Learn more about these different git repos.
Other Git URLs
I just face a weird crash during ekiga and evolution packages build in koji, where the i686 build crashes, while the x86_64 build succeeds fine. I do not know why it is happening, but it is reproducible.
The ekiga error [1][2] is: {{{ GCONF_CONFIG_SOURCE=xml:merged:/etc/gconf/gconf.xml.defaults gconftool-2 --makefile-install-rule ekiga.schemas >/dev/null 2>&1 /bin/sh: line 1: 19686 Segmentation fault (core dumped) GCONF_CONFIG_SOURCE=xml:merged:/etc/gconf/gconf.xml.defaults gconftool-2 --makefile-install-rule ekiga.schemas > /dev/null 2>&1 }}} The evolution error [3][4] is: {{{ DOC Introspecting gobjects sh: line 1: 4159 Segmentation fault (core dumped) ( /usr/bin/libtool --mode=execute ./evolution-mail-formatter-scan ) Scan failed: Makefile:739: recipe for target 'scan-build.stamp' failed }}} I cannot build dependencies of the evolution due to this error. The packages are: evolution-ews, evolution-mapi, evolution-rss and evolution-rspam. All these four packages have updated their .spec files, only fedpkg build is needed to be called once the evolution package in the rawhide's build root.
fedpkg build
[1] http://koji.fedoraproject.org/koji/taskinfo?taskID=14934565 \ [2] working x86_64 build: http://koji.fedoraproject.org/koji/taskinfo?taskID=14934500 \ [3] http://koji.fedoraproject.org/koji/taskinfo?taskID=14934439 \ [4] working x86_64 build: http://koji.fedoraproject.org/koji/taskinfo?taskID=14934582
filing a bug @ bugzilla.redhat.com against the most likely component would be the better course of action here (imo, this most likely isn't a releng issue).
And, unfortunately, due to lack of a verbose build in one of the examples: ... DOC Introspecting gobjects ... , doesn't give a good hint precisely what is crashing yet.
I thought I'll fill this against koji, which is under releng hands, at least as I understand it. I agree that the issue can be elsewhere, like in some of the packages related in the build, but I'm also clueless.
The code in the evolution crashed when the gtk-doc runs its own auto-generated code, thus it can be in the gtk-doc or somewhere around.
I filled a bug against GConf2, even the same version had been used a month ago, where the build worked without any issue: https://bugzilla.redhat.com/show_bug.cgi?id=1357600
and against gtk-doc, similarly the same version as the one used a month ago, which worked just fine: https://bugzilla.redhat.com/show_bug.cgi?id=1357606
I saw a report in #fedora-devel this morning that http://koji.fedoraproject.org/koji/taskinfo?taskID=14931936 failed appstream-util segfaulted on 32 bit x86 also. something in teh package set changed. that broke 32 bit builds for at least something. filing a releng ticket is not really the way to go about getting it fixed.
I commented in the 2 referenced bz tickets one culprit could be a new glib2 build that landed yesterday.
Okay, thanks. It would be useful if the backtraces of the crash could be taken from the builds, thus it might be easier to get a clue. There probably isn't installed ABRT or anything like that, is it?
Metadata Update from @mcrha: - Issue set to the milestone: Fedora 25 Alpha
Login to comment on this ticket.