How feasible is it to port gold to mingw?

[Joseph S. Myers]

On Sun, 5 Dec 2010, Greatwolf wrote:

>   how feasible would it be to port gold over to windows assuming someone 
> is willing to take on this workload? 
>   
>   What part of the code base in gold would likely need to be modified? 

What you would need to do is define a way of representing all features of 
Windows binaries in ELF, create configurations of the compiler and 
assembler that generate ELF, have gold link this form of ELF and then have 
a postlinking tool that converts the final ELF executables and shared 
libraries into PE-COFF binaries that can run directly on Windows.  (See 
the model used for arm-symbianelf, for example; 
<http://www.martin.st/symbian/> has tools that I think may include one 
implementation of the postlinker.)  There might also be comparatively 
small and self-contained gold patches to run a postlinker automatically 
after the main ELF linking stage, or you might make GCC do that somehow.

>   How much work would need to be done? Doesn't need to be precise; I'm 
> looking for a ballpark figure here. eg. would 20%, 50%, or 80% of the 
> code need to be change due to platform specific details?
> 
>   What sort of technical knowledge would one need to be able to perform 
> this task?

In-depth technical knowledge of both Windows and ELF ABIs would be needed 
with experience of ABI design and specification work.

-- 
Joseph S. Myers
joseph@codesourcery.com