Re: [RFC PATCH] configure check for ld's --no-relax flag

[Andrei Borzenkov]

В Sun, 21 Sep 2014 19:03:08 +0200
Vladimir 'φ-coder/phcoder' Serbinenko <address@hidden> пишет:

> On 04.08.2014 16:53, Andrey Borzenkov wrote:
> > В Mon, 04 Aug 2014 10:45:22 +0400
> > Stanislav Kholmanskikh <address@hidden> пишет:
> > 
> >> Hi!
> >>
> >> On 08/01/2014 07:40 PM, Vladimir 'φ-coder/phcoder' Serbinenko wrote:
> >>> On 01.08.2014 17:35, Andrey Borzenkov wrote:
> >>>> В Fri,  1 Aug 2014 16:15:56 +0400
> >>>> Stanislav Kholmanskikh <address@hidden> пишет:
> >>>>
> >>>>> Early versions of binutils doesn't support --no-relax flag, so
> >>>>> commit 063f2a04d158ec1b275a925dfbae74b124708cde prevents building
> >>>>> with such versions.
> >>>>>
> >>>>> Signed-off-by: Stanislav Kholmanskikh <address@hidden>
> >>>>> ---
> >>>>>   conf/Makefile.common |    8 ++++++++
> >>>>>   configure.ac         |   10 ++++++++++
> >>>>>   2 files changed, 18 insertions(+), 0 deletions(-)
> >>>>>
> >>>>> diff --git a/conf/Makefile.common b/conf/Makefile.common
> >>>>> index e4c301f..5bda66f 100644
> >>>>> --- a/conf/Makefile.common
> >>>>> +++ b/conf/Makefile.common
> >>>>> @@ -8,11 +8,19 @@ unexport LC_ALL
> >>>>>   # Platform specific options
> >>>>>   if COND_sparc64_ieee1275
> >>>>>     CFLAGS_PLATFORM += -mno-app-regs
> >>>>> +if COND_LD_SUPPORTS_NO_RELAX
> >>>>>     LDFLAGS_PLATFORM = -Wl,-melf64_sparc -Wl,--no-relax
> >>>>> +else
> >>>>> +  LDFLAGS_PLATFORM = -Wl,-melf64_sparc -mno-relax
> >>>>> +endif
> >>>>
> >>>> TBO I think commit should simply be reverted. "Uniformity" is rather
> >>>> poor excuse for breaking existing systems.
> >>>>
> >>> This commit is needed for clang to compile for sparc64. Given that
> >>> sparc64 clang still doesn't really work I'm ok with reverting, at least
> >>> for now.
> >>
> >> But, it this case, maybe it would be better to consider 
> >> reviewing/applying this patch? Just to not return to this issue after 
> >> some time?
> >>
> >> Andrey, Vladimir, what do you think?
> >>
> > 
> > Yes, commit message was pretty confusing. This leaves the question,
> > whether combination of clang and binutils that do not support
> > -Wl,--no-relax exists though :) Otherwise I agree, we should use this
> > patch.
> > 
> I think we could try to push for clang to have -mno-relax. They're
> usually pretty responsive and we'll probably need few fixes for few
> other clang problems anyway. For now I just reverted it.

Did you push it?

signature.asc
Description: PGP signature