memcpy error

[sap00]

I have an application that will capture sound from a mic in real-time. The application runs but when I have finished recording the sound I am receiving an error with the memcpy command. The error outputs a messagebox that says:

"Unhandled exception at 0x004739ca in capturesound.exe: 0xC0000005: Access violation writing location 0x00000000."

This error is on the "memcpy.asm" which is on the line:

"rep movsd ;N - move all of our dwords"

Here is the part where I used memcpy:

//-----------------------------------------------------------------------------
// Name: RecordCapturedData()
// Desc: Copies data from the capture buffer to the output buffer
//-----------------------------------------------------------------------------
HRESULT RecordCapturedData()
{
HRESULT hr;
VOID* pbCaptureData = NULL;
DWORD dwCaptureLength;
VOID* pbCaptureData2 = NULL;
DWORD dwCaptureLength2;
DWORD dwDataWrote;
DWORD dwReadPos;
DWORD dwCapturePos;
LONG lLockSize;
void* buf;

if( NULL == g_pDSBCapture )
return S_FALSE;
//if( NULL == g_pWaveFile )
// return S_FALSE;

if( FAILED( hr = g_pDSBCapture->GetCurrentPosition( &dwCapturePos, &dwReadPos ) ) )
return DXTRACE_ERR_MSGBOX( TEXT("GetCurrentPosition"), hr );

lLockSize = dwReadPos - g_dwNextCaptureOffset;
if( lLockSize < 0 )
lLockSize += g_dwCaptureBufferSize;

// Block align lock size so that we are always write on a boundary
lLockSize -= (lLockSize % g_dwNotifySize);

if( lLockSize == 0 )
return S_FALSE;

// Lock the capture buffer down
if( FAILED( hr = g_pDSBCapture->Lock( g_dwNextCaptureOffset, lLockSize,
&pbCaptureData, &dwCaptureLength,
&pbCaptureData2, &dwCaptureLength2, 0L ) ) )
return DXTRACE_ERR_MSGBOX( TEXT("Lock"), hr );


if(pbCaptureData) memcpy(buf,pbCaptureData,dwCaptureLength);
if(pbCaptureData2) memcpy((char*)
buf+dwCaptureLength,pbCaptureData2,dwCaptureLength2);


// Move the capture offset along
g_dwNextCaptureOffset += dwCaptureLength;
g_dwNextCaptureOffset %= g_dwCaptureBufferSize; // Circular buffer


// Move the capture offset along
g_dwNextCaptureOffset += dwCaptureLength2;
g_dwNextCaptureOffset %= g_dwCaptureBufferSize; // Circular buffer
}

// Unlock the capture buffer
g_pDSBCapture->Unlock( pbCaptureData, dwCaptureLength,
pbCaptureData2, dwCaptureLength2 );


return S_OK;
}

[namezero111111]

Hi.

You're attempting to copy data to a pointer that points to NULL.
E.g. you haven't allocated memory for the ptr.


-Andy

[cilu]

Take a look at this FAQ.

[sap00]

I have opened the Call Stack and the green arrow points to this message: ">capturesound.exe!memcpy(unsigned char * dst=0x00000000, unsigned char * src=0x013b0028, unsigned long count=65000) Line 188 Asm"

I do not know how to debug this kind of error. Is this because of the allocation of the memory for the destination "buf"? If yes, how can I make a memory space for that. Can anyone help...

here is my code:

[source]
//-----------------------------------------------------------------------------
// Name: RecordCapturedData()
// Desc: Copies data from the capture buffer to the output buffer
//-----------------------------------------------------------------------------
HRESULT RecordCapturedData()
{
HRESULT hr;
VOID* pbCaptureData ; //= NULL;
DWORD dwCaptureLength;
VOID* pbCaptureData2 ;//= NULL;
DWORD dwCaptureLength2;
DWORD dwDataWrote;
DWORD dwReadPos;
DWORD dwCapturePos;
LONG lLockSize;
void* buf;

if( NULL == g_pDSBCapture )
return S_FALSE;

if( FAILED( hr = g_pDSBCapture->GetCurrentPosition( &dwCapturePos, &dwReadPos ) ) )
return DXTRACE_ERR_MSGBOX( TEXT("GetCurrentPosition"), hr );

lLockSize = dwReadPos - g_dwNextCaptureOffset;
if( lLockSize < 0 )
lLockSize += g_dwCaptureBufferSize;

// Block align lock size so that we are always write on a boundary
lLockSize -= (lLockSize % g_dwNotifySize);

if( lLockSize == 0 )
return S_FALSE;

// Lock the capture buffer down
if( FAILED( hr = g_pDSBCapture->Lock( g_dwNextCaptureOffset, lLockSize,
&pbCaptureData, &dwCaptureLength,
&pbCaptureData2, &dwCaptureLength2, 0L ) ) )
return DXTRACE_ERR_MSGBOX( TEXT("Lock"), hr );


if(pbCaptureData) memcpy(buf,pbCaptureData,dwCaptureLength);
if(pbCaptureData2) memcpy((char*)buf+dwCaptureLength,pbCaptureData2,dwCaptureLength2);

g_writefile = WriteFile(g_hCreateFile,
pbCaptureData,
dwCaptureLength,
&dwDataWrote,
NULL);

// Move the capture offset along
g_dwNextCaptureOffset += dwCaptureLength;
g_dwNextCaptureOffset %= g_dwCaptureBufferSize; // Circular buffer


g_writefile = WriteFile(g_hCreateFile,
pbCaptureData2,
dwCaptureLength2,
&dwDataWrote,
NULL);

// Move the capture offset along
g_dwNextCaptureOffset += dwCaptureLength2;
g_dwNextCaptureOffset %= g_dwCaptureBufferSize; // Circular buffer


// Unlock the capture buffer
g_pDSBCapture->Unlock( pbCaptureData, dwCaptureLength,
pbCaptureData2, dwCaptureLength2 );

return S_OK;

}

[/source]

[cilu]

With this:

VOID* pbCaptureData ; //= NULL;
DWORD dwCaptureLength;
VOID* pbCaptureData2 ;//= NULL;

you made the worst programming error: left your variables uninitialized.

It seems you don't understand the problem: you have to allocate memory for the buffers you are using, and failure to do that will end in a run-time failure of your application.

[sap00]

That part is originally written in this way...

VOID* pbCaptureData ; = NULL;
DWORD dwCaptureLength;
VOID* pbCaptureData2 ;= NULL;

I have confirmed that my buffers are working properly, not until that I want to copy the data in the capture buffer to a memory space for analysis lets say fourier transform of the data. I am passing the data to "buf" which I have initialized to "void* buf;" is memory allocated in this manner?

[Paul McKenzie]

I have confirmed that my buffers are working properly,

So everything is OK then, right? Or is there still a problem?

I am passing the data to "buf" which I have initialized to "void* buf;" is memory allocated in this manner?

All you've declared is a pointer -- nothing is allocated.

The pointer must point to memory that you've allocated dynamically, or a static buffer. Either way, both must be large enough to hold the data you're copying.

Regards,

Paul McKenzie

[Mogulbasher]

When you declare a pointer you must allocate memory to it..

eg...

char * pData;
pData = new char[20];

This is rather simple but it emphasizes the point that we have now allocated effectively 20 bytes of memory which we can then memcpy into.

Alternatively if you have a data structure to accept TGA files

struct TgaHeader {
		char    IDLen;         //length of ID field
		char    MapType;       //0=none, 1=present
		char    ImgType;       //0=none, 1=c-map, 2=true, 3=b&w, 9=rle c-map, 10=rle true, 11=rle b&w
		short   FirstColor;    //where to start in colormap
		short   MapLen;        //number of colors to set
		char    EntrySize;     //number of bits per color entry
		short   Xorigin;
		short   Width;
		short   Height;
		char    Depth;
		char    Descriptor;
	};

When we want to load a new TGA we will declare a pointer to the texture

TgaHeader * pTGA;

But we then must allocate memory to it

pTGA = new TgaHeader;

In your example above you declare void * buf but you never tell it how much memory you want to reserve for buf

Something like this might work...although my c++ skills are rather limited

unsigned uDecompressBuffer[ MAX_CAPTURE_SIZE];
		pbufPointer = (unsigned *)(((char *)&uDecompressBuffer) );
		buf = (void *)pbufPointer;

Where you can define a max capture size in the h file. Alternatively if you know what the capture size is replace MAX_CAPTURE_SIZE with a variable that is at least equal to the size of the capture stream