https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123976

--- Comment #187 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Peter Dyballa from comment #186)
> Last try was with:
> 
> env PKG_CONFIG_PATH=/usr/lib/pkgconfig
> PATH=/bin:/usr/bin:/sbin:/usr/sbin:/opt/local/libexec/gcc10-bootstrap/bin
> LANG=C ../gcc-16-branch-gcc-16-1-darwin/configure --disable-libgomp
> --disable-multilib CC=powerpc-apple-darwin9-gcc CXX=powerpc-apple-darwin9-g++
> 

> n/Objekte/./gmp/.libs

no need to get distracted by the GMP etc. stuff .. that's normL.


> -L/Users/pete/Quellen/Objekte/./mpfr/src/.libs
> -L/Users/pete/Quellen/Objekte/./mpc/src/.libs -lmpc -lmpfr -lgmp  
> -L./../zlib -lz 
> ld: bl out of range (-16824316 max is +/-16M) from
> __ZSt21__throw_bad_exceptionv at 0x03AC2C7C in __text_cold of
> /opt/local/libexec/gcc10-bootstrap/lib/gcc/powerpc-apple-darwin9/10.5.0/../..
> /../libstdc++.a(functexcept.o) to ___cxa_allocate_exception$island_2 at
> 0x02AB749C in __text of  cc1 in __ZSt21__throw_bad_exceptionv from
> /opt/local/libexec/gcc10-bootstrap/lib/gcc/powerpc-apple-darwin9/10.5.0/../..
> /../libstdc++.a(functexcept.o)
> collect2: error: ld returned 1 exit status
> make[3]: *** [cc1] Error 1
> make[2]: *** [all-stage1-gcc] Error 2
> make[1]: *** [stage1-bubble] Error 2
> make: *** [all] Error 2
> 
> I do remember having had 'bl out of range' two years ago when trying to
> build MuPDF on PPC Tiger, Mac OS X 10.4.11, darwin8, Xcode 2.5:
> https://trac.macports.org/ticket/69879. Six weeks later I sent an eMail to
> gcc-help list – no answer. I reported that to Artifex,
> https://bugs.ghostscript.com/show_bug.cgi?id=707844, where it was argued
> that this "compiler/linker issue" happened because "GCC 7.5 is quite old".
> 
> I have in Apple Mail an eMail discussion with Gordon Steemson, probably from
> gcc-help, who tried to help, with the recommendation to use '"-mlongcall" or
> its synonym "-mlong-branch"', but no solution in the end (and MacPorts
> stopped supporting Tiger).

This is caused by a bug in ld64 (there's a series of partial fixes and
work-arounds in various versions of ld64) ... I do have a fixed version but
that would mean you building some other code.

I have some other things that need addressing - but will take a closer look
later (note that I do not see this issue in my build - and I am also using a
version of ld64 with the bug) . . I wonder if you are somehow pulling in a
different ld64 ...

Reply via email to