Any version of gnome-vfs-2.2.x creates a sandbox error when doing make install. Reproducible: Always Steps to Reproduce: 1. emerge =gnome-vfs-2.2.1 2. emerge =gnome-vfs-2.2.2 3. emerge =gnome-vfs-2.2.3 Actual Results: >>> Completed installing into /var/tmp/portage/gnome-vfs-2.2.1/image/ --------------------------- ACCESS VIOLATION SUMMARY --------------------------- LOG FILE = "/tmp/sandbox-gnome-vfs-2.2.1-26487.log" mkdir: /root/.gconfd -------------------------------------------------------------------------------- Expected Results: Installed without a sandbox violation System uname: 2.4.20-sparc-r3 sparc64 sun4u GENTOO_MIRRORS="http://gentoo.oregonstate.edu/ http://www.ibiblio.org/pub/Linux/distributions/gentoo" CONFIG_PROTECT="/etc /var/qmail/control /usr/kde/2/share/config /usr/kde/3/share/config /usr/X11R6/lib/X11/xkb /usr/kde/3.1/share/config /usr/share/config" CONFIG_PROTECT_MASK="/etc/gconf /etc/env.d" PORTDIR="/usr/portage" DISTDIR="/usr/portage/distfiles" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR_OVERLAY="/usr/local/portage" USE="sparc" COMPILER="gcc3" CHOST="sparc-unknown-linux-gnu" CFLAGS="-mcpu=ultrasparc -O3 -pipe" CXXFLAGS="-O2 -pipe" ACCEPT_KEYWORDS="sparc ~sparc" MAKEOPTS="-j2" AUTOCLEAN="yes" SYNC="rsync://rsync.gentoo.org/gentoo-portage" FEATURES="sandbox buildpkg ccache distcc userpriv usersandbox noclean noauto cvs keeptemp"
and if you emerge without the usersandbox and userpriv features ?
That seems to fix the problem. Any idea why those tripped up portage? I was emerging things as root.
hmm no, should be investigated ofcourse and we will :) but that was the only thing that was obviously different from how i emerge stuff.
*** Bug 18237 has been marked as a duplicate of this bug. ***
I also get a sandbox violation with the gnome-vfs-2.2.3 ebuild. I have tried with and without the userpriv and usersandbox features, but the result is the same. Based on the discussion of bug 18237, I manually created /root/.gconfd. This eliminated my problem.
was gconfd running at the time you got the error ?
I didn't have gconfd nor any GNOME or X components running at the time
*** Bug 21278 has been marked as a duplicate of this bug. ***
okay, i remembered to keep the original build error this time, before i fixed it. as such the original error message is: if test -z "" ; then \ for p in system_http_proxy.schemas ; do \ GCONF_CONFIG_SOURCE=xml::/etc/gconf/gconf.xml.defaults /usr/bin/gconftool-2 --makefile-install-rule ./$p; \ done \ fi GCONF_DISABLE_MAKEFILE_SCHEMA_INSTALL is set, not installing schemas ACCESS DENIED mkdir: /root/.gconfd Failed to create /root/.gconfd: Permission denied which is being called from schemas/Makefile There's no reason to execute gconftool-2 here, because the gnome2 eclass does this for us in the function gnome2_gconf_install( ). There's a simple configure flag to fix this as well, all we need is this, --disable-schemas-install, which if you look in schemas/Makefile.in, you'll see disables the above part of the Makefile. I havent had time to investigate yet, but this might also be relevant to bug 18237 and bug 21278.
Created attachment 13370 [details] an updated ebuild, just a one line diff the diff is basically. # okay we need to tell gnome-vfs to disable installation of # the gconf schemas, because we'll do it later ourselves G2CONF="${G2CONF} --disable-schemas-install"
sounds good to me .. maybe there are other packages that respect this flag? because i'm sure gnome-vfs isn't the only package that has the /root/.gconfd problem
bug 18237 also has these symptoms, and although configure describes the --disable-schema-install flag in that package (gnome-utils), it isn't actually implemented anywhere in the Makefile.in's. so while i'd hope it might be the case, it turns out there are inconsistencies across the packages. foser seemed to think there was more to this than just simply disabling gconftool-2 from running at inopportune times, more it was a sandbox, or usersandbox related bug. can either liquidx/foser comment on whether that'll be followed up? or should this be considered the close? thanks.
problem persists in 2.2.5
and if you apply the changes mentioned in #10 to the ebuild ?
i've applied to changes to gnome-vfs-2.2.[45]. would someone who had the problem before try and see if its fixed?
i assume it does