ci/freedreno: Move our skips lists over to being known-flakes lists.

This makes sure that we keep executing the tests so that we can get our
alerts in IRC and know whether the tests are still flaking.  It also keeps
us from having adjustments to the skip list causing failures/flakes to
move to different tests (as seen with a530 having to move some xfails
around after changing the skip list)

Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6392>

.gitlab-ci.yml

@@ -1054,7 +1054,7 @@ arm64_a630_gles2:
     BM_KERNEL: /lava-files/cheza-kernel
     BM_CMDLINE: "ip=dhcp console=ttyMSM0,115200n8 root=/dev/nfs rw nfsrootdebug nfsroot=,tcp,nfsvers=4.2 init=/init"
     DEQP_EXPECTED_FAILS: deqp-freedreno-a630-fails.txt
-    DEQP_SKIPS: deqp-freedreno-a630-skips.txt
+    DEQP_FLAKES: deqp-freedreno-a630-flakes.txt
     GIT_STRATEGY: none
     DEQP_EXPECTED_RENDERER: FD630
     DEQP_NO_SAVE_RESULTS: ""
@@ -1175,7 +1175,8 @@ arm64_a306_gles2:
     FLAKES_CHANNEL: "#freedreno-ci"
     BARE_METAL_TEST_SCRIPT: "/install/deqp-runner.sh"
     DEQP_EXPECTED_FAILS: deqp-freedreno-a307-fails.txt
-    DEQP_SKIPS: deqp-freedreno-a307-skips.txt
+    DEQP_SKIPS: deqp-default-skips.txt
+    DEQP_FLAKES: deqp-freedreno-a307-flakes.txt
     DEQP_VER: gles2
     DEQP_PARALLEL: 4
     DEQP_EXPECTED_RENDERER: FD307
@@ -1221,7 +1222,7 @@ arm64_a530_gles2:
     # current upstream kernel.
     BM_CMDLINE: "ip=dhcp console=ttyMSM0,115200n8 nosmp"
     DEQP_EXPECTED_FAILS: deqp-freedreno-a530-fails.txt
-    DEQP_SKIPS: deqp-freedreno-a530-skips.txt
+    DEQP_FLAKES: deqp-freedreno-a530-flakes.txt
     DEQP_EXPECTED_RENDERER: FD530
     NIR_VALIDATE: 0
   tags:

.gitlab-ci/deqp-freedreno-a307-flakes.txt

@@ -1,15 +1,6 @@
-# Note: skips lists for CI are just a list of lines that, when
+# Note: flakes lists for CI are just a list of lines that, when
 # non-zero-length and not starting with '#', will regex match to
 # delete lines from the test list.  Be careful.
-
-# Skip the perf/stress tests to keep runtime manageable
-dEQP-GLES[0-9]*.performance.*
-dEQP-GLES[0-9]*.stress.*
-
-# These are really slow on tiling architectures (including llvmpipe).
-dEQP-GLES[0-9]*.functional.flush_finish.*
-
-# Flaky results
 dEQP-GLES3.functional.occlusion_query.stencil_write
 dEQP-GLES3.functional.rasterization.fbo.rbo_.*
 dEQP-GLES3.functional.rasterization.fbo.texture_2d.interpolation.triangles

.gitlab-ci/deqp-freedreno-a530-fails.txt

@@ -56,10 +56,11 @@ dEQP-GLES31.functional.separate_shader.random.59
 dEQP-GLES31.functional.separate_shader.random.69
 dEQP-GLES31.functional.separate_shader.random.79
 dEQP-GLES31.functional.separate_shader.random.99
-dEQP-GLES31.functional.texture.border_clamp.formats.compressed_rgb8_punchthrough_alpha1_etc2.linear_size_tile_multiple
+dEQP-GLES31.functional.synchronization.inter_call.without_memory_barrier.ssbo_atomic_dispatch_100_calls_32k_invocations
-dEQP-GLES31.functional.texture.border_clamp.formats.luminance.nearest_size_pot
+dEQP-GLES31.functional.texture.border_clamp.formats.depth_component16.nearest_size_pot
-dEQP-GLES31.functional.texture.border_clamp.per_axis_wrap_mode.texture_2d.unorm_color.gather.s_mirrored_repeat_t_clamp_to_border_pot
+dEQP-GLES31.functional.texture.border_clamp.formats.rgba8ui.nearest_size_pot
-dEQP-GLES31.functional.texture.border_clamp.sampler.unorm_depth
+dEQP-GLES31.functional.texture.border_clamp.per_axis_wrap_mode.texture_3d.snorm_color.linear.s_repeat_t_mirrored_repeat_r_clamp_to_border_pot
+dEQP-GLES31.functional.texture.border_clamp.per_axis_wrap_mode.texture_3d.uint_color.nearest.s_clamp_to_border_t_clamp_to_border_r_repeat_pot
 dEQP-GLES31.functional.texture.texture_buffer.modify.bufferdata.buffer_size_131071
 dEQP-GLES31.functional.texture.texture_buffer.render.as_index_array_as_fragment_texture.offset_7_alignments
 dEQP-GLES31.functional.texture.texture_buffer.render.as_vertex_array_as_index_array_as_fragment_texture.offset_1_alignments

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

@@ -1,14 +1,7 @@
-# Note: skips lists for CI are just a list of lines that, when
+# Note: flakes lists for CI are just a list of lines that, when
 # non-zero-length and not starting with '#', will regex match to
 # delete lines from the test list.  Be careful.
 
-# Skip the perf/stress tests to keep runtime manageable
-dEQP-GLES[0-9]*.performance.*
-dEQP-GLES[0-9]*.stress.*
-
-# These are really slow on tiling architectures (including llvmpipe).
-dEQP-GLES[0-9]*.functional.flush_finish.*
-
 # Flakes reported more than once during Jan-Feb 2020
 dEQP-GLES31.functional.layout_binding.ssbo.fragment_binding_array
 
@@ -23,11 +16,6 @@ dEQP-VK.pipeline.spec_constant.compute.composite.matrix.mat3x2
 # Fails NIR_VALIDATE so probably flaky
 dEQP-VK.memory_model.write_after_read.core11.u32.coherent.fence_fence.atomicwrite.workgroup.payload_nonlocal.workgroup.guard_local.buffer.comp
 
-# Sysmem flake: this one is fairly frequent, but if you enable it then
-# it moves to dEQP-VK.renderpass.dedicated_allocation.attachment.3.393
-#
-#dEQP-VK.renderpass.suballocation.attachment_allocation.grow_shrink.89
-
 # At least some of the separate_channels tests fail on sysmem due to an
 # interaction of use of a UBWC buffer as both a render target and a
 # texture.  Stores are done through both paths in separate channels,