From patchwork Wed Aug 16 13:12:47 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Antony Antony X-Patchwork-Id: 13355165 X-Patchwork-Delegate: kuba@kernel.org Received: from lindbergh.monkeyblade.net (lindbergh.monkeyblade.net [23.128.96.19]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 1BEAB10967 for ; Wed, 16 Aug 2023 13:13:03 +0000 (UTC) Received: from a.mx.secunet.com (a.mx.secunet.com [62.96.220.36]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 8A805270C for ; Wed, 16 Aug 2023 06:12:55 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by a.mx.secunet.com (Postfix) with ESMTP id 5E5C5205ED; Wed, 16 Aug 2023 15:12:54 +0200 (CEST) X-Virus-Scanned: by secunet Received: from a.mx.secunet.com ([127.0.0.1]) by localhost (a.mx.secunet.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id kWRaxzbtPmhB; Wed, 16 Aug 2023 15:12:53 +0200 (CEST) Received: from mailout1.secunet.com (mailout1.secunet.com [62.96.220.44]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by a.mx.secunet.com (Postfix) with ESMTPS id B3BC520547; Wed, 16 Aug 2023 15:12:53 +0200 (CEST) Received: from cas-essen-02.secunet.de (unknown [10.53.40.202]) by mailout1.secunet.com (Postfix) with ESMTP id ADF9D80004A; Wed, 16 Aug 2023 15:12:53 +0200 (CEST) Received: from mbx-essen-02.secunet.de (10.53.40.198) by cas-essen-02.secunet.de (10.53.40.202) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.27; Wed, 16 Aug 2023 15:12:53 +0200 Received: from moon.secunet.de (172.18.149.1) by mbx-essen-02.secunet.de (10.53.40.198) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.27; Wed, 16 Aug 2023 15:12:52 +0200 Date: Wed, 16 Aug 2023 15:12:47 +0200 From: Antony Antony To: Steffen Klassert , Herbert Xu CC: Eyal Birger , , , Antony Antony Subject: [PATCH v5 ipsec-next 1/3] xfrm: Use the XFRM_GRO to indicate a GRO call on input Message-ID: <8fa66af337da070696cc5fe18ff3ee2b8fd82224.1692191034.git.antony.antony@secunet.com> Reply-To: References: <6dfd03c5fa0afb99f255f4a35772df19e33880db.1674156645.git.antony.antony@secunet.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: Precedence: first-class Priority: normal Organization: secunet X-ClientProxiedBy: cas-essen-02.secunet.de (10.53.40.202) To mbx-essen-02.secunet.de (10.53.40.198) X-EXCLAIMER-MD-CONFIG: 2c86f778-e09b-4440-8b15-867914633a10 X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_BLOCKED,SPF_HELO_NONE,SPF_PASS autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net X-Patchwork-Delegate: kuba@kernel.org From: Steffen Klassert This is needed to support GRO for ESP in UDP encapsulation. Signed-off-by: Steffen Klassert Co-developed-by: Antony Antony Signed-off-by: Antony Antony --- net/ipv4/esp4_offload.c | 2 +- net/ipv6/esp6_offload.c | 2 +- net/xfrm/xfrm_input.c | 6 ++---- 3 files changed, 4 insertions(+), 6 deletions(-) diff --git a/net/ipv4/esp4_offload.c b/net/ipv4/esp4_offload.c index 3969fa805679..77bb01032667 100644 --- a/net/ipv4/esp4_offload.c +++ b/net/ipv4/esp4_offload.c @@ -76,7 +76,7 @@ static struct sk_buff *esp4_gro_receive(struct list_head *head, /* We don't need to handle errors from xfrm_input, it does all * the error handling and frees the resources on error. */ - xfrm_input(skb, IPPROTO_ESP, spi, -2); + xfrm_input(skb, IPPROTO_ESP, spi, 0); return ERR_PTR(-EINPROGRESS); out_reset: diff --git a/net/ipv6/esp6_offload.c b/net/ipv6/esp6_offload.c index 75c02992c520..ee5f5abdb503 100644 --- a/net/ipv6/esp6_offload.c +++ b/net/ipv6/esp6_offload.c @@ -103,7 +103,7 @@ static struct sk_buff *esp6_gro_receive(struct list_head *head, /* We don't need to handle errors from xfrm_input, it does all * the error handling and frees the resources on error. */ - xfrm_input(skb, IPPROTO_ESP, spi, -2); + xfrm_input(skb, IPPROTO_ESP, spi, 0); return ERR_PTR(-EINPROGRESS); out_reset: diff --git a/net/xfrm/xfrm_input.c b/net/xfrm/xfrm_input.c index 39fb91ff23d9..7d4a0bb9ca8d 100644 --- a/net/xfrm/xfrm_input.c +++ b/net/xfrm/xfrm_input.c @@ -466,7 +466,7 @@ int xfrm_input(struct sk_buff *skb, int nexthdr, __be32 spi, int encap_type) struct xfrm_offload *xo = xfrm_offload(skb); struct sec_path *sp; - if (encap_type < 0) { + if (encap_type < 0 || (xo && xo->flags & XFRM_GRO)) { x = xfrm_input_state(skb); if (unlikely(x->km.state != XFRM_STATE_VALID)) { @@ -489,9 +489,7 @@ int xfrm_input(struct sk_buff *skb, int nexthdr, __be32 spi, int encap_type) seq = XFRM_SKB_CB(skb)->seq.input.low; goto resume; } - - /* encap_type < -1 indicates a GRO call. */ - encap_type = 0; + /* GRO call */ seq = XFRM_SPI_SKB_CB(skb)->seq; if (xo && (xo->flags & CRYPTO_DONE)) { From patchwork Wed Aug 16 13:13:03 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Antony Antony X-Patchwork-Id: 13355166 X-Patchwork-Delegate: kuba@kernel.org Received: from lindbergh.monkeyblade.net (lindbergh.monkeyblade.net [23.128.96.19]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 51C2A1096D for ; Wed, 16 Aug 2023 13:13:14 +0000 (UTC) Received: from a.mx.secunet.com (a.mx.secunet.com [62.96.220.36]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 35EF91FFF for ; Wed, 16 Aug 2023 06:13:12 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by a.mx.secunet.com (Postfix) with ESMTP id 09AE820849; Wed, 16 Aug 2023 15:13:11 +0200 (CEST) X-Virus-Scanned: by secunet Received: from a.mx.secunet.com ([127.0.0.1]) by localhost (a.mx.secunet.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hFnkMYG3ndaL; Wed, 16 Aug 2023 15:13:09 +0200 (CEST) Received: from mailout2.secunet.com (mailout2.secunet.com [62.96.220.49]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by a.mx.secunet.com (Postfix) with ESMTPS id E8062207B3; Wed, 16 Aug 2023 15:13:09 +0200 (CEST) Received: from cas-essen-02.secunet.de (unknown [10.53.40.202]) by mailout2.secunet.com (Postfix) with ESMTP id D900F80004A; Wed, 16 Aug 2023 15:13:09 +0200 (CEST) Received: from mbx-essen-02.secunet.de (10.53.40.198) by cas-essen-02.secunet.de (10.53.40.202) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.27; Wed, 16 Aug 2023 15:13:09 +0200 Received: from moon.secunet.de (172.18.149.1) by mbx-essen-02.secunet.de (10.53.40.198) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.27; Wed, 16 Aug 2023 15:13:09 +0200 Date: Wed, 16 Aug 2023 15:13:03 +0200 From: Antony Antony To: Steffen Klassert , Herbert Xu CC: Eyal Birger , , , Antony Antony Subject: [PATCH v5 ipsec-next 2/3] xfrm: Support GRO for IPv4 ESP in UDP encapsulation Message-ID: <97337d8e577e79724d687e3bffe5b69a5822decc.1692191034.git.antony.antony@secunet.com> Reply-To: References: <6dfd03c5fa0afb99f255f4a35772df19e33880db.1674156645.git.antony.antony@secunet.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: Precedence: first-class Priority: normal Organization: secunet X-ClientProxiedBy: cas-essen-01.secunet.de (10.53.40.201) To mbx-essen-02.secunet.de (10.53.40.198) X-EXCLAIMER-MD-CONFIG: 2c86f778-e09b-4440-8b15-867914633a10 X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_BLOCKED,SPF_HELO_NONE,SPF_PASS autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net X-Patchwork-Delegate: kuba@kernel.org From: Steffen Klassert This patch enables the GRO codepath for IPv4 ESP in UDP encapsulated packets. Decapsulation happens at L2 and saves a full round through the stack for each packet. This is also needed to support HW offload for ESP in UDP encapsulation. Signed-off-by: Steffen Klassert Co-developed-by: Antony Antony Signed-off-by: Antony Antony --- include/net/gro.h | 2 +- include/net/xfrm.h | 2 + net/ipv4/esp4_offload.c | 6 ++- net/ipv4/udp.c | 14 ++++++ net/ipv4/xfrm4_input.c | 94 +++++++++++++++++++++++++++++++++-------- 5 files changed, 98 insertions(+), 20 deletions(-) diff --git a/include/net/gro.h b/include/net/gro.h index a4fab706240d..41c12c5d1ea1 100644 --- a/include/net/gro.h +++ b/include/net/gro.h @@ -29,7 +29,7 @@ struct napi_gro_cb { /* Number of segments aggregated. */ u16 count; - /* Used in ipv6_gro_receive() and foo-over-udp */ + /* Used in ipv6_gro_receive() and foo-over-udp and esp-in-udp */ u16 proto; /* jiffies when first packet was created/queued */ diff --git a/include/net/xfrm.h b/include/net/xfrm.h index 33ee3f5936e6..3be5d11bb983 100644 --- a/include/net/xfrm.h +++ b/include/net/xfrm.h @@ -1711,6 +1711,8 @@ int xfrm6_output(struct net *net, struct sock *sk, struct sk_buff *skb); void xfrm6_local_rxpmtu(struct sk_buff *skb, u32 mtu); int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb); int xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb); +struct sk_buff *xfrm4_gro_udp_encap_rcv(struct sock *sk, struct list_head *head, + struct sk_buff *skb); int xfrm_user_policy(struct sock *sk, int optname, sockptr_t optval, int optlen); #else diff --git a/net/ipv4/esp4_offload.c b/net/ipv4/esp4_offload.c index 77bb01032667..34ebfdf0e986 100644 --- a/net/ipv4/esp4_offload.c +++ b/net/ipv4/esp4_offload.c @@ -32,6 +32,7 @@ static struct sk_buff *esp4_gro_receive(struct list_head *head, int offset = skb_gro_offset(skb); struct xfrm_offload *xo; struct xfrm_state *x; + int encap_type = 0; __be32 seq; __be32 spi; @@ -69,6 +70,9 @@ static struct sk_buff *esp4_gro_receive(struct list_head *head, xo->flags |= XFRM_GRO; + if (NAPI_GRO_CB(skb)->proto == IPPROTO_UDP) + encap_type = UDP_ENCAP_ESPINUDP; + XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4 = NULL; XFRM_SPI_SKB_CB(skb)->family = AF_INET; XFRM_SPI_SKB_CB(skb)->daddroff = offsetof(struct iphdr, daddr); @@ -76,7 +80,7 @@ static struct sk_buff *esp4_gro_receive(struct list_head *head, /* We don't need to handle errors from xfrm_input, it does all * the error handling and frees the resources on error. */ - xfrm_input(skb, IPPROTO_ESP, spi, 0); + xfrm_input(skb, IPPROTO_ESP, spi, encap_type); return ERR_PTR(-EINPROGRESS); out_reset: diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index aa32afd871ee..caf1aa733219 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -2681,6 +2681,17 @@ void udp_destroy_sock(struct sock *sk) } } +static void set_xfrm_gro_udp_encap_rcv(__u16 encap_type, unsigned short family, + struct udp_sock *up) +{ +#ifdef CONFIG_XFRM + if (up->gro_enabled && encap_type == UDP_ENCAP_ESPINUDP) { + if (family == AF_INET) + up->gro_receive = xfrm4_gro_udp_encap_rcv; + } +#endif +} + /* * Socket option code for UDP */ @@ -2730,6 +2741,8 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname, case 0: #ifdef CONFIG_XFRM case UDP_ENCAP_ESPINUDP: + set_xfrm_gro_udp_encap_rcv(val, sk->sk_family, up); + fallthrough; case UDP_ENCAP_ESPINUDP_NON_IKE: #if IS_ENABLED(CONFIG_IPV6) if (sk->sk_family == AF_INET6) @@ -2773,6 +2786,7 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname, udp_tunnel_encap_enable(sk->sk_socket); up->gro_enabled = valbool; up->accept_udp_l4 = valbool; + set_xfrm_gro_udp_encap_rcv(up->encap_type, sk->sk_family, up); release_sock(sk); break; diff --git a/net/ipv4/xfrm4_input.c b/net/ipv4/xfrm4_input.c index ad2afeef4f10..2f75b68d7db7 100644 --- a/net/ipv4/xfrm4_input.c +++ b/net/ipv4/xfrm4_input.c @@ -17,6 +17,8 @@ #include #include #include +#include +#include static int xfrm4_rcv_encap_finish2(struct net *net, struct sock *sk, struct sk_buff *skb) @@ -72,14 +74,7 @@ int xfrm4_transport_finish(struct sk_buff *skb, int async) return 0; } -/* If it's a keepalive packet, then just eat it. - * If it's an encapsulated packet, then pass it to the - * IPsec xfrm input. - * Returns 0 if skb passed to xfrm or was dropped. - * Returns >0 if skb should be passed to UDP. - * Returns <0 if skb should be resubmitted (-ret is protocol) - */ -int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) +static int __xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb, bool pull) { struct udp_sock *up = udp_sk(sk); struct udphdr *uh; @@ -110,7 +105,7 @@ int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) case UDP_ENCAP_ESPINUDP: /* Check if this is a keepalive packet. If so, eat it. */ if (len == 1 && udpdata[0] == 0xff) { - goto drop; + return -EINVAL; } else if (len > sizeof(struct ip_esp_hdr) && udpdata32[0] != 0) { /* ESP Packet without Non-ESP header */ len = sizeof(struct udphdr); @@ -121,7 +116,7 @@ int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) case UDP_ENCAP_ESPINUDP_NON_IKE: /* Check if this is a keepalive packet. If so, eat it. */ if (len == 1 && udpdata[0] == 0xff) { - goto drop; + return -EINVAL; } else if (len > 2 * sizeof(u32) + sizeof(struct ip_esp_hdr) && udpdata32[0] == 0 && udpdata32[1] == 0) { @@ -139,7 +134,7 @@ int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) * protocol to ESP, and then call into the transform receiver. */ if (skb_unclone(skb, GFP_ATOMIC)) - goto drop; + return -EINVAL; /* Now we can update and verify the packet length... */ iph = ip_hdr(skb); @@ -147,24 +142,87 @@ int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) iph->tot_len = htons(ntohs(iph->tot_len) - len); if (skb->len < iphlen + len) { /* packet is too small!?! */ - goto drop; + return -EINVAL; } /* pull the data buffer up to the ESP header and set the * transport header to point to ESP. Keep UDP on the stack * for later. */ - __skb_pull(skb, len); - skb_reset_transport_header(skb); + if (pull) { + __skb_pull(skb, len); + skb_reset_transport_header(skb); + } else { + skb_set_transport_header(skb, len); + } /* process ESP */ - return xfrm4_rcv_encap(skb, IPPROTO_ESP, 0, encap_type); - -drop: - kfree_skb(skb); return 0; } +/* If it's a keepalive packet, then just eat it. + * If it's an encapsulated packet, then pass it to the + * IPsec xfrm input. + * Returns 0 if skb passed to xfrm or was dropped. + * Returns >0 if skb should be passed to UDP. + * Returns <0 if skb should be resubmitted (-ret is protocol) + */ +int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) +{ + int ret; + + ret = __xfrm4_udp_encap_rcv(sk, skb, true); + if (!ret) + return xfrm4_rcv_encap(skb, IPPROTO_ESP, 0, + udp_sk(sk)->encap_type); + + if (ret < 0) { + kfree_skb(skb); + return 0; + } + + return ret; +} + +struct sk_buff *xfrm4_gro_udp_encap_rcv(struct sock *sk, struct list_head *head, + struct sk_buff *skb) +{ + int offset = skb_gro_offset(skb); + const struct net_offload *ops; + struct sk_buff *pp = NULL; + int ret; + + offset = offset - sizeof(struct udphdr); + + if (!pskb_pull(skb, offset)) + return NULL; + + rcu_read_lock(); + ops = rcu_dereference(inet_offloads[IPPROTO_ESP]); + if (!ops || !ops->callbacks.gro_receive) + goto out; + + ret = __xfrm4_udp_encap_rcv(sk, skb, false); + if (ret) + goto out; + + skb_push(skb, offset); + NAPI_GRO_CB(skb)->proto = IPPROTO_UDP; + + pp = call_gro_receive(ops->callbacks.gro_receive, head, skb); + rcu_read_unlock(); + + return pp; + +out: + rcu_read_unlock(); + skb_push(skb, offset); + NAPI_GRO_CB(skb)->same_flow = 0; + NAPI_GRO_CB(skb)->flush = 1; + + return NULL; +} + int xfrm4_rcv(struct sk_buff *skb) { return xfrm4_rcv_spi(skb, ip_hdr(skb)->protocol, 0); From patchwork Wed Aug 16 13:13:21 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Antony Antony X-Patchwork-Id: 13355167 X-Patchwork-Delegate: kuba@kernel.org Received: from lindbergh.monkeyblade.net (lindbergh.monkeyblade.net [23.128.96.19]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 74C061078E for ; Wed, 16 Aug 2023 13:13:32 +0000 (UTC) Received: from a.mx.secunet.com (a.mx.secunet.com [62.96.220.36]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 479541BFB for ; Wed, 16 Aug 2023 06:13:30 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by a.mx.secunet.com (Postfix) with ESMTP id 1D6DB205DD; Wed, 16 Aug 2023 15:13:29 +0200 (CEST) X-Virus-Scanned: by secunet Received: from a.mx.secunet.com ([127.0.0.1]) by localhost (a.mx.secunet.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id E0GTUixYtvsN; Wed, 16 Aug 2023 15:13:28 +0200 (CEST) Received: from mailout2.secunet.com (mailout2.secunet.com [62.96.220.49]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by a.mx.secunet.com (Postfix) with ESMTPS id 29FFD205ED; Wed, 16 Aug 2023 15:13:28 +0200 (CEST) Received: from cas-essen-01.secunet.de (unknown [10.53.40.201]) by mailout2.secunet.com (Postfix) with ESMTP id 1AA4680004A; Wed, 16 Aug 2023 15:13:28 +0200 (CEST) Received: from mbx-essen-02.secunet.de (10.53.40.198) by cas-essen-01.secunet.de (10.53.40.201) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.27; Wed, 16 Aug 2023 15:13:27 +0200 Received: from moon.secunet.de (172.18.149.1) by mbx-essen-02.secunet.de (10.53.40.198) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.27; Wed, 16 Aug 2023 15:13:27 +0200 Date: Wed, 16 Aug 2023 15:13:21 +0200 From: Antony Antony To: Steffen Klassert , Herbert Xu CC: Eyal Birger , , , Antony Antony Subject: [PATCH v5 ipsec-next 3/3] xfrm: Support GRO for IPv6 ESP in UDP encapsulation Message-ID: Reply-To: References: <6dfd03c5fa0afb99f255f4a35772df19e33880db.1674156645.git.antony.antony@secunet.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: Precedence: first-class Priority: normal Organization: secunet X-ClientProxiedBy: cas-essen-02.secunet.de (10.53.40.202) To mbx-essen-02.secunet.de (10.53.40.198) X-EXCLAIMER-MD-CONFIG: 2c86f778-e09b-4440-8b15-867914633a10 X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_BLOCKED,SPF_HELO_NONE,SPF_PASS autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net X-Patchwork-Delegate: kuba@kernel.org From: Steffen Klassert This patch enables the GRO codepath for IPv6 ESP in UDP encapsulated packets. Decapsulation happens at L2 and saves a full round through the stack for each packet. This is also needed to support HW offload for ESP in UDP encapsulation. Signed-off-by: Steffen Klassert Co-developed-by: Antony Antony Signed-off-by: Antony Antony --- include/net/ipv6_stubs.h | 3 ++ include/net/xfrm.h | 2 + net/ipv4/udp.c | 2 + net/ipv6/af_inet6.c | 1 + net/ipv6/esp6_offload.c | 10 ++++- net/ipv6/xfrm6_input.c | 94 ++++++++++++++++++++++++++++++++-------- 6 files changed, 92 insertions(+), 20 deletions(-) diff --git a/include/net/ipv6_stubs.h b/include/net/ipv6_stubs.h index c48186bf4737..887d35f716c7 100644 --- a/include/net/ipv6_stubs.h +++ b/include/net/ipv6_stubs.h @@ -60,6 +60,9 @@ struct ipv6_stub { #if IS_ENABLED(CONFIG_XFRM) void (*xfrm6_local_rxpmtu)(struct sk_buff *skb, u32 mtu); int (*xfrm6_udp_encap_rcv)(struct sock *sk, struct sk_buff *skb); + struct sk_buff *(*xfrm6_gro_udp_encap_rcv)(struct sock *sk, + struct list_head *head, + struct sk_buff *skb); int (*xfrm6_rcv_encap)(struct sk_buff *skb, int nexthdr, __be32 spi, int encap_type); #endif diff --git a/include/net/xfrm.h b/include/net/xfrm.h index 3be5d11bb983..6b133be77e3b 100644 --- a/include/net/xfrm.h +++ b/include/net/xfrm.h @@ -1713,6 +1713,8 @@ int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb); int xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb); struct sk_buff *xfrm4_gro_udp_encap_rcv(struct sock *sk, struct list_head *head, struct sk_buff *skb); +struct sk_buff *xfrm6_gro_udp_encap_rcv(struct sock *sk, struct list_head *head, + struct sk_buff *skb); int xfrm_user_policy(struct sock *sk, int optname, sockptr_t optval, int optlen); #else diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index caf1aa733219..a4c5e4b00dc8 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -2688,6 +2688,8 @@ static void set_xfrm_gro_udp_encap_rcv(__u16 encap_type, unsigned short family, if (up->gro_enabled && encap_type == UDP_ENCAP_ESPINUDP) { if (family == AF_INET) up->gro_receive = xfrm4_gro_udp_encap_rcv; + else if (IS_ENABLED(CONFIG_IPV6) && family == AF_INET6) + up->gro_receive = ipv6_stub->xfrm6_gro_udp_encap_rcv; } #endif } diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c index 2bbf13216a3d..0ba95226e07d 100644 --- a/net/ipv6/af_inet6.c +++ b/net/ipv6/af_inet6.c @@ -1046,6 +1046,7 @@ static const struct ipv6_stub ipv6_stub_impl = { #if IS_ENABLED(CONFIG_XFRM) .xfrm6_local_rxpmtu = xfrm6_local_rxpmtu, .xfrm6_udp_encap_rcv = xfrm6_udp_encap_rcv, + .xfrm6_gro_udp_encap_rcv = xfrm6_gro_udp_encap_rcv, .xfrm6_rcv_encap = xfrm6_rcv_encap, #endif .nd_tbl = &nd_tbl, diff --git a/net/ipv6/esp6_offload.c b/net/ipv6/esp6_offload.c index ee5f5abdb503..a10d1e1cf544 100644 --- a/net/ipv6/esp6_offload.c +++ b/net/ipv6/esp6_offload.c @@ -33,7 +33,9 @@ static __u16 esp6_nexthdr_esp_offset(struct ipv6hdr *ipv6_hdr, int nhlen) int off = sizeof(struct ipv6hdr); struct ipv6_opt_hdr *exthdr; - if (likely(ipv6_hdr->nexthdr == NEXTHDR_ESP)) + /* ESP or ESPINUDP */ + if (likely(ipv6_hdr->nexthdr == NEXTHDR_ESP || + ipv6_hdr->nexthdr == NEXTHDR_UDP)) return offsetof(struct ipv6hdr, nexthdr); while (off < nhlen) { @@ -53,10 +55,14 @@ static struct sk_buff *esp6_gro_receive(struct list_head *head, int offset = skb_gro_offset(skb); struct xfrm_offload *xo; struct xfrm_state *x; + int encap_type = 0; __be32 seq; __be32 spi; int nhoff; + if (NAPI_GRO_CB(skb)->proto == IPPROTO_UDP) + encap_type = UDP_ENCAP_ESPINUDP; + if (!pskb_pull(skb, offset)) return NULL; @@ -103,7 +109,7 @@ static struct sk_buff *esp6_gro_receive(struct list_head *head, /* We don't need to handle errors from xfrm_input, it does all * the error handling and frees the resources on error. */ - xfrm_input(skb, IPPROTO_ESP, spi, 0); + xfrm_input(skb, IPPROTO_ESP, spi, encap_type); return ERR_PTR(-EINPROGRESS); out_reset: diff --git a/net/ipv6/xfrm6_input.c b/net/ipv6/xfrm6_input.c index 04cbeefd8982..b77ab24f707b 100644 --- a/net/ipv6/xfrm6_input.c +++ b/net/ipv6/xfrm6_input.c @@ -16,6 +16,8 @@ #include #include #include +#include +#include int xfrm6_rcv_spi(struct sk_buff *skb, int nexthdr, __be32 spi, struct ip6_tnl *t) @@ -67,14 +69,7 @@ int xfrm6_transport_finish(struct sk_buff *skb, int async) return 0; } -/* If it's a keepalive packet, then just eat it. - * If it's an encapsulated packet, then pass it to the - * IPsec xfrm input. - * Returns 0 if skb passed to xfrm or was dropped. - * Returns >0 if skb should be passed to UDP. - * Returns <0 if skb should be resubmitted (-ret is protocol) - */ -int xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) +static int __xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb, bool pull) { struct udp_sock *up = udp_sk(sk); struct udphdr *uh; @@ -106,7 +101,7 @@ int xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) case UDP_ENCAP_ESPINUDP: /* Check if this is a keepalive packet. If so, eat it. */ if (len == 1 && udpdata[0] == 0xff) { - goto drop; + return -EINVAL; } else if (len > sizeof(struct ip_esp_hdr) && udpdata32[0] != 0) { /* ESP Packet without Non-ESP header */ len = sizeof(struct udphdr); @@ -117,7 +112,7 @@ int xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) case UDP_ENCAP_ESPINUDP_NON_IKE: /* Check if this is a keepalive packet. If so, eat it. */ if (len == 1 && udpdata[0] == 0xff) { - goto drop; + return -EINVAL; } else if (len > 2 * sizeof(u32) + sizeof(struct ip_esp_hdr) && udpdata32[0] == 0 && udpdata32[1] == 0) { @@ -135,31 +130,94 @@ int xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) * protocol to ESP, and then call into the transform receiver. */ if (skb_unclone(skb, GFP_ATOMIC)) - goto drop; + return -EINVAL; /* Now we can update and verify the packet length... */ ip6h = ipv6_hdr(skb); ip6h->payload_len = htons(ntohs(ip6h->payload_len) - len); if (skb->len < ip6hlen + len) { /* packet is too small!?! */ - goto drop; + return -EINVAL; } /* pull the data buffer up to the ESP header and set the * transport header to point to ESP. Keep UDP on the stack * for later. */ - __skb_pull(skb, len); - skb_reset_transport_header(skb); + if (pull) { + __skb_pull(skb, len); + skb_reset_transport_header(skb); + } else { + skb_set_transport_header(skb, len); + } /* process ESP */ - return xfrm6_rcv_encap(skb, IPPROTO_ESP, 0, encap_type); - -drop: - kfree_skb(skb); return 0; } +/* If it's a keepalive packet, then just eat it. + * If it's an encapsulated packet, then pass it to the + * IPsec xfrm input. + * Returns 0 if skb passed to xfrm or was dropped. + * Returns >0 if skb should be passed to UDP. + * Returns <0 if skb should be resubmitted (-ret is protocol) + */ +int xfrm6_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) +{ + int ret; + + ret = __xfrm6_udp_encap_rcv(sk, skb, true); + if (!ret) + return xfrm6_rcv_encap(skb, IPPROTO_ESP, 0, + udp_sk(sk)->encap_type); + + if (ret < 0) { + kfree_skb(skb); + return 0; + } + + return ret; +} + +struct sk_buff *xfrm6_gro_udp_encap_rcv(struct sock *sk, struct list_head *head, + struct sk_buff *skb) +{ + int offset = skb_gro_offset(skb); + const struct net_offload *ops; + struct sk_buff *pp = NULL; + int ret; + + offset = offset - sizeof(struct udphdr); + + if (!pskb_pull(skb, offset)) + return NULL; + + rcu_read_lock(); + ops = rcu_dereference(inet6_offloads[IPPROTO_ESP]); + if (!ops || !ops->callbacks.gro_receive) + goto out; + + ret = __xfrm6_udp_encap_rcv(sk, skb, false); + if (ret) + goto out; + + skb_push(skb, offset); + NAPI_GRO_CB(skb)->proto = IPPROTO_UDP; + + pp = call_gro_receive(ops->callbacks.gro_receive, head, skb); + rcu_read_unlock(); + + return pp; + +out: + rcu_read_unlock(); + skb_push(skb, offset); + NAPI_GRO_CB(skb)->same_flow = 0; + NAPI_GRO_CB(skb)->flush = 1; + + return NULL; +} + int xfrm6_rcv_tnl(struct sk_buff *skb, struct ip6_tnl *t) { return xfrm6_rcv_spi(skb, skb_network_header(skb)[IP6CB(skb)->nhoff],