PDN ring on macro grid, VDD/VSS not connecting despite ring + straps over macro footprint (gf180mcuC)

[RakshithSuresh2001]

Related to a similar issue I opened for IHP SG13G2 (#11087), but this is a genuinely different case: different PDK, different metal stack, and different fix status.

Setup

Integrating a real SRAM macro (gf180mcu_fd_ip_sram__sram512x8m8wm1, single-port, 512x8) into a design on gf180mcuC, via LibreLane. The macro's power pins (VDD, VSS) sit on Metal3. The design's main power grid uses Metal4 (vertical) / Metal5 (horizontal). No via layer needs bridging across a skipped layer here, Metal3 to Metal4 is a direct, adjacent hop.

What I've tried

Started with a macro-scoped add_pdn_ring (Metal3 <-> Metal4), width kept narrow (0.3um, close to this PDK's 0.28um MINWIDTH) based on a min-cut-density lesson from the IHP case. Got most connections working, unconnected node count dropped substantially, but a couple of floating shapes remained regardless of halo (tried 10um and 40um, identical result both times, same coordinates).

Per gadfort's guidance on the IHP thread, added add_pdn_stripe over the macro's own footprint (not just the ring around its perimeter), same Metal3 layer, same 0.3um width, in addition to the existing ring. Rerun produced byte-for-byte identical output, same coordinates, same violation count, same PSM-0069 failure. Wondering if I'm missing something in how the stripe needs to be scoped relative to the ring, or if there's a different mechanism needed here than what worked (partially) for IHP.

Question

Is there a known pattern for combining a macro-grid ring with stripe coverage correctly, specifically whether the stripe needs its own halo/extent settings distinct from the ring's, or whether there's an ordering dependency between the two calls that I'm missing?

Reproducer attached, confirmed to run standalone (no LibreLane needed, run_ol.sh eject) with bash run.sh.

gf180_reproducible_v2.tar.gz

[maliberty]

@gadfort thoughts?

[gadfort]

@maliberty it's not worth looking at until the IHP130 issues are resolved.