#include "brw_defines.h"
#include "brw_util.h"
#include "main/macros.h"
+#include "main/fbobject.h"
#include "intel_batchbuffer.h"
static void
upload_sbe_state(struct brw_context *brw)
{
- struct intel_context *intel = &brw->intel;
- struct gl_context *ctx = &intel->ctx;
- struct brw_vue_map vue_map;
- uint32_t urb_entry_read_length;
- /* CACHE_NEW_VS_PROG */
- GLbitfield64 vs_outputs_written = brw->vs.prog_data->outputs_written;
- /* 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;
+ struct gl_context *ctx = &brw->ctx;
+ /* BRW_NEW_FS_PROG_DATA */
+ uint32_t num_outputs = brw->wm.prog_data->num_varying_inputs;
+ uint32_t dw1;
+ uint32_t point_sprite_enables;
+ uint32_t flat_enables;
int i;
- int attr = 0, input_index = 0;
- /* _NEW_TRANSFORM */
- int urb_entry_read_offset = 1;
- bool userclip_active = (ctx->Transform.ClipPlanesEnabled != 0);
- uint16_t attr_overrides[FRAG_ATTRIB_MAX];
-
- brw_compute_vue_map(&vue_map, intel, userclip_active, vs_outputs_written);
- urb_entry_read_length = (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;
- }
+ const int urb_entry_read_offset = BRW_SF_URB_ENTRY_READ_OFFSET;
+ uint16_t attr_overrides[16];
+ /* _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 |
- urb_entry_read_length << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT |
- urb_entry_read_offset << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT;
-
- /* _NEW_POINT */
- if (ctx->Point.SpriteOrigin == GL_LOWER_LEFT)
- dw1 |= GEN6_SF_POINT_SPRITE_LOWERLEFT;
+ dw1 = GEN7_SBE_SWIZZLE_ENABLE | num_outputs << GEN7_SBE_NUM_OUTPUTS_SHIFT;
- dw10 = 0;
- dw11 = 0;
-
- /* Create the mapping from the FS inputs we produce to the VS outputs
- * they source from.
+ /* _NEW_POINT
+ *
+ * Window coordinates in an FBO are inverted, which means point
+ * sprite origin must be inverted.
*/
- 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);
-
- /* _NEW_LIGHT | _NEW_PROGRAM */
- attr_overrides[input_index++] =
- get_attr_override(&vue_map, urb_entry_read_offset, attr,
- ctx->VertexProgram._TwoSideEnabled);
+ 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;
- for (; attr < FRAG_ATTRIB_MAX; attr++)
- attr_overrides[input_index++] = 0;
+ /* BRW_NEW_VUE_MAP_GEOM_OUT | BRW_NEW_FRAGMENT_PROGRAM
+ * _NEW_POINT | _NEW_LIGHT | _NEW_PROGRAM | BRW_NEW_FS_PROG_DATA
+ */
+ uint32_t urb_entry_read_length;
+ calculate_attr_overrides(brw, attr_overrides, &point_sprite_enables,
+ &flat_enables, &urb_entry_read_length);
+ dw1 |= urb_entry_read_length << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT |
+ urb_entry_read_offset << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT;
BEGIN_BATCH(14);
OUT_BATCH(_3DSTATE_SBE << 16 | (14 - 2));
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(point_sprite_enables); /* dw10 */
+ OUT_BATCH(flat_enables);
OUT_BATCH(0); /* wrapshortest enables 0-7 */
OUT_BATCH(0); /* wrapshortest enables 8-15 */
ADVANCE_BATCH();
const struct brw_tracked_state gen7_sbe_state = {
.dirty = {
- .mesa = (_NEW_LIGHT |
- _NEW_POINT |
- _NEW_PROGRAM |
- _NEW_TRANSFORM),
- .brw = (BRW_NEW_CONTEXT |
- BRW_NEW_FRAGMENT_PROGRAM),
- .cache = CACHE_NEW_VS_PROG
+ .mesa = _NEW_BUFFERS |
+ _NEW_LIGHT |
+ _NEW_POINT |
+ _NEW_PROGRAM,
+ .brw = BRW_NEW_CONTEXT |
+ BRW_NEW_FRAGMENT_PROGRAM |
+ BRW_NEW_FS_PROG_DATA |
+ BRW_NEW_GEOMETRY_PROGRAM |
+ BRW_NEW_PRIMITIVE |
+ BRW_NEW_VUE_MAP_GEOM_OUT,
},
.emit = upload_sbe_state,
};
static void
upload_sf_state(struct brw_context *brw)
{
- struct intel_context *intel = &brw->intel;
- struct gl_context *ctx = &intel->ctx;
+ struct gl_context *ctx = &brw->ctx;
uint32_t dw1, dw2, dw3;
float point_size;
/* _NEW_BUFFERS */
- bool render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0;
+ bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
+ bool multisampled_fbo = ctx->DrawBuffer->Visual.samples > 1;
+
+ dw1 = GEN6_SF_STATISTICS_ENABLE;
- dw1 = GEN6_SF_STATISTICS_ENABLE | GEN6_SF_VIEWPORT_TRANSFORM_ENABLE;
+ if (brw->sf.viewport_transform_enable)
+ dw1 |= GEN6_SF_VIEWPORT_TRANSFORM_ENABLE;
/* _NEW_BUFFERS */
dw1 |= (brw_depthbuffer_format(brw) << GEN7_SF_DEPTH_BUFFER_SURFACE_FORMAT_SHIFT);
break;
default:
- assert(0);
- break;
+ unreachable("not reached");
}
switch (ctx->Polygon.BackMode) {
break;
default:
- assert(0);
- break;
+ unreachable("not reached");
}
dw2 = 0;
dw2 |= GEN6_SF_CULL_BOTH;
break;
default:
- assert(0);
- break;
+ unreachable("not reached");
}
} else {
dw2 |= GEN6_SF_CULL_NONE;
}
/* _NEW_SCISSOR */
- if (ctx->Scissor.Enabled)
+ if (ctx->Scissor.EnableFlags)
dw2 |= GEN6_SF_SCISSOR_ENABLE;
/* _NEW_LINE */
- dw2 |= U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7) <<
- GEN6_SF_LINE_WIDTH_SHIFT;
+ {
+ /* OpenGL dictates that line width should be rounded to the nearest
+ * integer
+ */
+ float line_width =
+ roundf(CLAMP(ctx->Line.Width, 0.0, ctx->Const.MaxLineWidth));
+ uint32_t line_width_u3_7 = U_FIXED(line_width, 7);
+ /* TODO: line width of 0 is not allowed when MSAA enabled */
+ if (line_width_u3_7 == 0)
+ line_width_u3_7 = 1;
+ dw2 |= line_width_u3_7 << GEN6_SF_LINE_WIDTH_SHIFT;
+ }
if (ctx->Line.SmoothFlag) {
dw2 |= GEN6_SF_LINE_AA_ENABLE;
- dw2 |= GEN6_SF_LINE_AA_MODE_TRUE;
dw2 |= GEN6_SF_LINE_END_CAP_WIDTH_1_0;
}
+ if (ctx->Line.StippleFlag && brw->is_haswell) {
+ dw2 |= HSW_SF_LINE_STIPPLE_ENABLE;
+ }
+ /* _NEW_MULTISAMPLE */
+ if (multisampled_fbo && ctx->Multisample.Enabled)
+ dw2 |= GEN6_SF_MSRAST_ON_PATTERN;
/* FINISHME: Last Pixel Enable? Vertex Sub Pixel Precision Select?
- * FINISHME: AA Line Distance Mode?
*/
- dw3 = 0;
+ dw3 = GEN6_SF_LINE_AA_MODE_TRUE;
- /* _NEW_POINT */
+ /* _NEW_PROGRAM | _NEW_POINT */
if (!(ctx->VertexProgram.PointSizeEnabled || ctx->Point._Attenuated))
dw3 |= GEN6_SF_USE_STATE_POINT_WIDTH;
OUT_BATCH(dw3);
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 */
+ OUT_BATCH_F(ctx->Polygon.OffsetClamp); /* global depth offset clamp */
ADVANCE_BATCH();
}
const struct brw_tracked_state gen7_sf_state = {
.dirty = {
- .mesa = (_NEW_LIGHT |
- _NEW_PROGRAM |
- _NEW_POLYGON |
- _NEW_LINE |
- _NEW_SCISSOR |
- _NEW_BUFFERS |
- _NEW_POINT),
- .brw = (BRW_NEW_CONTEXT),
- .cache = CACHE_NEW_VS_PROG
+ .mesa = _NEW_BUFFERS |
+ _NEW_LIGHT |
+ _NEW_LINE |
+ _NEW_MULTISAMPLE |
+ _NEW_POINT |
+ _NEW_POLYGON |
+ _NEW_PROGRAM |
+ _NEW_SCISSOR,
+ .brw = BRW_NEW_CONTEXT,
},
.emit = upload_sf_state,
};