[ntpwg] ntpwg Updated NTPv4 Protocol Specification/timestamplocation

[Kurt Roeckx]

On Sun, Feb 03, 2008 at 08:32:55AM -0800, TS Glassey wrote:
> I think that the problem is that unlike 1588, the NTP process depends on 
> some stack receiving and framing the data prior to its actually being 
> reliable data. Since there are no precision timer/receipt processes 
> available for logging in TCP, UDP,  or the NTP process to address this it 
> becomes a real issue in that the data could sit in the stack for any amount 
> of time...

You can ask the kernel when it received a packet.  This is currently
mostly done in the kernel itself without any help from hardware.  Ntpd
is using this for a long time now.  Complete software based PTP
does the same thing and has exactly the same problem as ntpd.

What could be changed to get better results is:
- Have more hardware support so that you can capture the time
  in the PHY for both send and receive of a packet;
- Have the kernel export that to user space;
- Change ntpd so that is can send a "Follow Up" message that
  contains the real time of when the previous packet was send
  as captured by the PHY.

But those are all just current implementation details.  This is not
a limitation of the NTP protocol.


Kurt