Can I write my own core dump that is readable by GDB?

Question

Is it possible to write core dumps manually? Lets assume I have a binary memory dump of stack, heap, code and all the processor registers.

Is it possible to create a core file that would look to GDB as if a real process has crashed and generated this core file?

The main usage for this is that I am running a VM on assembley code. Lets assume the architecture is PPC. Wn I encounter a problem I want to be able to write a core file that is readable by GDB on Linux to help debug the problem.

I am not sure what is the diffrence between ELF format core file and other formats. But what I am trying to find out is if its possible to write a module that gets data segment, code segment, stack segment and registers and produces a readable core file for GDB and if so, how complicated it that.

Answer 1

is it possible to create a core file that would look to GDB as if a real process has crashed and generated this core file?

It sure is. The format of the core is somewhat compilcated, and is system dependent (and you didn't say which system you want the solution for).

This answer gives possible ELF solutions.

If you are on non-ELF system, you'll have to tell us which one.