Skip to contentSkip to conversation

sctp: Add GSO support

PATCH In review Marcelo Ricardo Leitner 2 participants 9 messages Jun 2, 2016 – Jun 3, 2016

Subsystems (4)
Networking Network drivers IPv4/IPv6 SCTP protocol
lore
Share & tools
lore

Marcelo Ricardo Leitner

author maintainer
#0 lore raw

v3 cover sctp: Add GSO support

This patchset adds sctp GSO support.

Performance tests indicates that increases throughput by 10% if using bigger chunk sizes, specially if bigger than MTU. For small chunks, it doesn't help much if not using heavy firewall rules.

For small chunks it will probably be of more use once we get something like MSG_MORE as David Laight had suggested.

overall changes:
v1->v2:
Added support for receiving GSO frames on SCTP stack, as requested by Dave Miller.

v2->v3:
Consider sctphdr size in skb_gso_transport_seglen() rebased due to 5c7cdf339af5 ("gso: Remove arbitrary checks for unsupported GSO")

Marcelo Ricardo Leitner (7):
loopback: make use of NETIF_F_GSO_SOFTWARE
skbuff: export skb_gro_receive
sk_buff: allow segmenting based on frag sizes
skbuff: introduce skb_gso_validate_mtu
sctp: delay as much as possible skb_linearize
sctp: Add GSO support
sctp: improve debug message to also log curr pkt and new chunk size

 drivers/net/loopback.c          |   5 +-
 include/linux/netdev_features.h |   7 +-
 include/linux/netdevice.h       |   1 +
 include/linux/skbuff.h          |   8 +
 include/net/sctp/sctp.h         |   4 +
 include/net/sctp/structs.h      |   5 +
 net/core/ethtool.c              |   1 +
 net/core/skbuff.c               |  45 ++++-
 net/ipv4/ip_forward.c           |   2 +-
 net/ipv4/ip_output.c            |   2 +-
 net/ipv6/ip6_output.c           |   2 +-
 net/mpls/af_mpls.c              |   2 +-
 net/sctp/Makefile               |   3 +-
 net/sctp/input.c                |  57 ++++---
 net/sctp/inqueue.c              |  78 +++++++--
 net/sctp/offload.c              |  98 +++++++++++
 net/sctp/output.c               | 366 +++++++++++++++++++++++++++-------------
 net/sctp/protocol.c             |   3 +
 net/sctp/socket.c               |   2 +
 19 files changed, 524 insertions(+), 167 deletions(-)
 create mode 100644 net/sctp/offload.c
Signature · 1 line
2.5.5
author maintainer
#1 lore raw

PATCH v3 1/7 loopback: make use of NETIF_F_GSO_SOFTWARE

NETIF_F_GSO_SOFTWARE was defined to list all GSO software types, so lets make use of it in loopback code. Note that veth/vxlan/others already uses it.

Within this patch series, this patch causes lo to pick up SCTP GSO feature automatically (as it's added to NETIF_F_GSO_SOFTWARE) and thus avoiding segmentation if possible.

Signed-off-by Marcelo Ricardo Leitner <[email protected]>
Tested-by Xin Long <[email protected]>
---
 drivers/net/loopback.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/net/loopback.c +2 -3
index a400288cb37b9bfb6190f1bd7c64d02e97713956..6255973e3dda35fd41464ce51f0f9fb9f0b8364b 100644
@@ -169,10 +169,9 @@ static void loopback_setup(struct net_device *dev)
 	dev->flags		= IFF_LOOPBACK;
 	dev->priv_flags		|= IFF_LIVE_ADDR_CHANGE | IFF_NO_QUEUE;
 	netif_keep_dst(dev);
-	dev->hw_features	= NETIF_F_ALL_TSO | NETIF_F_UFO;
+	dev->hw_features	= NETIF_F_GSO_SOFTWARE;
 	dev->features 		= NETIF_F_SG | NETIF_F_FRAGLIST
-		| NETIF_F_ALL_TSO
-		| NETIF_F_UFO
+		| NETIF_F_GSO_SOFTWARE
 		| NETIF_F_HW_CSUM
 		| NETIF_F_RXCSUM
 		| NETIF_F_SCTP_CRC
Signature · 1 line
2.5.5
author maintainer
#2 lore raw

PATCH v3 2/7 skbuff: export skb_gro_receive

In reply to Marcelo Ricardo Leitner

sctp GSO requires it and sctp can be compiled as a module, so we need to export this function.

Signed-off-by Marcelo Ricardo Leitner <[email protected]>
Tested-by Xin Long <[email protected]>
---
 net/core/skbuff.c | 1 +
 1 file changed, 1 insertion(+)
net/core/skbuff.c +1 -0
index f2b77e549c03a771909cd9c87c40ec2b7826cd31..4724bcf9b0cae1cecbe5bc2c04e308bb70b3232a 100644
@@ -3438,6 +3438,7 @@ done:
 	NAPI_GRO_CB(skb)->same_flow = 1;
 	return 0;
 }
+EXPORT_SYMBOL_GPL(skb_gro_receive);
 
 void __init skb_init(void)
 {
Signature · 1 line
2.5.5
author maintainer
#3 lore raw

PATCH v3 3/7 sk_buff: allow segmenting based on frag sizes

In reply to Marcelo Ricardo Leitner

This patch allows segmenting a skb based on its frags sizes instead of based on a fixed value.

Signed-off-by Marcelo Ricardo Leitner <[email protected]>
Tested-by Xin Long <[email protected]>
---
 include/linux/skbuff.h |  5 +++++
 net/core/skbuff.c      | 10 +++++++---
 2 files changed, 12 insertions(+), 3 deletions(-)
include/linux/skbuff.h +5 -0
index ee38a41274759f279be1c0752a7fab63fac517c8..329a0a9ef67115cae03b7c1304de031116384148 100644
@@ -301,6 +301,11 @@ struct sk_buff;
 #endif
 extern int sysctl_max_skb_frags;
 
+/* Set skb_shinfo(skb)->gso_size to this in case you want skb_segment to
+ * segment using its current segmentation instead.
+ */
+#define GSO_BY_FRAGS	0xFFFF
+
 typedef struct skb_frag_struct skb_frag_t;
 
 struct skb_frag_struct {
net/core/skbuff.c +7 -3
index 4724bcf9b0cae1cecbe5bc2c04e308bb70b3232a..97c32c75e704af1f31b064e8f1e0475ff1505d67 100644
@@ -3116,9 +3116,13 @@ struct sk_buff *skb_segment(struct sk_buff *head_skb,
 		int hsize;
 		int size;
 
-		len = head_skb->len - offset;
-		if (len > mss)
-			len = mss;
+		if (unlikely(mss = GSO_BY_FRAGS)) {
+			len = list_skb->len;
+		} else {
+			len = head_skb->len - offset;
+			if (len > mss)
+				len = mss;
+		}
 
 		hsize = skb_headlen(head_skb) - offset;
 		if (hsize < 0)
Signature · 1 line
2.5.5
author maintainer
#4 lore raw

PATCH v3 4/7 skbuff: introduce skb_gso_validate_mtu

In reply to Marcelo Ricardo Leitner

skb_gso_network_seglen is not enough for checking fragment sizes if skb is using GSO_BY_FRAGS as we have to check frag per frag.

This patch introduces skb_gso_validate_mtu, based on the former, which will wrap the use case inside it as all calls to skb_gso_network_seglen were to validate if it fits on a given TMU, and improve the check.

Signed-off-by Marcelo Ricardo Leitner <[email protected]>
Tested-by Xin Long <[email protected]>
---
 include/linux/skbuff.h |  1 +
 net/core/skbuff.c      | 31 +++++++++++++++++++++++++++++++
 net/ipv4/ip_forward.c  |  2 +-
 net/ipv4/ip_output.c   |  2 +-
 net/ipv6/ip6_output.c  |  2 +-
 net/mpls/af_mpls.c     |  2 +-
 6 files changed, 36 insertions(+), 4 deletions(-)
include/linux/skbuff.h +1 -0
index 329a0a9ef67115cae03b7c1304de031116384148..aa3f9d7e8d5ca455387efa22d4a9d3a079a56f0c 100644
@@ -2992,6 +2992,7 @@ void skb_split(struct sk_buff *skb, struct sk_buff *skb1, const u32 len);
 int skb_shift(struct sk_buff *tgt, struct sk_buff *skb, int shiftlen);
 void skb_scrub_packet(struct sk_buff *skb, bool xnet);
 unsigned int skb_gso_transport_seglen(const struct sk_buff *skb);
+bool skb_gso_validate_mtu(const struct sk_buff *skb, unsigned int mtu);
 struct sk_buff *skb_segment(struct sk_buff *skb, netdev_features_t features);
 struct sk_buff *skb_vlan_untag(struct sk_buff *skb);
 int skb_ensure_writable(struct sk_buff *skb, int write_len);
net/core/skbuff.c +31 -0
index 97c32c75e704af1f31b064e8f1e0475ff1505d67..5ca562b56ec39d39e1225d96547e242732518ffe 100644
@@ -4392,6 +4392,37 @@ unsigned int skb_gso_transport_seglen(const struct sk_buff *skb)
 }
 EXPORT_SYMBOL_GPL(skb_gso_transport_seglen);
 
+/**
+ * skb_gso_validate_mtu - Return in case such skb fits a given MTU
+ *
+ * @skb: GSO skb
+ *
+ * skb_gso_validate_mtu validates if a given skb will fit a wanted MTU
+ * once split.
+ */
+bool skb_gso_validate_mtu(const struct sk_buff *skb, unsigned int mtu)
+{
+	const struct skb_shared_info *shinfo = skb_shinfo(skb);
+	const struct sk_buff *iter;
+	unsigned int hlen;
+
+	hlen = skb_gso_network_seglen(skb);
+
+	if (shinfo->gso_size != GSO_BY_FRAGS)
+		return hlen <= mtu;
+
+	/* Undo this so we can re-use header sizes */
+	hlen -= GSO_BY_FRAGS;
+
+	skb_walk_frags(skb, iter) {
+		if (hlen + skb_headlen(iter) > mtu)
+			return false;
+	}
+
+	return true;
+}
+EXPORT_SYMBOL_GPL(skb_gso_validate_mtu);
+
 static struct sk_buff *skb_reorder_vlan_header(struct sk_buff *skb)
 {
 	if (skb_cow(skb, skb_headroom(skb)) < 0) {
net/ipv4/ip_forward.c +1 -1
index cbfb1808fcc490b94dc0bbdab6142acb8fa37815..9f0a7b96646f368021d9cd51bc3f728ba49eed0d 100644
@@ -54,7 +54,7 @@ static bool ip_exceeds_mtu(const struct sk_buff *skb, unsigned int mtu)
 	if (skb->ignore_df)
 		return false;
 
-	if (skb_is_gso(skb) && skb_gso_network_seglen(skb) <= mtu)
+	if (skb_is_gso(skb) && skb_gso_validate_mtu(skb, mtu))
 		return false;
 
 	return true;
net/ipv4/ip_output.c +1 -1
index 124bf0a663283502deb03397343160d493a378b1..cbac493c913ac37b57a97314f9e7099b14b8246c 100644
@@ -225,7 +225,7 @@ static int ip_finish_output_gso(struct net *net, struct sock *sk,
 
 	/* common case: locally created skb or seglen is <= mtu */
 	if (((IPCB(skb)->flags & IPSKB_FORWARDED) = 0) ||
-	      skb_gso_network_seglen(skb) <= mtu)
+	      skb_gso_validate_mtu(skb, mtu))
 		return ip_finish_output2(net, sk, skb);
 
 	/* Slowpath -  GSO segment length is exceeding the dst MTU.
net/ipv6/ip6_output.c +1 -1
index cbf127ae7c676650cc626cbf12cd61b6b570ea43..6b2f60a5c1de3063bb65c07b2b77c13f33890af8 100644
@@ -368,7 +368,7 @@ static bool ip6_pkt_too_big(const struct sk_buff *skb, unsigned int mtu)
 	if (skb->ignore_df)
 		return false;
 
-	if (skb_is_gso(skb) && skb_gso_network_seglen(skb) <= mtu)
+	if (skb_is_gso(skb) && skb_gso_validate_mtu(skb, mtu))
 		return false;
 
 	return true;
net/mpls/af_mpls.c +1 -1
index 0b80a7140cc494d8c39bd3efba2423272d1b8844..7a4aa3450dd71039e73516bd711ba7392493eb5e 100644
@@ -91,7 +91,7 @@ bool mpls_pkt_too_big(const struct sk_buff *skb, unsigned int mtu)
 	if (skb->len <= mtu)
 		return false;
 
-	if (skb_is_gso(skb) && skb_gso_network_seglen(skb) <= mtu)
+	if (skb_is_gso(skb) && skb_gso_validate_mtu(skb, mtu))
 		return false;
 
 	return true;
Signature · 1 line
2.5.5
author maintainer
#5 lore raw

PATCH v3 5/7 sctp: delay as much as possible skb_linearize

In reply to Marcelo Ricardo Leitner

This patch is a preparation for the GSO one. In order to successfully handle GSO packets on rx path we must not call skb_linearize, otherwise it defeats any gain GSO may have had.

This patch thus delays as much as possible the call to skb_linearize, leaving it to sctp_inq_pop() moment. For that the sanity checks performed now know how to deal with fragments.

One positive side-effect of this is that if the socket is backlogged it will have the chance of doing it on backlog processing instead of during softirq.

With this move, it's evident that a check for non-linearity in sctp_inq_pop was ineffective and is now removed. Note that a similar check is performed a bit below this one.

Signed-off-by Marcelo Ricardo Leitner <[email protected]>
Tested-by Xin Long <[email protected]>
---
 net/sctp/input.c   | 45 +++++++++++++++++++++++++--------------------
 net/sctp/inqueue.c | 29 ++++++++++++++++++-----------
 2 files changed, 43 insertions(+), 31 deletions(-)
net/sctp/input.c +25 -20
index a701527a9480faff1b8d91257e1dbf3c0f09ed68..5cff2546c3dd6d3823b5a28bac1e72880cd57756 100644
@@ -112,7 +112,6 @@ int sctp_rcv(struct sk_buff *skb)
 	struct sctp_ep_common *rcvr;
 	struct sctp_transport *transport = NULL;
 	struct sctp_chunk *chunk;
-	struct sctphdr *sh;
 	union sctp_addr src;
 	union sctp_addr dest;
 	int family;

@@ -124,15 +123,18 @@ int sctp_rcv(struct sk_buff *skb)
 
 	__SCTP_INC_STATS(net, SCTP_MIB_INSCTPPACKS);
 
-	if (skb_linearize(skb))
+	/* If packet is too small to contain a single chunk, let's not
+	 * waste time on it anymore.
+	 */
+	if (skb->len < sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr) +
+		       skb_transport_offset(skb))
 		goto discard_it;
 
-	sh = sctp_hdr(skb);
+	if (!pskb_may_pull(skb, sizeof(struct sctphdr)))
+		goto discard_it;
 
-	/* Pull up the IP and SCTP headers. */
+	/* Pull up the IP header. */
 	__skb_pull(skb, skb_transport_offset(skb));
-	if (skb->len < sizeof(struct sctphdr))
-		goto discard_it;
 
 	skb->csum_valid = 0; /* Previous value not applicable */
 	if (skb_csum_unnecessary(skb))

@@ -141,11 +143,7 @@ int sctp_rcv(struct sk_buff *skb)
 		goto discard_it;
 	skb->csum_valid = 1;
 
-	skb_pull(skb, sizeof(struct sctphdr));
-
-	/* Make sure we at least have chunk headers worth of data left. */
-	if (skb->len < sizeof(struct sctp_chunkhdr))
-		goto discard_it;
+	__skb_pull(skb, sizeof(struct sctphdr));
 
 	family = ipver2af(ip_hdr(skb)->version);
 	af = sctp_get_af_specific(family);

@@ -230,7 +228,7 @@ int sctp_rcv(struct sk_buff *skb)
 	chunk->rcvr = rcvr;
 
 	/* Remember the SCTP header. */
-	chunk->sctp_hdr = sh;
+	chunk->sctp_hdr = sctp_hdr(skb);
 
 	/* Set the source and destination addresses of the incoming chunk.  */
 	sctp_init_addrs(chunk, &src, &dest);

@@ -660,19 +658,23 @@ out_unlock:
  */
 static int sctp_rcv_ootb(struct sk_buff *skb)
 {
-	sctp_chunkhdr_t *ch;
-	__u8 *ch_end;
-
-	ch = (sctp_chunkhdr_t *) skb->data;
+	sctp_chunkhdr_t *ch, _ch;
+	int ch_end, offset = 0;
 
 	/* Scan through all the chunks in the packet.  */
 	do {
+		/* Make sure we have at least the header there */
+		if (offset + sizeof(sctp_chunkhdr_t) > skb->len)
+			break;
+
+		ch = skb_header_pointer(skb, offset, sizeof(*ch), &_ch);
+
 		/* Break out if chunk length is less then minimal. */
 		if (ntohs(ch->length) < sizeof(sctp_chunkhdr_t))
 			break;
 
-		ch_end = ((__u8 *)ch) + WORD_ROUND(ntohs(ch->length));
-		if (ch_end > skb_tail_pointer(skb))
+		ch_end = offset + WORD_ROUND(ntohs(ch->length));
+		if (ch_end > skb->len)
 			break;
 
 		/* RFC 8.4, 2) If the OOTB packet contains an ABORT chunk, the

@@ -697,8 +699,8 @@ static int sctp_rcv_ootb(struct sk_buff *skb)
 		if (SCTP_CID_INIT = ch->type && (void *)ch != skb->data)
 			goto discard;
 
-		ch = (sctp_chunkhdr_t *) ch_end;
-	} while (ch_end < skb_tail_pointer(skb));
+		offset = ch_end;
+	} while (ch_end < skb->len);
 
 	return 0;
 

@@ -1173,6 +1175,9 @@ static struct sctp_association *__sctp_rcv_lookup_harder(struct net *net,
 {
 	sctp_chunkhdr_t *ch;
 
+	if (skb_linearize(skb))
+		return NULL;
+
 	ch = (sctp_chunkhdr_t *) skb->data;
 
 	/* The code below will attempt to walk the chunk and extract
net/sctp/inqueue.c +18 -10
index 9d87bba0ff1d34134093f34e4db3dc1b7e3dafd6..5ba08ceda3ab6cf61eb64c58bffb9ccd589e8664 100644
@@ -130,7 +130,8 @@ struct sctp_chunk *sctp_inq_pop(struct sctp_inq *queue)
 	 * at this time.
 	 */
 
-	if ((chunk = queue->in_progress)) {
+	chunk = queue->in_progress;
+	if (chunk) {
 		/* There is a packet that we have been working on.
 		 * Any post processing work to do before we move on?
 		 */

@@ -152,15 +153,29 @@ struct sctp_chunk *sctp_inq_pop(struct sctp_inq *queue)
 	if (!chunk) {
 		struct list_head *entry;
 
+next_chunk:
 		/* Is the queue empty?  */
 		if (list_empty(&queue->in_chunk_list))
 			return NULL;
 
 		entry = queue->in_chunk_list.next;
-		chunk = queue->in_progress -			list_entry(entry, struct sctp_chunk, list);
+		chunk = list_entry(entry, struct sctp_chunk, list);
 		list_del_init(entry);
 
+		/* Linearize if it's not GSO */
+		if (skb_is_nonlinear(chunk->skb)) {
+			if (skb_linearize(chunk->skb)) {
+				__SCTP_INC_STATS(dev_net(chunk->skb->dev), SCTP_MIB_IN_PKT_DISCARDS);
+				sctp_chunk_free(chunk);
+				goto next_chunk;
+			}
+
+			/* Update sctp_hdr as it probably changed */
+			chunk->sctp_hdr = sctp_hdr(chunk->skb);
+		}
+
+		queue->in_progress = chunk;
+
 		/* This is the first chunk in the packet.  */
 		chunk->singleton = 1;
 		ch = (sctp_chunkhdr_t *) chunk->skb->data;

@@ -172,14 +187,6 @@ struct sctp_chunk *sctp_inq_pop(struct sctp_inq *queue)
 
 	chunk->chunk_hdr = ch;
 	chunk->chunk_end = ((__u8 *)ch) + WORD_ROUND(ntohs(ch->length));
-	/* In the unlikely case of an IP reassembly, the skb could be
-	 * non-linear. If so, update chunk_end so that it doesn't go past
-	 * the skb->tail.
-	 */
-	if (unlikely(skb_is_nonlinear(chunk->skb))) {
-		if (chunk->chunk_end > skb_tail_pointer(chunk->skb))
-			chunk->chunk_end = skb_tail_pointer(chunk->skb);
-	}
 	skb_pull(chunk->skb, sizeof(sctp_chunkhdr_t));
 	chunk->subh.v = NULL; /* Subheader is no longer valid.  */
Signature · 1 line
2.5.5
author maintainer
#6 lore raw

PATCH v3 6/7 sctp: Add GSO support

In reply to Marcelo Ricardo Leitner

SCTP has this pecualiarity that its packets cannot be just segmented to (P)MTU. Its chunks must be contained in IP segments, padding respected. So we can't just generate a big skb, set gso_size to the fragmentation point and deliver it to IP layer.

This patch takes a different approach. SCTP will now build a skb as it would be if it was received using GRO. That is, there will be a cover skb with protocol headers and children ones containing the actual segments, already segmented to a way that respects SCTP RFCs.

With that, we can tell skb_segment() to just split based on frag_list, trusting its sizes are already in accordance.

This way SCTP can benefit from GSO and instead of passing several packets through the stack, it can pass a single large packet.

v2:
- Added support for receiving GSO frames, as requested by Dave Miller.
- Clear skb->cb if packet is GSO (otherwise it's not used by SCTP)
- Added heuristics similar to what we have in TCP for not generating single GSO packets that fills cwnd.
v3:
- consider sctphdr size in skb_gso_transport_seglen()
- rebased due to 5c7cdf339af5 ("gso: Remove arbitrary checks for unsupported GSO")

Signed-off-by Marcelo Ricardo Leitner <[email protected]>
Tested-by Xin Long <[email protected]>
---
 include/linux/netdev_features.h |   7 +-
 include/linux/netdevice.h       |   1 +
 include/linux/skbuff.h          |   2 +
 include/net/sctp/sctp.h         |   4 +
 include/net/sctp/structs.h      |   5 +
 net/core/ethtool.c              |   1 +
 net/core/skbuff.c               |   3 +
 net/sctp/Makefile               |   3 +-
 net/sctp/input.c                |  12 +-
 net/sctp/inqueue.c              |  51 +++++-
 net/sctp/offload.c              |  98 +++++++++++
 net/sctp/output.c               | 363 +++++++++++++++++++++++++++-------------
 net/sctp/protocol.c             |   3 +
 net/sctp/socket.c               |   2 +
 14 files changed, 429 insertions(+), 126 deletions(-)
 create mode 100644 net/sctp/offload.c
include/linux/netdev_features.h +5 -2
index aa7b2400f98c584d29e83f0eddf7bf13766cedd1..9c6c8ef2e9e704513cc4272b0a3ee2fec6809d46 100644
@@ -53,8 +53,9 @@ enum {
 					 *     headers in software.
 					 */
 	NETIF_F_GSO_TUNNEL_REMCSUM_BIT, /* ... TUNNEL with TSO & REMCSUM */
+	NETIF_F_GSO_SCTP_BIT,		/* ... SCTP fragmentation */
 	/**/NETIF_F_GSO_LAST =		/* last bit, see GSO_MASK */
-		NETIF_F_GSO_TUNNEL_REMCSUM_BIT,
+		NETIF_F_GSO_SCTP_BIT,
 
 	NETIF_F_FCOE_CRC_BIT,		/* FCoE CRC32 */
 	NETIF_F_SCTP_CRC_BIT,		/* SCTP checksum offload */

@@ -128,6 +129,7 @@ enum {
 #define NETIF_F_TSO_MANGLEID	__NETIF_F(TSO_MANGLEID)
 #define NETIF_F_GSO_PARTIAL	 __NETIF_F(GSO_PARTIAL)
 #define NETIF_F_GSO_TUNNEL_REMCSUM __NETIF_F(GSO_TUNNEL_REMCSUM)
+#define NETIF_F_GSO_SCTP	__NETIF_F(GSO_SCTP)
 #define NETIF_F_HW_VLAN_STAG_FILTER __NETIF_F(HW_VLAN_STAG_FILTER)
 #define NETIF_F_HW_VLAN_STAG_RX	__NETIF_F(HW_VLAN_STAG_RX)
 #define NETIF_F_HW_VLAN_STAG_TX	__NETIF_F(HW_VLAN_STAG_TX)

@@ -166,7 +168,8 @@ enum {
 				 NETIF_F_FSO)
 
 /* List of features with software fallbacks. */
-#define NETIF_F_GSO_SOFTWARE	(NETIF_F_ALL_TSO | NETIF_F_UFO)
+#define NETIF_F_GSO_SOFTWARE	(NETIF_F_ALL_TSO | NETIF_F_UFO | \
+				 NETIF_F_GSO_SCTP)
 
 /*
  * If one device supports one of these features, then enable them
include/linux/netdevice.h +1 -0
index f45929ce815725d868261e9a2585ac53d0c8f128..fa6df2699532e4ad6deb37f1bdcfafc71d2580cb 100644
@@ -4012,6 +4012,7 @@ static inline bool net_gso_ok(netdev_features_t features, int gso_type)
 	BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
 	BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
 	BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
+	BUILD_BUG_ON(SKB_GSO_SCTP    != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
 
 	return (features & feature) = feature;
 }
include/linux/skbuff.h +2 -0
index aa3f9d7e8d5ca455387efa22d4a9d3a079a56f0c..dc0fca747c5e1c5b23b1e52ce3e354667eb2a994 100644
@@ -487,6 +487,8 @@ enum {
 	SKB_GSO_PARTIAL = 1 << 13,
 
 	SKB_GSO_TUNNEL_REMCSUM = 1 << 14,
+
+	SKB_GSO_SCTP = 1 << 15,
 };
 
 #if BITS_PER_LONG > 32
include/net/sctp/sctp.h +4 -0
index b392ac8382f2bf0be118f797a4444cc0eb4ddeb5..632e205ca54bfe85124753e09445251056e19aa7 100644
@@ -186,6 +186,10 @@ void sctp_assocs_proc_exit(struct net *net);
 int sctp_remaddr_proc_init(struct net *net);
 void sctp_remaddr_proc_exit(struct net *net);
 
+/*
+ * sctp/offload.c
+ */
+int sctp_offload_init(void);
 
 /*
  * Module global variables
include/net/sctp/structs.h +5 -0
index 16b013a6191cf1c416e4dd1aeb1707a8569ea49b..83c5ec58b93a073fce845c453747191f23495ffb 100644
@@ -566,6 +566,9 @@ struct sctp_chunk {
 	/* This points to the sk_buff containing the actual data.  */
 	struct sk_buff *skb;
 
+	/* In case of GSO packets, this will store the head one */
+	struct sk_buff *head_skb;
+
 	/* These are the SCTP headers by reverse order in a packet.
 	 * Note that some of these may happen more than once.  In that
 	 * case, we point at the "current" one, whatever that means

@@ -696,6 +699,8 @@ struct sctp_packet {
 	size_t overhead;
 	/* This is the total size of all chunks INCLUDING padding.  */
 	size_t size;
+	/* This is the maximum size this packet may have */
+	size_t max_size;
 
 	/* The packet is destined for this transport address.
 	 * The function we finally use to pass down to the next lower
net/core/ethtool.c +1 -0
index f4034817d2559babb39b06c432facf14e3347890..977489820eb957098705c9ea1674ed3092a6cfe6 100644
@@ -89,6 +89,7 @@ static const char netdev_features_strings[NETDEV_FEATURE_COUNT][ETH_GSTRING_LEN]
 	[NETIF_F_GSO_UDP_TUNNEL_BIT] =	 "tx-udp_tnl-segmentation",
 	[NETIF_F_GSO_UDP_TUNNEL_CSUM_BIT] = "tx-udp_tnl-csum-segmentation",
 	[NETIF_F_GSO_PARTIAL_BIT] =	 "tx-gso-partial",
+	[NETIF_F_GSO_SCTP_BIT] =	 "tx-sctp-segmentation",
 
 	[NETIF_F_FCOE_CRC_BIT] =         "tx-checksum-fcoe-crc",
 	[NETIF_F_SCTP_CRC_BIT] =        "tx-checksum-sctp",
net/core/skbuff.c +3 -0
index 5ca562b56ec39d39e1225d96547e242732518ffe..b6e0f95bef3645659bf782adae8885c32670dde7 100644
@@ -49,6 +49,7 @@
 #include <linux/slab.h>
 #include <linux/tcp.h>
 #include <linux/udp.h>
+#include <linux/sctp.h>
 #include <linux/netdevice.h>
 #ifdef CONFIG_NET_CLS_ACT
 #include <net/pkt_sched.h>

@@ -4383,6 +4384,8 @@ unsigned int skb_gso_transport_seglen(const struct sk_buff *skb)
 			thlen += inner_tcp_hdrlen(skb);
 	} else if (likely(shinfo->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6))) {
 		thlen = tcp_hdrlen(skb);
+	} else if (unlikely(shinfo->gso_type & SKB_GSO_SCTP)) {
+		thlen = sizeof(struct sctphdr);
 	}
 	/* UFO sets gso_size to the size of the fragmentation
 	 * payload, i.e. the size of the L4 (UDP) header is already
net/sctp/Makefile +2 -1
index 0fca5824ad0e93c905e2cbd59ff2ff7e2077ca7c..6c4f7496cec612b52e1e69664a209b4d58763be5 100644
@@ -11,7 +11,8 @@ sctp-y := sm_statetable.o sm_statefuns.o sm_sideeffect.o \
 	  transport.o chunk.o sm_make_chunk.o ulpevent.o \
 	  inqueue.o outqueue.o ulpqueue.o \
 	  tsnmap.o bind_addr.o socket.o primitive.o \
-	  output.o input.o debug.o ssnmap.o auth.o
+	  output.o input.o debug.o ssnmap.o auth.o \
+	  offload.o
 
 sctp_probe-y := probe.o
net/sctp/input.c +11 -1
index 5cff2546c3dd6d3823b5a28bac1e72880cd57756..6f8e676d285ead987b0a1337beec3b29c34e0a8e 100644
@@ -139,7 +139,9 @@ int sctp_rcv(struct sk_buff *skb)
 	skb->csum_valid = 0; /* Previous value not applicable */
 	if (skb_csum_unnecessary(skb))
 		__skb_decr_checksum_unnecessary(skb);
-	else if (!sctp_checksum_disable && sctp_rcv_checksum(net, skb) < 0)
+	else if (!sctp_checksum_disable &&
+		 !(skb_shinfo(skb)->gso_type & SKB_GSO_SCTP) &&
+		 sctp_rcv_checksum(net, skb) < 0)
 		goto discard_it;
 	skb->csum_valid = 1;
 

@@ -1175,6 +1177,14 @@ static struct sctp_association *__sctp_rcv_lookup_harder(struct net *net,
 {
 	sctp_chunkhdr_t *ch;
 
+	/* We do not allow GSO frames here as we need to linearize and
+	 * then cannot guarantee frame boundaries. This shouldn't be an
+	 * issue as packets hitting this are mostly INIT or INIT-ACK and
+	 * those cannot be on GSO-style anyway.
+	 */
+	if ((skb_shinfo(skb)->gso_type & SKB_GSO_SCTP) = SKB_GSO_SCTP)
+		return NULL;
+
 	if (skb_linearize(skb))
 		return NULL;
net/sctp/inqueue.c +43 -8

79 lines not on this page · raw message

net/sctp/offload.c new file +98 -0

98 lines not on this page · raw message

net/sctp/output.c +248 -114

482 lines not on this page · raw message

net/sctp/protocol.c +3 -0
index d3d50daa248b06d7a4306d903b2dad89e9d2acbd..40022ee885d7e8d9fbce3c7d9df43f57f0bcfa0e 100644
@@ -1516,6 +1516,9 @@ static __init int sctp_init(void)
 	if (status)
 		goto err_v6_add_protocol;
 
+	if (sctp_offload_init() < 0)
+		pr_crit("%s: Cannot add SCTP protocol offload\n", __func__);
+
 out:
 	return status;
 err_v6_add_protocol:
net/sctp/socket.c +2 -0
index 67154b848aa9bae93ce2bc6941e7974acb678142..712fb2339baa24fc0a2e354e8aaea72de8acdc81 100644
@@ -4003,6 +4003,8 @@ static int sctp_init_sock(struct sock *sk)
 		return -ESOCKTNOSUPPORT;
 	}
 
+	sk->sk_gso_type = SKB_GSO_SCTP;
+
 	/* Initialize default send parameters. These parameters can be
 	 * modified with the SCTP_DEFAULT_SEND_PARAM socket option.
 	 */
Signature · 1 line
2.5.5
author maintainer
#7 lore raw

PATCH v3 7/7 sctp: improve debug message to also log curr pkt and new chunk size

In reply to Marcelo Ricardo Leitner

This is useful for debugging packet sizes.

Signed-off-by Marcelo Ricardo Leitner <[email protected]>
Tested-by Xin Long <[email protected]>
---
 net/sctp/output.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/sctp/output.c +2 -1
index 60499a69179d255c47da1fa19b73147917a050bf..90d2e125c2f5e0e1ecb33a7eab10772e5b39567c 100644
@@ -182,7 +182,8 @@ sctp_xmit_t sctp_packet_transmit_chunk(struct sctp_packet *packet,
 	sctp_xmit_t retval;
 	int error = 0;
 
-	pr_debug("%s: packet:%p chunk:%p\n", __func__, packet, chunk);
+	pr_debug("%s: packet:%p size:%lu chunk:%p size:%d\n", __func__,
+		 packet, packet->size, chunk, chunk->skb ? chunk->skb->len : -1);
 
 	switch ((retval = (sctp_packet_append_chunk(packet, chunk)))) {
 	case SCTP_XMIT_PMTU_FULL:
Signature · 1 line
2.5.5
maintainer
#8 lore raw
In reply to Marcelo Ricardo Leitner

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 2 Jun 2016 15:05:37 -0300

Quoted 14 lines

This patchset adds sctp GSO support.

Performance tests indicates that increases throughput by 10% if using bigger chunk sizes, specially if bigger than MTU. For small chunks, it doesn't help much if not using heavy firewall rules.

For small chunks it will probably be of more use once we get something like MSG_MORE as David Laight had suggested.

overall changes:
v1->v2:
Added support for receiving GSO frames on SCTP stack, as requested by Dave Miller.

Quoted 4 lines

v2->v3:
Consider sctphdr size in skb_gso_transport_seglen() rebased due to 5c7cdf339af5 ("gso: Remove arbitrary checks for unsupported GSO")

Series applied, thanks Marcelo.