/**
* Create the mapping from the FS inputs we produce to the previous pipeline
* stage (GS or VS) outputs they source from.
*/
void
calculate_attr_overrides(const struct brw_context *brw,
uint16_t *attr_overrides,
uint32_t *point_sprite_enables,
uint32_t *flat_enables,
uint32_t *urb_entry_read_length)
{
const int urb_entry_read_offset = BRW_SF_URB_ENTRY_READ_OFFSET;
uint32_t max_source_attr = 0;
*point_sprite_enables = 0;
*flat_enables = 0;
/* _NEW_LIGHT */
bool shade_model_flat = brw->ctx.Light.ShadeModel == GL_FLAT;
/* Initialize all the attr_overrides to 0. In the loop below we'll modify
* just the ones that correspond to inputs used by the fs.
*/
memset(attr_overrides, 0, 16*sizeof(*attr_overrides));
for (int attr = 0; attr < VARYING_SLOT_MAX; attr++) {
enum glsl_interp_qualifier interp_qualifier =
brw->fragment_program->InterpQualifier[attr];
bool is_gl_Color = attr == VARYING_SLOT_COL0 || attr == VARYING_SLOT_COL1;
/* CACHE_NEW_WM_PROG */
int input_index = brw->wm.prog_data->urb_setup[attr];
if (input_index < 0)
continue;
/* _NEW_POINT */
bool point_sprite = false;
if (brw->ctx.Point.PointSprite &&
(attr >= VARYING_SLOT_TEX0 && attr <= VARYING_SLOT_TEX7) &&
brw->ctx.Point.CoordReplace[attr - VARYING_SLOT_TEX0]) {
point_sprite = true;
}
if (attr == VARYING_SLOT_PNTC)
point_sprite = true;
if (point_sprite)
*point_sprite_enables |= (1 << input_index);
/* flat shading */
if (interp_qualifier == INTERP_QUALIFIER_FLAT ||
(shade_model_flat && is_gl_Color &&
interp_qualifier == INTERP_QUALIFIER_NONE))
*flat_enables |= (1 << input_index);
/* BRW_NEW_VUE_MAP_GEOM_OUT | _NEW_LIGHT | _NEW_PROGRAM */
uint16_t attr_override = point_sprite ? 0 :
get_attr_override(&brw->vue_map_geom_out,
urb_entry_read_offset, attr,
brw->ctx.VertexProgram._TwoSideEnabled,
&max_source_attr);
/* The hardware can only do the overrides on 16 overrides at a
* time, and the other up to 16 have to be lined up so that the
* input index = the output index. We'll need to do some
* tweaking to make sure that's the case.
*/
if (input_index < 16)
attr_overrides[input_index] = attr_override;
else
assert(attr_override == input_index);
}
/* From the Sandy Bridge PRM, Volume 2, Part 1, documentation for
* 3DSTATE_SF DWord 1 bits 15:11, "Vertex URB Entry Read Length":
*
* "This field should be set to the minimum length required to read the
* maximum source attribute. The maximum source attribute is indicated
* by the maximum value of the enabled Attribute # Source Attribute if
* Attribute Swizzle Enable is set, Number of Output Attributes-1 if
* enable is not set.
* read_length = ceiling((max_source_attr + 1) / 2)
*
* [errata] Corruption/Hang possible if length programmed larger than
* recommended"
*
* Similar text exists for Ivy Bridge.
*/
*urb_entry_read_length = ALIGN(max_source_attr + 1, 2) / 2;
}
static void
upload_sbe_state(struct brw_context *brw)
{
struct intel_context *intel = &brw->intel;
struct gl_context *ctx = &intel->ctx;
uint32_t urb_entry_read_length;
/* BRW_NEW_FRAGMENT_PROGRAM */
uint32_t num_outputs = _mesa_bitcount_64(brw->fragment_program->Base.InputsRead);
/* _NEW_LIGHT */
bool shade_model_flat = ctx->Light.ShadeModel == GL_FLAT;
uint32_t dw1, dw10, dw11;
int i;
int attr = 0, input_index = 0;
int urb_entry_read_offset = 1;
uint16_t attr_overrides[FRAG_ATTRIB_MAX];
/* _NEW_BUFFERS */
bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
uint32_t point_sprite_origin;
/* CACHE_NEW_VS_PROG */
urb_entry_read_length = ((brw->vs.prog_data->vue_map.num_slots + 1) / 2 -
urb_entry_read_offset);
if (urb_entry_read_length == 0) {
/* Setting the URB entry read length to 0 causes undefined behavior, so
* if we have no URB data to read, set it to 1.
*/
urb_entry_read_length = 1;
}
/* FINISHME: Attribute Swizzle Control Mode? */
dw1 =
GEN7_SBE_SWIZZLE_ENABLE |
num_outputs << GEN7_SBE_NUM_OUTPUTS_SHIFT |
urb_entry_read_length << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT |
urb_entry_read_offset << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT;
/* _NEW_POINT
*
* Window coordinates in an FBO are inverted, which means point
* sprite origin must be inverted.
*/
if ((ctx->Point.SpriteOrigin == GL_LOWER_LEFT) != render_to_fbo) {
point_sprite_origin = GEN6_SF_POINT_SPRITE_LOWERLEFT;
} else {
point_sprite_origin = GEN6_SF_POINT_SPRITE_UPPERLEFT;
}
dw1 |= point_sprite_origin;
dw10 = 0;
dw11 = 0;
/* Create the mapping from the FS inputs we produce to the VS outputs
* they source from.
*/
for (; attr < FRAG_ATTRIB_MAX; attr++) {
enum glsl_interp_qualifier interp_qualifier =
brw->fragment_program->InterpQualifier[attr];
bool is_gl_Color = attr == FRAG_ATTRIB_COL0 || attr == FRAG_ATTRIB_COL1;
if (!(brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)))
continue;
if (ctx->Point.PointSprite &&
attr >= FRAG_ATTRIB_TEX0 && attr <= FRAG_ATTRIB_TEX7 &&
ctx->Point.CoordReplace[attr - FRAG_ATTRIB_TEX0]) {
dw10 |= (1 << input_index);
}
if (attr == FRAG_ATTRIB_PNTC)
dw10 |= (1 << input_index);
/* flat shading */
if (interp_qualifier == INTERP_QUALIFIER_FLAT ||
(shade_model_flat && is_gl_Color &&
interp_qualifier == INTERP_QUALIFIER_NONE))
dw11 |= (1 << input_index);
/* The hardware can only do the overrides on 16 overrides at a
* time, and the other up to 16 have to be lined up so that the
* input index = the output index. We'll need to do some
* tweaking to make sure that's the case.
*/
assert(input_index < 16 || attr == input_index);
/* CACHE_NEW_VS_PROG | _NEW_LIGHT | _NEW_PROGRAM */
attr_overrides[input_index++] =
get_attr_override(&brw->vs.prog_data->vue_map,
urb_entry_read_offset, attr,
ctx->VertexProgram._TwoSideEnabled);
}
for (; input_index < FRAG_ATTRIB_MAX; input_index++)
attr_overrides[input_index] = 0;
BEGIN_BATCH(14);
OUT_BATCH(_3DSTATE_SBE << 16 | (14 - 2));
OUT_BATCH(dw1);
/* Output dwords 2 through 9 */
for (i = 0; i < 8; i++) {
OUT_BATCH(attr_overrides[i * 2] | attr_overrides[i * 2 + 1] << 16);
}
OUT_BATCH(dw10); /* point sprite texcoord bitmask */
OUT_BATCH(dw11); /* constant interp bitmask */
OUT_BATCH(0); /* wrapshortest enables 0-7 */
OUT_BATCH(0); /* wrapshortest enables 8-15 */
ADVANCE_BATCH();
}
static void
upload_sf_state(struct brw_context *brw)
{
struct intel_context *intel = &brw->intel;
struct gl_context *ctx = &intel->ctx;
uint32_t urb_entry_read_length;
/* BRW_NEW_FRAGMENT_PROGRAM */
uint32_t num_outputs = _mesa_bitcount_64(brw->fragment_program->Base.InputsRead);
/* _NEW_LIGHT */
bool shade_model_flat = ctx->Light.ShadeModel == GL_FLAT;
uint32_t dw1, dw2, dw3, dw4, dw16, dw17;
int i;
/* _NEW_BUFFER */
bool render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0;
int attr = 0, input_index = 0;
int urb_entry_read_offset = 1;
float point_size;
uint16_t attr_overrides[FRAG_ATTRIB_MAX];
uint32_t point_sprite_origin;
/* CACHE_NEW_VS_PROG */
urb_entry_read_length = ((brw->vs.prog_data->vue_map.num_slots + 1) / 2 -
urb_entry_read_offset);
if (urb_entry_read_length == 0) {
/* Setting the URB entry read length to 0 causes undefined behavior, so
* if we have no URB data to read, set it to 1.
*/
urb_entry_read_length = 1;
}
dw1 =
GEN6_SF_SWIZZLE_ENABLE |
num_outputs << GEN6_SF_NUM_OUTPUTS_SHIFT |
urb_entry_read_length << GEN6_SF_URB_ENTRY_READ_LENGTH_SHIFT |
urb_entry_read_offset << GEN6_SF_URB_ENTRY_READ_OFFSET_SHIFT;
dw2 = GEN6_SF_STATISTICS_ENABLE |
GEN6_SF_VIEWPORT_TRANSFORM_ENABLE;
dw3 = 0;
dw4 = 0;
dw16 = 0;
dw17 = 0;
/* _NEW_POLYGON */
if ((ctx->Polygon.FrontFace == GL_CCW) ^ render_to_fbo)
dw2 |= GEN6_SF_WINDING_CCW;
if (ctx->Polygon.OffsetFill)
dw2 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_SOLID;
if (ctx->Polygon.OffsetLine)
dw2 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_WIREFRAME;
if (ctx->Polygon.OffsetPoint)
dw2 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_POINT;
switch (ctx->Polygon.FrontMode) {
case GL_FILL:
dw2 |= GEN6_SF_FRONT_SOLID;
break;
case GL_LINE:
dw2 |= GEN6_SF_FRONT_WIREFRAME;
break;
case GL_POINT:
dw2 |= GEN6_SF_FRONT_POINT;
break;
default:
assert(0);
break;
}
switch (ctx->Polygon.BackMode) {
case GL_FILL:
dw2 |= GEN6_SF_BACK_SOLID;
break;
case GL_LINE:
dw2 |= GEN6_SF_BACK_WIREFRAME;
break;
case GL_POINT:
dw2 |= GEN6_SF_BACK_POINT;
break;
default:
assert(0);
break;
}
/* _NEW_SCISSOR */
if (ctx->Scissor.Enabled)
dw3 |= GEN6_SF_SCISSOR_ENABLE;
/* _NEW_POLYGON */
if (ctx->Polygon.CullFlag) {
switch (ctx->Polygon.CullFaceMode) {
case GL_FRONT:
dw3 |= GEN6_SF_CULL_FRONT;
break;
case GL_BACK:
dw3 |= GEN6_SF_CULL_BACK;
break;
case GL_FRONT_AND_BACK:
dw3 |= GEN6_SF_CULL_BOTH;
break;
default:
assert(0);
break;
}
} else {
dw3 |= GEN6_SF_CULL_NONE;
}
/* _NEW_LINE */
dw3 |= U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7) <<
GEN6_SF_LINE_WIDTH_SHIFT;
if (ctx->Line.SmoothFlag) {
dw3 |= GEN6_SF_LINE_AA_ENABLE;
dw3 |= GEN6_SF_LINE_AA_MODE_TRUE;
dw3 |= GEN6_SF_LINE_END_CAP_WIDTH_1_0;
}
/* _NEW_PROGRAM | _NEW_POINT */
if (!(ctx->VertexProgram.PointSizeEnabled ||
ctx->Point._Attenuated))
dw4 |= GEN6_SF_USE_STATE_POINT_WIDTH;
/* Clamp to ARB_point_parameters user limits */
point_size = CLAMP(ctx->Point.Size, ctx->Point.MinSize, ctx->Point.MaxSize);
/* Clamp to the hardware limits and convert to fixed point */
dw4 |= U_FIXED(CLAMP(point_size, 0.125, 255.875), 3);
/*
* Window coordinates in an FBO are inverted, which means point
* sprite origin must be inverted, too.
*/
if ((ctx->Point.SpriteOrigin == GL_LOWER_LEFT) != render_to_fbo) {
point_sprite_origin = GEN6_SF_POINT_SPRITE_LOWERLEFT;
} else {
point_sprite_origin = GEN6_SF_POINT_SPRITE_UPPERLEFT;
}
dw1 |= point_sprite_origin;
/* _NEW_LIGHT */
if (ctx->Light.ProvokingVertex != GL_FIRST_VERTEX_CONVENTION) {
dw4 |=
(2 << GEN6_SF_TRI_PROVOKE_SHIFT) |
(2 << GEN6_SF_TRIFAN_PROVOKE_SHIFT) |
(1 << GEN6_SF_LINE_PROVOKE_SHIFT);
} else {
dw4 |=
(1 << GEN6_SF_TRIFAN_PROVOKE_SHIFT);
}
/* Create the mapping from the FS inputs we produce to the VS outputs
* they source from.
*/
for (; attr < FRAG_ATTRIB_MAX; attr++) {
enum glsl_interp_qualifier interp_qualifier =
brw->fragment_program->InterpQualifier[attr];
bool is_gl_Color = attr == FRAG_ATTRIB_COL0 || attr == FRAG_ATTRIB_COL1;
if (!(brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)))
continue;
/* _NEW_POINT */
if (ctx->Point.PointSprite &&
(attr >= FRAG_ATTRIB_TEX0 && attr <= FRAG_ATTRIB_TEX7) &&
ctx->Point.CoordReplace[attr - FRAG_ATTRIB_TEX0]) {
dw16 |= (1 << input_index);
}
if (attr == FRAG_ATTRIB_PNTC)
dw16 |= (1 << input_index);
/* flat shading */
if (interp_qualifier == INTERP_QUALIFIER_FLAT ||
(shade_model_flat && is_gl_Color &&
interp_qualifier == INTERP_QUALIFIER_NONE))
dw17 |= (1 << input_index);
/* The hardware can only do the overrides on 16 overrides at a
* time, and the other up to 16 have to be lined up so that the
* input index = the output index. We'll need to do some
* tweaking to make sure that's the case.
*/
assert(input_index < 16 || attr == input_index);
/* CACHE_NEW_VS_PROG | _NEW_LIGHT | _NEW_PROGRAM */
attr_overrides[input_index++] =
get_attr_override(&brw->vs.prog_data->vue_map,
urb_entry_read_offset, attr,
ctx->VertexProgram._TwoSideEnabled);
}
for (; input_index < FRAG_ATTRIB_MAX; input_index++)
attr_overrides[input_index] = 0;
BEGIN_BATCH(20);
OUT_BATCH(_3DSTATE_SF << 16 | (20 - 2));
OUT_BATCH(dw1);
OUT_BATCH(dw2);
OUT_BATCH(dw3);
OUT_BATCH(dw4);
OUT_BATCH_F(ctx->Polygon.OffsetUnits * 2); /* constant. copied from gen4 */
OUT_BATCH_F(ctx->Polygon.OffsetFactor); /* scale */
OUT_BATCH_F(0.0); /* XXX: global depth offset clamp */
for (i = 0; i < 8; i++) {
OUT_BATCH(attr_overrides[i * 2] | attr_overrides[i * 2 + 1] << 16);
}
OUT_BATCH(dw16); /* point sprite texcoord bitmask */
OUT_BATCH(dw17); /* constant interp bitmask */
OUT_BATCH(0); /* wrapshortest enables 0-7 */
OUT_BATCH(0); /* wrapshortest enables 8-15 */
ADVANCE_BATCH();
}
static void
upload_sf_state(struct brw_context *brw)
{
struct intel_context *intel = &brw->intel;
struct gl_context *ctx = &intel->ctx;
/* BRW_NEW_FRAGMENT_PROGRAM */
uint32_t num_outputs = _mesa_bitcount_64(brw->fragment_program->Base.InputsRead);
/* _NEW_LIGHT */
bool shade_model_flat = ctx->Light.ShadeModel == GL_FLAT;
uint32_t dw1, dw2, dw3, dw4, dw16, dw17;
int i;
/* _NEW_BUFFER */
bool render_to_fbo = _mesa_is_user_fbo(brw->intel.ctx.DrawBuffer);
bool multisampled_fbo = ctx->DrawBuffer->Visual.samples > 1;
int attr = 0, input_index = 0;
int urb_entry_read_offset = 1;
float point_size;
uint16_t attr_overrides[FRAG_ATTRIB_MAX];
uint32_t point_sprite_origin;
dw1 = GEN6_SF_SWIZZLE_ENABLE | num_outputs << GEN6_SF_NUM_OUTPUTS_SHIFT;
dw2 = GEN6_SF_STATISTICS_ENABLE |
GEN6_SF_VIEWPORT_TRANSFORM_ENABLE;
dw3 = 0;
dw4 = 0;
dw16 = 0;
dw17 = 0;
/* _NEW_POLYGON */
if ((ctx->Polygon.FrontFace == GL_CCW) ^ render_to_fbo)
dw2 |= GEN6_SF_WINDING_CCW;
if (ctx->Polygon.OffsetFill)
dw2 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_SOLID;
if (ctx->Polygon.OffsetLine)
dw2 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_WIREFRAME;
if (ctx->Polygon.OffsetPoint)
dw2 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_POINT;
switch (ctx->Polygon.FrontMode) {
case GL_FILL:
dw2 |= GEN6_SF_FRONT_SOLID;
break;
case GL_LINE:
dw2 |= GEN6_SF_FRONT_WIREFRAME;
break;
case GL_POINT:
dw2 |= GEN6_SF_FRONT_POINT;
break;
default:
assert(0);
break;
}
switch (ctx->Polygon.BackMode) {
case GL_FILL:
dw2 |= GEN6_SF_BACK_SOLID;
break;
case GL_LINE:
dw2 |= GEN6_SF_BACK_WIREFRAME;
break;
case GL_POINT:
dw2 |= GEN6_SF_BACK_POINT;
break;
default:
assert(0);
break;
}
/* _NEW_SCISSOR */
if (ctx->Scissor.Enabled)
dw3 |= GEN6_SF_SCISSOR_ENABLE;
/* _NEW_POLYGON */
if (ctx->Polygon.CullFlag) {
switch (ctx->Polygon.CullFaceMode) {
case GL_FRONT:
dw3 |= GEN6_SF_CULL_FRONT;
break;
case GL_BACK:
dw3 |= GEN6_SF_CULL_BACK;
break;
case GL_FRONT_AND_BACK:
dw3 |= GEN6_SF_CULL_BOTH;
break;
default:
assert(0);
break;
}
} else {
dw3 |= GEN6_SF_CULL_NONE;
}
/* _NEW_LINE */
{
uint32_t line_width_u3_7 = U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7);
/* TODO: line width of 0 is not allowed when MSAA enabled */
if (line_width_u3_7 == 0)
line_width_u3_7 = 1;
dw3 |= line_width_u3_7 << GEN6_SF_LINE_WIDTH_SHIFT;
}
if (ctx->Line.SmoothFlag) {
dw3 |= GEN6_SF_LINE_AA_ENABLE;
dw3 |= GEN6_SF_LINE_AA_MODE_TRUE;
dw3 |= GEN6_SF_LINE_END_CAP_WIDTH_1_0;
}
/* _NEW_MULTISAMPLE */
if (multisampled_fbo && ctx->Multisample.Enabled)
dw3 |= GEN6_SF_MSRAST_ON_PATTERN;
/* _NEW_PROGRAM | _NEW_POINT */
if (!(ctx->VertexProgram.PointSizeEnabled ||
ctx->Point._Attenuated))
dw4 |= GEN6_SF_USE_STATE_POINT_WIDTH;
/* Clamp to ARB_point_parameters user limits */
point_size = CLAMP(ctx->Point.Size, ctx->Point.MinSize, ctx->Point.MaxSize);
/* Clamp to the hardware limits and convert to fixed point */
dw4 |= U_FIXED(CLAMP(point_size, 0.125, 255.875), 3);
/*
* Window coordinates in an FBO are inverted, which means point
* sprite origin must be inverted, too.
*/
if ((ctx->Point.SpriteOrigin == GL_LOWER_LEFT) != render_to_fbo) {
point_sprite_origin = GEN6_SF_POINT_SPRITE_LOWERLEFT;
} else {
point_sprite_origin = GEN6_SF_POINT_SPRITE_UPPERLEFT;
}
dw1 |= point_sprite_origin;
/* _NEW_LIGHT */
if (ctx->Light.ProvokingVertex != GL_FIRST_VERTEX_CONVENTION) {
dw4 |=
(2 << GEN6_SF_TRI_PROVOKE_SHIFT) |
(2 << GEN6_SF_TRIFAN_PROVOKE_SHIFT) |
(1 << GEN6_SF_LINE_PROVOKE_SHIFT);
} else {
dw4 |=
(1 << GEN6_SF_TRIFAN_PROVOKE_SHIFT);
}
/* Create the mapping from the FS inputs we produce to the VS outputs
* they source from.
*/
uint32_t max_source_attr = 0;
for (; attr < FRAG_ATTRIB_MAX; attr++) {
enum glsl_interp_qualifier interp_qualifier =
brw->fragment_program->InterpQualifier[attr];
bool is_gl_Color = attr == FRAG_ATTRIB_COL0 || attr == FRAG_ATTRIB_COL1;
if (!(brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)))
continue;
/* _NEW_POINT */
if (ctx->Point.PointSprite &&
(attr >= FRAG_ATTRIB_TEX0 && attr <= FRAG_ATTRIB_TEX7) &&
ctx->Point.CoordReplace[attr - FRAG_ATTRIB_TEX0]) {
dw16 |= (1 << input_index);
}
if (attr == FRAG_ATTRIB_PNTC)
dw16 |= (1 << input_index);
/* flat shading */
if (interp_qualifier == INTERP_QUALIFIER_FLAT ||
(shade_model_flat && is_gl_Color &&
interp_qualifier == INTERP_QUALIFIER_NONE))
dw17 |= (1 << input_index);
/* The hardware can only do the overrides on 16 overrides at a
* time, and the other up to 16 have to be lined up so that the
* input index = the output index. We'll need to do some
* tweaking to make sure that's the case.
*/
assert(input_index < 16 || attr == input_index);
/* CACHE_NEW_VS_PROG | _NEW_LIGHT | _NEW_PROGRAM */
attr_overrides[input_index++] =
get_attr_override(&brw->vs.prog_data->vue_map,
urb_entry_read_offset, attr,
ctx->VertexProgram._TwoSideEnabled,
&max_source_attr);
}
for (; input_index < FRAG_ATTRIB_MAX; input_index++)
attr_overrides[input_index] = 0;
/* From the Sandy Bridge PRM, Volume 2, Part 1, documentation for
* 3DSTATE_SF DWord 1 bits 15:11, "Vertex URB Entry Read Length":
*
* "This field should be set to the minimum length required to read the
* maximum source attribute. The maximum source attribute is indicated
* by the maximum value of the enabled Attribute # Source Attribute if
* Attribute Swizzle Enable is set, Number of Output Attributes-1 if
* enable is not set.
* read_length = ceiling((max_source_attr + 1) / 2)
*
* [errata] Corruption/Hang possible if length programmed larger than
* recommended"
*/
uint32_t urb_entry_read_length = ALIGN(max_source_attr + 1, 2) / 2;
dw1 |= urb_entry_read_length << GEN6_SF_URB_ENTRY_READ_LENGTH_SHIFT |
urb_entry_read_offset << GEN6_SF_URB_ENTRY_READ_OFFSET_SHIFT;
BEGIN_BATCH(20);
OUT_BATCH(_3DSTATE_SF << 16 | (20 - 2));
OUT_BATCH(dw1);
OUT_BATCH(dw2);
OUT_BATCH(dw3);
OUT_BATCH(dw4);
OUT_BATCH_F(ctx->Polygon.OffsetUnits * 2); /* constant. copied from gen4 */
OUT_BATCH_F(ctx->Polygon.OffsetFactor); /* scale */
OUT_BATCH_F(0.0); /* XXX: global depth offset clamp */
for (i = 0; i < 8; i++) {
OUT_BATCH(attr_overrides[i * 2] | attr_overrides[i * 2 + 1] << 16);
}
OUT_BATCH(dw16); /* point sprite texcoord bitmask */
OUT_BATCH(dw17); /* constant interp bitmask */
OUT_BATCH(0); /* wrapshortest enables 0-7 */
OUT_BATCH(0); /* wrapshortest enables 8-15 */
ADVANCE_BATCH();
}
static void
upload_sbe_state(struct brw_context *brw)
{
struct intel_context *intel = &brw->intel;
struct gl_context *ctx = &intel->ctx;
/* BRW_NEW_FRAGMENT_PROGRAM */
uint32_t num_outputs = _mesa_bitcount_64(brw->fragment_program->Base.InputsRead);
/* _NEW_LIGHT */
bool shade_model_flat = ctx->Light.ShadeModel == GL_FLAT;
uint32_t dw1, dw10, dw11;
int i;
int attr = 0, input_index = 0;
int urb_entry_read_offset = 1;
uint16_t attr_overrides[VARYING_SLOT_MAX];
/* _NEW_BUFFERS */
bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
uint32_t point_sprite_origin;
/* FINISHME: Attribute Swizzle Control Mode? */
dw1 = GEN7_SBE_SWIZZLE_ENABLE | num_outputs << GEN7_SBE_NUM_OUTPUTS_SHIFT;
/* _NEW_POINT
*
* Window coordinates in an FBO are inverted, which means point
* sprite origin must be inverted.
*/
if ((ctx->Point.SpriteOrigin == GL_LOWER_LEFT) != render_to_fbo) {
point_sprite_origin = GEN6_SF_POINT_SPRITE_LOWERLEFT;
} else {
point_sprite_origin = GEN6_SF_POINT_SPRITE_UPPERLEFT;
}
dw1 |= point_sprite_origin;
dw10 = 0;
dw11 = 0;
/* Create the mapping from the FS inputs we produce to the VS outputs
* they source from.
*/
uint32_t max_source_attr = 0;
for (; attr < VARYING_SLOT_MAX; attr++) {
enum glsl_interp_qualifier interp_qualifier =
brw->fragment_program->InterpQualifier[attr];
bool is_gl_Color = attr == VARYING_SLOT_COL0 || attr == VARYING_SLOT_COL1;
if (!(brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)))
continue;
if (ctx->Point.PointSprite &&
attr >= VARYING_SLOT_TEX0 && attr <= VARYING_SLOT_TEX7 &&
ctx->Point.CoordReplace[attr - VARYING_SLOT_TEX0]) {
dw10 |= (1 << input_index);
}
if (attr == VARYING_SLOT_PNTC)
dw10 |= (1 << input_index);
/* flat shading */
if (interp_qualifier == INTERP_QUALIFIER_FLAT ||
(shade_model_flat && is_gl_Color &&
interp_qualifier == INTERP_QUALIFIER_NONE))
dw11 |= (1 << input_index);
/* The hardware can only do the overrides on 16 overrides at a
* time, and the other up to 16 have to be lined up so that the
* input index = the output index. We'll need to do some
* tweaking to make sure that's the case.
*/
assert(input_index < 16 || attr == input_index);
/* BRW_NEW_VUE_MAP_GEOM_OUT | _NEW_LIGHT | _NEW_PROGRAM */
attr_overrides[input_index++] =
get_attr_override(&brw->vue_map_geom_out,
urb_entry_read_offset, attr,
ctx->VertexProgram._TwoSideEnabled,
&max_source_attr);
}
/* From the Ivy Bridge PRM, Volume 2, Part 1, documentation for
* 3DSTATE_SBE DWord 1 bits 15:11, "Vertex URB Entry Read Length":
*
* "This field should be set to the minimum length required to read the
* maximum source attribute. The maximum source attribute is indicated
* by the maximum value of the enabled Attribute # Source Attribute if
* Attribute Swizzle Enable is set, Number of Output Attributes-1 if
* enable is not set.
*
* read_length = ceiling((max_source_attr + 1) / 2)"
*/
uint32_t urb_entry_read_length = ALIGN(max_source_attr + 1, 2) / 2;
dw1 |= urb_entry_read_length << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT |
urb_entry_read_offset << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT;
for (; input_index < VARYING_SLOT_MAX; input_index++)
attr_overrides[input_index] = 0;
BEGIN_BATCH(14);
OUT_BATCH(_3DSTATE_SBE << 16 | (14 - 2));
OUT_BATCH(dw1);
/* Output dwords 2 through 9 */
for (i = 0; i < 8; i++) {
OUT_BATCH(attr_overrides[i * 2] | attr_overrides[i * 2 + 1] << 16);
}
OUT_BATCH(dw10); /* point sprite texcoord bitmask */
OUT_BATCH(dw11); /* constant interp bitmask */
OUT_BATCH(0); /* wrapshortest enables 0-7 */
OUT_BATCH(0); /* wrapshortest enables 8-15 */
ADVANCE_BATCH();
}
static void
upload_sf_state(struct brw_context *brw)
{
struct intel_context *intel = &brw->intel;
GLcontext *ctx = &intel->ctx;
/* CACHE_NEW_VS_PROG */
uint32_t num_inputs = brw_count_bits(brw->vs.prog_data->outputs_written);
uint32_t num_outputs = brw_count_bits(brw->fragment_program->Base.InputsRead);
uint32_t dw1, dw2, dw3, dw4, dw16;
int i;
/* _NEW_BUFFER */
GLboolean render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0;
int attr = 0;
dw1 =
num_outputs << GEN6_SF_NUM_OUTPUTS_SHIFT |
(num_inputs + 1) / 2 << GEN6_SF_URB_ENTRY_READ_LENGTH_SHIFT |
1 << GEN6_SF_URB_ENTRY_READ_OFFSET_SHIFT;
dw2 = GEN6_SF_VIEWPORT_TRANSFORM_ENABLE |
GEN6_SF_STATISTICS_ENABLE;
dw3 = 0;
dw4 = 0;
dw16 = 0;
/* _NEW_POLYGON */
if ((ctx->Polygon.FrontFace == GL_CCW) ^ render_to_fbo)
dw2 |= GEN6_SF_WINDING_CCW;
if (ctx->Polygon.OffsetFill)
dw2 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_SOLID;
/* _NEW_SCISSOR */
if (ctx->Scissor.Enabled)
dw3 |= GEN6_SF_SCISSOR_ENABLE;
/* _NEW_POLYGON */
if (ctx->Polygon.CullFlag) {
switch (ctx->Polygon.CullFaceMode) {
case GL_FRONT:
dw3 |= GEN6_SF_CULL_FRONT;
break;
case GL_BACK:
dw3 |= GEN6_SF_CULL_BACK;
break;
case GL_FRONT_AND_BACK:
dw3 |= GEN6_SF_CULL_BOTH;
break;
default:
assert(0);
break;
}
} else {
dw3 |= GEN6_SF_CULL_NONE;
}
/* _NEW_LINE */
dw3 |= U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7) <<
GEN6_SF_LINE_WIDTH_SHIFT;
if (ctx->Line.SmoothFlag) {
dw3 |= GEN6_SF_LINE_AA_ENABLE;
dw3 |= GEN6_SF_LINE_AA_MODE_TRUE;
dw3 |= GEN6_SF_LINE_END_CAP_WIDTH_1_0;
}
/* _NEW_POINT */
if (ctx->Point._Attenuated)
dw4 |= GEN6_SF_USE_STATE_POINT_WIDTH;
dw4 |= U_FIXED(CLAMP(ctx->Point.Size, 0.125, 225.875), 3) <<
GEN6_SF_POINT_WIDTH_SHIFT;
if (ctx->Point.SpriteOrigin == GL_LOWER_LEFT)
dw1 |= GEN6_SF_POINT_SPRITE_LOWERLEFT;
/* _NEW_LIGHT */
if (ctx->Light.ProvokingVertex != GL_FIRST_VERTEX_CONVENTION) {
dw4 |=
(2 << GEN6_SF_TRI_PROVOKE_SHIFT) |
(2 << GEN6_SF_TRIFAN_PROVOKE_SHIFT) |
(1 << GEN6_SF_LINE_PROVOKE_SHIFT);
} else {
dw4 |=
(1 << GEN6_SF_TRIFAN_PROVOKE_SHIFT);
}
if (ctx->Point.PointSprite) {
for (i = 0; i < 8; i++) {
if (ctx->Point.CoordReplace[i])
dw16 |= (1 << i);
}
}
BEGIN_BATCH(20);
OUT_BATCH(CMD_3D_SF_STATE << 16 | (20 - 2));
OUT_BATCH(dw1);
OUT_BATCH(dw2);
OUT_BATCH(dw3);
OUT_BATCH(dw4);
OUT_BATCH_F(ctx->Polygon.OffsetUnits * 2); /* constant. copied from gen4 */
OUT_BATCH_F(ctx->Polygon.OffsetFactor); /* scale */
OUT_BATCH_F(0.0); /* XXX: global depth offset clamp */
for (i = 0; i < 8; i++) {
uint32_t attr_overrides = 0;
for (; attr < 64; attr++) {
if (brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)) {
attr_overrides |= get_attr_override(brw, attr);
attr++;
break;
}
}
for (; attr < 64; attr++) {
if (brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)) {
attr_overrides |= get_attr_override(brw, attr) << 16;
attr++;
break;
}
}
OUT_BATCH(attr_overrides);
}
OUT_BATCH(dw16); /* point sprite texcoord bitmask */
OUT_BATCH(0); /* constant interp bitmask */
OUT_BATCH(0); /* wrapshortest enables 0-7 */
OUT_BATCH(0); /* wrapshortest enables 8-15 */
ADVANCE_BATCH();
intel_batchbuffer_emit_mi_flush(intel->batch);
}
