- Ask a related questionWhat is a related question?A related question is a question created from another question. When the related question is created, it will be automatically linked to the original question.
I would like to get some clarification on a weird behavior regarding a modified PA_emacExample from pdk_C6670_1_0_0_17.
The objective of the modification is quite simple: to send packets from a PC and receive them in the TMDSEVM6670LE via the RJ45 connector.
1) To configure the PHY, I replaced the example's Init_Cpsw() function with Javier Malave-Bonet's SWSS_Setup_Swss() function, as instructed elsewhere in this forum. Sometimes this function will report that "PHY is running at 10mbps speed", and some times 1000mbps as well, though my PC's "mii-tool" will report "eth0: negotiated 1000baseT-HD flow-control, link ok" in all cases. I wonder which is correct and if I am actually able to get a full-duplex link.
2) I have modified the Add_MACAddress() function to route everything to pa_DEST_HOST (no filters), and have disabled the Add_IPAddress and Add_Port calls in Setup_PASS(). This way, every packet is captured in the interrupt service routine.
3) The interrupt service routine, Cpsw_RxISR, reads from the accumulator list and eventually calls VerifyPacket. All I did in the latter was disable the byte-wise check so that I can send any payload from the PC. This way, VerifyPacket will happily increment the receive counter and the ISR will terminate.
4) The main task, Cpsw_SingleCoreApp, will do a bunch of initializations and then send MAX_NUM_PACKETS to the wire. After that, there is a while loop which prints the number of received packets so far. I've turned this into an infinite loop and, here's my problem, I will only receive packets if SendPacket() runs at least once.
That's it. This is the relevant code:
/* Setup PA */
if (Setup_PASS () != 0)
{
System_printf ("PASS setup failed \n");
BIOS_exit (-1);
}
else
{
System_printf ("PASS setup successfully done \n");
}
if (SendPacket () != 0) {
System_printf ("Packet %d send failed \n", i);
BIOS_exit (-1);
}
while (gRxCounter >= 0) {
System_printf("Received %d packets so far...\n", gRxCounter);
CycleDelay (10000);
}
More specifically, this is the line inside SendPacket that must run:
Qmss_queuePush (gPaTxQHnd[8], pCppiDesc, dataBufferSize, SIZE_HOST_DESC, Qmss_Location_TAIL);
Anyway, in the main task, if I remove the SendPacket call, the ISR will never be called. I suspect the cause is that some queue has no descriptors and, after the PKTDMA recycles the descriptor which was sent to gPaTxQHnd[8], the whole thing can work again. I would like someone to suggest a way to remove this problem if possible, and an indication of a way to fix the problem in (1) is also very welcome.
