Major Tools
Other Tools
General Info

Issue #1267

Unlinked error on input-only tristate comparisons

Added by Alexis G 12 months ago. Updated 12 months ago.

% Done:



I am using basic Altera's IPcores such as 220model.v (for `lpm_mult`) and altera_mf.v (for `scfifo`).

Code snippet:
%Error: Internal Error: /home/work/gitmodelsim/sim_lib/220model.v:3009: ../V3Tristate.cpp:1123: Unlinked
%Error: Internal Error: See the manual and for more assistance.
%Error: Command Failed /home/extssd/git/verilator//bin/verilator_bin -Wall +1364-2001ext+v --cc /home/work/gitmodelsim/sim_lib/220model.v --top-module lpm_mult --exe lpm_mult.cpp
module lpm_mult (
    dataa,  // Multiplicand. (Required)
    datab,  // Multiplier. (Required)
    sum,    // Partial sum.
    aclr,   // Asynchronous clear for pipelined usage.
    sclr,   // Synchronous clear for pipelined usage.
    clock,  // Clock for pipelined usage.
    clken,  // Clock enable for pipelined usage.
    result  // result = dataa[] * datab[] + sum. The product LSB is aligned with the sum LSB.

    parameter lpm_widtha = 1; // Width of the dataa[] port. (Required)
    parameter lpm_widthb = 1; // Width of the datab[] port. (Required)
    parameter lpm_widthp = 1; // Width of the result[] port. (Required)
    parameter lpm_widths = 1; // Width of the sum[] port. (Required)
    parameter lpm_representation  = "UNSIGNED"; // Type of multiplication performed
    parameter lpm_pipeline  = 0; // Number of clock cycles of latency
    parameter lpm_type = "lpm_mult";
    parameter lpm_hint = "UNUSED";

    input  [lpm_widtha-1:0] dataa;
    input  [lpm_widthb-1:0] datab;
    input  [lpm_widths-1:0] sum;  =====================> here is the line 3009
    input  aclr;
    input  sclr;
    input  clock;
    input  clken;
The compilation of the lpm_mult only results in this error with --debugi-V3LinkDot 9:
- V3LinkDot.cpp:2191:    BEGIN 0x27a11b0 <e26718> {d3229} u1=0x295fcd0 [UNNAMED]
- V3LinkDot.cpp:2196:    cur=se0x295fcd0
- V3LinkDot.cpp:2200:    cur=se0x2967580
- V3LinkDot.cpp:2200:    cur=se0x296a6e0
- V3LinkDot.cpp:2200:    cur=se0x24e6b20
- V3LinkDot.cpp:1587:   MODULE 0x26ee430 <e69135> {d2985} u1=0x24d8d40 u4=0x1  TOP_lpm_mult  L1 [P]
- V3LinkDot.cpp:1632: (Backto) Link Cell: CELL 0x26f1620 <e69138> {d2985} u1=0x24e6b20  lpm_mult -> MODULE 0x2770dc0 <e69137> {d2985} u1=0x24e6b20 u4=0x1  lpm_mult  L2
%Error: Internal Error: /home/work/gitmodelsim/sim_lib/220model.v:3009: ../V3Tristate.cpp:1123: Unlinked
%Error: Internal Error: See the manual and for more assistance.
%Error: Command Failed /home/extssd/git/verilator//bin/verilator_bin -Wall +1364-2001ext+v --cc /home/work/gitmodelsim/sim_lib/220model.v --top-module lpm_mult --exe lpm_mult.cpp --debugi-V3LinkDot 9

I will try to investigate by myself but if you have any idea, you're welcome.


#1 Updated by Todd Strader 12 months ago

The Altera libraries have some seemingly meaningless Verilog nastiness in them. We put these scripts together to get around some of these issues:

I'm not sure if it will work out of the box for you, but hopefully it gets you pointed in the right direction.

#2 Updated by Wilson Snyder 12 months ago

  • Category set to TranslationError
  • Status changed from New to AskedReporter

Todd, thanks for the pointer.

Alexis, as to this specific error, if you can provide a small completely standalone test that shows this issue, I'll take a look.

#3 Updated by Alexis G 12 months ago

Todd, thanks for that. Indeed, it doesn't work out of the box. I need to make some changes in the lpm_mult.v but I keep getting `GET_PARAMETER_VALUE` warnings that end in error.

Wilson, it's pretty basic, you can find the 220model.v attached (seems it doesn't work) and just run: `verilator -Wall +1364-2001ext+v --cc 220model.v --top-module lpm_mult --exe lpm_mult.cpp`

#4 Updated by Wilson Snyder 12 months ago

  • Subject changed from 220model.v:3009: ../V3Tristate.cpp:1123: Unlinked to Unlinked error on input-only tristate comparisons
  • Status changed from AskedReporter to Resolved
  • Assignee set to Wilson Snyder

Fixed in git towards 3.919. I also tested all other models in the library.

However, basically the tristate comparison they have will add an extra __en input to your model you might not expect. This also slows things down. It's much better to update Todd's script to remove the compare against Z in sum (they should have known better when writing it...)

#5 Updated by Wilson Snyder 12 months ago

  • Status changed from Resolved to Closed

In 3.920.

Also available in: Atom