static struct vtn_ssa_value *
matrix_inverse(struct vtn_builder *b, struct vtn_ssa_value *src)
{
nir_ssa_def *adj_col[4];
unsigned size = glsl_get_vector_elements(src->type);
/* Build up an adjugate matrix */
for (unsigned c = 0; c < size; c++) {
nir_ssa_def *elem[4];
for (unsigned r = 0; r < size; r++) {
elem[r] = build_mat_subdet(&b->nb, src, size, c, r);
if ((r + c) % 2)
elem[r] = nir_fneg(&b->nb, elem[r]);
}
adj_col[c] = nir_vec(&b->nb, elem, size);
}
nir_ssa_def *det_inv = nir_frcp(&b->nb, build_mat_det(b, src));
struct vtn_ssa_value *val = vtn_create_ssa_value(b, src->type);
for (unsigned i = 0; i < size; i++)
val->elems[i]->def = nir_fmul(&b->nb, adj_col[i], det_inv);
return val;
}
static bool
lower_offset(nir_builder *b, nir_tex_instr *tex)
{
int offset_index = nir_tex_instr_src_index(tex, nir_tex_src_offset);
if (offset_index < 0)
return false;
int coord_index = nir_tex_instr_src_index(tex, nir_tex_src_coord);
assert(coord_index >= 0);
assert(tex->src[offset_index].src.is_ssa);
assert(tex->src[coord_index].src.is_ssa);
nir_ssa_def *offset = tex->src[offset_index].src.ssa;
nir_ssa_def *coord = tex->src[coord_index].src.ssa;
b->cursor = nir_before_instr(&tex->instr);
nir_ssa_def *offset_coord;
if (nir_tex_instr_src_type(tex, coord_index) == nir_type_float) {
if (tex->sampler_dim == GLSL_SAMPLER_DIM_RECT) {
offset_coord = nir_fadd(b, coord, nir_i2f32(b, offset));
} else {
nir_ssa_def *txs = get_texture_size(b, tex);
nir_ssa_def *scale = nir_frcp(b, txs);
offset_coord = nir_fadd(b, coord,
nir_fmul(b,
nir_i2f32(b, offset),
scale));
}
} else {
offset_coord = nir_iadd(b, coord, offset);
}
if (tex->is_array) {
/* The offset is not applied to the array index */
if (tex->coord_components == 2) {
offset_coord = nir_vec2(b, nir_channel(b, offset_coord, 0),
nir_channel(b, coord, 1));
} else if (tex->coord_components == 3) {
offset_coord = nir_vec3(b, nir_channel(b, offset_coord, 0),
nir_channel(b, offset_coord, 1),
nir_channel(b, coord, 2));
} else {
unreachable("Invalid number of components");
}
}
nir_instr_rewrite_src(&tex->instr, &tex->src[coord_index].src,
nir_src_for_ssa(offset_coord));
nir_tex_instr_remove_src(tex, offset_index);
return true;
}
static void
lower_rect(nir_builder *b, nir_tex_instr *tex)
{
nir_ssa_def *txs = get_texture_size(b, tex);
nir_ssa_def *scale = nir_frcp(b, txs);
/* Walk through the sources normalizing the requested arguments. */
for (unsigned i = 0; i < tex->num_srcs; i++) {
if (tex->src[i].src_type != nir_tex_src_coord)
continue;
nir_ssa_def *coords =
nir_ssa_for_src(b, tex->src[i].src, tex->coord_components);
nir_instr_rewrite_src(&tex->instr,
&tex->src[i].src,
nir_src_for_ssa(nir_fmul(b, coords, scale)));
}
tex->sampler_dim = GLSL_SAMPLER_DIM_2D;
}
static void
project_src(nir_builder *b, nir_tex_instr *tex)
{
/* Find the projector in the srcs list, if present. */
int proj_index = nir_tex_instr_src_index(tex, nir_tex_src_projector);
if (proj_index < 0)
return;
b->cursor = nir_before_instr(&tex->instr);
nir_ssa_def *inv_proj =
nir_frcp(b, nir_ssa_for_src(b, tex->src[proj_index].src, 1));
/* Walk through the sources projecting the arguments. */
for (unsigned i = 0; i < tex->num_srcs; i++) {
switch (tex->src[i].src_type) {
case nir_tex_src_coord:
case nir_tex_src_comparator:
break;
default:
continue;
}
nir_ssa_def *unprojected =
nir_ssa_for_src(b, tex->src[i].src, nir_tex_instr_src_size(tex, i));
nir_ssa_def *projected = nir_fmul(b, unprojected, inv_proj);
/* Array indices don't get projected, so make an new vector with the
* coordinate's array index untouched.
*/
if (tex->is_array && tex->src[i].src_type == nir_tex_src_coord) {
switch (tex->coord_components) {
case 4:
projected = nir_vec4(b,
nir_channel(b, projected, 0),
nir_channel(b, projected, 1),
nir_channel(b, projected, 2),
nir_channel(b, unprojected, 3));
break;
case 3:
projected = nir_vec3(b,
nir_channel(b, projected, 0),
nir_channel(b, projected, 1),
nir_channel(b, unprojected, 2));
break;
case 2:
projected = nir_vec2(b,
nir_channel(b, projected, 0),
nir_channel(b, unprojected, 1));
break;
default:
unreachable("bad texture coord count for array");
break;
}
}
nir_instr_rewrite_src(&tex->instr,
&tex->src[i].src,
nir_src_for_ssa(projected));
}
nir_tex_instr_remove_src(tex, proj_index);
}
static void
convert_instr(nir_builder *bld, nir_alu_instr *alu)
{
nir_ssa_def *numer, *denom, *af, *bf, *a, *b, *q, *r;
nir_op op = alu->op;
bool is_signed;
if ((op != nir_op_idiv) &&
(op != nir_op_udiv) &&
(op != nir_op_umod))
return;
is_signed = (op == nir_op_idiv);
bld->cursor = nir_before_instr(&alu->instr);
numer = nir_ssa_for_alu_src(bld, alu, 0);
denom = nir_ssa_for_alu_src(bld, alu, 1);
if (is_signed) {
af = nir_i2f(bld, numer);
bf = nir_i2f(bld, denom);
af = nir_fabs(bld, af);
bf = nir_fabs(bld, bf);
a = nir_iabs(bld, numer);
b = nir_iabs(bld, denom);
} else {
af = nir_u2f(bld, numer);
bf = nir_u2f(bld, denom);
a = numer;
b = denom;
}
/* get first result: */
bf = nir_frcp(bld, bf);
bf = nir_isub(bld, bf, nir_imm_int(bld, 2)); /* yes, really */
q = nir_fmul(bld, af, bf);
if (is_signed) {
q = nir_f2i(bld, q);
} else {
q = nir_f2u(bld, q);
}
/* get error of first result: */
r = nir_imul(bld, q, b);
r = nir_isub(bld, a, r);
r = nir_u2f(bld, r);
r = nir_fmul(bld, r, bf);
r = nir_f2u(bld, r);
/* add quotients: */
q = nir_iadd(bld, q, r);
/* correction: if modulus >= divisor, add 1 */
r = nir_imul(bld, q, b);
r = nir_isub(bld, a, r);
r = nir_uge(bld, r, b);
r = nir_b2i(bld, r);
q = nir_iadd(bld, q, r);
if (is_signed) {
/* fix the sign: */
r = nir_ixor(bld, numer, denom);
r = nir_ushr(bld, r, nir_imm_int(bld, 31));
r = nir_i2b(bld, r);
b = nir_ineg(bld, q);
q = nir_bcsel(bld, r, b, q);
}
if (op == nir_op_umod) {
/* division result in q */
r = nir_imul(bld, q, b);
q = nir_isub(bld, a, r);
}
assert(alu->dest.dest.is_ssa);
nir_ssa_def_rewrite_uses(&alu->dest.dest.ssa, nir_src_for_ssa(q));
}