static void
fetch_four(struct ureg_program *shader, struct ureg_dst m[2], struct ureg_src addr[2],
struct ureg_src sampler, bool resource3d)
{
ureg_TEX(shader, m[0], resource3d ? TGSI_TEXTURE_3D : TGSI_TEXTURE_2D, addr[0], sampler);
ureg_TEX(shader, m[1], resource3d ? TGSI_TEXTURE_3D : TGSI_TEXTURE_2D, addr[1], sampler);
}
static void *
fd_prog_blit(struct pipe_context *pctx, int rts, bool depth)
{
int i;
struct ureg_src tc;
struct ureg_program *ureg;
debug_assert(rts <= MAX_RENDER_TARGETS);
ureg = ureg_create(PIPE_SHADER_FRAGMENT);
if (!ureg)
return NULL;
tc = ureg_DECL_fs_input(
ureg, TGSI_SEMANTIC_GENERIC, 0, TGSI_INTERPOLATE_PERSPECTIVE);
for (i = 0; i < rts; i++)
ureg_TEX(ureg, ureg_DECL_output(ureg, TGSI_SEMANTIC_COLOR, i),
TGSI_TEXTURE_2D, tc, ureg_DECL_sampler(ureg, i));
if (depth)
ureg_TEX(ureg,
ureg_writemask(
ureg_DECL_output(ureg, TGSI_SEMANTIC_POSITION, 0),
TGSI_WRITEMASK_Z),
TGSI_TEXTURE_2D, tc, ureg_DECL_sampler(ureg, rts));
ureg_END(ureg);
return ureg_create_shader_and_destroy(ureg, pctx);
}
static INLINE void
xrender_tex(struct ureg_program *ureg,
struct ureg_dst dst,
struct ureg_src coords,
struct ureg_src sampler,
struct ureg_src imm0,
boolean repeat_none, boolean swizzle, boolean set_alpha)
{
if (repeat_none) {
struct ureg_dst tmp0 = ureg_DECL_temporary(ureg);
struct ureg_dst tmp1 = ureg_DECL_temporary(ureg);
ureg_SGT(ureg, tmp1, ureg_swizzle(coords,
TGSI_SWIZZLE_X,
TGSI_SWIZZLE_Y,
TGSI_SWIZZLE_X,
TGSI_SWIZZLE_Y), ureg_scalar(imm0,
TGSI_SWIZZLE_X));
ureg_SLT(ureg, tmp0,
ureg_swizzle(coords, TGSI_SWIZZLE_X, TGSI_SWIZZLE_Y,
TGSI_SWIZZLE_X, TGSI_SWIZZLE_Y), ureg_scalar(imm0,
TGSI_SWIZZLE_W));
ureg_MIN(ureg, tmp0, ureg_src(tmp0), ureg_src(tmp1));
ureg_MIN(ureg, tmp0, ureg_scalar(ureg_src(tmp0), TGSI_SWIZZLE_X),
ureg_scalar(ureg_src(tmp0), TGSI_SWIZZLE_Y));
ureg_TEX(ureg, tmp1, TGSI_TEXTURE_2D, coords, sampler);
if (swizzle)
ureg_MOV(ureg, tmp1, ureg_swizzle(ureg_src(tmp1),
TGSI_SWIZZLE_Z,
TGSI_SWIZZLE_Y, TGSI_SWIZZLE_X,
TGSI_SWIZZLE_W));
if (set_alpha)
ureg_MOV(ureg,
ureg_writemask(tmp1, TGSI_WRITEMASK_W),
ureg_scalar(imm0, TGSI_SWIZZLE_W));
ureg_MUL(ureg, dst, ureg_src(tmp1), ureg_src(tmp0));
ureg_release_temporary(ureg, tmp0);
ureg_release_temporary(ureg, tmp1);
} else {
if (swizzle) {
struct ureg_dst tmp = ureg_DECL_temporary(ureg);
ureg_TEX(ureg, tmp, TGSI_TEXTURE_2D, coords, sampler);
ureg_MOV(ureg, dst, ureg_swizzle(ureg_src(tmp),
TGSI_SWIZZLE_Z,
TGSI_SWIZZLE_Y, TGSI_SWIZZLE_X,
TGSI_SWIZZLE_W));
ureg_release_temporary(ureg, tmp);
} else {
ureg_TEX(ureg, dst, TGSI_TEXTURE_2D, coords, sampler);
}
if (set_alpha)
ureg_MOV(ureg,
ureg_writemask(dst, TGSI_WRITEMASK_W),
ureg_scalar(imm0, TGSI_SWIZZLE_W));
}
}
static void *
create_yuv_shader(struct pipe_context *pipe, struct ureg_program *ureg)
{
struct ureg_src y_sampler, u_sampler, v_sampler;
struct ureg_src pos;
struct ureg_src matrow0, matrow1, matrow2;
struct ureg_dst y, u, v, rgb;
struct ureg_dst out = ureg_DECL_output(ureg,
TGSI_SEMANTIC_COLOR,
0);
pos = ureg_DECL_fs_input(ureg,
TGSI_SEMANTIC_GENERIC, 0,
TGSI_INTERPOLATE_PERSPECTIVE);
rgb = ureg_DECL_temporary(ureg);
y = ureg_DECL_temporary(ureg);
u = ureg_DECL_temporary(ureg);
v = ureg_DECL_temporary(ureg);
y_sampler = ureg_DECL_sampler(ureg, 0);
u_sampler = ureg_DECL_sampler(ureg, 1);
v_sampler = ureg_DECL_sampler(ureg, 2);
matrow0 = ureg_DECL_constant(ureg, 0);
matrow1 = ureg_DECL_constant(ureg, 1);
matrow2 = ureg_DECL_constant(ureg, 2);
ureg_TEX(ureg, y, TGSI_TEXTURE_2D, pos, y_sampler);
ureg_TEX(ureg, u, TGSI_TEXTURE_2D, pos, u_sampler);
ureg_TEX(ureg, v, TGSI_TEXTURE_2D, pos, v_sampler);
ureg_SUB(ureg, u, ureg_src(u), ureg_scalar(matrow0, TGSI_SWIZZLE_W));
ureg_SUB(ureg, v, ureg_src(v), ureg_scalar(matrow0, TGSI_SWIZZLE_W));
ureg_MUL(ureg, rgb, ureg_scalar(ureg_src(y), TGSI_SWIZZLE_X), matrow0);
ureg_MAD(ureg, rgb,
ureg_scalar(ureg_src(u), TGSI_SWIZZLE_X), matrow1, ureg_src(rgb));
ureg_MAD(ureg, rgb,
ureg_scalar(ureg_src(v), TGSI_SWIZZLE_X), matrow2, ureg_src(rgb));
/* rgb.a = 1; */
ureg_MOV(ureg, ureg_writemask(rgb, TGSI_WRITEMASK_W),
ureg_scalar(matrow0, TGSI_SWIZZLE_X));
ureg_MOV(ureg, out, ureg_src(rgb));
ureg_release_temporary(ureg, rgb);
ureg_release_temporary(ureg, y);
ureg_release_temporary(ureg, u);
ureg_release_temporary(ureg, v);
ureg_END(ureg);
return ureg_create_shader_and_destroy(ureg, pipe);
}
static void *
create_frag_shader(struct vl_matrix_filter *filter, unsigned num_offsets,
struct vertex2f *offsets, const float *matrix_values)
{
struct ureg_program *shader;
struct ureg_src i_vtex;
struct ureg_src sampler;
struct ureg_dst tmp;
struct ureg_dst t_sum;
struct ureg_dst o_fragment;
unsigned i;
shader = ureg_create(PIPE_SHADER_FRAGMENT);
if (!shader) {
return NULL;
}
i_vtex = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX, TGSI_INTERPOLATE_LINEAR);
sampler = ureg_DECL_sampler(shader, 0);
ureg_DECL_sampler_view(shader, 0, TGSI_TEXTURE_2D,
TGSI_RETURN_TYPE_FLOAT,
TGSI_RETURN_TYPE_FLOAT,
TGSI_RETURN_TYPE_FLOAT,
TGSI_RETURN_TYPE_FLOAT);
tmp = ureg_DECL_temporary(shader);
t_sum = ureg_DECL_temporary(shader);
o_fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
ureg_MOV(shader, t_sum, ureg_imm1f(shader, 0.0f));
for (i = 0; i < num_offsets; ++i) {
if (matrix_values[i] == 0.0f)
continue;
if (!is_vec_zero(offsets[i])) {
ureg_ADD(shader, ureg_writemask(tmp, TGSI_WRITEMASK_XY),
i_vtex, ureg_imm2f(shader, offsets[i].x, offsets[i].y));
ureg_MOV(shader, ureg_writemask(tmp, TGSI_WRITEMASK_ZW),
ureg_imm1f(shader, 0.0f));
ureg_TEX(shader, tmp, TGSI_TEXTURE_2D, ureg_src(tmp), sampler);
} else {
ureg_TEX(shader, tmp, TGSI_TEXTURE_2D, i_vtex, sampler);
}
ureg_MAD(shader, t_sum, ureg_src(tmp), ureg_imm1f(shader, matrix_values[i]),
ureg_src(t_sum));
}
ureg_MOV(shader, o_fragment, ureg_src(t_sum));
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, filter->pipe);
}
/**
* Make a simple fragment texture shader which reads the texture unit 0 and 1
* and writes it as depth and stencil, respectively.
*/
void *
util_make_fragment_tex_shader_writedepthstencil(struct pipe_context *pipe,
unsigned tex_target,
unsigned interp_mode)
{
struct ureg_program *ureg;
struct ureg_src depth_sampler, stencil_sampler;
struct ureg_src tex;
struct ureg_dst out, depth, stencil;
struct ureg_src imm;
ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
if (ureg == NULL)
return NULL;
depth_sampler = ureg_DECL_sampler( ureg, 0 );
stencil_sampler = ureg_DECL_sampler( ureg, 1 );
tex = ureg_DECL_fs_input( ureg,
TGSI_SEMANTIC_GENERIC, 0,
interp_mode );
out = ureg_DECL_output( ureg,
TGSI_SEMANTIC_COLOR,
0 );
depth = ureg_DECL_output( ureg,
TGSI_SEMANTIC_POSITION,
0 );
stencil = ureg_DECL_output( ureg,
TGSI_SEMANTIC_STENCIL,
0 );
imm = ureg_imm4f( ureg, 0, 0, 0, 1 );
ureg_MOV( ureg, out, imm );
ureg_TEX( ureg,
ureg_writemask(depth, TGSI_WRITEMASK_Z),
tex_target, tex, depth_sampler );
ureg_TEX( ureg,
ureg_writemask(stencil, TGSI_WRITEMASK_Y),
tex_target, tex, stencil_sampler );
ureg_END( ureg );
return ureg_create_shader_and_destroy( ureg, pipe );
}
static void *
create_frag_shader_palette(struct vl_compositor *c, bool include_cc)
{
struct ureg_program *shader;
struct ureg_src csc[3];
struct ureg_src tc;
struct ureg_src sampler;
struct ureg_src palette;
struct ureg_dst texel;
struct ureg_dst fragment;
unsigned i;
shader = ureg_create(TGSI_PROCESSOR_FRAGMENT);
if (!shader)
return false;
for (i = 0; include_cc && i < 3; ++i)
csc[i] = ureg_DECL_constant(shader, i);
tc = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX, TGSI_INTERPOLATE_LINEAR);
sampler = ureg_DECL_sampler(shader, 0);
palette = ureg_DECL_sampler(shader, 1);
texel = ureg_DECL_temporary(shader);
fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
/*
* texel = tex(tc, sampler)
* fragment.xyz = tex(texel, palette) * csc
* fragment.a = texel.a
*/
ureg_TEX(shader, texel, TGSI_TEXTURE_2D, tc, sampler);
ureg_MOV(shader, ureg_writemask(fragment, TGSI_WRITEMASK_W), ureg_src(texel));
if (include_cc) {
ureg_TEX(shader, texel, TGSI_TEXTURE_1D, ureg_src(texel), palette);
for (i = 0; i < 3; ++i)
ureg_DP4(shader, ureg_writemask(fragment, TGSI_WRITEMASK_X << i), csc[i], ureg_src(texel));
} else {
ureg_TEX(shader, ureg_writemask(fragment, TGSI_WRITEMASK_XYZ),
TGSI_TEXTURE_1D, ureg_src(texel), palette);
}
ureg_release_temporary(shader, texel);
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, c->pipe);
}
static void *
create_copy_frag_shader(struct vl_deint_filter *filter, unsigned field)
{
struct ureg_program *shader;
struct ureg_src i_vtex;
struct ureg_src sampler;
struct ureg_dst o_fragment;
struct ureg_dst t_tex;
shader = ureg_create(PIPE_SHADER_FRAGMENT);
if (!shader) {
return NULL;
}
t_tex = ureg_DECL_temporary(shader);
i_vtex = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX, TGSI_INTERPOLATE_LINEAR);
sampler = ureg_DECL_sampler(shader, 2);
o_fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
ureg_MOV(shader, t_tex, i_vtex);
if (field) {
ureg_MOV(shader, ureg_writemask(t_tex, TGSI_WRITEMASK_ZW),
ureg_imm4f(shader, 0, 0, 1.0f, 0));
} else {
ureg_MOV(shader, ureg_writemask(t_tex, TGSI_WRITEMASK_ZW),
ureg_imm1f(shader, 0));
}
ureg_TEX(shader, o_fragment, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_tex), sampler);
ureg_release_temporary(shader, t_tex);
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, filter->pipe);
}
static void *
create_frag_shader_rgba(struct vl_compositor *c)
{
struct ureg_program *shader;
struct ureg_src tc, color, sampler;
struct ureg_dst texel, fragment;
shader = ureg_create(TGSI_PROCESSOR_FRAGMENT);
if (!shader)
return false;
tc = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX, TGSI_INTERPOLATE_LINEAR);
color = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_COLOR, VS_O_COLOR, TGSI_INTERPOLATE_LINEAR);
sampler = ureg_DECL_sampler(shader, 0);
texel = ureg_DECL_temporary(shader);
fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
/*
* fragment = tex(tc, sampler)
*/
ureg_TEX(shader, texel, TGSI_TEXTURE_2D, tc, sampler);
ureg_MUL(shader, fragment, ureg_src(texel), color);
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, c->pipe);
}
static void
linear_gradient(struct ureg_program *ureg,
struct ureg_dst out,
struct ureg_src pos,
struct ureg_src sampler,
struct ureg_src coords,
struct ureg_src const0124,
struct ureg_src matrow0,
struct ureg_src matrow1, struct ureg_src matrow2)
{
struct ureg_dst temp0 = ureg_DECL_temporary(ureg);
struct ureg_dst temp1 = ureg_DECL_temporary(ureg);
struct ureg_dst temp2 = ureg_DECL_temporary(ureg);
struct ureg_dst temp3 = ureg_DECL_temporary(ureg);
struct ureg_dst temp4 = ureg_DECL_temporary(ureg);
struct ureg_dst temp5 = ureg_DECL_temporary(ureg);
ureg_MOV(ureg, ureg_writemask(temp0, TGSI_WRITEMASK_XY), pos);
ureg_MOV(ureg,
ureg_writemask(temp0, TGSI_WRITEMASK_Z),
ureg_scalar(const0124, TGSI_SWIZZLE_Y));
ureg_DP3(ureg, temp1, matrow0, ureg_src(temp0));
ureg_DP3(ureg, temp2, matrow1, ureg_src(temp0));
ureg_DP3(ureg, temp3, matrow2, ureg_src(temp0));
ureg_RCP(ureg, temp3, ureg_src(temp3));
ureg_MUL(ureg, temp1, ureg_src(temp1), ureg_src(temp3));
ureg_MUL(ureg, temp2, ureg_src(temp2), ureg_src(temp3));
ureg_MOV(ureg, ureg_writemask(temp4, TGSI_WRITEMASK_X), ureg_src(temp1));
ureg_MOV(ureg, ureg_writemask(temp4, TGSI_WRITEMASK_Y), ureg_src(temp2));
ureg_MUL(ureg, temp0,
ureg_scalar(coords, TGSI_SWIZZLE_Y),
ureg_scalar(ureg_src(temp4), TGSI_SWIZZLE_Y));
ureg_MAD(ureg, temp1,
ureg_scalar(coords, TGSI_SWIZZLE_X),
ureg_scalar(ureg_src(temp4), TGSI_SWIZZLE_X), ureg_src(temp0));
ureg_MUL(ureg, temp2, ureg_src(temp1), ureg_scalar(coords, TGSI_SWIZZLE_Z));
ureg_TEX(ureg, out, TGSI_TEXTURE_1D, ureg_src(temp2), sampler);
ureg_release_temporary(ureg, temp0);
ureg_release_temporary(ureg, temp1);
ureg_release_temporary(ureg, temp2);
ureg_release_temporary(ureg, temp3);
ureg_release_temporary(ureg, temp4);
ureg_release_temporary(ureg, temp5);
}
static void *
create_frag_shader_video_buffer(struct vl_compositor *c)
{
struct ureg_program *shader;
struct ureg_src tc;
struct ureg_src csc[3];
struct ureg_src sampler[3];
struct ureg_dst texel;
struct ureg_dst fragment;
unsigned i;
shader = ureg_create(TGSI_PROCESSOR_FRAGMENT);
if (!shader)
return false;
tc = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX, TGSI_INTERPOLATE_LINEAR);
for (i = 0; i < 3; ++i) {
csc[i] = ureg_DECL_constant(shader, i);
sampler[i] = ureg_DECL_sampler(shader, i);
}
texel = ureg_DECL_temporary(shader);
fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
/*
* texel.xyz = tex(tc, sampler[i])
* fragment = csc * texel
*/
for (i = 0; i < 3; ++i)
ureg_TEX(shader, ureg_writemask(texel, TGSI_WRITEMASK_X << i), TGSI_TEXTURE_3D, tc, sampler[i]);
ureg_MOV(shader, ureg_writemask(texel, TGSI_WRITEMASK_W), ureg_imm1f(shader, 1.0f));
for (i = 0; i < 3; ++i)
ureg_DP4(shader, ureg_writemask(fragment, TGSI_WRITEMASK_X << i), csc[i], ureg_src(texel));
ureg_MOV(shader, ureg_writemask(fragment, TGSI_WRITEMASK_W), ureg_imm1f(shader, 1.0f));
ureg_release_temporary(shader, texel);
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, c->pipe);
}
/**
* Make simple fragment texture shader:
* IMM {0,0,0,1} // (if writemask != 0xf)
* MOV OUT[0], IMM[0] // (if writemask != 0xf)
* TEX OUT[0].writemask, IN[0], SAMP[0], 2D;
* END;
*
* \param tex_target one of PIPE_TEXTURE_x
* \parma interp_mode either TGSI_INTERPOLATE_LINEAR or PERSPECTIVE
* \param writemask mask of TGSI_WRITEMASK_x
*/
void *
util_make_fragment_tex_shader_writemask(struct pipe_context *pipe,
unsigned tex_target,
unsigned interp_mode,
unsigned writemask )
{
struct ureg_program *ureg;
struct ureg_src sampler;
struct ureg_src tex;
struct ureg_dst out;
assert(interp_mode == TGSI_INTERPOLATE_LINEAR ||
interp_mode == TGSI_INTERPOLATE_PERSPECTIVE);
ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
if (ureg == NULL)
return NULL;
sampler = ureg_DECL_sampler( ureg, 0 );
tex = ureg_DECL_fs_input( ureg,
TGSI_SEMANTIC_GENERIC, 0,
interp_mode );
out = ureg_DECL_output( ureg,
TGSI_SEMANTIC_COLOR,
0 );
if (writemask != TGSI_WRITEMASK_XYZW) {
struct ureg_src imm = ureg_imm4f( ureg, 0, 0, 0, 1 );
ureg_MOV( ureg, out, imm );
}
ureg_TEX( ureg,
ureg_writemask(out, writemask),
tex_target, tex, sampler );
ureg_END( ureg );
return ureg_create_shader_and_destroy( ureg, pipe );
}
static void
mc_frag_shader_callback(void *priv, struct vl_mc *mc,
struct ureg_program *shader,
unsigned first_input,
struct ureg_dst dst)
{
struct vl_mpeg12_decoder *dec = priv;
struct ureg_src src, sampler;
assert(priv && mc);
assert(shader);
if (dec->base.entrypoint <= PIPE_VIDEO_ENTRYPOINT_IDCT) {
struct vl_idct *idct = mc == &dec->mc_y ? &dec->idct_y : &dec->idct_c;
vl_idct_stage2_frag_shader(idct, shader, first_input, dst);
} else {
src = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, first_input, TGSI_INTERPOLATE_LINEAR);
sampler = ureg_DECL_sampler(shader, 0);
ureg_TEX(shader, dst, TGSI_TEXTURE_2D, src, sampler);
}
}
static void
ureg_load_tex(struct ureg_program *ureg, struct ureg_dst out,
struct ureg_src coord, struct ureg_src sampler,
unsigned tex_target, bool load_level_zero, bool use_txf)
{
if (use_txf) {
struct ureg_dst temp = ureg_DECL_temporary(ureg);
ureg_F2I(ureg, temp, coord);
if (load_level_zero)
ureg_TXF_LZ(ureg, out, tex_target, ureg_src(temp), sampler);
else
ureg_TXF(ureg, out, tex_target, ureg_src(temp), sampler);
} else {
if (load_level_zero)
ureg_TEX_LZ(ureg, out, tex_target, coord, sampler);
else
ureg_TEX(ureg, out, tex_target, coord, sampler);
}
}
static void *
create_ref_frag_shader(struct vl_mc *r)
{
const float y_scale =
r->buffer_height / 2 *
r->macroblock_size / VL_MACROBLOCK_HEIGHT;
struct ureg_program *shader;
struct ureg_src tc[2], sampler;
struct ureg_dst ref, field;
struct ureg_dst fragment;
unsigned label;
shader = ureg_create(TGSI_PROCESSOR_FRAGMENT);
if (!shader)
return NULL;
tc[0] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTOP, TGSI_INTERPOLATE_LINEAR);
tc[1] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VBOTTOM, TGSI_INTERPOLATE_LINEAR);
sampler = ureg_DECL_sampler(shader, 0);
ref = ureg_DECL_temporary(shader);
fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
field = calc_line(shader);
/*
* ref = field.z ? tc[1] : tc[0]
*
* // Adjust tc acording to top/bottom field selection
* if (|ref.z|) {
* ref.y *= y_scale
* ref.y = floor(ref.y)
* ref.y += ref.z
* ref.y /= y_scale
* }
* fragment.xyz = tex(ref, sampler[0])
*/
ureg_CMP(shader, ureg_writemask(ref, TGSI_WRITEMASK_XYZ),
ureg_negate(ureg_scalar(ureg_src(field), TGSI_SWIZZLE_Y)),
tc[1], tc[0]);
ureg_CMP(shader, ureg_writemask(fragment, TGSI_WRITEMASK_W),
ureg_negate(ureg_scalar(ureg_src(field), TGSI_SWIZZLE_Y)),
tc[1], tc[0]);
ureg_IF(shader, ureg_scalar(ureg_src(ref), TGSI_SWIZZLE_Z), &label);
ureg_MUL(shader, ureg_writemask(ref, TGSI_WRITEMASK_Y),
ureg_src(ref), ureg_imm1f(shader, y_scale));
ureg_FLR(shader, ureg_writemask(ref, TGSI_WRITEMASK_Y), ureg_src(ref));
ureg_ADD(shader, ureg_writemask(ref, TGSI_WRITEMASK_Y),
ureg_src(ref), ureg_scalar(ureg_src(ref), TGSI_SWIZZLE_Z));
ureg_MUL(shader, ureg_writemask(ref, TGSI_WRITEMASK_Y),
ureg_src(ref), ureg_imm1f(shader, 1.0f / y_scale));
ureg_fixup_label(shader, label, ureg_get_instruction_number(shader));
ureg_ENDIF(shader);
ureg_TEX(shader, ureg_writemask(fragment, TGSI_WRITEMASK_XYZ), TGSI_TEXTURE_2D, ureg_src(ref), sampler);
ureg_release_temporary(shader, ref);
ureg_release_temporary(shader, field);
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, r->pipe);
}
static void *
create_deint_frag_shader(struct vl_deint_filter *filter, unsigned field,
struct vertex2f *sizes, bool spatial_filter)
{
struct ureg_program *shader;
struct ureg_src i_vtex;
struct ureg_src sampler_cur;
struct ureg_src sampler_prevprev;
struct ureg_src sampler_prev;
struct ureg_src sampler_next;
struct ureg_dst o_fragment;
struct ureg_dst t_tex;
struct ureg_dst t_comp_top, t_comp_bot;
struct ureg_dst t_diff;
struct ureg_dst t_a, t_b;
struct ureg_dst t_weave, t_linear;
shader = ureg_create(PIPE_SHADER_FRAGMENT);
if (!shader) {
return NULL;
}
t_tex = ureg_DECL_temporary(shader);
t_comp_top = ureg_DECL_temporary(shader);
t_comp_bot = ureg_DECL_temporary(shader);
t_diff = ureg_DECL_temporary(shader);
t_a = ureg_DECL_temporary(shader);
t_b = ureg_DECL_temporary(shader);
t_weave = ureg_DECL_temporary(shader);
t_linear = ureg_DECL_temporary(shader);
i_vtex = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX, TGSI_INTERPOLATE_LINEAR);
sampler_prevprev = ureg_DECL_sampler(shader, 0);
sampler_prev = ureg_DECL_sampler(shader, 1);
sampler_cur = ureg_DECL_sampler(shader, 2);
sampler_next = ureg_DECL_sampler(shader, 3);
o_fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
// we don't care about ZW interpolation (allows better optimization)
ureg_MOV(shader, t_tex, i_vtex);
ureg_MOV(shader, ureg_writemask(t_tex, TGSI_WRITEMASK_ZW),
ureg_imm1f(shader, 0));
// sample between texels for cheap lowpass
ureg_ADD(shader, t_comp_top, ureg_src(t_tex),
ureg_imm4f(shader, sizes->x * 0.5f, sizes->y * -0.5f, 0, 0));
ureg_ADD(shader, t_comp_bot, ureg_src(t_tex),
ureg_imm4f(shader, sizes->x * -0.5f, sizes->y * 0.5f, 1.0f, 0));
if (field == 0) {
/* interpolating top field -> current field is a bottom field */
// cur vs prev2
ureg_TEX(shader, t_a, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_bot), sampler_cur);
ureg_TEX(shader, t_b, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_bot), sampler_prevprev);
ureg_ADD(shader, ureg_writemask(t_diff, TGSI_WRITEMASK_X), ureg_src(t_a), ureg_negate(ureg_src(t_b)));
// prev vs next
ureg_TEX(shader, t_a, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_top), sampler_prev);
ureg_TEX(shader, t_b, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_top), sampler_next);
ureg_ADD(shader, ureg_writemask(t_diff, TGSI_WRITEMASK_Y), ureg_src(t_a), ureg_negate(ureg_src(t_b)));
} else {
/* interpolating bottom field -> current field is a top field */
// cur vs prev2
ureg_TEX(shader, t_a, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_top), sampler_cur);
ureg_TEX(shader, t_b, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_top), sampler_prevprev);
ureg_ADD(shader, ureg_writemask(t_diff, TGSI_WRITEMASK_X), ureg_src(t_a), ureg_negate(ureg_src(t_b)));
// prev vs next
ureg_TEX(shader, t_a, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_bot), sampler_prev);
ureg_TEX(shader, t_b, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_bot), sampler_next);
ureg_ADD(shader, ureg_writemask(t_diff, TGSI_WRITEMASK_Y), ureg_src(t_a), ureg_negate(ureg_src(t_b)));
}
// absolute maximum of differences
ureg_MAX(shader, ureg_writemask(t_diff, TGSI_WRITEMASK_X), ureg_abs(ureg_src(t_diff)),
ureg_scalar(ureg_abs(ureg_src(t_diff)), TGSI_SWIZZLE_Y));
if (field == 0) {
/* weave with prev top field */
ureg_TEX(shader, t_weave, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_tex), sampler_prev);
/* get linear interpolation from current bottom field */
ureg_ADD(shader, t_comp_top, ureg_src(t_tex), ureg_imm4f(shader, 0, sizes->y * -1.0f, 1.0f, 0));
ureg_TEX(shader, t_linear, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_top), sampler_cur);
} else {
/* weave with prev bottom field */
ureg_ADD(shader, t_comp_bot, ureg_src(t_tex), ureg_imm4f(shader, 0, 0, 1.0f, 0));
ureg_TEX(shader, t_weave, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_bot), sampler_prev);
/* get linear interpolation from current top field */
ureg_ADD(shader, t_comp_bot, ureg_src(t_tex), ureg_imm4f(shader, 0, sizes->y * 1.0f, 0, 0));
ureg_TEX(shader, t_linear, TGSI_TEXTURE_2D_ARRAY, ureg_src(t_comp_bot), sampler_cur);
}
// mix between weave and linear
// fully weave if diff < 6 (0.02353), fully interpolate if diff > 14 (0.05490)
ureg_ADD(shader, ureg_writemask(t_diff, TGSI_WRITEMASK_X), ureg_src(t_diff),
ureg_imm4f(shader, -0.02353f, 0, 0, 0));
ureg_MUL(shader, ureg_saturate(ureg_writemask(t_diff, TGSI_WRITEMASK_X)),
ureg_src(t_diff), ureg_imm4f(shader, 31.8750f, 0, 0, 0));
ureg_LRP(shader, ureg_writemask(t_tex, TGSI_WRITEMASK_X), ureg_src(t_diff),
ureg_src(t_linear), ureg_src(t_weave));
ureg_MOV(shader, o_fragment, ureg_scalar(ureg_src(t_tex), TGSI_SWIZZLE_X));
ureg_release_temporary(shader, t_tex);
ureg_release_temporary(shader, t_comp_top);
ureg_release_temporary(shader, t_comp_bot);
ureg_release_temporary(shader, t_diff);
ureg_release_temporary(shader, t_a);
ureg_release_temporary(shader, t_b);
ureg_release_temporary(shader, t_weave);
ureg_release_temporary(shader, t_linear);
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, filter->pipe);
}
static void *
create_frag_shader_weave(struct vl_compositor *c)
{
struct ureg_program *shader;
struct ureg_src i_tc[2];
struct ureg_src csc[3];
struct ureg_src sampler[3];
struct ureg_dst t_tc[2];
struct ureg_dst t_texel[2];
struct ureg_dst o_fragment;
unsigned i, j;
shader = ureg_create(TGSI_PROCESSOR_FRAGMENT);
if (!shader)
return false;
i_tc[0] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTOP, TGSI_INTERPOLATE_LINEAR);
i_tc[1] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VBOTTOM, TGSI_INTERPOLATE_LINEAR);
for (i = 0; i < 3; ++i) {
csc[i] = ureg_DECL_constant(shader, i);
sampler[i] = ureg_DECL_sampler(shader, i);
}
for (i = 0; i < 2; ++i) {
t_tc[i] = ureg_DECL_temporary(shader);
t_texel[i] = ureg_DECL_temporary(shader);
}
o_fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
/* calculate the texture offsets
* t_tc.x = i_tc.x
* t_tc.y = (round(i_tc.y) + 0.5) / height * 2
*/
for (i = 0; i < 2; ++i) {
ureg_MOV(shader, ureg_writemask(t_tc[i], TGSI_WRITEMASK_X), i_tc[i]);
ureg_ROUND(shader, ureg_writemask(t_tc[i], TGSI_WRITEMASK_YZ), i_tc[i]);
ureg_MOV(shader, ureg_writemask(t_tc[i], TGSI_WRITEMASK_W),
ureg_imm1f(shader, i ? 0.75f : 0.25f));
ureg_ADD(shader, ureg_writemask(t_tc[i], TGSI_WRITEMASK_YZ),
ureg_src(t_tc[i]), ureg_imm1f(shader, 0.5f));
ureg_MUL(shader, ureg_writemask(t_tc[i], TGSI_WRITEMASK_Y),
ureg_src(t_tc[i]), ureg_scalar(i_tc[0], TGSI_SWIZZLE_W));
ureg_MUL(shader, ureg_writemask(t_tc[i], TGSI_WRITEMASK_Z),
ureg_src(t_tc[i]), ureg_scalar(i_tc[1], TGSI_SWIZZLE_W));
}
/* fetch the texels
* texel[0..1].x = tex(t_tc[0..1][0])
* texel[0..1].y = tex(t_tc[0..1][1])
* texel[0..1].z = tex(t_tc[0..1][2])
*/
for (i = 0; i < 2; ++i)
for (j = 0; j < 3; ++j) {
struct ureg_src src = ureg_swizzle(ureg_src(t_tc[i]),
TGSI_SWIZZLE_X, j ? TGSI_SWIZZLE_Z : TGSI_SWIZZLE_Y, TGSI_SWIZZLE_W, TGSI_SWIZZLE_W);
ureg_TEX(shader, ureg_writemask(t_texel[i], TGSI_WRITEMASK_X << j),
TGSI_TEXTURE_3D, src, sampler[j]);
}
/* calculate linear interpolation factor
* factor = |round(i_tc.y) - i_tc.y| * 2
*/
ureg_ROUND(shader, ureg_writemask(t_tc[0], TGSI_WRITEMASK_YZ), i_tc[0]);
ureg_ADD(shader, ureg_writemask(t_tc[0], TGSI_WRITEMASK_YZ),
ureg_src(t_tc[0]), ureg_negate(i_tc[0]));
ureg_MUL(shader, ureg_writemask(t_tc[0], TGSI_WRITEMASK_XY),
ureg_abs(ureg_src(t_tc[0])), ureg_imm1f(shader, 2.0f));
ureg_LRP(shader, t_texel[0], ureg_swizzle(ureg_src(t_tc[0]),
TGSI_SWIZZLE_Y, TGSI_SWIZZLE_Z, TGSI_SWIZZLE_Z, TGSI_SWIZZLE_Z),
ureg_src(t_texel[1]), ureg_src(t_texel[0]));
/* and finally do colour space transformation
* fragment = csc * texel
*/
ureg_MOV(shader, ureg_writemask(t_texel[0], TGSI_WRITEMASK_W), ureg_imm1f(shader, 1.0f));
for (i = 0; i < 3; ++i)
ureg_DP4(shader, ureg_writemask(o_fragment, TGSI_WRITEMASK_X << i), csc[i], ureg_src(t_texel[0]));
ureg_MOV(shader, ureg_writemask(o_fragment, TGSI_WRITEMASK_W), ureg_imm1f(shader, 1.0f));
for (i = 0; i < 2; ++i) {
ureg_release_temporary(shader, t_texel[i]);
ureg_release_temporary(shader, t_tc[i]);
}
ureg_END(shader);
return ureg_create_shader_and_destroy(shader, c->pipe);
}
static void *
create_frag_shader(struct vl_zscan *zscan)
{
struct ureg_program *shader;
struct ureg_src *vtex;
struct ureg_src samp_src, samp_scan, samp_quant;
struct ureg_dst *tmp;
struct ureg_dst quant, fragment;
unsigned i;
shader = ureg_create(PIPE_SHADER_FRAGMENT);
if (!shader)
return NULL;
vtex = MALLOC(zscan->num_channels * sizeof(struct ureg_src));
tmp = MALLOC(zscan->num_channels * sizeof(struct ureg_dst));
for (i = 0; i < zscan->num_channels; ++i)
vtex[i] = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX + i, TGSI_INTERPOLATE_LINEAR);
samp_src = ureg_DECL_sampler(shader, 0);
samp_scan = ureg_DECL_sampler(shader, 1);
samp_quant = ureg_DECL_sampler(shader, 2);
for (i = 0; i < zscan->num_channels; ++i)
tmp[i] = ureg_DECL_temporary(shader);
quant = ureg_DECL_temporary(shader);
fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
/*
* tmp.x = tex(vtex, 1)
* tmp.y = vtex.z
* fragment = tex(tmp, 0) * quant
*/
for (i = 0; i < zscan->num_channels; ++i)
ureg_TEX(shader, ureg_writemask(tmp[i], TGSI_WRITEMASK_X), TGSI_TEXTURE_2D, vtex[i], samp_scan);
for (i = 0; i < zscan->num_channels; ++i)
ureg_MOV(shader, ureg_writemask(tmp[i], TGSI_WRITEMASK_Y), ureg_scalar(vtex[i], TGSI_SWIZZLE_W));
for (i = 0; i < zscan->num_channels; ++i) {
ureg_TEX(shader, ureg_writemask(tmp[0], TGSI_WRITEMASK_X << i), TGSI_TEXTURE_2D, ureg_src(tmp[i]), samp_src);
ureg_TEX(shader, ureg_writemask(quant, TGSI_WRITEMASK_X << i), TGSI_TEXTURE_3D, vtex[i], samp_quant);
}
ureg_MUL(shader, quant, ureg_src(quant), ureg_imm1f(shader, 16.0f));
ureg_MUL(shader, fragment, ureg_src(tmp[0]), ureg_src(quant));
for (i = 0; i < zscan->num_channels; ++i)
ureg_release_temporary(shader, tmp[i]);
ureg_END(shader);
FREE(vtex);
FREE(tmp);
return ureg_create_shader_and_destroy(shader, zscan->pipe);
}
static void *
create_frag_shader(struct vl_median_filter *filter,
struct vertex2f *offsets,
unsigned num_offsets)
{
struct pipe_screen *screen = filter->pipe->screen;
struct ureg_program *shader;
struct ureg_src i_vtex;
struct ureg_src sampler;
struct ureg_dst *t_array = MALLOC(sizeof(struct ureg_dst) * num_offsets);
struct ureg_dst o_fragment;
const unsigned median = num_offsets >> 1;
unsigned i, j;
assert(num_offsets & 1); /* we need an odd number of offsets */
if (!(num_offsets & 1)) { /* yeah, we REALLY need an odd number of offsets!!! */
FREE(t_array);
return NULL;
}
if (num_offsets > screen->get_shader_param(
screen, PIPE_SHADER_FRAGMENT, PIPE_SHADER_CAP_MAX_TEMPS)) {
FREE(t_array);
return NULL;
}
shader = ureg_create(PIPE_SHADER_FRAGMENT);
if (!shader) {
FREE(t_array);
return NULL;
}
i_vtex = ureg_DECL_fs_input(shader, TGSI_SEMANTIC_GENERIC, VS_O_VTEX, TGSI_INTERPOLATE_LINEAR);
sampler = ureg_DECL_sampler(shader, 0);
ureg_DECL_sampler_view(shader, 0, TGSI_TEXTURE_2D,
TGSI_RETURN_TYPE_FLOAT,
TGSI_RETURN_TYPE_FLOAT,
TGSI_RETURN_TYPE_FLOAT,
TGSI_RETURN_TYPE_FLOAT);
for (i = 0; i < num_offsets; ++i)
t_array[i] = ureg_DECL_temporary(shader);
o_fragment = ureg_DECL_output(shader, TGSI_SEMANTIC_COLOR, 0);
/*
* t_array[0..*] = vtex + offset[0..*]
* t_array[0..*] = tex(t_array[0..*], sampler)
* result = partial_bubblesort(t_array)[mid]
*/
for (i = 0; i < num_offsets; ++i) {
if (!is_vec_zero(offsets[i])) {
ureg_ADD(shader, ureg_writemask(t_array[i], TGSI_WRITEMASK_XY),
i_vtex, ureg_imm2f(shader, offsets[i].x, offsets[i].y));
ureg_MOV(shader, ureg_writemask(t_array[i], TGSI_WRITEMASK_ZW),
ureg_imm1f(shader, 0.0f));
}
}
for (i = 0; i < num_offsets; ++i) {
struct ureg_src src = is_vec_zero(offsets[i]) ? i_vtex : ureg_src(t_array[i]);
ureg_TEX(shader, t_array[i], TGSI_TEXTURE_2D, src, sampler);
}
// TODO: Couldn't this be improved even more?
for (i = 0; i <= median; ++i) {
for (j = 1; j < (num_offsets - i - 1); ++j) {
struct ureg_dst tmp = ureg_DECL_temporary(shader);
ureg_MOV(shader, tmp, ureg_src(t_array[j]));
ureg_MAX(shader, t_array[j], ureg_src(t_array[j]), ureg_src(t_array[j - 1]));
ureg_MIN(shader, t_array[j - 1], ureg_src(tmp), ureg_src(t_array[j - 1]));
ureg_release_temporary(shader, tmp);
}
if (i == median)
ureg_MAX(shader, t_array[j], ureg_src(t_array[j]), ureg_src(t_array[j - 1]));
else
ureg_MIN(shader, t_array[j - 1], ureg_src(t_array[j]), ureg_src(t_array[j - 1]));
}
ureg_MOV(shader, o_fragment, ureg_src(t_array[median]));
ureg_END(shader);
FREE(t_array);
return ureg_create_shader_and_destroy(shader, filter->pipe);
}
