[lld] r262026 - Add comment on AMDGPU that the difference has no obvious reason.

[Rafael Espíndola via llvm-commits]

> We can probably model this within ELF by just putting each section inside a
> PT_LOAD (or whatever getAmdgpuPhdr returns) with vaddr of 0.

I am not sure we could call that an "address". The fundamental problem
that needs to be solved is that there shouldn't be a special case for
a program handling amd gpu or the XYZ cpu. It has to be possible to
just read PT_LOAD and mmap them as usual.

Cheers,
Rafael