go-ethereum/p2p/peer.go
Felix Lange 995fab2ebc p2p: fix yet another disconnect hang
Peer.readLoop will only terminate if the connection is closed. Fix the
hang by closing the connection before waiting for readLoop to terminate.

This also removes the british disconnect procedure where we're waiting
for the remote end to close the connection. I have confirmed with
@subtly that cpp-ethereum doesn't adhere to it either.
2015-04-13 17:34:08 +02:00

306 lines
7.0 KiB
Go

package p2p
import (
"errors"
"fmt"
"io"
"net"
"sort"
"sync"
"time"
"github.com/ethereum/go-ethereum/logger"
"github.com/ethereum/go-ethereum/p2p/discover"
"github.com/ethereum/go-ethereum/rlp"
)
const (
baseProtocolVersion = 3
baseProtocolLength = uint64(16)
baseProtocolMaxMsgSize = 10 * 1024 * 1024
pingInterval = 15 * time.Second
)
const (
// devp2p message codes
handshakeMsg = 0x00
discMsg = 0x01
pingMsg = 0x02
pongMsg = 0x03
getPeersMsg = 0x04
peersMsg = 0x05
)
// Peer represents a connected remote node.
type Peer struct {
// Peers have all the log methods.
// Use them to display messages related to the peer.
*logger.Logger
conn net.Conn
rw *conn
running map[string]*protoRW
wg sync.WaitGroup
protoErr chan error
closed chan struct{}
disc chan DiscReason
}
// NewPeer returns a peer for testing purposes.
func NewPeer(id discover.NodeID, name string, caps []Cap) *Peer {
pipe, _ := net.Pipe()
msgpipe, _ := MsgPipe()
conn := &conn{msgpipe, &protoHandshake{ID: id, Name: name, Caps: caps}}
peer := newPeer(pipe, conn, nil)
close(peer.closed) // ensures Disconnect doesn't block
return peer
}
// ID returns the node's public key.
func (p *Peer) ID() discover.NodeID {
return p.rw.ID
}
// Name returns the node name that the remote node advertised.
func (p *Peer) Name() string {
return p.rw.Name
}
// Caps returns the capabilities (supported subprotocols) of the remote peer.
func (p *Peer) Caps() []Cap {
// TODO: maybe return copy
return p.rw.Caps
}
// RemoteAddr returns the remote address of the network connection.
func (p *Peer) RemoteAddr() net.Addr {
return p.conn.RemoteAddr()
}
// LocalAddr returns the local address of the network connection.
func (p *Peer) LocalAddr() net.Addr {
return p.conn.LocalAddr()
}
// Disconnect terminates the peer connection with the given reason.
// It returns immediately and does not wait until the connection is closed.
func (p *Peer) Disconnect(reason DiscReason) {
select {
case p.disc <- reason:
case <-p.closed:
}
}
// String implements fmt.Stringer.
func (p *Peer) String() string {
return fmt.Sprintf("Peer %.8x %v", p.rw.ID[:], p.RemoteAddr())
}
func newPeer(fd net.Conn, conn *conn, protocols []Protocol) *Peer {
logtag := fmt.Sprintf("Peer %.8x %v", conn.ID[:], fd.RemoteAddr())
protomap := matchProtocols(protocols, conn.Caps, conn)
p := &Peer{
Logger: logger.NewLogger(logtag),
conn: fd,
rw: conn,
running: protomap,
disc: make(chan DiscReason),
protoErr: make(chan error, len(protomap)+1), // protocols + pingLoop
closed: make(chan struct{}),
}
return p
}
func (p *Peer) run() DiscReason {
readErr := make(chan error, 1)
p.wg.Add(2)
go p.readLoop(readErr)
go p.pingLoop()
p.startProtocols()
// Wait for an error or disconnect.
var reason DiscReason
select {
case err := <-readErr:
if r, ok := err.(DiscReason); ok {
reason = r
} else {
// Note: We rely on protocols to abort if there is a write
// error. It might be more robust to handle them here as well.
p.DebugDetailf("Read error: %v\n", err)
reason = DiscNetworkError
}
case err := <-p.protoErr:
reason = discReasonForError(err)
case reason = <-p.disc:
}
close(p.closed)
p.politeDisconnect(reason)
p.wg.Wait()
p.Debugf("Disconnected: %v\n", reason)
return reason
}
func (p *Peer) politeDisconnect(reason DiscReason) {
if reason != DiscNetworkError {
SendItems(p.rw, discMsg, uint(reason))
}
p.conn.Close()
}
func (p *Peer) pingLoop() {
ping := time.NewTicker(pingInterval)
defer p.wg.Done()
defer ping.Stop()
for {
select {
case <-ping.C:
if err := SendItems(p.rw, pingMsg); err != nil {
p.protoErr <- err
return
}
case <-p.closed:
return
}
}
}
func (p *Peer) readLoop(errc chan<- error) {
defer p.wg.Done()
for {
msg, err := p.rw.ReadMsg()
if err != nil {
errc <- err
return
}
if err = p.handle(msg); err != nil {
errc <- err
return
}
}
}
func (p *Peer) handle(msg Msg) error {
switch {
case msg.Code == pingMsg:
msg.Discard()
go SendItems(p.rw, pongMsg)
case msg.Code == discMsg:
var reason [1]DiscReason
// This is the last message. We don't need to discard or
// check errors because, the connection will be closed after it.
rlp.Decode(msg.Payload, &reason)
p.Debugf("Disconnect requested: %v\n", reason[0])
return DiscRequested
case msg.Code < baseProtocolLength:
// ignore other base protocol messages
return msg.Discard()
default:
// it's a subprotocol message
proto, err := p.getProto(msg.Code)
if err != nil {
return fmt.Errorf("msg code out of range: %v", msg.Code)
}
select {
case proto.in <- msg:
return nil
case <-p.closed:
return io.EOF
}
}
return nil
}
// matchProtocols creates structures for matching named subprotocols.
func matchProtocols(protocols []Protocol, caps []Cap, rw MsgReadWriter) map[string]*protoRW {
sort.Sort(capsByName(caps))
offset := baseProtocolLength
result := make(map[string]*protoRW)
outer:
for _, cap := range caps {
for _, proto := range protocols {
if proto.Name == cap.Name && proto.Version == cap.Version && result[cap.Name] == nil {
result[cap.Name] = &protoRW{Protocol: proto, offset: offset, in: make(chan Msg), w: rw}
offset += proto.Length
continue outer
}
}
}
return result
}
func (p *Peer) startProtocols() {
p.wg.Add(len(p.running))
for _, proto := range p.running {
proto := proto
proto.closed = p.closed
p.DebugDetailf("Starting protocol %s/%d\n", proto.Name, proto.Version)
go func() {
err := proto.Run(p, proto)
if err == nil {
p.DebugDetailf("Protocol %s/%d returned\n", proto.Name, proto.Version)
err = errors.New("protocol returned")
} else {
p.DebugDetailf("Protocol %s/%d error: %v\n", proto.Name, proto.Version, err)
}
p.protoErr <- err
p.wg.Done()
}()
}
}
// getProto finds the protocol responsible for handling
// the given message code.
func (p *Peer) getProto(code uint64) (*protoRW, error) {
for _, proto := range p.running {
if code >= proto.offset && code < proto.offset+proto.Length {
return proto, nil
}
}
return nil, newPeerError(errInvalidMsgCode, "%d", code)
}
// writeProtoMsg sends the given message on behalf of the given named protocol.
// this exists because of Server.Broadcast.
func (p *Peer) writeProtoMsg(protoName string, msg Msg) error {
proto, ok := p.running[protoName]
if !ok {
return fmt.Errorf("protocol %s not handled by peer", protoName)
}
if msg.Code >= proto.Length {
return newPeerError(errInvalidMsgCode, "code %x is out of range for protocol %q", msg.Code, protoName)
}
msg.Code += proto.offset
return p.rw.WriteMsg(msg)
}
type protoRW struct {
Protocol
in chan Msg
closed <-chan struct{}
offset uint64
w MsgWriter
}
func (rw *protoRW) WriteMsg(msg Msg) error {
if msg.Code >= rw.Length {
return newPeerError(errInvalidMsgCode, "not handled")
}
msg.Code += rw.offset
return rw.w.WriteMsg(msg)
}
func (rw *protoRW) ReadMsg() (Msg, error) {
select {
case msg := <-rw.in:
msg.Code -= rw.offset
return msg, nil
case <-rw.closed:
return Msg{}, io.EOF
}
}