Prevent never-ending loop due to stratum mode autodetection (#1335)

Prevent never-ending loop on first connection

Author: AndreaLanfranchi

libpoolprotocols/PoolClient.h

@@ -22,6 +22,7 @@ namespace dev
 			void setConnection(URI &conn)
 			{
 				m_conn = &conn;
+				m_canconnect.store(false, std::memory_order_relaxed);
 			}
 
 			virtual void connect() = 0;
@@ -49,6 +50,7 @@ namespace dev
 			std::atomic<bool> m_subscribed = { false };
 			std::atomic<bool> m_authorized = { false };
 			std::atomic<bool> m_connected = { false };
+			std::atomic<bool> m_canconnect = { false };
 
 			boost::asio::ip::basic_endpoint<boost::asio::ip::tcp> m_endpoint;
 

libpoolprotocols/stratum/EthStratumClient.cpp

@@ -257,7 +257,9 @@ void EthStratumClient::disconnect_finalize() {
 
     // If we got disconnected during autodetection phase
     // reissue a connect lowering stratum mode checks
-    if (!m_conn->StratumModeConfirmed())
+    // m_canconnect flag is used to prevent never-ending loop when
+    // remote endpoint rejects connections attempts persistently since the first
+    if (!m_conn->StratumModeConfirmed() && m_canconnect.load(std::memory_order_relaxed))
     {
         unsigned l = m_conn->StratumMode();
         if (l > 0)
@@ -440,6 +442,7 @@ void EthStratumClient::connect_handler(const boost::system::error_code& ec)
 
 		// Immediately set connecting flag to prevent 
 		// occurrence of subsequents timeouts (if any)
+		m_canconnect.store(true, std::memory_order_relaxed);
 		m_connecting.store(false, std::memory_order_relaxed);
 		m_conntimer.cancel();