verific: Fixes incorrect aldff inference in verific importer

[QuantamHD]

The following SV module at HEAD imported with verific,

    module my_module(
      input logic [4:0] a,
      input logic clk,
      input logic enable,
      output logic [4:0] z
    );

    reg [4:0] pipeline_register;
    always @(posedge clk) begin
      pipeline_register <= enable ? a : pipeline_register;
    end
    assign z = pipeline_register;

    endmodule : my_module

results in the following output verilog

/* Generated by 0.36 */

(* top =  1  *)
(* hdlname = "my_module" *)
(* src = "/tmp/temp_directory_zTwd0l/my_input.v:2.12-2.21" *)
module my_module(clk, enable, a, z);
  wire [4:0] _0_;
  (* src = "/tmp/temp_directory_zTwd0l/my_input.v:3.25-3.26" *)
  input [4:0] a;
  wire [4:0] a;
  (* src = "/tmp/temp_directory_zTwd0l/my_input.v:4.19-4.22" *)
  input clk;
  wire clk;
  (* src = "/tmp/temp_directory_zTwd0l/my_input.v:5.19-5.25" *)
  input enable;
  wire enable;
  (* src = "/tmp/temp_directory_zTwd0l/my_input.v:6.26-6.27" *)
  output [4:0] z;
  wire [4:0] z;
  (* src = "/tmp/temp_directory_zTwd0l/my_input.v:10.12-12.8" *)
  \$aldff  #(
    .ALOAD_POLARITY(32'd1),
    .CLK_POLARITY(32'd1),
    .WIDTH(32'd5)
  ) _1_ (
    .AD(5'hxx),
    .ALOAD(1'h0),
    .CLK(clk),
    .D(_0_),
    .Q(z)
  );
  (* src = "/tmp/temp_directory_zTwd0l/my_input.v:11.28-11.58" *)
  \$mux  #(
    .WIDTH(32'd5)
  ) _2_ (
    .A(z),
    .B(a),
    .S(enable),
    .Y(_0_)
  );
endmodule

Yosys is incorrectly infering aldffs due to an incorrect conversion of logical 1 and 0 SigBits.

My PR unifies the conversion of Verific::Net objects into SigBits using Yosys' internal representation of special signals like 0,1,x,z. After my PR these signals are correctly converted into DFFs.

[mmicko]

Thanks. Just to point that current inference is not invalid, it produces a correct $aldff cell that can be optimized later with opt pass (mainly opt_dff but also some wire optimization is needed as well). If you run opt pass after import, you will get same $dffe with current code and your changes as well.
Anyway this is improvement since it is able to do some optimization earlier in the process.

[QuantamHD]

Is this ready to be merged?

[mmicko]

@QuantamHD Yesterday, during the review process, there was outage of canonical (ubuntu) services so CI was failing, that is where delay comes from.

[QuantamHD]

Ah, cool. Just wanted to make sure there wasn't any feedback that needed to be addressed. Thanks!