This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] Put GNU_UNIQUE symbols into the dynamic section.
- From: Paul Pluzhnikov <ppluzhnikov at google dot com>
- To: Cary Coutant <ccoutant at gmail dot com>
- Cc: Ian Coolidge <icoolidge at google dot com>, Binutils <binutils at sourceware dot org>, kmackay at google dot com, Han Shen <shenhan at google dot com>
- Date: Fri, 17 Apr 2015 13:31:23 -0700
- Subject: Re: [PATCH] Put GNU_UNIQUE symbols into the dynamic section.
- Authentication-results: sourceware.org; auth=none
- References: <1429255182-11847-1-git-send-email-icoolidge at google dot com> <CAJimCsFyNXziWQBTuxfpnYxKuOvV8tg21_VbduCJhNbMcsTBJw at mail dot gmail dot com> <CALoOobP4Fc1pmRjt-=2qNEoVkzfNzy-x6bnjsV9eNbpgUT_tiw at mail dot gmail dot com> <CAJimCsEFXR7Ac2_o30ij95f7XQpZok=QTMuE22DJaJAcA8z3wA at mail dot gmail dot com>
On Fri, Apr 17, 2015 at 1:14 PM, Cary Coutant <ccoutant@gmail.com> wrote:
>
> >> Gold has a --no-gnu-unique option that converts STB_GNU_UNIQUE to
> >> STB_GLOBAL while writing the dynamic symbol table. This was to work
> >> around an issue with older dynamic loaders.
> >
> > Older loaders are not at all the reason for this flag:
> > https://sourceware.org/ml/binutils/2011-10/msg00276.html
>
> Oh, yeah, forgot about that (should have searched instead of relying
> on my poor memory!). That, however, was the second time that
> STB_GNU_UNIQUE had caused problems; I think the first time really was
> because we had a loader that didn't understand the binding at all, but
> that reason probably has fallen by the wayside.
AFAICT, GNU_UNIQUE was added in glibc-2.11 in 2009:
0122f28eb (Ulrich Drepper 2009-07-21 08:32:54 -0700 872) *
STB_GNU_UNIQUE support added.
That is only a year after public release of Gold. Also, gcc/configure
tests for support for GNU_UNIQUE, and doesn't emit it if the target
libc doesn't support it. So I am not sure how you could run into
problems here. But you may well be correct ;-)
--
Paul Pluzhnikov