07:10 <xobs> Hi, I'm trying to debug a FSM. Is Migen's FSM object documented anywhere? Also, how can I output the current state to some pins, so I can monitor it with an LA?

07:13 <xobs> Or is there no way to get the state, and I just have to manually set signals as I enter and leave states?

07:18 <kay2> xobs: why don't you do it in simulation first ?

07:18 <xobs> kay2: this particular design works fine in simulation, but behaves oddly on real hardware.

07:19 <kay2> rarely happen ;)

07:19 <kay2> you have fsm.state then in this case

07:19 <kay2> you are using a SoC ?

07:20 <xobs> I am.

07:20 <kay2> misoc or litex ?

07:20 <xobs> When I tried that, Python said "AttributeError: 'FSM' object has no attribute 'state"

07:20 <xobs> Litex

07:20 <kay2> then concider using litescope

07:21 <xobs> I don't think I have enough room on this FPGA to add in litescope. It's 90% full already...

07:22 <kay2> xobs: maybe it's there in the finalize ?

07:23 <kay2> what fpga is it ?

07:23 <xobs> IC40UP5K

07:25 <kay2> https://github.com/m-labs/migen/blob/master/migen/genlib/fsm.py#L215

07:25 <kay2> state exists once finalized

07:27 <kay2> xobs: and adding a signal yourself would end up doing https://github.com/m-labs/migen/blob/master/migen/genlib/fsm.py#L175

07:27 <xobs> What does "finalized" mean? Should I call `do_finalize()` myself on the FSM in order to access the state?

07:27 <kay2> migen does it on 2nd pass

07:29 <xobs> Is the FSM documented anywhere? I just see documentation on `FSM.act()`, but I see a lot more functions used. For example, what does NextValue() do?

07:30 <kay2> allows you to synchroniously set a value

07:30 <kay2> what are you trying to do with this fsm ?

07:30 <xobs> I understand that it sets a particular Signal to a given Value. Does that only happen for the next clock cycle? Or does it permanently change the value and it goes back to 0 in two cycles? Or does it set the value and keep it there until it's transitioned to another state?

07:30 <kay2> it does set the value synchroniously !

07:31 <kay2> as if you were in a self.sync

07:31 <kay2> while the rest of the fsm is done in comb

07:31 <kay2> try to explain more what you are trying to achieve with ur fsm

07:31 <xobs> I'm trying to get a USB stack working.

07:33 <xobs> There are two shift registers: One for Tx and one for Rx. I can transmit data just fine, and I can receive data just fine, and this state machine that glues them together works fine in simulation, but gives random and corrupt values during execution on real hardware.

07:33 <xobs> I'm trying to understand whether it's getting into a bad state due to some async logic bugs, or maybe it's just skipping states altogether.

07:50 <kay2> hard to tell

07:56 <xobs> Yeah, I really think I'm not understanding what a Migen FSM is. I brought out one particular state to an external pin, and I'm seeing that it spends 4ns in this state. That... is probably the source of my problems.

08:00 <xobs> Are before/after_entering/leaving() documented anywhere?

08:13 <kay2> in the example?

08:13 <kay2> what usb core are you using ? valentyusb ?

08:22 <xobs> Yeah, I'm porting valentyusb to Fomu.

08:25 <kay2> i thought mitro was doing that

08:26 <xobs> I took it over from him.