gallium/radeon: set VPORT_ZMIN/MAX registers correctly
[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 #define SI_NUM_GRAPHICS_SHADERS (PIPE_SHADER_TESS_EVAL+1)
34 #define SI_NUM_SHADERS (PIPE_SHADER_COMPUTE+1)
35
36 #define SI_MAX_ATTRIBS 16
37 #define SI_NUM_VERTEX_BUFFERS SI_MAX_ATTRIBS
38 #define SI_NUM_SAMPLERS 32 /* OpenGL textures units per shader */
39 #define SI_NUM_CONST_BUFFERS 16
40 #define SI_NUM_IMAGES 16
41 #define SI_NUM_SHADER_BUFFERS 16
42
43 struct si_screen;
44 struct si_shader;
45
46 struct si_state_blend {
47 struct si_pm4_state pm4;
48 uint32_t cb_target_mask;
49 bool alpha_to_coverage;
50 bool alpha_to_one;
51 bool dual_src_blend;
52 /* Set 0xf or 0x0 (4 bits) per render target if the following is
53 * true. ANDed with spi_shader_col_format.
54 */
55 unsigned blend_enable_4bit;
56 unsigned need_src_alpha_4bit;
57 };
58
59 struct si_state_rasterizer {
60 struct si_pm4_state pm4;
61 /* poly offset states for 16-bit, 24-bit, and 32-bit zbuffers */
62 struct si_pm4_state pm4_poly_offset[3];
63 bool flatshade;
64 bool two_side;
65 bool multisample_enable;
66 bool force_persample_interp;
67 bool line_stipple_enable;
68 unsigned sprite_coord_enable;
69 unsigned pa_sc_line_stipple;
70 unsigned pa_cl_clip_cntl;
71 unsigned clip_plane_enable;
72 bool poly_stipple_enable;
73 bool line_smooth;
74 bool poly_smooth;
75 bool uses_poly_offset;
76 bool clamp_fragment_color;
77 bool rasterizer_discard;
78 bool scissor_enable;
79 bool clip_halfz;
80 };
81
82 struct si_dsa_stencil_ref_part {
83 uint8_t valuemask[2];
84 uint8_t writemask[2];
85 };
86
87 struct si_state_dsa {
88 struct si_pm4_state pm4;
89 unsigned alpha_func;
90 struct si_dsa_stencil_ref_part stencil_ref;
91 };
92
93 struct si_stencil_ref {
94 struct r600_atom atom;
95 struct pipe_stencil_ref state;
96 struct si_dsa_stencil_ref_part dsa_part;
97 };
98
99 struct si_vertex_element
100 {
101 unsigned count;
102 uint32_t rsrc_word3[SI_MAX_ATTRIBS];
103 uint32_t format_size[SI_MAX_ATTRIBS];
104 struct pipe_vertex_element elements[SI_MAX_ATTRIBS];
105 };
106
107 union si_state {
108 struct {
109 struct si_state_blend *blend;
110 struct si_state_rasterizer *rasterizer;
111 struct si_state_dsa *dsa;
112 struct si_pm4_state *poly_offset;
113 struct si_pm4_state *ls;
114 struct si_pm4_state *hs;
115 struct si_pm4_state *es;
116 struct si_pm4_state *gs;
117 struct si_pm4_state *vgt_shader_config;
118 struct si_pm4_state *vs;
119 struct si_pm4_state *ps;
120 } named;
121 struct si_pm4_state *array[0];
122 };
123
124 union si_state_atoms {
125 struct {
126 /* The order matters. */
127 struct r600_atom *cache_flush;
128 struct r600_atom *render_cond;
129 struct r600_atom *streamout_begin;
130 struct r600_atom *streamout_enable; /* must be after streamout_begin */
131 struct r600_atom *framebuffer;
132 struct r600_atom *msaa_sample_locs;
133 struct r600_atom *db_render_state;
134 struct r600_atom *msaa_config;
135 struct r600_atom *sample_mask;
136 struct r600_atom *cb_render_state;
137 struct r600_atom *blend_color;
138 struct r600_atom *clip_regs;
139 struct r600_atom *clip_state;
140 struct r600_atom *shader_userdata;
141 struct r600_atom *scissors;
142 struct r600_atom *viewports;
143 struct r600_atom *stencil_ref;
144 struct r600_atom *spi_map;
145 } s;
146 struct r600_atom *array[0];
147 };
148
149 #define SI_NUM_ATOMS (sizeof(union si_state_atoms)/sizeof(struct r600_atom*))
150
151 struct si_shader_data {
152 struct r600_atom atom;
153 uint32_t sh_base[SI_NUM_SHADERS];
154 };
155
156 /* Private read-write buffer slots. */
157 enum {
158 SI_HS_RING_TESS_FACTOR,
159 SI_HS_RING_TESS_OFFCHIP,
160
161 SI_ES_RING_ESGS,
162 SI_GS_RING_ESGS,
163
164 SI_GS_RING_GSVS0,
165 SI_GS_RING_GSVS1,
166 SI_GS_RING_GSVS2,
167 SI_GS_RING_GSVS3,
168 SI_VS_RING_GSVS,
169
170 SI_VS_STREAMOUT_BUF0,
171 SI_VS_STREAMOUT_BUF1,
172 SI_VS_STREAMOUT_BUF2,
173 SI_VS_STREAMOUT_BUF3,
174
175 SI_HS_CONST_DEFAULT_TESS_LEVELS,
176 SI_VS_CONST_CLIP_PLANES,
177 SI_PS_CONST_POLY_STIPPLE,
178 SI_PS_CONST_SAMPLE_POSITIONS,
179
180 SI_NUM_RW_BUFFERS,
181 };
182
183 /* Indices into sctx->descriptors, laid out so that gfx and compute pipelines
184 * are contiguous:
185 *
186 * 0 - rw buffers
187 * 1 - vertex const buffers
188 * 2 - vertex shader buffers
189 * ...
190 * 5 - fragment const buffers
191 * ...
192 * 21 - compute const buffers
193 * ...
194 */
195 #define SI_SHADER_DESCS_CONST_BUFFERS 0
196 #define SI_SHADER_DESCS_SHADER_BUFFERS 1
197 #define SI_SHADER_DESCS_SAMPLERS 2
198 #define SI_SHADER_DESCS_IMAGES 3
199 #define SI_NUM_SHADER_DESCS 4
200
201 #define SI_DESCS_RW_BUFFERS 0
202 #define SI_DESCS_FIRST_SHADER 1
203 #define SI_DESCS_FIRST_COMPUTE (SI_DESCS_FIRST_SHADER + \
204 PIPE_SHADER_COMPUTE * SI_NUM_SHADER_DESCS)
205 #define SI_NUM_DESCS (SI_DESCS_FIRST_SHADER + \
206 SI_NUM_SHADERS * SI_NUM_SHADER_DESCS)
207
208 /* This represents descriptors in memory, such as buffer resources,
209 * image resources, and sampler states.
210 */
211 struct si_descriptors {
212 /* The list of descriptors in malloc'd memory. */
213 uint32_t *list;
214 /* The size of one descriptor. */
215 unsigned element_dw_size;
216 /* The maximum number of descriptors. */
217 unsigned num_elements;
218
219 /* The buffer where the descriptors have been uploaded. */
220 struct r600_resource *buffer;
221 unsigned buffer_offset;
222
223 /* Offset in CE RAM */
224 unsigned ce_offset;
225
226 /* elements of the list that are changed and need to be uploaded */
227 unsigned dirty_mask;
228
229 /* Whether the CE ram is dirty and needs to be reinitialized entirely
230 * before we can do partial updates. */
231 bool ce_ram_dirty;
232
233 /* The shader userdata offset within a shader where the 64-bit pointer to the descriptor
234 * array will be stored. */
235 unsigned shader_userdata_offset;
236 /* Whether the pointer should be re-emitted. */
237 bool pointer_dirty;
238 };
239
240 struct si_sampler_views {
241 struct pipe_sampler_view *views[SI_NUM_SAMPLERS];
242 void *sampler_states[SI_NUM_SAMPLERS];
243
244 /* The i-th bit is set if that element is enabled (non-NULL resource). */
245 unsigned enabled_mask;
246 };
247
248 struct si_buffer_resources {
249 enum radeon_bo_usage shader_usage; /* READ, WRITE, or READWRITE */
250 enum radeon_bo_priority priority;
251 struct pipe_resource **buffers; /* this has num_buffers elements */
252
253 /* The i-th bit is set if that element is enabled (non-NULL resource). */
254 unsigned enabled_mask;
255 };
256
257 #define si_pm4_block_idx(member) \
258 (offsetof(union si_state, named.member) / sizeof(struct si_pm4_state *))
259
260 #define si_pm4_state_changed(sctx, member) \
261 ((sctx)->queued.named.member != (sctx)->emitted.named.member)
262
263 #define si_pm4_bind_state(sctx, member, value) \
264 do { \
265 (sctx)->queued.named.member = (value); \
266 } while(0)
267
268 #define si_pm4_delete_state(sctx, member, value) \
269 do { \
270 if ((sctx)->queued.named.member == (value)) { \
271 (sctx)->queued.named.member = NULL; \
272 } \
273 si_pm4_free_state(sctx, (struct si_pm4_state *)(value), \
274 si_pm4_block_idx(member)); \
275 } while(0)
276
277 /* si_descriptors.c */
278 void si_ce_reinitialize_all_descriptors(struct si_context *sctx);
279 void si_ce_enable_loads(struct radeon_winsys_cs *ib);
280 void si_set_mutable_tex_desc_fields(struct r600_texture *tex,
281 const struct radeon_surf_level *base_level_info,
282 unsigned base_level, unsigned first_level,
283 unsigned block_width, bool is_stencil,
284 uint32_t *state);
285 void si_set_ring_buffer(struct pipe_context *ctx, uint slot,
286 struct pipe_resource *buffer,
287 unsigned stride, unsigned num_records,
288 bool add_tid, bool swizzle,
289 unsigned element_size, unsigned index_stride, uint64_t offset);
290 void si_init_all_descriptors(struct si_context *sctx);
291 bool si_upload_vertex_buffer_descriptors(struct si_context *sctx);
292 bool si_upload_graphics_shader_descriptors(struct si_context *sctx);
293 bool si_upload_compute_shader_descriptors(struct si_context *sctx);
294 void si_release_all_descriptors(struct si_context *sctx);
295 void si_all_descriptors_begin_new_cs(struct si_context *sctx);
296 void si_upload_const_buffer(struct si_context *sctx, struct r600_resource **rbuffer,
297 const uint8_t *ptr, unsigned size, uint32_t *const_offset);
298 void si_update_all_texture_descriptors(struct si_context *sctx);
299 void si_shader_change_notify(struct si_context *sctx);
300 void si_update_compressed_colortex_masks(struct si_context *sctx);
301 void si_emit_graphics_shader_userdata(struct si_context *sctx,
302 struct r600_atom *atom);
303 void si_emit_compute_shader_userdata(struct si_context *sctx);
304 void si_set_rw_buffer(struct si_context *sctx,
305 uint slot, const struct pipe_constant_buffer *input);
306 /* si_state.c */
307 struct si_shader_selector;
308
309 void si_init_atom(struct si_context *sctx, struct r600_atom *atom,
310 struct r600_atom **list_elem,
311 void (*emit_func)(struct si_context *ctx, struct r600_atom *state));
312 void si_init_state_functions(struct si_context *sctx);
313 void si_init_screen_state_functions(struct si_screen *sscreen);
314 void
315 si_make_buffer_descriptor(struct si_screen *screen, struct r600_resource *buf,
316 enum pipe_format format,
317 unsigned offset, unsigned size,
318 uint32_t *state);
319 void
320 si_make_texture_descriptor(struct si_screen *screen,
321 struct r600_texture *tex,
322 bool sampler,
323 enum pipe_texture_target target,
324 enum pipe_format pipe_format,
325 const unsigned char state_swizzle[4],
326 unsigned first_level, unsigned last_level,
327 unsigned first_layer, unsigned last_layer,
328 unsigned width, unsigned height, unsigned depth,
329 uint32_t *state,
330 uint32_t *fmask_state);
331 struct pipe_sampler_view *
332 si_create_sampler_view_custom(struct pipe_context *ctx,
333 struct pipe_resource *texture,
334 const struct pipe_sampler_view *state,
335 unsigned width0, unsigned height0,
336 unsigned force_level);
337
338 /* si_state_shader.c */
339 bool si_update_shaders(struct si_context *sctx);
340 void si_init_shader_functions(struct si_context *sctx);
341 bool si_init_shader_cache(struct si_screen *sscreen);
342 void si_destroy_shader_cache(struct si_screen *sscreen);
343 void si_init_shader_selector_async(void *job, int thread_index);
344
345 /* si_state_draw.c */
346 void si_emit_cache_flush(struct si_context *sctx, struct r600_atom *atom);
347 void si_ce_pre_draw_synchronization(struct si_context *sctx);
348 void si_ce_post_draw_synchronization(struct si_context *sctx);
349 void si_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *dinfo);
350 void si_trace_emit(struct si_context *sctx);
351
352
353 static inline unsigned
354 si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool stencil)
355 {
356 if (stencil)
357 return rtex->surface.stencil_tiling_index[level];
358 else
359 return rtex->surface.tiling_index[level];
360 }
361
362 #endif