Date: Mon, 24 May 2010 15:14:26 +0100 From: Chris Vine <chris@cvine.freeserve.co.uk> To: Daniel Eischen <deischen@freebsd.org> Cc: freebsd-threads@freebsd.org Subject: Re: Monotonic clocks Message-ID: <20100524151426.7ec75b1a@boulder.homenet> In-Reply-To: <Pine.GSO.4.64.1005240944500.5996@sea.ntplx.net> References: <20100524140513.6475cb2c@boulder.homenet> <Pine.GSO.4.64.1005240944500.5996@sea.ntplx.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 24 May 2010 09:46:39 -0400 (EDT)
Daniel Eischen <deischen@freebsd.org> wrote:
> On Mon, 24 May 2010, Chris Vine wrote:
>
> > Hi,
> >
> > I am the upstream maintainer of efax-gtk and I am told recent
> > versions of the program will not compile on the FreeBSD port of
> > debian, because FreeBSD does not appear to provide
> > pthread_condattr_setclock() to set a monotonic clock on condition
> > variables.
> >
> > The program configuration script calls up sysconf() to determine
> > whether the POSIX advanced realtime option (_SC_MONOTONIC_CLOCK) is
> > provided, and FreeBSD's sysconf() indicates that it is. Is anything
> > else needed to enable monotonic clocks on BSD, or is sysconf()
> > incorrectly advertising the implementation of monotonic clocks in
> > BSD's libc?
>
> What version of FreeBSD are you using? The code looks like it
> tries to use CLOCK_MONOTONIC if specified. From
> src/lib/libthr/thread/thr_init.c:
>
> int
> _pthread_condattr_setclock(pthread_condattr_t *attr, clockid_t
> clock_id) {
> if (attr == NULL || *attr == NULL)
> return (EINVAL);
> if (clock_id != CLOCK_REALTIME &&
> clock_id != CLOCK_VIRTUAL &&
> clock_id != CLOCK_PROF &&
> clock_id != CLOCK_MONOTONIC) {
> return (EINVAL);
> }
> (*attr)->c_clockid = clock_id;
> return (0);
> }
>
> Are you getting an error from pthread_condattr_setclock() or is
> it just not functioning correctly?
The function is entirely missing from the pthread.h header. I don't use
FreeBSD but the problem arises on the version of FreeBSD used by
debian. I don't know what that is, I am afraid.
I am hopeful that checking for_POSIX_CLOCK_SELECTION and
_SC_CLOCK_SELECTION, as well as _POSIX_MONOTONIC_CLOCK and
_SC_MONOTONIC_CLOCK will address the problem. (This means that the
system clock will be used instead.)
Chris
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100524151426.7ec75b1a>