Im using stm32f103RB on nucleo64 board, to read SPI sensor with half duplex SPI master mode.
SPI blocking recive function do not disable SPI properly, it should disable it and then finish recive one more time, as user manual RM0008 specify(25.3.8), but function only disable SPI. Next recive will be broken with partially read byte in DR.
Current function code is
HAL_StatusTypeDef HAL_SPI_Receive(SPI_HandleTypeDef *hspi, uint8_t *pData, uint16_t Size, uint32_t Timeout)
{
...
/* Check the end of the transaction */
if((hspi->Init.Mode == SPI_MODE_MASTER)&&((hspi->Init.Direction == SPI_DIRECTION_1LINE)||(hspi->Init.Direction == SPI_DIRECTION_2LINES_RXONLY)))
{
/* Disable SPI peripheral */
__HAL_SPI_DISABLE(hspi);
}
...
}
Ah that comment, but no one care, now one!
It should be like that
volatile uint8_t ttt;
if((hspi->Init.Mode == SPI_MODE_MASTER)&&((hspi->Init.Direction == SPI_DIRECTION_1LINE)||(hspi->Init.Direction == SPI_DIRECTION_2LINES_RXONLY)))
{
//
/* Disable SPI peripheral */
__HAL_SPI_DISABLE(hspi);
//fix according to 25.3.8
while(!__HAL_SPI_GET_FLAG(hspi, SPI_FLAG_RXNE)){};
ttt = *(__IO uint8_t *)&hspi->Instance->DR;
}
Im very disappointed to see such bug in year 2019.
***
That bug lead to very confusing behavior of SPI.
For example.
Shifted recived data
https://i.imgur.com/3rsLKaN.png
DR is partially filed with data which get ther because SPI wasn disabled imideatly or wated for RXNE and DR cleared after disabling.
Another example, looked like totaly wrong data, but it all same source - SPI internals was not reset after last recive(disable/enable do not do it, awesome premature optimization thank you very much). So blocking recive return control even befor requred amoint of data is recived and clock is active.
https://i.imgur.com/JHtG5Rs.png
