timerfd: Manage cancelable timers in timerfd

Peter is concerned about the extra scan of CLOCK_REALTIME_COS in the
timer interrupt. Yes, I did not think about it, because the solution
was so elegant. I didn't like the extra list in timerfd when it was
proposed some time ago, but with a rcu based list the list walk it's
less horrible than the original global lock, which was held over the
list iteration.

Requested-by: Peter Zijlstra <[e-mail]>
Signed-off-by: Thomas Gleixner <[e-mail]>
Reviewed-by: Peter Zijlstra <[e-mail]>

fs/timerfd.c

···
 #include <linux/anon_inodes.h>
 #include <linux/timerfd.h>
 #include <linux/syscalls.h>
+#include <linux/rcupdate.h>
 
 struct timerfd_ctx {
 	struct hrtimer tmr;
···
 	u64 ticks;
 	int expired;
 	int clockid;
+	struct rcu_head rcu;
+	struct list_head clist;
 	bool might_cancel;
 };
+
+static LIST_HEAD(cancel_list);
+static DEFINE_SPINLOCK(cancel_lock);
 
 /*
  * This gets called when the timer event triggers. We set the "expired"
···
 	return HRTIMER_NORESTART;
 }
 
+/*
+ * Called when the clock was set to cancel the timers in the cancel
+ * list.
+ */
+void timerfd_clock_was_set(void)
+{
+	ktime_t moffs = ktime_get_monotonic_offset();
+	struct timerfd_ctx *ctx;
+	unsigned long flags;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(ctx, &cancel_list, clist) {
+		if (!ctx->might_cancel)
+			continue;
+		spin_lock_irqsave(&ctx->wqh.lock, flags);
+		if (ctx->moffs.tv64 != moffs.tv64) {
+			ctx->moffs.tv64 = KTIME_MAX;
+			wake_up_locked(&ctx->wqh);
+		}
+		spin_unlock_irqrestore(&ctx->wqh.lock, flags);
+	}
+	rcu_read_unlock();
+}
+
+static void timerfd_remove_cancel(struct timerfd_ctx *ctx)
+{
+	if (ctx->might_cancel) {
+		ctx->might_cancel = false;
+		spin_lock(&cancel_lock);
+		list_del_rcu(&ctx->clist);
+		spin_unlock(&cancel_lock);
+	}
+}
+
+static bool timerfd_canceled(struct timerfd_ctx *ctx)
+{
+	if (!ctx->might_cancel || ctx->moffs.tv64 != KTIME_MAX)
+		return false;
+	ctx->moffs = ktime_get_monotonic_offset();
+	return true;
+}
+
+static void timerfd_setup_cancel(struct timerfd_ctx *ctx, int flags)
+{
+	if (ctx->clockid == CLOCK_REALTIME && (flags & TFD_TIMER_ABSTIME) &&
+	    (flags & TFD_TIMER_CANCEL_ON_SET)) {
+		if (!ctx->might_cancel) {
+			ctx->might_cancel = true;
+			spin_lock(&cancel_lock);
+			list_add_rcu(&ctx->clist, &cancel_list);
+			spin_unlock(&cancel_lock);
+		}
+	} else if (ctx->might_cancel) {
+		timerfd_remove_cancel(ctx);
+	}
+}
+
 static ktime_t timerfd_get_remaining(struct timerfd_ctx *ctx)
 {
 	ktime_t remaining;
 
 	remaining = hrtimer_expires_remaining(&ctx->tmr);
 	return remaining.tv64 < 0 ? ktime_set(0, 0): remaining;
-}
-
-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,
···
 
 	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;
···
 {
 	struct timerfd_ctx *ctx = file->private_data;
 
+	timerfd_remove_cancel(ctx);
 	hrtimer_cancel(&ctx->tmr);
-	kfree(ctx);
+	kfree_rcu(ctx, rcu);
 	return 0;
 }
 
···
 	else
 		res = wait_event_interruptible_locked_irq(ctx->wqh, 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)) {
+		ctx->ticks = 0;
+		ctx->expired = 0;
+		res = -ECANCELED;
+	}
+
 	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 (IS_ERR(file))
 		return PTR_ERR(file);
 	ctx = file->private_data;
+
+	timerfd_setup_cancel(ctx, flags);
 
 	/*
 	 * We need to stop the existing timer before reprogramming

include/linux/hrtimer.h

···
 	HRTIMER_BASE_REALTIME,
 	HRTIMER_BASE_MONOTONIC,
 	HRTIMER_BASE_BOOTTIME,
-	HRTIMER_BASE_REALTIME_COS,
 	HRTIMER_MAX_CLOCK_BASES,
 };
 
···
 #endif
 
 extern void clock_was_set(void);
+#ifdef CONFIG_TIMERFD
+extern void timerfd_clock_was_set(void);
+#else
+static inline void timerfd_clock_was_set(void) { }
+#endif
 extern void hrtimers_resume(void);
 
 extern ktime_t ktime_get(void);

include/linux/time.h

···
  * 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

include/linux/timerfd.h

···
  * shared O_* flags.
  */
 #define TFD_TIMER_ABSTIME (1 << 0)
-#define TFD_TIMER_CANCELON_SET (1 << 1)
+#define TFD_TIMER_CANCEL_ON_SET (1 << 1)
 #define TFD_CLOEXEC O_CLOEXEC
 #define TFD_NONBLOCK O_NONBLOCK
 
···
 /* Flags for timerfd_create.  */
 #define TFD_CREATE_FLAGS TFD_SHARED_FCNTL_FLAGS
 /* Flags for timerfd_settime.  */
-#define TFD_SETTIME_FLAGS (TFD_TIMER_ABSTIME | TFD_TIMER_CANCELON_SET)
+#define TFD_SETTIME_FLAGS (TFD_TIMER_ABSTIME | TFD_TIMER_CANCEL_ON_SET)
 
 #endif /* _LINUX_TIMERFD_H */

kernel/hrtimer.c

···
 			.get_time = &ktime_get_boottime,
 			.resolution = KTIME_LOW_RES,
 		},
-		{
-			.index = CLOCK_REALTIME_COS,
-			.get_time = &ktime_get_real,
-			.resolution = KTIME_LOW_RES,
-		},
 	}
 };
 
···
 	[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)
···
 	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;
 }
 
 /*
···
 	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
 
···
 	return 0;
 }
 
-static void retrigger_next_event(void *arg);
+/*
+ * 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())
+		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);
+
+	hrtimer_force_reprogram(base, 0);
+	raw_spin_unlock(&base->lock);
+}
 
 /*
  * Switch to high resolution mode
···
 	return 0;
 }
 static inline void hrtimer_init_hres(struct hrtimer_cpu_base *base) { }
+static inline void retrigger_next_event(void *arg) { }
 
 #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
···
  */
 void clock_was_set(void)
 {
+#ifdef CONFIG_HIGHRES_TIMERS
 	/* Retrigger the CPU local events everywhere */
 	on_each_cpu(retrigger_next_event, NULL, 1);
+#endif
+	timerfd_clock_was_set();
 }
 
 /*
···
 		  KERN_INFO "hrtimers_resume() called with IRQs enabled!");
 
 	retrigger_next_event(NULL);
+	timerfd_clock_was_set();
 }
 
 static inline void timer_stats_hrtimer_set_start_info(struct hrtimer *timer)
···
 	WARN_ON_ONCE(!(timer->state & HRTIMER_STATE_CALLBACK));
 
 	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