Search Results

Search found 54 results on 3 pages for 'threadsafe'.

Page 3/3 | < Previous Page | 1 2 3 

  • Lock free multiple readers single writer

    - by dummzeuch
    I have got an in memory data structure that is read by multiple threads and written by only one thread. Currently I am using a critical section to make this access threadsafe. Unfortunately this has the effect of blocking readers even though only another reader is accessing it. There are two options to remedy this: use TMultiReadExclusiveWriteSynchronizer do away with any blocking by using a lock free approach For 2. I have got the following so far (any code that doesn't matter has been left out): type TDataManager = class private FAccessCount: integer; FData: TDataClass; public procedure Read(out _Some: integer; out _Data: double); procedure Write(_Some: integer; _Data: double); end; procedure TDataManager.Read(out _Some: integer; out _Data: double); var Data: TDAtaClass; begin InterlockedIncrement(FAccessCount); try // make sure we get both values from the same TDataClass instance Data := FData; // read the actual data _Some := Data.Some; _Data := Data.Data; finally InterlockedDecrement(FAccessCount); end; end; procedure TDataManager.Write(_Some: integer; _Data: double); var NewData: TDataClass; OldData: TDataClass; ReaderCount: integer; begin NewData := TDataClass.Create(_Some, _Data); InterlockedIncrement(FAccessCount); OldData := TDataClass(InterlockedExchange(integer(FData), integer(NewData)); // now FData points to the new instance but there might still be // readers that got the old one before we exchanged it. ReaderCount := InterlockedDecrement(FAccessCount); if ReaderCount = 0 then // no active readers, so we can safely free the old instance FreeAndNil(OldData) else begin /// here is the problem end; end; Unfortunately there is the small problem of getting rid of the OldData instance after it has been replaced. If no other thread is currently within the Read method (ReaderCount=0), it can safely be disposed and that's it. But what can I do if that's not the case? I could just store it until the next call and dispose it there, but Windows scheduling could in theory let a reader thread sleep while it is within the Read method and still has got a reference to OldData. If you see any other problem with the above code, please tell me about it. This is to be run on computers with multiple cores and the above methods are to be called very frequently. In case this matters: I am using Delphi 2007 with the builtin memory manager. I am aware that the memory manager probably enforces some lock anyway when creating a new class but I want to ignore that for the moment. Edit: It may not have been clear from the above: For the full lifetime of the TDataManager object there is only one thread that writes to the data, not several that might compete for write access. So this is a special case of MREW.

    Read the article

  • C++ Mutexes and STL Lists Across Subclasses

    - by Genesis
    I am currently writing a multi-threaded C++ server using Poco and am now at the point where I need to be keeping information on which users are connected, how many connections each of them have, and given it is a proxy server, where each of those connections are proxying through to. For this purpose I have created a ServerStats class which holds an STL list of ServerUser objects. The ServerStats class includes functions which can add and remove objects from the list as well as find a user in the list an return a pointer to them so I can access member functions within any given ServerUser object in the list. The ServerUser class contains an STL list of ServerConnection objects and much like the ServerStats class it contains functions to add, remove and find elements within this list. Now all of the above is working but I am now trying to make it threadsafe. I have defined a Poco::FastMutex within the ServerStats class and can lock/unlock this in the appropriate places so that STL containers are not modified at the same time as being searched for example. I am however having an issue setting up mutexes within the ServerUser class and am getting the following compiler error: /root/poco/Foundation/include/Poco/Mutex.h: In copy constructor âServerUser::ServerUser(const ServerUser&)â: src/SocksServer.cpp:185: instantiated from âvoid __gnu_cxx::new_allocator<_Tp::construct(_Tp*, const _Tp&) [with _Tp = ServerUser]â /usr/include/c++/4.4/bits/stl_list.h:464: instantiated from âstd::_List_node<_Tp* std::list<_Tp, _Alloc::_M_create_node(const _Tp&) [with _Tp = ServerUser, _Alloc = std::allocator]â /usr/include/c++/4.4/bits/stl_list.h:1407: instantiated from âvoid std::list<_Tp, _Alloc::_M_insert(std::_List_iterator<_Tp, const _Tp&) [with _Tp = ServerUser, _Alloc = std::allocator]â /usr/include/c++/4.4/bits/stl_list.h:920: instantiated from âvoid std::list<_Tp, _Alloc::push_back(const _Tp&) [with _Tp = ServerUser, _Alloc = std::allocator]â src/SocksServer.cpp:301: instantiated from here /root/poco/Foundation/include/Poco/Mutex.h:164: error: âPoco::FastMutex::FastMutex(const Poco::FastMutex&)â is private src/SocksServer.cpp:185: error: within this context In file included from /usr/include/c++/4.4/x86_64-linux-gnu/bits/c++allocator.h:34, from /usr/include/c++/4.4/bits/allocator.h:48, from /usr/include/c++/4.4/string:43, from /root/poco/Foundation/include/Poco/Bugcheck.h:44, from /root/poco/Foundation/include/Poco/Foundation.h:147, from /root/poco/Net/include/Poco/Net/Net.h:45, from /root/poco/Net/include/Poco/Net/TCPServerParams.h:43, from src/SocksServer.cpp:1: /usr/include/c++/4.4/ext/new_allocator.h: In member function âvoid __gnu_cxx::new_allocator<_Tp::construct(_Tp*, const _Tp&) [with _Tp = ServerUser]â: /usr/include/c++/4.4/ext/new_allocator.h:105: note: synthesized method âServerUser::ServerUser(const ServerUser&)â first required here src/SocksServer.cpp: At global scope: src/SocksServer.cpp:118: warning: âstd::string getWord(std::string)â defined but not used make: * [/root/poco/SocksServer/obj/Linux/x86_64/debug_shared/SocksServer.o] Error 1 The code for the ServerStats, ServerUser and ServerConnection classes is below: class ServerConnection { public: bool continue_connection; int bytes_in; int bytes_out; string source_address; string destination_address; ServerConnection() { continue_connection = true; } ~ServerConnection() { } }; class ServerUser { public: string username; int connection_count; string client_ip; ServerUser() { } ~ServerUser() { } ServerConnection* addConnection(string source_address, string destination_address) { //FastMutex::ScopedLock lock(_connection_mutex); ServerConnection connection; connection.source_address = source_address; connection.destination_address = destination_address; client_ip = getWord(source_address, ":"); _connections.push_back(connection); connection_count++; return &_connections.back(); } void removeConnection(string source_address) { //FastMutex::ScopedLock lock(_connection_mutex); for(list<ServerConnection>::iterator it = _connections.begin(); it != _connections.end(); it++) { if(it->source_address == source_address) { it = _connections.erase(it); connection_count--; } } } void disconnect() { //FastMutex::ScopedLock lock(_connection_mutex); for(list<ServerConnection>::iterator it = _connections.begin(); it != _connections.end(); it++) { it->continue_connection = false; } } list<ServerConnection>* getConnections() { return &_connections; } private: list<ServerConnection> _connections; //UNCOMMENTING THIS LINE BREAKS IT: //mutable FastMutex _connection_mutex; }; class ServerStats { public: int current_users; ServerStats() { current_users = 0; } ~ServerStats() { } ServerUser* addUser(string username) { FastMutex::ScopedLock lock(_user_mutex); for(list<ServerUser>::iterator it = _users.begin(); it != _users.end(); it++) { if(it->username == username) { return &(*it); } } ServerUser newUser; newUser.username = username; _users.push_back(newUser); current_users++; return &_users.back(); } void removeUser(string username) { FastMutex::ScopedLock lock(_user_mutex); for(list<ServerUser>::iterator it = _users.begin(); it != _users.end(); it++) { if(it->username == username) { _users.erase(it); current_users--; break; } } } ServerUser* getUser(string username) { FastMutex::ScopedLock lock(_user_mutex); for(list<ServerUser>::iterator it = _users.begin(); it != _users.end(); it++) { if(it->username == username) { return &(*it); } } return NULL; } private: list<ServerUser> _users; mutable FastMutex _user_mutex; }; Now I have never used C++ for a project of this size or mutexes for that matter so go easy please :) Firstly, can anyone tell me why the above is causing a compiler error? Secondly, can anyone suggest a better way of storing the information I require? Bear in mind that I need to update this info whenever connections come or go and it needs to be global to the whole server.

    Read the article

  • Azure Diagnostics wrt Custom Logs and honoring scheduledTransferPeriod

    - by kjsteuer
    I have implemented my own TraceListener similar to http://blogs.technet.com/b/meamcs/archive/2013/05/23/diagnostics-of-cloud-services-custom-trace-listener.aspx . One thing I noticed is that that logs show up immediately in My Azure Table Storage. I wonder if this is expected with Custom Trace Listeners or because I am in a development environment. My diagnosics.wadcfg <?xml version="1.0" encoding="utf-8"?> <DiagnosticMonitorConfiguration configurationChangePollInterval="PT1M""overallQuotaInMB="4096" xmlns="http://schemas.microsoft.com/ServiceHosting/2010/10/DiagnosticsConfiguration"> <DiagnosticInfrastructureLogs scheduledTransferLogLevelFilter="Information" /> <Directories scheduledTransferPeriod="PT1M"> <IISLogs container="wad-iis-logfiles" /> <CrashDumps container="wad-crash-dumps" /> </Directories> <Logs bufferQuotaInMB="0" scheduledTransferPeriod="PT30M" scheduledTransferLogLevelFilter="Information" /> </DiagnosticMonitorConfiguration> I have changed my approach a bit. Now I am defining in the web config of my webrole. I notice when I set autoflush to true in the webconfig, every thing works but scheduledTransferPeriod is not honored because the flush method pushes to the table storage. I would like to have scheduleTransferPeriod trigger the flush or trigger flush after a certain number of log entries like the buffer is full. Then I can also flush on server shutdown. Is there any method or event on the CustomTraceListener where I can listen to the scheduleTransferPeriod? <system.diagnostics> <!--http://msdn.microsoft.com/en-us/library/sk36c28t(v=vs.110).aspx By default autoflush is false. By default useGlobalLock is true. While we try to be threadsafe, we keep this default for now. Later if we would like to increase performance we can remove this. see http://msdn.microsoft.com/en-us/library/system.diagnostics.trace.usegloballock(v=vs.110).aspx --> <trace> <listeners> <add name="TableTraceListener" type="Pos.Services.Implementation.TableTraceListener, Pos.Services.Implementation" /> <remove name="Default" /> </listeners> </trace> </system.diagnostics> I have modified the custom trace listener to the following: namespace Pos.Services.Implementation { class TableTraceListener : TraceListener { #region Fields //connection string for azure storage readonly string _connectionString; //Custom sql storage table for logs. //TODO put in config readonly string _diagnosticsTable; [ThreadStatic] static StringBuilder _messageBuffer; readonly object _initializationSection = new object(); bool _isInitialized; CloudTableClient _tableStorage; readonly object _traceLogAccess = new object(); readonly List<LogEntry> _traceLog = new List<LogEntry>(); #endregion #region Constructors public TableTraceListener() : base("TableTraceListener") { _connectionString = RoleEnvironment.GetConfigurationSettingValue("DiagConnection"); _diagnosticsTable = RoleEnvironment.GetConfigurationSettingValue("DiagTableName"); } #endregion #region Methods /// <summary> /// Flushes the entries to the storage table /// </summary> public override void Flush() { if (!_isInitialized) { lock (_initializationSection) { if (!_isInitialized) { Initialize(); } } } var context = _tableStorage.GetTableServiceContext(); context.MergeOption = MergeOption.AppendOnly; lock (_traceLogAccess) { _traceLog.ForEach(entry => context.AddObject(_diagnosticsTable, entry)); _traceLog.Clear(); } if (context.Entities.Count > 0) { context.BeginSaveChangesWithRetries(SaveChangesOptions.None, (ar) => context.EndSaveChangesWithRetries(ar), null); } } /// <summary> /// Creates the storage table object. This class does not need to be locked because the caller is locked. /// </summary> private void Initialize() { var account = CloudStorageAccount.Parse(_connectionString); _tableStorage = account.CreateCloudTableClient(); _tableStorage.GetTableReference(_diagnosticsTable).CreateIfNotExists(); _isInitialized = true; } public override bool IsThreadSafe { get { return true; } } #region Trace and Write Methods /// <summary> /// Writes the message to a string buffer /// </summary> /// <param name="message">the Message</param> public override void Write(string message) { if (_messageBuffer == null) _messageBuffer = new StringBuilder(); _messageBuffer.Append(message); } /// <summary> /// Writes the message with a line breaker to a string buffer /// </summary> /// <param name="message"></param> public override void WriteLine(string message) { if (_messageBuffer == null) _messageBuffer = new StringBuilder(); _messageBuffer.AppendLine(message); } /// <summary> /// Appends the trace information and message /// </summary> /// <param name="eventCache">the Event Cache</param> /// <param name="source">the Source</param> /// <param name="eventType">the Event Type</param> /// <param name="id">the Id</param> /// <param name="message">the Message</param> public override void TraceEvent(TraceEventCache eventCache, string source, TraceEventType eventType, int id, string message) { base.TraceEvent(eventCache, source, eventType, id, message); AppendEntry(id, eventType, eventCache); } /// <summary> /// Adds the trace information to a collection of LogEntry objects /// </summary> /// <param name="id">the Id</param> /// <param name="eventType">the Event Type</param> /// <param name="eventCache">the EventCache</param> private void AppendEntry(int id, TraceEventType eventType, TraceEventCache eventCache) { if (_messageBuffer == null) _messageBuffer = new StringBuilder(); var message = _messageBuffer.ToString(); _messageBuffer.Length = 0; if (message.EndsWith(Environment.NewLine)) message = message.Substring(0, message.Length - Environment.NewLine.Length); if (message.Length == 0) return; var entry = new LogEntry() { PartitionKey = string.Format("{0:D10}", eventCache.Timestamp >> 30), RowKey = string.Format("{0:D19}", eventCache.Timestamp), EventTickCount = eventCache.Timestamp, Level = (int)eventType, EventId = id, Pid = eventCache.ProcessId, Tid = eventCache.ThreadId, Message = message }; lock (_traceLogAccess) _traceLog.Add(entry); } #endregion #endregion } }

    Read the article

  • Cumulative +1/-1 Cointoss crashes on 1000 iterations. Please advise; c++ boost random libraries

    - by user1731972
    following some former advice Multithreaded application, am I doing it right? I think I have a threadsafe number generator using boost, but my program crashes when I input 1000 iterations. The output .csv file when graphed looks right, but I'm not sure why it's crashing. It's using _beginthread, and everyone is telling me I should use the more (convoluted) _beingthreadex, which I'm not familiar with. If someone could recommend an example, I would greatly appreciate it. Also... someone pointed out I should be applying a second parameter to my _beginthread for the array counting start positions, but I have no idea how to pass more than one parameter, other than attempting to use a structure, and I've read structure's and _beginthread don't get along (although, I could just use the boost threads...) #include <process.h> #include <windows.h> #include <iostream> #include <fstream> #include <time.h> #include <random> #include <boost/random.hpp> //for srand48_r(time(NULL), &randBuffer); which doesn't work #include <stdio.h> #include <stdlib.h> //#include <thread> using namespace std; using namespace boost; using namespace boost::random; void myThread0 (void *dummy ); void myThread1 (void *dummy ); void myThread2 (void *dummy ); void myThread3 (void *dummy ); //for random seeds void initialize(); //from http://stackoverflow.com/questions/7114043/random-number-generation-in-c11-how-to-generate-how-do-they-work uniform_int_distribution<> two(1,2); typedef std::mt19937 MyRNG; // the Mersenne Twister with a popular choice of parameters uint32_t seed_val; // populate somehow MyRNG rng1; // e.g. keep one global instance (per thread) MyRNG rng2; // e.g. keep one global instance (per thread) MyRNG rng3; // e.g. keep one global instance (per thread) MyRNG rng4; // e.g. keep one global instance (per thread) //only needed for shared variables //CRITICAL_SECTION cs1,cs2,cs3,cs4; // global int main() { ofstream myfile; myfile.open ("coinToss.csv"); int rNum; long numRuns; long count = 0; int divisor = 1; float fHolder = 0; long counter = 0; float percent = 0.0; //? //unsigned threadID; //HANDLE hThread; initialize(); HANDLE hThread[4]; const int size = 100000; int array[size]; printf ("Runs (uses multiple of 100,000) "); cin >> numRuns; for (int a = 0; a < numRuns; a++) { hThread[0] = (HANDLE)_beginthread( myThread0, 0, (void*)(array) ); hThread[1] = (HANDLE)_beginthread( myThread1, 0, (void*)(array) ); hThread[2] = (HANDLE)_beginthread( myThread2, 0, (void*)(array) ); hThread[3] = (HANDLE)_beginthread( myThread3, 0, (void*)(array) ); //waits for threads to finish before continuing WaitForMultipleObjects(4, hThread, TRUE, INFINITE); //closes handles I guess? CloseHandle( hThread[0] ); CloseHandle( hThread[1] ); CloseHandle( hThread[2] ); CloseHandle( hThread[3] ); //dump array into calculations //average array into fHolder //this could be split into threads as well for (int p = 0; p < size; p++) { counter += array[p] == 2 ? 1 : -1; //cout << array[p] << endl; //cout << counter << endl; } //this fHolder calculation didn't work //fHolder = counter / size; //so I had to use this cout << counter << endl; fHolder = counter; fHolder = fHolder / size; myfile << fHolder << endl; } } void initialize() { //seed value needs to be supplied //rng1.seed(seed_val*1); rng1.seed((unsigned int)time(NULL)); rng2.seed(((unsigned int)time(NULL))*2); rng3.seed(((unsigned int)time(NULL))*3); rng4.seed(((unsigned int)time(NULL))*4); }; void myThread0 (void *param) { //EnterCriticalSection(&cs1); //aquire the critical section object int *i = (int *)param; for (int x = 0; x < 25000; x++) { //doesn't work, part of merssene twister //i[x] = next(); i[x] = two(rng1); //original srand //i[x] = rand() % 2 + 1; //doesn't work for some reason. //uint_dist2(rng); //i[x] = qrand() % 2 + 1; //cout << i[x] << endl; } //LeaveCriticalSection(&cs1); // release the critical section object } void myThread1 (void *param) { //EnterCriticalSection(&cs2); //aquire the critical section object int *i = (int *)param; for (int x = 25000; x < 50000; x++) { //param[x] = rand() % 2 + 1; i[x] = two(rng2); //i[x] = rand() % 2 + 1; //cout << i[x] << endl; } //LeaveCriticalSection(&cs2); // release the critical section object } void myThread2 (void *param) { //EnterCriticalSection(&cs3); //aquire the critical section object int *i = (int *)param; for (int x = 50000; x < 75000; x++) { i[x] = two(rng3); //i[x] = rand() % 2 + 1; //cout << i[x] << endl; } //LeaveCriticalSection(&cs3); // release the critical section object } void myThread3 (void *param) { //EnterCriticalSection(&cs4); //aquire the critical section object int *i = (int *)param; for (int x = 75000; x < 100000; x++) { i[x] = two(rng4); //i[x] = rand() % 2 + 1; //cout << i[x] << endl; } //LeaveCriticalSection(&cs4); // release the critical section object }

    Read the article

< Previous Page | 1 2 3