Freeing dynamic allocated memory of structure

[darekg11]

Hello guys, I have qusetion, I have following structure:

typedef struct
{
	GtkWidget* PoziomaLinijka; <--GTK
	GtkWidget* PionowaLinijka; <--GTK
	GtkWidget* Label1; <--GTK
	GtkWidget* Label2; <--GTK
	gint x,y; <--GTK
} StrukturaDrawing;

And now I dynamicly allocate memory for it:

StrukturaDrawing* Wsk;
  Wsk = (StrukturaDrawing*)malloc(sizeof(StrukturaDrawing));
  if (!Wsk)
  {
	std::cout << "Error" << std::endl; 
  }

It works great, it allocates it. Now I use GTK callbacks to do stuff with it, which works great but when I want to free memory of it:
Callback to function: Wyjscie:

 g_signal_connect(G_OBJECT(Okno), "destroy", G_CALLBACK(Wyjscie), Wsk);

void Wyjscie(GtkWindow* window, GdkEvent* event, StrukturaDrawing* data)
{
	gtk_main_quit();
	free(data);
	data = NULL;
}

I get:
First-chance exception at 0x7c96df51 in Ekseprymetnowanie.exe: 0xC0000005: Access violation reading location 0xfffffff9.

But if I write data = NULL and then free(data) which is nonsense, right? I get no warnings, also if I comment out freeing memory I get no error but then I will get memory leaks.

Thanks in advance guys.

[Bayint Naung]

Is it C or C++? If it's C, use printf, not cout and remove the cast from malloc().
Use a debugger. You can use valgrind also.

[\007]

GTK is done in C, and has C++ bindings. What are you doing exactly? Looks like a mix of C/C++.

[darekg11]

std::cout was just for test because I use Visual C++ Compilator because I got used to it and I need to use cast on malloc because it's c++ compilator. And I use C version of GTK.

[tabstop]

So 0xfffffff9 is, what, -7? Somehow along the way, the pointer you malloc and the pointer you try to free are not the same. I don't know enough about GTK to know what the parameters to pass in to a destroy handler are supposed to be. (EDIT: My point there, which I appear to have forgotten to mention, is that if the number/type of things your function expects don't match what a destroy handler is supposed to take, then you can wind up in bad places, since you aren't calling the function yourself -- the main loop is calling it for you, and it will pass in whatever it thinks a destroy handler is supposed to need.)

Not that it matters here, since destroy should probably only be called once, but you do realize that data = NULL has no effect? If you're expecting Wsk to be changed in the calling function, it won't.

[laserlight]

std::cout was just for test because I use Visual C++ Compilator because I got used to it and I need to use cast on malloc because it's c++ compilator. And I use C version of GTK.

That still leaves the question unanswered: is this C or C++? If you really want to code in C, then you should also test using C specific constructs, compiling with a C compiler (in this case this means configuring your compiler to compile the source code as C).

[darekg11]

I code in C, I know that I use C++ compiler.

I will go deeper into GTK documentations about main lopp of it and see what I can find.

[tabstop]

If you're coding in C, why not use a C compiler? There's one in VS, somewhere.

Anyway, another thing I thought of as you were looking at your code, related to the data=NULL line above: you are passing your pointer to the function by value. So, based on my limited understanding of how the gtk library works, the registration you do in the g_signal_connect function would then be based on the value Wsk holds at the instant you register the callback. If you do the malloc later, then that doesn't help, since that won't change the value that the callback function has. In this case, you would actually want

void Wyjscie(GtkWindow* window, GdkEvent* event, StrukturaDrawing** data)
{
	gtk_main_quit();
	free(*data);
	*data = NULL;
}

and pass in &Wsk to your g_connect_signal thing.

[darekg11]

I must say that I tried that before and with that I get:
0xC0000005: Access violation reading location 0x00000001.

EDIT: While debugging I have looked at data structure at the point where it crashed and members of structure have that error:
The first one has: PoziomaLinijka CXX0017: Error: symbol "" not found
And later the whole rest have: PionowaLinijka CXX0030: Error: expression cannot be evaluated

[darekg11]

It's fixed. Strucutre of callback function was wrong, now it runs without any violations but I have one question:

void Wyjscie(GtkObject* window, StrukturaDrawing* data)
{
	gtk_main_quit();
	free(data);
	data = NULL;
}

Data won't be set to NULL right? Adress will be set to NULL, that function should look like:

void Wyjscie(GtkObject* window, StrukturaDrawing** data)
{
	gtk_main_quit();
	free(*data);
	*data = NULL;
}

And I pass to function &Wsk instead of Wsk, am I right?

[User Name:]

To tell VC you want it to compile like a C compiler, change the extension of the file to .c and you'll get an (almost) proper C compiler.

[tabstop]

It's fixed. Strucutre of callback function was wrong, now it runs without any violations but I have one question:

void Wyjscie(GtkObject* window, StrukturaDrawing* data)
{
	gtk_main_quit();
	free(data);
	data = NULL;
}

Data won't be set to NULL right? Adress will be set to NULL, that function should look like:

void Wyjscie(GtkObject* window, StrukturaDrawing** data)
{
	gtk_main_quit();
	free(*data);
	*data = NULL;
}

And I pass to function &Wsk instead of Wsk, am I right?

That is how you get changes to appear in the calling function as well.

[darekg11]

Yeah so it's right way to do it?

[laserlight]

Yes.

[darekg11]

I guess I will leave it as it was before because when doing it with this new function:

void Wyjscie(GtkObject* window, StrukturaDrawing** data)
{
	gtk_main_quit();
	free(*data);
	*data = NULL;
}

I get again acces violation and when I do it in old way I don't get any warnings or errors.

Anyway thanks for help.