This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[PING] [PATCH][BZ #16145] Reduce lock contention in __tz_convert()


Pinging this for review.  (Should I be CCing a particular reviewer?)

    - Kevin

On Thu, Feb 12, 2015 at 12:37:19AM +1100, Kevin Easton wrote:
> This patch is an "easy win" partial fix for BZ #16145, which notes
> the heavy contention on tzset_lock when multiple threads are converting
> times with localtime_r().
> 
> In __tz_convert(), the lock does not need to be held after
> __tzfile_compute() / __tz_compute() have been called, so we can move the
> unlock up.  At this point there is still significant work to be done in
> __offtime(), so we see some improvement (in my testing with 8 cores
> banging on localtime_r(), ~20% improvement in throughput).
> 
> 	[BZ #16145] (partial fix)
> 	* time/tzset.c (__tz_convert): Unlock tzset_lock earlier
> 	to reduce lock contention.
> 
> diff --git a/time/tzset.c b/time/tzset.c
> index 8bc7a2e..82324ca 100644
> --- a/time/tzset.c
> +++ b/time/tzset.c
> @@ -644,6 +644,8 @@ __tz_convert (const time_t *timer, int use_localtime, struct tm *tp)
>        leap_extra_secs = 0;
>      }
> 
> +  __libc_lock_unlock (tzset_lock);
> +
>    if (tp)
>      {
>        if (! use_localtime)
> @@ -659,8 +661,6 @@ __tz_convert (const time_t *timer, int use_localtime, struct tm *tp)
>         tp = NULL;
>      }
> 
> -  __libc_lock_unlock (tzset_lock);
> -
>    return tp;
>  }
> 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]