svn commit: r354899 - head/usr.sbin/jail

[Warner Losh]

On Wed, Nov 20, 2019 at 9:58 AM Li-Wen Hsu <lwhsu at freebsd.org> wrote:

> On Thu, Nov 21, 2019 at 12:42 AM John Baldwin <jhb at freebsd.org> wrote:
> >
> > On 11/20/19 8:35 AM, Li-Wen Hsu wrote:
> > > Author: lwhsu
> > > Date: Wed Nov 20 16:35:58 2019
> > > New Revision: 354899
> > > URL: https://svnweb.freebsd.org/changeset/base/354899
> > >
> > > Log:
> > >   Limit the workaround to riscv only
> > >
> > >   PR:         242109
> > >   Sponsored by:       The FreeBSD Foundation
> > >
> > > Modified:
> > >   head/usr.sbin/jail/Makefile
> > >
> > > Modified: head/usr.sbin/jail/Makefile
> > >
> ==============================================================================
> > > --- head/usr.sbin/jail/Makefile       Wed Nov 20 16:32:13 2019
> (r354898)
> > > +++ head/usr.sbin/jail/Makefile       Wed Nov 20 16:35:58 2019
> (r354899)
> > > @@ -18,7 +18,9 @@ CFLAGS+=-I. -I${.CURDIR}
> > >  # workaround for GNU ld (GNU Binutils) 2.33.1:
> > >  #   relocation truncated to fit: R_RISCV_GPREL_I against `.LANCHOR2'
> > >  # https://bugs.freebsd.org/242109
> > > +.if ${MACHINE_ARCH} == "riscv"
> > >  CFLAGS+=-Wl,--no-relax
> > > +.endif
> >
> > Eh, will that work?  I think MACHINE and MACHINE_CPUARCH are riscv,
> > but MACHINE_ARCH is riscv64 and riscv64sf.
>
> No it doesn't. I was too lazy to do scp from test machine to commit
> machine and caused a copy-n-paste error, but ${MACHINE} should be
> better since it covers more.
>

No. MACHINE_CPUARCH is better because it is userland. This isn't a kernel
interface thing, so MACHINE is the wrong thing to use.

Warner


> > Also, it would be good to wrap this in .if ${LINKER_TYPE} == "bfd" I
> think.
>
> Also added in r354900. Thanks for the suggestion.
>
> > (I was able to build a world + kernel with lld earlier this week, though
> it
> >  doesn't yet boot)
>
> Looking forward to building world & kernel with llvm toolchain.
>
> Thanks,
> Li-Wen
>