diff --git a/xen/arch/x86/acpi/cpu_idle.c b/xen/arch/x86/acpi/cpu_idle.c
index f8302d1f54..40263873e8 100644
--- a/xen/arch/x86/acpi/cpu_idle.c
+++ b/xen/arch/x86/acpi/cpu_idle.c
@@ -195,6 +195,15 @@ static void acpi_processor_idle(void)
     int sleep_ticks = 0;
     u32 t1, t2 = 0;
 
+    sched_tick_suspend();
+    /*
+     * sched_tick_suspend may raise TIMER_SOFTIRQ by __stop_timer,
+     * which will break the later assumption of no sofirq pending,
+     * so add do_softirq
+     */
+    if ( softirq_pending(smp_processor_id()) )
+        do_softirq();
+
     /*
      * Interrupts must be disabled during bus mastering calculations and
      * for C2/C3 transitions.
@@ -204,6 +213,7 @@ static void acpi_processor_idle(void)
     if ( softirq_pending(smp_processor_id()) )
     {
         local_irq_enable();
+        sched_tick_resume();
         return;
     }
 
@@ -223,6 +233,7 @@ static void acpi_processor_idle(void)
             pm_idle_save();
         else
             acpi_safe_halt();
+        sched_tick_resume();
         return;
     }
 
@@ -329,6 +340,7 @@ static void acpi_processor_idle(void)
 
     default:
         local_irq_enable();
+        sched_tick_resume();
         return;
     }
 
@@ -339,6 +351,8 @@ static void acpi_processor_idle(void)
         cx->time += sleep_ticks;
     }
 
+    sched_tick_resume();
+
     if ( cpuidle_current_governor->reflect )
         cpuidle_current_governor->reflect(power);
 }
diff --git a/xen/common/sched_credit.c b/xen/common/sched_credit.c
index d724293bc1..c554c8b8b9 100644
--- a/xen/common/sched_credit.c
+++ b/xen/common/sched_credit.c
@@ -154,6 +154,7 @@ struct csched_private {
     spinlock_t lock;
     struct list_head active_sdom;
     uint32_t ncpus;
+    struct timer  master_ticker;
     unsigned int master;
     cpumask_t idlers;
     uint32_t weight;
@@ -757,7 +758,7 @@ csched_runq_sort(unsigned int cpu)
 }
 
 static void
-csched_acct(void)
+csched_acct(void* dummy)
 {
     unsigned long flags;
     struct list_head *iter_vcpu, *next_vcpu;
@@ -792,7 +793,7 @@ csched_acct(void)
         csched_priv.credit_balance = 0;
         spin_unlock_irqrestore(&csched_priv.lock, flags);
         CSCHED_STAT_CRANK(acct_no_work);
-        return;
+        goto out;
     }
 
     CSCHED_STAT_CRANK(acct_run);
@@ -950,6 +951,10 @@ csched_acct(void)
 
     /* Inform each CPU that its runq needs to be sorted */
     csched_priv.runq_sort++;
+
+out:
+    set_timer( &csched_priv.master_ticker, NOW() +
+            MILLISECS(CSCHED_MSECS_PER_TICK) * CSCHED_TICKS_PER_ACCT );
 }
 
 static void
@@ -966,18 +971,6 @@ csched_tick(void *_cpu)
     if ( !is_idle_vcpu(current) )
         csched_vcpu_acct(cpu);
 
-    /*
-     * Host-wide accounting duty
-     *
-     * Note: Currently, this is always done by the master boot CPU. Eventually,
-     * we could distribute or at the very least cycle the duty.
-     */
-    if ( (csched_priv.master == cpu) &&
-         (spc->tick % CSCHED_TICKS_PER_ACCT) == 0 )
-    {
-        csched_acct();
-    }
-
     /*
      * Check if runq needs to be sorted
      *
@@ -1310,10 +1303,35 @@ static __init int csched_start_tickers(void)
         set_timer(&spc->ticker, NOW() + MILLISECS(CSCHED_MSECS_PER_TICK));
     }
 
+    init_timer( &csched_priv.master_ticker, csched_acct, NULL,
+                    csched_priv.master);
+
+    set_timer( &csched_priv.master_ticker, NOW() +
+            MILLISECS(CSCHED_MSECS_PER_TICK) * CSCHED_TICKS_PER_ACCT );
+
     return 0;
 }
 __initcall(csched_start_tickers);
 
+static void csched_tick_suspend(void)
+{
+    struct csched_pcpu *spc;
+
+    spc = CSCHED_PCPU(smp_processor_id());
+
+    stop_timer(&spc->ticker);
+}
+
+static void csched_tick_resume(void)
+{
+    struct csched_pcpu *spc;
+    uint64_t now = NOW();
+
+    spc = CSCHED_PCPU(smp_processor_id());
+
+    set_timer(&spc->ticker, now + MILLISECS(CSCHED_MSECS_PER_TICK)
+            - now % MILLISECS(CSCHED_MSECS_PER_TICK) );
+}
 
 struct scheduler sched_credit_def = {
     .name           = "SMP Credit Scheduler",
@@ -1337,4 +1355,7 @@ struct scheduler sched_credit_def = {
     .dump_cpu_state = csched_dump_pcpu,
     .dump_settings  = csched_dump,
     .init           = csched_init,
+
+    .tick_suspend   = csched_tick_suspend,
+    .tick_resume    = csched_tick_resume,
 };
diff --git a/xen/common/schedule.c b/xen/common/schedule.c
index 5e91f6c85d..f1121938da 100644
--- a/xen/common/schedule.c
+++ b/xen/common/schedule.c
@@ -964,6 +964,16 @@ void dump_runq(unsigned char key)
     local_irq_restore(flags);
 }
 
+void sched_tick_suspend(void)
+{
+    SCHED_OP(tick_suspend);
+}
+
+void sched_tick_resume(void)
+{
+    SCHED_OP(tick_resume);
+}
+
 #ifdef CONFIG_COMPAT
 #include "compat/schedule.c"
 #endif
diff --git a/xen/include/xen/sched-if.h b/xen/include/xen/sched-if.h
index f6ba9faa5c..5caf8245d2 100644
--- a/xen/include/xen/sched-if.h
+++ b/xen/include/xen/sched-if.h
@@ -77,6 +77,9 @@ struct scheduler {
                                     struct xen_domctl_scheduler_op *);
     void         (*dump_settings)  (void);
     void         (*dump_cpu_state) (int);
+
+    void         (*tick_suspend)    (void);
+    void         (*tick_resume)     (void);
 };
 
 #endif /* __XEN_SCHED_IF_H__ */
diff --git a/xen/include/xen/sched.h b/xen/include/xen/sched.h
index efaec7e9b7..206bf6de4e 100644
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -428,6 +428,8 @@ int  sched_init_domain(struct domain *d);
 void sched_destroy_domain(struct domain *d);
 long sched_adjust(struct domain *, struct xen_domctl_scheduler_op *);
 int  sched_id(void);
+void sched_tick_suspend(void);
+void sched_tick_resume(void);
 void vcpu_wake(struct vcpu *d);
 void vcpu_sleep_nosync(struct vcpu *d);
 void vcpu_sleep_sync(struct vcpu *d);