bpf: Enforce that struct_ops programs be GPL-only
authorToke Høiland-Jørgensen <toke@redhat.com>
Fri, 26 Mar 2021 10:03:13 +0000 (11:03 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 14 Apr 2021 06:42:00 +0000 (08:42 +0200)
commitb7004ecafade02c6e413b8360802b1a524cee97f
tree263c5f82dcae3e70bf3bb479854496d872e3b472
parent3015db3de71545d385416a4a4e1242509343cceb
bpf: Enforce that struct_ops programs be GPL-only

commit 12aa8a9467b354ef893ce0fc5719a4de4949a9fb upstream.

With the introduction of the struct_ops program type, it became possible to
implement kernel functionality in BPF, making it viable to use BPF in place
of a regular kernel module for these particular operations.

Thus far, the only user of this mechanism is for implementing TCP
congestion control algorithms. These are clearly marked as GPL-only when
implemented as modules (as seen by the use of EXPORT_SYMBOL_GPL for
tcp_register_congestion_control()), so it seems like an oversight that this
was not carried over to BPF implementations. Since this is the only user
of the struct_ops mechanism, just enforcing GPL-only for the struct_ops
program type seems like the simplest way to fix this.

Fixes: 0baf26b0fcd7 ("bpf: tcp: Support tcp_congestion_ops in bpf")
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Link: https://lore.kernel.org/bpf/20210326100314.121853-1-toke@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
kernel/bpf/verifier.c