Hi,
a quick summary of what's yet to be done in the ports you listed.
Mark Kettenis <mark.kette...@xs4all.nl> writes:
> Hi Folks,
>
> robert@ has been so kind to do a bulk build with binutils 2.17 on
> amd64, sent me a list of ports that failed to build and gave me access
> to the build logs. I've analyzed most of these now. I'm looking into
> the harder ones myself, but below is a list of ports with "trivial"
> issues. These basically fall into two categories:
>
> 1. Ports that link by invoking ld directly:
>
> lang/tinyscheme
> net/honeyd
> security/ADMfzap
I propose to kill it. Weird use case, clunky licensing and build system,
probably already broken, hardcodes path to libc.so.23.1, etc etc
ok?
> www/mongrel2
> x11/pinot
Fixed recently, but depends on graphics/exiv2 which is broken too. Are you
already looking at it?
> audio/swh-plugins (-nostartfiles)
I'm not sure how to test the resulting package after such a horrible
diff.
Index: Makefile
===================================================================
RCS file: /cvs/ports/audio/swh-plugins/Makefile,v
retrieving revision 1.9
diff -u -p -r1.9 Makefile
--- Makefile 1 Jun 2013 12:15:48 -0000 1.9
+++ Makefile 21 May 2015 11:05:34 -0000
@@ -37,6 +37,11 @@ CONFIGURE_ARGS += --disable-nls
AUTO_ENV = AUTOCONF_VERSION=${AUTOCONF_VERSION} \
AUTOMAKE_VERSION=${AUTOMAKE_VERSION}
+post-extract:
+ find ${WRKSRC} -type f -name '*.c' -print -execdir perl -pi \
+ -e 's/^void\s+_init\(\)\s+{/void __init()
__attribute__((constructor));\nvoid __init() {/;' \
+ -e 's/^void\s+_fini\(\)\s+{/void __fini()
__attribute__((destructor));\nvoid __fini() {/;' \
+ {} \;
post-patch:
rm ${WRKSRC}/acconfig.h ${WRKSRC}/ladspa.h
cd ${WRKSRC}; ${AUTO_ENV} aclocal -I . -I m4
Index: patches/patch-Makefile_am
===================================================================
RCS file: /cvs/ports/audio/swh-plugins/patches/patch-Makefile_am,v
retrieving revision 1.2
diff -u -p -r1.2 patch-Makefile_am
--- patches/patch-Makefile_am 1 Jun 2013 12:15:48 -0000 1.2
+++ patches/patch-Makefile_am 21 May 2015 11:11:38 -0000
@@ -1,15 +1,16 @@
$OpenBSD: patch-Makefile_am,v 1.2 2013/06/01 12:15:48 ajacoutot Exp $
+- don't use -nostartfiles: relocation R_X86_64_PC32 against `__guard_local' ...
- use "libtool archives" (PIC) instead of static archives
---- Makefile.am.orig Tue Aug 8 08:56:35 2006
-+++ Makefile.am Fri Oct 22 11:45:47 2010
+--- Makefile.am.orig Tue Aug 8 17:56:35 2006
++++ Makefile.am Thu May 21 13:06:06 2015
@@ -43,15 +43,15 @@ plugindir = @prefix@/lib/ladspa
# Uncomment below for Darwin support. Or add a conditional for this.
#AM_CFLAGS = -fno-common -flat_namespace -bundle -undefined suppress
-lbundle1.o
-AM_LDFLAGS = -module -avoid-version -Wc,-nostartfiles
-+AM_LDFLAGS = -module -avoid-version -nostartfiles
++AM_LDFLAGS = -module -avoid-version
DESTDIR = $(INSTALL_ROOT)
DISTFN = $(distdir)
> These ports dont include /usr/lib/crtbeginS.o and therefore run into
> an issue referencing __guard_local, resulting in messages like:
>
> ld: foobar.o: relocation R_X86_64_PC32 against `__guard_local' can
> not be used when making a shared object; recompile with -fPIC
>
> The solution here is to patch the port to link by invoking cc.
>
> 2. Ports that attempt to link code compiled without -fPIC into a
> shared library:
>
> devel/valgrind
This one looks complicated. :)
cc'd Masao in case he was not aware.
> textproc/mupdf
> www/vteplugin
> x11/cool-retro-term
> x11/kmplayer
>
> ld: foobar.o: relocation R_X86_64_PC32 against `foobar' can not be
> used when making a shared object; recompile with -fPIC
>
> The solution here is to make sure the relevant code gets compiled with -fPIC.
>
> I've already fixed a couple of important instances of both categories.
> I invite maintainers/porters to fix the others on the list, or declare
> them low-priority. espie@ already indicated that x11/kmplayer isn't a
> priority.
>
> Thanks,
>
> Mark
--
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
