Hello
I'm trying to setup UART2 for 115200 8n1 configuration and hardware flow control on an OMAP-L138 processor. The purpose of this configuration to communicate with a TI GPS/Bluetooth chip. This is not an EVM board, but our own custom board. I have read through a few posts already on the e2e forums, and found this one which is close, but doesn't have a good solution.
I am running linux v3.3 on the OMAP and trying to use the serial port as a tty device (/dev/ttyS1).
I have verified the following information:
* Pinmuxing the CTS/RTS registers, following code in da850.c:
+ MUX_CFG(DA850, NUART2_CTS, 0, 28, 15, 4, false) + MUX_CFG(DA850, NUART2_RTS, 0, 24, 15, 4, false)
* Enabling UART2 with the davinci_serial_init function and setting bit 2 in the config struct
* Current register dump right before transmit:
UART8250: fed0d000 = reg_0 = 0x00000000 UART8250: fed0d004 = reg_1 = 0x00000007 UART8250: fed0d008 = reg_2 = 0x000000c1 UART8250: fed0d00c = reg_3 = 0x00000013 UART8250: fed0d010 = reg_4 = 0x00000023 UART8250: fed0d014 = reg_5 = 0x00000000 UART8250: fed0d018 = reg_6 = 0x00000000 UART8250: fed0d01c = reg_7 = 0x00000000 UART8250: fed0d020 = reg_8 = 0x000000b6 UART8250: fed0d024 = reg_9 = 0x00000001 UART8250: fed0d028 = reg_10 = 0x44141102 UART8250: fed0d02c = reg_11 = 0x00000000 UART8250: fed0d030 = reg_12 = 0x00006003 UART8250: fed0d034 = reg_13 = 0x00000000
* In my debugging I have made a few "hacks" to 8250.c (marked with "//todo ajr") and have attached the modified file
* * These hacks allowed me to bypass tty->hw_stopped checks and print various information
* * Additionally I modified the code set the RTS bit in the MCR when setting cflag CRTSCTS
The current behavior that I see when trying to write a small buffer to /dev/ttyS1 is that only the first byte is written out, but nothing else. Using print statements (in 8250.c and serial_core.c), I have verified that these values are being written to the THR, but not all are being written out. I am using an oscilloscope on the tx line to verify that data is being written out. However, I never see the CTS line go low, so that may be an additional issue (with the NL5500L chip). I have followed all startup procedure for the 5500 chip, including supplying stable clocks, power, and deasserting the shutdown line.
Here is the test application I am running to write to the tty device:
#include <stdio.h>
#include <unistd.h>
#include <stdint.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/ioctl.h>
#include <termios.h>
#define RX_SZ 11
#define MSG_SZ 6
static const uint8_t msg[MSG_SZ] = {0x09, 0x02, 0x02, 0x00, 0x00, 0x00};
int main(void)
{
int i, j;
uint8_t rx[RX_SZ];
struct termios sp_options;
int serial_port = open("/dev/ttyS1", O_RDWR | O_NOCTTY);
speed_t speed = 115200;
if (serial_port < 0)
{
printf("error opening serial port\n");
return -1;
}
else
{
printf("setting sp options\n");
tcgetattr(serial_port, &sp_options);
cfmakeraw(&sp_options);
cfsetispeed(&sp_options, speed);
cfsetospeed(&sp_options, speed);
sp_options.c_cflag |= CRTSCTS;
tcsetattr(serial_port, TCSANOW, &sp_options);
}
i = write(serial_port, msg, MSG_SZ);
if (i < 0) {
printf("error writing to serial port, %d\n", i);
return -1;
} else {
j = tcdrain(serial_port);
printf("tcdrain = %d, wrote %d bytes\n", j, i);
}
i = read(serial_port, rx, RX_SZ);
if (i < 0) {
printf("error reading serail port, %d\n", i);
return -1;
}
for (i=0; i<RX_SZ; i++)
{
printf("%02x : ", rx[i]);
}
return 0;
}
I have been stuck on this simple issues for a few days now am starting to get frustrated and run out of ideas on what could be going wrong. Please let me know if further information is needed.
Austin
