Re: [PATCH bpf-next v5 3/7] bpf: lsm: provide attachment points for BPF LSM programs

From: Andrii Nakryiko
Date: Tue Mar 24 2020 - 17:27:10 EST


On Tue, Mar 24, 2020 at 9:12 AM KP Singh <kpsingh@xxxxxxxxxxxx> wrote:
>
> On 24-MÃr 11:39, KP Singh wrote:
> > On 23-MÃr 12:59, Andrii Nakryiko wrote:
> > > On Mon, Mar 23, 2020 at 9:45 AM KP Singh <kpsingh@xxxxxxxxxxxx> wrote:
> > > >
> > > > From: KP Singh <kpsingh@xxxxxxxxxx>
> > > >
> > > > When CONFIG_BPF_LSM is enabled, nops functions, bpf_lsm_<hook_name>, are
> > > > generated for each LSM hook. These nops are initialized as LSM hooks in
> > > > a subsequent patch.
> > > >
> > > > Signed-off-by: KP Singh <kpsingh@xxxxxxxxxx>
> > > > Reviewed-by: Brendan Jackman <jackmanb@xxxxxxxxxx>
> > > > Reviewed-by: Florent Revest <revest@xxxxxxxxxx>
> > > > ---
> > > > include/linux/bpf_lsm.h | 21 +++++++++++++++++++++
> > > > kernel/bpf/bpf_lsm.c | 19 +++++++++++++++++++
> > > > 2 files changed, 40 insertions(+)
> > > > create mode 100644 include/linux/bpf_lsm.h
> > > >
> > > > diff --git a/include/linux/bpf_lsm.h b/include/linux/bpf_lsm.h
> > > > new file mode 100644
> > > > index 000000000000..c6423a140220
> > > > --- /dev/null
> > > > +++ b/include/linux/bpf_lsm.h
> > > > @@ -0,0 +1,21 @@
> > > > +/* SPDX-License-Identifier: GPL-2.0 */
> > > > +
> > > > +/*
> > > > + * Copyright (C) 2020 Google LLC.
> > > > + */
> > > > +
> > > > +#ifndef _LINUX_BPF_LSM_H
> > > > +#define _LINUX_BPF_LSM_H
> > > > +
> > > > +#include <linux/bpf.h>
> > > > +#include <linux/lsm_hooks.h>
> > > > +
> > > > +#ifdef CONFIG_BPF_LSM
> > > > +
> > > > +#define LSM_HOOK(RET, NAME, ...) RET bpf_lsm_##NAME(__VA_ARGS__);
> > > > +#include <linux/lsm_hook_names.h>
> > > > +#undef LSM_HOOK
> > > > +
> > > > +#endif /* CONFIG_BPF_LSM */
> > > > +
> > > > +#endif /* _LINUX_BPF_LSM_H */
> > > > diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
> > > > index 82875039ca90..530d137f7a84 100644
> > > > --- a/kernel/bpf/bpf_lsm.c
> > > > +++ b/kernel/bpf/bpf_lsm.c
> > > > @@ -7,6 +7,25 @@
> > > > #include <linux/filter.h>
> > > > #include <linux/bpf.h>
> > > > #include <linux/btf.h>
> > > > +#include <linux/lsm_hooks.h>
> > > > +#include <linux/bpf_lsm.h>
> > > > +
> > > > +/* For every LSM hook that allows attachment of BPF programs, declare a NOP
> > > > + * function where a BPF program can be attached as an fexit trampoline.
> > > > + */
> > > > +#define LSM_HOOK(RET, NAME, ...) LSM_HOOK_##RET(NAME, __VA_ARGS__)
> > > > +
> > > > +#define LSM_HOOK_int(NAME, ...) \
> > > > +noinline __weak int bpf_lsm_##NAME(__VA_ARGS__) \
> > > > +{ \
> > > > + return 0; \
> > > > +}
> > > > +
> > > > +#define LSM_HOOK_void(NAME, ...) \
> > > > +noinline __weak void bpf_lsm_##NAME(__VA_ARGS__) {}
> > > > +
> > >
> > > Could unify with:
> > >
> > > #define LSM_HOOK(RET, NAME, ...) noinline __weak RET bpf_lsm_##NAME(__VA_ARGS__)
> > > {
> > > return (RET)0;
> > > }
> > >
> > > then you don't need LSM_HOOK_int and LSM_HOOK_void.
> >
> > Nice.
> >
> > But, given that we are adding default values and that
> > they are only needed for int hooks, we will need to keep the macros
> > separate for int and void. Or, Am I missing a trick here?
> >
> > - KP
>
> Actually, was able to get it work. not setting a default for void
> hooks makes the macros messier. So i just set it void. For example:
>
> LSM_HOOK(void, void, bprm_committing_creds, struct linux_binprm *bprm)

surprised this works, was going to propose to specify `(void)0` as
default value :)

>
> This also allows me to use the cleanup you suggested and not having
> to split every usage into int and void.
>

Nice, one of the reasons for proposing this.

> - KP
>
> >
> > >
> > > > +#include <linux/lsm_hook_names.h>
> > > > +#undef LSM_HOOK
> > > >
> > > > const struct bpf_prog_ops lsm_prog_ops = {
> > > > };
> > > > --
> > > > 2.20.1
> > > >