1. 01 May, 2018 3 commits
    • Dylan Baker's avatar
      profile: use gz to compress profiles · 30b992bd
      Dylan Baker authored
      This results in substantially smaller profiles and doesn't seem to
      affect runtime.
      
      v2: - install xml and xml.gz files. This is needed so that meta profiles
            will be installed.
      Tested-by: Rafael Antognolli's avatarRafael Antognolli <rafael.antognolli@intel.com>
      30b992bd
    • Dylan Baker's avatar
      Generate xml for builtin profiles · 2f02cf0d
      Dylan Baker authored
      This creates and installs xml for all builtin profiles. Using these
      profiles I see startup times of ~1 second as opposed to more than 10
      with current master, and runtimes that drop by ~1 minute.
      
      v2: - Fix compilation with make. v1 only worked with ninja.
      Tested-by: Rafael Antognolli's avatarRafael Antognolli <rafael.antognolli@intel.com>
      2f02cf0d
    • Dylan Baker's avatar
      tests: use meta profiles · 400dcd5c
      Dylan Baker authored
      This is a rather sizeable commit, I'm sorry for that. Basically there
      was no clean way I could come up with to make the change from a single
      python file to multiple xml files.
      
      The reason we want this is to to avoid duplication amongst profiles (and
      very long build times), by allowing a single profile to be included in
      multiple other profiles. This means that the all profile is now composed
      of the "glslparser", "shader", and "opengl" (everything in all that's
      not glslparser, asmparser, or shader tests). Running `piglit run all
      foo` still produces the same results. This also allows us to mix native
      and non-native tests so that traditional targets like quick_cl still
      work.
      Tested-by: Rafael Antognolli's avatarRafael Antognolli <rafael.antognolli@intel.com>
      400dcd5c
  2. 31 Jul, 2017 1 commit
  3. 12 Jul, 2017 1 commit
  4. 29 Jun, 2017 1 commit
  5. 24 Jun, 2017 1 commit
  6. 16 Jun, 2017 1 commit
  7. 07 Jun, 2017 1 commit
    • Daniel Diaz's avatar
      CMake: define GBM_BO_MAP only when symbol is found · 2217871a
      Daniel Diaz authored
      gbm_bo_map() and _unmap() have been added recently to Mesa,
      and this update may not have reached all implementations of
      GBM, such as the one provided by Mali r6, where said
      definitions can be found in the header file but not in the
      library itself. This leads to errors like the following when
      linking:
        ../../../../lib/libpiglitutil_gl.so.0: undefined reference to `gbm_bo_unmap'
        ../../../../lib/libpiglitutil_gl.so.0: undefined reference to `gbm_bo_map'
        collect2: error: ld returned 1 exit status
        make[2]: *** [bin/point-sprite] Error 1
      
      Instead of relying on the header file, actually try to link
      using that symbol to determine if PIGLIT_HAS_GBM_BO_MAP
      should be defined.
      Signed-off-by: 's avatarDaniel Díaz <daniel.diaz@linaro.org>
      Reviewed-by: Jan Vesely's avatarJan Vesely <jan.vesely@rutgers.edu>
      Reviewed-by: Dylan Baker's avatarDylan Baker <dylan@pnwbakers.com>
      2217871a
  8. 14 Apr, 2017 1 commit
  9. 24 Jan, 2017 1 commit
  10. 18 Nov, 2016 1 commit
  11. 26 Sep, 2016 1 commit
    • Rob Clark's avatar
      dmabuf: fix YUV tests for drivers other than intel · 0222f5db
      Rob Clark authored
      Ok, so the basic problem with the YUV tests is that they currently
      completely ignore driver/hw pitch requirements, since the code that
      allocates the buffer doesn't know the pixel format, only the 'cpp'.
      
      The yuv test creates a small 4x4 yuv eglimage.  If, say, the hardware
      requires the pitch to be aligned to, say, 32pixels, everything is fine
      for the Y plane, but the subsampled U/V or U+V plane has half as many
      pixels.  (This did help me catch a bug in driver, not rejecting the
      dmabuf import with invalid pitch, but that doesn't help to get the
      piglit tests running.)
      
      The best approach I could come up with to fix this is to pass the
      fourcc all the way down to the code that creates the dmabuf (and copies
      src data into the dmabuf).  Unfortunately this makes the patch a bit
      bigger than I was hoping, and not really sure a good way to split it
      up.
      
      This is tested on i965 (with the intel dma-buf backend) and freedreno
      (with the gbm dma-buf backend).  In the gbm case, it requires new
      gbm format values for R8 and GR88, which is on mesa master as of
      this morning.  (So I bumped the gbm version dependency to 12.1.)
      Signed-off-by: Rob Clark's avatarRob Clark <robdclark@gmail.com>
      Reviewed-by: Eric Anholt's avatarEric Anholt <eric@anholt.net>
      Reviewed-by: Kristian H. Kristensen's avatarKristian Høgsberg <krh@bitplanet.net>
      0222f5db
  12. 23 Sep, 2016 1 commit
    • Vinson Lee's avatar
      cmake: Require egl 11.0. · 65de2254
      Vinson Lee authored
      mesa 11.0 is needed for EGLDeviceEXT symbol.
      
      egl_ext_device_query.c: In function ‘main’:
      egl_ext_device_query.c:33:2: error: unknown type name ‘EGLDeviceEXT’
        EGLDeviceEXT device = EGL_NO_DEVICE_EXT;
        ^
      Signed-off-by: Vinson Lee's avatarVinson Lee <vlee@freedesktop.org>
      65de2254
  13. 31 Aug, 2016 1 commit
  14. 22 Aug, 2016 1 commit
  15. 18 Aug, 2016 1 commit
  16. 13 Apr, 2016 1 commit
  17. 19 Mar, 2016 1 commit
  18. 03 Mar, 2016 1 commit
  19. 18 Feb, 2016 2 commits
  20. 15 Feb, 2016 1 commit
  21. 11 Feb, 2016 1 commit
    • Dylan Baker's avatar
      python: function with six version 1.5.2 · c4a5618a
      Dylan Baker authored
      CMake actually marks that we require six 1.4.0, however, I can't find
      any packages anywhere for 1.4.0, and the lowest version I've seen
      requested is 1.5.2.
      
      This fixes requirements for working with six 1.5.2, and sets tox to use
      1.5.2 (and a suitable version of mock). Primarily there are a few things
      we're using that are not available: six.moves.getcwd, six.viewvalues,
      six.python_2_unicode_compatible.
      Signed-off-by: 's avatarDylan Baker <dylanx.c.baker@intel.com>
      Tested-by: Brian Paul's avatarBrian Paul <brianp@vmware.com>
      c4a5618a
  22. 08 Feb, 2016 1 commit
  23. 26 Jan, 2016 1 commit
  24. 11 Dec, 2015 1 commit
  25. 29 Sep, 2015 1 commit
  26. 02 Sep, 2015 1 commit
  27. 15 Jul, 2015 1 commit
  28. 08 Jun, 2015 2 commits
  29. 06 Jun, 2015 1 commit
  30. 05 Jun, 2015 1 commit
  31. 24 Mar, 2015 1 commit
  32. 04 Mar, 2015 1 commit
  33. 03 Mar, 2015 1 commit
  34. 27 Feb, 2015 1 commit
  35. 26 Feb, 2015 1 commit
  36. 23 Feb, 2015 1 commit
    • Dylan Baker's avatar
      Add dependency on Python six · 3780c734
      Dylan Baker authored
      Six is a module that provides a clean, standardized interface for
      handling python2 and python3 from the same code base. This adds a
      requirement on six as a build-time dependency, the plan is to use it
      only for python generators (those called during build time)
      
      While it certainly is possible to reimplement much of what six does
      scratch and not add another python dependency, I think it's better to
      just use six. For one thing a large number of python modules already
      depend on six, so the chances are good that most people already have it
      installed. Second, it's the de facto standard for supporting complex
      code bases in 2 and 3, so it's familiar and many of the corner cases
      have already been addressed.
      
      This adds the necessary cmake boilerplate to ensure that six is
      available. At this time I don't know of a specific version being
      required, but I am currently using 1.9.0
      
      v2: - update README with six dependency (Jordan)
      Signed-off-by: 's avatarDylan Baker <dylanx.c.baker@intel.com>
      Reviewed-by: Jordan Justen's avatarJordan Justen <jordan.l.justen@intel.com>
      3780c734