@User: Okay, found the problem in the last commit; cif/CIFgen.c line 4972 should be

*bls.temps

, not

bls.temps

. I'm not sure why that doesn't just crash immediately, or how it survived the test case I ran before committing the erroneous code. But anyway, it's fixed now (master branch on opencircuitdesign.com ; I have another issue to track down but then I am going to force the github mirror to update rather than leave a bad commit up on the site).

Ok, rebuilding now.

I'm thinking about implementing some kind of method for "ignore this area" when doing DRC checks. It's a little bit tricky in Magic.

Yeah that'd be nice. I think I'll move to setting

MAGIC_DRC_USE_GDS

 to 0 for the time being. I guess that should still provide decent checking since presumably (1) the cells themselves are good and (2) no reason openlane would generate anything in diffusion or poly that could interfere with the abstract view.

@tnt: It's what happens when I am not paying attention and put a compute-intensive routine inside a loop that doesn't need to be there. . .

😄