glibc-2.27.9000 and struct statx_timestamp
by Sandro Mani
Hi
I'm hitting the following with qt5-qtbase:
/usr/include/linux/stat.h:56:8: error: redefinition of 'struct statx_timestamp'
struct statx_timestamp {
^~~~~~~~~~~~~~~
In file included from /usr/include/sys/stat.h:446,
from /builddir/build/BUILD/qtbase-everywhere-src-5.11.1/mkspecs/linux-g++/qplatformdefs.h:75,
from /builddir/build/BUILD/qtbase-everywhere-src-5.11.1/src/corelib/io/qfilesystemengine_unix.cpp:42:
/usr/include/bits/statx.h:25:8: note: previous definition of 'struct statx_timestamp'
struct statx_timestamp
^~~~~~~~~~~~~~~
I.e. struct statx_timestamp in linux/stat.h (included directly by qfilesystemengine_unix.cpp) is conflicting with struct statx_timestamp in bits/statx.h included via sys/stat.h.
This has begun with glibc-2.27.9000.
Any ideas?
Thanks
Sandro
5 years, 10 months
F29 Mass rebuild - Help needed to fix failed build
by Zoltan Kota
Hi All,
The Mass rebuild of pybliographer failed. See below:
----
Fwd: releng's pybliographer-1.2.18-4.fc29 failed to build
Notification time stamped 2018-07-15 03:04:40 UTC
releng's pybliographer-1.2.18-4.fc29 failed to build
http://koji.fedoraproject.org/koji/buildinfo?buildID=1121364
----
It seems the .configure script does not find 'python'. The script uses
Python variable to store python path (configure.ac: AC_PATH_PROG(Python,
python, no) ).
Can I add/override this variable in the spec file?
eg. %configure Python=%{__python2}
or what is the correct syntax/way to fix the issue?
Thanks,
Zoltan
5 years, 10 months
Lots a permission denied activity
by Steve Grubb
Hello,
I have been testing a new set of audit rules and have run across some
processes that are doing things that might out to be changed. Typically,
audit users expect a normally functioning system to not be noisy. There is a
requirement to audit failed file access due to permission denied. What I'm
finding is that two processes are generating tens of thousands of events
every day.
There is a /usr/libexec/tracker-extract process that searches my directories
about every 11 seconds. I can imagine on a laptop that would be a lot of disk
activity. Sometimes I use root in my home directory and accidentally create
files owned by root. This leads to a lots of events on my system. Does it
really need to run with this frequency?
But I also see one that I just don't understand. Every 12 seconds, /usr/lib/
systemd/systemd calls openat with write flags to open
/sys/fs/cgroup/cpu/cgroup.procs
/sys/fs/cgroup/cpuacct/cgroup.procs
/sys/fs/cgroup/blkio/cgroup.procs
/sys/fs/cgroup/memory/user.slice/user-4325.slice/user(a)4325.service/
cgroup.procs
/sys/fs/cgroup/memory/user.slice/user-4325.slice/cgroup.procs
/sys/fs/cgroup/memory/user.slice/cgroup.procs
/sys/fs/cgroup/memory/cgroup.procs
/sys/fs/cgroup/devices/user.slice/cgroup.procs
/sys/fs/cgroup/devices/cgroup.procs
/sys/fs/cgroup/pids/user.slice/user-4325.slice/user(a)4325.service/cgroup.procs
/sys/fs/cgroup/pids/user.slice/user-4325.slice/cgroup.procs
/sys/fs/cgroup/pids/user.slice/cgroup.procs
/sys/fs/cgroup/pids/cgroup.procs
Which are all root owned files. This adds up to about 45,000 events a day. Is
there a purpose to opening those files? And if that was truly needed, should
it be logging failures? Are the permissions wrong? If the failures are
benign, why is it doing it at all?
Thanks,
-Steve
5 years, 10 months
GCC 8/9 ABI change: call for rebuilds
by Marek Polacek
Recently a serious bug in the compiler was discovered whereby we miscompiled
several packages. The problem only occurs in C++ programs that contain a class
with a trivial move constructor and deleted copy constructor. For such
programs, the calling convention changed inappropriately, in the sense that an
object was passed via memory rather than via registers. See
<https://gcc.gnu.org/PR86094>.
I did a scratch mass rebuild with a specially-tweaked gcc in order to find out
which packages need to be rebuilt. Since the problem occurred in both GCC 8
and GCC 9, we will need to rebuild packages in F28 and F29:
ceph-12.2.4-1.fc29.src.rpm
chromium-67.0.3396.79-1.fc29.src.rpm
firefox-60.0.1-6.fc29.src.rpm
gdb-8.1.50.20180605-18.fc29.src.rpm
insight-8.1.50.20180216-2.fc29.src.rpm
mame-0.198-1.fc29.src.rpm
v8-6.7.17-4.fc29.src.rpm
polymake-3.2r3-1.fc29.src.rpm
pybind11-2.2.2-3.fc29.src.rpm
qt5-qtwebengine-5.10.1-7.fc29.src.rpm
cpprest-2.10.2-2.fc28.src.rpm
waylandpp-0.2.3-1.fc29.src.rpm
webkit2gtk3-2.21.3-1.fc29.src.rpm
webrtc-audio-processing-0.3-7.fc29.src.rpm
The problem is fixed in gcc-8.1.1-2.fc28 and gcc-8.1.1-2.fc29. But it seems
the F28 buildroots still have the unpatched version:
$ koji latest-pkg f28-build gcc
Build Tag Built by
---------------------------------------- -------------------- ----------------
gcc-8.1.1-1.fc28 f28-updates jakub
So perhaps the rebuild in F28 needs to wait a bit.
Marek
5 years, 10 months
Mass rebuild, mass Golang packages failures
by Robert-André Mauchin
Hello,
Since the mass rebuild and maybe because of Golang 1.11 beta 1, all the Golang
package containing a binary fail because the debuginfo are not generated:
RPM build errors:
error: Empty %files file /builddir/build/BUILD/rclone-1.42/
debugsourcefiles.list
Empty %files file /builddir/build/BUILD/rclone-1.42/debugsourcefiles.list
Child return code was: 1
whereas it was working correctly before.
Has anyone any idea what is causing this and how it can be fixed?
Best regards,
Robert-André
5 years, 10 months
Updating R-qvalue and R-xtable
by Pierre-Yves Chibon
Good Morning Everyone,
A few days ago, I sent this to the r-sig-fedora(a)r-project.org list but without
reactions, so I'm sending it here as well in case there are more people
interested here :)
Thanks,
Pierre
------
Good Morning Everyone,
I was going through my list of FTBFS packages today and I fixed all my R
packages but 2: qvalue and xtable.
qvalue requires ggplot2
xtable requires: lsmeans, spdep, splm, sphet, plm
I am not doing any R anymore these days and in fact spot has been the one
maintaining most of my R packages these days (thanks spot!!), so I am not really
interested in maintaining more R packages.
Would someone be interested in packaging these libraries?
Otherwise, I think both qvalue and xtable will get retired in a nearish future.
Thanks for your help :)
Pierre
5 years, 10 months
Audit-3.0 pre-release coming to rawhide
by Steve Grubb
Hello,
This is to let everyone know that audit-3.0 pre-release is coming to rawhide.
The big change that is prompting this email is that there is a config change
that people might need to be aware of. One of the improvements is to drop
audispd (realtime audit event dispatcher) and merge its functionality into
auditd. This will eliminate one source of overflow messages and decrease the
time from event occurrence to plugin seeing it. But since audispd doesn't
exist anymore, I think that the location for the plugin directory should be
moved from /etc/audisp/plugins.d/ to /etc/audit/plugins.d/. This way we have
all audit config items in one place for the first time.
I have already coordinated this with the selinux team. They are moving the
selinux troubleshooter plugin and adjusting selinux policy for the new
locations and transitions from auditd to the plugins. I don't know if there
are any people beyond the selinux team that makes use of the audit dispatcher
plugin capabilities for real time audit analysis. But that is why I'm doing a
pre-release and making this announcment.
Audit-3.0 is only intended for F29 and higher. It will never be backported to
anything less because of these changes.
Thanks,
-Steve
5 years, 10 months