summaryrefslogtreecommitdiffhomepage
path: root/pkg/tcpip/transport
diff options
context:
space:
mode:
Diffstat (limited to 'pkg/tcpip/transport')
-rw-r--r--pkg/tcpip/transport/icmp/endpoint.go46
-rw-r--r--pkg/tcpip/transport/icmp/protocol.go7
-rw-r--r--pkg/tcpip/transport/raw/BUILD1
-rw-r--r--pkg/tcpip/transport/raw/endpoint.go135
-rw-r--r--pkg/tcpip/transport/raw/protocol.go32
-rw-r--r--pkg/tcpip/transport/tcp/endpoint.go49
-rw-r--r--pkg/tcpip/transport/tcp/endpoint_state.go1
-rw-r--r--pkg/tcpip/transport/tcp/testing/context/context.go6
-rw-r--r--pkg/tcpip/transport/udp/endpoint.go75
-rw-r--r--pkg/tcpip/transport/udp/endpoint_state.go2
-rw-r--r--pkg/tcpip/transport/udp/udp_test.go339
11 files changed, 478 insertions, 215 deletions
diff --git a/pkg/tcpip/transport/icmp/endpoint.go b/pkg/tcpip/transport/icmp/endpoint.go
index 33cefd937..ba6671c26 100644
--- a/pkg/tcpip/transport/icmp/endpoint.go
+++ b/pkg/tcpip/transport/icmp/endpoint.go
@@ -291,7 +291,7 @@ func (e *endpoint) Write(p tcpip.Payload, opts tcpip.WriteOptions) (uintptr, <-c
switch e.netProto {
case header.IPv4ProtocolNumber:
- err = e.send4(route, v)
+ err = send4(route, e.id.LocalPort, v)
case header.IPv6ProtocolNumber:
err = send6(route, e.id.LocalPort, v)
@@ -314,6 +314,22 @@ func (e *endpoint) SetSockOpt(opt interface{}) *tcpip.Error {
return nil
}
+// GetSockOptInt implements tcpip.Endpoint.GetSockOptInt.
+func (e *endpoint) GetSockOptInt(opt tcpip.SockOpt) (int, *tcpip.Error) {
+ switch opt {
+ case tcpip.ReceiveQueueSizeOption:
+ v := 0
+ e.rcvMu.Lock()
+ if !e.rcvList.Empty() {
+ p := e.rcvList.Front()
+ v = p.data.Size()
+ }
+ e.rcvMu.Unlock()
+ return v, nil
+ }
+ return -1, tcpip.ErrUnknownProtocolOption
+}
+
// GetSockOpt implements tcpip.Endpoint.GetSockOpt.
func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
switch o := opt.(type) {
@@ -332,17 +348,6 @@ func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
e.rcvMu.Unlock()
return nil
- case *tcpip.ReceiveQueueSizeOption:
- e.rcvMu.Lock()
- if e.rcvList.Empty() {
- *o = 0
- } else {
- p := e.rcvList.Front()
- *o = tcpip.ReceiveQueueSizeOption(p.data.Size())
- }
- e.rcvMu.Unlock()
- return nil
-
case *tcpip.KeepaliveEnabledOption:
*o = 0
return nil
@@ -352,20 +357,20 @@ func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
}
}
-func (e *endpoint) send4(r *stack.Route, data buffer.View) *tcpip.Error {
- if len(data) < header.ICMPv4EchoMinimumSize {
+func send4(r *stack.Route, ident uint16, data buffer.View) *tcpip.Error {
+ if len(data) < header.ICMPv4MinimumSize {
return tcpip.ErrInvalidEndpointState
}
// Set the ident to the user-specified port. Sequence number should
// already be set by the user.
- binary.BigEndian.PutUint16(data[header.ICMPv4MinimumSize:], e.id.LocalPort)
+ binary.BigEndian.PutUint16(data[header.ICMPv4PayloadOffset:], ident)
- hdr := buffer.NewPrependable(header.ICMPv4EchoMinimumSize + int(r.MaxHeaderLength()))
+ hdr := buffer.NewPrependable(header.ICMPv4MinimumSize + int(r.MaxHeaderLength()))
- icmpv4 := header.ICMPv4(hdr.Prepend(header.ICMPv4EchoMinimumSize))
+ icmpv4 := header.ICMPv4(hdr.Prepend(header.ICMPv4MinimumSize))
copy(icmpv4, data)
- data = data[header.ICMPv4EchoMinimumSize:]
+ data = data[header.ICMPv4MinimumSize:]
// Linux performs these basic checks.
if icmpv4.Type() != header.ICMPv4Echo || icmpv4.Code() != 0 {
@@ -422,6 +427,11 @@ func (e *endpoint) Connect(addr tcpip.FullAddress) *tcpip.Error {
e.mu.Lock()
defer e.mu.Unlock()
+ if addr.Addr == "" {
+ // AF_UNSPEC isn't supported.
+ return tcpip.ErrAddressFamilyNotSupported
+ }
+
nicid := addr.NIC
localPort := uint16(0)
switch e.state {
diff --git a/pkg/tcpip/transport/icmp/protocol.go b/pkg/tcpip/transport/icmp/protocol.go
index c89538131..7fdba5d56 100644
--- a/pkg/tcpip/transport/icmp/protocol.go
+++ b/pkg/tcpip/transport/icmp/protocol.go
@@ -90,19 +90,18 @@ func (p *protocol) NewRawEndpoint(stack *stack.Stack, netProto tcpip.NetworkProt
func (p *protocol) MinimumPacketSize() int {
switch p.number {
case ProtocolNumber4:
- return header.ICMPv4EchoMinimumSize
+ return header.ICMPv4MinimumSize
case ProtocolNumber6:
return header.ICMPv6EchoMinimumSize
}
panic(fmt.Sprint("unknown protocol number: ", p.number))
}
-// ParsePorts returns the source and destination ports stored in the given icmp
-// packet.
+// ParsePorts in case of ICMP sets src to 0, dst to ICMP ID, and err to nil.
func (p *protocol) ParsePorts(v buffer.View) (src, dst uint16, err *tcpip.Error) {
switch p.number {
case ProtocolNumber4:
- return 0, binary.BigEndian.Uint16(v[header.ICMPv4MinimumSize:]), nil
+ return 0, binary.BigEndian.Uint16(v[header.ICMPv4PayloadOffset:]), nil
case ProtocolNumber6:
return 0, binary.BigEndian.Uint16(v[header.ICMPv6MinimumSize:]), nil
}
diff --git a/pkg/tcpip/transport/raw/BUILD b/pkg/tcpip/transport/raw/BUILD
index 34a14bf7f..bc4b255b4 100644
--- a/pkg/tcpip/transport/raw/BUILD
+++ b/pkg/tcpip/transport/raw/BUILD
@@ -21,6 +21,7 @@ go_library(
"endpoint.go",
"endpoint_state.go",
"packet_list.go",
+ "protocol.go",
],
importpath = "gvisor.dev/gvisor/pkg/tcpip/transport/raw",
imports = ["gvisor.dev/gvisor/pkg/tcpip/buffer"],
diff --git a/pkg/tcpip/transport/raw/endpoint.go b/pkg/tcpip/transport/raw/endpoint.go
index 03f495e48..b633cd9d8 100644
--- a/pkg/tcpip/transport/raw/endpoint.go
+++ b/pkg/tcpip/transport/raw/endpoint.go
@@ -16,7 +16,7 @@
// sockets allow applications to:
//
// * manually write and inspect transport layer headers and payloads
-// * receive all traffic of a given transport protcol (e.g. ICMP or UDP)
+// * receive all traffic of a given transport protocol (e.g. ICMP or UDP)
// * optionally write and inspect network layer and link layer headers for
// packets
//
@@ -67,6 +67,7 @@ type endpoint struct {
netProto tcpip.NetworkProtocolNumber
transProto tcpip.TransportProtocolNumber
waiterQueue *waiter.Queue
+ associated bool
// The following fields are used to manage the receive queue and are
// protected by rcvMu.
@@ -97,8 +98,12 @@ type endpoint struct {
}
// NewEndpoint returns a raw endpoint for the given protocols.
-// TODO(b/129292371): IP_HDRINCL, IPPROTO_RAW, and AF_PACKET.
+// TODO(b/129292371): IP_HDRINCL and AF_PACKET.
func NewEndpoint(stack *stack.Stack, netProto tcpip.NetworkProtocolNumber, transProto tcpip.TransportProtocolNumber, waiterQueue *waiter.Queue) (tcpip.Endpoint, *tcpip.Error) {
+ return newEndpoint(stack, netProto, transProto, waiterQueue, true /* associated */)
+}
+
+func newEndpoint(stack *stack.Stack, netProto tcpip.NetworkProtocolNumber, transProto tcpip.TransportProtocolNumber, waiterQueue *waiter.Queue, associated bool) (tcpip.Endpoint, *tcpip.Error) {
if netProto != header.IPv4ProtocolNumber {
return nil, tcpip.ErrUnknownProtocol
}
@@ -110,6 +115,16 @@ func NewEndpoint(stack *stack.Stack, netProto tcpip.NetworkProtocolNumber, trans
waiterQueue: waiterQueue,
rcvBufSizeMax: 32 * 1024,
sndBufSize: 32 * 1024,
+ associated: associated,
+ }
+
+ // Unassociated endpoints are write-only and users call Write() with IP
+ // headers included. Because they're write-only, We don't need to
+ // register with the stack.
+ if !associated {
+ ep.rcvBufSizeMax = 0
+ ep.waiterQueue = nil
+ return ep, nil
}
if err := ep.stack.RegisterRawTransportEndpoint(ep.registeredNIC, ep.netProto, ep.transProto, ep); err != nil {
@@ -124,7 +139,7 @@ func (ep *endpoint) Close() {
ep.mu.Lock()
defer ep.mu.Unlock()
- if ep.closed {
+ if ep.closed || !ep.associated {
return
}
@@ -142,8 +157,11 @@ func (ep *endpoint) Close() {
if ep.connected {
ep.route.Release()
+ ep.connected = false
}
+ ep.closed = true
+
ep.waiterQueue.Notify(waiter.EventHUp | waiter.EventErr | waiter.EventIn | waiter.EventOut)
}
@@ -152,6 +170,10 @@ func (ep *endpoint) ModerateRecvBuf(copied int) {}
// Read implements tcpip.Endpoint.Read.
func (ep *endpoint) Read(addr *tcpip.FullAddress) (buffer.View, tcpip.ControlMessages, *tcpip.Error) {
+ if !ep.associated {
+ return buffer.View{}, tcpip.ControlMessages{}, tcpip.ErrInvalidOptionValue
+ }
+
ep.rcvMu.Lock()
// If there's no data to read, return that read would block or that the
@@ -192,6 +214,33 @@ func (ep *endpoint) Write(payload tcpip.Payload, opts tcpip.WriteOptions) (uintp
return 0, nil, tcpip.ErrInvalidEndpointState
}
+ payloadBytes, err := payload.Get(payload.Size())
+ if err != nil {
+ ep.mu.RUnlock()
+ return 0, nil, err
+ }
+
+ // If this is an unassociated socket and callee provided a nonzero
+ // destination address, route using that address.
+ if !ep.associated {
+ ip := header.IPv4(payloadBytes)
+ if !ip.IsValid(payload.Size()) {
+ ep.mu.RUnlock()
+ return 0, nil, tcpip.ErrInvalidOptionValue
+ }
+ dstAddr := ip.DestinationAddress()
+ // Update dstAddr with the address in the IP header, unless
+ // opts.To is set (e.g. if sendto specifies a specific
+ // address).
+ if dstAddr != tcpip.Address([]byte{0, 0, 0, 0}) && opts.To == nil {
+ opts.To = &tcpip.FullAddress{
+ NIC: 0, // NIC is unset.
+ Addr: dstAddr, // The address from the payload.
+ Port: 0, // There are no ports here.
+ }
+ }
+ }
+
// Did the user caller provide a destination? If not, use the connected
// destination.
if opts.To == nil {
@@ -216,12 +265,12 @@ func (ep *endpoint) Write(payload tcpip.Payload, opts tcpip.WriteOptions) (uintp
return 0, nil, tcpip.ErrInvalidEndpointState
}
- n, ch, err := ep.finishWrite(payload, savedRoute)
+ n, ch, err := ep.finishWrite(payloadBytes, savedRoute)
ep.mu.Unlock()
return n, ch, err
}
- n, ch, err := ep.finishWrite(payload, &ep.route)
+ n, ch, err := ep.finishWrite(payloadBytes, &ep.route)
ep.mu.RUnlock()
return n, ch, err
}
@@ -248,7 +297,7 @@ func (ep *endpoint) Write(payload tcpip.Payload, opts tcpip.WriteOptions) (uintp
return 0, nil, err
}
- n, ch, err := ep.finishWrite(payload, &route)
+ n, ch, err := ep.finishWrite(payloadBytes, &route)
route.Release()
ep.mu.RUnlock()
return n, ch, err
@@ -256,7 +305,7 @@ func (ep *endpoint) Write(payload tcpip.Payload, opts tcpip.WriteOptions) (uintp
// finishWrite writes the payload to a route. It resolves the route if
// necessary. It's really just a helper to make defer unnecessary in Write.
-func (ep *endpoint) finishWrite(payload tcpip.Payload, route *stack.Route) (uintptr, <-chan struct{}, *tcpip.Error) {
+func (ep *endpoint) finishWrite(payloadBytes []byte, route *stack.Route) (uintptr, <-chan struct{}, *tcpip.Error) {
// We may need to resolve the route (match a link layer address to the
// network address). If that requires blocking (e.g. to use ARP),
// return a channel on which the caller can wait.
@@ -269,13 +318,14 @@ func (ep *endpoint) finishWrite(payload tcpip.Payload, route *stack.Route) (uint
}
}
- payloadBytes, err := payload.Get(payload.Size())
- if err != nil {
- return 0, nil, err
- }
-
switch ep.netProto {
case header.IPv4ProtocolNumber:
+ if !ep.associated {
+ if err := route.WriteHeaderIncludedPacket(buffer.View(payloadBytes).ToVectorisedView()); err != nil {
+ return 0, nil, err
+ }
+ break
+ }
hdr := buffer.NewPrependable(len(payloadBytes) + int(route.MaxHeaderLength()))
if err := route.WritePacket(nil /* gso */, hdr, buffer.View(payloadBytes).ToVectorisedView(), ep.transProto, route.DefaultTTL()); err != nil {
return 0, nil, err
@@ -298,6 +348,11 @@ func (ep *endpoint) Connect(addr tcpip.FullAddress) *tcpip.Error {
ep.mu.Lock()
defer ep.mu.Unlock()
+ if addr.Addr == "" {
+ // AF_UNSPEC isn't supported.
+ return tcpip.ErrAddressFamilyNotSupported
+ }
+
if ep.closed {
return tcpip.ErrInvalidEndpointState
}
@@ -330,15 +385,17 @@ func (ep *endpoint) Connect(addr tcpip.FullAddress) *tcpip.Error {
}
defer route.Release()
- // Re-register the endpoint with the appropriate NIC.
- if err := ep.stack.RegisterRawTransportEndpoint(addr.NIC, ep.netProto, ep.transProto, ep); err != nil {
- return err
+ if ep.associated {
+ // Re-register the endpoint with the appropriate NIC.
+ if err := ep.stack.RegisterRawTransportEndpoint(addr.NIC, ep.netProto, ep.transProto, ep); err != nil {
+ return err
+ }
+ ep.stack.UnregisterRawTransportEndpoint(ep.registeredNIC, ep.netProto, ep.transProto, ep)
+ ep.registeredNIC = nic
}
- ep.stack.UnregisterRawTransportEndpoint(ep.registeredNIC, ep.netProto, ep.transProto, ep)
- // Save the route and NIC we've connected via.
+ // Save the route we've connected via.
ep.route = route.Clone()
- ep.registeredNIC = nic
ep.connected = true
return nil
@@ -381,14 +438,16 @@ func (ep *endpoint) Bind(addr tcpip.FullAddress) *tcpip.Error {
return tcpip.ErrBadLocalAddress
}
- // Re-register the endpoint with the appropriate NIC.
- if err := ep.stack.RegisterRawTransportEndpoint(addr.NIC, ep.netProto, ep.transProto, ep); err != nil {
- return err
+ if ep.associated {
+ // Re-register the endpoint with the appropriate NIC.
+ if err := ep.stack.RegisterRawTransportEndpoint(addr.NIC, ep.netProto, ep.transProto, ep); err != nil {
+ return err
+ }
+ ep.stack.UnregisterRawTransportEndpoint(ep.registeredNIC, ep.netProto, ep.transProto, ep)
+ ep.registeredNIC = addr.NIC
+ ep.boundNIC = addr.NIC
}
- ep.stack.UnregisterRawTransportEndpoint(ep.registeredNIC, ep.netProto, ep.transProto, ep)
- ep.registeredNIC = addr.NIC
- ep.boundNIC = addr.NIC
ep.boundAddr = addr.Addr
ep.bound = true
@@ -428,6 +487,23 @@ func (ep *endpoint) SetSockOpt(opt interface{}) *tcpip.Error {
return nil
}
+// GetSockOptInt implements tcpip.Endpoint.GetSockOptInt.
+func (ep *endpoint) GetSockOptInt(opt tcpip.SockOpt) (int, *tcpip.Error) {
+ switch opt {
+ case tcpip.ReceiveQueueSizeOption:
+ v := 0
+ ep.rcvMu.Lock()
+ if !ep.rcvList.Empty() {
+ p := ep.rcvList.Front()
+ v = p.data.Size()
+ }
+ ep.rcvMu.Unlock()
+ return v, nil
+ }
+
+ return -1, tcpip.ErrUnknownProtocolOption
+}
+
// GetSockOpt implements tcpip.Endpoint.GetSockOpt.
func (ep *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
switch o := opt.(type) {
@@ -446,17 +522,6 @@ func (ep *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
ep.rcvMu.Unlock()
return nil
- case *tcpip.ReceiveQueueSizeOption:
- ep.rcvMu.Lock()
- if ep.rcvList.Empty() {
- *o = 0
- } else {
- p := ep.rcvList.Front()
- *o = tcpip.ReceiveQueueSizeOption(p.data.Size())
- }
- ep.rcvMu.Unlock()
- return nil
-
case *tcpip.KeepaliveEnabledOption:
*o = 0
return nil
diff --git a/pkg/tcpip/transport/raw/protocol.go b/pkg/tcpip/transport/raw/protocol.go
new file mode 100644
index 000000000..783c21e6b
--- /dev/null
+++ b/pkg/tcpip/transport/raw/protocol.go
@@ -0,0 +1,32 @@
+// Copyright 2019 The gVisor Authors.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package raw
+
+import (
+ "gvisor.dev/gvisor/pkg/tcpip"
+ "gvisor.dev/gvisor/pkg/tcpip/stack"
+ "gvisor.dev/gvisor/pkg/waiter"
+)
+
+type factory struct{}
+
+// NewUnassociatedRawEndpoint implements stack.UnassociatedEndpointFactory.
+func (factory) NewUnassociatedRawEndpoint(stack *stack.Stack, netProto tcpip.NetworkProtocolNumber, transProto tcpip.TransportProtocolNumber, waiterQueue *waiter.Queue) (tcpip.Endpoint, *tcpip.Error) {
+ return newEndpoint(stack, netProto, transProto, waiterQueue, false /* associated */)
+}
+
+func init() {
+ stack.RegisterUnassociatedFactory(factory{})
+}
diff --git a/pkg/tcpip/transport/tcp/endpoint.go b/pkg/tcpip/transport/tcp/endpoint.go
index ee60ebf58..89154391b 100644
--- a/pkg/tcpip/transport/tcp/endpoint.go
+++ b/pkg/tcpip/transport/tcp/endpoint.go
@@ -117,6 +117,7 @@ const (
notifyDrain
notifyReset
notifyKeepaliveChanged
+ notifyMSSChanged
)
// SACKInfo holds TCP SACK related information for a given endpoint.
@@ -218,8 +219,6 @@ type endpoint struct {
mu sync.RWMutex `state:"nosave"`
id stack.TransportEndpointID
- // state endpointState `state:".(endpointState)"`
- // pState ProtocolState
state EndpointState `state:".(EndpointState)"`
isPortReserved bool `state:"manual"`
@@ -313,6 +312,10 @@ type endpoint struct {
// in SYN-RCVD state.
synRcvdCount int
+ // userMSS if non-zero is the MSS value explicitly set by the user
+ // for this endpoint using the TCP_MAXSEG setsockopt.
+ userMSS int
+
// The following fields are used to manage the send buffer. When
// segments are ready to be sent, they are added to sndQueue and the
// protocol goroutine is signaled via sndWaker.
@@ -917,6 +920,17 @@ func (e *endpoint) SetSockOpt(opt interface{}) *tcpip.Error {
}
return nil
+ case tcpip.MaxSegOption:
+ userMSS := v
+ if userMSS < header.TCPMinimumMSS || userMSS > header.TCPMaximumMSS {
+ return tcpip.ErrInvalidOptionValue
+ }
+ e.mu.Lock()
+ e.userMSS = int(userMSS)
+ e.mu.Unlock()
+ e.notifyProtocolGoroutine(notifyMSSChanged)
+ return nil
+
case tcpip.ReceiveBufferSizeOption:
// Make sure the receive buffer size is within the min and max
// allowed.
@@ -1086,6 +1100,15 @@ func (e *endpoint) readyReceiveSize() (int, *tcpip.Error) {
return e.rcvBufUsed, nil
}
+// GetSockOptInt implements tcpip.Endpoint.GetSockOptInt.
+func (e *endpoint) GetSockOptInt(opt tcpip.SockOpt) (int, *tcpip.Error) {
+ switch opt {
+ case tcpip.ReceiveQueueSizeOption:
+ return e.readyReceiveSize()
+ }
+ return -1, tcpip.ErrUnknownProtocolOption
+}
+
// GetSockOpt implements tcpip.Endpoint.GetSockOpt.
func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
switch o := opt.(type) {
@@ -1096,6 +1119,14 @@ func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
e.lastErrorMu.Unlock()
return err
+ case *tcpip.MaxSegOption:
+ // This is just stubbed out. Linux never returns the user_mss
+ // value as it either returns the defaultMSS or returns the
+ // actual current MSS. Netstack just returns the defaultMSS
+ // always for now.
+ *o = header.TCPDefaultMSS
+ return nil
+
case *tcpip.SendBufferSizeOption:
e.sndBufMu.Lock()
*o = tcpip.SendBufferSizeOption(e.sndBufSize)
@@ -1108,15 +1139,6 @@ func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
e.rcvListMu.Unlock()
return nil
- case *tcpip.ReceiveQueueSizeOption:
- v, err := e.readyReceiveSize()
- if err != nil {
- return err
- }
-
- *o = tcpip.ReceiveQueueSizeOption(v)
- return nil
-
case *tcpip.DelayOption:
*o = 0
if v := atomic.LoadUint32(&e.delay); v != 0 {
@@ -1271,6 +1293,11 @@ func (e *endpoint) checkV4Mapped(addr *tcpip.FullAddress) (tcpip.NetworkProtocol
// Connect connects the endpoint to its peer.
func (e *endpoint) Connect(addr tcpip.FullAddress) *tcpip.Error {
+ if addr.Addr == "" && addr.Port == 0 {
+ // AF_UNSPEC isn't supported.
+ return tcpip.ErrAddressFamilyNotSupported
+ }
+
return e.connect(addr, true, true)
}
diff --git a/pkg/tcpip/transport/tcp/endpoint_state.go b/pkg/tcpip/transport/tcp/endpoint_state.go
index ec61a3886..b93959034 100644
--- a/pkg/tcpip/transport/tcp/endpoint_state.go
+++ b/pkg/tcpip/transport/tcp/endpoint_state.go
@@ -342,6 +342,7 @@ func loadError(s string) *tcpip.Error {
tcpip.ErrNoBufferSpace,
tcpip.ErrBroadcastDisabled,
tcpip.ErrNotPermitted,
+ tcpip.ErrAddressFamilyNotSupported,
}
messageToError = make(map[string]*tcpip.Error)
diff --git a/pkg/tcpip/transport/tcp/testing/context/context.go b/pkg/tcpip/transport/tcp/testing/context/context.go
index 630dd7925..bcc0f3e28 100644
--- a/pkg/tcpip/transport/tcp/testing/context/context.go
+++ b/pkg/tcpip/transport/tcp/testing/context/context.go
@@ -271,7 +271,7 @@ func (c *Context) GetPacketNonBlocking() []byte {
// SendICMPPacket builds and sends an ICMPv4 packet via the link layer endpoint.
func (c *Context) SendICMPPacket(typ header.ICMPv4Type, code uint8, p1, p2 []byte, maxTotalSize int) {
// Allocate a buffer data and headers.
- buf := buffer.NewView(header.IPv4MinimumSize + header.ICMPv4MinimumSize + len(p1) + len(p2))
+ buf := buffer.NewView(header.IPv4MinimumSize + header.ICMPv4PayloadOffset + len(p1) + len(p2))
if len(buf) > maxTotalSize {
buf = buf[:maxTotalSize]
}
@@ -291,8 +291,8 @@ func (c *Context) SendICMPPacket(typ header.ICMPv4Type, code uint8, p1, p2 []byt
icmp.SetType(typ)
icmp.SetCode(code)
- copy(icmp[header.ICMPv4MinimumSize:], p1)
- copy(icmp[header.ICMPv4MinimumSize+len(p1):], p2)
+ copy(icmp[header.ICMPv4PayloadOffset:], p1)
+ copy(icmp[header.ICMPv4PayloadOffset+len(p1):], p2)
// Inject packet.
c.linkEP.Inject(ipv4.ProtocolNumber, buf.ToVectorisedView())
diff --git a/pkg/tcpip/transport/udp/endpoint.go b/pkg/tcpip/transport/udp/endpoint.go
index 99fdfb795..91f89a781 100644
--- a/pkg/tcpip/transport/udp/endpoint.go
+++ b/pkg/tcpip/transport/udp/endpoint.go
@@ -189,7 +189,6 @@ func (e *endpoint) Read(addr *tcpip.FullAddress) (buffer.View, tcpip.ControlMess
p := e.rcvList.Front()
e.rcvList.Remove(p)
e.rcvBufSize -= p.data.Size()
-
e.rcvMu.Unlock()
if addr != nil {
@@ -253,7 +252,7 @@ func (e *endpoint) connectRoute(nicid tcpip.NICID, addr tcpip.FullAddress) (stac
if nicid == 0 {
nicid = e.multicastNICID
}
- if localAddr == "" {
+ if localAddr == "" && nicid == 0 {
localAddr = e.multicastAddr
}
}
@@ -539,6 +538,22 @@ func (e *endpoint) SetSockOpt(opt interface{}) *tcpip.Error {
return nil
}
+// GetSockOptInt implements tcpip.Endpoint.GetSockOptInt.
+func (e *endpoint) GetSockOptInt(opt tcpip.SockOpt) (int, *tcpip.Error) {
+ switch opt {
+ case tcpip.ReceiveQueueSizeOption:
+ v := 0
+ e.rcvMu.Lock()
+ if !e.rcvList.Empty() {
+ p := e.rcvList.Front()
+ v = p.data.Size()
+ }
+ e.rcvMu.Unlock()
+ return v, nil
+ }
+ return -1, tcpip.ErrUnknownProtocolOption
+}
+
// GetSockOpt implements tcpip.Endpoint.GetSockOpt.
func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
switch o := opt.(type) {
@@ -573,17 +588,6 @@ func (e *endpoint) GetSockOpt(opt interface{}) *tcpip.Error {
}
return nil
- case *tcpip.ReceiveQueueSizeOption:
- e.rcvMu.Lock()
- if e.rcvList.Empty() {
- *o = 0
- } else {
- p := e.rcvList.Front()
- *o = tcpip.ReceiveQueueSizeOption(p.data.Size())
- }
- e.rcvMu.Unlock()
- return nil
-
case *tcpip.MulticastTTLOption:
e.mu.Lock()
*o = tcpip.MulticastTTLOption(e.multicastTTL)
@@ -671,6 +675,9 @@ func sendUDP(r *stack.Route, data buffer.VectorisedView, localPort, remotePort u
func (e *endpoint) checkV4Mapped(addr *tcpip.FullAddress, allowMismatch bool) (tcpip.NetworkProtocolNumber, *tcpip.Error) {
netProto := e.netProto
+ if len(addr.Addr) == 0 {
+ return netProto, nil
+ }
if header.IsV4MappedAddress(addr.Addr) {
// Fail if using a v4 mapped address on a v6only endpoint.
if e.v6only {
@@ -698,8 +705,44 @@ func (e *endpoint) checkV4Mapped(addr *tcpip.FullAddress, allowMismatch bool) (t
return netProto, nil
}
+func (e *endpoint) disconnect() *tcpip.Error {
+ e.mu.Lock()
+ defer e.mu.Unlock()
+
+ if e.state != stateConnected {
+ return nil
+ }
+ id := stack.TransportEndpointID{}
+ // Exclude ephemerally bound endpoints.
+ if e.bindNICID != 0 || e.id.LocalAddress == "" {
+ var err *tcpip.Error
+ id = stack.TransportEndpointID{
+ LocalPort: e.id.LocalPort,
+ LocalAddress: e.id.LocalAddress,
+ }
+ id, err = e.registerWithStack(e.regNICID, e.effectiveNetProtos, id)
+ if err != nil {
+ return err
+ }
+ e.state = stateBound
+ } else {
+ e.state = stateInitial
+ }
+
+ e.stack.UnregisterTransportEndpoint(e.regNICID, e.effectiveNetProtos, ProtocolNumber, e.id, e)
+ e.id = id
+ e.route.Release()
+ e.route = stack.Route{}
+ e.dstPort = 0
+
+ return nil
+}
+
// Connect connects the endpoint to its peer. Specifying a NIC is optional.
func (e *endpoint) Connect(addr tcpip.FullAddress) *tcpip.Error {
+ if addr.Addr == "" {
+ return e.disconnect()
+ }
if addr.Port == 0 {
// We don't support connecting to port zero.
return tcpip.ErrInvalidEndpointState
@@ -734,12 +777,16 @@ func (e *endpoint) Connect(addr tcpip.FullAddress) *tcpip.Error {
defer r.Release()
id := stack.TransportEndpointID{
- LocalAddress: r.LocalAddress,
+ LocalAddress: e.id.LocalAddress,
LocalPort: localPort,
RemotePort: addr.Port,
RemoteAddress: r.RemoteAddress,
}
+ if e.state == stateInitial {
+ id.LocalAddress = r.LocalAddress
+ }
+
// Even if we're connected, this endpoint can still be used to send
// packets on a different network protocol, so we register both even if
// v6only is set to false and this is an ipv6 endpoint.
diff --git a/pkg/tcpip/transport/udp/endpoint_state.go b/pkg/tcpip/transport/udp/endpoint_state.go
index 701bdd72b..18e786397 100644
--- a/pkg/tcpip/transport/udp/endpoint_state.go
+++ b/pkg/tcpip/transport/udp/endpoint_state.go
@@ -92,8 +92,6 @@ func (e *endpoint) afterLoad() {
if err != nil {
panic(*err)
}
-
- e.id.LocalAddress = e.route.LocalAddress
} else if len(e.id.LocalAddress) != 0 { // stateBound
if e.stack.CheckLocalAddress(e.regNICID, netProto, e.id.LocalAddress) == 0 {
panic(tcpip.ErrBadLocalAddress)
diff --git a/pkg/tcpip/transport/udp/udp_test.go b/pkg/tcpip/transport/udp/udp_test.go
index 75129a2ff..958d5712e 100644
--- a/pkg/tcpip/transport/udp/udp_test.go
+++ b/pkg/tcpip/transport/udp/udp_test.go
@@ -847,9 +847,7 @@ func TestWriteIncrementsPacketsSent(t *testing.T) {
}
}
-func TestTTL(t *testing.T) {
- payload := tcpip.SlicePayload(buffer.View(newPayload()))
-
+func setSockOptVariants(t *testing.T, optFunc func(*testing.T, string, tcpip.NetworkProtocolNumber, string)) {
for _, name := range []string{"v4", "v6", "dual"} {
t.Run(name, func(t *testing.T) {
var networkProtocolNumber tcpip.NetworkProtocolNumber
@@ -874,134 +872,219 @@ func TestTTL(t *testing.T) {
for _, variant := range variants {
t.Run(variant, func(t *testing.T) {
- for _, typ := range []string{"unicast", "multicast"} {
- t.Run(typ, func(t *testing.T) {
- var addr tcpip.Address
- var port uint16
- switch typ {
- case "unicast":
- port = testPort
- switch variant {
- case "v4":
- addr = testAddr
- case "mapped":
- addr = testV4MappedAddr
- case "v6":
- addr = testV6Addr
- default:
- t.Fatal("unknown test variant")
- }
- case "multicast":
- port = multicastPort
- switch variant {
- case "v4":
- addr = multicastAddr
- case "mapped":
- addr = multicastV4MappedAddr
- case "v6":
- addr = multicastV6Addr
- default:
- t.Fatal("unknown test variant")
- }
- default:
- t.Fatal("unknown test variant")
- }
-
- c := newDualTestContext(t, defaultMTU)
- defer c.cleanup()
-
- var err *tcpip.Error
- c.ep, err = c.s.NewEndpoint(udp.ProtocolNumber, networkProtocolNumber, &c.wq)
- if err != nil {
- c.t.Fatalf("NewEndpoint failed: %v", err)
- }
-
- switch name {
- case "v4":
- case "v6":
- if err := c.ep.SetSockOpt(tcpip.V6OnlyOption(1)); err != nil {
- c.t.Fatalf("SetSockOpt failed: %v", err)
- }
- case "dual":
- if err := c.ep.SetSockOpt(tcpip.V6OnlyOption(0)); err != nil {
- c.t.Fatalf("SetSockOpt failed: %v", err)
- }
- default:
- t.Fatal("unknown test variant")
- }
+ optFunc(t, name, networkProtocolNumber, variant)
+ })
+ }
+ })
+ }
+}
- const multicastTTL = 42
- if err := c.ep.SetSockOpt(tcpip.MulticastTTLOption(multicastTTL)); err != nil {
- c.t.Fatalf("SetSockOpt failed: %v", err)
- }
+func TestTTL(t *testing.T) {
+ payload := tcpip.SlicePayload(buffer.View(newPayload()))
- n, _, err := c.ep.Write(payload, tcpip.WriteOptions{To: &tcpip.FullAddress{Addr: addr, Port: port}})
- if err != nil {
- c.t.Fatalf("Write failed: %v", err)
- }
- if n != uintptr(len(payload)) {
- c.t.Fatalf("got c.ep.Write(...) = %d, want = %d", n, len(payload))
- }
+ setSockOptVariants(t, func(t *testing.T, name string, networkProtocolNumber tcpip.NetworkProtocolNumber, variant string) {
+ for _, typ := range []string{"unicast", "multicast"} {
+ t.Run(typ, func(t *testing.T) {
+ var addr tcpip.Address
+ var port uint16
+ switch typ {
+ case "unicast":
+ port = testPort
+ switch variant {
+ case "v4":
+ addr = testAddr
+ case "mapped":
+ addr = testV4MappedAddr
+ case "v6":
+ addr = testV6Addr
+ default:
+ t.Fatal("unknown test variant")
+ }
+ case "multicast":
+ port = multicastPort
+ switch variant {
+ case "v4":
+ addr = multicastAddr
+ case "mapped":
+ addr = multicastV4MappedAddr
+ case "v6":
+ addr = multicastV6Addr
+ default:
+ t.Fatal("unknown test variant")
+ }
+ default:
+ t.Fatal("unknown test variant")
+ }
+
+ c := newDualTestContext(t, defaultMTU)
+ defer c.cleanup()
+
+ var err *tcpip.Error
+ c.ep, err = c.s.NewEndpoint(udp.ProtocolNumber, networkProtocolNumber, &c.wq)
+ if err != nil {
+ c.t.Fatalf("NewEndpoint failed: %v", err)
+ }
+
+ switch name {
+ case "v4":
+ case "v6":
+ if err := c.ep.SetSockOpt(tcpip.V6OnlyOption(1)); err != nil {
+ c.t.Fatalf("SetSockOpt failed: %v", err)
+ }
+ case "dual":
+ if err := c.ep.SetSockOpt(tcpip.V6OnlyOption(0)); err != nil {
+ c.t.Fatalf("SetSockOpt failed: %v", err)
+ }
+ default:
+ t.Fatal("unknown test variant")
+ }
+
+ const multicastTTL = 42
+ if err := c.ep.SetSockOpt(tcpip.MulticastTTLOption(multicastTTL)); err != nil {
+ c.t.Fatalf("SetSockOpt failed: %v", err)
+ }
+
+ n, _, err := c.ep.Write(payload, tcpip.WriteOptions{To: &tcpip.FullAddress{Addr: addr, Port: port}})
+ if err != nil {
+ c.t.Fatalf("Write failed: %v", err)
+ }
+ if n != uintptr(len(payload)) {
+ c.t.Fatalf("got c.ep.Write(...) = %d, want = %d", n, len(payload))
+ }
+
+ checkerFn := checker.IPv4
+ switch variant {
+ case "v4", "mapped":
+ case "v6":
+ checkerFn = checker.IPv6
+ default:
+ t.Fatal("unknown test variant")
+ }
+ var wantTTL uint8
+ var multicast bool
+ switch typ {
+ case "unicast":
+ multicast = false
+ switch variant {
+ case "v4", "mapped":
+ ep, err := ipv4.NewProtocol().NewEndpoint(0, "", nil, nil, nil)
+ if err != nil {
+ t.Fatal(err)
+ }
+ wantTTL = ep.DefaultTTL()
+ ep.Close()
+ case "v6":
+ ep, err := ipv6.NewProtocol().NewEndpoint(0, "", nil, nil, nil)
+ if err != nil {
+ t.Fatal(err)
+ }
+ wantTTL = ep.DefaultTTL()
+ ep.Close()
+ default:
+ t.Fatal("unknown test variant")
+ }
+ case "multicast":
+ wantTTL = multicastTTL
+ multicast = true
+ default:
+ t.Fatal("unknown test variant")
+ }
+
+ var networkProtocolNumber tcpip.NetworkProtocolNumber
+ switch variant {
+ case "v4", "mapped":
+ networkProtocolNumber = ipv4.ProtocolNumber
+ case "v6":
+ networkProtocolNumber = ipv6.ProtocolNumber
+ default:
+ t.Fatal("unknown test variant")
+ }
+
+ b := c.getPacket(networkProtocolNumber, multicast)
+ checkerFn(c.t, b,
+ checker.TTL(wantTTL),
+ checker.UDP(
+ checker.DstPort(port),
+ ),
+ )
+ })
+ }
+ })
+}
- checkerFn := checker.IPv4
- switch variant {
- case "v4", "mapped":
- case "v6":
- checkerFn = checker.IPv6
- default:
- t.Fatal("unknown test variant")
- }
- var wantTTL uint8
- var multicast bool
- switch typ {
- case "unicast":
- multicast = false
- switch variant {
- case "v4", "mapped":
- ep, err := ipv4.NewProtocol().NewEndpoint(0, "", nil, nil, nil)
- if err != nil {
- t.Fatal(err)
- }
- wantTTL = ep.DefaultTTL()
- ep.Close()
- case "v6":
- ep, err := ipv6.NewProtocol().NewEndpoint(0, "", nil, nil, nil)
- if err != nil {
- t.Fatal(err)
- }
- wantTTL = ep.DefaultTTL()
- ep.Close()
- default:
- t.Fatal("unknown test variant")
- }
- case "multicast":
- wantTTL = multicastTTL
- multicast = true
- default:
- t.Fatal("unknown test variant")
+func TestMulticastInterfaceOption(t *testing.T) {
+ setSockOptVariants(t, func(t *testing.T, name string, networkProtocolNumber tcpip.NetworkProtocolNumber, variant string) {
+ for _, bindTyp := range []string{"bound", "unbound"} {
+ t.Run(bindTyp, func(t *testing.T) {
+ for _, optTyp := range []string{"use local-addr", "use NICID", "use local-addr and NIC"} {
+ t.Run(optTyp, func(t *testing.T) {
+ var mcastAddr, localIfAddr tcpip.Address
+ switch variant {
+ case "v4":
+ mcastAddr = multicastAddr
+ localIfAddr = stackAddr
+ case "mapped":
+ mcastAddr = multicastV4MappedAddr
+ localIfAddr = stackAddr
+ case "v6":
+ mcastAddr = multicastV6Addr
+ localIfAddr = stackV6Addr
+ default:
+ t.Fatal("unknown test variant")
+ }
+
+ var ifoptSet tcpip.MulticastInterfaceOption
+ switch optTyp {
+ case "use local-addr":
+ ifoptSet.InterfaceAddr = localIfAddr
+ case "use NICID":
+ ifoptSet.NIC = 1
+ case "use local-addr and NIC":
+ ifoptSet.InterfaceAddr = localIfAddr
+ ifoptSet.NIC = 1
+ default:
+ t.Fatal("unknown test variant")
+ }
+
+ c := newDualTestContext(t, defaultMTU)
+ defer c.cleanup()
+
+ var err *tcpip.Error
+ c.ep, err = c.s.NewEndpoint(udp.ProtocolNumber, networkProtocolNumber, &c.wq)
+ if err != nil {
+ c.t.Fatalf("NewEndpoint failed: %v", err)
+ }
+
+ if bindTyp == "bound" {
+ // Bind the socket by connecting to the multicast address.
+ // This may have an influence on how the multicast interface
+ // is set.
+ addr := tcpip.FullAddress{
+ Addr: mcastAddr,
+ Port: multicastPort,
}
-
- var networkProtocolNumber tcpip.NetworkProtocolNumber
- switch variant {
- case "v4", "mapped":
- networkProtocolNumber = ipv4.ProtocolNumber
- case "v6":
- networkProtocolNumber = ipv6.ProtocolNumber
- default:
- t.Fatal("unknown test variant")
+ if err := c.ep.Connect(addr); err != nil {
+ c.t.Fatalf("Connect failed: %v", err)
}
-
- b := c.getPacket(networkProtocolNumber, multicast)
- checkerFn(c.t, b,
- checker.TTL(wantTTL),
- checker.UDP(
- checker.DstPort(port),
- ),
- )
- })
- }
- })
- }
- })
- }
+ }
+
+ if err := c.ep.SetSockOpt(ifoptSet); err != nil {
+ c.t.Fatalf("SetSockOpt failed: %v", err)
+ }
+
+ // Verify multicast interface addr and NIC were set correctly.
+ // Note that NIC must be 1 since this is our outgoing interface.
+ ifoptWant := tcpip.MulticastInterfaceOption{NIC: 1, InterfaceAddr: ifoptSet.InterfaceAddr}
+ var ifoptGot tcpip.MulticastInterfaceOption
+ if err := c.ep.GetSockOpt(&ifoptGot); err != nil {
+ c.t.Fatalf("GetSockOpt failed: %v", err)
+ }
+ if ifoptGot != ifoptWant {
+ c.t.Errorf("got GetSockOpt() = %#v, want = %#v", ifoptGot, ifoptWant)
+ }
+ })
+ }
+ })
+ }
+ })
}
generated by cgit v1.2.3 (git 2.39.1) at 2025-04-12 14:28:18 +0000