[PATCH 00/11] 3.2-stable: Fix for leapsecond caused hrtimer/futex issue

[PATCH 00/11] 3.2-stable: Fix for leapsecond caused hrtimer/futex issue

[John Stultz]

Here is backport of the leapsecond fixes to 3.2-stable. These are less
straight forward, and should get closer review.

This patch set addresses two issues:

1) Deadlock leapsecond issue that a few reports described.

I spent some time over the weekend trying to find a way to reproduce
the hard-hang issue some folks were reporting after the leapsecond.
Initially I didn't think the 6b43ae8a619d17 leap-second hrimter livelock
patch needed to be backported since, I assumed it required the ntp_lock
split for it to be triggered, but looking again I found that the same
issue could occur prior to splitting out the ntp_lock. So I've backported
that fix (and its follow-on fixups) as well as created a test case
to reproduce the hard-hang deadlock.


2) Early hrtimer/futex expiration issue that was more widely observed

This is the load-spike issue that a number of folks saw that did not
hard hang most boxes (although some reports did show nmi-watchdogs
triggering due to sudden spinning in tight loops).

I've booted and tested this entire patchset on two boxes and run through a
number of leapsecond related stress tests. However, additional testing and
review would be appreciated.

The original commits backported in this set are:

Deadlock issue fixes:
---------------------
6b43ae8a619d17c4935c3320d2ef9e92bdeed05d    ntp: Fix leap-second hrtimer livelock
dd48d708ff3e917f6d6b6c2b696c3f18c019feed    ntp: Correct TAI offset during leap second
fad0c66c4bb836d57a5f125ecd38bed653ca863a    timekeeping: Fix CLOCK_MONOTONIC inconsistency during leapsecond

Helper change: (allows the following fixes to backport more easily):
--------------------------------------------------------------------
cc06268c6a87db156af2daed6e96a936b955cc82    time: Move common updates to a function

Hrtimer early-expiration issue fixes:
-------------------------------
f55a6faa384304c89cfef162768e88374d3312cb    hrtimer: Provide clock_was_set_delayed()
4873fa070ae84a4115f0b3c9dfabc224f1bc7c51    timekeeping: Fix leapsecond triggered load spike issue
5b9fe759a678e05be4937ddf03d50e950207c1c0    timekeeping: Maintain ktime_t based offsets for hrtimers
196951e91262fccda81147d2bcf7fdab08668b40    hrtimers: Move lock held region in hrtimer_interrupt()
f6c06abfb3972ad4914cef57d8348fcb2932bc3b    timekeeping: Provide hrtimer update function
5baefd6d84163443215f4a99f6a20f054ef11236    hrtimer: Update hrtimer base offsets each hrtimer_interrupt
3e997130bd2e8c6f5aaa49d6e3161d4d29b43ab0    timekeeping: Add missing update call in timekeeping_resume()


I've already done backports to all the stable kernels to 2.6.32, and
will send out the rest tomorrow.

Please let me know if you have any comments or feedback. 

thanks
-john

Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>

John Stultz (5):
  3.2.x: ntp: Fix leap-second hrtimer livelock
  3.2.x: timekeeping: Fix CLOCK_MONOTONIC inconsistency during
    leapsecond
  3.2.x: hrtimer: Provide clock_was_set_delayed()
  3.2.x: timekeeping: Fix leapsecond triggered load spike issue
  3.2.x: hrtimer: Update hrtimer base offsets each hrtimer_interrupt

Richard Cochran (1):
  3.2.x: ntp: Correct TAI offset during leap second

Thomas Gleixner (5):
  3.2.x: time: Move common updates to a function
  3.2.x: timekeeping: Maintain ktime_t based offsets for hrtimers
  3.2.x: hrtimers: Move lock held region in hrtimer_interrupt()
  3.2.x: timekeeping: Provide hrtimer update function
  3.2.x: timekeeping: Add missing update call in timekeeping_resume()

 include/linux/hrtimer.h   |   10 +++-
 include/linux/timex.h     |    2 +-
 kernel/hrtimer.c          |   53 +++++++++++++------
 kernel/time/ntp.c         |  124 +++++++++++++++------------------------------
 kernel/time/timekeeping.c |  115 +++++++++++++++++++++++++++++++----------
 5 files changed, 175 insertions(+), 129 deletions(-)

-- 
1.7.9.5

[PATCH 01/11] 3.2.x: ntp: Fix leap-second hrtimer livelock

[John Stultz]

From: John Stultz <john.stultz@linaro.org>

This is a backport of 6b43ae8a619d17c4935c3320d2ef9e92bdeed05d

This should have been backported when it was commited, but I
mistook the problem as requiring the ntp_lock changes
that landed in 3.4 in order for it to occur.

Unfortunately the same issue can happen (with only one cpu)
as follows:
do_adjtimex()
 write_seqlock_irq(&xtime_lock);
  process_adjtimex_modes()
   process_adj_status()
    ntp_start_leap_timer()
     hrtimer_start()
      hrtimer_reprogram()
       tick_program_event()
        clockevents_program_event()
         ktime_get()
          seq = req_seqbegin(xtime_lock); [DEADLOCK]

This deadlock will no always occur, as it requires the
leap_timer to force a hrtimer_reprogram which only happens
if its set and there's no sooner timer to expire.

NOTE: This patch, being faithful to the original commit,
introduces a bug (we don't update wall_to_monotonic),
which will be resovled by backporting a following fix.

Original commit message below:

Since commit 7dffa3c673fbcf835cd7be80bb4aec8ad3f51168 the ntp
subsystem has used an hrtimer for triggering the leapsecond
adjustment. However, this can cause a potential livelock.

Thomas diagnosed this as the following pattern:
CPU 0                                                    CPU 1
do_adjtimex()
  spin_lock_irq(&ntp_lock);
    process_adjtimex_modes();				 timer_interrupt()
      process_adj_status();                                do_timer()
        ntp_start_leap_timer();                             write_lock(&xtime_lock);
          hrtimer_start();                                  update_wall_time();
             hrtimer_reprogram();                            ntp_tick_length()
               tick_program_event()                            spin_lock(&ntp_lock);
                 clockevents_program_event()
		   ktime_get()
                     seq = req_seqbegin(xtime_lock);

This patch tries to avoid the problem by reverting back to not using
an hrtimer to inject leapseconds, and instead we handle the leapsecond
processing in the second_overflow() function.

The downside to this change is that on systems that support highres
timers, the leap second processing will occur on a HZ tick boundary,
(ie: ~1-10ms, depending on HZ)  after the leap second instead of
possibly sooner (~34us in my tests w/ x86_64 lapic).

This patch applies on top of tip/timers/core.

CC: Sasha Levin <levinsasha928@gmail.com>
CC: Thomas Gleixner <tglx@linutronix.de>
Reported-by: Sasha Levin <levinsasha928@gmail.com>
Diagnoised-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Sasha Levin <levinsasha928@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
---
 include/linux/timex.h     |    2 +-
 kernel/time/ntp.c         |  122 +++++++++++++++------------------------------
 kernel/time/timekeeping.c |   18 +++----
 3 files changed, 48 insertions(+), 94 deletions(-)

diff --git a/include/linux/timex.h b/include/linux/timex.h
index aa60fe7..08e90fb 100644
--- a/include/linux/timex.h
+++ b/include/linux/timex.h
@@ -266,7 +266,7 @@ static inline int ntp_synced(void)
 /* Returns how long ticks are at present, in ns / 2^NTP_SCALE_SHIFT. */
 extern u64 tick_length;
 
-extern void second_overflow(void);
+extern int second_overflow(unsigned long secs);
 extern void update_ntp_one_tick(void);
 extern int do_adjtimex(struct timex *);
 extern void hardpps(const struct timespec *, const struct timespec *);
diff --git a/kernel/time/ntp.c b/kernel/time/ntp.c
index 4b85a7a..4508f7f 100644
--- a/kernel/time/ntp.c
+++ b/kernel/time/ntp.c
@@ -31,8 +31,6 @@ unsigned long			tick_nsec;
 u64				tick_length;
 static u64			tick_length_base;
 
-static struct hrtimer		leap_timer;
-
 #define MAX_TICKADJ		500LL		/* usecs */
 #define MAX_TICKADJ_SCALED \
 	(((MAX_TICKADJ * NSEC_PER_USEC) << NTP_SCALE_SHIFT) / NTP_INTERVAL_FREQ)
@@ -350,60 +348,60 @@ void ntp_clear(void)
 }
 
 /*
- * Leap second processing. If in leap-insert state at the end of the
- * day, the system clock is set back one second; if in leap-delete
- * state, the system clock is set ahead one second.
+ * this routine handles the overflow of the microsecond field
+ *
+ * The tricky bits of code to handle the accurate clock support
+ * were provided by Dave Mills (Mills@UDEL.EDU) of NTP fame.
+ * They were originally developed for SUN and DEC kernels.
+ * All the kudos should go to Dave for this stuff.
+ *
+ * Also handles leap second processing, and returns leap offset
  */
-static enum hrtimer_restart ntp_leap_second(struct hrtimer *timer)
+int second_overflow(unsigned long secs)
 {
-	enum hrtimer_restart res = HRTIMER_NORESTART;
-
-	write_seqlock(&xtime_lock);
+	int leap = 0;
+	s64 delta;
 
+	/*
+	 * Leap second processing. If in leap-insert state at the end of the
+	 * day, the system clock is set back one second; if in leap-delete
+	 * state, the system clock is set ahead one second.
+	 */
 	switch (time_state) {
 	case TIME_OK:
+		if (time_status & STA_INS)
+			time_state = TIME_INS;
+		else if (time_status & STA_DEL)
+			time_state = TIME_DEL;
 		break;
 	case TIME_INS:
-		timekeeping_leap_insert(-1);
-		time_state = TIME_OOP;
-		printk(KERN_NOTICE
-			"Clock: inserting leap second 23:59:60 UTC\n");
-		hrtimer_add_expires_ns(&leap_timer, NSEC_PER_SEC);
-		res = HRTIMER_RESTART;
+		if (secs % 86400 == 0) {
+			leap = -1;
+			time_state = TIME_OOP;
+			printk(KERN_NOTICE
+				"Clock: inserting leap second 23:59:60 UTC\n");
+		}
 		break;
 	case TIME_DEL:
-		timekeeping_leap_insert(1);
-		time_tai--;
-		time_state = TIME_WAIT;
-		printk(KERN_NOTICE
-			"Clock: deleting leap second 23:59:59 UTC\n");
+		if ((secs + 1) % 86400 == 0) {
+			leap = 1;
+			time_tai--;
+			time_state = TIME_WAIT;
+			printk(KERN_NOTICE
+				"Clock: deleting leap second 23:59:59 UTC\n");
+		}
 		break;
 	case TIME_OOP:
 		time_tai++;
 		time_state = TIME_WAIT;
-		/* fall through */
+		break;
+
 	case TIME_WAIT:
 		if (!(time_status & (STA_INS | STA_DEL)))
 			time_state = TIME_OK;
 		break;
 	}
 
-	write_sequnlock(&xtime_lock);
-
-	return res;
-}
-
-/*
- * this routine handles the overflow of the microsecond field
- *
- * The tricky bits of code to handle the accurate clock support
- * were provided by Dave Mills (Mills@UDEL.EDU) of NTP fame.
- * They were originally developed for SUN and DEC kernels.
- * All the kudos should go to Dave for this stuff.
- */
-void second_overflow(void)
-{
-	s64 delta;
 
 	/* Bump the maxerror field */
 	time_maxerror += MAXFREQ / NSEC_PER_USEC;
@@ -423,23 +421,25 @@ void second_overflow(void)
 	pps_dec_valid();
 
 	if (!time_adjust)
-		return;
+		goto out;
 
 	if (time_adjust > MAX_TICKADJ) {
 		time_adjust -= MAX_TICKADJ;
 		tick_length += MAX_TICKADJ_SCALED;
-		return;
+		goto out;
 	}
 
 	if (time_adjust < -MAX_TICKADJ) {
 		time_adjust += MAX_TICKADJ;
 		tick_length -= MAX_TICKADJ_SCALED;
-		return;
+		goto out;
 	}
 
 	tick_length += (s64)(time_adjust * NSEC_PER_USEC / NTP_INTERVAL_FREQ)
 							 << NTP_SCALE_SHIFT;
 	time_adjust = 0;
+out:
+	return leap;
 }
 
 #ifdef CONFIG_GENERIC_CMOS_UPDATE
@@ -501,27 +501,6 @@ static void notify_cmos_timer(void)
 static inline void notify_cmos_timer(void) { }
 #endif
 
-/*
- * Start the leap seconds timer:
- */
-static inline void ntp_start_leap_timer(struct timespec *ts)
-{
-	long now = ts->tv_sec;
-
-	if (time_status & STA_INS) {
-		time_state = TIME_INS;
-		now += 86400 - now % 86400;
-		hrtimer_start(&leap_timer, ktime_set(now, 0), HRTIMER_MODE_ABS);
-
-		return;
-	}
-
-	if (time_status & STA_DEL) {
-		time_state = TIME_DEL;
-		now += 86400 - (now + 1) % 86400;
-		hrtimer_start(&leap_timer, ktime_set(now, 0), HRTIMER_MODE_ABS);
-	}
-}
 
 /*
  * Propagate a new txc->status value into the NTP state:
@@ -546,22 +525,6 @@ static inline void process_adj_status(struct timex *txc, struct timespec *ts)
 	time_status &= STA_RONLY;
 	time_status |= txc->status & ~STA_RONLY;
 
-	switch (time_state) {
-	case TIME_OK:
-		ntp_start_leap_timer(ts);
-		break;
-	case TIME_INS:
-	case TIME_DEL:
-		time_state = TIME_OK;
-		ntp_start_leap_timer(ts);
-	case TIME_WAIT:
-		if (!(time_status & (STA_INS | STA_DEL)))
-			time_state = TIME_OK;
-		break;
-	case TIME_OOP:
-		hrtimer_restart(&leap_timer);
-		break;
-	}
 }
 /*
  * Called with the xtime lock held, so we can access and modify
@@ -643,9 +606,6 @@ int do_adjtimex(struct timex *txc)
 		    (txc->tick <  900000/USER_HZ ||
 		     txc->tick > 1100000/USER_HZ))
 			return -EINVAL;
-
-		if (txc->modes & ADJ_STATUS && time_state != TIME_OK)
-			hrtimer_cancel(&leap_timer);
 	}
 
 	if (txc->modes & ADJ_SETOFFSET) {
@@ -967,6 +927,4 @@ __setup("ntp_tick_adj=", ntp_tick_adj_setup);
 void __init ntp_init(void)
 {
 	ntp_clear();
-	hrtimer_init(&leap_timer, CLOCK_REALTIME, HRTIMER_MODE_ABS);
-	leap_timer.function = ntp_leap_second;
 }
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 2378413..4780a7d 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -169,15 +169,6 @@ static struct timespec raw_time;
 /* flag for if timekeeping is suspended */
 int __read_mostly timekeeping_suspended;
 
-/* must hold xtime_lock */
-void timekeeping_leap_insert(int leapsecond)
-{
-	xtime.tv_sec += leapsecond;
-	wall_to_monotonic.tv_sec -= leapsecond;
-	update_vsyscall(&xtime, &wall_to_monotonic, timekeeper.clock,
-			timekeeper.mult);
-}
-
 /**
  * timekeeping_forward_now - update clock to the current time
  *
@@ -942,9 +933,11 @@ static cycle_t logarithmic_accumulation(cycle_t offset, int shift)
 
 	timekeeper.xtime_nsec += timekeeper.xtime_interval << shift;
 	while (timekeeper.xtime_nsec >= nsecps) {
+		int leap;
 		timekeeper.xtime_nsec -= nsecps;
 		xtime.tv_sec++;
-		second_overflow();
+		leap = second_overflow(xtime.tv_sec);
+		xtime.tv_sec += leap;
 	}
 
 	/* Accumulate raw time */
@@ -1050,9 +1043,12 @@ static void update_wall_time(void)
 	 * xtime.tv_nsec isn't larger then NSEC_PER_SEC
 	 */
 	if (unlikely(xtime.tv_nsec >= NSEC_PER_SEC)) {
+		int leap;
 		xtime.tv_nsec -= NSEC_PER_SEC;
 		xtime.tv_sec++;
-		second_overflow();
+		leap = second_overflow(xtime.tv_sec);
+		xtime.tv_sec += leap;
+
 	}
 
 	/* check to see if there is a new clocksource to use */
-- 
1.7.9.5

[PATCH 02/11] 3.2.x: ntp: Correct TAI offset during leap second

[John Stultz]

From: Richard Cochran <richardcochran@gmail.com>

This is a backport of dd48d708ff3e917f6d6b6c2b696c3f18c019feed

When repeating a UTC time value during a leap second (when the UTC
time should be 23:59:60), the TAI timescale should not stop. The kernel
NTP code increments the TAI offset one second too late. This patch fixes
the issue by incrementing the offset during the leap second itself.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
---
 kernel/time/ntp.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/time/ntp.c b/kernel/time/ntp.c
index 4508f7f..f1eb182 100644
--- a/kernel/time/ntp.c
+++ b/kernel/time/ntp.c
@@ -378,6 +378,7 @@ int second_overflow(unsigned long secs)
 		if (secs % 86400 == 0) {
 			leap = -1;
 			time_state = TIME_OOP;
+			time_tai++;
 			printk(KERN_NOTICE
 				"Clock: inserting leap second 23:59:60 UTC\n");
 		}
@@ -392,7 +393,6 @@ int second_overflow(unsigned long secs)
 		}
 		break;
 	case TIME_OOP:
-		time_tai++;
 		time_state = TIME_WAIT;
 		break;
 
-- 
1.7.9.5

[PATCH 03/11] 3.2.x: timekeeping: Fix CLOCK_MONOTONIC inconsistency during leapsecond

[John Stultz]

From: John Stultz <john.stultz@linaro.org>

This is a backport of fad0c66c4bb836d57a5f125ecd38bed653ca863a
which resolves a bug the previous commit.

Commit 6b43ae8a61 (ntp: Fix leap-second hrtimer livelock) broke the
leapsecond update of CLOCK_MONOTONIC. The missing leapsecond update to
wall_to_monotonic causes discontinuities in CLOCK_MONOTONIC.

Adjust wall_to_monotonic when NTP inserted a leapsecond.

Reported-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Tested-by: Richard Cochran <richardcochran@gmail.com>
Cc: stable@kernel.org
Link: http://lkml.kernel.org/r/1338400497-12420-1-git-send-email-john.stultz@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 kernel/time/timekeeping.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 4780a7d..5c9b67e 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -938,6 +938,7 @@ static cycle_t logarithmic_accumulation(cycle_t offset, int shift)
 		xtime.tv_sec++;
 		leap = second_overflow(xtime.tv_sec);
 		xtime.tv_sec += leap;
+		wall_to_monotonic.tv_sec -= leap;
 	}
 
 	/* Accumulate raw time */
@@ -1048,7 +1049,7 @@ static void update_wall_time(void)
 		xtime.tv_sec++;
 		leap = second_overflow(xtime.tv_sec);
 		xtime.tv_sec += leap;
-
+		wall_to_monotonic.tv_sec -= leap;
 	}
 
 	/* check to see if there is a new clocksource to use */
-- 
1.7.9.5

[PATCH 04/11] 3.2.x: time: Move common updates to a function

[John Stultz]

From: Thomas Gleixner <tglx@linutronix.de>

This is a backport of cc06268c6a87db156af2daed6e96a936b955cc82

While not a bugfix itself, it allows following fixes to backport
in a more straightforward manner.

CC: Thomas Gleixner <tglx@linutronix.de>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
---
 kernel/time/timekeeping.c |   34 +++++++++++++++++-----------------
 1 file changed, 17 insertions(+), 17 deletions(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 5c9b67e..5d55185 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -166,6 +166,19 @@ static struct timespec total_sleep_time;
  */
 static struct timespec raw_time;
 
+/* must hold write on xtime_lock */
+static void timekeeping_update(bool clearntp)
+{
+	if (clearntp) {
+		timekeeper.ntp_error = 0;
+		ntp_clear();
+	}
+	update_vsyscall(&xtime, &wall_to_monotonic,
+			 timekeeper.clock, timekeeper.mult);
+}
+
+
+
 /* flag for if timekeeping is suspended */
 int __read_mostly timekeeping_suspended;
 
@@ -366,11 +379,7 @@ int do_settimeofday(const struct timespec *tv)
 
 	xtime = *tv;
 
-	timekeeper.ntp_error = 0;
-	ntp_clear();
-
-	update_vsyscall(&xtime, &wall_to_monotonic, timekeeper.clock,
-				timekeeper.mult);
+	timekeeping_update(true);
 
 	write_sequnlock_irqrestore(&xtime_lock, flags);
 
@@ -403,11 +412,7 @@ int timekeeping_inject_offset(struct timespec *ts)
 	xtime = timespec_add(xtime, *ts);
 	wall_to_monotonic = timespec_sub(wall_to_monotonic, *ts);
 
-	timekeeper.ntp_error = 0;
-	ntp_clear();
-
-	update_vsyscall(&xtime, &wall_to_monotonic, timekeeper.clock,
-				timekeeper.mult);
+	timekeeping_update(true);
 
 	write_sequnlock_irqrestore(&xtime_lock, flags);
 
@@ -636,10 +641,7 @@ void timekeeping_inject_sleeptime(struct timespec *delta)
 
 	__timekeeping_inject_sleeptime(delta);
 
-	timekeeper.ntp_error = 0;
-	ntp_clear();
-	update_vsyscall(&xtime, &wall_to_monotonic, timekeeper.clock,
-				timekeeper.mult);
+	timekeeping_update(true);
 
 	write_sequnlock_irqrestore(&xtime_lock, flags);
 
@@ -1052,9 +1054,7 @@ static void update_wall_time(void)
 		wall_to_monotonic.tv_sec -= leap;
 	}
 
-	/* check to see if there is a new clocksource to use */
-	update_vsyscall(&xtime, &wall_to_monotonic, timekeeper.clock,
-				timekeeper.mult);
+	timekeeping_update(false);
 }
 
 /**
-- 
1.7.9.5

[PATCH 05/11] 3.2.x: hrtimer: Provide clock_was_set_delayed()

[John Stultz]

This is a backport of f55a6faa384304c89cfef162768e88374d3312cb

clock_was_set() cannot be called from hard interrupt context because
it calls on_each_cpu().

For fixing the widely reported leap seconds issue it is necessary to
call it from hard interrupt context, i.e. the timer tick code, which
does the timekeeping updates.

Provide a new function which denotes it in the hrtimer cpu base
structure of the cpu on which it is called and raise the hrtimer
softirq. We then execute the clock_was_set() notificiation from
softirq context in run_hrtimer_softirq(). The hrtimer softirq is
rarely used, so polling the flag there is not a performance issue.

[ tglx: Made it depend on CONFIG_HIGH_RES_TIMERS. We really should get
  rid of all this ifdeffery ASAP ]

Signed-off-by: John Stultz <johnstul@us.ibm.com>
Reported-by: Jan Engelhardt <jengelh@inai.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Prarit Bhargava <prarit@redhat.com>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/1341960205-56738-2-git-send-email-johnstul@us.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 include/linux/hrtimer.h |    9 ++++++++-
 kernel/hrtimer.c        |   20 ++++++++++++++++++++
 2 files changed, 28 insertions(+), 1 deletion(-)

diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index fd0dc30..c9ec940 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -165,6 +165,7 @@ enum  hrtimer_base_type {
  * @lock:		lock protecting the base and associated clock bases
  *			and timers
  * @active_bases:	Bitfield to mark bases with active timers
+ * @clock_was_set:	Indicates that clock was set from irq context.
  * @expires_next:	absolute time of the next event which was scheduled
  *			via clock_set_next_event()
  * @hres_active:	State of high resolution mode
@@ -177,7 +178,8 @@ enum  hrtimer_base_type {
  */
 struct hrtimer_cpu_base {
 	raw_spinlock_t			lock;
-	unsigned long			active_bases;
+	unsigned int			active_bases;
+	unsigned int			clock_was_set;
 #ifdef CONFIG_HIGH_RES_TIMERS
 	ktime_t				expires_next;
 	int				hres_active;
@@ -286,6 +288,8 @@ extern void hrtimer_peek_ahead_timers(void);
 # define MONOTONIC_RES_NSEC	HIGH_RES_NSEC
 # define KTIME_MONOTONIC_RES	KTIME_HIGH_RES
 
+extern void clock_was_set_delayed(void);
+
 #else
 
 # define MONOTONIC_RES_NSEC	LOW_RES_NSEC
@@ -306,6 +310,9 @@ static inline int hrtimer_is_hres_active(struct hrtimer *timer)
 {
 	return 0;
 }
+
+static inline void clock_was_set_delayed(void) { }
+
 #endif
 
 extern void clock_was_set(void);
diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
index ae34bf5..3c24fb2 100644
--- a/kernel/hrtimer.c
+++ b/kernel/hrtimer.c
@@ -717,6 +717,19 @@ static int hrtimer_switch_to_hres(void)
 	return 1;
 }
 
+/*
+ * Called from timekeeping code to reprogramm the hrtimer interrupt
+ * device. If called from the timer interrupt context we defer it to
+ * softirq context.
+ */
+void clock_was_set_delayed(void)
+{
+	struct hrtimer_cpu_base *cpu_base = &__get_cpu_var(hrtimer_bases);
+
+	cpu_base->clock_was_set = 1;
+	__raise_softirq_irqoff(HRTIMER_SOFTIRQ);
+}
+
 #else
 
 static inline int hrtimer_hres_active(void) { return 0; }
@@ -1395,6 +1408,13 @@ void hrtimer_peek_ahead_timers(void)
 
 static void run_hrtimer_softirq(struct softirq_action *h)
 {
+	struct hrtimer_cpu_base *cpu_base = &__get_cpu_var(hrtimer_bases);
+
+	if (cpu_base->clock_was_set) {
+		cpu_base->clock_was_set = 0;
+		clock_was_set();
+	}
+
 	hrtimer_peek_ahead_timers();
 }
 
-- 
1.7.9.5

[PATCH 06/11] 3.2.x: timekeeping: Fix leapsecond triggered load spike issue

[John Stultz]

This is a backport of 4873fa070ae84a4115f0b3c9dfabc224f1bc7c51

The timekeeping code misses an update of the hrtimer subsystem after a
leap second happened. Due to that timers based on CLOCK_REALTIME are
either expiring a second early or late depending on whether a leap
second has been inserted or deleted until an operation is initiated
which causes that update. Unless the update happens by some other
means this discrepancy between the timekeeping and the hrtimer data
stays forever and timers are expired either early or late.

The reported immediate workaround - $ data -s "`date`" - is causing a
call to clock_was_set() which updates the hrtimer data structures.
See: http://www.sheeri.com/content/mysql-and-leap-second-high-cpu-and-fix

Add the missing clock_was_set() call to update_wall_time() in case of
a leap second event. The actual update is deferred to softirq context
as the necessary smp function call cannot be invoked from hard
interrupt context.

Signed-off-by: John Stultz <johnstul@us.ibm.com>
Reported-by: Jan Engelhardt <jengelh@inai.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Prarit Bhargava <prarit@redhat.com>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/1341960205-56738-3-git-send-email-johnstul@us.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 kernel/time/timekeeping.c |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 5d55185..8958ad7 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -941,6 +941,8 @@ static cycle_t logarithmic_accumulation(cycle_t offset, int shift)
 		leap = second_overflow(xtime.tv_sec);
 		xtime.tv_sec += leap;
 		wall_to_monotonic.tv_sec -= leap;
+		if (leap)
+			clock_was_set_delayed();
 	}
 
 	/* Accumulate raw time */
@@ -1052,6 +1054,8 @@ static void update_wall_time(void)
 		leap = second_overflow(xtime.tv_sec);
 		xtime.tv_sec += leap;
 		wall_to_monotonic.tv_sec -= leap;
+		if (leap)
+			clock_was_set_delayed();
 	}
 
 	timekeeping_update(false);
-- 
1.7.9.5

[PATCH 07/11] 3.2.x: timekeeping: Maintain ktime_t based offsets for hrtimers

[John Stultz]

From: Thomas Gleixner <tglx@linutronix.de>

This is a backport of 5b9fe759a678e05be4937ddf03d50e950207c1c0

We need to update the hrtimer clock offsets from the hrtimer interrupt
context. To avoid conversions from timespec to ktime_t maintain a
ktime_t based representation of those offsets in the timekeeper. This
puts the conversion overhead into the code which updates the
underlying offsets and provides fast accessible values in the hrtimer
interrupt.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Prarit Bhargava <prarit@redhat.com>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/1341960205-56738-4-git-send-email-johnstul@us.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 kernel/time/timekeeping.c |   25 ++++++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 8958ad7..d5d0e5d 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -161,18 +161,34 @@ static struct timespec xtime __attribute__ ((aligned (16)));
 static struct timespec wall_to_monotonic __attribute__ ((aligned (16)));
 static struct timespec total_sleep_time;
 
+/* Offset clock monotonic -> clock realtime */
+static ktime_t offs_real;
+
+/* Offset clock monotonic -> clock boottime */
+static ktime_t offs_boot;
+
 /*
  * The raw monotonic time for the CLOCK_MONOTONIC_RAW posix clock.
  */
 static struct timespec raw_time;
 
 /* must hold write on xtime_lock */
+static void update_rt_offset(void)
+{
+	struct timespec tmp, *wtm = &wall_to_monotonic;
+
+	set_normalized_timespec(&tmp, -wtm->tv_sec, -wtm->tv_nsec);
+	offs_real = timespec_to_ktime(tmp);
+}
+
+/* must hold write on xtime_lock */
 static void timekeeping_update(bool clearntp)
 {
 	if (clearntp) {
 		timekeeper.ntp_error = 0;
 		ntp_clear();
 	}
+	update_rt_offset();
 	update_vsyscall(&xtime, &wall_to_monotonic,
 			 timekeeper.clock, timekeeper.mult);
 }
@@ -587,6 +603,7 @@ void __init timekeeping_init(void)
 	}
 	set_normalized_timespec(&wall_to_monotonic,
 				-boot.tv_sec, -boot.tv_nsec);
+	update_rt_offset();
 	total_sleep_time.tv_sec = 0;
 	total_sleep_time.tv_nsec = 0;
 	write_sequnlock_irqrestore(&xtime_lock, flags);
@@ -595,6 +612,12 @@ void __init timekeeping_init(void)
 /* time in seconds when suspend began */
 static struct timespec timekeeping_suspend_time;
 
+static void update_sleep_time(struct timespec t)
+{
+	total_sleep_time = t;
+	offs_boot = timespec_to_ktime(t);
+}
+
 /**
  * __timekeeping_inject_sleeptime - Internal function to add sleep interval
  * @delta: pointer to a timespec delta value
@@ -612,7 +635,7 @@ static void __timekeeping_inject_sleeptime(struct timespec *delta)
 
 	xtime = timespec_add(xtime, *delta);
 	wall_to_monotonic = timespec_sub(wall_to_monotonic, *delta);
-	total_sleep_time = timespec_add(total_sleep_time, *delta);
+	update_sleep_time(timespec_add(total_sleep_time, *delta));
 }
 
 
-- 
1.7.9.5

[PATCH 08/11] 3.2.x: hrtimers: Move lock held region in hrtimer_interrupt()

[John Stultz]

From: Thomas Gleixner <tglx@linutronix.de>

This is a backport of 196951e91262fccda81147d2bcf7fdab08668b40

We need to update the base offsets from this code and we need to do
that under base->lock. Move the lock held region around the
ktime_get() calls. The ktime_get() calls are going to be replaced with
a function which gets the time and the offsets atomically.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Prarit Bhargava <prarit@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: John Stultz <johnstul@us.ibm.com>
Link: http://lkml.kernel.org/r/1341960205-56738-6-git-send-email-johnstul@us.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 kernel/hrtimer.c |    5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
index 3c24fb2..8f320af 100644
--- a/kernel/hrtimer.c
+++ b/kernel/hrtimer.c
@@ -1263,11 +1263,10 @@ void hrtimer_interrupt(struct clock_event_device *dev)
 	cpu_base->nr_events++;
 	dev->next_event.tv64 = KTIME_MAX;
 
+	raw_spin_lock(&cpu_base->lock);
 	entry_time = now = ktime_get();
 retry:
 	expires_next.tv64 = KTIME_MAX;
-
-	raw_spin_lock(&cpu_base->lock);
 	/*
 	 * We set expires_next to KTIME_MAX here with cpu_base->lock
 	 * held to prevent that a timer is enqueued in our queue via
@@ -1344,6 +1343,7 @@ retry:
 	 * interrupt routine. We give it 3 attempts to avoid
 	 * overreacting on some spurious event.
 	 */
+	raw_spin_lock(&cpu_base->lock);
 	now = ktime_get();
 	cpu_base->nr_retries++;
 	if (++retries < 3)
@@ -1356,6 +1356,7 @@ retry:
 	 */
 	cpu_base->nr_hangs++;
 	cpu_base->hang_detected = 1;
+	raw_spin_unlock(&cpu_base->lock);
 	delta = ktime_sub(now, entry_time);
 	if (delta.tv64 > cpu_base->max_hang_time.tv64)
 		cpu_base->max_hang_time = delta;
-- 
1.7.9.5

[PATCH 09/11] 3.2.x: timekeeping: Provide hrtimer update function

[John Stultz]

From: Thomas Gleixner <tglx@linutronix.de>

This is a backport of f6c06abfb3972ad4914cef57d8348fcb2932bc3b

To finally fix the infamous leap second issue and other race windows
caused by functions which change the offsets between the various time
bases (CLOCK_MONOTONIC, CLOCK_REALTIME and CLOCK_BOOTTIME) we need a
function which atomically gets the current monotonic time and updates
the offsets of CLOCK_REALTIME and CLOCK_BOOTTIME with minimalistic
overhead. The previous patch which provides ktime_t offsets allows us
to make this function almost as cheap as ktime_get() which is going to
be replaced in hrtimer_interrupt().

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Prarit Bhargava <prarit@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: John Stultz <johnstul@us.ibm.com>
Link: http://lkml.kernel.org/r/1341960205-56738-7-git-send-email-johnstul@us.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 include/linux/hrtimer.h   |    1 +
 kernel/time/timekeeping.c |   34 ++++++++++++++++++++++++++++++++++
 2 files changed, 35 insertions(+)

diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index c9ec940..cc07d27 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -327,6 +327,7 @@ 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);
+extern ktime_t ktime_get_update_offsets(ktime_t *offs_real, ktime_t *offs_boot);
 
 DECLARE_PER_CPU(struct tick_device, tick_cpu_device);
 
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index d5d0e5d..4938c5e 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1240,6 +1240,40 @@ void get_xtime_and_monotonic_and_sleep_offset(struct timespec *xtim,
 	} while (read_seqretry(&xtime_lock, seq));
 }
 
+#ifdef CONFIG_HIGH_RES_TIMERS
+/**
+ * ktime_get_update_offsets - hrtimer helper
+ * @real:	pointer to storage for monotonic -> realtime offset
+ * @_boot:	pointer to storage for monotonic -> boottime offset
+ *
+ * Returns current monotonic time and updates the offsets
+ * Called from hrtimer_interupt() or retrigger_next_event()
+ */
+ktime_t ktime_get_update_offsets(ktime_t *real, ktime_t *boot)
+{
+	ktime_t now;
+	unsigned int seq;
+	u64 secs, nsecs;
+
+	do {
+		seq = read_seqbegin(&xtime_lock);
+
+		secs = xtime.tv_sec;
+		nsecs = xtime.tv_nsec;
+		nsecs += timekeeping_get_ns();
+		/* If arch requires, add in gettimeoffset() */
+		nsecs += arch_gettimeoffset();
+
+		*real = offs_real;
+		*boot = offs_boot;
+	} while (read_seqretry(&xtime_lock, seq));
+
+	now = ktime_add_ns(ktime_set(secs, 0), nsecs);
+	now = ktime_sub(now, *real);
+	return now;
+}
+#endif
+
 /**
  * ktime_get_monotonic_offset() - get wall_to_monotonic in ktime_t format
  */
-- 
1.7.9.5

[PATCH 10/11] 3.2.x: hrtimer: Update hrtimer base offsets each hrtimer_interrupt

[John Stultz]

This is a backport of 5baefd6d84163443215f4a99f6a20f054ef11236

The update of the hrtimer base offsets on all cpus cannot be made
atomically from the timekeeper.lock held and interrupt disabled region
as smp function calls are not allowed there.

clock_was_set(), which enforces the update on all cpus, is called
either from preemptible process context in case of do_settimeofday()
or from the softirq context when the offset modification happened in
the timer interrupt itself due to a leap second.

In both cases there is a race window for an hrtimer interrupt between
dropping timekeeper lock, enabling interrupts and clock_was_set()
issuing the updates. Any interrupt which arrives in that window will
see the new time but operate on stale offsets.

So we need to make sure that an hrtimer interrupt always sees a
consistent state of time and offsets.

ktime_get_update_offsets() allows us to get the current monotonic time
and update the per cpu hrtimer base offsets from hrtimer_interrupt()
to capture a consistent state of monotonic time and the offsets. The
function replaces the existing ktime_get() calls in hrtimer_interrupt().

The overhead of the new function vs. ktime_get() is minimal as it just
adds two store operations.

This ensures that any changes to realtime or boottime offsets are
noticed and stored into the per-cpu hrtimer base structures, prior to
any hrtimer expiration and guarantees that timers are not expired early.

Signed-off-by: John Stultz <johnstul@us.ibm.com>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Prarit Bhargava <prarit@redhat.com>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/1341960205-56738-8-git-send-email-johnstul@us.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 kernel/hrtimer.c |   28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)

diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
index 8f320af..6db7a5e 100644
--- a/kernel/hrtimer.c
+++ b/kernel/hrtimer.c
@@ -657,6 +657,14 @@ static inline int hrtimer_enqueue_reprogram(struct hrtimer *timer,
 	return 0;
 }
 
+static inline ktime_t hrtimer_update_base(struct hrtimer_cpu_base *base)
+{
+	ktime_t *offs_real = &base->clock_base[HRTIMER_BASE_REALTIME].offset;
+	ktime_t *offs_boot = &base->clock_base[HRTIMER_BASE_BOOTTIME].offset;
+
+	return ktime_get_update_offsets(offs_real, offs_boot);
+}
+
 /*
  * Retrigger next event is called after clock was set
  *
@@ -665,22 +673,12 @@ static inline int hrtimer_enqueue_reprogram(struct hrtimer *timer,
 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_update_base(base);
 	hrtimer_force_reprogram(base, 0);
 	raw_spin_unlock(&base->lock);
 }
@@ -710,7 +708,6 @@ static int hrtimer_switch_to_hres(void)
 		base->clock_base[i].resolution = KTIME_HIGH_RES;
 
 	tick_setup_sched_timer();
-
 	/* "Retrigger" the interrupt to get things going */
 	retrigger_next_event(NULL);
 	local_irq_restore(flags);
@@ -1264,7 +1261,7 @@ void hrtimer_interrupt(struct clock_event_device *dev)
 	dev->next_event.tv64 = KTIME_MAX;
 
 	raw_spin_lock(&cpu_base->lock);
-	entry_time = now = ktime_get();
+	entry_time = now = hrtimer_update_base(cpu_base);
 retry:
 	expires_next.tv64 = KTIME_MAX;
 	/*
@@ -1342,9 +1339,12 @@ retry:
 	 * We need to prevent that we loop forever in the hrtimer
 	 * interrupt routine. We give it 3 attempts to avoid
 	 * overreacting on some spurious event.
+	 *
+	 * Acquire base lock for updating the offsets and retrieving
+	 * the current time.
 	 */
 	raw_spin_lock(&cpu_base->lock);
-	now = ktime_get();
+	now = hrtimer_update_base(cpu_base);
 	cpu_base->nr_retries++;
 	if (++retries < 3)
 		goto retry;
-- 
1.7.9.5

[PATCH 11/11] 3.2.x: timekeeping: Add missing update call in timekeeping_resume()

[John Stultz]

From: Thomas Gleixner <tglx@linutronix.de>

This is a backport of 3e997130bd2e8c6f5aaa49d6e3161d4d29b43ab0

The leap second rework unearthed another issue of inconsistent data.

On timekeeping_resume() the timekeeper data is updated, but nothing
calls timekeeping_update(), so now the update code in the timer
interrupt sees stale values.

This has been the case before those changes, but then the timer
interrupt was using stale data as well so this went unnoticed for quite
some time.

Add the missing update call, so all the data is consistent everywhere.

Reported-by: Andreas Schwab <schwab@linux-m68k.org>
Reported-and-tested-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Reported-and-tested-by: Martin Steigerwald <Martin@lichtvoll.de>
Cc: LKML <linux-kernel@vger.kernel.org>
Cc: Linux PM list <linux-pm@vger.kernel.org>
Cc: John Stultz <johnstul@us.ibm.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>,
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linux Kernel <linux-kernel@vger.kernel.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
---
 kernel/time/timekeeping.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 4938c5e..03e67d4 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -699,6 +699,7 @@ static void timekeeping_resume(void)
 	timekeeper.clock->cycle_last = timekeeper.clock->read(timekeeper.clock);
 	timekeeper.ntp_error = 0;
 	timekeeping_suspended = 0;
+	timekeeping_update(false);
 	write_sequnlock_irqrestore(&xtime_lock, flags);
 
 	touch_softlockup_watchdog();
-- 
1.7.9.5

Re: [PATCH 00/11] 3.2-stable: Fix for leapsecond caused hrtimer/futex issue

[John Stultz]

[-- Attachment #1: Type: text/plain, Size: 872 bytes --]

On 07/17/2012 12:05 AM, John Stultz wrote:
> 1) Deadlock leapsecond issue that a few reports described.
>
> I spent some time over the weekend trying to find a way to reproduce
> the hard-hang issue some folks were reporting after the leapsecond.
> Initially I didn't think the 6b43ae8a619d17 leap-second hrimter livelock
> patch needed to be backported since, I assumed it required the ntp_lock
> split for it to be triggered, but looking again I found that the same
> issue could occur prior to splitting out the ntp_lock. So I've backported
> that fix (and its follow-on fixups) as well as created a test case
> to reproduce the hard-hang deadlock.

Attached is the test case I used to reproduce and test the solution to 
the hard-hang deadlock.

WARNING: THIS TEST WILL LIKELY HARD LOCK YOUR BOX IN IRQ CONTEXT!
YOU MAY LOSE DATA! RUN AT YOUR OWN RISK!

thanks
-john


[-- Attachment #2: leapcrash.c --]
[-- Type: text/x-csrc, Size: 1928 bytes --]

/* Demo leapsecond deadlock
 *              by: john stultz (johnstul@us.ibm.com)
 *              (C) Copyright IBM 2012
 *              Licensed under the GPL
 *
 * This test demonstrates leapsecond deadlock that is possibe
 * on kernels from 2.6.26 to 3.3.
 *
 * WARNING: THIS WILL LIKELY HARDHANG SYSTEMS AND MAY LOSE DATA
 * RUN AT YOUR OWN RISK!
 *  To build:
 *	$ gcc leapcrash.c -o leapcrash -lrt
 */



#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <sys/time.h>
#include <sys/timex.h>
#include <string.h>
#include <signal.h>



/* clear NTP time_status & time_state */
void clear_time_state(void)
{
	struct timex tx;
	int ret;

	/*
	 * XXX - The fact we have to call this twice seems
	 * to point to a slight issue in the kernel's ntp state
	 * managment. Needs to be investigated further.
	 */

	tx.modes = ADJ_STATUS;
	tx.status = STA_PLL;
	ret = adjtimex(&tx);

	tx.modes = ADJ_STATUS;
	tx.status = 0;
	ret = adjtimex(&tx);
}

/* Make sure we cleanup on ctrl-c */
void handler(int unused)
{
	clear_time_state();
	exit(0);
}


int  main(void)
{
	struct timex tx;
	struct timespec ts;
	time_t next_leap;
	int count =0;

	setbuf(stdout, NULL);

	signal(SIGINT, handler);
	signal(SIGKILL, handler);
	printf("This runs continuously. Press ctrl-c to stop\n");

	clear_time_state();


	/* Get the current time */
	clock_gettime(CLOCK_REALTIME, &ts);

	/* Calculate the next possible leap second 23:59:60 GMT */
	next_leap = ts.tv_sec;
	next_leap += 86400 - (next_leap % 86400);

	while (1) {
		struct timeval tv;


		/* set the time to 2 seconds before the leap */
		tv.tv_sec = next_leap - 2;
		tv.tv_usec = 0;
		settimeofday(&tv, NULL);

		adjtimex(&tx);

		/* hammer on adjtime w/ STA_INS */
		while (tx.time.tv_sec < next_leap + 1) {
			/* Set the leap second insert flag */
			tx.modes = ADJ_STATUS;
			tx.status = STA_INS;
			adjtimex(&tx);
		}
		clear_time_state();
		printf(".");
	}
	return 0;
}

Re: [PATCH 00/11] 3.2-stable: Fix for leapsecond caused hrtimer/futex issue

[Christoph Biedl]

John Stultz wrote...

> Attached is the test case I used to reproduce and test the solution
> to the hard-hang deadlock.

I was wondering whether anybody managed to crash a virtualbox guest
using your program. No avail, using version 4.1.18 on the host and the
guest kernel running several 3.0.x (x < 38) kernels on both x32 and
x64, the guest utilies were stopped. Rather a fun fact I guess but I
wanted to let you know.

All real hardware tested, including a dockstar on armel, crashed as
predicted, while 3.0.38-rc1 was immune.

    Christoph

Re: [PATCH 00/11] 3.2-stable: Fix for leapsecond caused hrtimer/futex issue

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 290 bytes --]

On Tue, 2012-07-17 at 03:05 -0400, John Stultz wrote:
> Here is backport of the leapsecond fixes to 3.2-stable. These are less
> straight forward, and should get closer review.
[...]

All queued up for 3.2, thanks.

Ben.

-- 
Ben Hutchings
73.46% of all statistics are made up.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: [PATCH 00/11] 3.2-stable: Fix for leapsecond caused hrtimer/futex issue

[John Stultz]

On 07/19/2012 01:48 PM, Christoph Biedl wrote:
> John Stultz wrote...
>
>> Attached is the test case I used to reproduce and test the solution
>> to the hard-hang deadlock.
> I was wondering whether anybody managed to crash a virtualbox guest
> using your program. No avail, using version 4.1.18 on the host and the
> guest kernel running several 3.0.x (x < 38) kernels on both x32 and
> x64, the guest utilies were stopped. Rather a fun fact I guess but I
> wanted to let you know.

I've been able to crash a kvm guest with an unpatched kernel with my 
test.  The issue requires that the adding of the hrtimer causes the 
clockevent to be reprogrammed. This usually happens if there's no timers 
that expire sooner then the leapsecond timer. So if there are drivers 
that set frequent timers, or set timers right before the leapsecond, it 
may be difficult to trigger this issue.

Lowering HZ or adding more vcpus might help if you really want to be 
able to trigger the issue.


> All real hardware tested, including a dockstar on armel, crashed as
> predicted, while 3.0.38-rc1 was immune.
Great to hear!

thanks
-john

Re: [PATCH 00/11] 3.2-stable: Fix for leapsecond caused hrtimer/futex issue

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1336 bytes --]

On Mon, 2012-07-23 at 12:51 -0700, John Stultz wrote:
> On 07/19/2012 01:48 PM, Christoph Biedl wrote:
> > John Stultz wrote...
> >
> >> Attached is the test case I used to reproduce and test the solution
> >> to the hard-hang deadlock.
> > I was wondering whether anybody managed to crash a virtualbox guest
> > using your program. No avail, using version 4.1.18 on the host and the
> > guest kernel running several 3.0.x (x < 38) kernels on both x32 and
> > x64, the guest utilies were stopped. Rather a fun fact I guess but I
> > wanted to let you know.
> 
> I've been able to crash a kvm guest with an unpatched kernel with my 
> test.  The issue requires that the adding of the hrtimer causes the 
> clockevent to be reprogrammed. This usually happens if there's no timers 
> that expire sooner then the leapsecond timer. So if there are drivers 
> that set frequent timers, or set timers right before the leapsecond, it 
> may be difficult to trigger this issue.
> 
> Lowering HZ or adding more vcpus might help if you really want to be 
> able to trigger the issue.
[...]

Your test program also made Linux 3.2.23 (or it may have been .21)
lock-up for me in a KVM guest, while 3.2.24-rc1 seemed immune.

Ben.

-- 
Ben Hutchings
If more than one person is responsible for a bug, no one is at fault.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]