process time < thread time?

If someone who understands our thread/process time implementation can look into this, I'd appreciate it.

Attached below is a watered-down version of rt/tst-cpuclock2.c from GLIBC. Just build it with "gcc -o test test.c -lpthread -lrt" or similar.

Run it several times, and you will see cases where the main thread will measure a process clock difference before and after the nanosleep which is smaller than the cpu-burner thread's individual thread clock difference. This doesn't make any sense since the cpu-burner thread is part of the top-level process's thread group.

I've reproduced this on both x86-64 and sparc64 (using both 32-bit and 64-bit binaries).

For example:

[davem@boricha build-x86_64-linux]$ ./test

process: before(0.001221967) after(0.498624371) diff(497402404)
thread:  before(0.000081692) after(0.498316431) diff(498234739)
self:    before(0.001223521) after(0.001240219) diff(16698)

[davem@boricha build-x86_64-linux]$

The diff of 'process' should always be >= the diff of 'thread'.

I make sure to wrap the 'thread' clock measurements the most tightly around the nanosleep() call, and that the 'process' clock measurements are the outer-most ones.

I suspect this might be some kind of artifact of how the partial runqueue ->clock and ->clock_task updates work? Maybe some weird interaction with ->skip_clock_update?

Or is this some known issue?

Thanks!

--------------------
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <fcntl.h>
#include <string.h>
#include <errno.h>
#include <pthread.h>

static pthread_barrier_t barrier;

static void *chew_cpu(void *arg)

{
	pthread_barrier_wait(&barrier);
	while (1)
		__asm__ __volatile__("" : : : "memory");
	return NULL;
}

int main(void)

{
	clockid_t process_clock, my_thread_clock, th_clock;
	struct timespec process_before, process_after;
	struct timespec me_before, me_after;
	struct timespec th_before, th_after;
	struct timespec sleeptime;
	unsigned long diff;
	pthread_t th;
	int err;

	err = clock_getcpuclockid(0, &process_clock);
	if (err)
		return 1;

	err = pthread_getcpuclockid(pthread_self(), &my_thread_clock);
	if (err)
		return 1;

	pthread_barrier_init(&barrier, NULL, 2);
	err = pthread_create(&th, NULL, chew_cpu, NULL);
	if (err)
		return 1;

	err = pthread_getcpuclockid(th, &th_clock);
	if (err)
		return 1;

	pthread_barrier_wait(&barrier);

	err = clock_gettime(process_clock, &process_before);
	if (err)
		return 1;

	err = clock_gettime(my_thread_clock, &me_before);
	if (err)
		return 1;

	err = clock_gettime(th_clock, &th_before);
	if (err)
		return 1;

	sleeptime.tv_sec = 0;
	sleeptime.tv_nsec = 500000000;
	nanosleep(&sleeptime, NULL);

	err = clock_gettime(th_clock, &th_after);
	if (err)
		return 1;

	err = clock_gettime(my_thread_clock, &me_after);
	if (err)
		return 1;

	err = clock_gettime(process_clock, &process_after);
	if (err)
		return 1;

	diff = process_after.tv_nsec - process_before.tv_nsec;
	printf("process: before(%lu.%.9lu) after(%lu.%.9lu) diff(%lu)\n",
	       process_before.tv_sec, process_before.tv_nsec,
	       process_after.tv_sec, process_after.tv_nsec, diff);
	diff = th_after.tv_nsec - th_before.tv_nsec;
	printf("thread:  before(%lu.%.9lu) after(%lu.%.9lu) diff(%lu)\n",
	       th_before.tv_sec, th_before.tv_nsec,
	       th_after.tv_sec, th_after.tv_nsec, diff);
	diff = me_after.tv_nsec - me_before.tv_nsec;
	printf("self:    before(%lu.%.9lu) after(%lu.%.9lu) diff(%lu)\n",
	       me_before.tv_sec, me_before.tv_nsec,
	       me_after.tv_sec, me_after.tv_nsec, diff);

	return 0;
}

Dave,

That's an SMP artifact. If you run "taskset 01 ./test" the result is always correct.

The reason why this shows deviations on SMP is how the thread times are accumulated in thread_group_cputime(). We sum t->se.sum_exec_runtime of all threads. So if the hog thread is currently running on the other core (which is likely) then the runtime field of that thread is not up to date.

The untested patch below should cure this.

Thanks,

tglx

index 58f405b..42378cb 100644
@@ -250,7 +250,7 @@ void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times)
 	do {
 		times->utime = cputime_add(times->utime, t->utime);
 		times->stime = cputime_add(times->stime, t->stime);
-		times->sum_exec_runtime += t->se.sum_exec_runtime;
+		times->sum_exec_runtime += task_sched_runtime(t);
 	} while_each_thread(tsk, t);
 out:
 	rcu_read_unlock();

I think we just deadlocked:

cpu_clock_sample_group()
CPUCLOCK_SCHED: thread_group_sched_runtime()

    task_rq_lock()        <-.
    thread_group_cputime()  |
      task_sched_runtime()  |
       task_rq_lock() ------'

Also, having to take locks here is sad, but yeah if we want to cure this there's not much we can do about that.

That's why I said untested :)

Yeah :(

the below seems to build boot and give consistent output..

---
Subject: posix-cpu-timers: Cure SMP wobbles
From: Peter Zijlstra <[email protected]>
Date: Thu Sep 01 12:42:04 CEST 2011

David reported:

Attached below is a watered-down version of rt/tst-cpuclock2.c from GLIBC. Just build it with "gcc -o test test.c -lpthread -lrt" or similar.

Run it several times, and you will see cases where the main thread will measure a process clock difference before and after the nanosleep which is smaller than the cpu-burner thread's individual thread clock difference. This doesn't make any sense since the cpu-burner thread is part of the top-level process's thread group.

I've reproduced this on both x86-64 and sparc64 (using both 32-bit and 64-bit binaries).

For example:

[davem@boricha build-x86_64-linux]$ ./test
process: before(0.001221967) after(0.498624371) diff(497402404)
thread: before(0.000081692) after(0.498316431) diff(498234739)
self: before(0.001223521) after(0.001240219) diff(16698)
[davem@boricha build-x86_64-linux]$

The diff of 'process' should always be >= the diff of 'thread'.

I make sure to wrap the 'thread' clock measurements the most tightly around the nanosleep() call, and that the 'process' clock measurements are the outer-most ones.

  ---
  #include <unistd.h>
  #include <stdio.h>
  #include <stdlib.h>
  #include <time.h>
  #include <fcntl.h>
  #include <string.h>
  #include <errno.h>
  #include <pthread.h>

static pthread_barrier_t barrier;

static void *chew_cpu(void *arg)

  {
	  pthread_barrier_wait(&barrier);
	  while (1)
		  __asm__ __volatile__("" : : : "memory");
	  return NULL;

}

int main(void)

  {
	  clockid_t process_clock, my_thread_clock, th_clock;
	  struct timespec process_before, process_after;
	  struct timespec me_before, me_after;
	  struct timespec th_before, th_after;
	  struct timespec sleeptime;
	  unsigned long diff;
	  pthread_t th;
	  int err;

	  err = clock_getcpuclockid(0, &process_clock);
	  if (err)
		  return 1;

	  err = pthread_getcpuclockid(pthread_self(), &my_thread_clock);
	  if (err)
		  return 1;

	  pthread_barrier_init(&barrier, NULL, 2);
	  err = pthread_create(&th, NULL, chew_cpu, NULL);
	  if (err)
		  return 1;

	  err = pthread_getcpuclockid(th, &th_clock);
	  if (err)
		  return 1;

	  pthread_barrier_wait(&barrier);

	  err = clock_gettime(process_clock, &process_before);
	  if (err)
		  return 1;

	  err = clock_gettime(my_thread_clock, &me_before);
	  if (err)
		  return 1;

	  err = clock_gettime(th_clock, &th_before);
	  if (err)
		  return 1;

	  sleeptime.tv_sec = 0;
	  sleeptime.tv_nsec = 500000000;
	  nanosleep(&sleeptime, NULL);

	  err = clock_gettime(th_clock, &th_after);
	  if (err)
		  return 1;

	  err = clock_gettime(my_thread_clock, &me_after);
	  if (err)
		  return 1;

	  err = clock_gettime(process_clock, &process_after);
	  if (err)
		  return 1;

	  diff = process_after.tv_nsec - process_before.tv_nsec;
	  printf("process: before(%lu.%.9lu) after(%lu.%.9lu) diff(%lu)\n",
		 process_before.tv_sec, process_before.tv_nsec,
		 process_after.tv_sec, process_after.tv_nsec, diff);
	  diff = th_after.tv_nsec - th_before.tv_nsec;
	  printf("thread:  before(%lu.%.9lu) after(%lu.%.9lu) diff(%lu)\n",
		 th_before.tv_sec, th_before.tv_nsec,
		 th_after.tv_sec, th_after.tv_nsec, diff);
	  diff = me_after.tv_nsec - me_before.tv_nsec;
	  printf("self:    before(%lu.%.9lu) after(%lu.%.9lu) diff(%lu)\n",
		 me_before.tv_sec, me_before.tv_nsec,
		 me_after.tv_sec, me_after.tv_nsec, diff);

	  return 0;

}

This is due to us using p->se.sum_exec_runtime in thread_group_cputime() where we iterate the thread group and sum all data. This does not take time since the last schedule operation (tick or otherwise) into account. We can cure this by using task_sched_runtime() at the cost of having to take locks.

This also means we can (and must) do away with thread_group_sched_runtime() since the modified thread_group_cputime() is now more accurate and would deadlock when called from thread_group_sched_runtime().

---
 include/linux/sched.h     |    1 -
 kernel/posix-cpu-timers.c |    5 +++--
 kernel/sched.c            |   24 ------------------------
 3 files changed, 3 insertions(+), 27 deletions(-)

Index: linux-2.6/include/linux/sched.h
===================================================================
@@ -1956,7 +1956,6 @@ static inline void disable_sched_clock_i
 
 extern unsigned long long
 task_sched_runtime(struct task_struct *task);
-extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
 
 /* sched_exec is called by processes performing an exec */
 #ifdef CONFIG_SMP

Index: linux-2.6/kernel/posix-cpu-timers.c
===================================================================
@@ -250,7 +250,7 @@ void thread_group_cputime(struct task_st
 	do {
 		times->utime = cputime_add(times->utime, t->utime);
 		times->stime = cputime_add(times->stime, t->stime);
-		times->sum_exec_runtime += t->se.sum_exec_runtime;
+		times->sum_exec_runtime += task_sched_runtime(t);
 	} while_each_thread(tsk, t);
 out:
 	rcu_read_unlock();

@@ -312,7 +312,8 @@ static int cpu_clock_sample_group(const 
 		cpu->cpu = cputime.utime;
 		break;
 	case CPUCLOCK_SCHED:
-		cpu->sched = thread_group_sched_runtime(p);
+		thread_group_cputime(p, &cputime);
+		cpu->sched = cputime.sum_exec_runtime;
 		break;
 	}
 	return 0;

Index: linux-2.6/kernel/sched.c
===================================================================
@@ -3866,30 +3866,6 @@ unsigned long long task_sched_runtime(st
 }
 
 /*
- * Return sum_exec_runtime for the thread group.
- * In case the task is currently running, return the sum plus current's
- * pending runtime that have not been accounted yet.
- *
- * Note that the thread group might have other running tasks as well,
- * so the return value not includes other pending runtime that other
- * running tasks might have.
- */
-unsigned long long thread_group_sched_runtime(struct task_struct *p)
-{
-	struct task_cputime totals;
-	unsigned long flags;
-	struct rq *rq;
-	u64 ns;
-
-	rq = task_rq_lock(p, &flags);
-	thread_group_cputime(p, &totals);
-	ns = totals.sum_exec_runtime + do_task_delta_exec(p, rq);
-	task_rq_unlock(rq, p, &flags);
-
-	return ns;
-}
-
-/*
  * 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

From: Peter Zijlstra <[email protected]>
Date: Thu, 01 Sep 2011 12:54:19 +0200

Works here too.