lib/fdelay-pulse-tom: dump start/end/delta timestamps. Default pulse width to 250 ns if not specified
Signed-off-by: Luis Fernando Ruiz Gago <lfruiz@cern.ch>

Signed-off-by: Luis Fernando Ruiz Gago <lfruiz@cern.ch>

The new calibration is now there, but some details are still missing,
like using the DAC value for the VCXO and validating against the
installed gateware.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This is a massive change, spread across several files.
     - the calibration structure is changed
     - it doesn't live at offset 0x1800 any more
     - it must beloaded by SDB structures (not there yet)
     - all calibration users are fixed
     - calibration.c is a new file dealing with it
     - i2c.c has some material less, because of calibration.c
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

The SDB library is an exact copy of the sdbfs/lib directory of repo:
   git://ohwr.org/hdl-core-lib/fpga-config-space.git
at version "sdb-v1.1" (commit de2c98f).

It actually is the same code that is already in fmc-bus/sdb-lib, but
until I finalize eeprom SDB support within fmc-bus I need a copy here.
This directory will disappear when fmc-bus is completed in that respect,
but we need a release with SDB in the eeprom now for fine-delay.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

This old FIXME is now fixed. Unfortunately the calculation involved is
very boring, given the range of values we deal with.

With this in place, we can now generate a 10MHz square wave and
simulate the inputs for a WR grandmaster.
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: Federico Vaga <federico.vaga@gmail.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>

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 commit uses a non-constant i2c address so a multi-slot
carrier can be supported.  Also, all constant register offsets are
removed, because they are retrieved from SDB.  Unfortunately
this  doesn't yet support several cores with the same device ID.
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>

Also, this turns the sdb part to spec-specific code, hopefully
as a temporary measure.

There are 4 possible combinations for device names.

They can be in /dev or in /dev/zio according to which udev version
you run: older ones didn't create the directory.

Moreover the device name can be called "zio-fd" or just "fd" according
to which version of ZIO you run: when we backported to 2.6.24 we dropped
the leading "zio-" part because old kernels had a short limit on device
names. Unfortunatelty fine-delay is being used on 2.6.24 as well.

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

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

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