So I built a compiler in college, but we were generating MASM assembly like:

proc FooBar

... do something


proc FooBar2
call FooBar


Now I'm doing a job where their compiler is emitting x86 assembly, but is also emitting relocations. I'm confused as to how it is emitting relocations? How does it know where to put relocations anyway?

Isn't that the assembler/linker's job?

Is there like a book that explains this? It seems that is a very platform specific detail and OS specific detail so maybe there are multiple books?

P.S. - Sorry for asking such a vague question. I'm sort of lost.

Maybe you should revisit the PE format documentation.

DOS MZ will support one very simly relocation table in that it, I recommend that you maybe work in DOS MZ format. The reaosn is that the DOS MZ format support one very simply relocation table., it's easy than PE format; and second reason is that DOS MZ format only support 16bit insturction , but PE format uses 32BIT and 64bit instructions, it maybe block you to understand it.

Thanks.

Kenneth Zheng