This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: Disable R_X86_64_PLT32 generation as branch marker on Solaris/x86
- From: "H.J. Lu" <hjl dot tools at gmail dot com>
- To: Rainer Orth <ro at cebitec dot uni-bielefeld dot de>
- Cc: Binutils <binutils at sourceware dot org>
- Date: Wed, 3 Apr 2019 08:13:44 -0700
- Subject: Re: Disable R_X86_64_PLT32 generation as branch marker on Solaris/x86
- References: <yddo95wz2fi.fsf@CeBiTec.Uni-Bielefeld.DE> <CAMe9rOo-vsKNPS=_opinfCG=tiTOMa-5PauT2kmJbCQwd61wzQ@mail.gmail.com> <ydda7h7qi9a.fsf@CeBiTec.Uni-Bielefeld.DE> <CAMe9rOru1LJC=r7vY-+AcLa71fv+-NWYLDgp=-riVjQ0YQ9sdg@mail.gmail.com> <ydd5zrvqh13.fsf@CeBiTec.Uni-Bielefeld.DE>
On Wed, Apr 3, 2019 at 7:50 AM Rainer Orth <ro@cebitec.uni-bielefeld.de> wrote:
>
> Hi H.J.,
>
> >> > Please create a subdirectory, solaris, with copies of those tests and
> >> > skip original tests for Solaris targets so that the original tests can be
> >> > updated without affecting Solaris targets.
> >>
> >> How would this help? Do you want me to create copies of the affected .s
> >> files and move the *-solaris.d files both to that solaris subdirectory?
> >
> > Yes.
> >
> >> If gas produces different output for identical input, the solaris .d
> >> files have to be changed anyway, irrespective of their location.
> >
> > Not necessarily. Initially both inputs are the same. But it may
> > change over time.
>
> True, but in the overwhelming majority of cases, gas will produce the
> same output on both Linux and Solaris. When we stay with only one set
> of .s files, one immediately notices the need to adapt the Solaris .d
> files the next time make check is run, without any impact on other
> targets. We can still fork the inputs should it really becomes
> necessary at some point.
You can initially use "#source:" in solaris directory and make a
copy later when needed. I want the original tests not tied to Solaris.
--
H.J.