GCC 3.4.4-r1 just went stable, but it looks like the singular_value_decomposition function in dev-python/numeric breaks with this compiler. The attached test script illustrates the problem. It should run to completion, but hangs in singular_value_decomposition. I have extracted this from a program that worked a few days ago, but after rebuilding everything with the new gcc it hangs. My CFLAGS are CFLAGS="-O2 -march=pentium4 -fomit-frame-pointer -pipe" Reproducible: Always Steps to Reproduce: 1. Run the attached Python script 2. 3. Actual Results: It hangs. demokrit ~ # gcc-config -c i686-pc-linux-gnu-3.4.4 demokrit ~ # etcat -v gcc [ Results for search key : gcc ] [ Candidate applications found : 17 ] Only printing found installed programs. * sys-devel/gcc : [M ] 2.95.3-r9 (2.95) [M ] 3.1.1-r2 (3.1) [M ] 3.2.2 (3.2) [M ] 3.2.3-r4 (3.2) [M ] 3.3.2-r7 (3.3) [ ] 3.3.5-r1 (3.3) [ ] 3.3.5.20050130-r1 (3.3) [M~ ] 3.3.5.20050130-r2 (3.3) [ I] 3.3.6 (3.3) [M~ ] 3.4.1-r3 (3.4) [M~ ] 3.4.3-r1 (3.4) [M~ ] 3.4.3.20050110-r2 (3.4) [ I] 3.4.4-r1 (3.4) [M ] 3.4.5 (3.4) [M ] 4.0.1 (4.0) [M ] 4.0.2-r1 (4.0) [M ] 4.1.0_beta20051202 (4.1) demokrit ~ # emerge info Portage 2.0.51.22-r3 (default-linux/x86/2005.0, gcc-3.4.4, glibc-2.3.5-r2, 2.6.14-gentoo-r2 i686) ================================================================= System uname: 2.6.14-gentoo-r2 i686 Mobile Intel(R) Pentium(R) 4 - M CPU 2.00GHz Gentoo Base System version 1.6.13 dev-lang/python: 2.3.5-r2, 2.4.2 sys-apps/sandbox: 1.2.12 sys-devel/autoconf: 2.13, 2.59-r6 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r1 sys-devel/binutils: 2.16.1 sys-devel/libtool: 1.5.20 virtual/os-headers: 2.6.11-r2 ACCEPT_KEYWORDS="x86" AUTOCLEAN="yes" CBUILD="i686-pc-linux-gnu" CFLAGS="-O2 -march=pentium4 -fomit-frame-pointer -pipe" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3.4/env /usr/kde/3.4/share/config /usr/kde/3.4/shutdown /usr/kde/3/share/config /usr/lib/X11/xkb /usr/lib/mozilla/defaults/pref /usr/share/config /usr/share/texmf/dvipdfm/config/ /usr/share/texmf/dvips/config/ /usr/share/texmf/tex/generic/config/ /usr/share/texmf/tex/platex/config/ /usr/share/texmf/xdvi/ /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/env.d" CXXFLAGS="-O2 -march=pentium4 -fomit-frame-pointer -pipe" DISTDIR="/var/portage/distfiles" FEATURES="autoconfig distlocks fixpackages sandbox sfperms strict" GENTOO_MIRRORS="http://trumpetti.atm.tut.fi/gentoo/ ftp://trumpetti.atm.tut.fi/gentoo/ ftp://ftp.rhnet.is/pub/gentoo/ ftp://gd.tuwien.ac.at/opsys/linux/gentoo/" LINGUAS="dk" MAKEOPTS="-j2" PKGDIR="/var/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/home/schiotz/development/CamposEbuilds/portage" SYNC="rsync://rsync.europe.gentoo.org/gentoo-portage" USE="x86 X Xaw3d acpi alsa atlas audiofile avi berkdb bitmap-fonts bzip2 cdr crypt cups curl dga dvd eds emacs emboss encode exif expat fam ffmpeg flac font-server foomaticdb fortran gd gdbm gif glut gpm gstreamer gtk gtk2 idn imagemagick imap imlib java jpeg kde kdeenablefinal kdexdeltas lcms libg++ libwww mad maildir mbox mikmod mime mng motif mozilla mp3 mpeg mule ncurses netcdf nls nptl ogg oggvorbis opengl oss pam pcmcia pcre pdflib perl png pnp ppds python qt quicktime readline real scanner sdl smime spell sse sse2 ssl tcltk tcpd tetex tiff truetype truetype-fonts trusted type1-fonts udev usb vorbis win32codecs wxwindows xine xml2 xmms xv xvid zlib linguas_dk userland_GNU kernel_linux elibc_glibc" Unset: ASFLAGS, CTARGET, LANG, LC_ALL, LDFLAGS
Created attachment 74315 [details] Test script.
I did some more testing. The fomit-frame-pointer flag has no effect, it still hangs when I remove it. But recompiling numeric with i686-pc-linux-gnu-3.3.6 fixes the problem. I will keep the old compiler around just for the purpose of compiling numeric. This bug is assigned to the python herd, but it is of course possible that it is a GCC problem. /Jakob
Other solution would be to fix numeric ebuilds to use syatem lapack for LinearAlgebra module. It has been discussed (http://bugs.gentoo.org/show_bug.cgi?id=81520) but unfortunately it seems that no ebuilds were added to portage as a result. I have numeric-23.8 compiled with gcc-3.4.4, linked with lapack-atlas-3.6.0 and no problems with your script.
(In reply to comment #3) > Other solution would be to fix numeric ebuilds to use syatem lapack for > LinearAlgebra module. I did an ebuild for numeric-24.2 that links with blas/lapack, in the Gentoo Scientific overlay: http://gentooscience.org.
I agree with comment #4. I just wanted to add, that the bug is also present when compiling with GCC 4.1.1. /Jakob
fixed in >=dev-python/numeric-24.2-r1, now in cvs. Thanks for testing!