/*
* TSC-warp measurement loop running on both CPUs:
*/
static __cpuinit void check_tsc_warp(void)
{
cycles_t start, now, prev, end;
int i;
rdtsc_barrier();
start = get_cycles();
rdtsc_barrier();
/*
* The measurement runs for 20 msecs:
*/
end = start + tsc_khz * 20ULL;
now = start;
for (i = 0; ; i++) {
/*
* We take the global lock, measure TSC, save the
* previous TSC that was measured (possibly on
* another CPU) and update the previous TSC timestamp.
*/
__raw_spin_lock(&sync_lock);
prev = last_tsc;
rdtsc_barrier();
now = get_cycles();
rdtsc_barrier();
last_tsc = now;
__raw_spin_unlock(&sync_lock);
/*
* Be nice every now and then (and also check whether
* measurement is done [we also insert a 10 million
* loops safety exit, so we dont lock up in case the
* TSC readout is totally broken]):
*/
if (unlikely(!(i & 7))) {
if (now > end || i > 10000000)
break;
cpu_relax();
touch_nmi_watchdog();
}
/*
* Outside the critical section we can now see whether
* we saw a time-warp of the TSC going backwards:
*/
if (unlikely(prev > now)) {
__raw_spin_lock(&sync_lock);
max_warp = max(max_warp, prev - now);
nr_warps++;
__raw_spin_unlock(&sync_lock);
}
}
WARN(!(now-start),
"Warning: zero tsc calibration delta: %Ld [max: %Ld]\n",
now-start, end-start);
}
static void lock_double_clock(struct sched_clock_data *data1,
struct sched_clock_data *data2)
{
if (data1 < data2) {
__raw_spin_lock(&data1->lock);
__raw_spin_lock(&data2->lock);
} else {
__raw_spin_lock(&data2->lock);
__raw_spin_lock(&data1->lock);
}
}
/*
* Test the trace buffer to see if all the elements
* are still sane.
*/
static int trace_test_buffer(struct trace_array *tr, unsigned long *count)
{
unsigned long flags, cnt = 0;
int cpu, ret = 0;
/* Don't allow flipping of max traces now */
local_irq_save(flags);
__raw_spin_lock(&ftrace_max_lock);
cnt = ring_buffer_entries(tr->buffer);
/*
* The trace_test_buffer_cpu runs a while loop to consume all data.
* If the calling tracer is broken, and is constantly filling
* the buffer, this will run forever, and hard lock the box.
* We disable the ring buffer while we do this test to prevent
* a hard lock up.
*/
tracing_off();
for_each_possible_cpu(cpu) {
ret = trace_test_buffer_cpu(tr, cpu);
if (ret)
break;
}
tracing_on();
__raw_spin_unlock(&ftrace_max_lock);
local_irq_restore(flags);
if (count)
*count = cnt;
return ret;
}
unsigned __kprobes long oops_begin(void)
{
int cpu;
unsigned long flags;
/* notify the hw-branch tracer so it may disable tracing and
add the last trace to the trace buffer -
the earlier this happens, the more useful the trace. */
trace_hw_branch_oops();
oops_enter();
/* racy, but better than risking deadlock. */
raw_local_irq_save(flags);
cpu = smp_processor_id();
if (!__raw_spin_trylock(&die_lock)) {
if (cpu == die_owner)
/* nested oops. should stop eventually */;
else
__raw_spin_lock(&die_lock);
}
die_nest_count++;
die_owner = cpu;
console_verbose();
bust_spinlocks(1);
return flags;
}
static void
probe_wakeup(struct rq *rq, struct task_struct *p, int success)
{
struct trace_array_cpu *data;
int cpu = smp_processor_id();
unsigned long flags;
long disabled;
int pc;
if (likely(!tracer_enabled))
return;
tracing_record_cmdline(p);
tracing_record_cmdline(current);
if ((wakeup_rt && !rt_task(p)) ||
p->prio >= wakeup_prio ||
p->prio >= current->prio)
return;
pc = preempt_count();
disabled = atomic_inc_return(&wakeup_trace->data[cpu]->disabled);
if (unlikely(disabled != 1))
goto out;
/* interrupts should be off from try_to_wake_up */
__raw_spin_lock(&wakeup_lock);
/* check for races. */
if (!tracer_enabled || p->prio >= wakeup_prio)
goto out_locked;
/* reset the trace */
__wakeup_reset(wakeup_trace);
wakeup_cpu = task_cpu(p);
wakeup_current_cpu = wakeup_cpu;
wakeup_prio = p->prio;
wakeup_task = p;
get_task_struct(wakeup_task);
local_save_flags(flags);
data = wakeup_trace->data[wakeup_cpu];
data->preempt_timestamp = ftrace_now(cpu);
tracing_sched_wakeup_trace(wakeup_trace, p, current, flags, pc);
/*
* We must be careful in using CALLER_ADDR2. But since wake_up
* is not called by an assembly function (where as schedule is)
* it should be safe to use it here.
*/
trace_function(wakeup_trace, CALLER_ADDR1, CALLER_ADDR2, flags, pc);
out_locked:
__raw_spin_unlock(&wakeup_lock);
out:
atomic_dec(&wakeup_trace->data[cpu]->disabled);
}
unsigned __kprobes long oops_begin(void)
{
int cpu;
unsigned long flags;
trace_hw_branch_oops();
oops_enter();
raw_local_irq_save(flags);
cpu = smp_processor_id();
if (!__raw_spin_trylock(&die_lock)) {
if (cpu == die_owner)
;
else
__raw_spin_lock(&die_lock);
}
die_nest_count++;
die_owner = cpu;
console_verbose();
bust_spinlocks(1);
return flags;
}
static inline void __lock_kernel(void)
{
preempt_disable();
if (unlikely(!__raw_spin_trylock(&kernel_flag))) {
/*
* If preemption was disabled even before this
* was called, there's nothing we can be polite
* about - just spin.
*/
if (preempt_count() > 1) {
__raw_spin_lock(&kernel_flag);
return;
}
/*
* Otherwise, let's wait for the kernel lock
* with preemption enabled..
*/
do {
preempt_enable();
while (spin_is_locked(&kernel_flag))
cpu_relax();
preempt_disable();
} while (!__raw_spin_trylock(&kernel_flag));
}
}
static int trace_test_buffer(struct trace_array *tr, unsigned long *count)
{
unsigned long flags, cnt = 0;
int cpu, ret = 0;
local_irq_save(flags);
__raw_spin_lock(&ftrace_max_lock);
cnt = ring_buffer_entries(tr->buffer);
tracing_off();
for_each_possible_cpu(cpu) {
ret = trace_test_buffer_cpu(tr, cpu);
if (ret)
break;
}
tracing_on();
__raw_spin_unlock(&ftrace_max_lock);
local_irq_restore(flags);
if (count)
*count = cnt;
return ret;
}
u64 sched_clock_cpu(int cpu)
{
u64 now, clock, this_clock, remote_clock;
struct sched_clock_data *scd;
if (sched_clock_stable)
return sched_clock();
scd = cpu_sdc(cpu);
/*
* Normally this is not called in NMI context - but if it is,
* trying to do any locking here is totally lethal.
*/
if (unlikely(in_nmi()))
return scd->clock;
if (unlikely(!sched_clock_running))
return 0ull;
WARN_ON_ONCE(!irqs_disabled());
now = sched_clock();
if (cpu != raw_smp_processor_id()) {
struct sched_clock_data *my_scd = this_scd();
lock_double_clock(scd, my_scd);
this_clock = __update_sched_clock(my_scd, now);
remote_clock = scd->clock;
/*
* Use the opportunity that we have both locks
* taken to couple the two clocks: we take the
* larger time as the latest time for both
* runqueues. (this creates monotonic movement)
*/
if (likely((s64)(remote_clock - this_clock) < 0)) {
clock = this_clock;
scd->clock = clock;
} else {
/*
* Should be rare, but possible:
*/
clock = remote_clock;
my_scd->clock = remote_clock;
}
__raw_spin_unlock(&my_scd->lock);
} else {
__raw_spin_lock(&scd->lock);
clock = __update_sched_clock(scd, now);
}
__raw_spin_unlock(&scd->lock);
return clock;
}
void __cpuinit rtas_take_timebase(void)
{
while (!timebase)
barrier();
__raw_spin_lock(&timebase_lock);
set_tb(timebase >> 32, timebase & 0xffffffff);
timebase = 0;
__raw_spin_unlock(&timebase_lock);
}
static void wakeup_reset(struct trace_array *tr)
{
unsigned long flags;
local_irq_save(flags);
__raw_spin_lock(&wakeup_lock);
__wakeup_reset(tr);
__raw_spin_unlock(&wakeup_lock);
local_irq_restore(flags);
}
static inline void rgb_lock (char *name, rgb_s *lock)
{
#if DEBUG
printf("%s lock %s\n", name, lock->name);
#endif
#if MUTEX
pthread_mutex_lock( &lock->lock);
#elif SPIN
__raw_spin_lock( &lock->lock);
#elif TSPIN
pthread_spin_lock( &lock->lock);
#endif
}
static void
probe_wakeup(struct rq *rq, struct task_struct *p, int success)
{
int cpu = smp_processor_id();
unsigned long flags;
long disabled;
int pc;
if (likely(!tracer_enabled))
return;
tracing_record_cmdline(p);
tracing_record_cmdline(current);
if (likely(!rt_task(p)) ||
p->prio >= wakeup_prio ||
p->prio >= current->prio)
return;
pc = preempt_count();
disabled = atomic_inc_return(&wakeup_trace->data[cpu]->disabled);
if (unlikely(disabled != 1))
goto out;
/* interrupts should be off from try_to_wake_up */
__raw_spin_lock(&wakeup_lock);
/* check for races. */
if (!tracer_enabled || p->prio >= wakeup_prio)
goto out_locked;
/* reset the trace */
__wakeup_reset(wakeup_trace);
wakeup_cpu = task_cpu(p);
wakeup_prio = p->prio;
wakeup_task = p;
get_task_struct(wakeup_task);
local_save_flags(flags);
wakeup_trace->data[wakeup_cpu]->preempt_timestamp = ftrace_now(cpu);
trace_function(wakeup_trace, wakeup_trace->data[wakeup_cpu],
CALLER_ADDR1, CALLER_ADDR2, flags, pc);
out_locked:
__raw_spin_unlock(&wakeup_lock);
out:
atomic_dec(&wakeup_trace->data[cpu]->disabled);
}
void __cpuinit rtas_give_timebase(void)
{
unsigned long flags;
local_irq_save(flags);
hard_irq_disable();
__raw_spin_lock(&timebase_lock);
rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL);
timebase = get_tb();
__raw_spin_unlock(&timebase_lock);
while (timebase)
barrier();
rtas_call(rtas_token("thaw-time-base"), 0, 1, NULL);
local_irq_restore(flags);
}
/*
* irqsoff uses its own tracer function to keep the overhead down:
*/
static void
wakeup_tracer_call(unsigned long ip, unsigned long parent_ip)
{
struct trace_array *tr = wakeup_trace;
struct trace_array_cpu *data;
unsigned long flags;
long disabled;
int resched;
int cpu;
int pc;
if (likely(!wakeup_task))
return;
pc = preempt_count();
resched = ftrace_preempt_disable();
cpu = raw_smp_processor_id();
data = tr->data[cpu];
disabled = atomic_inc_return(&data->disabled);
if (unlikely(disabled != 1))
goto out;
local_irq_save(flags);
__raw_spin_lock(&wakeup_lock);
if (unlikely(!wakeup_task))
goto unlock;
/*
* The task can't disappear because it needs to
* wake up first, and we have the wakeup_lock.
*/
if (task_cpu(wakeup_task) != cpu)
goto unlock;
trace_function(tr, ip, parent_ip, flags, pc);
unlock:
__raw_spin_unlock(&wakeup_lock);
local_irq_restore(flags);
out:
atomic_dec(&data->disabled);
ftrace_preempt_enable(resched);
}
/*
* We just idled delta nanoseconds (called with irqs disabled):
*/
void sched_clock_idle_wakeup_event(u64 delta_ns)
{
struct sched_clock_data *scd = this_scd();
u64 now = sched_clock();
/*
* Override the previous timestamp and ignore all
* sched_clock() deltas that occured while we idled,
* and use the PM-provided delta_ns to advance the
* rq clock:
*/
__raw_spin_lock(&scd->lock);
scd->prev_raw = now;
scd->clock += delta_ns;
__raw_spin_unlock(&scd->lock);
touch_softlockup_watchdog();
}
void sched_clock_tick(void)
{
struct sched_clock_data *scd = this_scd();
u64 now, now_gtod;
if (unlikely(!sched_clock_running))
return;
WARN_ON_ONCE(!irqs_disabled());
now_gtod = ktime_to_ns(ktime_get());
now = sched_clock();
__raw_spin_lock(&scd->lock);
scd->tick_raw = now;
scd->tick_gtod = now_gtod;
__update_sched_clock(scd, now);
__raw_spin_unlock(&scd->lock);
}
unsigned __kprobes long oops_begin(void)
{
unsigned long flags;
oops_enter();
if (die_owner != raw_smp_processor_id()) {
console_verbose();
raw_local_irq_save(flags);
__raw_spin_lock(&die_lock);
die_owner = smp_processor_id();
die_nest_count = 0;
bust_spinlocks(1);
} else {
raw_local_irq_save(flags);
}
die_nest_count++;
return flags;
}
u64 sched_clock_cpu(int cpu)
{
struct sched_clock_data *scd = cpu_sdc(cpu);
u64 now, clock;
if (unlikely(!sched_clock_running))
return 0ull;
WARN_ON_ONCE(!irqs_disabled());
now = sched_clock();
if (cpu != raw_smp_processor_id()) {
/*
* in order to update a remote cpu's clock based on our
* unstable raw time rebase it against:
* tick_raw (offset between raw counters)
* tick_gotd (tick offset between cpus)
*/
struct sched_clock_data *my_scd = this_scd();
lock_double_clock(scd, my_scd);
now -= my_scd->tick_raw;
now += scd->tick_raw;
now += my_scd->tick_gtod;
now -= scd->tick_gtod;
__raw_spin_unlock(&my_scd->lock);
__update_sched_clock(scd, now, &clock);
__raw_spin_unlock(&scd->lock);
} else {
__raw_spin_lock(&scd->lock);
__update_sched_clock(scd, now, NULL);
clock = scd->clock;
__raw_spin_unlock(&scd->lock);
}
return clock;
}
void sched_clock_tick(void)
{
struct sched_clock_data *scd = this_scd();
unsigned long now_jiffies = jiffies;
s64 mult, delta_gtod, delta_raw;
u64 now, now_gtod;
if (unlikely(!sched_clock_running))
return;
WARN_ON_ONCE(!irqs_disabled());
now_gtod = ktime_to_ns(ktime_get());
now = sched_clock();
__raw_spin_lock(&scd->lock);
__update_sched_clock(scd, now, NULL);
/*
* update tick_gtod after __update_sched_clock() because that will
* already observe 1 new jiffy; adding a new tick_gtod to that would
* increase the clock 2 jiffies.
*/
delta_gtod = now_gtod - scd->tick_gtod;
delta_raw = now - scd->tick_raw;
if ((long)delta_raw > 0) {
mult = delta_gtod << MULTI_SHIFT;
do_div(mult, delta_raw);
scd->multi = mult;
if (scd->multi > MAX_MULTI)
scd->multi = MAX_MULTI;
else if (scd->multi < MIN_MULTI)
scd->multi = MIN_MULTI;
} else
scd->multi = 1 << MULTI_SHIFT;
scd->tick_raw = now;
scd->tick_gtod = now_gtod;
scd->tick_jiffies = now_jiffies;
__raw_spin_unlock(&scd->lock);
}
unsigned __kprobes long oops_begin(void)
{
int cpu;
unsigned long flags;
oops_enter();
/* racy, but better than risking deadlock. */
raw_local_irq_save(flags);
cpu = smp_processor_id();
if (!__raw_spin_trylock(&die_lock)) {
if (cpu == die_owner)
/* nested oops. should stop eventually */;
else
__raw_spin_lock(&die_lock);
}
die_nest_count++;
die_owner = cpu;
console_verbose();
bust_spinlocks(1);
return flags;
}
void sched_clock_tick(void)
{
struct sched_clock_data *scd = this_scd();
u64 now, now_gtod;
if (unlikely(!sched_clock_running))
return;
WARN_ON_ONCE(!irqs_disabled());
now = sched_clock();
now_gtod = ktime_to_ns(ktime_get());
__raw_spin_lock(&scd->lock);
__update_sched_clock(scd, now);
/*
* update tick_gtod after __update_sched_clock() because that will
* already observe 1 new jiffy; adding a new tick_gtod to that would
* increase the clock 2 jiffies.
*/
scd->tick_raw = now;
scd->tick_gtod = now_gtod;
__raw_spin_unlock(&scd->lock);
}
u64 notrace trace_clock_global(void)
{
unsigned long flags;
int this_cpu;
u64 now;
raw_local_irq_save(flags);
this_cpu = raw_smp_processor_id();
now = cpu_clock(this_cpu);
/*
* If in an NMI context then dont risk lockups and return the
* cpu_clock() time:
*/
if (unlikely(in_nmi()))
goto out;
__raw_spin_lock(&trace_clock_struct.lock);
/*
* TODO: if this happens often then maybe we should reset
* my_scd->clock to prev_time+1, to make sure
* we start ticking with the local clock from now on?
*/
if ((s64)(now - trace_clock_struct.prev_time) < 0)
now = trace_clock_struct.prev_time + 1;
trace_clock_struct.prev_time = now;
__raw_spin_unlock(&trace_clock_struct.lock);
out:
raw_local_irq_restore(flags);
return now;
}
void _remote_spin_lock(_remote_spinlock_t *lock)
{
__raw_spin_lock((raw_spinlock_t *) (*lock));
}
/*
* Non-preemption case - just get the spinlock
*/
static inline void __lock_kernel(void)
{
__raw_spin_lock(&kernel_flag);
}
void __lockfunc _raw_spin_lock(raw_spinlock_t *lock)
{
__raw_spin_lock(lock);
}
static inline void
default_spin_lock_flags(raw_spinlock_t *lock, unsigned long flags)
{
__raw_spin_lock(lock);
}
static void notrace
probe_wakeup_sched_switch(struct rq *rq, struct task_struct *prev,
struct task_struct *next)
{
struct trace_array_cpu *data;
cycle_t T0, T1, delta;
unsigned long flags;
long disabled;
int cpu;
int pc;
tracing_record_cmdline(prev);
if (unlikely(!tracer_enabled))
return;
/*
* When we start a new trace, we set wakeup_task to NULL
* and then set tracer_enabled = 1. We want to make sure
* that another CPU does not see the tracer_enabled = 1
* and the wakeup_task with an older task, that might
* actually be the same as next.
*/
smp_rmb();
if (next != wakeup_task)
return;
pc = preempt_count();
/* disable local data, not wakeup_cpu data */
cpu = raw_smp_processor_id();
disabled = atomic_inc_return(&wakeup_trace->data[cpu]->disabled);
if (likely(disabled != 1))
goto out;
local_irq_save(flags);
__raw_spin_lock(&wakeup_lock);
/* We could race with grabbing wakeup_lock */
if (unlikely(!tracer_enabled || next != wakeup_task))
goto out_unlock;
/* The task we are waiting for is waking up */
data = wakeup_trace->data[wakeup_cpu];
trace_function(wakeup_trace, CALLER_ADDR0, CALLER_ADDR1, flags, pc);
tracing_sched_switch_trace(wakeup_trace, prev, next, flags, pc);
T0 = data->preempt_timestamp;
T1 = ftrace_now(cpu);
delta = T1-T0;
if (!report_latency(delta))
goto out_unlock;
if (likely(!is_tracing_stopped())) {
tracing_max_latency = delta;
update_max_tr(wakeup_trace, wakeup_task, wakeup_cpu);
}
out_unlock:
__wakeup_reset(wakeup_trace);
__raw_spin_unlock(&wakeup_lock);
local_irq_restore(flags);
out:
atomic_dec(&wakeup_trace->data[cpu]->disabled);
}
