Using fstream with USB Serial Port

[mo1492]

If it works in a terminal program I really don't have any help for you except this. I don't know if it is this issue but these lines could be a potential problem depending on how COMPORT functions: COMPORT << "UPLOAD NEW MAP\n"; COMPORT >> number; There might be a timing issue if all data has not been received from the pic when COMPORT >> number; is called. As a test you might want to put a delay in to give it time. COMPORT << "UPLOAD NEW MAP\n"; delay? COMPORT >> number;

[OscardelaGrouch]

Thanks for the reply and suggestion. I think that a delay might help if I am trying to get a string from the PIC to the PC app. I would HOPE that the Getline function would do the waiting, but ??

speedbump99 wrote:

As a test you might want to put a delay in to give it time.

I've thought about trying that, I don't see a delay function in C++ though. Is there one? In PIC C there is a __delay_us(10); function that would delay, in this case 10 microseconds based on the current clock speed. Seems that C++ ought to have one. Maybe have to just do a loop? OdlG

[mo1492]

I kind of doubt that getline will wait for data as it has no way of knowing how much data to expect. A quick search for Linux and Delay turned up a sleep() command that may be what you need. Also a search of "Linux synchronous serial port PIC" showed several of examples where they went through the process of setting the parameters for the serial port and running a loop to retrieve data. Something like (pseudo code): while (1) { read(1 char); if char == "\n" break; } You would probably read the text and convert back to integer. You might try that search. Hope this helps.

[OscardelaGrouch]

Thanks for the reply,

speedbump99 wrote:

I kind of doubt that getline will wait for data as it has no way of knowing how much data to expect.

I think it will. In fact there are several versions of the getline function. In one you can specify a char array, max number of chars and a delimiter char. The default delim is '\n'. I have tried doing it char by char and didn't have any better results. I may try that again as it seems the operator >> only inputs numbers (of various types) and not strings. I think getline must be used for that. For right now I'm going to get my scope out and see why my reply jumps by two's instead of one with each command sent to the PIC. I will keep this thread updated. Thanks all, OdlG

[OscardelaGrouch]

Just an update on things I found using my oscilloscope and looking at what is being sent back and forth between the PC and PIC. I wish I could upload a screen shot from the scope. The PC is sending out three (sometimes two) data bursts. The first is obviously the "UPLOAD NEW MAP\n" command to the PIC. What the other two data burst are, I don't know yet. The PIC is replying twice, as would be expected if the data bursts contain the '\n' char. So looks like my first objective is to find out why

COMPORT << "UPLOAD NEW MAP" << endl;

is sending data out more than once. Any ideas would be appreciated. OdlG

[mo1492]

Last suggestion. If you have another pc with a com port, connect the two pc's together and run the terminal program you have on one and your application on the other. Then you can see exactly what your application is sending out. I have also done this on a single computer with two com ports. Good Luck

[mo1492]

Last suggestion. If you have another pc with a com port, connect the two pc's together with a loopback cable, run your terminal program on one and your test application on the other. Then you can see what your test app is sending. I have also done this on a single computer with two com ports. Good Luck

[OscardelaGrouch]

Did a little debugging and it seems that the PC app is echoing the data from the PIC back to the PIC. I wonder why and how to stop that. OdlG

[OscardelaGrouch]

I did a little debugging on the PIC side and see that the PC app is echoing back the data it gets from the PIC. Any ideas why and how to stop that? Thanks? OdlG

[Jochen Arndt]

That is probably because the serial device has echoing enabled. Such is used with serial terminals. When a character is received, it is send back to be displayed in the terminal (which usually does not display the character locally before sending it but waits for the echoed character to display that). To avoid that you have to disable echoing by configuring the serial device initially:

stty -F /dev/ttyACM0 raw pass8 -echo -hup -clocal 9600

Note that it might be necessary to specify also other settings, remove some used above, and change the baud rate. See stty(1): change/print terminal line settings - Linux man page[^] for the available options. Before changing the options I suggest to execute

stty -F /dev/ttyACM0

to check the current settings. As already mentioned by others it is rather uncommon to use C++ streams with serial ports. The common method is using termios and the IO functions from the standard C library, or a library like Serial Ports - 1.65.0[^] from Boost.Asio - 1.66.0[^] . Related reads: Serial HOWTO[^] Serial Programming/Serial Linux - Wikibooks, open books for an open world[^]

[OscardelaGrouch]

Jochen Arndt wrote:

The common method is using termios and the IO functions from the standard C library

Thank you for the reply. It is ironic that I ran across this yesterday and was able to write something that used this successfully. I will use this in my final code I am sure. My issue now is the write command, it seems to be char based, so I'm working on sending a char buffer. Once again, thanks to you and the forum, I really appreciate it. OdlG

[OscardelaGrouch]

Thanks for the help so far. I have written a small PC App program to try termios out. I am having two issues. First is the first time I send data to the PIC, the return response is junk. Second is the response is out of sync with the sent command, it takes sending the command a second time for the response string to be correct. Would appreciate a second pair of eyes to see what is the problem. PC app code ..

int main(int argc, char** argv)
{
ComPort MyComPort;
char TXBuf[255];
char RXBuf[255];
char c;

while(1)
{
    cout << "(U)load New Map, (N)ext Map, (Q)uit .." << endl;
    cin >> c;
    c = (char)toupper(c);
    if (c == 'Q') return 0;
    if (c == 'U')
    {
        strcpy(TXBuf, "UPLOAD NEW MAP\\n");
        MyComPort.WriteComPortDataMsg(TXBuf, strlen(TXBuf));
        MyComPort.ReadComPortDataMsg(RXBuf);
        cout << "RXBuf = " << RXBuf << endl;
    }
    else if (c == 'N')
    {
        strcpy(TXBuf, "NEXT MAP\\n");
        MyComPort.WriteComPortDataMsg(TXBuf, strlen(TXBuf));
        MyComPort.ReadComPortDataMsg(RXBuf);
        cout << "RXBuf = " << RXBuf << endl;
    }
    cout << endl;
}
return 0;

}

class defs ..

/***** Class Declaration ******************************************************/
class ComPort
{
public:
ComPort();
~ComPort();
void WriteComPortDataMsg(char *data, char num);
void ReadComPortDataMsg(char *data, char dlimc);
void ReadComPortDataMsg(char *data);
void SetDlimiter(char dlimc);
private:
char dlim;
struct termios config;
int hComPort;
bool Error;
};

void ComPort::WriteComPortDataMsg(char *data, char num)
{
for (int cnt = 0; cnt < num; cnt++)
{
write(hComPort, &data[cnt], 1); // write char to Com Port
}
tcdrain(hComPort);
}

void ComPort::ReadComPortDataMsg(char *data)
{
int cnt = 0;
char c;

data\[0\] = 0x00;
do
{
    read(hComPort, &c, 1);
    if ((c >= 0x20) && (c <= 0x7E)) data\[cnt++\] = c;
} while ((c != dlim) && (cnt < 255));
data\[cnt++\] = '\\0';

}

responses ..

(U)load New Map, (N)ext Map, (Q)uit ..
u
RXBuf = MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM

[Jochen Arndt]

I did not see any code that initialises the serial port. So it will still use the current settings as shown by the stty command line tool. You have to use tcsetattr to select the same basic settings as on the PIC (probably 8N1; baud rate seems to mach already), disable all kinds of flow control (software and hardware), and all additional features like echoing. Example:

memset(&config, 0, sizeof(config));
config.c_cflag = CS8 | CREAD | CLOCAL; // 8N1, enable receiver, ignore ctrl lines
cfsetospeed(&config, B9600); // baud rate used by the PIC
cfsetispeed(&config, B9600);
tcsetattr(hComPort, TCSANOW, &config);

With new line terminated data transfers, you may also use the canonical mode (ICANON).

[OscardelaGrouch]

Jochen Arndt wrote:

I did not see any code that initialises the serial port.

There is, I was having trouble with the forum editor, thought maybe my code was too long. The first part is here ..

/***** Class Declaration ******************************************************/
class ComPort
{
public:
ComPort();
~ComPort();
void WriteComPortDataMsg(char *data, char num);
void ReadComPortDataMsg(char *data, char dlimc);
void ReadComPortDataMsg(char *data);
void SetDlimiter(char dlimc);
private:
char dlim;
struct termios config;
int hComPort;
bool Error;
};

ComPort::ComPort()
{
dlim = '\n';
char c;

Error = false;
hComPort = open("/dev/ttyACM0", O\_RDWR | O\_NOCTTY | O\_NDELAY);
if (hComPort == -1)
{
    cout << "failed to open port\\n" << endl;
    cout << "Hit  to Stop .." << endl;
    cin >> c;
    c = (char)toupper(c);
    if (c == 'S') Error = true;
}

if(!isatty(hComPort))
{
    cout << "file descriptor is  not pointing to a TTY device\\n" << endl;
    cout << "Hit  to Stop .." << endl;
    cin.get(c);
    if ((c == 's' || c == 'S')) Error = true;
}

if(tcgetattr(hComPort, &config) < 0)
{
    cout << "configuration of the serial interface not found\\n" << endl;
    cout << "Hit  to Stop .." << endl;
    cin >> c;
    c = (char)toupper(c);
    if (c == 'S') Error = true;
}

//
// Input flags - Turn off input processing
//
// convert break to null byte, no CR to NL translation,
// no NL to CR translation, don't mark parity errors or breaks
// no input parity check, don't strip high bit off,
// no XON/XOFF software flow control
//
config.c\_iflag &= ~(IGNBRK | BRKINT | ICRNL | INLCR | PARMRK | INPCK | ISTRIP | IXON);

// Output flags - Turn off output processing
//
// no CR to NL translation, no NL to CR-NL translation,
// no NL to CR translation, no column 0 CR suppression,
// no Ctrl-D suppression, no fill characters, no case mapping,
// no local output processing
//
// config.c\_oflag &= ~(OCRNL | ONLCR | ONLRET |
//                     ONOCR | ONOEOT| OFILL | OLCUC | OPOST);
config.c\_oflag = 0;

//
// No line processing
//
// echo off, echo newline off, canonical mode off, 
// extended input processing off, signal ch

[Jochen Arndt]

Use the code from my post. It sets all while you are modifying the existing where something might be still set / cleared. CREAD is probably set but what about CLOCAL?

[OscardelaGrouch]

Jochen Arndt wrote:

CREAD is probably set but what about CLOCAL?

I had the same code except .. memset

if(tcgetattr(hComPort, &config) < 0)
{
cout << "configuration of the serial interface not found\n" << endl;
cout << "Hit to Stop .." << endl;
cin >> c;
c = (char)toupper(c);
if (c == 'S') Error = true;
}
memset(&config, 0, sizeof(config));

this is baudrate code ..

config.c_cflag = CS8 | CREAD | CLOCAL; // 8N1, enable receiver, ignore ctrl lines

if(cfsetispeed(&config, B9600) < 0 || cfsetospeed(&config, B9600) < 0)
{
    cout << "Baud Rate did not set correctly\\n" << endl;
    cout << "Hit  to Stop .." << endl;
    cin >> c;
    c = (char)toupper(c);
    if (c == 'S') Error = true;
}

I get the same result as before. OdlG

[Jochen Arndt]

The source is probably here:

read(hComPort, &c, 1);
if ((c >= 0x20) && (c <= 0x7E)) data[cnt++] = c;

See read(2): read from file descriptor - Linux man page[^]:

Quote:

Return Value On success, the number of bytes read is returned (zero indicates end of file), and the file position is advanced by this number. It is not an error if this number is smaller than the number of bytes requested; this may happen for example because fewer bytes are actually available right now (maybe because we were close to end-of-file, or because we are reading from a pipe, or from a terminal), or because read() was interrupted by a signal. On error, -1 is returned, and errno is set appropriately. In this case it is left unspecified whether the file position (if any) changes.

read() will return zero when there are no data available. So you have either to clear your c variable before, or - better - check the return value:

int received = read(hComPort, &c, 1);
if (received == 1 && c >= 0x20 && c <= 0x7E)
data[cnt++] = c;
else if (received < 0)
// handle error here

Another option is setting blocking mode (attributes c_cc[VMIN] to non zero and c_cc[VTIME] to the timeout value). Then you will not get zero return values but still have to check for negative return values indicating timeout or other errors.

[OscardelaGrouch]

I was beginning to think I needed some way to wait on the USB/Com Port, but wasn't sure how to do that. This works great. My code change ..

void ComPort::ReadComPortDataMsg(char *data)
{
int cnt = 0;
int received;
char c;

data\[0\] = 0x00;
do
{
    received = read(hComPort, &c, 1);
    if ((received == 1) && (c >= 0x20) && (c <= 0x7E))
    {
        data\[cnt++\] = c;
    }
    else if (received < 0)
    {
        
    }
} while ((c != dlim) && (cnt < 255));
data\[cnt++\] = '\\0';

}

Not sure how to deal with received < 0. Do you know where can I read up on

struct termios

and

global errno variable

? Thanks!!! OdlG

[Jochen Arndt]

For all standard C library features you can get the description from the shell command prompt as man page:

man errno

will shown the content of the errno(3) - Linux manual page[^]. Similar for termios which is covered by a single man page: termios(3) - Linux man page[^]. If you want to get the above links, just use the same "man " with your favorite search engine. Other good sources for the standard C and C++ libraries are cplusplus.com - The C++ Resources Network[^] and cppreference.com[^].

[OscardelaGrouch]

Question on how to error handle. I am falling into the received < 0 but if I just ignore it and keep looking for chars, I get a complete, correct message. Is it safe to do that or should I clear the error? And How? Thanks in advance ;) OdlG