[v3,bpf-next,2/5] bpf: Mark cgroups and dfl_cgrp fields as trusted.

Commit Message

Alexei Starovoitov Feb. 28, 2023, 4:01 a.m. UTC

From: Alexei Starovoitov <[e-mail]>

bpf programs sometimes do:
bpf_cgrp_storage_get(&map, task->cgroups->dfl_cgrp, ...);
It is safe to do, because cgroups->dfl_cgrp pointer is set diring init and
never changes. The task->cgroups is also never NULL. It is also set during init
and will change when task switches cgroups. For any trusted task pointer
dereference of cgroups and dfl_cgrp should yield trusted pointers. The verifier
wasn't aware of this. Hence in gcc compiled kernels task->cgroups dereference
was producing PTR_TO_BTF_ID without modifiers while in clang compiled kernels
the verifier recognizes __rcu tag in cgroups field and produces
PTR_TO_BTF_ID | MEM_RCU | MAYBE_NULL.
Tag cgroups and dfl_cgrp as trusted to equalize clang and gcc behavior.
When GCC supports btf_type_tag such tagging will done directly in the type.

Signed-off-by: Alexei Starovoitov <[e-mail]>
---
 kernel/bpf/verifier.c | 5 +++++
 1 file changed, 5 insertions(+)

Comments

David Vernet Feb. 28, 2023, 4:45 p.m. UTC | #1

On Mon, Feb 27, 2023 at 08:01:18PM -0800, Alexei Starovoitov wrote:
> From: Alexei Starovoitov <[e-mail]>
> 
> bpf programs sometimes do:
> bpf_cgrp_storage_get(&map, task->cgroups->dfl_cgrp, ...);
> It is safe to do, because cgroups->dfl_cgrp pointer is set diring init and
> never changes. The task->cgroups is also never NULL. It is also set during init
> and will change when task switches cgroups. For any trusted task pointer
> dereference of cgroups and dfl_cgrp should yield trusted pointers. The verifier
> wasn't aware of this. Hence in gcc compiled kernels task->cgroups dereference
> was producing PTR_TO_BTF_ID without modifiers while in clang compiled kernels
> the verifier recognizes __rcu tag in cgroups field and produces
> PTR_TO_BTF_ID | MEM_RCU | MAYBE_NULL.
> Tag cgroups and dfl_cgrp as trusted to equalize clang and gcc behavior.
> When GCC supports btf_type_tag such tagging will done directly in the type.
> 
> Signed-off-by: Alexei Starovoitov <[e-mail]>

Acked-by: David Vernet <[e-mail]>

Tejun Heo Feb. 28, 2023, 7:32 p.m. UTC | #2

On Mon, Feb 27, 2023 at 08:01:18PM -0800, Alexei Starovoitov wrote:
> From: Alexei Starovoitov <[e-mail]>
> 
> bpf programs sometimes do:
> bpf_cgrp_storage_get(&map, task->cgroups->dfl_cgrp, ...);
> It is safe to do, because cgroups->dfl_cgrp pointer is set diring init and
                                                              ^
							      u

> never changes. The task->cgroups is also never NULL. It is also set during init
> and will change when task switches cgroups. For any trusted task pointer
> dereference of cgroups and dfl_cgrp should yield trusted pointers. The verifier
> wasn't aware of this. Hence in gcc compiled kernels task->cgroups dereference
> was producing PTR_TO_BTF_ID without modifiers while in clang compiled kernels
> the verifier recognizes __rcu tag in cgroups field and produces
> PTR_TO_BTF_ID | MEM_RCU | MAYBE_NULL.
> Tag cgroups and dfl_cgrp as trusted to equalize clang and gcc behavior.
> When GCC supports btf_type_tag such tagging will done directly in the type.
> 
> Signed-off-by: Alexei Starovoitov <[e-mail]>

Acked-by: Tejun Heo <[e-mail]>

Thanks.

Patch

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 5cb8b623f639..e4234266e76d 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -5011,6 +5011,10 @@  static int bpf_map_direct_read(struct bpf_map *map, int off, int size, u64 *val)
 
 BTF_TYPE_SAFE_NESTED(struct task_struct) {
 	const cpumask_t *cpus_ptr;
+	struct css_set *cgroups;
+};
+BTF_TYPE_SAFE_NESTED(struct css_set) {
+	struct cgroup *dfl_cgrp;
 };
 
 static bool nested_ptr_is_trusted(struct bpf_verifier_env *env,
@@ -5022,6 +5026,7 @@  static bool nested_ptr_is_trusted(struct bpf_verifier_env *env,
 		return false;
 
 	BTF_TYPE_EMIT(BTF_TYPE_SAFE_NESTED(struct task_struct));
+	BTF_TYPE_EMIT(BTF_TYPE_SAFE_NESTED(struct css_set));
 
 	return btf_nested_type_is_trusted(&env->log, reg, off);
 }