Many thanks for your fast answer and suggestions.
I did a quick test with the RPi2 (1.1 of April 10, 2015) image on the Optr.link repository.
This appears to be built on Debian (that is OK), I had to install the PICNC drivers and the MF70 files.
To be short: it works now. The status led now blinks at twice the speed when Machinekit is running. I pushed a few buttons on Axis GUI and all seems to work, stepper pulse and direction signals are generated as they should. I"l do some more test later today.
Now I have the following problem: my stepper driver board needs a active high enable signal, currently it is active low. I could reverse the polarity of this signal (in the *.hal file or in the firmware) but this would not help since the output of pin6 of the PIC32 (PB2, signal EN) is also used to enable the 74HCT367 drivers. Any advise here ? Is there a particular reason to have the outputs in tri-state during idle ?
See the picture of my set-up, left is the stepper driver board, home built based on TB6560 chip (it is fed independently and includes a power sequencer). The PICNC board is in the middle and the RasPi2 is at the right (obviously).
As you will appreciate, the connection between the PICNC and the RasPi2 is really short, it is a small board with nothing more than a 8pin and a 40pin female connector. The SPI signals are wired between both connectors together with the the GND (the ground plane) and the +5V power (to feed the PICNC).
An other thing if I may. My previous set-up (with a parallel port out of a Ubuntu desktop) uses the spindle-PWM signal, which seems to lack here. I wonder if it is possible to use one of the flood/mist signals (which are not used anyway) to output a PWM signal of say 100Hz.
I could probably manage the firmware myself for the PWM generation part using the still remaining resources of the PIC32MX150F128B chip.
Where I need guidance is on how to adapt the SPI communication part (an additional uint8 variable), the PICNC driver part and the *.hal part.
I am quite happy to modify the several files and especially to test them, I rather not reverse engineer e.g. the PICNC driver on my own.
Also, they may be other and more clever ways to implement to implement the pindle-PWM signal (the speed and accuracy not being that important), e.g. in the RasPi itself.
I am very happy with the progress so far, many thanks again for all the work on this fine project and for your help.