BPF Recorder: Exclude Container Initialization from Recorded Profile

[mhils]

What type of PR is this?

/kind feature

What this PR does / why we need it:

This PR makes us 1) detect when containers are (about to be) initialized, and 2) makes sure the permissions exercised before this point are not included in the recorded profile.

Which issue(s) this PR fixes:

Fixes #2576

Does this PR have test?

Covered by existing E2E tests.

Special notes for your reviewer:

99% sure that some kernels won't support sched_prepare_exec, but let's see if it works in the first place before we revise.

Switched from sched_prepare_exec to an execve tracepoint.

Switched from execve to setsid to make sure we capture the last part of runc startup for seccomp.

Switched from execve to getppid in an attempt to make things more reliable.

Switched to an approach where we detect the initial unshare and then start suppressing events until runc reaches getppid. This avoids data races on the Go side.

Switched back to an approach where we clean the mount namespace instead of suppressing events.

Does this PR introduce a user-facing change?

The BPF profile recorder now excludes unnecessary permissions exercised during container init.

[k8s-ci-robot]

Hi @mhils. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 0% with 17 lines in your changes missing coverage. Please review.

Project coverage is 41.36%. Comparing base (11d77f4) to head (ff68cef).
Report is 602 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2623      +/-   ##
==========================================
- Coverage   45.50%   41.36%   -4.14%     
==========================================
  Files          79      109      +30     
  Lines        7782    18231   +10449     
==========================================
+ Hits         3541     7541    +4000     
- Misses       4099    10192    +6093     
- Partials      142      498     +356     

[ccojocar]

It is an impressive reduction in permissions in both apapmror and secocmp. 🤞 that this works!

[ccojocar]

/test pull-security-profiles-operator-test-e2e

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ccojocar, mhils

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [ccojocar]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[ccojocar]

/test pull-security-profiles-operator-test-e2e