Contributing to Debian is part of Freexian s mission. This article covers the latest achievements of Freexian and their collaborators. All of this is made possible by organizations subscribing to our Long Term Support contracts and consulting services.

Upcoming Improvements to Salsa CI, by Santiago Ruano Rinc n Santiago started picking up the work made by Outreachy Intern, Enock Kashada (a big thanks to him!), to solve some long-standing issues in Salsa CI. Currently, the first job in a Salsa CI pipeline is the extract-source job, used to produce a debianize source tree of the project. This job was introduced to make it possible to build the projects on different architectures, on the subsequent build jobs. However, that extract-source approach is sub-optimal: not only it increases the execution time of the pipeline by some minutes, but also projects whose source tree is too large are not able to use the pipeline. The debianize source tree is passed as an artifact to the build jobs, and for those large projects, the size of their source tree exceeds the Salsa s limits. This is specific issue is documented as issue #195, and the proposed solution is to get rid of the extract-source job, relying on sbuild in the very build job (see issue #296). Switching to sbuild would also help to improve the build source job, solving issues such as #187 and #298. The current work-in-progress is very preliminary, but it has already been possible to run the build (amd64), build-i386 and build-source job using sbuild with the unshare mode. The image on the right shows a pipeline that builds grep. All the test jobs use the artifacts of the new build job. There is a lot of remaining work, mainly making the integration with ccache work. This change could break some things, it will also be important to test how the new pipeline works with complex projects. Also, thanks to Emmanuel Arias, we are proposing a Google Summer of Code 2024 project to improve Salsa CI. As part of the ongoing work in preparation for the GSoC 2024 project, Santiago has proposed a merge request to make more efficient how contributors can test their changes on the Salsa CI pipeline.

/usr-move, by Helmut Grohne In January, we sent most of the moving patches for the set of packages involved with debootstrap. Notably missing is glibc, which turns out harder than anticipated via dumat, because it has Conflicts between different architectures, which dumat does not analyze. Patches for diversion mitigations have been updated in a way to not exhibit any loss anymore. The main change here is that packages which are being diverted now support the diverting packages in transitioning their diversions. We also supported a few packages with non-trivial changes such as netplan.io. dumat has been enhanced to better support derivatives such as Ubuntu.

Miscellaneous contributions

• Python 3.12 migration trundles on. Stefano Rivera helped port several new packages to support 3.12.
• Stefano updated the Sphinx configuration of DebConf Video Team s documentation, which was broken by Sphinx 7.
• Stefano published the videos from the Cambridge MiniDebConf to YouTube and PeerTube.
• DebConf 24 planning has begun, and Stefano & Utkarsh have started work on this.
• Utkarsh re-sponsored the upload of golang-github-prometheus-community-pgbouncer-exporter for Lena.
• Colin Watson added Incus support to autopkgtest.
• Colin discovered Perl::Critic and used it to tidy up some poor practices in several of his packages, including debconf.
• Colin did some overdue debconf maintenance, mainly around tidying up error message handling in several places (1, 2, 3).
• Colin figured out how to update the mirror size documentation in debmirror, last updated in 2010. It should now be much easier to keep it up to date regularly.
• Colin issued a man-db buster update to clean up some irritations due to strict sandboxing.
• Thorsten Alteholz adopted two more packages, magicfilter and ifhp, for the debian-printing team. Those packages are the last ones of the latest round of adoptions to preserve the old printing protocol within Debian. If you know of other packages that should be retained, please don t hesitate to contact Thorsten.
• Enrico participated in /usr-merge discussions with Helmut.
• Helmut sent patches for 16 cross build failures.
• Helmut supported Matthias Klose (not affiliated with Freexian) with adding -for-host support to gcc-defaults.
• Helmut uploaded dput-ng enabling dcut migrate and merging two MRs of Ben Hutchings.
• Santiago took part in the discussions relating to the EU Cyber Resilience Act (CRA) and the Debian public statement that was published last year. He participated in a meeting with Members of the European Parliament (MEPs), Marcel Kolaja and Karen Melchior, and their teams to clarify some points about the impact of the CRA and Debian and downstream projects, and the improvements in the last version of the proposed regulation.

Welcome to the January 2024 report from the Reproducible Builds project. In these reports we outline the most important things that we have been up to over the past month. If you are interested in contributing to the project, please visit our Contribute page on our website.

---

How we executed a critical supply chain attack on PyTorch John Stawinski and Adnan Khan published a lengthy blog post detailing how they executed a supply-chain attack against PyTorch, a popular machine learning platform used by titans like Google, Meta, Boeing, and Lockheed Martin :

Our exploit path resulted in the ability to upload malicious PyTorch releases to GitHub, upload releases to [Amazon Web Services], potentially add code to the main repository branch, backdoor PyTorch dependencies the list goes on. In short, it was bad. Quite bad.

The attack pivoted on PyTorch s use of self-hosted runners as well as submitting a pull request to address a trivial typo in the project s README file to gain access to repository secrets and API keys that could subsequently be used for malicious purposes.

New Arch Linux forensic filesystem tool On our mailing list this month, long-time Reproducible Builds developer kpcyrd announced a new tool designed to forensically analyse Arch Linux filesystem images. Called archlinux-userland-fs-cmp, the tool is supposed to be used from a rescue image (any Linux) with an Arch install mounted to, [for example], /mnt. Crucially, however, at no point is any file from the mounted filesystem eval d or otherwise executed. Parsers are written in a memory safe language. More information about the tool can be found on their announcement message, as well as on the tool s homepage. A GIF of the tool in action is also available.

Issues with our SOURCE_DATE_EPOCH code? Chris Lamb started a thread on our mailing list summarising some potential problems with the source code snippet the Reproducible Builds project has been using to parse the SOURCE_DATE_EPOCH environment variable:

I m not 100% sure who originally wrote this code, but it was probably sometime in the ~2015 era, and it must be in a huge number of codebases by now. Anyway, Alejandro Colomar was working on the shadow security tool and pinged me regarding some potential issues with the code. You can see this conversation here.

Chris ends his message with a request that those with intimate or low-level knowledge of time_t, C types, overflows and the various parsing libraries in the C standard library (etc.) contribute with further info.

Distribution updates In Debian this month, Roland Clobus posted another detailed update of the status of reproducible ISO images on our mailing list. In particular, Roland helpfully summarised that all major desktops build reproducibly with bullseye, bookworm, trixie and sid provided they are built for a second time within the same DAK run (i.e. [within] 6 hours) . Additionally 7 of the 8 bookworm images from the official download link build reproducibly at any later time. In addition to this, three reviews of Debian packages were added, 17 were updated and 15 were removed this month adding to our knowledge about identified issues. Elsewhere, Bernhard posted another monthly update for his work elsewhere in openSUSE.

Community updates There were made a number of improvements to our website, including Bernhard M. Wiedemann fixing a number of typos of the term nondeterministic . [ ] and Jan Zerebecki adding a substantial and highly welcome section to our page about SOURCE_DATE_EPOCH to document its interaction with distribution rebuilds. [ ].
diffoscope is our in-depth and content-aware diff utility that can locate and diagnose reproducibility issues. This month, Chris Lamb made a number of changes such as uploading versions 254 and 255 to Debian but focusing on triaging and/or merging code from other contributors. This included adding support for comparing eXtensible ARchive (.XAR/.PKG) files courtesy of Seth Michael Larson [ ][ ], as well considerable work from Vekhir in order to fix compatibility between various and subtle incompatible versions of the progressbar libraries in Python [ ][ ][ ][ ]. Thanks!

Reproducibility testing framework The Reproducible Builds project operates a comprehensive testing framework (available at tests.reproducible-builds.org) in order to check packages and other artifacts for reproducibility. In January, a number of changes were made by Holger Levsen:

• Debian-related changes:
  • Reduce the number of arm64 architecture workers from 24 to 16. [ ]
  • Use diffoscope from the Debian release being tested again. [ ]
  • Improve the handling when killing unwanted processes [ ][ ][ ] and be more verbose about it, too [ ].
  • Don t mark a job as failed if process marked as to-be-killed is already gone. [ ]
  • Display the architecture of builds that have been running for more than 48 hours. [ ]
  • Reboot arm64 nodes when they hit an OOM (out of memory) state. [ ]
• Package rescheduling changes:
  • Reduce IRC notifications to 1 when rescheduling due to package status changes. [ ]
  • Correctly set SUDO_USER when rescheduling packages. [ ]
  • Automatically reschedule packages regressing to FTBFS (build failure) or FTBR (build success, but unreproducible). [ ]
• OpenWrt-related changes:
  • Install the python3-dev and python3-pyelftools packages as they are now needed for the sunxi target. [ ][ ]
  • Also install the libpam0g-dev which is needed by some OpenWrt hardware targets. [ ]
• Misc:
  • As it s January, set the real_year variable to 2024 [ ] and bump various copyright years as well [ ].
  • Fix a large (!) number of spelling mistakes in various scripts. [ ][ ][ ]
  • Prevent Squid and Systemd processes from being killed by the kernel s OOM killer. [ ]
  • Install the iptables tool everywhere, else our custom rc.local script fails. [ ]
  • Cleanup the /srv/workspace/pbuilder directory on boot. [ ]
  • Automatically restart Squid if it fails. [ ]
  • Limit the execution of chroot-installation jobs to a maximum of 4 concurrent runs. [ ][ ]

Significant amounts of node maintenance was performed by Holger Levsen (eg. [ ][ ][ ][ ][ ][ ][ ] etc.) and Vagrant Cascadian (eg. [ ][ ][ ][ ][ ][ ][ ][ ]). Indeed, Vagrant Cascadian handled an extended power outage for the network running the Debian armhf architecture test infrastructure. This provided the incentive to replace the UPS batteries and consolidate infrastructure to reduce future UPS load. [ ] Elsewhere in our infrastructure, however, Holger Levsen also adjusted the email configuration for @reproducible-builds.org to deal with a new SMTP email attack. [ ]

Upstream patches The Reproducible Builds project tries to detects, dissects and fix as many (currently) unreproducible packages as possible. We endeavour to send all of our patches upstream where appropriate. This month, we wrote a large number of such patches, including:

• Bernhard M. Wiedemann:
  • cython (nondeterminstic path issue)
  • deluge (issue with modification time of .egg file)
  • gap-ferret, gap-semigroups & gap-simpcomp (nondeterministic config.log file)
  • grpc (filesystem ordering issue )
  • hub (random)
  • kubernetes1.22 & kubernetes1.23 (sort-related issue)
  • kubernetes1.24 & kubernetes1.25 (go -trimpath vs random issue)
  • libjcat (drop test files with random bytes)
  • luajit (Use new d option for deterministic bytecode output)
  • meson [ ][ ] (sort the results from Python filesystem call)
  • python-rjsmin (drop GCC instrumentation artifacts)
  • qt6-virtualkeyboard+others (bug parallelism/race)
  • SoapySDR (parallelism-related issue)
  • systemd (sorting problem)
  • warewulf (CPIO modification time issue, etc.)
• Chris Lamb:
  • #1060254 filed against mumble.
• James Addison:
  • guake ( Schroedinger file due to race condition)
  • qhelpgenerator-qt5 (timezone localization; fix also merged upstream for QT6)
  • sphinx (search index doctitle sorting)

Separate to this, Vagrant Cascadian followed up with the relevant maintainers when reproducibility fixes were not included in newly-uploaded versions of the mm-common package in Debian this was quickly fixed, however. [ ]

---

If you are interested in contributing to the Reproducible Builds project, please visit our Contribute page on our website. However, you can get in touch with us via:

• IRC: #reproducible-builds on irc.oftc.net.
• Mailing list: rb-general@lists.reproducible-builds.org
• Mastodon: @reproducible_builds
• Twitter: @ReproBuilds

New Year, Same Great People! Our Debian User Group met for the first of our 2024 bi-monthly meetings on February 4th and it was loads of fun. Around twelve different people made it this time to Koumbit, where the meeting happened. As a reminder, our meetings are called "Debian & Stuff" because we want to be as open as possible and welcome people that want to work on "other stuff" than Debian. Here is what we did: pollo:

• tested a laptop that had a defective battery with a known good one (the problem was indeed with the battery :D)
• renewed his expiring OpenPGP key
• worked on removing trapperkeeper-scheduler-clojure from the DebCI reject_list
• helped anarcat with packaging sigal
• managed to feed most of the people present :)

LeLutin:

• worked with lavamind to upload new upstream release of smokeping

mjeanson:

• migrated from lxd to incus on his servers
• helped anarcat with flashing his AirGradient

lavamind:

• submitted a bug report on r-cran-rserve (promptly fixed/uploaded by maintainer!)
• reviewed and uploaded smokeping
• bug triaged the facter package
• worked on puppet-agent new upstream version 8.4.0

viashimo:

• updated puppet-strings from 2.9.0 to 4.2.1
• reported upstream test failures on puppet-strings with recent versions of mdl

tvaz & tassia:

• drafted a call & request for funding for the Vanier College FLOSS Club hardware marathon at Eastern Bloc
• worked on an application to conduct research at Vanier College on Debian usability
• babysitted :-)

joeDoe:

• worked on his AirGradient
• debugged the WiFi and VPN setup on his new laptop

anarcat:

• worked on packaging sigal
• worked on flashing his AirGradient boards
• rewrote a Prometheus exporter for Airgradient devices
• brought over a car full of old computer gear to donate

Pictures I was pretty busy this time around and ended up not taking a lot of pictures. Here's a bad one of the ceiling at Koumbit I took, and a picture by anarcat of the content of his boxes of loot:

Roadkiller was a Soekris net5501 router I used as my main gateway between 2010 and 2016 (for r seau and t l phone). It was upgraded to FreeBSD 8.4-p12 (2014-06-06) and pkgng. It was retired in favor of octavia around 2016. Roughly 10 years later (2024-01-24), I found it in a drawer and, to my surprised, it booted. After wrangling with a RS-232 USB adapter, a null modem cable, and bit rates, I even logged in:

comBIOS ver. 1.33  20070103  Copyright (C) 2000-2007 Soekris Engineering.
net5501
0512 Mbyte Memory                        CPU Geode LX 500 Mhz 
Pri Mas  WDC WD800VE-00HDT0              LBA Xlt 1024-255-63  78 Gbyte
Slot   Vend Dev  ClassRev Cmd  Stat CL LT HT  Base1    Base2   Int 
-------------------------------------------------------------------
0:01:2 1022 2082 10100000 0006 0220 08 00 00 A0000000 00000000 10
0:06:0 1106 3053 02000096 0117 0210 08 40 00 0000E101 A0004000 11
0:07:0 1106 3053 02000096 0117 0210 08 40 00 0000E201 A0004100 05
0:08:0 1106 3053 02000096 0117 0210 08 40 00 0000E301 A0004200 09
0:09:0 1106 3053 02000096 0117 0210 08 40 00 0000E401 A0004300 12
0:20:0 1022 2090 06010003 0009 02A0 08 40 80 00006001 00006101 
0:20:2 1022 209A 01018001 0005 02A0 08 00 00 00000000 00000000 
0:21:0 1022 2094 0C031002 0006 0230 08 00 80 A0005000 00000000 15
0:21:1 1022 2095 0C032002 0006 0230 08 00 00 A0006000 00000000 15
 4 Seconds to automatic boot.   Press Ctrl-P for entering Monitor.
 
                                            
                                                  ______
                                                    ____  __ ___  ___ 
            Welcome to FreeBSD!                     __   '__/ _ \/ _ \
                                                    __       __/  __/
                                                                      
    1. Boot FreeBSD [default]                     _     _   \___ \___ 
    2. Boot FreeBSD with ACPI enabled             ____   _____ _____
    3. Boot FreeBSD in Safe Mode                    _ \ / ____   __ \
    4. Boot FreeBSD in single user mode             _)   (___         
    5. Boot FreeBSD with verbose logging            _ < \___ \        
    6. Escape to loader prompt                      _)  ____)    __   
    7. Reboot                                                         
                                                  ____/ _____/ _____/
                                            
                                            
                                            
    Select option, [Enter] for default      
    or [Space] to pause timer  5            
  
Copyright (c) 1992-2013 The FreeBSD Project.
Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994
        The Regents of the University of California. All rights reserved.
FreeBSD is a registered trademark of The FreeBSD Foundation.
FreeBSD 8.4-RELEASE-p12 #5: Fri Jun  6 02:43:23 EDT 2014
    root@roadkiller.anarc.at:/usr/obj/usr/src/sys/ROADKILL i386
gcc version 4.2.2 20070831 prerelease [FreeBSD]
Timecounter "i8254" frequency 1193182 Hz quality 0
CPU: Geode(TM) Integrated Processor by AMD PCS (499.90-MHz 586-class CPU)
  Origin = "AuthenticAMD"  Id = 0x5a2  Family = 5  Model = a  Stepping = 2
  Features=0x88a93d<FPU,DE,PSE,TSC,MSR,CX8,SEP,PGE,CMOV,CLFLUSH,MMX>
  AMD Features=0xc0400000<MMX+,3DNow!+,3DNow!>
real memory  = 536870912 (512 MB)
avail memory = 506445824 (482 MB)
kbd1 at kbdmux0
K6-family MTRR support enabled (2 registers)
ACPI Error: A valid RSDP was not found (20101013/tbxfroot-309)
ACPI: Table initialisation failed: AE_NOT_FOUND
ACPI: Try disabling either ACPI or apic support.
cryptosoft0: <software crypto> on motherboard
pcib0 pcibus 0 on motherboard
pci0: <PCI bus> on pcib0
Geode LX: Soekris net5501 comBIOS ver. 1.33 20070103 Copyright (C) 2000-2007
pci0: <encrypt/decrypt, entertainment crypto> at device 1.2 (no driver attached)
vr0: <VIA VT6105M Rhine III 10/100BaseTX> port 0xe100-0xe1ff mem 0xa0004000-0xa00040ff irq 11 at device 6.0 on pci0
vr0: Quirks: 0x2
vr0: Revision: 0x96
miibus0: <MII bus> on vr0
ukphy0: <Generic IEEE 802.3u media interface> PHY 1 on miibus0
ukphy0:  none, 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto, auto-flow
vr0: Ethernet address: 00:00:24:cc:93:44
vr0: [ITHREAD]
vr1: <VIA VT6105M Rhine III 10/100BaseTX> port 0xe200-0xe2ff mem 0xa0004100-0xa00041ff irq 5 at device 7.0 on pci0
vr1: Quirks: 0x2
vr1: Revision: 0x96
miibus1: <MII bus> on vr1
ukphy1: <Generic IEEE 802.3u media interface> PHY 1 on miibus1
ukphy1:  none, 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto, auto-flow
vr1: Ethernet address: 00:00:24:cc:93:45
vr1: [ITHREAD]
vr2: <VIA VT6105M Rhine III 10/100BaseTX> port 0xe300-0xe3ff mem 0xa0004200-0xa00042ff irq 9 at device 8.0 on pci0
vr2: Quirks: 0x2
vr2: Revision: 0x96
miibus2: <MII bus> on vr2
ukphy2: <Generic IEEE 802.3u media interface> PHY 1 on miibus2
ukphy2:  none, 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto, auto-flow
vr2: Ethernet address: 00:00:24:cc:93:46
vr2: [ITHREAD]
vr3: <VIA VT6105M Rhine III 10/100BaseTX> port 0xe400-0xe4ff mem 0xa0004300-0xa00043ff irq 12 at device 9.0 on pci0
vr3: Quirks: 0x2
vr3: Revision: 0x96
miibus3: <MII bus> on vr3
ukphy3: <Generic IEEE 802.3u media interface> PHY 1 on miibus3
ukphy3:  none, 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto, auto-flow
vr3: Ethernet address: 00:00:24:cc:93:47
vr3: [ITHREAD]
isab0: <PCI-ISA bridge> at device 20.0 on pci0
isa0: <ISA bus> on isab0
atapci0: <AMD CS5536 UDMA100 controller> port 0x1f0-0x1f7,0x3f6,0x170-0x177,0x376,0xe000-0xe00f at device 20.2 on pci0
ata0: <ATA channel> at channel 0 on atapci0
ata0: [ITHREAD]
ata1: <ATA channel> at channel 1 on atapci0
ata1: [ITHREAD]
ohci0: <OHCI (generic) USB controller> mem 0xa0005000-0xa0005fff irq 15 at device 21.0 on pci0
ohci0: [ITHREAD]
usbus0 on ohci0
ehci0: <AMD CS5536 (Geode) USB 2.0 controller> mem 0xa0006000-0xa0006fff irq 15 at device 21.1 on pci0
ehci0: [ITHREAD]
usbus1: EHCI version 1.0
usbus1 on ehci0
cpu0 on motherboard
pmtimer0 on isa0
orm0: <ISA Option ROM> at iomem 0xc8000-0xd27ff pnpid ORM0000 on isa0
atkbdc0: <Keyboard controller (i8042)> at port 0x60,0x64 on isa0
atkbd0: <AT Keyboard> irq 1 on atkbdc0
kbd0 at atkbd0
atkbd0: [GIANT-LOCKED]
atkbd0: [ITHREAD]
atrtc0: <AT Real Time Clock> at port 0x70 irq 8 on isa0
ppc0: parallel port not found.
uart0: <16550 or compatible> at port 0x3f8-0x3ff irq 4 flags 0x10 on isa0
uart0: [FILTER]
uart0: console (19200,n,8,1)
uart1: <16550 or compatible> at port 0x2f8-0x2ff irq 3 on isa0
uart1: [FILTER]
Timecounter "TSC" frequency 499903982 Hz quality 800
Timecounters tick every 1.000 msec
IPsec: Initialized Security Association Processing.
usbus0: 12Mbps Full Speed USB v1.0
usbus1: 480Mbps High Speed USB v2.0
ad0: 76319MB <WDC WD800VE-00HDT0 09.07D09> at ata0-master UDMA100 
ugen0.1: <AMD> at usbus0
uhub0: <AMD OHCI root HUB, class 9/0, rev 1.00/1.00, addr 1> on usbus0
ugen1.1: <AMD> at usbus1
uhub1: <AMD EHCI root HUB, class 9/0, rev 2.00/1.00, addr 1> on usbus1
GEOM: ad0s1: geometry does not match label (255h,63s != 16h,63s).
uhub0: 4 ports with 4 removable, self powered
Root mount waiting for: usbus1
Root mount waiting for: usbus1
uhub1: 4 ports with 4 removable, self powered
Trying to mount root from ufs:/dev/ad0s1a

The last log rotation is from 2016:

[root@roadkiller /var/log]# stat /var/log/wtmp      
65 61783 -rw-r--r-- 1 root wheel 208219 1056 "Nov  1 05:00:01 2016" "Jan 18 22:29:16 2017" "Jan 18 22:29:16 2017" "Nov  1 05:00:01 2016" 16384 4 0 /var/log/wtmp

Interestingly, I switched between eicat and teksavvy on December 11th. Which year? Who knows!

Dec 11 16:38:40 roadkiller mpd: [eicatL0] LCP: authorization successful
Dec 11 16:41:15 roadkiller mpd: [teksavvyL0] LCP: authorization successful

Never realized those good old logs had a "oh dear forgot the year" issue (that's something like Y2K except just "Y", I guess). That was probably 2015, because the log dates from 2017, and the last entry is from November of the year after the above:

[root@roadkiller /var/log]# stat mpd.log 
65 47113 -rw-r--r-- 1 root wheel 193008 71939195 "Jan 18 22:39:18 2017" "Jan 18 22:39:59 2017" "Jan 18 22:39:59 2017" "Apr  2 10:41:37 2013" 16384 140640 0 mpd.log

It looks like the system was installed in 2010:

[root@roadkiller /var/log]# stat /
63 2 drwxr-xr-x 21 root wheel 2120 512 "Jan 18 22:34:43 2017" "Jan 18 22:28:12 2017" "Jan 18 22:28:12 2017" "Jul 18 22:25:00 2010" 16384 4 0 /

... so it lived for about 6 years, but still works after almost 14 years, which I find utterly amazing. Another amazing thing is that there's tuptime installed on that server! That is a software I thought I discovered later and then sponsored in Debian, but turns out I was already using it then!

[root@roadkiller /var]# tuptime 
System startups:        19   since   21:20:16 11/07/15
System shutdowns:       0 ok   -   18 bad
System uptime:          85.93 %   -   1 year, 11 days, 10 hours, 3 minutes and 36 seconds
System downtime:        14.07 %   -   61 days, 15 hours, 22 minutes and 45 seconds
System life:            1 year, 73 days, 1 hour, 26 minutes and 20 seconds
Largest uptime:         122 days, 9 hours, 17 minutes and 6 seconds   from   08:17:56 02/02/16
Shortest uptime:        5 minutes and 4 seconds   from   21:55:00 01/18/17
Average uptime:         19 days, 19 hours, 28 minutes and 37 seconds
Largest downtime:       57 days, 1 hour, 9 minutes and 59 seconds   from   20:45:01 11/22/16
Shortest downtime:      -1 years, 364 days, 23 hours, 58 minutes and 12 seconds   from   22:30:01 01/18/17
Average downtime:       3 days, 5 hours, 51 minutes and 43 seconds
Current uptime:         18 minutes and 23 seconds   since   22:28:13 01/18/17

Actual up/down times:

[root@roadkiller /var]# tuptime -t
No.        Startup Date                                         Uptime       Shutdown Date   End                                                  Downtime
1     21:20:16 11/07/15      1 day, 0 hours, 40 minutes and 12 seconds   22:00:28 11/08/15   BAD                                  2 minutes and 37 seconds
2     22:03:05 11/08/15      1 day, 9 hours, 41 minutes and 57 seconds   07:45:02 11/10/15   BAD                                  3 minutes and 24 seconds
3     07:48:26 11/10/15    20 days, 2 hours, 41 minutes and 34 seconds   10:30:00 11/30/15   BAD                        4 hours, 50 minutes and 21 seconds
4     15:20:21 11/30/15                      19 minutes and 40 seconds   15:40:01 11/30/15   BAD                                   6 minutes and 5 seconds
5     15:46:06 11/30/15                      53 minutes and 55 seconds   16:40:01 11/30/15   BAD                           1 hour, 1 minute and 38 seconds
6     17:41:39 11/30/15     6 days, 16 hours, 3 minutes and 22 seconds   09:45:01 12/07/15   BAD                4 days, 6 hours, 53 minutes and 11 seconds
7     16:38:12 12/11/15   50 days, 17 hours, 56 minutes and 49 seconds   10:35:01 01/31/16   BAD                                 10 minutes and 52 seconds
8     10:45:53 01/31/16     1 day, 21 hours, 28 minutes and 16 seconds   08:14:09 02/02/16   BAD                                  3 minutes and 48 seconds
9     08:17:56 02/02/16    122 days, 9 hours, 17 minutes and 6 seconds   18:35:02 06/03/16   BAD                                 10 minutes and 16 seconds
10    18:45:18 06/03/16   29 days, 17 hours, 14 minutes and 43 seconds   12:00:01 07/03/16   BAD                                 12 minutes and 34 seconds
11    12:12:35 07/03/16   31 days, 17 hours, 17 minutes and 26 seconds   05:30:01 08/04/16   BAD                                 14 minutes and 25 seconds
12    05:44:26 08/04/16     15 days, 1 hour, 55 minutes and 35 seconds   07:40:01 08/19/16   BAD                                  6 minutes and 51 seconds
13    07:46:52 08/19/16     7 days, 5 hours, 23 minutes and 10 seconds   13:10:02 08/26/16   BAD                                  3 minutes and 45 seconds
14    13:13:47 08/26/16   27 days, 21 hours, 36 minutes and 14 seconds   10:50:01 09/23/16   BAD                                  2 minutes and 14 seconds
15    10:52:15 09/23/16   60 days, 10 hours, 52 minutes and 46 seconds   20:45:01 11/22/16   BAD                 57 days, 1 hour, 9 minutes and 59 seconds
16    21:55:00 01/18/17                        5 minutes and 4 seconds   22:00:04 01/18/17   BAD                                 11 minutes and 15 seconds
17    22:11:19 01/18/17                       8 minutes and 42 seconds   22:20:01 01/18/17   BAD                                   1 minute and 20 seconds
18    22:21:21 01/18/17                       8 minutes and 40 seconds   22:30:01 01/18/17   BAD   -1 years, 364 days, 23 hours, 58 minutes and 12 seconds
19    22:28:13 01/18/17                      20 minutes and 17 seconds

The last few entries are actually the tests I'm running now, it seems this machine thinks we're now on 2017-01-18 at ~22:00, while we're actually 2024-01-24 at ~12:00 local:

Wed Jan 18 23:05:38 EST 2017
FreeBSD/i386 (roadkiller.anarc.at) (ttyu0)
login: root
Password:
Jan 18 23:07:10 roadkiller login: ROOT LOGIN (root) ON ttyu0
Last login: Wed Jan 18 22:29:16 on ttyu0
Copyright (c) 1992-2013 The FreeBSD Project.
Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994
        The Regents of the University of California. All rights reserved.
FreeBSD 8.4-RELEASE-p12 (ROADKILL) #5: Fri Jun  6 02:43:23 EDT 2014
Reminders:
 * commit stuff in /etc
 * reload firewall (in screen!):
    pfctl -f /etc/pf.conf ; sleep 1
 * vim + syn on makes pf.conf more readable
 * monitoring the PPPoE uplink:
   tail -f /var/log/mpd.log
Current problems:
 * sometimes pf doesn't start properly on boot, if pppoe failed to come up, use
   this to resume:
     /etc/rc.d/pf start
   it will kill your shell, but fix NAT (2012-08-10)
 * babel fails to start on boot (2013-06-15):
     babeld -D -g 33123 tap0 vr3
 * DNS often fails, tried messing with unbound.conf (2014-10-05) and updating
   named.root (2016-01-28) and performance tweaks (ee63689)
 * asterisk and mpd4 are deprecated and should be uninstalled when we're sure
   their replacements (voipms + ata and mpd5) are working (2015-01-13)
 * if IPv6 fails, it's because netblocks are not being routed upstream. DHCPcd
   should do this, but doesn't start properly, use this to resume (2015-12-21):
     /usr/local/sbin/dhcpcd -6 --persistent --background --timeout 0 -C resolv.conf ng0
This machine is doomed to be replaced with the new omnia router, Indiegogo
campaign should ship in april 2016: http://igg.me/at/turris-omnia/x

(I really like the motd I left myself there. In theory, I guess this could just start connecting to the internet again if I still had the same PPPoE/ADSL link I had almost a decade ago; obviously, I do not.) Not sure how the system figured the 2017 time: the onboard clock itself believes we're in 1980, so clearly the CMOS battery has (understandably) failed:

> ?
comBIOS Monitor Commands
boot [drive][:partition] INT19 Boot
reboot                   cold boot
download                 download a file using XMODEM/CRC
flashupdate              update flash BIOS with downloaded file
time [HH:MM:SS]          show or set time
date [YYYY/MM/DD]        show or set date
d[b w d] [adr]           dump memory bytes/words/dwords
e[b w d] adr value [...] enter bytes/words/dwords
i[b w d] port            input from 8/16/32-bit port
o[b w d] port value      output to 8/16/32-bit port
run adr                  execute code at adr
cmosread [adr]           read CMOS RAM data
cmoswrite adr byte [...] write CMOS RAM data
cmoschecksum             update CMOS RAM Checksum
set parameter=value      set system parameter to value
show [parameter]         show one or all system parameters
?/help                   show this help
> show
ConSpeed = 19200
ConLock = Enabled
ConMute = Disabled
BIOSentry = Enabled
PCIROMS = Enabled
PXEBoot = Enabled
FLASH = Primary
BootDelay = 5
FastBoot = Disabled
BootPartition = Disabled
BootDrive = 80 81 F0 FF 
ShowPCI = Enabled
Reset = Hard
CpuSpeed = Default
> time
Current Date and Time is: 1980/01/01 00:56:47

Another bit of archeology: I had documented various outages with my ISP... back in 2003!

[root@roadkiller ~/bin]# cat ppp_stats/downtimes.txt
11/03/2003 18:24:49 218
12/03/2003 09:10:49 118
12/03/2003 10:05:57 680
12/03/2003 10:14:50 106
12/03/2003 10:16:53 6
12/03/2003 10:35:28 146
12/03/2003 10:57:26 393
12/03/2003 11:16:35 5
12/03/2003 11:16:54 11
13/03/2003 06:15:57 18928
13/03/2003 09:43:36 9730
13/03/2003 10:47:10 23
13/03/2003 10:58:35 5
16/03/2003 01:32:36 338
16/03/2003 02:00:33 120
16/03/2003 11:14:31 14007
19/03/2003 00:56:27 11179
19/03/2003 00:56:43 5
19/03/2003 00:56:53 0
19/03/2003 00:56:55 1
19/03/2003 00:57:09 1
19/03/2003 00:57:10 1
19/03/2003 00:57:24 1
19/03/2003 00:57:25 1
19/03/2003 00:57:39 1
19/03/2003 00:57:40 1
19/03/2003 00:57:44 3
19/03/2003 00:57:53 0
19/03/2003 00:57:55 0
19/03/2003 00:58:08 0
19/03/2003 00:58:10 0
19/03/2003 00:58:23 0
19/03/2003 00:58:25 0
19/03/2003 00:58:39 1
19/03/2003 00:58:42 2
19/03/2003 00:58:58 5
19/03/2003 00:59:35 2
19/03/2003 00:59:47 3
19/03/2003 01:00:34 3
19/03/2003 01:00:39 0
19/03/2003 01:00:54 0
19/03/2003 01:01:11 2
19/03/2003 01:01:25 1
19/03/2003 01:01:48 1
19/03/2003 01:02:03 1
19/03/2003 01:02:10 2
19/03/2003 01:02:20 3
19/03/2003 01:02:44 3
19/03/2003 01:03:45 3
19/03/2003 01:04:39 2
19/03/2003 01:05:40 2
19/03/2003 01:06:35 2
19/03/2003 01:07:36 2
19/03/2003 01:08:31 2
19/03/2003 01:08:38 2
19/03/2003 01:10:07 3
19/03/2003 01:11:05 2
19/03/2003 01:12:03 3
19/03/2003 01:13:01 3
19/03/2003 01:13:58 2
19/03/2003 01:14:59 5
19/03/2003 01:15:54 2
19/03/2003 01:16:55 2
19/03/2003 01:17:50 2
19/03/2003 01:18:51 3
19/03/2003 01:19:46 2
19/03/2003 01:20:46 2
19/03/2003 01:21:42 3
19/03/2003 01:22:42 3
19/03/2003 01:23:37 2
19/03/2003 01:24:38 3
19/03/2003 01:25:33 2
19/03/2003 01:26:33 2
19/03/2003 01:27:30 3
19/03/2003 01:28:55 2
19/03/2003 01:29:56 2
19/03/2003 01:30:50 2
19/03/2003 01:31:42 3
19/03/2003 01:32:36 3
19/03/2003 01:33:27 2
19/03/2003 01:34:21 2
19/03/2003 01:35:22 2
19/03/2003 01:36:17 3
19/03/2003 01:37:18 2
19/03/2003 01:38:13 3
19/03/2003 01:39:39 2
19/03/2003 01:40:39 2
19/03/2003 01:41:35 3
19/03/2003 01:42:35 3
19/03/2003 01:43:31 3
19/03/2003 01:44:31 3
19/03/2003 01:45:53 3
19/03/2003 01:46:48 3
19/03/2003 01:47:48 2
19/03/2003 01:48:44 3
19/03/2003 01:49:44 2
19/03/2003 01:50:40 3
19/03/2003 01:51:39 1
19/03/2003 11:04:33 19   
19/03/2003 18:39:36 2833 
19/03/2003 18:54:05 825  
19/03/2003 19:04:00 454  
19/03/2003 19:08:11 210  
19/03/2003 19:41:44 272  
19/03/2003 21:18:41 208  
24/03/2003 04:51:16 6
27/03/2003 04:51:20 5
30/03/2003 04:51:25 5
31/03/2003 08:30:31 255  
03/04/2003 08:30:36 5
06/04/2003 01:16:00 621  
06/04/2003 22:18:08 17   
06/04/2003 22:32:44 13   
09/04/2003 22:33:12 28   
12/04/2003 22:33:17 6
15/04/2003 22:33:22 5
17/04/2003 15:03:43 18   
20/04/2003 15:03:48 5
23/04/2003 15:04:04 16   
23/04/2003 21:08:30 339  
23/04/2003 21:18:08 13   
23/04/2003 23:34:20 253  
26/04/2003 23:34:45 25   
29/04/2003 23:34:49 5
02/05/2003 13:10:01 185  
05/05/2003 13:10:06 5
08/05/2003 13:10:11 5
09/05/2003 14:00:36 63928
09/05/2003 16:58:52 2
11/05/2003 23:08:48 2
14/05/2003 23:08:53 6
17/05/2003 23:08:58 5
20/05/2003 23:09:03 5
23/05/2003 23:09:08 5
26/05/2003 23:09:14 5
29/05/2003 23:00:10 3
29/05/2003 23:03:01 10   
01/06/2003 23:03:05 4
04/06/2003 23:03:10 5
07/06/2003 23:03:38 28   
10/06/2003 23:03:50 12   
13/06/2003 23:03:55 6
14/06/2003 07:42:20 3
14/06/2003 14:37:08 3
15/06/2003 20:08:34 3
18/06/2003 20:08:39 6
21/06/2003 20:08:45 6
22/06/2003 03:05:19 138  
22/06/2003 04:06:28 3
25/06/2003 04:06:58 31   
28/06/2003 04:07:02 4
01/07/2003 04:07:06 4
04/07/2003 04:07:11 5
07/07/2003 04:07:16 5
12/07/2003 04:55:20 6
12/07/2003 19:09:51 1158 
12/07/2003 22:14:49 8025 
15/07/2003 22:14:54 6
16/07/2003 05:43:06 18   
19/07/2003 05:43:12 6
22/07/2003 05:43:17 5
23/07/2003 18:18:55 183  
23/07/2003 18:19:55 9
23/07/2003 18:29:15 158  
23/07/2003 19:48:44 4604 
23/07/2003 20:16:27 3
23/07/2003 20:37:29 1079 
23/07/2003 20:43:12 342  
23/07/2003 22:25:51 6158

Fascinating. I suspect the (IDE!) hard drive might be failing as I saw two new files created in /var that I didn't remember seeing before:

-rw-r--r--   1 root    wheel        0 Jan 18 22:55 3@T3
-rw-r--r--   1 root    wheel        0 Jan 18 22:55 DY5

So I shutdown the machine, possibly for the last time:

Waiting (max 60 seconds) for system process  bufdaemon' to stop...done
Waiting (max 60 seconds) for system process  syncer' to stop...
Syncing disks, vnodes remaining...3 3 0 1 1 0 0 done
All buffers synced.
Uptime: 36m43s
usbus0: Controller shutdown
uhub0: at usbus0, port 1, addr 1 (disconnected)
usbus0: Controller shutdown complete
usbus1: Controller shutdown
uhub1: at usbus1, port 1, addr 1 (disconnected)
usbus1: Controller shutdown complete
The operating system has halted.
Please press any key to reboot.

I'll finally note this was the last FreeBSD server I personally operated. I also used FreeBSD to setup the core routers at Koumbit but those were replaced with Debian recently as well. Thanks Soekris, that was some sturdy hardware. Hopefully this new Protectli router will live up to that "decade plus" challenge. Not sure what the fate of this device will be: I'll bring it to the next Montreal Debian & Stuff to see if anyone's interested, contact me if you can't show up and want this thing.

If you ve perused the ActivityPub feed of certificates whose keys are known to be compromised, and clicked on the Show More button to see the name of the certificate issuer, you may have noticed that some issuers seem to come up again and again. This might make sense after all, if a CA is issuing a large volume of certificates, they ll be seen more often in a list of compromised certificates. In an attempt to see if there is anything that we can learn from this data, though, I did a bit of digging, and came up with some illuminating results.

The Procedure I started off by finding all the unexpired certificates logged in Certificate Transparency (CT) logs that have a key that is in the pwnedkeys database as having been publicly disclosed. From this list of certificates, I removed duplicates by matching up issuer/serial number tuples, and then reduced the set by counting the number of unique certificates by their issuer. This gave me a list of the issuers of these certificates, which looks a bit like this:

/C=BE/O=GlobalSign nv-sa/CN=AlphaSSL CA - SHA256 - G4
/C=GB/ST=Greater Manchester/L=Salford/O=Sectigo Limited/CN=Sectigo RSA Domain Validation Secure Server CA
/C=GB/ST=Greater Manchester/L=Salford/O=Sectigo Limited/CN=Sectigo RSA Organization Validation Secure Server CA
/C=US/ST=Arizona/L=Scottsdale/O=GoDaddy.com, Inc./OU=http://certs.godaddy.com/repository//CN=Go Daddy Secure Certificate Authority - G2
/C=US/ST=Arizona/L=Scottsdale/O=Starfield Technologies, Inc./OU=http://certs.starfieldtech.com/repository//CN=Starfield Secure Certificate Authority - G2
/C=AT/O=ZeroSSL/CN=ZeroSSL RSA Domain Secure Site CA
/C=BE/O=GlobalSign nv-sa/CN=GlobalSign GCC R3 DV TLS CA 2020

Rather than try to work with raw issuers (because, as Andrew Ayer says, The SSL Certificate Issuer Field is a Lie), I mapped these issuers to the organisations that manage them, and summed the counts for those grouped issuers together.

The Data

Insert obligatory "not THAT data" comment here

The end result of this work is the following table, sorted by the count of certificates which have been compromised by exposing their private key:

Issuer	Compromised Count
Sectigo	170
ISRG (Let's Encrypt)	161
GoDaddy	141
DigiCert	81
GlobalSign	46
Entrust	3
SSL.com	1

If you re familiar with the CA ecosystem, you ll probably recognise that the organisations with large numbers of compromised certificates are also those who issue a lot of certificates. So far, nothing particularly surprising, then. Let s look more closely at the relationships, though, to see if we can get more useful insights.

Volume Control Using the issuance volume report from crt.sh, we can compare issuance volumes to compromise counts, to come up with a compromise rate . I m using the Unexpired Precertificates colume from the issuance volume report, as I feel that s the number that best matches the certificate population I m examining to find compromised certificates. To maintain parity with the previous table, this one is still sorted by the count of certificates that have been compromised.

Issuer	Issuance Volume	Compromised Count	Compromise Rate
Sectigo	88,323,068	170	1 in 519,547
ISRG (Let's Encrypt)	315,476,402	161	1 in 1,959,480
GoDaddy	56,121,429	141	1 in 398,024
DigiCert	144,713,475	81	1 in 1,786,586
GlobalSign	1,438,485	46	1 in 31,271
Entrust	23,166	3	1 in 7,722
SSL.com	171,816	1	1 in 171,816

If we now sort this table by compromise rate, we can see which organisations have the most (and least) leakiness going on from their customers:

Issuer	Issuance Volume	Compromised Count	Compromise Rate
Entrust	23,166	3	1 in 7,722
GlobalSign	1,438,485	46	1 in 31,271
SSL.com	171,816	1	1 in 171,816
GoDaddy	56,121,429	141	1 in 398,024
Sectigo	88,323,068	170	1 in 519,547
DigiCert	144,713,475	81	1 in 1,786,586
ISRG (Let's Encrypt)	315,476,402	161	1 in 1,959,480

By grouping by order-of-magnitude in the compromise rate, we can identify three bands :

• The Super Leakers: Customers of Entrust and GlobalSign seem to love to lose control of their private keys. For Entrust, at least, though, the small volumes involved make the numbers somewhat untrustworthy. The three compromised certificates could very well belong to just one customer, for instance. I m not aware of anything that GlobalSign does that would make them such an outlier, either, so I m inclined to think they just got unlucky with one or two customers, but as CAs don t include customer IDs in the certificates they issue, it s not possible to say whether that s the actual cause or not.
• The Regular Leakers: Customers of SSL.com, GoDaddy, and Sectigo all have compromise rates in the 1-in-hundreds-of-thousands range. Again, the low volumes of SSL.com make the numbers somewhat unreliable, but the other two organisations in this group have large enough numbers that we can rely on that data fairly well, I think.
• The Low Leakers: Customers of DigiCert and Let s Encrypt are at least three times less likely than customers of the regular leakers to lose control of their private keys. Good for them!

Now we have some useful insights we can think about.

Why Is It So?

If you don't know who Professor Julius Sumner Miller is, I highly recommend finding out

All of the organisations on the list, with the exception of Let s Encrypt, are what one might term traditional CAs. To a first approximation, it s reasonable to assume that the vast majority of the customers of these traditional CAs probably manage their certificates the same way they have for the past two decades or more. That is, they generate a key and CSR, upload the CSR to the CA to get a certificate, then copy the cert and key somewhere. Since humans are handling the keys, there s a higher risk of the humans using either risky practices, or making a mistake, and exposing the private key to the world. Let s Encrypt, on the other hand, issues all of its certificates using the ACME (Automatic Certificate Management Environment) protocol, and all of the Let s Encrypt documentation encourages the use of software tools to generate keys, issue certificates, and install them for use. Given that Let s Encrypt has 161 compromised certificates currently in the wild, it s clear that the automation in use is far from perfect, but the significantly lower compromise rate suggests to me that lifecycle automation at least reduces the rate of key compromise, even though it doesn t eliminate it completely.

Sidebar: ACME Does Not Currently Rule The World It is true that all of the organisations in this analysis also provide ACME issuance workflows, should customers desire it. However, the traditional CA companies have been around a lot longer than ACME has, and so they probably acquired many of their customers before ACME existed. Given that it s incredibly hard to get humans to change the way they do things, once they have a way that works , it seems reasonable to assume that most of the certificates issued by these CAs are handled in the same human-centric, error-prone manner they always have been. If organisations would like to refute this assumption, though, by sharing their data on ACME vs legacy issuance rates, I m sure we d all be extremely interested.

Explaining the Outlier The difference in presumed issuance practices would seem to explain the significant difference in compromise rates between Let s Encrypt and the other organisations, if it weren t for one outlier. This is a largely traditional CA, with the manual-handling issues that implies, but with a compromise rate close to that of Let s Encrypt. We are, of course, talking about DigiCert. The thing about DigiCert, that doesn t show up in the raw numbers from crt.sh, is that DigiCert manages the issuance of certificates for several of the biggest hosted TLS providers, such as CloudFlare and AWS. When these services obtain a certificate from DigiCert on their customer s behalf, the private key is kept locked away, and no human can (we hope) get access to the private key. This is supported by the fact that no certificates identifiably issued to either CloudFlare or AWS appear in the set of certificates with compromised keys. When we ask for all certificates issued by DigiCert , we get both the certificates issued to these big providers, which are very good at keeping their keys under control, as well as the certificates issued to everyone else, whose key handling practices may not be quite so stringent. It s possible, though not trivial, to account for certificates issued to these hosted TLS providers, because the certificates they use are issued from intermediates branded to those companies. With the crt.sh psql interface we can run this query to get the total number of unexpired precertificates issued to these managed services:

SELECT SUM(sub.NUM_ISSUED[2] - sub.NUM_EXPIRED[2])
  FROM (
    SELECT ca.name, max(coalesce(coalesce(nullif(trim(cc.SUBORDINATE_CA_OWNER), ''), nullif(trim(cc.CA_OWNER), '')), cc.INCLUDED_CERTIFICATE_OWNER)) as OWNER,
           ca.NUM_ISSUED, ca.NUM_EXPIRED
      FROM ccadb_certificate cc, ca_certificate cac, ca
     WHERE cc.CERTIFICATE_ID = cac.CERTIFICATE_ID
       AND cac.CA_ID = ca.ID
  GROUP BY ca.ID
  ) sub
 WHERE sub.name ILIKE '%Amazon%' OR sub.name ILIKE '%CloudFlare%' AND sub.owner = 'DigiCert';

The number I get from running that query is 104,316,112, which should be subtracted from DigiCert s total issuance figures to get a more accurate view of what DigiCert s regular customers do with their private keys. When I do this, the compromise rates table, sorted by the compromise rate, looks like this:

Issuer	Issuance Volume	Compromised Count	Compromise Rate
Entrust	23,166	3	1 in 7,722
GlobalSign	1,438,485	46	1 in 31,271
SSL.com	171,816	1	1 in 171,816
GoDaddy	56,121,429	141	1 in 398,024
"Regular" DigiCert	40,397,363	81	1 in 498,732
Sectigo	88,323,068	170	1 in 519,547
All DigiCert	144,713,475	81	1 in 1,786,586
ISRG (Let's Encrypt)	315,476,402	161	1 in 1,959,480

In short, it appears that DigiCert s regular customers are just as likely as GoDaddy or Sectigo customers to expose their private keys.

What Does It All Mean? The takeaway from all this is fairly straightforward, and not overly surprising, I believe.

The less humans have to do with certificate issuance, the less likely they are to compromise that certificate by exposing the private key. While it may not be surprising, it is nice to have some empirical evidence to back up the common wisdom. Fully-managed TLS providers, such as CloudFlare, AWS Certificate Manager, and whatever Azure s thing is called, is the platonic ideal of this principle: never give humans any opportunity to expose a private key. I m not saying you should use one of these providers, but the security approach they have adopted appears to be the optimal one, and should be emulated universally. The ACME protocol is the next best, in that there are a variety of standardised tools widely available that allow humans to take themselves out of the loop, but it s still possible for humans to handle (and mistakenly expose) key material if they try hard enough. Legacy issuance methods, which either cannot be automated, or require custom, per-provider automation to be developed, appear to be at least four times less helpful to the goal of avoiding compromise of the private key associated with a certificate.

Humans Are, Of Course, The Problem

No thanks, Bender, I'm busy tonight

This observation that if you don t let humans near keys, they don t get leaked is further supported by considering the biggest issuers by volume who have not issued any certificates whose keys have been compromised: Google Trust Services (fourth largest issuer overall, with 57,084,529 unexpired precertificates), and Microsoft Corporation (sixth largest issuer overall, with 22,852,468 unexpired precertificates). It appears that somewhere between most and basically all of the certificates these organisations issue are to customers of their public clouds, and my understanding is that the keys for these certificates are managed in same manner as CloudFlare and AWS the keys are locked away where humans can t get to them. It should, of course, go without saying that if a human can never have access to a private key, it makes it rather difficult for a human to expose it. More broadly, if you are building something that handles sensitive or secret data, the more you can do to keep humans out of the loop, the better everything will be.

Your Support is Appreciated If you d like to see more analysis of how key compromise happens, and the lessons we can learn from examining billions of certificates, please show your support by buying me a refreshing beverage. Trawling CT logs is thirsty work.

Appendix: Methodology Limitations In the interests of clarity, I feel it s important to describe ways in which my research might be flawed. Here are the things I know of that may have impacted the accuracy, that I couldn t feasibly account for.

• Time Periods: Because time never stops, there is likely to be some slight mismatches in the numbers obtained from the various data sources, because they weren t collected at exactly the same moment.
• Issuer-to-Organisation Mapping: It s possible that the way I mapped issuers to organisations doesn t match exactly with how crt.sh does it, meaning that counts might be skewed. I tried to minimise that by using the same data sources (the CCADB AllCertificates report) that I believe that crt.sh uses for its mapping, but I cannot be certain of a perfect match.
• Unwarranted Grouping: I ve drawn some conclusions about the practices of the various organisations based on their general approach to certificate issuance. If a particular subordinate CA that I ve grouped into the parent organisation is managed in some unusual way, that might cause my conclusions to be erroneous. I was able to fairly easily separate out CloudFlare, AWS, and Azure, but there are almost certainly others that I didn t spot, because hoo boy there are a lot of intermediate CAs out there.

In my previous blog post Providing online reference documentation for debputy, I made a point about how debhelper documentation was suboptimal on account of being static rather than online. The thing is that debhelper is not alone in this problem space, even if it is a major contributor to the number of packaging files you have to to know about. If we look at the "competition" here such as Fedora and Arch Linux, they tend to only have one packaging file. While most Debian people will tell you a long list of cons about having one packaging file (such a Fedora's spec file being 3+ domain specific languages "mashed" into one file), one major advantage is that there is only "the one packaging file". You only need to remember where to find the documentation for one file, which is great when you are running on wetware with limited storage capacity. Which means as a newbie, you can dedicate less mental resources to tracking multiple files and how they interact and more effort understanding the "one file" at hand. I started by asking myself how can we in Debian make the packaging stack more accessible to newcomers? Spoiler alert, I dug myself into rabbit hole and ended up somewhere else than where I thought I was going. I started by wanting to scan the debian directory and annotate all files that I could with documentation links. The logic was that if debputy could do that for you, then you could spend more mental effort elsewhere. So I combined debputy's packager provided files detection with a static list of files and I quickly had a good starting point for debputy-based packages.

After more than one a year, a new minor FAI version is available, but it includes some interesting new features. Here a the items from the NEWS file: fai (6.2) unstable; urgency=low

• fai-cd can now create live images
• Use systemd during installation
• New feature: run FAI inside a screen or tmux session
• fai-diskimage: do not use compression of qemu-img which is slow instead provide .qcow2.zst, add option -C
• fai-kvm: add support for booting from USB storage
• new tool mk-data-partition adds a data partition to an ISO
• easy installation of packages from /pkgs/<CLASS> directories
• new helper functions for creating custom list of disks
• new method detect:// for FAI_CONFIG_SRC

In the past the command fai-cd was only used for creating installation ISOs, that could be used from CD or USB stick. Now it possible to create a live ISO. Therefore you create your live chroot environment using 'fai dirinstall' and then convert it to a bootable live ISO using fai-cd. See man fai-cd(8) for an example. Years ago I had the idea to use the remaining disk space on an USB stick after copying an ISO onto it. I've blogged about this recently: https://blog.fai-project.org/posts/extending-iso-images/ The new FAI version includes the tool mk-data-partition for adding a data partition to the ISO itself or to an USB stick. FAI detects this data partition, mounts it to /media/data and can then use various configurations from it. You may want to copy your own set of .deb packages or your whole FAI config space to this partition. FAI now automatically searches this partition for usable FAI configuration data and packages. FAI will install all packages from pkgs/<CLASSNAME> if the equivalent class is defined. Setting FAI_CONFIG_SRC=detect:// now looks into the data partition for the subdirectory 'config' and uses this as the config space. So it's now possible to modify an existing ISO (that is read-only) and make changes to the config space. If there's no config directory in the data partition FAI uses the default location on the ISO. The tool fai-kvm, which starts virtual machines can now boot an ISO not only as CD but also as USB stick. Sometimes users want to adjust the list of disks before the partitioning is startet. Therefore FAI provides several new functions including

• smallestdisk()
• largestdisk()
• matchdisks()

You can select individual disks by their model name or even the serial number. Two new FAI flags were added (tmux and screen) that make it easy to run FAI inside a tmux or screen session. And finally FAI uses systemd. Yeah! This technical change was waiting since 2015 in a merge request from Moritz 'Morty' Str be, that would enable using systemd during the installation. Before FAI still was using old-style SYSV init scripts and did not started systemd. I didn't tried to apply the patch, because I was afraid that it would need much time to make it work. But then in may 2023 Juri Grabowski just gave it a try at MiniDebConf Hamburg, and voil it just works! Many, many thanks to Moritz and Juri for their bravery. The whole changelog can be found at https://tracker.debian.org/media/packages/f/fai/changelog-6.2 New ISOs for FAI are also available including an example of a Xfce desktop live ISO: https://fai-project.org/fai-cd/ The FAIme service for creating customized installation ISOs will get its update later. The new packages are available for bookworm by adding this line to your sources.list: deb https://fai-project.org/download bookworm koeln

I ve been using rsync for years and still did not know its full powers. I just wanted a quick and dirty simple backup but realised that rsnapshot is not in Debian anymore. However you can do much of rsnapshot with rsync alone nowadays. The --link-dest option (manpage) solves the part of creating hardlinks to a previous backup (found here). So my backup program becomes this shell script in ~/backups/backup.sh:

#!/bin/sh
SERVER="$ 1 "
BACKUP="$ HOME /backups/$ SERVER "
SNAPSHOTS="$ BACKUP /snapshots"
FOLDER=$(date --utc +%F_%H-%M-%S)
DEST="$ SNAPSHOTS /$ FOLDER "
LAST=$(ls -d1 $ SNAPSHOTS /????-??-??_??-??-?? tail -n 1)
rsync \
  --rsh="ssh -i $ BACKUP /sshkey -o ControlPath=none -o ForwardAgent=no" \
  -rlpt \
  --delete --link-dest="$ LAST " \
  $ SERVER ::backup "$ DEST "

The script connects to rsync in daemon mode as outlined in section USING RSYNC-DAEMON FEATURES VIA A REMOTE-SHELL CONNECTION in the rsync manpage. This allows to reference a module as the source that is defined on the server side as follows:

[backup]
path = /
read only = true
exclude from = /srv/rsyncbackup/excludelist
uid = root
gid = root

The important bit is the read only setting that protects the server against somebody with access to the ssh key to overwrit files on the server via rsync and thus gaining full root access. Finally the command prefix in ~/.ssh/authorized_keys runs rsync as daemon with sudo and the specified config file:

command="sudo rsync --config=/srv/rsyncbackup/config --server --daemon ."

The sudo setup is left as an exercise for the reader as mine is rather opinionated. Unfortunately I have not managed to configure systemd timers in the way I wanted and therefor opened an issue: Allow retry of timer triggered oneshot services with failed conditions or asserts . Any help there is welcome!

Using NetworkManager and systemd-resolved together in Debian bookworm does not work out of the box. The first sign of trouble was these constant messages in my logs:

avahi-daemon[pid]: Host name conflict, retrying with hostname-2

Then I realized that CUPS printer discovery didn't work: my network printer could not be found. Since this discovery now relies on Multicast DNS, it would make sense that both problems are related to an incompatibility between NetworkManager and Avahi.

What didn't work The first attempt I made at fixing this was to look for known bugs in Avahi. Neither of the work-arounds I found worked:

• the one proposed in https://github.com/avahi/avahi/issues/117#issuecomment-1651475104:

    [publish]
    publish-aaaa-on-ipv4=no
    publish-a-on-ipv6=no
  

• nor the one proposed in https://github.com/avahi/avahi/issues/117#issuecomment-442201162:

    [server]
    cache-entries-max=0
  

What worked The real problem turned out to be the fact that NetworkManager turns on full mDNS support in systemd-resolved which conflicts with the mDNS support in avahi-daemon. You can see this in the output of resolvectl status:

Global
       Protocols: -LLMNR +mDNS -DNSOverTLS DNSSEC=no/unsupported
resolv.conf mode: stub
Link 2 (enp6s0)
    Current Scopes: DNS mDNS/IPv4 mDNS/IPv6
         Protocols: +DefaultRoute -LLMNR +mDNS -DNSOverTLS
                    DNSSEC=no/unsupported
Current DNS Server: 192.168.1.1
       DNS Servers: 192.168.1.1
        DNS Domain: lan

which includes +mDNS for the main network adapter. I initially thought that I could just uninstall avahi-daemon and rely on the systemd-resolved mDNS stack, but it's not actually compatible with CUPS. The solution was to tell NetworkManager to set mDNS to resolve-only mode in systemd-resolved by adding the following to /etc/NetworkManager/conf.d/mdns.conf:

[connection]
connection.mdns=1

leaving /etc/avahi/avahi-daemon.conf to the default Debian configuration.

Verifying the configuration After rebooting, resolvectl status now shows the following:

Global
       Protocols: -LLMNR +mDNS -DNSOverTLS DNSSEC=no/unsupported
resolv.conf mode: stub
Link 2 (enp6s0)
    Current Scopes: DNS mDNS/IPv4 mDNS/IPv6
         Protocols: +DefaultRoute -LLMNR mDNS=resolve -DNSOverTLS
                    DNSSEC=no/unsupported
Current DNS Server: 192.168.1.1
       DNS Servers: 192.168.1.1
        DNS Domain: lan

Avahi finally sees my printer (called hp in the output below):

$ avahi-browse -at   grep Printer
+ enp6s0 IPv6 hp @ myprintserver   Secure Internet Printer local
+ enp6s0 IPv4 hp @ myprintserver   Secure Internet Printer local
+ enp6s0 IPv6 hp @ myprintserver   Internet Printer        local
+ enp6s0 IPv4 hp @ myprintserver   Internet Printer        local
+ enp6s0 IPv6 hp @ myprintserver   UNIX Printer            local
+ enp6s0 IPv4 hp @ myprintserver   UNIX Printer            local

and so does CUPS:

$ sudo lpinfo --include-schemes dnssd -v
network dnssd://myprintserver%20%40%20hp._ipp._tcp.local/cups?uuid=d46942a2-b730-11ee-b05c-a75251a34287

Firewall rules Since printer discovery in CUPS relies on mDNS, another thing to double-check is that the correct ports are open on the firewall. This is what I have in /etc/network/iptables.up.rules:

# Allow mDNS for local service discovery
-A INPUT -d 100.64.0.0/10 -p udp --dport 5353 -j ACCEPT
-A INPUT -d 192.168.1.0/24 -p udp --dport 5353 -j ACCEPT

and in etc/network/ip6tables.up.rules:

# Allow mDNS for local service discovery
-A INPUT -d ff02::/16 -p udp --dport 5353 -j ACCEPT

I just decided to upgrade the core of my home network from 1Gbit to 2.5Gbit. I didn t really need to do this, it was only about 5 years ago that I upgrade from 100Mbit to 1Gbit. but it s cheap and seemed interesting. I decided to do it because a 2.5Gbit switch was listed as cheap on Ozbargain Computing [1], that was $40.94 delivered. If you are in Australia and like computers then Ozbargain is a site worth polling, every day there s interesting things at low prices. The seller of the switch is KeeplinkStore [2] who distinguished themselves by phoning me from China to inform me that I had ordered a switch with a UK plug for delivery to Australia and suggesting that I cancel the order and make a new order with an Australian plug. It wouldn t have been a big deal if I had received a UK plug as I ve got a collection of adaptors but it was still nice of them to make it convenient for me. The switch basically does what it s expected to do and has no fan so it s quiet. I got a single port 2.5Gbit PCIe card for $18.77 and a dual port card for $34.07. Those cards are a little expensive when compared to 1Gbit cards but very cheap when compared to the computers they are installed in. These cards use the Realtek RTL8125 chipset and work well. I got a USB-3 2.5Gbit device for $17.43. I deliberately didn t get USB-C because I still use laptops without USB-C and most of the laptops with USB-C only have a single USB-C port which is used for power. I don t plan to stop using my 100Mbit USB ethernet device because most of the time I don t need a lot of speed. But sometimes I do things like testing auto-install on laptops and then having something faster than Gigabit is good. This card worked at 1Gbit speed on a 1Gbit network when used with a system running Debian/Bookworm with kernel 6.1 and worked at 2.5Gbit speed when connected to my LicheePi RISC-V system running Linux 5.10, but it would only do 100Mbit on my laptop running Debian/Unstable with kernel 6.6 (Debian Bug #1061095) [3]. It s a little disappointing but not many people have such hardware so it probably doesn t get a lot of testing. For the moment I plan to just use a 1Gbit USB Ethernet device most of the time and if I really need the speed I ll just use an older kernel. I did some tests with wget and curl to see if I could get decent speeds. When using wget 1.21.3 on Debian/Bookworm I got transfer speeds of 103MB/s and 18.8s of system CPU time out of 23.6s of elapsed time. Curl on Debian/Bookworm did 203MB/s and took 10.7s of system CPU time out of 11.8s elapsed time. The difference is that curl was using 100KB read buffers and a mix of 12K and 4K write buffers while wget was using 8KB read buffers and 4KB write buffers. On Debian/Unstable wget 1.21.4 uses 64K read buffers and a mix of 4K and 60K write buffers and gets a speed of 208MB/s. As an experiment I changed the read buffer size for wget to 256K and that got the speed up to around 220MB/s but it was difficult to measure as the occasional packet loss slowed things down. The pattern of writing 4K and then writing the rest continued, it seemed related to fwrite() buffering. For anyone else who wants to experiment with the code, the wget code is simpler (due to less features) and the package builds a lot faster (due to fewer tests) so that s the one to work on. The client machine for these tests has a E5-2696 v3 CPU, this doesn t compare well to some of the recent AMD CPUs on single-core performance but is still a decently powerful system. Getting good performance at Gigabit speeds on an ARM or RISC-V system is probably going to be a lot harder than getting good performance at 2.5Gbit speeds on this system. In conclusion 2.5Gbit basically works apart from a problem with new kernels and a problem with the old version of wget. I expect that when Debian/Trixie is released (probably mid 2025) things will work well. For good transfer rates use wget version 1.21.4 or newer or use curl. As an aside I use a 1500byte MTU because I have some 100baseT systems on my LAN and the settings regarding TCP acceleration etc are all the defaults.

• [1] https://www.ozbargain.com.au/cat/computing
• [2] https://keeplink.aliexpress.com/store/911812257
• [3] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1061095

I Just bought a LicheePi 4A RISC-V embedded computer (like a RaspberryPi but with a RISC-V CPU) for $322.68 from Aliexpress (the official site for buying LicheePi devices). Here is the Sipheed web page about it and their other recent offerings [1]. I got the version with 16G of RAM and 128G of storage, I probably don t need that much storage (I can use NFS or USB) but 16G of RAM is good for VMs. Here is the Wiki about this board [2]. Configuration When you get one of these devices you should make setting up ssh server your first priority. I found the HDMI output to be very unreliable. The first monitor I tried was a Samsung 4K monitor dating from when 4K was a new thing, the LicheePi initially refused to operate at a resolution higher than 1024*768 but later on switched to 4K resolution when resuming from screen-blank for no apparent reason (and the window manager didn t support this properly). On the Dell 4K monitor I use on my main workstation it sometimes refused to talk to it and occasionally worked. I got it running at 1920*1080 without problems and then switched it to 4K and it lost video sync and never talked to that monitor again. On my Desklab portabable 4K monitor I got it to display in 4K resolution but only the top left 1/4 of the screen displayed. The issues with HDMI monitor support greatly limit the immediate potential for using this as a workstation. It doesn t make it impossible but would be fiddly at best. It s quite likely that a future OS update will fix this. But at the moment it s best used as a server. The LicheePi has a custom Linux distribution based on Ubuntu so you want too put something like the following in /etc/network/interfaces to make it automatically connect to the ethernet when plugged in:

auto end0
iface end0 inet dhcp

Then to get sshd to start you have to run the following commands to generate ssh host keys that aren t zero bytes long:

rm /etc/ssh/ssh_host_*
systemctl restart ssh.service

It appears to have wifi hardware but the OS doesn t recognise it. This isn t a priority for me as I mostly want to use it as a server. Performance For the first test of performance I created a 100MB file from /dev/urandom and then tried compressing it on various systems. With zstd -9 it took 16.893 user seconds on the LicheePi4A, 0.428s on my Thinkpad X1 Carbon Gen5 with a i5-6300U CPU (Debian/Unstable), 1.288s on my E5-2696 v3 workstation (Debian/Bookworm), 0.467s on the E5-2696 v3 running Debian/Unstable, 2.067s on a E3-1271 v3 server, and 7.179s on the E3-1271 v3 system emulating a RISC-V system via QEMU running Debian/Unstable. It s very impressive that the QEMU emulation is fast enough that emulating a different CPU architecture is only 3.5* slower for this test (or maybe 10* slower if it was running Debian/Unstable on the AMD64 code)! The emulated RISC-V is also more than twice as fast as real RISC-V hardware and probably of comparable speed to real RISC-V hardware when running the same versions (and might be slightly slower if running the same version of zstd) which is a tribute to the quality of emulation. One performance issue that most people don t notice is the time taken to negotiate ssh sessions. It s usually not noticed because the common CPUs have got faster at about the same rate as the algorithms for encryption and authentication have become more complex. On my i5-6300U laptop it takes 0m0.384s to run ssh -i ~/.ssh/id_ed25519 localhost id with the below server settings (taken from advice on ssh-audit.com [3] for a secure ssh configuration). On the E3-1271 v3 server it is 0.336s, on the QMU system it is 28.022s, and on the LicheePi it is 0.592s. By this metric the LicheePi is about 80% slower than decent x86 systems and the QEMU emulation of RISC-V is 73* slower than the x86 system it runs on. Does crypto depend on instructions that are difficult to emulate?

HostKey /etc/ssh/ssh_host_ed25519_key
KexAlgorithms -ecdh-sha2-nistp256,ecdh-sha2-nistp384,ecdh-sha2-nistp521,diffie-hellman-group14-sha256
MACs -umac-64-etm@openssh.com,hmac-sha1-etm@openssh.com,umac-64@openssh.com,umac-128@openssh.com,hmac-sha2-256,hmac-sha2-512,hmac-sha1

I haven t yet tested the performance of Ethernet (what routing speed can you get through the 2 gigabit ports?), emmc storage, and USB. At the moment I ve been focused on using RISC-V as a test and development platform. My conclusion is that I m glad I don t plan to compile many kernels or anything large like LibreOffice. But that for typical development that I do it will be quite adequate. The speed of Chromium seems adequate in basic tests, but the video output hasn t worked reliably enough to do advanced tests. Hardware Features Having two Gigabit Ethernet ports, 4 USB-3 ports, and Wifi on board gives some great options for using this as a router. It s disappointing that they didn t go with 2.5Gbit as everyone seems to be doing that nowadays but Gigabit is enough for most things. Having only a single HDMI port and not supporting USB-C docks (the USB-C port appears to be power only) limits what can be done for workstation use and for controlling displays. I know of people using small ARM computers attached to the back of large TVs for advertising purposes and that isn t going to be a great option for this. The CPU and RAM apparently uses a lot of power (which is relative the entire system draws up to 2A at 5V so the CPU would be something below 5W). To get this working a cooling fan has to be stuck to the CPU and RAM chips via a layer of thermal stuff that resembles a fine sheet of blu-tack in both color and stickyness. I am disappointed that there isn t any more solid form of construction, to mount this on a wall or ceiling some extra hardware would be needed to secure this. Also if they just had a really big copper heatsink I think that would be better. 80386 CPUs with similar TDP were able to run without a fan. I wonder how things would work with all USB ports in use. It s expected that a USB port can supply a minimum of 2.5W which means that all the ports could require 10W if they were active. Presumably something significantly less than 5W is available for the USB ports. Other Devices Sipheed has a range of other devices in the works. They currently sell the LicheeCluster4A which support 7 compute modules for a cluster in a box. This has some interesting potential for testing and demonstrating cluster software but you could probably buy an AMD64 system with more compute power for less money. The Lichee Console 4A is a tiny laptop which could be useful for people who like the 7 laptop form factor, unfortunately it only has a 1280*800 display if it had the same resolution display as a typical 7 phone I would have bought one. The next device that appeals to me is the soon to be released Lichee Pad 4A which is a 10.1 tablet with 1920*1200 display, Wifi6, Bluetooth 5.4, and 16G of RAM. It also has 1 USB-C connection, 2*USB-3 sockets, and support for an external card with 2*Gigabit ethernet. It s a tablet as a laptop without keyboard instead of the more common larger phone design model. They are also about to release the LicheePadMax4A which is similar to the other tablet but with a 14 2240*1400 display and which ships with a keyboard to make it essentially a laptop with detachable keyboard. Conclusion At this time I wouldn t recommend that this device be used as a workstation or laptop, although the people who want to do such things will probably do it anyway regardless of my recommendations. I think it will be very useful as a test system for RISC-V development. I have some friends who are interested in this sort of thing and I can give them VMs. It is a bit expensive. The Sipheed web site boasts about the LicheePi4 being faster than the RaspberryPi4, but it s not a lot faster and the RaspberryPi4 is much cheaper ($127 or $129 for one with 8G of RAM). The RaspberryPi4 has two HDMI ports but a limit of 8G of RAM while the LicheePi has up to 16G of RAM and two Gigabit Ethernet ports but only a single HDMI port. It seems that the RaspberryPi4 might win if you want a cheap low power desktop system. At this time I think the reason for this device is testing out RISC-V as an alternative to the AMD64 and ARM64 architectures. An open CPU architecture goes well with free software, but it isn t just people who are into FOSS who are testing such things. I know some corporations are trying out RISC-V as a way of getting other options for embedded systems that don t involve paying monopolists. The Lichee Console 4A is probably a usable tiny laptop if the resolution is sufficient for your needs. As an aside I predict that the tiny laptop or pocket computer segment will take off in the near future. There are some AMD64 systems the size of a phone but thicker that run Windows and go for reasonable prices on AliExpress. Hopefully in the near future this device will have better video drivers and be usable as a small and quiet workstation. I won t rule out the possibility of making this my main workstation in the not too distant future, all it needs is reliable 4K display and the ability to decode 4K video. It s performance for web browsing and as an ssh client seems adequate, and that s what matters for my workstation use. But for the moment it s just for server use.

• [1] https://sipeed.com/licheepi4a
• [2] https://wiki.sipeed.com/hardware/en/lichee/th1520/lp4a.html
• [3] https://www.ssh-audit.com/

As well as the collection and distribution of compromised keys, the pwnedkeys project also matches those pwned keys against issued SSL certificates. I m excited to announce that, as of the beginning of 2024, all matched certificates are now being published on the Fediverse, thanks to the botsin.space Mastodon server. Want to know which sites are susceptible to interception and interference, in (near-)real time? Do you have a burning desire to know who is issuing certificates to people that post their private keys in public? Now you can.

How It Works The process for publishing pwned certs is, roughly, as follows:

1. All the certificates in Certificate Transparency (CT) logs are hoovered up (using my scrape-ct-log tool, the fastest log scraper in the west!), and the fingerprint of the public key of each certificate is stored in an LMDB datafile.
2. As new private keys are identified as having been compromised, the fingerprint of that key is checked against all the LMDB files, which map key fingerprints to certificates (actually to CT log entry IDs, from which the certificates themselves are retrieved).
3. If one or more matches are found, then the certificates using the compromised key are forwarded to the tooter , which publishes them for the world to marvel at.

This makes it sound all very straightforward, and it is in theory. The trick comes in optimising the pipeline so that the five million or so new certificates every day can get indexed on the one slightly middle-aged server I ve got, without getting backlogged.

Why Don t You Just Have the Certificates Revoked? Funny story about that I used to notify CAs of certificates they d issued using compromised keys, which had the effect of requiring them to revoke the associated certificates. However, several CAs disliked having to revoke all those certificates, because it cost them staff time (and hence money) to do so. They went so far as to change their procedures from the standard way of accepting problem reports (emailing a generic attestation of compromise), and instead required CA-specific hoop-jumping to notify them of compromised keys. Since the effectiveness of revocation in the WebPKI is, shall we say, homeopathic at best, I decided I couldn t be bothered to play whack-a-mole with CAs that just wanted to be difficult, and I stopped sending compromised key notifications to CAs. Instead, now I m publishing the details of compromised certificates to everyone, so that users can protect themselves directly should they choose to.

Further Work The astute amongst you may have noticed, in the above How It Works description, a bit of a gap in my scanning coverage. CAs can (and do!) issue certificates for keys that are already compromised, including weak keys that have been known about for a decade or more (1, 2, 3). However, as currently implemented, the pwnedkeys certificate checker does not automatically find such certificates. My plan is to augment the CT scraping / cert processing pipeline to check all incoming certificates against the existing (2M+) set of pwned keys. Though, with over five million new certificates to check every day, it s not necessarily as simple as just hit the pwnedkeys API for every new cert . The poor old API server might not like that very much.

Support My Work If you d like to see this extra matching happen a bit quicker, I ve setup a ko-fi supporters page, where you can support my work on pwnedkeys and the other open source software and projects I work on by buying me a refreshing beverage. I would be very appreciative, and your support lets me know I should do more interesting things with the giant database of compromised keys I ve accumulated.

Contributing to Debian is part of Freexian s mission. This article covers the latest achievements of Freexian and their collaborators. All of this is made possible by organizations subscribing to our Long Term Support contracts and consulting services.

LXD/Incus backend bug in autopkgtest by Stefano Rivera While working on the Python 3.12 transition, Stefano repeatedly ran into a bug in autopkgtest when using LXD (or in the future Incus), that caused it to hang when running cython s multi-hour autopkgtests. After some head-banging, the bug turned out to be fairly straightforward: LXD didn t shut down on receiving a SIGTERM, so when a testsuite timed out, it would hang forever. A simple fix has been applied.

/usr-merge, by Helmut Grohne Thanks to Christian Hofstaedtler and others, the effort is moving into a community effort and the work funded by Freexian becomes more difficult to separate from non-funded work. In particular, since the community fully handled all issues around lost udev rules, dh_installudev now installs rules to /usr. The story around diversions took another detour. We learned that conflicts do not reliably prevent concurrent unpack and the reiterated mitigation for molly-guard triggered this. After a bit of back and forth and consultation with the developer mailing list, we concluded that avoiding the problematic behavior when using apt or an apt-based upgrader combined with a loss mitigation would be good enough. The involved packages bfh-container, molly-guard, progress-linux-container and systemd have since been uploaded to unstable and the matter seems finally solved except that it doesn t quite work with sysvinit yet. The same approach is now being proposed for the diversions of zutils for gzip. We thank involved maintainers for their timely cooperation.

gcc-for-host, by Helmut Grohne Since forever, it has been difficult to correctly express a toolchain build dependency. This can be seen in the Build-Depends of the linux source package for instance. While this has been solved for binutils a while back, the patches for gcc have been unfinished. With lots of constructive feedback from gcc package maintainer Matthias Klose, Helmut worked on finalizing and testing these patches. Patch stacks are now available for gcc-13 and gcc-14 and Matthias already included parts of them in test builds for Ubuntu noble. Finishing this work would enable us to resolve around 1000 cross build dependency satisfiability issues in unstable.

Miscellaneous contributions

• Stefano continued work on the Python 3.12 transition, including uploads of cython, pycxx, numpy, python-greenlet, twisted, foolscap and dh-python.
• Stefano reviewed and selected from a new round of DebConf 24 bids, as part of the DebConf Committee. Busan, South Korea was selected.
• For debian-printing Thorsten uploaded hplip to unstable to fix a /usr-merge bug and cups to Bookworm to fix bugs related to printing in color.
• Utkarsh helped newcomers in mentoring and reviewing their packaging; eg: golang-github-prometheus-community-pgbouncer-exporter.
• Helmut sent patches for 42 cross build failures unrelated to the gcc-for-host work.
• Helmut continues to maintain rebootstrap. In December, blt started depending on libjpeg and this poses a dependency loop. Ideally, Python would stop depending on blt. Also linux-libc-dev having become Multi-Arch: foreign poses non-trivial issues that are not fully resolved yet.
• Enrico participated in /usr-merge discussions with Helmut.

Release 0.6.34 of the digest package arrived at CRAN today and has also been uploaded to Debian already. digest creates hash digests of arbitrary R objects. It can use a number different hashing algorithms (md5, sha-1, sha-256, sha-512, crc32, xxhash32, xxhash64, murmur32, spookyhash, blake3, and crc32c), and ebales easy comparison of (potentially large and nested) R language objects as it relies on the native serialization in R. It is a mature and widely-used package (with 63.8 million downloads just on the partial cloud mirrors of CRAN which keep logs) as many tasks may involve caching of objects for which it provides convenient general-purpose hash key generation to quickly identify the various objects. (Oh and we also just passed the 20th anniversary of the initial CRAN upload. Time flies, as they say.) This release contains small (build-focussed) enhancements contributed by Michael Chirico, and another set of fixed for printf format warnings this time on Windows. My CRANberries provides a summary of changes to the previous version. For questions or comments use the issue tracker off the GitHub repo. If you like this or other open-source work I do, you can now sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

Having had success running Trisquel on the ppc64 Talos II, I felt ready to get an arm64 machine running Trisquel. I have a Ampere Altra Developer Platform from ADLINK, which is a fairly powerful desktop machine. While there were some issues during installation, I m happy to say the machine is stable and everything appears to work fine.

ISO images for non-amd64 platforms are unfortunately still hidden from the main Trisquel download area, so you will have to use the following procedure to download and extract a netinst ISO image (using debian-installer) and write it to a USB memory device. Another unfortunate problem is that there are no OpenPGP signatures or hash checksums, but below I publish one checksum.

wget -q http://builds.trisquel.org/debian-installer-images/debian-installer-images_20210731+deb11u9+11.0trisquel15_arm64.tar.gz
tar xfa debian-installer-images_20210731+deb11u9+11.0trisquel15_arm64.tar.gz ./installer-arm64/20210731+deb11u9+11/images/netboot/mini.iso
echo '311732519cc8c7c1bb2fe873f134fdafb211ef3bcb5b0d2ecdc6ea4e3b336357  installer-arm64/20210731+deb11u9+11/images/netboot/mini.iso'   sha256sum -c
sudo wipefs -a /dev/sdX
sudo dd if=installer-arm64/20210731+deb11u9+11/images/netboot/mini.iso of=/dev/sdX conv=sync status=progress

Insert the USB stick in a USB slot in the machine, and power up. Press ESCAPE at the BIOS prompt and select the USB device as the boot device. The first problem that hit me was that translations didn t work, I selected Swedish but the strings were garbled. Rebooting and selecting the default English worked fine. For installation, you need Internet connectivity and I use the RJ45 port closest to VGA/serial which is available as enP5p1s0 in the installer. I wouldn t connect the BMC RJ45 port to anything unless you understand the security implications. During installation you have to create a EFI partition for booting, and I ended up with one 1GB EFI partition, one 512GB ext4 partition for / with discard/noatime options, and a 32GB swap partition. The installer did not know about any Trisquel mirrors, but only had the default archive.trisquel.org, so if you need to use a mirror, take a note of the necessary details. The installation asks me about which kernel to install, and I went with the default linux-generic which results in a 5.15 linux-libre kernel. At the end of installation, unfortunately grub failed with a mysterious error message: Unable to install GRUB in dummy. Executing 'grub-install dummy' failed. On another console there is a better error message: failed to register the EFI boot entry. There are some references to file descriptor issues. Perhaps I partitioned the disk in a bad way, or this is a real bug in the installer for this platform. I continued installation, and it appears the installer was able to write GRUB to the device, but not add the right boot menu. So I was able to finish the installation properly, and then reboot and manually type the following GRUB commands: linux (hd0,gpt2)/boot/vmlinuz initrd (hd0,gpt2)/boot/initrd.img boot. Use the GRUB ls command to find the right device. See images below for more information.

Booting and installing GRUB again manually works fine:

root@ampel:~# update-grub
Sourcing file  /etc/default/grub'
Sourcing file  /etc/default/grub.d/background.cfg'
Sourcing file  /etc/default/grub.d/init-select.cfg'
Generating grub configuration file ...
Found linux image: /boot/vmlinuz-5.15.0-91-generic
Found initrd image: /boot/initrd.img-5.15.0-91-generic
Found linux image: /boot/vmlinuz-5.15.0-58-generic
Found initrd image: /boot/initrd.img-5.15.0-58-generic
Warning: os-prober will not be executed to detect other bootable partitions.
Systems on them will not be added to the GRUB boot configuration.
Check GRUB_DISABLE_OS_PROBER documentation entry.
Adding boot menu entry for UEFI Firmware Settings ...
done
root@ampel:~# 

During installation I tend to avoid selecting any tasksel components, in part because it didn t use a local mirror to gain network speed, and in part because I don t want to generate OpenSSH keys in a possibly outdated environment that is harder to audit and reproducible rebuild than the finally installed system. When I selected the OpenSSH and GNOME tasksel, I get an error, but fortunately using apt get directly is simple.

root@ampel:~# tasksel
Tasksel GNOME failed:
tasksel: apt-get failed (100)
root@ampel:~# apt-get install trisquel-gnome ssh

Graphics in GNOME was slow using the built-in ASPEED AST2500 VGA controller with linux-libre 5.15. There are kernels labeled 64k but I haven t tested them, and I m not sure they would bring any significant advantage. I simply upgraded to a more recent linux-libre 6.2 kernel via the linux-image-generic-hwe-11.0 virtual package. After a reboot, graphics in GNOME is usable.

root@ampel:~# apt-get install linux-image-generic-hwe-11.0

There seems to be some issue with power-saving inside GNOME, since the machine becomes unresponsive after 20 minutes, and I m unable to make it resume via keyboard or power button. Disabling the inactivity power setting in GNOME works fine to resolve this. I will now put this machine to some more heavy use and see how it handles it. I hope to find more suitable arm64-based servers to complement my ppc64el-based servers in the future, as this ADLINK Ampere Altra Developer Platform with liquid-cooling is more of a toy than a serious server for use in a datacentre. Happy Trisquel-on-arm64 Hacking!

I ended 2022 with a musical retrospective and very much enjoyed writing that blog post. As such, I have decided to do the same for 2023! From now on, this will probably be an annual thing :) Albums In 2023, I added 73 new albums to my collection nearly 2 albums every three weeks! I listed them below in the order in which I acquired them. I purchased most of these albums when I could and borrowed the rest at libraries. If you want to browse though, I added links to the album covers pointing either to websites where you can buy them or to Discogs when digital copies weren't available. Once again this year, it seems that Punk (mostly O !) and Metal dominate my list, mostly fueled by Angry Metal Guy and the amazing Montr al Skinhead/Punk concert scene. Concerts A trend I started in 2022 was to go to as many concerts of artists I like as possible. I'm happy to report I went to around 80% more concerts in 2023 than in 2022! Looking back at my list, April was quite a busy month... Here are the concerts I went to in 2023:

• March 8th: Godspeed You! Black Emperor
• April 11th: Alexandra Str liski
• April 12th: Bikini Kill
• April 21th: Brigada Flores Magon, Union Thugs
• April 28th: Komintern Sect, The Outcasts, Violent Way, Ultra Razzia, Over the Hill
• May 3rd: First Fragment
• May 12th: Rhapsody of Fire, Wind Rose
• May 13th: Aeternam
• June 2nd: Mortier, La Gachette
• June 17th: Ultra Razzia, Total Nada, BLEMISH
• June 30th: Avishai Cohen Trio
• July 9th: Richard Galliano
• August 18th: Gojira, Mastodon, Lorna Shore
• September 14th: Jinjer
• September 22nd: CUIR, Salvaje Punk, Hysteric Polemix, Perestroika, Ultra Razzia, Ilusion, Over the Hill, Asbestos
• October 6th: Rancoeur, Street Code, Tenaz, Mortimer, Guernica, High Anxiety

Although metalfinder continues to work as intended, I'm very glad to have discovered the Montr al underground scene has departed from Facebook/Instagram and adopted en masse Gancio, a FOSS community agenda that supports ActivityPub. Our local instance, askapunk.net is pretty much all I could ask for :) That's it for 2023!

So this blog is now celebrating its 21st birthday (or 20 if you count from zero, or 18 if you want to be pedantic), and I figured I would do this yearly thing of reviewing how that went.

Number of posts 2022 was the official 20th anniversary in any case, and that was one of my best years on record, with 46 posts, surpassed only by the noisy 2005 (62) and matching 2006 (46). 2023, in comparison, was underwhelming: a feeble 11 posts! What happened! Well, I was busy with other things, mostly away from keyboard, that I will not bore you with here... The other thing that happened is that the one-liner I used to collect stats was broken (it counted folders and other unrelated files) and wildly overestimated 2022! Turns out I didn't write that much then:

anarc.at$ ls blog   grep '^[0-9][0-9][0-9][0-9].*.md'   se
d s/-.*//   sort   uniq -c    sort -n -k2
     57 2005
     43 2006
     20 2007
     20 2008
      7 2009
     13 2010
     16 2011
     11 2012
     13 2013
      5 2014
     13 2015
     18 2016
     29 2017
     27 2018
     17 2019
     18 2020
     14 2021
     28 2022
     10 2023
      1 2024

But even that is inaccurate because, in ikiwiki, I can tag any page as being featured on the blog. So we actually need to process the HTML itself because we don't have much better on hand without going through ikiwiki's internals:

anarcat@angela:anarc.at$ curl -sSL https://anarc.at/blog/   grep 'href="\./'   grep -o 20[0-9][0-9]   sort   uniq -c 
     56 2005
     42 2006
     19 2007
     18 2008
      6 2009
     12 2010
     15 2011
     10 2012
     11 2013
      3 2014
     15 2015
     32 2016
     50 2017
     37 2018
     19 2019
     19 2020
     15 2021
     28 2022
     13 2023

Which puts the top 10 years at:

$ curl -sSL https://anarc.at/blog/   grep 'href="\./'   grep -o 20[0-9][0-9]   sort   uniq -c    sort -nr   head -10
     56 2005
     50 2017
     42 2006
     37 2018
     32 2016
     28 2022
     19 2020
     19 2019
     19 2007
     18 2008

Anyway. 2023 is certainly not a glorious year in that regard, in any case.

Visitors In terms of visits, however, we had quite a few hits. According to Goatcounter, I had 122 300 visits in 2023! 2022, in comparison, had 89 363, so that's quite a rise.

What you read I seem to have hit the Hacker News front page at least twice. I say "seem" because it's actually pretty hard to tell what the HN frontpage actually is on any given day. I had 22k visits on 2023-03-13, in any case, and you can't see me on the front that day. We do see a post of mine on 2023-09-02, all the way down there, which seem to have generated another 10k visits. In any case, here were the most popular stories for you fine visitors:

• Framework 12th gen laptop review: 24k visits, which is surprising for a 13k words article "without images", as some critics have complained. 15k referred by Hacker News. Good reference and time-consuming benchmarks, slowly bit-rotting. That is, by far, my most popular article ever. A popular article in 2021 or 2022 was around 6k to 9k, so that's a big one. I suspect it will keep getting traffic for a long while.
• Calibre replacement considerations: 15k visits, most of which without a referrer. Was actually an old article, but I suspect HN brought it back to light. I keep updating that wiki page regularly when I find new things, but I'm still using Calibre to import ebooks.
• Hacking my Kobo Clara HD: is not new but always gathering more and more hits, it had 1800 hits in the first year, 4600 hits last year and now brought 6400 visitors to the blog! Not directly related, but this iFixit battery replacement guide I wrote also seem to be quite popular

Everything else was published before 2023. Replacing Smokeping with Prometheus is still around and Looking at Wayland terminal emulators makes an entry in the top five.

Where you've been People send less and less private information when they browse the web. The number of visitors without referrers was 41% in 2021, it rose to 44% in 2023. Most of the remaining traffic comes from Google, but Hacker News is now a significant chunk, almost as big as Google. In 2021, Google represented 23% of my traffic, in 2022, it was down to 15% so 18% is actually a rise from last year, even if it seems much smaller than what I usually think of.

Ratio	Referrer	Visits
18%	Google	22 098
13%	Hacker News	16 003
2%	duckduckgo.com	2 640
1%	community.frame.work	1 090
1%	missing.csail.mit.edu	918

Note that Facebook and Twitter do not appear at all in my referrers.

Where you are Unsurprisingly, most visits still come from the US:

Ratio	Country	Visits
26%	United States	32 010
14%	France	17 046
10%	Germany	11 650
6%	Canada	7 425
5%	United Kingdom	6 473
3%	Netherlands	3 436

Those ratios are nearly identical to last year, but quite different from 2021, where Germany and France were more or less reversed. Back in 2021, I mentioned there was a long tail of countries with at least one visit, with 160 countries listed. I expanded that and there's now 182 countries in that list, almost all of the 193 member states in the UN.

What you were Chrome's dominance continues to expand, even on readers of this blog, gaining two percentage points from Firefox compared to 2021.

Ratio	Browser	Visits
49%	Firefox	60 126
36%	Chrome	44 052
14%	Safari	17 463
1%	Others	N/A

It seems like, unfortunately, my Lynx and Haiku users have not visited in the past year. It seems like trying to read those metrics is like figuring out tea leaves... In terms of operating systems:

Ratio	OS	Visits
28%	Linux	34 010
23%	macOS	28 728
21%	Windows	26 303
17%	Android	20 614
10%	iOS	11 741

Again, Linux and Mac are over-represented, and Android and iOS are under-represented.

What is next I hope to write more next year. I've been thinking about a few posts I could write for work, about how things work behind the scenes at Tor, that could be informative for many people. We run a rather old setup, but things hold up pretty well for what we throw at it, and it's worth sharing that with the world... So anyway, thanks for coming, faithful reader, and see you in the coming 2024 year...

This year was hard from a personal and work point of view, which impacted the amount of Free Software bits I ended up doing - even when I had the time I often wasn t in the right head space to make progress on things. However writing this annual recap up has been a useful exercise, as I achieved more than I realised. For previous years see 2019, 2020, 2021 + 2022.

Conferences The only Free Software related conference I made it to this year was DebConf23 in Kochi, India. Changes with projects at work meant I couldn t justify anything work related. This year I m planning to make it to FOSDEM, and haven t made a decision on DebConf24 yet.

Debian Most of my contributions to Free software continue to happen within Debian. I started the year working on retrogaming with Kodi on Debian. I got this to a much better state for bookworm, with it being possible to run the bsnes-mercury emulator under Kodi using RetroArch. There are a few other libretro backends available for RetroArch, but Kodi needs some extra controller mappings packaged up first. Plenty of uploads were involved, though some of this was aligning all the dependencies and generally cleaning things up in iterations.

• retroarch: 1.13.0-dfsg+3, 1.14.0+dfsg-1, 1.15.0+dfsg-1, 1.16.0.3+dfgs-1
• libretro-core-info uploads: 1.14.0-1, 1.15.0-1.
• libretro-bsnes-mercury: 094+git20220807-2, 094+git20220807-3, 094+git20220807-4, 094+git20220807-5, 094+git20220807-6, 094+git20220807-7, 094+git20220807-8.
• kodi-game-libretro: 20.2.0-2, 20.2.2-1, 20.2.2-2, 20.2.4-1, 20.2.4-2
• rcheevos: 10.5.0-2, 10.6.0-1, 10.7.0-1, 10.7.1-1.

I continued to work on a few packages within the Debian Electronics Packaging Team. OpenOCD produced a new release in time for the bookworm release, so I uploaded 0.12.0-1. There were a few minor sigrok cleanups - sigrok 0.3, libsigrokdecode 0.5.3-4 + libsigrok 0.5.2-4 / 0.5.2-5. While I didn t manage to get the work completed I did some renaming of the ESP8266 related packages - gcc-xtensa-lx106 (which saw a 13 upload pre-bookworm) has become gcc-xtensa (with 14) and binutils-xtensa-lx106 has become binutils-xtensa (with 6). Binary packages remain the same, but this is intended to allow for the generation of ESP32 compiler toolchains from the same source. onak saw 0.6.3-1 uploaded to match the upstream release. I also uploaded libgpg-error 1.47-1 (though I can claim no credit for any of the work in preparing the package) to help move things forward on updating gnupg2 in Debian. I NMUed tpm2-pkcs11 1.9.0-0.1 to fix some minor issues pre-bookworm release; I use this package myself to store my SSH key within my laptop TPM, so I care about it being in a decent state. sg3-utils also saw a bit of love with 1.46-2 + 1.46-3 - I don t work in the storage space these days, but I m still listed as an uploaded and there was an RC bug around the library package naming that I was qualified to fix and test pre-bookworm. Related to my retroarch work I sponsored uploads of mgba for Ryan Tandy: 0.10.0+dfsg-1, 0.10.0+dfsg-2, 0.10.1+dfsg-1, 0.10.2+dfsg-1, mgba 0.10.1+dfsg-1+deb12u1. As part of the Data Protection Team I responded to various inbound queries to that team, both from project members and those external to the project. I continue to keep an eye on Debian New Members, even though I m mostly inactive as an application manager - we generally seem to have enough available recently. Mostly my involvement is via Front Desk activities, helping out with queries to the team alias, and contributing to internal discussions as well as our panel at DebConf23. Finally the 3 month rotation for Debian Keyring continues to operate smoothly. I dealt with 2023.03.24, 2023.06.26, 2023.06.29, 2023.09.10, 2023.09.24 + 2023.12.24.

Linux I had a few minor patches accepted to the kernel this year. A pair of safexcel cleanups (improved error logging for firmware load fail and cleanup on load failure) came out of upgrading the kernel running on my RB5009. The rest were related to my work on repurposing my C.H.I.P.. The AXP209 driver needed extended to support GPIO3 (with associated DT schema update). That allowed Bluetooth to be enabled. Adding the AXP209 internal temperature ADC as an iio-hwmon node means it can be tracked using the normal sensor monitoring framework. And finally I added the pinmux settings for mmc2, which I use to support an external microSD slot on my C.H.I.P.

Personal projects 2023 saw another minor release of onak, 0.6.3, which resulted in a corresponding Debian upload (0.6.3-1). It has a couple of bug fixes (including a particularly annoying, if minor, one around systemd socket activation that felt very satisfying to get to the bottom of), but I still lack the time to do any of the major changes I would like to. I wrote listadmin3 to allow easy manipulation of moderation queues for Mailman3. It s basic, but it s drastically improved my timeliness on dealing with held messages.

Work related This year only involved a single upstream related submission; a fix for tpm_tis interrupts with the Lenovo P620 that then got dropped when the change that caused the issue was reverted. That wraps up 2023. I ve got no particular goals for this year; looking around my desk I ve a few ARM based devices I d like to get running a mainline kernel. I need to play about a bit more with the retroarch bits (if I really had time I d do the migration for Kodi to PCRE2, as that s currently causing testing migration issues), perhaps getting some more controller mappings packaged. But no promises.

I needed to migrate an existing system to an Hetzner cloud VPS. While it is possible to attach KVM consoles and custom ISO images to dedicated servers, i didn t find any way to do so with regular cloud instances. For system migrations i usually use REAR, which has never failed me. (and also has saved my ass during recovery multiple times). It s an awesome utility! It s possible to do this using the Hetzner recovery console too, but using REAR is very convenient here, because it handles things like re-creating the partition layout and network settings automatically! The steps are:

• Create bootable REAR rescue image on the source system.
• Register a target system in Hetzner Cloud with at least the same disk size as the source system.
• Boot the REAR image s initrd and kernel from the running VPS system using kexec
• Make the REAR recovery console accessible via SSH (or use Hetzners console).
• Let REAR do its magic and re-partition the system.
• Restore the system data to the freshly partitioned disks
• Let REAR handle the bootloader and network re-configuration.

Example To create a rescue image on the source system:

apt install rear
echo OUTPUT=ISO > /etc/rear/local.conf
rear mkrescue -v
[..]
Wrote ISO image: /var/lib/rear/output/rear-debian12.iso (185M)

My source system had a 128 GB disk, so i registered an instance on Hetzner cloud with greater disk size to make things easier: Now copy the ISO image to the newly created instance and extract its data:

 apt install kexec-tools
 scp rear-debian12.iso root@49.13.193.226:/tmp/
 modprobe loop
 mount -o loop rear-debian12.iso /mnt/
 cp /mnt/isolinux/kernel /tmp/
 cp /mnt/isolinux/initrd.cgz /tmp/

Install kexec if not installed already:

 apt install kexec-tools

Note down the current gateway configuration, this is required later on to make the REAR recovery console reachable via SSH:

root@testme:~# ip route
default via 172.31.1.1 dev eth0
172.31.1.1 dev eth0 scope link

Reboot the running VPS instance into the REAR recovery image using somewhat the same kernel cmdline:

root@testme:~# cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-6.1.0-13-amd64 root=UUID=5174a81e-5897-47ca-8fe4-9cd19dc678c4 ro consoleblank=0 systemd.show_status=true console=tty1 console=ttyS0
kexec --initrd /tmp/initrd.cgz --command-line="consoleblank=0 systemd.show_status=true console=tty1 console=ttyS0" /tmp/kernel
Connection to 49.13.193.226 closed by remote host.
Connection to 49.13.193.226 closed

Now watch the system on the Console booting into the REAR system: Login the recovery console (root without password) and fix its default route to make it reachable:

ip addr
[..]
2: enp1s0
..
$ ip route add 172.31.1.1 dev enp1s0
$ ip route add default via 172.31.1.1
ping 49.13.193.226
64 bytes from 49.13.193.226: icmp_seq=83 ttl=52 time=27.7 ms

The network configuration might differ, the source system in this example used DHCP, as the target does. If REAR detects changed static network configuration it guides you through the setup pretty nicely. Login via SSH (REAR will store your ssh public keys in the image) and start the recovery process, follow the steps as suggested by REAR:

ssh -l root 49.13.193.226
Welcome to Relax-and-Recover. Run "rear recover" to restore your system !
RESCUE debian12:~ # rear recover
Relax-and-Recover 2.7 / Git
Running rear recover (PID 673 date 2024-01-04 19:20:22)
Using log file: /var/log/rear/rear-debian12.log
Running workflow recover within the ReaR rescue/recovery system
Will do driver migration (recreating initramfs/initrd)
Comparing disks
Device vda does not exist (manual configuration needed)
Switching to manual disk layout configuration (GiB sizes rounded down to integer)
/dev/vda had size 137438953472 (128 GiB) but it does no longer exist
/dev/sda was not used on the original system and has now 163842097152 (152 GiB)
Original disk /dev/vda does not exist (with same size) in the target system
Using /dev/sda (the only available of the disks) for recreating /dev/vda
Current disk mapping table (source => target):
  /dev/vda => /dev/sda
Confirm or edit the disk mapping
1) Confirm disk mapping and continue 'rear recover'
[..]
User confirmed recreated disk layout
[..]

This step re-recreates your original disk layout and mounts it to /mnt/local/ (this example uses a pretty lame layout, but usually REAR will handle things like lvm/btrfs just nicely):

mount
/dev/sda3 on /mnt/local type ext4 (rw,relatime,errors=remount-ro)
/dev/sda1 on /mnt/local/boot type ext4 (rw,relatime)

Now clone your source systems data to /mnt/local/ with whatever utility you like to use and exit the recovery step. After confirming everything went well, REAR will setup the bootloader (and all other config details like fstab entries and adjusted network configuration) for you as required:

rear> exit
Did you restore the backup to /mnt/local ? Are you ready to continue recovery ? yes
User confirmed restored files
Updated initramfs with new drivers for this system.
Skip installing GRUB Legacy boot loader because GRUB 2 is installed (grub-probe or grub2-probe exist).
Installing GRUB2 boot loader...
Determining where to install GRUB2 (no GRUB2_INSTALL_DEVICES specified)
Found possible boot disk /dev/sda - installing GRUB2 there
Finished 'recover'. The target system is mounted at '/mnt/local'.
Exiting rear recover (PID 7103) and its descendant processes ...
Running exit tasks

Now reboot the recovery console and watch it boot into your target systems configuration: Being able to use this procedure for complete disaster recovery within Hetzner cloud VPS (using off-site backups) gives me a better feeling, too.

I ve been getting annoyed with Raspberry Pi OS (Raspbian) for years now. It s a fork of Debian, but manages to omit some of the most useful things. So I ve decided to migrate all of my Pis to run pure Debian. These are my reasons:

1. Raspberry Pi OS has, for years now, specified that there is no upgrade path. That is, to get to a newer major release, it s a reinstall. While I have sometimes worked around this, for a device that is frequently installed in hard-to-reach locations, this is even more important than usual. It s common for me to upgrade machines for a decade or more across Debian releases and there s no reason that it should be so much more difficult with Raspbian.
2. As I noted in Consider Security First, the security situation for Raspberry Pi OS isn t as good as it is with Debian.
3. Raspbian lags behind Debian often times by 6 months or more for major releases, and days or weeks for bug fixes and security patches.
4. Raspbian has no direct backports support, though Raspberry Pi 3 and above can use Debian s backports (per my instructions as Installing Debian Backports on Raspberry Pi)
5. Raspbian uses a custom kernel without initramfs support

It turns out it is actually possible to do an in-place migration from Raspberry Pi OS bullseye to Debian bookworm. Here I will describe how. Even if you don t have a Raspberry Pi, this might still be instructive on how Raspbian and Debian packages work.

WARNINGS Before continuing, back up your system. This process isn t for the neophyte and it is entirely possible to mess up your boot device to the point that you have to do a fresh install to get your Pi to boot. This isn t a supported process at all.

Architecture Confusion Debian has three ARM-based architectures:

• armel, for the lowest-end 32-bit ARM devices without hardware floating point support
• armhf, for the higher-end 32-bit ARM devices with hardware float (hence hf )
• arm64, for 64-bit ARM devices (which all have hardware float)

Although the Raspberry Pi 0 and 1 do support hardware float, they lack support for other CPU features that Debian s armhf architecture assumes. Therefore, the Raspberry Pi 0 and 1 could only run Debian s armel architecture. Raspberry Pi 3 and above are capable of running 64-bit, and can run both armhf and arm64. Prior to the release of the Raspberry Pi 5 / Raspbian bookworm, Raspbian only shipped the armhf architecture. Well, it was an architecture they called armhf, but it was different from Debian s armhf in that everything was recompiled to work with the more limited set of features on the earlier Raspberry Pi boards. It was really somewhere between Debian s armel and armhf archs. You could run Debian armel on those, but it would run more slowly, due to doing floating point calculations without hardware support. Debian s raspi FAQ goes into this a bit. What I am going to describe here is going from Raspbian armhf to Debian armhf with a 64-bit kernel. Therefore, it will only work with Raspberry Pi 3 and above. It may theoretically be possible to take a Raspberry Pi 2 to Debian armhf with a 32-bit kernel, but I haven t tried this and it may be more difficult. I have seen conflicting information on whether armhf really works on a Pi 2. (If you do try it on a Pi 2, ignore everything about arm64 and 64-bit kernels below, and just go with the linux-image-armmp-lpae kernel per the ARMMP page) There is another wrinkle: Debian doesn t support running 32-bit ARM kernels on 64-bit ARM CPUs, though it does support running a 32-bit userland on them. So we will wind up with a system with kernel packages from arm64 and everything else from armhf. This is a perfectly valid configuration as the arm64 like x86_64 is multiarch (that is, the CPU can natively execute both the 32-bit and 64-bit instructions). (It is theoretically possible to crossgrade a system from 32-bit to 64-bit userland, but that felt like a rather heavy lift for dubious benefit on a Pi; nevertheless, if you want to make this process even more complicated, refer to the CrossGrading page.)

Prerequisites and Limitations In addition to the need for a Raspberry Pi 3 or above in order for this to work, there are a few other things to mention. If you are using the GPIO features of the Pi, I don t know if those work with Debian. I think Raspberry Pi OS modified the desktop environment more than other components. All of my Pis are headless, so I don t know if this process will work if you use a desktop environment. I am assuming you are booting from a MicroSD card as is typical in the Raspberry Pi world. The Pi s firmware looks for a FAT partition (MBR type 0x0c) and looks within it for boot information. Depending on how long ago you first installed an OS on your Pi, your /boot may be too small for Debian. Use df -h /boot to see how big it is. I recommend 200MB at minimum. If your /boot is smaller than that, stop now (or use some other system to shrink your root filesystem and rearrange your partitions; I ve done this, but it s outside the scope of this article.) You need to have stable power. Once you begin this process, your pi will mostly be left in a non-bootable state until you finish. (You did make a backup, right?)

Basic idea The basic idea here is that since bookworm has almost entirely newer packages then bullseye, we can just switch over to it and let the Debian packages replace the Raspbian ones as they are upgraded. Well, it s not quite that easy, but that s the main idea.

Preparation First, make a backup. Even an image of your MicroSD card might be nice. OK, I think I ve said that enough now. It would be a good idea to have a HDMI cable (with the appropriate size of connector for your particular Pi board) and a HDMI display handy so you can troubleshoot any bootup issues with a console.

Preparation: access The Raspberry Pi OS by default sets up a user named pi that can use sudo to gain root without a password. I think this is an insecure practice, but assuming you haven t changed it, you will need to ensure it still works once you move to Debian. Raspberry Pi OS had a patch in their sudo package to enable it, and that will be removed when Debian s sudo package is installed. So, put this in /etc/sudoers.d/010_picompat:

pi ALL=(ALL) NOPASSWD: ALL

Also, there may be no password set for the root account. It would be a good idea to set one; it makes it easier to log in at the console. Use the passwd command as root to do so.

Preparation: bluetooth Debian doesn t correctly identify the Bluetooth hardware address. You can save it off to a file by running hcitool dev > /root/bluetooth-from-raspbian.txt. I don t use Bluetooth, but this should let you develop a script to bring it up properly.

Preparation: Debian archive keyring You will next need to install Debian s archive keyring so that apt can authenticate packages from Debian. Go to the bookworm download page for debian-archive-keyring and copy the URL for one of the files, then download it on the pi. For instance:

wget http://http.us.debian.org/debian/pool/main/d/debian-archive-keyring/debian-archive-keyring_2023.3+deb12u1_all.deb

Use sha256sum to verify the checksum of the downloaded file, comparing it to the package page on the Debian site. Now, you ll install it with:

dpkg -i debian-archive-keyring_2023.3+deb12u1_all.deb

Package first steps From here on, we are making modifications to the system that can leave it in a non-bootable state. Examine /etc/apt/sources.list and all the files in /etc/apt/sources.list.d. Most likely you will want to delete or comment out all lines in all files there. Replace them with something like:

deb http://deb.debian.org/debian/ bookworm main non-free-firmware contrib non-free
deb http://security.debian.org/debian-security bookworm-security main non-free-firmware contrib non-free
deb https://deb.debian.org/debian bookworm-backports main non-free-firmware contrib non-free

(you might leave off contrib and non-free depending on your needs) Now, we re going to tell it that we ll support arm64 packages:

dpkg --add-architecture arm64

And finally, download the bookworm package lists:

apt-get update

If there are any errors from that command, fix them and don t proceed until you have a clean run of apt-get update.

Moving /boot to /boot/firmware The boot FAT partition I mentioned above is mounted at /boot by Raspberry Pi OS, but Debian s scripts assume it will be at /boot/firmware. We need to fix this. First:

umount /boot
mkdir /boot/firmware

Now, edit fstab and change the reference to /boot to be to /boot/firmware. Now:

mount -v /boot/firmware
cd /boot/firmware
mv -vi * ..

This mounts the filesystem at the new location, and moves all its contents back to where apt believes it should be. Debian s packages will populate /boot/firmware later.

Installing the first packages Now we start by installing the first of the needed packages. Eventually we will wind up with roughly the same set Debian uses.

apt-get install linux-image-arm64
apt-get install firmware-brcm80211=20230210-5
apt-get install raspi-firmware

If you get errors relating to firmware-brcm80211 from any commands, run that install firmware-brcm80211 command and then proceed. There are a few packages that Raspbian marked as newer than the version in bookworm (whether or not they really are), and that s one of them.

Configuring the bootloader We need to configure a few things in /etc/default/raspi-firmware before proceeding. Edit that file. First, uncomment (or add) a line like this:

KERNEL_ARCH="arm64"

Next, in /boot/cmdline.txt you can find your old Raspbian boot command line. It will say something like:

root=PARTUUID=...

Save off the bit starting with PARTUUID. Back in /etc/default/raspi-firmware, set a line like this:

ROOTPART=PARTUUID=abcdef00

(substituting your real value for abcdef00). This is necessary because the microSD card device name often changes from /dev/mmcblk0 to /dev/mmcblk1 when switching to Debian s kernel. raspi-firmware will encode the current device name in /boot/firmware/cmdline.txt by default, which will be wrong once you boot into Debian s kernel. The PARTUUID approach lets it work regardless of the device name.

Purging the Raspbian kernel Run:

dpkg --purge raspberrypi-kernel

Upgrading the system At this point, we are going to run the procedure beginning at section 4.4.3 of the Debian release notes. Generally, you will do:

apt-get -u upgrade
apt full-upgrade

Fix any errors at each step before proceeding to the next. Now, to remove some cruft, run:

apt-get --purge autoremove

Inspect the list to make sure nothing important isn t going to be removed.

Removing Raspbian cruft You can list some of the cruft with:

apt list '~o'

And remove it with:

apt purge '~o'

I also don t run Bluetooth, and it seemed to sometimes hang on boot becuase I didn t bother to fix it, so I did:

apt-get --purge remove bluez

Installing some packages This makes sure some basic Debian infrastructure is available:

apt-get install wpasupplicant parted dosfstools wireless-tools iw alsa-tools
apt-get --purge autoremove

Installing firmware Now run:

apt-get install firmware-linux

Resolving firmware package version issues If it gives an error about the installed version of a package, you may need to force it to the bookworm version. For me, this often happened with firmware-atheros, firmware-libertas, and firmware-realtek. Here s how to resolve it, with firmware-realtek as an example:

1. Go to https://packages.debian.org/PACKAGENAME for instance, https://packages.debian.org/firmware-realtek. Note the version number in bookworm in this case, 20230210-5.
2. Now, you will force the installation of that package at that version:

   apt-get install firmware-realtek=20230210-5
   

3. Repeat with every conflicting package until done.
4. Rerun apt-get install firmware-linux and make sure it runs cleanly.

Also, in the end you should be able to:

apt-get install firmware-atheros firmware-libertas firmware-realtek firmware-linux

Dealing with other Raspbian packages The Debian release notes discuss removing non-Debian packages. There will still be a few of those. Run:

apt list '?narrow(?installed, ?not(?origin(Debian)))'

Deal with them; mostly you will need to force the installation of a bookworm version using the procedure in the section Resolving firmware package version issues above (even if it s not for a firmware package). For non-firmware packages, you might possibly want to add --mark-auto to your apt-get install command line to allow the package to be autoremoved later if the things depending on it go away. If you aren t going to use Bluetooth, I recommend apt-get --purge remove bluez as well. Sometimes it can hang at boot if you don t fix it up as described above.

Set up networking We ll be switching to the Debian method of networking, so we ll create some files in /etc/network/interfaces.d. First, eth0 should look like this:

allow-hotplug eth0
iface eth0 inet dhcp
iface eth0 inet6 auto

And wlan0 should look like this:

allow-hotplug wlan0
iface wlan0 inet dhcp
    wpa-conf /etc/wpa_supplicant/wpa_supplicant.conf

Raspbian is inconsistent about using eth0/wlan0 or renamed interface. Run ifconfig or ip addr. If you see a long-named interface such as enx<something> or wlp<something>, copy the eth0 file to the one named after the enx interface, or the wlan0 file to the one named after the wlp interface, and edit the internal references to eth0/wlan0 in this new file to name the long interface name. If using wifi, verify that your SSIDs and passwords are in /etc/wpa_supplicant/wpa_supplicant.conf. It should have lines like:

network= 
   ssid="NetworkName"
   psk="passwordHere"
 

(This is where Raspberry Pi OS put them).

Deal with DHCP Raspberry Pi OS used dhcpcd, whereas bookworm normally uses isc-dhcp-client. Verify the system is in the correct state:

apt-get install isc-dhcp-client
apt-get --purge remove dhcpcd dhcpcd-base dhcpcd5 dhcpcd-dbus

Set up LEDs To set up the LEDs to trigger on MicroSD activity as they did with Raspbian, follow the Debian instructions. Run apt-get install sysfsutils. Then put this in a file at /etc/sysfs.d/local-raspi-leds.conf:

class/leds/ACT/brightness = 1
class/leds/ACT/trigger = mmc1

Prepare for boot To make sure all the /boot/firmware files are updated, run update-initramfs -u. Verify that root in /boot/firmware/cmdline.txt references the PARTUUID as appropriate. Verify that /boot/firmware/config.txt contains the lines arm_64bit=1 and upstream_kernel=1. If not, go back to the section on modifying /etc/default/raspi-firmware and fix it up.

The moment arrives Cross your fingers and try rebooting into your Debian system:

reboot

For some reason, I found that the first boot into Debian seems to hang for 30-60 seconds during bootstrap. I m not sure why; don t panic if that happens. It may be necessary to power cycle the Pi for this boot.

Troubleshooting If things don t work out, hook up the Pi to a HDMI display and see what s up. If I anticipated a particular problem, I would have documented it here (a lot of the things I documented here are because I ran into them!) So I can t give specific advice other than to watch boot messages on the console. If you don t even get kernel messages going, then there is some problem with your partition table or /boot/firmware FAT partition. Otherwise, you ve at least got the kernel going and can troubleshoot like usual from there.