[bpf-next,v3,1/3] bpf: Relax allowlist for css_task iter

Commit Message

Chuyi Zhou Oct. 25, 2023, 7:59 a.m. UTC

The newly added open-coded css_task iter would try to hold the global
css_set_lock in bpf_iter_css_task_new, so the bpf side has to be careful in
where it allows to use this iter. The mainly concern is dead locking on
css_set_lock. check_css_task_iter_allowlist() in verifier enforced css_task
can only be used in bpf_lsm hooks and sleepable bpf_iter.

This patch relax the allowlist for css_task iter. Any lsm and any iter
(even non-sleepable) and any sleepable are safe since they would not hold
the css_set_lock before entering BPF progs context.

This patch also fixes the misused BPF_TRACE_ITER in
check_css_task_iter_allowlist which compared bpf_prog_type with
bpf_attach_type.

Fixes: 9c66dc94b62ae ("bpf: Introduce css_task open-coded iterator kfuncs")
Signed-off-by: Chuyi Zhou <[e-mail]>
---
 kernel/bpf/verifier.c                            | 16 ++++++++++++----
 .../selftests/bpf/progs/iters_task_failure.c     |  4 ++--
 2 files changed, 14 insertions(+), 6 deletions(-)

Comments

Yonghong Song Oct. 31, 2023, 12:05 a.m. UTC | #1

On 10/25/23 12:59 AM, Chuyi Zhou wrote:
> The newly added open-coded css_task iter would try to hold the global
> css_set_lock in bpf_iter_css_task_new, so the bpf side has to be careful in
> where it allows to use this iter. The mainly concern is dead locking on
> css_set_lock. check_css_task_iter_allowlist() in verifier enforced css_task
> can only be used in bpf_lsm hooks and sleepable bpf_iter.
>
> This patch relax the allowlist for css_task iter. Any lsm and any iter
> (even non-sleepable) and any sleepable are safe since they would not hold
> the css_set_lock before entering BPF progs context.
>
> This patch also fixes the misused BPF_TRACE_ITER in
> check_css_task_iter_allowlist which compared bpf_prog_type with
> bpf_attach_type.
>
> Fixes: 9c66dc94b62ae ("bpf: Introduce css_task open-coded iterator kfuncs")
> Signed-off-by: Chuyi Zhou <[e-mail]>

Acked-by: Yonghong Song <[e-mail]>

Patch

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index e9bc5d4a25a1..9243b6ca4854 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -11088,6 +11088,12 @@  static int process_kf_arg_ptr_to_rbtree_node(struct bpf_verifier_env *env,
 						  &meta->arg_rbtree_root.field);
 }
 
+/*
+ * css_task iter allowlist is needed to avoid dead locking on css_set_lock.
+ * LSM hooks and iters (both sleepable and non-sleepable) are safe.
+ * Any sleepable progs are also safe since bpf_check_attach_target() enforce
+ * them can only be attached to some specific hook points.
+ */
 static bool check_css_task_iter_allowlist(struct bpf_verifier_env *env)
 {
 	enum bpf_prog_type prog_type = resolve_prog_type(env->prog);
@@ -11095,10 +11101,12 @@  static bool check_css_task_iter_allowlist(struct bpf_verifier_env *env)
 	switch (prog_type) {
 	case BPF_PROG_TYPE_LSM:
 		return true;
-	case BPF_TRACE_ITER:
-		return env->prog->aux->sleepable;
+	case BPF_PROG_TYPE_TRACING:
+		if (env->prog->expected_attach_type == BPF_TRACE_ITER)
+			return true;
+		fallthrough;
 	default:
-		return false;
+		return env->prog->aux->sleepable;
 	}
 }
 
@@ -11357,7 +11365,7 @@  static int check_kfunc_args(struct bpf_verifier_env *env, struct bpf_kfunc_call_
 		case KF_ARG_PTR_TO_ITER:
 			if (meta->func_id == special_kfunc_list[KF_bpf_iter_css_task_new]) {
 				if (!check_css_task_iter_allowlist(env)) {
-					verbose(env, "css_task_iter is only allowed in bpf_lsm and bpf iter-s\n");
+					verbose(env, "css_task_iter is only allowed in bpf_lsm, bpf_iter and sleepable progs\n");
 					return -EINVAL;
 				}
 			}
diff --git a/tools/testing/selftests/bpf/progs/iters_task_failure.c b/tools/testing/selftests/bpf/progs/iters_task_failure.c
index c3bf96a67dba..6b1588d70652 100644
--- a/tools/testing/selftests/bpf/progs/iters_task_failure.c
+++ b/tools/testing/selftests/bpf/progs/iters_task_failure.c
@@ -84,8 +84,8 @@  int BPF_PROG(iter_css_lock_and_unlock)
 	return 0;
 }
 
-SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
-__failure __msg("css_task_iter is only allowed in bpf_lsm and bpf iter-s")
+SEC("?fentry/" SYS_PREFIX "sys_getpgid")
+__failure __msg("css_task_iter is only allowed in bpf_lsm, bpf_iter and sleepable progs")
 int BPF_PROG(iter_css_task_for_each)
 {
 	u64 cg_id = bpf_get_current_cgroup_id();