Secondary PG constraints conflict or leave dual-rail cells unplaceable. What do you do?

Short Answer

Secondary PG constraints tell the placer where each secondary supply really has straps, and they conflict when user regions, tool-derived regions, the UPF and the netlist disagree. Run check_secondary_pg_placement_constraints (ICC2), read which pairs disagree, fix the source rather than relying on commit-time resolution, and re-commit. If dual-rail cells still cannot be placed, the allowed regions are too small for them, so widen the margin, add straps or relax the cell-type filter.

Technical Explanation

• Default: ICC2 assumes secondary straps exist everywhere in a VA; constraints narrow that to where they really are, so dual-rail cells sit near them.
• Two ways in: create_secondary_pg_placement_constraints (ICC2) needs -layers or -region; derive_secondary_pg_placement_constraints (ICC2) reads strap shapes on the layers you list.
• Exclusion: -exclude_supply makes a UPF supply physically unavailable, but not for the primary, a supply already used by a cell, or one named in a strategy.
• Commit resolves silently: for example, the same supply with both -supply and -exclude_supply keeps -supply with only a warning.
• Unplaceable cells: regions too narrow, a -lib_cell_types filter that bans the cell type, or no advanced legalizer, which these constraints require.
• Re-commit rule: after any change to constraints, floorplan or UPF, run commit_secondary_pg_placement_constraints (ICC2) again.

# [ICC2]  icc2_shell
derive_secondary_pg_placement_constraints -layers {M9} -margin 15
create_secondary_pg_placement_constraints -name ret_vdd1p0 -supply VDD1p0 -layers {M9} -margin 15 -voltage_areas VA_COP -lib_cell_types {retention_registers isolation_cells}
check_secondary_pg_placement_constraints
report_secondary_pg_placement_constraints
commit_secondary_pg_placement_constraints
check_legality

What To Check

• check_secondary_pg_placement_constraints reports no conflicts before commit.
• Every dual-rail cell type has an allowed region in its VA.
• Allowed region area comfortably exceeds the dual-rail cell area.
• Constraints were re-committed after the last floorplan or UPF change.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): No conflicts, and all 400 retention flops in VA_COP sit within 15 um of a VDD1p0 strap.
• Suspicious (illustrative): Commit printed warnings that it ignored an exclude constraint, and nobody checked which one.
• Hard stop: check_legality shows 60 retention flops outside any allowed region, or dual-rail cells sit where the supply is excluded.

What The Interviewer Is Testing

• Do you know why these constraints exist, not only the command names?
• Would you read commit warnings instead of trusting automatic resolution?

Follow-up Question & Model Response

"How do these constraints work in a hierarchical design?"

Candidate Model Response: You create and commit secondary PG constraints for each block, then at the top. At the top, commit_secondary_pg_placement_constraints -commit_subblocks (ICC2) also commits any uncommitted block constraints. report_secondary_pg_placement_constraints -all_blocks (ICC2) lists them all, and the check command takes -blocks to test specific blocks. Keep block strap plans and top constraints in step, or the block view and top view disagree.