1. 03 Aug, 2022 1 commit
  2. 30 Jul, 2022 2 commits
  3. 21 Jul, 2022 1 commit
  4. 06 Jul, 2022 1 commit
  5. 03 Jul, 2022 1 commit
  6. 02 Jul, 2022 1 commit
  7. 01 Jul, 2022 2 commits
  8. 27 Jun, 2022 2 commits
  9. 26 Jun, 2022 2 commits
  10. 24 Jun, 2022 2 commits
  11. 21 Jun, 2022 1 commit
  12. 20 Jun, 2022 2 commits
  13. 15 Jun, 2022 2 commits
    • David Edmundson's avatar
      Drop call to QApplication::desktop in x11locker · 71c1cde4
      David Edmundson authored
      We previously needed to create the root window so that we would be
      notified when the root window changed size.
      Since then API has moved to QScreen to get window sizes which has other
      ways of monitoring changes.
      This is for Qt6 porting.
      Testing done:
      - removed this whole block of code and confirmed it still does
       - removed just this one line and confirmed input is still grabbed
    • Fushan Wen's avatar
      Add `repaintNeeded` to fix warning · d3ff9f3c
      Fushan Wen authored
      This is to keep compatible with WallpaperInterface in plasma-framework.
      It doesn't have any practical use.
  14. 14 Jun, 2022 1 commit
  15. 10 Jun, 2022 2 commits
  16. 07 Jun, 2022 1 commit
  17. 01 Jun, 2022 1 commit
    • Arjen Hiemstra's avatar
      Handle fallback packages when checking for screenlocker API version · fdbba676
      Arjen Hiemstra authored
      The current lookandfeel package may not contain the
      "X-Plasma-ApiVersion" metadata key but at the same time also not define
      a custom lockscreen, instead relying on the fallback mechanism. In that
      case, we would end up setting the package path to an empty string,
      causing everything else that is using the package to treat it as
      To fix that, we need to check if the lockscreen comes from the current
      package and if not, check the API version of the fallback package
      instead. Additionally, we also should explicitly set the package path to
      the default of Breeze instead of using an empty string as path, which
      implies an invalid package rather than "use the default".
  18. 31 May, 2022 1 commit
    • David Edmundson's avatar
      Wait for screenlocker UI to call close · 8dbec4a1
      David Edmundson authored
      This is important as it allows the screenlocker to show a prompt for
      passwordless users before unlocking. For example mobile having a simple
      swipe to unlock.
      It also potentially allows a future successful login animation.
      An additional guard is added so rogue QML can't trivially bypass the
      user password; though pragamtically they could before if they emit the
      right signal in the authenticator.
  19. 28 May, 2022 1 commit
  20. 26 May, 2022 1 commit
    • Nate Graham's avatar
      Remove old non-compliant Ctrl+Alt+L shortcut that interferes with apps · 0ce96496
      Nate Graham authored
      KScreenLocker lets the user lock the screen with the Meta+L and
      Ctrl+Alt+L shortcuts by default. This latter shortcut has been reported
      by users to conflict with the Ctrl+Alt+L shortcut in at least one
      user-facing app. System components shouldn't break apps by masking
      their shortcuts.
      This commit removes the Ctrl+Alt+L shortcut, leaving only Meta+L. This
      un-breaks userspace apps using that shortcut.
      BUG: 454397
      FIXED-IN: 5.25
  21. 25 May, 2022 1 commit
  22. 22 May, 2022 1 commit
  23. 21 May, 2022 1 commit
  24. 19 May, 2022 2 commits
  25. 16 May, 2022 1 commit
    • David Edmundson's avatar
      Disable logind integration for kwin's unit tests · 67782537
      David Edmundson authored
      Using real logind causes issues with CI as it's dependent on external
      factors and it can leave the host machine is an unclean state.
      We can't mock it like we do in the internal tests as easily and it
      wouldn't be very useful to do so. We can't disable completly as
      lockscreentest relies on externally emitting a signal on this object.
  26. 12 May, 2022 1 commit
  27. 11 May, 2022 1 commit
    • David Edmundson's avatar
      Guard double authenticate calls · 84b2cf3c
      David Edmundson authored
      Whilst authenticating we emit signals and expect callbacks to prompts.
      The pam worker is inside pam_authenticate in the nested event loop.
      It is not expected for GUI clients to call authenticate twice without a
      cancel inbetween, but we should still guard against it as it causes a
  28. 10 May, 2022 1 commit
  29. 06 May, 2022 1 commit
  30. 05 May, 2022 1 commit
    • David Edmundson's avatar
      Cleanup kcheckpass · 132adacf
      David Edmundson authored
      kcheckpass existed because historically we needed to be root to check
      passwords. This hasn't been true for tens of years. There are no
      security benefits, all authentication is based on the exit status of the
      greeter application.
      This patch drops kcheckpass and brings everything in process, but in
      another thread.
      This also reports back more fine grained PAM control at the same time
      (following on from !29) forwarding all prompts and messages.
      The unit test has been replaced with one that actually checks against a
      real PAM using pam_wrapper to force a fake user.
  31. 03 May, 2022 1 commit