drm/vc4: Add fragment shader threading support

FS threading brings performance improvements of 0-20% in glmark2.

The validation code checks for thread switch signals and ensures that
the registers of the other thread are not touched, and that our clamps
are not live across thread switches.  It also checks that the
threading and branching instructions do not interfere.

(Original patch by Jonas, changes by anholt for style cleanup,
removing validation the kernel doesn't need to do, and adding the flag
for userspace).

v2: Minor style fixes from checkpatch.

Signed-off-by: Jonas Pfeil <pfeiljonas@gmx.de>
Signed-off-by: Eric Anholt <eric@anholt.net>

1 files changed, 12 insertions, 5 deletions

diff --git a/drivers/gpu/drm/vc4/vc4_validate.c b/drivers/gpu/drm/vc4/vc4_validate.c
index e18f88203d32..9fd171c361c2 100644
--- a/drivers/gpu/drm/vc4/vc4_validate.c
+++ b/drivers/gpu/drm/vc4/vc4_validate.c
@@ -789,11 +789,6 @@ validate_gl_shader_rec(struct drm_device *dev,
 	exec->shader_rec_v += roundup(packet_size, 16);
 	exec->shader_rec_size -= packet_size;
 
-	if (!(*(uint16_t *)pkt_u & VC4_SHADER_FLAG_FS_SINGLE_THREAD)) {
-		DRM_ERROR("Multi-threaded fragment shaders not supported.\n");
-		return -EINVAL;
-	}
-
 	for (i = 0; i < shader_reloc_count; i++) {
 		if (src_handles[i] > exec->bo_count) {
 			DRM_ERROR("Shader handle %d too big\n", src_handles[i]);
@@ -810,6 +805,18 @@ validate_gl_shader_rec(struct drm_device *dev,
 			return -EINVAL;
 	}
 
+	if (((*(uint16_t *)pkt_u & VC4_SHADER_FLAG_FS_SINGLE_THREAD) == 0) !=
+	    to_vc4_bo(&bo[0]->base)->validated_shader->is_threaded) {
+		DRM_ERROR("Thread mode of CL and FS do not match\n");
+		return -EINVAL;
+	}
+
+	if (to_vc4_bo(&bo[1]->base)->validated_shader->is_threaded ||
+	    to_vc4_bo(&bo[2]->base)->validated_shader->is_threaded) {
+		DRM_ERROR("cs and vs cannot be threaded\n");
+		return -EINVAL;
+	}
+
 	for (i = 0; i < shader_reloc_count; i++) {
 		struct vc4_validated_shader_info *validated_shader;
 		uint32_t o = shader_reloc_offsets[i];