One thing that could be added, for the sake of completeness: in the part "Attack begins", toward the end, when they are pushing for updating xz in the major distros, Ubuntu and Debian are mentioned but not Fedora.

Looks like the social engineering/pressuring for Fedora started at least weeks before 2024 March 04, according to a comment by @rwmj on HN [1]. I also found this thread on Fedora's devel list [2], but didn't dig too much.

[1] https://news.ycombinator.com/item?id=39866275

[2] https://lists.fedoraproject.org/archives/list/[email protected]...

Its time to stop accepting that the way we've done this in the past is the way we will continue doing it ad infinitum.

There might have been a time where it worked, but people started to use plugins for all kinds of reasons, quite good ones in most cases. Findbugs, code coverage, source code generation,...

Today, a maven project without plugins is rare. Maven brought us 95%, but there is a long tail left to cover.

There is a sizable movement in the Rust ecosystem to move all build-time scripts and procedural macros to (be combined to) WASM. This allows you to write turing-complete performant code to cover all use-cases people can reasonably think of, while also allowing trivially easy sandboxing.

It's not perfect, for example some build scripts download content from the internet, which can be abused for information extraction. And code generation scripts could generate different code depending on where it thinks it's running. But it's a lot better than the unsandboxed code execution that's present in most current build systems, without introducing the constraints of a pure config file.

I’d argue that Rust’s biggest advantage here is that build scripts and procedural macros are written in Rust themselves, making them easier to read and thus harder to hide things in than autotools’ m4-to-shell gunk.

But that means it’s important to build those things from source! Part of the movement you cite consists of watt, which ships proc macros as precompiled WebAssembly binaries mainly to save build time. But if watt were more widely adopted, it would actually make an attacker’s life much easier since they could backdoor the precompiled binary. Historically I’ve been sympathetic to dtolnay’s various attempts to use precompiled binaries this way (not just watt but also the serde thing that caused a hullabaloo); I hate slow builds. But after the xz backdoor I think this is untenable.

But determined actors with access are always going to try to thwart these schemes, so verification and testing is going to need to step up.

I’m not sure I follow. The bar for passing review and audits is “people can understand it”, at least to the point of preventing smuggling in a custom public keys and other shenanigans – as opposed to “all valid programs in the language”. While I agree Turing-completeness opens up undeniable complexity, it’s the best we got. And there are huge variations in obscurity across languages and tooling.

if you trust that they're giving you the correct hash, but not the correct binary, then you're not thinking clearly.

To put it another way: if you don't have reproducible builds and you're trying to replicate an upstream artifact then it's very hard to tell what the cause is. It might just be some ephemeral state, a race, or some machine aspect that caused a few fairly meaningless differences. But as soon as you have a reproducible build, then failure to reproduce instantly marks upstream as being suspect.

It's also useful when you're trying to do things like tweak a build - you can ensure you're importing upstream correctly by checking the hash with what you're making, and then even if your downstream version isn't a binary dependent on upstream (e.g. optimized differently, or with some extra plugins somewhere), you can be sure that changes between what you're building and upstream are intentional and not spurious.

It's clearly not a silver bullet, sure. But it's not entirely useless either, and it probably could help as part of a larger ecosystem shift; especially if conventional tooling created and published these by default such that bad actors trying to hide build processes actually stick out more glaringly.

Debian's response is looking to be mandating the source must come from the git repository, not a tar ball. And it will be done using a script. And the script must produce the same output every time, ie be reproducible. Currently Debian's idea of "reproducible" means it reflects the source Debian distributes. This change means it will reproducibly reflect the upstream sources. That doesn't mean it will be the same - just that it's derived in a reproducible way.

As for trusting the person who gave you the hash: that's not what it hinges on. It hinges on the majority installing a binary from a distro. That binary can be produced reproducibly, by an automated build system. Even now, Debian's binaries aren't created just once by Debian. There are Debian builders all around the planet creating the binary, and verifying it matches what's being distributed.

Thus the odds of the binary you are running not being derived from the source are vanishingly low.

If everything is done carefully enough with reproducible builds, I think using a binary whose hash can be checked shouldn't be a great extension of trust.

You could have multiple independent autobuilders verifying that particular source does indeed generate a binary with the claimed hash.

For the purpose of this conversation we mostly just care about the use case of someone grabbing the code and wanting to use it in their own project. For this use case, dev tools like findbugs and code coverage can be ignored, so it would suffice to have a version of the build system with plugins completely disabled.

Code generation is the thornier one, and we can at least be more principled about it than "run some arbitrary code", and at least it should be trivial to say "this codegen process gets absolutely no I/O access whatsoever; you're a dumb text pipeline". But at the end of the day, we have to Just Say No to things like this. Even if it makes the codebase grodier to check in generated code, if I can't inspect and audit the source code, that's a problem, and arbitrary build-time codegen prevents that. Some trade-offs are worth making.

I'm not a fan of generated code. It tends to cause misery, being in a no mans land between code and not-code. But it is usefull sometimes, e.g rust generating an API from the opengl XML specs.

Sandboxing seems the least worst option, but it will still be uninspected half code that one day ends up in production.

The code was only provided in a roundabout way that was deliberately done to evade manual inspection, so that's not a failure of checking in generated code, that's a failure of actually building a binary from the artifacts that we expect it to be built from. Suffice to say, cutting out the Turing-complete crap from our build systems is only one of many things that we need to fix.

I'm not sure how well isolated the containers are (probably not very – I think GitHub gives access to the Docker socket) and you'd have to make sure they don't share secret tokens etc., but at least it might make things simpler to audit, and isolation could be improved in the future.

I would like to see this on the language level, not just on the OS level.

Say I have a directory with a virtualenv and some code that needs some packages from PyPI. I would very much like to sandbox anything that runs in this virtualenv to just disk access inside that directory, and network access only to specifically whitelisted URLs. As a user I should only need to add "sandbox = True" to the pyproject.toml file, and optionally "network_whitelist = [...]".

From my cursory looking around, I believe Cloudflare sandbox utils, which are convenience wrappers around systemd Seccomp, might be the best starting point.

Edit: or just use Firejail, interesting...

You mention sandboxing on the language level, but I don't think it is the way. Apparently sandboxing within Python itself is a particularly nasty rabbit hole that is ultimately unfruitful because of Python's introspection capabilities. You will find many dire warnings on that path.

> 99.9% of all projects have the capability to be built with trivial declarative rulesets.

Only if you forbid bootstrapping, which all projects ultimately rely on at some point in their supply chain.

We could write an entire PhD thesis on the number of dire technical failings that would need to be addressed to stop this attack, so while this alone wouldn't have stopped it, it would have required the actor to come up with another vector of code injection which would have been easier to find.

> Only if you forbid bootstrapping

Codebases that bootstrap are the 0.1%. Those can be built via `bash build.sh` rather than deceptively hiding a Turing-complete environment behind a declarative one. Even if you need to have these in your trusted computing base somewhere, we can focus auditing resources there, especially once we've reduced the amount of auditing that we need to do on the other 99.9% of codebases now that we've systematically limited the build-time shenanigans that they can get up to.

At worst, you're actually creating more holes. The reason autoconf/automake exist and M4 scripts are innocuous in the first place is because the build system uses an underpowered language and developers have to turn to code generation to get around it.

If you kneecap the build system's language you're not solving problems. You're creating them.

> it would have required the actor to come up with another vector of code injection which would have been easier to find.

If make was standardized and could programmatically determine the environment its run under and write full programs then the attack vector wouldn't exist in the first place.

> Codebases that bootstrap are the 0.1%.

We have different experiences, because ime it's close to 100% especially when you include transitive dependencies. When you care about supply chain security you care about being able to bootstrap from sources for your code and all your dependencies, and it's almost guaranteed that one of your dependencies needs to be bootstrapped.

Sounds like it would be impossible but maybe my thinking is just enclosed and not free.

Who will pay for all the rewriting you want done? Or even just for the new frameworks that are "scrutable"? How do we guarantee that the result is not inscrutable to you or others?

There is so much knee-jerking in this xz debacle.

(And I say this / ask these questions with no love for autoconf/m4/sh.)

Being able to send xz patches to the Linux kernel would have been a nice point of leverage for Jia Tan's future work. We're not at trusting trust [1] levels yet, but it would be one step closer.

[1]: https://research.swtch.com/nih

Reading the timeline, the root cause is pure social engineering. The technical details could be swapped out with anything. Sure there are aspects unique to xz that were exploited such as the test directory full of binaries, but that's just because they happened to be the easiest target to implement their changes in an obfuscated way, that misses the point - once an attacker has gained maintainer-ship you are basically screwed - because they will find a way to insert something, somehow, eventually, no matter how many easy targets for obfuscation are removed.

Is the real problem here in handing over substantial trust to an anonymous contributor? If a person has more to lose than maintainership then would this have happened?

That someone can weave their way into a project under a pseudonym and eventually gain maintainership without ever risking their real reputation seems to set up quite a risk free target for attackers.

It could also be called an industry problem, where we rely on other's code without running proper checks. This seems to be an emerging realization, with services like socket.dev starting to emerge.

Unless there's some practical way of comprehensively solving The Real Problem Here, it makes a lot of sense to consider all reasonable mitigations, technical, social or otherwise.

> If a person has more to lose than maintainership then would this have happened?

I guess that's one possible mitigation, but what exactly would that look like in practice? Good luck getting open source contributors to accept any kind of liability. Except for the JiaTans of the world, who will be the first to accept it since they will have already planned to be untouchable.

It's heartblead all over again (though those were honest mistakes, not an intentional attack

It's not necessary to accept liability, that's waived in all FOSS licenses. What I'm suggesting would only risk reputation of non-malicious contributors, and from what I've seen, most of the major FOSS contributors and maintainers freely use their real identity or associate it with their pseudonym anyway, since that attribution comes with real life perks.

Disallowing anonymous pseudonyms would raise the bar quite a bit and require more effort from attackers to construct or steal plausible looking online identities for each attack, especially when they need to hold up for a long time as with this attack.

And all code that gets linked into security critical applications/libraries needs to be covered by under some sort of security focused code review.

So no patching the compression code that openSSL links to with random junk distribution maintainers.

Build infrastructure should be minimal, standardized, and not subject to endless special, undocumented fragility.

cmake, just, conan, meson, and bazel (and forks) exist.

But I've still yet to see a proper build system that does feature detection in parallel and concurrently, or supports live, incremental, continuous, cached rebuilding.

It doesn't matter if it's minified or obfuscated because you basically have to run unknown, untrusted code everywhere while browsing the web with JavaScript turned on. So the ecosystem and tooling is extremely resilient to most forms of malicious attacks no matter how minified or obfuscated the js you're running is. The complete opposite is true for bash and shell scripting in general

Another example of this was log4j: if you were still using the old 1.x log4j versions, you wouldn't have been vulnerable to the log4shell vulnerability, since it was introduced early in the 2.x series. The old 1.x log4j versions had other known vulnerabilities, but only if you were using less common appenders or an uncommon server mode or a built-in GUI log viewer (!); the most common use of log4j (logging into a local file) was not exposed to any of these, and in fact, you could remove the vulnerable classes and still have a functional log4j setup (see for instance https://www.petefreitag.com/blog/log4j-1x-mitigation/ which I just found on a quick web search).

Did log4shell (and a later vulnerability which could only be exploited if you were using Java 9 or later, because it depended on a new method which was introduced on Java 9) lead people to question whether always being on the "latest and greatest" was a good thing? No, AFAIK the opposite happened: people started to push even harder to keep everything on the latest release, "so that when another vulnerability happens, upgrading to a fixed version (which is assumed to be based on the latest release) will be easy".

Lol, this exact thing happened at my last gig. When I first learned of the vulnerability I panicked, until I found out we were so outdated it didn't affect us. We had a sad laugh about it.

> "so that when another vulnerability happens, upgrading to a fixed version (which is assumed to be based on the latest release) will be easy".

I think there is some truth to this motivation though - if you are on an ancient 1.X version and have to jump a major version of two, that almost always causes pain depending on how critical the service or library is. I don't pretend to know the right answer but I always tend to wait several versions before upgrading so any vulnerabilities or fixes can come by the time I get to the upgrade.

Now we're in a position where most projects update so quickly, that you don't really have a choice. If you need to update one component, there's a good chance it will require that many more as dependencies in your project will require an update to be compatible.

The industry as a whole sacrificed stability and some aspects of security for faster advancements and features. Overall I'd say the net benefit is positive, but its times like these that remind us that perhaps we need to slow things down just a little and do a bit of a course correction to bring things into a better balance.

A newly-introduced security issue tends to have very limited exploitability, because it's valuable, not-yet well understood, and public exploits are yet to be developed.

Compare to that a similar vulnerability in an older package: chances are that everything about it has been learned and is publicly known. Exploits have become a commodity and are now part of every offensive security distro on the planet. If you run that vulnerable version, there's a real risk that a non-targeted campaign will randomly bite you.

Except in the scenario that is this exact case: Supply chain attacks that are developed with the exploit in mind.

For an old vulnerability, however, any unpatched system is a target. So the individual risk for the average unpatched system is still orders of magnitude higher than in the former scenario.

Supply chain attacks are a real threat too, but being conservative about upgrading to improve your security posture is like saying "I'm worried someone with the key to my house might rob me, so I'm going to remove the lock from my door."

Everyone being on different versions results in something like a moving target

I highly encourage people to try freeBSD sometime. Give ports a try (although the modern sense is that poudrie is better even if you want custom-built packages). See how nicely everything works. All the system options you need go into rc.conf (almost uniformly). Everything is documented and you can basically operate the system out of the FreeBSD Handbook documentation (it's not at all comparable to the "how to use a window or a menu" level intro stuff the linux provides). You can't do that when everything is furiously churning every release. everything just works, everything is just documented, it's an experience when you're coming from linux.

and that forum post from 2007 on how to tweak a service script is probably still valid, because BSD hasn't had 3 different init systems over that timespan etc.

just like "engineering is knowing how to build a bridge that barely doesn't fall over", engineering here is knowing what not to churn, and fitting your own work and functionality extensions into the existing patterns etc. like it doesn't have to be even "don't make a bigger change than you have to", you just have to present a stable userland and stable kernel interface and stable init/services interface. the fact that linux doesn't present a stable kernel interface is actually fairly sketchy/poor engineering, it doesn't have to be that way, a large subset of kernel interfaces probably should be stable.

It's like seeing a "Bill Johnson, real American patriot" account on Twitter: would not believe it at face value.

If being rude and pushy doesn’t work, the next round will be kind and helpful. Don’t read too much into the cultural techniques used, because the cultural techniques will mirror the culture at the time.

I think we need to start paying more of these open source maintainers and have some staff/volunteers that can help them manage their git hub issue volume.

There's a certain kind of talented person who is all too conscious of their abilities and is arrogant, irascible, and demanding as a result. Linus Torvalds, Steve Jobs, Casey Muratori come to mind. Much as we might want these characters to be kinder, their irascibility is inseparable from their more admirable qualities.

Sometimes good things, even the best things, are made by difficult people, and we would lose a lot by making a community that alienates them.

[0] Funnily enough, just a week or two ago, I fixed an issue in an OS project that we introduced at work. It was an easy frontend fix even for someone like me who doesn't do frontend and barely knows how to spell Vue. And more importantly, in the issue description somebody already wrote exactly what causes the bug and what would need to change - the only thing left was finding the place where to make the (one-line) change. Somehow that issue had been open for 2 years but nobody of the several people who complained (nor the maintainer) had bothered to fix it. After I made a PR, it was merged within a day.

A user reported a safety issue, the maintainer said it was safe. Then it was proven that it was in fact unsafe, and the maintainer justified it with performance. Then a PR was filed which was safe and did not regress performance, and the maintainer rejected it with "this patch is boring"

The behavior of both sides was deeply unacceptable. If someone identifies a legitimate issue and files a PR to fix it, don't insult them by calling the patch "boring" and don't reject it solely on that basis.

We've complained about bots in social media for a long time, but how many people in open source discussions are shady manipulative entities?

I did an engineering/program manager role for 8 years and people pretty much always did what I asked if I showed up at their desk or bothered their boss.

"Squeaky wheel gets the grease?"

But I too like to think that I prioritize my children on merit rather than fuss level. For some reason they continue to cry despite me claiming I don't react to it.

As a hobbyist developer/maintainer of open source projects, I strive to remember that this is my gift to the world, and it comes with no strings attached. If people have any expectations about the software, it’s for them to manage; if they depend on it somehow, it’s their responsibility to ensure timely resolution of issues. None of this translates to obligations on my part, unless I explicitly make promises.

I empathize with Lasse having been slowed down by mental issues. I have, too. And we need to take good care of ourselves, and proactively prevent the burden of maintainership from exacerbating those issues.

[0]: https://gist.github.com/g1eny0ung/9e7d4d0f72547a8d156452e76f...

Anyone and everyone in the OSS world should be concerned about this too. You have nation state level actors out there with massive amounts of information on you. How much information have you leaked to data brokers? These groups will know how much debt you're in. The status of your relationships. Your health conditions and medications? It would not take much on their part to make your life worse and increase your stress levels. Just imagine things like fake calls from your bank saying that debt of yours has been put in collections.

If there is a law that would entitle a user to more, it is a bug in legislation that needs urgent fixing.

(1) xz and the lib are widely used in the wild including linux kernel, systemd, openSSH; (2) single maintainer, low rate of maintenance; (3) the original maintainer has other problems in his life distracting them from paying closer attention to the project.

I am wondering how many other OSS projects look similar and can be targeted in similar ways?

Controlling someone's lights is probably less important than Debian's build fleet but it's a scary proposition for the impacted individual who happens to use one of those long tail home assistant integrations or whatever.

Are you married? Have a house? Pets? Children? Sick parent? Gay? Trans? Mental health issues? Disabled? All of this can be used against you. Be careful where and how you share stuff like this. I know it's not "cool" to be mysterious online anymore, but it creates a much larger attack surface. People can still engage with groups around these things, but better to do it with various personas than to have one trackable identity with everything attached to it.

We're in a tech slowdown right now. There are people who got used to a certain lifestyle who now have "seeking work" on their LinkedIn profiles, and who have property taxes in arrears that are listed in county newspapers-of-record. If you're an intelligence operative in the Silicon Valley area, these guys should be easy pickings. An envelope full of cash to make some financial problems go away in exchange for a few commits on the FOSS projects they contribute to or maintain.

It's time for more of the big vendors who use these projects in their offerings to step up and give people running these small projects more resources and structure. $20k to have maintainers for each project actually meet twice a year at a conference is chump change for the biggest vendors, especially when compared against the cost of the audits they'll now be doing on everything Jia Tan and Co. touched.

Now maybe one could build a career out of OSS maintainerships, with work/time funded by lots of donations much smaller than a salary but amounting to a salary.

Furthermore, these organizations are in a place to put experienced, trustworthy contributors on projects that need maintainers if need be. If Lasse had been able to go to, idk, the Linux Foundation and say, "Listen, I'm getting burnt out, got anyone?" and they said "Sure, we've got this contributor with an established record who would love to help maintain your project", none of this is happening right now.

I wonder if it would be possible to crowdsource FOSS mailing list moderation.

First send a message to the mailing list as "Alice" providing a detailed bug report for a real bug, and a flawless patch to fix it.

Then you reply to the mailing list as "Bob" agreeing that it's a bug, thanking "Alice" for the patch and the time she spent producing such a detailed bug report, then explaining that unfortunately it won't be merged any time soon, then apologising and saying you know how frustrating it must be for Alice.

Your two characters have been model citizens: Alice has contributed a good quality bug report, and code. Bob has helped the project by confirming bug reports, and has never been rude or critical - merely straightforward, realistic and a bit overly polite.

For lesser/individual maintainers there is no way to survive this kind of mob attack. Corporate maintainers may be able to manage as it could be considered just as paid job and there are worse ways to make money.

We live in an adversarial environment, time to stop playing naively nice. Ideally it isn't the maintainer that has to do all this work.

Yeah, same. We should be much more willing to kick jerks out of our work spaces. The work is hard enough as it is without also being shit on while you do it.

Being mean is stressful and stops your brain from working properly. If someone doesn't allow you to be abusive, then they are not a mark. Predators look for prey that falls into certain patterns.

Simple ban and get the fuck out. Too often I've dealt with people trying to rationalize it as much as "o its just cultural, they don't understand". No, get the fuck out.

But hey I'm a NYer and telling people to fuck off is a past time.

I think this could be a genuine use of an AI: to go through all of the shit, and have it summarized in a fashion that the user wants: distant and objective, friendly, etc. It could provide an assessment on the general tone, aggregate the differently phrased requests, many things like that.

Crowdsourcing would works best with the reddit / hacker news model I feel, where discussion happens in tree styled threads, and users can react to messages in ways that are not text, but something meta, like a vote or a reaction indicating tone.

Both of these have significant downsides, but significant upsides too. People pick the mailing list in a similar way.

Trying to do something about it was called censorship.

I myself do not think that this is the actual problem. I think the actual problem is that many FOSS communities have fostered an idea that cracking down on certain types of behavior is censorship. Of course they might all agree to ban somebody who is saying certain well known explicit words or being really really aggressive, but there's a lot of behavior that I would consider bannable but they do not.

I'm trying to avoid mentioning specific cases because I don't wanna either reopen old wounds or cause another similar dustup. I can say that it reminds me a lot of the really old post about the 5 geek social fallacies https://plausiblydeniable.com/five-geek-social-fallacies/ In particular social fallacy #1

Obviously we're not talking about geek culture here (in general anyways). After rereading it, I saw something else that we actually do a see a lot even today.

This in particular has happened on many a mailing list, including the ones I'm not specifically mentioning

>> when nothing smacking of social selectiveness can be discussed in public, people inevitably begin to organize activities in secret.

Well-known annoyances get tolerated for years, and the people who do the work start moving their discussions to bug trackers, irc, or in personal emails to avoid interacting with them.

He was also turned on by multiple people who he considered close friends. In the book it did not seem that he had considered that it might not be a "them" problem.

*my details may be off here, I read it some time ago

This is the second time I've read this "observation", but this observation is just wrong? Jigar's email is "${name}${number}@${host}", yes, but Dennis's is just "${name}@${host}" — there's not a suffixed number. (There's a 3, but it's just a "leetcode" substitution for the E, i.e., it's semantically a letter.)

(They could still both be sockpuppets, of course. This doesn't prove or disprove anything. But … let's check our facts?)

So much for the folks advocating for binary (driver) blobs in OSS t support otherwise unsupported hardware.

It's either in source form and reproducable or it's not there.

Wanna know how I know you haven't read into the discussion much?

There are a whole lot of binary test cases in software. Especially when you're dealing with things like file formats and test cases that should specifically fail on bad data of particular types.

That's not how I read GP's point. If even binary blobs in test cases are a place where backdoors are, now as a matter of fact, hidden then, certainly, among the folks advocating for binary drivers in FOSS, there are some who are already --or planning to-- add backdoors there.

Binary blobs are all terrible, terrible, terrible ideas.

Builds should be 100% reproducible from source, bit for bit. At this point it's not open up for discussion anymore.

Instead of obscure constants, you use known constants, or at least simple methods to derive your constants.

You can do the same thing to come up with your test cases. Bad zip? Construct a good zip of 10 files, each containing the first 10,000 prime numbers. Then corrupt the zip by seeking to position (100/pi) and write a thousand zeroes there.

Bad JPEG? Use Imagemagick to render the first 1000 prime numbers as text into a JPEG file, then apply a simple nothing-up-my-sleeve corruption operation.

There are still cases where this approach isn't going to work: that new icon, helpfully proposed by a contributor, meant to be used in production, might contain malicious code, steganographically embedded. I think there's little you can do to prevent that.

[1]: https://en.wikipedia.org/wiki/Nothing-up-my-sleeve_number

That's the root cause. At some point a corp/gov consortium needs to adopt key projects and hire out the maintainers and give them real power and flexibility, perhaps similar to the way a nation might nationalize key infrastructure.

But this is against the core ethos at some level. Freedom and safety can be antagonistic.

At a certain point, companies need to step up and provide funding or engineering work or they should just keep expecting to get owned.

That doesn't mean we shouldn't try to help burnt out committers, but the problem seems very hard. As lcamtuf also says, many things don't need maintenance, and just paying people doesn't address what happens when they just don't want to do it anymore. In an alternate universe with different leadership, an organization like the FSF might use donated funds to pay a maintenance staff and an open-source maintainer might be able to lean on them. Of course, that still doesn't address the problem of Jia Tan getting a job with this organization.

https://lcamtuf.substack.com/p/technologist-vs-spy-the-xz-ba... https://lcamtuf.substack.com/p/oss-backdoors-the-allure-of-t...

I admit, the op has a "professional/polished vibe" to me as well, but we seem to know very little except for what work time/zones were preferred by the possibly collective/possibly singular human(s) behind the Jia Tan identity. Does anyone have slick linguistic tools to assess if the writer is a single author? Maybe an opportunity to show off.. It's sort of how they caught Ted Kaczynski.

It also absolutely makes sense to think of all the state actors (I agree including as you say the US/UK) as part of the ongoing threat model. If the KGB/Ministry of State Security/NSA/MI6 were not doing this before then they surely might in the future. Maybe with more gusto/funding now! They all seem to have an "information dominance at all costs" mentality, at least as agency collectives, whatever individuals inside think.

Anyway, I've yet to hear of anything in the xz work beyond the ability of 1-3 skilled black hats with a lot of post-COVID time on their hands. The NSA ensuring the Intel ME/AMT could be turned off seems another level of sophistication entirely, for example { a "higher pinnacle" in your nice phrasing :-) }.

In terms of sheer numbers, my impression is that the vast majority of attacks blocked at almost every sophistication level are more "criminals" than "states". Admittedly, that may partly be states just acquiring ability to act surgically rather than launching big compromise initiatives like botnets (or otherwise states going undetected). I'm sure it's hard to know.

Maybe we're just in a sad era of increasing attack sophistication measured in "Kilo-Sophistic-meters"? The AntiVirus industry has been having a boom lately.

It's probably already been mentioned many times, but besides economic & geopolitical incentives, maybe the attacker(s) was a hard core `systemd` or IBM/RedHat hater or one of the people who supposedly issued death threats to Lennart Poettering now at Microsoft or even an open source hater burnt out or wanting to burn the world down like The Joker in Batman. In light of that, Russ' Setting The Stage Prelude could perhaps profitably add the introduction of that lib dependency into `systemd` and also various distros defaulting to `systemd`.

Anyway, premature conclusions are to be cautioned against over & over. That's all I was trying to do. { And I'm not claiming you were concluding anything - you seem pretty open-minded about it all. I was always just amplifying parent comments. I absolutely agree long games "feel" more State Actor - part of what I meant by "vibe", but to quote the detective in V For Vendetta - "it's just a feeling". To Old People, 2 years famously doesn't seem as long as to a 25- or 15-year old. ;-) It's actually short on the 5+ year Linux distro maintenance time scales. }

If I had ever received a request from a well-known figure with a longstanding reputation, who's appeared in-person at talks and has verifiable employment, I might've been more receptive. But all the requests I got were from random internet identities that easily could've been fabricated, and in any case had no previous reputation. "Jia Tan" and their other sockpuppets very likely are not their real identities.

In this scenario, Jia Tan would not be a suitable replacement, since they don't friggin' exist.

Yes, there's problems with this approach. When money gets involved, incentives can get distorted. It limits the pool of acceptable maintainers to those employable by a Big Company. But I think these are solvable problems, especially if there's a strong culture of maintainer independence. It provides a real improvement over the current situation of putting so much pressure on individuals doing good work for no benefit.

Is it a solution at all? Say Oracle offer Lasse Collin $200k for maintaining xz but he doesn't want to work for Oracle so he refuses, then what? Amazon offer Lasse $200k but they require fulltime work on other open source packages which he isn't experienced with or interested in, so he refuses, then what? Google employ someone else for $200k but they can't force Lasse Collin to hand over commit rights to xz to Google, or force him to work with a fulltime Google employee pestering with many constant changes trying to justify their job, and they can't force Debian to accept a new Google fork of xz, so then what? And NetFlix, Microsoft, Facebook, Uber, they can't all employ an xz maintainer, xz doesn't need that many people, but if they just employ 'open source maintainers' scattering their attention over all kinds of random projects they have no prior experience with, how would they catch this kind of subtle multi-pronged long-term attack on some low-attention, slow moving, project?

Google already employ a very capable security team who find issues in all kinds of projects and publicise them, they didn't find this one. Is it likely this attack could have made its way into ChromeOS and Android if it wasn't noticed now, or would Google have noticed it?

> "1) don't make everyone so mad that everyone on the Internet asks for you to be fired"

So it's a sinecure position, doing nothing is the safest thing?

> "and 2) name a suitable replacement when you're ready to move on"

How could Lasse Collin have named a more suitable replacement than someone who seemed technically capable, interested in the project, and motivated to work on some of the boring details like the build system and tests and didn't seem to be doing it for the hype of saying "I improved compression by 10%" for their resume? Are they needing to be skilled in hiring and recruitment now?

> but if they just employ 'open source maintainers' scattering their attention over all kinds of random projects they have no prior experience with

They would pay the people who are doing the work now. Under this hypothetical, one of the Big Companies would have hired Lasse as the xz maintainer, for example. His job responsibilities are to maintain xz as he had been doing, and identify a successor when he's ready to move on. Nothing else.

> So it's a sinecure position, doing nothing is the safest thing?

No. Not doing the maintenance tasks would make everyone mad, violating one of the two job responsibilities.

> How could Lasse Collin have named a more suitable replacement than someone who seemed technically capable, interested in the project, and motivated to work on [it]

Lasse would suggest Tan as a suitable replacement. Big Company's hiring pipeline would approach Tan and start the hiring process (in person interviews, tax docs, etc etc). At some point they would realize Tan isn't a real person and not hire him. Or, the adversary would have to put up "a body" behind the profile to keep up the act, which is a much higher bar to clear than what actually happened.

> "Lasse would suggest Tan as a suitable replacement. Big Company's hiring pipeline would approach Tan and start the hiring process (in person interviews, tax docs, etc etc). At some point they would realize Tan isn't a real person and not hire him"

What if they find that Tan is a real person but he doesn't want to work for Amazon or legally can't (This is before knowing of his commits being malicious, we're assuming he's a fake profile but he could be a real person being blackmailed)? Collin can't leave? Collin has to pick someone else out of a candidate pool of people he's never heard of? Same question if they find Tan isn't a real person - what then; is there an obligation to review all of Tan's historical commits? Just committing under a pseudonym or pen name isn't a crime, is it? Would the new maintainer be obliged to review all historic commits or audit the codebase or anything? Would Amazon want their money back from Lasse once it became clear that he had let a bad actor commit changes which opened a serious security hole during his tenure as maintainer?

> "No. Not doing the maintenance tasks would make everyone mad, violating one of the two job responsibilities."

What he was doing already was apparently ignoring issues for months and making people like Jigar Kumar annoyed. Which is fine for a volunteer thing. If "Jigar Kumar" is a sock-puppet, nobody knew that at the time of their posts; Lasse Collins' hypothetical employer wouldn't have known and would surely be on his case about paying him lots of money for maintenance while complaints are flowing on the project's public mailing list, right? Either they're paying him to do what he was doing before (which involved apparently ignoring work and making some people mad) or they're paying him to do more than he was doing before (which is not what you said).

It doesn't seem like it would work - but if it did work it doesn't seem like it would have helped against this attack?

Well, there's a few issues I'm trying to target. I'm trying to work backwards from "how do we stop bad actor Tan from getting maintainer access to the project?" Creating an identify-verified relationship (employment) is a good fit for that, I think. And it nicely solves some other related issues with the current volunteer maintainership model. Lasse may not have felt the strong pressure/health issues if he was being paid to do the work. Or, if he was feeling burnt out, he may have felt more comfortable passing the torch earlier if there was a clear framework to do so, backed by an entity that can do some of the heavy lifting of naming/validating a successor.

> What if they find that Tan is a real person but he doesn't want to work for Amazon or legally can't

I think this would be a fairly rare occurrence, but it's one I called out as a potential problem in my original post, yeah ("smaller pool of possible maintainers"). If there isn't a clear successor, I think the maintainer could inform the Big Company that they'd like to move on in the next year or two, and Big Company could maybe find an internal engineer who wants to take over the role. Or maybe this more formal sponsored-maintainership arrangement would create incentives for outside contributors to aim for those positions, so there's more often someone waiting to take over (and then be verified by Big Company's hiring process).

> is there an obligation [for the maintainer] to review all of Tan's historical commits? Would the new maintainer be obliged to review all historic commits or audit the codebase or anything? Would Amazon [fire] Lasse once it became clear that he had let a bad actor commit changes which opened a serious security hole during his tenure as maintainer?

(I tweaked your questions a tiny bit to rephrase them as I interpreted them. I think the spirit of your questions was kept, I apologize if not.) If these tasks fall under the "don't make everyone mad" job responsibility, then yes. If not, then no, to all of these. There are no obligations other than the two I mentioned: don't piss off the community and help name a successor. It's up to the project's community to decide if the maintainer is not meeting their obligations, not the sponsoring Big Company.

> What he was doing already was apparently ignoring issues for months and making people like Jigar Kumar annoyed.

I'm not sure. It seems like Kumar was a bad actor. Was there actually a real maintenance issue? If so, maybe it could have been avoided in the first place by the sponsorship arrangement, like I mentioned at the top of this reply. Or, the community could raise the problem to Big Company, who can do the work of verifying that there is a problem and working with the maintainer to resolve it. Instead what happened here, which was for one burned out guy deciding to hand the keys over to some email address.

It would stop a sock puppet, but Jai Tan might be a real person, a real developer paid or blackmailed by a hostile group; Amazon might just have hired him and handed over maintainer access to him thinking it was above board, if a problem hadn't been found yet. I don't know where Jai Tan claimed to be from, but it's quite possible they would say "I don't have a passport", "I can't leave my family to travel to America for an in-person interview", "I'm not in good health to travel", "I don't speak English well enough for an in-person interview", "I live in a poor country without a functioning government and have no tax documents", or etc. etc. excuses which are quite plausible

> "Or, if he was feeling burnt out, he may have felt more comfortable passing the torch earlier if there was a clear framework to do so, backed by an entity that can do some of the heavy lifting of naming/validating a successor."

Your suggested $200k is equivalent to £160k GBP in the UK; look at this UK average salary list: https://uk.jobted.com/ no job comes close; not Managing Director, IT director, Finance Director, Aerospace engineer, DevOps engineer, neurosurgeon, nothing on the list is above £110k. Sure there are many people earning that much as a senior devops AI cloud security specialist in a fast paced London based fintech trading house, but the idea that someone would comfortably pass up a salary around the 98th percentile of incomes in the country for like 2 days a month of work because they're "feeling burnt out" is unthinkable. Anyone sensible would hold onto that until they pried it out of one's cold dead hands; American tech salaries are almost literally unbelievable. Even moreso if we consider a maintainer in a poorer country.

> "I tweaked your questions a tiny bit to rephrase them as I interpreted them. I think the spirit of your questions was kept, I apologize if not"

I started writing Tan, but then changed it. A lot of your reply is assuming that we know there were malicious patches and suspect Jigar Kumar was a bad actor and that the big company would be somewhat US friendly. But we can't plan to know all that for all situations like this. Some people will be speculating that the previous paid maintainer was complicit and all their work and merges are now suspect. The billion dollar company who hired Collins in this hypothetical maintainer could be Baidu or Yandex or Saudi Aramco, and then people would be suspicious. It's one thing to have your task be "don't make people mad" but doesn't that change if people getting mad can give you unbounded retrospective work and responsibility?

> "If these tasks fall under the "don't make everyone mad" job responsibility, then yes. [...] Was there actually a real maintenance issue? [...] Or, the community could raise the problem to Big Company, who can do the work of verifying that there is a problem and working with the maintainer to resolve it."

As soon as the internet becomes aware that they can get anything merged ASAP by threatening to get mad, everyone will be mad about everything all the time. Whom at the BigCo will do the work of verifying whether there is a problem? I mean, let's put Lasse Collins on a specific team along with other employees who are expected to work 40-80 hour weeks while he isn't. The pressure on the team manager to ditch the maintainer and distribute his salary among the other team members would be constant. If those other team members see him doing less work for similar or more money it would be a morale killer and they would want to leave. If they also have to know his project well enough to follow all the drama and things people are complaining about and tease out what is and isn't a real problem and coerce him to do his job, sorry 'work with him', well, they won't be very motivated to do that.

The guy in Nebraska is still not getting anything but will also not have the stress of becoming one of the main characters/victims in a huge attack.

Does that even matter these days?

Especially if we're talking nation state level stuff convincing histories are not hard to create to deflect casual observers.

>Be more paranoid.

Most people in OSS just want to write some code to do something, not defend the world against evil.

Yes, it would have prevented this attack. It isn’t totally sufficient but it’s quick and easy and would have prevented this attack.

“Most people don’t want …”

I get it. I think the issue is that pushing junk code from malicious contributors into your project causes more hassle in the long run. If you just want to code and make stuff work, you should probably be careful who you pull from. It’s not just for the benefit of others, it’s first and foremost to protect the code base and the time and sanity of other contributors.

The more paranoid walls you put up the more actual contributors getting into the movement say "eh, screw this, who wants to code anyway".

This isn't just a problems with OSS, this is a fundamental issue the internet as a whole is experiencing and no one has good answers that don't have terrible trade offs of their own.

If my speculation is correct then the the exact date on which access was granted must then first be known, after that a trusted backup of the repo from before that date is needed. Ideally Lasse Collin would have a daily backup of the repo.

Although perhaps the entire repo may have to be completely audited at this point.

I think it's a pretty reasonable assumption that this didn't happen, though it would be nice to see a testimony to that effect from someone trustworthy (e.g. "I restored a xz checkout from a backup taken before 5.6.0 and all the commit IDs match").

Isn't this part just a few pages of code, if that?

I agree that it will be blindingly obvious for the reasons you list.

[0] https://inversegravity.net/2019/web-of-trust-dead/

Let's say they have fake passports and physically appear at key signing parties. Now you're screwed because even your peers (that you thought know how to validate identities using passports) will get fooled.

Read more on GPG's trust levels: https://www.gnupg.org/gph/en/manual/x334.html

Github+ even has a scheme for signing artifacts such that you have some level of trust they came from inside their Actions system, derived from some git commit. This would allow the benefits of a modular build for a large product like a distro, while preserving a chain of trust in its component parts.

+Not advocating a dependency on Github per se -- the same sort of artifact attestation scheme could be implemented elsewhere.

Ideally, we would have a mechanism to verify that a given build _matches_ the source for a release. Then it wouldn't matter where it was built, we would be able to independently verify nothing funky happened.

One of the best things about Golang is that packages are shared direct via source repositories (Github) rather than a package repository containing mystery tarballs. I understand the appeal of package repositories, but without proper security constraints it's a security disaster waiting to happen.

I believe this means it was unauthorized, not unauthenticated.

For something like SSH which has authentication and authorization as features, I would expect to talk about an RCE in that context, and not the backdoor’s auth features.

This backdoor bypasses both authentication (not requiring an account password, authorized key, etc on the target system) as well as authorization (as it doesn’t check a user against any policy for what commands or users can log in).

The links you point to there are about "RCE attack" which also implies not authorized.

> in the introductory paragraph it reads "unauthenticated, targeted remote code execution ... I believe this means it was unauthorized, not unauthenticated.

You again:

> agree that RCE, unqualified, means unauthorized RCE

So you agree that your "unauthorized" qualification from your orignal post was unwarranted, since all unqualified RCE are unauthorized.

Now if you want to split hairs, you'll say "but the introductory paragraph said unauthenticated, which is qualified RCE, thus I am still right". ok then

All RCEs are classified in either unauthenticated or authenticated, the former being the worst (or best if you're a researcher/hacker).

Consider the impact if anybody else, not just this attacker, could have exploited it -- on the other hand, it may have been discovered sooner (had it also not been discovered by accident first).

Can somebody write a script which diffs the release tarballs from the git sources for all debian packages and detects whether there are any differences apart from the files added by autotools :)?

Reflections on Trusting Trust becomes even more interesting if you consider photolithography machines being backdoored.

I would think a "real identity" should be required by linux distros for all /major/ open source projects/library committers which are included in the distro, so that we can hold folks legally accountable

Google's Know, Prevent, Fix blog post floated the idea of stronger identity for open source in https://security.googleblog.com/2021/02/know-prevent-fix-fra... and there was very significant pushback. We learned a lot from that.

The fundamental problem with stronger identity is that spy agencies can create very convincing ones. How are distros going to detect those?

I realize, it's a hard problem. (And, thanks for the link to the "Know, Prevent, Fix" post.)

PS: FWIW, I "win my bread" by working for a company that "does" open source.

Edit: Some projects I know use in-person GPG key signing, or maintainer summits (Linux kernel), etc. None of them are perfect, but raises the bar for motivated anonymous contributors with malicious intent, wanting to become maintainers.

What I'm saying is, in context of critical-path software, the identity of maintainers vs non-maintainers matters more. I'm not naively claiming that it'll "solve" the problem at hand, just that it's another layer in defense. For a critical software, you shouldn't be able to simply submit a "patch"[1] such as:

  tests: Add-binary-blob-with-a-subtle-backdoor.xz

  Signed-off-by: "Anonymous Rabbit" 

It's a messy situation. How much, if at all, "clever tech" can mitigate this human "trust issue" is an open problem for now.

[1] https://git.tukaani.org/?p=xz.git;a=commitdiff;h=cf44e4b7f5d

And there are other reasons some people might not want to travel outside their nearby region. For instance, they might be taking care of an elderly relative. Or they might be the elderly relative, with travel counter-indicated for health reasons.

But you have a point. As an agency you can seed two jiatan's to serve diligently for a couple of years following the strict 2-person code reviews and then still poison the project. On the other hand, if the xz build process was automated and transparent and release artifacts were reproducible and verifiable even in this poor condition of xz-utils as a project it would have been much harder to squeeze in a rogue m4/build-to-host.m4

Source code is provided without warranty and this statement is clear in the license.

Putting an verified identity behind the source code publish is basically starting to twist said said no-warranty. Fuck that.