FreeBSD source tree development on the GitHub platform

Hey group, I always make sure to update my fork of base system on GitHub. I was inspecting and looking over the CI/CD pipeline when syncing my forked repository and I tabbed over because I noticed some security notifications. I guess CodeQL popped some bugs during the builds. What does this mean here essentially. Is there a problem in these programming languages' codebases or is it FreeBSD base system relevant

Screenshot (85).png
 
I just forked the CodeQL github repository to poke around. I'll look into it and if I find more information I'll follow up
 
I just forked the CodeQL github repository to poke around. I'll look into it and if I find more information I'll follow up
So I have been tracking codeql’s development over the past couple of days. A lot goes on with various different programming languages and frameworks too many to name. This is via issues and gh pr’s. This is my latest update
 
Last edited:
CodeQL is a static analysis tool, it primarily checks some coding style guidelines that it considers good practice to be conformed to, unless we explicitly configured it to do something else. We do not care what other people consider to be good style though, we have our own coding style guide lines codified in the style(9) man page. What comes on top of that is that alot of code that is part of FreeBSD is not required to abide to our style, because FreeBSD contains a lot of code that originates from external sources and is also developed and maintained by third parties, such as the LLVM project. We could modify their code to fit our style, but that would make it difficult to make quality bug reports to the upstream developers, as even if we know very well which branches are taken to cause misbehaviour, renaming their symbols, reindenting their code, adding/removing lines and other whitespace commit games will complicate communications with the upstream developers and make merging upstream commits more difficult, too, while not really providing us with any advantage either, so we try avoiding unneccessary modifications of third party code, even if that means that it does not conform to our coding style.

So, even in the event we have configured CodeQL to check our style, the whole contrib stuff will have to be considered false positive.
 
I think I have the solution for gh that works for the freebsd community. I have been poking at the ci stack on gh to see how it works.

I think a robust solution for base system that works for the community are either official FreeBSD self-hosted runners or an official FreeBSD modular ci/cd from the community that does not bind base system dev into vendor lock-in.

Some more information on gh’s ci/cd framework: GitHub Actions Runners
 
I believe I have traced it down to codeql’s autobuild.sh shell script seen here. I have been poking at it in the ci passes of gh’s stack 😂
IMG_5722.jpeg
 
Back
Top