mbox series

[v4,bpf-next,0/4] Add O_PATH-based BPF_OBJ_PIN and BPF_OBJ_GET support

Message ID 20230523170013.728457-1-andrii@kernel.org (mailing list archive)
Headers show
Series Add O_PATH-based BPF_OBJ_PIN and BPF_OBJ_GET support | expand

Message

Andrii Nakryiko May 23, 2023, 5 p.m. UTC
Add ability to specify pinning location within BPF FS using O_PATH-based FDs,
similar to openat() family of APIs. Patch #2 adds necessary kernel-side
changes. Patch #3 exposes this through libbpf APIs. Patch #4 uses new mount
APIs (fsopen, fsconfig, fsmount) to demonstrated how now it's possible to work
with detach-mounted BPF FS using new BPF_OBJ_PIN and BPF_OBJ_GET
functionality. We also add few more tests using various combinations of
path_fd and pathname to validate proper argument propagation in kernel code.

This feature is inspired as a result of recent conversations during
LSF/MM/BPF 2023 conference about shortcomings of being able to perform BPF
objects pinning only using lookup-based paths.

v3->v4:
  - libbpf v1.3 bump (Daniel);
v2->v3:
  - __s32 for path_fd in union bpf_attr (Christian);
  - added subtest for absolute/relative paths during pinning/getting;
  - added pre-patch moving LSM hook (security_path_mknod) around (Christian);
v1->v2:
  - add BPF_F_PATH_FD flag that should go along with path FD (Christian).

Andrii Nakryiko (4):
  bpf: Validate BPF object in BPF_OBJ_PIN before calling LSM
  libbpf: start v1.3 development cycle
  bpf: support O_PATH FDs in BPF_OBJ_PIN and BPF_OBJ_GET commands
  selftests/bpf: add path_fd-based BPF_OBJ_PIN and BPF_OBJ_GET tests

 include/linux/bpf.h                           |   4 +-
 include/uapi/linux/bpf.h                      |  10 +
 kernel/bpf/inode.c                            |  27 +-
 kernel/bpf/syscall.c                          |  25 +-
 tools/include/uapi/linux/bpf.h                |  10 +
 tools/lib/bpf/bpf.c                           |  17 +-
 tools/lib/bpf/bpf.h                           |  18 +-
 tools/lib/bpf/libbpf.map                      |   5 +
 tools/lib/bpf/libbpf_version.h                |   2 +-
 .../bpf/prog_tests/bpf_obj_pinning.c          | 268 ++++++++++++++++++
 10 files changed, 359 insertions(+), 27 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/bpf_obj_pinning.c

Comments

Daniel Borkmann May 23, 2023, 7:43 p.m. UTC | #1
On 5/23/23 7:00 PM, Andrii Nakryiko wrote:
> Add ability to specify pinning location within BPF FS using O_PATH-based FDs,
> similar to openat() family of APIs. Patch #2 adds necessary kernel-side
> changes. Patch #3 exposes this through libbpf APIs. Patch #4 uses new mount
> APIs (fsopen, fsconfig, fsmount) to demonstrated how now it's possible to work
> with detach-mounted BPF FS using new BPF_OBJ_PIN and BPF_OBJ_GET
> functionality. We also add few more tests using various combinations of
> path_fd and pathname to validate proper argument propagation in kernel code.
> 
> This feature is inspired as a result of recent conversations during
> LSF/MM/BPF 2023 conference about shortcomings of being able to perform BPF
> objects pinning only using lookup-based paths.
> 
> v3->v4:
>    - libbpf v1.3 bump (Daniel);

Looks good, applied 2-4 now. (Patch 1 was already in the tree.) Thanks!