Re: [PATCHv6 0/7] system time changes notification

[Alexander Shishkin <virtuoso@slind.org>]

On Wed, Nov 17, 2010 at 10:34:00PM +0100, Kay Sievers wrote:
> On Wed, Nov 17, 2010 at 22:29, Alexander Shishkin <virtuoso@slind.org> wrote:
> > On Wed, Nov 17, 2010 at 12:42:52PM -0800, Davide Libenzi wrote:
> >> On Wed, 17 Nov 2010, Alexander Shishkin wrote:
> >>
> >> > > But what you folks really want for this stuff is an extension to
> >> > > timerfd as you want to be able to poll, right?
> >> > >
> >> > > So what about the following:
> >> > >
> >> > > Add a new flag TDF_NOTIFY_CLOCK_WAS_SET to the timerfd flags. Now this
> >> > > flag adds the timer to a separate list, which gets woken up when the
> >> > > clock is set.
> >> > >
> >> > > No new syscall, just a few lines of code in fs/timerfd.c and
> >> > > clock_was_set().
> >> > >
> >> > > Thoughts ?
> >> >
> >> > Something like this (sans ugliness)?
> >>
> >> Oh, gosh, please.  This is interface-multiplexing-a-palooza.
> >
> > Thomas made a suggestion, I came up with how it might look like so that
> > pros and cons are clearer to everyone (or at least me) and can be discussed
> > on technical grounds. Code talks, sort of. I'm not convinced that a timer
> > that returns to userspace when the clock changes is such a bad idea, could
> > you please elaborate?
> 
> I like it.
> 
> It's all bout timers, and timerfd is fine to use, I think. It has
> nothing to do with "system-events", we ask for the timer to serve us,
> but if the *time* changes underneath, we need to know to re-calculate.
> 
> I think it's simple and fits very well in the current interface.

What would you say timerfd_gettime() should return for such a timer? Should
it be current clock's time in it_value?

Regards,
--
Alex