gallium/auxiliary/vl: add compute shaders for deint yuv
authorSonny Jiang <sonny.jiang@amd.com>
Fri, 7 Jun 2019 20:07:29 +0000 (16:07 -0400)
committerMarek Olšák <marek.olsak@amd.com>
Mon, 29 Jul 2019 21:52:49 +0000 (17:52 -0400)
Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Reviewed-by: Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
src/gallium/auxiliary/vl/vl_compositor.c
src/gallium/auxiliary/vl/vl_compositor.h
src/gallium/auxiliary/vl/vl_compositor_cs.c
src/gallium/auxiliary/vl/vl_compositor_cs.h

index 54a3a21d2cbd944a6f5feec61e2855e71aa8f62f..04808e80d84e525022971d786ff487b1ab204c08 100644 (file)
@@ -36,17 +36,9 @@ init_shaders(struct vl_compositor *c)
    assert(c);
 
    if (c->pipe_cs_composit_supported) {
-      c->cs_video_buffer = vl_compositor_cs_create_shader(c, compute_shader_video_buffer);
-      if (!c->cs_video_buffer) {
-         debug_printf("Unable to create video_buffer compute shader.\n");
+      if (!vl_compositor_cs_init_shaders(c))
          return false;
-      }
 
-      c->cs_weave_rgb = vl_compositor_cs_create_shader(c, compute_shader_weave);
-      if (!c->cs_weave_rgb) {
-         debug_printf("Unable to create weave_rgb compute shader.\n");
-         return false;
-      }
    } else if (c->pipe_gfx_supported) {
       c->fs_video_buffer = create_frag_shader_video_buffer(c);
       if (!c->fs_video_buffer) {
@@ -59,24 +51,24 @@ init_shaders(struct vl_compositor *c)
          debug_printf("Unable to create YCbCr-to-RGB weave fragment shader.\n");
          return false;
       }
-   }
-
-   if (c->pipe_gfx_supported) {
-      c->vs = create_vert_shader(c);
-      if (!c->vs) {
-         debug_printf("Unable to create vertex shader.\n");
-         return false;
-      }
 
       c->fs_yuv.weave.y = create_frag_shader_deint_yuv(c, true, true);
       c->fs_yuv.weave.uv = create_frag_shader_deint_yuv(c, false, true);
       c->fs_yuv.bob.y = create_frag_shader_deint_yuv(c, true, false);
       c->fs_yuv.bob.uv = create_frag_shader_deint_yuv(c, false, false);
       if (!c->fs_yuv.weave.y || !c->fs_yuv.weave.uv ||
-      !c->fs_yuv.bob.y || !c->fs_yuv.bob.uv) {
+          !c->fs_yuv.bob.y || !c->fs_yuv.bob.uv) {
          debug_printf("Unable to create YCbCr i-to-YCbCr p deint fragment shader.\n");
          return false;
       }
+   }
+
+   if (c->pipe_gfx_supported) {
+      c->vs = create_vert_shader(c);
+      if (!c->vs) {
+         debug_printf("Unable to create vertex shader.\n");
+         return false;
+      }
 
       c->fs_palette.yuv = create_frag_shader_palette(c, true);
       if (!c->fs_palette.yuv) {
@@ -112,19 +104,18 @@ static void cleanup_shaders(struct vl_compositor *c)
    assert(c);
 
    if (c->pipe_cs_composit_supported) {
-      c->pipe->delete_compute_state(c->pipe, c->cs_video_buffer);
-      c->pipe->delete_compute_state(c->pipe, c->cs_weave_rgb);
+      vl_compositor_cs_cleanup_shaders(c);
    } else if (c->pipe_gfx_supported) {
       c->pipe->delete_fs_state(c->pipe, c->fs_video_buffer);
       c->pipe->delete_fs_state(c->pipe, c->fs_weave_rgb);
-   }
-
-   if (c->pipe_gfx_supported) {
-      c->pipe->delete_vs_state(c->pipe, c->vs);
       c->pipe->delete_fs_state(c->pipe, c->fs_yuv.weave.y);
       c->pipe->delete_fs_state(c->pipe, c->fs_yuv.weave.uv);
       c->pipe->delete_fs_state(c->pipe, c->fs_yuv.bob.y);
       c->pipe->delete_fs_state(c->pipe, c->fs_yuv.bob.uv);
+   }
+
+   if (c->pipe_gfx_supported) {
+      c->pipe->delete_vs_state(c->pipe, c->vs);
       c->pipe->delete_fs_state(c->pipe, c->fs_palette.yuv);
       c->pipe->delete_fs_state(c->pipe, c->fs_palette.rgb);
       c->pipe->delete_fs_state(c->pipe, c->fs_rgb_yuv.y);
@@ -355,18 +346,27 @@ set_yuv_layer(struct vl_compositor_state *s, struct vl_compositor *c,
       s->layers[layer].zw.x = 0.0f;
       s->layers[layer].src.tl.y += half_a_line;
       s->layers[layer].src.br.y += half_a_line;
-      s->layers[layer].fs = (y) ? c->fs_yuv.bob.y : c->fs_yuv.bob.uv;
+      if (c->pipe_gfx_supported)
+          s->layers[layer].fs = (y) ? c->fs_yuv.bob.y : c->fs_yuv.bob.uv;
+      if (c->pipe_cs_composit_supported)
+          s->layers[layer].cs = (y) ? c->cs_yuv.bob.y : c->cs_yuv.bob.uv;
       break;
 
    case VL_COMPOSITOR_BOB_BOTTOM:
       s->layers[layer].zw.x = 1.0f;
       s->layers[layer].src.tl.y -= half_a_line;
       s->layers[layer].src.br.y -= half_a_line;
-      s->layers[layer].fs = (y) ? c->fs_yuv.bob.y : c->fs_yuv.bob.uv;
+      if (c->pipe_gfx_supported)
+          s->layers[layer].fs = (y) ? c->fs_yuv.bob.y : c->fs_yuv.bob.uv;
+      if (c->pipe_cs_composit_supported)
+          s->layers[layer].cs = (y) ? c->cs_yuv.bob.y : c->cs_yuv.bob.uv;
       break;
 
    default:
-      s->layers[layer].fs = (y) ? c->fs_yuv.weave.y : c->fs_yuv.weave.uv;
+      if (c->pipe_gfx_supported)
+          s->layers[layer].fs = (y) ? c->fs_yuv.weave.y : c->fs_yuv.weave.uv;
+      if (c->pipe_cs_composit_supported)
+          s->layers[layer].cs = (y) ? c->cs_yuv.weave.y : c->cs_yuv.weave.uv;
       break;
    }
 }
index 36f9bb3d4dc72c673e42539d7e2054150af3a4fb..596e055cb09af8c63d0fab6753827d799744118c 100644 (file)
@@ -135,6 +135,17 @@ struct vl_compositor
       } bob;
    } fs_yuv;
 
+   struct {
+      struct {
+         void *y;
+         void *uv;
+      } weave;
+      struct {
+         void *y;
+         void *uv;
+      } bob;
+   } cs_yuv;
+
    struct {
       void *rgb;
       void *yuv;
index abd2aa63f32142531b2ca60928240f96df459ac5..1970861271d60f3e09f15647ebcc8553d2964245 100644 (file)
@@ -264,6 +264,322 @@ const char *compute_shader_rgba =
 
       "END\n";
 
+static const char *compute_shader_yuv_weave_y =
+      "COMP\n"
+      "PROPERTY CS_FIXED_BLOCK_WIDTH 8\n"
+      "PROPERTY CS_FIXED_BLOCK_HEIGHT 8\n"
+      "PROPERTY CS_FIXED_BLOCK_DEPTH 1\n"
+
+      "DCL SV[0], THREAD_ID\n"
+      "DCL SV[1], BLOCK_ID\n"
+
+      "DCL CONST[0..5]\n"
+      "DCL SVIEW[0..2], 2D_ARRAY, FLOAT\n"
+      "DCL SAMP[0..2]\n"
+
+      "DCL IMAGE[0], 2D, WR\n"
+      "DCL TEMP[0..15]\n"
+
+      "IMM[0] UINT32 { 8, 8, 1, 0}\n"
+      "IMM[1] FLT32 { 1.0, 2.0, 0.0, 0.0}\n"
+      "IMM[2] UINT32 { 1, 2, 4, 0}\n"
+      "IMM[3] FLT32 { 0.25, 0.5, 0.125, 0.125}\n"
+
+      "UMAD TEMP[0], SV[1], IMM[0], SV[0]\n"
+
+      /* Drawn area check */
+      "USGE TEMP[1].xy, TEMP[0].xyxy, CONST[4].xyxy\n"
+      "USLT TEMP[1].zw, TEMP[0].xyxy, CONST[4].zwzw\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].yyyy\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].zzzz\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].wwww\n"
+
+      "UIF TEMP[1]\n"
+         "MOV TEMP[2], TEMP[0]\n"
+         /* Translate */
+         "UADD TEMP[2].xy, TEMP[2], -CONST[5].xyxy\n"
+
+         /* Top Y */
+         "U2F TEMP[2], TEMP[2]\n"
+         "DIV TEMP[2].y, TEMP[2].yyyy, IMM[1].yyyy\n"
+         /* Down Y */
+         "MOV TEMP[12], TEMP[2]\n"
+
+         /* Top UV */
+         "MOV TEMP[3], TEMP[2]\n"
+         "DIV TEMP[3].xy, TEMP[3], IMM[1].yyyy\n"
+         /* Down UV */
+         "MOV TEMP[13], TEMP[3]\n"
+
+         /* Texture offset */
+         "ADD TEMP[2].x, TEMP[2].xxxx, IMM[3].yyyy\n"
+         "ADD TEMP[2].y, TEMP[2].yyyy, IMM[3].xxxx\n"
+         "ADD TEMP[12].x, TEMP[12].xxxx, IMM[3].yyyy\n"
+         "ADD TEMP[12].y, TEMP[12].yyyy, IMM[3].xxxx\n"
+
+         "ADD TEMP[3].x, TEMP[3].xxxx, IMM[3].xxxx\n"
+         "ADD TEMP[3].y, TEMP[3].yyyy, IMM[3].wwww\n"
+         "ADD TEMP[13].x, TEMP[13].xxxx, IMM[3].xxxx\n"
+         "ADD TEMP[13].y, TEMP[13].yyyy, IMM[3].wwww\n"
+
+         /* Scale */
+         "DIV TEMP[2].xy, TEMP[2], CONST[3].zwzw\n"
+         "DIV TEMP[12].xy, TEMP[12], CONST[3].zwzw\n"
+         "DIV TEMP[3].xy, TEMP[3], CONST[3].zwzw\n"
+         "DIV TEMP[13].xy, TEMP[13], CONST[3].zwzw\n"
+
+         /* Weave offset */
+         "ADD TEMP[2].y, TEMP[2].yyyy, IMM[3].xxxx\n"
+         "ADD TEMP[12].y, TEMP[12].yyyy, -IMM[3].xxxx\n"
+         "ADD TEMP[3].y, TEMP[3].yyyy, IMM[3].xxxx\n"
+         "ADD TEMP[13].y, TEMP[13].yyyy, -IMM[3].xxxx\n"
+
+         /* Texture layer */
+         "MOV TEMP[14].x, TEMP[2].yyyy\n"
+         "MOV TEMP[14].yz, TEMP[3].yyyy\n"
+         "ROUND TEMP[15], TEMP[14]\n"
+         "ADD TEMP[14], TEMP[14], -TEMP[15]\n"
+         "MOV TEMP[14], |TEMP[14]|\n"
+         "MUL TEMP[14], TEMP[14], IMM[1].yyyy\n"
+
+         /* Normalize */
+         "DIV TEMP[2].xy, TEMP[2], CONST[5].zwzw\n"
+         "DIV TEMP[12].xy, TEMP[12], CONST[5].zwzw\n"
+         "DIV TEMP[15].xy, CONST[5].zwzw, IMM[1].yyyy\n"
+         "DIV TEMP[3].xy, TEMP[3], TEMP[15].xyxy\n"
+         "DIV TEMP[13].xy, TEMP[13], TEMP[15].xyxy\n"
+
+         /* Fetch texels */
+         "MOV TEMP[2].z, IMM[1].wwww\n"
+         "MOV TEMP[3].z, IMM[1].wwww\n"
+         "TEX_LZ TEMP[10].x, TEMP[2], SAMP[0], 2D_ARRAY\n"
+         "TEX_LZ TEMP[10].y, TEMP[3], SAMP[1], 2D_ARRAY\n"
+         "TEX_LZ TEMP[10].z, TEMP[3], SAMP[2], 2D_ARRAY\n"
+
+         "MOV TEMP[12].z, IMM[1].xxxx\n"
+         "MOV TEMP[13].z, IMM[1].xxxx\n"
+         "TEX_LZ TEMP[11].x, TEMP[12], SAMP[0], 2D_ARRAY\n"
+         "TEX_LZ TEMP[11].y, TEMP[13], SAMP[1], 2D_ARRAY\n"
+         "TEX_LZ TEMP[11].z, TEMP[13], SAMP[2], 2D_ARRAY\n"
+
+         "LRP TEMP[6], TEMP[14], TEMP[10], TEMP[11]\n"
+         "MOV TEMP[6].w, IMM[1].xxxx\n"
+
+         "STORE IMAGE[0], TEMP[0], TEMP[6], 2D\n"
+      "ENDIF\n"
+
+      "END\n";
+
+static const char *compute_shader_yuv_weave_uv =
+      "COMP\n"
+      "PROPERTY CS_FIXED_BLOCK_WIDTH 8\n"
+      "PROPERTY CS_FIXED_BLOCK_HEIGHT 8\n"
+      "PROPERTY CS_FIXED_BLOCK_DEPTH 1\n"
+
+      "DCL SV[0], THREAD_ID\n"
+      "DCL SV[1], BLOCK_ID\n"
+
+      "DCL CONST[0..5]\n"
+      "DCL SVIEW[0..2], 2D_ARRAY, FLOAT\n"
+      "DCL SAMP[0..2]\n"
+
+      "DCL IMAGE[0], 2D, WR\n"
+      "DCL TEMP[0..15]\n"
+
+      "IMM[0] UINT32 { 8, 8, 1, 0}\n"
+      "IMM[1] FLT32 { 1.0, 2.0, 0.0, 0.0}\n"
+      "IMM[2] UINT32 { 1, 2, 4, 0}\n"
+      "IMM[3] FLT32 { 0.25, 0.5, 0.125, 0.125}\n"
+
+      "UMAD TEMP[0], SV[1], IMM[0], SV[0]\n"
+
+      /* Drawn area check */
+      "USGE TEMP[1].xy, TEMP[0].xyxy, CONST[4].xyxy\n"
+      "USLT TEMP[1].zw, TEMP[0].xyxy, CONST[4].zwzw\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].yyyy\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].zzzz\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].wwww\n"
+
+      "UIF TEMP[1]\n"
+         "MOV TEMP[2], TEMP[0]\n"
+         /* Translate */
+         "UADD TEMP[2].xy, TEMP[2], -CONST[5].xyxy\n"
+
+         /* Top Y */
+         "U2F TEMP[2], TEMP[2]\n"
+         "DIV TEMP[2].y, TEMP[2].yyyy, IMM[1].yyyy\n"
+         /* Down Y */
+         "MOV TEMP[12], TEMP[2]\n"
+
+         /* Top UV */
+         "MOV TEMP[3], TEMP[2]\n"
+         "DIV TEMP[3].xy, TEMP[3], IMM[1].yyyy\n"
+         /* Down UV */
+         "MOV TEMP[13], TEMP[3]\n"
+
+         /* Texture offset */
+         "ADD TEMP[2].x, TEMP[2].xxxx, IMM[3].yyyy\n"
+         "ADD TEMP[2].y, TEMP[2].yyyy, IMM[3].xxxx\n"
+         "ADD TEMP[12].x, TEMP[12].xxxx, IMM[3].yyyy\n"
+         "ADD TEMP[12].y, TEMP[12].yyyy, IMM[3].xxxx\n"
+
+         "ADD TEMP[3].x, TEMP[3].xxxx, IMM[3].xxxx\n"
+         "ADD TEMP[3].y, TEMP[3].yyyy, IMM[3].wwww\n"
+         "ADD TEMP[13].x, TEMP[13].xxxx, IMM[3].xxxx\n"
+         "ADD TEMP[13].y, TEMP[13].yyyy, IMM[3].wwww\n"
+
+         /* Scale */
+         "DIV TEMP[2].xy, TEMP[2], CONST[3].zwzw\n"
+         "DIV TEMP[12].xy, TEMP[12], CONST[3].zwzw\n"
+         "DIV TEMP[3].xy, TEMP[3], CONST[3].zwzw\n"
+         "DIV TEMP[13].xy, TEMP[13], CONST[3].zwzw\n"
+
+         /* Weave offset */
+         "ADD TEMP[2].y, TEMP[2].yyyy, IMM[3].xxxx\n"
+         "ADD TEMP[12].y, TEMP[12].yyyy, -IMM[3].xxxx\n"
+         "ADD TEMP[3].y, TEMP[3].yyyy, IMM[3].xxxx\n"
+         "ADD TEMP[13].y, TEMP[13].yyyy, -IMM[3].xxxx\n"
+
+         /* Texture layer */
+         "MOV TEMP[14].x, TEMP[2].yyyy\n"
+         "MOV TEMP[14].yz, TEMP[3].yyyy\n"
+         "ROUND TEMP[15], TEMP[14]\n"
+         "ADD TEMP[14], TEMP[14], -TEMP[15]\n"
+         "MOV TEMP[14], |TEMP[14]|\n"
+         "MUL TEMP[14], TEMP[14], IMM[1].yyyy\n"
+
+         /* Normalize */
+         "DIV TEMP[2].xy, TEMP[2], CONST[5].zwzw\n"
+         "DIV TEMP[12].xy, TEMP[12], CONST[5].zwzw\n"
+         "DIV TEMP[15].xy, CONST[5].zwzw, IMM[1].yyyy\n"
+         "DIV TEMP[3].xy, TEMP[3], TEMP[15].xyxy\n"
+         "DIV TEMP[13].xy, TEMP[13], TEMP[15].xyxy\n"
+
+         /* Fetch texels */
+         "MOV TEMP[2].z, IMM[1].wwww\n"
+         "MOV TEMP[3].z, IMM[1].wwww\n"
+         "TEX_LZ TEMP[10].x, TEMP[2], SAMP[0], 2D_ARRAY\n"
+         "TEX_LZ TEMP[10].y, TEMP[3], SAMP[1], 2D_ARRAY\n"
+         "TEX_LZ TEMP[10].z, TEMP[3], SAMP[2], 2D_ARRAY\n"
+
+         "MOV TEMP[12].z, IMM[1].xxxx\n"
+         "MOV TEMP[13].z, IMM[1].xxxx\n"
+         "TEX_LZ TEMP[11].x, TEMP[12], SAMP[0], 2D_ARRAY\n"
+         "TEX_LZ TEMP[11].y, TEMP[13], SAMP[1], 2D_ARRAY\n"
+         "TEX_LZ TEMP[11].z, TEMP[13], SAMP[2], 2D_ARRAY\n"
+
+         "LRP TEMP[6], TEMP[14], TEMP[10], TEMP[11]\n"
+         "MOV TEMP[6].w, IMM[1].xxxx\n"
+
+         "MOV TEMP[7].xy, TEMP[6].yzww\n"
+
+         "STORE IMAGE[0], TEMP[0], TEMP[7], 2D\n"
+      "ENDIF\n"
+
+      "END\n";
+
+static const char *compute_shader_yuv_bob_y =
+      "COMP\n"
+      "PROPERTY CS_FIXED_BLOCK_WIDTH 8\n"
+      "PROPERTY CS_FIXED_BLOCK_HEIGHT 8\n"
+      "PROPERTY CS_FIXED_BLOCK_DEPTH 1\n"
+
+      "DCL SV[0], THREAD_ID\n"
+      "DCL SV[1], BLOCK_ID\n"
+
+      "DCL CONST[0..5]\n"
+      "DCL SVIEW[0..2], RECT, FLOAT\n"
+      "DCL SAMP[0..2]\n"
+
+      "DCL IMAGE[0], 2D, WR\n"
+      "DCL TEMP[0..4]\n"
+
+      "IMM[0] UINT32 { 8, 8, 1, 0}\n"
+      "IMM[1] FLT32 { 1.0, 2.0, 0.0, 0.0}\n"
+
+      "UMAD TEMP[0], SV[1], IMM[0], SV[0]\n"
+
+      /* Drawn area check */
+      "USGE TEMP[1].xy, TEMP[0].xyxy, CONST[4].xyxy\n"
+      "USLT TEMP[1].zw, TEMP[0].xyxy, CONST[4].zwzw\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].yyyy\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].zzzz\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].wwww\n"
+
+      "UIF TEMP[1]\n"
+         /* Translate */
+         "UADD TEMP[2].xy, TEMP[0], -CONST[5].xyxy\n"
+         "U2F TEMP[2], TEMP[2]\n"
+         "DIV TEMP[3], TEMP[2], IMM[1].yyyy\n"
+
+         /* Scale */
+         "DIV TEMP[2], TEMP[2], CONST[3].zwzw\n"
+         "DIV TEMP[3], TEMP[3], CONST[3].zwzw\n"
+
+         /* Fetch texels */
+         "TEX_LZ TEMP[4].x, TEMP[2], SAMP[0], RECT\n"
+         "TEX_LZ TEMP[4].y, TEMP[3], SAMP[1], RECT\n"
+         "TEX_LZ TEMP[4].z, TEMP[3], SAMP[2], RECT\n"
+
+         "MOV TEMP[4].w, IMM[1].xxxx\n"
+
+         "STORE IMAGE[0], TEMP[0], TEMP[4], 2D\n"
+      "ENDIF\n"
+
+      "END\n";
+
+static const char *compute_shader_yuv_bob_uv =
+      "COMP\n"
+      "PROPERTY CS_FIXED_BLOCK_WIDTH 8\n"
+      "PROPERTY CS_FIXED_BLOCK_HEIGHT 8\n"
+      "PROPERTY CS_FIXED_BLOCK_DEPTH 1\n"
+
+      "DCL SV[0], THREAD_ID\n"
+      "DCL SV[1], BLOCK_ID\n"
+
+      "DCL CONST[0..5]\n"
+      "DCL SVIEW[0..2], RECT, FLOAT\n"
+      "DCL SAMP[0..2]\n"
+
+      "DCL IMAGE[0], 2D, WR\n"
+      "DCL TEMP[0..5]\n"
+
+      "IMM[0] UINT32 { 8, 8, 1, 0}\n"
+      "IMM[1] FLT32 { 1.0, 2.0, 0.0, 0.0}\n"
+
+      "UMAD TEMP[0], SV[1], IMM[0], SV[0]\n"
+
+      /* Drawn area check */
+      "USGE TEMP[1].xy, TEMP[0].xyxy, CONST[4].xyxy\n"
+      "USLT TEMP[1].zw, TEMP[0].xyxy, CONST[4].zwzw\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].yyyy\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].zzzz\n"
+      "AND TEMP[1].x, TEMP[1].xxxx, TEMP[1].wwww\n"
+
+      "UIF TEMP[1]\n"
+         /* Translate */
+         "UADD TEMP[2].xy, TEMP[0], -CONST[5].xyxy\n"
+         "U2F TEMP[2], TEMP[2]\n"
+         "DIV TEMP[3], TEMP[2], IMM[1].yyyy\n"
+
+         /* Scale */
+         "DIV TEMP[2], TEMP[2], CONST[3].zwzw\n"
+         "DIV TEMP[3], TEMP[3], CONST[3].zwzw\n"
+
+         /* Fetch texels */
+         "TEX_LZ TEMP[4].x, TEMP[2], SAMP[0], RECT\n"
+         "TEX_LZ TEMP[4].y, TEMP[3], SAMP[1], RECT\n"
+         "TEX_LZ TEMP[4].z, TEMP[3], SAMP[2], RECT\n"
+
+         "MOV TEMP[4].w, IMM[1].xxxx\n"
+
+         "MOV TEMP[5].xy, TEMP[4].yzww\n"
+
+         "STORE IMAGE[0], TEMP[0], TEMP[5], 2D\n"
+      "ENDIF\n"
+
+      "END\n";
+
 static void
 cs_launch(struct vl_compositor *c,
           void                 *cs,
@@ -467,3 +783,50 @@ vl_compositor_cs_render(struct vl_compositor_state *s,
 
    draw_layers(c, s, dirty_area);
 }
+
+bool vl_compositor_cs_init_shaders(struct vl_compositor *c)
+{
+        assert(c);
+
+        c->cs_video_buffer = vl_compositor_cs_create_shader(c, compute_shader_video_buffer);
+        if (!c->cs_video_buffer) {
+                debug_printf("Unable to create video_buffer compute shader.\n");
+                return false;
+        }
+
+        c->cs_weave_rgb = vl_compositor_cs_create_shader(c, compute_shader_weave);
+        if (!c->cs_weave_rgb) {
+                debug_printf("Unable to create weave_rgb compute shader.\n");
+                return false;
+        }
+
+        c->cs_yuv.weave.y = vl_compositor_cs_create_shader(c, compute_shader_yuv_weave_y);
+        c->cs_yuv.weave.uv = vl_compositor_cs_create_shader(c, compute_shader_yuv_weave_uv);
+        c->cs_yuv.bob.y = vl_compositor_cs_create_shader(c, compute_shader_yuv_bob_y);
+        c->cs_yuv.bob.uv = vl_compositor_cs_create_shader(c, compute_shader_yuv_bob_uv);
+        if (!c->cs_yuv.weave.y || !c->cs_yuv.weave.uv ||
+            !c->cs_yuv.bob.y || !c->cs_yuv.bob.uv) {
+                debug_printf("Unable to create YCbCr i-to-YCbCr p deint compute shader.\n");
+                return false;
+        }
+
+        return true;
+}
+
+void vl_compositor_cs_cleanup_shaders(struct vl_compositor *c)
+{
+        assert(c);
+
+        if (c->cs_video_buffer)
+                c->pipe->delete_compute_state(c->pipe, c->cs_video_buffer);
+        if (c->cs_weave_rgb)
+                c->pipe->delete_compute_state(c->pipe, c->cs_weave_rgb);
+        if (c->cs_yuv.weave.y)
+                c->pipe->delete_compute_state(c->pipe, c->cs_yuv.weave.y);
+        if (c->cs_yuv.weave.uv)
+                c->pipe->delete_compute_state(c->pipe, c->cs_yuv.weave.uv);
+        if (c->cs_yuv.bob.y)
+                c->pipe->delete_compute_state(c->pipe, c->cs_yuv.bob.y);
+        if (c->cs_yuv.bob.uv)
+                c->pipe->delete_compute_state(c->pipe, c->cs_yuv.bob.uv);
+}
index 8521add5613b9eaff3fc864e72b549b2b2cc47b4..28059b6b6e3f483ef0010d58de1d776bcf001d0c 100644 (file)
 
 #include "vl_compositor.h"
 
-extern const char *compute_shader_video_buffer;
-extern const char *compute_shader_weave;
-extern const char *compute_shader_rgba;
-
 /**
  * create compute shader
  */
@@ -53,4 +49,6 @@ vl_compositor_cs_render(struct vl_compositor_state *s,
                         struct u_rect              *dirty_area,
                         bool                        clear_dirty);
 
+bool vl_compositor_cs_init_shaders(struct vl_compositor *c);
+void vl_compositor_cs_cleanup_shaders(struct vl_compositor *c);
 #endif /* vl_compositor_cs_h */