conn, device, tun: implement vectorized I/O on Linux

Implement TCP offloading via TSO and GRO for the Linux tun.Device, which
is made possible by virtio extensions in the kernel's TUN driver.

Delete conn.LinuxSocketEndpoint in favor of a collapsed conn.StdNetBind.
conn.StdNetBind makes use of recvmmsg() and sendmmsg() on Linux. All
platforms now fall under conn.StdNetBind, except for Windows, which
remains in conn.WinRingBind, which still needs to be adjusted to handle
multiple packets.

Also refactor sticky sockets support to eventually be applicable on
platforms other than just Linux. However Linux remains the sole platform
that fully implements it for now.

Co-authored-by: James Tucker <james@tailscale.com>
Signed-off-by: James Tucker <james@tailscale.com>
Signed-off-by: Jordan Whited <jordan@tailscale.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>

3 files changed, 13 insertions, 9 deletions

diff --git a/device/queueconstants_android.go b/device/queueconstants_android.go
index fc937b3..1158387 100644
--- a/device/queueconstants_android.go
+++ b/device/queueconstants_android.go
@@ -5,10 +5,12 @@
 
 package device
 
+import "golang.zx2c4.com/wireguard/conn"
+
 /* Reduce memory consumption for Android */
 
 const (
-	QueueStagedSize            = 128
+	QueueStagedSize            = conn.DefaultBatchSize
 	QueueOutboundSize          = 1024
 	QueueInboundSize           = 1024
 	QueueHandshakeSize         = 1024
diff --git a/device/queueconstants_default.go b/device/queueconstants_default.go
index 6b69150..7ed70a1 100644
--- a/device/queueconstants_default.go
+++ b/device/queueconstants_default.go
@@ -7,8 +7,10 @@
 
 package device
 
+import "golang.zx2c4.com/wireguard/conn"
+
 const (
-	QueueStagedSize            = 128
+	QueueStagedSize            = conn.DefaultBatchSize
 	QueueOutboundSize          = 1024
 	QueueInboundSize           = 1024
 	QueueHandshakeSize         = 1024
diff --git a/device/sticky_linux.go b/device/sticky_linux.go
index 7afdf28..3ce0769 100644
--- a/device/sticky_linux.go
+++ b/device/sticky_linux.go
@@ -25,7 +25,7 @@ import (
 )
 
 func (device *Device) startRouteListener(bind conn.Bind) (*rwcancel.RWCancel, error) {
-	if _, ok := bind.(*conn.LinuxSocketBind); !ok {
+	if _, ok := bind.(*conn.StdNetBind); !ok {
 		return nil, nil
 	}
 
@@ -112,11 +112,11 @@ func (device *Device) routineRouteListener(bind conn.Bind, netlinkSock int, netl
 									pePtr.peer.Unlock()
 									break
 								}
-								if uint32(pePtr.peer.endpoint.(*conn.LinuxSocketEndpoint).Src4().Ifindex) == ifidx {
+								if uint32(pePtr.peer.endpoint.(*conn.StdNetEndpoint).SrcIfidx()) == ifidx {
 									pePtr.peer.Unlock()
 									break
 								}
-								pePtr.peer.endpoint.(*conn.LinuxSocketEndpoint).ClearSrc()
+								pePtr.peer.endpoint.(*conn.StdNetEndpoint).ClearSrc()
 								pePtr.peer.Unlock()
 							}
 							attr = attr[attrhdr.Len:]
@@ -136,12 +136,12 @@ func (device *Device) routineRouteListener(bind conn.Bind, netlinkSock int, netl
 							peer.RUnlock()
 							continue
 						}
-						nativeEP, _ := peer.endpoint.(*conn.LinuxSocketEndpoint)
+						nativeEP, _ := peer.endpoint.(*conn.StdNetEndpoint)
 						if nativeEP == nil {
 							peer.RUnlock()
 							continue
 						}
-						if nativeEP.IsV6() || nativeEP.Src4().Ifindex == 0 {
+						if nativeEP.DstIP().Is6() || nativeEP.SrcIfidx() == 0 {
 							peer.RUnlock()
 							break
 						}
@@ -169,12 +169,12 @@ func (device *Device) routineRouteListener(bind conn.Bind, netlinkSock int, netl
 								Len:  8,
 								Type: unix.RTA_DST,
 							},
-							nativeEP.Dst4().Addr,
+							nativeEP.DstIP().As4(),
 							unix.RtAttr{
 								Len:  8,
 								Type: unix.RTA_SRC,
 							},
-							nativeEP.Src4().Src,
+							nativeEP.SrcIP().As4(),
 							unix.RtAttr{
 								Len:  8,
 								Type: unix.RTA_MARK,