...and it seems to be no workaround to this errormessage: >>> Test phase [test]: sys-fs/dazuko-2.3.2-r3 !! you must be root !! make: *** [test] Error 1 Portage 2.1.2_rc4-r5 (default-linux/x86/2006.1/desktop, gcc-4.1.1, glibc-2.5-r1, 2.6.19-gentoo-r2 i686) ================================================================= System uname: 2.6.19-gentoo-r2 i686 Intel(R) Pentium(R) 4 CPU 2.80GHz Gentoo Base System version 1.12.8 Last Sync: Wed, 03 Jan 2007 15:00:03 +0000 distcc 2.18.3 i686-pc-linux-gnu (protocols 1 and 2) (default port 3632) [enabled] ccache version 2.4 [enabled] dev-java/java-config: 1.3.7, 2.0.31 dev-lang/python: 2.5-r1 dev-python/pycrypto: 2.0.1-r5 dev-util/ccache: 2.4-r6 sys-apps/sandbox: 1.2.18.1 sys-devel/autoconf: 2.13, 2.61 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.17.50.0.8 sys-devel/gcc-config: 1.3.14 sys-devel/libtool: 1.5.22 virtual/os-headers: 2.6.19 ACCEPT_KEYWORDS="x86 ~x86" AUTOCLEAN="yes" CBUILD="i686-pc-linux-gnu" CFLAGS="-march=pentium4 -pipe -O2 -ggdb" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/share/X11/xkb" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/revdep-rebuild /etc/splash /etc/terminfo /etc/texmf/web2c" CXXFLAGS="-march=pentium4 -pipe -O2 -ggdb -fvisibility-inlines-hidden" DISTDIR="/usr/portage/distfiles" EMERGE_DEFAULT_OPTS=" " FEATURES="autoconfig ccache distcc distlocks parallel-fetch sandbox sfperms splitdebug strict test userfetch userpriv usersandbox" GENTOO_MIRRORS="ftp://ftp.csbnet.se/pub/linux/distributions/gentoo/ http://trumpetti.atm.tut.fi/gentoo/ http://pandemonium.tiscali.de/pub/gentoo/ http://mirror.pudas.net/gentoo" LANG="sv" LC_ALL="sv_SE.UTF-8" LDFLAGS="-Wl,-O1 -Wl,--as-needed -Wl,-znow -Wl,--hash-style=gnu" LINGUAS="sv" MAKEOPTS="-j5 -s" PKGDIR="/usr/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/overlays/portage /usr/local/overlays/nouveau /usr/portage/local/layman/musicbrainz /usr/portage/local/layman/nxsty /usr/portage/local/layman/gnome-experimental /usr/portage/local/layman/mozilla /usr/portage/local/layman/gentopia" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="X a52 aac acpi aiglx alsa asf audioscrobbler avahi bash bash-completion beagle berkdb bitmap-fonts branding bzip2 cairo cdr cli cracklib crypt cups dbus debug divx dlloader dmx dpms dri dvd dvdr eds eiglx emboss encode evolution fam fat ffmpeg firefox flac fortran freetype gd gdbm gif gimp glitz gmp gnome gnutls gpm gstreamer gtk gtk2 gtkhtml hal howl-compat iconv icu inotify ipv6 irmc isdnlog ithreads java jikes joystick jpeg lcms ldap libg++ libnotify libsexy lm_sensors logrotate mad matroska matrox mikmod mmx mng mono moznocompose moznoirc moznomail mp3 mpeg musepack musicbrainz nautilus ncurses network nls nptl nptlonly ntfs ntp numeric offensive ogg opengl pam pam_console pcre pdf perl pic png ppds pppd print pulseaudio python quicktime readline real reflection reiserfs rtc samba sdl sensord session smp sox spell spf spl sse sse2 ssh ssl startup-notification subtitles svg syslog tcltk tcpd test theora threads tiff truetype truetype-fonts tth type1-fonts udev unicode usb userlocales utf8 vorbis win32codecs wma wmf wv wxwindows x86 xcb xinerama xinetd xml xml2 xorg xosd xprint xulrunner xv xvid zlib" ALSA_CARDS="emu10k1 intel8x0" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" ELIBC="glibc" INPUT_DEVICES="evdev" KERNEL="linux" LINGUAS="sv" USERLAND="GNU" VIDEO_CARDS="none nvidia nouveau" Unset: CTARGET, INSTALL_MASK, PORTAGE_RSYNC_EXTRA_OPTS
Well... It is right... You cannot test kernel module if you are not root... What behavior do you expect? I can remove the test support... But I guess this is not the only package that require root for testing... And people may want to test...
...or make it possible to have portage make some elements as root when FEATURES="userpriv" is enabled? That would make this into a thing for the portage-team...
Hello Portage team, I will appreciate any thoughts on this one. I think that if user specify userpriv, he should be ensured that testing cannot damage his system... And if the user specified test, and the test cannot be conducted, then we should fail the test procedure, not ignore the issue. So actually what I am saying is that I am happy with current implementation... Thanks!
(In reply to comment #3) > Hello Portage team, > I will appreciate any thoughts on this one. > > I think that if user specify userpriv, he should be ensured that testing cannot > damage his system... > And if the user specified test, and the test cannot be conducted, then we > should fail the test procedure, not ignore the issue. > > So actually what I am saying is that I am happy with current implementation... > > Thanks! > We have a few options; one of course is not depriving during the test phase. We can add a finer grained RESTRICT to restrict to certain phases. I don't like either of these to be honest, but kernel modules are an exception to a normal build routine (where you can generally build and install as a non-root user). What if I'm building on SELINUX and I'm not in the correct security context to load kernel modules (=failure) so I also agree with the current behavior. Is there a way to patch the tests to run as a normal user (ie are there non-root tests we can perform, and just exclude the others?)
Thanks for the reply! (In reply to comment #4) > Is there a way to patch the tests to run as a normal user (ie are there > non-root tests we can perform, and just exclude the others?) Not in this one... kernel module must be loaded in order to succeed in any test... And performing these test may cause system to be unusable... So I am quite happy that test fails and not switch to root automatically...
Then maybe just restrict test if userpriv and leave a message telling no tests was ran due to missing permissons?
(In reply to comment #6) > Then maybe just restrict test if userpriv and leave a message telling no tests > was ran due to missing permissons? The problem is that there is currently no way to make RESTRICT="test" conditional on FEATURES="userpriv". One possible solution would be to put FEATURES into USE_EXPAND so that we could have RESTRICT="features_userpriv? ( test )" but that would require some changes to portage internals and some people are opposed to the general idea of having FEATURES extend into the USE conditional space (although it would be useful for this bug and in other cases like bug #69021).
How about something like having both src_test() and src_test_root_only() in the ebuild API, and portage having the logic to decide when each one should be invoked?
As workaround you could do a if ! has userpriv $FEATURES; then # perform root only tests else elog "Some tests skipped due to userpriv restrictions" fi Still not really comfortable with it (ebuilds shouldn't care about FEATURES) but works without any portage changes. How many packages share this problem anyway? If it's just a few it should be handled via an eclass IMHO.
It means that the test will not be preformed... I don't think ewarn is enough... But it can be simpler... No need to look-up features... if [ $(id -u) = 0 ]; then ...
(In reply to comment #10) > It means that the test will not be preformed... I don't think ewarn is > enough... > > But it can be simpler... No need to look-up features... > > if [ $(id -u) = 0 ]; then ... > +1, except try if [[ $EUID = 0 ]] ; then.. and save yourself a fork.
(In reply to comment #10) > It means that the test will not be preformed... I don't think ewarn is > enough... Well, replace ewarn with whatever you want. Point was that unless this is a common problem it the solution should be in the tree, not in portage. Closing this for now, reopen if it turns out to be a common problem (as in affects several dozen packages or so). > But it can be simpler... No need to look-up features... > > if [ $(id -u) = 0 ]; then ... That should work as well, yep.
Well this "bug" is not really fixed, so I think closing is wrong, better to reassign it back to maintainer of dazuko. If none of the fixes mentioned here does it right maybe go another way. If you take for example sys-devel/gcc it passes the test phase with a comment like "some test failed and it sucks" but it does not error out but merges the package. Maybe do the same for dazuko but give a message that the fails could be becouse of you where running it with userpriv? Apparently this is something that is not an error in the tests (as it needs root) and it is not an error in portage (as userpriv forces user portage as much as possible as it should) and I think it is wrong to RESTRICT the test but also wrong to force users to disable test/userpriv for just this package (and it is very annoying if you do a massive 'emerge world -uDN' for example.
OK. Added ewarn if not root on test stage (dazuko).
Maybe a "test" USE-flag in this case could be nice. The ebuild could behave like it now does when the flag is disabled and it could die when "test" is enabled. So the dying could be controlled by /ets/portage/*.
Thats not what the test flag is for. Better as it is now, and for the people it matters for they have the reasons in their ELOG.
Note that FEATURES=test now implies USE=test for portage-2.1.2 (see bug #69021). Also note the relevant discussions happening on the gentoo-dev mailing list: http://thread.gmane.org/gmane.linux.gentoo.devel/45241 http://thread.gmane.org/gmane.linux.gentoo.devel/45354
