rules/klayout/lvs: parse xschem netlist

[proppy]

• try multiple extension
• remove modeling equations
• parse X subckt elements
• reorder device signals

Fixes #98 #101

[atorkmabrains]

@proppy I advise against making this change in LVS. The LVS deal with X as subcircuit by default not as primitive. adding the device as X might lead to weird behavior in LVS.

[atorkmabrains]

@proppy Usually, the netlist for simulation use X for devices, but when you create a netlist for LVS purpose, it should use the primitive form of that device like: M, L, R, Q, etc...

[atorkmabrains]

The solution for that should happen in side xschem or any other tool that generates the netlist. LVS should only use primitive form for devices and X for real subcircuits that present a circuit not a device. Remember the purpose of LVS it check the equivalence of the circuit against layout not to simulate the netlist.

[atorkmabrains]

I believe @StefanSchippers has already addressed that #102.

Please don't change the LVS rule deck. It's working correctly.

[proppy]

@proppy I advise against making this change in LVS. The LVS deal with X as subcircuit by default not as primitive. adding the device as X might lead to weird behavior in LVS.

Yes, I agree with #102 the primitive gets correctly exported as M devices with the right pin ordering so I don't believe 94ae368 is necessary anymore.

[proppy]

Please don't change the LVS rule deck. It's working correctly.

I still believe that the part of 061d376#diff-5692c889f49e046ff484fad52b83b49f9adfd5f9f8258289418462282a350f2bR290-R296 that deals with multiple file extension could be useful for interactive usage.

What do you think?

[atorkmabrains]

I don't believe file extension is important for klayout. As long as the file adhere with the Standard Spice Format with the limitation mentioned here:
https://klayout.de/doc-qt4/manual/lvs_io.html#h2-146

[atorkmabrains]

@proppy The LVS:
https://github.com/efabless/globalfoundries-pdk-libs-gf180mcu_fd_pr/tree/main/rules/klayout/lvs
is working correctly. If there are differences between what we have there and here in Google's version, I would advise that you take Efabless version.

[proppy]

I don't believe file extension is important for klayout.

Yes, but the default produced by xschem (.spice) doesn't match with the default consumed by the lvs deck (.cdl).
https://github.com/efabless/globalfoundries-pdk-libs-gf180mcu_fd_pr/blob/main/rules/klayout/lvs/gf180mcu.lvs#L259

[atorkmabrains]

@proppy I get the point, we could change .cdl to .spice in the LVS.

[proppy]

@atorkmabrains yes, see efabless#26