.gitlab-ci/deqp-freedreno-a630-skips.txt

   1 # Note: skips lists for CI are just a list of lines that, when
   2 # non-zero-length and not starting with '#', will regex match to
   3 # delete lines from the test list.  Be careful.
   4
   5 # Skip the perf/stress tests to keep runtime manageable
   6 dEQP-GLES[0-9]*.performance.*
   7 dEQP-GLES[0-9]*.stress.*
   8
   9 # These are really slow on tiling architectures (including llvmpipe).
  10 dEQP-GLES[0-9]*.functional.flush_finish.*
  11
  12 # Flakes reported more than once during Jan-Feb 2020
  13 dEQP-GLES31.functional.layout_binding.ssbo.fragment_binding_array
  14
  15 # This started failing, despite passing locally (and generating identical
  16 # cmdstream as before.  Not sure what is going on, but adding it to skips
  17 # for now
  18 dEQP-GLES31.functional.compute.shared_var.atomic.compswap.lowp_int
  19
  20 # Non-sysmem flakes
  21 dEQP-VK.pipeline.spec_constant.compute.composite.matrix.mat3x2
  22
  23 # Fails NIR_VALIDATE so probably flaky
  24 dEQP-VK.memory_model.write_after_read.core11.u32.coherent.fence_fence.atomicwrite.workgroup.payload_nonlocal.workgroup.guard_local.buffer.comp
  25
  26 # Sysmem flake: this one is fairly frequent, but if you enable it then
  27 # it moves to dEQP-VK.renderpass.dedicated_allocation.attachment.3.393
  28 #
  29 #dEQP-VK.renderpass.suballocation.attachment_allocation.grow_shrink.89
  30
  31 # At least some of the separate_channels tests fail on sysmem due to an
  32 # interaction of use of a UBWC buffer as both a render target and a
  33 # texture.  Stores are done through both paths in separate channels,
  34 # and the UBWC updates don't get synced.  The current a650 blob also
  35 # fails these tests and qcom apparently noted the failure at one point
  36 # https://gitlab.khronos.org/Tracker/vk-gl-cts/-/issues/2017
  37 dEQP-VK.renderpass.*separate_channels.*