reduce @20211019_0: error: ld: symbol(s) not found for architecture arm64

[zzanderr]

Build fails on MacBook Pro, M2, Sonoma 14.1, Xcode 15.1 Beta

:info:build libtool: link: gcc -dynamiclib  -o .libs/libffi.7.dylib  src/.libs/prep_cif.o src/      .libs/types.o src/.libs/raw_api.o src/.libs/java_raw_api.o src/.libs/closures.o src/x86/.libs/      ffi64.o src/x86/.libs/unix64.o src/x86/.libs/ffiw64.o src/x86/.libs/win64.o    -arch x86_64 -a      rch x86_64 -arch x86_64   -install_name  /opt/local/var/macports/build/_opt_local_var_macports      _sources_rsync.macports.org_macports_release_tarballs_ports_math_reduce/reduce/work/Reduce-svn      6110-src/cslbuild/aarch64-apple-darwin23.1.0/libffi/build-x86_64/../x86_64/lib/libffi.7.dylib       -compatibility_version 9 -current_version 9.0
14522 :info:build gcc: warning: this compiler does not support x86 ('-arch' option ignored)
14523 :info:build Undefined symbols for architecture arm64:
14524 :info:build   "_ffi_call", referenced from:
14525 :info:build       _ffi_raw_call in raw_api.o
14526 :info:build       _ffi_java_raw_call in java_raw_api.o
14527 :info:build   "_ffi_prep_cif_machdep", referenced from:
14528 :info:build       _ffi_prep_cif_core in prep_cif.o
14529 :info:build   "_ffi_prep_closure_loc", referenced from:
14530 :info:build       _ffi_prep_closure in prep_cif.o
14531 :info:build       _ffi_prep_raw_closure_loc in raw_api.o
14532 :info:build       _ffi_prep_java_raw_closure_loc in java_raw_api.o
14533 :info:build ld: symbol(s) not found for architecture arm64
14534 :info:build collect2: error: ld returned 1 exit status
14535 :info:build gmake[7]: *** [Makefile:1070: libffi.la] Error 1

[zzanderr]

reduce build.log

[zzanderr]

I'm not really sure why the compiler claims not to support x86 if it's the Xcode compiler. I've built a few universal binaries with Xcode's Clang without any trouble.

There are additional warnings in the configure stage reporting many macros to be obsolete and that autoupdate should be run:

 9645 :info:configure /opt/local/share/aclocal/libvideogfx.m4:7:   or see https://www.gnu.org/softwa      re/automake/manual/automake.html#Extending-aclocal
 9646 :info:configure autoreconf: running: /opt/local/bin/autoconf --force
 9647 :info:configure configure.ac:8: warning: The macro `AC_CANONICAL_SYSTEM' is obsolete.
 9648 :info:configure configure.ac:8: You should run autoupdate.
 9649 :info:configure ./lib/autoconf/general.m4:2081: AC_CANONICAL_SYSTEM is expanded from...
 9650 :info:configure configure.ac:8: the top level
 9651 :info:configure configure.ac:41: warning: The macro `AC_PROG_LIBTOOL' is obsolete.
 9652 :info:configure configure.ac:41: You should run autoupdate.
 9653 :info:configure m4/libtool.m4:100: AC_PROG_LIBTOOL is expanded from...
 9654 :info:configure configure.ac:41: the top level
 9655 :info:configure configure.ac:68: warning: The macro `AC_TRY_COMPILE' is obsolete.
 9656 :info:configure configure.ac:68: You should run autoupdate.
 9657 :info:configure ./lib/autoconf/general.m4:2847: AC_TRY_COMPILE is expanded from...
 9658 :info:configure lib/m4sugar/m4sh.m4:692: _AS_IF_ELSE is expanded from...
 9659 :info:configure lib/m4sugar/m4sh.m4:699: AS_IF is expanded from...
 9660 :info:configure ./lib/autoconf/general.m4:2249: AC_CACHE_VAL is expanded from...
 9661 :info:configure ./lib/autoconf/general.m4:2270: AC_CACHE_CHECK is expanded from...
 9662 :info:configure acinclude.m4:3: AC_FUNC_MMAP_BLACKLIST is expanded from...
 9663 :info:configure configure.ac:68: the top level

and so on until line 9714.

[zzanderr]

Just to emphasize: I am running Xcode 15.1 _Beta_.

[zzanderr]

Tried to autoupdate configure.ac and got:

$ sudo autoupdate configure.ac
aclocal.m4:17: warning: this file was generated for autoconf 2.69.
You have another version of autoconf.  It may work, but is not guaranteed to.
If you have problems, you may need to regenerate the build system entirely.
To do so, use the procedure documented by the package, typically 'autoreconf'.

Note: I am not at all sure if I did this correctly (i.e., I don't know what I'm doing). Just thought I'd throw it in if it helps.

[zzanderr]

I am using autoconf @2.71_2.

[ryandesign (Ryan Carsten Schmidt)]

Nobody knows what gcc is on your system. This port is not UsingTheRightCompiler, so it could be anything.

Warnings from autoreconf are for the developer to fix. As long as autoreconf did not end in failure (which it didn't) I'm optimistic that the warnings are not the reason why the build failed.

[zzanderr]

I have MacPorts gcc12 and gcc13; however, I haven't set 'port select' for either of them, as I only have them for other ports that depend on them.

[zzanderr]

I deactivated both gcc12 and gcc13, and reduce built fine. Thanks for pointing me in the right direction.

I built reduce a few months ago with a late version of Xcode 14 and must have had MacPorts gcc12 activated at the time, but it wasn't pulled into the build. I don't think I had gcc13 installed at the time, but I'm not completely sure.

[zzanderr]

No, I didn't have gcc13 installed the last time reduce built successfully.

[barracuda156]

We should actually update this port.

[barracuda156]

I tried the latest version, but it is broken for ppc, or needs some different settings. Gonna return to this, but not today.

libtool: compile:  /opt/local/bin/gcc-mp-13 -DHAVE_CONFIG_H -I. -I/opt/local/var/macports/build/_opt_PPCSnowLeopardPorts_math_reduce/reduce/work/Reduce-svn6547-src/libraries/libffi -I. -I/opt/local/var/macports/build/_opt_PPCSnowLeopardPorts_math_reduce/reduce/work/Reduce-svn6547-src/libraries/libffi/include -Iinclude -I/opt/local/var/macports/build/_opt_PPCSnowLeopardPorts_math_reduce/reduce/work/Reduce-svn6547-src/libraries/libffi/src -isystem/opt/local/include/LegacySupport -I/opt/local/include -I/opt/local/include -O3 -fomit-frame-pointer -fstrict-aliasing -ffast-math -mcpu=970 -Wall -fexceptions -MT src/powerpc/ffi_darwin.lo -MD -MP -MF src/powerpc/.deps/ffi_darwin.Tpo -c /opt/local/var/macports/build/_opt_PPCSnowLeopardPorts_math_reduce/reduce/work/Reduce-svn6547-src/libraries/libffi/src/powerpc/ffi_darwin.c  -fno-common -DPIC -o src/powerpc/.libs/ffi_darwin.o
/opt/local/var/macports/build/_opt_PPCSnowLeopardPorts_math_reduce/reduce/work/Reduce-svn6547-src/libraries/libffi/src/powerpc/ffi_darwin.c:1114:22: error: unknown type name 'ffi_go_closure'; did you mean 'ffi_raw_closure'?
 1114 | ffi_prep_go_closure (ffi_go_closure* closure,
      |                      ^~~~~~~~~~~~~~
      |                      ffi_raw_closure
/opt/local/var/macports/build/_opt_PPCSnowLeopardPorts_math_reduce/reduce/work/Reduce-svn6547-src/libraries/libffi/src/powerpc/ffi_darwin.c:1172:31: error: unknown type name 'ffi_go_closure'; did you mean 'ffi_raw_closure'?
 1172 | ffi_go_closure_helper_DARWIN (ffi_go_closure*, void *,
      |                               ^~~~~~~~~~~~~~
      |                               ffi_raw_closure
/opt/local/var/macports/build/_opt_PPCSnowLeopardPorts_math_reduce/reduce/work/Reduce-svn6547-src/libraries/libffi/src/powerpc/ffi_darwin.c:1434:31: error: unknown type name 'ffi_go_closure'; did you mean 'ffi_raw_closure'?
 1434 | ffi_go_closure_helper_DARWIN (ffi_go_closure *closure, void *rvalue,
      |                               ^~~~~~~~~~~~~~
      |                               ffi_raw_closure
gmake[4]: *** [Makefile:1281: src/powerpc/ffi_darwin.lo] Error 1

[kencu (Ken)]

I fixed this ppc issue in our libffi port a few years ago:

​https://github.com/macports/macports-ports/commit/5287b4e585c1277c33540e78cc8c8ef626dd776e

It looks like reduce is using a bundled libffi, so doesn’t have that fix.

BTW, I updated and installed reduce on arm64 yesterday… it builds but crashed on launch, so needs some testing and may need some sortingbout.

[ryandesign (Ryan Carsten Schmidt)]

Making reduce use MacPorts libffi instead of a bundled copy would be a good idea too.