wafer.space Community - chore: update to LibreLane dev (#55)

8:05 a.m.

In general, we trust KLayout. Passing magic DRC is not required for tapeout. In some cases, such as for contacts, magic does not operate directly on the GDS layout (at least on gf180mcu), which can lead to false reports of DRC violations. That's why ERROR_ON_MAGIC_DRC is False in the LibreLane config. However, the issue you mentioned came up with the latest LibreLane (and thus magic) update. The errors are M2.2a/M3.2a/M4.2a violations on the bondpad, which look like false positives to me. I'll ask magic upstream about it. Thanks for reporting the issue!

1

is it possible to read the magic DRC logs/results from that #55 run?

You can reproduce it locally, open the layout in KLayout (make librelane-klayout), and then load the .lyrdb marker database from 67-magic-drc/reports/ of the run folder.

basically running gf180mcu-project-template/.github/workflows/ci.yml locally?

8:25 a.m.

oh, I see the root of the repo has run instructions

1

cool cool, if needed i can also attach the log files, but it is also reproducible in CI or from the gf180mcu template repo

Thanks, but I already managed to reproduce them locally.

1

it's odd that the errors only show up under 4 pads

KLayout has a limit on the number of markers it shows for a single rule. This is noted in the marker browser.

The false positive DRC errors reported by magic have now been fixed: https://github.com/wafer-space/gf180mcu-project-template/actions/runs/23190573468?pr=56 This was related to a recent regression in magic and how offsets were calculated in instance arrays.

chore: update LibreLane dev · wafer-space/gf180mcu-project-templat...

Project template for wafer.space MPW runs using the gf180mcu PDK - chore: update LibreLane dev · wafer-space/gf180mcu-project-template@50b46ea

1

great to hear!! i'll merge this into my template

1