[eder-matheus]

Use the new check_placement -fixed / -placeable options

Fixed instances (tapcells, endcaps, macros) are not moved after floorplan, so they are now checked once at the end of floorplan, and later stages check only placeable cells. A broken floorplan now fails at 2_4 instead of at detailed placement.

• pdn.tcl: runs check_placement -fixed after pdngen, because the blocked-layers check needs the power grid. On failure it writes 2_4_floorplan_pdn-failed.odb.
• The check uses CELL_PAD_IN_SITES_DETAIL_PLACEMENT padding, so padding conflicts between fixed instances are still caught. That variable is now also read in the floorplan stage, and the padding is reset after the check.
• Detailed placement, CTS and global route now use check_placement -placeable. Fixed cells are still treated as obstacles, so overlaps and padding conflicts between movable and fixed cells are still caught.
• The full check after filler placement is unchanged.

Testing: ran full flows for nangate45/gcd, nangate45/tinyRocket, asap7/mock-alu, asap7/riscv32i and asap7/tinyRocket. In all of them:

• the floorplan check passed;
• fixed instances were identical from floorplan to 6_final;
• -placeable gave the same result as the full check at every stage.

A tapcell deliberately pushed off the site grid makes 2_4 fail with DPL-41.

[gemini-code-assist]

Code Review

This pull request updates the flow scripts to check placement with the -placeable flag during CTS, detail placement, and global routing. It also introduces a placement check for fixed instances after power distribution network (PDN) generation, utilizing the CELL_PAD_IN_SITES_DETAIL_PLACEMENT variable which is now also enabled for the floorplan stage. The review feedback points out that in flow/scripts/pdn.tcl, resetting the global placement padding to 0 before writing the failure database makes debugging difficult, and suggests writing the database before resetting the padding.

[openroad-ci]

🔍 QoR check

Metrics reflect the PR merge build — i.e. what will land on the target branch.

Commit 39fc130 · Jenkins build #2 · Baseline: build · View build on dashboard

62 design(s) checked — 0 with regression(s), 0 without a comparable baseline.

[osamahammad21]

is this needed?

[eder-matheus]

It was inserted to follow one of the items from the MakefileContract.md file:

A step never leaves configuration for a later step. A command gets its options in its own log_cmd line; a setting it needs is set right before the call and reset right after. State that crosses steps is in the .odb or another result file. Breaks: the later step's log does not show what it ran with, and a checkpoint changes when the design did not.

In practice this is no-op.