Learned a few things about xdg and mimetype registration in the last week that could be helpful to have condensed in a single place.

No Need to Ship a Mailcap Mime File

If you already ship a .desktop file (that is what ends up in /usr/share/applications/) which has a MimeType declared, there is no need to also ship a mailcap file (that is what ends up in /usr/lib/mime/packages/). Some triggers will do the conversion work for you. See also Debian Policy 4.9.

Reverse DNS Naming Convention for .desktop Files

Seems to be a closely guarded secret, maybe mainly known inside the Gnome world, but it's in the spec. Also not very widely known inside Debian if I look at my local system as not very representative sample.

Your hicolor Theme App Icon can be a Mime Type Icon as Well

In case you didn't know the hicolor icon theme is the default fallback theme. Many of us already install application icons e.g. in /usr/share/icons/hicolor/48x48/apps/ which is used in conjunction with the Icon field in the .desktop file to locate the application icon. Now the next step, and there it seems quite of few us miss out, is to create a symlink to also provide a mime type icon, so it's displayed in graphical file managers for the application data files. The schema here is simple: Take the MimeType e.g. application/x-vymand replace the / with a - and use that as file name in e.g. /usr/share/icons/hicolor/48x48/mimetypes/. In the vym case that is /usr/share/icons/hicolor/48x48/mimetypes/application-x-vym.png. If you have one use a scalable .svg file instead of .png.

This seems to be an area where Debian lacks a bit of tooling to automatically convert application icons to all the different sizes and install it in all the appropriate places. What is already there is a trigger to run gtk-update-icon-cache when you install new icons into one of the icon theme folder so they're picked up.

No Priority or Order in .desktop Files

Likely something that hapens on all my fresh installations: Libreoffice is installed and xdg-open starts to open pdf files with Libreoffice instead of evince. Now I've to figure out again to run xdg-mime default org.gnome.Evince.desktop application/pdf to change that (at least for my user). Background here is that the desktop file spec explicitly mandates "Priority for applications is handled external to the .desktop files.". That's why we got in addition to all of that mimeapps.list files. And now, after running the xdg-mime command from above, we've a ~/.config/mimeapps.list defining

[Default Applications]
application/pdf=org.gnome.Evince.desktop

Debian as whole seems to be not very keen on shipping something like a sensible default mimeapps.list outside of desktop environment specific ones. A quick search gave me just

$ apt-file search mimeapps.list
cinnamon-desktop-data: /usr/share/applications/x-cinnamon-mimeapps.list
gdm3: /usr/share/gdm/greeter/applications/mimeapps.list
gnome-session-common: /usr/share/applications/gnome-mimeapps.list
plasma-workspace: /usr/share/applications/kde-mimeapps.list
sxmo-utils: /usr/share/applications/mimeapps.list
sxmo-utils: /usr/share/sxmo/xdg/mimeapps.list

While it's a bit anoying to run into that pdf vs Libreoffice thing every now and then, it's maybe better to not have long controversial threads about default pdf viewer, like the ones we already had about the default MTA choices. And while we're at it: everyone using Libreoffice should give a virtual hug to rene@ for taming that beast since 2010 and OpenOffice.org before.

Had a need for a mindmapping application and found view your mind in the archive. Works but the version is a bit rusty. Sadly my Debian packaging skills are a bit rusty as well, especially when it comes to bigger GUI applications. Thus I spent a good chunk of yesterday afternoon to rip out cdbs and package the last source release on github which is right now 2.9.22 (the release branch already has 2.9.27, still sorting that out).

Git repository and a amd64 build of the current state. It still deserves some additional love, e.g. creating a -common package for arch indep content.

Proposed a few changes upstream:

• spelling fixes
• .desktop file improvements
• small adjustments for the documentation installation via cmake

Also pinged pollux@ who uploaded vym up to 2019 if he'd be fine if I pick it up. If someone else is interested, I'm also fine to put it up on salsa in the general "Debian" group for shared maintenance. I guess I will use it in the future, but time is still a scarce resource for all of us.

I recently came a cross a x509 P(rivate)KI Root Certificate which had a pathLen constrain set on the (self signed) Root Certificate. Since that is not commonly seen I looked a bit around to get a better understanding about how the pathLen basic constrain should be used.

Primary source is RFC 5280 section 4.2.1.9

The pathLenConstraint field is meaningful only if the cA boolean is asserted and the key usage extension, if present, asserts the keyCertSign bit (Section 4.2.1.3). In this case, it gives the maximum number of non-self-issued intermediate certificates that may follow this certificate in a valid certification path

Since the Root is always self-issued it doesn't count towards the limit, and since it's the last certificate (or the first depending on how you count) in a chain, it's pretty much pointless to configure a pathLen constrain directly on a Root Certificate.

Another relevant resource are the Baseline Requirements of the CA/Browser Forum (currently v2.0.2). Section 7.1.2.1.4 "Root CA Basic Constraints" describes it as NOT RECOMMENDED for a Root CA.

Last but not least there is the awesome x509 Limbo project which has a section for validating pathLen constrains. Since the RFC 5280 based assumption is that self signed certs do not count, they do not check a case with such a constrain on the Root itself, and what the implementations do about it. So the assumption right now is that they properly ignore it.

Summary: It's pointless to set the pathLen constrain on the Root Certificate, so just don't do it.

Write it down before I forget about it again:

for x in $(gh api graphql --paginate -f query='query($endCursor:String) { organization(login:"myorg") {
    repositories(first: 100, after: $endCursor, isArchived:false) {
        pageInfo {
            hasNextPage
            endCursor
        }
        nodes {
            name
        }
    }
    }
    }' --jq '.data.organization.repositories.nodes[].name'); do

    secrets=$(gh secret list --json name --jq '.[].name' -R "myorg/${x}" | tr '\n' ',')
    if ! [ -z "${secrets}" ]; then
        echo "${x},${secrets}"
    fi
done

Requests a list of all not archived repositories in a GitHub org and queries repository secrets. If we find some we output the repo name and the secrets in a comma separated list. Not real CSV, but good enough for further processing. I've to admit it's kinda beautiful what you can do with the gh cli by now. Sadly it seems the secrets are not yet available via GraphQL (or I missed it in the docs), so I just use the gh cli to do the REST calls.

I stick to some very archaic workflows, e.g. to connect to some corp VPN I just run sudo vpnc-connect and later on sudo vpnc-disconnect. In the past that also managed to restore my resolv.conf, currently it doesn't. According to a colleague that's also the case for Ubuntu.

Taking a step back, the sane way would be to use the NetworkManager vpnc plugin, but that does not work with this specific case because we use uncool VPN tech which requires the Enable weak authentication setting for vpnc. There is a feature request open for that one at https://gitlab.gnome.org/GNOME/NetworkManager-vpnc/-/issues/11

Taking another step back I thought that it shouldn't be that hard to add some checkbox, a boolean and render out another config flag or line in a config file. Not as intuitive as I thought this mix of XML and C. So let's quickly look elsewhere.

What happens is that the backup files in /var/run/vpnc/ are created by the vpnc-scripts script called vpnc-script, but not moved back, because it adds some pid as a suffix and the pid is not the final pid of the vpnc process. Basically it can not find the backup when it tries to restore it. So I decided to replace the pid guessing code with a suffix made up of the gateway IP and the tun interface name. No idea if that is stable in all circumstance (someone with a vpn name DNS RR?) or several connections to different gateways. But good enough for myself, so here is my patch:

vpnc-scripts [master]$ cat debian/patches/replace-pid-detection 
Index: vpnc-scripts/vpnc-script
===================================================================
--- vpnc-scripts.orig/vpnc-script
+++ vpnc-scripts/vpnc-script
@@ -91,21 +91,15 @@ OS="`uname -s`"

 HOOKS_DIR=/etc/vpnc

-# Use the PID of the controlling process (vpnc or OpenConnect) to
-# uniquely identify this VPN connection. Normally, the parent process
-# is a shell, and the grandparent's PID is the relevant one.
-# OpenConnect v9.0+ provides VPNPID, so we don't need to determine it.
-if [ -z "$VPNPID" ]; then
-    VPNPID=$PPID
-    PCMD=`ps -c -o cmd= -p $PPID`
-    case "$PCMD" in
-        *sh) VPNPID=`ps -o ppid= -p $PPID` ;;
-    esac
+# This whole script is called twice via vpnc-connect. On the first run
+# the variables are empty. Catch that and move on when they're there.
+if [ -n "$VPNGATEWAY" ]; then
+    BACKUPID="${VPNGATEWAY}_${TUNDEV}"
+    DEFAULT_ROUTE_FILE=/var/run/vpnc/defaultroute.${BACKUPID}
+    DEFAULT_ROUTE_FILE_IPV6=/var/run/vpnc/defaultroute_ipv6.${BACKUPID}
+    RESOLV_CONF_BACKUP=/var/run/vpnc/resolv.conf-backup.${BACKUPID}
 fi

-DEFAULT_ROUTE_FILE=/var/run/vpnc/defaultroute.${VPNPID}
-DEFAULT_ROUTE_FILE_IPV6=/var/run/vpnc/defaultroute_ipv6.${VPNPID}
-RESOLV_CONF_BACKUP=/var/run/vpnc/resolv.conf-backup.${VPNPID}
 SCRIPTNAME=`basename $0`

 # some systems, eg. Darwin & FreeBSD, prune /var/run on boot

Or rolled into a debian package at https://sven.stormbind.net/debian/vpnc-scripts/

The colleague decided to stick to NetworkManager, moved the vpnc binary aside and added a wrapper which invokes vpnc with --enable-weak-authentication. The beauty is, all of this will break on updates, so at some point someone has to understand GTK4 to fix the NetworkManager plugin for good.

Always amusing to see some more or less famous open source tools on stage or in movies. Lately we watched THE ME (german only) which is mixing live playing of actors and pre recorded video material. In one of the early video sequences a fictional console interface is displayed, claiming to be running on a Macbook, and htop is used to look for a suspicious process.

PSA for those foolish enough to use Google Cloud and try to use private service connect: If you want to change the serviceAttachment your private service connect forwarding rule points at, you must delete the forwarding rule and create a new one. Updates are not supported. I've done that in the past via terraform, but lately encountered strange errors like this:

Error updating ForwardingRule: googleapi: Error 400: Invalid value for field 'target.target':
'<https://www.googleapis.com/compute/v1/projects/mydumbproject/regions/europe-west1/serviceAttachments/
k8s1-sa-xyz-abc>'. Unexpected resource collection 'serviceAttachments'., invalid

Worked around that with the help of terrraform_data and lifecycle:

resource "terraform_data" "replacement" {
    input = var.gcp_psc_data["target"]
}

resource "google_compute_forwarding_rule" "this" {
    count   = length(var.gcp_psc_data["target"]) > 0 ? 1 : 0
    name    = "${var.gcp_psc_name}-psc"
    region  = var.gcp_region
    project = var.gcp_project

    target                = var.gcp_psc_data["target"]
    load_balancing_scheme = "" # need to override EXTERNAL default when target is a service attachment
    network               = var.gcp_network
    ip_address            = google_compute_address.this.id

    lifecycle {
        replace_triggered_by = [
            terraform_data.replacement
        ]
    }
}

See also terraform data for replace_triggered_by.

I know a few people hold on to the exFAT fuse implementation due the support for timezone offsets, so here is a small update for you. Andrew released 1.4.0, which includes the timezone offset support, which was so far only part of the git master branch. It also fixes a, from my point of view very minor, security issue CVE-2022-29973. In addition to that it's the first build with fuse3 support. If you still use this driver, pick it up in experimental (we're in the bookworm freeze right now), and give it a try. I'm personally not using it anymore beyond a very basic "does it mount" test.

tl;dr; OpenSSL 3.0.1 leaks memory in ssl3_setup_write_buffer(), seems to be fixed in 3.0.5 3.0.2. The issue manifests at least in stunnel and keepalived on CentOS 9. In addition I learned the hard way that running a not so recent VirtualBox version on Debian bullseye let to dh parameter generation crashing in libcrypto in bn_sqr8x_internal().

A recent rabbit hole I went down. The actual bug in openssl was nailed down and documented by Quentin Armitage on GitHub in keepalived My bugreport with all back and forth in the RedHat Bugzilla is #2128412.

Act I - Hello stunnel, this is the OOMkiller Calling

We started to use stunnel on Google Cloud compute engine instances running CentOS 9. The loadbalancer in front of those instances used a TCP health check to validate the backend availability. A day or so later the stunnel instances got killed by the OOMkiller. Restarting stunnel and looking into /proc/<pid>/smaps showed a heap segment growing quite quickly.

Act II - Reproducing the Issue

While I'm not the biggest fan of VirtualBox and Vagrant I've to admit it's quite nice to just fire up a VM image, and give other people a chance to recreate that setup as well. Since VirtualBox is no longer released with Debian/stable I just recompiled what was available in unstable at the time of the bullseye release, and used that. That enabled me now to just start a CentOS 9 VM, setup stunnel with a minimal config, grab netcat and a for loop and watch the memory grow. E.g. while true; do nc -z localhost 2600; sleep 1; done To my surprise, in addition to the memory leak, I also observed some crashes but did not yet care too much about those.

Act III - Wrong Suspect, a Workaround and Bugreporting

Of course the first idea was that something must be wrong in stunnel itself. But I could not find any recent bugreports. My assumption is that there are still a few people around using CentOS and stunnel, so someone else should probably have seen it before. Just to be sure I recompiled the latest stunnel package from Fedora. Didn't change anything. Next I recompiled it without almost all the patches Fedora/RedHat carries. Nope, no progress. Next idea: Maybe this is related to the fact that we do not initiate a TLS context after connecting? So we changed the test case from nc to openssl s_client, and the loadbalancer healthcheck from TCP to a TLS based one. Tada, a workaround, no more memory leaking. In addition I gave Fedora a try (they have Vagrant Virtualbox images in the "Cloud" Spin, e.g. here for Fedora 36) and my local Debian installation a try. No leaks experienced on both. Next I reported #2128412.

Act IV - Crash in libcrypto and a VirtualBox Bug

When I moved with the test case from the Google Cloud compute instance to my local VM I encountered some crashes. That morphed into a real problem when I started to run stunnel with gdb and valgrind. All crashes happened in libcrypto bn_sqr8x_internal() when generating new dh parameter (stunnel does that for you if you do not use static dh parameter). I quickly worked around that by generating static dh parameter for stunnel. After some back and forth I suspected VirtualBox as the culprit. Recompiling the current VirtualBox version (6.1.38-dfsg-3) from unstable on bullseye works without any changes. Upgrading actually fixed that issue.

Epilog

I highly appreciate that RedHat, with all the bashing around the future of CentOS, still works on community contributed bugreports. My kudos go to Clemens Lang. Now that the root cause is clear, I guess RedHat will push out a fix for the openssl 3.0.1 based release they have in RHEL/CentOS 9. Until that is available at least stunnel and keepalived are known to be affected. If you run stunnel on something public it's not that pretty, because already a low rate of TCP connections will result in a DoS condition.

Update of my notes from 2020.

# Download a binary device tree file and matching kernel a good soul uploaded to github
wget https://github.com/vfdev-5/qemu-rpi2-vexpress/raw/master/kernel-qemu-4.4.1-vexpress
wget https://github.com/vfdev-5/qemu-rpi2-vexpress/raw/master/vexpress-v2p-ca15-tc1.dtb
# Download the official Rasbian image without X
wget https://downloads.raspberrypi.org/raspios_lite_armhf/images/raspios_lite_armhf-2022-04-07/2022-04-04-raspios-bullseye-armhf-lite.img.xz
unxz 2022-04-04-raspios-bullseye-armhf-lite.img.xz
# Convert it from the raw image to a qcow2 image and add some space
qemu-img convert -f raw -O qcow2 2022-04-04-raspios-bullseye-armhf-lite.img rasbian.qcow2
qemu-img resize rasbian.qcow2 4G
# make sure we get a user account setup
echo "me:$(echo 'test123'|openssl passwd -6 -stdin)" > userconf
sudo guestmount -a rasbian.qcow2 -m /dev/sda1 /mnt
sudo mv userconf /mnt
sudo guestunmount /mnt
# start qemu
qemu-system-arm -m 2048M -M vexpress-a15 -cpu cortex-a15 \
 -kernel kernel-qemu-4.4.1-vexpress -no-reboot \
 -smp 2 -serial stdio \
 -dtb vexpress-v2p-ca15-tc1.dtb -sd rasbian.qcow2 \
 -append "root=/dev/mmcblk0p2 rw rootfstype=ext4 console=ttyAMA0,15200 loglevel=8" \
 -nic user,hostfwd=tcp::5555-:22
# login at the serial console as user me with password test123
sudo -i
# enable ssh
systemctl enable ssh
systemctl start ssh
# resize partition and filesystem
parted /dev/mmcblk0 resizepart 2 100%
resize2fs /dev/mmcblk0p2

Now I can login via ssh and start to play:

ssh me@localhost -p 5555