Hello,
I currently try communicate with a different processor over mcspi. I'm using the provided bsp mcspi driver through the GIO interface layer and for the implementation I followed the given example. The TDA2px is configured as a spi slave and should run in interrupt mode, since polling seems to block all execution.
My problem is that I cant read any data in interrupt mode while I can read data in polling mode. I'm using the GIO Standard Model but have also tried the GIO Issue Reclaim Model this does not seem to make a difference. When I'm using a timeout the GIO_read function will fail with a timeout error, otherwise the function does not seem to return.
While debugging I noticed the function mcspiSlaveTransferStart() is only started if there is a transmission already scheduled.
if ((MCSPI_COMMMODE_SLAVE ==
instHandle->spiHWconfig.masterOrSlave) &&
(instHandle->opMode ==
MCSPI_OPMODE_INTERRUPT))
{
if ((Bool) TRUE ==
Queue_empty(Queue_handle(&(chanHandle->
queuePendingList))))
{
/* This is the first buffer just prime
*this */
tempCastIoPtr = (Void *) ioPacket;
Queue_put(Queue_handle(
&(chanHandle->
queuePendingList)),
(Queue_Elem *) tempCastIoPtr);
ioPacketProcess = NULL;
status = IOM_PENDING;
Hwi_restore(hwiKey);
}
else
{
/* This is the not the first buffer.
* Put current packet in queue. */
tempCastIoPtr = (Void *) ioPacket;
Queue_put(Queue_handle(
&(chanHandle->
queuePendingList)),
(Queue_Elem *) tempCastIoPtr);
/* Get the top element in queue. */
ioPacketProcess =
(IOM_Packet *) Queue_get(
Queue_handle(&(
chanHandle
->
queuePendingList)));
instHandle->currentActiveChannel =
(Mcspi_ChanObj *) chanHandle;
Hwi_restore(hwiKey);
status = mcspiSlaveTransferStart(
instHandle, chanHandle,
ioPacketProcess);
}
}
If I try to trigger the read a second time to execute mcspiSlaveTransferStart, GIO_read will again fail with a timeout but afterwards the interrupt handler will run.
The communication master is transmitting data in an interval which I confirmed with an oscilloscope.
