Other Parts Discussed in Thread: HALCOGEN
The floating point baud rate divisor was truncated instead of rounded - see below
/** @fn void sciSetBaudrate(sciBASE_t *sci, uint32 baud)
* @brief Change baudrate at runtime.
* @param[in] sci - sci module base address
* @param[in] baud - baudrate in Hz
*
* Change the SCI baudrate at runtime.
*/
/* SourceId : SCI_SourceId_003 */
/* DesignId : SCI_DesignId_003 */
/* Requirements : HL_SR232 */
void sciSetBaudrate(sciBASE_t *sci, uint32 baud)
{
float64 vclk = 80.000 * 1000000.0;
uint32 f = ((sci->GCR1 & 2U) == 2U) ? 16U : 1U;
uint32 temp;
float64 temp2;
/* USER CODE BEGIN (6) */
//the halcogen code below does not work when the baud rates start to get above 300 k baud
// the computation for temp2 is correct as a float. However, when it is converted to an int in the
// next line it is truncated instead of rounded. In one case the float was 10.99999 or so but the register
// was set to 10 instead of 11. This resulted in a rate that was far enough off so it didn't work.
// The ifdef below in the user section should eliminate the automatically generated halcogen code with the bug
// and replace it with the code in the next user section after the endif
#ifdef 0
/* USER CODE END */
/*SAFETYMCUSW 96 S MR:6.1 <APPROVED> "Calculations including int and float cannot be avoided" */
temp = (f*(baud));
temp2 = ((vclk)/((float64)temp))-1U;
sci->BRS = (uint32)((uint32)temp2 & 0x00FFFFFFU);
/* USER CODE BEGIN (7) */
#endif
temp = (f*(baud));
temp2 = ((vclk)/((float64)temp))-1U;
sci->BRS = (uint32)((uint32)(temp2 +.5) & 0x00FFFFFFU);
/* USER CODE END */
}
