On Mon, Mar 30, 2020 at 11:18 PM Fangrui Song <i@maskray.me> wrote:
>
> On 2020-03-30, H.J. Lu via Binutils wrote:
> >Since value of non-preemptible absolute symbol (SHN_ABS) won't change,
> >only S - A relocations against non-preemptible absolute symbol are
> >allowed in PIE and shared library.
>
> Thanks for the patch. At some point we should define some generic
> relocation categories (https://reviews.llvm.org/rL266158). For example,
>
> - R_ABS: R_X86_64_8, R_X86_64_16, R_X86_64_32, R_X86_64_64
>           R_AARCH64_ABS16, R_AARCH64_ABS32, R_AARCH64_ABS64
>           R_PPC64_ADDR16, R_PPC64_ADDR16_DS, R_PPC64_ADDR16_HA, ...
> - R_PCREL: R_X86_64_PC8, R_X86_64_PC16, R_X86_64_PC32, R_X86_64_PC64
>           R_AARCH64_PREL16, R_AARCH64_PREL32, R_AARCH64_PREL64
> - R_PLT_PCREL: R_X86_64_PLT32, R_AARCH64_CALL26
> - R_GOT_PCREL: R_X86_64_GOTPCREL, R_X86_64_GOTPCRELX
>
> Define a property is_preemptible as a more appropriate level of
> abstraction for GENERATE_DYNAMIC_RELOCATION_P and some use cases of bfd_link_pic and bfd_link_executable...

Current BFD linker delegates most, if not all, of relocation to each backend.
There are many similar codes in backends.  I am sharing as much codes
between i386 and x86-64 backends as I can.

This is the patch I am checking in.

-- 
H.J.