Tails

#6 Updated by intrigeri 2019-09-29 07:19:41

Thanks! I’ll run the full test suite locally (the branch hasn’t +force-all-tests so Jenkins won’t do that for us) and will merge if happy, as this blocks most other current FT work.

#8 Updated by segfault 2019-09-29 10:41:49

intrigeri wrote:
> Thanks! I’ll run the full test suite locally (the branch hasn’t +force-all-tests so Jenkins won’t do that for us) and will merge if happy, as this blocks most other current FT work.

Oh right, I forgot to add +force-all-tests.

> Status changed from In Progress to Needs Validation
>
> (So that our CI tests build reproducibility.)

Should I have done that before I got the Jenkins test results? If so, in which cases should I do that? Because I got used to waiting for the test results before setting the status to “Needs Validation” (as documented in contribute/merge_policy).

#9 Updated by intrigeri 2019-09-29 11:27:23

> Because I got used to waiting for the test results before setting the status to “Needs Validation” (as documented in contribute/merge_policy).

That’s fine! I just wanted to save some time and be able to merge faster, if/once I see the test suite pass.

#10 Updated by intrigeri 2019-09-29 12:18:34

> I’ll run the full test suite locally

“Scenario: Erasure of memory freed by killed userspace processes” failed twice in 2 runs locally. I see it failed on Jenkins too: https://jenkins.tails.boum.org/job/test_Tails_ISO_bugfix-17100-upgrade-linux-to-5.2.0-3/2/cucumber-html-reports/report-feature_3_2622904403.html. I’ve never seen this happen before so it looks very much like a regression.

Other than that, my 1st test suite run looks OK.

I don’t think this regression a blocker for merging this branch so I’ll go ahead once my 2nd test suite run finishes.

But we should file a dedicated ticket to investigate whether it’s a regression in Linux or an intended change of behaviour that our test suite needs to be adjusted for. In passing, /usr/share/doc/linux-image-5.2.0-3-amd64/changelog.gz has quite a few potentially relevant changes in the mm area.

#12 Updated by intrigeri 2019-09-29 12:58:03

intrigeri wrote:
> But we should file a dedicated ticket to investigate whether it’s a regression in Linux or an intended change of behaviour that our test suite needs to be adjusted for. In passing, /usr/share/doc/linux-image-5.2.0-3-amd64/changelog.gz has quite a few potentially relevant changes in the mm area.

See you on Bug #17104.

I’m now going to merge this branch so that devel-based work can resume :)