time, drivers/rtc: Don't bother with rtc_resume() for the nonstop clocksource

If a system does not provide a persistent_clock(), the time
will be updated on resume by rtc_resume(). With the addition
of the non-stop clocksources for suspend timing, those systems
set the time on resume in timekeeping_resume(), but may not
provide a valid persistent_clock().

This results in the rtc_resume() logic thinking no one has set
the time and it then will over-write the suspend time again,
which is not necessary and only increases clock error.

So, fix this for rtc_resume().

This patch also improves the name of persistent_clock_exist to
make it more grammatical.

Signed-off-by: Xunlei Pang <pang.xunlei@linaro.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1427945681-29972-19-git-send-email-john.stultz@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>

3 files changed, 54 insertions, 25 deletions

diff --git a/drivers/rtc/class.c b/drivers/rtc/class.c
index d40760aa4553..c29ba7e14304 100644
--- a/drivers/rtc/class.c
+++ b/drivers/rtc/class.c
@@ -55,7 +55,7 @@ static int rtc_suspend(struct device *dev)
 	struct timespec64	delta, delta_delta;
 	int err;
 
-	if (has_persistent_clock())
+	if (timekeeping_rtc_skipsuspend())
 		return 0;
 
 	if (strcmp(dev_name(&rtc->dev), CONFIG_RTC_HCTOSYS_DEVICE) != 0)
@@ -102,7 +102,7 @@ static int rtc_resume(struct device *dev)
 	struct timespec64	sleep_time;
 	int err;
 
-	if (has_persistent_clock())
+	if (timekeeping_rtc_skipresume())
 		return 0;
 
 	rtc_hctosys_ret = -ENODEV;
diff --git a/include/linux/timekeeping.h b/include/linux/timekeeping.h
index 7a2369d5b3f4..99176af216af 100644
--- a/include/linux/timekeeping.h
+++ b/include/linux/timekeeping.h
@@ -248,6 +248,9 @@ static inline void timekeeping_clocktai(struct timespec *ts)
 /*
  * RTC specific
  */
+extern bool timekeeping_rtc_skipsuspend(void);
+extern bool timekeeping_rtc_skipresume(void);
+
 extern void timekeeping_inject_sleeptime64(struct timespec64 *delta);
 
 /*
@@ -259,14 +262,8 @@ extern void getnstime_raw_and_real(struct timespec *ts_raw,
 /*
  * Persistent clock related interfaces
  */
-extern bool persistent_clock_exist;
 extern int persistent_clock_is_local;
 
-static inline bool has_persistent_clock(void)
-{
-	return persistent_clock_exist;
-}
-
 extern void read_persistent_clock(struct timespec *ts);
 extern void read_persistent_clock64(struct timespec64 *ts);
 extern void read_boot_clock(struct timespec *ts);
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index b7db4916415b..79b9bc6e7876 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -64,9 +64,6 @@ static struct tk_fast tk_fast_raw  ____cacheline_aligned;
 /* flag for if timekeeping is suspended */
 int __read_mostly timekeeping_suspended;
 
-/* Flag for if there is a persistent clock on this platform */
-bool __read_mostly persistent_clock_exist = false;
-
 static inline void tk_normalize_xtime(struct timekeeper *tk)
 {
 	while (tk->tkr_mono.xtime_nsec >= ((u64)NSEC_PER_SEC << tk->tkr_mono.shift)) {
@@ -1204,6 +1201,12 @@ void __weak read_boot_clock64(struct timespec64 *ts64)
 	*ts64 = timespec_to_timespec64(ts);
 }
 
+/* Flag for if timekeeping_resume() has injected sleeptime */
+static bool sleeptime_injected;
+
+/* Flag for if there is a persistent clock on this platform */
+static bool persistent_clock_exists;
+
 /*
  * timekeeping_init - Initializes the clocksource and common timekeeping values
  */
@@ -1221,7 +1224,7 @@ void __init timekeeping_init(void)
 		now.tv_sec = 0;
 		now.tv_nsec = 0;
 	} else if (now.tv_sec || now.tv_nsec)
-		persistent_clock_exist = true;
+		persistent_clock_exists = true;
 
 	read_boot_clock64(&boot);
 	if (!timespec64_valid_strict(&boot)) {
@@ -1282,11 +1285,47 @@ static void __timekeeping_inject_sleeptime(struct timekeeper *tk,
 
 #if defined(CONFIG_PM_SLEEP) && defined(CONFIG_RTC_HCTOSYS_DEVICE)
 /**
+ * We have three kinds of time sources to use for sleep time
+ * injection, the preference order is:
+ * 1) non-stop clocksource
+ * 2) persistent clock (ie: RTC accessible when irqs are off)
+ * 3) RTC
+ *
+ * 1) and 2) are used by timekeeping, 3) by RTC subsystem.
+ * If system has neither 1) nor 2), 3) will be used finally.
+ *
+ *
+ * If timekeeping has injected sleeptime via either 1) or 2),
+ * 3) becomes needless, so in this case we don't need to call
+ * rtc_resume(), and this is what timekeeping_rtc_skipresume()
+ * means.
+ */
+bool timekeeping_rtc_skipresume(void)
+{
+	return sleeptime_injected;
+}
+
+/**
+ * 1) can be determined whether to use or not only when doing
+ * timekeeping_resume() which is invoked after rtc_suspend(),
+ * so we can't skip rtc_suspend() surely if system has 1).
+ *
+ * But if system has 2), 2) will definitely be used, so in this
+ * case we don't need to call rtc_suspend(), and this is what
+ * timekeeping_rtc_skipsuspend() means.
+ */
+bool timekeeping_rtc_skipsuspend(void)
+{
+	return persistent_clock_exists;
+}
+
+/**
  * timekeeping_inject_sleeptime64 - Adds suspend interval to timeekeeping values
  * @delta: pointer to a timespec64 delta value
  *
  * This hook is for architectures that cannot support read_persistent_clock64
  * because their RTC/persistent clock is only accessible when irqs are enabled.
+ * and also don't have an effective nonstop clocksource.
  *
  * This function should only be called by rtc_resume(), and allows
  * a suspend offset to be injected into the timekeeping values.
@@ -1296,13 +1335,6 @@ void timekeeping_inject_sleeptime64(struct timespec64 *delta)
 	struct timekeeper *tk = &tk_core.timekeeper;
 	unsigned long flags;
 
-	/*
-	 * Make sure we don't set the clock twice, as timekeeping_resume()
-	 * already did it
-	 */
-	if (has_persistent_clock())
-		return;
-
 	raw_spin_lock_irqsave(&timekeeper_lock, flags);
 	write_seqcount_begin(&tk_core.seq);
 
@@ -1334,8 +1366,8 @@ void timekeeping_resume(void)
 	unsigned long flags;
 	struct timespec64 ts_new, ts_delta;
 	cycle_t cycle_now, cycle_delta;
-	bool suspendtime_found = false;
 
+	sleeptime_injected = false;
 	read_persistent_clock64(&ts_new);
 
 	clockevents_resume();
@@ -1381,13 +1413,13 @@ void timekeeping_resume(void)
 		nsec += ((u64) cycle_delta * mult) >> shift;
 
 		ts_delta = ns_to_timespec64(nsec);
-		suspendtime_found = true;
+		sleeptime_injected = true;
 	} else if (timespec64_compare(&ts_new, &timekeeping_suspend_time) > 0) {
 		ts_delta = timespec64_sub(ts_new, timekeeping_suspend_time);
-		suspendtime_found = true;
+		sleeptime_injected = true;
 	}
 
-	if (suspendtime_found)
+	if (sleeptime_injected)
 		__timekeeping_inject_sleeptime(tk, &ts_delta);
 
 	/* Re-base the last cycle value */
@@ -1421,14 +1453,14 @@ int timekeeping_suspend(void)
 	 * value returned, update the persistent_clock_exists flag.
 	 */
 	if (timekeeping_suspend_time.tv_sec || timekeeping_suspend_time.tv_nsec)
-		persistent_clock_exist = true;
+		persistent_clock_exists = true;
 
 	raw_spin_lock_irqsave(&timekeeper_lock, flags);
 	write_seqcount_begin(&tk_core.seq);
 	timekeeping_forward_now(tk);
 	timekeeping_suspended = 1;
 
-	if (has_persistent_clock()) {
+	if (persistent_clock_exists) {
 		/*
 		 * To avoid drift caused by repeated suspend/resumes,
 		 * which each can add ~1 second drift error,