libffi @3.3_1: error: expected ‘)’ before ‘*’ token

[TzunamiOSX]

Want to update ffmpeg and get an error on libffi

ffi_darwin.lo -MD -MP -MF src/powerpc/.deps/ffi_darwin.Tpo -c ../src/powerpc/ffi_darwin.c  -fno-common -DPIC -o src/powerpc/.libs/ffi_darwin.o
:info:build ../src/powerpc/ffi_darwin.c:1114: error: expected ‘)’ before ‘*’ token
:info:build ../src/powerpc/ffi_darwin.c:1172: error: expected ‘)’ before ‘*’ token
:info:build ../src/powerpc/ffi_darwin.c:1434: error: expected ‘)’ before ‘*’ token
:info:build make[3]: *** [src/powerpc/ffi_darwin.lo] Error 1
:info:build make[3]: Leaving directory `/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_devel_libffi/libffi/work/libffi-3.3/powerpc-apple-darwin9.8.0'
:info:build make[2]: *** [all-recursive] Error 1
:info:build make[2]: Leaving directory `/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_devel_libffi/libffi/work/libffi-3.3/powerpc-apple-darwin9.8.0'
:info:build make[1]: *** [all] Error 2
:info:build make[1]: Leaving directory `/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_devel_libffi/libffi/work/libffi-3.3/powerpc-apple-darwin9.8.0'
:info:build make: *** [all] Error 2
:info:build make: Leaving directory `/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_devel_libffi/libffi/work/libffi-3.3'
:info:build Command failed:  cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_devel_libffi/libffi/work/libffi-3.3" && /usr/bin/make -w all 
:info:build Exit code: 2
:error:build Failed to build libffi: command execution failed
:debug:build Error code: CHILDSTATUS 19223 2
:debug:build Backtrace: command execution failed
:debug:build     while executing
:debug:build "system {*}$notty {*}$nice $fullcmdstring"
:debug:build     invoked from within
:debug:build "command_exec build"
:debug:build     (procedure "portbuild::build_main" line 8)
:debug:build     invoked from within
:debug:build "$procedure $targetname"
:error:build See /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_devel_libffi/libffi/main.log for details.

[TzunamiOSX]

mainlog

[kencu (Ken)]

that error usually means you need a newer compiler to build it...try gcc7.

hopefully we can tweak it to build with gcc-4.2 as bootstrapping is more complicated otherwise.

[TzunamiOSX]

Replying to kencu:

that error usually means you need a newer compiler to build it...try gcc7.

hopefully we can tweak it to build with gcc-4.2 as bootstrapping is more complicated otherwise.

ok, i will try

see you in 3 days :P

Thx :)

[TzunamiOSX]

Replying to kencu:

that error usually means you need a newer compiler to build it...try gcc7.

hopefully we can tweak it to build with gcc-4.2 as bootstrapping is more complicated otherwise.

Cant install gcc7 because gcc7 wants libffi to install

[kencu (Ken)]

yep, that's the bootstrapping issue I was talking about.

turns out gcc7 doesn't fix it either, but at least it has better error messages to get us closer to the real issue:

libtool: compile:  /opt/local/bin/gcc-mp-7 -DHAVE_CONFIG_H -I. -I.. -I. -I../include -Iinclude -I../src -I/opt/local/include -pipe -Os -arch ppc -Wall -fexceptions -MT src/powerpc/ffi_darwin.lo -MD -MP -MF src/powerpc/.deps/ffi_darwin.Tpo -c ../src/powerpc/ffi_darwin.c  -fno-common -DPIC -o src/powerpc/.libs/ffi_darwin.o
../src/powerpc/ffi_darwin.c:1114:22: error: unknown type name 'ffi_go_closure'; did you mean 'ffi_raw_closure'?
 ffi_prep_go_closure (ffi_go_closure* closure,
                      ^~~~~~~~~~~~~~
                      ffi_raw_closure
../src/powerpc/ffi_darwin.c:1172:31: error: unknown type name 'ffi_go_closure'; did you mean 'ffi_raw_closure'?
 ffi_go_closure_helper_DARWIN (ffi_go_closure*, void *,
                               ^~~~~~~~~~~~~~
                               ffi_raw_closure
../src/powerpc/ffi_darwin.c:1434:31: error: unknown type name 'ffi_go_closure'; did you mean 'ffi_raw_closure'?
 ffi_go_closure_helper_DARWIN (ffi_go_closure *closure, void *rvalue,
                               ^~~~~~~~~~~~~~
                               ffi_raw_closure

[kencu (Ken)]

I thought perhaps deactivating the previous libffi might fix this, but no go.

So we need to figure out why ffi_go_closure is undefined, or otherwise fix that section of the file.

If this is relating somehow to the go port, then that doesn't build on PowerPC anyway...

[kencu (Ken)]

This commit <​https://github.com/libffi/libffi/commit/fa1040c111b3e423bc9c7e78d9af89470c0fa2fb#diff-8812a240988e12730090db5e550af6d2> is what we have to turn off somehow.

I can see us holding back libffi on PowerPC for a while to sort this out.

[ryandesign (Ryan Carsten Schmidt)]

Replying to kencu:

So we need to figure out why ffi_go_closure is undefined, or otherwise fix that section of the file.

ffi_go_closure should be defined in include/ffi.h.in, provided FFI_GO_CLOSURES is defined and truthy. FFI_GO_CLOSURES should be defined to 1 in src/powerpc/ffitarget.h, provided that POWERPC or POWERPC_FREEBSD are defined. I don't know where POWERPC or POWERPC_FREEBSD would get defined.

[kencu (Ken)]

Sadly not so easy -- I set it to "0" in the header, but there were yet more errors. Have to really fix it, unfortunately :>

That commit I referenced indicates there was a lot of work done there, and I doubt anyone has ever tried to build it with darwin 32bit PPC before now...

[ballapete (Peter "Pete" Dyballa)]

The same error happens on PPC Tiger, Mac OS X 10.4.11.

[TzunamiOSX]

I have installed gcc7 around 7 days ago without problems, but now there is the problem with libffi.

Im not sure this was needed for ffmpeg or gcc7 before.

[ryandesign (Ryan Carsten Schmidt)]

Replying to kencu:

Sadly not so easy -- I set it to "0" in the header, but there were yet more errors. Have to really fix it, unfortunately :>

You set what to "0"? I was suggesting that maybe POWERPC or POWERPC_FREEBSD needed to be defined.

[ryandesign (Ryan Carsten Schmidt)]

Replying to TzunamiOSX:

Im not sure this was needed for ffmpeg or gcc7 before.

Right, this problem is new in libffi 3.3.

[ballapete (Peter "Pete" Dyballa)]

The error is in libffi-3.3/powerpc-apple-darwin8.11.0/include/ffi.h. This C header file has a perfect type declaration:

  436	#if FFI_GO_CLOSURES
  437	
  438	typedef struct {
  439	  void      *tramp;
  440	  ffi_cif   *cif;
  441	  void     (*fun)(ffi_cif*,void*,void**,void*);
  442	} ffi_go_closure;
  443	
  444	FFI_API ffi_status ffi_prep_go_closure (ffi_go_closure*, ffi_cif *,
  445					void (*fun)(ffi_cif*,void*,void**,void*));
  446	
  447	FFI_API void ffi_call_go (ffi_cif *cif, void (*fn)(void), void *rvalue,
  448			  void **avalue, void *closure);
  449	
  450	#endif /* FFI_GO_CLOSURES */
  451	
  452	/* ---- Public interface definition -------------------------------------- */

It's just necessary to define FFI_GO_CLOSURES, I think…

[ballapete (Peter "Pete" Dyballa)]

Replying to ryandesign:

Replying to kencu:

So we need to figure out why ffi_go_closure is undefined, or otherwise fix that section of the file.

GCC always defines __POWERPC__ and similiar values, starting with two underscores and finishing with two underscores. So we should try to add

defined (POWERPC)

defined (POWERPC64)` to the line # 145:

#if defined (POWERPC) || defined (POWERPC_FREEBSD)

[ballapete (Peter "Pete" Dyballa)]

Replying to ballapete:

Replying to ryandesign:

Replying to kencu:

Last line of the configure script is:

cp ${srcdir}/src/$TARGETDIR/ffitarget.h include/ffitarget.h

The value of ${srcdir}/src/$TARGETDIR/ffitarget.h is ${worksrcpath}/src/powerpc. Here exists:

  -rw-r--r--  1 macports admin  6294 18. Nov 2019  ffitarget.h

So this file needs to be patched.

[ballapete (Peter "Pete" Dyballa)]

Patch file to add POWER and POWER64

[ballapete (Peter "Pete" Dyballa)]

New error:

libtool: link: ranlib .libs/libffi_convenience.a
libtool: link: ( cd ".libs" && rm -f "libffi_convenience.la" && ln -s "../libffi_convenience.la" "libffi_convenience.la" )
/bin/sh ./libtool  --tag=CC   --mode=link /opt/local/bin/gcc-apple-4.2  -pipe -Os -arch ppc -Wall -fexceptions -no-undefined -version-info `grep -v '^#' ../libtool-version`   '-L/opt/local/lib' '-Wl,-headerpad_max_install_names' '-arch' 'ppc'  -L/opt/local/lib -Wl,-headerpad_max_install_names -arch ppc -o libffi.la -rpath /opt/local/lib src/prep_cif.lo src/types.lo src/raw_api.lo src/java_raw_api.lo src/closures.lo  src/powerpc/ffi_darwin.lo src/powerpc/darwin.lo src/powerpc/darwin_closure.lo 
libtool: link: /opt/local/bin/gcc-apple-4.2 -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/powerpc/.libs/ffi_darwin.o src/powerpc/.libs/darwin.o src/powerpc/.libs/darwin_closure.o   -L/opt/local/lib  -Os -arch ppc -Wl,-headerpad_max_install_names -arch ppc -Wl,-headerpad_max_install_names -arch ppc   -install_name  /opt/local/lib/libffi.7.dylib -compatibility_version 9 -current_version 9.0 -Wl,-single_module
Undefined symbols:
  "_ffi_go_closure_ASM", referenced from:
      _ffi_go_closure_ASM$non_lazy_ptr in ffi_darwin.o
     (maybe you meant: _ffi_go_closure_ASM$non_lazy_ptr)
  "_ffi_call_go_AIX", referenced from:
      _ffi_call_go in ffi_darwin.o
  "_ffi_prep_cif_machdep_var", referenced from:
      _ffi_prep_cif_core in prep_cif.o
ld: symbol(s) not found
collect2: ld returned 1 exit status
make[3]: *** [libffi.la] Error 1
make[3]: Leaving directory `/opt/local/var/macports/build/_opt_local_var_macports_sources_nue.de.rsync.macports.org_macports_release_tarballs_ports_devel_libffi/libffi/work/libffi-3.3/powerpc-apple-darwin8.11.0'

[kencu (Ken)]

Replying to ryandesign:

Replying to kencu:

Sadly not so easy -- I set it to "0" in the header, but there were yet more errors. Have to really fix it, unfortunately :>

You set what to "0"? I was suggesting that maybe POWERPC or POWERPC_FREEBSD needed to be defined.

# define FFI_GO_CLOSURES 1

to

# define FFI_GO_CLOSURES 0

[kencu (Ken)]

For now, I have pegged libffi at 3.2.1 in my Tiger & Leopard PowerPC overlay repos.

[kencu (Ken)]

You can try this <​https://github.com/macports/macports-ports/pull/8379> which builds for me on 10.5.8 PowerPC and passes most of the tests.

For now I'm still pegging libffi at 3.2.1, but at least this PR gives us a starting point, perhaps, and does build through.

[ballapete (Peter "Pete" Dyballa)]

(Before testing Guile) Tiger is upgrading its software, Clang and llvm with old libffi, and port has decided to first upgrade some other ports (imlib2, python27, re2c, rsync, xterm)…

[kencu (Ken)]

Upstream PR: <​https://github.com/libffi/libffi/pull/583>

[kencu (Ken)]

In 5287b4e585c1277c33540e78cc8c8ef626dd776e/macports-ports (master):

libffi: fix darwin powerpc build

disable go bindings as they are not supported
on darwin powerpc at present

enable tests

closes: #61153

[ballapete (Peter "Pete" Dyballa)]

libffi @3.3_1 built on PPC Tiger with these test results:

# of unexpected failures	616
# of unresolved testcases	616
# of unsupported tests		30

[kencu (Ken)]

Yes -- you have to use gcc7 to test it, due to some unsupported warning flags that gcc-4.2 barfs on.

I just haven't bothered to fix the flags for gcc-4.2, as I tested it with gcc7. But if you feel like fixing the warning flags for gcc-4.2, I'll happily PR it.

[kencu (Ken)]

Here's the PPC test suite for 10.5 Leopard libffi version 3.2.1:

		=== libffi Summary ===

# of expected passes		1851
# of unexpected failures	12

and here it is for version 3.3

		=== libffi Summary ===

# of expected passes		1414
# of unexpected failures	17
# of unsupported tests		30