My application is a tcp/ip server, with main thread created only once & listening all the time. When new client connects, the main thread creates the new thread of TClientThread type. There is however no list of running Client threads, as that would make my app a bit complicated... is there any way to execute "terminate" method on all the threads, even if the thread is busy (in my case "busy" means it's waiting for the data, where the timeout set is about 30 sec ... so I have to kill it anyway, without waiting.)?
The simple closing application seems not to run "terminate" method on the threads, which ends up with memory leaks reported by FastMM...
Memory leaks on shutdown are nothing to worry about - going to the trouble of freeing memory before returning control to the operating system is a waste of time and needlessly slows down application exit. All you really need to do is ensure that all data has been saved, and all interprocess handles (such as semaphores and mutexes) correctly released, and exit away.
For notifying clients, the best you can do would be a strategy somewhat like this:
Add all client-handling threads to some list somewhere (with suitable locking on creation, destruction and iteration)
Make client threads remove themselves from the list upon termination, and have the last item removed from the list set an event (manual reset event, e.g. TEvent in SyncObjs) if the server is shutting down
Introduce polling (e.g. select or equivalent with a timeout) or other kind of interruption (e.g. SO_RCVTIMEO / SO_SNDTIMEO) in what would otherwise be long-running blocking routines, monitoring the Terminated property
On shutdown, lock the list and iterate through it, calling Terminate, and then wait for the event to be signaled; of course, the listening socket which adds items to the list should be closed and known to be closed before iterating through the list
Sounds like this article may help
What you'll see if you click that link:
Using Semaphores in Delphi, Part 2:
The Connection Pool
By: Cary Jensen
Abstract: Semaphores are used to
coordinate multiple threads and
processes. That semaphores provide
multiple threads with simultaneous
access to a shared resource is
highlighted by the
TFixedConnectionPool class described
in this article.
I use a KillThreadList: TList global.
I monitor it in my thread as:
while (Not Terminated) do
begin
inc(Inker);
if (WaitForSingleObject(FTick, finterval) = WAIT_TIMEOUT) then
Begin
if Inker >= 10 then
Begin
ProcessTables;
Inker := 0;
sleep(1000);
End;
if KillThreadList.Contains(ThreadID) = True then Terminate;
End;
end;
I also test for the KillThreadList in my processes to let me opt out of them before completion, where safe to do so.
I pass the OnTerminate event out to the Main thread and remove the ThreadID from the KillList there. I use this model extensively and it has not failed me yet.
procedure TfrmProcessQualcommLocations.OnTerminateThread;
var
ThreadID : Cardinal;
i : integer;
aStatusBar :TStatFrame;
begin
ThreadID := (Sender as Tthread).ThreadID;
for i := 0 to StatusBarList.Count -1 do
Begin
if StatusBarList.Items[i].ThreadID = ThreadID then
Begin
aStatusBar := StatusBarList.Items[i];
KillThreadList.Extract(ThreadID);
StatusBarList.Extract(aStatusBar);
aStatusBar.Free;
break;
End;
End;
self.Refresh;
end;
In the case above, I am also removing some GUI stuff.
Hope that helps.
SpringerRider
Related
I created a WindowProc to be notified for system time changes:
constructor TJJWScheduler.Create;
begin
fTimeChangeWnd := Classes.AllocateHWnd(TimeChangeWndProc);
end;
procedure TJJWScheduler.TimeChangeWndProc(var msg: TMessage);
var
i: integer;
begin
case msg.Msg of
WM_TIMECHANGE:
begin
// my things
end;
end;
end;
This code is running inside a Windows Service.
The problem is that it isn't fired when I change the system time!
Why not the broadcast message (WM_TIMECHANGE) isn't delivered to my window?
There is another way to do this without a loop?
EDIT
I don't known why, but I hardcoded the PeekMessage to process messages to that window, and everything comes to work fine. The code below solved my problem:
var
msg: TMsg;
if PeekMessage(msg, fTimeChangeWnd, WM_TIMECHANGE, WM_TIMECHANGE, PM_REMOVE) then
begin
TranslateMessage(msg);
DispatchMessage(msg);
end;
This workaround is very strange, because I already have others windows processing messages (by generic ProcessMessages), only this one isn't processing its messages.
The reason your window was not receiving the WM_TIMECHANGE messages is that your window is created from a secondary thread.
Each thread in a process has its own message queue. Synchronous messages are delivered when you service the message queue, so ever for a non-queued message like WM_TIMECHANGE you do need to service the secondary threads message queue in order for messages to be delivered.
For example, look at the documentation for GetMessage, the most common way to pull messages of the queue:
The function dispatches incoming sent messages until a posted message is available for retrieval.
The same is true for PeekMessage. It dispatches incoming sent messages before peeking the queue.
There are a handful of other ways for sent messages to be dispatched, but these are the primary ones.
Now, I suspect that it may be inconvenient for you to periodically dispatch messages from your secondary thread. If your secondary thread does nothing else then it can simply sit in the traditional GetMessage, TranslateMessage, DispatchMessage loop. And most of the time it will happily block in GetMessage dispatching any incoming sent messages. But if your secondary thread does more work then that's probably not a viable option.
You are already running and servicing a message queue on the main service thread. It may make more sense to make your listener window have affinity with the main service thread. Do that by creating it from code that runs in the main service thread.
Note also that AllocateHWnd is documented not to be thread-safe. You must not call it from any thread other than the main thread of the process. So, if you do wish to remain on a secondary thread, you'll need to use CreateWindow rather than AllocateHWnd. But this is perhaps yet another good reason to move this window onto the main thread.
Delphi XE2, Indy V10, Windows 7 Pro - but I think I have a general conceptual problem.
Indy's TCP client is synchronous - it uses blocking calls.
However parts of my application are asynchronous - I want to send data over TCP and wait for a response when A) the 3rd party serial port component reports input from the serial port (it appears to be asynchronous & run in it's own thread, posting messages to my application's main form's Windows message queue) and B) when one of several timers expires (also asynchronous)
My application's handling of these async events needs to make a blocking call to send data over TCP and get a response before the next TCP data can be sent. E.G.
procedure OnSerialPortRxChar(...);
begin
if SendTCpData(...) = 'OK' then ...
end;
procedure OnTimerExpiry(...);
begin
if SendTCpData(...) = 'OK' then ...
end;
These should not interrupt each other, but currently do.
Obviously, my function SendTCpData needs some sort of blocking mechanism to prevent reentrant calls, or a queuing mechanism. Given that the caller needs to know the result, is my best solution a mutex? The problem is that the TCP transaction is just one line in the 20 line SendTCpData function which those asynch events can invoke.
I hope that I have explained this comprehendably. If not, please request more information.
Thank you very much in advance for your help.
If your serial library is AsyncPro, I would go with a single-threaded solution. There is more than one way to skin a cat, and I am not saying that you could not use a multi-threaded solution. But given that the AsycPro events will be running in the main thread, and you are not currently using threading in your application, this may be the simplest way forward.
In this solution we use a re-entry gate and a queue. The following is a mix of Delphi and pseudo-code.
var
isInTCP: boolean = False;
function DoSendTCP: boolean;
begin
isInTCP := True;
try
result := SendTCpData(...) = 'OK'
finally
isInTCP := False
end
end;
procedure OnSerialPortRxChar(...);
// This is safely re-entrant.
begin
repeat
if isInTCP then
Push the event onto a queue
else if DoSendTCP then
// Calling DoSendTCP may cause re-entry.
etc...
;
if (queue is empty) or isInTCP then break;
Pop from head of queue
until False
end;
You can use TIdAntiFreeze. Just drop it on your main form. Your call will still be blocked but your GUI will not be blocked. You may want to use some timeouts with your client though.
I have an Indy Server TIdTCPServer which has 3 bindings for different ports. If I connect a client to those 3 ports, and then deactivate the server, it gets stuck in what appears to be a deadlock. No matter what I do, it won't respond to my click, it won't even report "not responding" to Windows. If I disconnect the client(s) before deactivating the server, everything goes just perfect. I mean "deactivating" as in Server.Active:= False;.
Has anyone else experienced this? What might be causing it? I have nothing happening in here which crosses over threads which could in turn cause a deadlock (for example GUI updates). I tried an Antifreeze component TIdAntiFreeze but no luck.
TIdTCPServer is a multi-threaded component. A deadlock during server deactivation means that one or more of its client threads is not terminating correctly. That usually means that your server event handlers are doing something they should not be doing, typically either catching and discarding Indy's internal exceptions to itself, synchronizing with the thread context that is busy terminating the server, or deadlocking on something else outside of Indy. Without seeing your actual code, there is no way to know for sure which is actually the case, but it is always user error that causes this kind of deadlock.
TIdAntiFreeze only affects Indy components that run in the context of the main thread. TIdTCPServer does not.
I added this code on Form.OnClose works good!
procedure TformSFTP.FormClose(Sender: TObject; var Action: TCloseAction);
var
iA : Integer;
Context: TidContext;
begin
if sftpServidorFTP.Active then
with sftpServidorFTP.Contexts.LockList do
try
for iA := Count - 1 downto 0 do
begin
Context := Items[iA];
if Context = nil then
Continue;
Context.Connection.IOHandler.WriteBufferClear;
Context.Connection.IOHandler.InputBuffer.Clear;
Context.Connection.IOHandler.Close;
if Context.Connection.Connected then
Context.Connection.Disconnect;
end;
finally
sftpServidorFTP.Contexts.UnlockList;
end;
if sftpServidorFTP.Active then
sftpServidorFTP.Active := False;
end;
Version used: Delphi 7.
I'm working on a program that does a simple for loop on a Virtual ListView. The data is stored in the following record:
type TList=record
Item:Integer;
SubItem1:String;
SubItem2:String;
end;
Item is the index. SubItem1 the status of the operations (success or not). SubItem2 the path to the file. The for loop loads each file, does a few operations and then, save it. The operations take place in a TStringList. Files are about 2mb each.
Now, if I do the operations on the main form, it works perfectly.
Multi-threaded, there is a huge memory problem. Somehow, the TStringList doesn't seem to be freed completely. After 3-4k files, I get an EOutofMemory exception. Sometimes, the software is stuck to 500-600mb, sometimes not. In any case, the TStringList always return an EOutofMemory exception and no file can be loaded anymore. On computers with more memory, it takes longer to get the exception.
The same thing happens with other components. For instance, if I use THTTPSend from Synapse, well, after a while, the software cannot create any new threads because the memory consumption is too high. It's around 500-600mb while it should be, max, 100mb. On the main form, everything works fine.
I guess the mistake is on my side. Maybe I don't understand threads enough. I tried to free everything on the Destroy event. I tried FreeAndNil procedure. I tried with only one thread at a time. I tried freeing the thread manually (no FreeOnTerminate...)
No luck.
So here is the thread code. It's only the basic idea; not the full code with all the operations. If I remove the LoadFile prodecure, everything works good. A thread is created for each file, according to a thread pool.
unit OperationsFiles;
interface
uses Classes, SysUtils, Windows;
type
TOperationFile = class(TThread)
private
Position : Integer;
TPath, StatusMessage: String;
FileStringList: TStringList;
procedure UpdateStatus;
procedure LoadFile;
protected
procedure Execute; override;
public
constructor Create(Path: String; LNumber: Integer);
end;
implementation
uses Form1;
procedure TOperationFile.LoadFile;
begin
try
FileStringList.LoadFromFile(TPath);
// Operations...
StatusMessage := 'Success';
except
on E : Exception do StatusMessage := E.ClassName;
end;
end;
constructor TOperationFile.Create(Path : String; LNumber: Integer);
begin
inherited Create(False);
TPath := Path;
Position := LNumber;
FreeOnTerminate := True;
end;
procedure TOperationFile.UpdateStatus;
begin
FileList[Position].SubItem1 := StatusMessage;
Form1.ListView4.UpdateItems(Position,Position);
end;
procedure TOperationFile.Execute;
begin
FileStringList:= TStringList.Create;
LoadFile;
Synchronize(UpdateStatus);
FileStringList.Free;
end;
end.
What could be the problem?
I thought at one point that, maybe, too many threads are created. If a user loads 1 million files, well, ultimately, 1 million threads is going to be created -- although, only 50 threads are created and running at the same time.
Thanks for your input.
There are (probably) no leaks in the code you show in the question.
I say probably because an exception raised during the Execute could result in a leak. The lifetime of the string list should be protected by a finally block.
FileStringList:= TStringList.Create;
try
LoadFile;
Synchronize(UpdateStatus);
finally
FileStringList.Free;
end;
That said, I expect the exception swallow in LoadFile means that you don't leak the string list.
You say that perhaps thousands of threads are created. Each thread reserves memory for its stack, and the default stack size is 1MB. Once you have thousands of 1MB stacks reserved, you can easily exhaust or fragment address space.
I've seen problems due to cavalier creation of threads in the past. For example I had a program that failed when it created and destroyed threads, with never more than 256 threads in existence. This was on a 16 core machine with 4GB address space. You probably have 2GB address space available.
Although you state that no more than 50 threads are in existence at any one moment, I'm not sure how you can be sure of that. Not least, because you have set FreeOnTerminate to True and thereby surrendered control over the lifetime of your threads.
My guess is that your problems are related to the number of threads you create. One thread per processor will suffice. Re-use your threads. It's expensive to create and destroy a thread for a small task.
If this is not enough to solve your problems then you will need to show the code that manages thread lifetime.
Finally, I wonder how much benefit you will extract from threading this app. If it is IO bound then the threaded version may well be slower!
Based on the information given, it's not possible to reproduce your error.
Some hints are made by Remy and David which might help you.
Looking at the structure of your program, the flow can be divided into two classical solutions.
The first part where you are delegating tasks to different threads, is a Single-Producer-Multiple-Consumer problem.
Here it can be solved by creating a small number of threads, passing them a thread-safe object queue.
The main thread then pushes the task objects into the queue. The consumer threads takes care of the individual file checking tasks.
The second part where the result is to be transfered to the main thread is a Multiple-Producer-Single-Consumer problem.
If you pass a second thread-safe object queue to the threads at initialization, they can easily put the results into the queue.
Drain the result queue from the main thread within a timer event.
In my multithread application
I use TThread.suspend and TThread.resume
Since moving my application to Delphi 2010 I get the following warring message
[DCC Warning] xxx.pas(277): W1000 Symbol ‘Resume’ is deprecated
If Resume is deprecated what should be used in place?
EDIT 1:
I use the Resume command to start the thread - as it is Created with 'CreateSuspended' set to True and Suspend before I terminate the thread.
EDIT 2:
Here is a link the delphi 2010 manual
Charles if do you read the code of TThread class , do you find the answer.
TThread = class
private type
..
..
..
public
constructor Create(CreateSuspended: Boolean);
destructor Destroy; override;
procedure AfterConstruction; override;
// This function is not intended to be used for thread synchronization.
procedure Resume; deprecated;
// Use Start after creating a suspended thread.
procedure Start;
// This function is not intended to be used for thread synchronization.
procedure Suspend; deprecated;
procedure Terminate;
See this link RAD Studio 2010: Community pulse: The day after. (Part 2)
Edit:
If you need to synchronize threads, you can use a scheme based on TMutex, TEvent and critical sections.
Bye.
Use TThread.Start instead of .Resume
--EDIT--
Start can of course only be used with Delphi 2010 (and later, presumably) to start a thread that was created suspended (where you would have used Resume before).
Using Resume/Suspend (or corresponding WinAPI functions) for thread synchronisation is NOT recommended. See the discussion here (have a look at Barry Kelly's comments).
Suspend and Resume were (or used to be) potentially broken in the TThread class (if you look at the source you will see that the Suspend method directly and unconditionally sets a Boolean to indicated thread suspended state rather than more robustly deriving this state from the execution count on the thread handle. Ironically the Resume method does use this more robust indicator to update the suspended state Boolean).
This is possibly why they have been deprecated. It's also why I implemented my own class to encapsulate a Windows thread with a more robust suspend and resume mechanism as well as the ability to Restart a thread once it had completed.
I'm not sure why their deprecation is supposedly related to synchronization. Suspending and resuming threads is not necessarily related to synchronization, although I can see how it might be. It's interesting to note that the equivalent methods in the .NET framework Thread class are similarly marked as obsoleted. And the same comments w.r.t synchronization appear in the Windows API documentation for thread suspend/resume.
If using deprecated methods makes you nervous and you still wish to suspend/resume you could always use the Windows API to suspend and resume the thread by reference to it's handle.
Just in case all you wanted to do was get rid of the compiler hints
(1) To get rid of the compiler hint when Starting a Thread ...
replace
MyThread := TMyThread.Create(True);
MyThread.Resume;
with
MyThread := TMyThread.Create(True);
MyThread.Start;
(2) To get rid of the compiler hint when Stopping a Thread ...
replace
MyThread.Suspend;
MyThread.Terminate;
with
MyThread.Terminate;
Not a big deal at all. Beware of attempted obfuscation.
Thread behavior control code should lie with in a thread procedure. Use appropriate sync objects and corresponding API calls in order to suspend/resume thread execution. Doing it from outside is a dangerous practice. So there was a decision to depricate it.
Use
Suspended := False; // Resume;
or
Start;
You should create the thread as follows:
constructor TSignalThread.Create;
begin
// create event handle first!
FEventHandle := CreateEvent(
{security} nil,
{bManualReset} true,
{bInitialState} false,
{name} nil);
FWaitTime := 10;
inherited Create({CreateSuspended}false);
end;
This way a call to Start is not required.
See http://www.gerixsoft.com/blog/delphi/creating-threads for an explanation why this code works.
#mghie (a little late, I know)
take for example madexcept and alike. If your application crashes and a bug report is being shown to user, that means that the dialog waits for use input. If it so happens that the crash is a result of a thread action (not necessarily a thread crashing), if you do not suspend the threads, the screen will be filled with bug report dialogs.
example 2: logging. for any particular reason, I at least, had the need to log some threads execution state. That includes the current stack trace. Now, as you (should) know, you cannot do that while the thread is running because during the time you collect information about it, the threads keeps doing stuff so by the time you finish collection, the information collected will not be consistent. Hence, you need to suspend the thread.
And I can go on with practical examples on thread management. Granted, these are not things you do in every day programming, but at least the first example I am sure that many of you are using, even if you are not aware of the internals of it. Debuggers? again, you use them.
But indeed, in all these cases, TThread is not used, since the work is done on thread handles. So, indeed, a valid example of TThread suspend usage is hard to come by. But threads in general, that's another story.