Hello.
On středa 23. února 2022 21:10:04 CET Florian Westphal wrote:
> There is no guarantee that state->sk refers to a full socket.
>
> If refcount transitions to 0, sock_put calls sk_free which then ends up
> with garbage fields.
>
> I'd like to thank Oleksandr Natalenko and Jiri Benc for considerable
> debug work and pointing out state->sk oddities.
No thank you for spotting the exact issue.
> Fixes: ca6fb0651883 ("tcp: attach SYNACK messages to request sockets instead of listener")
> Signed-off-by: Florian Westphal <fw@xxxxxxxxx>
Tested-by: Oleksandr Natalenko <oleksandr@xxxxxxxxxx>
> ---
> No reproducer, so there is a chance that the reported crash is caused by
> something else. That said, I don't see how sock_put use is safe here.
Using your reproducer from [1] I could trigger the issue on unpatched kernel and couldn't trigger the issue on patched kernel with the patch from below applied.
[1] https://patchwork.ozlabs.org/project/netfilter-devel/patch/20220224152118.20619-1-fw@xxxxxxxxx/
>
> net/netfilter/nf_queue.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/netfilter/nf_queue.c b/net/netfilter/nf_queue.c
> index 6d12afabfe8a..178742a110d9 100644
> --- a/net/netfilter/nf_queue.c
> +++ b/net/netfilter/nf_queue.c
> @@ -54,7 +54,7 @@ static void nf_queue_entry_release_refs(struct nf_queue_entry *entry)
> dev_put(state->in);
> dev_put(state->out);
> if (state->sk)
> - sock_put(state->sk);
> + sock_gen_put(state->sk);
>
> #if IS_ENABLED(CONFIG_BRIDGE_NETFILTER)
> dev_put(entry->physin);
>
--
Oleksandr Natalenko (post-factum)
Principal Software Maintenance Engineer
