svn commit: r332090 - head/stand/i386

[Warner Losh]

On Fri, Apr 6, 2018 at 11:54 AM, John Baldwin <jhb at freebsd.org> wrote:

> On Friday, April 06, 2018 02:57:58 AM Ed Maste wrote:
> > Author: emaste
> > Date: Fri Apr  6 02:57:58 2018
> > New Revision: 332090
> > URL: https://svnweb.freebsd.org/changeset/base/332090
> >
> > Log:
> >   stand: pass --no-rosegment for i386 bits when linking with lld
> >
> >   btxld does not correctly handle input with other than 2 PT_LOAD
> >   segments.  Passing --no-rosegment lets lld produce output eqivalent to
> >   ld.bfd: 2 PT_LOAD segments and no PT_GNU_RELRO.
> >
> >   PR:         225775
> >   MFC after:  3 weeks
> >   Sponsored by:       The FreeBSD Foundation
> >   Differential Revision:      https://reviews.freebsd.org/D14956
> >
> > Modified:
> >   head/stand/i386/Makefile.inc
> >
> > Modified: head/stand/i386/Makefile.inc
> > ============================================================
> ==================
> > --- head/stand/i386/Makefile.inc      Fri Apr  6 02:47:43 2018
> (r332089)
> > +++ head/stand/i386/Makefile.inc      Fri Apr  6 02:57:58 2018
> (r332090)
> > @@ -2,8 +2,13 @@
> >  #
> >  # $FreeBSD$
> >
> > +.sinclude <bsd.linker.mk>
> > +
> >  LOADER_ADDRESS?=0x200000
> >  LDFLAGS+=    -nostdlib
> > +.if defined(LINKER_TYPE) && ${LINKER_TYPE} == "lld"
> > +LDFLAGS+=    -Wl,--no-rosegment
> > +.endif
>
> Maybe we should support LDFLAGS.${LINKER_TYPE} as we do for CFLAGS, etc.?
>

I concur. It doesn't take too many of these for that to pay off big time,
especially if we have to do anything globally...

Warner