Project

General

Profile

[logo] 
 
Home
News
Activity
About/Contact
Major Tools
  Dinotrace
  Verilator
  Verilog-mode
  Verilog-Perl
Other Tools
  IPC::Locker
  Parallel::Forker
  Voneline
General Info
  Papers

Issue #1580

test_regress/t_prot_lib is unstable

Added by Wilson Snyder 17 days ago. Updated 14 days ago.

Status:
Closed
Priority:
Low
Assignee:
Category:
Tests
% Done:

0%


Description

Debugging a test makefile cleanup, I realized why t_prot_lib and t_noprot_lib seem to fail occasionally for me, but then passes on rerun.

Both run t_prot_lib_secret.pl, so there's up to three running (t_prot_lib's, noprot's and secret form the original driver). If they run simultaneously they break.

Suggest we leave t_prot_lib_secret.pl as is, but have t_prot_lib/t_noprot_lib instead of calling the secret.pl test, just run the verilator command standalone putting output in e.g. {obj_dir}/secret/.... (or otherwise). Then each test will be self-contained.

Let me know if you prefer if I do this.

History

#1 Updated by Wilson Snyder 17 days ago

  • Assignee changed from John Coiner to Todd Strader

Todd wrote it originally...

#2 Updated by Todd Strader 14 days ago

This should resolve the issue and make the tests self-contained: https://github.com/toddstrader/verilator-dev/tree/prot-lib-test-fix

I'm not sure why you would have seen this with t_noprot_lib as it does not call t_prot_lib_secret.pl (since it's building a monolithic version of the same thing). I looked around in Travis for an example of any of these reruns but couldn't find any. However, that may be a result of Travis' limited core count.

#3 Updated by Wilson Snyder 14 days ago

Thanks, feel free to push and close, no need for a Changes update.

#4 Updated by Todd Strader 14 days ago

  • Status changed from Assigned to Closed

Done.

Also available in: Atom