Merge branch 'cputime' of git://git390.osdl.marist.edu/pub/scm/linux-2.6

* 'cputime' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [PATCH] fast vdso implementation for CLOCK_THREAD_CPUTIME_ID
  [PATCH] improve idle cputime accounting
  [PATCH] improve precision of idle time detection.
  [PATCH] improve precision of process accounting.
  [PATCH] idle cputime accounting
  [PATCH] fix scaled & unscaled cputime accounting

arch/ia64/kernel/time.c

@@ -93,13 +93,14 @@ void ia64_account_on_switch(struct task_struct *prev, struct task_struct *next)
 	now = ia64_get_itc();
 
 	delta_stime = cycle_to_cputime(pi->ac_stime + (now - pi->ac_stamp));
-	account_system_time(prev, 0, delta_stime);
-	account_system_time_scaled(prev, delta_stime);
+	if (idle_task(smp_processor_id()) != prev)
+		account_system_time(prev, 0, delta_stime, delta_stime);
+	else
+		account_idle_time(delta_stime);
 
 	if (pi->ac_utime) {
 		delta_utime = cycle_to_cputime(pi->ac_utime);
-		account_user_time(prev, delta_utime);
-		account_user_time_scaled(prev, delta_utime);
+		account_user_time(prev, delta_utime, delta_utime);
 	}
 
 	pi->ac_stamp = ni->ac_stamp = now;
@@ -122,8 +123,10 @@ void account_system_vtime(struct task_struct *tsk)
 	now = ia64_get_itc();
 
 	delta_stime = cycle_to_cputime(ti->ac_stime + (now - ti->ac_stamp));
-	account_system_time(tsk, 0, delta_stime);
-	account_system_time_scaled(tsk, delta_stime);
+	if (irq_count() || idle_task(smp_processor_id()) != tsk)
+		account_system_time(tsk, 0, delta_stime, delta_stime);
+	else
+		account_idle_time(delta_stime);
 	ti->ac_stime = 0;
 
 	ti->ac_stamp = now;
@@ -143,8 +146,7 @@ void account_process_tick(struct task_struct *p, int user_tick)
 
 	if (ti->ac_utime) {
 		delta_utime = cycle_to_cputime(ti->ac_utime);
-		account_user_time(p, delta_utime);
-		account_user_time_scaled(p, delta_utime);
+		account_user_time(p, delta_utime, delta_utime);
 		ti->ac_utime = 0;
 	}
 }

arch/powerpc/kernel/process.c

@@ -33,6 +33,7 @@
 #include <linux/mqueue.h>
 #include <linux/hardirq.h>
 #include <linux/utsname.h>
+#include <linux/kernel_stat.h>
 
 #include <asm/pgtable.h>
 #include <asm/uaccess.h>

arch/powerpc/kernel/time.c

@@ -256,8 +256,10 @@ void account_system_vtime(struct task_struct *tsk)
 		delta += sys_time;
 		get_paca()->system_time = 0;
 	}
-	account_system_time(tsk, 0, delta);
-	account_system_time_scaled(tsk, deltascaled);
+	if (in_irq() || idle_task(smp_processor_id()) != tsk)
+		account_system_time(tsk, 0, delta, deltascaled);
+	else
+		account_idle_time(delta);
 	per_cpu(cputime_last_delta, smp_processor_id()) = delta;
 	per_cpu(cputime_scaled_last_delta, smp_processor_id()) = deltascaled;
 	local_irq_restore(flags);
@@ -275,10 +277,8 @@ void account_process_tick(struct task_struct *tsk, int user_tick)
 
 	utime = get_paca()->user_time;
 	get_paca()->user_time = 0;
-	account_user_time(tsk, utime);
-
 	utimescaled = cputime_to_scaled(utime);
-	account_user_time_scaled(tsk, utimescaled);
+	account_user_time(tsk, utime, utimescaled);
 }
 
 /*
@@ -338,8 +338,12 @@ void calculate_steal_time(void)
 	tb = mftb();
 	purr = mfspr(SPRN_PURR);
 	stolen = (tb - pme->tb) - (purr - pme->purr);
-	if (stolen > 0)
-		account_steal_time(current, stolen);
+	if (stolen > 0) {
+		if (idle_task(smp_processor_id()) != current)
+			account_steal_time(stolen);
+		else
+			account_idle_time(stolen);
+	}
 	pme->tb = tb;
 	pme->purr = purr;
 }

arch/s390/include/asm/cpu.h

@@ -14,7 +14,6 @@
 
 struct s390_idle_data {
 	spinlock_t lock;
-	unsigned int in_idle;
 	unsigned long long idle_count;
 	unsigned long long idle_enter;
 	unsigned long long idle_time;
@@ -22,12 +21,12 @@ struct s390_idle_data {
 
 DECLARE_PER_CPU(struct s390_idle_data, s390_idle);
 
-void s390_idle_leave(void);
+void vtime_start_cpu(void);
 
 static inline void s390_idle_check(void)
 {
-	if ((&__get_cpu_var(s390_idle))->in_idle)
-		s390_idle_leave();
+	if ((&__get_cpu_var(s390_idle))->idle_enter != 0ULL)
+		vtime_start_cpu();
 }
 
 #endif /* _ASM_S390_CPU_H_ */

arch/s390/include/asm/cputime.h

@@ -11,7 +11,7 @@
 
 #include <asm/div64.h>
 
-/* We want to use micro-second resolution. */
+/* We want to use full resolution of the CPU timer: 2**-12 micro-seconds. */
 
 typedef unsigned long long cputime_t;
 typedef unsigned long long cputime64_t;
@@ -53,9 +53,9 @@ __div(unsigned long long n, unsigned int base)
 #define cputime_ge(__a, __b)		((__a) >= (__b))
 #define cputime_lt(__a, __b)		((__a) <  (__b))
 #define cputime_le(__a, __b)		((__a) <= (__b))
-#define cputime_to_jiffies(__ct)	(__div((__ct), 1000000 / HZ))
+#define cputime_to_jiffies(__ct)	(__div((__ct), 4096000000ULL / HZ))
 #define cputime_to_scaled(__ct)		(__ct)
-#define jiffies_to_cputime(__hz)	((cputime_t)(__hz) * (1000000 / HZ))
+#define jiffies_to_cputime(__hz)	((cputime_t)(__hz) * (4096000000ULL / HZ))
 
 #define cputime64_zero			(0ULL)
 #define cputime64_add(__a, __b)		((__a) + (__b))
@@ -64,7 +64,7 @@ __div(unsigned long long n, unsigned int base)
 static inline u64
 cputime64_to_jiffies64(cputime64_t cputime)
 {
-	do_div(cputime, 1000000 / HZ);
+	do_div(cputime, 4096000000ULL / HZ);
 	return cputime;
 }
 
@@ -74,13 +74,13 @@ cputime64_to_jiffies64(cputime64_t cputime)
 static inline unsigned int
 cputime_to_msecs(const cputime_t cputime)
 {
-	return __div(cputime, 1000);
+	return __div(cputime, 4096000);
 }
 
 static inline cputime_t
 msecs_to_cputime(const unsigned int m)
 {
-	return (cputime_t) m * 1000;
+	return (cputime_t) m * 4096000;
 }
 
 /*
@@ -89,13 +89,13 @@ msecs_to_cputime(const unsigned int m)
 static inline unsigned int
 cputime_to_secs(const cputime_t cputime)
 {
-	return __div(cputime, 1000000);
+	return __div(cputime, 2048000000) >> 1;
 }
 
 static inline cputime_t
 secs_to_cputime(const unsigned int s)
 {
-	return (cputime_t) s * 1000000;
+	return (cputime_t) s * 4096000000ULL;
 }
 
 /*
@@ -104,7 +104,7 @@ secs_to_cputime(const unsigned int s)
 static inline cputime_t
 timespec_to_cputime(const struct timespec *value)
 {
-        return value->tv_nsec / 1000 + (u64) value->tv_sec * 1000000;
+	return value->tv_nsec * 4096 / 1000 + (u64) value->tv_sec * 4096000000ULL;
 }
 
 static inline void
@@ -114,12 +114,12 @@ cputime_to_timespec(const cputime_t cputime, struct timespec *value)
 	register_pair rp;
 
 	rp.pair = cputime >> 1;
-	asm ("dr %0,%1" : "+d" (rp) : "d" (1000000 >> 1));
-	value->tv_nsec = rp.subreg.even * 1000;
+	asm ("dr %0,%1" : "+d" (rp) : "d" (2048000000UL));
+	value->tv_nsec = rp.subreg.even * 1000 / 4096;
 	value->tv_sec = rp.subreg.odd;
 #else
-	value->tv_nsec = (cputime % 1000000) * 1000;
-	value->tv_sec = cputime / 1000000;
+	value->tv_nsec = (cputime % 4096000000ULL) * 1000 / 4096;
+	value->tv_sec = cputime / 4096000000ULL;
 #endif
 }
 
@@ -131,7 +131,7 @@ cputime_to_timespec(const cputime_t cputime, struct timespec *value)
 static inline cputime_t
 timeval_to_cputime(const struct timeval *value)
 {
-        return value->tv_usec + (u64) value->tv_sec * 1000000;
+	return value->tv_usec * 4096 + (u64) value->tv_sec * 4096000000ULL;
 }
 
 static inline void
@@ -141,12 +141,12 @@ cputime_to_timeval(const cputime_t cputime, struct timeval *value)
 	register_pair rp;
 
 	rp.pair = cputime >> 1;
-	asm ("dr %0,%1" : "+d" (rp) : "d" (1000000 >> 1));
-	value->tv_usec = rp.subreg.even;
+	asm ("dr %0,%1" : "+d" (rp) : "d" (2048000000UL));
+	value->tv_usec = rp.subreg.even / 4096;
 	value->tv_sec = rp.subreg.odd;
 #else
-	value->tv_usec = cputime % 1000000;
-	value->tv_sec = cputime / 1000000;
+	value->tv_usec = cputime % 4096000000ULL;
+	value->tv_sec = cputime / 4096000000ULL;
 #endif
 }
 
@@ -156,13 +156,13 @@ cputime_to_timeval(const cputime_t cputime, struct timeval *value)
 static inline clock_t
 cputime_to_clock_t(cputime_t cputime)
 {
-	return __div(cputime, 1000000 / USER_HZ);
+	return __div(cputime, 4096000000ULL / USER_HZ);
 }
 
 static inline cputime_t
 clock_t_to_cputime(unsigned long x)
 {
-	return (cputime_t) x * (1000000 / USER_HZ);
+	return (cputime_t) x * (4096000000ULL / USER_HZ);
 }
 
 /*
@@ -171,7 +171,7 @@ clock_t_to_cputime(unsigned long x)
 static inline clock_t
 cputime64_to_clock_t(cputime64_t cputime)
 {
-       return __div(cputime, 1000000 / USER_HZ);
+       return __div(cputime, 4096000000ULL / USER_HZ);
 }
 
 #endif /* _S390_CPUTIME_H */

arch/s390/include/asm/lowcore.h

@@ -67,11 +67,11 @@
 #define __LC_SYNC_ENTER_TIMER		0x248
 #define __LC_ASYNC_ENTER_TIMER		0x250
 #define __LC_EXIT_TIMER			0x258
-#define __LC_LAST_UPDATE_TIMER		0x260
-#define __LC_USER_TIMER			0x268
-#define __LC_SYSTEM_TIMER		0x270
-#define __LC_LAST_UPDATE_CLOCK		0x278
-#define __LC_STEAL_CLOCK		0x280
+#define __LC_USER_TIMER			0x260
+#define __LC_SYSTEM_TIMER		0x268
+#define __LC_STEAL_TIMER		0x270
+#define __LC_LAST_UPDATE_TIMER		0x278
+#define __LC_LAST_UPDATE_CLOCK		0x280
 #define __LC_RETURN_MCCK_PSW            0x288
 #define __LC_KERNEL_STACK               0xC40
 #define __LC_THREAD_INFO		0xC44
@@ -89,11 +89,11 @@
 #define __LC_SYNC_ENTER_TIMER		0x250
 #define __LC_ASYNC_ENTER_TIMER		0x258
 #define __LC_EXIT_TIMER			0x260
-#define __LC_LAST_UPDATE_TIMER		0x268
-#define __LC_USER_TIMER			0x270
-#define __LC_SYSTEM_TIMER		0x278
-#define __LC_LAST_UPDATE_CLOCK		0x280
-#define __LC_STEAL_CLOCK		0x288
+#define __LC_USER_TIMER			0x268
+#define __LC_SYSTEM_TIMER		0x270
+#define __LC_STEAL_TIMER		0x278
+#define __LC_LAST_UPDATE_TIMER		0x280
+#define __LC_LAST_UPDATE_CLOCK		0x288
 #define __LC_RETURN_MCCK_PSW            0x290
 #define __LC_KERNEL_STACK               0xD40
 #define __LC_THREAD_INFO		0xD48
@@ -106,8 +106,10 @@
 #define __LC_IPLDEV                     0xDB8
 #define __LC_CURRENT			0xDD8
 #define __LC_INT_CLOCK			0xDE8
+#define __LC_VDSO_PER_CPU		0xE38
 #endif /* __s390x__ */
 
+#define __LC_PASTE			0xE40
 
 #define __LC_PANIC_MAGIC		0xE00
 #ifndef __s390x__
@@ -252,11 +254,11 @@ struct _lowcore
 	__u64        sync_enter_timer;         /* 0x248 */
 	__u64        async_enter_timer;        /* 0x250 */
 	__u64        exit_timer;               /* 0x258 */
-	__u64        last_update_timer;        /* 0x260 */
-	__u64        user_timer;               /* 0x268 */
-	__u64        system_timer;             /* 0x270 */
-	__u64        last_update_clock;        /* 0x278 */
-	__u64        steal_clock;              /* 0x280 */
+	__u64	     user_timer;	       /* 0x260 */
+	__u64	     system_timer;	       /* 0x268 */
+	__u64	     steal_timer;	       /* 0x270 */
+	__u64	     last_update_timer;        /* 0x278 */
+	__u64	     last_update_clock;        /* 0x280 */
         psw_t        return_mcck_psw;          /* 0x288 */
 	__u8         pad8[0xc00-0x290];        /* 0x290 */
 
@@ -343,11 +345,11 @@ struct _lowcore
 	__u64        sync_enter_timer;         /* 0x250 */
 	__u64        async_enter_timer;        /* 0x258 */
 	__u64        exit_timer;               /* 0x260 */
-	__u64        last_update_timer;        /* 0x268 */
-	__u64        user_timer;               /* 0x270 */
-	__u64        system_timer;             /* 0x278 */
-	__u64        last_update_clock;        /* 0x280 */
-	__u64        steal_clock;              /* 0x288 */
+	__u64	     user_timer;	       /* 0x268 */
+	__u64	     system_timer;	       /* 0x270 */
+	__u64	     steal_timer;	       /* 0x278 */
+	__u64	     last_update_timer;        /* 0x280 */
+	__u64	     last_update_clock;        /* 0x288 */
         psw_t        return_mcck_psw;          /* 0x290 */
         __u8         pad8[0xc00-0x2a0];        /* 0x2a0 */
         /* System info area */
@@ -381,7 +383,12 @@ struct _lowcore
         /* whether the kernel died with panic() or not */
         __u32        panic_magic;              /* 0xe00 */
 
-	__u8         pad13[0x11b8-0xe04];      /* 0xe04 */
+	/* Per cpu primary space access list */
+	__u8	     pad_0xe04[0xe3c-0xe04];   /* 0xe04 */
+	__u32	     vdso_per_cpu_data;	       /* 0xe3c */
+	__u32	     paste[16];		       /* 0xe40 */
+
+	__u8	     pad13[0x11b8-0xe80];      /* 0xe80 */
 
 	/* 64 bit extparam used for pfault, diag 250 etc  */
 	__u64        ext_params2;               /* 0x11B8 */

arch/s390/include/asm/system.h

@@ -99,7 +99,7 @@ static inline void restore_access_regs(unsigned int *acrs)
 	prev = __switch_to(prev,next);					     \
 } while (0)
 
-extern void account_vtime(struct task_struct *);
+extern void account_vtime(struct task_struct *, struct task_struct *);
 extern void account_tick_vtime(struct task_struct *);
 extern void account_system_vtime(struct task_struct *);
 
@@ -121,7 +121,7 @@ static inline void cmma_init(void) { }
 
 #define finish_arch_switch(prev) do {					     \
 	set_fs(current->thread.mm_segment);				     \
-	account_vtime(prev);						     \
+	account_vtime(prev, current);					     \
 } while (0)
 
 #define nop() asm volatile("nop")

arch/s390/include/asm/thread_info.h

@@ -47,6 +47,8 @@ struct thread_info {
 	unsigned int		cpu;		/* current CPU */
 	int			preempt_count;	/* 0 => preemptable, <0 => BUG */
 	struct restart_block	restart_block;
+	__u64			user_timer;
+	__u64			system_timer;
 };
 
 /*

arch/s390/include/asm/timer.h

@@ -23,20 +23,18 @@ struct vtimer_list {
 	__u64 expires;
 	__u64 interval;
 
-	spinlock_t lock;
-	unsigned long magic;
-
 	void (*function)(unsigned long);
 	unsigned long data;
 };
 
-/* the offset value will wrap after ca. 71 years */
+/* the vtimer value will wrap after ca. 71 years */
 struct vtimer_queue {
 	struct list_head list;
 	spinlock_t lock;
-	__u64 to_expire;	  /* current event expire time */
-	__u64 offset;		  /* list offset to zero */
+	__u64 timer;		/* last programmed timer */
+	__u64 elapsed;		/* elapsed time of timer expire values */
 	__u64 idle;		/* temp var for idle */
+	int do_spt;		/* =1: reprogram cpu timer in idle */
 };
 
 extern void init_virt_timer(struct vtimer_list *timer);
@@ -48,8 +46,8 @@ extern int del_virt_timer(struct vtimer_list *timer);
 extern void init_cpu_vtimer(void);
 extern void vtime_init(void);
 
-extern void vtime_start_cpu_timer(void);
-extern void vtime_stop_cpu_timer(void);
+extern void vtime_stop_cpu(void);
+extern void vtime_start_leave(void);
 
 #endif /* __KERNEL__ */
 

arch/s390/include/asm/vdso.h

@@ -12,9 +12,9 @@
 #ifndef __ASSEMBLY__
 
 /*
- * Note about this structure:
+ * Note about the vdso_data and vdso_per_cpu_data structures:
  *
- * NEVER USE THIS IN USERSPACE CODE DIRECTLY. The layout of this
+ * NEVER USE THEM IN USERSPACE CODE DIRECTLY. The layout of the
  * structure is supposed to be known only to the function in the vdso
  * itself and may change without notice.
  */
@@ -28,10 +28,21 @@ struct vdso_data {
 	__u64 wtom_clock_nsec;		/*				0x28 */
 	__u32 tz_minuteswest;		/* Minutes west of Greenwich	0x30 */
 	__u32 tz_dsttime;		/* Type of dst correction	0x34 */
+	__u32 ectg_available;
+};
+
+struct vdso_per_cpu_data {
+	__u64 ectg_timer_base;
+	__u64 ectg_user_time;
 };
 
 extern struct vdso_data *vdso_data;
 
+#ifdef CONFIG_64BIT
+int vdso_alloc_per_cpu(int cpu, struct _lowcore *lowcore);
+void vdso_free_per_cpu(int cpu, struct _lowcore *lowcore);
+#endif
+
 #endif /* __ASSEMBLY__ */
 
 #endif /* __KERNEL__ */

arch/s390/kernel/asm-offsets.c

@@ -48,6 +48,11 @@ int main(void)
 	DEFINE(__VDSO_WTOM_SEC, offsetof(struct vdso_data, wtom_clock_sec));
 	DEFINE(__VDSO_WTOM_NSEC, offsetof(struct vdso_data, wtom_clock_nsec));
 	DEFINE(__VDSO_TIMEZONE, offsetof(struct vdso_data, tz_minuteswest));
+	DEFINE(__VDSO_ECTG_OK, offsetof(struct vdso_data, ectg_available));
+	DEFINE(__VDSO_ECTG_BASE,
+	       offsetof(struct vdso_per_cpu_data, ectg_timer_base));
+	DEFINE(__VDSO_ECTG_USER,
+	       offsetof(struct vdso_per_cpu_data, ectg_user_time));
 	/* constants used by the vdso */
 	DEFINE(CLOCK_REALTIME, CLOCK_REALTIME);
 	DEFINE(CLOCK_MONOTONIC, CLOCK_MONOTONIC);

arch/s390/kernel/entry.S

@@ -583,8 +583,8 @@ kernel_per:
 
 	.globl io_int_handler
 io_int_handler:
-	stpt	__LC_ASYNC_ENTER_TIMER
 	stck	__LC_INT_CLOCK
+	stpt	__LC_ASYNC_ENTER_TIMER
 	SAVE_ALL_BASE __LC_SAVE_AREA+16
 	SAVE_ALL_ASYNC __LC_IO_OLD_PSW,__LC_SAVE_AREA+16
 	CREATE_STACK_FRAME __LC_IO_OLD_PSW,__LC_SAVE_AREA+16
@@ -723,8 +723,8 @@ io_notify_resume:
 
 	.globl	ext_int_handler
 ext_int_handler:
-	stpt	__LC_ASYNC_ENTER_TIMER
 	stck	__LC_INT_CLOCK
+	stpt	__LC_ASYNC_ENTER_TIMER
 	SAVE_ALL_BASE __LC_SAVE_AREA+16
 	SAVE_ALL_ASYNC __LC_EXT_OLD_PSW,__LC_SAVE_AREA+16
 	CREATE_STACK_FRAME __LC_EXT_OLD_PSW,__LC_SAVE_AREA+16
@@ -750,6 +750,7 @@ __critical_end:
 
 	.globl mcck_int_handler
 mcck_int_handler:
+	stck	__LC_INT_CLOCK
 	spt	__LC_CPU_TIMER_SAVE_AREA	# revalidate cpu timer
 	lm	%r0,%r15,__LC_GPREGS_SAVE_AREA	# revalidate gprs
 	SAVE_ALL_BASE __LC_SAVE_AREA+32

arch/s390/kernel/entry64.S

@@ -177,8 +177,11 @@ _TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
 	.if !\sync
 	ni	\psworg+1,0xfd		# clear wait state bit
 	.endif
-	lmg	%r0,%r15,SP_R0(%r15)	# load gprs 0-15 of user
+	lg	%r14,__LC_VDSO_PER_CPU
+	lmg	%r0,%r13,SP_R0(%r15)	# load gprs 0-13 of user
 	stpt	__LC_EXIT_TIMER
+	mvc	__VDSO_ECTG_BASE(16,%r14),__LC_EXIT_TIMER
+	lmg	%r14,%r15,SP_R14(%r15)	# load grps 14-15 of user
 	lpswe	\psworg			# back to caller
 	.endm
 
@@ -559,8 +562,8 @@ kernel_per:
  */
 	.globl io_int_handler
 io_int_handler:
-	stpt	__LC_ASYNC_ENTER_TIMER
 	stck	__LC_INT_CLOCK
+	stpt	__LC_ASYNC_ENTER_TIMER
 	SAVE_ALL_BASE __LC_SAVE_AREA+32
 	SAVE_ALL_ASYNC __LC_IO_OLD_PSW,__LC_SAVE_AREA+32
 	CREATE_STACK_FRAME __LC_IO_OLD_PSW,__LC_SAVE_AREA+32
@@ -721,8 +724,8 @@ io_notify_resume:
  */
 	.globl	ext_int_handler
 ext_int_handler:
-	stpt	__LC_ASYNC_ENTER_TIMER
 	stck	__LC_INT_CLOCK
+	stpt	__LC_ASYNC_ENTER_TIMER
 	SAVE_ALL_BASE __LC_SAVE_AREA+32
 	SAVE_ALL_ASYNC __LC_EXT_OLD_PSW,__LC_SAVE_AREA+32
 	CREATE_STACK_FRAME __LC_EXT_OLD_PSW,__LC_SAVE_AREA+32
@@ -746,6 +749,7 @@ __critical_end:
  */
 	.globl mcck_int_handler
 mcck_int_handler:
+	stck	__LC_INT_CLOCK
 	la	%r1,4095		# revalidate r1
 	spt	__LC_CPU_TIMER_SAVE_AREA-4095(%r1)	# revalidate cpu timer
 	lmg	%r0,%r15,__LC_GPREGS_SAVE_AREA-4095(%r1)# revalidate gprs
@@ -979,23 +983,23 @@ cleanup_sysc_return:
 
 cleanup_sysc_leave:
 	clc	8(8,%r12),BASED(cleanup_sysc_leave_insn)
-	je	2f
-	mvc	__LC_EXIT_TIMER(8),__LC_ASYNC_ENTER_TIMER
+	je	3f
 	clc	8(8,%r12),BASED(cleanup_sysc_leave_insn+8)
-	je	2f
-	mvc	__LC_RETURN_PSW(16),SP_PSW(%r15)
+	jhe	0f
+	mvc	__LC_EXIT_TIMER(8),__LC_ASYNC_ENTER_TIMER
+0:	mvc	__LC_RETURN_PSW(16),SP_PSW(%r15)
 	cghi	%r12,__LC_MCK_OLD_PSW
-	jne	0f
+	jne	1f
 	mvc	__LC_SAVE_AREA+64(32),SP_R12(%r15)
-	j	1f
-0:	mvc	__LC_SAVE_AREA+32(32),SP_R12(%r15)
-1:	lmg	%r0,%r11,SP_R0(%r15)
+	j	2f
+1:	mvc	__LC_SAVE_AREA+32(32),SP_R12(%r15)
+2:	lmg	%r0,%r11,SP_R0(%r15)
 	lg	%r15,SP_R15(%r15)
-2:	la	%r12,__LC_RETURN_PSW
+3:	la	%r12,__LC_RETURN_PSW
 	br	%r14
 cleanup_sysc_leave_insn:
 	.quad	sysc_done - 4
-	.quad	sysc_done - 8
+	.quad	sysc_done - 16
 
 cleanup_io_return:
 	mvc	__LC_RETURN_PSW(8),0(%r12)
@@ -1005,23 +1009,23 @@ cleanup_io_return:
 
 cleanup_io_leave:
 	clc	8(8,%r12),BASED(cleanup_io_leave_insn)
-	je	2f
-	mvc	__LC_EXIT_TIMER(8),__LC_ASYNC_ENTER_TIMER
+	je	3f
 	clc	8(8,%r12),BASED(cleanup_io_leave_insn+8)
-	je	2f
-	mvc	__LC_RETURN_PSW(16),SP_PSW(%r15)
+	jhe	0f
+	mvc	__LC_EXIT_TIMER(8),__LC_ASYNC_ENTER_TIMER
+0:	mvc	__LC_RETURN_PSW(16),SP_PSW(%r15)
 	cghi	%r12,__LC_MCK_OLD_PSW
-	jne	0f
+	jne	1f
 	mvc	__LC_SAVE_AREA+64(32),SP_R12(%r15)
-	j	1f
-0:	mvc	__LC_SAVE_AREA+32(32),SP_R12(%r15)
-1:	lmg	%r0,%r11,SP_R0(%r15)
+	j	2f
+1:	mvc	__LC_SAVE_AREA+32(32),SP_R12(%r15)
+2:	lmg	%r0,%r11,SP_R0(%r15)
 	lg	%r15,SP_R15(%r15)
-2:	la	%r12,__LC_RETURN_PSW
+3:	la	%r12,__LC_RETURN_PSW
 	br	%r14
 cleanup_io_leave_insn:
 	.quad	io_done - 4
-	.quad	io_done - 8
+	.quad	io_done - 16
 
 /*
  * Integer constants

arch/s390/kernel/head64.S

@@ -87,6 +87,8 @@ startup_continue:
 	lg	%r12,.Lparmaddr-.LPG1(%r13)	# pointer to parameter area
 					# move IPL device to lowcore
 	mvc	__LC_IPLDEV(4),IPL_DEVICE+4-PARMAREA(%r12)
+	lghi	%r0,__LC_PASTE
+	stg	%r0,__LC_VDSO_PER_CPU
 #
 # Setup stack
 #

arch/s390/kernel/process.c

@@ -38,6 +38,7 @@
 #include <linux/utsname.h>
 #include <linux/tick.h>
 #include <linux/elfcore.h>
+#include <linux/kernel_stat.h>
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
 #include <asm/system.h>
@@ -45,7 +46,6 @@
 #include <asm/processor.h>
 #include <asm/irq.h>
 #include <asm/timer.h>
-#include <asm/cpu.h>
 #include "entry.h"
 
 asmlinkage void ret_from_fork(void) asm ("ret_from_fork");
@@ -75,36 +75,6 @@ unsigned long thread_saved_pc(struct task_struct *tsk)
 	return sf->gprs[8];
 }
 
-DEFINE_PER_CPU(struct s390_idle_data, s390_idle) = {
-	.lock = __SPIN_LOCK_UNLOCKED(s390_idle.lock)
-};
-
-static int s390_idle_enter(void)
-{
-	struct s390_idle_data *idle;
-
-	idle = &__get_cpu_var(s390_idle);
-	spin_lock(&idle->lock);
-	idle->idle_count++;
-	idle->in_idle = 1;
-	idle->idle_enter = get_clock();
-	spin_unlock(&idle->lock);
-	vtime_stop_cpu_timer();
-	return NOTIFY_OK;
-}
-
-void s390_idle_leave(void)
-{
-	struct s390_idle_data *idle;
-
-	vtime_start_cpu_timer();
-	idle = &__get_cpu_var(s390_idle);
-	spin_lock(&idle->lock);
-	idle->idle_time += get_clock() - idle->idle_enter;
-	idle->in_idle = 0;
-	spin_unlock(&idle->lock);
-}
-
 extern void s390_handle_mcck(void);
 /*
  * The idle loop on a S390...
@@ -117,10 +87,6 @@ static void default_idle(void)
 		local_irq_enable();
 		return;
 	}
-	if (s390_idle_enter() == NOTIFY_BAD) {
-		local_irq_enable();
-		return;
-	}
 #ifdef CONFIG_HOTPLUG_CPU
 	if (cpu_is_offline(smp_processor_id())) {
 		preempt_enable_no_resched();
@@ -130,7 +96,6 @@ static void default_idle(void)
 	local_mcck_disable();
 	if (test_thread_flag(TIF_MCCK_PENDING)) {
 		local_mcck_enable();
-		s390_idle_leave();
 		local_irq_enable();
 		s390_handle_mcck();
 		return;
@@ -138,9 +103,9 @@ static void default_idle(void)
 	trace_hardirqs_on();
 	/* Don't trace preempt off for idle. */
 	stop_critical_timings();
-	/* Wait for external, I/O or machine check interrupt. */
-	__load_psw_mask(psw_kernel_bits | PSW_MASK_WAIT |
-			PSW_MASK_IO | PSW_MASK_EXT);
+	/* Stop virtual timer and halt the cpu. */
+	vtime_stop_cpu();
+	/* Reenable preemption tracer. */
 	start_critical_timings();
 }
 

arch/s390/kernel/s390_ext.c

@@ -119,8 +119,8 @@ void do_extint(struct pt_regs *regs, unsigned short code)
 	struct pt_regs *old_regs;
 
 	old_regs = set_irq_regs(regs);
-	irq_enter();
 	s390_idle_check();
+	irq_enter();
 	if (S390_lowcore.int_clock >= S390_lowcore.clock_comparator)
 		/* Serve timer interrupts first. */
 		clock_comparator_work();

arch/s390/kernel/setup.c

@@ -427,6 +427,8 @@ setup_lowcore(void)
 		/* enable extended save area */
 		__ctl_set_bit(14, 29);
 	}
+#else
+	lc->vdso_per_cpu_data = (unsigned long) &lc->paste[0];
 #endif
 	set_prefix((u32)(unsigned long) lc);
 }

arch/s390/kernel/smp.c

@@ -47,6 +47,7 @@
 #include <asm/lowcore.h>
 #include <asm/sclp.h>
 #include <asm/cpu.h>
+#include <asm/vdso.h>
 #include "entry.h"
 
 /*
@@ -500,6 +501,9 @@ static int __cpuinit smp_alloc_lowcore(int cpu)
 			goto out;
 		lowcore->extended_save_area_addr = (u32) save_area;
 	}
+#else
+	if (vdso_alloc_per_cpu(cpu, lowcore))
+		goto out;
 #endif
 	lowcore_ptr[cpu] = lowcore;
 	return 0;
@@ -522,6 +526,8 @@ static void smp_free_lowcore(int cpu)
 #ifndef CONFIG_64BIT
 	if (MACHINE_HAS_IEEE)
 		free_page((unsigned long) lowcore->extended_save_area_addr);
+#else
+	vdso_free_per_cpu(cpu, lowcore);
 #endif
 	free_page(lowcore->panic_stack - PAGE_SIZE);
 	free_pages(lowcore->async_stack - ASYNC_SIZE, ASYNC_ORDER);
@@ -664,6 +670,7 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	lowcore = (void *) __get_free_pages(GFP_KERNEL | GFP_DMA, lc_order);
 	panic_stack = __get_free_page(GFP_KERNEL);
 	async_stack = __get_free_pages(GFP_KERNEL, ASYNC_ORDER);
+	BUG_ON(!lowcore || !panic_stack || !async_stack);
 #ifndef CONFIG_64BIT
 	if (MACHINE_HAS_IEEE)
 		save_area = get_zeroed_page(GFP_KERNEL);
@@ -677,6 +684,8 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 #ifndef CONFIG_64BIT
 	if (MACHINE_HAS_IEEE)
 		lowcore->extended_save_area_addr = (u32) save_area;
+#else
+	BUG_ON(vdso_alloc_per_cpu(smp_processor_id(), lowcore));
 #endif
 	set_prefix((u32)(unsigned long) lowcore);
 	local_mcck_enable();
@@ -845,9 +854,11 @@ static ssize_t show_idle_count(struct sys_device *dev,
 	unsigned long long idle_count;
 
 	idle = &per_cpu(s390_idle, dev->id);
-	spin_lock_irq(&idle->lock);
+	spin_lock(&idle->lock);
 	idle_count = idle->idle_count;
-	spin_unlock_irq(&idle->lock);
+	if (idle->idle_enter)
+		idle_count++;
+	spin_unlock(&idle->lock);
 	return sprintf(buf, "%llu\n", idle_count);
 }
 static SYSDEV_ATTR(idle_count, 0444, show_idle_count, NULL);
@@ -856,18 +867,17 @@ static ssize_t show_idle_time(struct sys_device *dev,
 				struct sysdev_attribute *attr, char *buf)
 {
 	struct s390_idle_data *idle;
-	unsigned long long new_time;
+	unsigned long long now, idle_time, idle_enter;
 
 	idle = &per_cpu(s390_idle, dev->id);
-	spin_lock_irq(&idle->lock);
-	if (idle->in_idle) {
-		new_time = get_clock();
-		idle->idle_time += new_time - idle->idle_enter;
-		idle->idle_enter = new_time;
-	}
-	new_time = idle->idle_time;
-	spin_unlock_irq(&idle->lock);
-	return sprintf(buf, "%llu\n", new_time >> 12);
+	spin_lock(&idle->lock);
+	now = get_clock();
+	idle_time = idle->idle_time;
+	idle_enter = idle->idle_enter;
+	if (idle_enter != 0ULL && idle_enter < now)
+		idle_time += now - idle_enter;
+	spin_unlock(&idle->lock);
+	return sprintf(buf, "%llu\n", idle_time >> 12);
 }
 static SYSDEV_ATTR(idle_time_us, 0444, show_idle_time, NULL);
 

arch/s390/kernel/vdso.c

@@ -31,9 +31,6 @@
 #include <asm/sections.h>
 #include <asm/vdso.h>
 
-/* Max supported size for symbol names */
-#define MAX_SYMNAME	64
-
 #if defined(CONFIG_32BIT) || defined(CONFIG_COMPAT)
 extern char vdso32_start, vdso32_end;
 static void *vdso32_kbase = &vdso32_start;
@@ -70,6 +67,119 @@ static union {
 } vdso_data_store __attribute__((__section__(".data.page_aligned")));
 struct vdso_data *vdso_data = &vdso_data_store.data;
 
+/*
+ * Setup vdso data page.
+ */
+static void vdso_init_data(struct vdso_data *vd)
+{
+	unsigned int facility_list;
+
+	facility_list = stfl();
+	vd->ectg_available = switch_amode && (facility_list & 1);
+}
+
+#ifdef CONFIG_64BIT
+/*
+ * Setup per cpu vdso data page.
+ */
+static void vdso_init_per_cpu_data(int cpu, struct vdso_per_cpu_data *vpcd)
+{
+}
+
+/*
+ * Allocate/free per cpu vdso data.
+ */
+#ifdef CONFIG_64BIT
+#define SEGMENT_ORDER	2
+#else
+#define SEGMENT_ORDER	1
+#endif
+
+int vdso_alloc_per_cpu(int cpu, struct _lowcore *lowcore)
+{
+	unsigned long segment_table, page_table, page_frame;
+	u32 *psal, *aste;
+	int i;
+
+	lowcore->vdso_per_cpu_data = __LC_PASTE;
+
+	if (!switch_amode || !vdso_enabled)
+		return 0;
+
+	segment_table = __get_free_pages(GFP_KERNEL, SEGMENT_ORDER);
+	page_table = get_zeroed_page(GFP_KERNEL | GFP_DMA);
+	page_frame = get_zeroed_page(GFP_KERNEL);
+	if (!segment_table || !page_table || !page_frame)
+		goto out;
+
+	clear_table((unsigned long *) segment_table, _SEGMENT_ENTRY_EMPTY,
+		    PAGE_SIZE << SEGMENT_ORDER);
+	clear_table((unsigned long *) page_table, _PAGE_TYPE_EMPTY,
+		    256*sizeof(unsigned long));
+
+	*(unsigned long *) segment_table = _SEGMENT_ENTRY + page_table;
+	*(unsigned long *) page_table = _PAGE_RO + page_frame;
+
+	psal = (u32 *) (page_table + 256*sizeof(unsigned long));
+	aste = psal + 32;
+
+	for (i = 4; i < 32; i += 4)
+		psal[i] = 0x80000000;
+
+	lowcore->paste[4] = (u32)(addr_t) psal;
+	psal[0] = 0x20000000;
+	psal[2] = (u32)(addr_t) aste;
+	*(unsigned long *) (aste + 2) = segment_table +
+		_ASCE_TABLE_LENGTH + _ASCE_USER_BITS + _ASCE_TYPE_SEGMENT;
+	aste[4] = (u32)(addr_t) psal;
+	lowcore->vdso_per_cpu_data = page_frame;
+
+	vdso_init_per_cpu_data(cpu, (struct vdso_per_cpu_data *) page_frame);
+	return 0;
+
+out:
+	free_page(page_frame);
+	free_page(page_table);
+	free_pages(segment_table, SEGMENT_ORDER);
+	return -ENOMEM;
+}
+
+#ifdef CONFIG_HOTPLUG_CPU
+void vdso_free_per_cpu(int cpu, struct _lowcore *lowcore)
+{
+	unsigned long segment_table, page_table, page_frame;
+	u32 *psal, *aste;
+
+	if (!switch_amode || !vdso_enabled)
+		return;
+
+	psal = (u32 *)(addr_t) lowcore->paste[4];
+	aste = (u32 *)(addr_t) psal[2];
+	segment_table = *(unsigned long *)(aste + 2) & PAGE_MASK;
+	page_table = *(unsigned long *) segment_table;
+	page_frame = *(unsigned long *) page_table;
+
+	free_page(page_frame);
+	free_page(page_table);
+	free_pages(segment_table, SEGMENT_ORDER);
+}
+#endif /* CONFIG_HOTPLUG_CPU */
+
+static void __vdso_init_cr5(void *dummy)
+{
+	unsigned long cr5;
+
+	cr5 = offsetof(struct _lowcore, paste);
+	__ctl_load(cr5, 5, 5);
+}
+
+static void vdso_init_cr5(void)
+{
+	if (switch_amode && vdso_enabled)
+		on_each_cpu(__vdso_init_cr5, NULL, 1);
+}
+#endif /* CONFIG_64BIT */
+
 /*
  * This is called from binfmt_elf, we create the special vma for the
  * vDSO and insert it into the mm struct tree
@@ -172,6 +282,9 @@ static int __init vdso_init(void)
 {
 	int i;
 
+	if (!vdso_enabled)
+		return 0;
+	vdso_init_data(vdso_data);
 #if defined(CONFIG_32BIT) || defined(CONFIG_COMPAT)
 	/* Calculate the size of the 32 bit vDSO */
 	vdso32_pages = ((&vdso32_end - &vdso32_start
@@ -208,6 +321,10 @@ static int __init vdso_init(void)
 	}
 	vdso64_pagelist[vdso64_pages - 1] = virt_to_page(vdso_data);
 	vdso64_pagelist[vdso64_pages] = NULL;
+#ifndef CONFIG_SMP
+	BUG_ON(vdso_alloc_per_cpu(0, S390_lowcore));
+#endif
+	vdso_init_cr5();
 #endif /* CONFIG_64BIT */
 
 	get_page(virt_to_page(vdso_data));

arch/s390/kernel/vdso64/clock_getres.S

@@ -22,7 +22,12 @@ __kernel_clock_getres:
 	cghi	%r2,CLOCK_REALTIME
 	je	0f
 	cghi	%r2,CLOCK_MONOTONIC
+	je	0f
+	cghi	%r2,-2		/* CLOCK_THREAD_CPUTIME_ID for this thread */
 	jne	2f
+	larl	%r5,_vdso_data
+	icm	%r0,15,__LC_ECTG_OK(%r5)
+	jz	2f
 0:	ltgr	%r3,%r3
 	jz	1f				/* res == NULL */
 	larl	%r1,3f

arch/s390/kernel/vdso64/clock_gettime.S

@@ -22,8 +22,10 @@ __kernel_clock_gettime:
 	larl	%r5,_vdso_data
 	cghi	%r2,CLOCK_REALTIME
 	je	4f
+	cghi	%r2,-2		/* CLOCK_THREAD_CPUTIME_ID for this thread */
+	je	9f
 	cghi	%r2,CLOCK_MONOTONIC
-	jne	9f
+	jne	12f
 
 	/* CLOCK_MONOTONIC */
 	ltgr	%r3,%r3
@@ -42,7 +44,7 @@ __kernel_clock_gettime:
 	alg	%r0,__VDSO_WTOM_SEC(%r5)
 	clg	%r4,__VDSO_UPD_COUNT(%r5)	/* check update counter */
 	jne	0b
-	larl	%r5,10f
+	larl	%r5,13f
 1:	clg	%r1,0(%r5)
 	jl	2f
 	slg	%r1,0(%r5)
@@ -68,7 +70,7 @@ __kernel_clock_gettime:
 	lg	%r0,__VDSO_XTIME_SEC(%r5)
 	clg	%r4,__VDSO_UPD_COUNT(%r5)	/* check update counter */
 	jne	5b
-	larl	%r5,10f
+	larl	%r5,13f
 6:	clg	%r1,0(%r5)
 	jl	7f
 	slg	%r1,0(%r5)
@@ -79,11 +81,38 @@ __kernel_clock_gettime:
 8:	lghi	%r2,0
 	br	%r14
 
+	/* CLOCK_THREAD_CPUTIME_ID for this thread */
+9:	icm	%r0,15,__VDSO_ECTG_OK(%r5)
+	jz	12f
+	ear	%r2,%a4
+	llilh	%r4,0x0100
+	sar	%a4,%r4
+	lghi	%r4,0
+	sacf	512				/* Magic ectg instruction */
+	.insn	ssf,0xc80100000000,__VDSO_ECTG_BASE(4),__VDSO_ECTG_USER(4),4
+	sacf	0
+	sar	%a4,%r2
+	algr	%r1,%r0				/* r1 = cputime as TOD value */
+	mghi	%r1,1000			/* convert to nanoseconds */
+	srlg	%r1,%r1,12			/* r1 = cputime in nanosec */
+	lgr	%r4,%r1
+	larl	%r5,13f
+	srlg	%r1,%r1,9			/* divide by 1000000000 */
+	mlg	%r0,8(%r5)
+	srlg	%r0,%r0,11			/* r0 = tv_sec */
+	stg	%r0,0(%r3)
+	msg	%r0,0(%r5)			/* calculate tv_nsec */
+	slgr	%r4,%r0				/* r4 = tv_nsec */
+	stg	%r4,8(%r3)
+	lghi	%r2,0
+	br	%r14
+
 	/* Fallback to system call */
-9:	lghi	%r1,__NR_clock_gettime
+12:	lghi	%r1,__NR_clock_gettime
 	svc	0
 	br	%r14
 
-10:	.quad	1000000000
+13:	.quad	1000000000
+14:	.quad	19342813113834067
 	.cfi_endproc
 	.size	__kernel_clock_gettime,.-__kernel_clock_gettime

arch/s390/kernel/vtime.c

@@ -23,19 +23,43 @@
 #include <asm/s390_ext.h>
 #include <asm/timer.h>
 #include <asm/irq_regs.h>
+#include <asm/cpu.h>
 
 static ext_int_info_t ext_int_info_timer;
+
 static DEFINE_PER_CPU(struct vtimer_queue, virt_cpu_timer);
 
+DEFINE_PER_CPU(struct s390_idle_data, s390_idle) = {
+	.lock = __SPIN_LOCK_UNLOCKED(s390_idle.lock)
+};
+
+static inline __u64 get_vtimer(void)
+{
+	__u64 timer;
+
+	asm volatile("STPT %0" : "=m" (timer));
+	return timer;
+}
+
+static inline void set_vtimer(__u64 expires)
+{
+	__u64 timer;
+
+	asm volatile ("  STPT %0\n"  /* Store current cpu timer value */
+		      "  SPT %1"     /* Set new value immediatly afterwards */
+		      : "=m" (timer) : "m" (expires) );
+	S390_lowcore.system_timer += S390_lowcore.last_update_timer - timer;
+	S390_lowcore.last_update_timer = expires;
+}
+
 /*
  * Update process times based on virtual cpu times stored by entry.S
  * to the lowcore fields user_timer, system_timer & steal_clock.
  */
-void account_process_tick(struct task_struct *tsk, int user_tick)
+static void do_account_vtime(struct task_struct *tsk, int hardirq_offset)
 {
-	cputime_t cputime;
-	__u64 timer, clock;
-	int rcu_user_flag;
+	struct thread_info *ti = task_thread_info(tsk);
+	__u64 timer, clock, user, system, steal;
 
 	timer = S390_lowcore.last_update_timer;
 	clock = S390_lowcore.last_update_clock;
@@ -44,50 +68,41 @@ void account_process_tick(struct task_struct *tsk, int user_tick)
 		      : "=m" (S390_lowcore.last_update_timer),
 		        "=m" (S390_lowcore.last_update_clock) );
 	S390_lowcore.system_timer += timer - S390_lowcore.last_update_timer;
-	S390_lowcore.steal_clock += S390_lowcore.last_update_clock - clock;
-
-	cputime = S390_lowcore.user_timer >> 12;
-	rcu_user_flag = cputime != 0;
-	S390_lowcore.user_timer -= cputime << 12;
-	S390_lowcore.steal_clock -= cputime << 12;
-	account_user_time(tsk, cputime);
-
-	cputime =  S390_lowcore.system_timer >> 12;
-	S390_lowcore.system_timer -= cputime << 12;
-	S390_lowcore.steal_clock -= cputime << 12;
-	account_system_time(tsk, HARDIRQ_OFFSET, cputime);
-
-	cputime = S390_lowcore.steal_clock;
-	if ((__s64) cputime > 0) {
-		cputime >>= 12;
-		S390_lowcore.steal_clock -= cputime << 12;
-		account_steal_time(tsk, cputime);
+	S390_lowcore.steal_timer += S390_lowcore.last_update_clock - clock;
+
+	user = S390_lowcore.user_timer - ti->user_timer;
+	S390_lowcore.steal_timer -= user;
+	ti->user_timer = S390_lowcore.user_timer;
+	account_user_time(tsk, user, user);
+
+	system = S390_lowcore.system_timer - ti->system_timer;
+	S390_lowcore.steal_timer -= system;
+	ti->system_timer = S390_lowcore.system_timer;
+	account_system_time(tsk, hardirq_offset, system, system);
+
+	steal = S390_lowcore.steal_timer;
+	if ((s64) steal > 0) {
+		S390_lowcore.steal_timer = 0;
+		account_steal_time(steal);
 	}
 }
 
-/*
- * Update process times based on virtual cpu times stored by entry.S
- * to the lowcore fields user_timer, system_timer & steal_clock.
- */
-void account_vtime(struct task_struct *tsk)
+void account_vtime(struct task_struct *prev, struct task_struct *next)
 {
-	cputime_t cputime;
-	__u64 timer;
-
-	timer = S390_lowcore.last_update_timer;
-	asm volatile ("  STPT %0"    /* Store current cpu timer value */
-		      : "=m" (S390_lowcore.last_update_timer) );
-	S390_lowcore.system_timer += timer - S390_lowcore.last_update_timer;
-
-	cputime = S390_lowcore.user_timer >> 12;
-	S390_lowcore.user_timer -= cputime << 12;
-	S390_lowcore.steal_clock -= cputime << 12;
-	account_user_time(tsk, cputime);
+	struct thread_info *ti;
+
+	do_account_vtime(prev, 0);
+	ti = task_thread_info(prev);
+	ti->user_timer = S390_lowcore.user_timer;
+	ti->system_timer = S390_lowcore.system_timer;
+	ti = task_thread_info(next);
+	S390_lowcore.user_timer = ti->user_timer;
+	S390_lowcore.system_timer = ti->system_timer;
+}
 
-	cputime =  S390_lowcore.system_timer >> 12;
-	S390_lowcore.system_timer -= cputime << 12;
-	S390_lowcore.steal_clock -= cputime << 12;
-	account_system_time(tsk, 0, cputime);
+void account_process_tick(struct task_struct *tsk, int user_tick)
+{
+	do_account_vtime(tsk, HARDIRQ_OFFSET);
 }
 
 /*
@@ -96,80 +111,131 @@ void account_vtime(struct task_struct *tsk)
  */
 void account_system_vtime(struct task_struct *tsk)
 {
-	cputime_t cputime;
-	__u64 timer;
+	struct thread_info *ti = task_thread_info(tsk);
+	__u64 timer, system;
 
 	timer = S390_lowcore.last_update_timer;
-	asm volatile ("  STPT %0"    /* Store current cpu timer value */
-		      : "=m" (S390_lowcore.last_update_timer) );
+	S390_lowcore.last_update_timer = get_vtimer();
 	S390_lowcore.system_timer += timer - S390_lowcore.last_update_timer;
 
-	cputime =  S390_lowcore.system_timer >> 12;
-	S390_lowcore.system_timer -= cputime << 12;
-	S390_lowcore.steal_clock -= cputime << 12;
-	account_system_time(tsk, 0, cputime);
+	system = S390_lowcore.system_timer - ti->system_timer;
+	S390_lowcore.steal_timer -= system;
+	ti->system_timer = S390_lowcore.system_timer;
+	account_system_time(tsk, 0, system, system);
 }
 EXPORT_SYMBOL_GPL(account_system_vtime);
 
-static inline void set_vtimer(__u64 expires)
+void vtime_start_cpu(void)
 {
-	__u64 timer;
-
-	asm volatile ("  STPT %0\n"  /* Store current cpu timer value */
-		      "  SPT %1"     /* Set new value immediatly afterwards */
-		      : "=m" (timer) : "m" (expires) );
-	S390_lowcore.system_timer += S390_lowcore.last_update_timer - timer;
-	S390_lowcore.last_update_timer = expires;
-
-	/* store expire time for this CPU timer */
-	__get_cpu_var(virt_cpu_timer).to_expire = expires;
-}
-
-void vtime_start_cpu_timer(void)
-{
-	struct vtimer_queue *vt_list;
-
-	vt_list = &__get_cpu_var(virt_cpu_timer);
-
-	/* CPU timer interrupt is pending, don't reprogramm it */
-	if (vt_list->idle & 1LL<<63)
-		return;
+	struct s390_idle_data *idle = &__get_cpu_var(s390_idle);
+	struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer);
+	__u64 idle_time, expires;
+
+	/* Account time spent with enabled wait psw loaded as idle time. */
+	idle_time = S390_lowcore.int_clock - idle->idle_enter;
+	account_idle_time(idle_time);
+	S390_lowcore.last_update_clock = S390_lowcore.int_clock;
+
+	/* Account system time spent going idle. */
+	S390_lowcore.system_timer += S390_lowcore.last_update_timer - vq->idle;
+	S390_lowcore.last_update_timer = S390_lowcore.async_enter_timer;
+
+	/* Restart vtime CPU timer */
+	if (vq->do_spt) {
+		/* Program old expire value but first save progress. */
+		expires = vq->idle - S390_lowcore.async_enter_timer;
+		expires += get_vtimer();
+		set_vtimer(expires);
+	} else {
+		/* Don't account the CPU timer delta while the cpu was idle. */
+		vq->elapsed -= vq->idle - S390_lowcore.async_enter_timer;
+	}
 
-	if (!list_empty(&vt_list->list))
-		set_vtimer(vt_list->idle);
+	spin_lock(&idle->lock);
+	idle->idle_time += idle_time;
+	idle->idle_enter = 0ULL;
+	idle->idle_count++;
+	spin_unlock(&idle->lock);
 }
 
-void vtime_stop_cpu_timer(void)
+void vtime_stop_cpu(void)
 {
-	struct vtimer_queue *vt_list;
-
-	vt_list = &__get_cpu_var(virt_cpu_timer);
+	struct s390_idle_data *idle = &__get_cpu_var(s390_idle);
+	struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer);
+	psw_t psw;
 
-	/* nothing to do */
-	if (list_empty(&vt_list->list)) {
-		vt_list->idle = VTIMER_MAX_SLICE;
-		goto fire;
-	}
-
-	/* store the actual expire value */
-	asm volatile ("STPT %0" : "=m" (vt_list->idle));
+	/* Wait for external, I/O or machine check interrupt. */
+	psw.mask = psw_kernel_bits | PSW_MASK_WAIT | PSW_MASK_IO | PSW_MASK_EXT;
 
+	/* Check if the CPU timer needs to be reprogrammed. */
+	if (vq->do_spt) {
+		__u64 vmax = VTIMER_MAX_SLICE;
 		/*
-	 * If the CPU timer is negative we don't reprogramm
-	 * it because we will get instantly an interrupt.
+		 * The inline assembly is equivalent to
+		 *	vq->idle = get_cpu_timer();
+		 *	set_cpu_timer(VTIMER_MAX_SLICE);
+		 *	idle->idle_enter = get_clock();
+		 *	__load_psw_mask(psw_kernel_bits | PSW_MASK_WAIT |
+		 *			   PSW_MASK_IO | PSW_MASK_EXT);
+		 * The difference is that the inline assembly makes sure that
+		 * the last three instruction are stpt, stck and lpsw in that
+		 * order. This is done to increase the precision.
 		 */
-	if (vt_list->idle & 1LL<<63)
-		return;
-
-	vt_list->offset += vt_list->to_expire - vt_list->idle;
-
+		asm volatile(
+#ifndef CONFIG_64BIT
+			"	basr	1,0\n"
+			"0:	ahi	1,1f-0b\n"
+			"	st	1,4(%2)\n"
+#else /* CONFIG_64BIT */
+			"	larl	1,1f\n"
+			"	stg	1,8(%2)\n"
+#endif /* CONFIG_64BIT */
+			"	stpt	0(%4)\n"
+			"	spt	0(%5)\n"
+			"	stck	0(%3)\n"
+#ifndef CONFIG_64BIT
+			"	lpsw	0(%2)\n"
+#else /* CONFIG_64BIT */
+			"	lpswe	0(%2)\n"
+#endif /* CONFIG_64BIT */
+			"1:"
+			: "=m" (idle->idle_enter), "=m" (vq->idle)
+			: "a" (&psw), "a" (&idle->idle_enter),
+			  "a" (&vq->idle), "a" (&vmax), "m" (vmax), "m" (psw)
+			: "memory", "cc", "1");
+	} else {
 		/*
-	 * We cannot halt the CPU timer, we just write a value that
-	 * nearly never expires (only after 71 years) and re-write
-	 * the stored expire value if we continue the timer
+		 * The inline assembly is equivalent to
+		 *	vq->idle = get_cpu_timer();
+		 *	idle->idle_enter = get_clock();
+		 *	__load_psw_mask(psw_kernel_bits | PSW_MASK_WAIT |
+		 *			   PSW_MASK_IO | PSW_MASK_EXT);
+		 * The difference is that the inline assembly makes sure that
+		 * the last three instruction are stpt, stck and lpsw in that
+		 * order. This is done to increase the precision.
 		 */
- fire:
-	set_vtimer(VTIMER_MAX_SLICE);
+		asm volatile(
+#ifndef CONFIG_64BIT
+			"	basr	1,0\n"
+			"0:	ahi	1,1f-0b\n"
+			"	st	1,4(%2)\n"
+#else /* CONFIG_64BIT */
+			"	larl	1,1f\n"
+			"	stg	1,8(%2)\n"
+#endif /* CONFIG_64BIT */
+			"	stpt	0(%4)\n"
+			"	stck	0(%3)\n"
+#ifndef CONFIG_64BIT
+			"	lpsw	0(%2)\n"
+#else /* CONFIG_64BIT */
+			"	lpswe	0(%2)\n"
+#endif /* CONFIG_64BIT */
+			"1:"
+			: "=m" (idle->idle_enter), "=m" (vq->idle)
+			: "a" (&psw), "a" (&idle->idle_enter),
+			  "a" (&vq->idle), "m" (psw)
+			: "memory", "cc", "1");
+	}
 }
 
 /*
@@ -195,30 +261,23 @@ static void list_add_sorted(struct vtimer_list *timer, struct list_head *head)
  */
 static void do_callbacks(struct list_head *cb_list)
 {
-	struct vtimer_queue *vt_list;
+	struct vtimer_queue *vq;
 	struct vtimer_list *event, *tmp;
-	void (*fn)(unsigned long);
-	unsigned long data;
 
 	if (list_empty(cb_list))
 		return;
 
-	vt_list = &__get_cpu_var(virt_cpu_timer);
+	vq = &__get_cpu_var(virt_cpu_timer);
 
 	list_for_each_entry_safe(event, tmp, cb_list, entry) {
-		fn = event->function;
-		data = event->data;
-		fn(data);
-
-		if (!event->interval)
-			/* delete one shot timer */
 		list_del_init(&event->entry);
-		else {
-			/* move interval timer back to list */
-			spin_lock(&vt_list->lock);
-			list_del_init(&event->entry);
-			list_add_sorted(event, &vt_list->list);
-			spin_unlock(&vt_list->lock);
+		(event->function)(event->data);
+		if (event->interval) {
+			/* Recharge interval timer */
+			event->expires = event->interval + vq->elapsed;
+			spin_lock(&vq->lock);
+			list_add_sorted(event, &vq->list);
+			spin_unlock(&vq->lock);
 		}
 	}
 }
@@ -228,64 +287,57 @@ static void do_callbacks(struct list_head *cb_list)
  */
 static void do_cpu_timer_interrupt(__u16 error_code)
 {
-	__u64 next, delta;
-	struct vtimer_queue *vt_list;
+	struct vtimer_queue *vq;
 	struct vtimer_list *event, *tmp;
-	struct list_head *ptr;
-	/* the callback queue */
-	struct list_head cb_list;
+	struct list_head cb_list;	/* the callback queue */
+	__u64 elapsed, next;
 
 	INIT_LIST_HEAD(&cb_list);
-	vt_list = &__get_cpu_var(virt_cpu_timer);
+	vq = &__get_cpu_var(virt_cpu_timer);
 
 	/* walk timer list, fire all expired events */
-	spin_lock(&vt_list->lock);
-
-	if (vt_list->to_expire < VTIMER_MAX_SLICE)
-		vt_list->offset += vt_list->to_expire;
-
-	list_for_each_entry_safe(event, tmp, &vt_list->list, entry) {
-		if (event->expires > vt_list->offset)
-			/* found first unexpired event, leave */
-			break;
-
-		/* re-charge interval timer, we have to add the offset */
-		if (event->interval)
-			event->expires = event->interval + vt_list->offset;
+	spin_lock(&vq->lock);
 
+	elapsed = vq->elapsed + (vq->timer - S390_lowcore.async_enter_timer);
+	BUG_ON((s64) elapsed < 0);
+	vq->elapsed = 0;
+	list_for_each_entry_safe(event, tmp, &vq->list, entry) {
+		if (event->expires < elapsed)
 			/* move expired timer to the callback queue */
 			list_move_tail(&event->entry, &cb_list);
+		else
+			event->expires -= elapsed;
 	}
-	spin_unlock(&vt_list->lock);
+	spin_unlock(&vq->lock);
+
+	vq->do_spt = list_empty(&cb_list);
 	do_callbacks(&cb_list);
 
 	/* next event is first in list */
-	spin_lock(&vt_list->lock);
-	if (!list_empty(&vt_list->list)) {
-		ptr = vt_list->list.next;
-		event = list_entry(ptr, struct vtimer_list, entry);
-		next = event->expires - vt_list->offset;
-
-		/* add the expired time from this interrupt handler
-		 * and the callback functions
-		 */
-		asm volatile ("STPT %0" : "=m" (delta));
-		delta = 0xffffffffffffffffLL - delta + 1;
-		vt_list->offset += delta;
-		next -= delta;
-	} else {
-		vt_list->offset = 0;
 	next = VTIMER_MAX_SLICE;
-	}
-	spin_unlock(&vt_list->lock);
-	set_vtimer(next);
+	spin_lock(&vq->lock);
+	if (!list_empty(&vq->list)) {
+		event = list_first_entry(&vq->list, struct vtimer_list, entry);
+		next = event->expires;
+	} else
+		vq->do_spt = 0;
+	spin_unlock(&vq->lock);
+	/*
+	 * To improve precision add the time spent by the
+	 * interrupt handler to the elapsed time.
+	 * Note: CPU timer counts down and we got an interrupt,
+	 *	 the current content is negative
+	 */
+	elapsed = S390_lowcore.async_enter_timer - get_vtimer();
+	set_vtimer(next - elapsed);
+	vq->timer = next - elapsed;
+	vq->elapsed = elapsed;
 }
 
 void init_virt_timer(struct vtimer_list *timer)
 {
 	timer->function = NULL;
 	INIT_LIST_HEAD(&timer->entry);
-	spin_lock_init(&timer->lock);
 }
 EXPORT_SYMBOL(init_virt_timer);
 
@@ -299,44 +351,40 @@ static inline int vtimer_pending(struct vtimer_list *timer)
  */
 static void internal_add_vtimer(struct vtimer_list *timer)
 {
+	struct vtimer_queue *vq;
 	unsigned long flags;
-	__u64 done;
-	struct vtimer_list *event;
-	struct vtimer_queue *vt_list;
+	__u64 left, expires;
 
-	vt_list = &per_cpu(virt_cpu_timer, timer->cpu);
-	spin_lock_irqsave(&vt_list->lock, flags);
+	vq = &per_cpu(virt_cpu_timer, timer->cpu);
+	spin_lock_irqsave(&vq->lock, flags);
 
 	BUG_ON(timer->cpu != smp_processor_id());
 
-	/* if list is empty we only have to set the timer */
-	if (list_empty(&vt_list->list)) {
-		/* reset the offset, this may happen if the last timer was
-		 * just deleted by mod_virt_timer and the interrupt
-		 * didn't happen until here
-		 */
-		vt_list->offset = 0;
-		goto fire;
+	if (list_empty(&vq->list)) {
+		/* First timer on this cpu, just program it. */
+		list_add(&timer->entry, &vq->list);
+		set_vtimer(timer->expires);
+		vq->timer = timer->expires;
+		vq->elapsed = 0;
+	} else {
+		/* Check progress of old timers. */
+		expires = timer->expires;
+		left = get_vtimer();
+		if (likely((s64) expires < (s64) left)) {
+			/* The new timer expires before the current timer. */
+			set_vtimer(expires);
+			vq->elapsed += vq->timer - left;
+			vq->timer = expires;
+		} else {
+			vq->elapsed += vq->timer - left;
+			vq->timer = left;
+		}
+		/* Insert new timer into per cpu list. */
+		timer->expires += vq->elapsed;
+		list_add_sorted(timer, &vq->list);
 	}
 
-	/* save progress */
-	asm volatile ("STPT %0" : "=m" (done));
-
-	/* calculate completed work */
-	done = vt_list->to_expire - done + vt_list->offset;
-	vt_list->offset = 0;
-
-	list_for_each_entry(event, &vt_list->list, entry)
-		event->expires -= done;
-
- fire:
-	list_add_sorted(timer, &vt_list->list);
-
-	/* get first element, which is the next vtimer slice */
-	event = list_entry(vt_list->list.next, struct vtimer_list, entry);
-
-	set_vtimer(event->expires);
-	spin_unlock_irqrestore(&vt_list->lock, flags);
+	spin_unlock_irqrestore(&vq->lock, flags);
 	/* release CPU acquired in prepare_vtimer or mod_virt_timer() */
 	put_cpu();
 }
@@ -381,14 +429,15 @@ EXPORT_SYMBOL(add_virt_timer_periodic);
  * If we change a pending timer the function must be called on the CPU
  * where the timer is running on, e.g. by smp_call_function_single()
  *
- * The original mod_timer adds the timer if it is not pending. For compatibility
- * we do the same. The timer will be added on the current CPU as a oneshot timer.
+ * The original mod_timer adds the timer if it is not pending. For
+ * compatibility we do the same. The timer will be added on the current
+ * CPU as a oneshot timer.
  *
  * returns whether it has modified a pending timer (1) or not (0)
  */
 int mod_virt_timer(struct vtimer_list *timer, __u64 expires)
 {
-	struct vtimer_queue *vt_list;
+	struct vtimer_queue *vq;
 	unsigned long flags;
 	int cpu;
 
@@ -404,17 +453,17 @@ int mod_virt_timer(struct vtimer_list *timer, __u64 expires)
 		return 1;
 
 	cpu = get_cpu();
-	vt_list = &per_cpu(virt_cpu_timer, cpu);
+	vq = &per_cpu(virt_cpu_timer, cpu);
 
 	/* check if we run on the right CPU */
 	BUG_ON(timer->cpu != cpu);
 
 	/* disable interrupts before test if timer is pending */
-	spin_lock_irqsave(&vt_list->lock, flags);
+	spin_lock_irqsave(&vq->lock, flags);
 
 	/* if timer isn't pending add it on the current CPU */
 	if (!vtimer_pending(timer)) {
-		spin_unlock_irqrestore(&vt_list->lock, flags);
+		spin_unlock_irqrestore(&vq->lock, flags);
 		/* we do not activate an interval timer with mod_virt_timer */
 		timer->interval = 0;
 		timer->expires = expires;
@@ -431,7 +480,7 @@ int mod_virt_timer(struct vtimer_list *timer, __u64 expires)
 		timer->interval = expires;
 
 	/* the timer can't expire anymore so we can release the lock */
-	spin_unlock_irqrestore(&vt_list->lock, flags);
+	spin_unlock_irqrestore(&vq->lock, flags);
 	internal_add_vtimer(timer);
 	return 1;
 }
@@ -445,25 +494,19 @@ EXPORT_SYMBOL(mod_virt_timer);
 int del_virt_timer(struct vtimer_list *timer)
 {
 	unsigned long flags;
-	struct vtimer_queue *vt_list;
+	struct vtimer_queue *vq;
 
 	/* check if timer is pending */
 	if (!vtimer_pending(timer))
 		return 0;
 
-	vt_list = &per_cpu(virt_cpu_timer, timer->cpu);
-	spin_lock_irqsave(&vt_list->lock, flags);
+	vq = &per_cpu(virt_cpu_timer, timer->cpu);
+	spin_lock_irqsave(&vq->lock, flags);
 
 	/* we don't interrupt a running timer, just let it expire! */
 	list_del_init(&timer->entry);
 
-	/* last timer removed */
-	if (list_empty(&vt_list->list)) {
-		vt_list->to_expire = 0;
-		vt_list->offset = 0;
-	}
-
-	spin_unlock_irqrestore(&vt_list->lock, flags);
+	spin_unlock_irqrestore(&vq->lock, flags);
 	return 1;
 }
 EXPORT_SYMBOL(del_virt_timer);
@@ -473,24 +516,19 @@ EXPORT_SYMBOL(del_virt_timer);
  */
 void init_cpu_vtimer(void)
 {
-	struct vtimer_queue *vt_list;
+	struct vtimer_queue *vq;
 
 	/* kick the virtual timer */
-	S390_lowcore.exit_timer = VTIMER_MAX_SLICE;
-	S390_lowcore.last_update_timer = VTIMER_MAX_SLICE;
-	asm volatile ("SPT %0" : : "m" (S390_lowcore.last_update_timer));
 	asm volatile ("STCK %0" : "=m" (S390_lowcore.last_update_clock));
+	asm volatile ("STPT %0" : "=m" (S390_lowcore.last_update_timer));
+
+	/* initialize per cpu vtimer structure */
+	vq = &__get_cpu_var(virt_cpu_timer);
+	INIT_LIST_HEAD(&vq->list);
+	spin_lock_init(&vq->lock);
 
 	/* enable cpu timer interrupts */
 	__ctl_set_bit(0,10);
-
-	vt_list = &__get_cpu_var(virt_cpu_timer);
-	INIT_LIST_HEAD(&vt_list->list);
-	spin_lock_init(&vt_list->lock);
-	vt_list->to_expire = 0;
-	vt_list->offset = 0;
-	vt_list->idle = 0;
-
 }
 
 void __init vtime_init(void)

arch/x86/xen/time.c

@@ -132,8 +132,7 @@ static void do_stolen_accounting(void)
 	*snap = state;
 
 	/* Add the appropriate number of ticks of stolen time,
-	   including any left-overs from last time.  Passing NULL to
-	   account_steal_time accounts the time as stolen. */
+	   including any left-overs from last time. */
 	stolen = runnable + offline + __get_cpu_var(residual_stolen);
 
 	if (stolen < 0)
@@ -141,11 +140,10 @@ static void do_stolen_accounting(void)
 
 	ticks = iter_div_u64_rem(stolen, NS_PER_TICK, &stolen);
 	__get_cpu_var(residual_stolen) = stolen;
-	account_steal_time(NULL, ticks);
+	account_steal_ticks(ticks);
 
 	/* Add the appropriate number of ticks of blocked time,
-	   including any left-overs from last time.  Passing idle to
-	   account_steal_time accounts the time as idle/wait. */
+	   including any left-overs from last time. */
 	blocked += __get_cpu_var(residual_blocked);
 
 	if (blocked < 0)
@@ -153,7 +151,7 @@ static void do_stolen_accounting(void)
 
 	ticks = iter_div_u64_rem(blocked, NS_PER_TICK, &blocked);
 	__get_cpu_var(residual_blocked) = blocked;
-	account_steal_time(idle_task(smp_processor_id()), ticks);
+	account_idle_ticks(ticks);
 }
 
 /*

drivers/s390/cio/cio.c

@@ -632,8 +632,8 @@ do_IRQ (struct pt_regs *regs)
 	struct pt_regs *old_regs;
 
 	old_regs = set_irq_regs(regs);
-	irq_enter();
 	s390_idle_check();
+	irq_enter();
 	if (S390_lowcore.int_clock >= S390_lowcore.clock_comparator)
 		/* Serve timer interrupts first. */
 		clock_comparator_work();

drivers/s390/s390mach.c

@@ -18,6 +18,7 @@
 #include <asm/etr.h>
 #include <asm/lowcore.h>
 #include <asm/cio.h>
+#include <asm/cpu.h>
 #include "s390mach.h"
 
 static struct semaphore m_sem;
@@ -369,6 +370,8 @@ s390_do_machine_check(struct pt_regs *regs)
 
 	lockdep_off();
 
+	s390_idle_check();
+
 	mci = (struct mci *) &S390_lowcore.mcck_interruption_code;
 	mcck = &__get_cpu_var(cpu_mcck);
 	umode = user_mode(regs);

include/linux/kernel_stat.h

@@ -79,10 +79,13 @@ static inline unsigned int kstat_irqs(unsigned int irq)
 }
 
 extern unsigned long long task_delta_exec(struct task_struct *);
-extern void account_user_time(struct task_struct *, cputime_t);
-extern void account_user_time_scaled(struct task_struct *, cputime_t);
-extern void account_system_time(struct task_struct *, int, cputime_t);
-extern void account_system_time_scaled(struct task_struct *, cputime_t);
-extern void account_steal_time(struct task_struct *, cputime_t);
+extern void account_user_time(struct task_struct *, cputime_t, cputime_t);
+extern void account_system_time(struct task_struct *, int, cputime_t, cputime_t);
+extern void account_steal_time(cputime_t);
+extern void account_idle_time(cputime_t);
+
+extern void account_process_tick(struct task_struct *, int user);
+extern void account_steal_ticks(unsigned long ticks);
+extern void account_idle_ticks(unsigned long ticks);
 
 #endif /* _LINUX_KERNEL_STAT_H */

include/linux/sched.h

@@ -284,7 +284,6 @@ long io_schedule_timeout(long timeout);
 
 extern void cpu_init (void);
 extern void trap_init(void);
-extern void account_process_tick(struct task_struct *task, int user);
 extern void update_process_times(int user);
 extern void scheduler_tick(void);
 

kernel/sched.c

@@ -4150,13 +4150,17 @@ unsigned long long task_delta_exec(struct task_struct *p)
  * Account user cpu time to a process.
  * @p: the process that the cpu time gets accounted to
  * @cputime: the cpu time spent in user space since the last update
+ * @cputime_scaled: cputime scaled by cpu frequency
  */
-void account_user_time(struct task_struct *p, cputime_t cputime)
+void account_user_time(struct task_struct *p, cputime_t cputime,
+		       cputime_t cputime_scaled)
 {
 	struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
 	cputime64_t tmp;
 
+	/* Add user time to process. */
 	p->utime = cputime_add(p->utime, cputime);
+	p->utimescaled = cputime_add(p->utimescaled, cputime_scaled);
 	account_group_user_time(p, cputime);
 
 	/* Add user time to cpustat. */
@@ -4173,51 +4177,48 @@ void account_user_time(struct task_struct *p, cputime_t cputime)
  * Account guest cpu time to a process.
  * @p: the process that the cpu time gets accounted to
  * @cputime: the cpu time spent in virtual machine since the last update
+ * @cputime_scaled: cputime scaled by cpu frequency
  */
-static void account_guest_time(struct task_struct *p, cputime_t cputime)
+static void account_guest_time(struct task_struct *p, cputime_t cputime,
+			       cputime_t cputime_scaled)
 {
 	cputime64_t tmp;
 	struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
 
 	tmp = cputime_to_cputime64(cputime);
 
+	/* Add guest time to process. */
 	p->utime = cputime_add(p->utime, cputime);
+	p->utimescaled = cputime_add(p->utimescaled, cputime_scaled);
 	account_group_user_time(p, cputime);
 	p->gtime = cputime_add(p->gtime, cputime);
 
+	/* Add guest time to cpustat. */
 	cpustat->user = cputime64_add(cpustat->user, tmp);
 	cpustat->guest = cputime64_add(cpustat->guest, tmp);
 }
 
-/*
- * Account scaled user cpu time to a process.
- * @p: the process that the cpu time gets accounted to
- * @cputime: the cpu time spent in user space since the last update
- */
-void account_user_time_scaled(struct task_struct *p, cputime_t cputime)
-{
-	p->utimescaled = cputime_add(p->utimescaled, cputime);
-}
-
 /*
  * Account system cpu time to a process.
  * @p: the process that the cpu time gets accounted to
  * @hardirq_offset: the offset to subtract from hardirq_count()
  * @cputime: the cpu time spent in kernel space since the last update
+ * @cputime_scaled: cputime scaled by cpu frequency
  */
 void account_system_time(struct task_struct *p, int hardirq_offset,
-			 cputime_t cputime)
+			 cputime_t cputime, cputime_t cputime_scaled)
 {
 	struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
-	struct rq *rq = this_rq();
 	cputime64_t tmp;
 
 	if ((p->flags & PF_VCPU) && (irq_count() - hardirq_offset == 0)) {
-		account_guest_time(p, cputime);
+		account_guest_time(p, cputime, cputime_scaled);
 		return;
 	}
 
+	/* Add system time to process. */
 	p->stime = cputime_add(p->stime, cputime);
+	p->stimescaled = cputime_add(p->stimescaled, cputime_scaled);
 	account_group_system_time(p, cputime);
 
 	/* Add system time to cpustat. */
@@ -4226,48 +4227,84 @@ void account_system_time(struct task_struct *p, int hardirq_offset,
 		cpustat->irq = cputime64_add(cpustat->irq, tmp);
 	else if (softirq_count())
 		cpustat->softirq = cputime64_add(cpustat->softirq, tmp);
-	else if (p != rq->idle)
-		cpustat->system = cputime64_add(cpustat->system, tmp);
-	else if (atomic_read(&rq->nr_iowait) > 0)
-		cpustat->iowait = cputime64_add(cpustat->iowait, tmp);
 	else
-		cpustat->idle = cputime64_add(cpustat->idle, tmp);
+		cpustat->system = cputime64_add(cpustat->system, tmp);
+
 	/* Account for system time used */
 	acct_update_integrals(p);
 }
 
 /*
- * Account scaled system cpu time to a process.
- * @p: the process that the cpu time gets accounted to
- * @hardirq_offset: the offset to subtract from hardirq_count()
- * @cputime: the cpu time spent in kernel space since the last update
+ * Account for involuntary wait time.
+ * @steal: the cpu time spent in involuntary wait
  */
-void account_system_time_scaled(struct task_struct *p, cputime_t cputime)
+void account_steal_time(cputime_t cputime)
 {
-	p->stimescaled = cputime_add(p->stimescaled, cputime);
+	struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
+	cputime64_t cputime64 = cputime_to_cputime64(cputime);
+
+	cpustat->steal = cputime64_add(cpustat->steal, cputime64);
 }
 
 /*
- * Account for involuntary wait time.
- * @p: the process from which the cpu time has been stolen
- * @steal: the cpu time spent in involuntary wait
+ * Account for idle time.
+ * @cputime: the cpu time spent in idle wait
  */
-void account_steal_time(struct task_struct *p, cputime_t steal)
+void account_idle_time(cputime_t cputime)
 {
 	struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
-	cputime64_t tmp = cputime_to_cputime64(steal);
+	cputime64_t cputime64 = cputime_to_cputime64(cputime);
 	struct rq *rq = this_rq();
 
-	if (p == rq->idle) {
-		p->stime = cputime_add(p->stime, steal);
 	if (atomic_read(&rq->nr_iowait) > 0)
-			cpustat->iowait = cputime64_add(cpustat->iowait, tmp);
+		cpustat->iowait = cputime64_add(cpustat->iowait, cputime64);
 	else
-			cpustat->idle = cputime64_add(cpustat->idle, tmp);
-	} else
-		cpustat->steal = cputime64_add(cpustat->steal, tmp);
+		cpustat->idle = cputime64_add(cpustat->idle, cputime64);
 }
 
+#ifndef CONFIG_VIRT_CPU_ACCOUNTING
+
+/*
+ * Account a single tick of cpu time.
+ * @p: the process that the cpu time gets accounted to
+ * @user_tick: indicates if the tick is a user or a system tick
+ */
+void account_process_tick(struct task_struct *p, int user_tick)
+{
+	cputime_t one_jiffy = jiffies_to_cputime(1);
+	cputime_t one_jiffy_scaled = cputime_to_scaled(one_jiffy);
+	struct rq *rq = this_rq();
+
+	if (user_tick)
+		account_user_time(p, one_jiffy, one_jiffy_scaled);
+	else if (p != rq->idle)
+		account_system_time(p, HARDIRQ_OFFSET, one_jiffy,
+				    one_jiffy_scaled);
+	else
+		account_idle_time(one_jiffy);
+}
+
+/*
+ * Account multiple ticks of steal time.
+ * @p: the process from which the cpu time has been stolen
+ * @ticks: number of stolen ticks
+ */
+void account_steal_ticks(unsigned long ticks)
+{
+	account_steal_time(jiffies_to_cputime(ticks));
+}
+
+/*
+ * Account multiple ticks of idle time.
+ * @ticks: number of stolen ticks
+ */
+void account_idle_ticks(unsigned long ticks)
+{
+	account_idle_time(jiffies_to_cputime(ticks));
+}
+
+#endif
+
 /*
  * Use precise platform statistics if available:
  */

kernel/time/tick-sched.c

@@ -419,7 +419,9 @@ void tick_nohz_restart_sched_tick(void)
 {
 	int cpu = smp_processor_id();
 	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
+#ifndef CONFIG_VIRT_CPU_ACCOUNTING
 	unsigned long ticks;
+#endif
 	ktime_t now;
 
 	local_irq_disable();
@@ -441,6 +443,7 @@ void tick_nohz_restart_sched_tick(void)
 	tick_do_update_jiffies64(now);
 	cpumask_clear_cpu(cpu, nohz_cpu_mask);
 
+#ifndef CONFIG_VIRT_CPU_ACCOUNTING
 	/*
 	 * We stopped the tick in idle. Update process times would miss the
 	 * time we slept as update_process_times does only a 1 tick
@@ -450,12 +453,9 @@ void tick_nohz_restart_sched_tick(void)
 	/*
 	 * We might be one off. Do not randomly account a huge number of ticks!
 	 */
-	if (ticks && ticks < LONG_MAX) {
-		add_preempt_count(HARDIRQ_OFFSET);
-		account_system_time(current, HARDIRQ_OFFSET,
-				    jiffies_to_cputime(ticks));
-		sub_preempt_count(HARDIRQ_OFFSET);
-	}
+	if (ticks && ticks < LONG_MAX)
+		account_idle_ticks(ticks);
+#endif
 
 	touch_softlockup_watchdog();
 	/*

kernel/timer.c

@@ -1018,21 +1018,6 @@ unsigned long get_next_timer_interrupt(unsigned long now)
 }
 #endif
 
-#ifndef CONFIG_VIRT_CPU_ACCOUNTING
-void account_process_tick(struct task_struct *p, int user_tick)
-{
-	cputime_t one_jiffy = jiffies_to_cputime(1);
-
-	if (user_tick) {
-		account_user_time(p, one_jiffy);
-		account_user_time_scaled(p, cputime_to_scaled(one_jiffy));
-	} else {
-		account_system_time(p, HARDIRQ_OFFSET, one_jiffy);
-		account_system_time_scaled(p, cputime_to_scaled(one_jiffy));
-	}
-}
-#endif
-
 /*
  * Called from the timer interrupt handler to charge one tick to the current
  * process.  user_tick is 1 if the tick is user time, 0 for system.