Upgrade Linux to 4.12
Linux 4.11 is now in Debian testing. It's likely it (or 4.12) will make it into stretch-backports in time for Tails 3.2. Let's get ready for it.
#11 Updated by intrigeri over 2 years ago
- Subject changed from Upgrade Linux to 4.11 to Upgrade Linux to 4.12
stretch-backports now has 4.11.6-1~bpo9+1 (which might not be affected by the dkms issue that's been blocking this ticket so far), but it's lagging behind sid (4.11.11-1) due to RC bugs preventing transition to testing, so I'm not confident with pulling it in Tails. And 4.12.6-1 just got uploaded to sid, so let's bet on that one instead. I'll update my branch once binary packages are available.
#16 Updated by intrigeri over 2 years ago
- % Done changed from 10 to 20
Some tests fail on Jenkins for reasons that seem related to the testing environment more than to this branch; I'll investigate.
As a workaround I've grown
/tmp/TailsToaster on isotester*.lizard. I'm now testing if that's enough & doesn't cause regressions elsewhere (worst case I'll also give more RAM to our isotesters).
It seems the problem is that every snapshot (
*.memstate) takes 2GB on "disk" when the guest VM runs Linux 4.12, regardless of how much RAM is really used in the VM, while with Linux 4.9 they're smaller (e.g. I see 955MB for
tails-greeter-snapshot.memstate and 1.4GB for
no-network-logged-in-snapshot.memstate). I suspect that some new hardening feature (e.g. the kernel initializing memory when it's allocated?) marks all memory as used, therefore preventing QEMU from saving a smaller snapshot, and then there's nothing we can do about it: we simply need to live with it and allocate more space to
/tmp/TailsToaster. anonym, as our snapshots expert, it would be nice if you can have a look at this, but don't spend more than 1 hour on it since the workaround is trivial.
#17 Updated by intrigeri over 2 years ago
Woohoo, test suite passed on Jenkins https://jenkins.tails.boum.org/view/Tails_ISO/job/test_Tails_ISO_feature-12732-linux-4.12/16/ except one issue that's recurring on other branches anyway (I expect I'll flag it as fragile when triaging August false positives).
#20 Updated by intrigeri over 2 years ago
- % Done changed from 20 to 30
I've seen the full test suite pass locally on this branch (except the failure that's expected due to #14586). The subset of the test suite we run on lizard passed 3 times in a row. Now running the full test suite on lizard (we can't compare the results with anything but it might highlight regressions I've missed so far). Once I'm happy with these results I'll submit for QA (with our without the commit to pull the Linux packages from stretch-backports, depending on https://buildd.debian.org/status/package.php?p=linux&suite=stretch-backports).
#21 Updated by intrigeri over 2 years ago
- Assignee changed from intrigeri to anonym
- % Done changed from 30 to 50
- QA Check set to Ready for QA
Full test suite on lizard => only 3 failures (https://jenkins.tails.boum.org/job/test_Tails_ISO_feature-12732-linux-4.12-force-all-tests/1/cucumberTestReport/): #14586 + 2 * #11592. Please review'n'merge :)