Comments

144 Comments
karma

Not only "su -" is affected, also "su USERNAME".

There's also SELinux errors. Since the downgrade works, I assume something is broken in this test update.

Sep 19 00:39:10 noname audit[4498]: AVC avc: denied { mounton } for pid=4498 comm="(fprintd)" path="/var/lib/fprint" dev="sda5" ino=793064 scontext=system_u:system_r:init_t:s0 tcontext=system_u:object_r:fprin Sep 19 00:39:10 noname audit: SELINUX_ERR op=security_bounded_transition seresult=denied oldcontext=system_u:system_r:init_t:s0 newcontext=system_u:system_r:fprintd_t:s0 Sep 19 00:39:10 noname audit[4498]: AVC avc: denied { map } for pid=4498 comm="fprintd" path="/usr/libexec/fprintd" dev="sda5" ino=1704831 scontext=system_u:system_r:init_t:s0 tcontext=system_u:object_r:fprin Sep 19 00:39:10 noname audit[4498]: ANOM_ABEND auid=4294967295 uid=0 gid=0 ses=4294967295 subj=system_u:system_r:init_t:s0 pid=4498 comm="fprintd" exe="/usr/libexec/fprintd" sig=11 res=1

SELinux is preventing (fprintd) from mounton access on the directory /var/lib/fprint.

* Plugin catchall (100. confidence) suggests ******

If you believe that (fprintd) should be allowed mounton access on the fprint directory by default. Then you should report this as a bug. You can generate a local policy module to allow this access. Do allow this access for now by executing:

ausearch -c '(fprintd)' --raw | audit2allow -M my-fprintd

semodule -X 300 -i my-fprintd.pp

Additional Information: Source Context system_u:system_r:init_t:s0 Target Context system_u:object_r:fprintd_var_lib_t:s0 Target Objects /var/lib/fprint [ dir ] Source (fprintd) Source Path (fprintd) Port <Unknown> Host localhost.localdomain Source RPM Packages
Target RPM Packages fprintd-0.8.0-1.fc27.x86_64 Policy RPM selinux-policy-3.13.1-283.fc27.noarch Selinux Enabled True Policy Type targeted Enforcing Mode Enforcing Host Name noname Platform Linux noname 4.13.0-0.rc7.git0.1.fc27.x86_64 #1 SMP Mon Aug 28 02:33:21 UTC 2017 x86_64 x86_64 Alert Count 1 First Seen 2017-09-18 01:29:51 CEST Last Seen 2017-09-18 01:29:51 CEST Local ID 4aeb873a-c1c8-49bf-9c12-7dbc75c68ec5

Raw Audit Messages type=AVC msg=audit(1505690991.218:591): avc: denied { mounton } for pid=22890 comm="(fprintd)" path="/var/lib/fprint" dev="sda5" ino=793064 scontext=system_u:system_r:init_t:s0 tcontext=system_u:object_r:fprintd_var_lib_t:s0 tclass=dir permissive=0

Hash: (fprintd),init_t,fprintd_var_lib_t,dir,mounton

After installing this update, "su -" in gnome-terminal suffers from a 15 seconds delay. Downgrading to 0.7.0-4.fc27 fixes it.

Would be enough for a -1, but maybe I'm missing something.

Also fixes the currently crashing 3.15.0 in F27:

folder.c:3967 Condition path != NULL failed
folder.c:3991 Condition path != NULL failed
msgcache.c:586 Condition cache_file != NULL failed
(claws-mail:21030): Claws-Mail-CRITICAL **: imap_item_get_path: assertion 'folder_path != NULL' failed
Segmentation fault (core dumped)

LAME works for MP3 encoding in Audacious' filewriter output plugin.

BZ#1449467 Review Request: lame - MP3 encoder

yum-deprecated showing different behaviour is a start. Good. See bug 1439690 I've assigned to "dnf".

Of course, users with update problems have got 3rd party packages installed. a52dec has not been offered by Fedora before! Those are the packages you try to handle in your %changelog, btw.

The fundamental problem is that you're replacing a52dec with liba52, replacing executables with only libraries. Whether the various package tools, such as DNF, show warnings or errors, is implementation dependent. All replace a52dec because of your Obsoletes tags and remove executables from users' installations accidentally.

That is reproducible by you, too, with pkcon or dnf or other tools, provided that you do start with the older package releases such as -21.fc25, regardless of whether single-arch or multilib. That doesn't matter.

The reason why dnf runs into broken deps or conflicts may be your circular Obsoletes as pointed out before. You can't update a52dec with a new releases, because the new a52dec requires liba52, which in turn obsoletes a52dec. No, the versions in the tags don't do what you possibly hope they would do. You lose a52dec and only get liba52.

karma

Impressive, Nicolas, how much you care about your package users and feedback from fellow Fedora contributors. Hopefuly you do understand that by pointing the finger at libsolv/dnf you've just closed the door completely and won't get any support or bugzilla ticket from me.

Notice the topic on users@ list. Update problems with older a52dec updates affects various dist releases.

I've given you a clear test-case for F25 and pointed out several mistakes in your update. I don't have the time to provide you with many more details, if you are not willing to test your own updates painstakingly.

You cannot assume that any user has already applied a previous update. Your -26.fc25 update also failed to replace -21.fc25 correctly, flags a52dec as broken and removes it because of the obsoletes. The entire upgrade path for these updates is broken.

Oh, there's also multiple Obsoletes tags for a52dec in different packages, and both cover -21.fc25. That won't work, since only either one would be chosen.

Also, since a52dec-0.7.4-21.fc25 contains libs and executables, letting liba52 replace it isn't correct. User would get only the lib, not the executables.

You've introduced a circular Requires/Obsoletes pair. a52dec requires liba52, which obsoletes a52dec.

karma

DNF for this update does not manage to replace a52dec-0.7.4-21.fc25 from rpmfusion. Not for x86_64, and not for x86_64/i686 multilib either.

zdenek : Please investigate further. That smells like some conflict in wxGTK compat packages, but isn't reproducible here. I've come from "dnf install audacity" to "dnf --enablerepo=updates-testing update audacity" without any such problems.

rpm -qa|grep -i wx

wxBase3-3.0.2-26.fc25.x86_64 compat-wxBase3-gtk2-3.0.2-7.fc24.x86_64 wxGTK3-3.0.2-26.fc25.x86_64 compat-wxGTK3-gtk2-3.0.2-7.fc24.x86_64

BZ#1347053 Audacity UI has rendering glitches

In case anyone is annoyed by the 'next_on_delete' change, the hidden preference setting can be changed with the "clawsker" tool. Its latest version can be found in updates-testing on its way into stable updates.

Been using this for some time, and the only interruption is the changed passphrase storage method, which requires re-entering passphrases for accounts once.

Unfortunately, you need another release, because now you've put both files into the runtime package, which is exactly the opposite of the previous bug. With this new version you've packaged, there is a new versioned runtime library, and you need to ship the .so symlink in the -devel package and the versioned .so.1 in the runtime package: https://fedoraproject.org/wiki/Packaging:Guidelines#Devel_Packages

BZ#1177213 Mispackaged .so library

Unfortunately, you need another release, because now you've put both files into the runtime package, which is exactly the opposite of the previous bug. With this new version you've packaged, there is a new versioned runtime library, and you need to ship the .so symlink in the -devel package and the versioned .so.1 in the runtime package: https://fedoraproject.org/wiki/Packaging:Guidelines#Devel_Packages

BZ#1177213 Mispackaged .so library
karma

It doesn't fix bug 1177213, because it still ships the runtime lib in the -devel package as pointed out.

BZ#1177213 Mispackaged .so library
karma

It fixes #1291903 -- and is an interesting case, since priority-based detection of file formats really is problematic if the implemented signatures aren't 100%.

BZ#1291903 Since file-5.19-2.fc22, a clawsker Perl script is misrecognized

No vote, because since approx. two days, when logging in into GNOME Shell, I receive a notification about failed network connection. dhclient reports "Network is down" and fails to send packets. NetworkManager reports "Activation: failed for connection 'Wired connection 1'", but that's the last activity in the logs, and the connection is up. I only click away the notification and need not do anything else.

First thing I will try is a downgrade to 1.0.6.