Unpacked 2-D Array Not Supported as Mailbox Type

[mrstrike007]

Can you attach an example that shows the issue? (Must be openly licensed, ideally in test_regress format.)

`timescale 1ns / 1ps

class genericClass;

    localparam DWIDTH = 6;

    typedef int my_type_t [2**DWIDTH];

    mailbox #(my_type_t) foo_mbx;

    function new();
        this.foo_mbx = new(1);
    endfunction

endclass

module tb_top();

    genericClass genCls;

    initial begin
        genCls = new();

        $finish();
    end

endmodule

The error shows the following:

- V3LinkCells.cpp:210:Link --top-module: MODULE 0x1083460 <e923#> {d17ai} u1=0x1085530 u4=0x1083580  tb_top  L0 [1ns]
- V3LinkLevel.cpp:42: modSortByLevel()
- V3Ast.cpp:1323:     Dumping obj_dir/Vtb_top_001_cells.tree
- V3LinkDot.cpp:4179: linkDotPrimary:
- V3LinkJump.cpp:375: linkJump:
- V3Ast.cpp:1323:     Dumping obj_dir/Vtb_top_006_linkjump.tree
- V3LinkInc.cpp:306:  linkIncrements:
- V3Ast.cpp:1323:     Dumping obj_dir/Vtb_top_007_linkinc.tree
- V3Param.cpp:1517:   param:
%Error: tb_top.sv:9:5: Missing type parameter: 'T'
                     : ... note: In instance 'tb_top'
-node: CLASSREFDTYPE 0x108a690 <e1331> {d9af} @dt=this@(w0)class:mailbox  mailbox cpkg=0x105c6c0 -> CLASS 0x105c6c0 <e9> {c29ae} u3=0x108e9a0  mailbox  L5 [1ns]
    9 |     mailbox #(my_type_t) foo_mbx;
      |     ^~~~~~~

Please note that when changing the typedef to a packed 2D array of type logic the code compiles.

What 'verilator' command line do we use to run your example?

verilator --top-module tb_top --debug --binary --assert -Wall -Wno-fatal --timescale 1ns/1ps --trace --trace-structs ./tb_top.sv

What 'verilator --version' are you using? Did you try it with the git master version?

Verilator 5.030 2024-10-27 rev v5.030-45-g2cb1a8de7

What OS and distribution are you using?

Fedora

May we assist you in trying to fix this in Verilator yourself?

Yes.

[wsnyder]

The first problem here is that V3Param thinks a Void data type matches the mailbox. Bypassing that there's a problem with hashing these nodes, then bypassing that there's a SIDEEFFECT problem in the std:: package to fix.

Will need to get back to this.