#19 core dump for systemverilog

open
nobody
None
5
2012-11-14
2012-11-14
Matthew Li
No

Hi ,

First thanks!!

Second, I experienced "core dump" when the sv code has bug. like "begin;" VCS did not catch this problem. :)

Matt

Discussion

  • Matthew Li
    Matthew Li
    2012-11-15

    I did not put detail enough. Here is more detail.

    My "problem" sv code
    516 if (this.tcam_cfg.key_hash.exists(key_str)) begin
    517 idx = this.tcam_cfg.key_hash[key_str];
    518 // Only abort if in the current side of the TCAM
    519 if (idx[8] == dest_core[0]) begin;
    520 `uvm_info(flow.get_name(), $psprintf("%s: collision, duplicate key_hash[%s] = 0x%0x, using existing index",this.get_name(),key_str,idx), UVM_NONE );
    521 abort = 1;
    522 end
    523 end

    the second if has a extra ";" after begin

    The log file showed it pop the if tag. It should not pop the if tags. it should wait until see the keyword end not ";".

    findTag: Finding tag for keyword if (kind = control) at line 519
    findTag: Candidate-Scope = [if,else,while,for,foreach,repeat,forever:end] under Parent-Scope = [if,else,while,for,foreach,repeat,forever:end] (Path = user_check_t
    cam_duplicate_keys:if_516)
    findTag: Found begin-scope identifier: if,else,while,for,foreach,repeat,forever
    Pushing if(start=if,else,while,for,foreach,repeat,forever) onto scope-stack of size 2
    processBlockAssocTag: Extracting current scope = if_519
    Read Identifier =
    processBlockAssocTag: Saw parentheses
    Read Identifier = begin
    Associating sequential block with current scope
    Read Identifier =
    processBlockAssocTag: Saw semicolon
    processBlockAssocTag: Restored char ; into character-stream
    Updating TagPropertySet for control with if_519
    TPS_DUMP :: ID : 130 UPDATES : 0 KIND : control SCOPE : 0
    TPS_DUMP :: NAME : if_519

     
  • Vikram Khosa
    Vikram Khosa
    2012-12-03

    hi Matt, thanks for providing the example. Please refer to the main tracker thread -- I believe this is the same issue that Johannes ran into -- I'm pasting my reply to him below :

    Johannes, I debugged your segfault -- the problem is that I was exiting
    the loop (early) on encountering a semicolon within a block scope. I think
    commenting out line 1112 : "loop = FALSE" inside the ";" case of
    processBlockAssocTag(...) in systemverilog.c should work as a simple fix
    for now (although the "clk" identifier will still not be tagged with the
    implicit "reg" or "logic" type). I will need to test this a bit more
    thoroughly and include it in a consolidated patch later.