Re: [PATCH 2/2] bpf: guard sock_ops rtt_min against non-locked tcp_sock

[Martin KaFai Lau]

On Fri, Apr 17, 2026 at 02:31:26AM +0000, Werner Kasselman wrote:
> diff --git a/net/core/filter.c b/net/core/filter.c
> index e8ad062f63bc..9c43193a5c39 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -10827,14 +10827,12 @@ static u32 sock_ops_convert_ctx_access(enum bpf_access_type type,
>  			     sizeof(struct minmax));
>  		BUILD_BUG_ON(sizeof(struct minmax) <
>  			     sizeof(struct minmax_sample));
> +		BUILD_BUG_ON(offsetof(struct tcp_sock, rtt_min) +
> +			     offsetof(struct minmax_sample, v) > S16_MAX);

This doesn't look like a test that is added by human.
Will sizeof(tcp_sock) ever reach S16_MAX? It is unnecessarily defensive and
inconsistent with other tcp_sock field loads.

> diff --git a/tools/testing/selftests/bpf/prog_tests/sock_ops_get_sk.c b/tools/testing/selftests/bpf/prog_tests/sock_ops_get_sk.c
> index 343d92c4df30..1aea4c97d5d3 100644
> --- a/tools/testing/selftests/bpf/prog_tests/sock_ops_get_sk.c
> +++ b/tools/testing/selftests/bpf/prog_tests/sock_ops_get_sk.c

Separate the test in its own patch.

Also tag and add revision to subject, "[PATCH v3 bpf...]".
Take a look at how other patches are posted in the bpf mailing list.

pw-bot: cr