Bugzilla Bug 114745

cc -O3 -march=pentium4 -mmmx -msse -msse2 -msse3 -pipe   -c -o util_dp_sim.o
util_dp_sim.c
util_dp_sim.c: In function `sim_pair_wise_lalign':
util_dp_sim.c:1016: internal compiler error: in simplify_gen_subreg, at
simplify-rtx.c:3558
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/t-coffee-3.18/temp/ccFN9EEB.out
file, please attach this to your bugreport.
make: *** [util_dp_sim.o] Error 1

!!! ERROR: sci-biology/t-coffee-3.18 failed.
!!! Function src_compile, Line 31, Exitcode 2


!!! ERROR: sci-biology/t-coffee-3.18 failed.

# emerge info
Portage 2.0.53 (default-linux/x86/2005.1, gcc-3.4.4, glibc-2.3.5-r3, 2.6.13.2 i686)
=================================================================
System uname: 2.6.13.2 i686 Intel(R) Pentium(R) 4 CPU 3.00GHz
Gentoo Base System version 1.12.0_pre11
dev-lang/python:     2.3.5, 2.4.2
sys-apps/sandbox:    1.2.17
sys-devel/autoconf:  2.13, 2.59-r7
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-r1
sys-devel/libtool:   1.5.20-r1
virtual/os-headers:  2.6.11-r3
ACCEPT_KEYWORDS="x86 ~x86"
AUTOCLEAN="yes"
CBUILD="i686-pc-linux-gnu"
CFLAGS="-O3 -march=pentium4 -mmmx -msse -msse2 -msse3 -pipe"
CHOST="i686-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3/share/config
/usr/lib/X11/xkb /usr/lib/mozilla/defaults/pref /usr/share/config /usr/spool/PBS
/var/qmail/control"
CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/texmf/web2c /etc/env.d"
CXXFLAGS="-O2 -mcpu=i686 -pipe"
DISTDIR="/usr/portage/distfiles"
FEATURES="autoconfig distlocks sandbox sfperms strict"
GENTOO_MIRRORS="ftp://ftp.muni.cz/pub/linux/gentoo"
PKGDIR="/usr/portage/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="x86 FFmpeg X Xaw3d aalib acpi alsa apache2 apm avi berkdb bidi bitmap-fonts
bzip2 caca cdr crypt cscope curl dba divx divx4 divx4linux divx5 divx5linux dvd
dvdr dvdread eds emacs emacs-w3 emboss encode exif expat f77 faad faad2 fam fame
ffmpeg flash foomaticdb fortran fvwm fvwm2 gb gd gdbm ggi gif glut gpm gstreamer
gtk gtk2 gtkhtml i8x0 icc imagemagick imlib imlib2 innodb ipv6 ithreads java
javascript jpeg lcms leim libg++ libwww live lzo mad mcal mesa mikmod mmx mmx2
mng motif mozilla mp3 mpeg mule mysql ncurses network nls nptl ogg opengl oss
pam pda pdflib perl plotutils png ppds pthread pthreads python qt qtx quicktime
readline rtc samba sdl slang slp spell sse sse2 sse3 ssl tcltk tcpd tetex theora
thread threads tiff truetype truetype-fonts type1-fonts udev unicode usb v4l
v4l2 vorbis win32 winvidix wmf xml xml2 xmms xosd xv xvid xvmc zeo zlib
video_cards_radeon userland_GNU kernel_linux elibc_glibc"
Unset:  ASFLAGS, CTARGET, LANG, LC_ALL, LDFLAGS, LINGUAS, MAKEOPTS, PORTDIR_OVERLAY

#

Using -O2 does not help.

Created an attachment (id=74230) [edit]
ccFN9EEB.out

Thanks for reporting. Using -O2 will not solve the problem, but -O0 actually    
will. (I have not benchmarked how slow t-coffee actually is when not optimised  
at all, but it might be a problem when extracting the aligned sequences and    
iterating the alignment multiple times.)    

Since this is a reproducable GCC 3.4 specific compiler error, I am CCing the    
toolchain team. It is not specific to version 3.18. (The latest version,   
3.27, is also affected.)   

I have added a warning message to the ebuild for version 3.27. I checked   
different CFLAGS combinations, and it seems users of the stable version (2.66)   
are not affected at all. Unfortunately, this version is now masked since the  
archive may no longer be downloaded. Curse restricting downloading for grant 
purposes. :\ 

Yeah, I had to use CFLAGS="" emerge -u t-coffee

:(

This bug should be really filed to gcc guys. The preprocessed file should make
them able to fix it in a moment.

(In reply to comment #2)
> Thanks for reporting. Using -O2 will not solve the problem, but -O0 actually    
> will.  

Please note that "-O1" also works fine. "-O1 -funit-at-a-time" fails however.

This is a problem in gcc-3.4 all the way up to mainline.  Putting to
reso->upstream until we get a patch.

ribosome, can you make it strip all -O[123] flags in the meanwhile? It's a bit
annoying to just have the compile be broken.

Anything new on this? I am still getting this problem with gcc-3.4.5 but not
with gcc-3.3.6. Maybe running temporarily gcc-config(1) would be a better
temporary fix for this.

ribosome - I suggest adding -fno-unit-at-a-time with gcc-3.4.x; e.g. adding:

[[ $(gcc-version) == "3.4" ]] && append-flags -fno-unit-at-a-time

in src_compile() prior to the make.  This gets it to compile at -O2 for me; you
may also want to try -O3 and -Os as they're popular.  Since I'm not a biologist
I have no idea what the application does so can't check that it actually works
8-)

(In reply to comment #9)
> ribosome - I suggest adding -fno-unit-at-a-time with gcc-3.4.x; e.g. adding:
> 
> [[ $(gcc-version) == "3.4" ]] && append-flags -fno-unit-at-a-time
> 
> in src_compile() prior to the make.  This gets it to compile at -O2 for me; you
> may also want to try -O3 and -Os as they're popular.  Since I'm not a biologist
> I have no idea what the application does so can't check that it actually works
> 8-)
> 

Added to the ebuild. Thanks for the suggestion.