Bug when using a second Triple-Speed-Ethernet IP in a system.

Altera_Forum · ‎07-30-2010

I all!

I have a basic system

nios ii/f

main_mem

avalon_mm_pipeline_bridge

timer_sys_clk

sysid

jtag_uart

sgdma_tx

sgdma_rx

descriptor_memory

tse_mac

that works well (except that I can't use Avalon-MM pipeline bridge, cf. http://www.alteraforum.com/forum/showthread.php?t=24531 (http://www.alteraforum.com/forum/showthread.php?t=24531))

Now I have instanciated a second triple-speed-ethernet IP.

I connected config avalon MM bus to CPU data bus.

I connected RGMII bus to corresponding IOs of my board

I connected RX and TX avalon streaming buses to plugs.

In these conditions, the mac is not usable, so I did not even added software to handle it. System should work in the exactly same manner as before instanciation.

but now, I soon as I ping the tse_mac interface, the system crashes. I observed the config bus of the new mac, and there is not traffic there. So I suspect there is a bug in the way the syslib or the software is generated.

Has anyone ever encountered this issue?

Thanks

-------------------------------------------------------------------------

P.S: the way the system crashes is very weird:

My main() function displays a message <WELCOME_MESSAGE>, clears the timer, launches the initial task, and starts the OS.

The initial tasks inits the NicheStack, and starts a monitoring task which is my only task (except for the interniche tasks dealing with the tse interface) and self-destructs.

The monitoring tasks basically does:

while (1) {

printf monitor message;

tksleep(5000); // sleep 5s

} [/INDENT]

So when I ping the tse, the <WELCOME_MESSAGE> is displayed again (after all the monitoring traces), and then nothing happens. So it seems that for some reason the CPU jumps to main() again, and crashes immediatly after that...