1900,7 @@ extern unsigned int sysctl_sched_migration_cost; extern unsigned int sysctl_sched_nr_migrate; extern unsigned int sysctl_sched_time_avg; extern unsigned int sysctl_timer_migration; +extern unsigned int sysctl_sched_shares_window; int sched_proc_update_handler(struct ctl_table *table, int write, void __user *buffer, size_t *length, diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index b320753aa6c9..6c84439ce987 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -89,6 +89,13 @@ unsigned int normalized_sysctl_sched_wakeup_granularity = 1000000UL; const_debug unsigned int sysctl_sched_migration_cost = 500000UL; +/* + * The exponential sliding window over which load is averaged for shares + * distribution. + * (default: 10msec) + */ +unsigned int __read_mostly sysctl_sched_shares_window = 10000000UL; + static const struct sched_class fair_sched_class; /************************************************************** @@ -688,7 +695,7 @@ account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) #if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED static void update_cfs_load(struct cfs_rq *cfs_rq) { - u64 period = sched_avg_period(); + u64 period = sysctl_sched_shares_window; u64 now, delta; unsigned long load = cfs_rq->load.weight; diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 3132b25193db..9b520d74f052 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -332,6 +332,13 @@ static struct ctl_table kern_table[] = { .mode = 0644, .proc_handler = proc_dointvec, }, + { + .procname = "sched_shares_window", + .data = &sysctl_sched_shares_window, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec, + }, { .procname = "timer_migration", .data = &sysctl_timer_migration, -- cgit v1.2.3 From c66eaf619c0c7937e9ded160ae83b5a7a6b19b56 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:07 -0800 Subject: sched: Update shares on idle_balance Since shares updates are no longer expensive and effectively local, update them at idle_balance(). This allows us to more quickly redistribute shares to another cpu when our load becomes idle. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234938.204191702@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 6c84439ce987..33f941dcf88c 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -3343,6 +3343,7 @@ static void idle_balance(int this_cpu, struct rq *this_rq) */ raw_spin_unlock(&this_rq->lock); + update_shares(this_cpu); for_each_domain(this_cpu, sd) { unsigned long interval; int balance = 1; -- cgit v1.2.3 From 3b3d190ec3683d568fd2ebaead5e1ec7f97b6e37 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:08 -0800 Subject: sched: Implement demand based update_cfs_load() When the system is busy, dilation of rq->next_balance makes lb->update_shares() insufficiently frequent for threads which don't sleep (no dequeue/enqueue updates). Adjust for this by making demand based updates based on the accumulation of execution time sufficient to wrap our averaging window. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234938.291159744@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 9 ++++++++- kernel/sched_fair.c | 12 ++++++++++++ 2 files changed, 20 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index dadab4d13875..e914a716e1d4 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -353,9 +353,16 @@ struct cfs_rq { */ unsigned long h_load; + /* + * Maintaining per-cpu shares distribution for group scheduling + * + * load_stamp is the last time we updated the load average + * load_last is the last time we updated the load average and saw load + * load_unacc_exec_time is currently unaccounted execution time + */ u64 load_avg; u64 load_period; - u64 load_stamp, load_last; + u64 load_stamp, load_last, load_unacc_exec_time; unsigned long load_contribution; #endif diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 33f941dcf88c..e7e2f08e6d01 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -539,6 +539,9 @@ static u64 sched_vslice(struct cfs_rq *cfs_rq, struct sched_entity *se) return calc_delta_fair(sched_slice(cfs_rq, se), se); } +static void update_cfs_load(struct cfs_rq *cfs_rq); +static void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta); + /* * Update the current task's runtime statistics. Skip current tasks that * are not in our scheduling class. @@ -558,6 +561,14 @@ __update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr, curr->vruntime += delta_exec_weighted; update_min_vruntime(cfs_rq); + +#ifdef CONFIG_FAIR_GROUP_SCHED + cfs_rq->load_unacc_exec_time += delta_exec; + if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) { + update_cfs_load(cfs_rq); + update_cfs_shares(cfs_rq, 0); + } +#endif } static void update_curr(struct cfs_rq *cfs_rq) @@ -713,6 +724,7 @@ static void update_cfs_load(struct cfs_rq *cfs_rq) } cfs_rq->load_stamp = now; + cfs_rq->load_unacc_exec_time = 0; cfs_rq->load_period += delta; if (load) { cfs_rq->load_last = now; -- cgit v1.2.3 From d6b5591829bd348a5fbe1c428d28dea00621cdba Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:09 -0800 Subject: sched: Allow update_cfs_load() to update global load Refactor the global load updates from update_shares_cpu() so that update_cfs_load() can update global load when it is more than ~10% out of sync. The new global_load parameter allows us to force an update, regardless of the error factor so that we can synchronize w/ update_shares(). Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234938.377473595@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 44 +++++++++++++++++++++++++++++--------------- 1 file changed, 29 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index e7e2f08e6d01..390ce30ff2d0 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -539,7 +539,7 @@ static u64 sched_vslice(struct cfs_rq *cfs_rq, struct sched_entity *se) return calc_delta_fair(sched_slice(cfs_rq, se), se); } -static void update_cfs_load(struct cfs_rq *cfs_rq); +static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update); static void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta); /* @@ -565,7 +565,7 @@ __update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr, #ifdef CONFIG_FAIR_GROUP_SCHED cfs_rq->load_unacc_exec_time += delta_exec; if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) { - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); update_cfs_shares(cfs_rq, 0); } #endif @@ -704,7 +704,22 @@ account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) } #if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED -static void update_cfs_load(struct cfs_rq *cfs_rq) +static void update_cfs_rq_load_contribution(struct cfs_rq *cfs_rq, + int global_update) +{ + struct task_group *tg = cfs_rq->tg; + long load_avg; + + load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); + load_avg -= cfs_rq->load_contribution; + + if (global_update || abs(load_avg) > cfs_rq->load_contribution / 8) { + atomic_add(load_avg, &tg->load_weight); + cfs_rq->load_contribution += load_avg; + } +} + +static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) { u64 period = sysctl_sched_shares_window; u64 now, delta; @@ -731,6 +746,11 @@ static void update_cfs_load(struct cfs_rq *cfs_rq) cfs_rq->load_avg += delta * load; } + /* consider updating load contribution on each fold or truncate */ + if (global_update || cfs_rq->load_period > period + || !cfs_rq->load_period) + update_cfs_rq_load_contribution(cfs_rq, global_update); + while (cfs_rq->load_period > period) { /* * Inline assembly required to prevent the compiler @@ -790,7 +810,7 @@ static void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta) reweight_entity(cfs_rq_of(se), se, shares); } #else /* CONFIG_FAIR_GROUP_SCHED */ -static inline void update_cfs_load(struct cfs_rq *cfs_rq) +static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) { } @@ -920,7 +940,7 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) * Update run-time statistics of the 'current'. */ update_curr(cfs_rq); - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); update_cfs_shares(cfs_rq, se->load.weight); account_entity_enqueue(cfs_rq, se); @@ -981,7 +1001,7 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) if (se != cfs_rq->curr) __dequeue_entity(cfs_rq, se); se->on_rq = 0; - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); account_entity_dequeue(cfs_rq, se); update_min_vruntime(cfs_rq); update_cfs_shares(cfs_rq, 0); @@ -1216,7 +1236,7 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags) for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); update_cfs_shares(cfs_rq, 0); } @@ -1246,7 +1266,7 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags) for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); - update_cfs_load(cfs_rq); + update_cfs_load(cfs_rq, 0); update_cfs_shares(cfs_rq, 0); } @@ -2052,7 +2072,6 @@ static int update_shares_cpu(struct task_group *tg, int cpu) struct cfs_rq *cfs_rq; unsigned long flags; struct rq *rq; - long load_avg; if (!tg->se[cpu]) return 0; @@ -2063,12 +2082,7 @@ static int update_shares_cpu(struct task_group *tg, int cpu) raw_spin_lock_irqsave(&rq->lock, flags); update_rq_clock(rq); - update_cfs_load(cfs_rq); - - load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); - load_avg -= cfs_rq->load_contribution; - atomic_add(load_avg, &tg->load_weight); - cfs_rq->load_contribution += load_avg; + update_cfs_load(cfs_rq, 1); /* * We need to update shares after updating tg->load_weight in -- cgit v1.2.3 From 9437178f623a19af5951808d880a8599f66ac150 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 15 Nov 2010 15:47:10 -0800 Subject: sched: Update tg->shares after cpu.shares write Formerly sched_group_set_shares would force a rebalance by overflowing domain share sums. Now that per-cpu averages are maintained we can set the true value by issuing an update_cfs_shares() following a tg->shares update. Also initialize tg se->load to 0 for consistency since we'll now set correct weights on enqueue. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101115234938.465521344@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 42 +++++++++++------------------------------- 1 file changed, 11 insertions(+), 31 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index e914a716e1d4..550cf3a02377 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -7646,7 +7646,7 @@ static void init_tg_cfs_entry(struct task_group *tg, struct cfs_rq *cfs_rq, se->cfs_rq = parent->my_q; se->my_q = cfs_rq; - update_load_set(&se->load, tg->shares); + update_load_set(&se->load, 0); se->parent = parent; } #endif @@ -8274,37 +8274,12 @@ void sched_move_task(struct task_struct *tsk) #endif /* CONFIG_CGROUP_SCHED */ #ifdef CONFIG_FAIR_GROUP_SCHED -static void __set_se_shares(struct sched_entity *se, unsigned long shares) -{ - struct cfs_rq *cfs_rq = se->cfs_rq; - int on_rq; - - on_rq = se->on_rq; - if (on_rq) - dequeue_entity(cfs_rq, se, 0); - - update_load_set(&se->load, shares); - - if (on_rq) - enqueue_entity(cfs_rq, se, 0); -} - -static void set_se_shares(struct sched_entity *se, unsigned long shares) -{ - struct cfs_rq *cfs_rq = se->cfs_rq; - struct rq *rq = cfs_rq->rq; - unsigned long flags; - - raw_spin_lock_irqsave(&rq->lock, flags); - __set_se_shares(se, shares); - raw_spin_unlock_irqrestore(&rq->lock, flags); -} - static DEFINE_MUTEX(shares_mutex); int sched_group_set_shares(struct task_group *tg, unsigned long shares) { int i; + unsigned long flags; /* * We can't change the weight of the root cgroup. @@ -8323,10 +8298,15 @@ int sched_group_set_shares(struct task_group *tg, unsigned long shares) tg->shares = shares; for_each_possible_cpu(i) { - /* - * force a rebalance - */ - set_se_shares(tg->se[i], shares); + struct rq *rq = cpu_rq(i); + struct sched_entity *se; + + se = tg->se[i]; + /* Propagate contribution to hierarchy */ + raw_spin_lock_irqsave(&rq->lock, flags); + for_each_sched_entity(se) + update_cfs_shares(group_cfs_rq(se), 0); + raw_spin_unlock_irqrestore(&rq->lock, flags); } done: -- cgit v1.2.3 From 84e1c6bb38eb318e456558b610396d9f1afaabf0 Mon Sep 17 00:00:00 2001 From: matthieu castet Date: Tue, 16 Nov 2010 22:35:16 +0100 Subject: x86: Add RO/NX protection for loadable kernel modules This patch is a logical extension of the protection provided by CONFIG_DEBUG_RODATA to LKMs. The protection is provided by splitting module_core and module_init into three logical parts each and setting appropriate page access permissions for each individual section: 1. Code: RO+X 2. RO data: RO+NX 3. RW data: RW+NX In order to achieve proper protection, layout_sections() have been modified to align each of the three parts mentioned above onto page boundary. Next, the corresponding page access permissions are set right before successful exit from load_module(). Further, free_module() and sys_init_module have been modified to set module_core and module_init as RW+NX right before calling module_free(). By default, the original section layout and access flags are preserved. When compiled with CONFIG_DEBUG_SET_MODULE_RONX=y, the patch will page-align each group of sections to ensure that each page contains only one type of content and will enforce RO/NX for each group of pages. -v1: Initial proof-of-concept patch. -v2: The patch have been re-written to reduce the number of #ifdefs and to make it architecture-agnostic. Code formatting has also been corrected. -v3: Opportunistic RO/NX protection is now unconditional. Section page-alignment is enabled when CONFIG_DEBUG_RODATA=y. -v4: Removed most macros and improved coding style. -v5: Changed page-alignment and RO/NX section size calculation -v6: Fixed comments. Restricted RO/NX enforcement to x86 only -v7: Introduced CONFIG_DEBUG_SET_MODULE_RONX, added calls to set_all_modules_text_rw() and set_all_modules_text_ro() in ftrace -v8: updated for compatibility with linux 2.6.33-rc5 -v9: coding style fixes -v10: more coding style fixes -v11: minor adjustments for -tip -v12: minor adjustments for v2.6.35-rc2-tip -v13: minor adjustments for v2.6.37-rc1-tip Signed-off-by: Siarhei Liakh Signed-off-by: Xuxian Jiang Acked-by: Arjan van de Ven Reviewed-by: James Morris Signed-off-by: H. Peter Anvin Cc: Andi Kleen Cc: Rusty Russell Cc: Stephen Rothwell Cc: Dave Jones Cc: Kees Cook Cc: Linus Torvalds LKML-Reference: <4CE2F914.9070106@free.fr> [ minor cleanliness edits, -v14: build failure fix ] Signed-off-by: Ingo Molnar --- arch/x86/Kconfig.debug | 11 +++ arch/x86/kernel/ftrace.c | 3 + include/linux/module.h | 11 ++- kernel/module.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++- 4 files changed, 193 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug index b59ee765414e..45143bbcfe5e 100644 --- a/arch/x86/Kconfig.debug +++ b/arch/x86/Kconfig.debug @@ -117,6 +117,17 @@ config DEBUG_RODATA_TEST feature as well as for the change_page_attr() infrastructure. If in doubt, say "N" +config DEBUG_SET_MODULE_RONX + bool "Set loadable kernel module data as NX and text as RO" + depends on MODULES + ---help--- + This option helps catch unintended modifications to loadable + kernel module's text and read-only data. It also prevents execution + of module data. Such protection may interfere with run-time code + patching and dynamic kernel tracing - and they might also protect + against certain classes of kernel exploits. + If in doubt, say "N". + config DEBUG_NX_TEST tristate "Testcase for the NX non-executable stack feature" depends on DEBUG_KERNEL && m diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 3afb33f14d2d..298448656b60 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -19,6 +19,7 @@ #include #include #include +#include #include @@ -49,6 +50,7 @@ static DEFINE_PER_CPU(int, save_modifying_code); int ftrace_arch_code_modify_prepare(void) { set_kernel_text_rw(); + set_all_modules_text_rw(); modifying_code = 1; return 0; } @@ -56,6 +58,7 @@ int ftrace_arch_code_modify_prepare(void) int ftrace_arch_code_modify_post_process(void) { modifying_code = 0; + set_all_modules_text_ro(); set_kernel_text_ro(); return 0; } diff --git a/include/linux/module.h b/include/linux/module.h index b29e7458b966..ddaa689d71bd 100644 --- a/include/linux/module.h +++ b/include/linux/module.h @@ -308,6 +308,9 @@ struct module /* The size of the executable code in each section. */ unsigned int init_text_size, core_text_size; + /* Size of RO sections of the module (text+rodata) */ + unsigned int init_ro_size, core_ro_size; + /* Arch-specific module values */ struct mod_arch_specific arch; @@ -672,7 +675,6 @@ static inline int module_get_iter_tracepoints(struct tracepoint_iter *iter) { return 0; } - #endif /* CONFIG_MODULES */ #ifdef CONFIG_SYSFS @@ -687,6 +689,13 @@ extern int module_sysfs_initialized; #define __MODULE_STRING(x) __stringify(x) +#ifdef CONFIG_DEBUG_SET_MODULE_RONX +extern void set_all_modules_text_rw(void); +extern void set_all_modules_text_ro(void); +#else +static inline void set_all_modules_text_rw(void) { } +static inline void set_all_modules_text_ro(void) { } +#endif #ifdef CONFIG_GENERIC_BUG void module_bug_finalize(const Elf_Ehdr *, const Elf_Shdr *, diff --git a/kernel/module.c b/kernel/module.c index 437a74a7524a..ba421e6b4ada 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -56,6 +56,7 @@ #include #include #include +#include #define CREATE_TRACE_POINTS #include @@ -70,6 +71,26 @@ #define ARCH_SHF_SMALL 0 #endif +/* + * Modules' sections will be aligned on page boundaries + * to ensure complete separation of code and data, but + * only when CONFIG_DEBUG_SET_MODULE_RONX=y + */ +#ifdef CONFIG_DEBUG_SET_MODULE_RONX +# define debug_align(X) ALIGN(X, PAGE_SIZE) +#else +# define debug_align(X) (X) +#endif + +/* + * Given BASE and SIZE this macro calculates the number of pages the + * memory regions occupies + */ +#define MOD_NUMBER_OF_PAGES(BASE, SIZE) (((SIZE) > 0) ? \ + (PFN_DOWN((unsigned long)(BASE) + (SIZE) - 1) - \ + PFN_DOWN((unsigned long)BASE) + 1) \ + : (0UL)) + /* If this is set, the section belongs in the init part of the module */ #define INIT_OFFSET_MASK (1UL << (BITS_PER_LONG-1)) @@ -1542,6 +1563,115 @@ static int __unlink_module(void *_mod) return 0; } +#ifdef CONFIG_DEBUG_SET_MODULE_RONX +/* + * LKM RO/NX protection: protect module's text/ro-data + * from modification and any data from execution. + */ +void set_page_attributes(void *start, void *end, int (*set)(unsigned long start, int num_pages)) +{ + unsigned long begin_pfn = PFN_DOWN((unsigned long)start); + unsigned long end_pfn = PFN_DOWN((unsigned long)end); + + if (end_pfn > begin_pfn) + set(begin_pfn << PAGE_SHIFT, end_pfn - begin_pfn); +} + +static void set_section_ro_nx(void *base, + unsigned long text_size, + unsigned long ro_size, + unsigned long total_size) +{ + /* begin and end PFNs of the current subsection */ + unsigned long begin_pfn; + unsigned long end_pfn; + + /* + * Set RO for module text and RO-data: + * - Always protect first page. + * - Do not protect last partial page. + */ + if (ro_size > 0) + set_page_attributes(base, base + ro_size, set_memory_ro); + + /* + * Set NX permissions for module data: + * - Do not protect first partial page. + * - Always protect last page. + */ + if (total_size > text_size) { + begin_pfn = PFN_UP((unsigned long)base + text_size); + end_pfn = PFN_UP((unsigned long)base + total_size); + if (end_pfn > begin_pfn) + set_memory_nx(begin_pfn << PAGE_SHIFT, end_pfn - begin_pfn); + } +} + +/* Setting memory back to RW+NX before releasing it */ +void unset_section_ro_nx(struct module *mod, void *module_region) +{ + unsigned long total_pages; + + if (mod->module_core == module_region) { + /* Set core as NX+RW */ + total_pages = MOD_NUMBER_OF_PAGES(mod->module_core, mod->core_size); + set_memory_nx((unsigned long)mod->module_core, total_pages); + set_memory_rw((unsigned long)mod->module_core, total_pages); + + } else if (mod->module_init == module_region) { + /* Set init as NX+RW */ + total_pages = MOD_NUMBER_OF_PAGES(mod->module_init, mod->init_size); + set_memory_nx((unsigned long)mod->module_init, total_pages); + set_memory_rw((unsigned long)mod->module_init, total_pages); + } +} + +/* Iterate through all modules and set each module's text as RW */ +void set_all_modules_text_rw() +{ + struct module *mod; + + mutex_lock(&module_mutex); + list_for_each_entry_rcu(mod, &modules, list) { + if ((mod->module_core) && (mod->core_text_size)) { + set_page_attributes(mod->module_core, + mod->module_core + mod->core_text_size, + set_memory_rw); + } + if ((mod->module_init) && (mod->init_text_size)) { + set_page_attributes(mod->module_init, + mod->module_init + mod->init_text_size, + set_memory_rw); + } + } + mutex_unlock(&module_mutex); +} + +/* Iterate through all modules and set each module's text as RO */ +void set_all_modules_text_ro() +{ + struct module *mod; + + mutex_lock(&module_mutex); + list_for_each_entry_rcu(mod, &modules, list) { + if ((mod->module_core) && (mod->core_text_size)) { + set_page_attributes(mod->module_core, + mod->module_core + mod->core_text_size, + set_memory_ro); + } + if ((mod->module_init) && (mod->init_text_size)) { + set_page_attributes(mod->module_init, + mod->module_init + mod->init_text_size, + set_memory_ro); + } + } + mutex_unlock(&module_mutex); +} +#else +static inline void set_section_ro_nx(void *base, unsigned long text_size, unsigned long ro_size, unsigned long total_size) { } +static inline void unset_section_ro_nx(struct module *mod, void *module_region) { } +#endif + /* Free a module, remove from lists, etc. */ static void free_module(struct module *mod) { @@ -1566,6 +1696,7 @@ static void free_module(struct module *mod) destroy_params(mod->kp, mod->num_kp); /* This may be NULL, but that's OK */ + unset_section_ro_nx(mod, mod->module_init); module_free(mod, mod->module_init); kfree(mod->args); percpu_modfree(mod); @@ -1574,6 +1705,7 @@ static void free_module(struct module *mod) lockdep_free_key_range(mod->module_core, mod->core_size); /* Finally, free the core (containing the module structure) */ + unset_section_ro_nx(mod, mod->module_core); module_free(mod, mod->module_core); #ifdef CONFIG_MPU @@ -1777,8 +1909,19 @@ static void layout_sections(struct module *mod, struct load_info *info) s->sh_entsize = get_offset(mod, &mod->core_size, s, i); DEBUGP("\t%s\n", name); } - if (m == 0) + switch (m) { + case 0: /* executable */ + mod->core_size = debug_align(mod->core_size); mod->core_text_size = mod->core_size; + break; + case 1: /* RO: text and ro-data */ + mod->core_size = debug_align(mod->core_size); + mod->core_ro_size = mod->core_size; + break; + case 3: /* whole core */ + mod->core_size = debug_align(mod->core_size); + break; + } } DEBUGP("Init section allocation order:\n"); @@ -1796,8 +1939,19 @@ static void layout_sections(struct module *mod, struct load_info *info) | INIT_OFFSET_MASK); DEBUGP("\t%s\n", sname); } - if (m == 0) + switch (m) { + case 0: /* executable */ + mod->init_size = debug_align(mod->init_size); mod->init_text_size = mod->init_size; + break; + case 1: /* RO: text and ro-data */ + mod->init_size = debug_align(mod->init_size); + mod->init_ro_size = mod->init_size; + break; + case 3: /* whole init */ + mod->init_size = debug_align(mod->init_size); + break; + } } } @@ -2650,6 +2804,18 @@ static struct module *load_module(void __user *umod, kfree(info.strmap); free_copy(&info); + /* Set RO and NX regions for core */ + set_section_ro_nx(mod->module_core, + mod->core_text_size, + mod->core_ro_size, + mod->core_size); + + /* Set RO and NX regions for init */ + set_section_ro_nx(mod->module_init, + mod->init_text_size, + mod->init_ro_size, + mod->init_size); + /* Done! */ trace_module_load(mod); return mod; @@ -2753,6 +2919,7 @@ SYSCALL_DEFINE3(init_module, void __user *, umod, mod->symtab = mod->core_symtab; mod->strtab = mod->core_strtab; #endif + unset_section_ro_nx(mod, mod->module_init); module_free(mod, mod->module_init); mod->module_init = NULL; mod->init_size = 0; -- cgit v1.2.3 From 28d0686cf7b14e30243096bd874d3f80591ed392 Mon Sep 17 00:00:00 2001 From: Erik Gilling Date: Fri, 19 Nov 2010 18:08:51 -0800 Subject: sched: Make task dump print all 15 chars of proc comm Signed-off-by: Erik Gilling Signed-off-by: John Stultz Signed-off-by: Peter Zijlstra LKML-Reference: <1290218934-8544-3-git-send-email-john.stultz@linaro.org> Signed-off-by: Ingo Molnar --- kernel/sched.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 550cf3a02377..324afce0e223 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5249,7 +5249,7 @@ void sched_show_task(struct task_struct *p) unsigned state; state = p->state ? __ffs(p->state) + 1 : 0; - printk(KERN_INFO "%-13.13s %c", p->comm, + printk(KERN_INFO "%-15.15s %c", p->comm, state < sizeof(stat_nam) - 1 ? stat_nam[state] : '?'); #if BITS_PER_LONG == 32 if (state == TASK_RUNNING) -- cgit v1.2.3 From 70caf8a6c13c2279b35f2ad6b644815533d6c476 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sat, 20 Nov 2010 00:53:51 +0100 Subject: sched: Fix UP build breakage The recent cgroup-scheduling rework caused a UP build problem. Cc: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 390ce30ff2d0..82fd884b4e33 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -562,7 +562,7 @@ __update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr, curr->vruntime += delta_exec_weighted; update_min_vruntime(cfs_rq); -#ifdef CONFIG_FAIR_GROUP_SCHED +#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED cfs_rq->load_unacc_exec_time += delta_exec; if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) { update_cfs_load(cfs_rq, 0); -- cgit v1.2.3 From 2e01f4740a874b6085da6ebf541e7ffde9a72bf2 Mon Sep 17 00:00:00 2001 From: Dhaval Giani Date: Thu, 18 Nov 2010 15:44:54 +0100 Subject: cpu: Remove unused variable GCC warns us about: kernel/cpu.c: In function ‘take_cpu_down’: kernel/cpu.c:200:15: warning: unused variable ‘cpu’ This variable is unused since param->hcpu is directly used later on in cpu_notify. Signed-off-by: Dhaval Giani Signed-off-by: Peter Zijlstra LKML-Reference: <1290091494.1145.5.camel@gondor.retis> Signed-off-by: Ingo Molnar --- kernel/cpu.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/cpu.c b/kernel/cpu.c index 8615aa65d927..3945066fc01d 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -197,7 +197,6 @@ struct take_cpu_down_param { static int __ref take_cpu_down(void *_param) { struct take_cpu_down_param *param = _param; - unsigned int cpu = (unsigned long)param->hcpu; int err; /* Ensure this CPU doesn't handle any more interrupts. */ -- cgit v1.2.3 From 51a96c77815e7f139892a6e9c8275a50e9baebdf Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 19 Nov 2010 20:37:53 +0100 Subject: cpu: Remove incorrect BUG_ON Oleg mentioned that there is no actual guarantee the dying cpu's migration thread is actually finished running when we get there, so replace the BUG_ON() with a spinloop waiting for it. Reported-by: Oleg Nesterov Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/cpu.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/cpu.c b/kernel/cpu.c index 3945066fc01d..cb7a1efa9c2b 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -249,8 +249,11 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen) * The migration_call() CPU_DYING callback will have removed all * runnable tasks from the cpu, there's only the idle task left now * that the migration thread is done doing the stop_machine thing. + * + * Wait for the stop thread to go away. */ - BUG_ON(!idle_cpu(cpu)); + while (!idle_cpu(cpu)) + cpu_relax(); /* This actually kills the CPU. */ __cpu_die(cpu); -- cgit v1.2.3 From 5bb6b1ea67a73f0665a41726dd7138977b992c6c Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 19 Nov 2010 21:11:09 +0100 Subject: sched: Add some clock info to sched_debug Add more clock information to /proc/sched_debug, Thomas wanted to see the sched_clock_stable state. Requested-by: Thomas Gleixner Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/sched_clock.c | 2 +- kernel/sched_debug.c | 31 +++++++++++++++++++++++++++---- 2 files changed, 28 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_clock.c b/kernel/sched_clock.c index 52f1a149bfb1..9d8af0b3fb64 100644 --- a/kernel/sched_clock.c +++ b/kernel/sched_clock.c @@ -79,7 +79,7 @@ unsigned long long __attribute__((weak)) sched_clock(void) } EXPORT_SYMBOL_GPL(sched_clock); -static __read_mostly int sched_clock_running; +__read_mostly int sched_clock_running; #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK __read_mostly int sched_clock_stable; diff --git a/kernel/sched_debug.c b/kernel/sched_debug.c index e6590e7312e8..e95b77414a99 100644 --- a/kernel/sched_debug.c +++ b/kernel/sched_debug.c @@ -250,6 +250,8 @@ void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq) #undef P } +extern __read_mostly int sched_clock_running; + static void print_cpu(struct seq_file *m, int cpu) { struct rq *rq = cpu_rq(cpu); @@ -321,21 +323,42 @@ static const char *sched_tunable_scaling_names[] = { static int sched_debug_show(struct seq_file *m, void *v) { - u64 now = ktime_to_ns(ktime_get()); + u64 ktime, sched_clk, cpu_clk; + unsigned long flags; int cpu; - SEQ_printf(m, "Sched Debug Version: v0.09, %s %.*s\n", + local_irq_save(flags); + ktime = ktime_to_ns(ktime_get()); + sched_clk = sched_clock(); + cpu_clk = local_clock(); + local_irq_restore(flags); + + SEQ_printf(m, "Sched Debug Version: v0.10, %s %.*s\n", init_utsname()->release, (int)strcspn(init_utsname()->version, " "), init_utsname()->version); - SEQ_printf(m, "now at %Lu.%06ld msecs\n", SPLIT_NS(now)); +#define P(x) \ + SEQ_printf(m, "%-40s: %Ld\n", #x, (long long)(x)) +#define PN(x) \ + SEQ_printf(m, "%-40s: %Ld.%06ld\n", #x, SPLIT_NS(x)) + PN(ktime); + PN(sched_clk); + PN(cpu_clk); + P(jiffies); +#ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK + P(sched_clock_stable); +#endif +#undef PN +#undef P + + SEQ_printf(m, "\n"); + SEQ_printf(m, "sysctl_sched\n"); #define P(x) \ SEQ_printf(m, " .%-40s: %Ld\n", #x, (long long)(x)) #define PN(x) \ SEQ_printf(m, " .%-40s: %Ld.%06ld\n", #x, SPLIT_NS(x)) - P(jiffies); PN(sysctl_sched_latency); PN(sysctl_sched_min_granularity); PN(sysctl_sched_wakeup_granularity); -- cgit v1.2.3 From 335d7afbfb71faac833734a94240c1e07cf0ead8 Mon Sep 17 00:00:00 2001 From: Gerald Schaefer Date: Mon, 22 Nov 2010 15:47:36 +0100 Subject: mutexes, sched: Introduce arch_mutex_cpu_relax() The spinning mutex implementation uses cpu_relax() in busy loops as a compiler barrier. Depending on the architecture, cpu_relax() may do more than needed in this specific mutex spin loops. On System z we also give up the time slice of the virtual cpu in cpu_relax(), which prevents effective spinning on the mutex. This patch replaces cpu_relax() in the spinning mutex code with arch_mutex_cpu_relax(), which can be defined by each architecture that selects HAVE_ARCH_MUTEX_CPU_RELAX. The default is still cpu_relax(), so this patch should not affect other architectures than System z for now. Signed-off-by: Gerald Schaefer Signed-off-by: Peter Zijlstra LKML-Reference: <1290437256.7455.4.camel@thinkpad> Signed-off-by: Ingo Molnar --- arch/Kconfig | 3 +++ arch/s390/Kconfig | 1 + arch/s390/include/asm/mutex.h | 2 ++ include/linux/mutex.h | 4 ++++ kernel/mutex.c | 2 +- kernel/sched.c | 3 ++- 6 files changed, 13 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/arch/Kconfig b/arch/Kconfig index 8bf0fa652eb6..f78c2be4242b 100644 --- a/arch/Kconfig +++ b/arch/Kconfig @@ -175,4 +175,7 @@ config HAVE_PERF_EVENTS_NMI config HAVE_ARCH_JUMP_LABEL bool +config HAVE_ARCH_MUTEX_CPU_RELAX + bool + source "kernel/gcov/Kconfig" diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig index e0b98e71ff47..6c6d7b339aae 100644 --- a/arch/s390/Kconfig +++ b/arch/s390/Kconfig @@ -99,6 +99,7 @@ config S390 select HAVE_KERNEL_LZMA select HAVE_KERNEL_LZO select HAVE_GET_USER_PAGES_FAST + select HAVE_ARCH_MUTEX_CPU_RELAX select ARCH_INLINE_SPIN_TRYLOCK select ARCH_INLINE_SPIN_TRYLOCK_BH select ARCH_INLINE_SPIN_LOCK diff --git a/arch/s390/include/asm/mutex.h b/arch/s390/include/asm/mutex.h index 458c1f7fbc18..688271f5f2e4 100644 --- a/arch/s390/include/asm/mutex.h +++ b/arch/s390/include/asm/mutex.h @@ -7,3 +7,5 @@ */ #include + +#define arch_mutex_cpu_relax() barrier() diff --git a/include/linux/mutex.h b/include/linux/mutex.h index f363bc8fdc74..94b48bd40dd7 100644 --- a/include/linux/mutex.h +++ b/include/linux/mutex.h @@ -160,4 +160,8 @@ extern int mutex_trylock(struct mutex *lock); extern void mutex_unlock(struct mutex *lock); extern int atomic_dec_and_mutex_lock(atomic_t *cnt, struct mutex *lock); +#ifndef CONFIG_HAVE_ARCH_MUTEX_CPU_RELAX +#define arch_mutex_cpu_relax() cpu_relax() +#endif + #endif diff --git a/kernel/mutex.c b/kernel/mutex.c index 200407c1502f..a5889fb28ecf 100644 --- a/kernel/mutex.c +++ b/kernel/mutex.c @@ -199,7 +199,7 @@ __mutex_lock_common(struct mutex *lock, long state, unsigned int subclass, * memory barriers as we'll eventually observe the right * values at the cost of a few extra spins. */ - cpu_relax(); + arch_mutex_cpu_relax(); } #endif spin_lock_mutex(&lock->wait_lock, flags); diff --git a/kernel/sched.c b/kernel/sched.c index 3e8a7db951a6..abe7aec55763 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -75,6 +75,7 @@ #include #include +#include #include "sched_cpupri.h" #include "workqueue_sched.h" @@ -3888,7 +3889,7 @@ int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner) if (task_thread_info(rq->curr) != owner || need_resched()) return 0; - cpu_relax(); + arch_mutex_cpu_relax(); } return 1; -- cgit v1.2.3 From b7a2b39d9b7703ccf068f549c8dc3465fc41d015 Mon Sep 17 00:00:00 2001 From: Nikanth Karthikesan Date: Fri, 26 Nov 2010 12:37:09 +0530 Subject: sched: Remove unused argument dest_cpu to migrate_task() Remove unused argument, 'dest_cpu' of migrate_task(), and pass runqueue, as it is always known at the call site. Signed-off-by: Nikanth Karthikesan Signed-off-by: Peter Zijlstra LKML-Reference: <201011261237.09187.knikanth@suse.de> Signed-off-by: Ingo Molnar --- kernel/sched.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index abe7aec55763..35a6373f1265 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2061,10 +2061,8 @@ static int migration_cpu_stop(void *data); * The task's runqueue lock must be held. * Returns true if you have to wait for migration thread. */ -static bool migrate_task(struct task_struct *p, int dest_cpu) +static bool migrate_task(struct task_struct *p, struct rq *rq) { - struct rq *rq = task_rq(p); - /* * If the task is not on a runqueue (and not running), then * the next wake-up will properly place the task. @@ -3224,7 +3222,7 @@ void sched_exec(void) * select_task_rq() can race against ->cpus_allowed */ if (cpumask_test_cpu(dest_cpu, &p->cpus_allowed) && - likely(cpu_active(dest_cpu)) && migrate_task(p, dest_cpu)) { + likely(cpu_active(dest_cpu)) && migrate_task(p, rq)) { struct migration_arg arg = { p, dest_cpu }; task_rq_unlock(rq, &flags); @@ -5504,7 +5502,7 @@ again: goto out; dest_cpu = cpumask_any_and(cpu_active_mask, new_mask); - if (migrate_task(p, dest_cpu)) { + if (migrate_task(p, rq)) { struct migration_arg arg = { p, dest_cpu }; /* Need help from migration thread: drop lock and wait. */ task_rq_unlock(rq, &flags); -- cgit v1.2.3 From 24278d148316d2180be6df40e06db013d8b232b8 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Mon, 27 Sep 2010 17:25:23 -0700 Subject: rcu: priority boosting for TINY_PREEMPT_RCU Add priority boosting, but only for TINY_PREEMPT_RCU. This is enabled by the default-off RCU_BOOST kernel parameter. The priority to which to boost preempted RCU readers is controlled by the RCU_BOOST_PRIO kernel parameter (defaulting to real-time priority 1) and the time to wait before boosting the readers blocking a given grace period is controlled by the RCU_BOOST_DELAY kernel parameter (defaulting to 500 milliseconds). Signed-off-by: Paul E. McKenney Signed-off-by: Paul E. McKenney --- include/linux/init_task.h | 9 +- include/linux/sched.h | 11 ++- init/Kconfig | 39 +++++++++ kernel/rcutiny.c | 66 ++++++--------- kernel/rcutiny_plugin.h | 208 +++++++++++++++++++++++++++++++++++++++++++--- 5 files changed, 280 insertions(+), 53 deletions(-) (limited to 'kernel') diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 2fea6c8ef6ba..69f91aacdeee 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -81,6 +81,12 @@ extern struct group_info init_groups; */ # define CAP_INIT_BSET CAP_FULL_SET +#ifdef CONFIG_RCU_BOOST +#define INIT_TASK_RCU_BOOST() \ + .rcu_boost_mutex = NULL, +#else +#define INIT_TASK_RCU_BOOST() +#endif #ifdef CONFIG_TREE_PREEMPT_RCU #define INIT_TASK_RCU_TREE_PREEMPT() \ .rcu_blocked_node = NULL, @@ -92,7 +98,8 @@ extern struct group_info init_groups; .rcu_read_lock_nesting = 0, \ .rcu_read_unlock_special = 0, \ .rcu_node_entry = LIST_HEAD_INIT(tsk.rcu_node_entry), \ - INIT_TASK_RCU_TREE_PREEMPT() + INIT_TASK_RCU_TREE_PREEMPT() \ + INIT_TASK_RCU_BOOST() #else #define INIT_TASK_RCU_PREEMPT(tsk) #endif diff --git a/include/linux/sched.h b/include/linux/sched.h index e18473f0eb78..ed1a9bc52b2f 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1210,6 +1210,9 @@ struct task_struct { #ifdef CONFIG_TREE_PREEMPT_RCU struct rcu_node *rcu_blocked_node; #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */ +#ifdef CONFIG_RCU_BOOST + struct rt_mutex *rcu_boost_mutex; +#endif /* #ifdef CONFIG_RCU_BOOST */ #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) struct sched_info sched_info; @@ -1745,7 +1748,8 @@ extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t * #ifdef CONFIG_PREEMPT_RCU #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */ -#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */ +#define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */ +#define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */ static inline void rcu_copy_process(struct task_struct *p) { @@ -1753,7 +1757,10 @@ static inline void rcu_copy_process(struct task_struct *p) p->rcu_read_unlock_special = 0; #ifdef CONFIG_TREE_PREEMPT_RCU p->rcu_blocked_node = NULL; -#endif +#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */ +#ifdef CONFIG_RCU_BOOST + p->rcu_boost_mutex = NULL; +#endif /* #ifdef CONFIG_RCU_BOOST */ INIT_LIST_HEAD(&p->rcu_node_entry); } diff --git a/init/Kconfig b/init/Kconfig index a619a1ac7f4c..48efefcac12a 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -450,6 +450,45 @@ config TREE_RCU_TRACE TREE_PREEMPT_RCU implementations, permitting Makefile to trivially select kernel/rcutree_trace.c. +config RCU_BOOST + bool "Enable RCU priority boosting" + depends on RT_MUTEXES && TINY_PREEMPT_RCU + default n + help + This option boosts the priority of preempted RCU readers that + block the current preemptible RCU grace period for too long. + This option also prevents heavy loads from blocking RCU + callback invocation for all flavors of RCU. + + Say Y here if you are working with real-time apps or heavy loads + Say N here if you are unsure. + +config RCU_BOOST_PRIO + int "Real-time priority to boost RCU readers to" + range 1 99 + depends on RCU_BOOST + default 1 + help + This option specifies the real-time priority to which preempted + RCU readers are to be boosted. If you are working with CPU-bound + real-time applications, you should specify a priority higher then + the highest-priority CPU-bound application. + + Specify the real-time priority, or take the default if unsure. + +config RCU_BOOST_DELAY + int "Milliseconds to delay boosting after RCU grace-period start" + range 0 3000 + depends on RCU_BOOST + default 500 + help + This option specifies the time to wait after the beginning of + a given grace period before priority-boosting preempted RCU + readers blocking that grace period. Note that any RCU reader + blocking an expedited RCU grace period is boosted immediately. + + Accept the default if unsure. + endmenu # "RCU Subsystem" config IKCONFIG diff --git a/kernel/rcutiny.c b/kernel/rcutiny.c index 86eef29cdfb2..93d166582cbb 100644 --- a/kernel/rcutiny.c +++ b/kernel/rcutiny.c @@ -36,38 +36,16 @@ #include #include -/* Global control variables for rcupdate callback mechanism. */ -struct rcu_ctrlblk { - struct rcu_head *rcucblist; /* List of pending callbacks (CBs). */ - struct rcu_head **donetail; /* ->next pointer of last "done" CB. */ - struct rcu_head **curtail; /* ->next pointer of last CB. */ -}; - -/* Definition for rcupdate control block. */ -static struct rcu_ctrlblk rcu_sched_ctrlblk = { - .donetail = &rcu_sched_ctrlblk.rcucblist, - .curtail = &rcu_sched_ctrlblk.rcucblist, -}; - -static struct rcu_ctrlblk rcu_bh_ctrlblk = { - .donetail = &rcu_bh_ctrlblk.rcucblist, - .curtail = &rcu_bh_ctrlblk.rcucblist, -}; - -#ifdef CONFIG_DEBUG_LOCK_ALLOC -int rcu_scheduler_active __read_mostly; -EXPORT_SYMBOL_GPL(rcu_scheduler_active); -#endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ - -/* Controls for rcu_cbs() kthread, replacing RCU_SOFTIRQ used previously. */ -static struct task_struct *rcu_cbs_task; -static DECLARE_WAIT_QUEUE_HEAD(rcu_cbs_wq); -static unsigned long have_rcu_cbs; -static void invoke_rcu_cbs(void); +/* Controls for rcu_kthread() kthread, replacing RCU_SOFTIRQ used previously. */ +static struct task_struct *rcu_kthread_task; +static DECLARE_WAIT_QUEUE_HEAD(rcu_kthread_wq); +static unsigned long have_rcu_kthread_work; +static void invoke_rcu_kthread(void); /* Forward declarations for rcutiny_plugin.h. */ +struct rcu_ctrlblk; static void rcu_process_callbacks(struct rcu_ctrlblk *rcp); -static int rcu_cbs(void *arg); +static int rcu_kthread(void *arg); static void __call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu), struct rcu_ctrlblk *rcp); @@ -130,7 +108,7 @@ void rcu_sched_qs(int cpu) { if (rcu_qsctr_help(&rcu_sched_ctrlblk) + rcu_qsctr_help(&rcu_bh_ctrlblk)) - invoke_rcu_cbs(); + invoke_rcu_kthread(); } /* @@ -139,7 +117,7 @@ void rcu_sched_qs(int cpu) void rcu_bh_qs(int cpu) { if (rcu_qsctr_help(&rcu_bh_ctrlblk)) - invoke_rcu_cbs(); + invoke_rcu_kthread(); } /* @@ -201,37 +179,41 @@ static void rcu_process_callbacks(struct rcu_ctrlblk *rcp) * This is a kthread, but it is never stopped, at least not until * the system goes down. */ -static int rcu_cbs(void *arg) +static int rcu_kthread(void *arg) { unsigned long work; + unsigned long morework; unsigned long flags; for (;;) { - wait_event(rcu_cbs_wq, have_rcu_cbs != 0); + wait_event(rcu_kthread_wq, have_rcu_kthread_work != 0); + morework = rcu_boost(); local_irq_save(flags); - work = have_rcu_cbs; - have_rcu_cbs = 0; + work = have_rcu_kthread_work; + have_rcu_kthread_work = morework; local_irq_restore(flags); if (work) { rcu_process_callbacks(&rcu_sched_ctrlblk); rcu_process_callbacks(&rcu_bh_ctrlblk); rcu_preempt_process_callbacks(); } + schedule_timeout_interruptible(1); /* Leave CPU for others. */ } return 0; /* Not reached, but needed to shut gcc up. */ } /* - * Wake up rcu_cbs() to process callbacks now eligible for invocation. + * Wake up rcu_kthread() to process callbacks now eligible for invocation + * or to boost readers. */ -static void invoke_rcu_cbs(void) +static void invoke_rcu_kthread(void) { unsigned long flags; local_irq_save(flags); - have_rcu_cbs = 1; - wake_up(&rcu_cbs_wq); + have_rcu_kthread_work = 1; + wake_up(&rcu_kthread_wq); local_irq_restore(flags); } @@ -327,7 +309,11 @@ EXPORT_SYMBOL_GPL(rcu_barrier_sched); */ static int __init rcu_spawn_kthreads(void) { - rcu_cbs_task = kthread_run(rcu_cbs, NULL, "rcu_cbs"); + struct sched_param sp; + + rcu_kthread_task = kthread_run(rcu_kthread, NULL, "rcu_kthread"); + sp.sched_priority = RCU_BOOST_PRIO; + sched_setscheduler_nocheck(rcu_kthread_task, SCHED_FIFO, &sp); return 0; } early_initcall(rcu_spawn_kthreads); diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h index 95f9239df512..24f43165f222 100644 --- a/kernel/rcutiny_plugin.h +++ b/kernel/rcutiny_plugin.h @@ -24,6 +24,29 @@ #include +/* Global control variables for rcupdate callback mechanism. */ +struct rcu_ctrlblk { + struct rcu_head *rcucblist; /* List of pending callbacks (CBs). */ + struct rcu_head **donetail; /* ->next pointer of last "done" CB. */ + struct rcu_head **curtail; /* ->next pointer of last CB. */ +}; + +/* Definition for rcupdate control block. */ +static struct rcu_ctrlblk rcu_sched_ctrlblk = { + .donetail = &rcu_sched_ctrlblk.rcucblist, + .curtail = &rcu_sched_ctrlblk.rcucblist, +}; + +static struct rcu_ctrlblk rcu_bh_ctrlblk = { + .donetail = &rcu_bh_ctrlblk.rcucblist, + .curtail = &rcu_bh_ctrlblk.rcucblist, +}; + +#ifdef CONFIG_DEBUG_LOCK_ALLOC +int rcu_scheduler_active __read_mostly; +EXPORT_SYMBOL_GPL(rcu_scheduler_active); +#endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ + #ifdef CONFIG_TINY_PREEMPT_RCU #include @@ -48,17 +71,27 @@ struct rcu_preempt_ctrlblk { struct list_head *gp_tasks; /* Pointer to the first task blocking the */ /* current grace period, or NULL if there */ - /* is not such task. */ + /* is no such task. */ struct list_head *exp_tasks; /* Pointer to first task blocking the */ /* current expedited grace period, or NULL */ /* if there is no such task. If there */ /* is no current expedited grace period, */ /* then there cannot be any such task. */ +#ifdef CONFIG_RCU_BOOST + struct list_head *boost_tasks; + /* Pointer to first task that needs to be */ + /* priority-boosted, or NULL if no priority */ + /* boosting is needed. If there is no */ + /* current or expedited grace period, there */ + /* can be no such task. */ +#endif /* #ifdef CONFIG_RCU_BOOST */ u8 gpnum; /* Current grace period. */ u8 gpcpu; /* Last grace period blocked by the CPU. */ u8 completed; /* Last grace period completed. */ /* If all three are equal, RCU is idle. */ + s8 boosted_this_gp; /* Has boosting already happened? */ + unsigned long boost_time; /* When to start boosting (jiffies) */ }; static struct rcu_preempt_ctrlblk rcu_preempt_ctrlblk = { @@ -123,6 +156,130 @@ static int rcu_preempt_gp_in_progress(void) return rcu_preempt_ctrlblk.completed != rcu_preempt_ctrlblk.gpnum; } +/* + * Advance a ->blkd_tasks-list pointer to the next entry, instead + * returning NULL if at the end of the list. + */ +static struct list_head *rcu_next_node_entry(struct task_struct *t) +{ + struct list_head *np; + + np = t->rcu_node_entry.next; + if (np == &rcu_preempt_ctrlblk.blkd_tasks) + np = NULL; + return np; +} + +#ifdef CONFIG_RCU_BOOST + +#include "rtmutex_common.h" + +/* + * Carry out RCU priority boosting on the task indicated by ->boost_tasks, + * and advance ->boost_tasks to the next task in the ->blkd_tasks list. + */ +static int rcu_boost(void) +{ + unsigned long flags; + struct rt_mutex mtx; + struct list_head *np; + struct task_struct *t; + + if (rcu_preempt_ctrlblk.boost_tasks == NULL) + return 0; /* Nothing to boost. */ + raw_local_irq_save(flags); + rcu_preempt_ctrlblk.boosted_this_gp++; + t = container_of(rcu_preempt_ctrlblk.boost_tasks, struct task_struct, + rcu_node_entry); + np = rcu_next_node_entry(t); + rt_mutex_init_proxy_locked(&mtx, t); + t->rcu_boost_mutex = &mtx; + t->rcu_read_unlock_special |= RCU_READ_UNLOCK_BOOSTED; + raw_local_irq_restore(flags); + rt_mutex_lock(&mtx); + rt_mutex_unlock(&mtx); + return rcu_preempt_ctrlblk.boost_tasks != NULL; +} + +/* + * Check to see if it is now time to start boosting RCU readers blocking + * the current grace period, and, if so, tell the rcu_kthread_task to + * start boosting them. If there is an expedited boost in progress, + * we wait for it to complete. + */ +static void rcu_initiate_boost(void) +{ + if (rcu_preempt_ctrlblk.gp_tasks != NULL && + rcu_preempt_ctrlblk.boost_tasks == NULL && + rcu_preempt_ctrlblk.boosted_this_gp == 0 && + ULONG_CMP_GE(jiffies, rcu_preempt_ctrlblk.boost_time)) { + rcu_preempt_ctrlblk.boost_tasks = rcu_preempt_ctrlblk.gp_tasks; + invoke_rcu_kthread(); + } +} + +/* + * Initiate boosting for an expedited grace period. + */ +static void rcu_initiate_expedited_boost(void) +{ + unsigned long flags; + + raw_local_irq_save(flags); + if (!list_empty(&rcu_preempt_ctrlblk.blkd_tasks)) { + rcu_preempt_ctrlblk.boost_tasks = + rcu_preempt_ctrlblk.blkd_tasks.next; + rcu_preempt_ctrlblk.boosted_this_gp = -1; + invoke_rcu_kthread(); + } + raw_local_irq_restore(flags); +} + +#define RCU_BOOST_DELAY_JIFFIES DIV_ROUND_UP(CONFIG_RCU_BOOST_DELAY * HZ, 1000); + +/* + * Do priority-boost accounting for the start of a new grace period. + */ +static void rcu_preempt_boost_start_gp(void) +{ + rcu_preempt_ctrlblk.boost_time = jiffies + RCU_BOOST_DELAY_JIFFIES; + if (rcu_preempt_ctrlblk.boosted_this_gp > 0) + rcu_preempt_ctrlblk.boosted_this_gp = 0; +} + +#else /* #ifdef CONFIG_RCU_BOOST */ + +/* + * If there is no RCU priority boosting, we don't boost. + */ +static int rcu_boost(void) +{ + return 0; +} + +/* + * If there is no RCU priority boosting, we don't initiate boosting. + */ +static void rcu_initiate_boost(void) +{ +} + +/* + * If there is no RCU priority boosting, we don't initiate expedited boosting. + */ +static void rcu_initiate_expedited_boost(void) +{ +} + +/* + * If there is no RCU priority boosting, nothing to do at grace-period start. + */ +static void rcu_preempt_boost_start_gp(void) +{ +} + +#endif /* else #ifdef CONFIG_RCU_BOOST */ + /* * Record a preemptible-RCU quiescent state for the specified CPU. Note * that this just means that the task currently running on the CPU is @@ -150,12 +307,14 @@ static void rcu_preempt_cpu_qs(void) rcu_preempt_ctrlblk.gpcpu = rcu_preempt_ctrlblk.gpnum; current->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS; - /* - * If there is no GP, or if blocked readers are still blocking GP, - * then there is nothing more to do. - */ + /* If there is no GP then there is nothing more to do. */ if (!rcu_preempt_gp_in_progress() || rcu_preempt_blocked_readers_cgp()) return; + /* If there are blocked readers, go check up on boosting. */ + if (rcu_preempt_blocked_readers_cgp()) { + rcu_initiate_boost(); + return; + } /* Advance callbacks. */ rcu_preempt_ctrlblk.completed = rcu_preempt_ctrlblk.gpnum; @@ -168,7 +327,7 @@ static void rcu_preempt_cpu_qs(void) /* If there are done callbacks, cause them to be invoked. */ if (*rcu_preempt_ctrlblk.rcb.donetail != NULL) - invoke_rcu_cbs(); + invoke_rcu_kthread(); } /* @@ -186,6 +345,9 @@ static void rcu_preempt_start_gp(void) rcu_preempt_ctrlblk.gp_tasks = rcu_preempt_ctrlblk.blkd_tasks.next; + /* Set up for RCU priority boosting. */ + rcu_preempt_boost_start_gp(); + /* If there is no running reader, CPU is done with GP. */ if (!rcu_preempt_running_reader()) rcu_preempt_cpu_qs(); @@ -306,14 +468,16 @@ static void rcu_read_unlock_special(struct task_struct *t) */ empty = !rcu_preempt_blocked_readers_cgp(); empty_exp = rcu_preempt_ctrlblk.exp_tasks == NULL; - np = t->rcu_node_entry.next; - if (np == &rcu_preempt_ctrlblk.blkd_tasks) - np = NULL; + np = rcu_next_node_entry(t); list_del(&t->rcu_node_entry); if (&t->rcu_node_entry == rcu_preempt_ctrlblk.gp_tasks) rcu_preempt_ctrlblk.gp_tasks = np; if (&t->rcu_node_entry == rcu_preempt_ctrlblk.exp_tasks) rcu_preempt_ctrlblk.exp_tasks = np; +#ifdef CONFIG_RCU_BOOST + if (&t->rcu_node_entry == rcu_preempt_ctrlblk.boost_tasks) + rcu_preempt_ctrlblk.boost_tasks = np; +#endif /* #ifdef CONFIG_RCU_BOOST */ INIT_LIST_HEAD(&t->rcu_node_entry); /* @@ -333,6 +497,14 @@ static void rcu_read_unlock_special(struct task_struct *t) if (!empty_exp && rcu_preempt_ctrlblk.exp_tasks == NULL) rcu_report_exp_done(); } +#ifdef CONFIG_RCU_BOOST + /* Unboost self if was boosted. */ + if (special & RCU_READ_UNLOCK_BOOSTED) { + t->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_BOOSTED; + rt_mutex_unlock(t->rcu_boost_mutex); + t->rcu_boost_mutex = NULL; + } +#endif /* #ifdef CONFIG_RCU_BOOST */ local_irq_restore(flags); } @@ -376,7 +548,7 @@ static void rcu_preempt_check_callbacks(void) rcu_preempt_cpu_qs(); if (&rcu_preempt_ctrlblk.rcb.rcucblist != rcu_preempt_ctrlblk.rcb.donetail) - invoke_rcu_cbs(); + invoke_rcu_kthread(); if (rcu_preempt_gp_in_progress() && rcu_cpu_blocking_cur_gp() && rcu_preempt_running_reader()) @@ -534,6 +706,7 @@ void synchronize_rcu_expedited(void) /* Wait for tail of ->blkd_tasks list to drain. */ if (rcu_preempted_readers_exp()) + rcu_initiate_expedited_boost(); wait_event(sync_rcu_preempt_exp_wq, !rcu_preempted_readers_exp()); @@ -574,6 +747,15 @@ void exit_rcu(void) #else /* #ifdef CONFIG_TINY_PREEMPT_RCU */ +/* + * Because preemptible RCU does not exist, it is never necessary to + * boost preempted RCU readers. + */ +static int rcu_boost(void) +{ + return 0; +} + /* * Because preemptible RCU does not exist, it never has any callbacks * to check. @@ -614,3 +796,9 @@ void __init rcu_scheduler_starting(void) } #endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ + +#ifdef CONFIG_RCU_BOOST +#define RCU_BOOST_PRIO CONFIG_RCU_BOOST_PRIO +#else /* #ifdef CONFIG_RCU_BOOST */ +#define RCU_BOOST_PRIO 1 +#endif /* #else #ifdef CONFIG_RCU_BOOST */ -- cgit v1.2.3 From 9e571a82f0cb205a65a0ea41657f19f22b7fabb8 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Thu, 30 Sep 2010 21:26:52 -0700 Subject: rcu: add tracing for TINY_RCU and TINY_PREEMPT_RCU Add tracing for the tiny RCU implementations, including statistics on boosting in the case of TINY_PREEMPT_RCU and RCU_BOOST. Signed-off-by: Paul E. McKenney Signed-off-by: Paul E. McKenney --- init/Kconfig | 1 - kernel/rcutiny.c | 4 + kernel/rcutiny_plugin.h | 232 +++++++++++++++++++++++++++++++++++++++++++++--- 3 files changed, 226 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/init/Kconfig b/init/Kconfig index 48efefcac12a..929adf6cb6b4 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -384,7 +384,6 @@ config PREEMPT_RCU config RCU_TRACE bool "Enable tracing for RCU" - depends on TREE_RCU || TREE_PREEMPT_RCU help This option provides tracing in RCU which presents stats in debugfs for debugging RCU implementation. diff --git a/kernel/rcutiny.c b/kernel/rcutiny.c index 93d166582cbb..034493724749 100644 --- a/kernel/rcutiny.c +++ b/kernel/rcutiny.c @@ -144,6 +144,7 @@ static void rcu_process_callbacks(struct rcu_ctrlblk *rcp) { struct rcu_head *next, *list; unsigned long flags; + RCU_TRACE(int cb_count = 0); /* If no RCU callbacks ready to invoke, just return. */ if (&rcp->rcucblist == rcp->donetail) @@ -169,7 +170,9 @@ static void rcu_process_callbacks(struct rcu_ctrlblk *rcp) list->func(list); local_bh_enable(); list = next; + RCU_TRACE(cb_count++); } + RCU_TRACE(rcu_trace_sub_qlen(rcp, cb_count)); } /* @@ -252,6 +255,7 @@ static void __call_rcu(struct rcu_head *head, local_irq_save(flags); *rcp->curtail = head; rcp->curtail = &head->next; + RCU_TRACE(rcp->qlen++); local_irq_restore(flags); } diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h index 24f43165f222..f4e0df082d3c 100644 --- a/kernel/rcutiny_plugin.h +++ b/kernel/rcutiny_plugin.h @@ -23,12 +23,21 @@ */ #include +#include +#include + +#ifdef CONFIG_RCU_TRACE +#define RCU_TRACE(stmt) stmt +#else /* #ifdef CONFIG_RCU_TRACE */ +#define RCU_TRACE(stmt) +#endif /* #else #ifdef CONFIG_RCU_TRACE */ /* Global control variables for rcupdate callback mechanism. */ struct rcu_ctrlblk { struct rcu_head *rcucblist; /* List of pending callbacks (CBs). */ struct rcu_head **donetail; /* ->next pointer of last "done" CB. */ struct rcu_head **curtail; /* ->next pointer of last CB. */ + RCU_TRACE(long qlen); /* Number of pending CBs. */ }; /* Definition for rcupdate control block. */ @@ -90,8 +99,26 @@ struct rcu_preempt_ctrlblk { u8 gpcpu; /* Last grace period blocked by the CPU. */ u8 completed; /* Last grace period completed. */ /* If all three are equal, RCU is idle. */ +#ifdef CONFIG_RCU_BOOST s8 boosted_this_gp; /* Has boosting already happened? */ unsigned long boost_time; /* When to start boosting (jiffies) */ +#endif /* #ifdef CONFIG_RCU_BOOST */ +#ifdef CONFIG_RCU_TRACE + unsigned long n_grace_periods; +#ifdef CONFIG_RCU_BOOST + unsigned long n_tasks_boosted; + unsigned long n_exp_boosts; + unsigned long n_normal_boosts; + unsigned long n_normal_balk_blkd_tasks; + unsigned long n_normal_balk_gp_tasks; + unsigned long n_normal_balk_boost_tasks; + unsigned long n_normal_balk_boosted; + unsigned long n_normal_balk_notyet; + unsigned long n_normal_balk_nos; + unsigned long n_exp_balk_blkd_tasks; + unsigned long n_exp_balk_nos; +#endif /* #ifdef CONFIG_RCU_BOOST */ +#endif /* #ifdef CONFIG_RCU_TRACE */ }; static struct rcu_preempt_ctrlblk rcu_preempt_ctrlblk = { @@ -170,6 +197,65 @@ static struct list_head *rcu_next_node_entry(struct task_struct *t) return np; } +#ifdef CONFIG_RCU_TRACE + +#ifdef CONFIG_RCU_BOOST +static void rcu_initiate_boost_trace(void); +static void rcu_initiate_exp_boost_trace(void); +#endif /* #ifdef CONFIG_RCU_BOOST */ + +/* + * Dump additional statistice for TINY_PREEMPT_RCU. + */ +static void show_tiny_preempt_stats(struct seq_file *m) +{ + seq_printf(m, "rcu_preempt: qlen=%ld gp=%lu g%u/p%u/c%u tasks=%c%c%c\n", + rcu_preempt_ctrlblk.rcb.qlen, + rcu_preempt_ctrlblk.n_grace_periods, + rcu_preempt_ctrlblk.gpnum, + rcu_preempt_ctrlblk.gpcpu, + rcu_preempt_ctrlblk.completed, + "T."[list_empty(&rcu_preempt_ctrlblk.blkd_tasks)], + "N."[!rcu_preempt_ctrlblk.gp_tasks], + "E."[!rcu_preempt_ctrlblk.exp_tasks]); +#ifdef CONFIG_RCU_BOOST + seq_printf(m, " ttb=%c btg=", + "B."[!rcu_preempt_ctrlblk.boost_tasks]); + switch (rcu_preempt_ctrlblk.boosted_this_gp) { + case -1: + seq_puts(m, "exp"); + break; + case 0: + seq_puts(m, "no"); + break; + case 1: + seq_puts(m, "done"); + break; + default: + seq_printf(m, "?%d?", rcu_preempt_ctrlblk.boosted_this_gp); + } + seq_printf(m, " ntb=%lu neb=%lu nnb=%lu j=%04x bt=%04x\n", + rcu_preempt_ctrlblk.n_tasks_boosted, + rcu_preempt_ctrlblk.n_exp_boosts, + rcu_preempt_ctrlblk.n_normal_boosts, + (int)(jiffies & 0xffff), + (int)(rcu_preempt_ctrlblk.boost_time & 0xffff)); + seq_printf(m, " %s: nt=%lu gt=%lu bt=%lu b=%lu ny=%lu nos=%lu\n", + "normal balk", + rcu_preempt_ctrlblk.n_normal_balk_blkd_tasks, + rcu_preempt_ctrlblk.n_normal_balk_gp_tasks, + rcu_preempt_ctrlblk.n_normal_balk_boost_tasks, + rcu_preempt_ctrlblk.n_normal_balk_boosted, + rcu_preempt_ctrlblk.n_normal_balk_notyet, + rcu_preempt_ctrlblk.n_normal_balk_nos); + seq_printf(m, " exp balk: bt=%lu nos=%lu\n", + rcu_preempt_ctrlblk.n_exp_balk_blkd_tasks, + rcu_preempt_ctrlblk.n_exp_balk_nos); +#endif /* #ifdef CONFIG_RCU_BOOST */ +} + +#endif /* #ifdef CONFIG_RCU_TRACE */ + #ifdef CONFIG_RCU_BOOST #include "rtmutex_common.h" @@ -197,6 +283,7 @@ static int rcu_boost(void) t->rcu_read_unlock_special |= RCU_READ_UNLOCK_BOOSTED; raw_local_irq_restore(flags); rt_mutex_lock(&mtx); + RCU_TRACE(rcu_preempt_ctrlblk.n_tasks_boosted++); rt_mutex_unlock(&mtx); return rcu_preempt_ctrlblk.boost_tasks != NULL; } @@ -206,16 +293,27 @@ static int rcu_boost(void) * the current grace period, and, if so, tell the rcu_kthread_task to * start boosting them. If there is an expedited boost in progress, * we wait for it to complete. + * + * If there are no blocked readers blocking the current grace period, + * return 0 to let the caller know, otherwise return 1. Note that this + * return value is independent of whether or not boosting was done. */ -static void rcu_initiate_boost(void) +static int rcu_initiate_boost(void) { + if (!rcu_preempt_blocked_readers_cgp()) { + RCU_TRACE(rcu_preempt_ctrlblk.n_normal_balk_blkd_tasks++); + return 0; + } if (rcu_preempt_ctrlblk.gp_tasks != NULL && rcu_preempt_ctrlblk.boost_tasks == NULL && rcu_preempt_ctrlblk.boosted_this_gp == 0 && ULONG_CMP_GE(jiffies, rcu_preempt_ctrlblk.boost_time)) { rcu_preempt_ctrlblk.boost_tasks = rcu_preempt_ctrlblk.gp_tasks; invoke_rcu_kthread(); - } + RCU_TRACE(rcu_preempt_ctrlblk.n_normal_boosts++); + } else + RCU_TRACE(rcu_initiate_boost_trace()); + return 1; } /* @@ -231,7 +329,9 @@ static void rcu_initiate_expedited_boost(void) rcu_preempt_ctrlblk.blkd_tasks.next; rcu_preempt_ctrlblk.boosted_this_gp = -1; invoke_rcu_kthread(); - } + RCU_TRACE(rcu_preempt_ctrlblk.n_exp_boosts++); + } else + RCU_TRACE(rcu_initiate_exp_boost_trace()); raw_local_irq_restore(flags); } @@ -258,10 +358,13 @@ static int rcu_boost(void) } /* - * If there is no RCU priority boosting, we don't initiate boosting. + * If there is no RCU priority boosting, we don't initiate boosting, + * but we do indicate whether there are blocked readers blocking the + * current grace period. */ -static void rcu_initiate_boost(void) +static int rcu_initiate_boost(void) { + return rcu_preempt_blocked_readers_cgp(); } /* @@ -308,13 +411,14 @@ static void rcu_preempt_cpu_qs(void) current->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS; /* If there is no GP then there is nothing more to do. */ - if (!rcu_preempt_gp_in_progress() || rcu_preempt_blocked_readers_cgp()) + if (!rcu_preempt_gp_in_progress()) return; - /* If there are blocked readers, go check up on boosting. */ - if (rcu_preempt_blocked_readers_cgp()) { - rcu_initiate_boost(); + /* + * Check up on boosting. If there are no readers blocking the + * current grace period, leave. + */ + if (rcu_initiate_boost()) return; - } /* Advance callbacks. */ rcu_preempt_ctrlblk.completed = rcu_preempt_ctrlblk.gpnum; @@ -339,6 +443,7 @@ static void rcu_preempt_start_gp(void) /* Official start of GP. */ rcu_preempt_ctrlblk.gpnum++; + RCU_TRACE(rcu_preempt_ctrlblk.n_grace_periods++); /* Any blocked RCU readers block new GP. */ if (rcu_preempt_blocked_readers_any()) @@ -591,6 +696,7 @@ void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) local_irq_save(flags); *rcu_preempt_ctrlblk.nexttail = head; rcu_preempt_ctrlblk.nexttail = &head->next; + RCU_TRACE(rcu_preempt_ctrlblk.rcb.qlen++); rcu_preempt_start_gp(); /* checks to see if GP needed. */ local_irq_restore(flags); } @@ -747,6 +853,18 @@ void exit_rcu(void) #else /* #ifdef CONFIG_TINY_PREEMPT_RCU */ +#ifdef CONFIG_RCU_TRACE + +/* + * Because preemptible RCU does not exist, it is not necessary to + * dump out its statistics. + */ +static void show_tiny_preempt_stats(struct seq_file *m) +{ +} + +#endif /* #ifdef CONFIG_RCU_TRACE */ + /* * Because preemptible RCU does not exist, it is never necessary to * boost preempted RCU readers. @@ -802,3 +920,97 @@ void __init rcu_scheduler_starting(void) #else /* #ifdef CONFIG_RCU_BOOST */ #define RCU_BOOST_PRIO 1 #endif /* #else #ifdef CONFIG_RCU_BOOST */ + +#ifdef CONFIG_RCU_TRACE + +#ifdef CONFIG_RCU_BOOST + +static void rcu_initiate_boost_trace(void) +{ + if (rcu_preempt_ctrlblk.gp_tasks == NULL) + rcu_preempt_ctrlblk.n_normal_balk_gp_tasks++; + else if (rcu_preempt_ctrlblk.boost_tasks != NULL) + rcu_preempt_ctrlblk.n_normal_balk_boost_tasks++; + else if (rcu_preempt_ctrlblk.boosted_this_gp != 0) + rcu_preempt_ctrlblk.n_normal_balk_boosted++; + else if (!ULONG_CMP_GE(jiffies, rcu_preempt_ctrlblk.boost_time)) + rcu_preempt_ctrlblk.n_normal_balk_notyet++; + else + rcu_preempt_ctrlblk.n_normal_balk_nos++; +} + +static void rcu_initiate_exp_boost_trace(void) +{ + if (list_empty(&rcu_preempt_ctrlblk.blkd_tasks)) + rcu_preempt_ctrlblk.n_exp_balk_blkd_tasks++; + else + rcu_preempt_ctrlblk.n_exp_balk_nos++; +} + +#endif /* #ifdef CONFIG_RCU_BOOST */ + +static void rcu_trace_sub_qlen(struct rcu_ctrlblk *rcp, int n) +{ + unsigned long flags; + + raw_local_irq_save(flags); + rcp->qlen -= n; + raw_local_irq_restore(flags); +} + +/* + * Dump statistics for TINY_RCU, such as they are. + */ +static int show_tiny_stats(struct seq_file *m, void *unused) +{ + show_tiny_preempt_stats(m); + seq_printf(m, "rcu_sched: qlen: %ld\n", rcu_sched_ctrlblk.qlen); + seq_printf(m, "rcu_bh: qlen: %ld\n", rcu_bh_ctrlblk.qlen); + return 0; +} + +static int show_tiny_stats_open(struct inode *inode, struct file *file) +{ + return single_open(file, show_tiny_stats, NULL); +} + +static const struct file_operations show_tiny_stats_fops = { + .owner = THIS_MODULE, + .open = show_tiny_stats_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static struct dentry *rcudir; + +static int __init rcutiny_trace_init(void) +{ + struct dentry *retval; + + rcudir = debugfs_create_dir("rcu", NULL); + if (!rcudir) + goto free_out; + retval = debugfs_create_file("rcudata", 0444, rcudir, + NULL, &show_tiny_stats_fops); + if (!retval) + goto free_out; + return 0; +free_out: + debugfs_remove_recursive(rcudir); + return 1; +} + +static void __exit rcutiny_trace_cleanup(void) +{ + debugfs_remove_recursive(rcudir); +} + +module_init(rcutiny_trace_init); +module_exit(rcutiny_trace_cleanup); + +MODULE_AUTHOR("Paul E. McKenney"); +MODULE_DESCRIPTION("Read-Copy Update tracing for tiny implementation"); +MODULE_LICENSE("GPL"); + +#endif /* #ifdef CONFIG_RCU_TRACE */ -- cgit v1.2.3 From e940cc804ec212e483f91167b93d1740c2fd3415 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Thu, 4 Nov 2010 14:55:26 -0700 Subject: rcu: Distinguish between boosting and boosted RCU priority boosting's tracing did not distinguish between ongoing boosting and completion of boosting. This commit therefore adds this capability. Signed-off-by: Paul E. McKenney Signed-off-by: Paul E. McKenney --- kernel/rcutiny_plugin.h | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h index f4e0df082d3c..015abaea962a 100644 --- a/kernel/rcutiny_plugin.h +++ b/kernel/rcutiny_plugin.h @@ -229,6 +229,9 @@ static void show_tiny_preempt_stats(struct seq_file *m) seq_puts(m, "no"); break; case 1: + seq_puts(m, "begun"); + break; + case 2: seq_puts(m, "done"); break; default: @@ -284,6 +287,7 @@ static int rcu_boost(void) raw_local_irq_restore(flags); rt_mutex_lock(&mtx); RCU_TRACE(rcu_preempt_ctrlblk.n_tasks_boosted++); + rcu_preempt_ctrlblk.boosted_this_gp++; rt_mutex_unlock(&mtx); return rcu_preempt_ctrlblk.boost_tasks != NULL; } -- cgit v1.2.3 From deb7a41815a8a32d4f9ea2af7a48ed1175222cec Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Thu, 30 Sep 2010 21:33:32 -0700 Subject: rcu: get rid of obsolete "classic" names in TREE_RCU tracing The TREE_RCU tracing had obsolete rcuclassic_trace_init() and rcuclassic_trace_cleanup() function names. This commit brings them up to date: rcutree_trace_init() and rcutree_trace_cleanup(), respectively. Signed-off-by: Paul E. McKenney --- kernel/rcutree_trace.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree_trace.c b/kernel/rcutree_trace.c index d15430b9d122..78ad3c35b683 100644 --- a/kernel/rcutree_trace.c +++ b/kernel/rcutree_trace.c @@ -300,7 +300,7 @@ static const struct file_operations rcu_pending_fops = { static struct dentry *rcudir; -static int __init rcuclassic_trace_init(void) +static int __init rcutree_trace_init(void) { struct dentry *retval; @@ -337,14 +337,14 @@ free_out: return 1; } -static void __exit rcuclassic_trace_cleanup(void) +static void __exit rcutree_trace_cleanup(void) { debugfs_remove_recursive(rcudir); } -module_init(rcuclassic_trace_init); -module_exit(rcuclassic_trace_cleanup); +module_init(rcutree_trace_init); +module_exit(rcutree_trace_cleanup); MODULE_AUTHOR("Paul E. McKenney"); MODULE_DESCRIPTION("Read-Copy Update tracing for hierarchical implementation"); -- cgit v1.2.3 From 7b27d5475f86186914e54e4a6bb994e9a985337b Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Thu, 21 Oct 2010 11:29:05 +0800 Subject: rcu,cleanup: move synchronize_sched_expedited() out of sched.c The first version of synchronize_sched_expedited() used the migration code in the scheduler, and was therefore implemented in kernel/sched.c. However, the more recent version of this code no longer uses the migration code, so this commit moves it to the main RCU source files. Signed-off-by: Lai Jiangshan Signed-off-by: Paul E. McKenney --- include/linux/rcupdate.h | 1 - include/linux/rcutiny.h | 5 ++++ include/linux/rcutree.h | 1 + kernel/rcutree_plugin.h | 71 ++++++++++++++++++++++++++++++++++++++++++++++++ kernel/sched.c | 69 ---------------------------------------------- 5 files changed, 77 insertions(+), 70 deletions(-) (limited to 'kernel') diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h index 7142ee3304ab..49e8e16308e1 100644 --- a/include/linux/rcupdate.h +++ b/include/linux/rcupdate.h @@ -66,7 +66,6 @@ extern void call_rcu_sched(struct rcu_head *head, extern void synchronize_sched(void); extern void rcu_barrier_bh(void); extern void rcu_barrier_sched(void); -extern void synchronize_sched_expedited(void); extern int sched_expedited_torture_stats(char *page); static inline void __rcu_read_lock_bh(void) diff --git a/include/linux/rcutiny.h b/include/linux/rcutiny.h index ea025a611fcc..30ebd7c8d874 100644 --- a/include/linux/rcutiny.h +++ b/include/linux/rcutiny.h @@ -60,6 +60,11 @@ static inline void synchronize_rcu_bh_expedited(void) synchronize_sched(); } +static inline void synchronize_sched_expedited(void) +{ + synchronize_sched(); +} + #ifdef CONFIG_TINY_RCU static inline void rcu_preempt_note_context_switch(void) diff --git a/include/linux/rcutree.h b/include/linux/rcutree.h index c0e96833aa73..3a933482734a 100644 --- a/include/linux/rcutree.h +++ b/include/linux/rcutree.h @@ -48,6 +48,7 @@ static inline void exit_rcu(void) #endif /* #else #ifdef CONFIG_TREE_PREEMPT_RCU */ extern void synchronize_rcu_bh(void); +extern void synchronize_sched_expedited(void); extern void synchronize_rcu_expedited(void); static inline void synchronize_rcu_bh_expedited(void) diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 71a4147473f9..21df7f3e7273 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -25,6 +25,7 @@ */ #include +#include /* * Check the RCU kernel configuration parameters and print informative @@ -1014,6 +1015,76 @@ static void __init __rcu_init_preempt(void) #endif /* #else #ifdef CONFIG_TREE_PREEMPT_RCU */ +#ifndef CONFIG_SMP + +void synchronize_sched_expedited(void) +{ + cond_resched(); +} +EXPORT_SYMBOL_GPL(synchronize_sched_expedited); + +#else /* #ifndef CONFIG_SMP */ + +static atomic_t synchronize_sched_expedited_count = ATOMIC_INIT(0); + +static int synchronize_sched_expedited_cpu_stop(void *data) +{ + /* + * There must be a full memory barrier on each affected CPU + * between the time that try_stop_cpus() is called and the + * time that it returns. + * + * In the current initial implementation of cpu_stop, the + * above condition is already met when the control reaches + * this point and the following smp_mb() is not strictly + * necessary. Do smp_mb() anyway for documentation and + * robustness against future implementation changes. + */ + smp_mb(); /* See above comment block. */ + return 0; +} + +/* + * Wait for an rcu-sched grace period to elapse, but use "big hammer" + * approach to force grace period to end quickly. This consumes + * significant time on all CPUs, and is thus not recommended for + * any sort of common-case code. + * + * Note that it is illegal to call this function while holding any + * lock that is acquired by a CPU-hotplug notifier. Failing to + * observe this restriction will result in deadlock. + */ +void synchronize_sched_expedited(void) +{ + int snap, trycount = 0; + + smp_mb(); /* ensure prior mod happens before capturing snap. */ + snap = atomic_read(&synchronize_sched_expedited_count) + 1; + get_online_cpus(); + while (try_stop_cpus(cpu_online_mask, + synchronize_sched_expedited_cpu_stop, + NULL) == -EAGAIN) { + put_online_cpus(); + if (trycount++ < 10) + udelay(trycount * num_online_cpus()); + else { + synchronize_sched(); + return; + } + if (atomic_read(&synchronize_sched_expedited_count) - snap > 0) { + smp_mb(); /* ensure test happens before caller kfree */ + return; + } + get_online_cpus(); + } + atomic_inc(&synchronize_sched_expedited_count); + smp_mb__after_atomic_inc(); /* ensure post-GP actions seen after GP. */ + put_online_cpus(); +} +EXPORT_SYMBOL_GPL(synchronize_sched_expedited); + +#endif /* #else #ifndef CONFIG_SMP */ + #if !defined(CONFIG_RCU_FAST_NO_HZ) /* diff --git a/kernel/sched.c b/kernel/sched.c index ae8f75a5ceb4..d1e8889872a1 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -9131,72 +9131,3 @@ struct cgroup_subsys cpuacct_subsys = { }; #endif /* CONFIG_CGROUP_CPUACCT */ -#ifndef CONFIG_SMP - -void synchronize_sched_expedited(void) -{ - barrier(); -} -EXPORT_SYMBOL_GPL(synchronize_sched_expedited); - -#else /* #ifndef CONFIG_SMP */ - -static atomic_t synchronize_sched_expedited_count = ATOMIC_INIT(0); - -static int synchronize_sched_expedited_cpu_stop(void *data) -{ - /* - * There must be a full memory barrier on each affected CPU - * between the time that try_stop_cpus() is called and the - * time that it returns. - * - * In the current initial implementation of cpu_stop, the - * above condition is already met when the control reaches - * this point and the following smp_mb() is not strictly - * necessary. Do smp_mb() anyway for documentation and - * robustness against future implementation changes. - */ - smp_mb(); /* See above comment block. */ - return 0; -} - -/* - * Wait for an rcu-sched grace period to elapse, but use "big hammer" - * approach to force grace period to end quickly. This consumes - * significant time on all CPUs, and is thus not recommended for - * any sort of common-case code. - * - * Note that it is illegal to call this function while holding any - * lock that is acquired by a CPU-hotplug notifier. Failing to - * observe this restriction will result in deadlock. - */ -void synchronize_sched_expedited(void) -{ - int snap, trycount = 0; - - smp_mb(); /* ensure prior mod happens before capturing snap. */ - snap = atomic_read(&synchronize_sched_expedited_count) + 1; - get_online_cpus(); - while (try_stop_cpus(cpu_online_mask, - synchronize_sched_expedited_cpu_stop, - NULL) == -EAGAIN) { - put_online_cpus(); - if (trycount++ < 10) - udelay(trycount * num_online_cpus()); - else { - synchronize_sched(); - return; - } - if (atomic_read(&synchronize_sched_expedited_count) - snap > 0) { - smp_mb(); /* ensure test happens before caller kfree */ - return; - } - get_online_cpus(); - } - atomic_inc(&synchronize_sched_expedited_count); - smp_mb__after_atomic_inc(); /* ensure post-GP actions seen after GP. */ - put_online_cpus(); -} -EXPORT_SYMBOL_GPL(synchronize_sched_expedited); - -#endif /* #else #ifndef CONFIG_SMP */ -- cgit v1.2.3 From 29494be71afe2a16ad04e344306a620d7cc22d06 Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Wed, 20 Oct 2010 14:13:06 +0800 Subject: rcu,cleanup: simplify the code when cpu is dying When we handle the CPU_DYING notifier, the whole system is stopped except for the current CPU. We therefore need no synchronization with the other CPUs. This allows us to move any orphaned RCU callbacks directly to the list of any online CPU without needing to run them through the global orphan lists. These global orphan lists can therefore be dispensed with. This commit makes thes changes, though currently victimizes CPU 0 @@@. Signed-off-by: Lai Jiangshan Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 81 +++++++++++++------------------------------------ kernel/rcutree.h | 16 +++------- kernel/rcutree_plugin.h | 8 ++--- kernel/rcutree_trace.c | 4 +-- 4 files changed, 31 insertions(+), 78 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index ccdc04c47981..669d7fe049d1 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -67,9 +67,6 @@ static struct lock_class_key rcu_node_class[NUM_RCU_LVLS]; .gpnum = -300, \ .completed = -300, \ .onofflock = __RAW_SPIN_LOCK_UNLOCKED(&structname.onofflock), \ - .orphan_cbs_list = NULL, \ - .orphan_cbs_tail = &structname.orphan_cbs_list, \ - .orphan_qlen = 0, \ .fqslock = __RAW_SPIN_LOCK_UNLOCKED(&structname.fqslock), \ .n_force_qs = 0, \ .n_force_qs_ngp = 0, \ @@ -984,53 +981,31 @@ rcu_check_quiescent_state(struct rcu_state *rsp, struct rcu_data *rdp) #ifdef CONFIG_HOTPLUG_CPU /* - * Move a dying CPU's RCU callbacks to the ->orphan_cbs_list for the - * specified flavor of RCU. The callbacks will be adopted by the next - * _rcu_barrier() invocation or by the CPU_DEAD notifier, whichever - * comes first. Because this is invoked from the CPU_DYING notifier, - * irqs are already disabled. + * Move a dying CPU's RCU callbacks to online CPU's callback list. + * Synchronization is not required because this function executes + * in stop_machine() context. */ -static void rcu_send_cbs_to_orphanage(struct rcu_state *rsp) +static void rcu_send_cbs_to_online(struct rcu_state *rsp) { int i; + /* current DYING CPU is cleared in the cpu_online_mask */ + int receive_cpu = cpumask_any(cpu_online_mask); struct rcu_data *rdp = this_cpu_ptr(rsp->rda); + struct rcu_data *receive_rdp = per_cpu_ptr(rsp->rda, receive_cpu); if (rdp->nxtlist == NULL) return; /* irqs disabled, so comparison is stable. */ - raw_spin_lock(&rsp->onofflock); /* irqs already disabled. */ - *rsp->orphan_cbs_tail = rdp->nxtlist; - rsp->orphan_cbs_tail = rdp->nxttail[RCU_NEXT_TAIL]; + + *receive_rdp->nxttail[RCU_NEXT_TAIL] = rdp->nxtlist; + receive_rdp->nxttail[RCU_NEXT_TAIL] = rdp->nxttail[RCU_NEXT_TAIL]; + receive_rdp->qlen += rdp->qlen; + receive_rdp->n_cbs_adopted += rdp->qlen; + rdp->n_cbs_orphaned += rdp->qlen; + rdp->nxtlist = NULL; for (i = 0; i < RCU_NEXT_SIZE; i++) rdp->nxttail[i] = &rdp->nxtlist; - rsp->orphan_qlen += rdp->qlen; - rdp->n_cbs_orphaned += rdp->qlen; rdp->qlen = 0; - raw_spin_unlock(&rsp->onofflock); /* irqs remain disabled. */ -} - -/* - * Adopt previously orphaned RCU callbacks. - */ -static void rcu_adopt_orphan_cbs(struct rcu_state *rsp) -{ - unsigned long flags; - struct rcu_data *rdp; - - raw_spin_lock_irqsave(&rsp->onofflock, flags); - rdp = this_cpu_ptr(rsp->rda); - if (rsp->orphan_cbs_list == NULL) { - raw_spin_unlock_irqrestore(&rsp->onofflock, flags); - return; - } - *rdp->nxttail[RCU_NEXT_TAIL] = rsp->orphan_cbs_list; - rdp->nxttail[RCU_NEXT_TAIL] = rsp->orphan_cbs_tail; - rdp->qlen += rsp->orphan_qlen; - rdp->n_cbs_adopted += rsp->orphan_qlen; - rsp->orphan_cbs_list = NULL; - rsp->orphan_cbs_tail = &rsp->orphan_cbs_list; - rsp->orphan_qlen = 0; - raw_spin_unlock_irqrestore(&rsp->onofflock, flags); } /* @@ -1081,8 +1056,6 @@ static void __rcu_offline_cpu(int cpu, struct rcu_state *rsp) raw_spin_unlock_irqrestore(&rnp->lock, flags); if (need_report & RCU_OFL_TASKS_EXP_GP) rcu_report_exp_rnp(rsp, rnp); - - rcu_adopt_orphan_cbs(rsp); } /* @@ -1100,11 +1073,7 @@ static void rcu_offline_cpu(int cpu) #else /* #ifdef CONFIG_HOTPLUG_CPU */ -static void rcu_send_cbs_to_orphanage(struct rcu_state *rsp) -{ -} - -static void rcu_adopt_orphan_cbs(struct rcu_state *rsp) +static void rcu_send_cbs_to_online(struct rcu_state *rsp) { } @@ -1702,10 +1671,7 @@ static void _rcu_barrier(struct rcu_state *rsp, * early. */ atomic_set(&rcu_barrier_cpu_count, 1); - preempt_disable(); /* stop CPU_DYING from filling orphan_cbs_list */ - rcu_adopt_orphan_cbs(rsp); on_each_cpu(rcu_barrier_func, (void *)call_rcu_func, 1); - preempt_enable(); /* CPU_DYING can again fill orphan_cbs_list */ if (atomic_dec_and_test(&rcu_barrier_cpu_count)) complete(&rcu_barrier_completion); wait_for_completion(&rcu_barrier_completion); @@ -1831,18 +1797,13 @@ static int __cpuinit rcu_cpu_notify(struct notifier_block *self, case CPU_DYING: case CPU_DYING_FROZEN: /* - * preempt_disable() in _rcu_barrier() prevents stop_machine(), - * so when "on_each_cpu(rcu_barrier_func, (void *)type, 1);" - * returns, all online cpus have queued rcu_barrier_func(). - * The dying CPU clears its cpu_online_mask bit and - * moves all of its RCU callbacks to ->orphan_cbs_list - * in the context of stop_machine(), so subsequent calls - * to _rcu_barrier() will adopt these callbacks and only - * then queue rcu_barrier_func() on all remaining CPUs. + * The whole machine is "stopped" except this cpu, so we can + * touch any data without introducing corruption. And we send + * the callbacks to an attribute chosen online cpu. */ - rcu_send_cbs_to_orphanage(&rcu_bh_state); - rcu_send_cbs_to_orphanage(&rcu_sched_state); - rcu_preempt_send_cbs_to_orphanage(); + rcu_send_cbs_to_online(&rcu_bh_state); + rcu_send_cbs_to_online(&rcu_sched_state); + rcu_preempt_send_cbs_to_online(); break; case CPU_DEAD: case CPU_DEAD_FROZEN: diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 91d4170c5c13..1a54be2a902f 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -203,8 +203,8 @@ struct rcu_data { long qlen_last_fqs_check; /* qlen at last check for QS forcing */ unsigned long n_cbs_invoked; /* count of RCU cbs invoked. */ - unsigned long n_cbs_orphaned; /* RCU cbs sent to orphanage. */ - unsigned long n_cbs_adopted; /* RCU cbs adopted from orphanage. */ + unsigned long n_cbs_orphaned; /* RCU cbs orphaned by dying CPU */ + unsigned long n_cbs_adopted; /* RCU cbs adopted from dying CPU */ unsigned long n_force_qs_snap; /* did other CPU force QS recently? */ long blimit; /* Upper limit on a processed batch */ @@ -309,15 +309,7 @@ struct rcu_state { /* End of fields guarded by root rcu_node's lock. */ raw_spinlock_t onofflock; /* exclude on/offline and */ - /* starting new GP. Also */ - /* protects the following */ - /* orphan_cbs fields. */ - struct rcu_head *orphan_cbs_list; /* list of rcu_head structs */ - /* orphaned by all CPUs in */ - /* a given leaf rcu_node */ - /* going offline. */ - struct rcu_head **orphan_cbs_tail; /* And tail pointer. */ - long orphan_qlen; /* Number of orphaned cbs. */ + /* starting new GP. */ raw_spinlock_t fqslock; /* Only one task forcing */ /* quiescent states. */ unsigned long jiffies_force_qs; /* Time at which to invoke */ @@ -390,7 +382,7 @@ static void rcu_report_exp_rnp(struct rcu_state *rsp, struct rcu_node *rnp); static int rcu_preempt_pending(int cpu); static int rcu_preempt_needs_cpu(int cpu); static void __cpuinit rcu_preempt_init_percpu_data(int cpu); -static void rcu_preempt_send_cbs_to_orphanage(void); +static void rcu_preempt_send_cbs_to_online(void); static void __init __rcu_init_preempt(void); static void rcu_needs_cpu_flush(void); diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 21df7f3e7273..0de359be5b41 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -774,11 +774,11 @@ static void __cpuinit rcu_preempt_init_percpu_data(int cpu) } /* - * Move preemptable RCU's callbacks to ->orphan_cbs_list. + * Move preemptable DYING RCU's callbacks to other online CPU. */ -static void rcu_preempt_send_cbs_to_orphanage(void) +static void rcu_preempt_send_cbs_to_online(void) { - rcu_send_cbs_to_orphanage(&rcu_preempt_state); + rcu_send_cbs_to_online(&rcu_preempt_state); } /* @@ -1002,7 +1002,7 @@ static void __cpuinit rcu_preempt_init_percpu_data(int cpu) /* * Because there is no preemptable RCU, there are no callbacks to move. */ -static void rcu_preempt_send_cbs_to_orphanage(void) +static void rcu_preempt_send_cbs_to_online(void) { } diff --git a/kernel/rcutree_trace.c b/kernel/rcutree_trace.c index 78ad3c35b683..c8e97853b970 100644 --- a/kernel/rcutree_trace.c +++ b/kernel/rcutree_trace.c @@ -166,13 +166,13 @@ static void print_one_rcu_state(struct seq_file *m, struct rcu_state *rsp) gpnum = rsp->gpnum; seq_printf(m, "c=%lu g=%lu s=%d jfq=%ld j=%x " - "nfqs=%lu/nfqsng=%lu(%lu) fqlh=%lu oqlen=%ld\n", + "nfqs=%lu/nfqsng=%lu(%lu) fqlh=%lu\n", rsp->completed, gpnum, rsp->signaled, (long)(rsp->jiffies_force_qs - jiffies), (int)(jiffies & 0xffff), rsp->n_force_qs, rsp->n_force_qs_ngp, rsp->n_force_qs - rsp->n_force_qs_ngp, - rsp->n_force_qs_lh, rsp->orphan_qlen); + rsp->n_force_qs_lh); for (rnp = &rsp->node[0]; rnp - &rsp->node[0] < NUM_RCU_NODES; rnp++) { if (rnp->level != level) { seq_puts(m, "\n"); -- cgit v1.2.3 From 2d999e03b7c8305b4385dd20992e4ed3e827177b Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Wed, 20 Oct 2010 12:06:18 -0700 Subject: rcu: update documentation/comments for Lai's adoption patch Lai's RCU-callback immediate-adoption patch changes the RCU tracing output, so update tracing.txt. Also update a few comments to clarify the synchronization design. Signed-off-by: Paul E. McKenney --- Documentation/RCU/trace.txt | 12 ++++-------- kernel/rcutree.c | 10 ++++++---- kernel/rcutree_plugin.h | 2 +- 3 files changed, 11 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/Documentation/RCU/trace.txt b/Documentation/RCU/trace.txt index ff6d3f10c82f..6a8c73f55b80 100644 --- a/Documentation/RCU/trace.txt +++ b/Documentation/RCU/trace.txt @@ -134,7 +134,8 @@ o "ci" is the number of RCU callbacks that have been invoked for been registered in absence of CPU-hotplug activity. o "co" is the number of RCU callbacks that have been orphaned due to - this CPU going offline. + this CPU going offline. These orphaned callbacks have been moved + to an arbitrarily chosen online CPU. o "ca" is the number of RCU callbacks that have been adopted due to other CPUs going offline. Note that ci+co-ca+ql is the number of @@ -172,12 +173,12 @@ o "gpnum" is the number of grace periods that have started. It is The output of "cat rcu/rcuhier" looks as follows, with very long lines: -c=6902 g=6903 s=2 jfq=3 j=72c7 nfqs=13142/nfqsng=0(13142) fqlh=6 oqlen=0 +c=6902 g=6903 s=2 jfq=3 j=72c7 nfqs=13142/nfqsng=0(13142) fqlh=6 1/1 .>. 0:127 ^0 3/3 .>. 0:35 ^0 0/0 .>. 36:71 ^1 0/0 .>. 72:107 ^2 0/0 .>. 108:127 ^3 3/3f .>. 0:5 ^0 2/3 .>. 6:11 ^1 0/0 .>. 12:17 ^2 0/0 .>. 18:23 ^3 0/0 .>. 24:29 ^4 0/0 .>. 30:35 ^5 0/0 .>. 36:41 ^0 0/0 .>. 42:47 ^1 0/0 .>. 48:53 ^2 0/0 .>. 54:59 ^3 0/0 .>. 60:65 ^4 0/0 .>. 66:71 ^5 0/0 .>. 72:77 ^0 0/0 .>. 78:83 ^1 0/0 .>. 84:89 ^2 0/0 .>. 90:95 ^3 0/0 .>. 96:101 ^4 0/0 .>. 102:107 ^5 0/0 .>. 108:113 ^0 0/0 .>. 114:119 ^1 0/0 .>. 120:125 ^2 0/0 .>. 126:127 ^3 rcu_bh: -c=-226 g=-226 s=1 jfq=-5701 j=72c7 nfqs=88/nfqsng=0(88) fqlh=0 oqlen=0 +c=-226 g=-226 s=1 jfq=-5701 j=72c7 nfqs=88/nfqsng=0(88) fqlh=0 0/1 .>. 0:127 ^0 0/3 .>. 0:35 ^0 0/0 .>. 36:71 ^1 0/0 .>. 72:107 ^2 0/0 .>. 108:127 ^3 0/3f .>. 0:5 ^0 0/3 .>. 6:11 ^1 0/0 .>. 12:17 ^2 0/0 .>. 18:23 ^3 0/0 .>. 24:29 ^4 0/0 .>. 30:35 ^5 0/0 .>. 36:41 ^0 0/0 .>. 42:47 ^1 0/0 .>. 48:53 ^2 0/0 .>. 54:59 ^3 0/0 .>. 60:65 ^4 0/0 .>. 66:71 ^5 0/0 .>. 72:77 ^0 0/0 .>. 78:83 ^1 0/0 .>. 84:89 ^2 0/0 .>. 90:95 ^3 0/0 .>. 96:101 ^4 0/0 .>. 102:107 ^5 0/0 .>. 108:113 ^0 0/0 .>. 114:119 ^1 0/0 .>. 120:125 ^2 0/0 .>. 126:127 ^3 @@ -216,11 +217,6 @@ o "fqlh" is the number of calls to force_quiescent_state() that exited immediately (without even being counted in nfqs above) due to contention on ->fqslock. -o "oqlen" is the number of callbacks on the "orphan" callback - list. RCU callbacks are placed on this list by CPUs going - offline, and are "adopted" either by the CPU helping the outgoing - CPU or by the next rcu_barrier*() call, whichever comes first. - o Each element of the form "1/1 0:127 ^0" represents one struct rcu_node. Each line represents one level of the hierarchy, from root to leaves. It is best to think of the rcu_data structures diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 669d7fe049d1..120820ffc657 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -1668,7 +1668,9 @@ static void _rcu_barrier(struct rcu_state *rsp, * decrement rcu_barrier_cpu_count -- otherwise the first CPU * might complete its grace period before all of the other CPUs * did their increment, causing this function to return too - * early. + * early. Note that on_each_cpu() disables irqs, which prevents + * any CPUs from coming online or going offline until each online + * CPU has queued its RCU-barrier callback. */ atomic_set(&rcu_barrier_cpu_count, 1); on_each_cpu(rcu_barrier_func, (void *)call_rcu_func, 1); @@ -1797,9 +1799,9 @@ static int __cpuinit rcu_cpu_notify(struct notifier_block *self, case CPU_DYING: case CPU_DYING_FROZEN: /* - * The whole machine is "stopped" except this cpu, so we can - * touch any data without introducing corruption. And we send - * the callbacks to an attribute chosen online cpu. + * The whole machine is "stopped" except this CPU, so we can + * touch any data without introducing corruption. We send the + * dying CPU's callbacks to an arbitrarily chosen online CPU. */ rcu_send_cbs_to_online(&rcu_bh_state); rcu_send_cbs_to_online(&rcu_sched_state); diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 0de359be5b41..643c8f650dd0 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -774,7 +774,7 @@ static void __cpuinit rcu_preempt_init_percpu_data(int cpu) } /* - * Move preemptable DYING RCU's callbacks to other online CPU. + * Move preemptable RCU's callbacks from dying CPU to other online CPU. */ static void rcu_preempt_send_cbs_to_online(void) { -- cgit v1.2.3 From db3a8920995484e5e9a0abaf3bad2c7311b163db Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Mon, 25 Oct 2010 07:39:22 -0700 Subject: rcu: fix race condition in synchronize_sched_expedited() The new (early 2010) implementation of synchronize_sched_expedited() uses try_stop_cpu() to force a context switch on every CPU. It also permits concurrent calls to synchronize_sched_expedited() to share a single call to try_stop_cpu() through use of an atomically incremented synchronize_sched_expedited_count variable. Unfortunately, this is subject to failure as follows: o Task A invokes synchronize_sched_expedited(), try_stop_cpus() succeeds, but Task A is preempted before getting to the atomic increment of synchronize_sched_expedited_count. o Task B also invokes synchronize_sched_expedited(), with exactly the same outcome as Task A. o Task C also invokes synchronize_sched_expedited(), again with exactly the same outcome as Tasks A and B. o Task D also invokes synchronize_sched_expedited(), but only gets as far as acquiring the mutex within try_stop_cpus() before being preempted, interrupted, or otherwise delayed. o Task E also invokes synchronize_sched_expedited(), but only gets to the snapshotting of synchronize_sched_expedited_count. o Tasks A, B, and C all increment synchronize_sched_expedited_count. o Task E fails to get the mutex, so checks the new value of synchronize_sched_expedited_count. It finds that the value has increased, so (wrongly) assumes that its work has been done, returning despite there having been no expedited grace period since it began. The solution is to have the lowest-numbered CPU atomically increment the synchronize_sched_expedited_count variable within the synchronize_sched_expedited_cpu_stop() function, which is under the protection of the mutex acquired by try_stop_cpus(). However, this also requires that piggybacking tasks wait for three rather than two instances of try_stop_cpu(), because we cannot control the order in which the per-CPU callback function occur. Cc: Tejun Heo Cc: Lai Jiangshan Signed-off-by: Paul E. McKenney --- kernel/rcutree_plugin.h | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 643c8f650dd0..c22c4ef2a0d0 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -1041,6 +1041,8 @@ static int synchronize_sched_expedited_cpu_stop(void *data) * robustness against future implementation changes. */ smp_mb(); /* See above comment block. */ + if (cpumask_first(cpu_online_mask) == smp_processor_id()) + atomic_inc(&synchronize_sched_expedited_count); return 0; } @@ -1053,13 +1055,26 @@ static int synchronize_sched_expedited_cpu_stop(void *data) * Note that it is illegal to call this function while holding any * lock that is acquired by a CPU-hotplug notifier. Failing to * observe this restriction will result in deadlock. + * + * The synchronize_sched_expedited_cpu_stop() function is called + * in stop-CPU context, but in order to keep overhead down to a dull + * roar, we don't force this function to wait for its counterparts + * on other CPUs. One instance of this function will increment the + * synchronize_sched_expedited_count variable per call to + * try_stop_cpus(), but there is no guarantee what order this instance + * will occur in. The worst case is that it is last on one call + * to try_stop_cpus(), and the first on the next call. This means + * that piggybacking requires that synchronize_sched_expedited_count + * be incremented by 3: this guarantees that the piggybacking + * task has waited through an entire cycle of context switches, + * even in the worst case. */ void synchronize_sched_expedited(void) { int snap, trycount = 0; smp_mb(); /* ensure prior mod happens before capturing snap. */ - snap = atomic_read(&synchronize_sched_expedited_count) + 1; + snap = atomic_read(&synchronize_sched_expedited_count) + 2; get_online_cpus(); while (try_stop_cpus(cpu_online_mask, synchronize_sched_expedited_cpu_stop, @@ -1077,7 +1092,6 @@ void synchronize_sched_expedited(void) } get_online_cpus(); } - atomic_inc(&synchronize_sched_expedited_count); smp_mb__after_atomic_inc(); /* ensure post-GP actions seen after GP. */ put_online_cpus(); } -- cgit v1.2.3 From 46fdb0937f26124700fc9fc80da4776330cc00d3 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Tue, 26 Oct 2010 02:11:40 -0700 Subject: rcu: Make synchronize_srcu_expedited() fast if running readers The synchronize_srcu_expedited() function is currently quick if there are no active readers, but will delay a full jiffy if there are any. If these readers leave their SRCU read-side critical sections quickly, this is way too long to wait. So this commit first waits ten microseconds, and only then falls back to jiffy-at-a-time waiting. Reported-by: Avi Kivity Reported-by: Marcelo Tosatti Tested-by: Takuya Yoshikawa Signed-off-by: Paul E. McKenney --- init/Kconfig | 15 +++++++++++++++ kernel/srcu.c | 8 +++++++- 2 files changed, 22 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/init/Kconfig b/init/Kconfig index 929adf6cb6b4..35518243c4bd 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -488,6 +488,21 @@ config RCU_BOOST_DELAY Accept the default if unsure. +config SRCU_SYNCHRONIZE_DELAY + int "Microseconds to delay before waiting for readers" + range 0 20 + default 10 + help + This option controls how long SRCU delays before entering its + loop waiting on SRCU readers. The purpose of this loop is + to avoid the unconditional context-switch penalty that would + otherwise be incurred if there was an active SRCU reader, + in a manner similar to adaptive locking schemes. This should + be set to be a bit longer than the common-case SRCU read-side + critical-section overhead. + + Accept the default if unsure. + endmenu # "RCU Subsystem" config IKCONFIG diff --git a/kernel/srcu.c b/kernel/srcu.c index c71e07500536..98d8c1e80edb 100644 --- a/kernel/srcu.c +++ b/kernel/srcu.c @@ -31,6 +31,7 @@ #include #include #include +#include #include static int init_srcu_struct_fields(struct srcu_struct *sp) @@ -203,9 +204,14 @@ static void __synchronize_srcu(struct srcu_struct *sp, void (*sync_func)(void)) * all srcu_read_lock() calls using the old counters have completed. * Their corresponding critical sections might well be still * executing, but the srcu_read_lock() primitives themselves - * will have finished executing. + * will have finished executing. We initially give readers + * an arbitrarily chosen 10 microseconds to get out of their + * SRCU read-side critical sections, then loop waiting 1/HZ + * seconds per iteration. */ + if (srcu_readers_active_idx(sp, idx)) + udelay(CONFIG_SRCU_SYNCHRONIZE_DELAY); while (srcu_readers_active_idx(sp, idx)) schedule_timeout_interruptible(1); -- cgit v1.2.3 From 822bc180a7f7a7bc5fcaaea195f41b487cc8cae8 Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Mon, 29 Nov 2010 16:55:40 -0800 Subject: sched: Fix unregister_fair_sched_group() In the flipping and flopping between calling unregister_fair_sched_group() on a per-cpu versus per-group basis we ended up in a bad state. Remove from the list for the passed cpu as opposed to some arbitrary index. ( This fixes explosions w/ autogroup as well as a group creation/destruction stress test. ) Reported-by: Stephen Rothwell Signed-off-by: Paul Turner Cc: Peter Zijlstra Cc: Mike Galbraith LKML-Reference: <20101130005740.080828123@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 35a6373f1265..66ef5790d932 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -8085,7 +8085,6 @@ static inline void unregister_fair_sched_group(struct task_group *tg, int cpu) { struct rq *rq = cpu_rq(cpu); unsigned long flags; - int i; /* * Only empty task groups can be destroyed; so we can speculatively @@ -8095,7 +8094,7 @@ static inline void unregister_fair_sched_group(struct task_group *tg, int cpu) return; raw_spin_lock_irqsave(&rq->lock, flags); - list_del_leaf_cfs_rq(tg->cfs_rq[i]); + list_del_leaf_cfs_rq(tg->cfs_rq[cpu]); raw_spin_unlock_irqrestore(&rq->lock, flags); } #else /* !CONFG_FAIR_GROUP_SCHED */ -- cgit v1.2.3 From 5091faa449ee0b7d73bc296a93bca9540fc51d0a Mon Sep 17 00:00:00 2001 From: Mike Galbraith Date: Tue, 30 Nov 2010 14:18:03 +0100 Subject: sched: Add 'autogroup' scheduling feature: automated per session task groups A recurring complaint from CFS users is that parallel kbuild has a negative impact on desktop interactivity. This patch implements an idea from Linus, to automatically create task groups. Currently, only per session autogroups are implemented, but the patch leaves the way open for enhancement. Implementation: each task's signal struct contains an inherited pointer to a refcounted autogroup struct containing a task group pointer, the default for all tasks pointing to the init_task_group. When a task calls setsid(), a new task group is created, the process is moved into the new task group, and a reference to the preveious task group is dropped. Child processes inherit this task group thereafter, and increase it's refcount. When the last thread of a process exits, the process's reference is dropped, such that when the last process referencing an autogroup exits, the autogroup is destroyed. At runqueue selection time, IFF a task has no cgroup assignment, its current autogroup is used. Autogroup bandwidth is controllable via setting it's nice level through the proc filesystem: cat /proc//autogroup Displays the task's group and the group's nice level. echo > /proc//autogroup Sets the task group's shares to the weight of nice task. Setting nice level is rate limited for !admin users due to the abuse risk of task group locking. The feature is enabled from boot by default if CONFIG_SCHED_AUTOGROUP=y is selected, but can be disabled via the boot option noautogroup, and can also be turned on/off on the fly via: echo [01] > /proc/sys/kernel/sched_autogroup_enabled ... which will automatically move tasks to/from the root task group. Signed-off-by: Mike Galbraith Acked-by: Linus Torvalds Acked-by: Peter Zijlstra Cc: Markus Trippelsdorf Cc: Mathieu Desnoyers Cc: Paul Turner Cc: Oleg Nesterov [ Removed the task_group_path() debug code, and fixed !EVENTFD build failure. ] Signed-off-by: Ingo Molnar LKML-Reference: <1290281700.28711.9.camel@maggy.simson.net> Signed-off-by: Ingo Molnar --- Documentation/kernel-parameters.txt | 2 + fs/proc/base.c | 79 +++++++++++++ include/linux/sched.h | 23 ++++ init/Kconfig | 13 ++ kernel/fork.c | 5 +- kernel/sched.c | 13 +- kernel/sched_autogroup.c | 229 ++++++++++++++++++++++++++++++++++++ kernel/sched_autogroup.h | 32 +++++ kernel/sched_debug.c | 47 +------- kernel/sys.c | 4 +- kernel/sysctl.c | 11 ++ 11 files changed, 409 insertions(+), 49 deletions(-) create mode 100644 kernel/sched_autogroup.c create mode 100644 kernel/sched_autogroup.h (limited to 'kernel') diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index 92e83e53148f..86820a727b0b 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -1622,6 +1622,8 @@ and is between 256 and 4096 characters. It is defined in the file noapic [SMP,APIC] Tells the kernel to not make use of any IOAPICs that may be present in the system. + noautogroup Disable scheduler automatic task group creation. + nobats [PPC] Do not use BATs for mapping kernel lowmem on "Classic" PPC cores. diff --git a/fs/proc/base.c b/fs/proc/base.c index f3d02ca461ec..2fa0ce29b6dc 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -1407,6 +1407,82 @@ static const struct file_operations proc_pid_sched_operations = { #endif +#ifdef CONFIG_SCHED_AUTOGROUP +/* + * Print out autogroup related information: + */ +static int sched_autogroup_show(struct seq_file *m, void *v) +{ + struct inode *inode = m->private; + struct task_struct *p; + + p = get_proc_task(inode); + if (!p) + return -ESRCH; + proc_sched_autogroup_show_task(p, m); + + put_task_struct(p); + + return 0; +} + +static ssize_t +sched_autogroup_write(struct file *file, const char __user *buf, + size_t count, loff_t *offset) +{ + struct inode *inode = file->f_path.dentry->d_inode; + struct task_struct *p; + char buffer[PROC_NUMBUF]; + long nice; + int err; + + memset(buffer, 0, sizeof(buffer)); + if (count > sizeof(buffer) - 1) + count = sizeof(buffer) - 1; + if (copy_from_user(buffer, buf, count)) + return -EFAULT; + + err = strict_strtol(strstrip(buffer), 0, &nice); + if (err) + return -EINVAL; + + p = get_proc_task(inode); + if (!p) + return -ESRCH; + + err = nice; + err = proc_sched_autogroup_set_nice(p, &err); + if (err) + count = err; + + put_task_struct(p); + + return count; +} + +static int sched_autogroup_open(struct inode *inode, struct file *filp) +{ + int ret; + + ret = single_open(filp, sched_autogroup_show, NULL); + if (!ret) { + struct seq_file *m = filp->private_data; + + m->private = inode; + } + return ret; +} + +static const struct file_operations proc_pid_sched_autogroup_operations = { + .open = sched_autogroup_open, + .read = seq_read, + .write = sched_autogroup_write, + .llseek = seq_lseek, + .release = single_release, +}; + +#endif /* CONFIG_SCHED_AUTOGROUP */ + static ssize_t comm_write(struct file *file, const char __user *buf, size_t count, loff_t *offset) { @@ -2732,6 +2808,9 @@ static const struct pid_entry tgid_base_stuff[] = { INF("limits", S_IRUGO, proc_pid_limits), #ifdef CONFIG_SCHED_DEBUG REG("sched", S_IRUGO|S_IWUSR, proc_pid_sched_operations), +#endif +#ifdef CONFIG_SCHED_AUTOGROUP + REG("autogroup", S_IRUGO|S_IWUSR, proc_pid_sched_autogroup_operations), #endif REG("comm", S_IRUGO|S_IWUSR, proc_pid_set_comm_operations), #ifdef CONFIG_HAVE_ARCH_TRACEHOOK diff --git a/include/linux/sched.h b/include/linux/sched.h index a5b92c70c737..9c2d46da486e 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -509,6 +509,8 @@ struct thread_group_cputimer { spinlock_t lock; }; +struct autogroup; + /* * NOTE! "signal_struct" does not have it's own * locking, because a shared signal_struct always @@ -576,6 +578,9 @@ struct signal_struct { struct tty_struct *tty; /* NULL if no tty */ +#ifdef CONFIG_SCHED_AUTOGROUP + struct autogroup *autogroup; +#endif /* * Cumulative resource counters for dead threads in the group, * and for reaped dead child processes forked by this group. @@ -1927,6 +1932,24 @@ int sched_rt_handler(struct ctl_table *table, int write, extern unsigned int sysctl_sched_compat_yield; +#ifdef CONFIG_SCHED_AUTOGROUP +extern unsigned int sysctl_sched_autogroup_enabled; + +extern void sched_autogroup_create_attach(struct task_struct *p); +extern void sched_autogroup_detach(struct task_struct *p); +extern void sched_autogroup_fork(struct signal_struct *sig); +extern void sched_autogroup_exit(struct signal_struct *sig); +#ifdef CONFIG_PROC_FS +extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m); +extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice); +#endif +#else +static inline void sched_autogroup_create_attach(struct task_struct *p) { } +static inline void sched_autogroup_detach(struct task_struct *p) { } +static inline void sched_autogroup_fork(struct signal_struct *sig) { } +static inline void sched_autogroup_exit(struct signal_struct *sig) { } +#endif + #ifdef CONFIG_RT_MUTEXES extern int rt_mutex_getprio(struct task_struct *p); extern void rt_mutex_setprio(struct task_struct *p, int prio); diff --git a/init/Kconfig b/init/Kconfig index 88c10468db46..f1bba0a1b051 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -728,6 +728,19 @@ config NET_NS endif # NAMESPACES +config SCHED_AUTOGROUP + bool "Automatic process group scheduling" + select EVENTFD + select CGROUPS + select CGROUP_SCHED + select FAIR_GROUP_SCHED + help + This option optimizes the scheduler for common desktop workloads by + automatically creating and populating task groups. This separation + of workloads isolates aggressive CPU burners (like build jobs) from + desktop applications. Task group autogeneration is currently based + upon task session. + config MM_OWNER bool diff --git a/kernel/fork.c b/kernel/fork.c index 3b159c5991b7..b6f2475f1e83 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -174,8 +174,10 @@ static inline void free_signal_struct(struct signal_struct *sig) static inline void put_signal_struct(struct signal_struct *sig) { - if (atomic_dec_and_test(&sig->sigcnt)) + if (atomic_dec_and_test(&sig->sigcnt)) { + sched_autogroup_exit(sig); free_signal_struct(sig); + } } void __put_task_struct(struct task_struct *tsk) @@ -904,6 +906,7 @@ static int copy_signal(unsigned long clone_flags, struct task_struct *tsk) posix_cpu_timers_init_group(sig); tty_audit_fork(sig); + sched_autogroup_fork(sig); sig->oom_adj = current->signal->oom_adj; sig->oom_score_adj = current->signal->oom_score_adj; diff --git a/kernel/sched.c b/kernel/sched.c index 66ef5790d932..b646dad4a40e 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -79,6 +79,7 @@ #include "sched_cpupri.h" #include "workqueue_sched.h" +#include "sched_autogroup.h" #define CREATE_TRACE_POINTS #include @@ -271,6 +272,10 @@ struct task_group { struct task_group *parent; struct list_head siblings; struct list_head children; + +#ifdef CONFIG_SCHED_AUTOGROUP + struct autogroup *autogroup; +#endif }; #define root_task_group init_task_group @@ -603,11 +608,14 @@ static inline int cpu_of(struct rq *rq) */ static inline struct task_group *task_group(struct task_struct *p) { + struct task_group *tg; struct cgroup_subsys_state *css; css = task_subsys_state_check(p, cpu_cgroup_subsys_id, lockdep_is_held(&task_rq(p)->lock)); - return container_of(css, struct task_group, css); + tg = container_of(css, struct task_group, css); + + return autogroup_task_group(p, tg); } /* Change a task's cfs_rq and parent entity if it moves across CPUs/groups */ @@ -1869,6 +1877,7 @@ static void sched_irq_time_avg_update(struct rq *rq, u64 curr_irq_time) { } #include "sched_idletask.c" #include "sched_fair.c" #include "sched_rt.c" +#include "sched_autogroup.c" #include "sched_stoptask.c" #ifdef CONFIG_SCHED_DEBUG # include "sched_debug.c" @@ -7750,7 +7759,7 @@ void __init sched_init(void) #ifdef CONFIG_CGROUP_SCHED list_add(&init_task_group.list, &task_groups); INIT_LIST_HEAD(&init_task_group.children); - + autogroup_init(&init_task); #endif /* CONFIG_CGROUP_SCHED */ for_each_possible_cpu(i) { diff --git a/kernel/sched_autogroup.c b/kernel/sched_autogroup.c new file mode 100644 index 000000000000..57a7ac286a02 --- /dev/null +++ b/kernel/sched_autogroup.c @@ -0,0 +1,229 @@ +#ifdef CONFIG_SCHED_AUTOGROUP + +#include +#include +#include +#include + +unsigned int __read_mostly sysctl_sched_autogroup_enabled = 1; +static struct autogroup autogroup_default; +static atomic_t autogroup_seq_nr; + +static void autogroup_init(struct task_struct *init_task) +{ + autogroup_default.tg = &init_task_group; + init_task_group.autogroup = &autogroup_default; + kref_init(&autogroup_default.kref); + init_rwsem(&autogroup_default.lock); + init_task->signal->autogroup = &autogroup_default; +} + +static inline void autogroup_free(struct task_group *tg) +{ + kfree(tg->autogroup); +} + +static inline void autogroup_destroy(struct kref *kref) +{ + struct autogroup *ag = container_of(kref, struct autogroup, kref); + + sched_destroy_group(ag->tg); +} + +static inline void autogroup_kref_put(struct autogroup *ag) +{ + kref_put(&ag->kref, autogroup_destroy); +} + +static inline struct autogroup *autogroup_kref_get(struct autogroup *ag) +{ + kref_get(&ag->kref); + return ag; +} + +static inline struct autogroup *autogroup_create(void) +{ + struct autogroup *ag = kzalloc(sizeof(*ag), GFP_KERNEL); + struct task_group *tg; + + if (!ag) + goto out_fail; + + tg = sched_create_group(&init_task_group); + + if (IS_ERR(tg)) + goto out_free; + + kref_init(&ag->kref); + init_rwsem(&ag->lock); + ag->id = atomic_inc_return(&autogroup_seq_nr); + ag->tg = tg; + tg->autogroup = ag; + + return ag; + +out_free: + kfree(ag); +out_fail: + if (printk_ratelimit()) { + printk(KERN_WARNING "autogroup_create: %s failure.\n", + ag ? "sched_create_group()" : "kmalloc()"); + } + + return autogroup_kref_get(&autogroup_default); +} + +static inline bool +task_wants_autogroup(struct task_struct *p, struct task_group *tg) +{ + if (tg != &root_task_group) + return false; + + if (p->sched_class != &fair_sched_class) + return false; + + /* + * We can only assume the task group can't go away on us if + * autogroup_move_group() can see us on ->thread_group list. + */ + if (p->flags & PF_EXITING) + return false; + + return true; +} + +static inline struct task_group * +autogroup_task_group(struct task_struct *p, struct task_group *tg) +{ + int enabled = ACCESS_ONCE(sysctl_sched_autogroup_enabled); + + if (enabled && task_wants_autogroup(p, tg)) + return p->signal->autogroup->tg; + + return tg; +} + +static void +autogroup_move_group(struct task_struct *p, struct autogroup *ag) +{ + struct autogroup *prev; + struct task_struct *t; + unsigned long flags; + + BUG_ON(!lock_task_sighand(p, &flags)); + + prev = p->signal->autogroup; + if (prev == ag) { + unlock_task_sighand(p, &flags); + return; + } + + p->signal->autogroup = autogroup_kref_get(ag); + + t = p; + do { + sched_move_task(t); + } while_each_thread(p, t); + + unlock_task_sighand(p, &flags); + autogroup_kref_put(prev); +} + +/* Allocates GFP_KERNEL, cannot be called under any spinlock */ +void sched_autogroup_create_attach(struct task_struct *p) +{ + struct autogroup *ag = autogroup_create(); + + autogroup_move_group(p, ag); + /* drop extra refrence added by autogroup_create() */ + autogroup_kref_put(ag); +} +EXPORT_SYMBOL(sched_autogroup_create_attach); + +/* Cannot be called under siglock. Currently has no users */ +void sched_autogroup_detach(struct task_struct *p) +{ + autogroup_move_group(p, &autogroup_default); +} +EXPORT_SYMBOL(sched_autogroup_detach); + +void sched_autogroup_fork(struct signal_struct *sig) +{ + struct task_struct *p = current; + + spin_lock_irq(&p->sighand->siglock); + sig->autogroup = autogroup_kref_get(p->signal->autogroup); + spin_unlock_irq(&p->sighand->siglock); +} + +void sched_autogroup_exit(struct signal_struct *sig) +{ + autogroup_kref_put(sig->autogroup); +} + +static int __init setup_autogroup(char *str) +{ + sysctl_sched_autogroup_enabled = 0; + + return 1; +} + +__setup("noautogroup", setup_autogroup); + +#ifdef CONFIG_PROC_FS + +/* Called with siglock held. */ +int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice) +{ + static unsigned long next = INITIAL_JIFFIES; + struct autogroup *ag; + int err; + + if (*nice < -20 || *nice > 19) + return -EINVAL; + + err = security_task_setnice(current, *nice); + if (err) + return err; + + if (*nice < 0 && !can_nice(current, *nice)) + return -EPERM; + + /* this is a heavy operation taking global locks.. */ + if (!capable(CAP_SYS_ADMIN) && time_before(jiffies, next)) + return -EAGAIN; + + next = HZ / 10 + jiffies; + ag = autogroup_kref_get(p->signal->autogroup); + + down_write(&ag->lock); + err = sched_group_set_shares(ag->tg, prio_to_weight[*nice + 20]); + if (!err) + ag->nice = *nice; + up_write(&ag->lock); + + autogroup_kref_put(ag); + + return err; +} + +void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m) +{ + struct autogroup *ag = autogroup_kref_get(p->signal->autogroup); + + down_read(&ag->lock); + seq_printf(m, "/autogroup-%ld nice %d\n", ag->id, ag->nice); + up_read(&ag->lock); + + autogroup_kref_put(ag); +} +#endif /* CONFIG_PROC_FS */ + +#ifdef CONFIG_SCHED_DEBUG +static inline 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 */ diff --git a/kernel/sched_autogroup.h b/kernel/sched_autogroup.h new file mode 100644 index 000000000000..5358e241cb20 --- /dev/null +++ b/kernel/sched_autogroup.h @@ -0,0 +1,32 @@ +#ifdef CONFIG_SCHED_AUTOGROUP + +struct autogroup { + struct kref kref; + struct task_group *tg; + struct rw_semaphore lock; + unsigned long id; + int nice; +}; + +static inline struct task_group * +autogroup_task_group(struct task_struct *p, struct task_group *tg); + +#else /* !CONFIG_SCHED_AUTOGROUP */ + +static inline void autogroup_init(struct task_struct *init_task) { } +static inline void autogroup_free(struct task_group *tg) { } + +static inline struct task_group * +autogroup_task_group(struct task_struct *p, struct task_group *tg) +{ + return tg; +} + +#ifdef CONFIG_SCHED_DEBUG +static inline int autogroup_path(struct task_group *tg, char *buf, int buflen) +{ + return 0; +} +#endif + +#endif /* CONFIG_SCHED_AUTOGROUP */ diff --git a/kernel/sched_debug.c b/kernel/sched_debug.c index e95b77414a99..1dfae3d014b5 100644 --- a/kernel/sched_debug.c +++ b/kernel/sched_debug.c @@ -54,8 +54,7 @@ static unsigned long nsec_low(unsigned long long nsec) #define SPLIT_NS(x) nsec_high(x), nsec_low(x) #ifdef CONFIG_FAIR_GROUP_SCHED -static void print_cfs_group_stats(struct seq_file *m, int cpu, - struct task_group *tg) +static void print_cfs_group_stats(struct seq_file *m, int cpu, struct task_group *tg) { struct sched_entity *se = tg->se[cpu]; if (!se) @@ -110,16 +109,6 @@ print_task(struct seq_file *m, struct rq *rq, struct task_struct *p) 0LL, 0LL, 0LL, 0L, 0LL, 0L, 0LL, 0L); #endif -#ifdef CONFIG_CGROUP_SCHED - { - char path[64]; - - rcu_read_lock(); - cgroup_path(task_group(p)->css.cgroup, path, sizeof(path)); - rcu_read_unlock(); - SEQ_printf(m, " %s", path); - } -#endif SEQ_printf(m, "\n"); } @@ -147,19 +136,6 @@ static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu) read_unlock_irqrestore(&tasklist_lock, flags); } -#if defined(CONFIG_CGROUP_SCHED) && \ - (defined(CONFIG_FAIR_GROUP_SCHED) || defined(CONFIG_RT_GROUP_SCHED)) -static void task_group_path(struct task_group *tg, char *buf, int buflen) -{ - /* may be NULL if the underlying cgroup isn't fully-created yet */ - if (!tg->css.cgroup) { - buf[0] = '\0'; - return; - } - cgroup_path(tg->css.cgroup, buf, buflen); -} -#endif - void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) { s64 MIN_vruntime = -1, min_vruntime, max_vruntime = -1, @@ -168,16 +144,7 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) struct sched_entity *last; unsigned long flags; -#if defined(CONFIG_CGROUP_SCHED) && defined(CONFIG_FAIR_GROUP_SCHED) - char path[128]; - struct task_group *tg = cfs_rq->tg; - - task_group_path(tg, path, sizeof(path)); - - SEQ_printf(m, "\ncfs_rq[%d]:%s\n", cpu, path); -#else SEQ_printf(m, "\ncfs_rq[%d]:\n", cpu); -#endif SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "exec_clock", SPLIT_NS(cfs_rq->exec_clock)); @@ -215,7 +182,7 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) SEQ_printf(m, " .%-30s: %ld\n", "load_contrib", cfs_rq->load_contribution); SEQ_printf(m, " .%-30s: %d\n", "load_tg", - atomic_read(&tg->load_weight)); + atomic_read(&cfs_rq->tg->load_weight)); #endif print_cfs_group_stats(m, cpu, cfs_rq->tg); @@ -224,17 +191,7 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq) { -#if defined(CONFIG_CGROUP_SCHED) && defined(CONFIG_RT_GROUP_SCHED) - char path[128]; - struct task_group *tg = rt_rq->tg; - - task_group_path(tg, path, sizeof(path)); - - SEQ_printf(m, "\nrt_rq[%d]:%s\n", cpu, path); -#else SEQ_printf(m, "\nrt_rq[%d]:\n", cpu); -#endif - #define P(x) \ SEQ_printf(m, " .%-30s: %Ld\n", #x, (long long)(rt_rq->x)) diff --git a/kernel/sys.c b/kernel/sys.c index 7f5a0cd296a9..2745dcdb6c6c 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1080,8 +1080,10 @@ SYSCALL_DEFINE0(setsid) err = session; out: write_unlock_irq(&tasklist_lock); - if (err > 0) + if (err > 0) { proc_sid_connector(group_leader); + sched_autogroup_create_attach(group_leader); + } return err; } diff --git a/kernel/sysctl.c b/kernel/sysctl.c index a00fdefd24ce..121e4fff03d1 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -370,6 +370,17 @@ static struct ctl_table kern_table[] = { .mode = 0644, .proc_handler = proc_dointvec, }, +#ifdef CONFIG_SCHED_AUTOGROUP + { + .procname = "sched_autogroup_enabled", + .data = &sysctl_sched_autogroup_enabled, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec, + .extra1 = &zero, + .extra2 = &one, + }, +#endif #ifdef CONFIG_PROVE_LOCKING { .procname = "prove_locking", -- cgit v1.2.3 From 806c09a7db457be3758e14b1f152761135d89af5 Mon Sep 17 00:00:00 2001 From: Dario Faggioli Date: Tue, 30 Nov 2010 19:51:33 +0100 Subject: sched: Make pushable_tasks CONFIG_SMP dependant As noted by Peter Zijlstra at https://lkml.org/lkml/2010/11/10/391 (while reviewing other stuff, though), tracking pushable tasks only makes sense on SMP systems. Signed-off-by: Dario Faggioli Acked-by: Steven Rostedt Acked-by: Gregory Haskins Signed-off-by: Peter Zijlstra LKML-Reference: <1291143093.2697.298.camel@Palantir> Signed-off-by: Ingo Molnar --- include/linux/init_task.h | 9 ++++++++- include/linux/sched.h | 2 ++ kernel/sched.c | 2 ++ 3 files changed, 12 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 1f8c06ce0fa6..6ed8812bfe2d 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -12,6 +12,13 @@ #include #include +#ifdef CONFIG_SMP +# define INIT_PUSHABLE_TASKS(tsk) \ + .pushable_tasks = PLIST_NODE_INIT(tsk.pushable_tasks, MAX_PRIO), +#else +# define INIT_PUSHABLE_TASKS(tsk) +#endif + extern struct files_struct init_files; extern struct fs_struct init_fs; @@ -137,7 +144,7 @@ extern struct cred init_cred; .nr_cpus_allowed = NR_CPUS, \ }, \ .tasks = LIST_HEAD_INIT(tsk.tasks), \ - .pushable_tasks = PLIST_NODE_INIT(tsk.pushable_tasks, MAX_PRIO), \ + INIT_PUSHABLE_TASKS(tsk) \ .ptraced = LIST_HEAD_INIT(tsk.ptraced), \ .ptrace_entry = LIST_HEAD_INIT(tsk.ptrace_entry), \ .real_parent = &tsk, \ diff --git a/include/linux/sched.h b/include/linux/sched.h index 9c2d46da486e..4f92a239c14d 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1240,7 +1240,9 @@ struct task_struct { #endif struct list_head tasks; +#ifdef CONFIG_SMP struct plist_node pushable_tasks; +#endif struct mm_struct *mm, *active_mm; #if defined(SPLIT_RSS_COUNTING) diff --git a/kernel/sched.c b/kernel/sched.c index b646dad4a40e..3925a1bbf5dd 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2595,7 +2595,9 @@ void sched_fork(struct task_struct *p, int clone_flags) /* Want to start with kernel preemption disabled. */ task_thread_info(p)->preempt_count = 1; #endif +#ifdef CONFIG_SMP plist_node_init(&p->pushable_tasks, MAX_PRIO); +#endif put_cpu(); } -- cgit v1.2.3 From 40dc11ffb35e8c4e8fa71092048e0f8de9db758c Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 26 Nov 2010 17:22:16 +0100 Subject: printk: Use this_cpu_{read|write} api on printk_pending __get_cpu_var() is a bit inefficient, lets use __this_cpu_read() and __this_cpu_write() to manipulate printk_pending. printk_needs_cpu(cpu) is called only for the current cpu : Use faster __this_cpu_read(). Remove the redundant unlikely on (cpu_is_offline(cpu)) test: # size kernel/printk.o* text data bss dec hex filename 9942 756 263488 274186 42f0a kernel/printk.o.new 9990 756 263488 274234 42f3a kernel/printk.o.old Signed-off-by: Eric Dumazet Cc: Heiko Carstens Cc: H. Peter Anvin Cc: Christoph Lameter Signed-off-by: Peter Zijlstra LKML-Reference: <1290788536.2855.237.camel@edumazet-laptop> Signed-off-by: Ingo Molnar --- kernel/printk.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index a23315dc4498..ab3ffc5b3b64 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1074,17 +1074,17 @@ static DEFINE_PER_CPU(int, printk_pending); void printk_tick(void) { - if (__get_cpu_var(printk_pending)) { - __get_cpu_var(printk_pending) = 0; + if (__this_cpu_read(printk_pending)) { + __this_cpu_write(printk_pending, 0); wake_up_interruptible(&log_wait); } } int printk_needs_cpu(int cpu) { - if (unlikely(cpu_is_offline(cpu))) + if (cpu_is_offline(cpu)) printk_tick(); - return per_cpu(printk_pending, cpu); + return __this_cpu_read(printk_pending); } void wake_up_klogd(void) -- cgit v1.2.3 From 998adc3dda59f811966b3ccb21eb223680b25ec4 Mon Sep 17 00:00:00 2001 From: John Stultz Date: Mon, 20 Sep 2010 19:19:17 -0700 Subject: hrtimers: Convert hrtimers to use timerlist infrastructure Converts the hrtimer code to use the new timerlist infrastructure Signed-off-by: John Stultz LKML Reference: <1290136329-18291-3-git-send-email-john.stultz@linaro.org> Reviewed-by: Thomas Gleixner CC: Alessandro Zummo CC: Thomas Gleixner CC: Richard Cochran --- include/linux/hrtimer.h | 32 +++++++++--------- kernel/hrtimer.c | 86 +++++++++++++++++------------------------------- kernel/time/timer_list.c | 8 ++--- 3 files changed, 49 insertions(+), 77 deletions(-) (limited to 'kernel') diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h index dd9954b79342..330586ffffbb 100644 --- a/include/linux/hrtimer.h +++ b/include/linux/hrtimer.h @@ -22,7 +22,7 @@ #include #include #include - +#include struct hrtimer_clock_base; struct hrtimer_cpu_base; @@ -79,8 +79,8 @@ enum hrtimer_restart { /** * struct hrtimer - the basic hrtimer structure - * @node: red black tree node for time ordered insertion - * @_expires: the absolute expiry time in the hrtimers internal + * @node: timerqueue node, which also manages node.expires, + * the absolute expiry time in the hrtimers internal * representation. The time is related to the clock on * which the timer is based. Is setup by adding * slack to the _softexpires value. For non range timers @@ -101,8 +101,7 @@ enum hrtimer_restart { * The hrtimer structure must be initialized by hrtimer_init() */ struct hrtimer { - struct rb_node node; - ktime_t _expires; + struct timerqueue_node node; ktime_t _softexpires; enum hrtimer_restart (*function)(struct hrtimer *); struct hrtimer_clock_base *base; @@ -141,8 +140,7 @@ struct hrtimer_sleeper { struct hrtimer_clock_base { struct hrtimer_cpu_base *cpu_base; clockid_t index; - struct rb_root active; - struct rb_node *first; + struct timerqueue_head active; ktime_t resolution; ktime_t (*get_time)(void); ktime_t softirq_time; @@ -183,43 +181,43 @@ struct hrtimer_cpu_base { static inline void hrtimer_set_expires(struct hrtimer *timer, ktime_t time) { - timer->_expires = time; + timer->node.expires = time; timer->_softexpires = time; } static inline void hrtimer_set_expires_range(struct hrtimer *timer, ktime_t time, ktime_t delta) { timer->_softexpires = time; - timer->_expires = ktime_add_safe(time, delta); + timer->node.expires = ktime_add_safe(time, delta); } static inline void hrtimer_set_expires_range_ns(struct hrtimer *timer, ktime_t time, unsigned long delta) { timer->_softexpires = time; - timer->_expires = ktime_add_safe(time, ns_to_ktime(delta)); + timer->node.expires = ktime_add_safe(time, ns_to_ktime(delta)); } static inline void hrtimer_set_expires_tv64(struct hrtimer *timer, s64 tv64) { - timer->_expires.tv64 = tv64; + timer->node.expires.tv64 = tv64; timer->_softexpires.tv64 = tv64; } static inline void hrtimer_add_expires(struct hrtimer *timer, ktime_t time) { - timer->_expires = ktime_add_safe(timer->_expires, time); + timer->node.expires = ktime_add_safe(timer->node.expires, time); timer->_softexpires = ktime_add_safe(timer->_softexpires, time); } static inline void hrtimer_add_expires_ns(struct hrtimer *timer, u64 ns) { - timer->_expires = ktime_add_ns(timer->_expires, ns); + timer->node.expires = ktime_add_ns(timer->node.expires, ns); timer->_softexpires = ktime_add_ns(timer->_softexpires, ns); } static inline ktime_t hrtimer_get_expires(const struct hrtimer *timer) { - return timer->_expires; + return timer->node.expires; } static inline ktime_t hrtimer_get_softexpires(const struct hrtimer *timer) @@ -229,7 +227,7 @@ static inline ktime_t hrtimer_get_softexpires(const struct hrtimer *timer) static inline s64 hrtimer_get_expires_tv64(const struct hrtimer *timer) { - return timer->_expires.tv64; + return timer->node.expires.tv64; } static inline s64 hrtimer_get_softexpires_tv64(const struct hrtimer *timer) { @@ -238,12 +236,12 @@ static inline s64 hrtimer_get_softexpires_tv64(const struct hrtimer *timer) static inline s64 hrtimer_get_expires_ns(const struct hrtimer *timer) { - return ktime_to_ns(timer->_expires); + return ktime_to_ns(timer->node.expires); } static inline ktime_t hrtimer_expires_remaining(const struct hrtimer *timer) { - return ktime_sub(timer->_expires, timer->base->get_time()); + return ktime_sub(timer->node.expires, timer->base->get_time()); } #ifdef CONFIG_HIGH_RES_TIMERS diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c index ce669174f355..93976ad42f5a 100644 --- a/kernel/hrtimer.c +++ b/kernel/hrtimer.c @@ -516,10 +516,13 @@ hrtimer_force_reprogram(struct hrtimer_cpu_base *cpu_base, int skip_equal) for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++, base++) { struct hrtimer *timer; + struct timerqueue_node *next; - if (!base->first) + next = timerqueue_getnext(&base->active); + if (!next) continue; - timer = rb_entry(base->first, struct hrtimer, node); + timer = container_of(next, struct hrtimer, node); + expires = ktime_sub(hrtimer_get_expires(timer), base->offset); /* * clock_was_set() has changed base->offset so the @@ -840,48 +843,17 @@ EXPORT_SYMBOL_GPL(hrtimer_forward); static int enqueue_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base) { - struct rb_node **link = &base->active.rb_node; - struct rb_node *parent = NULL; - struct hrtimer *entry; - int leftmost = 1; - debug_activate(timer); - /* - * Find the right place in the rbtree: - */ - while (*link) { - parent = *link; - entry = rb_entry(parent, struct hrtimer, node); - /* - * We dont care about collisions. Nodes with - * the same expiry time stay together. - */ - if (hrtimer_get_expires_tv64(timer) < - hrtimer_get_expires_tv64(entry)) { - link = &(*link)->rb_left; - } else { - link = &(*link)->rb_right; - leftmost = 0; - } - } + timerqueue_add(&base->active, &timer->node); - /* - * Insert the timer to the rbtree and check whether it - * replaces the first pending timer - */ - if (leftmost) - base->first = &timer->node; - - rb_link_node(&timer->node, parent, link); - rb_insert_color(&timer->node, &base->active); /* * HRTIMER_STATE_ENQUEUED is or'ed to the current state to preserve the * state of a possibly running callback. */ timer->state |= HRTIMER_STATE_ENQUEUED; - return leftmost; + return (&timer->node == base->active.next); } /* @@ -901,12 +873,7 @@ static void __remove_hrtimer(struct hrtimer *timer, if (!(timer->state & HRTIMER_STATE_ENQUEUED)) goto out; - /* - * Remove the timer from the rbtree and replace the first - * entry pointer if necessary. - */ - if (base->first == &timer->node) { - base->first = rb_next(&timer->node); + if (&timer->node == timerqueue_getnext(&base->active)) { #ifdef CONFIG_HIGH_RES_TIMERS /* Reprogram the clock event device. if enabled */ if (reprogram && hrtimer_hres_active()) { @@ -919,7 +886,7 @@ static void __remove_hrtimer(struct hrtimer *timer, } #endif } - rb_erase(&timer->node, &base->active); + timerqueue_del(&base->active, &timer->node); out: timer->state = newstate; } @@ -1123,11 +1090,13 @@ ktime_t hrtimer_get_next_event(void) if (!hrtimer_hres_active()) { for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++, base++) { struct hrtimer *timer; + struct timerqueue_node *next; - if (!base->first) + next = timerqueue_getnext(&base->active); + if (!next) continue; - timer = rb_entry(base->first, struct hrtimer, node); + timer = container_of(next, struct hrtimer, node); delta.tv64 = hrtimer_get_expires_tv64(timer); delta = ktime_sub(delta, base->get_time()); if (delta.tv64 < mindelta.tv64) @@ -1157,6 +1126,7 @@ static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id, timer->base = &cpu_base->clock_base[clock_id]; hrtimer_init_timer_hres(timer); + timerqueue_init(&timer->node); #ifdef CONFIG_TIMER_STATS timer->start_site = NULL; @@ -1270,14 +1240,14 @@ retry: for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++) { ktime_t basenow; - struct rb_node *node; + struct timerqueue_node *node; basenow = ktime_add(now, base->offset); - while ((node = base->first)) { + while ((node = timerqueue_getnext(&base->active))) { struct hrtimer *timer; - timer = rb_entry(node, struct hrtimer, node); + timer = container_of(node, struct hrtimer, node); /* * The immediate goal for using the softexpires is @@ -1433,7 +1403,7 @@ void hrtimer_run_pending(void) */ void hrtimer_run_queues(void) { - struct rb_node *node; + struct timerqueue_node *node; struct hrtimer_cpu_base *cpu_base = &__get_cpu_var(hrtimer_bases); struct hrtimer_clock_base *base; int index, gettime = 1; @@ -1442,9 +1412,11 @@ void hrtimer_run_queues(void) return; for (index = 0; index < HRTIMER_MAX_CLOCK_BASES; index++) { - base = &cpu_base->clock_base[index]; + struct timerqueue_node *next; - if (!base->first) + base = &cpu_base->clock_base[index]; + next = timerqueue_getnext(&base->active); + if (!next) continue; if (gettime) { @@ -1454,10 +1426,10 @@ void hrtimer_run_queues(void) raw_spin_lock(&cpu_base->lock); - while ((node = base->first)) { + while ((node = next)) { struct hrtimer *timer; - timer = rb_entry(node, struct hrtimer, node); + timer = container_of(node, struct hrtimer, node); if (base->softirq_time.tv64 <= hrtimer_get_expires_tv64(timer)) break; @@ -1622,8 +1594,10 @@ static void __cpuinit init_hrtimers_cpu(int cpu) raw_spin_lock_init(&cpu_base->lock); - for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++) + for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++) { cpu_base->clock_base[i].cpu_base = cpu_base; + timerqueue_init_head(&cpu_base->clock_base[i].active); + } hrtimer_init_hres(cpu_base); } @@ -1634,10 +1608,10 @@ static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base, struct hrtimer_clock_base *new_base) { struct hrtimer *timer; - struct rb_node *node; + struct timerqueue_node *node; - while ((node = rb_first(&old_base->active))) { - timer = rb_entry(node, struct hrtimer, node); + while ((node = timerqueue_getnext(&old_base->active))) { + timer = container_of(node, struct hrtimer, node); BUG_ON(hrtimer_callback_running(timer)); debug_deactivate(timer); diff --git a/kernel/time/timer_list.c b/kernel/time/timer_list.c index ab8f5e33fa92..32a19f9397fc 100644 --- a/kernel/time/timer_list.c +++ b/kernel/time/timer_list.c @@ -79,26 +79,26 @@ print_active_timers(struct seq_file *m, struct hrtimer_clock_base *base, { struct hrtimer *timer, tmp; unsigned long next = 0, i; - struct rb_node *curr; + struct timerqueue_node *curr; unsigned long flags; next_one: i = 0; raw_spin_lock_irqsave(&base->cpu_base->lock, flags); - curr = base->first; + curr = timerqueue_getnext(&base->active); /* * Crude but we have to do this O(N*N) thing, because * we have to unlock the base when printing: */ while (curr && i < next) { - curr = rb_next(curr); + curr = timerqueue_iterate_next(curr); i++; } if (curr) { - timer = rb_entry(curr, struct hrtimer, node); + timer = container_of(curr, struct hrtimer, node); tmp = *timer; raw_spin_unlock_irqrestore(&base->cpu_base->lock, flags); -- cgit v1.2.3 From b007c389d3e09b823eccda1503390fa2a9adca0d Mon Sep 17 00:00:00 2001 From: John Stultz Date: Fri, 10 Dec 2010 22:19:53 -0800 Subject: hrtimer: fix timerqueue conversion flub In converting the hrtimers to timerqueue, I missed a spot in hrtimer_run_queues where we loop running timers. We end up not pulling the new next value out and instead just use the last next value, causing boot time hangs in some cases. The proper fix is to pull timerqueue_getnext each iteration instead of using a local next value. Reported-by: Ingo Molnar Signed-off-by: John Stultz --- kernel/hrtimer.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c index 93976ad42f5a..7a7a2061c24d 100644 --- a/kernel/hrtimer.c +++ b/kernel/hrtimer.c @@ -1412,11 +1412,8 @@ void hrtimer_run_queues(void) return; for (index = 0; index < HRTIMER_MAX_CLOCK_BASES; index++) { - struct timerqueue_node *next; - base = &cpu_base->clock_base[index]; - next = timerqueue_getnext(&base->active); - if (!next) + if (!timerqueue_getnext(&base->active)) continue; if (gettime) { @@ -1426,7 +1423,7 @@ void hrtimer_run_queues(void) raw_spin_lock(&cpu_base->lock); - while ((node = next)) { + while ((node = timerqueue_getnext(&base->active))) { struct hrtimer *timer; timer = container_of(node, struct hrtimer, node); -- cgit v1.2.3 From 7496351ad87e61e96b49dd7b43c6534e3401f566 Mon Sep 17 00:00:00 2001 From: Christoph Lameter Date: Tue, 30 Nov 2010 14:05:53 -0600 Subject: timers: Use this_cpu_read Eric asked for this. [tglx: Because it generates faster code according to Erics ] Signed-off-by: Christoph Lameter Cc: Pekka Enberg Cc: Eric Dumazet Cc: Mathieu Desnoyers Cc: Tejun Heo Cc: linux-mm@kvack.org LKML-Reference: Signed-off-by: Thomas Gleixner --- kernel/timer.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/timer.c b/kernel/timer.c index 483e54ba5c93..beb97fd11ac2 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -1227,7 +1227,7 @@ static unsigned long cmp_next_hrtimer_event(unsigned long now, */ unsigned long get_next_timer_interrupt(unsigned long now) { - struct tvec_base *base = __get_cpu_var(tvec_bases); + struct tvec_base *base = __this_cpu_read(tvec_bases); unsigned long expires; spin_lock(&base->lock); @@ -1267,7 +1267,7 @@ void update_process_times(int user_tick) */ static void run_timer_softirq(struct softirq_action *h) { - struct tvec_base *base = __get_cpu_var(tvec_bases); + struct tvec_base *base = __this_cpu_read(tvec_bases); hrtimer_run_pending(); -- cgit v1.2.3 From 3fb82d56ad003e804923185316236f26b30dfdd5 Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Tue, 23 Nov 2010 16:11:40 -0800 Subject: x86, suspend: Avoid unnecessary smp alternatives switch during suspend/resume During suspend, we disable all the non boot cpus. And during resume we bring them all back again. So no need to do alternatives_smp_switch() in between. On my core 2 based laptop, this speeds up the suspend path by 15msec and the resume path by 5 msec (suspend/resume speed up differences can be attributed to the different P-states that the cpu is in during suspend/resume). Signed-off-by: Suresh Siddha LKML-Reference: <1290557500.4946.8.camel@sbsiddha-MOBL3.sc.intel.com> Cc: Rafael J. Wysocki Signed-off-by: H. Peter Anvin --- arch/x86/include/asm/alternative.h | 1 + arch/x86/kernel/alternative.c | 3 ++- arch/x86/kernel/smpboot.c | 14 ++++++++++++++ kernel/cpu.c | 11 +++++++++++ 4 files changed, 28 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/arch/x86/include/asm/alternative.h b/arch/x86/include/asm/alternative.h index 76561d20ea2f..01171f6c2c37 100644 --- a/arch/x86/include/asm/alternative.h +++ b/arch/x86/include/asm/alternative.h @@ -66,6 +66,7 @@ extern void alternatives_smp_module_add(struct module *mod, char *name, extern void alternatives_smp_module_del(struct module *mod); extern void alternatives_smp_switch(int smp); extern int alternatives_text_reserved(void *start, void *end); +extern bool skip_smp_alternatives; #else static inline void alternatives_smp_module_add(struct module *mod, char *name, void *locks, void *locks_end, diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c index 5079f24c955a..9f98eb400fef 100644 --- a/arch/x86/kernel/alternative.c +++ b/arch/x86/kernel/alternative.c @@ -353,6 +353,7 @@ void __init_or_module alternatives_smp_module_del(struct module *mod) mutex_unlock(&smp_alt); } +bool skip_smp_alternatives; void alternatives_smp_switch(int smp) { struct smp_alt_module *mod; @@ -368,7 +369,7 @@ void alternatives_smp_switch(int smp) printk("lockdep: fixing up alternatives.\n"); #endif - if (noreplace_smp || smp_alt_once) + if (noreplace_smp || smp_alt_once || skip_smp_alternatives) return; BUG_ON(!smp && (num_online_cpus() > 1)); diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 083e99d1b7df..837c81e99edf 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1166,6 +1166,20 @@ out: preempt_enable(); } +void arch_disable_nonboot_cpus_begin(void) +{ + /* + * Avoid the smp alternatives switch during the disable_nonboot_cpus(). + * In the suspend path, we will be back in the SMP mode shortly anyways. + */ + skip_smp_alternatives = true; +} + +void arch_disable_nonboot_cpus_end(void) +{ + skip_smp_alternatives = false; +} + void arch_enable_nonboot_cpus_begin(void) { set_mtrr_aps_delayed_init(); diff --git a/kernel/cpu.c b/kernel/cpu.c index f6e726f18491..8ccc182069ec 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -386,6 +386,14 @@ out: #ifdef CONFIG_PM_SLEEP_SMP static cpumask_var_t frozen_cpus; +void __weak arch_disable_nonboot_cpus_begin(void) +{ +} + +void __weak arch_disable_nonboot_cpus_end(void) +{ +} + int disable_nonboot_cpus(void) { int cpu, first_cpu, error = 0; @@ -397,6 +405,7 @@ int disable_nonboot_cpus(void) * with the userspace trying to use the CPU hotplug at the same time */ cpumask_clear(frozen_cpus); + arch_disable_nonboot_cpus_begin(); printk("Disabling non-boot CPUs ...\n"); for_each_online_cpu(cpu) { @@ -412,6 +421,8 @@ int disable_nonboot_cpus(void) } } + arch_disable_nonboot_cpus_end(); + if (!error) { BUG_ON(num_online_cpus() > 1); /* Make sure the CPUs won't be enabled by someone else */ -- cgit v1.2.3 From e27fc9641e8ddc8146f8e01f06e5eba2469698de Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 22 Nov 2010 21:36:11 -0800 Subject: rcu: increase synchronize_sched_expedited() batching The fix in commit #6a0cc49 requires more than three concurrent instances of synchronize_sched_expedited() before batching is possible. This patch uses a ticket-counter-like approach that is also not unrelated to Lai Jiangshan's Ring RCU to allow sharing of expedited grace periods even when there are only two concurrent instances of synchronize_sched_expedited(). This commit builds on Tejun's original posting, which may be found at http://lkml.org/lkml/2010/11/9/204, adding memory barriers, avoiding overflow of signed integers (other than via atomic_t), and fixing the detection of batching. Signed-off-by: Tejun Heo Signed-off-by: Paul E. McKenney --- include/linux/rcupdate.h | 2 ++ kernel/rcutree_plugin.h | 82 ++++++++++++++++++++++++++++++++++++------------ 2 files changed, 64 insertions(+), 20 deletions(-) (limited to 'kernel') diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h index 49e8e16308e1..af5614856285 100644 --- a/include/linux/rcupdate.h +++ b/include/linux/rcupdate.h @@ -47,6 +47,8 @@ extern int rcutorture_runnable; /* for sysctl */ #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */ +#define UINT_CMP_GE(a, b) (UINT_MAX / 2 >= (a) - (b)) +#define UINT_CMP_LT(a, b) (UINT_MAX / 2 < (a) - (b)) #define ULONG_CMP_GE(a, b) (ULONG_MAX / 2 >= (a) - (b)) #define ULONG_CMP_LT(a, b) (ULONG_MAX / 2 < (a) - (b)) diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index c22c4ef2a0d0..a3638710dc67 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -1025,7 +1025,8 @@ EXPORT_SYMBOL_GPL(synchronize_sched_expedited); #else /* #ifndef CONFIG_SMP */ -static atomic_t synchronize_sched_expedited_count = ATOMIC_INIT(0); +static atomic_t sync_sched_expedited_started = ATOMIC_INIT(0); +static atomic_t sync_sched_expedited_done = ATOMIC_INIT(0); static int synchronize_sched_expedited_cpu_stop(void *data) { @@ -1041,8 +1042,6 @@ static int synchronize_sched_expedited_cpu_stop(void *data) * robustness against future implementation changes. */ smp_mb(); /* See above comment block. */ - if (cpumask_first(cpu_online_mask) == smp_processor_id()) - atomic_inc(&synchronize_sched_expedited_count); return 0; } @@ -1056,43 +1055,86 @@ static int synchronize_sched_expedited_cpu_stop(void *data) * lock that is acquired by a CPU-hotplug notifier. Failing to * observe this restriction will result in deadlock. * - * The synchronize_sched_expedited_cpu_stop() function is called - * in stop-CPU context, but in order to keep overhead down to a dull - * roar, we don't force this function to wait for its counterparts - * on other CPUs. One instance of this function will increment the - * synchronize_sched_expedited_count variable per call to - * try_stop_cpus(), but there is no guarantee what order this instance - * will occur in. The worst case is that it is last on one call - * to try_stop_cpus(), and the first on the next call. This means - * that piggybacking requires that synchronize_sched_expedited_count - * be incremented by 3: this guarantees that the piggybacking - * task has waited through an entire cycle of context switches, - * even in the worst case. + * This implementation can be thought of as an application of ticket + * locking to RCU, with sync_sched_expedited_started and + * sync_sched_expedited_done taking on the roles of the halves + * of the ticket-lock word. Each task atomically increments + * sync_sched_expedited_started upon entry, snapshotting the old value, + * then attempts to stop all the CPUs. If this succeeds, then each + * CPU will have executed a context switch, resulting in an RCU-sched + * grace period. We are then done, so we use atomic_cmpxchg() to + * update sync_sched_expedited_done to match our snapshot -- but + * only if someone else has not already advanced past our snapshot. + * + * On the other hand, if try_stop_cpus() fails, we check the value + * of sync_sched_expedited_done. If it has advanced past our + * initial snapshot, then someone else must have forced a grace period + * some time after we took our snapshot. In this case, our work is + * done for us, and we can simply return. Otherwise, we try again, + * but keep our initial snapshot for purposes of checking for someone + * doing our work for us. + * + * If we fail too many times in a row, we fall back to synchronize_sched(). */ void synchronize_sched_expedited(void) { - int snap, trycount = 0; + int firstsnap, s, snap, trycount = 0; - smp_mb(); /* ensure prior mod happens before capturing snap. */ - snap = atomic_read(&synchronize_sched_expedited_count) + 2; + /* Note that atomic_inc_return() implies full memory barrier. */ + firstsnap = snap = atomic_inc_return(&sync_sched_expedited_started); get_online_cpus(); + + /* + * Each pass through the following loop attempts to force a + * context switch on each CPU. + */ while (try_stop_cpus(cpu_online_mask, synchronize_sched_expedited_cpu_stop, NULL) == -EAGAIN) { put_online_cpus(); + + /* No joy, try again later. Or just synchronize_sched(). */ if (trycount++ < 10) udelay(trycount * num_online_cpus()); else { synchronize_sched(); return; } - if (atomic_read(&synchronize_sched_expedited_count) - snap > 0) { + + /* Check to see if someone else did our work for us. */ + s = atomic_read(&sync_sched_expedited_done); + if (UINT_CMP_GE((unsigned)s, (unsigned)firstsnap)) { smp_mb(); /* ensure test happens before caller kfree */ return; } + + /* + * Refetching sync_sched_expedited_started allows later + * callers to piggyback on our grace period. We subtract + * 1 to get the same token that the last incrementer got. + * We retry after they started, so our grace period works + * for them, and they started after our first try, so their + * grace period works for us. + */ get_online_cpus(); + snap = atomic_read(&sync_sched_expedited_started) - 1; + smp_mb(); /* ensure read is before try_stop_cpus(). */ } - smp_mb__after_atomic_inc(); /* ensure post-GP actions seen after GP. */ + + /* + * Everyone up to our most recent fetch is covered by our grace + * period. Update the counter, but only if our work is still + * relevant -- which it won't be if someone who started later + * than we did beat us to the punch. + */ + do { + s = atomic_read(&sync_sched_expedited_done); + if (UINT_CMP_GE((unsigned)s, (unsigned)snap)) { + smp_mb(); /* ensure test happens before caller kfree */ + break; + } + } while (atomic_cmpxchg(&sync_sched_expedited_done, s, snap) != s); + put_online_cpus(); } EXPORT_SYMBOL_GPL(synchronize_sched_expedited); -- cgit v1.2.3 From 20377f32dcb77941d450728da18cce5b1a7faec5 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 10 Dec 2010 22:11:10 +0100 Subject: rcu: Stop chasing QS if another CPU did it for us When a CPU is idle and others CPUs handled its extended quiescent state to complete grace periods on its behalf, it will catch up with completed grace periods numbers when it wakes up. But at this point there might be no more grace period to complete, but still the woken CPU always keeps its stale qs_pending value and will then continue to chase quiescent states even if its not needed anymore. This results in clusters of spurious softirqs until a new real grace period is started. Because if we continue to chase quiescent states but we have completed every grace periods, rcu_report_qs_rdp() is puzzled and makes that state run into infinite loops. As suggested by Lai Jiangshan, just reset qs_pending if someone completed every grace periods on our behalf. Suggested-by: Lai Jiangshan Signed-off-by: Frederic Weisbecker Cc: Paul E. McKenney Cc: Ingo Molnar Cc: Thomas Gleixner Cc: Peter Zijlstra Cc: Steven Rostedt Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 120820ffc657..916f42b39f1e 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -678,6 +678,14 @@ __rcu_process_gp_end(struct rcu_state *rsp, struct rcu_node *rnp, struct rcu_dat /* Remember that we saw this grace-period completion. */ rdp->completed = rnp->completed; + + /* + * If another CPU handled our extended quiescent states and + * we have no more grace period to complete yet, then stop + * chasing quiescent states. + */ + if (rdp->completed == rnp->gpnum) + rdp->qs_pending = 0; } } -- cgit v1.2.3 From 5ff8e6f0535fe730e921ca347bc38dcb9e01791a Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 10 Dec 2010 22:11:11 +0100 Subject: rcu: Keep gpnum and completed fields synchronized When a CPU that was in an extended quiescent state wakes up and catches up with grace periods that remote CPUs completed on its behalf, we update the completed field but not the gpnum that keeps a stale value of a backward grace period ID. Later, note_new_gpnum() will interpret the shift between the local CPU and the node grace period ID as some new grace period to handle and will then start to hunt quiescent state. But if every grace periods have already been completed, this interpretation becomes broken. And we'll be stuck in clusters of spurious softirqs because rcu_report_qs_rdp() will make this broken state run into infinite loop. The solution, as suggested by Lai Jiangshan, is to ensure that the gpnum and completed fields are well synchronized when we catch up with completed grace periods on their behalf by other cpus. This way we won't start noting spurious new grace periods. Suggested-by: Lai Jiangshan Signed-off-by: Frederic Weisbecker Cc: Paul E. McKenney Cc: Ingo Molnar Cc: Thomas Gleixner Cc: Peter Zijlstra Cc: Steven Rostedt --- kernel/rcutree.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 916f42b39f1e..8105271fc10e 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -679,6 +679,15 @@ __rcu_process_gp_end(struct rcu_state *rsp, struct rcu_node *rnp, struct rcu_dat /* Remember that we saw this grace-period completion. */ rdp->completed = rnp->completed; + /* + * If we were in an extended quiescent state, we may have + * missed some grace periods that others CPUs took care on + * our behalf. Catch up with this state to avoid noting + * spurious new grace periods. + */ + if (rdp->completed > rdp->gpnum) + rdp->gpnum = rdp->completed; + /* * If another CPU handled our extended quiescent states and * we have no more grace period to complete yet, then stop -- cgit v1.2.3 From 121dfc4b3eba9e2f3c42d35205a3510cc65b9931 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Fri, 10 Dec 2010 15:02:47 -0800 Subject: rcu: fine-tune grace-period begin/end checks Use the CPU's bit in rnp->qsmask to determine whether or not the CPU should try to report a quiescent state. Handle overflow in the check for rdp->gpnum having fallen behind. Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 27 ++++++++++++++++++--------- 1 file changed, 18 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 8105271fc10e..c39ec5b4ae82 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -617,9 +617,17 @@ static void __init check_cpu_stall_init(void) static void __note_new_gpnum(struct rcu_state *rsp, struct rcu_node *rnp, struct rcu_data *rdp) { if (rdp->gpnum != rnp->gpnum) { - rdp->qs_pending = 1; - rdp->passed_quiesc = 0; + /* + * If the current grace period is waiting for this CPU, + * set up to detect a quiescent state, otherwise don't + * go looking for one. + */ rdp->gpnum = rnp->gpnum; + if (rnp->qsmask & rdp->grpmask) { + rdp->qs_pending = 1; + rdp->passed_quiesc = 0; + } else + rdp->qs_pending = 0; } } @@ -681,19 +689,20 @@ __rcu_process_gp_end(struct rcu_state *rsp, struct rcu_node *rnp, struct rcu_dat /* * If we were in an extended quiescent state, we may have - * missed some grace periods that others CPUs took care on + * missed some grace periods that others CPUs handled on * our behalf. Catch up with this state to avoid noting - * spurious new grace periods. + * spurious new grace periods. If another grace period + * has started, then rnp->gpnum will have advanced, so + * we will detect this later on. */ - if (rdp->completed > rdp->gpnum) + if (ULONG_CMP_LT(rdp->gpnum, rdp->completed)) rdp->gpnum = rdp->completed; /* - * If another CPU handled our extended quiescent states and - * we have no more grace period to complete yet, then stop - * chasing quiescent states. + * If RCU does not need a quiescent state from this CPU, + * then make sure that this CPU doesn't go looking for one. */ - if (rdp->completed == rnp->gpnum) + if ((rnp->qsmask & rdp->grpmask) == 0) rdp->qs_pending = 0; } } -- cgit v1.2.3 From 0209f6490b030f35349a2bb71294f3fd75b0f36d Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Tue, 14 Dec 2010 16:07:52 -0800 Subject: rcu: limit rcu_node leaf-level fanout Some recent benchmarks have indicated possible lock contention on the leaf-level rcu_node locks. This commit therefore limits the number of CPUs per leaf-level rcu_node structure to 16, in other words, there can be at most 16 rcu_data structures fanning into a given rcu_node structure. Prior to this, the limit was 32 on 32-bit systems and 64 on 64-bit systems. Note that the fanout of non-leaf rcu_node structures is unchanged. The organization of accesses to the rcu_node tree is such that references to non-leaf rcu_node structures are much less frequent than to the leaf structures. Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 3 ++- kernel/rcutree.h | 45 +++++++++++++++++++++++++-------------------- 2 files changed, 27 insertions(+), 21 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index c39ec5b4ae82..01c8ad33c510 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -1869,8 +1869,9 @@ static void __init rcu_init_levelspread(struct rcu_state *rsp) { int i; - for (i = NUM_RCU_LVLS - 1; i >= 0; i--) + for (i = NUM_RCU_LVLS - 1; i > 0; i--) rsp->levelspread[i] = CONFIG_RCU_FANOUT; + rsp->levelspread[0] = RCU_FANOUT_LEAF; } #else /* #ifdef CONFIG_RCU_FANOUT_EXACT */ static void __init rcu_init_levelspread(struct rcu_state *rsp) diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 1a54be2a902f..e8f057e44e3e 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -31,46 +31,51 @@ /* * Define shape of hierarchy based on NR_CPUS and CONFIG_RCU_FANOUT. * In theory, it should be possible to add more levels straightforwardly. - * In practice, this has not been tested, so there is probably some - * bug somewhere. + * In practice, this did work well going from three levels to four. + * Of course, your mileage may vary. */ #define MAX_RCU_LVLS 4 -#define RCU_FANOUT (CONFIG_RCU_FANOUT) -#define RCU_FANOUT_SQ (RCU_FANOUT * RCU_FANOUT) -#define RCU_FANOUT_CUBE (RCU_FANOUT_SQ * RCU_FANOUT) -#define RCU_FANOUT_FOURTH (RCU_FANOUT_CUBE * RCU_FANOUT) - -#if NR_CPUS <= RCU_FANOUT +#if CONFIG_RCU_FANOUT > 16 +#define RCU_FANOUT_LEAF 16 +#else /* #if CONFIG_RCU_FANOUT > 16 */ +#define RCU_FANOUT_LEAF (CONFIG_RCU_FANOUT) +#endif /* #else #if CONFIG_RCU_FANOUT > 16 */ +#define RCU_FANOUT_1 (RCU_FANOUT_LEAF) +#define RCU_FANOUT_2 (RCU_FANOUT_1 * CONFIG_RCU_FANOUT) +#define RCU_FANOUT_3 (RCU_FANOUT_2 * CONFIG_RCU_FANOUT) +#define RCU_FANOUT_4 (RCU_FANOUT_3 * CONFIG_RCU_FANOUT) + +#if NR_CPUS <= RCU_FANOUT_1 # define NUM_RCU_LVLS 1 # define NUM_RCU_LVL_0 1 # define NUM_RCU_LVL_1 (NR_CPUS) # define NUM_RCU_LVL_2 0 # define NUM_RCU_LVL_3 0 # define NUM_RCU_LVL_4 0 -#elif NR_CPUS <= RCU_FANOUT_SQ +#elif NR_CPUS <= RCU_FANOUT_2 # define NUM_RCU_LVLS 2 # define NUM_RCU_LVL_0 1 -# define NUM_RCU_LVL_1 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT) +# define NUM_RCU_LVL_1 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_1) # define NUM_RCU_LVL_2 (NR_CPUS) # define NUM_RCU_LVL_3 0 # define NUM_RCU_LVL_4 0 -#elif NR_CPUS <= RCU_FANOUT_CUBE +#elif NR_CPUS <= RCU_FANOUT_3 # define NUM_RCU_LVLS 3 # define NUM_RCU_LVL_0 1 -# define NUM_RCU_LVL_1 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_SQ) -# define NUM_RCU_LVL_2 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT) -# define NUM_RCU_LVL_3 NR_CPUS +# define NUM_RCU_LVL_1 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_2) +# define NUM_RCU_LVL_2 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_1) +# define NUM_RCU_LVL_3 (NR_CPUS) # define NUM_RCU_LVL_4 0 -#elif NR_CPUS <= RCU_FANOUT_FOURTH +#elif NR_CPUS <= RCU_FANOUT_4 # define NUM_RCU_LVLS 4 # define NUM_RCU_LVL_0 1 -# define NUM_RCU_LVL_1 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_CUBE) -# define NUM_RCU_LVL_2 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_SQ) -# define NUM_RCU_LVL_3 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT) -# define NUM_RCU_LVL_4 NR_CPUS +# define NUM_RCU_LVL_1 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_3) +# define NUM_RCU_LVL_2 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_2) +# define NUM_RCU_LVL_3 DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_1) +# define NUM_RCU_LVL_4 (NR_CPUS) #else # error "CONFIG_RCU_FANOUT insufficient for NR_CPUS" -#endif /* #if (NR_CPUS) <= RCU_FANOUT */ +#endif /* #if (NR_CPUS) <= RCU_FANOUT_1 */ #define RCU_SUM (NUM_RCU_LVL_0 + NUM_RCU_LVL_1 + NUM_RCU_LVL_2 + NUM_RCU_LVL_3 + NUM_RCU_LVL_4) #define NUM_RCU_NODES (RCU_SUM - NR_CPUS) -- cgit v1.2.3 From b52573d2796274f7f31cfeff7185c320adcd4f12 Mon Sep 17 00:00:00 2001 From: Paul E. McKenney Date: Tue, 14 Dec 2010 17:36:02 -0800 Subject: rcu: reduce __call_rcu()-induced contention on rcu_node structures When the current __call_rcu() function was written, the expedited APIs did not exist. The __call_rcu() implementation therefore went to great lengths to detect the end of old grace periods and to start new ones, all in the name of reducing grace-period latency. Now the expedited APIs do exist, and the usage of __call_rcu() has increased considerably. This commit therefore causes __call_rcu() to avoid worrying about grace periods unless there are a large number of RCU callbacks stacked up on the current CPU. Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 38 +++++++++++++++++++++----------------- 1 file changed, 21 insertions(+), 17 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 01c8ad33c510..d0ddfea6579d 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -1435,22 +1435,11 @@ __call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu), */ local_irq_save(flags); rdp = this_cpu_ptr(rsp->rda); - rcu_process_gp_end(rsp, rdp); - check_for_new_grace_period(rsp, rdp); /* Add the callback to our list. */ *rdp->nxttail[RCU_NEXT_TAIL] = head; rdp->nxttail[RCU_NEXT_TAIL] = &head->next; - /* Start a new grace period if one not already started. */ - if (!rcu_gp_in_progress(rsp)) { - unsigned long nestflag; - struct rcu_node *rnp_root = rcu_get_root(rsp); - - raw_spin_lock_irqsave(&rnp_root->lock, nestflag); - rcu_start_gp(rsp, nestflag); /* releases rnp_root->lock. */ - } - /* * Force the grace period if too many callbacks or too long waiting. * Enforce hysteresis, and don't invoke force_quiescent_state() @@ -1459,12 +1448,27 @@ __call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu), * is the only one waiting for a grace period to complete. */ if (unlikely(++rdp->qlen > rdp->qlen_last_fqs_check + qhimark)) { - rdp->blimit = LONG_MAX; - if (rsp->n_force_qs == rdp->n_force_qs_snap && - *rdp->nxttail[RCU_DONE_TAIL] != head) - force_quiescent_state(rsp, 0); - rdp->n_force_qs_snap = rsp->n_force_qs; - rdp->qlen_last_fqs_check = rdp->qlen; + + /* Are we ignoring a completed grace period? */ + rcu_process_gp_end(rsp, rdp); + check_for_new_grace_period(rsp, rdp); + + /* Start a new grace period if one not already started. */ + if (!rcu_gp_in_progress(rsp)) { + unsigned long nestflag; + struct rcu_node *rnp_root = rcu_get_root(rsp); + + raw_spin_lock_irqsave(&rnp_root->lock, nestflag); + rcu_start_gp(rsp, nestflag); /* rlses rnp_root->lock */ + } else { + /* Give the grace period a kick. */ + rdp->blimit = LONG_MAX; + if (rsp->n_force_qs == rdp->n_force_qs_snap && + *rdp->nxttail[RCU_DONE_TAIL] != head) + force_quiescent_state(rsp, 0); + rdp->n_force_qs_snap = rsp->n_force_qs; + rdp->qlen_last_fqs_check = rdp->qlen; + } } else if (ULONG_CMP_LT(ACCESS_ONCE(rsp->jiffies_force_qs), jiffies)) force_quiescent_state(rsp, 1); local_irq_restore(flags); -- cgit v1.2.3 From 43365bd7ff37979d2afdccbe953299ed64a4649b Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Wed, 15 Dec 2010 19:10:17 -0800 Subject: sched: Move periodic share updates to entity_tick() Long running entities that do not block (dequeue) require periodic updates to maintain accurate share values. (Note: group entities with several threads are quite likely to be non-blocking in many circumstances). By virtue of being long-running however, we will see entity ticks (otherwise the required update occurs in dequeue/put and we are done). Thus we can move the detection (and associated work) for these updates into the periodic path. This restores the 'atomicity' of update_curr() with respect to accounting. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra LKML-Reference: <20101216031038.067028969@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 21 +++++++++++++++++---- 1 file changed, 17 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index c88671718bc9..16ee398d8a4e 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -564,10 +564,6 @@ __update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr, #if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED cfs_rq->load_unacc_exec_time += delta_exec; - if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) { - update_cfs_load(cfs_rq, 0); - update_cfs_shares(cfs_rq, 0); - } #endif } @@ -809,6 +805,14 @@ static void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta) reweight_entity(cfs_rq_of(se), se, shares); } + +static void update_entity_shares_tick(struct cfs_rq *cfs_rq) +{ + if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) { + update_cfs_load(cfs_rq, 0); + update_cfs_shares(cfs_rq, 0); + } +} #else /* CONFIG_FAIR_GROUP_SCHED */ static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) { @@ -817,6 +821,10 @@ static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) static inline void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta) { } + +static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq) +{ +} #endif /* CONFIG_FAIR_GROUP_SCHED */ static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se) @@ -1133,6 +1141,11 @@ entity_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr, int queued) */ update_curr(cfs_rq); + /* + * Update share accounting for long-running entities. + */ + update_entity_shares_tick(cfs_rq); + #ifdef CONFIG_SCHED_HRTICK /* * queued ticks are scheduled to match the slice, so don't bother -- cgit v1.2.3 From 19e5eebb8eaa5ca3ff8aa18cb57ccb7a9f67277d Mon Sep 17 00:00:00 2001 From: Paul Turner Date: Wed, 15 Dec 2010 19:10:18 -0800 Subject: sched: Fix interactivity bug by charging unaccounted run-time on entity re-weight Mike Galbraith reported poor interactivity[*] when the new shares distribution code was combined with autogroups. The root cause turns out to be a mis-ordering of accounting accrued execution time and shares updates. Since update_curr() is issued hierarchically, updating the parent entity weights to reflect child enqueue/dequeue results in the parent's unaccounted execution time then being accrued (vs vruntime) at the new weight as opposed to the weight present at accumulation. While this doesn't have much effect on processes with timeslices that cross a tick, it is particularly problematic for an interactive process (e.g. Xorg) which incurs many (tiny) timeslices. In this scenario almost all updates are at dequeue which can result in significant fairness perturbation (especially if it is the only thread, resulting in potential {tg->shares, MIN_SHARES} transitions). Correct this by ensuring unaccounted time is accumulated prior to manipulating an entity's weight. [*] http://xkcd.com/619/ is perversely Nostradamian here. Signed-off-by: Paul Turner Signed-off-by: Peter Zijlstra Cc: Mike Galbraith Cc: Linus Torvalds LKML-Reference: <20101216031038.159704378@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 16ee398d8a4e..c62ebae65cf0 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -765,8 +765,12 @@ static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, unsigned long weight) { - if (se->on_rq) + if (se->on_rq) { + /* commit outstanding execution time */ + if (cfs_rq->curr == se) + update_curr(cfs_rq); account_entity_dequeue(cfs_rq, se); + } update_load_set(&se->load, weight); -- cgit v1.2.3 From 94462ad3b14739d158a1ab87bb30008c1e5a6bc1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 29 Nov 2010 13:15:42 -0500 Subject: module: Move RO/NX module protection to after ftrace module update The commit: 84e1c6bb38eb318e456558b610396d9f1afaabf0 x86: Add RO/NX protection for loadable kernel modules Broke the function tracer with this output: ------------[ cut here ]------------ WARNING: at kernel/trace/ftrace.c:1014 ftrace_bug+0x114/0x171() Hardware name: Precision WorkStation 470 Modules linked in: i2c_core(+) Pid: 86, comm: modprobe Not tainted 2.6.37-rc2+ #68 Call Trace: [] warn_slowpath_common+0x85/0x9d [] ? __process_new_adapter+0x7/0x34 [i2c_core] [] ? __process_new_adapter+0x7/0x34 [i2c_core] [] warn_slowpath_null+0x1a/0x1c [] ftrace_bug+0x114/0x171 [] ? __process_new_adapter+0x7/0x34 [i2c_core] [] ftrace_process_locs+0x1ae/0x274 [] ? __process_new_adapter+0x7/0x34 [i2c_core] [] ftrace_module_notify+0x39/0x44 [] notifier_call_chain+0x37/0x63 [] __blocking_notifier_call_chain+0x46/0x5b [] blocking_notifier_call_chain+0x14/0x16 [] sys_init_module+0x73/0x1f3 [] system_call_fastpath+0x16/0x1b ---[ end trace 2aff4f4ca53ec746 ]--- ftrace faulted on writing [] __process_new_adapter+0x7/0x34 [i2c_core] The cause was that the module text was set to read only before ftrace could convert the calls to mcount to nops. Thus, the conversions failed due to not being able to write to the text locations. The simple fix is to move setting the module to read only after the module notifiers are called (where ftrace sets the module mcounts to nops). Reported-by: Peter Zijlstra Acked-by: Rusty Russell Signed-off-by: Steven Rostedt --- kernel/module.c | 24 ++++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/module.c b/kernel/module.c index 562f665c721f..34e00b708fad 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -2816,18 +2816,6 @@ static struct module *load_module(void __user *umod, kfree(info.strmap); free_copy(&info); - /* Set RO and NX regions for core */ - set_section_ro_nx(mod->module_core, - mod->core_text_size, - mod->core_ro_size, - mod->core_size); - - /* Set RO and NX regions for init */ - set_section_ro_nx(mod->module_init, - mod->init_text_size, - mod->init_ro_size, - mod->init_size); - /* Done! */ trace_module_load(mod); return mod; @@ -2888,6 +2876,18 @@ SYSCALL_DEFINE3(init_module, void __user *, umod, blocking_notifier_call_chain(&module_notify_list, MODULE_STATE_COMING, mod); + /* Set RO and NX regions for core */ + set_section_ro_nx(mod->module_core, + mod->core_text_size, + mod->core_ro_size, + mod->core_size); + + /* Set RO and NX regions for init */ + set_section_ro_nx(mod->module_init, + mod->init_text_size, + mod->init_ro_size, + mod->init_size); + do_mod_ctors(mod); /* Start the module */ if (mod->init != NULL) -- cgit v1.2.3 From 6706125e291bd3dddd269e043323a6ab93ccd5fb Mon Sep 17 00:00:00 2001 From: Yong Zhang Date: Fri, 31 Dec 2010 21:58:58 +0800 Subject: sched: Remove redundant CONFIG_CGROUP_SCHED ifdef CONFIG_[FAIR|RT]_GROUP_SCHED always means CONFIG_CGROUP_SCHED Signed-off-by: Yong Zhang Cc: Peter Zijlstra LKML-Reference: <1293803938-8157-1-git-send-email-yong.zhang0@gmail.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 3925a1bbf5dd..9f9dd8dda53c 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -7777,7 +7777,6 @@ void __init sched_init(void) #ifdef CONFIG_FAIR_GROUP_SCHED init_task_group.shares = init_task_group_load; INIT_LIST_HEAD(&rq->leaf_cfs_rq_list); -#ifdef CONFIG_CGROUP_SCHED /* * How much cpu bandwidth does init_task_group get? * @@ -7798,15 +7797,12 @@ void __init sched_init(void) * directly in rq->cfs (i.e init_task_group->se[] = NULL). */ init_tg_cfs_entry(&init_task_group, &rq->cfs, NULL, i, NULL); -#endif #endif /* CONFIG_FAIR_GROUP_SCHED */ rq->rt.rt_runtime = def_rt_bandwidth.rt_runtime; #ifdef CONFIG_RT_GROUP_SCHED INIT_LIST_HEAD(&rq->leaf_rt_rq_list); -#ifdef CONFIG_CGROUP_SCHED init_tg_rt_entry(&init_task_group, &rq->rt, NULL, i, NULL); -#endif #endif for (j = 0; j < CPU_LOAD_IDX_MAX; j++) -- cgit v1.2.3 From 4f8219875a0dad2cfad9e93a3fafcd9626db98d2 Mon Sep 17 00:00:00 2001 From: Mike Galbraith Date: Thu, 16 Dec 2010 15:09:52 +0100 Subject: sched, autogroup: Fix potential access to freed memory Oleg pointed out that the /proc interface kref_get() useage may race with the final put during autogroup_move_group(). A signal->autogroup assignment may be in flight when the /proc interface dereference, leaving them taking a reference to an already dead group. Reported-by: Oleg Nesterov Signed-off-by: Mike Galbraith Signed-off-by: Peter Zijlstra LKML-Reference: <1292508592.5940.28.camel@maggy.simson.net> Signed-off-by: Ingo Molnar --- kernel/sched_autogroup.c | 25 +++++++++++++++++-------- 1 file changed, 17 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_autogroup.c b/kernel/sched_autogroup.c index 57a7ac286a02..c80fedcd476b 100644 --- a/kernel/sched_autogroup.c +++ b/kernel/sched_autogroup.c @@ -41,6 +41,20 @@ static inline struct autogroup *autogroup_kref_get(struct autogroup *ag) return ag; } +static inline struct autogroup *autogroup_task_get(struct task_struct *p) +{ + struct autogroup *ag; + unsigned long flags; + + if (!lock_task_sighand(p, &flags)) + return autogroup_kref_get(&autogroup_default); + + ag = autogroup_kref_get(p->signal->autogroup); + unlock_task_sighand(p, &flags); + + return ag; +} + static inline struct autogroup *autogroup_create(void) { struct autogroup *ag = kzalloc(sizeof(*ag), GFP_KERNEL); @@ -149,11 +163,7 @@ EXPORT_SYMBOL(sched_autogroup_detach); void sched_autogroup_fork(struct signal_struct *sig) { - struct task_struct *p = current; - - spin_lock_irq(&p->sighand->siglock); - sig->autogroup = autogroup_kref_get(p->signal->autogroup); - spin_unlock_irq(&p->sighand->siglock); + sig->autogroup = autogroup_task_get(current); } void sched_autogroup_exit(struct signal_struct *sig) @@ -172,7 +182,6 @@ __setup("noautogroup", setup_autogroup); #ifdef CONFIG_PROC_FS -/* Called with siglock held. */ int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice) { static unsigned long next = INITIAL_JIFFIES; @@ -194,7 +203,7 @@ int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice) return -EAGAIN; next = HZ / 10 + jiffies; - ag = autogroup_kref_get(p->signal->autogroup); + ag = autogroup_task_get(p); down_write(&ag->lock); err = sched_group_set_shares(ag->tg, prio_to_weight[*nice + 20]); @@ -209,7 +218,7 @@ int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice) void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m) { - struct autogroup *ag = autogroup_kref_get(p->signal->autogroup); + struct autogroup *ag = autogroup_task_get(p); down_read(&ag->lock); seq_printf(m, "/autogroup-%ld nice %d\n", ag->id, ag->nice); -- cgit v1.2.3 From 101e5f77bf35679809586e250b6c62193d2ed179 Mon Sep 17 00:00:00 2001 From: Mike Galbraith Date: Fri, 31 Dec 2010 09:32:30 +0100 Subject: sched, autogroup: Fix reference leak The cgroup exit mess also uncovered a struct autogroup reference leak. copy_process() was simply freeing vs putting the signal_struct, stranding a reference. Signed-off-by: Mike Galbraith Signed-off-by: Peter Zijlstra Cc: Oleg Nesterov LKML-Reference: <1293784350.6839.2.camel@marge.simson.net> Signed-off-by: Ingo Molnar --- kernel/fork.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index b6f2475f1e83..067244495966 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1317,7 +1317,7 @@ bad_fork_cleanup_mm: } bad_fork_cleanup_signal: if (!(clone_flags & CLONE_THREAD)) - free_signal_struct(p->signal); + put_signal_struct(p->signal); bad_fork_cleanup_sighand: __cleanup_sighand(p->sighand); bad_fork_cleanup_fs: -- cgit v1.2.3 From 6bf4123760a5aece6e4829ce90b70b6ffd751d65 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Wed, 5 Jan 2011 12:50:16 +1100 Subject: sched: Change wait_for_completion_*_timeout() to return a signed long wait_for_completion_*_timeout() can return: 0: if the wait timed out -ve: if the wait was interrupted +ve: if the completion was completed. As they currently return an 'unsigned long', the last two cases are not easily distinguished which can easily result in buggy code, as is the case for the recently added wait_for_completion_interruptible_timeout() call in net/sunrpc/cache.c So change them both to return 'long'. As MAX_SCHEDULE_TIMEOUT is LONG_MAX, a large +ve return value should never overflow. Signed-off-by: NeilBrown Cc: Peter Zijlstra Cc: J. Bruce Fields Cc: Andrew Morton Cc: Linus Torvalds LKML-Reference: <20110105125016.64ccab0e@notabene.brown> Signed-off-by: Ingo Molnar --- include/linux/completion.h | 8 ++++---- kernel/sched.c | 4 ++-- 2 files changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/include/linux/completion.h b/include/linux/completion.h index 36d57f74cd01..51494e6b5548 100644 --- a/include/linux/completion.h +++ b/include/linux/completion.h @@ -81,10 +81,10 @@ extern int wait_for_completion_interruptible(struct completion *x); extern int wait_for_completion_killable(struct completion *x); extern unsigned long wait_for_completion_timeout(struct completion *x, unsigned long timeout); -extern unsigned long wait_for_completion_interruptible_timeout( - struct completion *x, unsigned long timeout); -extern unsigned long wait_for_completion_killable_timeout( - struct completion *x, unsigned long timeout); +extern long wait_for_completion_interruptible_timeout( + struct completion *x, unsigned long timeout); +extern long wait_for_completion_killable_timeout( + struct completion *x, unsigned long timeout); extern bool try_wait_for_completion(struct completion *x); extern bool completion_done(struct completion *x); diff --git a/kernel/sched.c b/kernel/sched.c index f2f914e0c47c..114a0deb2b01 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4395,7 +4395,7 @@ EXPORT_SYMBOL(wait_for_completion_interruptible); * This waits for either a completion of a specific task to be signaled or for a * specified timeout to expire. It is interruptible. The timeout is in jiffies. */ -unsigned long __sched +long __sched wait_for_completion_interruptible_timeout(struct completion *x, unsigned long timeout) { @@ -4428,7 +4428,7 @@ EXPORT_SYMBOL(wait_for_completion_killable); * signaled or for a specified timeout to expire. It can be * interrupted by a kill signal. The timeout is in jiffies. */ -unsigned long __sched +long __sched wait_for_completion_killable_timeout(struct completion *x, unsigned long timeout) { -- cgit v1.2.3 From 1dbd1951f39e13da579ffe879cce19586d0462de Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Thu, 9 Dec 2010 15:47:56 +0800 Subject: tracing: Fix preempt count leak While running my ftrace stress test, this showed up: BUG: sleeping function called from invalid context at mm/mmap.c:233 ... note: cat[3293] exited with preempt_count 1 The bug was introduced by commit 91e86e560d0b3ce4c5fc64fd2bbb99f856a30a4e ("tracing: Fix recursive user stack trace") Cc: Signed-off-by: Li Zefan LKML-Reference: <4D0089AC.1020802@cn.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index f8cf959bad45..dc53ecb80589 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1313,12 +1313,10 @@ ftrace_trace_userstack(struct ring_buffer *buffer, unsigned long flags, int pc) __this_cpu_inc(user_stack_count); - - event = trace_buffer_lock_reserve(buffer, TRACE_USER_STACK, sizeof(*entry), flags, pc); if (!event) - return; + goto out_drop_count; entry = ring_buffer_event_data(event); entry->tgid = current->tgid; @@ -1333,8 +1331,8 @@ ftrace_trace_userstack(struct ring_buffer *buffer, unsigned long flags, int pc) if (!filter_check_discard(call, entry, buffer, event)) ring_buffer_unlock_commit(buffer, event); + out_drop_count: __this_cpu_dec(user_stack_count); - out: preempt_enable(); } -- cgit v1.2.3 From 870915e047a2da695be118d32dd5a900f0c3e933 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Thu, 28 Oct 2010 11:31:17 -0400 Subject: tracing: Fix TRACE_EVENT power tracepoint creation DEFINE_TRACE should also exist when CONFIG_EVENT_TRACING=n. Otherwise, setting only TRACEPOINTS=y is broken. Acked-by: Arjan van de Ven Signed-off-by: Mathieu Desnoyers LKML-Reference: <20101028153117.GA4051@Krystal> Signed-off-by: Steven Rostedt --- kernel/Makefile | 1 + kernel/trace/Makefile | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 0b5ff083fa22..e0f2831634b4 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -100,6 +100,7 @@ obj-$(CONFIG_FUNCTION_TRACER) += trace/ obj-$(CONFIG_TRACING) += trace/ obj-$(CONFIG_X86_DS) += trace/ obj-$(CONFIG_RING_BUFFER) += trace/ +obj-$(CONFIG_TRACEPOINTS) += trace/ obj-$(CONFIG_SMP) += sched_cpupri.o obj-$(CONFIG_IRQ_WORK) += irq_work.o obj-$(CONFIG_PERF_EVENTS) += perf_event.o diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 53f338190b26..761c510a06c5 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -52,7 +52,7 @@ obj-$(CONFIG_EVENT_TRACING) += trace_event_perf.o endif obj-$(CONFIG_EVENT_TRACING) += trace_events_filter.o obj-$(CONFIG_KPROBE_EVENT) += trace_kprobe.o -obj-$(CONFIG_EVENT_TRACING) += power-traces.o +obj-$(CONFIG_TRACEPOINTS) += power-traces.o ifeq ($(CONFIG_TRACING),y) obj-$(CONFIG_KGDB_KDB) += trace_kdb.o endif -- cgit v1.2.3