Merge remote-tracking branch 'origin/master' into pipe-video
[mesa.git] / src / mesa / drivers / dri / i965 / gen6_wm_state.c
1 /*
2 * Copyright © 2009 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 * Authors:
24 * Eric Anholt <eric@anholt.net>
25 *
26 */
27
28 #include "brw_context.h"
29 #include "brw_state.h"
30 #include "brw_defines.h"
31 #include "brw_util.h"
32 #include "brw_wm.h"
33 #include "program/prog_parameter.h"
34 #include "program/prog_statevars.h"
35 #include "intel_batchbuffer.h"
36
37 static void
38 gen6_prepare_wm_push_constants(struct brw_context *brw)
39 {
40 struct intel_context *intel = &brw->intel;
41 struct gl_context *ctx = &intel->ctx;
42 const struct brw_fragment_program *fp =
43 brw_fragment_program_const(brw->fragment_program);
44
45 /* Updates the ParameterValues[i] pointers for all parameters of the
46 * basic type of PROGRAM_STATE_VAR.
47 */
48 /* XXX: Should this happen somewhere before to get our state flag set? */
49 _mesa_load_state_parameters(ctx, fp->program.Base.Parameters);
50
51 if (brw->wm.prog_data->nr_params != 0) {
52 float *constants;
53 unsigned int i;
54
55 constants = brw_state_batch(brw,
56 brw->wm.prog_data->nr_params *
57 sizeof(float),
58 32, &brw->wm.push_const_offset);
59
60 for (i = 0; i < brw->wm.prog_data->nr_params; i++) {
61 constants[i] = convert_param(brw->wm.prog_data->param_convert[i],
62 *brw->wm.prog_data->param[i]);
63 }
64
65 if (0) {
66 printf("WM constants:\n");
67 for (i = 0; i < brw->wm.prog_data->nr_params; i++) {
68 if ((i & 7) == 0)
69 printf("g%d: ", brw->wm.prog_data->first_curbe_grf + i / 8);
70 printf("%8f ", constants[i]);
71 if ((i & 7) == 7)
72 printf("\n");
73 }
74 if ((i & 7) != 0)
75 printf("\n");
76 printf("\n");
77 }
78 }
79 }
80
81 const struct brw_tracked_state gen6_wm_constants = {
82 .dirty = {
83 .mesa = _NEW_PROGRAM_CONSTANTS,
84 .brw = (BRW_NEW_BATCH |
85 BRW_NEW_FRAGMENT_PROGRAM),
86 .cache = 0,
87 },
88 .prepare = gen6_prepare_wm_push_constants,
89 };
90
91 static void
92 upload_wm_state(struct brw_context *brw)
93 {
94 struct intel_context *intel = &brw->intel;
95 struct gl_context *ctx = &intel->ctx;
96 const struct brw_fragment_program *fp =
97 brw_fragment_program_const(brw->fragment_program);
98 uint32_t dw2, dw4, dw5, dw6;
99
100 /* CACHE_NEW_WM_PROG */
101 if (brw->wm.prog_data->nr_params == 0) {
102 /* Disable the push constant buffers. */
103 BEGIN_BATCH(5);
104 OUT_BATCH(_3DSTATE_CONSTANT_PS << 16 | (5 - 2));
105 OUT_BATCH(0);
106 OUT_BATCH(0);
107 OUT_BATCH(0);
108 OUT_BATCH(0);
109 ADVANCE_BATCH();
110 } else {
111 BEGIN_BATCH(5);
112 OUT_BATCH(_3DSTATE_CONSTANT_PS << 16 |
113 GEN6_CONSTANT_BUFFER_0_ENABLE |
114 (5 - 2));
115 /* Pointer to the WM constant buffer. Covered by the set of
116 * state flags from gen6_prepare_wm_constants
117 */
118 OUT_BATCH(brw->wm.push_const_offset +
119 ALIGN(brw->wm.prog_data->nr_params,
120 brw->wm.prog_data->dispatch_width) / 8 - 1);
121 OUT_BATCH(0);
122 OUT_BATCH(0);
123 OUT_BATCH(0);
124 ADVANCE_BATCH();
125 }
126
127 dw2 = dw4 = dw5 = dw6 = 0;
128 dw4 |= GEN6_WM_STATISTICS_ENABLE;
129 dw5 |= GEN6_WM_LINE_AA_WIDTH_1_0;
130 dw5 |= GEN6_WM_LINE_END_CAP_AA_WIDTH_0_5;
131
132 /* OpenGL non-ieee floating point mode */
133 dw2 |= GEN6_WM_FLOATING_POINT_MODE_ALT;
134
135 /* BRW_NEW_NR_WM_SURFACES */
136 dw2 |= brw->wm.nr_surfaces << GEN6_WM_BINDING_TABLE_ENTRY_COUNT_SHIFT;
137
138 /* CACHE_NEW_SAMPLER */
139 dw2 |= (ALIGN(brw->wm.sampler_count, 4) / 4) << GEN6_WM_SAMPLER_COUNT_SHIFT;
140 dw4 |= (brw->wm.prog_data->first_curbe_grf <<
141 GEN6_WM_DISPATCH_START_GRF_SHIFT_0);
142 dw4 |= (brw->wm.prog_data->first_curbe_grf_16 <<
143 GEN6_WM_DISPATCH_START_GRF_SHIFT_2);
144
145 dw5 |= (brw->wm_max_threads - 1) << GEN6_WM_MAX_THREADS_SHIFT;
146
147 /* CACHE_NEW_WM_PROG */
148 if (brw->wm.prog_data->dispatch_width == 8) {
149 dw5 |= GEN6_WM_8_DISPATCH_ENABLE;
150 if (brw->wm.prog_data->prog_offset_16)
151 dw5 |= GEN6_WM_16_DISPATCH_ENABLE;
152 } else {
153 dw5 |= GEN6_WM_16_DISPATCH_ENABLE;
154 }
155
156 /* _NEW_LINE */
157 if (ctx->Line.StippleFlag)
158 dw5 |= GEN6_WM_LINE_STIPPLE_ENABLE;
159
160 /* _NEW_POLYGONSTIPPLE */
161 if (ctx->Polygon.StippleFlag)
162 dw5 |= GEN6_WM_POLYGON_STIPPLE_ENABLE;
163
164 /* BRW_NEW_FRAGMENT_PROGRAM */
165 if (fp->program.Base.InputsRead & (1 << FRAG_ATTRIB_WPOS))
166 dw5 |= GEN6_WM_USES_SOURCE_DEPTH | GEN6_WM_USES_SOURCE_W;
167 if (fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH))
168 dw5 |= GEN6_WM_COMPUTED_DEPTH;
169
170 /* _NEW_COLOR */
171 if (fp->program.UsesKill || ctx->Color.AlphaEnabled)
172 dw5 |= GEN6_WM_KILL_ENABLE;
173
174 if (brw_color_buffer_write_enabled(brw) ||
175 dw5 & (GEN6_WM_KILL_ENABLE | GEN6_WM_COMPUTED_DEPTH)) {
176 dw5 |= GEN6_WM_DISPATCH_ENABLE;
177 }
178
179 dw6 |= GEN6_WM_PERSPECTIVE_PIXEL_BARYCENTRIC;
180
181 dw6 |= brw_count_bits(brw->fragment_program->Base.InputsRead) <<
182 GEN6_WM_NUM_SF_OUTPUTS_SHIFT;
183
184 BEGIN_BATCH(9);
185 OUT_BATCH(_3DSTATE_WM << 16 | (9 - 2));
186 OUT_RELOC(brw->wm.prog_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
187 OUT_BATCH(dw2);
188 if (brw->wm.prog_data->total_scratch) {
189 OUT_RELOC(brw->wm.scratch_bo, I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
190 ffs(brw->wm.prog_data->total_scratch) - 11);
191 } else {
192 OUT_BATCH(0);
193 }
194 OUT_BATCH(dw4);
195 OUT_BATCH(dw5);
196 OUT_BATCH(dw6);
197 OUT_BATCH(0); /* kernel 1 pointer */
198 if (brw->wm.prog_data->prog_offset_16) {
199 OUT_RELOC(brw->wm.prog_bo, I915_GEM_DOMAIN_INSTRUCTION, 0,
200 brw->wm.prog_data->prog_offset_16);
201 } else {
202 OUT_BATCH(0); /* kernel 2 pointer */
203 }
204 ADVANCE_BATCH();
205 }
206
207 const struct brw_tracked_state gen6_wm_state = {
208 .dirty = {
209 .mesa = (_NEW_LINE | _NEW_POLYGONSTIPPLE | _NEW_COLOR | _NEW_BUFFERS |
210 _NEW_PROGRAM_CONSTANTS | _NEW_POLYGON),
211 .brw = (BRW_NEW_CURBE_OFFSETS |
212 BRW_NEW_FRAGMENT_PROGRAM |
213 BRW_NEW_NR_WM_SURFACES |
214 BRW_NEW_URB_FENCE |
215 BRW_NEW_BATCH),
216 .cache = (CACHE_NEW_SAMPLER |
217 CACHE_NEW_WM_PROG)
218 },
219 .emit = upload_wm_state,
220 };