Search Results: "cts"

18 December 2022

Bastian Venthur: The State of Python Packaging in 2022

Every year or so, I revisit the current best practices for Python packaging. This was my summary for 2021 here s the update for 2022. PyPA PyPA is still the place to go for information, best practices and tutorials for packaging Python projects. My only criticism from last year, namely that PyPA was heavily biased towards their own tooling (e.g. pipenv), has been addressed: the tool recommendations section lists now several tools for the same purpose with their own ones not necessarily being the first anymore. setup.py, setup.cfg, requirements.txt, Pipfile, pyproject.toml oh my! This is the reason why I m revisiting the documentation every year, to see what s the current way to go. Good progress has been made since last year: Bye setup.py and setup.cfg hello pyproject.toml pyproject.toml finally got mature enough to replace setup.py and setup.cfg in most cases. Recent versions of setuptools and pip now fully support pyproject.toml and even PyPA s packaging tutorial completely switched their example project from away setup.py towards pyproject.toml, making it an official recommendation. So, now you can replace your setup.py with pyproject.toml. If you had already some kind of declarative configuration in setup.cfg you can move that as well into pyproject.toml. Most tools, like pypy or pytest also support configuration in pyproject.toml (flake8 being a notable exception ) so there s no reason to keep setup.cfg around anymore. Actually, if you migrate to pyproject.toml it is best to do it properly and remove setup.py and setup.cfg as setuptools behaves a bit buggy when building a package that has either of them and the pyproject.toml. requirements.txt requirements.txt are still needed if you develop a deployable application (vs. a library) and want to provide pinned dependencies, i.e. with the specific versions that you ve tested your application with. Usually, the list of requirements in requirements.txt is the same as defined in pyproject.toml, but with pinned versions. Pipfile + Pipfile.lock I still do completely ignore Pipfile and Pipfile.lock as they are only supported by pipenv and not backed by any standard. Summary The major change this year was the proper support of pyproject.toml. I am slowly replacing all setup.py and setup.cfg in my projects with pyproject.toml and haven t discovered any issues yet. Even packaging those packages as Debian packages is well-supported by Debian s tooling. I m still running a quite boring stack based on pyproject.toml and requirements.txt, ignoring more advanced tools like poetry or such for dependency management. My build-system defined in pyproject.toml requires setuptools and I m using build for building and twine for uploading. Since PyPA changed the packaging tutorial towards pyproject.toml and away from setup.py, I think we will slowly see setup.py and setup.cfg go away over the years. Speaking of PyPA, I m happy that they changed their attitude towards a more unbiased recommendation of tooling.

15 December 2022

Reproducible Builds: Supporter spotlight: David A. Wheeler on supply chain security

The Reproducible Builds project relies on several projects, supporters and sponsors for financial support, but they are also valued as ambassadors who spread the word about our project and the work that we do. This is the sixth instalment in a series featuring the projects, companies and individuals who support the Reproducible Builds project. We started this series by featuring the Civil Infrastructure Platform project and followed this up with a post about the Ford Foundation as well as a recent ones about ARDC, the Google Open Source Security Team (GOSST), Jan Nieuwenhuizen on Bootstrappable Builds, GNU Mes and GNU Guix and Hans-Christoph Steiner of the F-Droid project. Today, however, we will be talking with David A. Wheeler, the Director of Open Source Supply Chain Security at the Linux Foundation.

Holger Levsen: Welcome, David, thanks for taking the time to talk with us today. First, could you briefly tell me about yourself? David: Sure! I m David A. Wheeler and I work for the Linux Foundation as the Director of Open Source Supply Chain Security. That just means that my job is to help open source software projects improve their security, including its development, build, distribution, and incorporation in larger works, all the way out to its eventual use by end-users. In my copious free time I also teach at George Mason University (GMU); in particular, I teach a graduate course on how to design and implement secure software. My background is technical. I have a Bachelor s in Electronics Engineering, a Master s in Computer Science and a PhD in Information Technology. My PhD dissertation is connected to reproducible builds. My PhD dissertation was on countering the Trusting Trust attack, an attack that subverts fundamental build system tools such as compilers. The attack was discovered by Karger & Schell in the 1970s, and later demonstrated & popularized by Ken Thompson. In my dissertation on trusting trust I showed that a process called Diverse Double-Compiling (DDC) could detect trusting trust attacks. That process is a specialized kind of reproducible build specifically designed to detect trusting trust style attacks. In addition, countering the trusting trust attack primarily becomes more important only when reproducible builds become more common. Reproducible builds enable detection of build-time subversions. Most attackers wouldn t bother with a trusting trust attack if they could just directly use a build-time subversion of the software they actually want to subvert.
Holger: Thanks for taking the time to introduce yourself to us. What do you think are the biggest challenges today in computing? There are many big challenges in computing today. For example:
Holger: Do you think reproducible builds are an important part in secure computing today already? David: Yes, but first let s put things in context. Today, when attackers exploit software vulnerabilities, they re primarily exploiting unintentional vulnerabilities that were created by the software developers. There are a lot of efforts to counter this: We re just starting to get better at this, which is good. However, attackers always try to attack the easiest target. As our deployed software has started to be hardened against attack, attackers have dramatically increased their attacks on the software supply chain (Sonatype found in 2022 that there s been a 742% increase year-over-year). The software supply chain hasn t historically gotten much attention, making it the easy target. There are simple supply chain attacks with simple solutions: Unfortunately, attackers know there are other lines of attack. One of the most dangerous is subverted build systems, as demonstrated by the subversion of SolarWinds Orion system. In a subverted build system, developers can review the software source code all day and see no problem, because there is no problem there. Instead, the process to convert source code into the code people run, called the build system , is subverted by an attacker. One solution for countering subverted build systems is to make the build systems harder to attack. That s a good thing to do, but you can never be confident that it was good enough . How can you be sure it s not subverted, if there s no way to know? A stronger defense against subverted build systems is the idea of verified reproducible builds. A build is reproducible if given the same source code, build environment and build instructions, any party can recreate bit-by-bit identical copies of all specified artifacts. A build is verified if multiple different parties verify that they get the same result for that situation. When you have a verified reproducible build, either all the parties colluded (and you could always double-check it yourself), or the build process isn t subverted. There is one last turtle: What if the build system tools or machines are subverted themselves? This is not a common attack today, but it s important to know if we can address them when the time comes. The good news is that we can address this. For some situations reproducible builds can also counter such attacks. If there s a loop (that is, a compiler is used to generate itself), that s called the trusting trust attack, and that is more challenging. Thankfully, the trusting trust attack has been known about for decades and there are known solutions. The diverse double-compiling (DDC) process that I explained in my PhD dissertation, as well as the bootstrappable builds process, can both counter trusting trust attacks in the software space. So there is no reason to lose hope: there is a bottom turtle , as it were.
Holger: Thankfully, this has all slowly started to change and supply chain issues are now widely discussed, as evident by efforts like Securing the Software Supply Chain: Recommended Practices Guide for Developers which you shared on our mailing list. In there, Reproducible Builds are mentioned as recommended advanced practice, which is both pretty cool (we ve come a long way!), but to me it also sounds like this will take another decade until it s become standard normal procedure. Do you agree on that timeline? David: I don t think there will be any particular timeframe. Different projects and ecosystems will move at different speeds. I wouldn t be surprised if it took a decade or so for them to become relatively common there are good reasons for that. Today the most common kinds of attacks based on software vulnerabilities still involve unintentional vulnerabilities in operational systems. Attackers are starting to apply supply chain attacks, but the top such attacks today are typosquatting (creating packages with similar names) and dependency confusion) (convincing projects to download packages from the wrong repositories). Reproducible builds don t counter those kinds of attacks, they counter subverted builds. It s important to eventually have verified reproducible builds, but understandably other issues are currently getting prioritized first. That said, reproducible builds are important long term. Many people are working on countering unintentional vulnerabilities and the most common kinds of supply chain attacks. As these other threats are countered, attackers will increasingly target build systems. Attackers always go for the weakest link. We will eventually need verified reproducible builds in many situations, and it ll take a while to get build systems able to widely perform reproducible builds, so we need to start that work now. That s true for anything where you know you ll need it but it will take a long time to get ready you need to start now.
Holger: What are your suggestions to accelerate adoption? David: Reproducible builds need to be: I think there s a snowball effect. Once many projects packages are reproducible, it will be easier to convince other projects to make their packages reproducible. I also think there should be some prioritization. If a package is in wide use (e.g., part of minimum set of packages for a widely-used Linux distribution or framework), its reproducibility should be a special focus. If a package is vital for supporting some societally important critical infrastructure (e.g., running dams), it should also be considered important. You can then work on the ones that are less important over time.
Holger: How is the Best Practices Badge going? How many projects are participating and how many are missing? David: It s going very well. You can see some automatically-generated statistics, showing we have over 5,000 projects, adding more than 1/day on average. We have more than 900 projects that have earned at least the passing badge level.
Holger: How many of the projects participating in the Best Practices badge engaging with reproducible builds? David: As of this writing there are 168 projects that report meeting the reproducible builds criterion. That s a relatively small percentage of projects. However, note that this criterion (labelled build_reproducible) is only required for the gold badge. It s not required for the passing or silver level badge. Currently we ve been strategically focused on getting projects to at least earn a passing badge, and less on earning silver or gold badges. We would love for all projects to get earn a silver or gold badge, of course, but our theory is that projects that can t even earn a passing badge present the most risk to their users. That said, there are some projects we especially want to see implementing higher badge levels. Those include projects that are very widely used, so that vulnerabilities in them can impact many systems. Examples of such projects include the Linux kernel and curl. In addition, some projects are used within systems where it s important to society that they not have serious security vulnerabilities. Examples include projects used by chemical manufacturers, financial systems and weapons. We definitely encourage any of those kinds of projects to earn higher badge levels.
Holger: Many thanks for this interview, David, and for all of your work at the Linux Foundation and elsewhere!



For more information about the Reproducible Builds project, please see our website at reproducible-builds.org. If you are interested in ensuring the ongoing security of the software that underpins our civilisation and wish to sponsor the Reproducible Builds project, please reach out to the project by emailing contact@reproducible-builds.org.

14 December 2022

Russ Allbery: Review: Contact

Review: Contact, by Carl Sagan
Publisher: Pocket Books
Copyright: 1985
Printing: October 1986
ISBN: 0-671-43422-5
Format: Mass market
Pages: 434
Contact is a standalone first-contact science fiction novel. Carl Sagan (1934 1996) was best known as a non-fiction writer, astronomer, and narrator of the PBS non-fiction program Cosmos. This is his first and only novel. Ellie Arroway is the director of Project Argus, a radio telescope array in the New Mexico desert whose primary mission is SETI: the search for extra-terrestrial intelligence by scanning the skies for unexpected radio signals. Its assignment to SETI is controversial; there are radio astronomy projects waiting, and although 25% of the telescope time is assigned to non-SETI projects, some astronomers think the SETI mission should be scrapped and Argus fully diverted to more useful research. That changes overnight when Argus picks up a signal from Vega, binary pulses representing the sequence of prime numbers. The signal of course doesn't stop when the Earth rotates, so Ellie and her team quickly notify every radio observatory they can get hold of to follow the signal as it passes out of their line of sight. Before long, nearly every country with a radio telescope is involved, and Russian help is particularly vital since they have ship-mounted equipment. The US military and intelligence establishment isn't happy about this and make a few attempts to shove the genie back into the bottle and keep any further discoveries secret, without a lot of success. (Sagan did not anticipate the end of the Cold War, and yet ironically relations with the Russians in his version of the 1990s are warmer by far than they are today. Not that this makes the military types any happier.) For better or worse, making sense of the alien signal becomes a global project. You may be familiar with this book through its 1997 movie adaptation starring Jodie Foster. What I didn't know before reading this book is that it started life as a movie treatment, co-written with Ann Druyan, in 1979. When the movie stalled, Sagan expanded it into a novel. (Given the thanks to Druyan in the author's note, it may not be far wrong to name her as a co-author.) If you've seen the movie, you will have a good idea of what will happen, but the book gives the project a more realistic international scope. Ellie has colleagues carefully selected from all over the world, including for the climactic moment of the story. The biggest problem with Contact as a novel is that Sagan is a non-fiction writer who didn't really know how to write a novel. The long, detailed descriptions of the science and the astronomical equipment fit a certain type of SF story, but the descriptions of the characters, even Ellie, are equally detailed and yet use the same style. The book starts with an account of Ellie's childhood and path into science written like a biography or a magazine profile, not like a novel in which she's the protagonist. The same is true of the other characters: we get characterization of a sort, but the tone ranges from Wikipedia article to long-form essay and never quite feels like a story. Ellie is the most interesting character in the book, partly because the way Sagan writes her is both distant but oddly compelling. Sagan (or perhaps Druyan?) tries hard to show what life is like for a woman born in the middle of the 20th century who is interested in science and I think mostly succeeds, although Ellie's reactions to sexism felt weirdly emotionless. The descriptions of her relationships are even odder and the parts where this book felt the least like a novel, but Sagan does sell some of that tone as reflective of Ellie's personality. She's a scientist, the work is the center of her life, and everything else, even when important, is secondary. It doesn't entirely make the writing style work, but it helps. Sagan does attempt to give Ellie a personal development arc related to her childhood and her relationships with her father and step-father. I thought the conclusion to that was neither believable nor anywhere near as important as Sagan thought it was, which was off-putting. Better were her ongoing arguments with evangelical Christians, one of whom is a close-minded ass and the other of which is a far more interesting character. They felt wedged into this book, and I'm dubious a realistic version of Ellie would have been the person to have those debates, but it's a subject Sagan clearly has deep personal interest in and that shows in how they're written. The other problem with Contact as a novel is that Sagan does not take science fiction seriously as a genre, instead treating it as a way to explore a thought experiment. To a science fiction reader, or at least to this science fiction reader, the interesting bits of this story involve the aliens. Those are not the bits Sagan is interested in. His attention is on how this sort of contact, and project, would affect humanity and human politics. We do get some more traditional science fiction near the end of the book, but Sagan then immediately backs away from it, minimizes that part of the story, and focuses exclusively on the emotional and philosophical implications for humans of his thought experiment. Since I found his philosophical musings about agnosticism and wonder and discovery less interesting than the actual science fiction bits, I found this somewhat annoying. The ending felt a bit more like a cheap trick than a satisfying conclusion. Interestingly, this entire novel is set in an alternate universe, for reasons entirely unexplained (at least that I noticed) in the book. It's set in the late 1990s but was written in 1985, so of course this is an alternate future, but the 1985 of this world still isn't ours. Yuri Gagarin was the first man to set foot on the moon, and the space program and the Cold War developed in subtly different ways. I'm not sure why Sagan made that choice, but it felt to me like he was separating his thought experiment farther from our world to give the ending more plausible deniability. There are, at the time of the novel, permanent orbital colonies for (mostly) rich people, because living in space turns out to greatly extend human lifespans. That gives Sagan an opportunity to wax poetic about the life-altering effects of seeing Earth from space, which in his alternate timeline rapidly sped up nuclear disarmament and made the rich more protective of the planet. This is an old bit of space boosterism that isn't as common these days, mostly because it's become abundantly clear that human psychology doesn't work this way. Sadly, rich sociopaths remain sociopaths even when you send them into space. I was a bit torn between finding Sagan's predictions charmingly hopeful and annoyingly daft. I don't think this novel is very successful as a novel. It's much longer than it needs to be and long parts of it drag. But it's still oddly readable; even knowing the rough shape of the ending in advance, I found it hard to put down once the plot properly kicks into gear about two-thirds of the way through. There's a lot in here that I'd argue with Sagan about, but he's thoughtful and makes a serious attempt to work out the political and scientific ramifications of such a discovery in detail. Despite the dry tone, he does a surprisingly good job capturing the anticipation and excitement of a very expensive and risky scientific experiment. I'm not sure I would recommend this book to anyone, but I'm also the person who found Gregory Benford's Timescape to be boring and tedious, despite its rave reviews as a science fiction novel about the practice of science. If that sort of book is more your jam, you may like Contact better than I did. Rating: 6 out of 10

12 December 2022

Dirk Eddelbuettel: digest 0.6.31 on CRAN: snprintf Update

Release 0.6.31 of the digest package arrived at CRAN this weekend, and is being uploaded to Debian as well. digest creates hash digests of arbitrary R objects (using the md5, sha-1, sha-256, sha-512, crc32, xxhash32, xxhash64, murmur32, spookyhash, and blake3 algorithms) permitting easy comparison of R language objects. It is a mature and widely-used as many tasks may involve caching of objects for which it provides convenient general-purpose hash key generation to quickly identify the various objects. This release contains the tiny change (in maybe a dozen places) triggered recently by macOS and now checked for by r-devel which consists of replacing (v)sprintf calls with (v)snprintf. No more, no less. My CRANberries provides the usual 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.

Jonathan McDowell: Setting up FreshRSS in a subdirectory

Ever since the demise of Google Reader I have been looking for a suitable replacement RSS reader. In the past I used to use Liferea but that was when I used a single desktop machine; these days I want to be able to read on my phone and multiple machines. I moved to Feedly and it s been mostly ok, but I m hitting the limit of feeds available in the free tier, and $72/year is a bit more than I can justify to myself. Especially when I have machines already available to me where I could self host something. The problem, of course, is what to host. It seems the best options are all written in PHP, so I had to get over my adverse knee-jerk reaction to that. I ended up on FreshRSS but if it hadn t worked out I d have tried TinyTinyRSS. Of course I m hosting on Debian, and the machine I chose to use was already running nginx and PostgreSQL. So I needed to install PHP: 
$ sudo apt install php7.4-fpm php-curl php-gmp php-intl php-mbstring \
	php-pgsql php-xml php-zip
I put my FreshRSS install in /srv/freshrss so I grabbed the 1.20.2 release from GitHub (actually 1.20.1 at the time, but I ve upgraded to the latest since) and untared it in there. I gave www-data access to the data directory (sudo chown -R www-data /srv/freshrss/data) (yes, yes, I could have created a new user specifically for FreshRSS, but I ve chosen not to for now). There s no actual need to configure things up on the filesystem, you can do the initial setup from the web interface. Which is where the trouble came. I ve been an Apache user since 1998 and as a result it s what I know and what I go to. nginx is new to me. And I wanted my FreshRSS instance to live in a subdirectory of an existing TLS-enabled host, rather than have it s own hostname. Now, at least FreshRSS copes with this (unlike far too many other projects), you just have to configure your webserver correctly. Which took me more experimentation than I d like, but I ve ended up with the following snippet: 
    # PHP files handling
    location ~ ^/freshrss/.+?\.php(/.*)?$  
        root /srv/freshrss/p;
        fastcgi_pass unix:/run/php/php-fpm.sock;
        fastcgi_split_path_info ^/freshrss(/.+\.php)(/.*)?$;
        set $path_info $fastcgi_path_info;
        fastcgi_param PATH_INFO $path_info;
        include fastcgi_params;
        fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
     
    location ~ ^/freshrss(/.*)?$  
        root /srv/freshrss/p;
        try_files $1 /freshrss$1/index.php$is_args$args;
Other than the addition of the freshrss prefix this ends up differing slightly from the FreshRSS webserver configuration example. I ended up having to make the path info on the fastcgi_split_path_info optional, and my try_files in the bare directory location directive needed $is_args$args added or I just ended up in a redirect loop because the session parameters didn t get passed through. I m sure there s a better way to do it, but I did a bunch of searching and this is how I ended up making it work. Before firing up the web configuration I created a suitable database: 
$ sudo -Hu postgres psql
psql (13.8 (Debian 13.8-0+deb11u1))
Type "help" for help.
postgres=# create database freshrss;
CREATE DATABASE
postgres=# create user freshrss with encrypted password 'hunter2';
CREATE ROLE
postgres=# grant all privileges on database freshrss to freshrss;
GRANT
postgres=# \q
I ran through the local configuration, creating myself a user and adding some feeds, then created a cronjob to fetch updates hourly and keep a log: 
# mkdir /var/log/freshrss
# chown :www-data /var/log/freshrss
# chmod 775 /var/log/freshrss
# cat > /etc/cron.d/freshrss-refresh <EOF
33 * * * * www-data /srv/freshrss/app/actualize_script.php > /var/log/freshrss/update-$(date --iso-8601=minutes).log 2>&1
EOF
Experiences so far? Reasonably happy. The interface seems snappy enough, and works well both on mobile and desktop. I m only running a single user instance at present, but am considering opening it up to some other folk and will see how that scales. And it clearly indicated a number of my feeds that were broken, so I ve cleaned some up that are still around and deleted the missing ones. Now I just need to figure out what else I should be subscribed to that I ve been putting off due to the Feedly limit!

Russ Allbery: Review: The Unbroken

Review: The Unbroken, by C.L. Clark
Series: Magic of the Lost #1
Publisher: Orbit
Copyright: March 2021
ISBN: 0-316-54267-9
Format: Kindle
Pages: 490
The Unbroken is the first book of a projected fantasy trilogy. It is C.L. Clark's first novel. Lieutenant Touraine is one of the Sands, the derogatory name for the Balladairan Colonial Brigade. She, like the others of her squad, are conscript soldiers, kidnapped by the Balladairan Empire from their colonies as children and beaten into "civilized" behavior by Balladairan training. They fought in the Balladairan war against the Taargens. Now, they've been reassigned to El-Wast, capital city of Qaz l, the foremost of the southern colonies. The place where Touraine was born, from which she was taken at the age of five. Balladaire is not France and Qaz l is not Algeria, but the parallels are obvious and strongly implied by the map and the climates. Touraine and her squad are part of the forces accompanying Princess Luca, the crown princess of the Balladairan Empire, who has been sent to take charge of Qaz l and quell a rebellion. Luca's parents died in the Withering, the latest round of a recurrent plague that haunts Balladaire. She is the rightful heir, but her uncle rules as regent and is reluctant to give her the throne. Qaz l is where she is to prove herself. If she can bring the colony in line, she can prove that she's ready to rule: her birthright and her destiny. The Qaz li are uninterested in being part of Luca's grand plan of personal accomplishment. She steps off her ship into an assassination attempt, foiled by Touraine's sharp eyes and quick reactions, which brings the Sand to the princess's attention. Touraine's reward is to be assigned the execution of the captured rebels, one of whom recognizes her and names her mother before he dies. This sets up the core of the plot: Qaz li rebellion against an oppressive colonial empire, Luca's attempt to use the colony as a political stepping stone, and Touraine caught in between. One of the reasons why I am happy to see increased diversity in SFF authors is that the way we tell stories is shaped by our cultural upbringing. I was taught to tell stories about colonialism and rebellion in a specific ideological shape. It's hard to describe briefly, but the core idea is that being under the rule of someone else is unnatural as well as being an injustice. It's a deviation from the way the world should work, something unexpected that is inherently unstable. Once people unite to overthrow their oppressors, eventual success is inevitable; it's not only right or moral, it's the natural path of history. This is what you get when you try to peel the supremacy part away from white supremacy but leave the unshakable self-confidence and bedrock assumption that the universe cares what we think. We were also taught that rebellion is primarily ideological. One may be motivated by personal injustice, but the correct use of that injustice is to subsume it into concepts such as freedom and democracy. Those concepts are more "real" in some foundational sense, more central to the right functioning of the world, than individual circumstance. When the now-dominant group tells stories of long-ago revolution, there is no personal experience of oppression and survival in which to ground the story; instead, it's linked to anticipatory fear in the reader, to the idea that one's privileges could be taken away by a foreign oppressor and that the counter to this threat is ideological unity. Obviously, not every white fantasy author uses this story shape, but the tendency runs deep because we're taught it young. You can see it everywhere in fantasy, from Lord of the Rings to Tigana. The Unbroken uses a much different story shape, and I don't think it's a coincidence that the author is Black. Touraine is not sympathetic to the Qaz li. These are not her people and this is not her life. She went through hell in Balladairan schools, but she won a place, however tenuous. Her personal role model is General Cantic, the Balladairan Blood General who was also one of her instructors. Cantic is hard as nails, unforgiving, unbending, and probably a war criminal, but also the embodiment of a military ethic. She is tough but fair with the conscript soldiers. She doesn't put a stop to their harassment by the regular Balladairan troops, but neither does she let it go too far. Cantic has power, she knows how to keep it, and there is a place for Touraine in Cantic's world. And, critically, that place is not just hers: it's one she shares with her squad. Touraine's primary loyalty is not to Balladaire or to Qaz l. It's to the Sands. Her soldiers are neither one thing nor the other, and they disagree vehemently among themselves about what Qaz l and their other colonial homes should be to them, but they learned together, fought together, and died together. That theme is woven throughout The Unbroken: personal bonds, third and fourth loyalties, and practical ethics of survival that complicate and contradict simple dichotomies of oppressor and oppressed. Touraine is repeatedly offered ideological motives that the protagonist in the typical story shape would adopt. And she repeatedly rejects them for personal bonds: trying to keep her people safe, in a world that is not looking out for them. The consequence is that this book tears Touraine apart. She tries to walk a precarious path between Luca, the Qaz li, Cantic, and the Sands, and she falls off that path a lot. Each time I thought I knew where this book was going, there's another reversal, often brutal. I tend to be a happily-ever-after reader who wants the protagonist to get everything they need, so this isn't my normal fare. The amount of hell that Touraine goes through made for difficult reading, worse because much of it is due to her own mistakes or betrayals. But Clark makes those decisions believable given the impossible position Touraine is in and the lack of role models she has for making other choices. She's set up to fail, and the price of small victories is to have no one understand the decisions that she makes, or to believe her motives. Luca is the other viewpoint character of the book (and yes, this is also a love affair, which complicates both of their loyalties). She is the heroine of a more typical genre fantasy novel: the outsider princess with a physical disability and a razor-sharp mind, ambitious but fair (at least in her own mind), with a trusted bodyguard advisor who also knew her father and a sincere desire to be kinder and more even-handed in her governance of the colony. All of this is real; Luca is a protagonist, and the reader is not being set up to dislike her. But compared to Touraine's grappling with identity, loyalty, and ethics, Luca is never in any real danger, and her concerns start to feel too calculated and superficial. It's hard to be seriously invested in whether Luca proves herself or gets her throne when people are being slaughtered and abused. This, I think, is the best part of this book. Clark tells a traditional ideological fantasy of learning to be a good ruler, but she puts it alongside a much deeper and more complex story of multi-faceted oppression. She has the two protagonists fall in love with each other and challenges them to understand each other, and Luca does not come off well in this comparison. Touraine is frustrated, impulsive, physical, and sometimes has catastrophically poor judgment. Luca is analytical and calculating, and in most ways understands the political dynamics far better than Touraine. We know how this story usually goes: Luca sees Touraine's brilliance and lifts her out of the ranks into a role of importance and influence, which Touraine should reward with loyalty. But Touraine's world is more real, more grounded, and more authentic, and both Touraine and the reader know what Luca could offer is contingent and comes with a higher price than Luca understands. (Incidentally, the cover of The Unbroken, designed by Lauren Panepinto with art by Tommy Arnold, is astonishingly good at capturing both Touraine's character and the overall feeling of the book. Here's a larger version.) The writing is good but uneven. Clark loves reversals, and they did keep me reading, but I think there were too many of them. By the end of the book, the escalation of betrayals and setbacks was more exhausting than exciting, and I'd stopped trusting anything good would last. (Admittedly, this is an accurate reflection of how Touraine felt.) Touraine's inner monologue also gets a bit repetitive when she's thrashing in the jaws of an emotional trap. I think some of this is first-novel problems of over-explaining emotional states and character reasoning, but these problems combine to make the book feel a bit over-long. I'm also not in love with the ending. It's perhaps the one place in the book where I am more cynical about the politics than Clark is, although she does lay the groundwork for it. But this book is also full of places small and large where it goes a different direction than most fantasy and is better for it. I think my favorite small moment is Touraine's quiet refusal to defend herself against certain insinuations. This is such a beautiful bit of characterization; she knows she won't be believed anyway, and refuses to demean herself by trying. I'm not sure I can recommend this book unconditionally, since I think you have to be in the mood for it, but it's one of the most thoughtful and nuanced looks at colonialism and rebellion I can remember seeing in fantasy. I found it frustrating in places, but I'm also still thinking about it. If you're looking for a political fantasy with teeth, you could do a lot worse, although expect to come out the other side a bit battered and bruised. Followed by The Faithless, and I have no idea where Clark is going to go with the second book. I suppose I'll have to read and find out. Content note: In addition to a lot of violence, gore, and death, including significant character death, there's also a major plague. If you're not feeling up to reading about panic caused by contageous illness, proceed with caution. Rating: 7 out of 10

11 December 2022

Vincent Bernat: Akvorado: a flow collector, enricher, and visualizer

Earlier this year, we released Akvorado, a flow collector, enricher, and visualizer. It receives network flows from your routers using either NetFlow v9, IPFIX, or sFlow. Several pieces of information are added, like GeoIP and interface names. The flows are exported to Apache Kafka, a distributed queue, then stored inside ClickHouse, a column-oriented database. A web frontend is provided to run queries. A live version is available for you to play.
Akvorado web interface displays the result of a simple query using stacked areas
Akvorado s web frontend
Several alternatives exist: Akvorado differentiates itself from these solutions because: The proposed deployment solution relies on Docker Compose to set up Akvorado, Zookeeper, Kafka, and ClickHouse. I hope it should be enough for anyone to get started quickly. Akvorado is performant enough to handle 100 000 flows per second with 64 GB of RAM and 24 vCPU. With 2 TB of disk, you should expect to keep data for a few years. I spent some time writing a fairly complete documentation. It seems redundant to repeat its content in this blog post. There is also a section about its internal design if you are interested in how it is built. I also did a FRnOG presentation earlier this year, and a ClickHouse meetup presentation, which focuses more on how ClickHouse is used. I plan to write more detailed articles on specific aspects of Akvorado. Stay tuned!
  1. While the collector could write directly to the database, the queue buffers flows if the database is unavailable. It also enables you to process flows with another piece of software (like an anti-DDoS system).

10 December 2022

Simon Josefsson: Trisquel 11 on NV41PZ: First impressions

My NovaCustom NV41PZ laptop arrived a couple of days ago, and today I had some time to install it. You may want to read about my purchasing decision process first. I expected a rough ride to get it to work, given the number of people claiming that modern laptops can t run fully free operating systems. I first tried the Trisquel 10 live DVD and it booted fine including network, but the mouse trackpad did not work. Before investigating it, I noticed a forum thread about Trisquel 11 beta3 images, and being based on Ubuntu 22.04 LTS and has Linux-libre 5.15 it seemed better to start with more modern software. After installing through the live DVD successfully, I realized I didn t like MATE but wanted to keep using GNOME. I reverted back to installing a minimal environment through the netinst image, and manually installed GNOME (apt-get install gnome) since I prefer that over MATE, together with a bunch of other packages. I ve been running it for a couple of hours now, and here is a brief summary of the hardware components that works.
CPUAlder Lake Intel i7-1260P
Memory2x32GB Kingston DDR4 SODIMM 3200MHz
StorageSamsung 980 Pro 2TB NVME
BIOSDasharo Coreboot
GraphicsIntel Xe
Screen (internal)14 1920 1080
Screen (HDMI)Connected to Dell 27 2560 1440
Screen (USB-C)Connected to Dell 27 2560 1440 via Wavlink port extender
WebcamBuiltin 1MP Camera
MicrophoneIntel Alder Lake
KeyboardISO layout, all function keys working
MouseTrackpad, tap clicking and gestures
Ethernet RJ45Realtek RTL8111/8168/8411 with r8169 driver
Memory cardO2 Micro comes up as /dev/mmcblk0
Docking stationWavlink 4xUSB, 2xHDMI, DP, RJ45,
ConnectivityUSB-A, USB-C
AudioIntel Alder Lake
Hardware components and status
So what s not working? Unfortunately, NovaCustom does not offer any WiFi or Bluetooth module that is compatible with Trisquel, so the AX211 (1675x) Wifi/Bluetooth card in it is just dead weight. I imagine it would be possible to get the card to work if non-free firmware is loaded. I don t need Bluetooth right now, and use the Technoetic N-150 USB WiFi dongle when I m not connected to wired network. Compared against my X201, the following factors have improved. I m still unhappy about the following properties with both the NV41PZ and the X201. Hopefully my next laptop will have improved on this further. I hope to be able to resolve the WiFi part by replacing the WiFi module, there appears to be options available but I have not tested them on this laptop yet. Does anyone know of a combined WiFi and Bluetooth M.2 module that would work on Trisquel? While I haven t put the laptop to heavy testing yet, everything that I would expect a laptop to be able to do seems to work fine. Including writing this blog post!

Matthew Garrett: On-device WebAuthn and what makes it hard to do well

WebAuthn improves login security a lot by making it significantly harder for a user's credentials to be misused - a WebAuthn token will only respond to a challenge if it's issued by the site a secret was issued to, and in general will only do so if the user provides proof of physical presence[1]. But giving people tokens is tedious and also I have a new laptop which only has USB-C but does have a working fingerprint reader and I hate the aesthetics of the Yubikey 5C Nano, so I've been thinking about what WebAuthn looks like done without extra hardware.

Let's talk about the broad set of problems first. For this to work you want to be able to generate a key in hardware (so it can't just be copied elsewhere if the machine is compromised), prove to a remote site that it's generated in hardware (so the remote site isn't confused about what security assertions you're making), and tie use of that key to the user being physically present (which may range from "I touched this object" to "I presented biometric evidence of identity"). What's important here is that a compromised OS shouldn't be able to just fake a response. For that to be possible, the chain between proof of physical presence to the secret needs to be outside the control of the OS.

For a physical security token like a Yubikey, this is pretty easy. The communication protocol involves the OS passing a challenge and the source of the challenge to the token. The token then waits for a physical touch, verifies that the source of the challenge corresponds to the secret it's being asked to respond to the challenge with, and provides a response. At the point where keys are being enrolled, the token can generate a signed attestation that it generated the key, and a remote site can then conclude that this key is legitimately sequestered away from the OS. This all takes place outside the control of the OS, meeting all the goals described above.

How about Macs? The easiest approach here is to make use of the secure enclave and TouchID. The secure enclave is a separate piece of hardware built into either a support chip (for x86-based Macs) or directly on the SoC (for ARM-based Macs). It's capable of generating keys and also capable of producing attestations that said key was generated on an Apple secure enclave ("Apple Anonymous Attestation", which has the interesting property of attesting that it was generated on Apple hardware, but not which Apple hardware, avoiding a lot of privacy concerns). These keys can have an associated policy that says they're only usable if the user provides a legitimate touch on the fingerprint sensor, which means it can not only assert physical presence of a user, it can assert physical presence of an authorised user. Communication between the fingerprint sensor and the secure enclave is a private channel that the OS can't meaningfully interfere with, which means even a compromised OS can't fake physical presence responses (eg, the OS can't record a legitimate fingerprint press and then send that to the secure enclave again in order to mimic the user being present - the secure enclave requires that each response from the fingerprint sensor be unique). This achieves our goals.

The PC space is more complicated. In the Mac case, communication between the biometric sensors (be that TouchID or FaceID) occurs in a controlled communication channel where all the hardware involved knows how to talk to the other hardware. In the PC case, the typical location where we'd store secrets is in the TPM, but TPMs conform to a standardised spec that has no understanding of this sort of communication, and biometric components on PCs have no way to communicate with the TPM other than via the OS. We can generate keys in the TPM, and the TPM can attest to those keys being TPM-generated, which means an attacker can't exfiltrate those secrets and mimic the user's token on another machine. But in the absence of any explicit binding between the TPM and the physical presence indicator, the association needs to be up to code running on the CPU. If that's in the OS, an attacker who compromises the OS can simply ask the TPM to respond to an challenge it wants, skipping the biometric validation entirely.

Windows solves this problem in an interesting way. The Windows Hello Enhanced Signin doesn't add new hardware, but relies on the use of virtualisation. The agent that handles WebAuthn responses isn't running in the OS, it's running in another VM that's entirely isolated from the OS. Hardware that supports this model has a mechanism for proving its identity to the local code (eg, fingerprint readers that support this can sign their responses with a key that has a certificate that chains back to Microsoft). Additionally, the secrets that are associated with the TPM can be held in this VM rather than in the OS, meaning that the OS can't use them directly. This means we have a flow where a browser asks for a WebAuthn response, that's passed to the VM, the VM asks the biometric device for proof of user presence (including some sort of random value to prevent the OS just replaying that), receives it, and then asks the TPM to generate a response to the challenge. Compromising the OS doesn't give you the ability to forge the responses between the biometric device and the VM, and doesn't give you access to the secrets in the TPM, so again we meet all our goals.

On Linux (and other free OSes), things are less good. Projects like tpm-fido generate keys on the TPM, but there's no secure channel between that code and whatever's providing proof of physical presence. An attacker who compromises the OS may not be able to copy the keys to their own system, but while they're on the compromised system they can respond to as many challenges as they like. That's not the same security assertion we have in the other cases.

Overall, Apple's approach is the simplest - having binding between the various hardware components involved means you can just ignore the OS entirely. Windows doesn't have the luxury of having as much control over what the hardware landscape looks like, so has to rely on virtualisation to provide a security barrier against a compromised OS. And in Linux land, we're fucked. Who do I have to pay to write a lightweight hypervisor that runs on commodity hardware and provides an environment where we can run this sort of code?


[1] As I discussed recently there are scenarios where these assertions are less strong, but even so

comment count unavailable comments

9 December 2022

Russell Coker: USB-PD and GaN

photo of 2 USB-PD chargers A recent development is cheap Gallium Nitride based power supplies that provide better efficiency in a smaller space than other technologies. Kogan recently had a special on such devices so I decided to try them out with my new Thinkpad X1 Carbon Gen 5 [1]. Google searches for power supplies for that Thinkpad included results for 30W PSUs which implies that any 30W USB-C PSU should work. I bought a 30W charger for $10 that can supply 15V/2A or 20V/1.5A on a single USB-C port or 15W on the USB-C port and 15W on the USB-2 port at the same time and expected it to work as a laptop charger. Unfortunately it didn t, I don t know whether the adverts for 30W Thinkpad PSUs were false or whether the claim of the GaN charger I bought being 30W was false, all I know is that the KDE power applet said that the PSU couldn t supply enough power. I then bought a 68W charger for $28 that can supply 20.0V/3.0A on a single USB-C port if the USB-2 port isn t used and 50W on the USB-C port if the USB-2 port is also being used. This worked well which wasn t a great surprise as I had previously run the laptop on 45W PSUs. If I connect a phone to the USB-2 port while the laptop is being charged then the laptop will be briefly cut off, presumably the voltage and current are being renegotiated when that happens. As you can see the 68W charger is significantly larger than the 30W charger, but still small enough to easily fit in a jacket pocket and smaller than a regular laptop charger. One of my uses for this will be to put it in a jacket pocket when I have my laptop in another pocket. Another use will be for charging in my car as the cables from the inverter to convert 12VDC to 240VAC takes enough space. I will probably get a ~50W USB-PD charger that connects to a car cigarette lighter socket when a GaN version of such a charger becomes available.
Related posts:
  1. Thinkpad X1 Carbon Gen 6 In February I reviewed a Thinkpad X1 Carbon Gen 1...
  2. Thinkpad X1 Carbon Gen5 Gen1 Since February 2018 I have been using a Thinkpad...
  3. Power Supplies and Wires For some time I ve been wondering how the wire size...

Dirk Eddelbuettel: AsioHeaders 1.22.1-2 on CRAN: Small Update

An new minor revision of the AsioHeaders package arrived at CRAN earlier today. Asio provides a cross-platform C++ library for network and low-level I/O programming. It is also included in Boost but requires linking when used as part of Boost. This standalone version of Asio is a header-only C++ library which can be used without linking (just like our BH package with parts of Boost). This minor update avoid use of (v)sprintf which CRAN now flags in r-devel (for all R builds), following the decision by Apple to deprecated it for macOS. Winston had notified me about email he had gotten for for his websocket package also highlighting the issue for both iptools and ipaddress. As the issue ticket dialog shows I was initially a little hamfisted about replicating, falsely thinking I would need an updated compiler. But this really is just a change in r-devel once again scanning shared libraries for symbols now warned about. Upstream has newer minor releases but they did not yet cover this; however I found a commit mentioning Xcode and snprint from three days ago which I essentially ported. We needed one more change, and that addressed the issue in websocket. But as it is good to increase the number of random acts of kindness, I also looked into iptools and ipaddress as CRAN has its eyes on them too for this. Turned out they needed simple and limited changes from sprint to snprintf so made those and sent them PRs: iptools PR #42 and ipaddress PR #79. The short NEWS entry for AsioHeaders follows.

Changes in version 1.22.1-2 (2022-12-07)
  • Switch a handful of upstream (v)sprint uses to (v)snprintf.

Thanks to my CRANberries, there is also a diffstat report relative to the previous release. Comments and suggestions about AsioHeaders are welcome via the issue tracker at 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.

8 December 2022

Reproducible Builds: Reproducible Builds in November 2022

Welcome to yet another report from the Reproducible Builds project, this time for November 2022. In all of these reports (which we have been publishing regularly since May 2015) we attempt to outline the most important things that we have been up to over the past month. As always, if you interested in contributing to the project, please visit our Contribute page on our website.

Reproducible Builds Summit 2022 Following-up from last month s report about our recent summit in Venice, Italy, a comprehensive report from the meeting has not been finalised yet watch this space! As a very small preview, however, we can link to several issues that were filed about the website during the summit (#38, #39, #40, #41, #42, #43, etc.) and collectively learned about Software Bill of Materials (SBOM) s and how .buildinfo files can be seen/used as SBOMs. And, no less importantly, the Reproducible Builds t-shirt design has been updated

Reproducible Builds at European Cyber Week 2022 During the European Cyber Week 2022, a Capture The Flag (CTF) cybersecurity challenge was created by Fr d ric Pierret on the subject of Reproducible Builds. The challenge consisted in a pedagogical sense based on how to make a software release reproducible. To progress through the challenge issues that affect the reproducibility of build (such as build path, timestamps, file ordering, etc.) were to be fixed in steps in order to get the final flag in order to win the challenge. At the end of the competition, five people succeeded in solving the challenge, all of whom were awarded with a shirt. Fr d ric Pierret intends to create similar challenge in the form of a how to in the Reproducible Builds documentation, but two of the 2022 winners are shown here:

On business adoption and use of reproducible builds Simon Butler announced on the rb-general mailing list that the Software Quality Journal published an article called On business adoption and use of reproducible builds for open and closed source software. This article is an interview-based study which focuses on the adoption and uses of Reproducible Builds in industry, with a focus on investigating the reasons why organisations might not have adopted them:
[ ] industry application of R-Bs appears limited, and we seek to understand whether awareness is low or if significant technical and business reasons prevent wider adoption.
This is achieved through interviews with software practitioners and business managers, and touches on both the business and technical reasons supporting the adoption (or not) of Reproducible Builds. The article also begins with an excellent explanation and literature review, and even introduces a new helpful analogy for reproducible builds:
[Users are] able to perform a bitwise comparison of the two binaries to verify that they are identical and that the distributed binary is indeed built from the source code in the way the provider claims. Applied in this manner, R-Bs function as a canary, a mechanism that indicates when something might be wrong, and offer an improvement in security over running unverified binaries on computer systems.
The full paper is available to download on an open access basis. Elsewhere in academia, Beatriz Michelson Reichert and Rafael R. Obelheiro have published a paper proposing a systematic threat model for a generic software development pipeline identifying possible mitigations for each threat (PDF). Under the Tampering rubric of their paper, various attacks against Continuous Integration (CI) processes:
An attacker may insert a backdoor into a CI or build tool and thus introduce vulnerabilities into the software (resulting in an improper build). To avoid this threat, it is the developer s responsibility to take due care when making use of third-party build tools. Tampered compilers can be mitigated using diversity, as in the diverse double compiling (DDC) technique. Reproducible builds, a recent research topic, can also provide mitigation for this problem. (PDF)

Misc news
On our mailing list this month:

Debian & other Linux distributions Over 50 reviews of Debian packages were added this month, another 48 were updated and almost 30 were removed, all of which adds to our knowledge about identified issues. Two new issue types were added as well. [ ][ ]. Vagrant Cascadian announced on our mailing list another online sprint to help clear the huge backlog of reproducible builds patches submitted by performing NMUs (Non-Maintainer Uploads). The first such sprint took place on September 22nd, but others were held on October 6th and October 20th. There were two additional sprints that occurred in November, however, which resulted in the following progress: Lastly, Roland Clobus posted his latest update of the status of reproducible Debian ISO images on our mailing list. This reports that all major desktops build reproducibly with bullseye, bookworm and sid as well as that no custom patches needed to applied to Debian unstable for this result to occur. During November, however, Roland proposed some modifications to live-setup and the rebuild script has been adjusted to fix the failing Jenkins tests for Debian bullseye [ ][ ].
In other news, Miro Hron ok proposed a change to clamp build modification times to the value of SOURCE_DATE_EPOCH. This was initially suggested and discussed on a devel@ mailing list post but was later written up on the Fedora Wiki as well as being officially proposed to Fedora Engineering Steering Committee (FESCo).

Upstream patches The Reproducible Builds project detects, dissects and attempts to 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:

diffoscope diffoscope is our in-depth and content-aware diff utility. Not only can it locate and diagnose reproducibility issues, it can provide human-readable diffs from many kinds of binary formats. This month, Chris Lamb prepared and uploaded versions 226 and 227 to Debian:
  • Support both python3-progressbar and python3-progressbar2, two modules providing the progressbar Python module. [ ]
  • Don t run Python decompiling tests on Python bytecode that file(1) cannot detect yet and Python 3.11 cannot unmarshal. (#1024335)
  • Don t attempt to attach text-only differences notice if there are no differences to begin with. (#1024171)
  • Make sure we recommend apksigcopier. [ ]
  • Tidy generation of os_list. [ ]
  • Make the code clearer around generating the Debian substvars . [ ]
  • Use our assert_diff helper in test_lzip.py. [ ]
  • Drop other copyright notices from lzip.py and test_lzip.py. [ ]
In addition to this, Christopher Baines added lzip support [ ], and FC Stegerman added an optimisation whereby we don t run apktool if no differences are detected before the signing block [ ].
A significant number of changes were made to the Reproducible Builds website and documentation this month, including Chris Lamb ensuring the openEuler logo is correctly visible with a white background [ ], FC Stegerman de-duplicated by email address to avoid listing some contributors twice [ ], Herv Boutemy added Apache Maven to the list of affiliated projects [ ] and boyska updated our Contribute page to remark that the Reproducible Builds presence on salsa.debian.org is not just the Git repository but is also for creating issues [ ][ ]. In addition to all this, however, Holger Levsen made the following changes:
  • Add a number of existing publications [ ][ ] and update metadata for some existing publications as well [ ].
  • Hide draft posts on the website homepage. [ ]
  • Add the Warpforge build tool as a participating project of the summit. [ ]
  • Clarify in the footer that we welcome patches to the website repository. [ ]

Testing framework The Reproducible Builds project operates a comprehensive testing framework at tests.reproducible-builds.org in order to check packages and other artifacts for reproducibility. In October, the following changes were made by Holger Levsen:
  • Improve the generation of meta package sets (used in grouping packages for reporting/statistical purposes) to treat Debian bookworm as equivalent to Debian unstable in this specific case [ ] and to parse the list of packages used in the Debian cloud images [ ][ ][ ].
  • Temporarily allow Frederic to ssh(1) into our snapshot server as the jenkins user. [ ]
  • Keep some reproducible jobs Jenkins logs much longer [ ] (later reverted).
  • Improve the node health checks to detect failures to update the Debian cloud image package set [ ][ ] and to improve prioritisation of some kernel warnings [ ].
  • Always echo any IRC output to Jenkins output as well. [ ]
  • Deal gracefully with problems related to processing the cloud image package set. [ ]
Finally, Roland Clobus continued his work on testing Live Debian images, including adding support for specifying the origin of the Debian installer [ ] and to warn when the image has unmet dependencies in the package list (e.g. due to a transition) [ ].
If you are interested in contributing to the Reproducible Builds project, please visit our Contribute page on our website. You can get in touch with us via:

3 December 2022

Vincent Bernat: Broken commit diff on Cisco IOS XR

TL;DR Never trust show commit changes diff on Cisco IOS XR.

Cisco IOS XR is the operating system running for the Cisco ASR, NCS, and 8000 routers. Compared to Cisco IOS, it features a candidate configuration and a running configuration. In configuration mode, you can modify the first one and issue the commit command to apply it to the running configuration.1 This is a common concept for many NOS. Before committing the candidate configuration to the running configuration, you may want to check the changes that have accumulated until now. That s where the show commit changes diff command2 comes up. Its goal is to show the difference between the running configuration (show running-configuration) and the candidate configuration (show configuration merge). How hard can it be? Let s put an interface down on IOS XR 7.6.2 (released in August 2022): 
RP/0/RP0/CPU0:router(config)#int Hu0/1/0/1 shut
RP/0/RP0/CPU0:router(config)#show commit changes diff
Wed Nov 23 11:08:30.275 CET
Building configuration...
!! IOS XR Configuration 7.6.2
+  interface HundredGigE0/1/0/1
+   shutdown
   !
end
The + sign before interface HundredGigE0/1/0/1 makes it look like you did create a new interface. Maybe there was a typo? No, the diff is just broken. If you look at the candidate configuration, everything is like you expect: 
RP/0/RP0/CPU0:router(config)#show configuration merge int Hu0/1/0/1
Wed Nov 23 11:08:43.360 CET
interface HundredGigE0/1/0/1
 description PNI: (some description)
 bundle id 4000 mode active
 lldp
  receive disable
  transmit disable
 !
 shutdown
 load-interval 30
Here is a more problematic example on IOS XR 7.2.2 (released in January 2021). We want to unconfigure three interfaces: 
RP/0/RP0/CPU0:router(config)#no int GigabitEthernet 0/0/0/5
RP/0/RP0/CPU0:router(config)#int TenGigE 0/0/0/5 shut
RP/0/RP0/CPU0:router(config)#no int TenGigE 0/0/0/28
RP/0/RP0/CPU0:router(config)#int TenGigE 0/0/0/28 shut
RP/0/RP0/CPU0:router(config)#no int TenGigE 0/0/0/29
RP/0/RP0/CPU0:router(config)#int TenGigE 0/0/0/29 shut
RP/0/RP0/CPU0:router(config)#show commit changes diff
Mon Nov  7 15:07:22.990 CET
Building configuration...
!! IOS XR Configuration 7.2.2
-  interface GigabitEthernet0/0/0/5
-   shutdown
   !
+  interface TenGigE0/0/0/5
+   shutdown
   !
   interface TenGigE0/0/0/28
-   description Trunk (some description)
-   bundle id 2 mode active
   !
end
The two first commands are correctly represented by the first two chunks of the diff: we remove GigabitEthernet0/0/0/5 and create TenGigE0/0/0/5. The two next commands are also correctly represented by the last chunk of the diff. TenGigE0/0/0/28 was already shut down, so it is expected that only description and bundle id are removed. However, the diff command forgets about the modifications for TenGigE0/0/0/29. The diff should include a chunk similar to the last one. 
RP/0/RP0/CPU0:router(config)#show run int TenGigE 0/0/0/29
Mon Nov  7 15:07:43.571 CET
interface TenGigE0/0/0/29
 description Trunk to other router
 bundle id 2 mode active
 shutdown
!
RP/0/RP0/CPU0:router(config)#show configuration merge int TenGigE 0/0/0/29
Mon Nov  7 15:07:53.584 CET
interface TenGigE0/0/0/29
 shutdown
!
How can the diff be correct for TenGigE0/0/0/28 but incorrect for TenGigE0/0/0/29 while they have the same configuration? How can you trust the diff command if it forgets part of the configuration? Do you remember the last time you ran an Ansible playbook and discovered the whole router ospf block disappeared without a warning? If you use automation tools, you should check how the diff is assembled. Automation tools should build it from the result of show running-config and show configuration merge. This is what NAPALM does. This is not what cisco.iosxr collection for Ansible does. The problem is not limited to the interface directives. You can get similar issues for other parts of the configuration. For example, here is what we get when removing inactive BGP neighbors on IOS XR 7.2.2: 
RP/0/RP0/CPU0:router(config)#router bgp 65400
RP/0/RP0/CPU0:router(config-bgp)#vrf public
RP/0/RP0/CPU0:router(config-bgp-vrf)#no neighbor 217.29.66.1
RP/0/RP0/CPU0:router(config-bgp-vrf)#no neighbor 217.29.66.75
RP/0/RP0/CPU0:router(config-bgp-vrf)#no neighbor 217.29.66.110
RP/0/RP0/CPU0:router(config-bgp-vrf)#no neighbor 217.29.66.112
RP/0/RP0/CPU0:router(config-bgp-vrf)#no neighbor 217.29.66.158
RP/0/RP0/CPU0:router(config-bgp-vrf)#show commit changes diff
Tue Aug  2 13:58:02.536 CEST
Building configuration...
!! IOS XR Configuration 7.2.2
   router bgp 65400
    vrf public
-    neighbor 217.29.66.1
-     remote-as 16004
-     use neighbor-group MIX_IPV4_PUBLIC
-     description MIX: MIX-IT
     !
-    neighbor 217.29.66.75
-     remote-as 49367
-     use neighbor-group MIX_IPV4_PUBLIC
     !
-    neighbor 217.29.67.10
-     remote-as 19679
     !
-    neighbor 217.29.67.15
-    neighbor 217.29.66.112
-     remote-as 8075
-     use neighbor-group MIX_IPV4_PUBLIC
-     description MIX: Microsoft
-     address-family ipv4 unicast
-      maximum-prefix 1500 95 restart 5
      !
     !
-    neighbor 217.29.66.158
-     remote-as 24482
-     use neighbor-group MIX_IPV4_PUBLIC
-     description MIX: SG.GS
-     address-family ipv4 unicast
     !
     !
    !
   !
end
The only correct chunk is for neighbor 217.29.66.112. All the others are missing some of the removed lines. 217.29.67.15 is even missing all of them. How bad is the code providing such a diff? I could go all day with examples such as these. Cisco TAC is happy to open a case in DDTS, their bug tracker, to fix specific occurrences of this bug.3 However, I fail to understand why the XR team is not just providing the diff between show run and show configuration merge. The output would always be correct!
  1. IOS XR has several limitations. The most inconvenient one is the inability to change the AS number in the router bgp directive. Such a limitation is a great pain for both operations and automation.
  2. This command could have been just show commit, as show commit changes diff is the only valid command you can execute from this point. Starting from IOS XR 7.5.1, show commit changes diff precise is also a valid command. However, I have failed to find any documentation about it and it seems to provide the same output as show commit changes diff. That s how clunky IOS XR can be.
  3. See CSCwa26251 as an example of a fix for something I reported earlier this year. You need a valid Cisco support contract to be able to see its content.

30 November 2022

Matthew Garrett: Making unphishable 2FA phishable

One of the huge benefits of WebAuthn is that it makes traditional phishing attacks impossible. An attacker sends you a link to a site that looks legitimate but isn't, and you type in your credentials. With SMS or TOTP-based 2FA, you type in your second factor as well, and the attacker now has both your credentials and a legitimate (if time-limited) second factor token to log in with. WebAuthn prevents this by verifying that the site it's sending the secret to is the one that issued it in the first place - visit an attacker-controlled site and said attacker may get your username and password, but they won't be able to obtain a valid WebAuthn response.

But what if there was a mechanism for an attacker to direct a user to a legitimate login page, resulting in a happy WebAuthn flow, and obtain valid credentials for that user anyway? This seems like the lead-in to someone saying "The Aristocrats", but unfortunately it's (a) real, (b) RFC-defined, and (c) implemented in a whole bunch of places that handle sensitive credentials. The villain of this piece is RFC 8628, and while it exists for good reasons it can be used in a whole bunch of ways that have unfortunate security consequences.

What is the RFC 8628-defined Device Authorization Grant, and why does it exist? Imagine a device that you don't want to type a password into - either it has no input devices at all (eg, some IoT thing) or it's awkward to type a complicated password (eg, a TV with an on-screen keyboard). You want that device to be able to access resources on behalf of a user, so you want to ensure that that user authenticates the device. RFC 8628 describes an approach where the device requests the credentials, and then presents a code to the user (either on screen or over Bluetooth or something), and starts polling an endpoint for a result. The user visits a URL and types in that code (or is given a URL that has the code pre-populated) and is then guided through a standard auth process. The key distinction is that if the user authenticates correctly, the issued credentials are passed back to the device rather than the user - on successful auth, the endpoint the device is polling will return an oauth token.

But what happens if it's not a device that requests the credentials, but an attacker? What if said attacker obfuscates the URL in some way and tricks a user into clicking it? The user will be presented with their legitimate ID provider login screen, and if they're using a WebAuthn token for second factor it'll work correctly (because it's genuinely talking to the real ID provider!). The user will then typically be prompted to approve the request, but in every example I've seen the language used here is very generic and doesn't describe what's going on or ask the user. AWS simply says "An application or device requested authorization using your AWS sign-in" and has a big "Allow" button, giving the user no indication at all that hitting "Allow" may give a third party their credentials.

This isn't novel! Christoph Tafani-Dereeper has an excellent writeup on this topic from last year, which builds on Nestori Syynimaa's earlier work. But whenever I've talked about this, people seem surprised at the consequences. WebAuthn is supposed to protect against phishing attacks, but this approach subverts that protection by presenting the user with a legitimate login page and then handing their credentials to someone else.

RFC 8628 actually recognises this vector and presents a set of mitigations. Unfortunately nobody actually seems to implement these, and most of the mitigations are based around the idea that this flow will only be used for physical devices. Sadly, AWS uses this for initial authentication for the aws-cli tool, so there's no device in that scenario. Another mitigation is that there's a relatively short window where the code is valid, and so sending a link via email is likely to result in it expiring before the user clicks it. An attacker could avoid this by directing the user to a domain under their control that triggers the flow and then redirects the user to the login page, ensuring that the code is only generated after the user has clicked the link.

Can this be avoided? The best way to do so is to ensure that you don't support this token issuance flow anywhere, or if you do then ensure that any tokens issued that way are extremely narrowly scoped. Unfortunately if you're an AWS user, that's probably not viable - this flow is required for the cli tool to perform SSO login, and users are going to end up with broadly scoped tokens as a result. The logs are also not terribly useful.

The infuriating thing is that this isn't necessary for CLI tooling. The reason this approach is taken is that you need a way to get the token to a local process even if the user is doing authentication in a browser. This can be avoided by having the process listen on localhost, and then have the login flow redirect to localhost (including the token) on successful completion. In this scenario the attacker can't get access to the token without having access to the user's machine, and if they have that they probably have access to the token anyway.

There's no real moral here other than "Security is hard". Sorry.

comment count unavailable comments

Russell Coker: Links November 2022

Here s the US Senate Statement of Frances Haugen who used to work for Facebook countering misinformation and espionage [1]. She believes that Facebook is capable of dealing with the online radicalisation and promotion of bad things on it s platform but is unwilling to do so for financial reasons. We need strong regulation of Facebook and it probably needs to be broken up. Interesting article from The Atlantic about filtered cigarettes being more unhealthy than unfiltered [2]. Every time I think I know how evil tobacco companies are I get surprised by some new evidence. Cory Doctorow wrote an insightful article about resistance to rubber hose cryptanalysis [3]. Cory Doctorow wrote an interesting article When Automation Becomes Enforcement with a new way of thinking about Snapchat etc [4]. Cory Doctorow wrote an insightful and informative article Big Tech Isn t Stealing News Publishers Content, It s Stealing Their Money [5] which should be read by politicians from all countries that are trying to restrict quoting news on the Internet. Interesting articl;e on Santiago Genoves who could be considered as a pioneer of reality TV for deliberately creating disputes between a group of young men and women on a raft in the Atlantic for 3 months [6]. Matthew Garrett wrote an interesting review of the Freedom Phone, seems that it s not good for privacy and linked to some companies doing weird stuff [7]. Definitely worth reading. Cory Doctorow wrote an interesting and amusing article about backdoors for machine learning [8] Petter Reinholdtsen wrote an informative post on how to make a bootable USB stick image from an ISO file [9]. Apparently Lenovo provides ISO images to update laptops that don t have DVD drives. :( Barry Gander wrote an interesting article about the fall of Rome and the decline of the US [10]. It s a great concern that the US might fail in the same way as Rome. Ethan Siegel wrote an interesting article about Iapetus, a moon of Saturn that is one of the strangest objects in the solar system [11]. Cory Doctorow s article Revenge of the Chickenized Reverse-Centaurs has some good insights into the horrible things that companies like Amazon are doing to their employees and how we can correct that [12]. Charles Stross wrote an insightful blog post about Billionaires [13]. They can t do much for themselves with the extra money beyond about $10m or $100m (EG Steve Jobs was unable to extend his own life much when he had cancer) and their money is trivial when compared to the global economy. They are however effective parasites capable of performing great damage to the country that hosts them. Cory Doctorow has an interesting article about how John Deere is being evil again [14]. This time with potentially catastrophic results.
Related posts:
  1. Links September 2022 Tony Kern wrote an insightful document about the crash of...
  2. Links November 2021 The Guardian has an amusing article by Sophie Elmhirst about...
  3. Links September 2020 MD5 cracker, find plain text that matches MD5 hash [1]....

29 November 2022

Sam Hartman: Introducing Carthage

For the past four years, I ve been working on Carthage, a free-software Infrastructure as Code framework. We ve finally reached a point where it makes sense to talk about Carthage and what it can do. This is the first in a series of blog posts to introduce Carthage, discuss what it can do and show how it works. Why Another IAC Framework It seems everywhere you look, there are products designed to support the IAC pattern. On the simple side, you could check a Containerfile into Git. Products like Terraform and Vagrant allow you to template cloud infrastructure and VMs. There are more commercial offerings than I can keep up with. We were disappointed by what was out there when we started Carthage. Other products have improved, but for many of our applications we re happy with what Carthage can build. The biggest challenge we ran into is that products wanted us to specify things at the wrong level. For some of our cyber training work we wanted to say things like We want 3 blue teams, each with a couple defended networks, a red team, and some neutral infrastructure for red to exploit. Yet the tools we were trying to use wanted to lay things out at the individual machine/container level. We found ourselves contemplating writing a program to generate input for some other IAC tool. Things were worse for our internal testing. Sometimes we d be shipping hardware to a customer. But sometimes we d be virtualizing that build out in a lab. Sometimes we d be doing a mixture. So we wanted to completely separate the descriptions of machines, networks, and software from any of the information about whether that was realized on hardware, VMs, containers, or a mixture. Dimensional Breakdown In discussing Carthage with Enrico Zini, he pointed me at Cognitive Dimensions of notation as a way to think about how Carthage approaches the IAC problem. I m more interested in the idea of breaking down a design along the idea of dimensions that allow examining the design space than I am particular adherence to Green s original dimensions. Low Viscosity, High Abstraction Reuse One of the guiding principles is that we want to be able to reuse different components at different scales and in different environments. These include being able to do things like: Hidden Dependencies To accomplish these abstraction goals, dependencies need to be non-local. For example, a software role might need to integrate with a directory if a directory is present in the environment. When writing the role, no one is going to know which directory to use, nor whether a directory is present. Taking that as an explicit input into the role is error-prone when the role is combined into large abstract units (bigger roles or collections of machines). Instead it is better to have a non-local dependency, and to find the directory if it is available. We accomplish this using dependency injection. In addition to being non-local, dependencies are sometimes hidden. It is very easy to overwhelm our cognitive capacity with even a fairly simple IAC description. An effective notation allows us to focus on the parts that matter when working with a particular part of the description. I ve found hiding dependencies, especially indirect dependencies, to be essential in building complex descriptions. Obviously, tools are required for examining these dependencies as part of debugging. First Class Modeling Clearly one of the goals of IAC descriptions is to actually build and manage infrastructure. It turns out that there are all sorts of things you want to do with the description well before you instantiate the infrastructure. You might want to query the description to build network diagrams, understand interdependencies, or even build inventory/bill of materials. We often find ourselves building Ansible inventory, switch configurations, DNS zones, and all sorts of configuration artifacts. These artifacts may be installed into infrastructure that is instantiated by the description, but they may be consumed in other ways. Allowing the artifacts to be consumed externally means that you can avoid pre-commitment and focus on whatever part of the description you originally want to work on. You may use an existing network at first. Later the IAC description may replace that, or perhaps it never will. As a result, Carthage separates modeling from instantiation. The model can generally be built and queried without needing to interact with clouds, VMs, or containers. We ve actually found it useful to build Carthage layouts that cannot ever be fully instantiated, for example because they never specify details like whether a model should be instantiated on a container or VM, or what kind of technology will realize a modeled network. This allows developing roles before the machines that use them or focusing on how machines will interact and how the network will be laid out before the details of installing on specific hardware. The modeling separation is by far the difference I value most between Carthage and other systems. A Tool for Experts. In Neal Stephenson s essay In the Beginning Was the Command Line , Stephenson points out that the kind of tools experts need are not the same tools that beginners need. The illustration of why a beginner might not be satisfied with a Hole Hog drill caught my attention. Carthage is a tool for experts. Despite what cloud providers will tell you, IAC is not easy. Doubly so when you start making reusable components. Trying to hide that or focus on making things easy to get started can make it harder for experts to efficiently solve the problems they are facing. When we have faced trade offs between making Carthage easy to pick up and making it powerful for expert users, we have chosen to support the experts. That said, Carthage today is harder to pick up than it needs to be. It s a relatively new project with few external users as of this time. Our documentation and examples need improvement, just like every project at this level of maturity. Similarly, as the set of things people try to do expand, we will doubtless run into bugs that our current test cases don t cover. So Carthage absolutely will get easier to learn and use than it is today. Also, we ve already had success building beginner-focused applications on top of Carthage. For our cyber training, we built web applications on top of Carthage that made rebuilding and exploring infrastructure easy. We ve had success using relatively understood tools like Ansible as integration and customization points for Carthage layouts. But in all these cases, when the core layout had significant reusable components and significant complexity in the networking, only an IAC expert was going to be able to maintain and develop that layout. What Carthage can do. Carthage has a number of capabilities today. One of Carthage s strengths is its extensible design. Abstract interfaces make it easy to add new virtualization platforms, cloud services, and support for various ways of managing real hardware. This approach has been validated by incrementally adding support for virtualization architectures and cloud services. As development has progressed, adding new integrations continues to get faster because we are able to reuse existing infrastructure. Today, Carthage can model: Carthage has excellent facilities for dealing with images on which VMs and Containers can be based, although it does have a bit of a Debian/Ubuntu bias in how it thinks about images: When instantiating infrastructure, Carthage can work with: We have also looked at Oracle Cloud and I believe Openstack, although that code is not merged. Future posts will talk about core Carthage concepts and how to use Carthage to build infrastructure.

comment count unavailable comments

26 November 2022

Matthew Garrett: Poking a mobile hotspot

I've been playing with an Orbic Speed, a relatively outdated device that only speaks LTE Cat 4, but the towers I can see from here are, uh, not well provisioned so throughput really isn't a concern (and refurbs are $18, so). As usual I'm pretty terrible at just buying devices and using them for their intended purpose, and in this case it has the irritating behaviour that if there's a power cut and the battery runs out it doesn't boot again when power returns, so here's what I've learned so far.

First, it's clearly running Linux (nmap indicates that, as do the headers from the built-in webserver). The login page for the web interface has some text reading "Open Source Notice" that highlights when you move the mouse over it, but that's it - there's code to make the text light up, but it's not actually a link. There's no exposed license notices at all, although there is a copy on the filesystem that doesn't seem to be reachable from anywhere. The notice tells you to email them to receive source code, but doesn't actually provide an email address.

Still! Let's see what else we can figure out. There's no open ports other than the web server, but there is an update utility that includes some interesting components. First, there's a copy of adb, the Android Debug Bridge. That doesn't mean the device is running Android, it's common for embedded devices from various vendors to use a bunch of Android infrastructure (including the bootloader) while having a non-Android userland on top. But this is still slightly surprising, because the device isn't exposing an adb interface over USB. There's also drivers for various Qualcomm endpoints that are, again, not exposed. Running the utility under Windows while the modem is connected results in the modem rebooting and Windows talking about new hardware being detected, and watching the device manager shows a bunch of COM ports being detected and bound by Qualcomm drivers. So, what's it doing?

Sticking the utility into Ghidra and looking for strings that correspond to the output that the tool conveniently leaves in the logs subdirectory shows that after finding a device it calls vendor_device_send_cmd(). This is implemented in a copy of libusb-win32 that, again, has no offer for source code. But it's also easy to drop that into Ghidra and discover thatn vendor_device_send_cmd() is just a wrapper for usb_control_msg(dev,0xc0,0xa0,0,0,NULL,0,1000);. Sending that from Linux results in the device rebooting and suddenly exposing some more USB endpoints, including a functional adb interface. Although, annoyingly, the rndis interface that enables USB tethering via the modem is now missing.

Unfortunately the adb user is unprivileged, but most files on the system are world-readable. data/logs/atfwd.log is especially interesting. This modem has an application processor built into the modem chipset itself, and while the modem implements the Hayes Command Set there's also a mechanism for userland to register that certain AT commands should be pushed up to userland. These are handled by the atfwd_daemon that runs as root, and conveniently logs everything it's up to. This includes having logged all the communications executed when the update tool was run earlier, so let's dig into that.

The system sends a bunch of AT+SYSCMD= commands, each of which is in the form of echo (stuff) >>/usrdata/sec/chipid. Once that's all done, it sends AT+CHIPID, receives a response of CHIPID:PASS, and then AT+SER=3,1, at which point the modem reboots back into the normal mode - adb is gone, but rndis is back. But the logs also reveal that between the CHIPID request and the response is a security check that involves RSA. The logs on the client side show that the text being written to the chipid file is a single block of base64 encoded data. Decoding it just gives apparently random binary. Heading back to Ghidra shows that atfwd_daemon is reading the chipid file and then decrypting it with an RSA key. The key is obtained by calling a series of functions, each of which returns a long base64-encoded string. Decoding each of these gives 1028 bytes of high entropy data, which is then passed to another function that decrypts it using AES CBC using a key of 000102030405060708090a0b0c0d0e0f and an initialization vector of all 0s. This is somewhat weird, since there's 1028 bytes of data and 128 bit AES works on blocks of 16 bytes. The behaviour of OpenSSL is apparently to just pad the data out to a multiple of 16 bytes, but that implies that we're going to end up with a block of garbage at the end. It turns out not to matter - despite the fact that we decrypt 1028 bytes of input only the first 200 bytes mean anything, with the rest just being garbage. Concatenating all of that together gives us a PKCS#8 private key blob in PEM format. Which means we have not only the private key, but also the public key.

So, what's in the encrypted data, and where did it come from in the first place? It turns out to be a JSON blob that contains the IMEI and the serial number of the modem. This is information that can be read from the modem in the first place, so it's not secret. The modem decrypts it, compares the values in the blob to its own values, and if they match sets a flag indicating that validation has succeeeded. But what encrypted it in the first place? It turns out that the json blob is just POSTed to http://pro.w.ifelman.com/api/encrypt and an encrypted blob returned. Of course, the fact that it's being encrypted on the server with the public key and sent to the modem that decrypted with the private key means that having access to the modem gives us the public key as well, which means we can just encrypt our own blobs.

What does that buy us? Digging through the code shows the only case that it seems to matter is when parsing the AT+SER command. The first argument to this is the serial mode to transition to, and the second is whether this should be a temporary transition or a permanent one. Once parsed, these arguments are passed to /sbin/usb/compositions/switch_usb which just writes the mode out to /usrdata/mode.cfg (if permanent) or /usrdata/mode_tmp.cfg (if temporary). On boot, /data/usb/boot_hsusb_composition reads the number from this file and chooses which USB profile to apply. This requires no special permissions, except if the number is 3 - if so, the RSA verification has to be performed first. This is somewhat strange, since mode 9 gives the same rndis functionality as mode 3, but also still leaves the debug and diagnostic interfaces enabled.

So what's the point of all of this? I'm honestly not sure! It doesn't seem like any sort of effective enforcement mechanism (even ignoring the fact that you can just create your own blobs, if you change the IMEI on the device somehow, you can just POST the new values to the server and get back a new blob), so the best I've been able to come up with is to ensure that there's some mapping between IMEI and serial number before the device can be transitioned into production mode during manufacturing.

But, uh, we can just ignore all of this anyway. Remember that AT+SYSCMD= stuff that was writing the data to /usrdata/sec/chipid in the first place? Anything that's passed to AT+SYSCMD is just executed as root. Which means we can just write a new value (including 3) to /usrdata/mode.cfg in the first place, without needing to jump through any of these hoops. Which also means we can just adb push a shell onto there and then use the AT interface to make it suid root, which avoids needing to figure out how to exploit any of the bugs that are just sitting there given it's running a 3.18.48 kernel.

Anyway, I've now got a modem that's got working USB tethering and also exposes a working adb interface, and I've got root on it. Which let me dump the bootloader and discover that it implements fastboot and has an oem off-mode-charge command which solves the problem I wanted to solve of having the device boot when it gets power again. Unfortunately I still need to get into fastboot mode. I haven't found a way to do it through software (adb reboot bootloader doesn't do anything), but this post suggests it's just a matter of grounding a test pad, at which point I should just be able to run fastboot oem off-mode-charge and it'll be all set. But that's a job for tomorrow.

Edit: Got into fastboot mode and ran fastboot oem off-mode-charge 0 but sadly it doesn't actually do anything, so I guess next is going to involve patching the bootloader binary. Since it's signed with a cert titled "General Use Test Key (for testing only)" it apparently doesn't have secure boot enabled, so this should be easy enough.

comment count unavailable comments

21 November 2022

Dirk Eddelbuettel: RcppClassic 0.9.13 on CRAN: Minor Update

A maintenance release 0.9.14 of the RcppClassic package arrived earlier today on CRAN. This package provides a maintained version of the otherwise deprecated initial Rcpp API which no new projects should use as the normal Rcpp API is so much better. The changes is. CRAN was reporting (for all four macOS builds, and only there) that an absolute path was embedded, so we updated the (old ) call to install_name_tool use on that (and only that) OS. No other changes were made. CRANberries also reports the changes relative to the previous release from nearly three years ago. Questions, comments etc should go to the rcpp-devel mailing list off the R-Forge page. 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.

19 November 2022

Joerg Jaspert: From QNAP QTS to TrueNAS Scale

History, Setup So for quite some time I have a QNAP TS-873x here, equipped with 8 Western Digital Red 10 TB disks, plus 2 WD Blue 500G M2 SSDs. The QNAP itself has an AMD Embedded R-Series RX-421MD with 4 cores and was equipped with 48G RAM. Initially I had been quite happy, the system is nice. It was fast, it was easy to get to run and the setup of things I wanted was simple enough. All in a web interface that tries to imitate a kind of workstation feeling and also tries to hide that it is actually a webinterface. Natually with that amount of disks I had a RAID6 for the disks, plus RAID1 for the SSDs. And then configured as a big storage pool with the RAID1 as cache. Below the hood QNAP uses MDADM Raid and LVM (if you want, with thin provisioning), in some form of emdedded linux. The interface allows for regular snapshots of your storage with flexible enough schedules to create them, so it all appears pretty good.

QNAP slow Fast forward some time and it gets annoying. First off you really should have regular raid resyncs scheduled, and while you can set priorities on them and have them low priority, they make the whole system feel very sluggish, quite annoying. And sure, power failure (rare, but can happen) means another full resync run. Also, it appears all of the snapshots are always mounted to some /mnt/snapshot/something place (df on the system gets quite unusable). Second, the reboot times. QNAP seems to be affected by the more features, fuck performance virus, and bloat their OS with more and more features while completly ignoring the performance. Everytime they do an upgrade it feels worse. Lately reboot times went up to 10 to 15 minutes - and then it still hadn t started the virtual machines / docker containers one might run on. Another 5 to 10 minutes for those. Opening the file explorer - ages on calculating what to show. Trying to get the storage setup shown? Go get a coffee, but please fetch the beans directly from the plantation, or you are too fast. Annoying it was. And no, no broken disks or fan or anything, it all checks out fine.

Replace QNAPs QTS system So I started looking around what to do. More RAM may help a little bit, but I already had 48G, the system itself appears to only do 64G maximum, so not much chance of it helping enough. Hardware is all fine and working, so software needs to be changed. Sounds hard, but turns out, it is not.

TrueNAS And I found that multiple people replaced the QNAPs own system with a TrueNAS installation and generally had been happy. Looking further I found that TrueNAS has a variant called Scale - which is based on Debian. Doubly good, that, so I went off checking what I may need for it.

Requirements Heck, that was a step back. To install TrueNAS you need an HDMI out and a disk to put it on. The one that QTS uses is too small, so no option.
QNAPs internal USB disk QNAPs original internal USB drive, DOM
So either use one of the SSDs that played cache (and should do so again in TrueNAS, or get the QNAP original replaced. HDMI out is simple, get a cheap card and put it into one of the two PCIe-4x slots, done. The disk thing looked more complicated, as QNAP uses some internal usb stick thing . Turns out it is just a USB stick that has an 8+1pin connector. Couldn t find anything nice as replacement, but hey, there are 9-pin to USB-A adapters.
9PIN to USB A a 9pin to USB A adapter
With that adapter, one can take some random M2 SSD and an M2-to-USB case, plus some cabling, and voila, we have a nice system disk.
USB 9pin to USB-A cable connected to Motherboard and some more cable 9pin adapter to USB-A connected with some more cable
Obviously there isn t a good place to put this SSD case and cable, but the QNAP case is large enough to find space and use some cable ties to store it safely. Space enough to get the cable from the side, where the mainboard is to the place I mounted it, so all fine.
Mounted SSD in external case, also shows the video card Mounted SSD in its external case
The next best M2 SSD was a Western Digital Red with 500G - and while this is WAY too much for TrueNAS, it works. And hey, only using a tiny fraction? Oh so much more cells available internally to use when others break. Or something Together with the Asus card mounted I was able to install TrueNAS. Which is simple, their installer is easy enough to follow, just make sure to select the right disk to put it on.

Preserving data during the move Switching from QNAP QTS to TrueNAS Scale means changing from MDADM Raid with LVM and ext4 on top to ZFS and as such all data on it gets erased. So a backup first is helpful, and I got myself two external Seagate USB Disks of 6TB each - enough for the data I wanted to keep. Copying things all over took ages, especially as the QNAP backup thingie sucks, it was breaking quite often. Also, for some reason I did not investigate, the performance of it was real bad. It started at a maximum of 50MB/s, but the last terabyte of data was copied at MUCH less than that, and so it took much longer than I anticipated. Copying back was slow too, but much less so. Of course reading things usually is faster than writing, with it going around 100MB/s most of the time, which is quite a bit more - still not what USB3 can actually do, but I guess the AMD chip doesn t want to go that fast.

TrueNAS experience The installation went mostly smooth, the only real trouble had been on my side. Turns out that a bad network cable does NOT help the network setup, who would have thought. Other than that it is the usual set of questions you would expect, a reboot, and then some webinterface. And here the differences start. The whole system boots up much faster. Not even a third of the time compared to QTS. One important thing: As TrueNAS scale is Debian based, and hence a linux kernel, it automatically detects and assembles the old RAID arrays that QTS put on. Which TrueNAS can do nothing with, so it helps to manually stop them and wipe the disks. Afterwards I put ZFS on the disks, with a similar setup to what I had before. The spinning rust are the data disks in a RAIDZ2 setup, the two SSDs are added as cache devices. Unlike MDADM, ZFS does not have a long sync process. Also unlike the MDADM/LVM/EXT4 setup from before, ZFS works different. It manages the raid thing but it also does the volume and filesystem parts. Quite different handling, and I m still getting used to it, so no, I won t write some ZFS introduction now.

Features The two systems can not be compared completly, they are having a pretty different target audience. QNAP is more for the user that wants some network storage that offers a ton of extra features easily available via a clickable interface. While TrueNAS appears more oriented to people that want a fast but reliable storage system. TrueNAS does not offer all the extra bloat the QNAP delivers. Still, you have the ability to run virtual machines and it seems it comes with Rancher, so some kubernetes/container ability is there. It lacks essential features like assigning PCI devices to virtual machines, so is not useful right now, but I assume that will come in a future version. I am still exploring it all, but I like what I have right now. Still rebuilding my setup to have all shares exported and used again, but the most important are working already.

16 November 2022

Ian Jackson: Stop writing Rust linked list libraries!

tl;dr: Don t write a Rust linked list library: they are hard to do well, and usually useless. Use VecDeque, which is great. If you actually need more than VecDeque can do, use one of the handful of libraries that actually offer a significantly more useful API. If you are writing your own data structure, check if someone has done it already, and consider slotmap or generation_arena, (or maybe Rc/Arc). Contents Survey of Rust linked list libraries I have updated my Survey of Rust linked list libraries. Background In 2019 I was writing plag-mangler, a tool for planar graph layout. I needed a data structure. Naturally I looked for a library to help. I didn t find what I needed, so I wrote rc-dlist-deque. However, on the way I noticed an inordinate number of linked list libraries written in Rust. Most all of these had no real reason for existing. Even the one in the Rust standard library is useless. Results Now I have redone the survey. The results are depressing. In 2019 there were 5 libraries which, in my opinion, were largely useless. In late 2022 there are now thirteen linked list libraries that ought probably not ever to be used. And, a further eight libraries for which there are strictly superior alternatives. Many of these have the signs of projects whose authors are otherwise competent: proper documentation, extensive APIs, and so on. There is one new library which is better for some applications than those available in 2019. (I m referring to generational_token_list, which makes a plausible alternative to dlv-list which I already recommended in 2019.) Why are there so many poor Rust linked list libraries ? Linked lists and Rust do not go well together. But (and I m guessing here) I presume many people are taught in programming school that a linked list is a fundamental data structure; people are often even asked to write one as a teaching exercise. This is a bad idea in Rust. Or maybe they ve heard that writing linked lists in Rust is hard and want to prove they can do it. Double-ended queues One of the main applications for a linked list in a language like C, is a queue, where you put items in at one end, and take them out at the other. The Rust standard library has a data structure for that, VecDeque. Five of the available libraries: For these you could, and should, just use VecDeque instead. The Cursor concept A proper linked list lets you identify and hold onto an element in the middle of the list, and cheaply insert and remove elements there. Rust s ownership and borrowing rules make this awkward. One idea that people have many times reinvented and reimplemented, is to have a Cursor type, derived from the list, which is a reference to an element, and permits insertion and removal there. Eight libraries have implemented this in the obvious way. However, there is a serious API limitation: To prevent a cursor being invalidated (e.g. by deletion of the entry it points to) you can t modify the list while the cursor exists. You can only have one cursor (that can be used for modification) at a time. The practical effect of this is that you cannot retain cursors. You can make and use such a cursor for a particular operation, but you must dispose of it soon. Attempts to do otherwise will see you losing a battle with the borrow checker. If that s good enough, then you could just use a VecDeque and use array indices instead of the cursors. It s true that deleting or adding elements in the middle involves a lot of copying, but your algorithm is O(n) even with the single-cursor list libraries, because it must first walk the cursor to the desired element. Formally, I believe any algorithm using these exclusive cursors can be rewritten, in an obvious way, to simply iterate and/or copy from the start or end (as one can do with VecDeque) without changing the headline O() performance characteristics. IMO the savings available from avoiding extra copies etc. are not worth the additional dependency, unsafe code, and so on, especially as there are other ways of helping with that (e.g. boxing the individual elements). Even if you don t find that convincing, generational_token_list and dlv_list are strictly superior since they offer a more flexible and convenient API and better performance, and rely on much less unsafe code. Rustic approaches to pointers-to-and-between-nodes data structures Most of the time a VecDeque is great. But if you actually want to hold onto (perhaps many) references to the middle of the list, and later modify it through those references, you do need something more. This is a specific case of a general class of problems where the naive approach (use Rust references to the data structure nodes) doesn t work well. But there is a good solution: Keep all the nodes in an array (a Vec<Option<T>> or similar) and use the index in the array as your node reference. This is fast, and quite ergonomic, and neatly solves most of the problems. If you are concerned that bare indices might cause confusion, as newly inserted elements would reuse indices, add a per-index generation count. These approaches have been neatly packaged up in libraries like slab, slotmap, generational-arena and thunderdome. And they have been nicely applied to linked lists by the authors of generational_token_list. and dlv-list. The alternative for nodey data structures in safe Rust: Rc/Arc Of course, you can just use Rust s interior mutability and reference counting smart pointers, to directly implement the data structure of your choice. In many applications, a single-threaded data structure is fine, in which case Rc and Cell/RefCell will let you write safe code, with cheap refcount updates and runtime checks inserted to defend against unexpected aliasing, use-after-free, etc. I took this approach in rc-dlist-deque, because I wanted each node to be able to be on multiple lists. Rust s package ecosystem demonstrating software s NIH problem The Rust ecosystem is full of NIH libraries of all kinds. In my survey, there are: five good options; seven libraries which are plausible, but just not as good as the alternatives; and fourteen others. There is a whole rant I could have about how the whole software and computing community is pathologically neophilic. Often we seem to actively resist reusing ideas, let alone code; and are ignorant and dismissive of what has gone before. As a result, we keep solving the same problems, badly - making the same mistakes over and over again. In some subfields, working software, or nearly working software, is frequently replaced with something worse, maybe more than once. One aspect of this is a massive cultural bias towards rewriting rather than reusing, let alone fixing and using. Many people can come out of a degree, trained to be a programmer, and have no formal training in selecting and evaluating software; this is even though working effectively with computers requires making good use of everyone else s work. If one isn t taught these skills (when and how to search for prior art, how to choose between dependencies, and so on) one must learn it on the job. The result is usually an ad-hoc and unsystematic approach, often dominated by fashion rather than engineering. The package naming paradox The more experienced and competent programmer is aware of all the other options that exist - after all they have evaluated other choices before writing their own library. So they will call their library something like generational_token_list or vecdeque-stableix. Whereas the novice straight out of a pre-Rust programming course just thinks what they are doing is the one and only obvious thing (even though it s a poor idea) and hasn t even searched for a previous implementation. So they call their package something obvious like linked list . As a result, the most obvious names seem to refer to the least useful libraries.
Edited 2022-11-16 23:55 UTC to update numbers of libraries in various categories following updates to the survey (including updates prompted by feedback received after this post first published).


comment count unavailable comments

Next.

Previous.