Fix a subclass covergroup that reads 0% because the base class keeps sampling its own covergroup of the same name.
Hardware context: A base monitor and a burst monitor that extends it. Make the subclass covergroup the one that observe() samples.
This is a Fix problem: the class in the editor has one realistic defect. Find it and fix it.
The base monitor mon covers lengths with automatic bins. The burst monitor burst_mon extends it and declares its own cov with the plan's four length classes. Tests sample through observe(), and the report reads m.cov on a burst_mon handle m.
Every burst test reports 0% for the burst monitor, however many bursts it sends.
Signals in the Timeline
| Signal | Width | Role |
|---|---|---|
samples | 8 | input of each check |
Full specification
// The base monitor (given): one covergroup, sampled by observe().
class mon;
bit [3:0] len;
covergroup cov;
coverpoint len;
endgroup
function new();
cov = new;
endfunction
virtual function void observe(bit [3:0] l);
len = l;
cov.sample();
endfunction
endclass
| Bin of burst_mon.cov | Lengths |
|---|---|
| single | 1 |
| short_ | 2 to 4 |
| long_ | 5 to 8 |
| max | 9 to 15 |
Each case builds a fresh burst_mon m, calls m.observe(len) for each length, and reads m.cov.get_inst_coverage() times 100, rounded. The full check samples more lists than the examples show.
Examples
Example 1
One Single Beat- Input
lengths: 1
- Output
coverage_percent_x100 = 2500
single is hit: 1 of 4, 25.00%.
Example 2
Short and Long- Input
lengths: 3 6
- Output
coverage_percent_x100 = 5000
3 is short_, 6 is long_: 2 of 4.
Example 3
Length 0- Input
lengths: 1 2 8 0
- Output
coverage_percent_x100 = 7500
0 is in no bin. 1, 2 and 8 cover single, short_ and long_: 3 of 4.
Constraints and Contract
- Keep the class name, the covergroup name
covand its bins. - Do not change the base class.
Hints
Hint 1
Inside mon::observe(), which cov does the name refer to?
Hint 2
A subclass covergroup of the same name hides the base one only for code that sees the subclass.
Hint 3
IEEE 1800-2023 covergroup extends would fix this at the source, but the coverage build rejects it (lesson C19).
What the grader checks: A fresh burst_mon per case is sampled through observe() with 18 lists of lengths. The subclass covergroup's instance coverage is compared with the value IEEE 1800-2023 clause 19 gives. Three lists are shown.
18 checks on Verilator coverage build · server. Run the examples first, then check the full regression. Full checks use the server judge. Downloads contain public samples only. A passing result establishes the tested cases, not a proof for all possible inputs.
See all 3 steps and the final capstone