Re: Leap second insertion causes futex to repeatedly timeout

[Markus Trippelsdorf <markus@trippelsdorf.de>]

On 2012.07.01 at 02:57 +0200, Jan Engelhardt wrote:
> 
> This year's leap second insertion has had the strange effect on at least 
> Linux versions 3.4.4 (my end) and 3.5-rc4 (Simon's box, Cc) that certain 
> processes use up all CPU power, because of futexes repeatedly timing 
> out. This seems to only affect certain processes.
> 
> Simon observes - http://s85.org/owXfmLvt - that 
> Firefox/Thunderbird/Chrome/Java are affected.

On my ntp synchronized Linux 3.5 testbox I saw a similar issue.

The leap second was successfully inserted last night:
Jul  1 01:59:59 x4 kernel: Clock: inserting leap second 23:59:60 UTC

Because glibc-2.16 was released yesterday I've build it this morning.
During "make check" I saw several mutex related tests failures. For
example /nptl/tst-mutex5a.c failed with a "premature timeout difference"
of one second.

tst-mutex5a.c:
#define TYPE PTHREAD_MUTEX_ADAPTIVE_NP
#include "tst-mutex5.c"

from tst-mutex5.c:
// ...
  gettimeofday (&tv, NULL);
  TIMEVAL_TO_TIMESPEC (&tv, &ts);

  ts.tv_sec += 2;	/* Wait 2 seconds.  */

  err = pthread_mutex_timedlock (&m, &ts);
// ...
    {
      int clk_tck = sysconf (_SC_CLK_TCK);

      gettimeofday (&tv2, NULL);

      tv2.tv_sec -= tv.tv_sec;
      tv2.tv_usec -= tv.tv_usec;
      if (tv2.tv_usec < 0)
	{
	  tv2.tv_usec += 1000000;
	  tv2.tv_sec -= 1;
	}

      /* Be a bit tolerant, add one CLK_TCK.  */
      tv2.tv_usec += 1000000 / clk_tck;
      if (tv2.tv_usec >= 1000000)
	{
	  tv2.tv_usec -= 1000000;
	  ++tv2.tv_sec;
	}

      if (tv2.tv_sec < 2)
	{
	  printf ("premature timeout: %ld.%06ld difference\n",
		  tv2.tv_sec, tv2.tv_usec);
	  return 1;
	}
    }
// ...

-- 
Markus