We have an embedded system (ES) which was built with CCS 3.1. ES 32c067 c code is running on a system that is not accessible by jtag. The ES code is designed to respond to remote commands and load data at a specified address. We have been using this load facility to patch ES code for some time by inserting patch code/data into unused sections of memory. The patch code is created by compiling with CCS 3.1, loading into a local jtag accessible processor, and dumping the patch area with CCS.
We've been limited to patching code that calls either none or a very small number of functions in the original code.
It would be highly desirable to build a memory image on the local processor that, except for patched areas, duplicates the ES code's memory layout. That would allow patch code to access ES functions without manual address relocation computations.
Attempts at forcing the linker to allocate sections at the ES locations have not suceeded; specifying link order doesn't seem to affect anything. Also, sections such as .pip and .hst that are shown in the map file as part of 'project'cfg.obj are loaded at different addresses and I can't find any directives that change that.
Suggestions on how to proceed would be greatly appreciated.
