danicoin/contrib/epee/include/net/abstract_tcp_server2.inl
moneromooo-monero 6bca9a8ef4
abstract_tcp_server2: avoid deadlock waiting for send queue to drain
If we reach the send queue size limit, we need to release the lock,
or we will deadlock and it will never drain.

If we reach that limit, it's likely there's another problem in the
first place though, so it will probably not drain in practice either,
unless some kind of transient network timeout.
2016-04-17 11:45:38 +01:00

1158 lines
47 KiB
C++

/**
@file
@author from CrypoNote (see copyright below; Andrey N. Sabelnikov)
@monero rfree
@brief the connection templated-class for one peer connection
*/
// Copyright (c) 2006-2013, Andrey N. Sabelnikov, www.sabelnikov.net
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
// * Neither the name of the Andrey N. Sabelnikov nor the
// names of its contributors may be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
// ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
// WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
// DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER BE LIABLE FOR ANY
// DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
// (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
// LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
// ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
//#include "net_utils_base.h"
#include <boost/lambda/bind.hpp>
#include <boost/foreach.hpp>
#include <boost/lambda/lambda.hpp>
#include <boost/uuid/random_generator.hpp>
#include <boost/chrono.hpp>
#include <boost/utility/value_init.hpp>
#include <boost/asio/deadline_timer.hpp>
#include <boost/date_time/posix_time/posix_time.hpp> // TODO
#include <boost/thread/v2/thread.hpp> // TODO
#include "misc_language.h"
#include "pragma_comp_defs.h"
#include <sstream>
#include <iomanip>
#include <algorithm>
#include "../../../../src/cryptonote_core/cryptonote_core.h" // e.g. for the send_stop_signal()
#include "../../../../contrib/otshell_utils/utils.hpp"
#include "../../../../src/p2p/data_logger.hpp"
using namespace nOT::nUtils; // TODO
#define CONNECTION_CLEANUP_TIME 30 // seconds
PRAGMA_WARNING_PUSH
namespace epee
{
namespace net_utils
{
/************************************************************************/
/* */
/************************************************************************/
PRAGMA_WARNING_DISABLE_VS(4355)
template<class t_protocol_handler>
connection<t_protocol_handler>::connection( boost::asio::io_service& io_service,
typename t_protocol_handler::config_type& config,
std::atomic<long> &ref_sock_count, // the ++/-- counter
std::atomic<long> &sock_number, // the only increasing ++ number generator
i_connection_filter* &pfilter
,t_connection_type connection_type
)
:
connection_basic(io_service, ref_sock_count, sock_number),
m_protocol_handler(this, config, context),
m_pfilter( pfilter ),
m_connection_type( connection_type ),
m_throttle_speed_in("speed_in", "throttle_speed_in"),
m_throttle_speed_out("speed_out", "throttle_speed_out")
{
_info_c("net/sleepRPC", "test, connection constructor set m_connection_type="<<m_connection_type);
}
PRAGMA_WARNING_DISABLE_VS(4355)
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
connection<t_protocol_handler>::~connection()
{
if(!m_was_shutdown)
{
_dbg3("[sock " << socket_.native_handle() << "] Socket destroyed without shutdown.");
shutdown();
}
_dbg3("[sock " << socket_.native_handle() << "] Socket destroyed");
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
boost::asio::ip::tcp::socket& connection<t_protocol_handler>::socket()
{
return socket_;
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
boost::shared_ptr<connection<t_protocol_handler> > connection<t_protocol_handler>::safe_shared_from_this()
{
try
{
return connection<t_protocol_handler>::shared_from_this();
}
catch (const boost::bad_weak_ptr&)
{
// It happens when the connection is being deleted
return boost::shared_ptr<connection<t_protocol_handler> >();
}
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::start(bool is_income, bool is_multithreaded)
{
TRY_ENTRY();
// Use safe_shared_from_this, because of this is public method and it can be called on the object being deleted
auto self = safe_shared_from_this();
if(!self)
return false;
m_is_multithreaded = is_multithreaded;
boost::system::error_code ec;
auto remote_ep = socket_.remote_endpoint(ec);
CHECK_AND_NO_ASSERT_MES(!ec, false, "Failed to get remote endpoint: " << ec.message() << ':' << ec.value());
auto local_ep = socket_.local_endpoint(ec);
CHECK_AND_NO_ASSERT_MES(!ec, false, "Failed to get local endpoint: " << ec.message() << ':' << ec.value());
context = boost::value_initialized<t_connection_context>();
long ip_ = boost::asio::detail::socket_ops::host_to_network_long(remote_ep.address().to_v4().to_ulong());
// create a random uuid
boost::uuids::uuid random_uuid;
// that stuff turns out to be included, even though it's from src... Taking advantage
random_uuid = crypto::rand<boost::uuids::uuid>();
context.set_details(random_uuid, ip_, remote_ep.port(), is_income);
_dbg3("[sock " << socket_.native_handle() << "] new connection from " << print_connection_context_short(context) <<
" to " << local_ep.address().to_string() << ':' << local_ep.port() <<
", total sockets objects " << m_ref_sock_count);
if(m_pfilter && !m_pfilter->is_remote_ip_allowed(context.m_remote_ip))
{
_dbg2("[sock " << socket_.native_handle() << "] ip denied " << string_tools::get_ip_string_from_int32(context.m_remote_ip) << ", shutdowning connection");
close();
return false;
}
m_protocol_handler.after_init_connection();
socket_.async_read_some(boost::asio::buffer(buffer_),
strand_.wrap(
boost::bind(&connection<t_protocol_handler>::handle_read, self,
boost::asio::placeholders::error,
boost::asio::placeholders::bytes_transferred)));
#if !defined(_WIN32) || !defined(__i686)
// not supported before Windows7, too lazy for runtime check
// Just exclude for 32bit windows builds
//set ToS flag
int tos = get_tos_flag();
boost::asio::detail::socket_option::integer< IPPROTO_IP, IP_TOS >
optionTos( tos );
socket_.set_option( optionTos );
//_dbg1("Set ToS flag to " << tos);
#endif
boost::asio::ip::tcp::no_delay noDelayOption(false);
socket_.set_option(noDelayOption);
return true;
CATCH_ENTRY_L0("connection<t_protocol_handler>::start()", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::request_callback()
{
TRY_ENTRY();
_dbg2("[" << print_connection_context_short(context) << "] request_callback");
// Use safe_shared_from_this, because of this is public method and it can be called on the object being deleted
auto self = safe_shared_from_this();
if(!self)
return false;
strand_.post(boost::bind(&connection<t_protocol_handler>::call_back_starter, self));
CATCH_ENTRY_L0("connection<t_protocol_handler>::request_callback()", false);
return true;
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
boost::asio::io_service& connection<t_protocol_handler>::get_io_service()
{
return socket_.get_io_service();
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::add_ref()
{
TRY_ENTRY();
//_dbg3("[sock " << socket_.native_handle() << "] add_ref, m_peer_number=" << mI->m_peer_number);
CRITICAL_REGION_LOCAL(m_self_refs_lock);
//_dbg3("[sock " << socket_.native_handle() << "] add_ref 2, m_peer_number=" << mI->m_peer_number);
// Use safe_shared_from_this, because of this is public method and it can be called on the object being deleted
auto self = safe_shared_from_this();
if(!self)
return false;
if(m_was_shutdown)
return false;
m_self_refs.push_back(self);
return true;
CATCH_ENTRY_L0("connection<t_protocol_handler>::add_ref()", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::release()
{
TRY_ENTRY();
boost::shared_ptr<connection<t_protocol_handler> > back_connection_copy;
LOG_PRINT_L4("[sock " << socket_.native_handle() << "] release");
CRITICAL_REGION_BEGIN(m_self_refs_lock);
CHECK_AND_ASSERT_MES(m_self_refs.size(), false, "[sock " << socket_.native_handle() << "] m_self_refs empty at connection<t_protocol_handler>::release() call");
//erasing from container without additional copy can cause start deleting object, including m_self_refs
back_connection_copy = m_self_refs.back();
m_self_refs.pop_back();
CRITICAL_REGION_END();
return true;
CATCH_ENTRY_L0("connection<t_protocol_handler>::release()", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void connection<t_protocol_handler>::call_back_starter()
{
TRY_ENTRY();
_dbg2("[" << print_connection_context_short(context) << "] fired_callback");
m_protocol_handler.handle_qued_callback();
CATCH_ENTRY_L0("connection<t_protocol_handler>::call_back_starter()", void());
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void connection<t_protocol_handler>::save_dbg_log()
{
std::string address, port;
boost::system::error_code e;
boost::asio::ip::tcp::endpoint endpoint = socket_.remote_endpoint(e);
if (e)
{
address = "<not connected>";
port = "<not connected>";
}
else
{
address = endpoint.address().to_string();
port = boost::lexical_cast<std::string>(endpoint.port());
}
_mark_c("net/kind" ,
" connection type " << to_string( m_connection_type ) << " "
<< socket_.local_endpoint().address().to_string() << ":" << socket_.local_endpoint().port()
<< " <--> " << address << ":" << port);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void connection<t_protocol_handler>::handle_read(const boost::system::error_code& e,
std::size_t bytes_transferred)
{
TRY_ENTRY();
//_info("[sock " << socket_.native_handle() << "] Async read calledback.");
if (!e)
{
{
CRITICAL_REGION_LOCAL(m_throttle_speed_in_mutex);
m_throttle_speed_in.handle_trafic_exact(bytes_transferred);
context.m_current_speed_down = m_throttle_speed_in.get_current_speed();
}
{
CRITICAL_REGION_LOCAL( epee::net_utils::network_throttle_manager::network_throttle_manager::m_lock_get_global_throttle_in );
epee::net_utils::network_throttle_manager::network_throttle_manager::get_global_throttle_in().handle_trafic_exact(bytes_transferred * 1024);
}
double delay=0; // will be calculated - how much we should sleep to obey speed limit etc
if (speed_limit_is_enabled()) {
do // keep sleeping if we should sleep
{
{ //_scope_dbg1("CRITICAL_REGION_LOCAL");
CRITICAL_REGION_LOCAL( epee::net_utils::network_throttle_manager::m_lock_get_global_throttle_in );
delay = epee::net_utils::network_throttle_manager::get_global_throttle_in().get_sleep_time_after_tick( bytes_transferred ); // decission from global throttle
}
delay *= 0.5;
if (delay > 0) {
long int ms = (long int)(delay * 100);
epee::net_utils::data_logger::get_instance().add_data("sleep_down", ms);
boost::this_thread::sleep_for(boost::chrono::milliseconds(ms));
}
} while(delay > 0);
} // any form of sleeping
//_info("[sock " << socket_.native_handle() << "] RECV " << bytes_transferred);
logger_handle_net_read(bytes_transferred);
context.m_last_recv = time(NULL);
context.m_recv_cnt += bytes_transferred;
bool recv_res = m_protocol_handler.handle_recv(buffer_.data(), bytes_transferred);
if(!recv_res)
{
//_info("[sock " << socket_.native_handle() << "] protocol_want_close");
//some error in protocol, protocol handler ask to close connection
boost::interprocess::ipcdetail::atomic_write32(&m_want_close_connection, 1);
bool do_shutdown = false;
CRITICAL_REGION_BEGIN(m_send_que_lock);
if(!m_send_que.size())
do_shutdown = true;
CRITICAL_REGION_END();
if(do_shutdown)
shutdown();
}else
{
socket_.async_read_some(boost::asio::buffer(buffer_),
strand_.wrap(
boost::bind(&connection<t_protocol_handler>::handle_read, connection<t_protocol_handler>::shared_from_this(),
boost::asio::placeholders::error,
boost::asio::placeholders::bytes_transferred)));
//_info("[sock " << socket_.native_handle() << "]Async read requested.");
}
}else
{
_dbg3("[sock " << socket_.native_handle() << "] Some not success at read: " << e.message() << ':' << e.value());
if(e.value() != 2)
{
_dbg3("[sock " << socket_.native_handle() << "] Some problems at read: " << e.message() << ':' << e.value());
shutdown();
}
}
// If an error occurs then no new asynchronous operations are started. This
// means that all shared_ptr references to the connection object will
// disappear and the object will be destroyed automatically after this
// handler returns. The connection class's destructor closes the socket.
CATCH_ENTRY_L0("connection<t_protocol_handler>::handle_read", void());
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::call_run_once_service_io()
{
TRY_ENTRY();
if(!m_is_multithreaded)
{
//single thread model, we can wait in blocked call
size_t cnt = socket_.get_io_service().run_one();
if(!cnt)//service is going to quit
return false;
}else
{
//multi thread model, we can't(!) wait in blocked call
//so we make non blocking call and releasing CPU by calling sleep(0);
//if no handlers were called
//TODO: Maybe we need to have have critical section + event + callback to upper protocol to
//ask it inside(!) critical region if we still able to go in event wait...
size_t cnt = socket_.get_io_service().poll_one();
if(!cnt)
misc_utils::sleep_no_w(0);
}
return true;
CATCH_ENTRY_L0("connection<t_protocol_handler>::call_run_once_service_io", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::do_send(const void* ptr, size_t cb) {
TRY_ENTRY();
// Use safe_shared_from_this, because of this is public method and it can be called on the object being deleted
auto self = safe_shared_from_this();
if (!self) return false;
if (m_was_shutdown) return false;
// TODO avoid copy
const double factor = 32; // TODO config
typedef long long signed int t_safe; // my t_size to avoid any overunderflow in arithmetic
const t_safe chunksize_good = (t_safe)( 1024 * std::max(1.0,factor) );
const t_safe chunksize_max = chunksize_good * 2 ;
const bool allow_split = (m_connection_type == e_connection_type_RPC) ? false : true; // do not split RPC data
ASRT(! (chunksize_max<0) ); // make sure it is unsigned before removin sign with cast:
long long unsigned int chunksize_max_unsigned = static_cast<long long unsigned int>( chunksize_max ) ;
if (allow_split && (cb > chunksize_max_unsigned)) {
{ // LOCK: chunking
epee::critical_region_t<decltype(m_chunking_lock)> send_guard(m_chunking_lock); // *** critical ***
_dbg3_c("net/out/size", "do_send() will SPLIT into small chunks, from packet="<<cb<<" B for ptr="<<ptr);
t_safe all = cb; // all bytes to send
t_safe pos = 0; // current sending position
// 01234567890
// ^^^^ (pos=0, len=4) ; pos:=pos+len, pos=4
// ^^^^ (pos=4, len=4) ; pos:=pos+len, pos=8
// ^^^ (pos=8, len=4) ;
// const size_t bufsize = chunksize_good; // TODO safecast
// char* buf = new char[ bufsize ];
bool all_ok = true;
while (pos < all) {
t_safe lenall = all-pos; // length from here to end
t_safe len = std::min( chunksize_good , lenall); // take a smaller part
ASRT(len<=chunksize_good);
// pos=8; len=4; all=10; len=3;
ASRT(! (len<0) ); // check before we cast away sign:
unsigned long long int len_unsigned = static_cast<long long int>( len );
ASRT(len>0); // (redundand)
ASRT(len_unsigned < std::numeric_limits<size_t>::max()); // yeap we want strong < then max size, to be sure
void *chunk_start = ((char*)ptr) + pos;
_fact_c("net/out/size","chunk_start="<<chunk_start<<" ptr="<<ptr<<" pos="<<pos);
ASRT(chunk_start >= ptr); // not wrapped around address?
//std::memcpy( (void*)buf, chunk_start, len);
_dbg3_c("net/out/size", "part of " << lenall << ": pos="<<pos << " len="<<len);
bool ok = do_send_chunk(chunk_start, len); // <====== ***
all_ok = all_ok && ok;
if (!all_ok) {
_dbg1_c("net/out/size", "do_send() DONE ***FAILED*** from packet="<<cb<<" B for ptr="<<ptr);
_dbg1("do_send() SEND was aborted in middle of big package - this is mostly harmless "
<< " (e.g. peer closed connection) but if it causes trouble tell us at #monero-dev. " << cb);
return false; // partial failure in sending
}
pos = pos+len; ASRT(pos >0);
// (in catch block, or uniq pointer) delete buf;
} // each chunk
_dbg3_c("net/out/size", "do_send() DONE SPLIT from packet="<<cb<<" B for ptr="<<ptr);
_dbg3 ( "do_send() DONE SPLIT from packet="<<cb<<" B for ptr="<<ptr);
_info_c("net/sleepRPC", "do_send() m_connection_type = " << m_connection_type);
return all_ok; // done - e.g. queued - all the chunks of current do_send call
} // LOCK: chunking
} // a big block (to be chunked) - all chunks
else { // small block
return do_send_chunk(ptr,cb); // just send as 1 big chunk
}
CATCH_ENTRY_L0("connection<t_protocol_handler>::do_send", false);
} // do_send()
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::do_send_chunk(const void* ptr, size_t cb)
{
TRY_ENTRY();
// Use safe_shared_from_this, because of this is public method and it can be called on the object being deleted
auto self = safe_shared_from_this();
if(!self)
return false;
if(m_was_shutdown)
return false;
{
CRITICAL_REGION_LOCAL(m_throttle_speed_out_mutex);
m_throttle_speed_out.handle_trafic_exact(cb);
context.m_current_speed_up = m_throttle_speed_out.get_current_speed();
}
//_info("[sock " << socket_.native_handle() << "] SEND " << cb);
context.m_last_send = time(NULL);
context.m_send_cnt += cb;
//some data should be wrote to stream
//request complete
if (speed_limit_is_enabled()) {
sleep_before_packet(cb, 1, 1);
}
m_send_que_lock.lock(); // *** critical ***
epee::misc_utils::auto_scope_leave_caller scope_exit_handler = epee::misc_utils::create_scope_leave_handler([&](){m_send_que_lock.unlock();});
long int retry=0;
const long int retry_limit = 5*4;
while (m_send_que.size() > ABSTRACT_SERVER_SEND_QUE_MAX_COUNT)
{
retry++;
/* if ( ::cryptonote::core::get_is_stopping() ) { // TODO re-add fast stop
_fact("ABORT queue wait due to stopping");
return false; // aborted
}*/
long int ms = 250 + (rand()%50);
_info_c("net/sleep", "Sleeping because QUEUE is FULL, in " << __FUNCTION__ << " for " << ms << " ms before packet_size="<<cb); // XXX debug sleep
m_send_que_lock.unlock();
boost::this_thread::sleep(boost::posix_time::milliseconds( ms ) );
m_send_que_lock.lock();
_dbg1("sleep for queue: " << ms);
if (retry > retry_limit) {
_erro("send que size is more than ABSTRACT_SERVER_SEND_QUE_MAX_COUNT(" << ABSTRACT_SERVER_SEND_QUE_MAX_COUNT << "), shutting down connection");
// _dbg1_c("net/sleep", "send que size is more than ABSTRACT_SERVER_SEND_QUE_MAX_COUNT(" << ABSTRACT_SERVER_SEND_QUE_MAX_COUNT << "), shutting down connection");
close();
return false;
}
}
m_send_que.resize(m_send_que.size()+1);
m_send_que.back().assign((const char*)ptr, cb);
if(m_send_que.size() > 1)
{ // active operation should be in progress, nothing to do, just wait last operation callback
auto size_now = cb;
_info_c("net/out/size", "do_send() NOW just queues: packet="<<size_now<<" B, is added to queue-size="<<m_send_que.size());
//do_send_handler_delayed( ptr , size_now ); // (((H))) // empty function
LOG_PRINT_L4("[sock " << socket_.native_handle() << "] Async send requested " << m_send_que.front().size());
}
else
{ // no active operation
if(m_send_que.size()!=1)
{
_erro("Looks like no active operations, but send que size != 1!!");
return false;
}
auto size_now = m_send_que.front().size();
_dbg1_c("net/out/size", "do_send() NOW SENSD: packet="<<size_now<<" B");
if (speed_limit_is_enabled())
do_send_handler_write( ptr , size_now ); // (((H)))
ASRT( size_now == m_send_que.front().size() );
boost::asio::async_write(socket_, boost::asio::buffer(m_send_que.front().data(), size_now ) ,
//strand_.wrap(
boost::bind(&connection<t_protocol_handler>::handle_write, self, _1, _2)
//)
);
//_dbg3("(chunk): " << size_now);
//logger_handle_net_write(size_now);
//_info("[sock " << socket_.native_handle() << "] Async send requested " << m_send_que.front().size());
}
//do_send_handler_stop( ptr , cb ); // empty function
return true;
CATCH_ENTRY_L0("connection<t_protocol_handler>::do_send", false);
} // do_send_chunk
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::shutdown()
{
// Initiate graceful connection closure.
boost::system::error_code ignored_ec;
socket_.shutdown(boost::asio::ip::tcp::socket::shutdown_both, ignored_ec);
m_was_shutdown = true;
m_protocol_handler.release_protocol();
return true;
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::close()
{
TRY_ENTRY();
//_info("[sock " << socket_.native_handle() << "] Que Shutdown called.");
size_t send_que_size = 0;
CRITICAL_REGION_BEGIN(m_send_que_lock);
send_que_size = m_send_que.size();
CRITICAL_REGION_END();
boost::interprocess::ipcdetail::atomic_write32(&m_want_close_connection, 1);
if(!send_que_size)
{
shutdown();
}
return true;
CATCH_ENTRY_L0("connection<t_protocol_handler>::close", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool connection<t_protocol_handler>::cancel()
{
return close();
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void connection<t_protocol_handler>::handle_write(const boost::system::error_code& e, size_t cb)
{
TRY_ENTRY();
LOG_PRINT_L4("[sock " << socket_.native_handle() << "] Async send calledback " << cb);
if (e)
{
_dbg1("[sock " << socket_.native_handle() << "] Some problems at write: " << e.message() << ':' << e.value());
shutdown();
return;
}
logger_handle_net_write(cb);
if (speed_limit_is_enabled()) {
sleep_before_packet(cb, 1, 1);
}
bool do_shutdown = false;
CRITICAL_REGION_BEGIN(m_send_que_lock);
if(m_send_que.empty())
{
_erro("[sock " << socket_.native_handle() << "] m_send_que.size() == 0 at handle_write!");
return;
}
m_send_que.pop_front();
if(m_send_que.empty())
{
if(boost::interprocess::ipcdetail::atomic_read32(&m_want_close_connection))
{
do_shutdown = true;
}
}else
{
//have more data to send
auto size_now = m_send_que.front().size();
_dbg1_c("net/out/size", "handle_write() NOW SENDS: packet="<<size_now<<" B" <<", from queue size="<<m_send_que.size());
if (speed_limit_is_enabled())
do_send_handler_write_from_queue(e, m_send_que.front().size() , m_send_que.size()); // (((H)))
ASRT( size_now == m_send_que.front().size() );
boost::asio::async_write(socket_, boost::asio::buffer(m_send_que.front().data(), size_now) ,
// strand_.wrap(
boost::bind(&connection<t_protocol_handler>::handle_write, connection<t_protocol_handler>::shared_from_this(), _1, _2)
// )
);
//_dbg3("(normal)" << size_now);
}
CRITICAL_REGION_END();
if(do_shutdown)
{
shutdown();
}
CATCH_ENTRY_L0("connection<t_protocol_handler>::handle_write", void());
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void connection<t_protocol_handler>::setRpcStation()
{
m_connection_type = e_connection_type_RPC;
_fact_c("net/sleepRPC", "set m_connection_type = RPC ");
_info_c("net/kind", "set m_connection_type = RPC ");
}
template<class t_protocol_handler>
bool connection<t_protocol_handler>::speed_limit_is_enabled() const {
return m_connection_type != e_connection_type_RPC ;
}
/************************************************************************/
/* */
/************************************************************************/
template<class t_protocol_handler>
boosted_tcp_server<t_protocol_handler>::boosted_tcp_server( t_connection_type connection_type ) :
m_io_service_local_instance(new boost::asio::io_service()),
io_service_(*m_io_service_local_instance.get()),
acceptor_(io_service_),
m_stop_signal_sent(false), m_port(0),
m_sock_count(0), m_sock_number(0), m_threads_count(0),
m_pfilter(NULL), m_thread_index(0),
m_connection_type( connection_type ),
new_connection_(new connection<t_protocol_handler>(io_service_, m_config, m_sock_count, m_sock_number, m_pfilter, m_connection_type))
{
create_server_type_map();
m_thread_name_prefix = "NET";
}
template<class t_protocol_handler>
boosted_tcp_server<t_protocol_handler>::boosted_tcp_server(boost::asio::io_service& extarnal_io_service, t_connection_type connection_type) :
io_service_(extarnal_io_service),
acceptor_(io_service_),
m_stop_signal_sent(false), m_port(0),
m_sock_count(0), m_sock_number(0), m_threads_count(0),
m_pfilter(NULL), m_thread_index(0),
m_connection_type(connection_type),
new_connection_(new connection<t_protocol_handler>(io_service_, m_config, m_sock_count, m_sock_number, m_pfilter, connection_type))
{
create_server_type_map();
m_thread_name_prefix = "NET";
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
boosted_tcp_server<t_protocol_handler>::~boosted_tcp_server()
{
this->send_stop_signal();
timed_wait_server_stop(10000);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void boosted_tcp_server<t_protocol_handler>::create_server_type_map()
{
server_type_map["NET"] = e_connection_type_NET;
server_type_map["RPC"] = e_connection_type_RPC;
server_type_map["P2P"] = e_connection_type_P2P;
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::init_server(uint32_t port, const std::string address)
{
TRY_ENTRY();
m_stop_signal_sent = false;
m_port = port;
m_address = address;
// Open the acceptor with the option to reuse the address (i.e. SO_REUSEADDR).
boost::asio::ip::tcp::resolver resolver(io_service_);
boost::asio::ip::tcp::resolver::query query(address, boost::lexical_cast<std::string>(port));
boost::asio::ip::tcp::endpoint endpoint = *resolver.resolve(query);
acceptor_.open(endpoint.protocol());
acceptor_.set_option(boost::asio::ip::tcp::acceptor::reuse_address(true));
acceptor_.bind(endpoint);
acceptor_.listen();
boost::asio::ip::tcp::endpoint binded_endpoint = acceptor_.local_endpoint();
m_port = binded_endpoint.port();
_fact_c("net/RPClog", "start accept");
acceptor_.async_accept(new_connection_->socket(),
boost::bind(&boosted_tcp_server<t_protocol_handler>::handle_accept, this,
boost::asio::placeholders::error));
return true;
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::init_server", false);
}
//-----------------------------------------------------------------------------
PUSH_WARNINGS
DISABLE_GCC_WARNING(maybe-uninitialized)
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::init_server(const std::string port, const std::string& address)
{
uint32_t p = 0;
if (port.size() && !string_tools::get_xtype_from_string(p, port)) {
LOG_ERROR("Failed to convert port no = " << port);
return false;
}
return this->init_server(p, address);
}
POP_WARNINGS
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::worker_thread()
{
TRY_ENTRY();
uint32_t local_thr_index = boost::interprocess::ipcdetail::atomic_inc32(&m_thread_index);
std::string thread_name = std::string("[") + m_thread_name_prefix;
thread_name += boost::to_string(local_thr_index) + "]";
log_space::log_singletone::set_thread_log_prefix(thread_name);
// _fact("Thread name: " << m_thread_name_prefix);
while(!m_stop_signal_sent)
{
try
{
io_service_.run();
}
catch(const std::exception& ex)
{
_erro("Exception at server worker thread, what=" << ex.what());
}
catch(...)
{
_erro("Exception at server worker thread, unknown execption");
}
}
//_info("Worker thread finished");
return true;
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::worker_thread", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void boosted_tcp_server<t_protocol_handler>::set_threads_prefix(const std::string& prefix_name)
{
m_thread_name_prefix = prefix_name;
auto it = server_type_map.find(m_thread_name_prefix);
if (it==server_type_map.end()) throw std::runtime_error("Unknown prefix/server type:" + std::string(prefix_name));
auto connection_type = it->second; // the value of type
_info_c("net/RPClog", "Set server type to: " << connection_type << " from name: " << m_thread_name_prefix);
_info_c("net/RPClog", "prefix_name = " << prefix_name);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void boosted_tcp_server<t_protocol_handler>::set_connection_filter(i_connection_filter* pfilter)
{
m_pfilter = pfilter;
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::run_server(size_t threads_count, bool wait, const boost::thread::attributes& attrs)
{
TRY_ENTRY();
m_threads_count = threads_count;
m_main_thread_id = boost::this_thread::get_id();
log_space::log_singletone::set_thread_log_prefix("[SRV_MAIN]");
add_idle_handler(boost::bind(&boosted_tcp_server::cleanup_connections, this), 5000);
while(!m_stop_signal_sent)
{
// Create a pool of threads to run all of the io_services.
CRITICAL_REGION_BEGIN(m_threads_lock);
for (std::size_t i = 0; i < threads_count; ++i)
{
boost::shared_ptr<boost::thread> thread(new boost::thread(
attrs, boost::bind(&boosted_tcp_server<t_protocol_handler>::worker_thread, this)));
_note("Run server thread name: " << m_thread_name_prefix);
m_threads.push_back(thread);
}
CRITICAL_REGION_END();
// Wait for all threads in the pool to exit.
if (wait) // && ! ::cryptonote::core::get_is_stopping()) // TODO fast_exit
{
_fact("JOINING all threads");
for (std::size_t i = 0; i < m_threads.size(); ++i) {
m_threads[i]->join();
}
_fact("JOINING all threads - almost");
m_threads.clear();
_fact("JOINING all threads - DONE");
}
else {
_dbg1("Reiniting OK.");
return true;
}
if(wait && !m_stop_signal_sent)
{
//some problems with the listening socket ?..
_dbg1("Net service stopped without stop request, restarting...");
if(!this->init_server(m_port, m_address))
{
_dbg1("Reiniting service failed, exit.");
return false;
}else
{
_dbg1("Reiniting OK.");
}
}
}
return true;
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::run_server", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::is_thread_worker()
{
TRY_ENTRY();
CRITICAL_REGION_LOCAL(m_threads_lock);
BOOST_FOREACH(boost::shared_ptr<boost::thread>& thp, m_threads)
{
if(thp->get_id() == boost::this_thread::get_id())
return true;
}
if(m_threads_count == 1 && boost::this_thread::get_id() == m_main_thread_id)
return true;
return false;
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::is_thread_worker", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::timed_wait_server_stop(uint64_t wait_mseconds)
{
TRY_ENTRY();
boost::chrono::milliseconds ms(wait_mseconds);
for (std::size_t i = 0; i < m_threads.size(); ++i)
{
if(m_threads[i]->joinable() && !m_threads[i]->try_join_for(ms))
{
_dbg1("Interrupting thread " << m_threads[i]->native_handle());
m_threads[i]->interrupt();
}
}
return true;
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::timed_wait_server_stop", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void boosted_tcp_server<t_protocol_handler>::send_stop_signal()
{
if (::cryptonote::core::get_fast_exit() == true)
{
detach_threads();
}
m_stop_signal_sent = true;
TRY_ENTRY();
connections_mutex.lock();
for (auto &c: connections_)
{
c.second->cancel();
}
connections_.clear();
connections_mutex.unlock();
io_service_.stop();
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::send_stop_signal()", void());
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::cleanup_connections()
{
connections_mutex.lock();
boost::system_time cutoff = boost::get_system_time() - boost::posix_time::seconds(CONNECTION_CLEANUP_TIME);
while (!connections_.empty() && connections_.front().first < cutoff)
{
connections_.pop_front();
}
connections_mutex.unlock();
return true;
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::is_stop_signal_sent()
{
return m_stop_signal_sent;
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void boosted_tcp_server<t_protocol_handler>::handle_accept(const boost::system::error_code& e)
{
_fact_c("net/RPClog", "handle_accept");
TRY_ENTRY();
if (!e)
{
if (m_connection_type == e_connection_type_RPC) {
_note_c("net/rpc", "New server for RPC connections");
_fact_c("net/RPClog", "New server for RPC connections");
new_connection_->setRpcStation(); // hopefully this is not needed actually
}
connection_ptr conn(std::move(new_connection_));
new_connection_.reset(new connection<t_protocol_handler>(io_service_, m_config, m_sock_count, m_sock_number, m_pfilter, m_connection_type));
acceptor_.async_accept(new_connection_->socket(),
boost::bind(&boosted_tcp_server<t_protocol_handler>::handle_accept, this,
boost::asio::placeholders::error));
conn->start(true, 1 < m_threads_count);
conn->save_dbg_log();
}else
{
_erro("Some problems at accept: " << e.message() << ", connections_count = " << m_sock_count);
}
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::handle_accept", void());
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
bool boosted_tcp_server<t_protocol_handler>::connect(const std::string& adr, const std::string& port, uint32_t conn_timeout, t_connection_context& conn_context, const std::string& bind_ip)
{
TRY_ENTRY();
connection_ptr new_connection_l(new connection<t_protocol_handler>(io_service_, m_config, m_sock_count, m_sock_number, m_pfilter, m_connection_type) );
connections_mutex.lock();
connections_.push_back(std::make_pair(boost::get_system_time(), new_connection_l));
LOG_PRINT_L2("connections_ size now " << connections_.size());
connections_mutex.unlock();
boost::asio::ip::tcp::socket& sock_ = new_connection_l->socket();
//////////////////////////////////////////////////////////////////////////
boost::asio::ip::tcp::resolver resolver(io_service_);
boost::asio::ip::tcp::resolver::query query(boost::asio::ip::tcp::v4(), adr, port);
boost::asio::ip::tcp::resolver::iterator iterator = resolver.resolve(query);
boost::asio::ip::tcp::resolver::iterator end;
if(iterator == end)
{
_erro("Failed to resolve " << adr);
return false;
}
//////////////////////////////////////////////////////////////////////////
//boost::asio::ip::tcp::endpoint remote_endpoint(boost::asio::ip::address::from_string(addr.c_str()), port);
boost::asio::ip::tcp::endpoint remote_endpoint(*iterator);
sock_.open(remote_endpoint.protocol());
if(bind_ip != "0.0.0.0" && bind_ip != "0" && bind_ip != "" )
{
boost::asio::ip::tcp::endpoint local_endpoint(boost::asio::ip::address::from_string(adr.c_str()), 0);
sock_.bind(local_endpoint);
}
/*
NOTICE: be careful to make sync connection from event handler: in case if all threads suddenly do sync connect, there will be no thread to dispatch events from io service.
*/
boost::system::error_code ec = boost::asio::error::would_block;
//have another free thread(s), work in wait mode, without event handling
struct local_async_context
{
boost::system::error_code ec;
boost::mutex connect_mut;
boost::condition_variable cond;
};
boost::shared_ptr<local_async_context> local_shared_context(new local_async_context());
local_shared_context->ec = boost::asio::error::would_block;
boost::unique_lock<boost::mutex> lock(local_shared_context->connect_mut);
auto connect_callback = [](boost::system::error_code ec_, boost::shared_ptr<local_async_context> shared_context)
{
shared_context->connect_mut.lock(); shared_context->ec = ec_; shared_context->connect_mut.unlock(); shared_context->cond.notify_one();
};
sock_.async_connect(remote_endpoint, boost::bind<void>(connect_callback, _1, local_shared_context));
while(local_shared_context->ec == boost::asio::error::would_block)
{
bool r = local_shared_context->cond.timed_wait(lock, boost::get_system_time() + boost::posix_time::milliseconds(conn_timeout));
if (m_stop_signal_sent)
{
if (sock_.is_open())
sock_.close();
return false;
}
if(local_shared_context->ec == boost::asio::error::would_block && !r)
{
//timeout
sock_.close();
_dbg3("Failed to connect to " << adr << ":" << port << ", because of timeout (" << conn_timeout << ")");
return false;
}
}
ec = local_shared_context->ec;
if (ec || !sock_.is_open())
{
_dbg3("Some problems at connect, message: " << ec.message());
if (sock_.is_open())
sock_.close();
return false;
}
_dbg3("Connected success to " << adr << ':' << port);
bool r = new_connection_l->start(false, 1 < m_threads_count);
if (r)
{
new_connection_l->get_context(conn_context);
//new_connection_l.reset(new connection<t_protocol_handler>(io_service_, m_config, m_sock_count, m_pfilter));
}
else
{
_erro("[sock " << new_connection_->socket().native_handle() << "] Failed to start connection, connections_count = " << m_sock_count);
}
new_connection_l->save_dbg_log();
return r;
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::connect", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler> template<class t_callback>
bool boosted_tcp_server<t_protocol_handler>::connect_async(const std::string& adr, const std::string& port, uint32_t conn_timeout, t_callback cb, const std::string& bind_ip)
{
TRY_ENTRY();
connection_ptr new_connection_l(new connection<t_protocol_handler>(io_service_, m_config, m_sock_count, m_sock_number, m_pfilter, m_connection_type) );
connections_mutex.lock();
connections_.push_back(std::make_pair(boost::get_system_time(), new_connection_l));
LOG_PRINT_L2("connections_ size now " << connections_.size());
connections_mutex.unlock();
boost::asio::ip::tcp::socket& sock_ = new_connection_l->socket();
//////////////////////////////////////////////////////////////////////////
boost::asio::ip::tcp::resolver resolver(io_service_);
boost::asio::ip::tcp::resolver::query query(boost::asio::ip::tcp::v4(), adr, port);
boost::asio::ip::tcp::resolver::iterator iterator = resolver.resolve(query);
boost::asio::ip::tcp::resolver::iterator end;
if(iterator == end)
{
_erro("Failed to resolve " << adr);
return false;
}
//////////////////////////////////////////////////////////////////////////
boost::asio::ip::tcp::endpoint remote_endpoint(*iterator);
sock_.open(remote_endpoint.protocol());
if(bind_ip != "0.0.0.0" && bind_ip != "0" && bind_ip != "" )
{
boost::asio::ip::tcp::endpoint local_endpoint(boost::asio::ip::address::from_string(adr.c_str()), 0);
sock_.bind(local_endpoint);
}
boost::shared_ptr<boost::asio::deadline_timer> sh_deadline(new boost::asio::deadline_timer(io_service_));
//start deadline
sh_deadline->expires_from_now(boost::posix_time::milliseconds(conn_timeout));
sh_deadline->async_wait([=](const boost::system::error_code& error)
{
if(error != boost::asio::error::operation_aborted)
{
_dbg3("Failed to connect to " << adr << ':' << port << ", because of timeout (" << conn_timeout << ")");
new_connection_l->socket().close();
}
});
//start async connect
sock_.async_connect(remote_endpoint, [=](const boost::system::error_code& ec_)
{
t_connection_context conn_context = AUTO_VAL_INIT(conn_context);
boost::system::error_code ignored_ec;
boost::asio::ip::tcp::socket::endpoint_type lep = new_connection_l->socket().local_endpoint(ignored_ec);
if(!ec_)
{//success
if(!sh_deadline->cancel())
{
cb(conn_context, boost::asio::error::operation_aborted);//this mean that deadline timer already queued callback with cancel operation, rare situation
}else
{
_dbg3("[sock " << new_connection_l->socket().native_handle() << "] Connected success to " << adr << ':' << port <<
" from " << lep.address().to_string() << ':' << lep.port());
bool r = new_connection_l->start(false, 1 < m_threads_count);
if (r)
{
new_connection_l->get_context(conn_context);
cb(conn_context, ec_);
}
else
{
_dbg3("[sock " << new_connection_l->socket().native_handle() << "] Failed to start connection to " << adr << ':' << port);
cb(conn_context, boost::asio::error::fault);
}
}
}else
{
_dbg3("[sock " << new_connection_l->socket().native_handle() << "] Failed to connect to " << adr << ':' << port <<
" from " << lep.address().to_string() << ':' << lep.port() << ": " << ec_.message() << ':' << ec_.value());
cb(conn_context, ec_);
}
});
return true;
CATCH_ENTRY_L0("boosted_tcp_server<t_protocol_handler>::connect_async", false);
}
//---------------------------------------------------------------------------------
template<class t_protocol_handler>
void boosted_tcp_server<t_protocol_handler>::detach_threads()
{
for (auto thread : m_threads)
thread->detach();
}
} // namespace
} // namespace
PRAGMA_WARNING_POP