[RFC,v2,net-next] net/tls: Implement getsockopt SOL_TLS TLS_RX

From: Yutaro Hayakawa <yhayakawa3720@gmail.com>

Implement the getsockopt SOL_TLS TLS_RX which is currently missing. The
primary usecase is to use it in conjunction with TCP_REPAIR to
checkpoint/restore the TLS record layer state.

TLS connection state usually exists on the user space library. So
basically we can easily extract it from there, but when the TLS
connections are delegated to the kTLS, it is not the case. We need to
have a way to extract the TLS state from the kernel for both of TX and
RX side.

The new TLS_RX getsockopt copies the crypto_info to user in the same
way as TLS_TX does.

We have described use cases in our research work in Netdev 0x14
Transport Workshop [1].

Also, there is an TLS implementation called tlse [2] which supports
TLS connection migration. They have support of kTLS and their code
shows that they are expecting the future support of this option.

[1] https://speakerdeck.com/yutarohayakawa/prism-proxies-without-the-pain
[2] https://github.com/eduardsui/tlse

Signed-off-by: Yutaro Hayakawa <yhayakawa3720@gmail.com>
---
Changes in v2:
- Remove duplicated memcpy for each cipher suites

Thanks for your reply. Reflected the comments.

 net/tls/tls_main.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)

--
1.8.3.1

Apology. Kernel Test Robot just pointed out that the v2 patch has a
compilation error. My local compilation tests have passed because I
didn't enable the kTLS kernel option. Let me resubmit the fixed
version soon.

Yutaro

2020年8月30日(日) 23:02 Yutaro Hayakawa <yhayakawa3720@gmail.com>:
>
> From: Yutaro Hayakawa <yhayakawa3720@gmail.com>
>
> Implement the getsockopt SOL_TLS TLS_RX which is currently missing. The
> primary usecase is to use it in conjunction with TCP_REPAIR to
> checkpoint/restore the TLS record layer state.
>
> TLS connection state usually exists on the user space library. So
> basically we can easily extract it from there, but when the TLS
> connections are delegated to the kTLS, it is not the case. We need to
> have a way to extract the TLS state from the kernel for both of TX and
> RX side.
>
> The new TLS_RX getsockopt copies the crypto_info to user in the same
> way as TLS_TX does.
>
> We have described use cases in our research work in Netdev 0x14
> Transport Workshop [1].
>
> Also, there is an TLS implementation called tlse [2] which supports
> TLS connection migration. They have support of kTLS and their code
> shows that they are expecting the future support of this option.
>
> [1] https://speakerdeck.com/yutarohayakawa/prism-proxies-without-the-pain
> [2] https://github.com/eduardsui/tlse
>
> Signed-off-by: Yutaro Hayakawa <yhayakawa3720@gmail.com>
> ---
> Changes in v2:
> - Remove duplicated memcpy for each cipher suites
>
> Thanks for your reply. Reflected the comments.
>
>  net/tls/tls_main.c | 25 +++++++++++++++++--------
>  1 file changed, 17 insertions(+), 8 deletions(-)
>
> diff --git a/net/tls/tls_main.c b/net/tls/tls_main.c
> index bbc52b0..0271441 100644
> --- a/net/tls/tls_main.c
> +++ b/net/tls/tls_main.c
> @@ -330,12 +330,13 @@ static void tls_sk_proto_close(struct sock *sk, long timeout)
>                 tls_ctx_free(sk, ctx);
>  }
>
> -static int do_tls_getsockopt_tx(struct sock *sk, char __user *optval,
> -                               int __user *optlen)
> +static int do_tls_getsockopt_conf(struct sock *sk, char __user *optval,
> +                                 int __user *optlen, int tx)
>  {
>         int rc = 0;
>         struct tls_context *ctx = tls_get_ctx(sk);
>         struct tls_crypto_info *crypto_info;
> +       struct tls_cipher_context *cctx;
>         int len;
>
>         if (get_user(len, optlen))
> @@ -352,7 +353,13 @@ static int do_tls_getsockopt_tx(struct sock *sk, char __user *optval,
>         }
>
>         /* get user crypto info */
> -       crypto_info = &ctx->crypto_send.info;
> +       if (tx) {
> +               crypto_info = &ctx->crypto_send.info;
> +               cctx = &ctx->tx;
> +       } else {
> +               crypto_info = &ctx->crypto_recv.info;
> +               cctx = &ctx->rx;
> +       }
>
>         if (!TLS_CRYPTO_INFO_READY(crypto_info)) {
>                 rc = -EBUSY;
> @@ -379,9 +386,9 @@ static int do_tls_getsockopt_tx(struct sock *sk, char __user *optval,
>                 }
>                 lock_sock(sk);
>                 memcpy(crypto_info_aes_gcm_128->iv,
> -                      ctx->tx.iv + TLS_CIPHER_AES_GCM_128_SALT_SIZE,
> +                      cctx->iv + TLS_CIPHER_AES_GCM_128_SALT_SIZE,
>                        TLS_CIPHER_AES_GCM_128_IV_SIZE);
> -               memcpy(crypto_info_aes_gcm_128->rec_seq, ctx->tx.rec_seq,
> +               memcpy(crypto_info_aes_gcm_128->rec_seq, cctx->rec_seq,
>                        TLS_CIPHER_AES_GCM_128_REC_SEQ_SIZE);
>                 release_sock(sk);
>                 if (copy_to_user(optval,
> @@ -403,9 +410,9 @@ static int do_tls_getsockopt_tx(struct sock *sk, char __user *optval,
>                 }
>                 lock_sock(sk);
>                 memcpy(crypto_info_aes_gcm_256->iv,
> -                      ctx->tx.iv + TLS_CIPHER_AES_GCM_256_SALT_SIZE,
> +                      cctx->iv + TLS_CIPHER_AES_GCM_256_SALT_SIZE,
>                        TLS_CIPHER_AES_GCM_256_IV_SIZE);
> -               memcpy(crypto_info_aes_gcm_256->rec_seq, ctx->tx.rec_seq,
> +               memcpy(crypto_info_aes_gcm_256->rec_seq, cctx->rec_seq,
>                        TLS_CIPHER_AES_GCM_256_REC_SEQ_SIZE);
>                 release_sock(sk);
>                 if (copy_to_user(optval,
> @@ -429,7 +436,9 @@ static int do_tls_getsockopt(struct sock *sk, int optname,
>
>         switch (optname) {
>         case TLS_TX:
> -               rc = do_tls_getsockopt_tx(sk, optval, optlen);
> +       case TLS_RX:
> +               rc = do_tls_getsockopt_conf(sk, optval, optlen,
> +                                           optname == TLS_TX);
>                 break;
>         default:
>                 rc = -ENOPROTOOPT;
> --
> 1.8.3.1