| Summary: | portage can't find cut | ||
|---|---|---|---|
| Product: | Portage Development | Reporter: | John Richard Moser <nigelenki> |
| Component: | Sandbox | Assignee: | Portage team <dev-portage> |
| Status: | VERIFIED NEEDINFO | ||
| Severity: | normal | CC: | base-system, hardened, manuel |
| Priority: | High | ||
| Version: | 2.0 | ||
| Hardware: | All | ||
| OS: | All | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Bug Depends on: | |||
| Bug Blocks: | 81031 | ||
| Attachments: |
my /etc/profile
Environment that shows the problem with "cut" under the sandbox. |
||
|
Description
John Richard Moser
2004-12-17 19:03:20 UTC
which version of baselayout do you have installed. 1.11.8 the binaries for cut are in /bin; there's a symlink to ../../bin/cut in /usr/bin/cut what version of coreutils ? does `cut --help` work ? Perhaps this needs clarification. my shell can find cut. I can run cut. if I run emerge, it complains it can't find cut, even though I just ran it successfully. Picture something like this: --cut here-- #!/bin/sh # foo.sh ls --cut here-- Imagine this script working like this: $ ls foo.sh $ ./foo.sh foo.sh: line 3: ls: command not found This is the scenario going on with portage here. cut is in my path, portage is probably clobbering the path. well, you still havent said WHERE your cut is located what does `which cut` show icebox ~ # which cut /usr/bin/cut cut is /bin/cut and /usr/bin/cut -> /bin/cut (for some odd reason). It works if I ctrl+Z emerge and try 'cut' or 'which cut' icebox ~ # cut cut: you must specify a list of bytes, characters, or fields Try `cut --help' for more information. icebox ~ # echo "1:2" | cut -d : -f 2 2 I added "echo $PATH" above line 730 in functions.sh
icebox ~ # echo $PATH
/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/bin:/usr/x86_64-pc-linux-gnu/gcc-bin/3.4.3:/usr/X11R6/bin:/opt/blackdown-jdk-1.4.2.01/bin:/opt/blackdown-jdk-1.4.2.01/jre/bin
icebox ~ # emerge portage
Calculating dependencies ...done!
>>> emerge (1 of 1) sys-apps/portage-2.0.51-r13 to /
/opt/bin:/usr/x86_64-pc-linux-gnu/gcc-bin/3.4.3:/usr/X11R6/bin:/opt/blackdown-jdk-1.4.2.01/bin:/opt/blackdown-jdk-1.4.2.01/jre/bin:/usr/games/bin:/opt/insight/bin
There's my path, and the path portage knows.
portage guys: any idea ? Hacking of /etc/profile? Created attachment 50031 [details]
my /etc/profile
hacking of /etc/profile ?
set -x at the top of ebuild.sh:
icebox ~ # emerge libsafe
Calculating dependencies ...done!
>>> emerge (1 of 1) sys-libs/libsafe-2.0_p16-r1 to /
+ export SANDBOX_PREDICT=:/proc/self/maps:/dev/console:/usr/lib/portage/pym:/dev/random
+ SANDBOX_PREDICT=:/proc/self/maps:/dev/console:/usr/lib/portage/pym:/dev/random
+ export SANDBOX_WRITE=:/dev/shm:/var/tmp
+ SANDBOX_WRITE=:/dev/shm:/var/tmp
+ export SANDBOX_READ=:/dev/shm:/var/tmp
+ SANDBOX_READ=:/dev/shm:/var/tmp
+ '[' '!' -z '' ']'
+ '[' clean '!=' depend ']'
+ '[' clean '!=' clean ']'
+ '[' -n 1 ']'
+ ARGS=clean
+ declare -rx EBUILD_PHASE=clean
+ unalias -a
+ unset GZIP BZIP BZIP2 CDPATH GREP_OPTIONS GREP_COLOR GLOBIGNORE
+ shopt -s expand_aliases
+ alias 'die=diefunc "$FUNCNAME" "$LINENO" "$?"'
+ alias 'assert=_pipestatus="${PIPESTATUS[*]}"; [[ "${_pipestatus// /}" -eq 0 ]] || diefunc "$FUNCNAME" "$LINENO"
"$_pipestatus"'
+ alias 'save_IFS=[ "${IFS:-unset}" != "unset" ] && old_IFS="${IFS}"'
+ alias 'restore_IFS=if [ "${old_IFS:-unset}" != "unset" ]; then IFS="${old_IFS}"; unset old_IFS; else unset IFS;
fi'
+ OCC=
+ OCXX=
+ '[' GNU == GNU ']'
+ source /etc/profile.env
+ '[' -f /etc/portage/bashrc ']'
+ source /etc/portage/bashrc
++ alias usev=useq
++ '[' /usr/lib/portage/bin/ebuild.sh = /usr/lib/portage/bin/ebuild.sh -o /usr/lib/portage/bin/ebuild.sh = /usr/lib/portage/bin/ebuild-daemon.sh ']'
++ source /sbin/functions.sh
+++ RC_GOT_FUNCTIONS=yes
+++ SVCDIR=/var/lib/supervise
+++ svcdir=/var/lib/init.d
+++ svclib=/lib/rcscripts
+++ svcmount=no
+++ svcfstype=tmpfs
+++ svcsize=1024
+++ deptypes='need use'
+++ ordtypes='before after'
+++ RC_QUIET_STDOUT=no
+++ RC_VERBOSE=no
+++ RC_NOCOLOR=no
+++ RC_ENDCOL=yes
+++ RC_TTY_NUMBER=11
+++ RC_NET_STRICT_CHECKING=no
+++ RC_PARALLEL_STARTUP=no
+++ RC_USE_CONFIG_PROFILE=yes
+++ RC_INDENTATION=
+++ RC_DEFAULT_INDENT=3
+++ RC_DOT_PATTERN=
+++ '[' -f /etc/conf.d/rc ']'
+++ source /etc/conf.d/rc
++++ RC_TTY_NUMBER=11
++++ RC_PARALLEL_STARTUP=yes
++++ RC_NET_STRICT_CHECKING=no
++++ RC_USE_FSTAB=no
++++ RC_DEVICES=auto
++++ RC_DEVICE_TARBALL=no
++++ RC_RETRY_KILL=yes
++++ RC_RETRY_TIMEOUT=1
++++ RC_RETRY_COUNT=5
++++ RC_FAIL_ON_ZOMBIE=no
++++ svcdir=/var/lib/init.d
++++ svcmount=yes
++++ svcfstype=tmpfs
++++ svcsize=4096
+++ '[' -f /sbin/splash-functions.sh ']'
+++ source /sbin/splash-functions.sh
+++ NET_FS_LIST='afs cifs coda ncpfs nfs nfs4 shfs smbfs'
+++ '[' -z /usr/portage/sys-libs/libsafe/libsafe-2.0_p16-r1.ebuild ']'
+++ [[ clean != *depend* ]]
++++ portageq envvar NOCOLOR
+++ RC_NOCOLOR=
+++ '[' '' = true ']'
+++ COLS=0
+++ (( COLS == 0 ))
++++ stty size
++++ cut '-d ' -f2
/sbin/functions.sh: line 730: cut: command not found
+++ COLS=
Try executing /bin/cut. I'm thinking you'll get "command not found". Then try running file /bin/cut and see if anything looks unusual. Whether it does or doesn't, if you do get "command not found" grab a binpkg of coreutils and merge that. Any ideas on this guys? could you try making sure your env is sane ? # env -i $SHELL # source /etc/profile # emerge blah Portage 2.0.51-r15 (default-linux/x86/2004.3, gcc-3.4.3, glibc-2.3.4.20041102-r0, 2.6.10-gentoo-r6 i686) ================================================================= System uname: 2.6.10-gentoo-r6 i686 AMD Athlon(tm) XP 1800+ Gentoo Base System version 1.6.9 Python: dev-lang/python-2.3.4-r1 [2.3.4 (#1, Feb 9 2005, 21:19:48)] ccache version 2.3 [enabled] dev-lang/python: 2.3.4-r1 sys-devel/autoconf: 2.13, 2.59-r6 sys-devel/automake: 1.5, 1.7.9-r1, 1.9.4, 1.6.3, 1.8.5-r3, 1.4_p6 sys-devel/binutils: 2.15.92.0.2-r2 sys-devel/libtool: 1.5.10-r4 virtual/os-headers: 2.6.8.1-r2 ACCEPT_KEYWORDS="x86 ~x86" AUTOCLEAN="yes" CFLAGS="-Os -march=athlon-xp -pipe -fomit-frame-pointer" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3/share/config /usr/lib/X11/xkb /usr/share/config /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/env.d" CXXFLAGS="-Os -march=athlon-xp -pipe -fomit-frame-pointer" DISTDIR="/usr/portage/distfiles" FEATURES="autoaddcvs autoconfig ccache distlocks sandbox sfperms" GENTOO_MIRRORS="ftp://ftp.wh2.tu-dresden.de/pub/mirrors/gentoo ftp://ftp-stud.fht-esslingen.de/pub/Mirrors/gentoo/ ftp://212.219.56.152/sites/www.ibiblio.org/gentoo/ http://mirrors.sec.informatik.tu-darmstadt.de/gentoo/" LANG="it_IT.utf8" MAKEOPTS="-j2" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="x86 3dnow 3dnowex X aalib alsa apm avi bash-completion berkdb bitmap-fonts cdr crypt cups encode f77 fam flac font-server foomaticdb fortran gdbm gif gpm gtk gtk2 guile imagemagick imlib java jpeg libg++ libwww mad mikmod mmx mmx2 motif mp3 mpeg ncurses nls nptl oggvorbis opengl pdflib perl png ppds python qt quicktime readline scanner sdl slang spell sse ssl svga tcltk tcpd tiff truetype truetype-fonts type1-fonts unicode usb xml xml2 xv zlib" Unset: ASFLAGS, CBUILD, CTARGET, LC_ALL, LDFLAGS i often get this when portage is emerging something: /usr/portage/eclass/gcc.eclass: line 31: [: : integer expression expected if [ "$(${CC} -dumpversion | cut -f1 -d.)" -ne 3 ] && [ "${WANT_GCC_3}" = "yes" ]
^^^ line 31
It looks like cut is failing in this case too.
Is everybody that is having problems using LANG="XX.UTF-8"? *** Bug 85949 has been marked as a duplicate of this bug. *** In my case (see bug 85949) I'm using LANG=en_US.UTF-8. Also, I found that if set up an evironment with the same variables that are set when the ebuild is running, cut executes but returns an empty string. With either LANG, LD_PRELOAD, or SANDBOX_ACTIVE unset, cut returns the right value: manuel@ulthar ~ $ . /tmp/ebuild.env manuel@ulthar MPlayer-1.0pre6a $ echo 3.3.5 | cut -f1,2 -d. manuel@ulthar MPlayer-1.0pre6a $ unset SANDBOX_ACTIVE manuel@ulthar MPlayer-1.0pre6a $ echo 3.3.5 | cut -f1,2 -d. 3.3 (/tmp/ebuild.env sets the environment that matches the one in the failed ebuild - I have attached it to the bug). Created attachment 53912 [details]
Environment that shows the problem with "cut" under the sandbox.
Can we raise the priority/severity on this one? It struck again, this time while emerging firefox - due to the "cut" failure it failed to detect a valid version of pango, thus screwing up all my fonts. At the moment, the sandbox is useless to me - I have to turn it off to be able to trust any of my emerges. I've had some friends try this out - it seems that if glibc and coreutils are built without the nls USE flag this is not a problem. It seems like the NLS stuff doesn't run well under the sandbox. This may be a GCC bug. If I build coreutils with -O2 instead of -Os, then the problem does not occur. Either compiler bug or bad code in "cut" that's tweaked by optimizations. Just to confirm the situation I tried to emerge mplayer. Version mplayer-1.0_pre5-r5 was chosen as the last stable by emerge It failed with the following: * Applying mplayer-1.0_pre5-gcc_detection.patch ... [ ok ]>>> Source unpacked. * No LINGUAS given, defaulting to English Detected operating system: Linux Detected host architecture: i386 Checking for cc version ... v. ?.??, bad Checking for gcc version ... v. ?.??, bad Checking for gcc-3.3 version ... v. ?.??, bad Checking for gcc-3.2 version ... v. ?.??, bad Checking for gcc-3.1 version ... v. ?.??, bad Checking for gcc3 version ... v. ?.??, bad Checking for gcc-3.0 version ... v. ?.??, bad Checking for cc version ... v. ?.??, bad *** Please downgrade/upgrade C compiler to version gcc-2.95.x or gcc-3.x! *** Gcc is of correct version: dwarf / # gcc --version gcc (GCC) 3.3.5 (Gentoo Linux 3.3.5-r1, ssp-3.3.2-3, pie-8.7.7.1) LANG was set to: dwarf / # env |grep LANG LANG=en_US.UTF-8 After unsetting LANG, mplayer is emerging fine (went past the configure part already) My system has been built from stage 1 with the following CFLAGS and USE: CFLAGS="-Os -march=pentium3 -pipe" USE="-oss -arts -fortran -f77 -kde -gnome -qt -motif -svga dvd alsa esd apm bluetooth cdparanoia dvdread hal maildir offensive pcmcia truetype vorbis win32codecs wifi xvid ffmpeg unicode" /bin/cut does exist and can be run: dwarf portage # /bin/cut /bin/cut: you must specify a list of bytes, characters, or fields Note the following however. Though USE includes "unicode", I only added it after reading the Gentoo-UTF8-howto. At that time, I was already at stage 3 and not willing to go back and rebuild everything from scratch (pentium3, 600MHz machine). Following the howto, I did: emerge --oneshot --verbose --ask sys-libs/ncurses sys-libs/slang revdep-rebuild --soname libncurses.so.5 revdep-rebuild --soname libslang.so.1 Funny thing is, when I run gcc manually (normal shell environment), the version number is readable and cut seems to work: dwarf portage # env |grep LANG LANG=en_US.UTF-8 dwarf portage # gcc --version gcc (GCC) 3.3.5 (Gentoo Linux 3.3.5-r1, ssp-3.3.2-3, pie-8.7.7.1) dwarf portage # echo 3.3.5 | cut -f1,2 -d. 3.3 Still an issue with the new sandbox? Apparently, this is not a portage bug. sandbox? gcc? coreutils? Rebuild coreutils with sandbox-1.2.12 installed. If it is still an issue, please reopen. Closing due to old age |
