科研 Skill
RTL仿真覆盖率收敛
coverage-closure
Measures RTL simulation code coverage and iteratively closes it toward a target by adding testbench stimulus. Use for a sign-off-grade coverage report.
SKILL.md
Coverage Closure (Simulation Code Coverage)
Measure statement / branch / condition coverage from an xsim run, find the
exact file:line:type that the testbench never exercised, and close the gap by
adding stimulus — not by deleting code, lowering the bar, or pragma-ing the
problem away. Coverage is a necessary signoff signal, not a sufficient one:
100% code coverage with a weak testbench is still a weak testbench. This skill
makes the number honest.
When to use
- "What's my code coverage?" / "How much of the design is covered?"
- "Get coverage to 90% statement / 85% branch before I merge."
- "Which lines / branches are never hit by the testbench?"
- You need a downloadable coverage report for a code review or tape-out gate.
- A regression passes but you suspect it isn't exercising the whole design.
When NOT to use
- FSM state/transition coverage — xsim code coverage has no FSM
dimension (verified). Use SVA
cover property/ directed tests; that is out of scope for this skill. Say so explicitly rather than faking an FSM number. - Functional / assertion coverage (covergroups, SVA cover) —
sim_get_coveragereports code coverage only. Do not present code coverage as functional coverage. - Toggle coverage as a gate — toggle is reported (percentage only) but is routinely dragged to single digits by reserved/constant-zero bits. Do not gate on it.
- "Is timing met / CDC clean / lint clean?" — wrong skill (timing-closure, cdc-analysis, lint-cleanup). Coverage says nothing about those.
- The user wants you to delete untested RTL to raise the percentage — refuse; see Safety rails.
Prerequisites (verify first)
Run these before measuring. Do not assume state.
test_connection— confirm the Tcl server is live inside Vivado. If it fails, stop and tell the user to start Vivado + the Tcl server.get_license_status—sim_get_coverageis a PRO-tier tool. If FREE, stop and report that coverage requires a Pro license.get_project_info— confirm a project is open (oropen_projectfirst).list_simulation_files— confirm the testbench and DUT are in the sim fileset. Coverage is only as good as the TB that drives it.set_simulation_top("<tb_top>")— mandatory.sim_compileresolves the compile order relative to the sim-fileset top. If you skip this, xelab fails withCannot find design unit work.<tb>. (Pitfall #1, verified.)
Only after all five succeed do you instrument and run.
Methodology
One change-class per iteration so cause→effect stays attributable. Always measure with a tool before and after acting.
1. Establish the target (ask if unstated)
Coverage closure is meaningless without a target. If the user gave none, ask, or propose conventional pre-signoff gates and confirm:
- statement ≥ 90%, branch ≥ 85%, condition ≥ 80% (typical).
Record which metrics gate and which modules are in scope (third-party /
generated IP,
glbl, and verification-only modules are normally excluded).
2. Instrument the compile
set_simulation_top("<tb_top>") # step 5 above, re-assert
sim_compile(top_module="<tb_top>", coverage_types="sbc") # s=stmt b=branch c=cond
- Use
"sbc"for gating metrics. Addtonly if the user explicitly wants toggle visibility — never to gate. - If compile fails, read
sim_get_compile_log(stage="compile"), fix the RTL/TB or fileset/order issue, and recompile. A coverage number from a partial compile is invalid.
3. Run the simulation (coverage flushes automatically)
- Short TB (< ~30 s expected):
sim_run(). - Long TB:
sim_run_async()then pollsim_get_sim_status()untilCOMPLETED. Do not fetch coverage while it is stillRUNNING.
No special flag is needed to flush coverage — the normal run/quit path records it (verified). Confirm the run actually completed and the TB passed before trusting coverage; a TB that aborts early under-counts coverage and the number is a lie. If the TB has failures, fix those first — coverage of a broken run is meaningless.
4. Measure — get structured coverage
sim_get_coverage(detail_level=2, uncovered_limit=200)
Returns JSON:
summary— global statement/branch/condition (and toggle if compiled).per_module— per source module,{covered, total, pct}for each metric. Counts are a cross-instance union: a line counts as hit if any instance hit it (verified). Parameterized modules appear asname(PARAM=val).uncovered—[{file, line, type, context}]: the exact lines/branches/ conditions never exercised. This is your worklist.report_path—dashboard.htmlfor human review (download for PR/signoff).
Report the baseline before touching anything (Show your work).
5. Classify each gap before writing any stimulus
Read each uncovered entry against the source (read_file_lines) and classify
it using the table below. Do not write a single line of TB code until every
in-scope gap is classified. The fix is determined by the class, not by the
percentage.
6. Close the gap — smallest safe action first
In priority order:
- Reachable-but-untested → add directed stimulus to the testbench (recommend the edit; show the new case to the user before/while applying).
- Reachable-only-from-another-config → add a parameterization / second TB instance that exercises that branch.
- Genuinely unreachable defensive code / reserved bits → do not fake it.
xsim ignores inline
(* coverage_off *)and// xilinx coverage off(both verified to have zero effect), so you cannot pragma it away. Surface it as a documented exclusion for the user's CI exclusion list (file:line:type), stated as an explicit assumption — never silently drop it.
Apply one class of change per iteration (e.g., "added reset-deassert case"), then re-measure, so you can attribute the delta.
7. Re-measure (mandatory after every change)
Re-run steps 2→4 (sim_compile → run → sim_get_coverage). A coverage claim is
only valid against fresh sim_get_coverage output from the current TB+RTL.
Never report a projected or inferred percentage.
Classification table (uncovered → cause → safe action)
uncovered symptom | Likely cause | Smallest safe action |
|---|---|---|
Statement in else/error branch never hit | TB never drives the error/edge input | Add directed TB case that forces that input; re-measure |
Branch hit one direction only (e.g. if taken, else not) | Missing complementary stimulus | Drive both polarities of the condition in the TB |
| Condition term never both-T/both-F | Sub-expression of a compound &&/|| never toggled | Add stimulus toggling that specific operand |
| Reset/de-assert path uncovered | TB asserts reset but never exercises de-assert sequence | Extend TB reset sequence |
| Whole module 0% | DUT instance not driven, or TB top wrong | Re-check set_simulation_top, fileset, and that the module is instantiated/driven |
MISSING ELSE with line: null | if with no else (synthetic branch) | Normal; add else-direction stimulus or document as N/A |
| Reserved/constant-zero bits, dead defensive default | Genuinely unreachable | Document as exclusion for the user's CI list — do not pragma, do not gate; ask before excluding if reachability is uncertain |
glbl / generated IP / verification module uncovered | Out-of-scope module | Exclude from scope (state it); do not chase |
| Toggle very low | Reserved/idle bits | Expected; do not gate toggle |
Loop + STOP conditions
measure (sim_get_coverage)
└─ classify every in-scope uncovered entry
└─ pick ONE class of reachable gap
└─ add TB stimulus (recommend / apply)
└─ re-compile + re-run + re-measure
└─ compare before/after → repeat
STOP and hand back to the human when:
- All gating metrics meet the target on fresh
sim_get_coverageoutput — done. - Remaining uncovered items are all classified unreachable / out-of-scope, and meeting the literal target would require deleting code or excluding reachable paths — present the exclusion list + costs; do not force the number green.
- Two consecutive iterations add stimulus but the gating metric does not move (thrash) — stop, present the residual gaps and options, ask for direction.
- A gap can only be closed by an RTL change (genuinely dead/unreachable logic that the user may want removed) — recommend it; do not silently edit RTL.
- The TB itself is failing/aborting — coverage is invalid; escalate the TB bug before any coverage work.
Never loop silently. Each iteration reports before→after numbers and what changed.
Safety rails (do not violate)
- No fake-pass. Never raise the percentage by deleting RTL, narrowing scope to dodge a real path, lowering the agreed target, or excluding a reachable line/branch. Exclusions are only for provably-unreachable code, stated as an explicit assumption, and confirmed with the user when reachability is uncertain.
- Pragmas don't work here — don't pretend they do. xsim ignores
(* coverage_off *)and// xilinx coverage off(verified, zero effect). Do not insert them to "handle" reserved bits; that is a silent fake-pass. Exclude in the CI list instead. - Evidence before claims. "Coverage met" / "target reached" REQUIRES fresh
sim_get_coverageoutput from the current TB+RTL on a passing run. Never infer, never reuse a stale number, never project. - Coverage ≠ correctness. High code coverage does not mean the design is correct, timing-clean, or functionally verified. State this when reporting.
- Smallest safe change first. Close gaps by adding testbench stimulus before touching DUT RTL. Recommend RTL/architectural edits; do not apply them silently.
- One change-class per iteration so each delta is attributable.
- Don't gate toggle. Report it if asked; never let it block.
- FSM / functional coverage are out of scope — say so; do not substitute code coverage for them.
- Real tools only. Use exactly:
test_connection,get_license_status,get_project_info,open_project,list_simulation_files,set_simulation_top,sim_compile,sim_run,sim_run_async,sim_get_sim_status,sim_get_compile_log,sim_get_coverage,read_file_lines,update_file/replace_in_file(TB edits). If a needed capability (e.g. FSM coverage) doesn't exist, say so — never invent a tool.
Output
Report, in this shape:
- Scope & target — gating metrics + thresholds; modules in/out of scope.
- Baseline (with evidence) —
summarypercentages + per-metriccovered/totalfrom the firstsim_get_coverage; note the run passed. - Gap analysis — table of in-scope
uncoveredentries:file:line:type→ classification (reachable / config-only / unreachable / out-of-scope) → action. - Iterations — for each: what stimulus was added (one class), and the before→after delta on the gating metric (fresh tool output each time).
- Final state (with evidence) — final
sim_get_coveragesummary+ per-modulecovered/total; explicit PASS/FAIL vs each target. Linkreport_path(dashboard.html) for human review. - Documented exclusions — any unreachable items proposed for the CI
exclusion list (
file:line:type+ one-line justification), flagged as assumptions for the user to ratify. - Caveats — restate: code coverage ≠ functional correctness; no FSM coverage; toggle not gated. Surface any residual gaps and the options/costs if the literal target wasn't honestly reachable.
示例提示词
帮我看看当前代码覆盖率缺口在哪里,补充测试用例提升覆盖率。
所属插件
安装该插件会一并获得同领域的 63 个 Skill / MCP。
同插件的其他科研 Skill
- ANDES电力系统动态仿真Runs ANDES power-system dynamic studies via PowerMCP: power flow, eigenvalue analysis, and time-domain simulation. Use for damping, stability, or oscillation-mode questions.
- 电池选型与电源预算Builds an evidence-tagged power budget and sizes a manufacturable battery with pouch outline, seals, and derating. Use when sizing runtime, mAh, charging, or supply rails.
- FPGA比特流生成与烧录Generates a Xilinx FPGA bitstream and programs it over JTAG, or burns an MCS image into flash for power-on boot. Use when flashing a board or diagnosing boot failures.
- 硬件项目发布包组装Assembles a hardware release package (fab files, BOM/CPL, carts, assembly guide, hashed manifest) and runs a ready-to-build gate. Use before ordering prototypes.
- 时钟域交叉风险审查Audits clock-domain-crossing risk on an FPGA design and applies the smallest safe synchronizer or constraint fix. Use to resolve CDC violations before sign-off.
- 硬件产品认证合规规划Plans certifications a hardware product needs per target market (FCC, CE/RED, UKCA, battery/radio rules), with cost and lead-time estimates. Use before PCB or enclosure tooling.
- Vivado XDC时序约束编写Authors a Vivado XDC timing constraint set from scratch: clocks, I/O delays, and timing exceptions. Use first, before timing closure, on a design with no or incomplete constraints.
- N-1/N-2应急违规处置Playbook for N-1/N-2 contingency violations: ranks contingencies and proposes preventive and corrective actions. Use after a study shows voltage or thermal problems.
- 潮流计算不收敛诊断Playbook for power-flow convergence failures in PSS/E, PSLF, PowerWorld, pandapower, PyPSA, surge, or OpenDSS. Use when a case diverges or fails to solve.
- 硬件成本核算与采购Prices a hardware product at multiple volumes with one-time vs per-unit cost, landed cost per country, and sourcing leads. Use to estimate unit cost before a pitch.
- 配电网DER接纳能力治理Playbook for distribution DER hosting-capacity limits: voltage rise, reverse flow, and thermal limits from PV and storage. Use to estimate how much DER a feeder can host.
- 电子电路设计与审查Designs, audits, or corrects an electronic circuit from a brief, schematic, or netlist, checking power domains, reset sequencing, and ratings. Use for circuit review.
- 硬件可制造性红队审查Red-team audit that fix-loops a hardware project to ready-to-build, re-deriving parts, rails, battery physics, and cost. Use for a reality check before committing to a design.
- 电力系统动态稳定处置Playbook for dynamic stability problems: poor damping, angle instability, or sustained oscillations from eigenvalue or time-domain studies. Use after an unstable study result.
- Egret电力市场调度优化Runs Egret market/operations optimization (DC-OPF, AC-OPF, unit commitment) via PowerMCP. Use for cheapest dispatch, unit commitment, or congestion checks.
- 电子元器件选型核查Researches electronic components against primary sources: part numbers, datasheet limits, lifecycle, stock, and price breaks, one dossier per part. Use to verify or source a BOM.
- EMC电磁兼容预测试Plans grounding, shielding, and filtering for EMC/EMI, and runs an emissions pre-scan with a root-cause fix sequence. Use before a lab test or after a failed EMI scan.
- 产品外壳参数化建模Designs a hardware enclosure as parametric CAD with STEP/STL/GLB output, a DFM check, and Blender assembly validation. Use when designing a case or checking fit.
- 电网频率响应问题处置Playbook for frequency problems: low inertia, high RoCoF, weak primary response, or UFLS risk after loss of the largest unit. Use when a study flags frequency issues.
- FPGA RTL到比特流全流程Drives a Xilinx FPGA design end to end through RTL, simulation, synthesis, implementation, and bitstream, gated at every stage. Use to run the complete build flow in one pass.
- 硬件产品全流程总协调Leads a physical electronic product from idea to a doable design, coordinating specs, parts, battery, PCB, enclosure, render, and cost sub-skills. Use starting a hardware product.
- 硬件原理图与BOM审查Reviews hardware projects: checks schematics against datasheets, pin connections, and firmware-to-hardware alignment across KiCad/Altium files. Use for a design alignment check.
- FPGA ILA在线逻辑调试Inserts an Integrated Logic Analyzer, programs the board, arms a trigger, and reads captured waveforms on real FPGA hardware. Use when sim passes but the board fails.
- 新能源并网影响评估Playbook for screening the grid impact of a new generator, storage, or large-load interconnection. Use to check whether a given MW can connect and what upgrades it needs.
常见问题
RTL仿真覆盖率收敛能做什么?
Measures RTL simulation code coverage and iteratively closes it toward a target by adding testbench stimulus. Use for a sign-off-grade coverage report。
如何安装RTL仿真覆盖率收敛?
在科搭 Codar 左侧点击「插件」,在「组件」页签搜索「RTL仿真覆盖率收敛」并安装;在 Claude Code 或 Codex 中可以发送本页的安装提示词,或下载 Skill 包后解压到 ~/.claude/skills 或 ~/.codex/skills。
RTL仿真覆盖率收敛需要哪些权限?
该科研 Skill没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。
RTL仿真覆盖率收敛经过官方审核吗?
是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。