Merge branch 'timers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'timers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (34 commits)
  time: Prevent 32 bit overflow with set_normalized_timespec()
  clocksource: Delay clocksource down rating to late boot
  clocksource: clocksource_select must be called with mutex locked
  clocksource: Resolve cpu hotplug dead lock with TSC unstable, fix crash
  timers: Drop a function prototype
  clocksource: Resolve cpu hotplug dead lock with TSC unstable
  timer.c: Fix S/390 comments
  timekeeping: Fix invalid getboottime() value
  timekeeping: Fix up read_persistent_clock() breakage on sh
  timekeeping: Increase granularity of read_persistent_clock(), build fix
  time: Introduce CLOCK_REALTIME_COARSE
  x86: Do not unregister PIT clocksource on PIT oneshot setup/shutdown
  clocksource: Avoid clocksource watchdog circular locking dependency
  clocksource: Protect the watchdog rating changes with clocksource_mutex
  clocksource: Call clocksource_change_rating() outside of watchdog_lock
  timekeeping: Introduce read_boot_clock
  timekeeping: Increase granularity of read_persistent_clock()
  timekeeping: Update clocksource with stop_machine
  timekeeping: Add timekeeper read_clock helper functions
  timekeeping: Move NTP adjusted clock multiplier to struct timekeeper
  ...

Fix trivial conflict due to MIPS lemote -> loongson renaming.

20 files changed, 118 insertions, 94 deletions

diff --git a/arch/arm/plat-omap/common.c b/arch/arm/plat-omap/common.c
index ebcf0064..95587b6 100644
--- a/arch/arm/plat-omap/common.c
+++ b/arch/arm/plat-omap/common.c
@@ -253,11 +253,8 @@ static struct clocksource clocksource_32k = {
  */
 unsigned long long sched_clock(void)
 {
-	unsigned long long ret;
-
-	ret = (unsigned long long)clocksource_32k.read(&clocksource_32k);
-	ret = (ret * clocksource_32k.mult_orig) >> clocksource_32k.shift;
-	return ret;
+	return clocksource_cyc2ns(clocksource_32k.read(&clocksource_32k),
+				  clocksource_32k.mult, clocksource_32k.shift);
 }
 
 static int __init omap_init_clocksource_32k(void)
diff --git a/arch/m68knommu/kernel/time.c b/arch/m68knommu/kernel/time.c
index c2aa717..a90acf5 100644
--- a/arch/m68knommu/kernel/time.c
+++ b/arch/m68knommu/kernel/time.c
@@ -72,9 +72,10 @@ static unsigned long read_rtc_mmss(void)
 	return  mktime(year, mon, day, hour, min, sec);
 }
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
-	return read_rtc_mmss();
+	ts->tv_sec = read_rtc_mmss();
+	ts->tv_nsec = 0;
 }
 
 int update_persistent_clock(struct timespec now)
diff --git a/arch/mips/dec/time.c b/arch/mips/dec/time.c
index 463136e..02f505f 100644
--- a/arch/mips/dec/time.c
+++ b/arch/mips/dec/time.c
@@ -18,7 +18,7 @@
 #include <asm/dec/ioasic.h>
 #include <asm/dec/machtype.h>
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
 	unsigned int year, mon, day, hour, min, sec, real_year;
 	unsigned long flags;
@@ -53,7 +53,8 @@ unsigned long read_persistent_clock(void)
 
 	year += real_year - 72 + 2000;
 
-	return mktime(year, mon, day, hour, min, sec);
+	ts->tv_sec = mktime(year, mon, day, hour, min, sec);
+	ts->tv_nsec = 0;
 }
 
 /*
diff --git a/arch/mips/lasat/ds1603.c b/arch/mips/lasat/ds1603.c
index 52cb143..c6fd96f 100644
--- a/arch/mips/lasat/ds1603.c
+++ b/arch/mips/lasat/ds1603.c
@@ -135,7 +135,7 @@ static void rtc_end_op(void)
 	lasat_ndelay(1000);
 }
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
 	unsigned long word;
 	unsigned long flags;
@@ -147,7 +147,8 @@ unsigned long read_persistent_clock(void)
 	rtc_end_op();
 	spin_unlock_irqrestore(&rtc_lock, flags);
 
-	return word;
+	ts->tv_sec = word;
+	ts->tv_nsec = 0;
 }
 
 int rtc_mips_set_mmss(unsigned long time)
diff --git a/arch/mips/lasat/sysctl.c b/arch/mips/lasat/sysctl.c
index 8f88886..3f04d4c 100644
--- a/arch/mips/lasat/sysctl.c
+++ b/arch/mips/lasat/sysctl.c
@@ -92,10 +92,12 @@ static int rtctmp;
 int proc_dolasatrtc(ctl_table *table, int write, struct file *filp,
 		       void *buffer, size_t *lenp, loff_t *ppos)
 {
+	struct timespec ts;
 	int r;
 
 	if (!write) {
-		rtctmp = read_persistent_clock();
+		read_persistent_clock(&ts);
+		rtctmp = ts.tv_sec;
 		/* check for time < 0 and set to 0 */
 		if (rtctmp < 0)
 			rtctmp = 0;
@@ -134,9 +136,11 @@ int sysctl_lasat_rtc(ctl_table *table,
 		    void *oldval, size_t *oldlenp,
 		    void *newval, size_t newlen)
 {
+	struct timespec ts;
 	int r;
 
-	rtctmp = read_persistent_clock();
+	read_persistent_clock(&ts);
+	rtctmp = ts.tv_sec;
 	if (rtctmp < 0)
 		rtctmp = 0;
 	r = sysctl_intvec(table, oldval, oldlenp, newval, newlen);
diff --git a/arch/mips/loongson/common/time.c b/arch/mips/loongson/common/time.c
index b13d171..0edbef3 100644
--- a/arch/mips/loongson/common/time.c
+++ b/arch/mips/loongson/common/time.c
@@ -21,7 +21,8 @@ void __init plat_time_init(void)
 	mips_hpt_frequency = cpu_clock_freq / 2;
 }
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
-	return mc146818_get_cmos_time();
+	ts->tv_sec = return mc146818_get_cmos_time();
+	ts->tv_nsec = 0;
 }
diff --git a/arch/mips/mti-malta/malta-time.c b/arch/mips/mti-malta/malta-time.c
index 0b97d47..3c6f190 100644
--- a/arch/mips/mti-malta/malta-time.c
+++ b/arch/mips/mti-malta/malta-time.c
@@ -100,9 +100,10 @@ static unsigned int __init estimate_cpu_frequency(void)
 	return count;
 }
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
-	return mc146818_get_cmos_time();
+	ts->tv_sec = mc146818_get_cmos_time();
+	ts->tv_nsec = 0;
 }
 
 static void __init plat_perf_setup(void)
diff --git a/arch/mips/pmc-sierra/yosemite/setup.c b/arch/mips/pmc-sierra/yosemite/setup.c
index 2d3c0dc..3498ac9 100644
--- a/arch/mips/pmc-sierra/yosemite/setup.c
+++ b/arch/mips/pmc-sierra/yosemite/setup.c
@@ -70,7 +70,7 @@ void __init bus_error_init(void)
 }
 
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
 	unsigned int year, month, day, hour, min, sec;
 	unsigned long flags;
@@ -92,7 +92,8 @@ unsigned long read_persistent_clock(void)
 	m48t37_base->control = 0x00;
 	spin_unlock_irqrestore(&rtc_lock, flags);
 
-	return mktime(year, month, day, hour, min, sec);
+	ts->tv_sec = mktime(year, month, day, hour, min, sec);
+	ts->tv_nsec = 0;
 }
 
 int rtc_mips_set_time(unsigned long tim)
diff --git a/arch/mips/sibyte/swarm/setup.c b/arch/mips/sibyte/swarm/setup.c
index 672e45d..623ffc9 100644
--- a/arch/mips/sibyte/swarm/setup.c
+++ b/arch/mips/sibyte/swarm/setup.c
@@ -87,19 +87,26 @@ enum swarm_rtc_type {
 
 enum swarm_rtc_type swarm_rtc_type;
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
+	unsigned long sec;
+
 	switch (swarm_rtc_type) {
 	case RTC_XICOR:
-		return xicor_get_time();
+		sec = xicor_get_time();
+		break;
 
 	case RTC_M4LT81:
-		return m41t81_get_time();
+		sec = m41t81_get_time();
+		break;
 
 	case RTC_NONE:
 	default:
-		return mktime(2000, 1, 1, 0, 0, 0);
+		sec = mktime(2000, 1, 1, 0, 0, 0);
+		break;
 	}
+	ts->tv_sec = sec;
+	tv->tv_nsec = 0;
 }
 
 int rtc_mips_set_time(unsigned long sec)
diff --git a/arch/mips/sni/time.c b/arch/mips/sni/time.c
index 0d9ec1a..62df6a5 100644
--- a/arch/mips/sni/time.c
+++ b/arch/mips/sni/time.c
@@ -182,7 +182,8 @@ void __init plat_time_init(void)
 	setup_pit_timer();
 }
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
-	return -1;
+	ts->tv_sec = -1;
+	ts->tv_nsec = 0;
 }
diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
index a180b4f..465e498 100644
--- a/arch/powerpc/kernel/time.c
+++ b/arch/powerpc/kernel/time.c
@@ -774,11 +774,12 @@ int update_persistent_clock(struct timespec now)
 	return ppc_md.set_rtc_time(&tm);
 }
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
 	struct rtc_time tm;
 	static int first = 1;
 
+	ts->tv_nsec = 0;
 	/* XXX this is a litle fragile but will work okay in the short term */
 	if (first) {
 		first = 0;
@@ -786,14 +787,18 @@ unsigned long read_persistent_clock(void)
 			timezone_offset = ppc_md.time_init();
 
 		/* get_boot_time() isn't guaranteed to be safe to call late */
-		if (ppc_md.get_boot_time)
-			return ppc_md.get_boot_time() -timezone_offset;
+		if (ppc_md.get_boot_time) {
+			ts->tv_sec = ppc_md.get_boot_time() - timezone_offset;
+			return;
+		}
+	}
+	if (!ppc_md.get_rtc_time) {
+		ts->tv_sec = 0;
+		return;
 	}
-	if (!ppc_md.get_rtc_time)
-		return 0;
 	ppc_md.get_rtc_time(&tm);
-	return mktime(tm.tm_year+1900, tm.tm_mon+1, tm.tm_mday,
-		      tm.tm_hour, tm.tm_min, tm.tm_sec);
+	ts->tv_sec = mktime(tm.tm_year+1900, tm.tm_mon+1, tm.tm_mday,
+			    tm.tm_hour, tm.tm_min, tm.tm_sec);
 }
 
 /* clocksource code */
diff --git a/arch/s390/kernel/time.c b/arch/s390/kernel/time.c
index e3dc28b..34162a0 100644
--- a/arch/s390/kernel/time.c
+++ b/arch/s390/kernel/time.c
@@ -184,12 +184,14 @@ static void timing_alert_interrupt(__u16 code)
 static void etr_reset(void);
 static void stp_reset(void);
 
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
-	struct timespec ts;
+	tod_to_timeval(get_clock() - TOD_UNIX_EPOCH, ts);
+}
 
-	tod_to_timeval(get_clock() - TOD_UNIX_EPOCH, &ts);
-	return ts.tv_sec;
+void read_boot_clock(struct timespec *ts)
+{
+	tod_to_timeval(sched_clock_base_cc - TOD_UNIX_EPOCH, ts);
 }
 
 static cycle_t read_tod_clock(struct clocksource *cs)
@@ -207,6 +209,10 @@ static struct clocksource clocksource_tod = {
 	.flags		= CLOCK_SOURCE_IS_CONTINUOUS,
 };
 
+struct clocksource * __init clocksource_default_clock(void)
+{
+	return &clocksource_tod;
+}
 
 void update_vsyscall(struct timespec *wall_time, struct clocksource *clock)
 {
@@ -244,10 +250,6 @@ void update_vsyscall_tz(void)
  */
 void __init time_init(void)
 {
-	struct timespec ts;
-	unsigned long flags;
-	cycle_t now;
-
 	/* Reset time synchronization interfaces. */
 	etr_reset();
 	stp_reset();
@@ -263,26 +265,6 @@ void __init time_init(void)
 	if (clocksource_register(&clocksource_tod) != 0)
 		panic("Could not register TOD clock source");
 
-	/*
-	 * The TOD clock is an accurate clock. The xtime should be
-	 * initialized in a way that the difference between TOD and
-	 * xtime is reasonably small. Too bad that timekeeping_init
-	 * sets xtime.tv_nsec to zero. In addition the clock source
-	 * change from the jiffies clock source to the TOD clock
-	 * source add another error of up to 1/HZ second. The same
-	 * function sets wall_to_monotonic to a value that is too
-	 * small for /proc/uptime to be accurate.
-	 * Reset xtime and wall_to_monotonic to sane values.
-	 */
-	write_seqlock_irqsave(&xtime_lock, flags);
-	now = get_clock();
-	tod_to_timeval(now - TOD_UNIX_EPOCH, &xtime);
-	clocksource_tod.cycle_last = now;
-	clocksource_tod.raw_time = xtime;
-	tod_to_timeval(sched_clock_base_cc - TOD_UNIX_EPOCH, &ts);
-	set_normalized_timespec(&wall_to_monotonic, -ts.tv_sec, -ts.tv_nsec);
-	write_sequnlock_irqrestore(&xtime_lock, flags);
-
 	/* Enable TOD clock interrupts on the boot cpu. */
 	init_cpu_timer();
 
diff --git a/arch/sh/kernel/time.c b/arch/sh/kernel/time.c
index 9b352a1..0e0e858 100644
--- a/arch/sh/kernel/time.c
+++ b/arch/sh/kernel/time.c
@@ -39,11 +39,9 @@ void (*rtc_sh_get_time)(struct timespec *) = null_rtc_get_time;
 int (*rtc_sh_set_time)(const time_t) = null_rtc_set_time;
 
 #ifdef CONFIG_GENERIC_CMOS_UPDATE
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
-	struct timespec tv;
-	rtc_sh_get_time(&tv);
-	return tv.tv_sec;
+	rtc_sh_get_time(ts);
 }
 
 int update_persistent_clock(struct timespec now)
diff --git a/arch/x86/include/asm/vgtod.h b/arch/x86/include/asm/vgtod.h
index dc27a69..3d61e20 100644
--- a/arch/x86/include/asm/vgtod.h
+++ b/arch/x86/include/asm/vgtod.h
@@ -21,6 +21,7 @@ struct vsyscall_gtod_data {
 		u32	shift;
 	} clock;
 	struct timespec wall_to_monotonic;
+	struct timespec wall_time_coarse;
 };
 extern struct vsyscall_gtod_data __vsyscall_gtod_data
 __section_vsyscall_gtod_data;
diff --git a/arch/x86/kernel/i8253.c b/arch/x86/kernel/i8253.c
index 5cf36c0..23c1679 100644
--- a/arch/x86/kernel/i8253.c
+++ b/arch/x86/kernel/i8253.c
@@ -19,12 +19,6 @@
 DEFINE_SPINLOCK(i8253_lock);
 EXPORT_SYMBOL(i8253_lock);
 
-#ifdef CONFIG_X86_32
-static void pit_disable_clocksource(void);
-#else
-static inline void pit_disable_clocksource(void) { }
-#endif
-
 /*
  * HPET replaces the PIT, when enabled. So we need to know, which of
  * the two timers is used
@@ -57,12 +51,10 @@ static void init_pit_timer(enum clock_event_mode mode,
 			outb_pit(0, PIT_CH0);
 			outb_pit(0, PIT_CH0);
 		}
-		pit_disable_clocksource();
 		break;
 
 	case CLOCK_EVT_MODE_ONESHOT:
 		/* One shot setup */
-		pit_disable_clocksource();
 		outb_pit(0x38, PIT_MODE);
 		break;
 
@@ -200,17 +192,6 @@ static struct clocksource pit_cs = {
 	.shift		= 20,
 };
 
-static void pit_disable_clocksource(void)
-{
-	/*
-	 * Use mult to check whether it is registered or not
-	 */
-	if (pit_cs.mult) {
-		clocksource_unregister(&pit_cs);
-		pit_cs.mult = 0;
-	}
-}
-
 static int __init init_pit_clocksource(void)
 {
 	 /*
diff --git a/arch/x86/kernel/rtc.c b/arch/x86/kernel/rtc.c
index 5d465b2..bf67dcb 100644
--- a/arch/x86/kernel/rtc.c
+++ b/arch/x86/kernel/rtc.c
@@ -178,7 +178,7 @@ static int set_rtc_mmss(unsigned long nowtime)
 }
 
 /* not static: needed by APM */
-unsigned long read_persistent_clock(void)
+void read_persistent_clock(struct timespec *ts)
 {
 	unsigned long retval, flags;
 
@@ -186,7 +186,8 @@ unsigned long read_persistent_clock(void)
 	retval = get_wallclock();
 	spin_unlock_irqrestore(&rtc_lock, flags);
 
-	return retval;
+	ts->tv_sec = retval;
+	ts->tv_nsec = 0;
 }
 
 int update_persistent_clock(struct timespec now)
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index 71f4368..fc3672a 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -744,10 +744,16 @@ static cycle_t __vsyscall_fn vread_tsc(void)
 }
 #endif
 
+static void resume_tsc(void)
+{
+	clocksource_tsc.cycle_last = 0;
+}
+
 static struct clocksource clocksource_tsc = {
 	.name                   = "tsc",
 	.rating                 = 300,
 	.read                   = read_tsc,
+	.resume			= resume_tsc,
 	.mask                   = CLOCKSOURCE_MASK(64),
 	.shift                  = 22,
 	.flags                  = CLOCK_SOURCE_IS_CONTINUOUS |
@@ -761,12 +767,14 @@ void mark_tsc_unstable(char *reason)
 {
 	if (!tsc_unstable) {
 		tsc_unstable = 1;
-		printk("Marking TSC unstable due to %s\n", reason);
+		printk(KERN_INFO "Marking TSC unstable due to %s\n", reason);
 		/* Change only the rating, when not registered */
 		if (clocksource_tsc.mult)
-			clocksource_change_rating(&clocksource_tsc, 0);
-		else
+			clocksource_mark_unstable(&clocksource_tsc);
+		else {
+			clocksource_tsc.flags |= CLOCK_SOURCE_UNSTABLE;
 			clocksource_tsc.rating = 0;
+		}
 	}
 }
 
diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c
index 25ee06a..cf53a78 100644
--- a/arch/x86/kernel/vsyscall_64.c
+++ b/arch/x86/kernel/vsyscall_64.c
@@ -87,6 +87,7 @@ void update_vsyscall(struct timespec *wall_time, struct clocksource *clock)
 	vsyscall_gtod_data.wall_time_sec = wall_time->tv_sec;
 	vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
 	vsyscall_gtod_data.wall_to_monotonic = wall_to_monotonic;
+	vsyscall_gtod_data.wall_time_coarse = __current_kernel_time();
 	write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
 }
 
diff --git a/arch/x86/vdso/vclock_gettime.c b/arch/x86/vdso/vclock_gettime.c
index 6a40b78..ee55754 100644
--- a/arch/x86/vdso/vclock_gettime.c
+++ b/arch/x86/vdso/vclock_gettime.c
@@ -86,14 +86,47 @@ notrace static noinline int do_monotonic(struct timespec *ts)
 	return 0;
 }
 
+notrace static noinline int do_realtime_coarse(struct timespec *ts)
+{
+	unsigned long seq;
+	do {
+		seq = read_seqbegin(&gtod->lock);
+		ts->tv_sec = gtod->wall_time_coarse.tv_sec;
+		ts->tv_nsec = gtod->wall_time_coarse.tv_nsec;
+	} while (unlikely(read_seqretry(&gtod->lock, seq)));
+	return 0;
+}
+
+notrace static noinline int do_monotonic_coarse(struct timespec *ts)
+{
+	unsigned long seq, ns, secs;
+	do {
+		seq = read_seqbegin(&gtod->lock);
+		secs = gtod->wall_time_coarse.tv_sec;
+		ns = gtod->wall_time_coarse.tv_nsec;
+		secs += gtod->wall_to_monotonic.tv_sec;
+		ns += gtod->wall_to_monotonic.tv_nsec;
+	} while (unlikely(read_seqretry(&gtod->lock, seq)));
+	vset_normalized_timespec(ts, secs, ns);
+	return 0;
+}
+
 notrace int __vdso_clock_gettime(clockid_t clock, struct timespec *ts)
 {
-	if (likely(gtod->sysctl_enabled && gtod->clock.vread))
+	if (likely(gtod->sysctl_enabled))
 		switch (clock) {
 		case CLOCK_REALTIME:
-			return do_realtime(ts);
+			if (likely(gtod->clock.vread))
+				return do_realtime(ts);
+			break;
 		case CLOCK_MONOTONIC:
-			return do_monotonic(ts);
+			if (likely(gtod->clock.vread))
+				return do_monotonic(ts);
+			break;
+		case CLOCK_REALTIME_COARSE:
+			return do_realtime_coarse(ts);
+		case CLOCK_MONOTONIC_COARSE:
+			return do_monotonic_coarse(ts);
 		}
 	return vdso_fallback_gettime(clock, ts);
 }
diff --git a/arch/xtensa/kernel/time.c b/arch/xtensa/kernel/time.c
index 8848120..19085ff 100644
--- a/arch/xtensa/kernel/time.c
+++ b/arch/xtensa/kernel/time.c
@@ -59,9 +59,8 @@ static struct irqaction timer_irqaction = {
 
 void __init time_init(void)
 {
-	xtime.tv_nsec = 0;
-	xtime.tv_sec = read_persistent_clock();
-
+	/* FIXME: xtime&wall_to_monotonic are set in timekeeping_init. */
+	read_persistent_clock(&xtime);
 	set_normalized_timespec(&wall_to_monotonic,
 		-xtime.tv_sec, -xtime.tv_nsec);