New project, litehm2: a hostmot2 port to linsn rv901t

Thanks for the clear instructions. I will link to this post from the README, if that's okay.

Regarding the 5V-issue ... It looks like the Spartan 6 has no clamp diodes on the inputs. This is the reason why it does not immediately break down. So we might get away with this as long as the drivers are only powered when the spartan is also powered.

@sensille
I found there is exactly 72 pins.
If declared more, some lost. If declared less, got LinuxCNC error.
I found code below, but not understand it.
3*24=72, there are no other 72 string anywhere else in the code.
This match, but I don't know, can I change these values?

Do you really need more than 72 pins? It can be changed, but I'm not 100% sure about the implications ,if linuxcnc handles it without modifications.
In the long run we need this value to be dynamic.
If you need it now, I can look into it.

sensille wrote:

Do you really need more than 72 pins?
 

Not now, but I think about colorlight 5a-75e ...

Now I have got big problem.
I can compile litehm2, but board is unstable, even if compiled your config.
I got <hm2/hm2_lite.0: error finishing read! iter=2380>
(Your firmware gives no this error.)
LinuxCNC 2.9 patched and compiled from sources.
LinuxCNC 2.10 not works at all, of course due to Python.
I'm afraid your project is completely unusable - it works for you only ...

So if my firmware is stable for you, there is still something wrong with your build process?
How about the firmware I built for you?
You can also give me your board.conf and I build a firmware for you to double check.

Btw, I think as of 2 days ago litex now has all my patches in master, linuxcnc has it in master, only mesaflash merge request is pending. It should become easier to build and run once all patches make it into releases.

tuxcnc wrote:

I got <hm2/hm2_lite.0: error finishing read! iter=2380>

The dreaded error in all mesa setups, can be network latency. I had that, too. Turned out to be a bad router in the path. Now I run the card back-to-back on a dedicated network interface and it is rock solid.

The problem is only in toolchain.
Theoretically, mine and author's are the same, practically are different and makes different code.
Litex is Python and daily build, yesterday did not work, today works, and tomorrow will not work again...
In C you can compile code written tens of years ago, in Python you can't run code written a week ago...
I updated Litex, I see difference, and will tested.

The <hm2/hm2_lite.0: error finishing read! iter=2380> is NOT from the hardware. I use crossed cable, no switch or router, and got good pings.
Furthermore, if one firmware works but other no, there is no hardware problem.

One thing we could do is you give me your board.conf, I build it and you test if it works for you. If it does, I can give you all involved git commit ids. With that you should be able to reproduce the build.

Could you also please double check that the commit "mac: fix typo" is in your liteeth repo? It should be there after your update, but just to be sure.

I'm sorry building the project gives you so much trouble. I hope the improvements done during the process makes it at least easier for others to come. Thanks for your perseverance.