Conclusions from CVE-2024-3094 (libxz disaster)

Petr Štetiar ynezz at
Mon Apr 1 05:49:46 PDT 2024

Daniel Golle <daniel at> [2024-03-30 15:30:49]:


> In many ways, we are already better

I would probably avoid such bold statements and would be more humble, since
you never know why OpenWrt wasn't directly targeted.

> I believe that the current tendency to use tarballs rather than
> (reproducible!) git checkouts is also problematic to begin with.

Git checkouts are currently problematic as well, IIRC the build is going to
happily use whatever Git is happy with. I mean, if the hash of the downloaded
tarball source code doesn't match, then the tarball is removed and Git clone
is performed, new source code tarball is produced, but the tarball hash is not
going to be checked again.

Perhaps this package source code integrity checks should be mandatory, not

> So why not **always** use that instead of potentially shady and hard to
> verify tarballs?

In this case, they were targeting specific audience and this attack vector was
cheapest/fastest, so the source code origin doesn't really matter.

> Why do we need to rely one proprietary hacks such as Gibhub codeload
> just to safe a few megabytes of traffic and a few seconds of build
> time?

Ok, I don't like GH either, but I find this irrelevant, origin of the source
code is not a problem, the content is the problem.

> There are even too many problems to reproduce even those supposedly
> automated Github-generated tarballs. Nobody actually checks that.

FYI we do on the CI

> 9bd7d8b, c7c2257, 77368ec, 86994e1, 954142f, 4c5d910, 21f713d, ...
> Probably all of those have trivial causes and there isn't anything
> malicious going on there.

I agree, I guess, that in some cases it might point to a subtle bug somewhere
in the source code tarball packaging path (host kernel, tools, container?),
maybe another backdoor in the works/testing? :P

Anyway, we should perhaps consider treating this situations in supply chain
more seriously, so perhaps in this cases of package hash failures, we need to
document it better, with more details in the commit message and maybe even
better, gather always more evidence in a separate GH issue, so its possible to
reconstruct the complete picture if we really find out 2 years later, that it
was something malicious going on somewhere? Whatever it might be.

> Always using git checkouts instead of tarballs would also makes it
> much easier for maintainers to at least have a quick look at the
> changes made in an upstream project between versions (a quick scroll
> over  'git diff oldtag..newtag' or even just 'git log --stat
> oldtag..newtag' doesn't take much more time than manually validating a

Although I mostly (always?) doing that Git diff/log, during bumps/reviews, I'm
sorry, but I'm not able to spot such masked backdoors :-)

> release tarball GPG signature in most cases, if there even is any...).

And speaking about the signatures:

 openwrt.git $ ~/bin/ "1 year ago"

   Total commits: 3267 (since 1 year ago)
   Good signatures: 94 (2.00%)
   Unsigned commits: 2456 (75.00%)
   Bad signatures: 0 (0%)
   Unknown signature status: 717 (21.00%)

 (script source

> Hiding a malicious change in a commit is infinitely harder than hiding it in
> a tarball.

Harder? Yes. Impossible? No.



More information about the openwrt-adm mailing list