[PATCH v8 bpf-next 17/18] selftests/bpf: add BPF token-enabled tests

[Andrii Nakryiko]

On Tue, Oct 17, 2023 at 4:02 AM Jiri Olsa <olsajiri at gmail.com> wrote:
>
> On Mon, Oct 16, 2023 at 11:02:19AM -0700, Andrii Nakryiko wrote:
> > Add a selftest that attempts to conceptually replicate intended BPF
> > token use cases inside user namespaced container.
> >
> > Child process is forked. It is then put into its own userns and mountns.
> > Child creates BPF FS context object and sets it up as desired. This
> > ensures child userns is captures as owning userns for this instance of
> > BPF FS.
> >
> > This context is passed back to privileged parent process through Unix
> > socket, where parent creates and mounts it as a detached mount. This
> > mount FD is passed back to the child to be used for BPF token creation,
> > which allows otherwise privileged BPF operations to succeed inside
> > userns.
> >
> > We validate that all of token-enabled privileged commands (BPF_BTF_LOAD,
> > BPF_MAP_CREATE, and BPF_PROG_LOAD) work as intended. They should only
> > succeed inside the userns if a) BPF token is provided with proper
> > allowed sets of commands and types; and b) namespaces CAP_BPF and other
> > privileges are set. Lacking a) or b) should lead to -EPERM failures.
> >
> > Based on suggested workflow by Christian Brauner ([0]).
> >
> >   [0] https://lore.kernel.org/bpf/20230704-hochverdient-lehne-eeb9eeef785e@brauner/
> >
> > Signed-off-by: Andrii Nakryiko <andrii at kernel.org>
> > ---
> >  .../testing/selftests/bpf/prog_tests/token.c  | 629 ++++++++++++++++++
> >  1 file changed, 629 insertions(+)
> >  create mode 100644 tools/testing/selftests/bpf/prog_tests/token.c
> >
> > diff --git a/tools/testing/selftests/bpf/prog_tests/token.c b/tools/testing/selftests/bpf/prog_tests/token.c
> > new file mode 100644
> > index 000000000000..41cee6b4731e
> > --- /dev/null
> > +++ b/tools/testing/selftests/bpf/prog_tests/token.c
> > @@ -0,0 +1,629 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/* Copyright (c) 2023 Meta Platforms, Inc. and affiliates. */
> > +#define _GNU_SOURCE
> > +#include <test_progs.h>
> > +#include <bpf/btf.h>
> > +#include "cap_helpers.h"
> > +#include <fcntl.h>
> > +#include <sched.h>
> > +#include <signal.h>
> > +#include <unistd.h>
> > +#include <linux/filter.h>
> > +#include <linux/unistd.h>
> > +#include <sys/mount.h>
> > +#include <sys/socket.h>
> > +#include <sys/syscall.h>
> > +#include <sys/un.h>
> > +
> > +/* copied from include/uapi/linux/mount.h, as including it conflicts with
> > + * sys/mount.h include
> > + */
> > +enum fsconfig_command {
> > +     FSCONFIG_SET_FLAG       = 0,    /* Set parameter, supplying no value */
> > +     FSCONFIG_SET_STRING     = 1,    /* Set parameter, supplying a string value */
> > +     FSCONFIG_SET_BINARY     = 2,    /* Set parameter, supplying a binary blob value */
> > +     FSCONFIG_SET_PATH       = 3,    /* Set parameter, supplying an object by path */
> > +     FSCONFIG_SET_PATH_EMPTY = 4,    /* Set parameter, supplying an object by (empty) path */
> > +     FSCONFIG_SET_FD         = 5,    /* Set parameter, supplying an object by fd */
> > +     FSCONFIG_CMD_CREATE     = 6,    /* Invoke superblock creation */
> > +     FSCONFIG_CMD_RECONFIGURE = 7,   /* Invoke superblock reconfiguration */
> > +};
>
> I'm getting compilation fail, because fsconfig_command seems to be
> included through the sys/mount.h include, but CI is green hum :-\
>
> when I get -E output I can see:
>
>         ...
>         # 16 "./cap_helpers.h"
>         int cap_enable_effective(__u64 caps, __u64 *old_caps);
>         int cap_disable_effective(__u64 caps, __u64 *old_caps);
>         # 7 "/home/jolsa/kernel/linux-qemu/tools/testing/selftests/bpf/prog_tests/token.c" 2
>
>         # 1 "/usr/include/sys/mount.h" 1 3 4
>         # 27 "/usr/include/sys/mount.h" 3 4
>         # 1 "/usr/lib/gcc/x86_64-redhat-linux/13/include/stddef.h" 1 3 4
>         # 28 "/usr/include/sys/mount.h" 2 3 4
>
>         # 1 "/home/jolsa/kernel/linux-qemu/tools/include/uapi/linux/mount.h" 1 3 4
>         # 96 "/home/jolsa/kernel/linux-qemu/tools/include/uapi/linux/mount.h" 3 4
>
>         # 96 "/home/jolsa/kernel/linux-qemu/tools/include/uapi/linux/mount.h" 3 4
>         enum fsconfig_command {
>          FSCONFIG_SET_FLAG = 0,
>          FSCONFIG_SET_STRING = 1,
>          FSCONFIG_SET_BINARY = 2,
>          FSCONFIG_SET_PATH = 3,
>          FSCONFIG_SET_PATH_EMPTY = 4,
>          FSCONFIG_SET_FD = 5,
>          FSCONFIG_CMD_CREATE = 6,
>          FSCONFIG_CMD_RECONFIGURE = 7,
>         };
>
>
>         ...
>
>
>         # 21 "/home/jolsa/kernel/linux-qemu/tools/testing/selftests/bpf/prog_tests/token.c"
>         enum fsconfig_command {
>          FSCONFIG_SET_FLAG = 0,
>          FSCONFIG_SET_STRING = 1,
>          FSCONFIG_SET_BINARY = 2,
>          FSCONFIG_SET_PATH = 3,
>          FSCONFIG_SET_PATH_EMPTY = 4,
>          FSCONFIG_SET_FD = 5,
>          FSCONFIG_CMD_CREATE = 6,
>          FSCONFIG_CMD_RECONFIGURE = 7,
>         };
>
>
> it's probably included through this bit in the /usr/include/sys/mount.h:
>
>         #ifdef __has_include
>         # if __has_include ("linux/mount.h")
>         #  include "linux/mount.h"
>         # endif
>         #endif
>
> which was added 'recently' in https://sourceware.org/git/?p=glibc.git;a=commit;h=774058d72942249f71d74e7f2b639f77184160a6
>
> maybe you use older glibs headers? or perhaps it might be my build setup

No, I'm pretty sure I have older headers. I'd like this to work in
both environments, so I need to fix this. I'll try to make this work
with uapi header, I guess. Thanks for reporting!

>
> jirka
>

[...]