Serial programming on FreeBSD 6.0 RELEASE
Derek Ragona
derek at computinginnovations.com
Thu Jun 22 18:28:35 UTC 2006
Your code looked OK, but I would add some debugging print statements.
silo overflow is a stack overflow, so you should figure where this is
happening.
The getty may be managing the port IO, which is why I would disable
it. Getty is listening to the port for traffic.
Depending on the OS, some don't run getty on all physical ports, some do by
default. Some are more exclusive in the perms as well, or should be.
-Derek
At 01:19 PM 6/22/2006, Andrew Falanga wrote:
>Derek,
>
>No I didn't disable the getty on the port. To be honest, I didn't know
>one was running.
>
>Second, the errors I'm receiving are:
>
>sio1: 2 more silo overflows (total 9)
>sio1: 280221 more tty-level buffer overflows (total 576898)
>
>
>One question I have is, why would kermit able to receive/send data across
>the port? I don't know if said this in my first message, but I started
>kermit on both the FreeBSD and Linux machines and was able to send/receive
>data in either direction. I didn't disable getty before doing that.
>
>Second question is, does my code look ok? Am I initializing everythin
>accordingly?
>
>Andy
>
>On 6/22/06, Derek Ragona <<mailto:derek at computinginnovations.com>
>derek at computinginnovations.com> wrote:
>>Andy,
>>
>>Did you kill the getty running on the port?
>>
>>Are you getting any errors?
>>
>> -Derek
>>
>>
>>At 10:28 AM 6/22/2006, Andrew Falanga wrote:
>>Hello,
>>
>>I've got a case where I'm writing a simply serial program to send bytes from
>>one system to another over a serial cable. The program works in Linux, but
>>when I use it in FreeBSD nothing happens. The program starts and, in the
>>case of receiving, waits for data to appear on the /dev/cuad1 port and times
>>out in select. If I'm sending in FreeBSD, the send program believes it has
>>written all the data but nothing is received in Linux (either with running
>>my code on Linux or using kermit). On FreeBSD, however, if I use kermit to
>>monitor the /dev/cuad1 port and send, using my code, from Linux, all 2500
>>packets are transmitted ok. Also, when sending from Linux and receiving in
>>FreeBSD, the FreeBSD machine show silo and tty-level interrupt errors. The
>>man page for sio describes that as problems in the interrupt handler, for
>>silo overflows, and that data arrived faster than the application could
>>process (for the tty-level overflows).
>>
>>What am I doing wrong? What is it that Kermit does to allow data being read
>>from the /dev/cuad1 device that I'm not? I've been looking over the kermit
>>sources but to be honest, before Thursday of last week, I'd never programmed
>>for serial ports before let alone having any knowledge of termios.
>>Therefore, I'm looking at a very steep learning curve. Please look over my
>>ctor for the serial line class that is in my code. After much debugging,
>>I'm convinced that my problem is in how I'm configuring my port.
>>
>>sline::sline( std::string d, int opm ) : dev( d ), opMode( opm )
>>{
>> memset( &oldSettings, 0, sizeof( oldSettings ) );
>> memset( &ioPort, 0, sizeof( ioPort ) );
>> memset( recvBuf, 0, PACKETSIZE );
>> memset( &timeout, 0, sizeof( timeout ) );
>>
>> FD_ZERO( &incoming );
>>
>> timeout.tv_sec = TIMEOUT_SEC;
>> timeout.tv_usec = TIMEOUT_USEC;
>>
>> // create the packet
>> char asciiPrtChars = 32; // first printable ascii character in decimal
>> for( int i = 0; i < PACKETSIZE; i++ ) {
>> packet[i] = asciiPrtChars++;
>> if( asciiPrtChars == 127 ) // 127 is the delete character
>> asciiPrtChars = 32;
>> }
>>
>> // on the systems this code was meant to compile, LINUX and FreeBSD
>> // O_NDELAY and O_NONBLOCK are the same, however, this is not always
>> // true
>> fd = open( dev.c_str(), O_RDWR | O_NOCTTY | O_NDELAY );
>> if( fd < 0 )
>> throw init();
>>
>> tcgetattr( fd, &oldSettings );
>> tcgetattr( fd, &ioPort );
>>
>>#ifdef DEBUG
>> COUT << "Current input speed is " << cfgetispeed( &ioPort ) << ENDL;
>> COUT << "Current output speed is " << cfgetospeed( &ioPort ) << ENDL;
>>#endif
>>
>>#if 0
>> if( opMode == OPMODE_WRITE ) {
>> if( fcntl( fd, F_SETFL, 0 ) < 0 ) {
>> perror( "fcntl" );
>> throw init();
>> }
>> } else {
>> if( fcntl( fd, F_SETFL, FNDELAY ) < 0 ) {
>> perror( "fcntl" );
>> throw init();
>> }
>> FD_SET( fd, &incoming );
>> }
>>#endif
>>
>>#if 0
>> // configure control field, this should configure for 8N1
>> // first, disable flow control (may have to put it back in)
>> // ioPort.c_cflag &= ~CRTSCTS;
>> ioPort.c_cflag |= CRTSCTS | CLOCAL | CREAD;
>> ioPort.c_cflag &= ~PARENB;
>> ioPort.c_cflag &= ~CSTOPB;
>> ioPort.c_cflag &= ~CSIZE;
>> ioPort.c_cflag |= CS8;
>>
>> // configure local field, setting to RAW mode
>> ioPort.c_lflag |= ~( ICANON | ECHO | ECHOE | ISIG );
>>
>> // configure the input field, setting to ignore parity errors
>> // and disable software flow control
>> ioPort.c_iflag |= IGNPAR;
>> ioPort.c_iflag &= ~( IXON | IXOFF | IXANY );
>>
>> // configure output field, setting to RAW
>> ioPort.c_iflag &= ~OPOST;
>>#endif /* end of if 0 */
>>
>> // configure for raw data transfer
>> cfmakeraw( &ioPort );
>>
>> // set VMIN and VTIME parameters in c_cc array
>> ioPort.c_cc[VMIN] = PACKETSIZE;
>> ioPort.c_cc[VTIME] = 0;
>>
>> if( cfsetispeed( &ioPort, BAUDRATE ) < 0 ) {
>> perror( "cfsetispeed" );
>> throw init();
>> }
>>
>> if( cfsetospeed( &ioPort, BAUDRATE ) < 0 ) {
>> perror( "cfsetospeed" );
>> throw init();
>> }
>>
>> COUT << "flushing dev: " << dev << ENDL;
>> if( tcflush( fd, TCIOFLUSH ) < 0 ) {
>> perror( "tcflush" );
>> throw init();
>> }
>>
>> COUT << "Setting new parameters to: " << dev << ENDL;
>> if( tcsetattr( fd, TCSANOW, &ioPort ) < 0 ) {
>> perror( "tcsetattr" );
>> throw init();
>> }
>>
>>#if 0
>> if( ioctl( fd, TIOCMGET, &portStatus ) < 0 ) {
>> perror( "ioctl - get" );
>> throw init();
>> }
>>
>> // I believe we want to clearn the DCD bit
>> portStatus &= ~TIOCM_DTR;
>>
>> if( ioctl( fd, TIOCMSET, &portStatus ) < 0 ) {
>> perror( "ioctl - set" );
>> throw init();
>> }
>>#endif /* removed for debugging, still not sure I need it */
>>
>> if( fcntl( fd, F_SETFL, FNDELAY | O_NONBLOCK ) < 0 ) {
>> perror( "fcntl" );
>> throw init();
>> }
>>
>>#ifdef DEBUG
>> COUT << "New input speed is " << cfgetispeed( &ioPort ) << ENDL;
>> COUT << "New output speed is " << cfgetospeed( &ioPort ) << ENDL;
>>#endif
>>} // end of sline ctor
>>
>>Thanks for you help,
>>Andy
>>_______________________________________________
>><mailto:freebsd-questions at freebsd.org>freebsd-questions at freebsd.org
>>mailing list
>><http://lists.freebsd.org/mailman/listinfo/freebsd-questions>http://lists.freebsd.org/mailman/listinfo/freebsd-questions
>>To unsubscribe, send any mail to
>>"<mailto:freebsd-questions-unsubscribe at freebsd.org>freebsd-questions-unsubscribe at freebsd.org"
>>
>>--
>>This message has been scanned for viruses and
>>dangerous content by MailScanner, and is
>>believed to be clean.
>>MailScanner thanks transtec Computers for their support.
>>
>>
>>--
>>This message has been scanned for viruses and
>>dangerous content by <http://www.mailscanner.info/>MailScanner, and is
>>believed to be clean.
>>MailScanner thanks <http://www.transtec.co.uk/>transtec Computers for
>>their support.
>
>
>--
>This message has been scanned for viruses and
>dangerous content by <http://www.mailscanner.info/>MailScanner, and is
>believed to be clean.
>MailScanner thanks <http://www.transtec.co.uk/>transtec Computers for
>their support.
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
MailScanner thanks transtec Computers for their support.
More information about the freebsd-questions
mailing list