[S390] idle time accounting vs. machine checks

A machine check can interrupt the i/o and external interrupt handler
anytime. If the machine check occurs while the interrupt handler is
waking up from idle vtime_start_cpu can get executed a second time
and the int_clock / async_enter_timer values in the lowcore get
clobbered. This can confuse the cpu time accounting.
To fix this problem two changes are needed. First the machine check
handler has to use its own copies of int_clock and async_enter_timer,
named mcck_clock and mcck_enter_timer. Second the nested execution
of vtime_start_cpu has to be prevented. This is done in s390_idle_check
by checking the wait bit in the program status word.

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

2 files changed, 56 insertions, 51 deletions

diff --git a/arch/s390/include/asm/cputime.h b/arch/s390/include/asm/cputime.h
index 258ba88..8b1a52a 100644
--- a/arch/s390/include/asm/cputime.h
+++ b/arch/s390/include/asm/cputime.h
@@ -188,15 +188,16 @@ struct s390_idle_data {
 
 DECLARE_PER_CPU(struct s390_idle_data, s390_idle);
 
-void vtime_start_cpu(void);
+void vtime_start_cpu(__u64 int_clock, __u64 enter_timer);
 cputime64_t s390_get_idle_time(int cpu);
 
 #define arch_idle_time(cpu) s390_get_idle_time(cpu)
 
-static inline void s390_idle_check(void)
+static inline void s390_idle_check(struct pt_regs *regs, __u64 int_clock,
+				   __u64 enter_timer)
 {
-	if ((&__get_cpu_var(s390_idle))->idle_enter != 0ULL)
-		vtime_start_cpu();
+	if (regs->psw.mask & PSW_MASK_WAIT)
+		vtime_start_cpu(int_clock, enter_timer);
 }
 
 static inline int s390_nohz_delay(int cpu)
diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
index f7e78c7..2c02d46 100644
--- a/arch/s390/include/asm/lowcore.h
+++ b/arch/s390/include/asm/lowcore.h
@@ -104,37 +104,39 @@ struct _lowcore {
 	/* CPU time accounting values */
 	__u64	sync_enter_timer;		/* 0x0250 */
 	__u64	async_enter_timer;		/* 0x0258 */
-	__u64	exit_timer;			/* 0x0260 */
-	__u64	user_timer;			/* 0x0268 */
-	__u64	system_timer;			/* 0x0270 */
-	__u64	steal_timer;			/* 0x0278 */
-	__u64	last_update_timer;		/* 0x0280 */
-	__u64	last_update_clock;		/* 0x0288 */
+	__u64	mcck_enter_timer;		/* 0x0260 */
+	__u64	exit_timer;			/* 0x0268 */
+	__u64	user_timer;			/* 0x0270 */
+	__u64	system_timer;			/* 0x0278 */
+	__u64	steal_timer;			/* 0x0280 */
+	__u64	last_update_timer;		/* 0x0288 */
+	__u64	last_update_clock;		/* 0x0290 */
 
 	/* Current process. */
-	__u32	current_task;			/* 0x0290 */
-	__u32	thread_info;			/* 0x0294 */
-	__u32	kernel_stack;			/* 0x0298 */
+	__u32	current_task;			/* 0x0298 */
+	__u32	thread_info;			/* 0x029c */
+	__u32	kernel_stack;			/* 0x02a0 */
 
 	/* Interrupt and panic stack. */
-	__u32	async_stack;			/* 0x029c */
-	__u32	panic_stack;			/* 0x02a0 */
+	__u32	async_stack;			/* 0x02a4 */
+	__u32	panic_stack;			/* 0x02a8 */
 
 	/* Address space pointer. */
-	__u32	kernel_asce;			/* 0x02a4 */
-	__u32	user_asce;			/* 0x02a8 */
-	__u32	user_exec_asce;			/* 0x02ac */
+	__u32	kernel_asce;			/* 0x02ac */
+	__u32	user_asce;			/* 0x02b0 */
+	__u32	user_exec_asce;			/* 0x02b4 */
 
 	/* SMP info area */
-	__u32	cpu_nr;				/* 0x02b0 */
-	__u32	softirq_pending;		/* 0x02b4 */
-	__u32	percpu_offset;			/* 0x02b8 */
-	__u32	ext_call_fast;			/* 0x02bc */
-	__u64	int_clock;			/* 0x02c0 */
-	__u64	clock_comparator;		/* 0x02c8 */
-	__u32	machine_flags;			/* 0x02d0 */
-	__u32	ftrace_func;			/* 0x02d4 */
-	__u8	pad_0x02d8[0x0300-0x02d8];	/* 0x02d8 */
+	__u32	cpu_nr;				/* 0x02b8 */
+	__u32	softirq_pending;		/* 0x02bc */
+	__u32	percpu_offset;			/* 0x02c0 */
+	__u32	ext_call_fast;			/* 0x02c4 */
+	__u64	int_clock;			/* 0x02c8 */
+	__u64	mcck_clock;			/* 0x02d0 */
+	__u64	clock_comparator;		/* 0x02d8 */
+	__u32	machine_flags;			/* 0x02e0 */
+	__u32	ftrace_func;			/* 0x02e4 */
+	__u8	pad_0x02e8[0x0300-0x02e8];	/* 0x02e8 */
 
 	/* Interrupt response block */
 	__u8	irb[64];			/* 0x0300 */
@@ -232,38 +234,40 @@ struct _lowcore {
 	/* CPU accounting and timing values. */
 	__u64	sync_enter_timer;		/* 0x02a0 */
 	__u64	async_enter_timer;		/* 0x02a8 */
-	__u64	exit_timer;			/* 0x02b0 */
-	__u64	user_timer;			/* 0x02b8 */
-	__u64	system_timer;			/* 0x02c0 */
-	__u64	steal_timer;			/* 0x02c8 */
-	__u64	last_update_timer;		/* 0x02d0 */
-	__u64	last_update_clock;		/* 0x02d8 */
+	__u64	mcck_enter_timer;		/* 0x02b0 */
+	__u64	exit_timer;			/* 0x02b8 */
+	__u64	user_timer;			/* 0x02c0 */
+	__u64	system_timer;			/* 0x02c8 */
+	__u64	steal_timer;			/* 0x02d0 */
+	__u64	last_update_timer;		/* 0x02d8 */
+	__u64	last_update_clock;		/* 0x02e0 */
 
 	/* Current process. */
-	__u64	current_task;			/* 0x02e0 */
-	__u64	thread_info;			/* 0x02e8 */
-	__u64	kernel_stack;			/* 0x02f0 */
+	__u64	current_task;			/* 0x02e8 */
+	__u64	thread_info;			/* 0x02f0 */
+	__u64	kernel_stack;			/* 0x02f8 */
 
 	/* Interrupt and panic stack. */
-	__u64	async_stack;			/* 0x02f8 */
-	__u64	panic_stack;			/* 0x0300 */
+	__u64	async_stack;			/* 0x0300 */
+	__u64	panic_stack;			/* 0x0308 */
 
 	/* Address space pointer. */
-	__u64	kernel_asce;			/* 0x0308 */
-	__u64	user_asce;			/* 0x0310 */
-	__u64	user_exec_asce;			/* 0x0318 */
+	__u64	kernel_asce;			/* 0x0310 */
+	__u64	user_asce;			/* 0x0318 */
+	__u64	user_exec_asce;			/* 0x0320 */
 
 	/* SMP info area */
-	__u32	cpu_nr;				/* 0x0320 */
-	__u32	softirq_pending;		/* 0x0324 */
-	__u64	percpu_offset;			/* 0x0328 */
-	__u64	ext_call_fast;			/* 0x0330 */
-	__u64	int_clock;			/* 0x0338 */
-	__u64	clock_comparator;		/* 0x0340 */
-	__u64	vdso_per_cpu_data;		/* 0x0348 */
-	__u64	machine_flags;			/* 0x0350 */
-	__u64	ftrace_func;			/* 0x0358 */
-	__u8	pad_0x0368[0x0380-0x0360];	/* 0x0360 */
+	__u32	cpu_nr;				/* 0x0328 */
+	__u32	softirq_pending;		/* 0x032c */
+	__u64	percpu_offset;			/* 0x0330 */
+	__u64	ext_call_fast;			/* 0x0338 */
+	__u64	int_clock;			/* 0x0340 */
+	__u64	mcck_clock;			/* 0x0348 */
+	__u64	clock_comparator;		/* 0x0350 */
+	__u64	vdso_per_cpu_data;		/* 0x0358 */
+	__u64	machine_flags;			/* 0x0360 */
+	__u64	ftrace_func;			/* 0x0368 */
+	__u8	pad_0x0370[0x0380-0x0370];	/* 0x0370 */
 
 	/* Interrupt response block. */
 	__u8	irb[64];			/* 0x0380 */