This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH 0/5] Add support for R_386_GOT32X/R_X86_64_[REX_]GOTPCRELX
- From: "H.J. Lu" <hjl dot tools at gmail dot com>
- To: Rafael EspÃndola <rafael dot espindola at gmail dot com>
- Cc: Binutils <binutils at sourceware dot org>
- Date: Mon, 18 Jan 2016 08:53:27 -0800
- Subject: Re: [PATCH 0/5] Add support for R_386_GOT32X/R_X86_64_[REX_]GOTPCRELX
- Authentication-results: sourceware.org; auth=none
- References: <1445514099-20925-1-git-send-email-hjl dot tools at gmail dot com> <CAG3jReK+ViY071FL1EMKh5uqLydvDUgQjKL7EM--a1G23gof-g at mail dot gmail dot com> <CAMe9rOoTgGh3-z2KnVkZQvKeLeXgfEj-++9MpbxtVcUgPHZjKQ at mail dot gmail dot com> <CAG3jRe+LY9EMD3e8NrnpOv-HaQfRt1pYAgNMw-1cbxPKONLMow at mail dot gmail dot com>
On Mon, Jan 18, 2016 at 8:44 AM, Rafael EspÃndola
<rafael.espindola@gmail.com> wrote:
>>> Sorry if I missed this somewhere else, but could you explain why we
>>> need these new relocations? Is it just to move work from the linker to
>>> the assembler or there was some correctness issue optimizing a plain
>>> R_X86_64_GOTPCREL?
>>>
>>
>> They are used to convert memory operand of test and binop into
>> immediate operand, where binop is one of adc, add, and, cmp, or,
>> sbb, sub, xor instructions, when position-independent code is disabled.
>> The new relocation tells linker that it should try such optimization
>> since binop is one of mentioned above.
>
> But why not check the instruction in the linker? There is precedent,
> since the mov to lea transformation is done for R_X86_64_GOTPCREL.
>
Since for some binop, linker needs to rewrite the REX byte, it needs
to know if instruction has a REX byte. R_X86_64_GOTPCREL doesn't
have such information
--
H.J.