| Message ID | 20200304191853.1529-5-kpsingh@chromium.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Introduce BPF_MODIFY_RET tracing progs | expand |
On Wed, Mar 4, 2020 at 2:20 PM KP Singh <kpsingh@chromium.org> wrote: > > From: KP Singh <kpsingh@google.com> > > - Allow BPF_MODIFY_RETURN attachment only to functions that are: > > * Whitelisted for error injection by checking > within_error_injection_list. Similar discussions happened for the > bpf_override_return helper. > > * security hooks, this is expected to be cleaned up with the LSM > changes after the KRSI patches introduce the LSM_HOOK macro: > > https://lore.kernel.org/bpf/20200220175250.10795-1-kpsingh@chromium.org/ > > - The attachment is currently limited to functions that return an int. > This can be extended later other types (e.g. PTR). > > Signed-off-by: KP Singh <kpsingh@google.com> > Acked-by: Andrii Nakryiko <andriin@fb.com> > --- > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c > index 2460c8e6b5be..ae32517d4ccd 100644 > --- a/kernel/bpf/verifier.c > +++ b/kernel/bpf/verifier.c > @@ -9800,6 +9801,33 @@ static int check_struct_ops_btf_id(struct bpf_verifier_env *env) > > return 0; > } > +#define SECURITY_PREFIX "security_" > + > +static int check_attach_modify_return(struct bpf_verifier_env *env) > +{ > + struct bpf_prog *prog = env->prog; > + unsigned long addr = (unsigned long) prog->aux->trampoline->func.addr; > + > + if (within_error_injection_list(addr)) > + return 0; > + > + /* This is expected to be cleaned up in the future with the KRSI effort > + * introducing the LSM_HOOK macro for cleaning up lsm_hooks.h. > + */ > + if (!strncmp(SECURITY_PREFIX, prog->aux->attach_func_name, > + sizeof(SECURITY_PREFIX) - 1)) { > + > + if (!capable(CAP_MAC_ADMIN)) > + return -EPERM; CAP_MAC_ADMIN was originally introduced for Smack and is not all-powerful wrt SELinux, so this is not a sufficient check for SELinux. We would want an actual security hook called here so we can implement a specific check over userspace being able to attach BPF progs to LSM hooks. CAP_MAC_ADMIN has other connotations to SELinux (presently the ability to set/get file security labels that are not known to the currently loaded policy).
On Thu, Mar 05, 2020 at 08:43:11AM -0500, Stephen Smalley wrote: > On Wed, Mar 4, 2020 at 2:20 PM KP Singh <kpsingh@chromium.org> wrote: > > > > From: KP Singh <kpsingh@google.com> > > > > - Allow BPF_MODIFY_RETURN attachment only to functions that are: > > > > * Whitelisted for error injection by checking > > within_error_injection_list. Similar discussions happened for the > > bpf_override_return helper. > > > > * security hooks, this is expected to be cleaned up with the LSM > > changes after the KRSI patches introduce the LSM_HOOK macro: > > > > https://lore.kernel.org/bpf/20200220175250.10795-1-kpsingh@chromium.org/ > > > > - The attachment is currently limited to functions that return an int. > > This can be extended later other types (e.g. PTR). > > > > Signed-off-by: KP Singh <kpsingh@google.com> > > Acked-by: Andrii Nakryiko <andriin@fb.com> > > --- > > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c > > index 2460c8e6b5be..ae32517d4ccd 100644 > > --- a/kernel/bpf/verifier.c > > +++ b/kernel/bpf/verifier.c > > @@ -9800,6 +9801,33 @@ static int check_struct_ops_btf_id(struct bpf_verifier_env *env) > > > > return 0; > > } > > +#define SECURITY_PREFIX "security_" > > + > > +static int check_attach_modify_return(struct bpf_verifier_env *env) > > +{ > > + struct bpf_prog *prog = env->prog; > > + unsigned long addr = (unsigned long) prog->aux->trampoline->func.addr; > > + > > + if (within_error_injection_list(addr)) > > + return 0; > > + > > + /* This is expected to be cleaned up in the future with the KRSI effort > > + * introducing the LSM_HOOK macro for cleaning up lsm_hooks.h. > > + */ > > + if (!strncmp(SECURITY_PREFIX, prog->aux->attach_func_name, > > + sizeof(SECURITY_PREFIX) - 1)) { > > + > > + if (!capable(CAP_MAC_ADMIN)) > > + return -EPERM; > > CAP_MAC_ADMIN was originally introduced for Smack and is not > all-powerful wrt SELinux, so this is not a sufficient check for > SELinux. I think you're misunderstanding the intent here. This facility is just a faster version of kprobe based fault injection. It doesn't care about LSM. Security is not a focus here. It can fault inject in a lot of places in the kernel: syscalls, kmalloc, page_alloc, fs internals, etc I think above capable() check created this confusion and we should remove it.
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c index 30841fb8b3c0..50080add2ab9 100644 --- a/kernel/bpf/btf.c +++ b/kernel/bpf/btf.c @@ -3710,14 +3710,26 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type, nr_args--; } - if ((prog->expected_attach_type == BPF_TRACE_FEXIT || - prog->expected_attach_type == BPF_MODIFY_RETURN) && - arg == nr_args) { - if (!t) - /* Default prog with 5 args. 6th arg is retval. */ - return true; - /* function return type */ - t = btf_type_by_id(btf, t->type); + if (arg == nr_args) { + if (prog->expected_attach_type == BPF_TRACE_FEXIT) { + if (!t) + return true; + t = btf_type_by_id(btf, t->type); + } else if (prog->expected_attach_type == BPF_MODIFY_RETURN) { + /* For now the BPF_MODIFY_RETURN can only be attached to + * functions that return an int. + */ + if (!t) + return false; + + t = btf_type_skip_modifiers(btf, t->type, NULL); + if (!btf_type_is_int(t)) { + bpf_log(log, + "ret type %s not allowed for fmod_ret\n", + btf_kind_str[BTF_INFO_KIND(t->info)]); + return false; + } + } } else if (arg >= nr_args) { bpf_log(log, "func '%s' doesn't have %d-th argument\n", tname, arg + 1); diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 2460c8e6b5be..ae32517d4ccd 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -19,6 +19,7 @@ #include <linux/sort.h> #include <linux/perf_event.h> #include <linux/ctype.h> +#include <linux/error-injection.h> #include "disasm.h" @@ -9800,6 +9801,33 @@ static int check_struct_ops_btf_id(struct bpf_verifier_env *env) return 0; } +#define SECURITY_PREFIX "security_" + +static int check_attach_modify_return(struct bpf_verifier_env *env) +{ + struct bpf_prog *prog = env->prog; + unsigned long addr = (unsigned long) prog->aux->trampoline->func.addr; + + if (within_error_injection_list(addr)) + return 0; + + /* This is expected to be cleaned up in the future with the KRSI effort + * introducing the LSM_HOOK macro for cleaning up lsm_hooks.h. + */ + if (!strncmp(SECURITY_PREFIX, prog->aux->attach_func_name, + sizeof(SECURITY_PREFIX) - 1)) { + + if (!capable(CAP_MAC_ADMIN)) + return -EPERM; + + return 0; + } + + verbose(env, "fmod_ret attach_btf_id %u (%s) is not modifiable\n", + prog->aux->attach_btf_id, prog->aux->attach_func_name); + + return -EINVAL; +} static int check_attach_btf_id(struct bpf_verifier_env *env) { @@ -10000,6 +10028,9 @@ static int check_attach_btf_id(struct bpf_verifier_env *env) } tr->func.addr = (void *)addr; prog->aux->trampoline = tr; + + if (prog->expected_attach_type == BPF_MODIFY_RETURN) + ret = check_attach_modify_return(env); out: mutex_unlock(&tr->mutex); if (ret)