aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--fs/timerfd.c57
-rw-r--r--include/linux/hrtimer.h18
-rw-r--r--include/linux/time.h6
-rw-r--r--include/linux/timerfd.h3
-rw-r--r--kernel/hrtimer.c159
-rw-r--r--kernel/time/alarmtimer.c12
-rw-r--r--kernel/time/tick-broadcast.c16
-rw-r--r--kernel/time/timekeeping.c17
8 files changed, 194 insertions, 94 deletions
diff --git a/fs/timerfd.c b/fs/timerfd.c
index 8c4fc14..7e14c9e 100644
--- a/fs/timerfd.c
+++ b/fs/timerfd.c
@@ -26,10 +26,12 @@
struct timerfd_ctx {
struct hrtimer tmr;
ktime_t tintv;
+ ktime_t moffs;
wait_queue_head_t wqh;
u64 ticks;
int expired;
int clockid;
+ bool might_cancel;
};
/*
@@ -59,24 +61,52 @@ static ktime_t timerfd_get_remaining(struct timerfd_ctx *ctx)
return remaining.tv64 < 0 ? ktime_set(0, 0): remaining;
}
-static void timerfd_setup(struct timerfd_ctx *ctx, int flags,
- const struct itimerspec *ktmr)
+static bool timerfd_canceled(struct timerfd_ctx *ctx)
+{
+ ktime_t moffs;
+
+ if (!ctx->might_cancel)
+ return false;
+
+ moffs = ktime_get_monotonic_offset();
+
+ if (moffs.tv64 == ctx->moffs.tv64)
+ return false;
+
+ ctx->moffs = moffs;
+ return true;
+}
+
+static int timerfd_setup(struct timerfd_ctx *ctx, int flags,
+ const struct itimerspec *ktmr)
{
enum hrtimer_mode htmode;
ktime_t texp;
+ int clockid = ctx->clockid;
htmode = (flags & TFD_TIMER_ABSTIME) ?
HRTIMER_MODE_ABS: HRTIMER_MODE_REL;
+ ctx->might_cancel = false;
+ if (htmode == HRTIMER_MODE_ABS && ctx->clockid == CLOCK_REALTIME &&
+ (flags & TFD_TIMER_CANCELON_SET)) {
+ clockid = CLOCK_REALTIME_COS;
+ ctx->might_cancel = true;
+ }
+
texp = timespec_to_ktime(ktmr->it_value);
ctx->expired = 0;
ctx->ticks = 0;
ctx->tintv = timespec_to_ktime(ktmr->it_interval);
- hrtimer_init(&ctx->tmr, ctx->clockid, htmode);
+ hrtimer_init(&ctx->tmr, clockid, htmode);
hrtimer_set_expires(&ctx->tmr, texp);
ctx->tmr.function = timerfd_tmrproc;
- if (texp.tv64 != 0)
+ if (texp.tv64 != 0) {
hrtimer_start(&ctx->tmr, texp, htmode);
+ if (timerfd_canceled(ctx))
+ return -ECANCELED;
+ }
+ return 0;
}
static int timerfd_release(struct inode *inode, struct file *file)
@@ -118,8 +148,21 @@ static ssize_t timerfd_read(struct file *file, char __user *buf, size_t count,
res = -EAGAIN;
else
res = wait_event_interruptible_locked_irq(ctx->wqh, ctx->ticks);
+
if (ctx->ticks) {
ticks = ctx->ticks;
+
+ /*
+ * If clock has changed, we do not care about the
+ * ticks and we do not rearm the timer. Userspace must
+ * reevaluate anyway.
+ */
+ if (timerfd_canceled(ctx)) {
+ ticks = 0;
+ ctx->expired = 0;
+ res = -ECANCELED;
+ }
+
if (ctx->expired && ctx->tintv.tv64) {
/*
* If tintv.tv64 != 0, this is a periodic timer that
@@ -183,6 +226,7 @@ SYSCALL_DEFINE2(timerfd_create, int, clockid, int, flags)
init_waitqueue_head(&ctx->wqh);
ctx->clockid = clockid;
hrtimer_init(&ctx->tmr, clockid, HRTIMER_MODE_ABS);
+ ctx->moffs = ktime_get_monotonic_offset();
ufd = anon_inode_getfd("[timerfd]", &timerfd_fops, ctx,
O_RDWR | (flags & TFD_SHARED_FCNTL_FLAGS));
@@ -199,6 +243,7 @@ SYSCALL_DEFINE4(timerfd_settime, int, ufd, int, flags,
struct file *file;
struct timerfd_ctx *ctx;
struct itimerspec ktmr, kotmr;
+ int ret;
if (copy_from_user(&ktmr, utmr, sizeof(ktmr)))
return -EFAULT;
@@ -240,14 +285,14 @@ SYSCALL_DEFINE4(timerfd_settime, int, ufd, int, flags,
/*
* Re-program the timer to the new value ...
*/
- timerfd_setup(ctx, flags, &ktmr);
+ ret = timerfd_setup(ctx, flags, &ktmr);
spin_unlock_irq(&ctx->wqh.lock);
fput(file);
if (otmr && copy_to_user(otmr, &kotmr, sizeof(kotmr)))
return -EFAULT;
- return 0;
+ return ret;
}
SYSCALL_DEFINE2(timerfd_gettime, int, ufd, struct itimerspec __user *, otmr)
diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index 62f500c..eda4ccd 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -148,15 +148,14 @@ struct hrtimer_clock_base {
ktime_t resolution;
ktime_t (*get_time)(void);
ktime_t softirq_time;
-#ifdef CONFIG_HIGH_RES_TIMERS
ktime_t offset;
-#endif
};
enum hrtimer_base_type {
HRTIMER_BASE_REALTIME,
HRTIMER_BASE_MONOTONIC,
HRTIMER_BASE_BOOTTIME,
+ HRTIMER_BASE_REALTIME_COS,
HRTIMER_MAX_CLOCK_BASES,
};
@@ -256,8 +255,6 @@ static inline ktime_t hrtimer_expires_remaining(const struct hrtimer *timer)
#ifdef CONFIG_HIGH_RES_TIMERS
struct clock_event_device;
-extern void clock_was_set(void);
-extern void hres_timers_resume(void);
extern void hrtimer_interrupt(struct clock_event_device *dev);
/*
@@ -291,16 +288,8 @@ extern void hrtimer_peek_ahead_timers(void);
# define MONOTONIC_RES_NSEC LOW_RES_NSEC
# define KTIME_MONOTONIC_RES KTIME_LOW_RES
-/*
- * clock_was_set() is a NOP for non- high-resolution systems. The
- * time-sorted order guarantees that a timer does not expire early and
- * is expired in the next softirq when the clock was advanced.
- */
-static inline void clock_was_set(void) { }
static inline void hrtimer_peek_ahead_timers(void) { }
-static inline void hres_timers_resume(void) { }
-
/*
* In non high resolution mode the time reference is taken from
* the base softirq time variable.
@@ -316,10 +305,13 @@ static inline int hrtimer_is_hres_active(struct hrtimer *timer)
}
#endif
+extern void clock_was_set(void);
+extern void hrtimers_resume(void);
+
extern ktime_t ktime_get(void);
extern ktime_t ktime_get_real(void);
extern ktime_t ktime_get_boottime(void);
-
+extern ktime_t ktime_get_monotonic_offset(void);
DECLARE_PER_CPU(struct tick_device, tick_cpu_device);
diff --git a/include/linux/time.h b/include/linux/time.h
index b306178..a924277 100644
--- a/include/linux/time.h
+++ b/include/linux/time.h
@@ -302,6 +302,12 @@ struct itimerval {
* The IDs of various hardware clocks:
*/
#define CLOCK_SGI_CYCLE 10
+
+#ifdef __KERNEL__
+/* This clock is not exposed to user space */
+#define CLOCK_REALTIME_COS 15
+#endif
+
#define MAX_CLOCKS 16
#define CLOCKS_MASK (CLOCK_REALTIME | CLOCK_MONOTONIC)
#define CLOCKS_MONO CLOCK_MONOTONIC
diff --git a/include/linux/timerfd.h b/include/linux/timerfd.h
index 2d07929..e9571fc 100644
--- a/include/linux/timerfd.h
+++ b/include/linux/timerfd.h
@@ -19,6 +19,7 @@
* shared O_* flags.
*/
#define TFD_TIMER_ABSTIME (1 << 0)
+#define TFD_TIMER_CANCELON_SET (1 << 1)
#define TFD_CLOEXEC O_CLOEXEC
#define TFD_NONBLOCK O_NONBLOCK
@@ -26,6 +27,6 @@
/* Flags for timerfd_create. */
#define TFD_CREATE_FLAGS TFD_SHARED_FCNTL_FLAGS
/* Flags for timerfd_settime. */
-#define TFD_SETTIME_FLAGS TFD_TIMER_ABSTIME
+#define TFD_SETTIME_FLAGS (TFD_TIMER_ABSTIME | TFD_TIMER_CANCELON_SET)
#endif /* _LINUX_TIMERFD_H */
diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
index dbbbf7d..eabcbd7 100644
--- a/kernel/hrtimer.c
+++ b/kernel/hrtimer.c
@@ -78,6 +78,11 @@ DEFINE_PER_CPU(struct hrtimer_cpu_base, hrtimer_bases) =
.get_time = &ktime_get_boottime,
.resolution = KTIME_LOW_RES,
},
+ {
+ .index = CLOCK_REALTIME_COS,
+ .get_time = &ktime_get_real,
+ .resolution = KTIME_LOW_RES,
+ },
}
};
@@ -85,6 +90,7 @@ static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = {
[CLOCK_REALTIME] = HRTIMER_BASE_REALTIME,
[CLOCK_MONOTONIC] = HRTIMER_BASE_MONOTONIC,
[CLOCK_BOOTTIME] = HRTIMER_BASE_BOOTTIME,
+ [CLOCK_REALTIME_COS] = HRTIMER_BASE_REALTIME_COS,
};
static inline int hrtimer_clockid_to_base(clockid_t clock_id)
@@ -110,6 +116,7 @@ static void hrtimer_get_softirq_time(struct hrtimer_cpu_base *base)
base->clock_base[HRTIMER_BASE_REALTIME].softirq_time = xtim;
base->clock_base[HRTIMER_BASE_MONOTONIC].softirq_time = mono;
base->clock_base[HRTIMER_BASE_BOOTTIME].softirq_time = boot;
+ base->clock_base[HRTIMER_BASE_REALTIME_COS].softirq_time = xtim;
}
/*
@@ -479,6 +486,8 @@ static inline void debug_deactivate(struct hrtimer *timer)
trace_hrtimer_cancel(timer);
}
+static void hrtimer_expire_cancelable(struct hrtimer_cpu_base *cpu_base);
+
/* High resolution timer related functions */
#ifdef CONFIG_HIGH_RES_TIMERS
@@ -621,66 +630,6 @@ static int hrtimer_reprogram(struct hrtimer *timer,
return res;
}
-
-/*
- * Retrigger next event is called after clock was set
- *
- * Called with interrupts disabled via on_each_cpu()
- */
-static void retrigger_next_event(void *arg)
-{
- struct hrtimer_cpu_base *base;
- struct timespec realtime_offset, wtm, sleep;
-
- if (!hrtimer_hres_active())
- return;
-
- get_xtime_and_monotonic_and_sleep_offset(&realtime_offset, &wtm,
- &sleep);
- set_normalized_timespec(&realtime_offset, -wtm.tv_sec, -wtm.tv_nsec);
-
- base = &__get_cpu_var(hrtimer_bases);
-
- /* Adjust CLOCK_REALTIME offset */
- raw_spin_lock(&base->lock);
- base->clock_base[HRTIMER_BASE_REALTIME].offset =
- timespec_to_ktime(realtime_offset);
- base->clock_base[HRTIMER_BASE_BOOTTIME].offset =
- timespec_to_ktime(sleep);
-
- hrtimer_force_reprogram(base, 0);
- raw_spin_unlock(&base->lock);
-}
-
-/*
- * Clock realtime was set
- *
- * Change the offset of the realtime clock vs. the monotonic
- * clock.
- *
- * We might have to reprogram the high resolution timer interrupt. On
- * SMP we call the architecture specific code to retrigger _all_ high
- * resolution timer interrupts. On UP we just disable interrupts and
- * call the high resolution interrupt code.
- */
-void clock_was_set(void)
-{
- /* Retrigger the CPU local events everywhere */
- on_each_cpu(retrigger_next_event, NULL, 1);
-}
-
-/*
- * During resume we might have to reprogram the high resolution timer
- * interrupt (on the local CPU):
- */
-void hres_timers_resume(void)
-{
- WARN_ONCE(!irqs_disabled(),
- KERN_INFO "hres_timers_resume() called with IRQs enabled!");
-
- retrigger_next_event(NULL);
-}
-
/*
* Initialize the high resolution related parts of cpu_base
*/
@@ -714,12 +663,14 @@ static inline int hrtimer_enqueue_reprogram(struct hrtimer *timer,
return 0;
}
+static void retrigger_next_event(void *arg);
+
/*
* Switch to high resolution mode
*/
static int hrtimer_switch_to_hres(void)
{
- int cpu = smp_processor_id();
+ int i, cpu = smp_processor_id();
struct hrtimer_cpu_base *base = &per_cpu(hrtimer_bases, cpu);
unsigned long flags;
@@ -735,9 +686,8 @@ static int hrtimer_switch_to_hres(void)
return 0;
}
base->hres_active = 1;
- base->clock_base[HRTIMER_BASE_REALTIME].resolution = KTIME_HIGH_RES;
- base->clock_base[HRTIMER_BASE_MONOTONIC].resolution = KTIME_HIGH_RES;
- base->clock_base[HRTIMER_BASE_BOOTTIME].resolution = KTIME_HIGH_RES;
+ for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++)
+ base->clock_base[i].resolution = KTIME_HIGH_RES;
tick_setup_sched_timer();
@@ -764,6 +714,71 @@ static inline void hrtimer_init_hres(struct hrtimer_cpu_base *base) { }
#endif /* CONFIG_HIGH_RES_TIMERS */
+/*
+ * Retrigger next event is called after clock was set
+ *
+ * Called with interrupts disabled via on_each_cpu()
+ */
+static void retrigger_next_event(void *arg)
+{
+ struct hrtimer_cpu_base *base = &__get_cpu_var(hrtimer_bases);
+ struct timespec realtime_offset, xtim, wtm, sleep;
+
+ if (!hrtimer_hres_active()) {
+ raw_spin_lock(&base->lock);
+ hrtimer_expire_cancelable(base);
+ raw_spin_unlock(&base->lock);
+ return;
+ }
+
+ /* Optimized out for !HIGH_RES */
+ get_xtime_and_monotonic_and_sleep_offset(&xtim, &wtm, &sleep);
+ set_normalized_timespec(&realtime_offset, -wtm.tv_sec, -wtm.tv_nsec);
+
+ /* Adjust CLOCK_REALTIME offset */
+ raw_spin_lock(&base->lock);
+ base->clock_base[HRTIMER_BASE_REALTIME].offset =
+ timespec_to_ktime(realtime_offset);
+ base->clock_base[HRTIMER_BASE_BOOTTIME].offset =
+ timespec_to_ktime(sleep);
+ base->clock_base[HRTIMER_BASE_REALTIME_COS].offset =
+ timespec_to_ktime(realtime_offset);
+
+ hrtimer_expire_cancelable(base);
+
+ hrtimer_force_reprogram(base, 0);
+ raw_spin_unlock(&base->lock);
+}
+
+/*
+ * Clock realtime was set
+ *
+ * Change the offset of the realtime clock vs. the monotonic
+ * clock.
+ *
+ * We might have to reprogram the high resolution timer interrupt. On
+ * SMP we call the architecture specific code to retrigger _all_ high
+ * resolution timer interrupts. On UP we just disable interrupts and
+ * call the high resolution interrupt code.
+ */
+void clock_was_set(void)
+{
+ /* Retrigger the CPU local events everywhere */
+ on_each_cpu(retrigger_next_event, NULL, 1);
+}
+
+/*
+ * During resume we might have to reprogram the high resolution timer
+ * interrupt (on the local CPU):
+ */
+void hrtimers_resume(void)
+{
+ WARN_ONCE(!irqs_disabled(),
+ KERN_INFO "hrtimers_resume() called with IRQs enabled!");
+
+ retrigger_next_event(NULL);
+}
+
static inline void timer_stats_hrtimer_set_start_info(struct hrtimer *timer)
{
#ifdef CONFIG_TIMER_STATS
@@ -1225,6 +1240,22 @@ static void __run_hrtimer(struct hrtimer *timer, ktime_t *now)
timer->state &= ~HRTIMER_STATE_CALLBACK;
}
+static void hrtimer_expire_cancelable(struct hrtimer_cpu_base *cpu_base)
+{
+ struct timerqueue_node *node;
+ struct hrtimer_clock_base *base;
+ ktime_t now = ktime_get_real();
+
+ base = &cpu_base->clock_base[HRTIMER_BASE_REALTIME_COS];
+
+ while ((node = timerqueue_getnext(&base->active))) {
+ struct hrtimer *timer;
+
+ timer = container_of(node, struct hrtimer, node);
+ __run_hrtimer(timer, &now);
+ }
+}
+
#ifdef CONFIG_HIGH_RES_TIMERS
/*
diff --git a/kernel/time/alarmtimer.c b/kernel/time/alarmtimer.c
index 9265014..c6027fe 100644
--- a/kernel/time/alarmtimer.c
+++ b/kernel/time/alarmtimer.c
@@ -669,12 +669,20 @@ static int __init has_wakealarm(struct device *dev, void *name_ptr)
*/
static int __init alarmtimer_init_late(void)
{
+ struct device *dev;
char *str;
/* Find an rtc device and init the rtc_timer */
- class_find_device(rtc_class, NULL, &str, has_wakealarm);
- if (str)
+ dev = class_find_device(rtc_class, NULL, &str, has_wakealarm);
+ /* If we have a device then str is valid. See has_wakealarm() */
+ if (dev) {
rtcdev = rtc_class_open(str);
+ /*
+ * Drop the reference we got in class_find_device,
+ * rtc_open takes its own.
+ */
+ put_device(dev);
+ }
if (!rtcdev) {
printk(KERN_WARNING "No RTC device found, ALARM timers will"
" not wake from suspend");
diff --git a/kernel/time/tick-broadcast.c b/kernel/time/tick-broadcast.c
index 723c763..c7218d1 100644
--- a/kernel/time/tick-broadcast.c
+++ b/kernel/time/tick-broadcast.c
@@ -456,23 +456,27 @@ void tick_broadcast_oneshot_control(unsigned long reason)
unsigned long flags;
int cpu;
- raw_spin_lock_irqsave(&tick_broadcast_lock, flags);
-
/*
* Periodic mode does not care about the enter/exit of power
* states
*/
if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC)
- goto out;
+ return;
- bc = tick_broadcast_device.evtdev;
+ /*
+ * We are called with preemtion disabled from the depth of the
+ * idle code, so we can't be moved away.
+ */
cpu = smp_processor_id();
td = &per_cpu(tick_cpu_device, cpu);
dev = td->evtdev;
if (!(dev->features & CLOCK_EVT_FEAT_C3STOP))
- goto out;
+ return;
+
+ bc = tick_broadcast_device.evtdev;
+ raw_spin_lock_irqsave(&tick_broadcast_lock, flags);
if (reason == CLOCK_EVT_NOTIFY_BROADCAST_ENTER) {
if (!cpumask_test_cpu(cpu, tick_get_broadcast_oneshot_mask())) {
cpumask_set_cpu(cpu, tick_get_broadcast_oneshot_mask());
@@ -489,8 +493,6 @@ void tick_broadcast_oneshot_control(unsigned long reason)
tick_program_event(dev->next_event, 1);
}
}
-
-out:
raw_spin_unlock_irqrestore(&tick_broadcast_lock, flags);
}
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 8e6a05a..342408c 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -680,7 +680,7 @@ static void timekeeping_resume(void)
clockevents_notify(CLOCK_EVT_NOTIFY_RESUME, NULL);
/* Resume hrtimers */
- hres_timers_resume();
+ hrtimers_resume();
}
static int timekeeping_suspend(void)
@@ -1099,6 +1099,21 @@ void get_xtime_and_monotonic_and_sleep_offset(struct timespec *xtim,
}
/**
+ * ktime_get_monotonic_offset() - get wall_to_monotonic in ktime_t format
+ */
+ktime_t ktime_get_monotonic_offset(void)
+{
+ unsigned long seq;
+ struct timespec wtom;
+
+ do {
+ seq = read_seqbegin(&xtime_lock);
+ wtom = wall_to_monotonic;
+ } while (read_seqretry(&xtime_lock, seq));
+ return timespec_to_ktime(wtom);
+}
+
+/**
* xtime_update() - advances the timekeeping infrastructure
* @ticks: number of ticks, that have elapsed since the last call.
*