Since emerging this gcc I have some packages failing with ICE. One example: numeric.c: In function `S_mulexp10': numeric.c:815: internal compiler error: in try_split, at emit-rtl.c:3343 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://bugs.gentoo.org/> for instructions. Preprocessed source stored into /var/tmp/portage/libperl-5.8.6/temp/ccZigoZr.out file, please attach this to your bugreport. make: *** [numeric.o] Error 1 !!! ERROR: sys-devel/libperl-5.8.6 failed. I've also heard of similar problems from other users :-( Reproducible: Always Steps to Reproduce: 1.emerge gcc-3.4.3-20050110 2.emerge -e world 3. OMGWTFBBQ!!! It is teh b0rken! Portage 2.0.51-r13 (default-linux/x86/2004.3, gcc-3.4.3, glibc-2.3.4.20041102-r0, 2.6.10 i686) ================================================================= System uname: 2.6.10 i686 AMD Athlon(tm) XP 2600+ Gentoo Base System version 1.6.8 Python: dev-lang/python-2.3.4 [2.3.4 (#1, Jan 18 2005, 17:26:06)] ccache version 2.3 [enabled] dev-lang/python: 2.3.4 sys-devel/autoconf: 2.59-r6, 2.13 sys-devel/automake: 1.8.5-r2, 1.5, 1.4_p6, 1.6.3, 1.7.9, 1.9.4 sys-devel/binutils: 2.15.92.0.2-r2 sys-devel/libtool: 1.5.10-r2 virtual/os-headers: 2.6.8.1-r2 ACCEPT_KEYWORDS="x86 ~x86" AUTOCLEAN="yes" CFLAGS="-O2 -march=athlon-xp -fomit-frame-pointer" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/X11R6/lib/X11/xkb /usr/kde/2/share/config /usr/kde/3.3/env /usr/kde/3.3/share/config /usr/kde/3.3/shutdown /usr/kde/3/share/config /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=athlon-xp -fomit-frame-pointer" DISTDIR="/usr/portage/distfiles" FEATURES="autoaddcvs autoconfig ccache cvs distlocks sandbox sfperms" GENTOO_MIRRORS="ftp://sunsite.informatik.rwth-aachen.de/pub/Linux/gentoo http://gentoo.inode.at/ ftp://gentoo.inode.at/source/" LDFLAGS="-Wl,-O1" MAKEOPTS="-j2" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" SYNC="rsync://rsync.europe.gentoo.org/gentoo-portage" USE="x86 3dnow 3dnowex X aac aalib acpi alsa apache2 apm arts audiofile avi bash-completion berkdb bidi bigger-fonts bitmap-fonts bzlib cdr chroot cjk cross crypt css cups curl curlwrappers dbm dga directfb divx4linux doc dvd dvdr dvdread edl encode erandom esd ethereal exif f77 faac faad flac flexresp flood font-server foomaticdb fortran freetype ftp gd gdbm gif gimp glep gmp gmttria gpm hardened hardenedphp iconv icq imagemagick imlib inifile ipv6 jabber java jpeg jpeg2k junit kde ldap libcaca libg++ libwww live lzo mad makecheck mikmod mime mmx mmx2 mng monkey motif mozsvg mozxmlterm mp3 mpeg mpeg4 mpi mplayer mysql mysqli nagios-dns nagios-ntp nagios-ping nagios-ssh ncurses network nls no-old-linux nptl nptlonly ntlm nvidia offensive oggvorbis openal opengl oss pam parse-clocks pcre pdf pdflib pear-db perl pie png prelude print python qemu-fast qt quicktime readline real recode rtc samba sdl skey slang slp smime sndfile sockets spell sse ssl ssp svg svga tcltk tcpd tcpmd5 tetex tga theora threads tidy tiff truetype truetype-fonts type1 type1-fonts usb vhosts xml xml2 xmlrpc xmms xprint xv xvid xvmc yv12 zeo zlib" Unset: PORTDIR_OVERLAY
Created attachment 48944 [details] What the error message said to attach
I've masked bogus GCC this again at pappy's request. Can someone look into backporting the bugs that this revision was aimed to solve to our current gcc in a new -r? This would appear much more sane than a CVS snapshot. If not, at least we can go for better QA on the snapshot before unmasking. Thanks guys. (Closing this for now)
we did give the snapshot a bit of testing with some users before putting it into ~arch ... they reported it solved the sse* issues and didnt have any reports of any other problems ... obviously the snapshot does have issues, but there's no reason to go around assuming that a shitty QA job was done in terms of getting it into ~arch we look into fixing up the sse* code another way and try again, not a big deal
I am sorry and have to excuse for Brandon, who wrote that on behalf of me and did follow my advise to get the ebuild out of the wild. If you have problems with me doing that or you have got the impression that we made a "big deal" about it and tried to blame eradicator or you for "messing with the QA", which is as far as i am concerned a presumably high goal of you and your team, you are free to remove the related comments by tseng or me in this bug report. Let me restate the facts: The ebuild is broken, bonsaikitten told me to do something, i pulled the brake and now things are set for you continuing putting CVS gcc's into ~arch. My team and i will do my best and stop broken CVS gcc versions entering portage in the future again. Thank you for your cooperation and sorry for disturbing you, Alex
*** Bug 78718 has been marked as a duplicate of this bug. ***
tseng: FWIW, most of the gcc ebuilds ARE snapshots. I just changed the gcc versioning scheme to reflect the snapshot in ${PV} rather than have it set by the BRANCH_UPDATE variable in the ebuild. QA was done on this version including a complete bootstrap -> emerge system -> emerge gnome kde on amd64 by me. gcc even rebuilt itself when I did it, and others verified for me that it was working great for them and fixed the sse2/sse3 issue. I'll look into backporting just the sse2/sse3 fixes, but I think the best route will be to report this regression upstream and wait for their fix as it will be much easier to deal with on that route.
Jeremy: thanks for the clarification, I didnt mean to offend anyone, but clearly did. pappy and bonsaikitten were looking for someone to mask this asap, and somehow the buck stopped at me. Comments made here may have been a bit harsher than the should have been due to the apperant urgency of the situation as it was relayed to me. Appologies for any mal effects of my hasty action.
I can't cause it to ICE here... gcc -mno-sse2 -fstack-protector -fstack-protector-all -march=athlon-xp -O2 -Wall -fno-strict-aliasing -fomit-frame-pointer -fPIC -frandom-seed=0 -c test.c Does this file always cause it to ICE for you?
*** Bug 78772 has been marked as a duplicate of this bug. ***
I also cannot reproduce it with the attachment and the command line from the file: /usr/libexec/gcc/i686-pc-linux-gnu/3.4.3/cc1 -fpreprocessed test.c -mno-sse2 -fstack-protector -fstack-protector-all -quiet -dumpbase numeric.tmp.localhost.15213.i -march=athlon-xp -auxbase-strip /root/.ccache/tmp.hash.localhost.15213.o -O2 -Wall -fno-strict-aliasing -fomit-frame-pointer -fPIC -o - -frandom-seed=0 Seems to work fine here.
Created attachment 49226 [details] Preprocessed source that caused the internal compiler error.
Sorry, attached to wrong bug.
emerge -v sys-devel/libperl-5.8.6 work fine on amd64 + gcc 3.4.3-20050110