verification env for can controller using uvm

[swathi92]

in can controller we have many sub modules like

acceptance filter

registers

frame generators

synchronizers etc...

 

how to create verification env for it.

do i need to verify all these sub components using uvm env only or seperately.

each can controller acts as master and slave, if i have 4 can nodes, how many agents are required and does each node have both master and slave capabilities.

[tudor.timi]

It depends on what you have to do. I'm guessing you don't really need to verify each of the individual sub-modules individually, so one big UVM env would be enough. As for the number of agents, it again depends on your objectives. I've been playing around with a CAN controller that has 2 nodes and one agent was fine enough for me. CAN anyway has a broadcast architecture so you can communicate to one agent from either of your DUT's nodes.

[swathi92]

using one agent how can i verify a can controller.

 

i'm thinking that i should have atleast 2 agents.

1 agent is with microprocessor interface used to program can controller and send data to can controller by puttting data into transmit buffer.

1 agent is with can interface used as other node in can bus network.

if i have more nodes (suppose 5) i'm think i should have 5 agents with can interface.

 

is my understanding correct.

[tudor.timi]

I'd rather say you need one CAN agent per CAN bus (if you have multiple), not per CAN node.

[swathi92]

can you please tell how to verify using only 1 agent.

in different kinds of interfaces are present, diff agents are to be taken right?

host and can interfaces are different. so atleast 2 are needed. how to do with 1agent?

[tudor.timi]

Since the discussion is about CAN, wherever I said "1" agent I mean one CAN agent. Any other interfaces you might have (for programming your module, for example) should of course also have their own agents.