RingBuffer for data collection - c++

Question

I have a Ring Buffer implementation that I like to use to process an incoming data. Is the following approach safe and efficient to use considering synchronizations needed?.
void CMyDlg::MyButton1()
{
RingBuffer BufRing(10000);
unsigned char InputBuf[100];
unsigned char OutBuf[100];
for (int ii = 0; ii < 1000; ++ii)
{
for (int i = 0; i < 100; ++i)
{
InputBuf[i] = i;
}
BufRing.Write(InputBuf,100);
BufRing.Read(OutBuf,100);
AfxBeginThread(WorkerThreadProc,OutBuf,THREAD_PRIORITY_NORMAL,0,0,NULL);
}
}
UINT WorkerThreadProc( LPVOID Param )
{
unsigned char* pThreadBuf = (unsigned char*)Param;
for (int c = 0; c < 100; ++c)
{
TRACE("Loop %d elemnt %x\n",c,pThreadBuf[c]);
}
return TRUE;
}

Answer 1

Looks hazardous to me...
void CMyDlg::MyButton1()
{
// ...
unsigned char OutBuf[100];
for (int ii = 0; ii < 1000; ++ii)
{
// ...
BufRing.Read(OutBuf,100);
AfxBeginThread(WorkerThreadProc,OutBuf,THREAD_PRIORITY_NORMAL,0,0,NULL);
}
}
The problem that I see is that you're using a single buffer (OutBuf) to store data in, passing it to a worker thread, and then modifying that same buffer in the next iteration of your loop.
Your test code won't reveal this, because you're simply repopulating OutBuf with the same values in every iteration (as far as I can tell, anyway). If you changed InputBuf[i] = i; to InputBuf[i] = ii; and included a unique thread ID in your TRACE output, you'd probably see suspicious behaviour.