ae42e6673b58b7b7b71cea25d00f4c88742a809c
[mesa.git] / src / gallium / drivers / radeonsi / si_state.h
1 /*
2 * Copyright 2012 Advanced Micro Devices, Inc.
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 * on the rights to use, copy, modify, merge, publish, distribute, sub
8 * license, and/or sell copies of the Software, and to permit persons to whom
9 * the 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 NON-INFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
19 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
20 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
21 * USE OR OTHER DEALINGS IN THE SOFTWARE.
22 *
23 * Authors:
24 * Christian König <christian.koenig@amd.com>
25 */
26
27 #ifndef SI_STATE_H
28 #define SI_STATE_H
29
30 #include "si_pm4.h"
31 #include "../radeon/r600_pipe_common.h"
32
33 struct si_screen;
34
35 struct si_state_blend {
36 struct si_pm4_state pm4;
37 uint32_t cb_target_mask;
38 bool alpha_to_one;
39 };
40
41 struct si_state_viewport {
42 struct si_pm4_state pm4;
43 struct pipe_viewport_state viewport;
44 };
45
46 struct si_state_rasterizer {
47 struct si_pm4_state pm4;
48 bool flatshade;
49 bool two_side;
50 bool multisample_enable;
51 bool line_stipple_enable;
52 unsigned sprite_coord_enable;
53 unsigned pa_sc_line_stipple;
54 unsigned pa_su_sc_mode_cntl;
55 unsigned pa_cl_clip_cntl;
56 unsigned pa_cl_vs_out_cntl;
57 unsigned clip_plane_enable;
58 float offset_units;
59 float offset_scale;
60 };
61
62 struct si_state_dsa {
63 struct si_pm4_state pm4;
64 float alpha_ref;
65 unsigned alpha_func;
66 unsigned db_render_control;
67 uint8_t valuemask[2];
68 uint8_t writemask[2];
69 };
70
71 struct si_vertex_element
72 {
73 unsigned count;
74 uint32_t rsrc_word3[PIPE_MAX_ATTRIBS];
75 struct pipe_vertex_element elements[PIPE_MAX_ATTRIBS];
76 };
77
78 union si_state {
79 struct {
80 struct si_pm4_state *init;
81 struct si_state_blend *blend;
82 struct si_pm4_state *blend_color;
83 struct si_pm4_state *clip;
84 struct si_pm4_state *sample_mask;
85 struct si_pm4_state *scissor;
86 struct si_state_viewport *viewport;
87 struct si_state_rasterizer *rasterizer;
88 struct si_state_dsa *dsa;
89 struct si_pm4_state *fb_rs;
90 struct si_pm4_state *fb_blend;
91 struct si_pm4_state *dsa_stencil_ref;
92 struct si_pm4_state *es;
93 struct si_pm4_state *gs;
94 struct si_pm4_state *gs_rings;
95 struct si_pm4_state *gs_sampler;
96 struct si_pm4_state *gs_onoff;
97 struct si_pm4_state *vs;
98 struct si_pm4_state *vs_sampler;
99 struct si_pm4_state *ps;
100 struct si_pm4_state *ps_sampler;
101 struct si_pm4_state *spi;
102 struct si_pm4_state *vertex_buffers;
103 struct si_pm4_state *draw_info;
104 struct si_pm4_state *draw;
105 } named;
106 struct si_pm4_state *array[0];
107 };
108
109 #define SI_NUM_USER_SAMPLERS 16 /* AKA OpenGL textures units per shader */
110
111 /* User sampler views: 0..15
112 * FMASK sampler views: 16..31 (no sampler states)
113 */
114 #define SI_FMASK_TEX_OFFSET SI_NUM_USER_SAMPLERS
115 #define SI_NUM_SAMPLER_VIEWS (SI_FMASK_TEX_OFFSET + SI_NUM_USER_SAMPLERS)
116 #define SI_NUM_SAMPLER_STATES SI_NUM_USER_SAMPLERS
117
118 /* User constant buffers: 0..15
119 * Driver state constants: 16
120 */
121 #define SI_NUM_USER_CONST_BUFFERS 16
122 #define SI_DRIVER_STATE_CONST_BUF SI_NUM_USER_CONST_BUFFERS
123 #define SI_NUM_CONST_BUFFERS (SI_DRIVER_STATE_CONST_BUF + 1)
124
125 /* Read-write buffer slots.
126 *
127 * Ring buffers: 0..1
128 * Streamout buffers: 2..5
129 */
130 #define SI_RING_ESGS 0
131 #define SI_RING_GSVS 1
132 #define SI_NUM_RING_BUFFERS 2
133 #define SI_SO_BUF_OFFSET SI_NUM_RING_BUFFERS
134 #define SI_NUM_RW_BUFFERS (SI_SO_BUF_OFFSET + 4)
135
136 #define SI_NUM_VERTEX_BUFFERS 16
137
138
139 /* This represents resource descriptors in memory, such as buffer resources,
140 * image resources, and sampler states.
141 */
142 struct si_descriptors {
143 struct r600_atom atom;
144
145 /* The size of one resource descriptor. */
146 unsigned element_dw_size;
147 /* The maximum number of resource descriptors. */
148 unsigned num_elements;
149
150 /* The buffer where resource descriptors are stored. */
151 struct r600_resource *buffer;
152
153 /* The i-th bit is set if that element is dirty (changed but not emitted). */
154 unsigned dirty_mask;
155 /* The i-th bit is set if that element is enabled (non-NULL resource). */
156 unsigned enabled_mask;
157
158 /* We can't update descriptors directly because the GPU might be
159 * reading them at the same time, so we have to update them
160 * in a copy-on-write manner. Each such copy is called a context,
161 * which is just another array descriptors in the same buffer. */
162 unsigned current_context_id;
163 /* The size of a context, should be equal to 4*element_dw_size*num_elements. */
164 unsigned context_size;
165
166 /* The shader userdata register where the 64-bit pointer to the descriptor
167 * array will be stored. */
168 unsigned shader_userdata_reg;
169 };
170
171 struct si_sampler_views {
172 struct si_descriptors desc;
173 struct pipe_sampler_view *views[SI_NUM_SAMPLER_VIEWS];
174 uint32_t *desc_data[SI_NUM_SAMPLER_VIEWS];
175 };
176
177 struct si_buffer_resources {
178 struct si_descriptors desc;
179 unsigned num_buffers;
180 enum radeon_bo_usage shader_usage; /* READ, WRITE, or READWRITE */
181 enum radeon_bo_priority priority;
182 struct pipe_resource **buffers; /* this has num_buffers elements */
183 uint32_t *desc_storage; /* this has num_buffers*4 elements */
184 uint32_t **desc_data; /* an array of pointers pointing to desc_storage */
185 };
186
187 #define si_pm4_block_idx(member) \
188 (offsetof(union si_state, named.member) / sizeof(struct si_pm4_state *))
189
190 #define si_pm4_state_changed(sctx, member) \
191 ((sctx)->queued.named.member != (sctx)->emitted.named.member)
192
193 #define si_pm4_bind_state(sctx, member, value) \
194 do { \
195 (sctx)->queued.named.member = (value); \
196 } while(0)
197
198 #define si_pm4_delete_state(sctx, member, value) \
199 do { \
200 if ((sctx)->queued.named.member == (value)) { \
201 (sctx)->queued.named.member = NULL; \
202 } \
203 si_pm4_free_state(sctx, (struct si_pm4_state *)(value), \
204 si_pm4_block_idx(member)); \
205 } while(0)
206
207 #define si_pm4_set_state(sctx, member, value) \
208 do { \
209 if ((sctx)->queued.named.member != (value)) { \
210 si_pm4_free_state(sctx, \
211 (struct si_pm4_state *)(sctx)->queued.named.member, \
212 si_pm4_block_idx(member)); \
213 (sctx)->queued.named.member = (value); \
214 } \
215 } while(0)
216
217 /* si_descriptors.c */
218 void si_set_sampler_view(struct si_context *sctx, unsigned shader,
219 unsigned slot, struct pipe_sampler_view *view,
220 unsigned *view_desc);
221 void si_set_ring_buffer(struct pipe_context *ctx, uint shader, uint slot,
222 struct pipe_constant_buffer *input,
223 unsigned stride, unsigned num_records,
224 bool add_tid, bool swizzle,
225 unsigned element_size, unsigned index_stride);
226 void si_init_all_descriptors(struct si_context *sctx);
227 void si_release_all_descriptors(struct si_context *sctx);
228 void si_all_descriptors_begin_new_cs(struct si_context *sctx);
229 void si_copy_buffer(struct si_context *sctx,
230 struct pipe_resource *dst, struct pipe_resource *src,
231 uint64_t dst_offset, uint64_t src_offset, unsigned size);
232 void si_upload_const_buffer(struct si_context *sctx, struct r600_resource **rbuffer,
233 const uint8_t *ptr, unsigned size, uint32_t *const_offset);
234
235 /* si_state.c */
236 struct si_pipe_shader_selector;
237
238 boolean si_is_format_supported(struct pipe_screen *screen,
239 enum pipe_format format,
240 enum pipe_texture_target target,
241 unsigned sample_count,
242 unsigned usage);
243 int si_shader_select(struct pipe_context *ctx,
244 struct si_pipe_shader_selector *sel);
245 void si_init_state_functions(struct si_context *sctx);
246 void si_init_config(struct si_context *sctx);
247 unsigned cik_bank_wh(unsigned bankwh);
248 unsigned cik_db_pipe_config(struct si_screen *sscreen, unsigned tile_mode);
249 unsigned cik_macro_tile_aspect(unsigned macro_tile_aspect);
250 unsigned cik_tile_split(unsigned tile_split);
251 uint32_t si_num_banks(struct si_screen *sscreen, unsigned bpe, unsigned tile_split,
252 unsigned tile_mode_index);
253 unsigned si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool stencil);
254
255 /* si_state_draw.c */
256 extern const struct r600_atom si_atom_cache_flush;
257 extern const struct r600_atom si_atom_msaa_config;
258 void si_emit_cache_flush(struct r600_common_context *sctx, struct r600_atom *atom);
259 void si_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *dinfo);
260
261 /* si_commands.c */
262 void si_cmd_context_control(struct si_pm4_state *pm4);
263 void si_cmd_draw_index_2(struct si_pm4_state *pm4, uint32_t max_size,
264 uint64_t index_base, uint32_t index_count,
265 uint32_t initiator, bool predicate);
266 void si_cmd_draw_index_auto(struct si_pm4_state *pm4, uint32_t count,
267 uint32_t initiator, bool predicate);
268 void si_cmd_draw_indirect(struct si_pm4_state *pm4, uint64_t indirect_va,
269 uint32_t indirect_offset, uint32_t base_vtx_loc,
270 uint32_t start_inst_loc, bool predicate);
271 void si_cmd_draw_index_indirect(struct si_pm4_state *pm4, uint64_t indirect_va,
272 uint64_t index_va, uint32_t index_max_size,
273 uint32_t indirect_offset, uint32_t base_vtx_loc,
274 uint32_t start_inst_loc, bool predicate);
275 void si_cmd_surface_sync(struct si_pm4_state *pm4, uint32_t cp_coher_cntl);
276
277 #endif