#include <util/u_memory.h>
#include <util/u_keymap.h>
#include <util/u_sampler.h>
+#include <util/u_draw.h>
#include <tgsi/tgsi_ureg.h>
#define DEFAULT_BUF_ALIGNMENT 1
struct vertex_stream
{
- struct vertex2f pos;
+ struct vertex2s pos;
+ struct vertex2s mv[4];
struct {
- float y;
- float cr;
- float cb;
+ int8_t y;
+ int8_t cr;
+ int8_t cb;
+ int8_t flag;
} eb[2][2];
- float interlaced;
- float frame_pred;
- float ref_frames;
- float bkwd_pred;
- struct vertex2f mv[4];
};
enum VS_INPUT
{
VS_I_RECT,
VS_I_VPOS,
- VS_I_EB_0_0,
- VS_I_EB_0_1,
- VS_I_EB_1_0,
- VS_I_EB_1_1,
- VS_I_INTERLACED,
- VS_I_FRAME_PRED,
- VS_I_REF_FRAMES,
- VS_I_BKWD_PRED,
VS_I_MV0,
VS_I_MV1,
VS_I_MV2,
VS_I_MV3,
+ VS_I_EB_0_0,
+ VS_I_EB_0_1,
+ VS_I_EB_1_0,
+ VS_I_EB_1_1,
NUM_VS_INPUTS
};
VS_O_TEX2,
VS_O_EB_0,
VS_O_EB_1,
- VS_O_REF_FRAMES,
- VS_O_BKWD_PRED,
+ VS_O_INFO,
VS_O_MV0,
VS_O_MV1,
VS_O_MV2,
VS_O_MV3
};
+static const unsigned const_empty_block_mask_420[3][2][2] = {
+ { { 0x20, 0x10 }, { 0x08, 0x04 } },
+ { { 0x02, 0x02 }, { 0x02, 0x02 } },
+ { { 0x01, 0x01 }, { 0x01, 0x01 } }
+};
+
static void *
create_vert_shader(struct vl_mpeg12_mc_renderer *r)
{
struct ureg_program *shader;
- struct ureg_src scale;
+ struct ureg_src block_scale, mv_scale;
struct ureg_src vrect, vpos, eb[2][2], vmv[4];
- struct ureg_src interlaced, frame_pred, ref_frames, bkwd_pred;
struct ureg_dst t_vpos, t_vtex, t_vmv;
- struct ureg_dst o_vpos, o_line, o_vtex[3], o_eb[2], o_vmv[4];
- struct ureg_dst o_ref_frames, o_bkwd_pred;
+ struct ureg_dst o_vpos, o_line, o_vtex[3], o_eb[2], o_vmv[4], o_info;
unsigned i, label;
shader = ureg_create(TGSI_PROCESSOR_VERTEX);
eb[1][0] = ureg_DECL_vs_input(shader, VS_I_EB_1_0);
eb[0][1] = ureg_DECL_vs_input(shader, VS_I_EB_0_1);
eb[1][1] = ureg_DECL_vs_input(shader, VS_I_EB_1_1);
- interlaced = ureg_DECL_vs_input(shader, VS_I_INTERLACED);
- frame_pred = ureg_DECL_vs_input(shader, VS_I_FRAME_PRED);
- ref_frames = ureg_DECL_vs_input(shader, VS_I_REF_FRAMES);
- bkwd_pred = ureg_DECL_vs_input(shader, VS_I_BKWD_PRED);
o_vpos = ureg_DECL_output(shader, TGSI_SEMANTIC_POSITION, VS_O_VPOS);
- o_line = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_LINE);
+ o_line = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_LINE);
o_vtex[0] = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_TEX0);
o_vtex[1] = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_TEX1);
- o_vtex[2] = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_TEX2);
+ o_vtex[2] = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_TEX2);
o_eb[0] = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_EB_0);
o_eb[1] = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_EB_1);
- o_ref_frames = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_REF_FRAMES);
- o_bkwd_pred = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_BKWD_PRED);
-
+ o_info = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_INFO);
+
for (i = 0; i < 4; ++i) {
vmv[i] = ureg_DECL_vs_input(shader, VS_I_MV0 + i);
o_vmv[i] = ureg_DECL_output(shader, TGSI_SEMANTIC_GENERIC, VS_O_MV0 + i);
}
/*
- * scale = (MACROBLOCK_WIDTH, MACROBLOCK_HEIGHT) / (dst.width, dst.height)
+ * block_scale = (MACROBLOCK_WIDTH, MACROBLOCK_HEIGHT) / (dst.width, dst.height)
+ * mv_scale = 0.5 / (dst.width, dst.height);
*
- * t_vpos = (vpos + vrect) * scale
+ * t_vpos = (vpos + vrect) * block_scale
* o_vpos.xy = t_vpos
* o_vpos.zw = vpos
*
+ * o_eb[0..1] = vrect.x ? eb[0..1][1] : eb[0..1][0]
+ *
+ * o_frame_pred = frame_pred
+ * o_info.x = ref_frames
+ * o_info.y = ref_frames > 0
+ * o_info.z = bkwd_pred
+ *
+ * // Apply motion vectors
+ * o_vmv[0..count] = t_vpos + vmv[0..count] * mv_scale
+ *
* o_line.xy = vrect * 8
* o_line.z = interlaced
*
- * if(interlaced) {
+ * if(eb[0][0].w) { //interlaced
* t_vtex.x = vrect.x
* t_vtex.y = vrect.y * 0.5
* t_vtex += vpos
*
- * o_vtex[0].xy = t_vtex * scale
+ * o_vtex[0].xy = t_vtex * block_scale
*
* t_vtex.y += 0.5
- * o_vtex[1].xy = t_vtex * scale
+ * o_vtex[1].xy = t_vtex * block_scale
* } else {
* o_vtex[0..1].xy = t_vpos
* }
* o_vtex[2].xy = t_vpos
*
- * o_eb[0..1] = vrect.x ? eb[0..1][1] : eb[0..1][0]
- *
- * o_frame_pred = frame_pred
- * o_ref_frames = ref_frames
- * o_bkwd_pred = bkwd_pred
- *
- * // Apply motion vectors
- * scale = 0.5 / (dst.width, dst.height);
- * o_vmv[0..count] = t_vpos + vmv[0..count] * scale
- *
*/
- scale = ureg_imm2f(shader,
+ block_scale = ureg_imm2f(shader,
(float)MACROBLOCK_WIDTH / r->buffer_width,
(float)MACROBLOCK_HEIGHT / r->buffer_height);
+ mv_scale = ureg_imm2f(shader,
+ 0.5f / r->buffer_width,
+ 0.5f / r->buffer_height);
+
ureg_ADD(shader, ureg_writemask(t_vpos, TGSI_WRITEMASK_XY), vpos, vrect);
- ureg_MUL(shader, ureg_writemask(t_vpos, TGSI_WRITEMASK_XY), ureg_src(t_vpos), scale);
+ ureg_MUL(shader, ureg_writemask(t_vpos, TGSI_WRITEMASK_XY), ureg_src(t_vpos), block_scale);
ureg_MOV(shader, ureg_writemask(o_vpos, TGSI_WRITEMASK_XY), ureg_src(t_vpos));
ureg_MOV(shader, ureg_writemask(o_vpos, TGSI_WRITEMASK_ZW), vpos);
+ ureg_CMP(shader, ureg_writemask(o_eb[0], TGSI_WRITEMASK_XYZ),
+ ureg_negate(ureg_scalar(vrect, TGSI_SWIZZLE_X)),
+ eb[0][1], eb[0][0]);
+ ureg_CMP(shader, ureg_writemask(o_eb[1], TGSI_WRITEMASK_XYZ),
+ ureg_negate(ureg_scalar(vrect, TGSI_SWIZZLE_X)),
+ eb[1][1], eb[1][0]);
+
+ ureg_MOV(shader, ureg_writemask(o_info, TGSI_WRITEMASK_X),
+ ureg_scalar(eb[1][1], TGSI_SWIZZLE_W));
+ ureg_SGE(shader, ureg_writemask(o_info, TGSI_WRITEMASK_Y),
+ ureg_scalar(eb[1][1], TGSI_SWIZZLE_W), ureg_imm1f(shader, 0.0f));
+ ureg_MOV(shader, ureg_writemask(o_info, TGSI_WRITEMASK_Z),
+ ureg_scalar(eb[1][0], TGSI_SWIZZLE_W));
+
+ ureg_MAD(shader, ureg_writemask(o_vmv[0], TGSI_WRITEMASK_XY), mv_scale, vmv[0], ureg_src(t_vpos));
+ ureg_MAD(shader, ureg_writemask(o_vmv[2], TGSI_WRITEMASK_XY), mv_scale, vmv[2], ureg_src(t_vpos));
+
+ ureg_CMP(shader, ureg_writemask(t_vmv, TGSI_WRITEMASK_XY),
+ ureg_negate(ureg_scalar(eb[0][1], TGSI_SWIZZLE_W)),
+ vmv[0], vmv[1]);
+ ureg_MAD(shader, ureg_writemask(o_vmv[1], TGSI_WRITEMASK_XY), mv_scale, ureg_src(t_vmv), ureg_src(t_vpos));
+
+ ureg_CMP(shader, ureg_writemask(t_vmv, TGSI_WRITEMASK_XY),
+ ureg_negate(ureg_scalar(eb[0][1], TGSI_SWIZZLE_W)),
+ vmv[2], vmv[3]);
+ ureg_MAD(shader, ureg_writemask(o_vmv[3], TGSI_WRITEMASK_XY), mv_scale, ureg_src(t_vmv), ureg_src(t_vpos));
+
ureg_MOV(shader, ureg_writemask(o_vtex[0], TGSI_WRITEMASK_XY), ureg_src(t_vpos));
ureg_MOV(shader, ureg_writemask(o_vtex[1], TGSI_WRITEMASK_XY), ureg_src(t_vpos));
ureg_MOV(shader, ureg_writemask(o_vtex[2], TGSI_WRITEMASK_XY), ureg_src(t_vpos));
ureg_MOV(shader, ureg_writemask(o_line, TGSI_WRITEMASK_X), ureg_scalar(vrect, TGSI_SWIZZLE_Y));
- ureg_MUL(shader, ureg_writemask(o_line, TGSI_WRITEMASK_Y),
+ ureg_MUL(shader, ureg_writemask(o_line, TGSI_WRITEMASK_Y),
vrect, ureg_imm1f(shader, MACROBLOCK_HEIGHT / 2));
- ureg_IF(shader, ureg_scalar(interlaced, TGSI_SWIZZLE_X), &label);
+ ureg_IF(shader, ureg_scalar(eb[0][0], TGSI_SWIZZLE_W), &label);
ureg_MOV(shader, ureg_writemask(t_vtex, TGSI_WRITEMASK_X), vrect);
ureg_MUL(shader, ureg_writemask(t_vtex, TGSI_WRITEMASK_Y), vrect, ureg_imm1f(shader, 0.5f));
ureg_ADD(shader, ureg_writemask(t_vtex, TGSI_WRITEMASK_XY), vpos, ureg_src(t_vtex));
- ureg_MUL(shader, ureg_writemask(o_vtex[0], TGSI_WRITEMASK_XY), ureg_src(t_vtex), scale);
+ ureg_MUL(shader, ureg_writemask(o_vtex[0], TGSI_WRITEMASK_XY), ureg_src(t_vtex), block_scale);
ureg_ADD(shader, ureg_writemask(t_vtex, TGSI_WRITEMASK_Y), ureg_src(t_vtex), ureg_imm1f(shader, 0.5f));
- ureg_MUL(shader, ureg_writemask(o_vtex[1], TGSI_WRITEMASK_XY), ureg_src(t_vtex), scale);
+ ureg_MUL(shader, ureg_writemask(o_vtex[1], TGSI_WRITEMASK_XY), ureg_src(t_vtex), block_scale);
ureg_MUL(shader, ureg_writemask(o_line, TGSI_WRITEMASK_X),
ureg_scalar(vrect, TGSI_SWIZZLE_Y),
ureg_imm1f(shader, MACROBLOCK_HEIGHT / 2));
+ ureg_fixup_label(shader, label, ureg_get_instruction_number(shader));
ureg_ENDIF(shader);
- ureg_CMP(shader, ureg_writemask(o_eb[0], TGSI_WRITEMASK_XYZ),
- ureg_negate(ureg_scalar(vrect, TGSI_SWIZZLE_X)),
- eb[0][1], eb[0][0]);
- ureg_CMP(shader, ureg_writemask(o_eb[1], TGSI_WRITEMASK_XYZ),
- ureg_negate(ureg_scalar(vrect, TGSI_SWIZZLE_X)),
- eb[1][1], eb[1][0]);
-
- ureg_MOV(shader, ureg_writemask(o_ref_frames, TGSI_WRITEMASK_X), ref_frames);
- ureg_MOV(shader, ureg_writemask(o_bkwd_pred, TGSI_WRITEMASK_X), bkwd_pred);
-
- scale = ureg_imm2f(shader,
- 0.5f / r->buffer_width,
- 0.5f / r->buffer_height);
-
- ureg_MAD(shader, ureg_writemask(o_vmv[0], TGSI_WRITEMASK_XY), scale, vmv[0], ureg_src(t_vpos));
- ureg_MAD(shader, ureg_writemask(o_vmv[2], TGSI_WRITEMASK_XY), scale, vmv[2], ureg_src(t_vpos));
-
- ureg_CMP(shader, ureg_writemask(t_vmv, TGSI_WRITEMASK_XY),
- ureg_negate(ureg_scalar(frame_pred, TGSI_SWIZZLE_X)),
- vmv[0], vmv[1]);
- ureg_MAD(shader, ureg_writemask(o_vmv[1], TGSI_WRITEMASK_XY), scale, ureg_src(t_vmv), ureg_src(t_vpos));
-
- ureg_CMP(shader, ureg_writemask(t_vmv, TGSI_WRITEMASK_XY),
- ureg_negate(ureg_scalar(frame_pred, TGSI_SWIZZLE_X)),
- vmv[2], vmv[3]);
- ureg_MAD(shader, ureg_writemask(o_vmv[3], TGSI_WRITEMASK_XY), scale, ureg_src(t_vmv), ureg_src(t_vpos));
-
ureg_release_temporary(shader, t_vtex);
ureg_release_temporary(shader, t_vpos);
ureg_release_temporary(shader, t_vmv);
fetch_ycbcr(struct vl_mpeg12_mc_renderer *r, struct ureg_program *shader, struct ureg_dst field)
{
struct ureg_src tc[3], sampler[3], eb[2];
- struct ureg_dst texel, t_tc, t_eb_info, tmp;
+ struct ureg_dst texel, t_tc, t_eb_info;
unsigned i, label;
texel = ureg_DECL_temporary(shader);
t_tc = ureg_DECL_temporary(shader);
t_eb_info = ureg_DECL_temporary(shader);
- tmp = ureg_DECL_temporary(shader);
tc[0] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_TEX0, TGSI_INTERPOLATE_LINEAR);
tc[1] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_TEX1, TGSI_INTERPOLATE_LINEAR);
ureg_TEX(shader, ureg_writemask(texel, TGSI_WRITEMASK_X << i), TGSI_TEXTURE_3D, tc[2], sampler[i]);
}
+ ureg_fixup_label(shader, label, ureg_get_instruction_number(shader));
ureg_ENDIF(shader);
}
ureg_release_temporary(shader, t_tc);
ureg_release_temporary(shader, t_eb_info);
- ureg_release_temporary(shader, tmp);
return texel;
}
static struct ureg_dst
fetch_ref(struct ureg_program *shader, struct ureg_dst field)
{
- struct ureg_src ref_frames, bkwd_pred;
+ struct ureg_src info;
struct ureg_src tc[4], sampler[2];
- struct ureg_dst ref[2], tmp, result;
+ struct ureg_dst ref[2], result;
unsigned i, intra_label, bi_label, label;
- ref_frames = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_REF_FRAMES, TGSI_INTERPOLATE_CONSTANT);
- bkwd_pred = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_BKWD_PRED, TGSI_INTERPOLATE_CONSTANT);
+ info = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_INFO, TGSI_INTERPOLATE_CONSTANT);
for (i = 0; i < 4; ++i)
tc[i] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_MV0 + i, TGSI_INTERPOLATE_LINEAR);
ref[i] = ureg_DECL_temporary(shader);
}
- tmp = ureg_DECL_temporary(shader);
result = ureg_DECL_temporary(shader);
- ureg_MOV(shader, result, ureg_imm1f(shader, 0.5f));
+ ureg_MOV(shader, ureg_writemask(result, TGSI_WRITEMASK_XYZ), ureg_imm1f(shader, 0.5f));
- ureg_SGE(shader, ureg_writemask(tmp, TGSI_WRITEMASK_X), ref_frames, ureg_imm1f(shader, 0.0f));
- ureg_IF(shader, ureg_scalar(ureg_src(tmp), TGSI_SWIZZLE_X), &intra_label);
- ureg_CMP(shader, ureg_writemask(tmp, TGSI_WRITEMASK_XY),
+ ureg_IF(shader, ureg_scalar(info, TGSI_SWIZZLE_Y), &intra_label);
+ ureg_CMP(shader, ureg_writemask(ref[0], TGSI_WRITEMASK_XY),
ureg_negate(ureg_scalar(ureg_src(field), TGSI_SWIZZLE_Y)),
tc[1], tc[0]);
- ureg_IF(shader, ureg_scalar(ref_frames, TGSI_SWIZZLE_X), &bi_label);
+ ureg_IF(shader, ureg_scalar(info, TGSI_SWIZZLE_X), &bi_label);
/*
* result = tex(field.z ? tc[1] : tc[0], sampler[bkwd_pred ? 1 : 0])
*/
- ureg_IF(shader, bkwd_pred, &label);
- ureg_TEX(shader, result, TGSI_TEXTURE_2D, ureg_src(tmp), sampler[1]);
+ ureg_IF(shader, ureg_scalar(info, TGSI_SWIZZLE_Z), &label);
+ ureg_TEX(shader, result, TGSI_TEXTURE_2D, ureg_src(ref[0]), sampler[1]);
+ ureg_fixup_label(shader, label, ureg_get_instruction_number(shader));
ureg_ELSE(shader, &label);
- ureg_TEX(shader, result, TGSI_TEXTURE_2D, ureg_src(tmp), sampler[0]);
+ ureg_TEX(shader, result, TGSI_TEXTURE_2D, ureg_src(ref[0]), sampler[0]);
+ ureg_fixup_label(shader, label, ureg_get_instruction_number(shader));
ureg_ENDIF(shader);
+ ureg_fixup_label(shader, bi_label, ureg_get_instruction_number(shader));
ureg_ELSE(shader, &bi_label);
/*
* else
* ref[0..1] = tex(tc[2..3], sampler[0..1])
*/
- ureg_TEX(shader, ref[0], TGSI_TEXTURE_2D, ureg_src(tmp), sampler[0]);
-
- ureg_CMP(shader, ureg_writemask(tmp, TGSI_WRITEMASK_XY),
+ ureg_CMP(shader, ureg_writemask(ref[1], TGSI_WRITEMASK_XY),
ureg_negate(ureg_scalar(ureg_src(field), TGSI_SWIZZLE_Y)),
tc[3], tc[2]);
- ureg_TEX(shader, ref[1], TGSI_TEXTURE_2D, ureg_src(tmp), sampler[1]);
+ ureg_TEX(shader, ref[0], TGSI_TEXTURE_2D, ureg_src(ref[0]), sampler[0]);
+ ureg_TEX(shader, ref[1], TGSI_TEXTURE_2D, ureg_src(ref[1]), sampler[1]);
- ureg_LRP(shader, result, ureg_scalar(ureg_imm1f(shader, 0.5f), TGSI_SWIZZLE_X),
+ ureg_LRP(shader, ureg_writemask(result, TGSI_WRITEMASK_XYZ), ureg_imm1f(shader, 0.5f),
ureg_src(ref[0]), ureg_src(ref[1]));
+ ureg_fixup_label(shader, bi_label, ureg_get_instruction_number(shader));
ureg_ENDIF(shader);
+ ureg_fixup_label(shader, intra_label, ureg_get_instruction_number(shader));
ureg_ENDIF(shader);
for (i = 0; i < 2; ++i)
ureg_release_temporary(shader, ref[i]);
- ureg_release_temporary(shader, tmp);
return result;
}
result = fetch_ref(shader, field);
- ureg_ADD(shader, fragment, ureg_src(texel), ureg_src(result));
+ ureg_ADD(shader, ureg_writemask(fragment, TGSI_WRITEMASK_XYZ), ureg_src(texel), ureg_src(result));
ureg_release_temporary(shader, field);
ureg_release_temporary(shader, texel);
r->quad = vl_vb_upload_quads(r->pipe, r->macroblocks_per_batch);
/* Position element */
- vertex_elems[VS_I_VPOS].src_format = PIPE_FORMAT_R32G32_FLOAT;
+ vertex_elems[VS_I_VPOS].src_format = PIPE_FORMAT_R16G16_SSCALED;
+
+ for (i = 0; i < 4; ++i)
+ /* motion vector 0..4 element */
+ vertex_elems[VS_I_MV0 + i].src_format = PIPE_FORMAT_R16G16_SSCALED;
/* y, cr, cb empty block element top left block */
- vertex_elems[VS_I_EB_0_0].src_format = PIPE_FORMAT_R32G32B32_FLOAT;
+ vertex_elems[VS_I_EB_0_0].src_format = PIPE_FORMAT_R8G8B8A8_SSCALED;
/* y, cr, cb empty block element top right block */
- vertex_elems[VS_I_EB_0_1].src_format = PIPE_FORMAT_R32G32B32_FLOAT;
+ vertex_elems[VS_I_EB_0_1].src_format = PIPE_FORMAT_R8G8B8A8_SSCALED;
/* y, cr, cb empty block element bottom left block */
- vertex_elems[VS_I_EB_1_0].src_format = PIPE_FORMAT_R32G32B32_FLOAT;
+ vertex_elems[VS_I_EB_1_0].src_format = PIPE_FORMAT_R8G8B8A8_SSCALED;
/* y, cr, cb empty block element bottom right block */
- vertex_elems[VS_I_EB_1_1].src_format = PIPE_FORMAT_R32G32B32_FLOAT;
-
- /* progressive=0.0f interlaced=1.0f */
- vertex_elems[VS_I_INTERLACED].src_format = PIPE_FORMAT_R32_FLOAT;
+ vertex_elems[VS_I_EB_1_1].src_format = PIPE_FORMAT_R8G8B8A8_SSCALED;
- /* frame=0.0f field=1.0f */
- vertex_elems[VS_I_FRAME_PRED].src_format = PIPE_FORMAT_R32_FLOAT;
-
- /* intra=-1.0f forward/backward=1.0f bi=0.0f */
- vertex_elems[VS_I_REF_FRAMES].src_format = PIPE_FORMAT_R32_FLOAT;
-
- /* forward=0.0f backward=1.0f */
- vertex_elems[VS_I_BKWD_PRED].src_format = PIPE_FORMAT_R32_FLOAT;
-
- for (i = 0; i < 4; ++i)
- /* motion vector 0..4 element */
- vertex_elems[VS_I_MV0 + i].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
- r->vertex_stream_stride = vl_vb_element_helper(&vertex_elems[VS_I_VPOS], 13, 1);
+ r->vertex_stream_stride = vl_vb_element_helper(&vertex_elems[VS_I_VPOS], 9, 1);
r->vertex_elems_state = r->pipe->create_vertex_elements_state(
r->pipe, NUM_VS_INPUTS, vertex_elems);
}
static void
-get_motion_vectors(struct pipe_mpeg12_macroblock *mb, struct vertex2f mv[4])
+get_motion_vectors(struct pipe_mpeg12_macroblock *mb, struct vertex2s mv[4])
{
switch (mb->mb_type) {
case PIPE_MPEG12_MACROBLOCK_TYPE_BI:
{
if (mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME) {
-
mv[2].x = mb->pmv[0][1][0];
mv[2].y = mb->pmv[0][1][1];
}
}
-static bool
-empty_block(enum pipe_video_chroma_format chroma_format,
- unsigned cbp, unsigned component,
- unsigned x, unsigned y)
-{
- /* TODO: Implement 422, 444 */
- assert(chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420);
-
- if(component == 0) /*luma*/
- return !(cbp & (1 << (5 - (x + y * 2))));
- else /*cr cb*/
- return !(cbp & (1 << (2 - component)));
-}
-
static void
grab_vectors(struct vl_mpeg12_mc_renderer *r,
struct vl_mpeg12_mc_buffer *buffer,
stream.pos.y = mb->mby;
for ( i = 0; i < 2; ++i) {
for ( j = 0; j < 2; ++j) {
- stream.eb[i][j].y = empty_block(r->chroma_format, mb->cbp, 0, j, i);
- stream.eb[i][j].cr = empty_block(r->chroma_format, mb->cbp, 1, j, i);
- stream.eb[i][j].cb = empty_block(r->chroma_format, mb->cbp, 2, j, i);
+ stream.eb[i][j].y = !(mb->cbp & (*r->empty_block_mask)[0][i][j]);
+ stream.eb[i][j].cr = !(mb->cbp & (*r->empty_block_mask)[1][i][j]);
+ stream.eb[i][j].cb = !(mb->cbp & (*r->empty_block_mask)[2][i][j]);
}
}
- stream.interlaced = mb->dct_type == PIPE_MPEG12_DCT_TYPE_FIELD ? 1.0f : 0.0f;
- stream.frame_pred = mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME ? 1.0f : 0.0f;
- stream.bkwd_pred = mb->mb_type == PIPE_MPEG12_MACROBLOCK_TYPE_BKWD ? 1.0f : 0.0f;
+ stream.eb[0][0].flag = mb->dct_type == PIPE_MPEG12_DCT_TYPE_FIELD;
+ stream.eb[0][1].flag = mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME;
+ stream.eb[1][0].flag = mb->mb_type == PIPE_MPEG12_MACROBLOCK_TYPE_BKWD;
switch (mb->mb_type) {
case PIPE_MPEG12_MACROBLOCK_TYPE_INTRA:
- stream.ref_frames = -1.0f;
+ stream.eb[1][1].flag = -1;
break;
case PIPE_MPEG12_MACROBLOCK_TYPE_FWD:
case PIPE_MPEG12_MACROBLOCK_TYPE_BKWD:
- stream.ref_frames = 1.0f;
+ stream.eb[1][1].flag = 1;
break;
-
+
case PIPE_MPEG12_MACROBLOCK_TYPE_BI:
- stream.ref_frames = 0.0f;
+ stream.eb[1][1].flag = 0;
break;
default:
}
get_motion_vectors(mb, stream.mv);
- vl_vb_add_block(&buffer->vertex_stream, (float*)&stream);
+ vl_vb_add_block(&buffer->vertex_stream, &stream);
}
static void
for (y = 0; y < 2; ++y) {
for (x = 0; x < 2; ++x, ++tb) {
- if (!empty_block(r->chroma_format, cbp, 0, x, y)) {
+ if (cbp & (*r->empty_block_mask)[0][y][x]) {
vl_idct_add_block(&buffer->idct_y, mbx * 2 + x, mby * 2 + y, blocks);
blocks += BLOCK_WIDTH * BLOCK_HEIGHT;
}
assert(r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420);
for (tb = 1; tb < 3; ++tb) {
- if (!empty_block(r->chroma_format, cbp, tb, 0, 0)) {
+ if (cbp & (*r->empty_block_mask)[tb][0][0]) {
if(tb == 1)
vl_idct_add_block(&buffer->idct_cb, mbx, mby, blocks);
else
renderer->chroma_format = chroma_format;
renderer->bufmode = bufmode;
+ /* TODO: Implement 422, 444 */
+ assert(chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420);
+ renderer->empty_block_mask = &const_empty_block_mask_420;
+
renderer->texview_map = util_new_keymap(sizeof(struct pipe_surface*), -1,
texview_map_delete);
if (!renderer->texview_map)
template.width0 = renderer->buffer_width;
template.height0 = renderer->buffer_height;
template.depth0 = 1;
- template.usage = PIPE_USAGE_DYNAMIC;
+ template.usage = PIPE_USAGE_STATIC;
template.bind = PIPE_BIND_SAMPLER_VIEW;
template.flags = 0;
pipe_resource_reference(&buffer->vertex_bufs.individual.quad.buffer, renderer->quad.buffer);
buffer->vertex_bufs.individual.stream = vl_vb_init(
- &buffer->vertex_stream, renderer->pipe, renderer->macroblocks_per_batch,
- sizeof(struct vertex_stream) / sizeof(float),
+ &buffer->vertex_stream, renderer->pipe, renderer->macroblocks_per_batch,
renderer->vertex_stream_stride);
return true;
projects / mesa.git / blobdiff