Possible bug in the UART SW lib

[vhdlnerd]

I'm writing some code in assembly and was looking at the neorv32_uart.c file for "inspiration" on how to interface to the UART. And, saw this C function:

/**********************************************************************//**
 * Check if UART0 has received a char.
 *
 * @note This function is non-blocking.
 * @note Use neorv32_uart0_char_received_get(void) to get the char.
 *
 * @return =!0 when a char has been received.
 **************************************************************************/
int neorv32_uart0_char_received(void) {

  if ((NEORV32_UART0.DATA & (1<<UART_DATA_AVAIL)) != 0) {
    return 1;
  }
  else {
    return 0;
  }
}

Although I did not test this, it seems wrong to me. If there are more than one chars in the UART RX FIFO, wouldn't neorv32_uart0_char_received() consume the first char (when checking the UART_DATA_AVAIL bit) in the FIFO and a later call to neorv32_uart0_char_received_get() would return the next char in the RX FIFO? (i.e., you would drop the first char).

When I coded this in assembly, I checked the UART_CTRL_RX_EMPTY bit in the NEORV32_UART0_CTRL control/status register instead -- it worked as expected.

Maybe, I'm missing something in the C code??

Thanks,
Brian

[stnolting]

Hey there!

If there are more than one chars in the UART RX FIFO, wouldn't neorv32_uart0_char_received() consume the first char

I think you are right. The function should check the RX FIFO flags in CTRL instead. Strange, that this has not caused any trouble so far... 😅

Thanks for identifying this bug!