On Thu, Nov 23, 2017 at 08:04:39AM -0700, Jason Gunthorpe wrote:

Like Russell has said already, the kernel is the only place that could
actually have all the information needed. User space would have to use
some kind of trail and error approach to figure out what the offset
should be.

Jason,

Now that -rc1 is out, and I'm rebasing the branches, what I find is
the patch that was merged is not the patch that I last tested.  The
last patch I had from you was 22 Sep 2017.

The most obvious difference is in drivers/rtc/class.c:

-+      /* Drivers can revise this default after allocating the device. */
++      /* Drivers can revise this default after allocating the device. It
++       * should be the value of wallclock tv_nsec that the driver needs in
++       * order to synchronize the second tick over during set.
++       */

but there's also considerable differences in kernel/time/ntp.c, which
seems to miss out on some fixes for bugs I reported along the way:

 +      getnstimeofday64(&now);
 +
-+      adjust = now;
++      now = adjust;

What's going on?

I've got a rebase mess here to sort out, and it's a horrid mess.

I think what's in -rc1 is rather fscked.

-- 
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 8.8Mbps down 630kbps up
According to speedtest.net: 8.21Mbps down 510kbps up