syntax changes from SDCC to KEIL

As already noted, porting from one compiler to another will always require a thorough study of the documentation for both compilers - in particular, the implementation-defined standard aspects and any non-standard extensions.

The usual approach is to "wrap" the dependencies in conditional compilation; eg,

#if defined SDCC
#define XDATA __xdata
#elif deinfed KEIL
#define XDATA xdata
#else
#error "Unknown compiler"
#endif

You will have to look up the actual symbols to identify the two compilers.

I would suggest that you maintain both versions - so that you can always go back to the SDCC version to check that your Keil version is still working as required.

If you're lucky, you might find that someone has already done this ...

If you're lucky, you might find that someone has already done this .

most (all?) SiLabs sample code can run either SDCC or Keil

you are wrong because it isn't just the hardware, it never is "just the hardware".

When one compiles a program in SDCC , he has to add a parameter which tells SDCC for what hardware SDCC is compiling. Than SDCC links the interrupts to the hardware addresses. But it is SDCC which does this for you. And I need to be able to it myself because of our FPGA chips. And this shouldn't be impossible.

Tnx. The manuals are quite usefull. With SDCC I never used a manual because I didn't had to. From the library files as well as an example program I could get what I needed with ease. I will take a look in Keil's as well.

I only don't onderstand the "using" keywoard yet, I never did. I've read the manual about it but I am not getting the purpose of the register banks and in partically what "using" does when used in combination with the interrupts. I only know that if you call a function from within a function, they should be in the same register bank to prevent strange behaviour

For my serial interrupt I use "using 2" because it was there in an example program. But I don't understand why or what.

Well, if SDCC doesn't have the feature then, presumably, none of your existing code uses it - so you don't need to worry about it at this point!

Start a separate, specific thread once you've got the basic port done.

Unless you say otherwise, we use register bank 0, but you could reserve another of the register banks for a function, so you can always have its registers ready to go.

the most useful use of register banking is ISRs