intel: Add a batch flush between front-buffer downsample and X protocol.
[mesa.git] / src / mesa / drivers / dri / i965 / gen7_sf_state.c
1 /*
2 * Copyright © 2011 Intel Corporation
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
21 * IN THE SOFTWARE.
22 */
23
24 #include "brw_context.h"
25 #include "brw_state.h"
26 #include "brw_defines.h"
27 #include "brw_util.h"
28 #include "main/macros.h"
29 #include "main/fbobject.h"
30 #include "intel_batchbuffer.h"
31
32 static void
33 upload_sbe_state(struct brw_context *brw)
34 {
35 struct gl_context *ctx = &brw->ctx;
36 /* BRW_NEW_FRAGMENT_PROGRAM */
37 uint32_t num_outputs = _mesa_bitcount_64(brw->fragment_program->Base.InputsRead);
38 /* _NEW_LIGHT */
39 bool shade_model_flat = ctx->Light.ShadeModel == GL_FLAT;
40 uint32_t dw1, dw10, dw11;
41 int i;
42 int attr = 0, input_index = 0;
43 int urb_entry_read_offset = 1;
44 uint16_t attr_overrides[VARYING_SLOT_MAX];
45 /* _NEW_BUFFERS */
46 bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
47 uint32_t point_sprite_origin;
48
49 /* FINISHME: Attribute Swizzle Control Mode? */
50 dw1 = GEN7_SBE_SWIZZLE_ENABLE | num_outputs << GEN7_SBE_NUM_OUTPUTS_SHIFT;
51
52 /* _NEW_POINT
53 *
54 * Window coordinates in an FBO are inverted, which means point
55 * sprite origin must be inverted.
56 */
57 if ((ctx->Point.SpriteOrigin == GL_LOWER_LEFT) != render_to_fbo) {
58 point_sprite_origin = GEN6_SF_POINT_SPRITE_LOWERLEFT;
59 } else {
60 point_sprite_origin = GEN6_SF_POINT_SPRITE_UPPERLEFT;
61 }
62 dw1 |= point_sprite_origin;
63
64
65 dw10 = 0;
66 dw11 = 0;
67
68 /* Create the mapping from the FS inputs we produce to the VS outputs
69 * they source from.
70 */
71 uint32_t max_source_attr = 0;
72 for (; attr < VARYING_SLOT_MAX; attr++) {
73 enum glsl_interp_qualifier interp_qualifier =
74 brw->fragment_program->InterpQualifier[attr];
75 bool is_gl_Color = attr == VARYING_SLOT_COL0 || attr == VARYING_SLOT_COL1;
76
77 if (!(brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)))
78 continue;
79
80 if (ctx->Point.PointSprite &&
81 attr >= VARYING_SLOT_TEX0 && attr <= VARYING_SLOT_TEX7 &&
82 ctx->Point.CoordReplace[attr - VARYING_SLOT_TEX0]) {
83 dw10 |= (1 << input_index);
84 }
85
86 if (attr == VARYING_SLOT_PNTC)
87 dw10 |= (1 << input_index);
88
89 /* flat shading */
90 if (interp_qualifier == INTERP_QUALIFIER_FLAT ||
91 (shade_model_flat && is_gl_Color &&
92 interp_qualifier == INTERP_QUALIFIER_NONE))
93 dw11 |= (1 << input_index);
94
95 /* The hardware can only do the overrides on 16 overrides at a
96 * time, and the other up to 16 have to be lined up so that the
97 * input index = the output index. We'll need to do some
98 * tweaking to make sure that's the case.
99 */
100 assert(input_index < 16 || attr == input_index);
101
102 /* BRW_NEW_VUE_MAP_GEOM_OUT | _NEW_LIGHT | _NEW_PROGRAM */
103 attr_overrides[input_index++] =
104 get_attr_override(&brw->vue_map_geom_out,
105 urb_entry_read_offset, attr,
106 ctx->VertexProgram._TwoSideEnabled,
107 &max_source_attr);
108 }
109
110 /* From the Ivy Bridge PRM, Volume 2, Part 1, documentation for
111 * 3DSTATE_SBE DWord 1 bits 15:11, "Vertex URB Entry Read Length":
112 *
113 * "This field should be set to the minimum length required to read the
114 * maximum source attribute. The maximum source attribute is indicated
115 * by the maximum value of the enabled Attribute # Source Attribute if
116 * Attribute Swizzle Enable is set, Number of Output Attributes-1 if
117 * enable is not set.
118 *
119 * read_length = ceiling((max_source_attr + 1) / 2)"
120 */
121 uint32_t urb_entry_read_length = ALIGN(max_source_attr + 1, 2) / 2;
122 dw1 |= urb_entry_read_length << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT |
123 urb_entry_read_offset << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT;
124
125 for (; input_index < VARYING_SLOT_MAX; input_index++)
126 attr_overrides[input_index] = 0;
127
128 BEGIN_BATCH(14);
129 OUT_BATCH(_3DSTATE_SBE << 16 | (14 - 2));
130 OUT_BATCH(dw1);
131
132 /* Output dwords 2 through 9 */
133 for (i = 0; i < 8; i++) {
134 OUT_BATCH(attr_overrides[i * 2] | attr_overrides[i * 2 + 1] << 16);
135 }
136
137 OUT_BATCH(dw10); /* point sprite texcoord bitmask */
138 OUT_BATCH(dw11); /* constant interp bitmask */
139 OUT_BATCH(0); /* wrapshortest enables 0-7 */
140 OUT_BATCH(0); /* wrapshortest enables 8-15 */
141 ADVANCE_BATCH();
142 }
143
144 const struct brw_tracked_state gen7_sbe_state = {
145 .dirty = {
146 .mesa = (_NEW_BUFFERS |
147 _NEW_LIGHT |
148 _NEW_POINT |
149 _NEW_PROGRAM),
150 .brw = (BRW_NEW_CONTEXT |
151 BRW_NEW_FRAGMENT_PROGRAM |
152 BRW_NEW_VUE_MAP_GEOM_OUT)
153 },
154 .emit = upload_sbe_state,
155 };
156
157 static void
158 upload_sf_state(struct brw_context *brw)
159 {
160 struct gl_context *ctx = &brw->ctx;
161 uint32_t dw1, dw2, dw3;
162 float point_size;
163 /* _NEW_BUFFERS */
164 bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
165 bool multisampled_fbo = ctx->DrawBuffer->Visual.samples > 1;
166
167 dw1 = GEN6_SF_STATISTICS_ENABLE |
168 GEN6_SF_VIEWPORT_TRANSFORM_ENABLE;
169
170 /* _NEW_BUFFERS */
171 dw1 |= (brw_depthbuffer_format(brw) << GEN7_SF_DEPTH_BUFFER_SURFACE_FORMAT_SHIFT);
172
173 /* _NEW_POLYGON */
174 if ((ctx->Polygon.FrontFace == GL_CCW) ^ render_to_fbo)
175 dw1 |= GEN6_SF_WINDING_CCW;
176
177 if (ctx->Polygon.OffsetFill)
178 dw1 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_SOLID;
179
180 if (ctx->Polygon.OffsetLine)
181 dw1 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_WIREFRAME;
182
183 if (ctx->Polygon.OffsetPoint)
184 dw1 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_POINT;
185
186 switch (ctx->Polygon.FrontMode) {
187 case GL_FILL:
188 dw1 |= GEN6_SF_FRONT_SOLID;
189 break;
190
191 case GL_LINE:
192 dw1 |= GEN6_SF_FRONT_WIREFRAME;
193 break;
194
195 case GL_POINT:
196 dw1 |= GEN6_SF_FRONT_POINT;
197 break;
198
199 default:
200 assert(0);
201 break;
202 }
203
204 switch (ctx->Polygon.BackMode) {
205 case GL_FILL:
206 dw1 |= GEN6_SF_BACK_SOLID;
207 break;
208
209 case GL_LINE:
210 dw1 |= GEN6_SF_BACK_WIREFRAME;
211 break;
212
213 case GL_POINT:
214 dw1 |= GEN6_SF_BACK_POINT;
215 break;
216
217 default:
218 assert(0);
219 break;
220 }
221
222 dw2 = 0;
223
224 if (ctx->Polygon.CullFlag) {
225 switch (ctx->Polygon.CullFaceMode) {
226 case GL_FRONT:
227 dw2 |= GEN6_SF_CULL_FRONT;
228 break;
229 case GL_BACK:
230 dw2 |= GEN6_SF_CULL_BACK;
231 break;
232 case GL_FRONT_AND_BACK:
233 dw2 |= GEN6_SF_CULL_BOTH;
234 break;
235 default:
236 assert(0);
237 break;
238 }
239 } else {
240 dw2 |= GEN6_SF_CULL_NONE;
241 }
242
243 /* _NEW_SCISSOR */
244 if (ctx->Scissor.Enabled)
245 dw2 |= GEN6_SF_SCISSOR_ENABLE;
246
247 /* _NEW_LINE */
248 {
249 uint32_t line_width_u3_7 = U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7);
250 /* TODO: line width of 0 is not allowed when MSAA enabled */
251 if (line_width_u3_7 == 0)
252 line_width_u3_7 = 1;
253 dw2 |= line_width_u3_7 << GEN6_SF_LINE_WIDTH_SHIFT;
254 }
255 if (ctx->Line.SmoothFlag) {
256 dw2 |= GEN6_SF_LINE_AA_ENABLE;
257 dw2 |= GEN6_SF_LINE_END_CAP_WIDTH_1_0;
258 }
259 if (ctx->Line.StippleFlag && brw->is_haswell) {
260 dw2 |= HSW_SF_LINE_STIPPLE_ENABLE;
261 }
262 /* _NEW_MULTISAMPLE */
263 if (multisampled_fbo && ctx->Multisample.Enabled)
264 dw2 |= GEN6_SF_MSRAST_ON_PATTERN;
265
266 /* FINISHME: Last Pixel Enable? Vertex Sub Pixel Precision Select?
267 */
268
269 dw3 = GEN6_SF_LINE_AA_MODE_TRUE;
270
271 /* _NEW_PROGRAM | _NEW_POINT */
272 if (!(ctx->VertexProgram.PointSizeEnabled || ctx->Point._Attenuated))
273 dw3 |= GEN6_SF_USE_STATE_POINT_WIDTH;
274
275 /* Clamp to ARB_point_parameters user limits */
276 point_size = CLAMP(ctx->Point.Size, ctx->Point.MinSize, ctx->Point.MaxSize);
277
278 /* Clamp to the hardware limits and convert to fixed point */
279 dw3 |= U_FIXED(CLAMP(point_size, 0.125, 255.875), 3);
280
281 /* _NEW_LIGHT */
282 if (ctx->Light.ProvokingVertex != GL_FIRST_VERTEX_CONVENTION) {
283 dw3 |=
284 (2 << GEN6_SF_TRI_PROVOKE_SHIFT) |
285 (2 << GEN6_SF_TRIFAN_PROVOKE_SHIFT) |
286 (1 << GEN6_SF_LINE_PROVOKE_SHIFT);
287 } else {
288 dw3 |= (1 << GEN6_SF_TRIFAN_PROVOKE_SHIFT);
289 }
290
291 BEGIN_BATCH(7);
292 OUT_BATCH(_3DSTATE_SF << 16 | (7 - 2));
293 OUT_BATCH(dw1);
294 OUT_BATCH(dw2);
295 OUT_BATCH(dw3);
296 OUT_BATCH_F(ctx->Polygon.OffsetUnits * 2); /* constant. copied from gen4 */
297 OUT_BATCH_F(ctx->Polygon.OffsetFactor); /* scale */
298 OUT_BATCH_F(0.0); /* XXX: global depth offset clamp */
299 ADVANCE_BATCH();
300 }
301
302 const struct brw_tracked_state gen7_sf_state = {
303 .dirty = {
304 .mesa = (_NEW_LIGHT |
305 _NEW_PROGRAM |
306 _NEW_POLYGON |
307 _NEW_LINE |
308 _NEW_SCISSOR |
309 _NEW_BUFFERS |
310 _NEW_POINT |
311 _NEW_MULTISAMPLE),
312 .brw = BRW_NEW_CONTEXT,
313 .cache = CACHE_NEW_VS_PROG
314 },
315 .emit = upload_sf_state,
316 };