Re: lang/sbcl build failure, ld issue

[Chavdar Ivanov <ci4ic4%gmail.com@localhost>]

On Thu, 31 Aug 2023 at 19:52, Robert Swindells <rjs%fdy2.co.uk@localhost> wrote:
>
>
> Chavdar Ivanov <ci4ic4%gmail.com@localhost> wrote:
> > As part of my current pkg_rr lang/sbcl came to be rebuilt - albeit I
> > already had the latest pkgsrc version 2.3.6; it only depends on
> > math/mpfr, which wasn't selected for an upgrade (as there isn't any),
> > so it is not clear for me why it had to be rebuilt.
>
> I added math/mpfr as a dependency as it is used by the sb-mpfr SBCL
> contrib module. It really needs math/gmp as well, was going to add
> this when I update the package to 2.3.8.
>
> SBCL is linking against the installed versions of these packages, not
> against the bl3 tree.
>
> > Anyway, the build failed with:
> > ....
> > gcc -g -Wl,--export-dynamic -export-dynamic -no-pie -o sbcl alloc.o
> > arena.o backtrace.o breakpoint.o coalesce.o coreparse.o dynbind.o
> > funcall.o gc-common.o globals.o hopscotch.o interr.o interrupt.o
> > largefile.o
> > main.o math.o monitor.o murmur_hash.o os-common.o parse.o print.o
> > regnames.o runtime.o safepoint.o save.o sc-offset.o search.o
> > stringspace.o thread.o time.o validate.o var-io.o vars.o wrap.o
> > run-program.o sprof.
> > o x86-64-arch.o bsd-os.o x86-64-bsd-os.o fullcgc.o gencgc.o
> > traceroot.o immobile-space.o elf.o x86-64-assem.o
> > ../../tlsf-bsd/tlsf/tlsf.o -lutil  -lm
> > ld: alloc.o: relocation R_X86_64_32 against `.rodata.str1.1' can not
> > be used when making a PIE object; recompile with -fPIE
> > ld: failed to set dynamic section sizes: bad value
> > make: *** [GNUmakefile:102: sbcl] Error 1
> > make: Leaving directory '/usr/pkgsrc/lang/sbcl/work/sbcl-2.3.6/src/runtime'
> > ....
>
> The same kind of error appears in some of the bulk build runs:
>
> <https://us-central.manta.mnx.io/pkgsrc/public/reports/NetBSD/trunk/x86_64/20230821.1117/sbcl-2.3.6/build.log>
>
> and
>
> <http://www.ki.nu/pkgsrc/reports/current/NetBSD-9.0/20230827.2307/sbcl-2.3.6/build.log>
>
> > My SBCL_BOOT_SYSTEM is sbcl, obviously (clisp is many times slower),
> > but just in case I repeated the build using clisp, to the same result.
>
> Allowing SBCL_BOOT_SYSTEM to point to something outside pkgsrc is
> deliberate, I have tested using this for all the recent updates that
> I did.
>
> > This is on -current amd64 10.99.7 from 20th of August. The existing
> > sbcl2.3.6 continues to work as expected.
>
> Looking at my /etc/mk.conf, I have PKGSRC_MKPIE=no set. I think I added
> this when the default changed as I didn't want to rebuild everything.
>
> Maybe we need to set MKPIE_SUPPORTED=no in lang/sbcl/Makefile.

It appears so - it builds OK with it. I am running the tests now, as usual.



-- 
----