Page MenuHomeFreeBSD

Start testing cloned interfaces
AbandonedPublic

Authored by freqlabs on May 6 2019, 6:49 AM.
Referenced Files
F80171859: D20168.id57084.diff
Thu, Mar 28, 8:34 PM
Unknown Object (File)
Jan 13 2024, 6:22 AM
Unknown Object (File)
Dec 28 2023, 4:47 AM
Unknown Object (File)
Dec 20 2023, 12:22 AM
Unknown Object (File)
Nov 15 2023, 10:09 AM
Unknown Object (File)
Jul 5 2023, 12:09 PM
Unknown Object (File)
May 28 2023, 9:28 PM
Unknown Object (File)
Apr 22 2023, 2:19 AM
Subscribers

Details

Reviewers
None
Group Reviewers
tests
network
Summary

Tests for cloned interfaces such as lagg and vlan have existed for a while but are not currently enabled. The entire sys/net test suite is not built and installed with the tests dist set, but even if it were, there are many test cases that are set to be skipped because they quickly panic the system. These panics are serious problems that need to be fixed and continuously tested moving forward to prevent regressions in the future.

Enable the existing sys/net tests. Missing tests for epair interfaces have been added, as well.

Sponsored by: iXsystems

Test Plan

Test early, test often.

Diff Detail

Repository
rS FreeBSD src repository - subversion
Lint
Lint Skipped
Unit
Tests Skipped

Event Timeline

Perhaps the panics currently requiring tests to be skipped should be fixed before the skipped tests are enabled. Multicast cleanup is pending in the following revisions that I am aware of:
https://reviews.freebsd.org/D19886
https://reviews.freebsd.org/D20070

D19886 at least fixes the if_freemulti: protospec not NULL panics (but currently introduces a memory leak). I have not yet tested with D20070.

I should also note that there are panics in the lagg tests, described in bugzilla here: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234135

That seems like a good idea. I'll try to bring it up during the BSDCan devsummit.

Definitely do not enable any panicing tests unless you're certain that the panic is fixed. A panic will ruin the entire test run.

There are sadly still a lot of panics triggered by these tests.

It may be worth splitting off the bits that add the epair tests, because we should be able to commit that. Those tests look to be stable.

I have created a new revision for just the epair tests here: https://reviews.freebsd.org/D20498