Skip to content

Compatibility matrix for eCapture's eBPF objects across kernels — useful as a CI lane? #1018

Description

@ErenAri

Hi @cfc4n / eCapture team,

I maintain bpfcompat (Apache-2.0): it boots real distro kernels in disposable QEMU/KVM VMs, load/attach-validates a compiled .bpf.o, and reports a per-kernel pass/fail matrix with classified failure reasons (missing BTF, CO-RE relocation, unsupported map/program/attach type).

eCapture's uprobe/kprobe CO-RE objects have to load across a very wide kernel range (and arches), which is exactly the surface bpfcompat targets. I'd be glad to run eCapture's compiled objects across a kernel spread and share the matrix — complementary evidence / a possible non-blocking CI lane, not a replacement for anything.

Context: a Falco modern_bpf reference matrix (it pinpoints the 5.8 ring-buffer boundary) and an enterprise/backported-kernel matrix (RHEL 8/9/10, Oracle UEK, Amazon Linux 2 incl. no-BTF 4.14, openSUSE — 14/14). Live demo.

Would a published "does eCapture load on these kernels" matrix (or a GH Actions lane) be useful? Happy to put one together against your build.

Independent test of a public project; not affiliated with or endorsed by eCapture.

Metadata

Metadata

Assignees

No one assigned

    Labels

    questionFurther information is requested

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions