static void
lower_implicit_lod(nir_builder *b, nir_tex_instr *tex)
{
assert(tex->op == nir_texop_tex || tex->op == nir_texop_txb);
assert(nir_tex_instr_src_index(tex, nir_tex_src_lod) < 0);
assert(nir_tex_instr_src_index(tex, nir_tex_src_ddx) < 0);
assert(nir_tex_instr_src_index(tex, nir_tex_src_ddy) < 0);
b->cursor = nir_before_instr(&tex->instr);
nir_ssa_def *lod = get_texture_lod(b, tex);
int bias_idx = nir_tex_instr_src_index(tex, nir_tex_src_bias);
if (bias_idx >= 0) {
/* If we have a bias, add it in */
lod = nir_fadd(b, lod, nir_ssa_for_src(b, tex->src[bias_idx].src, 1));
nir_tex_instr_remove_src(tex, bias_idx);
}
int min_lod_idx = nir_tex_instr_src_index(tex, nir_tex_src_min_lod);
if (min_lod_idx >= 0) {
/* If we have a minimum LOD, clamp LOD accordingly */
lod = nir_fmax(b, lod, nir_ssa_for_src(b, tex->src[min_lod_idx].src, 1));
nir_tex_instr_remove_src(tex, min_lod_idx);
}
nir_tex_instr_add_src(tex, nir_tex_src_lod, nir_src_for_ssa(lod));
tex->op = nir_texop_txl;
}
static nir_src
get_deref_reg_src(nir_deref_instr *deref, struct locals_to_regs_state *state)
{
nir_builder *b = &state->builder;
nir_src src;
src.is_ssa = false;
src.reg.reg = get_reg_for_deref(deref, state);
src.reg.base_offset = 0;
src.reg.indirect = NULL;
/* It is possible for a user to create a shader that has an array with a
* single element and then proceed to access it indirectly. Indirectly
* accessing a non-array register is not allowed in NIR. In order to
* handle this case we just convert it to a direct reference.
*/
if (src.reg.reg->num_array_elems == 0)
return src;
unsigned inner_array_size = 1;
for (const nir_deref_instr *d = deref; d; d = nir_deref_instr_parent(d)) {
if (d->deref_type != nir_deref_type_array)
continue;
if (nir_src_is_const(d->arr.index) && !src.reg.indirect) {
src.reg.base_offset += nir_src_as_uint(d->arr.index) *
inner_array_size;
} else {
if (src.reg.indirect) {
assert(src.reg.base_offset == 0);
} else {
src.reg.indirect = ralloc(b->shader, nir_src);
*src.reg.indirect =
nir_src_for_ssa(nir_imm_int(b, src.reg.base_offset));
src.reg.base_offset = 0;
}
assert(src.reg.indirect->is_ssa);
nir_ssa_def *index = nir_i2i(b, nir_ssa_for_src(b, d->arr.index, 1), 32);
src.reg.indirect->ssa =
nir_iadd(b, src.reg.indirect->ssa,
nir_imul(b, index, nir_imm_int(b, inner_array_size)));
}
inner_array_size *= glsl_get_length(nir_deref_instr_parent(d)->type);
}
return src;
}
/* Calculate the sampler index based on array indicies and also
* calculate the base uniform location for struct members.
*/
static void
calc_sampler_offsets(nir_deref *tail, nir_tex_instr *instr,
unsigned *array_elements, nir_ssa_def **indirect,
nir_builder *b, unsigned *location)
{
if (tail->child == NULL)
return;
switch (tail->child->deref_type) {
case nir_deref_type_array: {
nir_deref_array *deref_array = nir_deref_as_array(tail->child);
assert(deref_array->deref_array_type != nir_deref_array_type_wildcard);
calc_sampler_offsets(tail->child, instr, array_elements,
indirect, b, location);
instr->sampler_index += deref_array->base_offset * *array_elements;
if (deref_array->deref_array_type == nir_deref_array_type_indirect) {
nir_ssa_def *mul =
nir_imul(b, nir_imm_int(b, *array_elements),
nir_ssa_for_src(b, deref_array->indirect, 1));
nir_instr_rewrite_src(&instr->instr, &deref_array->indirect,
NIR_SRC_INIT);
if (*indirect) {
*indirect = nir_iadd(b, *indirect, mul);
} else {
*indirect = mul;
}
}
*array_elements *= glsl_get_length(tail->type);
break;
}
case nir_deref_type_struct: {
nir_deref_struct *deref_struct = nir_deref_as_struct(tail->child);
*location += glsl_get_record_location_offset(tail->type, deref_struct->index);
calc_sampler_offsets(tail->child, instr, array_elements,
indirect, b, location);
break;
}
default:
unreachable("Invalid deref type");
break;
}
}
static void
saturate_src(nir_builder *b, nir_tex_instr *tex, unsigned sat_mask)
{
b->cursor = nir_before_instr(&tex->instr);
/* Walk through the sources saturating 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 *src =
nir_ssa_for_src(b, tex->src[i].src, tex->coord_components);
/* split src into components: */
nir_ssa_def *comp[4];
for (unsigned j = 0; j < tex->coord_components; j++)
comp[j] = nir_channel(b, src, j);
/* clamp requested components, array index does not get clamped: */
unsigned ncomp = tex->coord_components;
if (tex->is_array)
ncomp--;
for (unsigned j = 0; j < ncomp; j++) {
if ((1 << j) & sat_mask) {
if (tex->sampler_dim == GLSL_SAMPLER_DIM_RECT) {
/* non-normalized texture coords, so clamp to texture
* size rather than [0.0, 1.0]
*/
nir_ssa_def *txs = get_texture_size(b, tex);
comp[j] = nir_fmax(b, comp[j], nir_imm_float(b, 0.0));
comp[j] = nir_fmin(b, comp[j], nir_channel(b, txs, j));
} else {
comp[j] = nir_fsat(b, comp[j]);
}
}
}
/* and move the result back into a single vecN: */
src = nir_vec(b, comp, tex->coord_components);
nir_instr_rewrite_src(&tex->instr,
&tex->src[i].src,
nir_src_for_ssa(src));
}
}
/* Multiply interp_var_at_offset's offset by transform.x to flip it. */
static void
lower_interp_var_at_offset(lower_wpos_ytransform_state *state,
nir_intrinsic_instr *interp)
{
nir_builder *b = &state->b;
nir_ssa_def *offset;
nir_ssa_def *flip_y;
b->cursor = nir_before_instr(&interp->instr);
offset = nir_ssa_for_src(b, interp->src[0], 2);
flip_y = nir_fmul(b, nir_channel(b, offset, 1),
nir_channel(b, get_transform(state), 0));
nir_instr_rewrite_src(&interp->instr, &interp->src[0],
nir_src_for_ssa(nir_vec2(b, nir_channel(b, offset, 0),
flip_y)));
}
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_src(bld, alu->src[0].src,
nir_ssa_alu_instr_src_components(alu, 0));
denom = nir_ssa_for_src(bld, alu->src[1].src,
nir_ssa_alu_instr_src_components(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_ige(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));
}
