2023-07-23 08:38:57

by Lin Ma

Subject: [PATCH v1] bpf: Add length check for SK_DIAG_BPF_STORAGE_REQ_MAP_FD parsing

The nla_for_each_nested parsing in function bpf_sk_storage_diag_alloc
does not check the length of the nested attribute. This can lead to an
out-of-attribute read and allow a malformed nlattr (e.g., length 0) to
be viewed as a 4 byte integer.

This patch adds additional check before the execution of nla_get_u32 to
make sure the attribute has enough length.

Fixes: 1ed4d92458a9 ("bpf: INET_DIAG support in bpf_sk_storage")
Signed-off-by: Lin Ma <[email protected]>
---
net/core/bpf_sk_storage.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/net/core/bpf_sk_storage.c b/net/core/bpf_sk_storage.c
index d4172534dfa8..6f1afbb394a6 100644
--- a/net/core/bpf_sk_storage.c
+++ b/net/core/bpf_sk_storage.c
@@ -511,6 +511,11 @@ bpf_sk_storage_diag_alloc(const struct nlattr *nla_stgs)
if (nla_type(nla) != SK_DIAG_BPF_STORAGE_REQ_MAP_FD)
continue;

+ if (nla_len(nla) < sizeof(map_fd)) {
+ err = -EINVAL;
+ goto err_free;
+ }
+
map_fd = nla_get_u32(nla);
map = bpf_map_get(map_fd);
if (IS_ERR(map)) {
--
2.17.1

2023-07-24 22:27:59

by Jakub Kicinski

Subject: Re: [PATCH v1] bpf: Add length check for SK_DIAG_BPF_STORAGE_REQ_MAP_FD parsing

On Sun, 23 Jul 2023 15:54:52 +0800 Lin Ma wrote:
> diff --git a/net/core/bpf_sk_storage.c b/net/core/bpf_sk_storage.c
> index d4172534dfa8..6f1afbb394a6 100644
> --- a/net/core/bpf_sk_storage.c
> +++ b/net/core/bpf_sk_storage.c
> @@ -511,6 +511,11 @@ bpf_sk_storage_diag_alloc(const struct nlattr *nla_stgs)
> if (nla_type(nla) != SK_DIAG_BPF_STORAGE_REQ_MAP_FD)
> continue;
>
> + if (nla_len(nla) < sizeof(map_fd)) {
> + err = -EINVAL;
> + goto err_free;
> + }

You can move this check earlier, when the attributes are getting
counted. That way we can avoid the alloc/free on error.

2023-07-25 00:53:03

by Lin Ma

Subject: Re: [PATCH v1] bpf: Add length check for SK_DIAG_BPF_STORAGE_REQ_MAP_FD parsing

Hi Jakub,

>
> You can move this check earlier, when the attributes are getting
> counted. That way we can avoid the alloc/free on error.

Good point, will fix that and prepare another patch

Thanks
Lin