clockevents: check broadcast tick device not the clock events device
[pandora-kernel.git] / kernel / hrtimer.c
index dea4c91..cdec83e 100644 (file)
@@ -43,6 +43,7 @@
 #include <linux/tick.h>
 #include <linux/seq_file.h>
 #include <linux/err.h>
+#include <linux/debugobjects.h>
 
 #include <asm/uaccess.h>
 
@@ -152,15 +153,6 @@ static void hrtimer_get_softirq_time(struct hrtimer_cpu_base *base)
                ktime_add(xtim, tomono);
 }
 
-/*
- * Helper function to check, whether the timer is running the callback
- * function
- */
-static inline int hrtimer_callback_running(struct hrtimer *timer)
-{
-       return timer->state & HRTIMER_STATE_CALLBACK;
-}
-
 /*
  * Functions and macros which are different for UP/SMP systems are kept in a
  * single place
@@ -308,11 +300,10 @@ EXPORT_SYMBOL_GPL(ktime_sub_ns);
  */
 u64 ktime_divns(const ktime_t kt, s64 div)
 {
-       u64 dclc, inc, dns;
+       u64 dclc;
        int sft = 0;
 
-       dclc = dns = ktime_to_ns(kt);
-       inc = div;
+       dclc = ktime_to_ns(kt);
        /* Make sure the divisor is less than 2^32: */
        while (div >> 32) {
                sft++;
@@ -342,6 +333,115 @@ ktime_t ktime_add_safe(const ktime_t lhs, const ktime_t rhs)
        return res;
 }
 
+#ifdef CONFIG_DEBUG_OBJECTS_TIMERS
+
+static struct debug_obj_descr hrtimer_debug_descr;
+
+/*
+ * fixup_init is called when:
+ * - an active object is initialized
+ */
+static int hrtimer_fixup_init(void *addr, enum debug_obj_state state)
+{
+       struct hrtimer *timer = addr;
+
+       switch (state) {
+       case ODEBUG_STATE_ACTIVE:
+               hrtimer_cancel(timer);
+               debug_object_init(timer, &hrtimer_debug_descr);
+               return 1;
+       default:
+               return 0;
+       }
+}
+
+/*
+ * fixup_activate is called when:
+ * - an active object is activated
+ * - an unknown object is activated (might be a statically initialized object)
+ */
+static int hrtimer_fixup_activate(void *addr, enum debug_obj_state state)
+{
+       switch (state) {
+
+       case ODEBUG_STATE_NOTAVAILABLE:
+               WARN_ON_ONCE(1);
+               return 0;
+
+       case ODEBUG_STATE_ACTIVE:
+               WARN_ON(1);
+
+       default:
+               return 0;
+       }
+}
+
+/*
+ * fixup_free is called when:
+ * - an active object is freed
+ */
+static int hrtimer_fixup_free(void *addr, enum debug_obj_state state)
+{
+       struct hrtimer *timer = addr;
+
+       switch (state) {
+       case ODEBUG_STATE_ACTIVE:
+               hrtimer_cancel(timer);
+               debug_object_free(timer, &hrtimer_debug_descr);
+               return 1;
+       default:
+               return 0;
+       }
+}
+
+static struct debug_obj_descr hrtimer_debug_descr = {
+       .name           = "hrtimer",
+       .fixup_init     = hrtimer_fixup_init,
+       .fixup_activate = hrtimer_fixup_activate,
+       .fixup_free     = hrtimer_fixup_free,
+};
+
+static inline void debug_hrtimer_init(struct hrtimer *timer)
+{
+       debug_object_init(timer, &hrtimer_debug_descr);
+}
+
+static inline void debug_hrtimer_activate(struct hrtimer *timer)
+{
+       debug_object_activate(timer, &hrtimer_debug_descr);
+}
+
+static inline void debug_hrtimer_deactivate(struct hrtimer *timer)
+{
+       debug_object_deactivate(timer, &hrtimer_debug_descr);
+}
+
+static inline void debug_hrtimer_free(struct hrtimer *timer)
+{
+       debug_object_free(timer, &hrtimer_debug_descr);
+}
+
+static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id,
+                          enum hrtimer_mode mode);
+
+void hrtimer_init_on_stack(struct hrtimer *timer, clockid_t clock_id,
+                          enum hrtimer_mode mode)
+{
+       debug_object_init_on_stack(timer, &hrtimer_debug_descr);
+       __hrtimer_init(timer, clock_id, mode);
+}
+
+void destroy_hrtimer_on_stack(struct hrtimer *timer)
+{
+       debug_object_free(timer, &hrtimer_debug_descr);
+}
+
+#else
+static inline void debug_hrtimer_init(struct hrtimer *timer) { }
+static inline void debug_hrtimer_activate(struct hrtimer *timer) { }
+static inline void debug_hrtimer_deactivate(struct hrtimer *timer) { }
+#endif
+
 /*
  * Check, whether the timer is on the callback pending list
  */
@@ -522,7 +622,7 @@ static void retrigger_next_event(void *arg)
 void clock_was_set(void)
 {
        /* Retrigger the CPU local events everywhere */
-       on_each_cpu(retrigger_next_event, NULL, 0, 1);
+       on_each_cpu(retrigger_next_event, NULL, 1);
 }
 
 /*
@@ -531,8 +631,6 @@ void clock_was_set(void)
  */
 void hres_timers_resume(void)
 {
-       WARN_ON_ONCE(num_online_cpus() > 1);
-
        /* Retrigger the CPU local events: */
        retrigger_next_event(NULL);
 }
@@ -567,20 +665,23 @@ static inline int hrtimer_enqueue_reprogram(struct hrtimer *timer,
                /* Timer is expired, act upon the callback mode */
                switch(timer->cb_mode) {
                case HRTIMER_CB_IRQSAFE_NO_RESTART:
+                       debug_hrtimer_deactivate(timer);
                        /*
                         * We can call the callback from here. No restart
                         * happens, so no danger of recursion
                         */
                        BUG_ON(timer->function(timer) != HRTIMER_NORESTART);
                        return 1;
-               case HRTIMER_CB_IRQSAFE_NO_SOFTIRQ:
+               case HRTIMER_CB_IRQSAFE_PERCPU:
+               case HRTIMER_CB_IRQSAFE_UNLOCKED:
                        /*
                         * This is solely for the sched tick emulation with
                         * dynamic tick support to ensure that we do not
                         * restart the tick right on the edge and end up with
                         * the tick timer in the softirq ! The calling site
-                        * takes care of this.
+                        * takes care of this. Also used for hrtimer sleeper !
                         */
+                       debug_hrtimer_deactivate(timer);
                        return 1;
                case HRTIMER_CB_IRQSAFE:
                case HRTIMER_CB_SOFTIRQ:
@@ -735,6 +836,8 @@ static void enqueue_hrtimer(struct hrtimer *timer,
        struct hrtimer *entry;
        int leftmost = 1;
 
+       debug_hrtimer_activate(timer);
+
        /*
         * Find the right place in the rbtree:
         */
@@ -831,6 +934,7 @@ remove_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base)
                 * reprogramming happens in the interrupt handler. This is a
                 * rare case and less expensive than a smp call.
                 */
+               debug_hrtimer_deactivate(timer);
                timer_stats_hrtimer_clear_start_info(timer);
                reprogram = base->cpu_base == &__get_cpu_var(hrtimer_bases);
                __remove_hrtimer(timer, base, HRTIMER_STATE_INACTIVE,
@@ -878,6 +982,7 @@ hrtimer_start(struct hrtimer *timer, ktime_t tim, const enum hrtimer_mode mode)
                tim = ktime_add_safe(tim, base->resolution);
 #endif
        }
+
        timer->expires = tim;
 
        timer_stats_hrtimer_set_start_info(timer);
@@ -896,10 +1001,18 @@ hrtimer_start(struct hrtimer *timer, ktime_t tim, const enum hrtimer_mode mode)
         */
        raise = timer->state == HRTIMER_STATE_PENDING;
 
+       /*
+        * We use preempt_disable to prevent this task from migrating after
+        * setting up the softirq and raising it. Otherwise, if me migrate
+        * we will raise the softirq on the wrong CPU.
+        */
+       preempt_disable();
+
        unlock_hrtimer_base(timer, &flags);
 
        if (raise)
                hrtimer_raise_softirq();
+       preempt_enable();
 
        return ret;
 }
@@ -971,7 +1084,7 @@ ktime_t hrtimer_get_remaining(const struct hrtimer *timer)
 }
 EXPORT_SYMBOL_GPL(hrtimer_get_remaining);
 
-#if defined(CONFIG_NO_IDLE_HZ) || defined(CONFIG_NO_HZ)
+#ifdef CONFIG_NO_HZ
 /**
  * hrtimer_get_next_event - get the time until next expiry event
  *
@@ -1011,14 +1124,8 @@ ktime_t hrtimer_get_next_event(void)
 }
 #endif
 
-/**
- * hrtimer_init - initialize a timer to the given clock
- * @timer:     the timer to be initialized
- * @clock_id:  the clock to be used
- * @mode:      timer mode abs/rel
- */
-void hrtimer_init(struct hrtimer *timer, clockid_t clock_id,
-                 enum hrtimer_mode mode)
+static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id,
+                          enum hrtimer_mode mode)
 {
        struct hrtimer_cpu_base *cpu_base;
 
@@ -1039,6 +1146,19 @@ void hrtimer_init(struct hrtimer *timer, clockid_t clock_id,
        memset(timer->start_comm, 0, TASK_COMM_LEN);
 #endif
 }
+
+/**
+ * hrtimer_init - initialize a timer to the given clock
+ * @timer:     the timer to be initialized
+ * @clock_id:  the clock to be used
+ * @mode:      timer mode abs/rel
+ */
+void hrtimer_init(struct hrtimer *timer, clockid_t clock_id,
+                 enum hrtimer_mode mode)
+{
+       debug_hrtimer_init(timer);
+       __hrtimer_init(timer, clock_id, mode);
+}
 EXPORT_SYMBOL_GPL(hrtimer_init);
 
 /**
@@ -1072,6 +1192,7 @@ static void run_hrtimer_pending(struct hrtimer_cpu_base *cpu_base)
                timer = list_entry(cpu_base->cb_pending.next,
                                   struct hrtimer, cb_entry);
 
+               debug_hrtimer_deactivate(timer);
                timer_stats_account_hrtimer(timer);
 
                fn = timer->function;
@@ -1120,11 +1241,13 @@ static void __run_hrtimer(struct hrtimer *timer)
        enum hrtimer_restart (*fn)(struct hrtimer *);
        int restart;
 
+       debug_hrtimer_deactivate(timer);
        __remove_hrtimer(timer, base, HRTIMER_STATE_CALLBACK, 0);
        timer_stats_account_hrtimer(timer);
 
        fn = timer->function;
-       if (timer->cb_mode == HRTIMER_CB_IRQSAFE_NO_SOFTIRQ) {
+       if (timer->cb_mode == HRTIMER_CB_IRQSAFE_PERCPU ||
+           timer->cb_mode == HRTIMER_CB_IRQSAFE_UNLOCKED) {
                /*
                 * Used for scheduler timers, avoid lock inversion with
                 * rq->lock and tasklist_lock.
@@ -1331,7 +1454,7 @@ void hrtimer_init_sleeper(struct hrtimer_sleeper *sl, struct task_struct *task)
        sl->timer.function = hrtimer_wakeup;
        sl->task = task;
 #ifdef CONFIG_HIGH_RES_TIMERS
-       sl->timer.cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ;
+       sl->timer.cb_mode = HRTIMER_CB_IRQSAFE_UNLOCKED;
 #endif
 }
 
@@ -1378,22 +1501,27 @@ long __sched hrtimer_nanosleep_restart(struct restart_block *restart)
 {
        struct hrtimer_sleeper t;
        struct timespec __user  *rmtp;
+       int ret = 0;
 
-       hrtimer_init(&t.timer, restart->nanosleep.index, HRTIMER_MODE_ABS);
+       hrtimer_init_on_stack(&t.timer, restart->nanosleep.index,
+                               HRTIMER_MODE_ABS);
        t.timer.expires.tv64 = restart->nanosleep.expires;
 
        if (do_nanosleep(&t, HRTIMER_MODE_ABS))
-               return 0;
+               goto out;
 
        rmtp = restart->nanosleep.rmtp;
        if (rmtp) {
-               int ret = update_rmtp(&t.timer, rmtp);
+               ret = update_rmtp(&t.timer, rmtp);
                if (ret <= 0)
-                       return ret;
+                       goto out;
        }
 
        /* The other values in restart are already filled in */
-       return -ERESTART_RESTARTBLOCK;
+       ret = -ERESTART_RESTARTBLOCK;
+out:
+       destroy_hrtimer_on_stack(&t.timer);
+       return ret;
 }
 
 long hrtimer_nanosleep(struct timespec *rqtp, struct timespec __user *rmtp,
@@ -1401,20 +1529,23 @@ long hrtimer_nanosleep(struct timespec *rqtp, struct timespec __user *rmtp,
 {
        struct restart_block *restart;
        struct hrtimer_sleeper t;
+       int ret = 0;
 
-       hrtimer_init(&t.timer, clockid, mode);
+       hrtimer_init_on_stack(&t.timer, clockid, mode);
        t.timer.expires = timespec_to_ktime(*rqtp);
        if (do_nanosleep(&t, mode))
-               return 0;
+               goto out;
 
        /* Absolute timers do not update the rmtp value and restart: */
-       if (mode == HRTIMER_MODE_ABS)
-               return -ERESTARTNOHAND;
+       if (mode == HRTIMER_MODE_ABS) {
+               ret = -ERESTARTNOHAND;
+               goto out;
+       }
 
        if (rmtp) {
-               int ret = update_rmtp(&t.timer, rmtp);
+               ret = update_rmtp(&t.timer, rmtp);
                if (ret <= 0)
-                       return ret;
+                       goto out;
        }
 
        restart = &current_thread_info()->restart_block;
@@ -1423,7 +1554,10 @@ long hrtimer_nanosleep(struct timespec *rqtp, struct timespec __user *rmtp,
        restart->nanosleep.rmtp = rmtp;
        restart->nanosleep.expires = t.timer.expires.tv64;
 
-       return -ERESTART_RESTARTBLOCK;
+       ret = -ERESTART_RESTARTBLOCK;
+out:
+       destroy_hrtimer_on_stack(&t.timer);
+       return ret;
 }
 
 asmlinkage long
@@ -1459,28 +1593,95 @@ static void __cpuinit init_hrtimers_cpu(int cpu)
 
 #ifdef CONFIG_HOTPLUG_CPU
 
-static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
-                               struct hrtimer_clock_base *new_base)
+static int migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
+                               struct hrtimer_clock_base *new_base, int dcpu)
 {
        struct hrtimer *timer;
        struct rb_node *node;
+       int raise = 0;
 
        while ((node = rb_first(&old_base->active))) {
                timer = rb_entry(node, struct hrtimer, node);
                BUG_ON(hrtimer_callback_running(timer));
-               __remove_hrtimer(timer, old_base, HRTIMER_STATE_INACTIVE, 0);
+               debug_hrtimer_deactivate(timer);
+
+               /*
+                * Should not happen. Per CPU timers should be
+                * canceled _before_ the migration code is called
+                */
+               if (timer->cb_mode == HRTIMER_CB_IRQSAFE_PERCPU) {
+                       __remove_hrtimer(timer, old_base,
+                                        HRTIMER_STATE_INACTIVE, 0);
+                       WARN(1, "hrtimer (%p %p)active but cpu %d dead\n",
+                            timer, timer->function, dcpu);
+                       continue;
+               }
+
+               /*
+                * Mark it as STATE_MIGRATE not INACTIVE otherwise the
+                * timer could be seen as !active and just vanish away
+                * under us on another CPU
+                */
+               __remove_hrtimer(timer, old_base, HRTIMER_STATE_MIGRATE, 0);
                timer->base = new_base;
                /*
                 * Enqueue the timer. Allow reprogramming of the event device
                 */
                enqueue_hrtimer(timer, new_base, 1);
+
+#ifdef CONFIG_HIGH_RES_TIMERS
+               /*
+                * Happens with high res enabled when the timer was
+                * already expired and the callback mode is
+                * HRTIMER_CB_IRQSAFE_UNLOCKED (hrtimer_sleeper). The
+                * enqueue code does not move them to the soft irq
+                * pending list for performance/latency reasons, but
+                * in the migration state, we need to do that
+                * otherwise we end up with a stale timer.
+                */
+               if (timer->state == HRTIMER_STATE_MIGRATE) {
+                       timer->state = HRTIMER_STATE_PENDING;
+                       list_add_tail(&timer->cb_entry,
+                                     &new_base->cpu_base->cb_pending);
+                       raise = 1;
+               }
+#endif
+               /* Clear the migration state bit */
+               timer->state &= ~HRTIMER_STATE_MIGRATE;
+       }
+       return raise;
+}
+
+#ifdef CONFIG_HIGH_RES_TIMERS
+static int migrate_hrtimer_pending(struct hrtimer_cpu_base *old_base,
+                                  struct hrtimer_cpu_base *new_base)
+{
+       struct hrtimer *timer;
+       int raise = 0;
+
+       while (!list_empty(&old_base->cb_pending)) {
+               timer = list_entry(old_base->cb_pending.next,
+                                  struct hrtimer, cb_entry);
+
+               __remove_hrtimer(timer, timer->base, HRTIMER_STATE_PENDING, 0);
+               timer->base = &new_base->clock_base[timer->base->index];
+               list_add_tail(&timer->cb_entry, &new_base->cb_pending);
+               raise = 1;
        }
+       return raise;
+}
+#else
+static int migrate_hrtimer_pending(struct hrtimer_cpu_base *old_base,
+                                  struct hrtimer_cpu_base *new_base)
+{
+       return 0;
 }
+#endif
 
 static void migrate_hrtimers(int cpu)
 {
        struct hrtimer_cpu_base *old_base, *new_base;
-       int i;
+       int i, raise = 0;
 
        BUG_ON(cpu_online(cpu));
        old_base = &per_cpu(hrtimer_bases, cpu);
@@ -1493,14 +1694,21 @@ static void migrate_hrtimers(int cpu)
        spin_lock_nested(&old_base->lock, SINGLE_DEPTH_NESTING);
 
        for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++) {
-               migrate_hrtimer_list(&old_base->clock_base[i],
-                                    &new_base->clock_base[i]);
+               if (migrate_hrtimer_list(&old_base->clock_base[i],
+                                        &new_base->clock_base[i], cpu))
+                       raise = 1;
        }
 
+       if (migrate_hrtimer_pending(old_base, new_base))
+               raise = 1;
+
        spin_unlock(&old_base->lock);
        spin_unlock(&new_base->lock);
        local_irq_enable();
        put_cpu_var(hrtimer_bases);
+
+       if (raise)
+               hrtimer_raise_softirq();
 }
 #endif /* CONFIG_HOTPLUG_CPU */
 
@@ -1541,7 +1749,7 @@ void __init hrtimers_init(void)
                          (void *)(long)smp_processor_id());
        register_cpu_notifier(&hrtimers_nb);
 #ifdef CONFIG_HIGH_RES_TIMERS
-       open_softirq(HRTIMER_SOFTIRQ, run_hrtimer_softirq, NULL);
+       open_softirq(HRTIMER_SOFTIRQ, run_hrtimer_softirq);
 #endif
 }