tracingpolicy: add BPF operations support #2943
Conversation
✅ Deploy Preview for tetragon ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
tixxdz
force-pushed
the
pr/tixxdz/2024-09-api-bpf-kprobes
branch
2 times, most recently
from
152045b to
1c50078
Compare
tixxdz
added
the
release-note/minor
tixxdz
force-pushed
the
pr/tixxdz/2024-09-api-bpf-kprobes
branch
from
1c50078 to
d72fdf0
Compare
|
so IIUC you want the event to show the bpf command numbers ... IMO the types should be just for kernel objects could we do that without changing the ebpf code? we just need to get the int value from kernel (which we already have support for, no need to add new) and transform that number into bpf command enum value I think we could factor the printer code to do that.. to have kernel type and user side type.. so you'd configure the rinter and argument to have int type value for kernel code and 'bpf_cmd' type for user space processing code |
Signed-off-by: Djalal Harouni <tixxdz@gmail.com>
tixxdz
force-pushed
the
pr/tixxdz/2024-09-api-bpf-kprobes
branch
from
d72fdf0 to
6c4bd3d
Compare
Signed-off-by: Djalal Harouni <tixxdz@gmail.com>
tixxdz
force-pushed
the
pr/tixxdz/2024-09-api-bpf-kprobes
branch
from
6c4bd3d to
944c6e5
Compare
Yes makes sense, can you please check again Jiri, thank you! |
Suggested by Jiri. Introduce userspace types that will be used for pretty printing. This allows to add different enum types into proto definition, have a high level representation that allows to pretty print data to users without propagating those same types into bpf part. We define the bpfCmd enum in proto, add its type in userspace, but we do not propagate this into kernel, we just keep using int types for kernel and we do the translation back into userspace where it makes sense. Suggested-by: Jiri Olsa <jolsa@kernel.org> Signed-off-by: Djalal Harouni <tixxdz@gmail.com>
Signed-off-by: Djalal Harouni <tixxdz@gmail.com>
Signed-off-by: Djalal Harouni <tixxdz@gmail.com>
tixxdz
force-pushed
the
pr/tixxdz/2024-09-api-bpf-kprobes
branch
from
944c6e5 to
f1a4175
Compare
| @@ -6970,3 +6970,67 @@ tetragon_missed_prog_probes_total{attach="wake_up_new_task",policy="__base__"} 0 | |||
| prometheus.BuildFQName(consts.MetricsNamespace, "", "missed_prog_probes_total"))) | |||
|
|
|||
| } | |||
|
|
|||
| func TestKprobeBpfCmd(t *testing.T) { | |||
There was a problem hiding this comment.
So I guess the idea of the test is to detect the tetragon bpf program / map loading?
There was a problem hiding this comment.
Yes kept it small, but it can be used to detect all bpf commands.
Description
Add BPF operations support.
Changelog