A10.2.1 Read the Failure

MediumDebugIcarus · browser samples / server checksReview the topic

Four logs from four runs. Say for each one whether the design, the reset, the assertion clock or a vacuous rule is to blame.

Hardware context: Four failing regression logs from a stall rule, a handshake, a mis-clocked check and an idle rule. Blame the design, the reset, the assertion clock or the rule itself, from the log and the code.

Each log below is a complete program and its rule. Read what the rule checks, when it fires, and why.

Write answer(q). For questions 3 to 6, return the cause code of Log A to Log D. The codes: 1 a design bug, 2 a reset mid-transfer, 3 the wrong assertion edge, 4 a rule that never triggered. Question 7: the edge where Log A's failing attempt began.

Questions 0 to 2 are worked examples on another program. Questions 3 to 7 are graded only by the full check.

Signals in the Timeline

SignalWidthRole
question4input of each check
Full specification
Log A, question 3
module tb;
  logic clk = 0; always #5 clk = ~clk;          // edges at 5, 15, 25, ...
  logic valid = 0, ready = 0; logic [7:0] data = 8'h00;
  stall_hold: assert property (@(posedge clk) valid && !ready |=> $stable(data))
    else $display("t=%0t stall_hold fails", $time);
  initial begin
    #12 valid = 1; data = 8'h3c;                // edge 15: valid, not ready
    #10 data = 8'h3d;                           // edge 25: the data moved during the stall
    #10 ready = 1;                              // edge 35
    #10 valid = 0; ready = 0;
    #10 $finish;
  end
endmodule
Log B, question 4
module tb;
  logic clk = 0; always #5 clk = ~clk;
  logic rst = 0, req = 0;
  // The responder: ack three edges after req, through a shift register that the reset clears.
  logic [2:0] sr = 0;
  always @(posedge clk) sr <= rst ? 3'b000 : {sr[1:0], req};
  wire ack = sr[2];
  hs: assert property (@(posedge clk) req |-> ##3 ack) else $display("t=%0t hs fails", $time);
  initial begin
    #12 req = 1;                                // edge 15: due at 45
    #10 req = 0; rst = 1;                       // t=22: reset over the edges at 25 and 35
    #20 rst = 0;                                // t=42
    #20 $finish;
  end
endmodule
Log C, question 5
module tb;
  logic clk = 0; always #5 clk = ~clk;          // rising at 5, 15, ...; falling at 10, 20, ...
  logic valid = 0, ready = 0; logic [7:0] data = 8'h00;
  // The design moves data at each rising edge after a handshake.
  always @(posedge clk) if (valid && ready) data <= data + 8'd1;
  hold_n: assert property (@(negedge clk) valid && !ready |=> $stable(data))
    else $display("t=%0t hold_n fails", $time);
  initial begin
    #12 valid = 1;                              // stalled from 12
    #10 ready = 1;                              // t=22: handshake at 25, data moves at 25
    #10 valid = 0; ready = 0;
    #10 $finish;
  end
endmodule
Log D, question 6
module tb;
  logic clk = 0; always #5 clk = ~clk;
  logic req = 0, ack = 0; logic [1:0] mode = 2'd0;
  int fails = 0;
  m3: assert property (@(posedge clk) req && mode == 2'd3 |=> ack) else fails++;
  initial begin
    #12 req = 1; mode = 2'd2;                   // mode 3 never comes
    #30 req = 0;
    #10 $display("m3 fails=%0d", fails);
    $finish;
  end
endmodule
The worked examples, questions 0 to 2
module tb;
  logic clk = 0; always #4 clk = ~clk;          // edges at 4, 12, 20, 28, 36
  logic go = 0, busy = 0;
  start: assert property (@(posedge clk) go |-> ##2 busy) else $display("t=%0t start fails", $time);
  initial begin
    #6  go = 1;                                 // edge 12: due at 28
    #8  go = 0;                                 // t=14: busy never rises: a design bug
    #26 $finish;
  end
endmodule
qReturn
0the time printed by the example
1the edge where its failing attempt began
2its cause code
3, 4, 5, 6the cause code of Log A, B, C, D
7the edge where Log A's failing attempt began

Simulator note: Every log was run on Verilator 5.052 (measured by tools/probes/sva-w1b/measure-keys.mjs). Your answer function is checked by a small bench on Icarus 12.

Examples

Example 1

When It Reports
Input
answer(0)
Output
answer = 28

The attempt needs busy two edges later; it is still low at 28.

Example 2

Walk Back
Input
answer(1)
Output
answer = 12

Two edges of 8 before 28 is 12, where go was sampled high.

Example 3

Blame
Input
answer(2)
Output
answer = 1

Nothing interrupted the design; it simply never raised busy: a design bug.

Constraints and Contract

  • Return one value per question from answer(q).
  • Keep the function header.

Hints

Hint 1

Find the failing attempt first: the failure time minus the delays of the rule.

Hint 2

Is the clock of the assertion the clock the design uses?

Hint 3

Zero failures can mean the rule never triggered.

What the grader checks: Eight constant answers: three worked examples and five graded questions. Failure times were measured on Verilator 5.052.

8 checks on Icarus · browser samples / server checks. 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

Report a problem with this page

Your code SystemVerilogSaved locally
Ln 1, Col 1Tab indents · Ctrl/⌘ + Enter runs samples
Not run yet

When It Reports

Input
answer(0)
Expected output
answer = 28

The attempt needs busy two edges later; it is still low at 28.

Run samples executes all 3 public checks. Check all tests runs 8 checks on the server.

Icarus · browser samples / server checks