bpf: Enforce RCU protection for KF_RCU_PROTECTED

Currently, KF_RCU_PROTECTED only applies to iterator APIs and that too
in a convoluted fashion: the presence of this flag on the kfunc is used
to set MEM_RCU in iterator type, and the lack of RCU protection results
in an error only later, once next() or destroy() methods are invoked on
the iterator. While there is no bug, this is certainly a bit
unintuitive, and makes the enforcement of the flag iterator specific.

In the interest of making this flag useful for other upcoming kfuncs,
e.g. scx_bpf_cpu_curr() [0][1], add enforcement for invoking the kfunc
in an RCU critical section in general.

This would also mean that iterator APIs using KF_RCU_PROTECTED will
error out earlier, instead of throwing an error for lack of RCU CS
protection when next() or destroy() methods are invoked.

In addition to this, if the kfuncs tagged KF_RCU_PROTECTED return a
pointer value, ensure that this pointer value is only usable in an RCU
critical section. There might be edge cases where the return value is
special and doesn't need to imply MEM_RCU semantics, but in general, the
assumption should hold for the majority of kfuncs, and we can revisit
things if necessary later.

  [0]: https://lore.kernel.org/all/20250903212311.369697-3-christian.loehle@arm.com
  [1]: https://lore.kernel.org/all/20250909195709.92669-1-arighi@nvidia.com

Tested-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20250917032755.4068726-2-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

Author: kkdwvd

Documentation/bpf/kfuncs.rst

@@ -335,9 +335,26 @@ consider doing refcnt != 0 check, especially when returning a KF_ACQUIRE
 pointer. Note as well that a KF_ACQUIRE kfunc that is KF_RCU should very likely
 also be KF_RET_NULL.
 
+2.4.8 KF_RCU_PROTECTED flag
+---------------------------
+
+The KF_RCU_PROTECTED flag is used to indicate that the kfunc must be invoked in
+an RCU critical section. This is assumed by default in non-sleepable programs,
+and must be explicitly ensured by calling ``bpf_rcu_read_lock`` for sleepable
+ones.
+
+If the kfunc returns a pointer value, this flag also enforces that the returned
+pointer is RCU protected, and can only be used while the RCU critical section is
+active.
+
+The flag is distinct from the ``KF_RCU`` flag, which only ensures that its
+arguments are at least RCU protected pointers. This may transitively imply that
+RCU protection is ensured, but it does not work in cases of kfuncs which require
+RCU protection but do not take RCU protected arguments.
+
 .. _KF_deprecated_flag:
 
-2.4.8 KF_DEPRECATED flag
+2.4.9 KF_DEPRECATED flag
 ------------------------
 
 The KF_DEPRECATED flag is used for kfuncs which are scheduled to be

kernel/bpf/verifier.c

@@ -13931,6 +13931,11 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 		return -EACCES;
 	}
 
+	if (is_kfunc_rcu_protected(&meta) && !in_rcu_cs(env)) {
+		verbose(env, "kernel func %s requires RCU critical section protection\n", func_name);
+		return -EACCES;
+	}
+
 	/* In case of release function, we get register number of refcounted
 	 * PTR_TO_BTF_ID in bpf_kfunc_arg_meta, do the release now.
 	 */
@@ -14044,6 +14049,9 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 			/* Ensures we don't access the memory after a release_reference() */
 			if (meta.ref_obj_id)
 				regs[BPF_REG_0].ref_obj_id = meta.ref_obj_id;
+
+			if (is_kfunc_rcu_protected(&meta))
+				regs[BPF_REG_0].type |= MEM_RCU;
 		} else {
 			mark_reg_known_zero(env, regs, BPF_REG_0);
 			regs[BPF_REG_0].btf = desc_btf;
@@ -14052,6 +14060,8 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 
 			if (meta.func_id == special_kfunc_list[KF_bpf_get_kmem_cache])
 				regs[BPF_REG_0].type |= PTR_UNTRUSTED;
+			else if (is_kfunc_rcu_protected(&meta))
+				regs[BPF_REG_0].type |= MEM_RCU;
 
 			if (is_iter_next_kfunc(&meta)) {
 				struct bpf_reg_state *cur_iter;

tools/testing/selftests/bpf/progs/cgroup_read_xattr.c

@@ -73,7 +73,7 @@ int BPF_PROG(use_css_iter_non_sleepable)
 }
 
 SEC("lsm.s/socket_connect")
-__failure __msg("expected an RCU CS")
+__failure __msg("kernel func bpf_iter_css_new requires RCU critical section protection")
 int BPF_PROG(use_css_iter_sleepable_missing_rcu_lock)
 {
 	u64 cgrp_id = bpf_get_current_cgroup_id();

tools/testing/selftests/bpf/progs/iters_task_failure.c

@@ -15,7 +15,7 @@ void bpf_rcu_read_lock(void) __ksym;
 void bpf_rcu_read_unlock(void) __ksym;
 
 SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
-__failure __msg("expected an RCU CS when using bpf_iter_task_next")
+__failure __msg("kernel func bpf_iter_task_new requires RCU critical section protection")
 int BPF_PROG(iter_tasks_without_lock)
 {
 	struct task_struct *pos;
@@ -27,7 +27,7 @@ int BPF_PROG(iter_tasks_without_lock)
 }
 
 SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
-__failure __msg("expected an RCU CS when using bpf_iter_css_next")
+__failure __msg("kernel func bpf_iter_css_new requires RCU critical section protection")
 int BPF_PROG(iter_css_without_lock)
 {
 	u64 cg_id = bpf_get_current_cgroup_id();