2a0462f3fc969b9482af9f206274f245adf9eade
[mesa.git] / src / mesa / drivers / dri / i965 / gen7_wm_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 <stdbool.h>
25 #include "brw_context.h"
26 #include "brw_state.h"
27 #include "brw_defines.h"
28 #include "brw_util.h"
29 #include "brw_wm.h"
30 #include "program/prog_parameter.h"
31 #include "program/prog_statevars.h"
32 #include "intel_batchbuffer.h"
33
34 static void
35 upload_wm_state(struct brw_context *brw)
36 {
37 struct intel_context *intel = &brw->intel;
38 struct gl_context *ctx = &intel->ctx;
39 const struct brw_fragment_program *fp =
40 brw_fragment_program_const(brw->fragment_program);
41 bool writes_depth = false;
42 bool multisampled = false;
43 uint32_t dw1, dw2;
44
45 /* _NEW_BUFFERS */
46 if (ctx->DrawBuffer->_ColorDrawBuffers[0])
47 multisampled = ctx->DrawBuffer->_ColorDrawBuffers[0]->NumSamples > 0;
48
49 dw1 = dw2 = 0;
50 dw1 |= GEN7_WM_STATISTICS_ENABLE;
51 dw1 |= GEN7_WM_LINE_AA_WIDTH_1_0;
52 dw1 |= GEN7_WM_LINE_END_CAP_AA_WIDTH_0_5;
53
54 /* _NEW_LINE */
55 if (ctx->Line.StippleFlag)
56 dw1 |= GEN7_WM_LINE_STIPPLE_ENABLE;
57
58 /* _NEW_POLYGON */
59 if (ctx->Polygon.StippleFlag)
60 dw1 |= GEN7_WM_POLYGON_STIPPLE_ENABLE;
61
62 /* BRW_NEW_FRAGMENT_PROGRAM */
63 if (fp->program.Base.InputsRead & FRAG_BIT_WPOS)
64 dw1 |= GEN7_WM_USES_SOURCE_DEPTH | GEN7_WM_USES_SOURCE_W;
65 if (fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
66 writes_depth = true;
67 dw1 |= GEN7_WM_PSCDEPTH_ON;
68 }
69 /* CACHE_NEW_WM_PROG */
70 dw1 |= brw->wm.prog_data->barycentric_interp_modes <<
71 GEN7_WM_BARYCENTRIC_INTERPOLATION_MODE_SHIFT;
72
73 /* _NEW_COLOR */
74 if (fp->program.UsesKill || ctx->Color.AlphaEnabled)
75 dw1 |= GEN7_WM_KILL_ENABLE;
76
77 /* _NEW_BUFFERS */
78 if (brw_color_buffer_write_enabled(brw) || writes_depth ||
79 dw1 & GEN7_WM_KILL_ENABLE) {
80 dw1 |= GEN7_WM_DISPATCH_ENABLE;
81 }
82 if (multisampled) {
83 dw1 |= GEN7_WM_MSRAST_ON_PATTERN;
84 dw2 |= GEN7_WM_MSDISPMODE_PERPIXEL;
85 } else {
86 dw1 |= GEN7_WM_MSRAST_OFF_PIXEL;
87 dw2 |= GEN7_WM_MSDISPMODE_PERSAMPLE;
88 }
89
90 BEGIN_BATCH(3);
91 OUT_BATCH(_3DSTATE_WM << 16 | (3 - 2));
92 OUT_BATCH(dw1);
93 OUT_BATCH(dw2);
94 ADVANCE_BATCH();
95 }
96
97 const struct brw_tracked_state gen7_wm_state = {
98 .dirty = {
99 .mesa = (_NEW_LINE | _NEW_POLYGON |
100 _NEW_COLOR | _NEW_BUFFERS),
101 .brw = (BRW_NEW_FRAGMENT_PROGRAM |
102 BRW_NEW_BATCH),
103 .cache = CACHE_NEW_WM_PROG,
104 },
105 .emit = upload_wm_state,
106 };
107
108 static void
109 upload_ps_state(struct brw_context *brw)
110 {
111 struct intel_context *intel = &brw->intel;
112 uint32_t dw2, dw4, dw5;
113 const int max_threads_shift = brw->intel.is_haswell ?
114 HSW_PS_MAX_THREADS_SHIFT : IVB_PS_MAX_THREADS_SHIFT;
115
116 /* BRW_NEW_PS_BINDING_TABLE */
117 BEGIN_BATCH(2);
118 OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_PS << 16 | (2 - 2));
119 OUT_BATCH(brw->wm.bind_bo_offset);
120 ADVANCE_BATCH();
121
122 /* CACHE_NEW_SAMPLER */
123 BEGIN_BATCH(2);
124 OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_PS << 16 | (2 - 2));
125 OUT_BATCH(brw->sampler.offset);
126 ADVANCE_BATCH();
127
128 /* CACHE_NEW_WM_PROG */
129 if (brw->wm.prog_data->nr_params == 0) {
130 /* Disable the push constant buffers. */
131 BEGIN_BATCH(7);
132 OUT_BATCH(_3DSTATE_CONSTANT_PS << 16 | (7 - 2));
133 OUT_BATCH(0);
134 OUT_BATCH(0);
135 OUT_BATCH(0);
136 OUT_BATCH(0);
137 OUT_BATCH(0);
138 OUT_BATCH(0);
139 ADVANCE_BATCH();
140 } else {
141 BEGIN_BATCH(7);
142 OUT_BATCH(_3DSTATE_CONSTANT_PS << 16 | (7 - 2));
143
144 OUT_BATCH(ALIGN(brw->wm.prog_data->nr_params,
145 brw->wm.prog_data->dispatch_width) / 8);
146 OUT_BATCH(0);
147 /* Pointer to the WM constant buffer. Covered by the set of
148 * state flags from gen6_upload_wm_push_constants.
149 */
150 OUT_BATCH(brw->wm.push_const_offset);
151 OUT_BATCH(0);
152 OUT_BATCH(0);
153 OUT_BATCH(0);
154 ADVANCE_BATCH();
155 }
156
157 dw2 = dw4 = dw5 = 0;
158
159 /* CACHE_NEW_SAMPLER */
160 dw2 |= (ALIGN(brw->sampler.count, 4) / 4) << GEN7_PS_SAMPLER_COUNT_SHIFT;
161
162 /* Use ALT floating point mode for ARB fragment programs, because they
163 * require 0^0 == 1. Even though _CurrentFragmentProgram is used for
164 * rendering, CurrentFragmentProgram is used for this check to
165 * differentiate between the GLSL and non-GLSL cases.
166 */
167 if (intel->ctx.Shader.CurrentFragmentProgram == NULL)
168 dw2 |= GEN7_PS_FLOATING_POINT_MODE_ALT;
169
170 if (intel->is_haswell)
171 dw4 |= SET_FIELD(1, HSW_PS_SAMPLE_MASK); /* 1 sample for now */
172
173 dw4 |= (brw->max_wm_threads - 1) << max_threads_shift;
174
175 /* CACHE_NEW_WM_PROG */
176 if (brw->wm.prog_data->nr_params > 0)
177 dw4 |= GEN7_PS_PUSH_CONSTANT_ENABLE;
178
179 /* BRW_NEW_FRAGMENT_PROGRAM */
180 if (brw->fragment_program->Base.InputsRead != 0)
181 dw4 |= GEN7_PS_ATTRIBUTE_ENABLE;
182
183 if (brw->wm.prog_data->dispatch_width == 8) {
184 dw4 |= GEN7_PS_8_DISPATCH_ENABLE;
185 if (brw->wm.prog_data->prog_offset_16)
186 dw4 |= GEN7_PS_16_DISPATCH_ENABLE;
187 } else {
188 dw4 |= GEN7_PS_16_DISPATCH_ENABLE;
189 }
190
191 dw5 |= (brw->wm.prog_data->first_curbe_grf <<
192 GEN7_PS_DISPATCH_START_GRF_SHIFT_0);
193 dw5 |= (brw->wm.prog_data->first_curbe_grf_16 <<
194 GEN7_PS_DISPATCH_START_GRF_SHIFT_2);
195
196 BEGIN_BATCH(8);
197 OUT_BATCH(_3DSTATE_PS << 16 | (8 - 2));
198 OUT_BATCH(brw->wm.prog_offset);
199 OUT_BATCH(dw2);
200 if (brw->wm.prog_data->total_scratch) {
201 OUT_RELOC(brw->wm.scratch_bo,
202 I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
203 ffs(brw->wm.prog_data->total_scratch) - 11);
204 } else {
205 OUT_BATCH(0);
206 }
207 OUT_BATCH(dw4);
208 OUT_BATCH(dw5);
209 OUT_BATCH(0); /* kernel 1 pointer */
210 OUT_BATCH(brw->wm.prog_offset + brw->wm.prog_data->prog_offset_16);
211 ADVANCE_BATCH();
212 }
213
214 const struct brw_tracked_state gen7_ps_state = {
215 .dirty = {
216 .mesa = _NEW_PROGRAM_CONSTANTS,
217 .brw = (BRW_NEW_FRAGMENT_PROGRAM |
218 BRW_NEW_PS_BINDING_TABLE |
219 BRW_NEW_BATCH),
220 .cache = (CACHE_NEW_SAMPLER |
221 CACHE_NEW_WM_PROG)
222 },
223 .emit = upload_ps_state,
224 };