separate PPS_out with time_valid.
    Disable transmitting packets when link is not up.

    command "ptp [mode] [port,df=0]" or "mode [mode] [port,df=0]"
  will set the ptp mode of port.
    Currently, if one port act as slave, its mode should be set
  after other ports.

  Now it works as below:
    1. default udp tx dst mac is the gateway of node's ip.
    2. It will send arp request after getting the udp tx dst ip.
    3. It stops sending arp request after getting the udp tx dst mac addr.

        tcpip tcp x , set the tcp server listen port
        tcpip udp txip x.x.x.x , set the udp tx destination ip addr.
        tcpip udp txport x , set the udp tx destination port, default
2000
        tcpip udp rxport x , set the udp rx destiniation port. default
(>256).
        tcpip udp txgw x.x.x.x , set the udp gateway, default x.x.x.1
        tcpip udp txsn x.x.x.x , set the subnet, default 255.255.255.0

through UDP packets.
   See related document for further help.

Please not that LLDP_TX_TICK_INTERVAL is not about ticks, but about
number of task executions. This should be changed to time.
Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Change indentation of switch statement.
Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

normally the ip is given after lldp has been initialized or is
modified maually

The optional TLV of LLDP has been added, leaving
two of them still as *todo* since there is not hw
support yet for getting info from the wr-core outer world.

LLDP a vendor-neutral link layer protocol used by network devices
for advertising their identity, capabilities, and neighbors.
Implements a transmit-only nodes, the node only transmit periodically
and doesn't process any LLDPUD.

This is a logic &, not an aritmetic one. So "&&" not "&".  The change
has no effect, because the two operands are verified to always be 0 or
1, but what we had was conceptually wrong and forced me to verify.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Under frame loss, most track-lost are just because of ptp timeouts.
But since the underlying pll is unaffected, and the RTT is almost constant,
we can tell in the logs whether the lost/resync was actually a time
jump or not.

This is what happens with moderate frame loss (fhe first track-lost is
real, due to a ptp restart, the following ones are not and max delta
between calculated setpoint and current setpoint is reported):

   Jan  1 00:00:02 192.168.128.84 (00: 26:7b:00:04:27) Node up since 1 seconds
   Jun 12 14:16:06 192.168.128.84 Tracking after 19.141 s
   Jun 12 14:17:14 192.168.128.84 Lost track
   Jun 12 14:17:20 192.168.128.84 2-th re-rtrack after 6.119 s
   Jun 12 14:17:43 192.168.128.84 Lost track
   Jun 12 14:17:46 192.168.128.84 3-th re-rtrack after 3.287 s (max delta 2 ps)
   Jun 12 14:18:48 192.168.128.84 Lost track
   Jun 12 14:19:00 192.168.128.84 4-th re-rtrack after 12.388 s (max delta 0 ps)
   Jun 12 14:19:14 192.168.128.84 Lost track
   Jun 12 14:19:18 192.168.128.84 5-th re-rtrack after 4.126 s (max delta 3 ps)
   Jun 12 14:19:25 192.168.128.84 Lost track
   Jun 12 14:19:31 192.168.128.84 6-th re-rtrack after 6.022 s (max delta 4 ps)
   Jun 12 14:19:36 192.168.128.84 Lost track
   Jun 12 14:19:38 192.168.128.84 7-th re-rtrack after 2.229 s (max delta 1 ps)
   Jun 12 14:19:52 192.168.128.84 Lost track
   Jun 12 14:20:03 192.168.128.84 8-th re-rtrack after 10.997 s (max delta 1 ps)
   Jun 12 14:21:02 192.168.128.84 Lost track
   Jun 12 14:21:21 192.168.128.84 9-th re-rtrack after 18.906 s (max delta 2 ps)
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This is my own fault introduced here:

   765443c7 udp: bugfix for checksum generation

It made no harm, because all send frames were oversized, but I'm now
using fixed-size buffers, and this overwrites the own IP address
thus corrupting the checksum and so on.

Pity me.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

we had "warning unused.variable", so let's use it
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>