From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: passt.top; dmarc=none (p=none dis=none) header.from=gibson.dropbear.id.au Authentication-Results: passt.top; dkim=pass (2048-bit key; secure) header.d=gibson.dropbear.id.au header.i=@gibson.dropbear.id.au header.a=rsa-sha256 header.s=202408 header.b=shiolDVy; dkim-atps=neutral Received: from mail.ozlabs.org (mail.ozlabs.org [IPv6:2404:9400:2221:ea00::3]) by passt.top (Postfix) with ESMTPS id 1A3AC5A004C for ; Mon, 16 Sep 2024 03:57:30 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gibson.dropbear.id.au; s=202408; t=1726451838; bh=baLvafkGkhvMtqQI3rrcTTaKnc8+hSnlKk5QpSPXnIY=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=shiolDVykhUG/DCWr0Opu/aJFRWvJ9EgjxHZ5w0STwBpJT0El3YSKwkDTIbOrKaoY f9Wln8i1m06BP2ga1DJ0FSK3SG+TxAPdZnz+7kHlYklyS1c+pxd2KSpEKQJfzc4+jE MM8hLnsm7zz8Md0xeGX9H/ZTnNjT0Ez/uKiQIhWfheb9kJOgeoHJQ+WF3kXUnJmPFy eVdGaju6NixWTQ5E7P56TGHXcnw137CJxcBwnqzmLYqHISbD5n2OgXGjoFve//sXvq Ca6HXAoHj2gYK3PdmQgb2Cfx4g2D5kSy/DubiZYMHuBttSQkRj0T74JokXsu698e79 UCD1bp0/IVWag== Received: by gandalf.ozlabs.org (Postfix, from userid 1007) id 4X6Sht0yMbz4xCV; Mon, 16 Sep 2024 11:57:18 +1000 (AEST) Date: Mon, 16 Sep 2024 11:57:01 +1000 From: David Gibson To: Jon Maloy Subject: Re: [PATCH v2 2/2] tcp: unify l2 TCPv4 and TCPv6 queues and structures Message-ID: References: <20240914003718.2871567-1-jmaloy@redhat.com> <20240914003718.2871567-3-jmaloy@redhat.com> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="Bu0niC5kaA4Z1ZUV" Content-Disposition: inline In-Reply-To: <20240914003718.2871567-3-jmaloy@redhat.com> Message-ID-Hash: RDBZT6AJFVSP6O2V6GPV66TJRACQFZGW X-Message-ID-Hash: RDBZT6AJFVSP6O2V6GPV66TJRACQFZGW X-MailFrom: dgibson@gandalf.ozlabs.org X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; emergency; loop; banned-address; member-moderation; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header CC: passt-dev@passt.top, sbrivio@redhat.com, lvivier@redhat.com, dgibson@redhat.com X-Mailman-Version: 3.3.8 Precedence: list List-Id: Development discussion and patches for passt Archived-At: Archived-At: List-Archive: List-Archive: List-Help: List-Owner: List-Post: List-Subscribe: List-Unsubscribe: --Bu0niC5kaA4Z1ZUV Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Fri, Sep 13, 2024 at 08:37:18PM -0400, Jon Maloy wrote: > Following the preparations in the previous commit, we can now remove > the payload and flag queues dedicated for TCPv6 and TCPv4 and move all > traffic into common queues handling both protocol types. >=20 > Apart from reducing code and memory footprint, this change reduces > a potential risk for TCPv4 traffic starving out TCPv6 traffic. > Since we always flush out the TCPv4 frame queue before the TCPv6 queue, > the latter will never be handled if the former fails to send all its > frames. >=20 > Tests with iperf3 shows no measurable change in performance after this > change. >=20 > Signed-off-by: Jon Maloy Mostly LGTM. I've noted a handful of things, some of which are important, but they should be easy to fix. =20 > --- > v2: Merged previous patch #3, where the tcp4_ prefix was changed to > tcp_ into this patch. > --- > tcp.c | 6 +- > tcp_buf.c | 241 +++++++++++++++++------------------------------------- > tcp_buf.h | 3 +- > 3 files changed, 75 insertions(+), 175 deletions(-) >=20 > diff --git a/tcp.c b/tcp.c > index f9fe1b9..294eeee 100644 > --- a/tcp.c > +++ b/tcp.c > @@ -2504,11 +2504,7 @@ int tcp_init(struct ctx *c) > { > ASSERT(!c->no_tcp); > =20 > - if (c->ifi4) > - tcp_sock4_iov_init(c); > - > - if (c->ifi6) > - tcp_sock6_iov_init(c); > + tcp_sock_iov_init(c); > =20 > memset(init_sock_pool4, 0xff, sizeof(init_sock_pool4)); > memset(init_sock_pool6, 0xff, sizeof(init_sock_pool6)); > diff --git a/tcp_buf.c b/tcp_buf.c > index af80cc5..ab55476 100644 > --- a/tcp_buf.c > +++ b/tcp_buf.c > @@ -67,59 +67,42 @@ struct tcp_flags_t { > } __attribute__ ((packed, aligned(__alignof__(unsigned int)))); > #endif > =20 > -/* Ethernet header for IPv4 frames */ > +/* Ethernet header for IPv4 and IPv6 frames */ > static struct ethhdr tcp4_eth_src; > +static struct ethhdr tcp6_eth_src; > + > +static struct tap_hdr tcp_payload_tap_hdr[TCP_FRAMES_MEM]; > =20 > -static struct tap_hdr tcp4_payload_tap_hdr[TCP_FRAMES_MEM]; > -/* IPv4 headers */ > -static struct iphdr tcp4_payload_ip[TCP_FRAMES_MEM]; > -/* TCP segments with payload for IPv4 frames */ > -static struct tcp_payload_t tcp4_payload[TCP_FRAMES_MEM]; > +/* IP headers for IPv4 and IPv6 */ > +struct iphdr tcp4_payload_ip[TCP_FRAMES_MEM]; > +struct ipv6hdr tcp6_payload_ip[TCP_FRAMES_MEM]; > =20 > -static_assert(MSS4 <=3D sizeof(tcp4_payload[0].data), "MSS4 is greater t= han 65516"); > +/* TCP segments with payload for IPv4 and IPv6 frames */ > +static struct tcp_payload_t tcp_payload[TCP_FRAMES_MEM]; > + > +static_assert(sizeof(tcp_payload[0].data) > MSS4, "MSS4 is greater than = 65516"); I'm not sure why you're reversing the order of the inequality here, seems like an unrelated change. If you do reverse it it should be >=3D not >. > /* References tracking the owner connection of frames in the tap outqueu= e */ > -static struct tcp_tap_conn *tcp4_frame_conns[TCP_FRAMES_MEM]; > -static unsigned int tcp4_payload_used; > +static struct tcp_tap_conn *tcp_frame_conns[TCP_FRAMES_MEM]; > +static unsigned int tcp_payload_used; > =20 > -static struct tap_hdr tcp4_flags_tap_hdr[TCP_FRAMES_MEM]; > +static struct tap_hdr tcp_flags_tap_hdr[TCP_FRAMES_MEM]; > /* IPv4 headers for TCP segment without payload */ > static struct iphdr tcp4_flags_ip[TCP_FRAMES_MEM]; > /* TCP segments without payload for IPv4 frames */ > -static struct tcp_flags_t tcp4_flags[TCP_FRAMES_MEM]; > - > -static unsigned int tcp4_flags_used; > +static struct tcp_flags_t tcp_flags[TCP_FRAMES_MEM]; > =20 > -/* Ethernet header for IPv6 frames */ > -static struct ethhdr tcp6_eth_src; > - > -static struct tap_hdr tcp6_payload_tap_hdr[TCP_FRAMES_MEM]; > -/* IPv6 headers */ > -static struct ipv6hdr tcp6_payload_ip[TCP_FRAMES_MEM]; > -/* TCP headers and data for IPv6 frames */ > -static struct tcp_payload_t tcp6_payload[TCP_FRAMES_MEM]; > +static unsigned int tcp_flags_used; > =20 > -static_assert(MSS6 <=3D sizeof(tcp6_payload[0].data), "MSS6 is greater t= han 65516"); AFAICT you've dropped this assert entirely, but it's still relevant. With the payload buffers unified, we still want to make sure each buffer is large enough for both the largest IPv4 and the largest IPv6 frame. This is, in a sense, redundant with the IPv4 check, since MSS4 will be bigger than MSS6, but I'd prefer to explcitly verify it rather than rely on that here. > -/* References tracking the owner connection of frames in the tap outqueu= e */ > -static struct tcp_tap_conn *tcp6_frame_conns[TCP_FRAMES_MEM]; > -static unsigned int tcp6_payload_used; > - > -static struct tap_hdr tcp6_flags_tap_hdr[TCP_FRAMES_MEM]; > /* IPv6 headers for TCP segment without payload */ > static struct ipv6hdr tcp6_flags_ip[TCP_FRAMES_MEM]; > -/* TCP segment without payload for IPv6 frames */ > -static struct tcp_flags_t tcp6_flags[TCP_FRAMES_MEM]; > - > -static unsigned int tcp6_flags_used; > =20 > /* recvmsg()/sendmsg() data for tap */ > static struct iovec iov_sock [TCP_FRAMES_MEM + 1]; > =20 > -static struct iovec tcp4_l2_iov [TCP_FRAMES_MEM][TCP_NUM_IOVS]; > -static struct iovec tcp6_l2_iov [TCP_FRAMES_MEM][TCP_NUM_IOVS]; > -static struct iovec tcp4_l2_flags_iov [TCP_FRAMES_MEM][TCP_NUM_IOVS]; > -static struct iovec tcp6_l2_flags_iov [TCP_FRAMES_MEM][TCP_NUM_IOVS]; > +static struct iovec tcp_l2_iov[TCP_FRAMES_MEM][TCP_NUM_IOVS]; > +static struct iovec tcp_l2_flags_iov[TCP_FRAMES_MEM][TCP_NUM_IOVS]; > + > /** > * tcp_update_l2_buf() - Update Ethernet header buffers with addresses > * @eth_d: Ethernet destination address, NULL if unchanged > @@ -132,87 +115,47 @@ void tcp_update_l2_buf(const unsigned char *eth_d, = const unsigned char *eth_s) > } > =20 > /** > - * tcp_sock4_iov_init() - Initialise scatter-gather L2 buffers for IPv4 = sockets > + * tcp_sock_iov_init() - Initialise scatter-gather L2 buffers for IPv4 s= ockets > * @c: Execution context > */ > -void tcp_sock4_iov_init(const struct ctx *c) > +void tcp_sock_iov_init(const struct ctx *c) > { > + struct ipv6hdr ip6 =3D L2_BUF_IP6_INIT(IPPROTO_TCP); > struct iphdr iph =3D L2_BUF_IP4_INIT(IPPROTO_TCP); > struct iovec *iov; > int i; > =20 > + tcp6_eth_src.h_proto =3D htons_constant(ETH_P_IPV6); > tcp4_eth_src.h_proto =3D htons_constant(ETH_P_IP); Aside: I guess we could make this a static initialiser, but that's not the job of this patch. > =20 > - for (i =3D 0; i < ARRAY_SIZE(tcp4_payload); i++) { > + for (i =3D 0; i < ARRAY_SIZE(tcp_payload); i++) { > + tcp6_payload_ip[i] =3D ip6; > tcp4_payload_ip[i] =3D iph; > - tcp4_payload[i].th.doff =3D sizeof(struct tcphdr) / 4; > - tcp4_payload[i].th.ack =3D 1; > + tcp_payload[i].th.doff =3D sizeof(struct tcphdr) / 4; > + tcp_payload[i].th.ack =3D 1; > } > =20 > - for (i =3D 0; i < ARRAY_SIZE(tcp4_flags); i++) { > + for (i =3D 0; i < ARRAY_SIZE(tcp_flags); i++) { > + tcp6_flags_ip[i] =3D ip6; > tcp4_flags_ip[i] =3D iph; > - tcp4_flags[i].th.doff =3D sizeof(struct tcphdr) / 4; > - tcp4_flags[i].th.ack =3D 1; > + tcp_flags[i].th.doff =3D sizeof(struct tcphdr) / 4; > + tcp_flags[i].th.ack =3D 1; > } > =20 > for (i =3D 0; i < TCP_FRAMES_MEM; i++) { > - iov =3D tcp4_l2_iov[i]; > + iov =3D tcp_l2_iov[i]; Aside: and it would probably make sense to have iov local to these blocks. > =20 > - iov[TCP_IOV_TAP] =3D tap_hdr_iov(c, &tcp4_payload_tap_hdr[i]); > + iov[TCP_IOV_TAP] =3D tap_hdr_iov(c, &tcp_payload_tap_hdr[i]); > iov[TCP_IOV_ETH].iov_len =3D sizeof(struct ethhdr); > - iov[TCP_IOV_PAYLOAD].iov_base =3D &tcp4_payload[i]; > + iov[TCP_IOV_PAYLOAD].iov_base =3D &tcp_payload[i]; > } > =20 > for (i =3D 0; i < TCP_FRAMES_MEM; i++) { > - iov =3D tcp4_l2_flags_iov[i]; > - > - iov[TCP_IOV_TAP] =3D tap_hdr_iov(c, &tcp4_flags_tap_hdr[i]); > - iov[TCP_IOV_ETH].iov_base =3D &tcp4_eth_src; > - iov[TCP_IOV_ETH] =3D IOV_OF_LVALUE(tcp4_eth_src); > - iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp4_flags_ip[i]); > - iov[TCP_IOV_PAYLOAD].iov_base =3D &tcp4_flags[i]; > - } > -} > - > -/** > - * tcp_sock6_iov_init() - Initialise scatter-gather L2 buffers for IPv6 = sockets > - * @c: Execution context > - */ > -void tcp_sock6_iov_init(const struct ctx *c) > -{ > - struct ipv6hdr ip6 =3D L2_BUF_IP6_INIT(IPPROTO_TCP); > - struct iovec *iov; > - int i; > - > - tcp6_eth_src.h_proto =3D htons_constant(ETH_P_IPV6); > + iov =3D tcp_l2_flags_iov[i]; > =20 > - for (i =3D 0; i < ARRAY_SIZE(tcp6_payload); i++) { > - tcp6_payload_ip[i] =3D ip6; > - tcp6_payload[i].th.doff =3D sizeof(struct tcphdr) / 4; > - tcp6_payload[i].th.ack =3D 1; > - } > - > - for (i =3D 0; i < ARRAY_SIZE(tcp6_flags); i++) { > - tcp6_flags_ip[i] =3D ip6; > - tcp6_flags[i].th.doff =3D sizeof(struct tcphdr) / 4; > - tcp6_flags[i].th .ack =3D 1; > - } > - > - for (i =3D 0; i < TCP_FRAMES_MEM; i++) { > - iov =3D tcp6_l2_iov[i]; > - > - iov[TCP_IOV_TAP] =3D tap_hdr_iov(c, &tcp6_payload_tap_hdr[i]); > + iov[TCP_IOV_TAP] =3D tap_hdr_iov(c, &tcp_flags_tap_hdr[i]); > iov[TCP_IOV_ETH].iov_len =3D sizeof(struct ethhdr); > - iov[TCP_IOV_PAYLOAD].iov_base =3D &tcp6_payload[i]; > - } > - > - for (i =3D 0; i < TCP_FRAMES_MEM; i++) { > - iov =3D tcp6_l2_flags_iov[i]; > - > - iov[TCP_IOV_TAP] =3D tap_hdr_iov(c, &tcp6_flags_tap_hdr[i]); > - iov[TCP_IOV_ETH] =3D IOV_OF_LVALUE(tcp6_eth_src); > - iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp6_flags_ip[i]); > - iov[TCP_IOV_PAYLOAD].iov_base =3D &tcp6_flags[i]; > + iov[TCP_IOV_PAYLOAD].iov_base =3D &tcp_flags[i]; > } > } > =20 > @@ -222,13 +165,9 @@ void tcp_sock6_iov_init(const struct ctx *c) > */ > void tcp_flags_flush(const struct ctx *c) > { > - tap_send_frames(c, &tcp6_l2_flags_iov[0][0], TCP_NUM_IOVS, > - tcp6_flags_used); > - tcp6_flags_used =3D 0; > - > - tap_send_frames(c, &tcp4_l2_flags_iov[0][0], TCP_NUM_IOVS, > - tcp4_flags_used); > - tcp4_flags_used =3D 0; > + tap_send_frames(c, &tcp_l2_flags_iov[0][0], TCP_NUM_IOVS, > + tcp_flags_used); > + tcp_flags_used =3D 0; > } > =20 > /** > @@ -267,21 +206,13 @@ void tcp_payload_flush(struct ctx *c) > { > size_t m; > =20 > - m =3D tap_send_frames(c, &tcp6_l2_iov[0][0], TCP_NUM_IOVS, > - tcp6_payload_used); > - if (m !=3D tcp6_payload_used) { > - tcp_revert_seq(c, &tcp6_frame_conns[m], &tcp6_l2_iov[m], > - tcp6_payload_used - m); > - } > - tcp6_payload_used =3D 0; > - > - m =3D tap_send_frames(c, &tcp4_l2_iov[0][0], TCP_NUM_IOVS, > - tcp4_payload_used); > - if (m !=3D tcp4_payload_used) { > - tcp_revert_seq(c, &tcp4_frame_conns[m], &tcp4_l2_iov[m], > - tcp4_payload_used - m); > + m =3D tap_send_frames(c, &tcp_l2_iov[0][0], TCP_NUM_IOVS, > + tcp_payload_used); > + if (m !=3D tcp_payload_used) { > + tcp_revert_seq(c, &tcp_frame_conns[m], &tcp_l2_iov[m], > + tcp_payload_used - m); > } > - tcp4_payload_used =3D 0; > + tcp_payload_used =3D 0; > } > =20 > /** > @@ -301,28 +232,23 @@ int tcp_buf_send_flag(struct ctx *c, struct tcp_tap= _conn *conn, int flags) > uint32_t seq; > int ret; > =20 > + iov =3D tcp_l2_flags_iov[tcp_flags_used]; > if (CONN_V4(conn)) { > - iov =3D tcp4_l2_flags_iov[tcp4_flags_used]; > - iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp4_flags_ip[tcp4_flags_used++]); > + iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp4_flags_ip[tcp_flags_used]); > iov[TCP_IOV_ETH].iov_base =3D &tcp4_eth_src; > } else { > - iov =3D tcp6_l2_flags_iov[tcp6_flags_used]; > - iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp6_flags_ip[tcp6_flags_used++]); > + iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp6_flags_ip[tcp_flags_used]); > iov[TCP_IOV_ETH].iov_base =3D &tcp6_eth_src; > } > - payload =3D iov[TCP_IOV_PAYLOAD].iov_base; > =20 > + payload =3D iov[TCP_IOV_PAYLOAD].iov_base; > seq =3D conn->seq_to_tap; > ret =3D tcp_prepare_flags(c, conn, flags, &payload->th, > payload->opts, &optlen); > - if (ret <=3D 0) { > - if (CONN_V4(conn)) > - tcp4_flags_used--; > - else > - tcp6_flags_used--; > + if (ret <=3D 0) > return ret; > - } > =20 > + tcp_flags_used++; > l4len =3D tcp_l2_buf_fill_headers(conn, iov, optlen, NULL, seq); > iov[TCP_IOV_PAYLOAD].iov_len =3D l4len; > =20 > @@ -330,28 +256,22 @@ int tcp_buf_send_flag(struct ctx *c, struct tcp_tap= _conn *conn, int flags) > struct iovec *dup_iov; > int i; > =20 > + dup_iov =3D tcp_l2_flags_iov[tcp_flags_used]; > if (CONN_V4(conn)) { > - dup_iov =3D tcp4_l2_flags_iov[tcp4_flags_used]; > - dup_iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp4_flags_ip[tcp4_flags_used++= ]); > + dup_iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp4_flags_ip[tcp_flags_used]); > dup_iov[TCP_IOV_ETH].iov_base =3D &tcp4_eth_src; > } else { > - dup_iov =3D tcp4_l2_flags_iov[tcp6_flags_used]; > - dup_iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp6_flags_ip[tcp6_flags_used++= ]); > + dup_iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp6_flags_ip[tcp_flags_used]); > dup_iov[TCP_IOV_ETH].iov_base =3D &tcp6_eth_src; > } > for (i =3D 0; i < TCP_NUM_IOVS; i++) > - memcpy(dup_iov[i].iov_base, iov[i].iov_base, > - iov[i].iov_len); > + memcpy(dup_iov[i].iov_base, iov[i].iov_base, iov[i].iov_len); This will conflict with a fix I have in the pipeline (we should exclude TCP_IOV_ETH here, or we'll do an overlapping memcpy()). That said.. since we're updating TCP_IOV_IP on the fly each time now, we could just point it at the previous frame's IP buffer, rather than memcpy()ing the whole header. That would mean we only need to copy the TCP_IOV_TAP and TCP_IOV_PAYLOAD buffers. > dup_iov[TCP_IOV_PAYLOAD].iov_len =3D iov[TCP_IOV_PAYLOAD].iov_len; > + tcp_flags_used++; > } > =20 > - if (CONN_V4(conn)) { > - if (tcp4_flags_used > TCP_FRAMES_MEM - 2) > - tcp_flags_flush(c); > - } else { > - if (tcp6_flags_used > TCP_FRAMES_MEM - 2) > - tcp_flags_flush(c); > - } > + if (tcp_flags_used > TCP_FRAMES_MEM - 2) > + tcp_flags_flush(c); > =20 > return 0; > } > @@ -367,40 +287,29 @@ int tcp_buf_send_flag(struct ctx *c, struct tcp_tap= _conn *conn, int flags) > static void tcp_data_to_tap(struct ctx *c, struct tcp_tap_conn *conn, > ssize_t dlen, int no_csum, uint32_t seq) > { > + const uint16_t *check =3D NULL; > struct iovec *iov; > size_t l4len; > =20 > conn->seq_to_tap =3D seq + dlen; > - > + tcp_frame_conns[tcp_payload_used] =3D conn; > + iov =3D tcp_l2_iov[tcp_payload_used]; > if (CONN_V4(conn)) { > - struct iovec *iov_prev =3D tcp4_l2_iov[tcp4_payload_used - 1]; > - const uint16_t *check =3D NULL; > - > if (no_csum) { > + struct iovec *iov_prev =3D tcp_l2_iov[tcp_payload_used - 1]; > struct iphdr *iph =3D iov_prev[TCP_IOV_IP].iov_base; > check =3D &iph->check; > } > - > - tcp4_frame_conns[tcp4_payload_used] =3D conn; > - > - iov =3D tcp4_l2_iov[tcp4_payload_used]; > - iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp4_payload_ip[tcp4_payload_used++]= ); > + iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp4_payload_ip[tcp_payload_used]); > iov[TCP_IOV_ETH].iov_base =3D &tcp4_eth_src; > - l4len =3D tcp_l2_buf_fill_headers(conn, iov, dlen, check, seq); > - iov[TCP_IOV_PAYLOAD].iov_len =3D l4len; > - if (tcp4_payload_used > TCP_FRAMES_MEM - 1) > - tcp_payload_flush(c); > } else if (CONN_V6(conn)) { > - tcp6_frame_conns[tcp6_payload_used] =3D conn; > - > - iov =3D tcp6_l2_iov[tcp6_payload_used]; > - iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp6_payload_ip[tcp6_payload_used++]= ); > + iov[TCP_IOV_IP] =3D IOV_OF_LVALUE(tcp6_payload_ip[tcp_payload_used]); > iov[TCP_IOV_ETH].iov_base =3D &tcp6_eth_src; > - l4len =3D tcp_l2_buf_fill_headers(conn, iov, dlen, NULL, seq); > - iov[TCP_IOV_PAYLOAD].iov_len =3D l4len; > - if (tcp6_payload_used > TCP_FRAMES_MEM - 1) > - tcp_payload_flush(c); > } > + l4len =3D tcp_l2_buf_fill_headers(conn, iov, dlen, check, seq); > + iov[TCP_IOV_PAYLOAD].iov_len =3D l4len; > + if (++tcp_payload_used > TCP_FRAMES_MEM - 1) > + tcp_payload_flush(c); > } > =20 > /** > @@ -464,19 +373,15 @@ int tcp_buf_data_from_sock(struct ctx *c, struct tc= p_tap_conn *conn) > mh_sock.msg_iovlen =3D fill_bufs; > } > =20 > - if (( v4 && tcp4_payload_used + fill_bufs > TCP_FRAMES_MEM) || > - (!v4 && tcp6_payload_used + fill_bufs > TCP_FRAMES_MEM)) { > + if ((v4 && tcp_payload_used + fill_bufs > TCP_FRAMES_MEM)) { I don't think you want the 'v4' here any more. > tcp_payload_flush(c); > =20 > /* Silence Coverity CWE-125 false positive */ > - tcp4_payload_used =3D tcp6_payload_used =3D 0; > + tcp_payload_used =3D 0; > } > =20 > for (i =3D 0, iov =3D iov_sock + 1; i < fill_bufs; i++, iov++) { > - if (v4) > - iov->iov_base =3D &tcp4_payload[tcp4_payload_used + i].data; > - else > - iov->iov_base =3D &tcp6_payload[tcp6_payload_used + i].data; > + iov->iov_base =3D &tcp_payload[tcp_payload_used + i].data; > iov->iov_len =3D mss; > } > if (iov_rem) > @@ -524,7 +429,7 @@ int tcp_buf_data_from_sock(struct ctx *c, struct tcp_= tap_conn *conn) > dlen =3D mss; > seq =3D conn->seq_to_tap; > for (i =3D 0; i < send_bufs; i++) { > - int no_csum =3D i && i !=3D send_bufs - 1 && tcp4_payload_used; > + int no_csum =3D i && i !=3D send_bufs - 1 && tcp_payload_used; > =20 > if (i =3D=3D send_bufs - 1) > dlen =3D last_len; > diff --git a/tcp_buf.h b/tcp_buf.h > index 3db4c56..03a2273 100644 > --- a/tcp_buf.h > +++ b/tcp_buf.h > @@ -6,8 +6,7 @@ > #ifndef TCP_BUF_H > #define TCP_BUF_H > =20 > -void tcp_sock4_iov_init(const struct ctx *c); > -void tcp_sock6_iov_init(const struct ctx *c); > +void tcp_sock_iov_init(const struct ctx *c); > void tcp_flags_flush(const struct ctx *c); > void tcp_payload_flush(struct ctx *c); > int tcp_buf_data_from_sock(struct ctx *c, struct tcp_tap_conn *conn); --=20 David Gibson (he or they) | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you, not the other way | around. http://www.ozlabs.org/~dgibson --Bu0niC5kaA4Z1ZUV Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- iQIzBAEBCAAdFiEEO+dNsU4E3yXUXRK2zQJF27ox2GcFAmbnkGwACgkQzQJF27ox 2GdX1w//bIAFr8lQEuk1jLFZgX+fzduKk24Fhx720JOoDXaCkfmHVUyOyrVFwngA QmutpVZQIMQk1IcEwVn7hSlgk5UeKUzoLIOACufTq50gOdCgTOB0SxIOSQa7VDdn dX5HArivAfiqHLfkkm5raugTi7YXbvnEgD9pmQO1WO/IxTnbvOvd4w9W/x7qgkSC JfySgU1x+xWD1uW0VPtFTVAwYiyApgV/hi6xmp4EBRbQ3RX0cVnZ7sQ3gtGbUBEe OboLxcEAWoMAvKUaWgLEWOGspt2vgq6W85z4gZ8CVs4mIP1F9/o+nozEBNsSXLkl cPoqwKo518EqtHPu+J/JdLzjUK9ZzClQXzRyHo+YFTvdiqTB3JqJPXRulCNfAmoE +H8+dsOUyz6WL83qCNzg0RR3C7BNuwYEpSBtUPG5EiZTV9YUMjSe4eXV6XYBpOXz wfilQodUJk/OI111lGy+AKgn7BezQd7VtBeP1aDoG/uvXk1QlnpPEmIeVblsIa+S cmhPGX7svwlOSkyBfQmIhvRY3HBzQZaw2gyiYxdfHPs/QICSED+jUklKrbD00d88 barmg15vA6g6+DWJX9OshQHCJojpOPuV772ppSfM7DJwKXw+142RW2w4ubsq7o6t 65CcxF52zOPLLjLpi97wLOO8jvG7B7N6Ep7c73ejIqnwPkzc5Go= =6SSy -----END PGP SIGNATURE----- --Bu0niC5kaA4Z1ZUV--