s390/vtime: steal time exponential moving average

To be able to judge the current overcommitment ratio for a CPU add
a lowcore field with the exponential moving average of the steal time.
The average is updated every tick.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

3 files changed, 45 insertions, 38 deletions

diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
index cc0947e08b6f..5b9f10b1e55d 100644
--- a/arch/s390/include/asm/lowcore.h
+++ b/arch/s390/include/asm/lowcore.h
@@ -91,52 +91,53 @@ struct lowcore {
 	__u64	hardirq_timer;			/* 0x02e8 */
 	__u64	softirq_timer;			/* 0x02f0 */
 	__u64	steal_timer;			/* 0x02f8 */
-	__u64	last_update_timer;		/* 0x0300 */
-	__u64	last_update_clock;		/* 0x0308 */
-	__u64	int_clock;			/* 0x0310 */
-	__u64	mcck_clock;			/* 0x0318 */
-	__u64	clock_comparator;		/* 0x0320 */
-	__u64	boot_clock[2];			/* 0x0328 */
+	__u64	avg_steal_timer;		/* 0x0300 */
+	__u64	last_update_timer;		/* 0x0308 */
+	__u64	last_update_clock;		/* 0x0310 */
+	__u64	int_clock;			/* 0x0318*/
+	__u64	mcck_clock;			/* 0x0320 */
+	__u64	clock_comparator;		/* 0x0328 */
+	__u64	boot_clock[2];			/* 0x0330 */
 
 	/* Current process. */
-	__u64	current_task;			/* 0x0338 */
-	__u64	kernel_stack;			/* 0x0340 */
+	__u64	current_task;			/* 0x0340 */
+	__u64	kernel_stack;			/* 0x0348 */
 
 	/* Interrupt, DAT-off and restartstack. */
-	__u64	async_stack;			/* 0x0348 */
-	__u64	nodat_stack;			/* 0x0350 */
-	__u64	restart_stack;			/* 0x0358 */
+	__u64	async_stack;			/* 0x0350 */
+	__u64	nodat_stack;			/* 0x0358 */
+	__u64	restart_stack;			/* 0x0360 */
 
 	/* Restart function and parameter. */
-	__u64	restart_fn;			/* 0x0360 */
-	__u64	restart_data;			/* 0x0368 */
-	__u64	restart_source;			/* 0x0370 */
+	__u64	restart_fn;			/* 0x0368 */
+	__u64	restart_data;			/* 0x0370 */
+	__u64	restart_source;			/* 0x0378 */
 
 	/* Address space pointer. */
-	__u64	kernel_asce;			/* 0x0378 */
-	__u64	user_asce;			/* 0x0380 */
-	__u64	vdso_asce;			/* 0x0388 */
+	__u64	kernel_asce;			/* 0x0380 */
+	__u64	user_asce;			/* 0x0388 */
+	__u64	vdso_asce;			/* 0x0390 */
 
 	/*
 	 * The lpp and current_pid fields form a
 	 * 64-bit value that is set as program
 	 * parameter with the LPP instruction.
 	 */
-	__u32	lpp;				/* 0x0390 */
-	__u32	current_pid;			/* 0x0394 */
+	__u32	lpp;				/* 0x0398 */
+	__u32	current_pid;			/* 0x039c */
 
 	/* SMP info area */
-	__u32	cpu_nr;				/* 0x0398 */
-	__u32	softirq_pending;		/* 0x039c */
-	__u32	preempt_count;			/* 0x03a0 */
-	__u32	spinlock_lockval;		/* 0x03a4 */
-	__u32	spinlock_index;			/* 0x03a8 */
-	__u32	fpu_flags;			/* 0x03ac */
-	__u64	percpu_offset;			/* 0x03b0 */
-	__u64	vdso_per_cpu_data;		/* 0x03b8 */
-	__u64	machine_flags;			/* 0x03c0 */
-	__u64	gmap;				/* 0x03c8 */
-	__u8	pad_0x03d0[0x0400-0x03d0];	/* 0x03d0 */
+	__u32	cpu_nr;				/* 0x03a0 */
+	__u32	softirq_pending;		/* 0x03a4 */
+	__u32	preempt_count;			/* 0x03a8 */
+	__u32	spinlock_lockval;		/* 0x03ac */
+	__u32	spinlock_index;			/* 0x03b0 */
+	__u32	fpu_flags;			/* 0x03b4 */
+	__u64	percpu_offset;			/* 0x03b8 */
+	__u64	vdso_per_cpu_data;		/* 0x03c0 */
+	__u64	machine_flags;			/* 0x03c8 */
+	__u64	gmap;				/* 0x03d0 */
+	__u8	pad_0x03d8[0x0400-0x03d8];	/* 0x03d8 */
 
 	/* br %r1 trampoline */
 	__u16	br_r1_trampoline;		/* 0x0400 */
diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index b198ece2aad6..b8eb99685546 100644
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -266,7 +266,8 @@ static void pcpu_prepare_secondary(struct pcpu *pcpu, int cpu)
 	lc->percpu_offset = __per_cpu_offset[cpu];
 	lc->kernel_asce = S390_lowcore.kernel_asce;
 	lc->machine_flags = S390_lowcore.machine_flags;
-	lc->user_timer = lc->system_timer = lc->steal_timer = 0;
+	lc->user_timer = lc->system_timer =
+		lc->steal_timer = lc->avg_steal_timer = 0;
 	__ctl_store(lc->cregs_save_area, 0, 15);
 	save_access_regs((unsigned int *) lc->access_regs_save_area);
 	memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list,
diff --git a/arch/s390/kernel/vtime.c b/arch/s390/kernel/vtime.c
index 98f850e00008..a69a0911ed0e 100644
--- a/arch/s390/kernel/vtime.c
+++ b/arch/s390/kernel/vtime.c
@@ -124,7 +124,7 @@ static void account_system_index_scaled(struct task_struct *p, u64 cputime,
  */
 static int do_account_vtime(struct task_struct *tsk)
 {
-	u64 timer, clock, user, guest, system, hardirq, softirq, steal;
+	u64 timer, clock, user, guest, system, hardirq, softirq;
 
 	timer = S390_lowcore.last_update_timer;
 	clock = S390_lowcore.last_update_clock;
@@ -182,12 +182,6 @@ static int do_account_vtime(struct task_struct *tsk)
 	if (softirq)
 		account_system_index_scaled(tsk, softirq, CPUTIME_SOFTIRQ);
 
-	steal = S390_lowcore.steal_timer;
-	if ((s64) steal > 0) {
-		S390_lowcore.steal_timer = 0;
-		account_steal_time(cputime_to_nsecs(steal));
-	}
-
 	return virt_timer_forward(user + guest + system + hardirq + softirq);
 }
 
@@ -213,8 +207,19 @@ void vtime_task_switch(struct task_struct *prev)
  */
 void vtime_flush(struct task_struct *tsk)
 {
+	u64 steal, avg_steal;
+
 	if (do_account_vtime(tsk))
 		virt_timer_expire();
+
+	steal = S390_lowcore.steal_timer;
+	avg_steal = S390_lowcore.avg_steal_timer / 2;
+	if ((s64) steal > 0) {
+		S390_lowcore.steal_timer = 0;
+		account_steal_time(steal);
+		avg_steal += steal;
+	}
+	S390_lowcore.avg_steal_timer = avg_steal;
 }
 
 /*