1. 17 Nov, 2020 1 commit
  2. 18 Sep, 2020 1 commit
  3. 11 Jul, 2019 2 commits
  4. 06 Jul, 2018 1 commit
  5. 05 Jul, 2018 1 commit
    • intrigeri's avatar
      Test suite: give 3 virtual CPUs to the system under test (refs: #15023). · 209de345
      intrigeri authored
      … otherwise the "Watching a WebM video" test fails ("No video with supported
      format and MIME type found.") and I see this in the Journal:
      
        Jul 04 09:25:39 amnesia tor-browser.desktop[9040]: [Child 9313, MediaPlayback #3] WARNING: Decoder=7e6a6a774200 state=DECODING_METADATA Decode metadata failed, shutting down decoder: file /var/tmp/build/firefox-a0efd2fcd6e9/dom/media/MediaDecoderStateMachine.cpp, line 379
        Jul 04 09:25:39 amnesia tor-browser.desktop[9040]: [Child 9313, MediaPlayback #3] WARNING: Decoder=7e6a6a774200 Decode error: NS_ERROR_DOM_MEDIA_METADATA_ERR (0x806e0006): file /var/tmp/build/firefox-a0efd2fcd6e9/dom/media/MediaDecoderStateMachine.cpp, line 3445
      
      We already had to bump the number of vCPUs to 2 for the exact same reason.
      I suspect that increased e10s usage has something to do with it.
      
      I can't reproduce the bug and that video plays just fine with only 2 vCPUs on
      a manual testing VM of mine, so I'll call this a test suite only issue.
      209de345
  6. 04 Dec, 2017 1 commit
  7. 09 Oct, 2017 1 commit
  8. 22 May, 2017 1 commit
  9. 16 Apr, 2017 1 commit
  10. 05 Apr, 2017 1 commit
  11. 29 Nov, 2016 1 commit
  12. 25 Oct, 2016 1 commit
  13. 24 Oct, 2016 1 commit
    • anonym's avatar
      Make the remote shell use a virtio channel for transport. · 054bcc84
      anonym authored
      The serial channel operated at 400 Kbaud = 50 KB/s (although I think it
      was significantly lower in practice) whereas we now should
      expect (depending on hardware) rates closer to 50 MB/s, i.e. 1000x
      faster.
      054bcc84
  14. 18 Oct, 2016 1 commit
    • bertagaz's avatar
      Plug CDROM only when used by the scenario. · 97710691
      bertagaz authored
      Since QEMU 2.6, VMs can't boot if the CDROM has no ISO file provided in
      the source field. We have to unplug and plug the whole CDROM device to
      workaround this.
      
      Will-fix: #11874
      97710691
  15. 23 Aug, 2016 1 commit
  16. 31 Jul, 2016 1 commit
  17. 29 Jul, 2016 1 commit
    • intrigeri's avatar
      Test suite: use more recent virtual hardware, i.e. USB 3.0 (nec-xhci) on a pc-i440fx-2.5 machine. · 499c6309
      intrigeri authored
      I have some vague hope that switching USB controllers might help with problems
      we see on Jenkins when booting from USB (refs: #11588). This change requires
      upgrading the machine type as well, QEMU otherwise won't boot from this USB
      3.0 controller. And while we're at it, let's migrate from IDE to SATA,
      in order to reflect better hardware Tails is being used with.
      
      Also, there are chances that more recent virtual hardware sees more testing
      these days, so it sounds potentially useful to "upgrade".
      
      Note that I've initially tried the more modern pc-q35-2.5 machine type,
      which worked fine when running the test suite on a sid host, but when
      running on Jessie the VM under testing crashed when logging into GNOME.
      I'll file a ticket about trying this again once Stretch is out.
      499c6309
  18. 21 Jul, 2016 1 commit
  19. 18 Jul, 2016 1 commit
    • intrigeri's avatar
      Revert "Give the VM used for the test suite a hardware RNG device." · 7bcaf91d
      intrigeri authored
      This reverts commit a47df202.
      
      Until libvirt 1.3.4, once can only use /dev/random (which is blocking)
      as a backend for virtio-rng, and as a result I see weird test suite
      failures in which the system under test freezes.
      
      Once we can use libvirt 1.3.4+ and want to actually implement tests
      that exercise rngd, we can revisit.
      
      refs: #5650
      7bcaf91d
  20. 08 Jun, 2016 2 commits
  21. 18 May, 2016 1 commit
  22. 29 Feb, 2016 1 commit
  23. 19 Jan, 2016 1 commit
  24. 06 Aug, 2015 1 commit
  25. 08 Jul, 2015 1 commit
  26. 19 May, 2015 1 commit
  27. 18 May, 2015 2 commits
    • anonym's avatar
      Blacklist the psmouse kernel module. · 2dd9895c
      anonym authored
      At least when two "relative" pointing devices are present at the same
      time, like a ps2 mouse and usb mouse, I've seen major breakage caused
      by host mouse events being used for each such device in virt-viewer,
      effectively doubling all mouse events inside the guest (i.e. one click
      => double click, mouse movements are doubled, etc). Hence it feels
      safer to disable the ps2 mouse and only rely on the ("absolute")
      tablet device (which has further improvements via Spice).
      
      Unfortunately it seems impossible to remove the ps2 mouse since it's
      part of all QEMU machine templates, including the one we use,
      presumably with the exception of 'none', which should be an empty
      machine. However, I've so far not been able to come up with a config
      that works based on 'none' (generally I get alias errors with the PCI
      controller). So we also have to blacklist the module.
      
      Will-fix: #9425
      2dd9895c
    • anonym's avatar
      Enable Spice in the guest. · b81076a1
      anonym authored
      This seems to help with the issues we have with lost mouse events, in
      particular in combination with an absolute pointing device, like the
      'tablet' we already have.
      
      There are some interesting bits about this mentioned in the Spice
      manual:
      
          https://elmarco.fedorapeople.org/manual.html
      
      So currently we use mouse mode='client' which "is appropriate for
      ... a loaded server, since cursor has smooth motion and
      responsiveness", which we have some data suggestion have been
      affecting us (see discussion on #8928). However, it also says "the
      cursor might lose synchronization (position and shape) for a while"
      but it's not mentioned what may cause this (when there's high
      load?). It could be interesting to test mouse mode='server' too, just
      for comparison.
      
      Will-fix: #9425
      b81076a1
  28. 14 Apr, 2015 2 commits
    • anonym's avatar
      Move domain/network name/uuid to configuration. · af6ad5ce
      anonym authored
      It's more convenient than parsing it from the XML, and it's one step
      close to making it possible to run multiple test suites on the same
      system by making them unique in local.yml.
      af6ad5ce
    • anonym's avatar
      Remove unnecessary device. · 8c121d48
      anonym authored
      It's not used, and I've seen it cause trouble when 2+ USB devices are
      plugged before booting the domain.
      8c121d48
  29. 10 Apr, 2015 1 commit
  30. 23 Jan, 2015 1 commit
  31. 18 Jul, 2014 1 commit
    • Tails developers's avatar
      Use qemu-system-x86_64 directly, instead of kvm, for running the automated... · ba0b1f97
      Tails developers authored
      Use qemu-system-x86_64 directly, instead of kvm, for running the automated test suite (Closes: #7605).
      
      This allows libvirt to correctly detect the qemu capabilities,
      and thus to avoid this error:
      
          Call to virDomainCreateWithFlags failed: unsupported configuration:
          ich9-usb-ehci1 not supported in this QEMU binary (Libvirt::Error)
      ba0b1f97
  32. 08 Jun, 2014 1 commit
  33. 07 May, 2014 1 commit
  34. 04 Apr, 2014 1 commit
  35. 27 Mar, 2013 1 commit
    • Tails developers's avatar
      Reorganise features/, unifying both test suites. · 375216ce
      Tails developers authored
      Now all .feature:s reside directly in the root of features/, and they
      are differentiated with tags: source tests are tagged '@source' and
      product (i.e. Tails ISO image) tests are tagged '@product'. These tags
      then set up the appropriate environment on a feature-by-feature basis.
      375216ce
  36. 26 Feb, 2013 1 commit