1. 28 Jan, 2019 1 commit
    • Andrey Orlov's avatar
      Moving m_export_mode flag definition on application level. · 14467db1
      Andrey Orlov authored
      In case of we don't need exported frames (pipeline without rendering),
      application still set export_mode flag to PRIME
      and this led to DEVICE_FAIL run2run issues.
      Behavior have been fixed and issue goes away.
      
      Issue: MDP-50654
      (cherry picked from commit 98dc339fcd27c3acf837117e46237c25fe799b57)
      14467db1
  2. 21 Jan, 2019 1 commit
  3. 20 Dec, 2018 2 commits
    • Vitaliy Ilichev's avatar
      [samples] fix drm drivers detection · 9b7c81e1
      Vitaliy Ilichev authored
      Current mechanism to detect Intel drm drivers uses the following schema:
      1) Walk through the list of PCI devices, get device properties
         (Vendor ID) and calculate number of PCI devices  that corresponds
         to ‘DISPLAY_CONTROLLER_CLASS’ (adapters_num)
      2) Use adapters_num to get path to /dev/dri/cardN and /dev/dri/renderDN
         with checking for Vendor ID.
      But this approach is not working in some cases. For example, if PCI device
      is present but corresponding driver is not set up. So it can be attempt to
      open wrong /dev/dri/renderN and /dev/dri/cardN  paths.
      
      New approach is based on checking for /dev/dri/renderD<N> (N=0..15) with
      drm request for driver name.
      9b7c81e1
    • Andrey Orlov's avatar
      [smt] cpu usage grow fix · 5ffc2423
      Andrey Orlov authored
      Issue: MDP-52379
      5ffc2423
  4. 10 Dec, 2018 1 commit
  5. 07 Dec, 2018 1 commit
  6. 06 Dec, 2018 1 commit
  7. 05 Dec, 2018 2 commits
  8. 04 Dec, 2018 1 commit
  9. 29 Nov, 2018 3 commits
  10. 28 Nov, 2018 3 commits
  11. 27 Nov, 2018 1 commit
  12. 26 Nov, 2018 1 commit
  13. 23 Nov, 2018 2 commits
  14. 21 Nov, 2018 2 commits
  15. 16 Nov, 2018 2 commits
  16. 14 Nov, 2018 2 commits
    • Vitaliy Ilichev's avatar
      [smt] fixed hang in threads waiting mechanism · 89459284
      Vitaliy Ilichev authored
      Old mechanism used pthread_cond_timedwait()/pthread_cond_signal() to wait till end
      of all threads executions. pthread_cond_signal() issued only once at the end of each
      thread execution but the signal has no effect if the main thread is not blocked on
      corresponding pthread_cond_timedwait. As a result smt could hang in the infinite loop.
      89459284
    • Ilya Sergeyev's avatar
      [JPEGE] BGR4 in vaapi_allocator · 07832b8f
      Ilya Sergeyev authored
      07832b8f
  17. 13 Nov, 2018 3 commits
  18. 07 Nov, 2018 3 commits
  19. 25 Oct, 2018 2 commits
  20. 24 Oct, 2018 2 commits
  21. 23 Oct, 2018 2 commits
  22. 17 Oct, 2018 2 commits