Re: [Flashforth-devel] K42 problem

[Mikael N. <mik...@fl...>]

Hi Attila.

The K42 is a bit problematic.

Interrupts do not work with FlashForth on K42.
It seems that just a simple TMR2 interrupt that only resets the 
interrupt flag,
occasionally corrupts the main program flow. Causing the effect you 
mention...
I have looked for a long time to find the problem in the code, but could 
not find it.
I suspect it is a fault in the K42 silicon that its HW context interrupt 
saving
does not work with some code sequence in FlashForth.

Therefore I decided to have the system ticks to be incremented by 
polling in PAUSE.
That works reasonably well.
With the code in git, I can compile any code and it works on my 26K42.

So I am bit surprised that you have this problem.
The UART interrupts could cause a problem if they occur in the middle of 
the
compilation, but during compilation of your one liner
there is normally no UART RX activity.

I use the ff-shell.py which has end-of-line flow control so that during 
the
interpreting/compilation of a line there is no UART RX activity.

Maybe the 25K42 has some difference to the 26K42 that I have not 
noticed.

BR Mikael