GNOME 46 is on its final stretch to be released. It’s been a custom to blog a little about the wallpaper selection, which is a big part of GNOME’s visual identity.

The first notable change in 46 is that we’re finally delivering on the promise of bringing you a next generation image file format. Lots of performance issues had to be addressed first, apologies for the delay. While efficiency and filesize requirements might not be too high on the list outside of the geek crowd, there is one aspect of JPEG-XL that I am very excited about.

JPEG-XL allows the use of client-side synthesized grain. A method pioneered by Netflix/AV1 I believe. Compression algorithms struggle with high frequency detail, which often introduce visible artifacts. JPEG-XL allows to decouple the grain component from the actual image data. This allows for significantly more efficient compression of images that inherently require noise, such as those in gnome-backgrounds — smooth gradients that would otherwise be susceptible to color banding. To achieve similar fidelity of the grain if it were baked in, a classic format like JPEG would need an order of magnitude larger filesize. Having the grain in the format itself also allows to skip various techniques in the rendering or compositing in the 3D software.

Instead of compressing a noisy image, JPEG-XL allows to generate film-like grain as part of the decoding process. This synthesized grain combats issues like color banding while allowing a much more efficient compression on the original image data.

In essence, client-side grain in JPEG-XL isn’t simply added noise, but a sophisticated strategy for achieving both efficient compression and visually pleasing image quality, especially for images that would otherwise require inherent noise.

The fresh batch of wallpapers includes evolutions of the existing assets as well as new additions. A few material/shape studies have been added as well as simple 2D shape textures. Thanks to the lovely JPEG-XL grain described earlier, it’s not just Inkscape and Blender that were used.

I hope you’re going to pick at least one of the wallpapers when GNOME 46 releases later next week as your favorite. Let me know on fediverse!

Previously, Previously, Previously, Previously, Previously

Warewulf

Preface

In High Performance Computing (HPC) computing tasks are usually distributed among many compute threads which are spread across multiples cores, sockets and machines. These threads are thightly coupled together. For this compute clusters consist of a number of largely identical machines that need to be managed to maintain a well defined and identical setup across all nodes. Once clusters scale up there are many scalability factors to overcome. Warewulf is there to address this ‘administrative scaling’.

Warewulf is an operating system agnostic installation and management system for HPC clusters.
It is quick and easy to learn aud use as many settings are pre-configured to sensible defaults. It still provides the flexibility allowing to finely tune the configuration to local needs. It is released under the BSD license. Its source code is available at https://github.com/warewulf/warewulf. This is where the development happens as well.

Installing warewulf4

Compute clusters consist of at least one management or head node which is usually multi-homed connecting both to an external network and to a cluster private network as well as multiple compute nodes which reside solely on the private network. Other private networks dedicated to high speed task like RDMA and storage access may exist as well. Warewulf gets installed to one of the management nodes to manage and oversee the installation and management of the compute nodes. To install Warewulf on a cluster is running SUSE Linux Enterprise HPC 15 SP5, openSUSE Leap 15.5 or openSUSE Tumbleweed, simpy run:

zypper install warewul4

on this management node to install the SUSE-provided Warewulf package. This package seamlessly integrates into a SUSE system and should therefore be preferred over packages provided on Github.

During the installation the actual network configuration is written to the /etc/warewulf/warewulf.conf. These settings should be verified, as for multi homed hosts a sensible pre-configuration is not always possible.

Check /etc/warewulf/warewulf.conf for the following values:

ipaddr: 172.16.16.250
netmask: 255.255.255.0
network: 172.16.16.0

where ipaddr should be the ip address of this host management host. Also check the values of netmask and network - these should match this network.

Additionally, you should configure the ip addresse range for dynamic/unknown hosts:

dhcp:
  range start: 172.16.26.21
  range end: 172.16.26.50

If the ISC dhcpd server is used (default on SUSE) make sure the value of DHCPD_INTERFACE in the file /etc/sysconfig/dhcpd has been set to the right value.

You are now ready to start the warewulfd service itself which delivers the images to the nodes:

systemctl enable --now warewulfd.service

Now wwctl can be used to configure the the remaining services needed by warewulf. Run

wwctl configure --all

which will configure all warewulf related services.

Adding nodes and profiles to warewulf

Warewulf uses the concept of profiles which hold the generalized settings of the individual nodes. It comes with a predefined profile default, to which all new node will be assigned, if not set otherwise. You may obtain the values of the default profile with:

wwctl profile list default

Now, a node can be added with the command assigning it an IP address:

wwctl node add node01 -I 172.16.16.101

the mac address is known for this node, you can specify this as well:

wwctl node add node01 -I 172.16.16.101 -H cc:aa:ff:ff:ee

For adding several nodes at once you may also use a node range which results e.g.

wwctl node add node[01-10] -I 172.16.16.101

this will add the nodes with ip addresses starting at the specified address and incremented by warewulf.

Importing a container

Warewulf uses a special ¹ container as the base system to build OS images for the compute nodes. This is self contained and independent of the operating system installed on the warewulf host.

To import an openSUSE Leap 15.5 container use the command

wwctl container import docker://registry.opensuse.org/science/warewulf/leap-15.5/containers/kernel:latest leap15.5 --setdefault

This will import the specified container for the default profile.

Alternative container sources

Alternative containers are available from the openSUSE registry under the science project at

https://registry.opensuse.org/cgi-bin/cooverview?srch_term=project%3D%5Escience%3A

or from the upstream warewulf community repository

https://github.com/orgs/warewulf/packages?repo_name=warewulf-node-images

It is also possible to import an image from a chroot by using the path to chroot as argument for wwctl import.

Booting nodes

As a final preparation you should rebuild the container image by running

wwctl container build leap15.5

as well as all the configuration overlays with the command

wwctl overlay build

just in case the build of the image may have failed earlier due to an error. If you didn’t assign a hardware address to a node before you should set the node into the discoverable state before powering it on. This is done with

wwctl node set node01 --discoverable

Now the node(s) can be powered on and will boot into assigned container.

For a convenient experience you should now log out of and back into the warewulf host, as this way an ssh key for password-less login to the compute nodes will be created on the warewulf host. Note, that this key is not pass-phrase protected. If you require a pass phrase, it is probably a good idea to set one now:

ssh-keygen -p -f $HOME/.ssh/cluster

Also you should run

wwctl configure hostlist

to add the new nodes to the file /etc/hosts.

Additional configuration

The configuration files for the nodes are managed as Golang text templates. The resulting files are overlayed over the node images. There are two ways of transport for the overlays to the compute node, the

• system overlay
• runtime overlay where the system overlay is baked into the boot image of the compute node while the runtime overlay is updated on the nodes on a regular base (1 minute per default) via the wwclient service.

In the default configuration the overlay called wwinit is used as system overlay. You can list the files in this overlays with the command:

wwctl overlay list wwinit -a

which will show a list of all the files in the overlays. Files ending with the suffix .ww are interpreted as template by warewulf and the suffix is removed in the rendered overlay. The content of the overlay can be shown using the command

wwctl overlay show wwinit /etc/issue.ww

To render the template using the values for node01 use:

wwctl overlay show wwinit /etc/issue.ww -r node01

The overlay template itself may be edited using the command

wwctl overlay edit wwinit /etc/issue.ww

Please note that after editing templates the overlays aren’t updated automatically and should be rebuild with the command

wwctl overlay build

The variables available in a template can be listed with

wwctl overlay show debug /warewulf/template-variables.md.ww

Modifying the container

The node container is a self contained operating system image. You can open a shell in the image with the command

wwctl container shell leap15.5

After you have opend a shell in the image additional software can be installed with zypper.

The shell command provides the option --bind which allows to mount arbitrary host directories into the container during the shell session.

Please note that if a command exits with a status other than zero the image won’t be rebuilt automatically. So its also advised to rebuild the container with

wwctl conainer build leap15.5

after any change.

Network configuration

Warewulf allows to configure multiple network interfaces for the compute nodes. You can add another network interface for example for infiniband using the command

wwctl node set node01 --netname infininet -I 172.16.17.101 --netdev ib0 --mtu 9000 --type infiniband

This will add the infiniband interface ib0 to the node node01. You can now list the network interfaces of the node:

wwctl node list -n

As changes in the settings are not propagated to all configuration files, the node overlays should be rebuilt after this change running the command:

wwctl overlay build

After a reboot these changes will be present on the nodes, in the avove case the Infiniband interface will be active on the node.

A more elegant way to get same result is to create a profile to hold the values which are the same for all interfaces. In this case these are mtu and the netdev. A new profile for an Infiniband network is created using the command

wwctl profile add infiniband-nodes --netname infininet --netdev ib0 --mtu 9000 --type infiniband

Once this has been created, you may add this profile to a node and remove the node specific settings which are now part of the common profile by executing:

wwctl node set node01 --netname infininet --netdev UNDEF --mtu UNDEF --type UNDEF --profiles default,infiniband-nodes

You may list the data in a profile using this command:

wwctl profile list -A infiniband-nodes

Secure Boot

Switch to grub boot

Per default warewulf boots nodes via iPXE, which isn’t signed by SUSE and can’t be used when secure boot is enabled. In order to switch to grub as boot method you will have add/change following value in /etc/warewulf/warewulf.conf

warewulf:
  grubboot: true

After this change you will have to reconfigure dhcpd and tftp executing

wwctl configure dhcp
wwctl configure tftp

and rebuild the overlays with the command

wwctl overlay build

Also make sure that the packages shim and grub2-x86_64-efi for x86-64 or grub2-arm64-efi for arm are installed in the container. shim is required by secure boot.

Cross product secure boot

If secure boot is enabled on the compute nodes and you want to boot different products make sure that the compute nodes boot with the so called ‘http’ boot method: For secure boot the signed shim needs to match the signature of the other pieces of the boot chain - including the kernel. The ‘http’ method is handled by warewulfd which will look up the image to boot and pick the shim from the image to deploy to this node. Otherwise, the initial shim for PXE boot, which is the default boot method, is extracted from the host running the warewulfd server. Make sure, the node container contains the shim package. The host system shim will also be used for nodes which are in discoverable state and subsequently have no hardware address assigned yet.

Disk management

It is possible to manage the disks of the compute nodes with warewulf. Here, warewulf itself doesn’t manage the disks, but creates a configuration and service files for ignition to do this job.

Prepare container

As ignition and its dependencies aren’t installed in most of the containers you should install the packages ignition and gptfdisk in the container

wwctl container exec <container_name> zypper -n in -y ignition gptdisk

Add disk to configuration

For storage devices all the necessary structures must be configured which are

• physical storage device(s) to be used
• partition(s) on the disks
• filesystem(s) to be used

Disks

The path to the device e.g. /dev/sda must be used As diskname. The only valid configuration for disks is diskwipe which should be self explanatory.

Partitions

The partname is the name to the partition whick iginition uses as the path for the device files, e.g. /dev/disk/by-partlabel/$PARTNAME.

Additionally the size and number of the partition need be specified for all but the last partition (the one with the highest number) in which case this partition will be extended to the maximal size possible.

You should also set the boolean variable --partcreate so that a parition is created if it doesn’t exist.

Filesystems

Filesystems are defined by the partition which contains them, so the name should have the format /dev/disk/by-partlabel/$PARTNAME. A filesystem needs to have a path if it is to be mounted, but its not mandatory.

Ignition will fail, if there no filesystem type is defined.

Examples

You can add a scratch partition with

wwctl node set node01 \
  --diskname /dev/vda --diskwipe \
  --partname scratch --partcreate \
  --fsname scratch --fsformat btrfs --fspath /scratch --fswipe

This will be the only (and last) partition, therefore it does not require a size. To add another partition as swap partition, you many run:

wwctl node set n01 \
  --diskname /dev/vda \
  --partname swap --partsize=1024 --partnumber 1 \
  --fsname swap --fsformat swap --fspath swap

This adds the partition number 1 which will be placed before the scratch partition.

Version 4.6 of syslog-ng introduced windows-eventlog-xml-parser(), a dedicated parser for XML-formatted event logs from Windows. It makes the EventData portion of log messages more useful, as it combines two arrays into a list of name-value pairs.

https://www.syslog-ng.com/community/b/blog/posts/dedicated-windows-xml-eventlog-parser-in-syslog-ng

Sometimes you have many log messages from an app, but none of them have the exact content you need. This is where the grouping-by() parser of syslog-ng can help. It allows you to aggregate information from multiple log messages into a single message.

In this blog, I will show you how to parse sshd logs using the patterndb parser of syslog-ng, and then create an aggregate message from the opening and closing log message using grouping-by.

https://www.syslog-ng.com/community/b/blog/posts/aggregating-messages-in-syslog-ng-using-grouping-by

There are several changes happening in openSUSE’s rolling release Tumbleweed on a daily basis and integrating systemd-boot into has been evolving.

A shift from the traditional GRUB boot loader is promising better system boot performance and security.

An all-systems-go presentation by Ludwig Nussel sheds some light on the motivations, challenges and future direction of systemd-boot integration in openSUSE.

The primary motivation behind adopting systemd-boot lies in its simplicity and efficiency, especially when handling full-disk encryption. Traditional boot loaders like GRUB require embedding decryption code and key derivation functions that can complicate the boot loader code and the boot process; this could slow down the system at startup. With systemd-boot, these responsibilities are delegated to the Linux Kernel and user space, which helps to streamline the boot process.

MicroOS and Tumbleweed’s use of Btrfs and its snapshotting capabilities adds complexity to the boot process; this is being addressed by integrating systemd-boot with the snapshot management system to ensure each snapshot boots successfully and that kernel updates are handled gracefully within this dynamic environment.

To facilitate this integration, new packaging scripts and tools like sdbootutil were introduced to manage kernel versions, snapshots and boot entries. This tool plays a crucial role in making systemd-boot a practical choice for openSUSE users, as the way that the snapshots are managed are different in MicroOS than in Tumbleweed.

Using sdbootutil, the system will create new Type #1 boot entries in the EFI System Partition (ESP) to represent all the avaliable boot options, which will copy from the snapshot the new installed kernels into this partition. It will also generate good initrd for those snapshots.

When using full disk encryption, sdbootutil will also call the different commands that will update the policies required to automatically unlock the new snapshots using the internal Trusted Platform Module 2.0 (TPM2) device of the system.

Ideally, in the future, some if not all of this functionallity will be exported into systemd itself or other systems components in the distribution. These changes make this tools a good place to experiment, validate or discard the different approaches that openSUSE is leading with systemd-boot.

At this point, systemd-boot support in openSUSE is still considered experimental. Both Tumbleweed and MicroOS yast installers offer systemd-boot as an alternative to grub for the brave. There are also ready-made appliances for qemu that use systemd-boot and full-disk encryption by default.

1) Introduction

The dnf5daemon-server component offers a collection of D-Bus interfaces to interact with the dnf5 package manager on the system. An openSUSE community packager wanted to add the additional D-Bus component to the openSUSE Tumbleweed distribution. New D-Bus system services require a review by the SUSE security team. In the course of this review I found the issues described in this report.

The version of dnf5 I reviewed for this is 5.1.9, and any source code references below are based on the corresponding version tag in the upstream Git repository.

2) D-Bus Interface Design

The dnf5daemon-server offers a main interface “org.rpm.dnf.v0.SessionManager” on which clients can create a new session, which results in a new dynamically allocated D-Bus object being registered on the bus. This session object provides a set of additional D-Bus interfaces for modifying package manager configuration, for installing or removing packages or for inspecting metadata about packages and repositories on the system.

The dnf5daemon-server is running as root and can be autostarted via the D-Bus system bus if it is not already running. Any other users with access to the D-Bus system bus may talk to it.

For certain privileged operations the D-Bus service implements Polkit authentication. Only three operations are protected by Polkit:

• org.rpm.dnf.v0.rpm.RepoConf.write
• org.rpm.dnf.v0.rpm.execute_transaction
• org.rpm.dnf.v0.rpm.Repo.confirm_key

These relate to changing the repository configuration, executing transactions or importing new trusted signing keys. Transactions cover all kinds of changes introduced through the package manager.

All of these operations require auth_admin privileges on Polkit level. The integration of the Polkit authorization logic is correct as far as I can tell.

The per-session D-Bus interface provided by dnf5daemon-server is rather large and many calls take additional key/value maps to tune the behaviour of the package manager logic contained in libdnf5. In summary, the libdnf5 library is attached very closely to the D-Bus system bus via dnf5daemon-server.

3) Local Root Exploit via Configuration Dictionary (CVE-2024-1929)

While the privileged operations mentioned above are correctly protected by Polkit, there are issues with the D-Bus interface long before Polkit is even invoked.

The org.rpm.dnf.v0.SessionManager.open_session method takes a key/value map of configuration entries. A sub-entry in this map, placed under the “config” key, is another key/value map. The configuration values found in it will be forwarded as configuration overrides to the libdnf5::Base configuration. The spot where this happens is found in session.cpp:63.

Practically all libdnf5 configuration aspects can be influenced here, as can be seen in the ConfigMain class in config_main.hpp. Already when opening the session via D-Bus, the libdnf5 will be initialized using these override configuration values. There is no sanity checking of the content of this “config” map, which is untrusted data.

There are surely a lot of different ways to exploit this possibility to influence the libdnf5 configuration. The simplest approach to get full root privileges I found is to trick the library into loading a plug-in shared library under control of the unprivileged user.

To do this, the “pluginpath” and “pluginconfpath” configuration entries need to be supplied and need to point to a user-controlled path. There the unprivileged user can place a configuration file that in turn points towards a user controlled shared library. The library will then dlopen() this user controlled shared library, which will lead to full code execution in the context of the root user.

3.1) Proof of Concept

A proof of concept I wrote shows this vulnerability in action. I successfully tested this exploit also on Fedora 39 using dnf5daemon-server version 5.1.10. The only precondition for this exploit is that the dnf5daemon-server package is installed on the system. Any local user, even nobody, can obtain root privileges this way.

3.2) Bugfix

To fix this, I suggested to enforce a whitelist of configuration parameters that are allowed to be overridden. Only a very small subset of values should be allowed for unprivileged clients.

In upstream commit e51bf2f0d such a whitelist enforcement has been implemented for dnf5daemon-server.

3.3) CVE Assignment

Red Hat Product Security assigned CVE-2024-1929 for this issue.

4) No Limit on Number of Open Sessions / Bad Session Close Behaviour (CVE-2024-1930)

There is no limit on how many sessions D-Bus clients may create using the open_session() D-Bus method. In my tests I was able to quickly create about 4,500 sessions and keep them open. For each session a thread is created in dnf5daemon-server. This spends a couple of hundred megabytes of memory in the process. Further connections will become impossible, likely because no more threads can be spawned by the D-Bus service.

In some cases I even managed to cause the D-Bus service to abort() as a result of hitting resource limits. If the service continues running and the client disconnects, then the cleanup code found in SessionManager::on_name_owner_changed() runs for each session that has been created. Each Session holds a ThreadsManager, where the thread associated with each session originates from. It is a thread that is busy-waiting to join other threads, the code for this is found in threads_manager.cpp:33. Since there is a sleep of one second in this thread’s loop it takes about ~4,500 seconds for all the threads from the 4,500 sessions to be joined one by one. The service will be unreachable for more than an hour.

4.1) Bugfix

To fix this, I suggested to limit the number of sessions for each unprivileged user in the system to a sensible value. Also the busy-wait loop in ThreadsManager seems ill-devised. Maybe using a condition variable to inform the cleanup thread of new work would be more efficient. Having a dedicated ThreadsManager and join-thread for each session seems a bit overkill, too.

In upstream commit c090ffeb79 the maximum number of sessions is limited to three sessions. The problematic thread joining behaviour has not been addressed as of now.

4.2) CVE Assignment

Red Hat Product Security assigned CVE-2024-1930 for this issue.

5) Untrusted locale Setting for each Session

Another part of the per-session setup is the use of an untrusted locale setting in session.cpp:54. This string is passed to the C library’s newlocale() function in threads_manager.cpp:92. A danger here is that arbitrary user controlled files or symlinks could be processed by the C library, which could lead to various issues like information leaks, denial of service or even code execution. I looked into the GNU glibc implementation of newlocale(), and luckily it already implements a very careful locale lookup algorithm, that prevents that arbitrary paths are accessed if crafted locale strings are passed to it. This outcome might change if a different C library is used.

Whether anything bad could happen, apart from file system issues, when exotic locales are selected for a thread running in dnf5daemon-server is another question that I did not investigate more closely.

For hardening purposes I suggested that dnf5daemon-server performs a sanity check of the locale string to prevent a string that contains e.g. a slash character / from being used. I don’t know of any upstream commits that address this yet, but upstream stated that they intend to work on this in the future.

6) dnfdaemon-client Demands Full Root

Although the D-Bus service implements Polkit for privileged operations, the dnf5daemon-client refuses to perform privileged operations for non-root users. For example:

user$ dnf5daemon-client distro-sync
This command has to be run with superuser privileges (under the root user on most systems).

The code for this is found in various sub-command implementations:

$ grep -r 'throw UnprivilegedUserError' -C 1
dnf5daemon-client/commands/downgrade/downgrade.cpp-    if (!libdnf5::utils::am_i_root()) {
dnf5daemon-client/commands/downgrade/downgrade.cpp:        throw UnprivilegedUserError();
dnf5daemon-client/commands/downgrade/downgrade.cpp-    }
--
dnf5daemon-client/commands/distro-sync/distro-sync.cpp-    if (!libdnf5::utils::am_i_root()) {
dnf5daemon-client/commands/distro-sync/distro-sync.cpp:        throw UnprivilegedUserError();
dnf5daemon-client/commands/distro-sync/distro-sync.cpp-    }
--
dnf5daemon-client/commands/install/install.cpp-    if (!libdnf5::utils::am_i_root()) {
dnf5daemon-client/commands/install/install.cpp:        throw UnprivilegedUserError();
dnf5daemon-client/commands/install/install.cpp-    }
--
dnf5daemon-client/commands/remove/remove.cpp-    if (!libdnf5::utils::am_i_root()) {
dnf5daemon-client/commands/remove/remove.cpp:        throw UnprivilegedUserError();
dnf5daemon-client/commands/remove/remove.cpp-    }
--
dnf5daemon-client/commands/upgrade/upgrade.cpp-    if (!libdnf5::utils::am_i_root()) {
dnf5daemon-client/commands/upgrade/upgrade.cpp:        throw UnprivilegedUserError();
dnf5daemon-client/commands/upgrade/upgrade.cpp-    }
--
dnf5daemon-client/commands/reinstall/reinstall.cpp-    if (!libdnf5::utils::am_i_root()) {
dnf5daemon-client/commands/reinstall/reinstall.cpp:        throw UnprivilegedUserError();
dnf5daemon-client/commands/reinstall/reinstall.cpp-    }

It doesn’t make sense that the client forces users to run as root (and thereby increase the attack surface by running also the client code as root) when the service could authenticate the user via Polkit.

I suggested to drop this root-check code in the client, and rely on the authentication logic in the service side code. If authentication fails, then the client code can still hint at the possibility to run the program as root. I am not aware on any upstream commits that address this yet, but upstream stated that they intend to work on this in the future.

7) General Review Summary

In summary my impression is that the libdnf5 library is too closely connected to the D-Bus system bus. The library itself is unaware of the fact that it is running with partially untrusted input. The dnf5daemon-server code needs to carefully filter untrusted input before it is passed to the generic library code.

8) Timeline

9) References

• dnf5 repository on GitHub
• dnf5 5.1.9 Git version tag (the version that has been reviewed)
• code that imports untrusted configuration settings into libdnf5
• bugfix commit for issue 3)
• bugfix commit for issue 4)