Fix a packet filter that counts almost every packet as a match and raises its alarm a cycle early. Both faults are in how operators were written.
Hardware context: A packet filter with a masked address compare, a match counter and an alarm flag; and a gain stage that scales a signed byte by an unsigned gain in quarters and saturates. Fix the compare that reads by precedence and the increment that makes the alarm early.
A packet filter watches a bus for one device. A packet matches when its address equals base on the bits that mask selects. The filter counts the matches and raises alarm one cycle after the count was 3.
In the lab the count ran far ahead of the real traffic: nearly every valid packet was counted. And when the test bench was tightened, alarm turned out to rise one cycle before the brief says.
This is a Fix problem. The design compiles, and each line reads as its author meant it. Find what the simulator reads instead, and change as little as you can.
Brief: rising-edge clk; rst_n asynchronous and active-low, clearing hits and alarm. match is combinational: valid is 1, and addr equals base on every bit where mask is 1. At an edge with match = 1, hits goes up by 1 and wraps from 15 to 0. At every edge, alarm loads 1 if hits was 3 before that edge, else 0. Synthesis check: the design must build on the site's synthesis flow with no latch. The cell and level counts come from an open-source flow and are a proxy for area and depth, not a foundry result.
Signals in the Timeline
| Signal | Width | Role |
|---|---|---|
base | 8 | input of each check |
mask | 8 | input of each check |
script | 108 | input of each check |
Full specification
module pkt_filter (
input logic clk,
input logic rst_n, // asynchronous, active-low
input logic valid, // 1: addr holds a packet address at this edge
input logic [7:0] addr,
input logic [7:0] base, // the address to match
input logic [7:0] mask, // 1: compare this bit; 0: ignore it
output logic match, // combinational: this packet matches
output logic [3:0] hits, // matches counted since reset; wraps from 15 to 0
output logic alarm // registered: 1 for the cycle after hits was 3
);
...
endmodule
| Output | Rule |
|---|---|
| match | valid = 1 and (addr and base agree on every bit where mask is 1). Bits where mask is 0 are ignored in both. |
| hits | At a rising edge out of reset with match = 1: hits + 1, wrapping from 15 to 0. |
| alarm | At every rising edge out of reset: 1 if hits was 3 before the edge, else 0. |
Each test sets base and mask, resets the filter and applies 12 packets, one per edge. A packet shown as -- has valid = 0. The result holds match, hits and alarm 1 ns after each edge.
hits is a register: it must change only after every block has read it at the edge. The fixture watches for an earlier change, and early_hits_write must stay false.
Examples
Example 1
The Top Nibble Selects a Device- Input
base=a0 mask=f0 (hex); packets: a3 53 af b0 a0 -- a1 0a -- a9 aa 5a
- Output
match_0 = yes, hits_0 = 1, alarm_0 = no, match_1 = no, hits_1 = 1, alarm_1 = no, match_2 = yes, hits_2 = 2, alarm_2 = no, match_3 = no, hits_3 = 2, alarm_3 = no, match_4 = yes, hits_4 = 3, alarm_4 = no, match_5 = no, hits_5 = 3, alarm_5 = yes, match_6 = yes, hits_6 = 4, alarm_6 = yes, match_7 = no, hits_7 = 4, alarm_7 = no, match_8 = no, hits_8 = 4, alarm_8 = no, match_9 = yes, hits_9 = 5, alarm_9 = no, match_10 = yes, hits_10 = 6, alarm_10 = no, match_11 = no, hits_11 = 6, alarm_11 = no, early_hits_write = no
Only addresses a0 to af match. Six of the ten valid packets do. hits is 3 after step 4, so alarm is 1 at steps 5 and 6. The fourth match at step 6 moves hits on, and alarm falls at step 7.
Example 2
Three Matches in a Row- Input
base=42 mask=ff (hex); packets: 42 42 42 -- -- 43 42 -- -- -- -- --
- Output
match_0 = yes, hits_0 = 1, alarm_0 = no, match_1 = yes, hits_1 = 2, alarm_1 = no, match_2 = yes, hits_2 = 3, alarm_2 = no, match_3 = no, hits_3 = 3, alarm_3 = yes, match_4 = no, hits_4 = 3, alarm_4 = yes, match_5 = no, hits_5 = 3, alarm_5 = yes, match_6 = yes, hits_6 = 4, alarm_6 = yes, match_7 = no, hits_7 = 4, alarm_7 = no, match_8 = no, hits_8 = 4, alarm_8 = no, match_9 = no, hits_9 = 4, alarm_9 = no, match_10 = no, hits_10 = 4, alarm_10 = no, match_11 = no, hits_11 = 4, alarm_11 = no, early_hits_write = no
hits is 3 after step 2 and stays 3 until the next match at step 6. alarm is 1 from step 3 to step 6.
Example 3
No Packet Is Valid- Input
base=00 mask=00 (hex); packets: -- -- -- -- -- -- -- -- -- -- -- --
- Output
match_0 = no, hits_0 = 0, alarm_0 = no, match_1 = no, hits_1 = 0, alarm_1 = no, match_2 = no, hits_2 = 0, alarm_2 = no, match_3 = no, hits_3 = 0, alarm_3 = no, match_4 = no, hits_4 = 0, alarm_4 = no, match_5 = no, hits_5 = 0, alarm_5 = no, match_6 = no, hits_6 = 0, alarm_6 = no, match_7 = no, hits_7 = 0, alarm_7 = no, match_8 = no, hits_8 = 0, alarm_8 = no, match_9 = no, hits_9 = 0, alarm_9 = no, match_10 = no, hits_10 = 0, alarm_10 = no, match_11 = no, hits_11 = 0, alarm_11 = no, early_hits_write = no
valid is 0 at every step, so nothing matches whatever addr holds, and hits stays 0.
Constraints and Contract
- Inputs are known 0/1 values and change while clk is low.
- Keep the module name and the ports.
Hints
Hint 1
Take the compare apart by precedence. Which binds first in that line: | or ==?
Hint 2
An expression used as a condition is true when it is not zero. What is the width of the compare line's result, and when is it zero?
Hint 3
For alarm, ask which value of hits the test reads at the edge where the third match arrives. An increment written with ++ is a blocking assignment.
What the grader checks: Seven directed and eleven seeded scripts of twelve packets each are compared with an independent model. It checks match, hits and alarm after every edge. The fixture also flags a blocking write to hits. Passing shows the filter follows the brief on these scripts; the synthesis check adds the latch count.
The submitted design must synthesize with 0 latches and no combinational loop. Counts come from the pinned open-source synthesis flow. They are proxies for logic size and depth, not physical area or timing.
18 checks on Icarus Verilog + Yosys · 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 2 steps and the final capstone