Back to For Beginners

void pointer addition MemoryCopy

For Beginners

Started by thestien March 21, 2012 11:50 AM

7 comments, last by SiCrane 12 years ago

thestien

102

Author

March 21, 2012 11:50 AM

Hi Guys im back with more problems from my new book...

in the book it makes a Line drawing class which looks ok but the render function it uses doesnt work i tried messing around with it but still no joy.

i think it is because of the MemoryCopy function ill show the rough code.





void* pData = 0;



r = pVB->Lock( 0, 0, (void**)&pData, 0);

    if( FAILED(r))

    {

        pVB->Release();

        return E_FAIL;

    }

    CopyMemory( pData, (void*)&m_StartPoint, sizeof( CZenVertex));

    CopyMemory( pData + sizeof(CZenVertex), (void*)&m_EndPoint, sizeof( CZenVertex));



    pVB->Unlock();



more code to draw primitive here...

the problem is it dont know how to count from the void pointer address?

there is some code missing but hopefully from this snippet you can understand what im trying to do and how i should be doing it

im not sure why the book makes void* i think its needed as a void** in another function before the memorycopy

many thanks in advance

EDIT: im sorry what i need is a way to find the address of pData and add the sizeof(Vertex) to that as the starting point to copy the Endpoint Vertex into the Vertex Buffer the start point goes in there easy enough though

Serapth

6,672

March 21, 2012 04:33 PM

You are copying memory into an unallocated pointer, unless CopyMemory is doing the allocation, this is a big bad no-no. Assuming that CopyMemory in question is this CopyMemory it certainly *doesnt* allocate the memory for you.

As to a counting from the void pointer address, it's as simple as pData+4 for example, this would return the position 4 bytes past the point that pData points at. Unliked typed pointers though, you can't assume the size of a void pointer, as a void pointer is sizeless.

Truth told, unless you have a very good reason, you shouldn't really be using void pointers.

As to finding the address of pData, since it is a pointer, it IS the address address in memory. That said, if you want to assigned a pointer, to a pointer to point to a pointer, it works like the following:

int p = 42;
int * pointerToP = &p;
int ** pointerToPointerToP = &pointerToP;

As to your wanting to copy data to a buffer, there are much simpler ways that do not involve pointers to pointers, without seeing the exact code you are working with there isn't much we can help with. Specifically what exactly are pVB, m_EndPoint, m_StartPoint, what is the error you are getting, and is CopyMemory just the Win32 api call?

Huge Collection of Game Development Tutorials || Getting Started? Start Here

Ectara

3,097

March 21, 2012 04:58 PM

It seems as if some sort of rendering context is mapping pData to some sort of data block in the locking process, as if it is copying a hardware buffer to memory for modification, to be copied back, judging by passing a pointer to pData earlier. Thus, it likely now points to valid data that shouldn't be referenced after the context is unlocked.

As for incrementing a void pointer, it might be taboo based on whatever rules you follow, but the easiest way to increment a pointer is to cast it.

CopyMemory( (void*)((char*)pData + sizeof(CZenVertex)), (void*)&m_EndPoint, sizeof( CZenVertex));

Evil Steve

2,021

March 21, 2012 05:00 PM

pData is allocated by the Lock() call, so this is fine.

What the OP probably wants is:



BYTE* pData = 0;

The problem is that adding an integer to a pointer adds X elements, not X bytes. Because pData is a void pointer, it's size is unknown. If you use a BYTE pointer instead, then the size of an element is 1 byte.

You could also use a CZenVertex pointer instead, and move forwards by 1 element.

21st Century Moose

13,459

March 21, 2012 07:38 PM

Evil Steve nailed it, so the only thing worth adding is: "void *pData = NULL;" not "0".

Direct3D has need of instancing, but we do not. We have plenty of glVertexAttrib calls.

thestien

102

Author

March 21, 2012 10:00 PM

Wow thanks for the replies guys its got my code Sort of working

i would post the entire code but its 2k+ lines so far and very messy the code above is fixed with the BYTE* what i wanted to do is at the address of pData copy one vertex worth of data then from the end of that data copy the last vertex data.

it is very strange i managed to fix this but my code is so messed up it still doesnt work as intended but that is for another topic

my code is one file at the moment and i am going to split it up into seperate files to make it more readable for me and anyone hopeing to help me.

im sure in a few weeks ill look back at this thread and realize how much of an idiot i am lol

thanks for the help guys much appreciated

Washu

7,836

March 21, 2012 10:05 PM

Evil Steve nailed it, so the only thing worth adding is: "void *pData = NULL;" not "0".

0 Is perfectly valid in C++ and evaluates to the null pointer value.

In time the project grows, the ignorance of its devs it shows, with many a convoluted function, it plunges into deep compunction, the price of failure is high, Washu's mirth is nigh.

ApochPiQ

23,138

March 21, 2012 10:33 PM

In fact NULL is a non-standard and potentially not-entirely-portable macro that evaluates to 0 on many platforms (chiefly MS Windows).

You should be using nullptr and a C++11 compliant compiler :-P

Wielder of the Sacred Wands
[Work - ArenaNet] [Epoch Language] [Scribblings]

SiCrane

11,840

March 21, 2012 11:54 PM

NULL is standard - just what it exactly evaluates to is implementation defined. All you know about it is that it can be used as a null pointer constant.

void pointer addition MemoryCopy

This topic is closed to new replies.

Popular Topics

Recommended Tutorials

void pointer addition MemoryCopy

This topic is closed to new replies.

Popular Topics

Recommended Tutorials

Reticulating splines