netstack/tcp: software segmentation offload

Right now, we send each tcp packet separately, we call one system
call per-packet. This patch allows to generate multiple tcp packets
and send them by sendmmsg.

The arguable part of this CL is a way how to handle multiple headers.
This CL adds the next field to the Prepandable buffer.

Nginx test results:

Server Software:        nginx/1.15.9
Server Hostname:        10.138.0.2
Server Port:            8080

Document Path:          /10m.txt
Document Length:        10485760 bytes

w/o gso:
Concurrency Level:      5
Time taken for tests:   5.491 seconds
Complete requests:      100
Failed requests:        0
Total transferred:      1048600200 bytes
HTML transferred:       1048576000 bytes
Requests per second:    18.21 [#/sec] (mean)
Time per request:       274.525 [ms] (mean)
Time per request:       54.905 [ms] (mean, across all concurrent requests)
Transfer rate:          186508.03 [Kbytes/sec] received

sw-gso:

Concurrency Level:      5
Time taken for tests:   3.852 seconds
Complete requests:      100
Failed requests:        0
Total transferred:      1048600200 bytes
HTML transferred:       1048576000 bytes
Requests per second:    25.96 [#/sec] (mean)
Time per request:       192.576 [ms] (mean)
Time per request:       38.515 [ms] (mean, across all concurrent requests)
Transfer rate:          265874.92 [Kbytes/sec] received

w/o gso:
$ ./tcp_benchmark --client --duration 15  --ideal
[SUM]  0.0-15.1 sec  2.20 GBytes  1.25 Gbits/sec

software gso:
$ tcp_benchmark --client --duration 15  --ideal --gso $((1<<16)) --swgso
[SUM]  0.0-15.1 sec  3.99 GBytes  2.26 Gbits/sec

PiperOrigin-RevId: 276112677

1 files changed, 6 insertions, 6 deletions

diff --git a/pkg/tcpip/link/fdbased/packet_dispatchers.go b/pkg/tcpip/link/fdbased/packet_dispatchers.go
index 9fb760891..12168a1dc 100644
--- a/pkg/tcpip/link/fdbased/packet_dispatchers.go
+++ b/pkg/tcpip/link/fdbased/packet_dispatchers.go
@@ -53,7 +53,7 @@ func newReadVDispatcher(fd int, e *endpoint) (linkDispatcher, error) {
 	d := &readVDispatcher{fd: fd, e: e}
 	d.views = make([]buffer.View, len(BufConfig))
 	iovLen := len(BufConfig)
-	if d.e.Capabilities()&stack.CapabilityGSO != 0 {
+	if d.e.Capabilities()&stack.CapabilityHardwareGSO != 0 {
 		iovLen++
 	}
 	d.iovecs = make([]syscall.Iovec, iovLen)
@@ -63,7 +63,7 @@ func newReadVDispatcher(fd int, e *endpoint) (linkDispatcher, error) {
 func (d *readVDispatcher) allocateViews(bufConfig []int) {
 	var vnetHdr [virtioNetHdrSize]byte
 	vnetHdrOff := 0
-	if d.e.Capabilities()&stack.CapabilityGSO != 0 {
+	if d.e.Capabilities()&stack.CapabilityHardwareGSO != 0 {
 		// The kernel adds virtioNetHdr before each packet, but
 		// we don't use it, so so we allocate a buffer for it,
 		// add it in iovecs but don't add it in a view.
@@ -106,7 +106,7 @@ func (d *readVDispatcher) dispatch() (bool, *tcpip.Error) {
 	if err != nil {
 		return false, err
 	}
-	if d.e.Capabilities()&stack.CapabilityGSO != 0 {
+	if d.e.Capabilities()&stack.CapabilityHardwareGSO != 0 {
 		// Skip virtioNetHdr which is added before each packet, it
 		// isn't used and it isn't in a view.
 		n -= virtioNetHdrSize
@@ -195,7 +195,7 @@ func newRecvMMsgDispatcher(fd int, e *endpoint) (linkDispatcher, error) {
 	}
 	d.iovecs = make([][]syscall.Iovec, MaxMsgsPerRecv)
 	iovLen := len(BufConfig)
-	if d.e.Capabilities()&stack.CapabilityGSO != 0 {
+	if d.e.Capabilities()&stack.CapabilityHardwareGSO != 0 {
 		// virtioNetHdr is prepended before each packet.
 		iovLen++
 	}
@@ -226,7 +226,7 @@ func (d *recvMMsgDispatcher) allocateViews(bufConfig []int) {
 	for k := 0; k < len(d.views); k++ {
 		var vnetHdr [virtioNetHdrSize]byte
 		vnetHdrOff := 0
-		if d.e.Capabilities()&stack.CapabilityGSO != 0 {
+		if d.e.Capabilities()&stack.CapabilityHardwareGSO != 0 {
 			// The kernel adds virtioNetHdr before each packet, but
 			// we don't use it, so so we allocate a buffer for it,
 			// add it in iovecs but don't add it in a view.
@@ -262,7 +262,7 @@ func (d *recvMMsgDispatcher) dispatch() (bool, *tcpip.Error) {
 	// Process each of received packets.
 	for k := 0; k < nMsgs; k++ {
 		n := int(d.msgHdrs[k].Len)
-		if d.e.Capabilities()&stack.CapabilityGSO != 0 {
+		if d.e.Capabilities()&stack.CapabilityHardwareGSO != 0 {
 			n -= virtioNetHdrSize
 		}
 		if n <= d.e.hdrSize {