Scope
This page covers post-route timing reports, clock-tree inspection, hold repair, and routing after an engineering-change-order step.
Hold-failure procedure
- Confirmed practice — 2025-12-03: Start with
54-openroad-stapostpnr/summary.rpt. Open the failing corner'smin.rptfor the detailed hold path. LibreLane's ECO flow can insert hold buffers. Evidence (Discord) - Proposal/opinion — 2025-12-03: If the paths suggest clock imbalance, inspect the clock tree in the OpenROAD GUI. This was a diagnostic suggestion for one case, not a general finding about all hold failures. Evidence (Discord)
- Confirmed failure mode — 2025-12-03:
Odb.InsertECOBufferscan add cells after routing. Without a following detailed-route step, the new buffer pins can remain unconnected. Add detailed routing after ECO insertion. Evidence (Discord) - Re-run post-route STA and verify that the new cells are connected. This is an inferred verification step from the recorded unconnected-pin failure. Evidence (Discord)
Decisions, proposals, and open points
- Proposal/opinion: A change from the default 7-track library to
gf180mcu_fd_sc_mcu9t5v0was proposed for one design. It did not by itself solve the reported problem. Do not treat 9-track as a default timing cure. Evidence (Discord) - Open question: No archive message establishes one maximum supported clock rate. The result depends on library, constraints, process corner, clock tree, routing, and ECO work. The recorded thread shows manual diagnosis was still required after a library suggestion. Evidence (Discord)
- Conflict to avoid: A buffer inserted by ECO is not proof of a repaired path. The route must include those new cells, and STA must be run again. Evidence (Discord)
Related pages
Sources
1445869250244902994— 2025-12-03T20:08:12.21+00:00 — Leo Moser (mole99) (@mole99) — OpenROAD GUI clock-tree inspection proposal. Evidence (Discord)1445872535160885409— 2025-12-03T20:21:15.395+00:00 — Lofty (@ravenslofty) — post-route timing report path and ECO guidance. Evidence (Discord)1445875652225011933— 2025-12-03T20:33:38.561+00:00 — Lofty (@ravenslofty) — case-specific 9-track library suggestion. Evidence (Discord)1445910109375959070— 2025-12-03T22:50:33.786+00:00 — Trevor Peyton (@trev5514) — detailed routing required after ECO buffer insertion. Evidence (Discord)