On 2 May 2014 03:52, Alexander Samilovskih <alexsamilovskih@xxxxxxxxx> wrote:
> There are several places where gettimeofday should be replaced by
> clock_gettime(CLOCK_MONOTONIC_RAW). According to manual pages
> gettimeofday affected by discontinues jumps made by system
> administrator, can be slewed by ntpd or kinda stuff. So when we
> subtract two different timestamps from gettimeofday we can get
> confusing results
>
> sys-utils/eject.c
> libmount/lock.c
> maybe some other places...
Hi Alexander,
The diff you sent does not seem to work.
fatal: corrupt patch at line 123
The format you get from
$ git format-patch HEAD~1
would be appreciated.
> diff --git a/include/timeutils.h b/include/timeutils.h
> index bcae613..7cf972d 100644
> --- a/include/timeutils.h
> +++ b/include/timeutils.h
> @@ -51,5 +51,6 @@ typedef uint64_t nsec_t;
> #define FORMAT_TIMESPAN_MAX 64
>
> int parse_timestamp(const char *t, usec_t *usec);
> +int gettime_monotonic(struct timeval *tv);
>
> #endif /* UTIL_LINUX_TIME_UTIL_H */
> diff --git a/lib/timeutils.c b/lib/timeutils.c
> index 7fe6218..06ae597 100644
> --- a/lib/timeutils.c
> +++ b/lib/timeutils.c
> @@ -336,3 +336,24 @@ int parse_timestamp(const char *t, usec_t *usec)
>
> return 0;
> }
> +
> +int gettime_monotonic(struct timeval *tv)
> +{
> +#ifdef CLOCK_MONOTONIC
> + //Can slew only by ntp and adjtime
Could you use /* comment */ style rather than // from C++ world.
> + int ret;
> + struct timespec ts;
> +#ifdef CLOCK_MONOTONIC_RAW
> + //Linux specific, cant slew
> + if (!(ret = clock_gettime(CLOCK_MONOTONIC_RAW, &ts))) {
> +#else
> + if (!(ret = clock_gettime(CLOCK_MONOTONIC, &ts))) {
> +#endif
> + tv->tv_sec = ts.tv_sec;
> + tv->tv_usec = ts.tv_nsec / 1000;
> + }
> + return ret;
> +#else
A comment could ease understanding what is going on, such as
#else /* CLOCK_MONOTONIC is not available */
> + return gettimeofday(tv, NULL);
> +#endif
> +}
Th rest looks pretty ok.
--
Sami Kerola
http://www.iki.fi/kerolasa/
--
To unsubscribe from this list: send the line "unsubscribe util-linux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
