v3d: request the kernel to flush caches when TMU is dirty

This adapts the v3d driver to the new CL submit ioctl interface that
allows the driver to request a flush of the caches after the render
job has completed. This seems to eliminate the kernel write violation
errors reported during CTS and Piglit excutions, fixing some CTS tests
and GPU resets along the way.

v2:
  - Adapt to changes in the kernel side.
  - Disable shader storage and shader images if the kernel doesn't
    implement cache flushing.

Fixes CTS tests:
KHR-GLES31.core.shader_image_size.basic-nonMS-fs-float
KHR-GLES31.core.shader_image_size.basic-nonMS-fs-int
KHR-GLES31.core.shader_image_size.basic-nonMS-fs-uint
KHR-GLES31.core.shader_image_size.advanced-nonMS-fs-float
KHR-GLES31.core.shader_image_size.advanced-nonMS-fs-int
KHR-GLES31.core.shader_image_size.advanced-nonMS-fs-uint
KHR-GLES31.core.shader_atomic_counters.advanced-usage-many-draw-calls2
KHR-GLES31.core.shader_atomic_counters.advanced-usage-draw-update-draw
KHR-GLES31.core.shader_storage_buffer_object.advanced-unsizedArrayLength-fs-int
KHR-GLES31.core.shader_storage_buffer_object.advanced-unsizedArrayLength-fs-std140-matR
KHR-GLES31.core.shader_storage_buffer_object.advanced-unsizedArrayLength-fs-std140-struct
KHR-GLES31.core.shader_storage_buffer_object.advanced-unsizedArrayLength-fs-std430-matC-pad
KHR-GLES31.core.shader_storage_buffer_object.advanced-unsizedArrayLength-fs-std430-vec

Reviewed-by: Eric Anholt <[email protected]>

5 files changed, 28 insertions, 22 deletions

diff --git a/src/gallium/drivers/v3d/v3d_job.c b/src/gallium/drivers/v3d/v3d_job.c
index 70176045545..0acd8b033c2 100644
--- a/src/gallium/drivers/v3d/v3d_job.c
+++ b/src/gallium/drivers/v3d/v3d_job.c
@@ -497,6 +497,10 @@ v3d_job_submit(struct v3d_context *v3d, struct v3d_job *job)
         job->submit.bcl_end = job->bcl.bo->offset + cl_offset(&job->bcl);
         job->submit.rcl_end = job->rcl.bo->offset + cl_offset(&job->rcl);
 
+        job->submit.flags = 0;
+        if (job->tmu_dirty_rcl && screen->has_cache_flush)
+                job->submit.flags |= DRM_V3D_SUBMIT_CL_FLUSH_CACHE;
+
         /* On V3D 4.1, the tile alloc/state setup moved to register writes
          * instead of binner packets.
          */
diff --git a/src/gallium/drivers/v3d/v3d_screen.c b/src/gallium/drivers/v3d/v3d_screen.c
index 957420a5311..2d63a43e2dd 100644
--- a/src/gallium/drivers/v3d/v3d_screen.c
+++ b/src/gallium/drivers/v3d/v3d_screen.c
@@ -175,7 +175,10 @@ v3d_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
                 return 4;
 
         case PIPE_CAP_SHADER_BUFFER_OFFSET_ALIGNMENT:
-                return 4;
+                if (screen->has_cache_flush)
+                        return 4;
+                else
+                        return 0; /* Disables shader storage */
 
         case PIPE_CAP_GLSL_FEATURE_LEVEL:
                 return 330;
@@ -356,16 +359,24 @@ v3d_screen_get_shader_param(struct pipe_screen *pscreen, unsigned shader,
                 return V3D_MAX_TEXTURE_SAMPLERS;
 
         case PIPE_SHADER_CAP_MAX_SHADER_BUFFERS:
-                if (shader == PIPE_SHADER_VERTEX)
-                        return 0;
+                if (screen->has_cache_flush) {
+                        if (shader == PIPE_SHADER_VERTEX)
+                                return 0;
 
-                return PIPE_MAX_SHADER_BUFFERS;
+                        return PIPE_MAX_SHADER_BUFFERS;
+                 } else {
+                        return 0;
+                 }
 
         case PIPE_SHADER_CAP_MAX_SHADER_IMAGES:
-                if (screen->devinfo.ver < 41)
+                if (screen->has_cache_flush) {
+                        if (screen->devinfo.ver < 41)
+                                return 0;
+                        else
+                                return PIPE_MAX_SHADER_IMAGES;
+                } else {
                         return 0;
-                else
-                        return PIPE_MAX_SHADER_IMAGES;
+                }
 
         case PIPE_SHADER_CAP_PREFERRED_IR:
                 return PIPE_SHADER_IR_NIR;
@@ -670,6 +681,8 @@ v3d_screen_create(int fd, const struct pipe_screen_config *config,
         slab_create_parent(&screen->transfer_pool, sizeof(struct v3d_transfer), 16);
 
         screen->has_csd = v3d_has_feature(screen, DRM_V3D_PARAM_SUPPORTS_CSD);
+        screen->has_cache_flush =
+                v3d_has_feature(screen, DRM_V3D_PARAM_SUPPORTS_CACHE_FLUSH);
 
         v3d_fence_init(screen);
 
diff --git a/src/gallium/drivers/v3d/v3d_screen.h b/src/gallium/drivers/v3d/v3d_screen.h
index 2f276ac9b0d..db2138d20be 100644
--- a/src/gallium/drivers/v3d/v3d_screen.h
+++ b/src/gallium/drivers/v3d/v3d_screen.h
@@ -78,6 +78,7 @@ struct v3d_screen {
         uint32_t bo_count;
 
         bool has_csd;
+        bool has_cache_flush;
         bool nonmsaa_texture_size_limit;
 
         struct v3d_simulator_file *sim_file;
diff --git a/src/gallium/drivers/v3d/v3dx_rcl.c b/src/gallium/drivers/v3d/v3dx_rcl.c
index e3ead90787f..0434fe84836 100644
--- a/src/gallium/drivers/v3d/v3dx_rcl.c
+++ b/src/gallium/drivers/v3d/v3dx_rcl.c
@@ -795,20 +795,5 @@ v3dX(emit_rcl)(struct v3d_job *job)
                 }
         }
 
-        if (job->tmu_dirty_rcl) {
-           cl_emit(&job->rcl, L1_CACHE_FLUSH_CONTROL, flush) {
-              flush.tmu_config_cache_clear = 0xf;
-              flush.tmu_data_cache_clear = 0xf;
-              flush.uniforms_cache_clear = 0xf;
-              flush.instruction_cache_clear = 0xf;
-           }
-
-           cl_emit(&job->rcl, L2T_CACHE_FLUSH_CONTROL, flush) {
-              flush.l2t_flush_mode = L2T_FLUSH_MODE_CLEAN;
-              flush.l2t_flush_start = cl_address(NULL, 0);
-              flush.l2t_flush_end = cl_address(NULL, ~0);
-           }
-        }
-
         cl_emit(&job->rcl, END_OF_RENDERING, end);
 }
diff --git a/src/gallium/drivers/v3d/v3dx_simulator.c b/src/gallium/drivers/v3d/v3dx_simulator.c
index 87bd2fd0e20..a9d3d8cd780 100644
--- a/src/gallium/drivers/v3d/v3dx_simulator.c
+++ b/src/gallium/drivers/v3d/v3dx_simulator.c
@@ -225,6 +225,9 @@ v3dX(simulator_get_param_ioctl)(struct v3d_hw *v3d,
         case DRM_V3D_PARAM_SUPPORTS_CSD:
                 args->value = V3D_VERSION >= 41;
                 return 0;
+        case DRM_V3D_PARAM_SUPPORTS_CACHE_FLUSH:
+                args->value = 1;
+                return 0;
         }
 
         if (args->param < ARRAY_SIZE(reg_map) && reg_map[args->param]) {