Bug#1070039: Acknowledgement (refpolicy: enforcing mode causes machine with GNOME desktop to crash)

2024-04-29 Thread Henrik Ahlgren
It seems the immediate crash was caused by gnome-shell trying to do execmem, I guess some JavaScript JIT thing. After enabling the allow_execmem boolean, gnome-shell no longer crashes. I am not sure how, but it would be good to have better out-of-the-box experience with SELinux on desktop

Bug#1070039: refpolicy: enforcing mode causes machine with GNOME desktop to crash

2024-04-29 Thread Henrik Ahlgren
Package: selinux-policy-default Version: 2:2.20221101-9 Severity: important Dear Maintainer, I am fully aware that selinux is not really considered a first class citizen in Debian, especially in graphical desktop use cases. Never had any trouble with AppArmor and I've had moderate success with

Bug#1043296: evolution: gdk_monitor_get_scale_factor: assertion 'GDK_IS_MONITOR (monitor)' failed

2023-08-08 Thread Henrik Ahlgren
Package: evolution Version: 3.46.4-2 Severity: normal X-Debbugs-Cc: none, Henrik Ahlgren Not sure if this bugs belongs to evolution or libgtk-3-0 package, but I mostly see this with Evolution (randomly also with LibreOffice?), so I don't think it's a Gtk bug affecting all software. Consider

Bug#472477: Still a problem in bookworm

2023-08-04 Thread Henrik Ahlgren
In a fresh install of bookworm with GNOME desktop, the problem of ssh-add -D not removing ed25519 keys still remains in 2023. When investigating this, I noticed that in the default configuration, there are at least FIVE separate SSH agent processes running: 1. gnome-keyring-daemon process (the

Bug#1022042: linux-image-amd64: linux-image-5.10.0-19-amd64 fails to boot on machines with AMD integrated graphics

2022-10-19 Thread Henrik Ahlgren
Same thing here with an AMD Ryzen 5 PRO 3400G system. Also gnome-shell have been complaining about "Failed to set CRTC gamma: drmModeCrtcSetGamma on CRTC 62 failed: Permission denied" for some time. I don't know what is CRTC 62, but clearly it is related.

Bug#675710: colord-sane still tries to access non-scanner USB devices

2022-02-17 Thread Henrik Ahlgren
The error message with version 1.4.5-3 in bullseye is slightly different, but colord-sane still outputs nasty errors with high severity to the system log when connecting and disconnecting USB storage devices: Feb 17 20:40:23 zanshin colord-sane[205757]: io/hpmud/musb.c 2101: Invalid usb_open:

Bug#895480: gnome-power-manager: Homepage under projects.gnome.org no longer exists

2021-11-08 Thread Henrik Ahlgren
In addition to the description being very much outdated and misleading and confusing, it does not help that the upstream homepage in package metadata https://projects.gnome.org/gnome-power-manager/ is now 404. Henrik

Bug#793675: hplip-gui: No system tray detected

2020-10-17 Thread Henrik Ahlgren
Any possibility the whole /etc/xdg/autostart/hplip-systray.desktop file could be dropped from future versions if it does not work at all? The "No system tray detected on this system." error dialog every user gets when logging in in default GNOME desktop is very annoying. Please fix this issue

Bug#513964: base-passwd: please add netdev and powerdev groups to group.master

2020-10-11 Thread Henrik Ahlgren
Is the netdev group really obsolete? It seems to me that at least network-manager, wpasupplicant and avahi-daemon rely on it (for dbus.1/polkit-1). All packages add the group in .postinst. Like with most of these old groups, the name is not optimal: it does not really deal with "devices". Only

Bug#972052: base-passwd: cups does not run as cupsys:lp

2020-10-11 Thread Henrik Ahlgren
reassign 972052 base-passwd retitle 972052 base-passwd: cups does not run as cupsys:lp thanks My bad, wrong package.

Bug#972052: base-files: cups does not run as cupsys:lp

2020-10-11 Thread Henrik Ahlgren
Package: cups-daemon Version: 2.2.10-6+deb10u3 Severity: normal The cupsys group was added to users-and-groups in 2005 (#290237). I don't remember if cups used to run as such user back then, but now it (very unfortunately) runs as root, and no such user is present in the system. This is at least

Bug#600700: base-passwd: sudo group's documented semantics don't match the sudo package

2020-10-11 Thread Henrik Ahlgren
I still feel sudo is such an important group, that it would be good if users-and-groups.txt would be further improved to more clearly document that this group is what allows users to not only run commands with sudo and pkexec, but also execute various other polkit-1 actions like install software

Bug#962420: /usr/local/share/fonts owned by group staff even if /etc/staff-group-for-usr-local not present

2020-10-10 Thread Henrik Ahlgren
I believe using dh_usrlocal(1) debhelper should do this automatically. Manpage: If a directory is owned by root:root, then ownership will be determined at install time. The ownership and permission bits will either be root:root mode 0755 or root:staff mode 02775. The actual choice depends on

Bug#821424: Groups for default user created by d-i

2020-10-10 Thread Henrik Ahlgren
I've always found it bit weird and confusing that the first user created during installation by d-i is "special" and belongs to a number of groups that apparently are mostly unecessary in the modern world. However, when you add a new user using the command line (useradd/adduser), or the GNOME

Bug#875890: Apparmor causes problems with Stretch upgrade

2017-11-06 Thread Henrik Ahlgren
Upgrading from Jessie to Stretch with apparmor enabled seems to fail: Setting up mariadb-server-10.1 (10.1.26-0+deb9u1) ... Installing new version of config file /etc/apparmor.d/usr.sbin.mysqld ... Installing new version of config file /etc/init.d/mysql ...

Bug#873581: certbot: Excessive logging

2017-08-29 Thread Henrik Ahlgren
Package: certbot Version: 0.10.2-1~bpo8+1 Severity: normal Certbog logs to /var/log/letsencrypt.log using DEBUG as the default log level. It rotates the log on each invocation, i.e. (at least) daily. If I understand correctly (main.py:setup_log_file_handler), 1000 log files are retained. On my

Bug#849227: Onionshare: CLI works but GUI doesn't

2017-05-26 Thread Henrik Ahlgren
Yes, the test case you described works as expected, but have you tried this: - Start onionshare GUI - Add a file - Uncheck the "Stop server automatically" setting - Click "Start server" and wait for the server to come up - Download the file using Tor Browser - Select "New tor circuit for this

Bug#814759: F1 is Help in GNOME Terminal

2017-02-21 Thread Henrik Ahlgren
F1 works in many terminals (e.g. plain xterm), but in GNOME it opens the help window, unless the shortcut is disabled. It would be useful if simple "1", "2", etc. would work instead of function keys.

Bug#849227: CLI works but GUI doesn't

2016-12-23 Thread Henrik Ahlgren
Oops, sorry about my brainfart. Of course it does not shut down the onion service in case I only download the HTML download page (using wget), but not the payload itself. After downloading "/download", onionshare says "Closing automatically because download finished" and the hidden service is gone

Bug#849227: onionshare: CLI never shuts down after download - GUI always does

2016-12-23 Thread Henrik Ahlgren
Package: onionshare Version: 0.6-3 Severity: important Manual page reads: "OnionShare's default behaviour is to shut down the hidden service and to stop once the file has been downloaded. You can prevent this behaviour by invoking the --stay-open option. This can be useful if you want multiple

Bug#771891: Removing the cache not a good workaround

2016-11-27 Thread Henrik Ahlgren
I disagree that this issue is not serious since there is a workaround. If the cache needs to be deleted every time you want to access your email, it basically makes the program unusable as an IMAP client. Also, sometimes certain emails do not appear in the inbox even after deleting the cache. They

Bug#313237: Fixed in tar 1.24

2016-10-14 Thread Henrik Ahlgren
I believe this has been fixed since GNU tar 1.24. >From NEWS: ** Symbolic link attributes When extracting symbolic links, tar now restores attributes such as last-modified time and link permissions, if the operating system supports this. For example, recent versions of the Linux kernel support

Bug#838377: apt-show-versions: Make sure cache files are world-readable

2016-09-20 Thread Henrik Ahlgren
Package: apt-show-versions Version: 0.22.4 Severity: wishlist If a strict umask in effect during cache file initialization, apt-show-versions creates the files without read permissions for normal users. If a non-root user runs apt-show-versions, this happens: can't open

Bug#780814: Included mp3 files are unusable

2016-08-24 Thread Henrik Ahlgren
If support for MP3 is not enabled (due to licensing/patent issues perhaps?), what is the point of shipping all the files in /usr/share/scratch/Media/Sounds? They just waste disk space. The file selection dialog does not even indicate file type (mp3/wav), so the user experience is frustrating

Bug#834803: macchanger: log should include timestamps

2016-08-19 Thread Henrik Ahlgren
Package: macchanger Version: 1.7.0-5.3 Severity: wishlist Dear Maintainer, ifupdown.sh logs to /var/log/macchanger.log simply by redirecting the output of macchanger and few echo commands. This means no timestamps are written to the log, making it difficult to debug problems. Please consider

Bug#808236: electrum: missing icon

2015-12-17 Thread Henrik Ahlgren
Package: electrum Version: 2.5.4-2 Severity: minor Dear Maintainer, 1.9.8-4 (jessie) had the application icon installed in rather strange location but it worked just fine: /usr/share/app-install/icons/electrum.png Version 2.5.4 does not appear to ship an icon file at all. However it says

Bug#791858: Does not happen with .txt

2015-11-30 Thread Henrik Ahlgren
This issue seems to only affect export to XML. Export to text file defaults to "All Files" in the file selection dialog, where XML defaults to *.xml. Otherwise the source code looks pretty similar for both: Export_KeePassX_Xml.cpp: bool Export_KeePassX_Xml::exportDatabase(QWidget*

Bug#791858: keepassx: XML export security bug

2015-11-29 Thread Henrik Ahlgren
severity 791858 grave tags 791858 security thanks How come this bug has not been marked as a pretty severe security issue? Just accessing a menu item, but canceling the export operation by hitting Esc or clicking Cancel silently creates a hidden (dotfile) cleartext copy of all of the user's

Bug#777559: [aufs-tools] auplink crashes

2015-08-13 Thread Henrik Ahlgren
Tags: patch When AuFin is called with errno = 0, error_at_line(3) does not exit. The attached patch sets errno to EINVAL following the pattern in most other similar error checks in aufs-tools. With this change auplink / flush outputs the following error message and exists without segfaulting:

Bug#771945: tor: abstractions/nameservice is unnecessary in AppArmor profile

2014-12-04 Thread Henrik Ahlgren
On Thu, Dec 04, 2014 at 09:56:18AM +0100, intrigeri wrote: If I remove that line, then the system Tor daemon fails to start here with: [notice] Tor v0.2.5.10 (git-43a5f3d91e726291) running on Linux with Libevent 2.0.21-stable, OpenSSL 1.0.1j and Zlib 1. [notice] Read configuration file

Bug#771945: tor: abstractions/nameservice is unnecessary in AppArmor profile

2014-12-03 Thread Henrik Ahlgren
Package: tor Version: 0.2.5.10-1 Severity: normal Please consider removing the line #include abstractions/nameservice from /etc/apparmor.d/abstractions/tor. In my testing tor seems to function just fine without it; I haven't seen any errors reported to audit.log or tor's own log. The

Bug#761116: git-annex webapp makes an annex repo out of home directory

2014-09-10 Thread Henrik Ahlgren
Package: git-annex Version: 5.20140831+b1 Severity: important Tags: upstream It is not really a common nor good practice, but sometimes users have their $HOME (dotfiles mostly) directly under git version control, i.e. ~/.git exists and ~ is the git working directory. Usually most files are

Bug#720289: more details

2013-08-21 Thread Henrik Ahlgren
On Tue, Aug 20, 2013 at 11:32:16PM -0700, Larry Doolittle wrote: I have not found this stanza anywhere in a Mozilla version control repository (I know git, but I don't claim to be a GitHub or Mozilla expert), so I don't know what its historical context is. Wikipedia tells me FireFox 17 ESR

Bug#704957: git-annex: git annex fsck reports bad file content for intact files

2013-04-08 Thread Henrik Ahlgren
Package: git-annex Version: 3.20120629 Severity: normal Dear Maintainer, The annex-fsck command appears to think some files are bad, even though there appears to be nothing wrong with them. What I did: I ran git annex fsck on two separate annex repositories (on the same machine), both

Bug#704957: maybe a problem with filenames containing many dots?

2013-04-08 Thread Henrik Ahlgren
After investigating a bit further, I noticed that a common thing with the files that fail the consistency check is that their names contains multiple dots, e.g. Fanni Joulukortti 2006.psd.gz and 16.01.1997-15.02.1997.pdf, something.tar.gz. The annex file contain two parts of the filename, e.g.

Bug#704957: already fixed

2013-04-08 Thread Henrik Ahlgren
Apologies--I did not read the changelog of more recent git-annex versions before reporting this bug, and I did not find any previous bug reports concerning this in BTS. I just noticed that this issue has already been fixed in version 3.20120825. Also, I forgot to mention that the repositories in

Bug#704957: already fixed

2013-04-08 Thread Henrik Ahlgren
On Mon, Apr 08, 2013 at 12:53:52PM -0400, Joey Hess wrote: mv .git/annex/bad . git annex add bad rm -rf bad Can you please explain how that should work? When I run that, it looks like git-annex just thinks I want to add new files, and git status reports this: # On branch master # Changes to

Bug#701102: iceweasel: PDF files are rendered in Iceweasel and are partly unreadable

2013-02-21 Thread Henrik Ahlgren
On Thu, Feb 21, 2013 at 04:57:00PM +0100, Vincent Lefevre wrote: No, I don't have the choice. Or is this something in the preferences? But Iceweasel normally lets the user choose by default. Edit. Preferences. Applications. Portable Document Format (PDF). Change from Preview in Iceweasel to

Bug#701102: iceweasel: PDF files are rendered in Iceweasel and are partly unreadable

2013-02-21 Thread Henrik Ahlgren
On Thu, Feb 21, 2013 at 06:40:33PM +0100, Vincent Lefevre wrote: OK, but it was Always ask with Iceweasel 18! An upgrade must preserve the preferences chosen by the user! That may be a RC bug: local changes must be preserved during a package upgrade. AFAIK the built-in viewer is the default

Bug#699159: Enable pam_tty_audit module

2013-01-28 Thread Henrik Ahlgren
it in Wheezy. Best regards, Henrik Ahlgren -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#689035: libcap2: List of capabilities not in sync with the linux kernel and libc6

2012-09-28 Thread Henrik Ahlgren
Package: libcap2 Version: 1:2.22-1.2 Severity: important Tags: upstream Dear Maintainer, The libcap2 source tree contains a copy of the header file linux/capability.h that is from an older kernel version and is missing the capability CAP_WAKE_ALARM (35). This causes e.g. lxc-start to fail when

Bug#686252: mozilla.pp: file context path hardcoded to xulrunner-1.9.1

2012-08-30 Thread Henrik Ahlgren
Package: selinux-policy-default Version: 2:0.2.20100524-7+squeeze1 Severity: normal While trying to use the mozilla.debian.net version of iceweasel (15.0) with selinux in enforcing mode, it crashes with segmentation fault. If iceweasel is running when I run setenforce 1, it crashes immediately,

Bug#552147: use dpkg-reconfigure to add modules for newly installed packages

2012-06-16 Thread Henrik Ahlgren
It took me a fair amount of time to understand that modules only get loaded at the time selinux-policy-default is installed, by the postinst Perl script where there is a hardcoded mapping between currently installed packages and the corresponding selinux modules. With similar Debian machines

Bug#661289: Is a bug and should be fixed in stable too

2012-06-15 Thread Henrik Ahlgren
When I log in, as a normal user, to a Debian Squeeze system using the standard Gnome display manager/login (which I believe is gdm3), id -Z reports system_u:system_r:initrc_t:s0 as the context. If I log in to the same machine from a text virtual console (Alt-Ctrl-F1), the context is