2 Copyright (C) Intel Corp. 2006. All Rights Reserved.
3 Intel funded Tungsten Graphics to
4 develop this 3D driver.
6 Permission is hereby granted, free of charge, to any person obtaining
7 a copy of this software and associated documentation files (the
8 "Software"), to deal in the Software without restriction, including
9 without limitation the rights to use, copy, modify, merge, publish,
10 distribute, sublicense, and/or sell copies of the Software, and to
11 permit persons to whom the Software is furnished to do so, subject to
12 the following conditions:
14 The above copyright notice and this permission notice (including the
15 next paragraph) shall be included in all copies or substantial
16 portions of the Software.
18 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
19 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
21 IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
22 LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
23 OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
24 WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 **********************************************************************/
29 * Keith Whitwell <keithw@vmware.com>
36 #include "brw_context.h"
42 enum intel_msaa_layout
;
44 extern const struct brw_tracked_state brw_blend_constant_color
;
45 extern const struct brw_tracked_state brw_cc_unit
;
46 extern const struct brw_tracked_state brw_clip_unit
;
47 extern const struct brw_tracked_state brw_vs_pull_constants
;
48 extern const struct brw_tracked_state brw_tcs_pull_constants
;
49 extern const struct brw_tracked_state brw_tes_pull_constants
;
50 extern const struct brw_tracked_state brw_gs_pull_constants
;
51 extern const struct brw_tracked_state brw_wm_pull_constants
;
52 extern const struct brw_tracked_state brw_cs_pull_constants
;
53 extern const struct brw_tracked_state brw_constant_buffer
;
54 extern const struct brw_tracked_state brw_curbe_offsets
;
55 extern const struct brw_tracked_state brw_invariant_state
;
56 extern const struct brw_tracked_state brw_gs_unit
;
57 extern const struct brw_tracked_state brw_binding_table_pointers
;
58 extern const struct brw_tracked_state brw_depthbuffer
;
59 extern const struct brw_tracked_state brw_recalculate_urb_fence
;
60 extern const struct brw_tracked_state brw_sf_vp
;
61 extern const struct brw_tracked_state brw_cs_texture_surfaces
;
62 extern const struct brw_tracked_state brw_vs_ubo_surfaces
;
63 extern const struct brw_tracked_state brw_vs_abo_surfaces
;
64 extern const struct brw_tracked_state brw_vs_image_surfaces
;
65 extern const struct brw_tracked_state brw_tcs_ubo_surfaces
;
66 extern const struct brw_tracked_state brw_tcs_abo_surfaces
;
67 extern const struct brw_tracked_state brw_tcs_image_surfaces
;
68 extern const struct brw_tracked_state brw_tes_ubo_surfaces
;
69 extern const struct brw_tracked_state brw_tes_abo_surfaces
;
70 extern const struct brw_tracked_state brw_tes_image_surfaces
;
71 extern const struct brw_tracked_state brw_gs_ubo_surfaces
;
72 extern const struct brw_tracked_state brw_gs_abo_surfaces
;
73 extern const struct brw_tracked_state brw_gs_image_surfaces
;
74 extern const struct brw_tracked_state brw_renderbuffer_surfaces
;
75 extern const struct brw_tracked_state brw_renderbuffer_read_surfaces
;
76 extern const struct brw_tracked_state brw_texture_surfaces
;
77 extern const struct brw_tracked_state brw_wm_binding_table
;
78 extern const struct brw_tracked_state brw_gs_binding_table
;
79 extern const struct brw_tracked_state brw_tes_binding_table
;
80 extern const struct brw_tracked_state brw_tcs_binding_table
;
81 extern const struct brw_tracked_state brw_vs_binding_table
;
82 extern const struct brw_tracked_state brw_wm_ubo_surfaces
;
83 extern const struct brw_tracked_state brw_wm_abo_surfaces
;
84 extern const struct brw_tracked_state brw_wm_image_surfaces
;
85 extern const struct brw_tracked_state brw_cs_ubo_surfaces
;
86 extern const struct brw_tracked_state brw_cs_abo_surfaces
;
87 extern const struct brw_tracked_state brw_cs_image_surfaces
;
88 extern const struct brw_tracked_state brw_wm_unit
;
90 extern const struct brw_tracked_state brw_psp_urb_cbs
;
92 extern const struct brw_tracked_state brw_indices
;
93 extern const struct brw_tracked_state brw_index_buffer
;
94 extern const struct brw_tracked_state gen7_cs_push_constants
;
95 extern const struct brw_tracked_state gen6_binding_table_pointers
;
96 extern const struct brw_tracked_state gen6_gs_binding_table
;
97 extern const struct brw_tracked_state gen6_renderbuffer_surfaces
;
98 extern const struct brw_tracked_state gen6_sampler_state
;
99 extern const struct brw_tracked_state gen6_sol_surface
;
100 extern const struct brw_tracked_state gen6_sf_vp
;
101 extern const struct brw_tracked_state gen6_urb
;
102 extern const struct brw_tracked_state gen7_depthbuffer
;
103 extern const struct brw_tracked_state gen7_l3_state
;
104 extern const struct brw_tracked_state gen7_push_constant_space
;
105 extern const struct brw_tracked_state gen7_urb
;
106 extern const struct brw_tracked_state gen8_pma_fix
;
107 extern const struct brw_tracked_state brw_cs_work_groups_surface
;
110 brw_state_dirty(const struct brw_context
*brw
,
111 GLuint mesa_flags
, uint64_t brw_flags
)
113 return ((brw
->NewGLState
& mesa_flags
) |
114 (brw
->ctx
.NewDriverState
& brw_flags
)) != 0;
117 /* brw_binding_tables.c */
118 void brw_upload_binding_table(struct brw_context
*brw
,
119 uint32_t packet_name
,
120 const struct brw_stage_prog_data
*prog_data
,
121 struct brw_stage_state
*stage_state
);
123 /* brw_misc_state.c */
124 void brw_upload_invariant_state(struct brw_context
*brw
);
126 brw_depthbuffer_format(struct brw_context
*brw
);
129 brw_convert_depth_value(mesa_format format
, float value
);
131 void brw_upload_state_base_address(struct brw_context
*brw
);
133 /* gen8_depth_state.c */
134 void gen8_write_pma_stall_bits(struct brw_context
*brw
,
135 uint32_t pma_stall_bits
);
137 /***********************************************************************
140 void brw_upload_render_state(struct brw_context
*brw
);
141 void brw_render_state_finished(struct brw_context
*brw
);
142 void brw_upload_compute_state(struct brw_context
*brw
);
143 void brw_compute_state_finished(struct brw_context
*brw
);
144 void brw_init_state(struct brw_context
*brw
);
145 void brw_destroy_state(struct brw_context
*brw
);
146 void brw_emit_select_pipeline(struct brw_context
*brw
,
147 enum brw_pipeline pipeline
);
150 brw_select_pipeline(struct brw_context
*brw
, enum brw_pipeline pipeline
)
152 if (unlikely(brw
->last_pipeline
!= pipeline
)) {
153 assert(pipeline
< BRW_NUM_PIPELINES
);
154 brw_emit_select_pipeline(brw
, pipeline
);
155 brw
->last_pipeline
= pipeline
;
159 /***********************************************************************
160 * brw_program_cache.c
163 void brw_upload_cache(struct brw_cache
*cache
,
164 enum brw_cache_id cache_id
,
171 uint32_t *out_offset
, void *out_aux
);
173 bool brw_search_cache(struct brw_cache
*cache
,
174 enum brw_cache_id cache_id
,
177 uint32_t *inout_offset
, void *inout_aux
);
179 const void *brw_find_previous_compile(struct brw_cache
*cache
,
180 enum brw_cache_id cache_id
,
181 unsigned program_string_id
);
183 void brw_program_cache_check_size(struct brw_context
*brw
);
185 void brw_init_caches( struct brw_context
*brw
);
186 void brw_destroy_caches( struct brw_context
*brw
);
188 void brw_print_program_cache(struct brw_context
*brw
);
190 /***********************************************************************
193 #define BRW_BATCH_STRUCT(brw, s) \
194 intel_batchbuffer_data(brw, (s), sizeof(*(s)), RENDER_RING)
196 void *brw_state_batch(struct brw_context
*brw
,
197 int size
, int alignment
, uint32_t *out_offset
);
198 uint32_t brw_state_batch_size(struct brw_context
*brw
, uint32_t offset
);
200 /* brw_wm_surface_state.c */
201 void gen4_init_vtable_surface_functions(struct brw_context
*brw
);
202 uint32_t brw_get_surface_tiling_bits(uint32_t tiling
);
203 uint32_t brw_get_surface_num_multisamples(unsigned num_samples
);
204 enum isl_format
brw_isl_format_for_mesa_format(mesa_format mesa_format
);
206 GLuint
translate_tex_target(GLenum target
);
208 enum isl_format
translate_tex_format(struct brw_context
*brw
,
209 mesa_format mesa_format
,
212 int brw_get_texture_swizzle(const struct gl_context
*ctx
,
213 const struct gl_texture_object
*t
);
215 void brw_emit_buffer_surface_state(struct brw_context
*brw
,
216 uint32_t *out_offset
,
218 unsigned buffer_offset
,
219 unsigned surface_format
,
220 unsigned buffer_size
,
224 void brw_update_texture_surface(struct gl_context
*ctx
,
225 unsigned unit
, uint32_t *surf_offset
,
226 bool for_gather
, uint32_t plane
);
228 uint32_t brw_update_renderbuffer_surface(struct brw_context
*brw
,
229 struct gl_renderbuffer
*rb
,
230 uint32_t flags
, unsigned unit
,
231 uint32_t surf_index
);
233 void brw_update_renderbuffer_surfaces(struct brw_context
*brw
,
234 const struct gl_framebuffer
*fb
,
235 uint32_t render_target_start
,
236 uint32_t *surf_offset
);
238 /* gen7_wm_surface_state.c */
239 void gen7_check_surface_setup(uint32_t *surf
, bool is_render_target
);
240 void gen7_init_vtable_surface_functions(struct brw_context
*brw
);
242 /* gen8_surface_state.c */
244 void gen8_init_vtable_surface_functions(struct brw_context
*brw
);
246 /* brw_sampler_state.c */
247 void brw_emit_sampler_state(struct brw_context
*brw
,
248 uint32_t *sampler_state
,
249 uint32_t batch_offset_for_sampler_state
,
253 unsigned max_anisotropy
,
254 unsigned address_rounding
,
262 unsigned shadow_function
,
263 bool non_normalized_coordinates
,
264 uint32_t border_color_offset
);
266 /* gen6_surface_state.c */
267 void gen6_init_vtable_surface_functions(struct brw_context
*brw
);
269 /* brw_vs_surface_state.c */
271 brw_upload_pull_constants(struct brw_context
*brw
,
272 GLbitfield64 brw_new_constbuf
,
273 const struct gl_program
*prog
,
274 struct brw_stage_state
*stage_state
,
275 const struct brw_stage_prog_data
*prog_data
);
277 /* gen7_vs_state.c */
279 gen7_upload_constant_state(struct brw_context
*brw
,
280 const struct brw_stage_state
*stage_state
,
281 bool active
, unsigned opcode
);
284 void brw_upload_clip_prog(struct brw_context
*brw
);
287 void brw_upload_sf_prog(struct brw_context
*brw
);
289 bool brw_is_drawing_points(const struct brw_context
*brw
);
290 bool brw_is_drawing_lines(const struct brw_context
*brw
);
292 /* gen7_l3_state.c */
294 gen7_restore_default_l3_config(struct brw_context
*brw
);
297 use_state_point_size(const struct brw_context
*brw
)
299 const struct gl_context
*ctx
= &brw
->ctx
;
301 /* Section 14.4 (Points) of the OpenGL 4.5 specification says:
303 * "If program point size mode is enabled, the derived point size is
304 * taken from the (potentially clipped) shader built-in gl_PointSize
307 * * the geometry shader, if active;
308 * * the tessellation evaluation shader, if active and no
309 * geometry shader is active;
310 * * the vertex shader, otherwise
312 * and clamped to the implementation-dependent point size range. If
313 * the value written to gl_PointSize is less than or equal to zero,
314 * or if no value was written to gl_PointSize, results are undefined.
315 * If program point size mode is disabled, the derived point size is
316 * specified with the command
318 * void PointSize(float size);
320 * size specifies the requested size of a point. The default value
323 * The rules for GLES come from the ES 3.2, OES_geometry_point_size, and
324 * OES_tessellation_point_size specifications. To summarize: if the last
325 * stage before rasterization is a GS or TES, then use gl_PointSize from
326 * the shader if written. Otherwise, use 1.0. If the last stage is a
327 * vertex shader, use gl_PointSize, or it is undefined.
329 * We can combine these rules into a single condition for both APIs.
330 * Using the state point size when the last shader stage doesn't write
331 * gl_PointSize satisfies GL's requirements, as it's undefined. Because
332 * ES doesn't have a PointSize() command, the state point size will
333 * remain 1.0, satisfying the ES default value in the GS/TES case, and
334 * the VS case (1.0 works for "undefined"). Mesa sets the program point
335 * mode flag to always-enabled in ES, so we can safely check that, and
336 * it'll be ignored for ES.
338 * _NEW_PROGRAM | _NEW_POINT
339 * BRW_NEW_VUE_MAP_GEOM_OUT
341 return (!ctx
->VertexProgram
.PointSizeEnabled
&& !ctx
->Point
._Attenuated
) ||
342 (brw
->vue_map_geom_out
.slots_valid
& VARYING_BIT_PSIZ
) == 0;
345 void brw_copy_pipeline_atoms(struct brw_context
*brw
,
346 enum brw_pipeline pipeline
,
347 const struct brw_tracked_state
**atoms
,
349 void gen4_init_atoms(struct brw_context
*brw
);
350 void gen45_init_atoms(struct brw_context
*brw
);
351 void gen5_init_atoms(struct brw_context
*brw
);
352 void gen6_init_atoms(struct brw_context
*brw
);
353 void gen7_init_atoms(struct brw_context
*brw
);
354 void gen75_init_atoms(struct brw_context
*brw
);
355 void gen8_init_atoms(struct brw_context
*brw
);
356 void gen9_init_atoms(struct brw_context
*brw
);
357 void gen10_init_atoms(struct brw_context
*brw
);
359 void upload_gs_state_for_tf(struct brw_context
*brw
);
361 /* Memory Object Control State:
362 * Specifying zero for L3 means "uncached in L3", at least on Haswell
363 * and Baytrail, since there are no PTE flags for setting L3 cacheability.
364 * On Ivybridge, the PTEs do have a cache-in-L3 bit, so setting MOCS to 0
365 * may still respect that.
367 #define GEN7_MOCS_L3 1
369 /* Ivybridge only: cache in LLC.
370 * Specifying zero here means to use the PTE values set by the kernel;
371 * non-zero overrides the PTE values.
373 #define IVB_MOCS_LLC (1 << 1)
375 /* Baytrail only: snoop in CPU cache */
376 #define BYT_MOCS_SNOOP (1 << 1)
378 /* Haswell only: LLC/eLLC controls (write-back or uncached).
379 * Specifying zero here means to use the PTE values set by the kernel,
380 * which is useful since it offers additional control (write-through
381 * cacheing and age). Non-zero overrides the PTE values.
383 #define HSW_MOCS_UC_LLC_UC_ELLC (1 << 1)
384 #define HSW_MOCS_WB_LLC_WB_ELLC (2 << 1)
385 #define HSW_MOCS_UC_LLC_WB_ELLC (3 << 1)
387 /* Broadwell: these defines always use all available caches (L3, LLC, eLLC),
388 * and let you force write-back (WB) or write-through (WT) caching, or leave
389 * it up to the page table entry (PTE) specified by the kernel.
391 #define BDW_MOCS_WB 0x78
392 #define BDW_MOCS_WT 0x58
393 #define BDW_MOCS_PTE 0x18
395 /* Skylake: MOCS is now an index into an array of 62 different caching
396 * configurations programmed by the kernel.
398 /* TC=LLC/eLLC, LeCC=WB, LRUM=3, L3CC=WB */
399 #define SKL_MOCS_WB (2 << 1)
400 /* TC=LLC/eLLC, LeCC=PTE, LRUM=3, L3CC=WB */
401 #define SKL_MOCS_PTE (1 << 1)
403 /* Cannonlake: MOCS is now an index into an array of 62 different caching
404 * configurations programmed by the kernel.
406 /* TC=LLC/eLLC, LeCC=WB, LRUM=3, L3CC=WB */
407 #define CNL_MOCS_WB (2 << 1)
408 /* TC=LLC/eLLC, LeCC=PTE, LRUM=3, L3CC=WB */
409 #define CNL_MOCS_PTE (1 << 1)