I have gone the other way and created a boot loader method wherein the loader code is assembled, compiled and linked with the applications code all in one program. The applications portion still uses the concept of the moved interrupt vector table just like you learned about from the Keil application note. Using this method it is easy to have shared data areas that both modules know about. At the same time it is also possible for the loader module to use much of RAM for its buffers and re-programming functions and still let the application program utilize RAM in just the way it requires (particularly important that C-51 can use the RAM area for building its subroutine variable overlay area). To acheive this I wrote the loader function in assembly language and the application code mostly in C.

After the compile and link are completed the resulting program code can end up in a single Intel Hex file or binary file. It is a simple matter to strip the first 2K, 4K etc off this file to leave a programming file that contains just the application program.

Here is a link to the source code for the loader portion of the programming I describe above.

Michael Karas