This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] clean up elf32-i386-vxworks hooking mechanism


On Wed, Jul 27, 2011 at 2:49 PM, Roland McGrath <mcgrathr@google.com> wrote:
> This cleans up the BFD elf32-i386 target vector to put its
> backend-specific parameters into a new struct hung off of struct
> elf_backend_data rather than overloading struct elf_i386_link_hash_table
> for this purpose. ?The latter is actual dynamic state. ?The former is
> just a constant table that varies by particular backend.
>
> I tested that this showed no regressions for --target=i686-wrs-vxworks
> compared to the unmodified trunk. ?However, since I don't actually have
> vxworks, I'm not sure how thorough my testing really was. ?To be able to
> do any tests at all, I built gcc --target=i686-wrs-vxworks from the 4.6
> branch. ?But I was not able to do a really usable build, i.e. no libgcc,
> just gcc itself (--enable-languages=c). ?The results (both before and
> after this bfd change) look like this:
>
> ? ? ? ? ? ? ? ? ? ? ? ?=== binutils Summary ===
>
> ? ? ? ?# of expected passes ? ? ? ? ? ?72
> ? ? ? ?# of expected failures ? ? ? ? ?1
> ? ? ? ?# of untested testcases ? ? ? ? 8
> ? ? ? ?# of unsupported tests ? ? ? ? ?9
>
> ? ? ? ? ? ? ? ? ? ? ? ?=== ld Summary ===
>
> ? ? ? ?# of expected passes ? ? ? ? ? ?58
> ? ? ? ?# of expected failures ? ? ? ? ?5
> ? ? ? ?# of untested testcases ? ? ? ? 12
> ? ? ? ?# of unsupported tests ? ? ? ? ?13
>
> Ok for trunk?
>
>
> Thanks,
> Roland
>
>
> bfd/
> 2011-07-27 ?Roland McGrath ?<mcgrathr@google.com>
>
> ? ? ? ?* elf32-i386.c (struct elf_i386_backend_data): New type.
> ? ? ? ?(get_elf_i386_backend_data): New macro.
> ? ? ? ?(elf_i386_arch_bed): New variable.
> ? ? ? ?(elf_backend_arch_data): New macro.
> ? ? ? ?(struct elf_i386_link_hash_table): Remove plt0_pad_byte and is_vxworks.
> ? ? ? ?(elf_i386_link_hash_table_create): Don't initialize them.
> ? ? ? ?(elf_i386_create_dynamic_sections): Find is_vxworks flags in
> ? ? ? ?elf_i386_backend_data, not elf_i386_link_hash_table.
> ? ? ? ?(elf_i386_adjust_dynamic_symbol): Likewise.
> ? ? ? ?(elf_i386_allocate_dynrelocs): Likewise.
> ? ? ? ?(elf_i386_readonly_dynrelocs): Likewise.
> ? ? ? ?(elf_i386_size_dynamic_sections): Likewise.
> ? ? ? ?(elf_i386_relocate_section): Likewise.
> ? ? ? ?(elf_i386_finish_dynamic_symbol): Likewise.
> ? ? ? ?(elf_i386_finish_dynamic_sections): Likewise. ?Same for plt0_pad_byte.
> ? ? ? ?(elf_i386_vxworks_link_hash_table_create): Function removed.
> ? ? ? ?(elf_i386_vxworks_arch_bed): New variable.
> ? ? ? ?(elf_backend_arch_data): New macro in elf32-i386-vxworks stanza.
>
> ? ? ? ?* elf-bfd.h (elf_backend_data): New member arch_backend_data.
> ? ? ? ?* elfxx-target.h (elf_backend_arch_data): New macro.
> ? ? ? ?(elfNN_bed): Use it as initializer for the new member.
>

elf32-i386.c change is OK.

Thanks.

-- 
H.J.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]