More time fixes. Now time is set properly in domU's, and

settimeofday() is simplified and fixed.
Signed-off-by: Keir Fraser <keir@xensource.com>

4 files changed, 64 insertions, 67 deletions

diff --git a/linux-2.6-xen-sparse/arch/xen/i386/kernel/time.c b/linux-2.6-xen-sparse/arch/xen/i386/kernel/time.c
index 4e30c2ec83..6b2fe1ccc5 100644
--- a/linux-2.6-xen-sparse/arch/xen/i386/kernel/time.c
+++ b/linux-2.6-xen-sparse/arch/xen/i386/kernel/time.c
@@ -227,33 +227,20 @@ static unsigned long get_usec_offset(struct shadow_time_info *shadow)
 	return scale_delta(delta, shadow->tsc_to_usec_mul, shadow->tsc_shift);
 }
 
-static void update_wallclock(void)
+static void __update_wallclock(time_t sec, long nsec)
 {
-	shared_info_t *s = HYPERVISOR_shared_info;
 	long wtm_nsec, xtime_nsec;
 	time_t wtm_sec, xtime_sec;
-	u64 tmp, nsec;
-
-	do {
-		shadow_tv_version = s->wc_version;
-		rmb();
-		shadow_tv.tv_sec  = s->wc_sec;
-		shadow_tv.tv_nsec = s->wc_nsec;
-		rmb();
-	}
-	while ((s->wc_version & 1) | (shadow_tv_version ^ s->wc_version));
-
-	if (independent_wallclock)
-		return;
+	u64 tmp, wc_nsec;
 
 	/* Adjust wall-clock time base based on wall_jiffies ticks. */
-	nsec = processed_system_time;
-	nsec += (u64)shadow_tv.tv_sec * 1000000000ULL;
-	nsec += (u64)shadow_tv.tv_nsec;
-	nsec -= (jiffies - wall_jiffies) * (u64)(NSEC_PER_SEC / HZ);
+	wc_nsec = processed_system_time;
+	wc_nsec += (u64)sec * 1000000000ULL;
+	wc_nsec += (u64)nsec;
+	wc_nsec -= (jiffies - wall_jiffies) * (u64)(NSEC_PER_SEC / HZ);
 
 	/* Split wallclock base into seconds and nanoseconds. */
-	tmp = nsec;
+	tmp = wc_nsec;
 	xtime_nsec = do_div(tmp, 1000000000);
 	xtime_sec  = (time_t)tmp;
 
@@ -262,6 +249,28 @@ static void update_wallclock(void)
 
 	set_normalized_timespec(&xtime, xtime_sec, xtime_nsec);
 	set_normalized_timespec(&wall_to_monotonic, wtm_sec, wtm_nsec);
+
+	time_adjust = 0;		/* stop active adjtime() */
+	time_status |= STA_UNSYNC;
+	time_maxerror = NTP_PHASE_LIMIT;
+	time_esterror = NTP_PHASE_LIMIT;
+}
+
+static void update_wallclock(void)
+{
+	shared_info_t *s = HYPERVISOR_shared_info;
+
+	do {
+		shadow_tv_version = s->wc_version;
+		rmb();
+		shadow_tv.tv_sec  = s->wc_sec;
+		shadow_tv.tv_nsec = s->wc_nsec;
+		rmb();
+	}
+	while ((s->wc_version & 1) | (shadow_tv_version ^ s->wc_version));
+
+	if (!independent_wallclock)
+		__update_wallclock(shadow_tv.tv_sec, shadow_tv.tv_nsec);
 }
 
 /*
@@ -408,19 +417,15 @@ EXPORT_SYMBOL(do_gettimeofday);
 
 int do_settimeofday(struct timespec *tv)
 {
-	time_t wtm_sec, sec = tv->tv_sec;
-	long wtm_nsec;
+	time_t sec;
 	s64 nsec;
-	struct timespec xentime;
 	unsigned int cpu;
 	struct shadow_time_info *shadow;
+	dom0_op_t op;
 
 	if ((unsigned long)tv->tv_nsec >= NSEC_PER_SEC)
 		return -EINVAL;
 
-	if (!independent_wallclock && !(xen_start_info.flags & SIF_INITDOMAIN))
-		return 0; /* Silent failure? */
-
 	cpu = get_cpu();
 	shadow = &per_cpu(shadow_time, cpu);
 
@@ -431,51 +436,30 @@ int do_settimeofday(struct timespec *tv)
 	 * overflows. If that were to happen then our shadow time values would
 	 * be stale, so we can retry with fresh ones.
 	 */
- again:
-	nsec = (s64)tv->tv_nsec - (s64)get_nsec_offset(shadow);
-	if (unlikely(!time_values_up_to_date(cpu))) {
+	for ( ; ; ) {
+		nsec = (s64)tv->tv_nsec - (s64)get_nsec_offset(shadow);
+		if (time_values_up_to_date(cpu))
+			break;
 		get_time_values_from_xen();
-		goto again;
 	}
-
+	sec = tv->tv_sec;
 	__normalize_time(&sec, &nsec);
-	set_normalized_timespec(&xentime, sec, nsec);
-
-	/*
-	 * This is revolting. We need to set "xtime" correctly. However, the
-	 * value in this location is the value at the most recent update of
-	 * wall time.  Discover what correction gettimeofday() would have
-	 * made, and then undo it!
-	 */
-	nsec -= (jiffies - wall_jiffies) * TICK_NSEC;
-
-	nsec -= (shadow->system_timestamp - processed_system_time);
-
-	__normalize_time(&sec, &nsec);
-	wtm_sec  = wall_to_monotonic.tv_sec + (xtime.tv_sec - sec);
-	wtm_nsec = wall_to_monotonic.tv_nsec + (xtime.tv_nsec - nsec);
-
-	set_normalized_timespec(&xtime, sec, nsec);
-	set_normalized_timespec(&wall_to_monotonic, wtm_sec, wtm_nsec);
 
-	time_adjust = 0;		/* stop active adjtime() */
-	time_status |= STA_UNSYNC;
-	time_maxerror = NTP_PHASE_LIMIT;
-	time_esterror = NTP_PHASE_LIMIT;
-
-#ifdef CONFIG_XEN_PRIVILEGED_GUEST
 	if ((xen_start_info.flags & SIF_INITDOMAIN) &&
 	    !independent_wallclock) {
-		dom0_op_t op;
 		op.cmd = DOM0_SETTIME;
-		op.u.settime.secs        = xentime.tv_sec;
-		op.u.settime.nsecs       = xentime.tv_nsec;
+		op.u.settime.secs        = sec;
+		op.u.settime.nsecs       = nsec;
 		op.u.settime.system_time = shadow->system_timestamp;
-		write_sequnlock_irq(&xtime_lock);
 		HYPERVISOR_dom0_op(&op);
-	} else
-#endif
-		write_sequnlock_irq(&xtime_lock);
+		update_wallclock();
+	} else if (independent_wallclock) {
+		nsec -= shadow->system_timestamp;
+		__normalize_time(&sec, &nsec);
+		__update_wallclock(sec, nsec);
+	}
+
+	write_sequnlock_irq(&xtime_lock);
 
 	put_cpu();
 
@@ -492,6 +476,9 @@ static int set_rtc_mmss(unsigned long nowtime)
 
 	WARN_ON(irqs_disabled());
 
+	if (!(xen_start_info.flags & SIF_INITDOMAIN))
+		return 0;
+
 	/* gets recalled with irq locally disabled */
 	spin_lock_irq(&rtc_lock);
 	if (efi_enabled)
@@ -603,8 +590,10 @@ static inline void do_timer_interrupt(int irq, void *dev_id,
 		profile_tick(CPU_PROFILING, regs);
 	}
 
-	if (unlikely(shadow_tv_version != HYPERVISOR_shared_info->wc_version))
+	if (shadow_tv_version != HYPERVISOR_shared_info->wc_version) {
 		update_wallclock();
+		clock_was_set();
+	}
 }
 
 /*
diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index dedebbbd6c..efd967464d 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -279,8 +279,6 @@ void arch_do_createdomain(struct vcpu *v)
     
     shadow_lock_init(d);        
     INIT_LIST_HEAD(&d->arch.free_shadow_frames);
-
-    init_domain_time(d);
 }
 
 void arch_do_boot_vcpu(struct vcpu *v)
@@ -481,7 +479,10 @@ int arch_set_info_guest(
     }
 
     update_pagetables(v);
-    
+
+    if ( v->vcpu_id == 0 )
+        init_domain_time(d);
+
     /* Don't redo final setup */
     set_bit(_VCPUF_initialised, &v->vcpu_flags);
 
diff --git a/xen/arch/x86/domain_build.c b/xen/arch/x86/domain_build.c
index 9ba5b4512b..b9941763c4 100644
--- a/xen/arch/x86/domain_build.c
+++ b/xen/arch/x86/domain_build.c
@@ -618,6 +618,8 @@ int construct_dom0(struct domain *d,
     /* DOM0 gets access to everything. */
     physdev_init_dom0(d);
 
+    init_domain_time(d);
+
     set_bit(_DOMF_constructed, &d->domain_flags);
 
     new_thread(v, dsi.v_kernentry, vstack_end, vstartinfo_start);
diff --git a/xen/arch/x86/time.c b/xen/arch/x86/time.c
index d978c8170e..81100e7a1c 100644
--- a/xen/arch/x86/time.c
+++ b/xen/arch/x86/time.c
@@ -44,6 +44,7 @@ spinlock_t rtc_lock = SPIN_LOCK_UNLOCKED;
 int timer_ack = 0;
 unsigned long volatile jiffies;
 static u32 wc_sec, wc_nsec; /* UTC time at last 'time update'. */
+static spinlock_t wc_lock = SPIN_LOCK_UNLOCKED;
 
 struct time_scale {
     int shift;
@@ -699,13 +700,14 @@ void do_settime(unsigned long secs, unsigned long nsecs, u64 system_time_base)
     struct domain *d;
     shared_info_t *s;
 
-    x = (secs * 1000000000ULL) + (u64)nsecs + system_time_base;
+    x = (secs * 1000000000ULL) + (u64)nsecs - system_time_base;
     y = do_div(x, 1000000000);
 
     wc_sec  = _wc_sec  = (u32)x;
     wc_nsec = _wc_nsec = (u32)y;
 
     read_lock(&domlist_lock);
+    spin_lock(&wc_lock);
 
     for_each_domain ( d )
     {
@@ -716,15 +718,18 @@ void do_settime(unsigned long secs, unsigned long nsecs, u64 system_time_base)
         version_update_end(&s->wc_version);
     }
 
+    spin_unlock(&wc_lock);
     read_unlock(&domlist_lock);
 }
 
 void init_domain_time(struct domain *d)
 {
+    spin_lock(&wc_lock);
     version_update_begin(&d->shared_info->wc_version);
     d->shared_info->wc_sec  = wc_sec;
     d->shared_info->wc_nsec = wc_nsec;
     version_update_end(&d->shared_info->wc_version);
+    spin_unlock(&wc_lock);
 }
 
 static void local_time_calibration(void *unused)