307 lines
15 KiB
C#
307 lines
15 KiB
C#
using System.Net;
|
||
using System.Net.Security;
|
||
using System.Net.Sockets;
|
||
using System.Security.Cryptography;
|
||
using System.Security.Cryptography.X509Certificates;
|
||
using System.Text.Json;
|
||
using Fleet.Core;
|
||
using Fleet.Identity;
|
||
using Fleet.Rpc;
|
||
|
||
namespace Fleet.Link;
|
||
|
||
/// <summary>一个已经认证的点对点会话：对端是谁、数据面使用的会话凭据、对端地址。</summary>
|
||
public sealed record PeerSessionInfo(string PeerId, SessionCredentials Credentials, EndPoint? Remote, string? Version = null);
|
||
|
||
public sealed class PeerListenerOptions
|
||
{
|
||
public required IPEndPoint Listen { get; init; }
|
||
|
||
public required X509Certificate2 Certificate { get; init; }
|
||
|
||
/// <summary>按对端 ID 查找允许连接的公钥（SPKI）。不在列表里的对端一律拒绝（期望状态里的 peers）。</summary>
|
||
public required Func<string, byte[]?> LookupPeerKey { get; init; }
|
||
|
||
/// <summary>预握手敲门：每个连接先过这一关（名册里的 Probe 才算得出敲门包），不通过就直接断开，什么都不写。</summary>
|
||
public required KnockGate Knock { get; init; }
|
||
|
||
public int MaxFrameBytes { get; init; } = 8 << 20;
|
||
|
||
public int MaxSessions { get; init; } = 8;
|
||
|
||
/// <summary>敲门、TLS、挑战应答整个握手的上限。线路差的时候（RTT 好几秒、还丢包）握手本来就要很久，10 秒做不完就会一直连不上；敲门已经挡住了不认识的连接，放宽不会放大被占连接的风险。</summary>
|
||
public TimeSpan HandshakeTimeout { get; init; } = PeerTimings.Handshake;
|
||
}
|
||
|
||
/// <summary>
|
||
/// 监听点对点会话（Agent 一侧）：TLS（自签名证书，对端按 SPKI 指纹固定）→ 挑战 → 对端用身份私钥签名（purpose = peer）
|
||
/// → 临时 ECDH 派生数据面凭据。认证通过之后把会话交给产品代码，由它挂上自己的路由。
|
||
/// </summary>
|
||
public sealed class PeerListener
|
||
{
|
||
private readonly PeerListenerOptions _options;
|
||
private readonly Func<PeerSessionInfo, IRpcTransport, CancellationToken, Task> _onSession;
|
||
private readonly INodeLog _log;
|
||
private readonly TimeProvider _time;
|
||
private int _sessions;
|
||
|
||
public PeerListener(PeerListenerOptions options, Func<PeerSessionInfo, IRpcTransport, CancellationToken, Task> onSession, INodeLog log, TimeProvider time)
|
||
{
|
||
_options = options;
|
||
_onSession = onSession;
|
||
_log = log;
|
||
_time = time;
|
||
}
|
||
|
||
/// <summary>实际监听的端口（配置为 0 时由系统分配）；<see cref="RunAsync"/> 开始监听之后有效。</summary>
|
||
public int Port { get; private set; }
|
||
|
||
public TaskCompletionSource Started { get; } = new(TaskCreationOptions.RunContinuationsAsynchronously);
|
||
|
||
public async Task RunAsync(CancellationToken ct)
|
||
{
|
||
var listener = new TcpListener(_options.Listen);
|
||
if (_options.Listen.Address.Equals(IPAddress.IPv6Any))
|
||
{
|
||
listener.Server.DualMode = true;
|
||
}
|
||
|
||
listener.Start();
|
||
Port = ((IPEndPoint)listener.LocalEndpoint).Port;
|
||
Started.TrySetResult();
|
||
try
|
||
{
|
||
while (!ct.IsCancellationRequested)
|
||
{
|
||
TcpClient client = await listener.AcceptTcpClientAsync(ct);
|
||
if (Interlocked.Increment(ref _sessions) > _options.MaxSessions)
|
||
{
|
||
Interlocked.Decrement(ref _sessions);
|
||
client.Dispose();
|
||
continue;
|
||
}
|
||
|
||
_ = Task.Run(() => HandleAsync(client, ct), CancellationToken.None);
|
||
}
|
||
}
|
||
catch (OperationCanceledException)
|
||
{
|
||
// 停止监听
|
||
}
|
||
finally
|
||
{
|
||
listener.Stop();
|
||
}
|
||
}
|
||
|
||
private async Task HandleAsync(TcpClient client, CancellationToken ct)
|
||
{
|
||
try
|
||
{
|
||
client.NoDelay = true;
|
||
PeerSockets.EnableKeepAlive(client.Client);
|
||
using var timeout = new CancellationTokenSource(_options.HandshakeTimeout, _time);
|
||
using CancellationTokenSource linked = CancellationTokenSource.CreateLinkedTokenSource(ct, timeout.Token);
|
||
Stream stream = client.GetStream();
|
||
if (!await _options.Knock.AdmitAsync(stream, (client.Client.RemoteEndPoint as IPEndPoint)?.Address, linked.Token))
|
||
{
|
||
return;
|
||
}
|
||
|
||
var tls = new SslStream(stream, leaveInnerStreamOpen: false);
|
||
await tls.AuthenticateAsServerAsync(
|
||
new SslServerAuthenticationOptions { ServerCertificate = _options.Certificate, ClientCertificateRequired = false, EnabledSslProtocols = System.Security.Authentication.SslProtocols.Tls12 | System.Security.Authentication.SslProtocols.Tls13 },
|
||
linked.Token);
|
||
var transport = new TlsFrameTransport(tls, _options.MaxFrameBytes);
|
||
PeerSessionInfo? info = await HandshakeAsync(transport, client.Client.RemoteEndPoint, linked.Token);
|
||
if (info is null)
|
||
{
|
||
await transport.DisposeAsync();
|
||
return;
|
||
}
|
||
|
||
_log.Info("peer session", ("probe", info.PeerId), ("result", "established"));
|
||
await using (transport)
|
||
{
|
||
await _onSession(info, transport, ct);
|
||
}
|
||
}
|
||
catch (Exception ex) when (ex is IOException or System.Security.Authentication.AuthenticationException or OperationCanceledException or SocketException or ObjectDisposedException)
|
||
{
|
||
// 握手失败或连接断开：这个会话结束，监听继续
|
||
}
|
||
finally
|
||
{
|
||
client.Dispose();
|
||
Interlocked.Decrement(ref _sessions);
|
||
}
|
||
}
|
||
|
||
private async Task<PeerSessionInfo?> HandshakeAsync(TlsFrameTransport transport, EndPoint? remote, CancellationToken ct)
|
||
{
|
||
using var ephemeral = new EphemeralKey();
|
||
byte[] nonceBytes = RandomNumberGenerator.GetBytes(32);
|
||
string nonce = Convert.ToBase64String(nonceBytes);
|
||
string serverPub = Convert.ToBase64String(ephemeral.PublicKeySpki);
|
||
await SendAsync(transport, RpcMessage.Notification(PeerMethods.Challenge, new PeerChallengeParams { Nonce = nonce, EcdhPublicKey = serverPub }, PeerRpcContext.Default.PeerChallengeParams), ct);
|
||
|
||
RpcFrame? frame = await transport.ReceiveAsync(ct);
|
||
RpcMessage? message = frame is { Kind: FrameKind.Text } f ? JsonSerializer.Deserialize(f.Data.Span, FleetRpcContext.Default.RpcMessage) : null;
|
||
if (message?.Method != PeerMethods.Auth || message.Id is null || message.Params is null)
|
||
{
|
||
return null;
|
||
}
|
||
|
||
PeerAuthParams? auth = null;
|
||
try
|
||
{
|
||
auth = message.Params.Value.Deserialize(PeerRpcContext.Default.PeerAuthParams);
|
||
}
|
||
catch (JsonException)
|
||
{
|
||
// 按缺失处理
|
||
}
|
||
|
||
byte[]? key = auth is null ? null : _options.LookupPeerKey(auth.PeerId);
|
||
byte[] clientPublic = [];
|
||
bool ok = key is not null
|
||
&& TryDecode(auth!.Signature, out byte[] signature)
|
||
&& TryDecode(auth.EcdhPublicKey, out clientPublic)
|
||
&& NodeIdentity.Verify(key, signature, SignaturePurpose.Peer, auth.PeerId, nonce, auth.EcdhPublicKey, serverPub);
|
||
if (!ok)
|
||
{
|
||
// 对端不在允许列表里和签名错误返回同样的错误
|
||
await SendAsync(transport, RpcMessage.Failure(message.Id, new FleetError(FleetErrors.Unauthorized, "peer authentication failed")), ct);
|
||
_log.Warn("peer rejected", ("result", "unauthorized"));
|
||
return null;
|
||
}
|
||
|
||
SessionCredentials credentials;
|
||
try
|
||
{
|
||
credentials = ephemeral.Derive(clientPublic, nonceBytes, clientPublic, ephemeral.PublicKeySpki);
|
||
}
|
||
catch (CryptographicException)
|
||
{
|
||
await SendAsync(transport, RpcMessage.Failure(message.Id, new FleetError(FleetErrors.InvalidParams, "invalid ecdh key")), ct);
|
||
return null;
|
||
}
|
||
|
||
await SendAsync(transport, RpcMessage.Success(message.Id.Value, new PeerAuthResult { SessionId = Convert.ToHexStringLower(credentials.SessionId) }, PeerRpcContext.Default.PeerAuthResult), ct);
|
||
return new PeerSessionInfo(auth!.PeerId, credentials, remote, auth.Version is { Length: <= 64 } version ? version : null);
|
||
}
|
||
|
||
private static ValueTask SendAsync(TlsFrameTransport transport, RpcMessage message, CancellationToken ct) =>
|
||
transport.SendAsync(new RpcFrame(FrameKind.Text, JsonSerializer.SerializeToUtf8Bytes(message, FleetRpcContext.Default.RpcMessage)), ct);
|
||
|
||
private static bool TryDecode(string text, out byte[] data)
|
||
{
|
||
try
|
||
{
|
||
data = Convert.FromBase64String(text);
|
||
return data.Length > 0;
|
||
}
|
||
catch (FormatException)
|
||
{
|
||
data = [];
|
||
return false;
|
||
}
|
||
}
|
||
}
|
||
|
||
/// <summary>发起点对点会话（Probe 一侧）：按 SPKI 指纹固定对端证书，用身份私钥签名，派生同样的数据面凭据。</summary>
|
||
/// <summary>点对点会话的套接字：开启 TCP keepalive（空闲 30 秒后每 10 秒探测，连续 6 次无响应才断开，约 90 秒），主机真的没了的时候连接不会一直挂着；
|
||
/// 但线路短暂中断（几十秒的黑洞）不能把会话杀掉——线路中断本来就是要测的东西，会话得扛过去，恢复之后接着测。</summary>
|
||
internal static class PeerSockets
|
||
{
|
||
public static void EnableKeepAlive(Socket socket)
|
||
{
|
||
try
|
||
{
|
||
socket.SetSocketOption(SocketOptionLevel.Socket, SocketOptionName.KeepAlive, true);
|
||
socket.SetSocketOption(SocketOptionLevel.Tcp, SocketOptionName.TcpKeepAliveTime, 30);
|
||
socket.SetSocketOption(SocketOptionLevel.Tcp, SocketOptionName.TcpKeepAliveInterval, 10);
|
||
socket.SetSocketOption(SocketOptionLevel.Tcp, SocketOptionName.TcpKeepAliveRetryCount, 6);
|
||
}
|
||
catch (Exception ex) when (ex is SocketException or ArgumentException or PlatformNotSupportedException)
|
||
{
|
||
// 个别平台不支持其中某个选项：退回系统默认
|
||
}
|
||
}
|
||
}
|
||
|
||
/// <summary>点对点会话的时间常数，Probe 和 Agent 两侧共用。</summary>
|
||
public static class PeerTimings
|
||
{
|
||
/// <summary>建立会话（连接、敲门、TLS、挑战应答）的总时限。</summary>
|
||
public static readonly TimeSpan Handshake = TimeSpan.FromSeconds(30);
|
||
}
|
||
|
||
public static class PeerConnector
|
||
{
|
||
public static async Task<(PeerSessionInfo Info, TlsFrameTransport Transport)> ConnectAsync(
|
||
string host, int port, byte[] agentPublicKey, NodeIdentity identity, string peerId, string? version, int maxFrameBytes, TimeProvider time, CancellationToken ct)
|
||
{
|
||
// 证书按公钥的指纹固定；敲门密钥由双方身份密钥派生
|
||
byte[] expectedSpkiSha256 = NodeIdentity.Fingerprint(agentPublicKey);
|
||
TcpClient client = IPAddress.TryParse(host, out IPAddress? address) ? new TcpClient(address.AddressFamily) : new TcpClient();
|
||
client.NoDelay = true;
|
||
PeerSockets.EnableKeepAlive(client.Client);
|
||
try
|
||
{
|
||
using var timeout = new CancellationTokenSource(PeerTimings.Handshake, time);
|
||
using CancellationTokenSource linked = CancellationTokenSource.CreateLinkedTokenSource(ct, timeout.Token);
|
||
await client.ConnectAsync(host, port, linked.Token);
|
||
Stream stream = client.GetStream();
|
||
await stream.WriteAsync(Knock.Build(Knock.DeriveKey(identity, agentPublicKey), time.GetUtcNow().ToUnixTimeMilliseconds()), linked.Token);
|
||
var tls = new SslStream(
|
||
stream, leaveInnerStreamOpen: false,
|
||
(_, certificate, _, _) => certificate is not null && SpkiMatches(certificate, expectedSpkiSha256));
|
||
await tls.AuthenticateAsClientAsync(new SslClientAuthenticationOptions { TargetHost = host }, linked.Token);
|
||
var transport = new TlsFrameTransport(tls, maxFrameBytes);
|
||
|
||
RpcFrame? first = await transport.ReceiveAsync(linked.Token);
|
||
RpcMessage? challengeMessage = first is { Kind: FrameKind.Text } f ? JsonSerializer.Deserialize(f.Data.Span, FleetRpcContext.Default.RpcMessage) : null;
|
||
if (challengeMessage?.Method != PeerMethods.Challenge || challengeMessage.Params is null)
|
||
{
|
||
throw new InvalidOperationException("the agent did not send a challenge");
|
||
}
|
||
|
||
PeerChallengeParams challenge = challengeMessage.Params.Value.Deserialize(PeerRpcContext.Default.PeerChallengeParams)!;
|
||
using var ephemeral = new EphemeralKey();
|
||
string clientPub = Convert.ToBase64String(ephemeral.PublicKeySpki);
|
||
string signature = Convert.ToBase64String(identity.Sign(SignaturePurpose.Peer, peerId, challenge.Nonce, clientPub, challenge.EcdhPublicKey));
|
||
RpcMessage request = RpcMessage.Request(1, PeerMethods.Auth, new PeerAuthParams { PeerId = peerId, EcdhPublicKey = clientPub, Signature = signature, Version = version }, PeerRpcContext.Default.PeerAuthParams);
|
||
await transport.SendAsync(new RpcFrame(FrameKind.Text, JsonSerializer.SerializeToUtf8Bytes(request, FleetRpcContext.Default.RpcMessage)), linked.Token);
|
||
|
||
RpcFrame? replyFrame = await transport.ReceiveAsync(linked.Token);
|
||
RpcMessage? reply = replyFrame is { Kind: FrameKind.Text } r ? JsonSerializer.Deserialize(r.Data.Span, FleetRpcContext.Default.RpcMessage) : null;
|
||
if (reply is null || reply.Error is not null)
|
||
{
|
||
throw new InvalidOperationException("peer authentication failed: " + reply?.Error?.Data?.Code);
|
||
}
|
||
|
||
SessionCredentials credentials = ephemeral.Derive(Convert.FromBase64String(challenge.EcdhPublicKey), Convert.FromBase64String(challenge.Nonce), ephemeral.PublicKeySpki, Convert.FromBase64String(challenge.EcdhPublicKey));
|
||
string announced = reply.Result!.Value.Deserialize(PeerRpcContext.Default.PeerAuthResult)!.SessionId;
|
||
if (announced != Convert.ToHexStringLower(credentials.SessionId))
|
||
{
|
||
throw new InvalidOperationException("session id mismatch: the key agreement did not produce the same credentials");
|
||
}
|
||
|
||
return (new PeerSessionInfo(peerId, credentials, client.Client.RemoteEndPoint), transport);
|
||
}
|
||
catch
|
||
{
|
||
client.Dispose();
|
||
throw;
|
||
}
|
||
}
|
||
|
||
/// <summary>固定证书：只看 SPKI 的 SHA-256 指纹，不看证书链和主机名（自签名证书）。</summary>
|
||
private static bool SpkiMatches(X509Certificate certificate, byte[] expected)
|
||
{
|
||
X509Certificate2 full = certificate as X509Certificate2 ?? X509CertificateLoader.LoadCertificate(certificate.GetRawCertData());
|
||
return CryptographicOperations.FixedTimeEquals(NodeCertificate.SpkiFingerprint(full), expected);
|
||
}
|
||
}
|