wrong data recieved via read

[Dushistov]

I want to connect with gps device via bluetooth on linux.

In console I did:

$ sudo rfcomm connect 0 10:C6:FC:FE:6D:D0 &
$ sudo cat /dev/rfcomm0
$GPGSA,A,3,08,27,11,69,14,18,77,32,,,,,2.5,1.6,1.9*35
$GPRMC,161659.5,A,5521.76350,N,03731.92747,E,000.57,058.2,040718,010.2,E,A*35
$GPGGA,161659.5,5521.76350,N,03731.92747,E,1,08,1.6,204.7,M,13.1,M,,*55
$GPVTG,058.2,T,048.0,M,000.57,N,0001.06,K,A*15

as you can see all data is is ascii, this is nmea protocol, it is text based.

Then I modify example.rs:

extern crate bluetooth_serial_port;
extern crate mio;
use bluetooth_serial_port::{BtProtocol, BtSocket};
use std::io::Read;

fn main() {
    // scan for devices
    let devices = bluetooth_serial_port::scan_devices().unwrap();
    if devices.len() == 0 {
        panic!("No devices found");
    }

    let device = devices.iter().position(|x| x.name.contains("GLO")).unwrap();
    let device = &devices[device];
    println!("Connecting to `{}` ({})", device.name, device.addr.to_string());

    // create and connect the RFCOMM socket
    let mut socket = BtSocket::new(BtProtocol::RFCOMM).unwrap();
    socket.connect(device.addr).unwrap();

    loop {
        let mut buffer: [u8; 20] = [0; 20];
        let num_bytes_read = socket.read(&mut buffer[..]).unwrap();
        if num_bytes_read == 0 {
            println!("read nothing, retry");
        }
        let s = &buffer[0..num_bytes_read];
        println!("raw bytes: {:?}", s);
    }
}

and output is:

raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]
raw bytes: [192, 1, 22, 0, 3, 1, 0, 37, 1, 10, 1, 128, 0, 232, 3, 232, 3, 184, 207, 4]
raw bytes: [0, 19]

this looks completly wrong for me, byte with 0 value, bytes with value 1 and so on.

[kaegi]

Thank you for your detailed bug report! Your code looks fine to me... Unfortunately I'm not working on bluetooth-serial-port anymore, and I don't see at first glance what the reason for this behavior could be. So I'm sorry to say that you're on your own for this problem. I will leave this issue open, so others can see it.

[Dushistov]

Looks like the problem in connection, I replace socket.connect(device.addr).unwrap(); with

    let mut socket = BtSocket::new(BtProtocol::RFCOMM).unwrap();
    let fd = socket.0.stream.into_raw_fd();

    let mut sock_addr = sockaddr_rc {
        rc_family: 31,
        rc_channel: 1,
        rc_bdaddr: bdaddr_t {
            b: [0xD0, 0x6D, 0xFE, 0xFC, 0xC6, 0x10],
        },
    };
    let status = unsafe { libc::connect(fd, mem::transmute(&mut sock_addr), mem::size_of::<sockaddr_rc>() as u32) };
    if status != 0 {
        eprintln!("connect failed");
        return;
    }

and all starts work as expected

[Dushistov]

@kaegi

I find out the reason, device has 3 rfcomm channels: 1,3,5.

The code in sdp.rs uses the last one 5, and via this channel really comes these strange data, if I force code to use channel 1, I got gps/nmea data.

What is your option about API change in crate should I made to handle this situtaion, here output from this test program:

rfcomm channel: 1 found service record 0x90001 rfcomm channel: 3 found service record 0x90003 rfcomm channel: 5 found service record 0x90005

[kaegi]

Probably adding a connect_to_channel to BtSocket. As I'm not interested in maintaining this project, you can have crate and GitHub publishing rights (or fork your own version) if you want to maintain it.

[Dushistov]

As I'm not interested in maintaining this project, you can have crate and GitHub publishing rights (or fork your own version) if you want to maintain it

I come back to this work. Therefore, if you are not interested in this project can you give me publish rights for crates.io ?

[kaegi]

Should be done now.

[Dushistov]

@kaegi Great, thanks!