dnsoverudp.go

package netxlite

//
// DNS-over-UDP transport
//

import (
	"context"
	"net"
	"time"

	"github.com/ooni/probe-engine/pkg/model"
)

// DNSOverUDPTransport is a DNS-over-UDP DNSTransport.
//
// To construct this type, either manually fill the fields marked as MANDATORY
// or just use the NewDNSOverUDPTransport factory directly.
//
// RoundTrip creates a new connected UDP socket for each outgoing query. Using a
// new socket is good because some censored environments will block the client UDP
// endpoint for several seconds when you query for blocked domains. We could also
// have used an unconnected UDP socket here, but:
//
// 1. connected UDP sockets are great because they get some ICMP errors to be
// translated into socket errors (among them, host_unreachable);
//
// 2. connected UDP sockets ignore responses from illegitimate IP addresses but
// most if not all DNS resolvers also do that, therefore this does not seem to
// be a realistic censorship vector. At the same time, connected sockets
// provide us for free with the feature that we don't need to bother with checking
// whether the reply comes from the expected server.
//
// Being able to observe some ICMP errors is good because it could possibly
// make this code suitable to implement parasitic traceroute.
//
// This transport by default listens for additional responses after the first
// one and makes them available using the context-configured trace.
type DNSOverUDPTransport struct {
	// Decoder is the MANDATORY DNSDecoder to use.
	Decoder model.DNSDecoder

	// Dialer is the MANDATORY dialer used to create the conn.
	Dialer model.Dialer

	// Endpoint is the MANDATORY server's endpoint (e.g., 1.1.1.1:53)
	Endpoint string

	// lateResponses is posted in nonblocking mode each time this
	// transport detects there was a late response for a query that had
	// already been answered. Use this channel for testing.
	lateResponses chan any
}

// NewUnwrappedDNSOverUDPTransport creates a DNSOverUDPTransport instance
// that has not been wrapped yet.
//
// Arguments:
//
// - dialer is any type that implements the Dialer interface;
//
// - address is the endpoint address (e.g., 8.8.8.8:53).
//
// If the address contains a domain name rather than an IP address
// (e.g., dns.google:53), we will end up using the first of the
// IP addresses returned by the underlying DNS lookup performed using
// the dialer. This usage pattern is NOT RECOMMENDED because we'll
// have less control over which IP address is being used.
func NewUnwrappedDNSOverUDPTransport(dialer model.Dialer, address string) *DNSOverUDPTransport {
	return &DNSOverUDPTransport{
		Decoder:       &DNSDecoderMiekg{},
		Dialer:        dialer,
		Endpoint:      address,
		lateResponses: nil, // not interested by default
	}
}

// RoundTrip sends a query and receives a response.
func (t *DNSOverUDPTransport) RoundTrip(
	ctx context.Context, query model.DNSQuery) (model.DNSResponse, error) {
	// QUIRK: the original DNS-over-UDP code had a five seconds timeout, which is
	// consistent with the Bionic implementation. Let's try to preserve such a
	// behavior by combining dialing and I/O timeout together.
	//
	// See https://labs.ripe.net/Members/baptiste_jonglez_1/persistent-dns-connections-for-reliability-and-performance
	const opTimeout = 5 * time.Second
	deadline := time.Now().Add(opTimeout)
	ctx, cancel := context.WithTimeout(ctx, opTimeout)
	defer cancel()
	rawQuery, err := query.Bytes()
	if err != nil {
		return nil, err
	}
	conn, err := t.Dialer.DialContext(ctx, "udp", t.Endpoint)
	if err != nil {
		return nil, err
	}
	conn.SetDeadline(deadline) // time to dial (usually ~zero) already factored in
	joinedch := make(chan bool)
	myaddr := conn.LocalAddr().String()
	if _, err := conn.Write(rawQuery); err != nil {
		conn.Close() // we still own the conn
		return nil, err
	}
	resp, err := t.recv(query, conn)
	if err != nil {
		conn.Close() // we still own the conn
		return nil, err
	}
	// start a goroutine to listen for any delayed DNS response and
	// TRANSFER the conn's OWNERSHIP to such a goroutine.
	go t.ownConnAndSendRecvLoop(ctx, conn, query, myaddr, joinedch)
	return resp, nil
}

// RequiresPadding returns false for UDP according to RFC8467.
func (t *DNSOverUDPTransport) RequiresPadding() bool {
	return false
}

// Network returns the transport network, i.e., "udp".
func (t *DNSOverUDPTransport) Network() string {
	return "udp"
}

// Address returns the upstream server address.
func (t *DNSOverUDPTransport) Address() string {
	return t.Endpoint
}

// CloseIdleConnections closes idle connections, if any.
func (t *DNSOverUDPTransport) CloseIdleConnections() {
	// The underlying dialer MAY have idle connections so let's
	// forward the call...
	t.Dialer.CloseIdleConnections()
}

var _ model.DNSTransport = &DNSOverUDPTransport{}

// ownConnAndSendRecvLoop listens for delayed DNS responses after we have returned the
// first response. As the name implies, this function TAKES OWNERSHIP of the [conn].
func (t *DNSOverUDPTransport) ownConnAndSendRecvLoop(ctx context.Context, conn net.Conn,
	query model.DNSQuery, myaddr string, eofch chan<- bool) {
	defer close(eofch) // synchronize with the caller
	defer conn.Close() // we own the conn
	trace := ContextTraceOrDefault(ctx)
	for {
		started := trace.TimeNow()
		resp, err := t.recv(query, conn)
		finished := trace.TimeNow()
		if err != nil {
			// We are going to consider all errors as fatal for now until we
			// hear of specific errs that it might have sense to ignore.
			//
			// Note that erroring out here includes the expiration of the conn's
			// I/O deadline, which we set above precisely because we want
			// the total runtime of this goroutine to be bounded.
			//
			// Also, we ARE NOT going to report any failure here as a delayed
			// DNS response because we only care about duplicate messages, since
			// this seems how censorship is implemented in, e.g., China.
			return
		}
		// if there's testing code waiting to be unblocked because we
		// received a delayed response, unblock it
		select {
		case t.lateResponses <- true:
		default:
			// there's no one waiting and it does not matter
		}
		addrs, err := resp.DecodeLookupHost()
		if err := trace.OnDelayedDNSResponse(started, t, query, resp, addrs, err, finished); err != nil {
			// This error typically indicates that the buffer on which we're
			// writing is now full, so there's no point in persisting.
			return
		}
	}
}

// recv receives a single response for the given query using the given conn.
func (t *DNSOverUDPTransport) recv(query model.DNSQuery, conn net.Conn) (model.DNSResponse, error) {
	const maxmessagesize = 1 << 17
	rawResponse := make([]byte, maxmessagesize)
	count, err := conn.Read(rawResponse)
	if err != nil {
		return nil, err
	}
	rawResponse = rawResponse[:count]
	return t.Decoder.DecodeResponse(rawResponse, query)
}