On Oct 2, 2013, at 14:02, Matthias Apitz <[email protected]> wrote: > After the unsuccessfull attempts in May/June I though it's time to give > it a try again to build the port x11/kde4 in > > $ uname -a > FreeBSD aurora.Sisis.de 10.0-ALPHA4 FreeBSD 10.0-ALPHA4 #0 r255948: Tue > Oct 1 09:00:53 CEST 2013 [email protected]:/usr/obj/usr/src/sys/GENERIC > i386 > > with /usr/ports as well from October 1st (r328930); it still fails, this > time in multimedia/ffmpeg1 (see attached nohup.out below); ... > Assertion failed: (N.getNode()->getNodeId() != NewNode && "Mapped to new > node!"), function RemapValue, file > /usr/src/lib/clang/libllvmselectiondag/../../../contrib/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp, > line 618. > 1 warning generated. > Stack dump: > 0. Program arguments: /usr/bin/cc -cc1 -triple i386-unknown-freebsd10.0 > -emit-obj -disable-free -main-file-name g723_1.c -mrelocation-model static > -relaxed-aliasing -masm-verbose -mconstructor-aliases -target-cpu pentium3 > -target-feature +sse -coverage-file > /usr/ports/multimedia/ffmpeg1/work/ffmpeg-1.2.3/libavcodec/g723_1.o > -resource-dir /usr/bin/../lib/clang/3.3 -dependency-file libavcodec/g723_1.d > -MT libavcodec/g723_1.o -D _ISOC99_SOURCE -D _FILE_OFFSET_BITS=64 -D > _LARGEFILE_SOURCE -D HAVE_AV_CONFIG_H -I . -I ./ -I /usr/local/include/vorbis > -I /usr/local/include -O3 -Wdeclaration-after-statement -Wall > -Wno-parentheses -Wno-switch -Wno-format-zero-length -Wdisabled-optimization > -Wpointer-arith -Wredundant-decls -Wno-pointer-sign -Wwrite-strings > -Wtype-limits -Wundef -Wmissing-prototypes -Wno-pointer-to-int-cast > -Wstrict-prototypes -Werror=implicit-function-declaration > -Werror=missing-prototypes -Werror=return-type -std=c99 -fconst-strings > -fdebug-compilation-dir /usr/ports/ > multimedia/ffmpeg1/work/ffmpeg-1.2.3 -ferror-limit 19 -fmessage-length 0 > -mstackrealign -fobjc-runtime=gnustep -fobjc-default-synthesize-properties > -fdiagnostics-show-option -backend-option -vectorize-loops -o > libavcodec/g723_1.o -x c libavcodec/g723_1.c
I have been able to reproduce this assertion, which only seems to occur with -march=pentium3. There is a fix from upstream, and I will probably commit it tomorrow. However, even with that fix, there is another problem in ac3dsp_init.c, which causes the inline assembler to run out of registers on i386. I am still looking into a fix for it. -Dimitry
signature.asc
Description: Message signed with OpenPGP using GPGMail
_______________________________________________ kde-freebsd mailing list [email protected] https://mail.kde.org/mailman/listinfo/kde-freebsd See also http://freebsd.kde.org/ for latest information
