Hello,
In using the CHIP lock-app example for the LP-CC2652R7, it seems as though using the SPI_transfer() function in slave-mode will lock up the board if a timeout is not specified. Is this expected?
In the example below, if params.transferTimeout = x; is not specified, the board appears to hang - i.e. there is no output on the serial debug.
Note that no data is being transferred over the SPI bus at the time:
void SpiSlaveTask(void * pvParameter)
{
SPI_Params params;
SPI_Transaction transaction;
uint8_t rxBuffer[8];
bool rc;
SPI_Params_init(¶ms);
params.bitRate = 4000000;
params.transferMode = SPI_MODE_BLOCKING;
params.transferTimeout = 500000; // this has to be specified
params.frameFormat = SPI_POL0_PHA1;
params.mode = SPI_SLAVE;
SPI_Handle spi = SPI_open(CONFIG_SPI_0, ¶ms);
vTaskDelay(500000);
PLAT_LOG("DEBUG: SPI-SLAVE running");
while ( true )
{
transaction.count = 1;
transaction.txBuf = nullptr;
transaction.rxBuf = (void*)rxBuffer;
memset(rxBuffer, 0, sizeof(rxBuffer));
rc = SPI_transfer(spi, &transaction);
if ( rc )
{
PLAT_LOG("DEBUG -> 0x%x", rxBuffer[0]);
}
PLAT_LOG("DEBUG: timed out ");
}
}
Actually, the SPI_transfer() doesn't seem to really work at all. Only a count size of "1" will receive some data. Sending more than one byte, and verifying those bytes are sent via a logic analyzer, yields no result whatsoever on the TI side. The SPI_transfer() simply times out and never indicates it hasn't received any data.
uint8_t rxBuffer[8];
. . .
transaction.count = sizeof(rxBuffer);
transaction.txBuf = nullptr;
transaction.rxBuf = (void*)rxBuffer;
