svn commit: r475256 - head/audio/deadbeef
Alexey Dokuchaev
danfe at FreeBSD.org
Tue Jul 24 14:00:51 UTC 2018
Author: danfe
Date: Tue Jul 24 14:00:50 2018
New Revision: 475256
URL: https://svnweb.freebsd.org/changeset/ports/475256
Log:
Tentatively try to unbreak the build on arm* architectures. There were
two kinds of problems:
1) Sloppy handling of 8-bit data and assuming that `char' is signed by
default in AdPlug and Commodore 64 SID plugins code;
2) Incorrect detection of SSE2 support. Contrary to GCC, Clang always
accepts `-msse2' command line option, even on !x86; it just tells that
it won't be used. As a result, SSE2 code was compiled whilst it could
not be, causing the build to fail.
While here, remove no longer relevant patching of the configure script.
Modified:
head/audio/deadbeef/Makefile
Modified: head/audio/deadbeef/Makefile
==============================================================================
--- head/audio/deadbeef/Makefile Tue Jul 24 13:57:46 2018 (r475255)
+++ head/audio/deadbeef/Makefile Tue Jul 24 14:00:50 2018 (r475256)
@@ -220,9 +220,16 @@ post-patch:
${WRKSRC}/plugins/supereq/Makefile.in
@${REINPLACE_CMD} -e 's,/etc/timidity++/timidity-freepats.cfg:/etc/timidity/freepats.cfg:/etc/timidity/,${LOCALBASE}/share/,' \
${WRKSRC}/plugins/wildmidi/wildmidiplug.c
- @${REINPLACE_CMD} \
- -e '/HAVE_OGG" = "yes"/ {n; n; /HAVE_VORBISPLUGIN=yes/d;}' \
+# Clang accepts `-msse2' command line option even on !x86, need better check
+ @${REINPLACE_CMD} -e '25177s,^,#include <xmmintrin.h>,' \
${WRKSRC}/configure
+ @${REINPLACE_CMD} -e '/chnresolv/s,const char,signed &,' \
+ ${WRKSRC}/plugins/adplug/adplug/s3m.cpp \
+ ${WRKSRC}/plugins/adplug/adplug/s3m.h
+ @${REINPLACE_CMD} -e "/sampleConvertTable/s,int8_t,u&, ; \
+ 86,+1 { s,',,g; s,\\\\x,0x,g; }" \
+ ${WRKSRC}/plugins/sid/sidplay-libs/libsidplay/src/xsid/xsid.cpp \
+ ${WRKSRC}/plugins/sid/sidplay-libs/libsidplay/src/xsid/xsid.h
# Fix build with libzip-1.0
@${REINPLACE_CMD} 's,zip_file_t,zip_file_DB,g' \
${WRKSRC}/plugins/vfs_zip/vfs_zip.c
More information about the svn-ports-head
mailing list