syncthing/lib/protocol/protocol.go
Jakob Borg e52be3d83e lib/connections, lib/model: Refactor connection close handling (fixes #3466)
So there were some issues here. The main problem was that
model.Close(deviceID) was overloaded to mean "the connection was closed
by the protocol layer" and "i want to close this connection". That meant
it could get called twice - once *to* close the connection and then once
more when the connection *was* closed.

After this refactor there is instead a Closed(conn) method that is the
callback. I didn't need to change the parameter in the end, but I think
it's clearer what it means when it takes the connection that was closed
instead of a device ID. To close a connection, the new close(deviceID)
method is used instead, which only closes the underlying connection and
leaves the cleanup to the Closed() callback.

I also changed how we do connection switching. Instead of the connection
service calling close and then adding the connection, it just adds the
new connection. The model knows that it already has a connection and
makes sure to close and clean out that one before adding the new
connection.

To make sure to sequence this properly I added a new map of channels
that get created on connection add and closed by Closed(), so that
AddConnection() can do the close and wait for the cleanup to happen
before proceeding.

GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/3490
2016-08-10 09:37:32 +00:00

822 lines
19 KiB
Go

// Copyright (C) 2014 The Protocol Authors.
package protocol
import (
"encoding/binary"
"errors"
"fmt"
"io"
"sync"
"time"
lz4 "github.com/bkaradzic/go-lz4"
)
const (
// BlockSize is the standard ata block size (128 KiB)
BlockSize = 128 << 10
// MaxMessageLen is the largest message size allowed on the wire. (500 MB)
MaxMessageLen = 500 * 1000 * 1000
hdrSize = 6
)
const (
stateInitial = iota
stateReady
)
// Request message flags
const (
FlagFromTemporary uint32 = 1 << iota
)
// ClusterConfigMessage.Folders flags
const (
FlagFolderReadOnly uint32 = 1 << 0
FlagFolderIgnorePerms = 1 << 1
FlagFolderIgnoreDelete = 1 << 2
FlagFolderDisabledTempIndexes = 1 << 3
FlagFolderAll = 1<<4 - 1
)
// ClusterConfigMessage.Folders.Devices flags
const (
FlagShareTrusted uint32 = 1 << 0
FlagShareReadOnly = 1 << 1
FlagIntroducer = 1 << 2
FlagShareBits = 0x000000ff
)
var (
ErrClosed = errors.New("connection closed")
ErrTimeout = errors.New("read timeout")
ErrSwitchingConnections = errors.New("switching connections")
errUnknownMessage = errors.New("unknown message")
)
type Model interface {
// An index was received from the peer device
Index(deviceID DeviceID, folder string, files []FileInfo)
// An index update was received from the peer device
IndexUpdate(deviceID DeviceID, folder string, files []FileInfo)
// A request was made by the peer device
Request(deviceID DeviceID, folder string, name string, offset int64, hash []byte, fromTemporary bool, buf []byte) error
// A cluster configuration message was received
ClusterConfig(deviceID DeviceID, config ClusterConfig)
// The peer device closed the connection
Closed(conn Connection, err error)
// The peer device sent progress updates for the files it is currently downloading
DownloadProgress(deviceID DeviceID, folder string, updates []FileDownloadProgressUpdate)
}
type Connection interface {
Start()
ID() DeviceID
Name() string
Index(folder string, files []FileInfo) error
IndexUpdate(folder string, files []FileInfo) error
Request(folder string, name string, offset int64, size int, hash []byte, fromTemporary bool) ([]byte, error)
ClusterConfig(config ClusterConfig)
DownloadProgress(folder string, updates []FileDownloadProgressUpdate)
Statistics() Statistics
Closed() bool
}
type rawConnection struct {
id DeviceID
name string
receiver Model
cr *countingReader
cw *countingWriter
awaiting map[int32]chan asyncResult
awaitingMut sync.Mutex
idxMut sync.Mutex // ensures serialization of Index calls
nextID int32
nextIDMut sync.Mutex
outbox chan asyncMessage
closed chan struct{}
once sync.Once
pool sync.Pool
compression Compression
}
type asyncResult struct {
val []byte
err error
}
type message interface {
ProtoSize() int
Marshal() ([]byte, error)
MarshalTo([]byte) (int, error)
Unmarshal([]byte) error
}
type asyncMessage struct {
msg message
done chan struct{} // done closes when we're done marshalling the message and it's contents can be reused
}
const (
// PingSendInterval is how often we make sure to send a message, by
// triggering pings if necessary.
PingSendInterval = 90 * time.Second
// ReceiveTimeout is the longest we'll wait for a message from the other
// side before closing the connection.
ReceiveTimeout = 300 * time.Second
)
// A buffer pool for global use. We don't allocate smaller buffers than 64k,
// in the hope of being able to reuse them later.
var buffers = bufferPool{
minSize: 64 << 10,
}
func NewConnection(deviceID DeviceID, reader io.Reader, writer io.Writer, receiver Model, name string, compress Compression) Connection {
cr := &countingReader{Reader: reader}
cw := &countingWriter{Writer: writer}
c := rawConnection{
id: deviceID,
name: name,
receiver: nativeModel{receiver},
cr: cr,
cw: cw,
awaiting: make(map[int32]chan asyncResult),
outbox: make(chan asyncMessage),
closed: make(chan struct{}),
pool: sync.Pool{
New: func() interface{} {
return make([]byte, BlockSize)
},
},
compression: compress,
}
return wireFormatConnection{&c}
}
// Start creates the goroutines for sending and receiving of messages. It must
// be called exactly once after creating a connection.
func (c *rawConnection) Start() {
go c.readerLoop()
go c.writerLoop()
go c.pingSender()
go c.pingReceiver()
}
func (c *rawConnection) ID() DeviceID {
return c.id
}
func (c *rawConnection) Name() string {
return c.name
}
// Index writes the list of file information to the connected peer device
func (c *rawConnection) Index(folder string, idx []FileInfo) error {
select {
case <-c.closed:
return ErrClosed
default:
}
c.idxMut.Lock()
c.send(&Index{
Folder: folder,
Files: idx,
}, nil)
c.idxMut.Unlock()
return nil
}
// IndexUpdate writes the list of file information to the connected peer device as an update
func (c *rawConnection) IndexUpdate(folder string, idx []FileInfo) error {
select {
case <-c.closed:
return ErrClosed
default:
}
c.idxMut.Lock()
c.send(&IndexUpdate{
Folder: folder,
Files: idx,
}, nil)
c.idxMut.Unlock()
return nil
}
// Request returns the bytes for the specified block after fetching them from the connected peer.
func (c *rawConnection) Request(folder string, name string, offset int64, size int, hash []byte, fromTemporary bool) ([]byte, error) {
c.nextIDMut.Lock()
id := c.nextID
c.nextID++
c.nextIDMut.Unlock()
c.awaitingMut.Lock()
if _, ok := c.awaiting[id]; ok {
panic("id taken")
}
rc := make(chan asyncResult, 1)
c.awaiting[id] = rc
c.awaitingMut.Unlock()
ok := c.send(&Request{
ID: id,
Folder: folder,
Name: name,
Offset: offset,
Size: int32(size),
Hash: hash,
FromTemporary: fromTemporary,
}, nil)
if !ok {
return nil, ErrClosed
}
res, ok := <-rc
if !ok {
return nil, ErrClosed
}
return res.val, res.err
}
// ClusterConfig send the cluster configuration message to the peer and returns any error
func (c *rawConnection) ClusterConfig(config ClusterConfig) {
c.send(&config, nil)
}
func (c *rawConnection) Closed() bool {
select {
case <-c.closed:
return true
default:
return false
}
}
// DownloadProgress sends the progress updates for the files that are currently being downloaded.
func (c *rawConnection) DownloadProgress(folder string, updates []FileDownloadProgressUpdate) {
c.send(&DownloadProgress{
Folder: folder,
Updates: updates,
}, nil)
}
func (c *rawConnection) ping() bool {
return c.send(&Ping{}, nil)
}
func (c *rawConnection) readerLoop() (err error) {
defer func() {
c.close(err)
}()
state := stateInitial
for {
select {
case <-c.closed:
return ErrClosed
default:
}
msg, err := c.readMessage()
if err == errUnknownMessage {
// Unknown message types are skipped, for future extensibility.
continue
}
if err != nil {
return err
}
switch msg := msg.(type) {
case *ClusterConfig:
l.Debugln("read ClusterConfig message")
if state != stateInitial {
return fmt.Errorf("protocol error: cluster config message in state %d", state)
}
c.receiver.ClusterConfig(c.id, *msg)
state = stateReady
case *Index:
l.Debugln("read Index message")
if state != stateReady {
return fmt.Errorf("protocol error: index message in state %d", state)
}
c.handleIndex(*msg)
state = stateReady
case *IndexUpdate:
l.Debugln("read IndexUpdate message")
if state != stateReady {
return fmt.Errorf("protocol error: index update message in state %d", state)
}
c.handleIndexUpdate(*msg)
state = stateReady
case *Request:
l.Debugln("read Request message")
if state != stateReady {
return fmt.Errorf("protocol error: request message in state %d", state)
}
// Requests are handled asynchronously
go c.handleRequest(*msg)
case *Response:
l.Debugln("read Response message")
if state != stateReady {
return fmt.Errorf("protocol error: response message in state %d", state)
}
c.handleResponse(*msg)
case *DownloadProgress:
l.Debugln("read DownloadProgress message")
if state != stateReady {
return fmt.Errorf("protocol error: response message in state %d", state)
}
c.receiver.DownloadProgress(c.id, msg.Folder, msg.Updates)
case *Ping:
l.Debugln("read Ping message")
if state != stateReady {
return fmt.Errorf("protocol error: ping message in state %d", state)
}
// Nothing
case *Close:
l.Debugln("read Close message")
return errors.New(msg.Reason)
default:
l.Debugf("read unknown message: %+T", msg)
return fmt.Errorf("protocol error: %s: unknown or empty message", c.id)
}
}
}
func (c *rawConnection) readMessage() (message, error) {
hdr, err := c.readHeader()
if err != nil {
return nil, err
}
return c.readMessageAfterHeader(hdr)
}
func (c *rawConnection) readMessageAfterHeader(hdr Header) (message, error) {
// First comes a 4 byte message length
buf := buffers.get(4)
if _, err := io.ReadFull(c.cr, buf); err != nil {
return nil, fmt.Errorf("reading message length: %v", err)
}
msgLen := int32(binary.BigEndian.Uint32(buf))
if msgLen < 0 {
return nil, fmt.Errorf("negative message length %d", msgLen)
}
// Then comes the message
buf = buffers.upgrade(buf, int(msgLen))
if _, err := io.ReadFull(c.cr, buf); err != nil {
return nil, fmt.Errorf("reading message: %v", err)
}
// ... which might be compressed
switch hdr.Compression {
case MessageCompressionNone:
// Nothing
case MessageCompressionLZ4:
decomp, err := c.lz4Decompress(buf)
buffers.put(buf)
if err != nil {
return nil, fmt.Errorf("decompressing message: %v", err)
}
buf = decomp
default:
return nil, fmt.Errorf("unknown message compression %d", hdr.Compression)
}
// ... and is then unmarshalled
msg, err := c.newMessage(hdr.Type)
if err != nil {
return nil, err
}
if err := msg.Unmarshal(buf); err != nil {
return nil, fmt.Errorf("unmarshalling message: %v", err)
}
buffers.put(buf)
return msg, nil
}
func (c *rawConnection) readHeader() (Header, error) {
// First comes a 2 byte header length
buf := buffers.get(2)
if _, err := io.ReadFull(c.cr, buf); err != nil {
return Header{}, fmt.Errorf("reading length: %v", err)
}
hdrLen := int16(binary.BigEndian.Uint16(buf))
if hdrLen < 0 {
return Header{}, fmt.Errorf("negative header length %d", hdrLen)
}
// Then comes the header
buf = buffers.upgrade(buf, int(hdrLen))
if _, err := io.ReadFull(c.cr, buf); err != nil {
return Header{}, fmt.Errorf("reading header: %v", err)
}
var hdr Header
if err := hdr.Unmarshal(buf); err != nil {
return Header{}, fmt.Errorf("unmarshalling header: %v", err)
}
buffers.put(buf)
return hdr, nil
}
func (c *rawConnection) handleIndex(im Index) {
l.Debugf("Index(%v, %v, %d file)", c.id, im.Folder, len(im.Files))
c.receiver.Index(c.id, im.Folder, filterIndexMessageFiles(im.Files))
}
func (c *rawConnection) handleIndexUpdate(im IndexUpdate) {
l.Debugf("queueing IndexUpdate(%v, %v, %d files)", c.id, im.Folder, len(im.Files))
c.receiver.IndexUpdate(c.id, im.Folder, filterIndexMessageFiles(im.Files))
}
func filterIndexMessageFiles(fs []FileInfo) []FileInfo {
var out []FileInfo
for i, f := range fs {
switch f.Name {
case "", ".", "..", "/": // A few obviously invalid filenames
l.Infof("Dropping invalid filename %q from incoming index", f.Name)
if out == nil {
// Most incoming updates won't contain anything invalid, so we
// delay the allocation and copy to output slice until we
// really need to do it, then copy all the so var valid files
// to it.
out = make([]FileInfo, i, len(fs)-1)
copy(out, fs)
}
default:
if out != nil {
out = append(out, f)
}
}
}
if out != nil {
return out
}
return fs
}
func (c *rawConnection) handleRequest(req Request) {
size := int(req.Size)
usePool := size <= BlockSize
var buf []byte
var done chan struct{}
if usePool {
buf = c.pool.Get().([]byte)[:size]
done = make(chan struct{})
} else {
buf = make([]byte, size)
}
err := c.receiver.Request(c.id, req.Folder, req.Name, int64(req.Offset), req.Hash, req.FromTemporary, buf)
if err != nil {
c.send(&Response{
ID: req.ID,
Data: nil,
Code: errorToCode(err),
}, done)
} else {
c.send(&Response{
ID: req.ID,
Data: buf,
Code: errorToCode(err),
}, done)
}
if usePool {
<-done
c.pool.Put(buf)
}
}
func (c *rawConnection) handleResponse(resp Response) {
c.awaitingMut.Lock()
if rc := c.awaiting[resp.ID]; rc != nil {
delete(c.awaiting, resp.ID)
rc <- asyncResult{resp.Data, codeToError(resp.Code)}
close(rc)
}
c.awaitingMut.Unlock()
}
func (c *rawConnection) send(msg message, done chan struct{}) bool {
select {
case c.outbox <- asyncMessage{msg, done}:
return true
case <-c.closed:
return false
}
}
func (c *rawConnection) writerLoop() {
for {
select {
case hm := <-c.outbox:
if err := c.writeMessage(hm); err != nil {
c.close(err)
return
}
case <-c.closed:
return
}
}
}
func (c *rawConnection) writeMessage(hm asyncMessage) error {
if c.shouldCompressMessage(hm.msg) {
return c.writeCompressedMessage(hm)
}
return c.writeUncompressedMessage(hm)
}
func (c *rawConnection) writeCompressedMessage(hm asyncMessage) error {
size := hm.msg.ProtoSize()
buf := buffers.get(size)
if _, err := hm.msg.MarshalTo(buf); err != nil {
return fmt.Errorf("marshalling message: %v", err)
}
if hm.done != nil {
close(hm.done)
}
compressed, err := c.lz4Compress(buf)
if err != nil {
return fmt.Errorf("compressing message: %v", err)
}
hdr := Header{
Type: c.typeOf(hm.msg),
Compression: MessageCompressionLZ4,
}
hdrSize := hdr.ProtoSize()
if hdrSize > 1<<16-1 {
panic("impossibly large header")
}
totSize := 2 + hdrSize + 4 + len(compressed)
buf = buffers.upgrade(buf, totSize)
// Header length
binary.BigEndian.PutUint16(buf, uint16(hdrSize))
// Header
if _, err := hdr.MarshalTo(buf[2:]); err != nil {
return fmt.Errorf("marshalling header: %v", err)
}
// Message length
binary.BigEndian.PutUint32(buf[2+hdrSize:], uint32(len(compressed)))
// Message
copy(buf[2+hdrSize+4:], compressed)
buffers.put(compressed)
n, err := c.cw.Write(buf)
buffers.put(buf)
l.Debugf("wrote %d bytes on the wire (2 bytes length, %d bytes header, 4 bytes message length, %d bytes message (%d uncompressed)), err=%v", n, hdrSize, len(compressed), size, err)
if err != nil {
return fmt.Errorf("writing message: %v", err)
}
return nil
}
func (c *rawConnection) writeUncompressedMessage(hm asyncMessage) error {
size := hm.msg.ProtoSize()
hdr := Header{
Type: c.typeOf(hm.msg),
}
hdrSize := hdr.ProtoSize()
if hdrSize > 1<<16-1 {
panic("impossibly large header")
}
totSize := 2 + hdrSize + 4 + size
buf := buffers.get(totSize)
// Header length
binary.BigEndian.PutUint16(buf, uint16(hdrSize))
// Header
if _, err := hdr.MarshalTo(buf[2:]); err != nil {
return fmt.Errorf("marshalling header: %v", err)
}
// Message length
binary.BigEndian.PutUint32(buf[2+hdrSize:], uint32(size))
// Message
if _, err := hm.msg.MarshalTo(buf[2+hdrSize+4:]); err != nil {
return fmt.Errorf("marshalling message: %v", err)
}
if hm.done != nil {
close(hm.done)
}
n, err := c.cw.Write(buf[:totSize])
buffers.put(buf)
l.Debugf("wrote %d bytes on the wire (2 bytes length, %d bytes header, 4 bytes message length, %d bytes message), err=%v", n, hdrSize, size, err)
if err != nil {
return fmt.Errorf("writing message: %v", err)
}
return nil
}
func (c *rawConnection) typeOf(msg message) MessageType {
switch msg.(type) {
case *ClusterConfig:
return messageTypeClusterConfig
case *Index:
return messageTypeIndex
case *IndexUpdate:
return messageTypeIndexUpdate
case *Request:
return messageTypeRequest
case *Response:
return messageTypeResponse
case *DownloadProgress:
return messageTypeDownloadProgress
case *Ping:
return messageTypePing
case *Close:
return messageTypeClose
default:
panic("bug: unknown message type")
}
}
func (c *rawConnection) newMessage(t MessageType) (message, error) {
switch t {
case messageTypeClusterConfig:
return new(ClusterConfig), nil
case messageTypeIndex:
return new(Index), nil
case messageTypeIndexUpdate:
return new(IndexUpdate), nil
case messageTypeRequest:
return new(Request), nil
case messageTypeResponse:
return new(Response), nil
case messageTypeDownloadProgress:
return new(DownloadProgress), nil
case messageTypePing:
return new(Ping), nil
case messageTypeClose:
return new(Close), nil
default:
return nil, errUnknownMessage
}
}
func (c *rawConnection) shouldCompressMessage(msg message) bool {
switch c.compression {
case CompressNever:
return false
case CompressAlways:
// Use compression for large enough messages
return msg.ProtoSize() >= compressionThreshold
case CompressMetadata:
_, isResponse := msg.(*Response)
// Compress if it's large enough and not a response message
return !isResponse && msg.ProtoSize() >= compressionThreshold
default:
panic("unknown compression setting")
}
}
func (c *rawConnection) close(err error) {
c.once.Do(func() {
l.Debugln("close due to", err)
close(c.closed)
c.awaitingMut.Lock()
for i, ch := range c.awaiting {
if ch != nil {
close(ch)
delete(c.awaiting, i)
}
}
c.awaitingMut.Unlock()
c.receiver.Closed(c, err)
})
}
// The pingSender makes sure that we've sent a message within the last
// PingSendInterval. If we already have something sent in the last
// PingSendInterval/2, we do nothing. Otherwise we send a ping message. This
// results in an effecting ping interval of somewhere between
// PingSendInterval/2 and PingSendInterval.
func (c *rawConnection) pingSender() {
ticker := time.Tick(PingSendInterval / 2)
for {
select {
case <-ticker:
d := time.Since(c.cw.Last())
if d < PingSendInterval/2 {
l.Debugln(c.id, "ping skipped after wr", d)
continue
}
l.Debugln(c.id, "ping -> after", d)
c.ping()
case <-c.closed:
return
}
}
}
// The pingReceiver checks that we've received a message (any message will do,
// but we expect pings in the absence of other messages) within the last
// ReceiveTimeout. If not, we close the connection with an ErrTimeout.
func (c *rawConnection) pingReceiver() {
ticker := time.Tick(ReceiveTimeout / 2)
for {
select {
case <-ticker:
d := time.Since(c.cr.Last())
if d > ReceiveTimeout {
l.Debugln(c.id, "ping timeout", d)
c.close(ErrTimeout)
}
l.Debugln(c.id, "last read within", d)
case <-c.closed:
return
}
}
}
type Statistics struct {
At time.Time
InBytesTotal int64
OutBytesTotal int64
}
func (c *rawConnection) Statistics() Statistics {
return Statistics{
At: time.Now(),
InBytesTotal: c.cr.Tot(),
OutBytesTotal: c.cw.Tot(),
}
}
func (c *rawConnection) lz4Compress(src []byte) ([]byte, error) {
var err error
buf := buffers.get(len(src))
buf, err = lz4.Encode(buf, src)
if err != nil {
return nil, err
}
binary.BigEndian.PutUint32(buf, binary.LittleEndian.Uint32(buf))
return buf, nil
}
func (c *rawConnection) lz4Decompress(src []byte) ([]byte, error) {
size := binary.BigEndian.Uint32(src)
binary.LittleEndian.PutUint32(src, size)
var err error
buf := buffers.get(int(size))
buf, err = lz4.Decode(buf, src)
if err != nil {
return nil, err
}
return buf, nil
}