C / C++ / MFC

19-Mar-09 10:45

I am currently writing a multi-threaded application using C++ and MFC. I am trying to write some code to keep track of my memory allocations (new) and my memory frees (delete). I have written aclass called MemoryChecker that is suppose to do this. To store the list of pointers that are allocated,
I am using STL which is not thread safe. To make it thread safe, I put a call to WaitForSingleObject
around the body of the function. I would expect to make this thread safe. Inside the member function
deleteItem, there is a static variable. It is initialized to true but later set to false. Before it is set to false, there is an assert statement that checks that the value is still true. However,
that assert statement is failing (sometimes). Therefore, I am thinking my lock is not working properly.

<br />
void<br />
MemoryChecker::deleteItem( void *ptr )<br />
{<br />
    if ( WaitForSingleObject(mutexDeleteHandle, INFINITE) == WAIT_OBJECT_0 ) {<br />
	if ( ptr == 0 ) {<br />
	    ReleaseMutex(mutexDeleteHandle);<br />
	    returneturn;<br />
	}<br />
	static bool go = true;<br />
	bool found = false;<br />
	NewList::iterator it = newList.begin();<br />
	while ( it != newList.end() ) {<br />
		if ( it->ptrValue == ptr ) {<br />
			found = true;<br />
			break;<br />
		}<br />
		it ++;<br />
	}<br />
	if ( found == false )<br />
		found = true;<br />
	else {<br />
		assert( found );<br />
		newList.erase( it );<br />
	}<br />
	assert( go );<br />
	go = false;<br />
	ReleaseMutex(mutexDeleteHandle);<br />
}<br />
}

I should also tell you that I create the lock, in another thread, with the following call:

mutexDeleteHandle = CreateMutex( NULL, FALSE, NULL );

Please note that the second parameter in CreateMutex is to specify who owns the lock and the
lock may not be owned by the thread making the call to WaitForSingleObject.

I am hoping that somebody here can tell me what I am doing wrong.

Thanks

Bob

modified on Thursday, March 19, 2009 6:26 PM

Code-o-mat19-Mar-09 12:31

Code-o-mat

19-Mar-09 12:31

Your 'go' variable is static, it will be set to true once, so after you do 'go = false' it will remain false in following calls too.

p.s: Please put all of your code between <pre> and </pre> tags to make it easier to read.

> The problem with computers is that they do what you tell them to do and not what you want them to do. <
> Life: great graphics, but the gameplay sux. <

BobInNJ19-Mar-09 12:39

19-Mar-09 12:39

Thanks for the response. You are 100% right. I am not awake today. I need to reinitialize the go flag every time to make it a valid test.

Now for your second point. I put the code between <pre> and </pre> but it did not preserve the white spacing. I do not know why. Any ideas?

Bob

bulg19-Mar-09 13:39

bulg

19-Mar-09 13:39

Sorry, I see one small <pre> tag, and a large, only-half-there <code> tag.

BobInNJ19-Mar-09 13:58

19-Mar-09 13:58

blug,

Thanks for the response. I am not sure. Before I include code, I click on the formatting option code block. That is the option just to the left of inline code and to the right of BIG. Maybe, I should click on inline code and paste my code in there. Not Sure.

Bob

Threads and Pointers

BobInNJ19-Mar-09 7:36

19-Mar-09 7:36

I am currently developing a multi-threaded application using C++ and MFC. Each thread calls new and delete to allocate and free memory. I am wondering if this is ok. In particular, if I have a routine that does the following, is it thread safe?

<br />
void<br />
WorkerClass::genRandomNumbers( double *randomNumbers, int count )<br />
{	<br />
    double *ranList = new double[6];<br />
<br />
    // Do some work, details not shown<br />
<br />
    delete []ranList;<br />
}<br />

I am thinking that if two different threads are executing this routine at the same time, one might be allocating memory and the other freeing memory. Is that ok? One possibility is that Microsoft has put locks in there code to make sure that new and delete are thread safe. I have been unable to find documentation saying that they have or have not done this. I am hoping somebody out there can enlighten me on this issue.

Thanks

Bob

led mike19-Mar-09 7:41

led mike

19-Mar-09 7:41

BobInNJ wrote:
I am thinking that

My advice is not to use your own thinking about how multi-threading works. I strongly urge you to find some material designed to explain or teach multi-threading concepts and study it.

If this is something you decide to do, I can tell you that back when was starting out with threading I found this book[^] (the chapters on threading) very helpful. It was the first edition and I can't verify for you that the new editions contain the same material.

Eytukan19-Mar-09 7:41

19-Mar-09 7:41

Perfectly safe. Unless you allocate something static. like

void
WorkerClass::genRandomNumbers( double *randomNumbers, int count )
{
 static int i =0; //Not safe.

 i = blahaha

}

He never answers anyone who replies to him. I've taken to calling him a retard, which is not fair to retards everywhere.-Christian Graus

led mike19-Mar-09 7:58

led mike

19-Mar-09 7:58

VuNic wrote:
Perfectly safe

Really? Even though you have no idea what is happening in the "details not shown"?

Eytukan19-Mar-09 8:12

19-Mar-09 8:12

Oops! that was not to negate your message up there.. as you see it's all posted at the same time. Smile | :)

"Perfectly-Safe" is perfectly restricted to that block of code he posted. Certainly I wanted to add a note asking him to look at the JR book you hrefed. but felt quite lazy to edit the text again Roll eyes | :rolleyes:

He never answers anyone who replies to him. I've taken to calling him a retard, which is not fair to retards everywhere.-Christian Graus

CPallini19-Mar-09 7:46

19-Mar-09 7:46

Automatic variables (like ranlist) are thread safe.
Smile | :)

If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.

This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke

[My articles]

led mike19-Mar-09 9:42

led mike

19-Mar-09 9:42

Everyone has there own definition of "mean", that seems far meaner than any of the offensive things I post to people! Laugh | :laugh:

Edit control disappears from dialog [modified]

cmk19-Mar-09 13:17

cmk

19-Mar-09 13:17

new will call malloc, which (eventually) calls HeapAlloc on the module CRT heap.

If you look at the Heap* API function docs you will see that they are thread safe by default (though this feature can be disabled) ... actually, to be precise, malloc will use a CRT lock (critical section) to lock the heap, but the effect is the same.

So yes, new and delete are thread safe.

As to your example; as led mike pointed out, it depends on what you are doing in 'Do some work, details not shown'.

...cmk

The idea that I can be presented with a problem, set out to logically solve it with the tools at hand, and wind up with a program that could not be legally used because someone else followed the same logical steps some years ago and filed for a patent on it is horrifying.
- John Carmack

rp_suman19-Mar-09 7:07

rp_suman

19-Mar-09 7:07

I have an edit control in a CFormview dialog.
The edit control is DDXed to a some special class which is giving hyperlink functionality.
The control is disappearing or its background becomes same as the dialog sometime and when mouse cursor goes on top of it, it appears from then.
It happens mostly when setting text on that control (by function call) and
the control's class calls "RedrawWindow()", "InvalidateRect" & "UpdateWindow()" for the parent dialog(CFormView) of control when setting text on it.

The problem is happening in Win XP system and not in Vista.

The Formview is using "CreateSolidBrush()" and "FillRect()" function inside OnPaint() to change its back color.

Any suggstion would be greatly appreciable..

--
"Programming is an art that fights back!"

modified on Thursday, March 19, 2009 1:14 PM

Re: Edit control disappears from dialog

Code-o-mat19-Mar-09 7:27

Code-o-mat

19-Mar-09 7:27

Could it be that your edit is overlapped by some other control like a static for example?

> The problem with computers is that they do what you tell them to do and not what you want them to do. <
> Life: great graphics, but the gameplay sux. <

C4244 conversion from 'double' to 'float' warning

penny black19-Mar-09 6:47

penny black

19-Mar-09 6:47

Running VS2008, writing an SDI in VC++. Here's the code in a nutshell:

float htmin;
CString value;
value=mslstr.Mid(parse_start, parse_stop-parse_start);
htmin=atof(value);

VS throws the C4244 warning at the htmin=atof(value); line.

Am I missing something stupid here?

Re: C4244 conversion from 'double' to 'float' warning

Eytukan19-Mar-09 6:56

Re: C4244 conversion from 'double' to 'float' warning

19-Mar-09 6:56

Nothing. atof returns you a double, and you are assigning it to a float. Nothing serious.

He never answers anyone who replies to him. I've taken to calling him a retard, which is not fair to retards everywhere.-Christian Graus

penny black19-Mar-09 7:02

penny black

19-Mar-09 7:02

Thx, stickler for no warnings. Is only remedy to initialize htmin as double?

Re: C4244 conversion from 'double' to 'float' warning [fixed]

CPallini19-Mar-09 7:21

Re: C4244 conversion from 'double' to 'float' warning

19-Mar-09 7:21

penny black wrote:
Is only remedy to initialize htmin as double?

An explicit cast will do the trick

htmin = static_cast<float>(atof(value));

BTW atof is just an axample of misnamed function.
Smile | :)

modified on Thursday, March 19, 2009 2:46 PM

Eytukan19-Mar-09 7:29

Re: C4244 conversion from 'double' to 'float' warning

19-Mar-09 7:29

another way to eat the warning Cool | :cool:

std::istringstream b("100.003133");
float f1;
b >> f1;

He never answers anyone who replies to him. I've taken to calling him a retard, which is not fair to retards everywhere.-Christian Graus

CPallini19-Mar-09 7:38

Re: C4244 conversion from 'double' to 'float' warning

19-Mar-09 7:38

very good, indeed.
Smile | :)

penny black19-Mar-09 7:36

penny black

19-Mar-09 7:36

static_cast gave me syntax error.

Re: C4244 conversion from 'double' to 'float' warning

CPallini19-Mar-09 7:41