add CI

[tharvik]

add GitHub actions support to automatically check that PRs lint/compile/build/...

• CI: lint C code, compile & build everything, verify a few
  • many TODO added, some uncompilable via make, some badly linking (missing lpm_* impl), every NET=dpdk need some unknown header
• automate venv generation, you only need to make verify-* and make will take care of its creation
• add a tool-test target, w/ venv auto-gen
• some nf/*/config were unvalid (missing ,)
• fix some compile issues

questions

• I removed some tests for the tool, as it was failing w/ "Incomplete, but sound" errors. I don't understand theses yet, was it intended to fail (wrap in try/expect instead)? can you take a look at it and see if there is some way to not remove theses?
• I'm unable to actually verify any NF (excepted nop), the CI bails out after 6h, locally it is been running for a while now. did I missed smth (like using my own riced z3)? some magic compiling option that simplifies angr's job?

[SolalPirelli]

https://github.com/angr/claripy/issues/271 is the shift issue

[SolalPirelli]

f402ee7dfe7db28adb885f3ac1ce845a8f684c12 should fix the verif issue... at the cost of verifying the iptables NF, but well, omelette, eggs, ...

[tharvik]

In general all of the bpf stuff isn't for general use, it was for an example in the paper

as discussed, removing bpf & experiments in a next PR

f402ee7 should fix the verif issue... at the cost of verifying the iptables NF, but well, omelette, eggs, ...

hoo nice, thanks for fix, it works (waaay faster) now :)

[SolalPirelli]

need to revert the "CI on push only" change so it triggers for this PR as well, I guess? e436377f405772dfd14b02447de09b818e91949b

[tharvik]

need to revert the "CI on push only" change so it triggers for this PR as well, I guess? e436377

ho right, forgot to drop the "TMP" commits, good catch

[tharvik]

sadly, the more I run the verify-nf, the more issues arises: dropping router because of the shift issue.

[SolalPirelli]

maybe if you open an issue or a PR for angr/claripy#271 they'll be more willing to take it since it affects >1 person 😇

(or maybe it's already fixed and the dependency versions need an update?)

[tharvik]

(or maybe it's already fixed and the dependency versions need an update?)

yep, that's what I'm trying currently :crossed_fingers:

maybe if you open an issue or a PR for angr/claripy#271 they'll be more willing to take it since it affects >1

and if need be, fallback on that, maybe with a PR to push it even more!

[tharvik]

(or maybe it's already fixed and the dependency versions need an update?)

yep, that's what I'm trying currently crossed_fingers

sadly, it isn't fixed w/ latest release :disappointed:

[tharvik]

maybe if you open an issue or a PR for angr/claripy#271 they'll be more willing to take it since it affects >1

and if need be, fallback on that, maybe with a PR to push it even more!

PR opened at https://github.com/angr/claripy/pull/297 :) waiting for merge and release, then bumping theses deps and then most of the NF can be verified :tada:

[tharvik]

alright, upstream released, deps bumped, one last review and we can merge it IMO :)

[SolalPirelli]

Does the tmp commit still need a revert? Or is there something else preventing CI from working on this PR?

[tharvik]

Does the tmp commit still need a revert?

a rebase & force push got rid of it

Or is there something else preventing CI from working on this PR?

hum, as an external contributor is adding the CI, maybe you need to enable the checks somehow? or you can see the action ran on the c4dt fork

[SolalPirelli]

Weird, GitHub's docs say I should see something about this, oh well.