This introduces "mode abscal", that forces a gm-lookalike mode in the
node, to send sync once per second. See "absolute calibration" document
by Peter Jansweijer from nikhef. The option is Kconfig'd, but on by
default.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Conflicts:
	ppsi

both "sec" and "hwrs->sec" are 64 bits wide: don't forcibly trim
to 31 bit the value being assigned.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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>

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

the configuration written to WB registers using commands in the
wr-streamers tools is used to override the default configuration
set with generics. one can enable/disable this overriding, meaning
that the original default configuration can be effective again.
this functions enables/disables overriding.

- for the qtag configuration to be effective, an appropriate override
  bit needs to be set, this was missing, fixed
- the function was first enabling the qtag-ing and then configuring it,
  this is not a good behavior, it is better to be enabled later (when
  the configuration is set, otherwise it will start sending frames
  tagged with some random config), fixed

these commands are not critical for the current release, the fix will take
some time, the codebase is good. let's make it work later

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

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

This commit adds a tool to send out, through ethernet, a complete log of
DAC values for the main oscillator. It allows to see if there are
oscillations (and there currently are) and what happens when tracking
is lost and recovered.

To activate the feature

  * add CONFIG_DAC_LOG (depends on CONFIG_DEVELOPER)

  * activate on the console: "daclog <ipaddr> <macaddr>"

  * collect data on your target host

     netcat -u -l -p 1050  | od -An -t u2 -v -w2 --endian=big > daclog.txt

  * stop it when done, on the console: "daclog off"
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

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>

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

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

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

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

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

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: Alessandro Rubini <rubini@gnudd.com>

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

- new command "ls" (leapsecond to overwrite the default value
- implement get_set_tx_local_mac, get_set_tx_remote_mac
  get_set_rx_ethertype get_set_rx_local_mac, get_set_rx_remote_mac,
Signed-off-by: Michel Arruat <michel.arruat@cern.ch>

With this patch named options follow the POSIX guidelines for the
command-line options, that is to say "short" and "long" option are
supported while so far only short ones was supported.
"cmd -a 0x1 --arg 0x2"
Signed-off-by: Michel Arruat <michel.arruat@cern.ch>