]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/commit
bpf: introduce BPF_F_ALLOW_OVERRIDE flag
authorAlexei Starovoitov <ast@fb.com>
Sat, 11 Feb 2017 04:28:24 +0000 (20:28 -0800)
committerDavid S. Miller <davem@davemloft.net>
Mon, 13 Feb 2017 02:52:19 +0000 (21:52 -0500)
commit7f677633379b4abb3281cdbe7e7006f049305c03
tree1e5e662e792467d09f8a4ebf5b96b2baa333ae25
parente722af6391949e8851310441bb0cec157d25611d
bpf: introduce BPF_F_ALLOW_OVERRIDE flag

If BPF_F_ALLOW_OVERRIDE flag is used in BPF_PROG_ATTACH command
to the given cgroup the descendent cgroup will be able to override
effective bpf program that was inherited from this cgroup.
By default it's not passed, therefore override is disallowed.

Examples:
1.
prog X attached to /A with default
prog Y fails to attach to /A/B and /A/B/C
Everything under /A runs prog X

2.
prog X attached to /A with allow_override.
prog Y fails to attach to /A/B with default (non-override)
prog M attached to /A/B with allow_override.
Everything under /A/B runs prog M only.

3.
prog X attached to /A with allow_override.
prog Y fails to attach to /A with default.
The user has to detach first to switch the mode.

In the future this behavior may be extended with a chain of
non-overridable programs.

Also fix the bug where detach from cgroup where nothing is attached
was not throwing error. Return ENOENT in such case.

Add several testcases and adjust libbpf.

Fixes: 3007098494be ("cgroup: add support for eBPF programs")
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: Daniel Mack <daniel@zonque.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/linux/bpf-cgroup.h
include/uapi/linux/bpf.h
kernel/bpf/cgroup.c
kernel/bpf/syscall.c
kernel/cgroup.c
samples/bpf/test_cgrp2_attach.c
samples/bpf/test_cgrp2_attach2.c
samples/bpf/test_cgrp2_sock.c
samples/bpf/test_cgrp2_sock2.c
tools/lib/bpf/bpf.c
tools/lib/bpf/bpf.h