Re: Reading Serial Port
"clinisbut" <clinisbut@gmail.com> wrote in message
news:2658f726-3a6b-46ff-9c5d-136d4370cf65@k2g2000hse.googlegroups.com...
Buff... I'm very lost, I think I'm using this class in a wrong way,
let me show you my class (resumed and some parts regarding to serial
proceses ommited ) and exactly how I'm using it:
MySerial::MySerial
{ //nothing here
}
~MySerial::MySerial
{ //nothing here
}
BOOL MySerial::InitInstance()
{
m_bAutoDelete = FALSE;
}
int MySerial::ExitInstance()
{
return CWinThread::ExitInstance();
}
//Next are my two only messages I send from GUI thread
BEGIN_MESSAGE_MAP(MySerial, CWinThread)
//{{AFX_MSG_MAP(MySerial)
ON_THREAD_MESSAGE( WM_MYSERIAL_WRITE, OnWriteData )
ON_THREAD_MESSAGE( WM_MYSERIAL_CLOSE, closePort )
END_MESSAGE_MAP()
int MySerial::Run()
{
TRACE("Running.");
return CWinThread::Run();
}
//Called when GUI sends WM_MYSERIAL_WRITE
void MySerial::OnWriteData( WPARAM wParam, LPARAM lParam )
{
//A group of instructions, no loop.
//Only called when, from my GUI I send
PostThreadMessage( WM_MY_SERIAL_WRITE )
}
UINT ReaderThread( LPVOID cT )
{
//A group of instructions, inside a loop.
//The loop exists when there is an error or ShutdownEvent is
signaled
while( reading )
{
//Blablablablabla
}
}
BOOL MySerial::openPort( int port )
{
WriteEvent = CreateEvent( NULL, TRUE, FALSE, NULL );
ReadEvent = CreateEvent( NULL, TRUE, FALSE, NULL );
ShutdownEvent = CreateEvent( NULL, TRUE, FALSE, NULL );
//Here I open my port and...
//Start ReaderThread to continuosly read Serial Port
AfxBeginThread( ReaderThread, this, THREAD_PRIORITY_NORMAL );
}
//This is my old Stop() method
void MySerial::closePort( WPARAM wparam, LPARAM lparam)
{
if( opened) //A flag set to true when I open the port
{
SetEvent( ShutdownEvent );
//WaitForSingleObject( ReaderStoped, INFINITE ); //No
ReaderStoped Event, no WFSO
::CloseHandle( WriteEvent );
::CloseHandle( ShutdownEvent );
::CloseHandle( ReadEvent );
::CloseHandle( hComm );
PostQuitMessage(0);
WaitForSingleObject( this->m_hThread, INFINITE );
}
}
This is basically my class MySerial.
Then here it is the way I use it, I think, after reading your last
post I'm using it wrongly:
First I create it:
(MySerial*) pMySerial = (MySerial*)
AfxBeginThread( RUNTIME_CLASS( MySerial ), THREAD_PRIORITY_NORMAL, 0,
CREATE_SUSPENDED );
Then I resume it and call OpenPort:
pMySerial->ResumeThread();
pMySerial->openPort( port_number );
This activates the ReaderThread automatically.
Then whenever I want send some data through serial port:
pMySerial->PostThreadMessage( WM_MYSERIAL_WRITE, buffer,
length );
And finally, when I want to close the port and let it ready for
another future open:
pMySerial->PostThreadMessage( WM_MYSERIAL_CLOSE, 0, 0);
I think I'm doing bad using pMySerial->openPort() directly, instead of
calling it from a message.
=======================
From the code that you show here it should not matter. It is OK to call it
directly instead of from a message. But the design would probably be better
and safer and more understandable if you call it from a message.
Steps 1 and 2 are correct, but not 3. First, PostQuitMessage is used only
with message driven threads (your writer thread). And, it must be called
from the thread that is quitting. You should call it in your writer
thread
when it detects the shutdown message/event.
Is every new message sended to MySerial creating a new Thread? I
though that MySerial is a single thread that manages the messages I'm
sending to it.
No, the writer thread waits after processing each message, then it resumes
when it gets the next message. MFC makes that happen in the Run method.
The only problem I see in this code is that in closePort you are waiting for
the wrong thread in WFSO, and when you start the reader thread you forgot to
save the thread pointer. First, save the pointer:
//Start ReaderThread to continuosly read Serial Port
pReaderThread = AfxBeginThread( ReaderThread, this);
Then in close port you would do this for the reader thread shut down:
SetEvent( ShutdownEvent );
WaitForSingleObject(pReaderThread->m_hThread, INFINITE );
Your code is very close to correct! Don't go making a lot of changes now :)
--
Scott McPhillips [VC++ MVP]