|
Description
r3lgar
2015-11-30 21:53:41 UTC
Created attachment 418248 [details]
media-libs/harfbuzz-1.0.6 log
Created attachment 418250 [details]
app-text/poppler-0.38.0 log
# emerge -pv media-libs/harfbuzz app-text/poppler
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild N ] media-libs/harfbuzz-1.0.6:0/0.9.18::gentoo USE="cairo fontconfig glib graphite icu introspection truetype -static-libs {-test}" ABI_X86="32 (64) (-x32)" 0 KiB
[ebuild N ] app-text/poppler-0.38.0:0/57::gentoo USE="cairo curl cxx introspection jpeg jpeg2k lcms png qt4 tiff utils -cjk -debug -doc (-qt5)" 0 KiB
i don't see what sandbox has to do with it. packages shouldn't access /root/. (In reply to SpanKY from comment #4) You're right. FEATURES=-sandbox gives same result. Created attachment 418390 [details]
/var/log/sandbox/sandbox-29794.log
Created attachment 418392 [details]
harfbuzz sandbox log
Created attachment 418394 [details]
poppler sandbox log
Could you paste the output of: $ getent passwd portage I guess something is using /root as its home while building and gobject-introspection uses /root/.cache to cache some of its work which results in this sandbox error. To solve this, the ebuilds can use xdg.eclass that should properly setup the environment to a temporary directory under portage control. (In reply to Gilles Dartiguelongue from comment #9) >getent passwd portage portage:x:250:250:portage:/var/tmp/portage:/bin/false I tried reproducing this problem using sudo su - or su - to get root access but without success. I don't see what's causing this. In both cases this is caused by the g-ir-scanner... :/ does that have global configuration files where things might go wrong? [ 97%] [34m[1mGenerating Poppler-0.18.gir[0m [ 98%] [34m[1mGenerating Poppler-0.18.gir[0m cd /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib && /usr/bin/g-ir-scanner "--add-include-path=/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib --warn-all" --namespace=Poppler --nsversion=0.18 --no-libtool --library=poppler-glib --include=GObject-2.0 --include=Gio-2.0 --include=cairo-1.0 --pkg-export=poppler-glib --c-include poppler.h -I/usr/include -I/usr/include -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0 -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/fofi -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/goo -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/poppler -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/poppler -I/usr/include/freetype2 -I/usr/include -I/usr/include -I/usr/include -I/usr/include -I/usr/include -I/usr/include/openjpeg-2.1 -I/usr/include -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -I/usr/include/cairo /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-action.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-date.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-document.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-page.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-attachment.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-form-field.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-annot.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-layer.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-movie.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-media.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-cached-file-loader.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-input-stream.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-structure-element.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-action.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-date.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-document.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-page.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-attachment.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-form-field.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-annot.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-layer.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-movie.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-media.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-structure-element.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/poppler-enums.c /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/poppler-enums.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/poppler-features.h --output /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/Poppler-0.18.gir cd /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib && /usr/bin/g-ir-scanner "--add-include-path=/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib --warn-all" --namespace=Poppler --nsversion=0.18 --no-libtool --library=poppler-glib --include=GObject-2.0 --include=Gio-2.0 --include=cairo-1.0 --pkg-export=poppler-glib --c-include poppler.h -I/usr/include -I/usr/include -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0 -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/fofi -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/goo -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/poppler -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/poppler -I/usr/include/freetype2 -I/usr/include -I/usr/include -I/usr/include -I/usr/include -I/usr/include -I/usr/include/openjpeg-2.1 -I/usr/include -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib -I/var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -I/usr/include/cairo /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-action.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-date.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-document.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-page.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-attachment.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-form-field.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-annot.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-layer.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-movie.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-media.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-cached-file-loader.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-input-stream.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-structure-element.cc /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-action.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-date.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-document.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-page.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-attachment.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-form-field.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-annot.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-layer.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-movie.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-media.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0/glib/poppler-structure-element.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/poppler-enums.c /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/poppler-enums.h /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/poppler-features.h --output /var/tmp/portage/app-text/poppler-0.38.0/work/poppler-0.38.0_build/glib/Poppler-0.18.gir * ACCESS DENIED: mkdir: /root/.cache * ACCESS DENIED: mkdir: /root/.cache Afaik, there is no configuration for g-ir-scanner. Could you dump the ebuild environment and attach it ? I have a similar error. I am assigning this to gnome as maintainer of gobject-introspection because I believe that to be the source of the issue. If you require more specific information from the reporter(s), then please specify exactly what from where. (which ebuild, where to find "environment", etc) All instruction are provided by portage when the build stops. Example of failing libgdiplus on my box: ^[[31;01m*^[[0m ERROR: dev-dotnet/libgdiplus-3.12::gentoo failed (compile phase): ^[[31;01m*^[[0m emake failed ^[[31;01m*^[[0m ^[[31;01m*^[[0m If you need support, post the output of `emerge --info '=dev-dotnet/libgdiplus-3.12::gentoo'`, ^[[31;01m*^[[0m the complete build log and the output of `emerge -pqv '=dev-dotnet/libgdiplus-3.12::gentoo'`. ^[[31;01m*^[[0m The complete build log is located at '/var/tmp/portage/portage/dev-dotnet/libgdiplus-3.12/temp/build.log'. ^[[31;01m*^[[0m The ebuild environment file is located at '/var/tmp/portage/portage/dev-dotnet/libgdiplus-3.12/temp/environment'. ^[[31;01m*^[[0m Working directory: '/var/tmp/portage/portage/dev-dotnet/libgdiplus-3.12/work/libgdiplus-3.12' ^[[31;01m*^[[0m S: '/var/tmp/portage/portage/dev-dotnet/libgdiplus-3.12/work/libgdiplus-3.12' (In reply to Gilles Dartiguelongue from comment #16) What do you mean? harfbuzz failing with this: * * The following package has failed to build, install, or execute postinst: * * (media-libs/harfbuzz-1.1.3:0/0.9.18::gentoo, ebuild scheduled for merge), Log file: * '/var/log/portage/pkg/build/media-libs/harfbuzz-1.1.3:20160115-065842.log' * I had the same error unset XDG_CACHE_HOME fixed the problem. I think this is somehow related: https://bugs.gentoo.org/show_bug.cgi?id=444568 Then, until this is covered at EAPI level it will be a problem to track all potential packages that could break with this :S (another reason for fixing bug 499288 some decade... but I am too pessimistic on that) (In reply to Pacho Ramos from comment #19) >track all potential packages that could break with this Some of them: media-libs/gexiv2-0.10.3 media-libs/gstreamer-1.6.2 x11-libs/gdk-pixbuf-2.32.3 x11-libs/libnotify-0.7.6-r1 media-libs/harfbuzz-1.1.3 gnome-base/librsvg-2.40.13 media-libs/gegl-0.3.4 Potentially all that binds to GObject Introspection. I think so. *** Bug 574772 has been marked as a duplicate of this bug. *** This looks to be caused by people running emerge with a "polluted" XDG_CACHE_HOME var (per duplicate bug it could be caused, for example, when people becomes root using "su" without the - and, then, getting the var set to their old home. I cannot run test now, but it seems that this could be reproduce if we set XDG_CACHE_HOME to a wrong value on purpose to detect what packages need to inherit xdg.eclass for cleaning their environments. If that is the case, I would try to ask Toralf to run a tinderbox with that for helping us to fix that packages and make them inherit the eclass. I also wonder if PMS team would finally accept to, at least, introduce the concrete vars that are being cleaned in xdg.eclass for eapi7 for, at least, cover them in the future :| (In reply to Pacho Ramos from comment #22) > This looks to be caused by people running emerge with a "polluted" > XDG_CACHE_HOME var (per duplicate bug it could be caused, for example, > when people becomes root using "su" without the - and, then, getting > the var set to their old home. XDG_CACHE_HOME is explicitly intended to be set by the user, and this issue occurs if it is set to anything at all, not just a "polluted" or "wrong" value inherited from the user env. I see that gobject-introspection-1.46.0 was just stabilized, so this problem just got that much more widespread. At least the following recently stabilized packages all fail now: - harfbuzz - libosinfo - libvirt-glib The problem does not occur with the previous stable version of gobject-introspection (1.44.0). Looking at the code of xdg-utils.eclass, the easiest way to fix this should be * inherit xdg-utils (has no phases) * call xdg_environment_reset at the start of src_configure a6a5345 x11-libs/libnotify: Clean up XDG environment, bug 567192 e1652c1 x11-libs/gdk-pixbuf: Clean up XDG environment, bug 567192 a8bd058 media-libs/gstreamer: Clean up XDG environment, bug 567192 08e2306 media-libs/gexiv2: Clean up XDG environment, bug 567192 be703cb app-emulation/libvirt-glib: Clean up XDG environment, bug 567192 8d4f779 sys-libs/libosinfo: Clean up XDG environment, bug 567192 1ce49ec app-text/poppler: Clean up XDG environment, bug 567192 7e3be4b media-libs/harfbuzz: Clean up XDG environment, bug 567192 ... always fixed in newest ~arch or in newly created ~arch revbump (plus live ebuild if it exists). Yes I know we need this in stable too, but let's wait first whether anything explodes. (In reply to Andreas K. Hüttel from comment #25) I do not understand how packages using gnome2.eclass (like gdk-pixbuf) were failing. They are already cleaning the enbvironment as gnome2.eclass does it using xdg.eclass :/ Was the breakage on that concrete packages really tested or was only following the advise of people talking about potential broken packages? I also wonder about revbumping for this build time change :| (In reply to Nick Bowler from comment #23) > XDG_CACHE_HOME is explicitly intended to be set by the user, and this > issue occurs if it is set to anything at all, not just a "polluted" or > "wrong" value inherited from the user env. If you keep intentionally changing the defaults of XDG_ (and not only XDG_) vars to random values you will get many random problems. That is the reason for trying to become root with su - instead of simply root because, as soon as package managers keep preserving most of the environment variables, build of multiple packages are prone to randomly fail as soon as they start to honor that variables that are set to uncontrolled values (that is a bit more developed in bug 499288) (In reply to Pacho Ramos from comment #26) > (In reply to Andreas K. Hüttel from comment #25) > > I do not understand how packages using gnome2.eclass (like gdk-pixbuf) were > failing. They are already cleaning the enbvironment as gnome2.eclass does it > using xdg.eclass :/ hmm interesting :/ feel free to revert > I also wonder about revbumping for this build time change :| well, I didnt want to change stable ebuilds in-place where I wasnt familiar with them. we can always just forward-port stable keywords from -r0 to -r1 once we know the change doesnt hurt. (In reply to Pacho Ramos from comment #27) > for trying to become root with su - instead of simply root because simply su I meant (In reply to Andreas K. Hüttel from comment #28) > hmm interesting :/ feel free to revert Bleh, I do not have access to my dev machine now, but I will try when I am able to (In reply to Andreas K. Hüttel from comment #25) [...] All this commits reviewed finally @Toralf, we would need to get an amd64 stable tinderbox run running emerge with something like: XDG_CACHE_HOME=/home/pacho/tmp/ emerge -1 ... Being the directory a folder in your real filesystem to see if it tries to write there. If that is the case, you will get a sandbox violation error and you will be able to make that bugs block this tracker. Thanks a lot (In reply to Pacho Ramos from comment #30) is /tmp a sufficient FS too or are the perms too weak there (rw for all) ? If I don't misremember, /tmp can be written and sandbox won't die As a rule of thumb, always use xdg.eclass, xdg-utils is really for cases where maintainer does not want any phase changes. xdg.eclass is meant to cover XDG_* variables and all other xdg spec that would need to be taken care of, now or in the future (desktop and mime files for now). I will add that info somewhere in the eclasses. *** Bug 576478 has been marked as a duplicate of this bug. *** *** Bug 579346 has been marked as a duplicate of this bug. *** as per IRC EAPI 7 solves this entirely: [21:51] <willikins> toralf, you have notes! [Apr-22 05:39] <mgorny> please either stop those XDG* runs or limit them to EAPI 7. there's no purpose in having people hack around this (rare) stuff when EAPI bump will usually solve it |
