2 * Copyright © 2011 Intel Corporation
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:
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
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
26 #include "intel_batchbuffer.h"
27 #include "intel_fbo.h"
28 #include "intel_mipmap_tree.h"
30 #include "brw_context.h"
31 #include "brw_defines.h"
32 #include "brw_state.h"
34 #include "brw_blorp.h"
35 #include "gen7_blorp.h"
43 * If the 3DSTATE_URB_VS is emitted, than the others must be also.
44 * From the Ivybridge PRM, Volume 2 Part 1, section 1.7.1 3DSTATE_URB_VS:
46 * 3DSTATE_URB_HS, 3DSTATE_URB_DS, and 3DSTATE_URB_GS must also be
47 * programmed in order for the programming of this state to be
51 gen7_blorp_emit_urb_config(struct brw_context
*brw
,
52 const brw_blorp_params
*params
)
54 /* The minimum valid value is 32. See 3DSTATE_URB_VS,
55 * Dword 1.15:0 "VS Number of URB Entries".
57 int num_vs_entries
= 32;
59 int vs_start
= 2; /* skip over push constants */
61 gen7_emit_urb_state(brw
, num_vs_entries
, vs_size
, vs_start
);
65 /* 3DSTATE_BLEND_STATE_POINTERS */
67 gen7_blorp_emit_blend_state_pointer(struct brw_context
*brw
,
68 const brw_blorp_params
*params
,
69 uint32_t cc_blend_state_offset
)
72 OUT_BATCH(_3DSTATE_BLEND_STATE_POINTERS
<< 16 | (2 - 2));
73 OUT_BATCH(cc_blend_state_offset
| 1);
78 /* 3DSTATE_CC_STATE_POINTERS */
80 gen7_blorp_emit_cc_state_pointer(struct brw_context
*brw
,
81 const brw_blorp_params
*params
,
82 uint32_t cc_state_offset
)
85 OUT_BATCH(_3DSTATE_CC_STATE_POINTERS
<< 16 | (2 - 2));
86 OUT_BATCH(cc_state_offset
| 1);
91 gen7_blorp_emit_cc_viewport(struct brw_context
*brw
,
92 const brw_blorp_params
*params
)
94 struct brw_cc_viewport
*ccv
;
95 uint32_t cc_vp_offset
;
97 ccv
= (struct brw_cc_viewport
*)brw_state_batch(brw
, AUB_TRACE_CC_VP_STATE
,
100 ccv
->min_depth
= 0.0;
101 ccv
->max_depth
= 1.0;
104 OUT_BATCH(_3DSTATE_VIEWPORT_STATE_POINTERS_CC
<< 16 | (2 - 2));
105 OUT_BATCH(cc_vp_offset
);
110 /* 3DSTATE_DEPTH_STENCIL_STATE_POINTERS
112 * The offset is relative to CMD_STATE_BASE_ADDRESS.DynamicStateBaseAddress.
115 gen7_blorp_emit_depth_stencil_state_pointers(struct brw_context
*brw
,
116 const brw_blorp_params
*params
,
117 uint32_t depthstencil_offset
)
120 OUT_BATCH(_3DSTATE_DEPTH_STENCIL_STATE_POINTERS
<< 16 | (2 - 2));
121 OUT_BATCH(depthstencil_offset
| 1);
126 /* SURFACE_STATE for renderbuffer or texture surface (see
127 * brw_update_renderbuffer_surface and brw_update_texture_surface)
130 gen7_blorp_emit_surface_state(struct brw_context
*brw
,
131 const brw_blorp_params
*params
,
132 const brw_blorp_surface_info
*surface
,
133 uint32_t read_domains
, uint32_t write_domain
,
134 bool is_render_target
)
136 uint32_t wm_surf_offset
;
137 uint32_t width
= surface
->width
;
138 uint32_t height
= surface
->height
;
139 /* Note: since gen7 uses INTEL_MSAA_LAYOUT_CMS or INTEL_MSAA_LAYOUT_UMS for
140 * color surfaces, width and height are measured in pixels; we don't need
141 * to divide them by 2 as we do for Gen6 (see
142 * gen6_blorp_emit_surface_state).
144 struct intel_region
*region
= surface
->mt
->region
;
145 uint32_t tile_x
, tile_y
;
146 uint8_t mocs
= brw
->is_haswell
? GEN7_MOCS_L3
: 0;
148 uint32_t tiling
= surface
->map_stencil_as_y_tiled
149 ? I915_TILING_Y
: region
->tiling
;
151 uint32_t *surf
= (uint32_t *)
152 brw_state_batch(brw
, AUB_TRACE_SURFACE_STATE
, 8 * 4, 32, &wm_surf_offset
);
153 memset(surf
, 0, 8 * 4);
155 surf
[0] = BRW_SURFACE_2D
<< BRW_SURFACE_TYPE_SHIFT
|
156 surface
->brw_surfaceformat
<< BRW_SURFACE_FORMAT_SHIFT
|
157 gen7_surface_tiling_mode(tiling
);
159 if (surface
->mt
->align_h
== 4)
160 surf
[0] |= GEN7_SURFACE_VALIGN_4
;
161 if (surface
->mt
->align_w
== 8)
162 surf
[0] |= GEN7_SURFACE_HALIGN_8
;
164 if (surface
->array_spacing_lod0
)
165 surf
[0] |= GEN7_SURFACE_ARYSPC_LOD0
;
167 surf
[0] |= GEN7_SURFACE_ARYSPC_FULL
;
171 surface
->compute_tile_offsets(&tile_x
, &tile_y
) + region
->bo
->offset
;
173 /* Note that the low bits of these fields are missing, so
174 * there's the possibility of getting in trouble.
176 assert(tile_x
% 4 == 0);
177 assert(tile_y
% 2 == 0);
178 surf
[5] = SET_FIELD(tile_x
/ 4, BRW_SURFACE_X_OFFSET
) |
179 SET_FIELD(tile_y
/ 2, BRW_SURFACE_Y_OFFSET
) |
180 SET_FIELD(mocs
, GEN7_SURFACE_MOCS
);
182 surf
[2] = SET_FIELD(width
- 1, GEN7_SURFACE_WIDTH
) |
183 SET_FIELD(height
- 1, GEN7_SURFACE_HEIGHT
);
185 uint32_t pitch_bytes
= region
->pitch
;
186 if (surface
->map_stencil_as_y_tiled
)
188 surf
[3] = pitch_bytes
- 1;
190 surf
[4] = gen7_surface_msaa_bits(surface
->num_samples
, surface
->msaa_layout
);
191 if (surface
->mt
->mcs_mt
) {
192 gen7_set_surface_mcs_info(brw
, surf
, wm_surf_offset
, surface
->mt
->mcs_mt
,
196 surf
[7] = surface
->mt
->fast_clear_color_value
;
198 if (brw
->is_haswell
) {
199 surf
[7] |= (SET_FIELD(HSW_SCS_RED
, GEN7_SURFACE_SCS_R
) |
200 SET_FIELD(HSW_SCS_GREEN
, GEN7_SURFACE_SCS_G
) |
201 SET_FIELD(HSW_SCS_BLUE
, GEN7_SURFACE_SCS_B
) |
202 SET_FIELD(HSW_SCS_ALPHA
, GEN7_SURFACE_SCS_A
));
205 /* Emit relocation to surface contents */
206 drm_intel_bo_emit_reloc(brw
->batch
.bo
,
209 surf
[1] - region
->bo
->offset
,
210 read_domains
, write_domain
);
212 gen7_check_surface_setup(surf
, is_render_target
);
214 return wm_surf_offset
;
219 * SAMPLER_STATE. See gen7_update_sampler_state().
222 gen7_blorp_emit_sampler_state(struct brw_context
*brw
,
223 const brw_blorp_params
*params
)
225 uint32_t sampler_offset
;
227 struct gen7_sampler_state
*sampler
= (struct gen7_sampler_state
*)
228 brw_state_batch(brw
, AUB_TRACE_SAMPLER_STATE
,
229 sizeof(struct gen7_sampler_state
),
230 32, &sampler_offset
);
231 memset(sampler
, 0, sizeof(*sampler
));
233 sampler
->ss0
.min_filter
= BRW_MAPFILTER_LINEAR
;
234 sampler
->ss0
.mip_filter
= BRW_MIPFILTER_NONE
;
235 sampler
->ss0
.mag_filter
= BRW_MAPFILTER_LINEAR
;
237 sampler
->ss3
.r_wrap_mode
= BRW_TEXCOORDMODE_CLAMP
;
238 sampler
->ss3
.s_wrap_mode
= BRW_TEXCOORDMODE_CLAMP
;
239 sampler
->ss3
.t_wrap_mode
= BRW_TEXCOORDMODE_CLAMP
;
241 // sampler->ss0.min_mag_neq = 1;
245 sampler
->ss0
.lod_bias
= 0;
247 sampler
->ss0
.lod_preclamp
= 1; /* OpenGL mode */
248 sampler
->ss0
.default_color_mode
= 0; /* OpenGL/DX10 mode */
250 /* Set BaseMipLevel, MaxLOD, MinLOD:
252 * XXX: I don't think that using firstLevel, lastLevel works,
253 * because we always setup the surface state as if firstLevel ==
254 * level zero. Probably have to subtract firstLevel from each of
257 sampler
->ss0
.base_level
= U_FIXED(0, 1);
259 sampler
->ss1
.max_lod
= U_FIXED(0, 8);
260 sampler
->ss1
.min_lod
= U_FIXED(0, 8);
262 sampler
->ss3
.non_normalized_coord
= 1;
264 sampler
->ss3
.address_round
|= BRW_ADDRESS_ROUNDING_ENABLE_U_MIN
|
265 BRW_ADDRESS_ROUNDING_ENABLE_V_MIN
|
266 BRW_ADDRESS_ROUNDING_ENABLE_R_MIN
;
267 sampler
->ss3
.address_round
|= BRW_ADDRESS_ROUNDING_ENABLE_U_MAG
|
268 BRW_ADDRESS_ROUNDING_ENABLE_V_MAG
|
269 BRW_ADDRESS_ROUNDING_ENABLE_R_MAG
;
271 return sampler_offset
;
277 * Disable vertex shader.
280 gen7_blorp_emit_vs_disable(struct brw_context
*brw
,
281 const brw_blorp_params
*params
)
284 OUT_BATCH(_3DSTATE_CONSTANT_VS
<< 16 | (7 - 2));
294 OUT_BATCH(_3DSTATE_VS
<< 16 | (6 - 2));
306 * Disable the hull shader.
309 gen7_blorp_emit_hs_disable(struct brw_context
*brw
,
310 const brw_blorp_params
*params
)
313 OUT_BATCH(_3DSTATE_CONSTANT_HS
<< 16 | (7 - 2));
323 OUT_BATCH(_3DSTATE_HS
<< 16 | (7 - 2));
336 * Disable the tesselation engine.
339 gen7_blorp_emit_te_disable(struct brw_context
*brw
,
340 const brw_blorp_params
*params
)
343 OUT_BATCH(_3DSTATE_TE
<< 16 | (4 - 2));
353 * Disable the domain shader.
356 gen7_blorp_emit_ds_disable(struct brw_context
*brw
,
357 const brw_blorp_params
*params
)
360 OUT_BATCH(_3DSTATE_CONSTANT_DS
<< 16 | (7 - 2));
370 OUT_BATCH(_3DSTATE_DS
<< 16 | (6 - 2));
381 * Disable the geometry shader.
384 gen7_blorp_emit_gs_disable(struct brw_context
*brw
,
385 const brw_blorp_params
*params
)
388 OUT_BATCH(_3DSTATE_CONSTANT_GS
<< 16 | (7 - 2));
398 OUT_BATCH(_3DSTATE_GS
<< 16 | (7 - 2));
413 gen7_blorp_emit_streamout_disable(struct brw_context
*brw
,
414 const brw_blorp_params
*params
)
417 OUT_BATCH(_3DSTATE_STREAMOUT
<< 16 | (3 - 2));
425 gen7_blorp_emit_sf_config(struct brw_context
*brw
,
426 const brw_blorp_params
*params
)
430 * Disable ViewportTransformEnable (dw1.1)
432 * From the SandyBridge PRM, Volume 2, Part 1, Section 1.3, "3D
433 * Primitives Overview":
434 * RECTLIST: Viewport Mapping must be DISABLED (as is typical with the
435 * use of screen- space coordinates).
437 * A solid rectangle must be rendered, so set FrontFaceFillMode (dw1.6:5)
438 * and BackFaceFillMode (dw1.4:3) to SOLID(0).
440 * From the Sandy Bridge PRM, Volume 2, Part 1, Section
441 * 6.4.1.1 3DSTATE_SF, Field FrontFaceFillMode:
442 * SOLID: Any triangle or rectangle object found to be front-facing
443 * is rendered as a solid object. This setting is required when
444 * (rendering rectangle (RECTLIST) objects.
448 OUT_BATCH(_3DSTATE_SF
<< 16 | (7 - 2));
449 OUT_BATCH(params
->depth_format
<<
450 GEN7_SF_DEPTH_BUFFER_SURFACE_FORMAT_SHIFT
);
451 OUT_BATCH(params
->num_samples
> 1 ? GEN6_SF_MSRAST_ON_PATTERN
: 0);
462 OUT_BATCH(_3DSTATE_SBE
<< 16 | (14 - 2));
463 OUT_BATCH((1 - 1) << GEN7_SBE_NUM_OUTPUTS_SHIFT
| /* only position */
464 1 << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT
|
465 0 << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT
);
466 for (int i
= 0; i
< 12; ++i
)
474 * Disable thread dispatch (dw5.19) and enable the HiZ op.
477 gen7_blorp_emit_wm_config(struct brw_context
*brw
,
478 const brw_blorp_params
*params
,
479 brw_blorp_prog_data
*prog_data
)
481 uint32_t dw1
= 0, dw2
= 0;
483 switch (params
->hiz_op
) {
484 case GEN6_HIZ_OP_DEPTH_CLEAR
:
485 dw1
|= GEN7_WM_DEPTH_CLEAR
;
487 case GEN6_HIZ_OP_DEPTH_RESOLVE
:
488 dw1
|= GEN7_WM_DEPTH_RESOLVE
;
490 case GEN6_HIZ_OP_HIZ_RESOLVE
:
491 dw1
|= GEN7_WM_HIERARCHICAL_DEPTH_RESOLVE
;
493 case GEN6_HIZ_OP_NONE
:
499 dw1
|= GEN7_WM_LINE_AA_WIDTH_1_0
;
500 dw1
|= GEN7_WM_LINE_END_CAP_AA_WIDTH_0_5
;
501 dw1
|= 0 << GEN7_WM_BARYCENTRIC_INTERPOLATION_MODE_SHIFT
; /* No interp */
502 if (params
->use_wm_prog
) {
503 dw1
|= GEN7_WM_KILL_ENABLE
; /* TODO: temporarily smash on */
504 dw1
|= GEN7_WM_DISPATCH_ENABLE
; /* We are rendering */
507 if (params
->num_samples
> 1) {
508 dw1
|= GEN7_WM_MSRAST_ON_PATTERN
;
509 if (prog_data
&& prog_data
->persample_msaa_dispatch
)
510 dw2
|= GEN7_WM_MSDISPMODE_PERSAMPLE
;
512 dw2
|= GEN7_WM_MSDISPMODE_PERPIXEL
;
514 dw1
|= GEN7_WM_MSRAST_OFF_PIXEL
;
515 dw2
|= GEN7_WM_MSDISPMODE_PERSAMPLE
;
519 OUT_BATCH(_3DSTATE_WM
<< 16 | (3 - 2));
529 * Pixel shader dispatch is disabled above in 3DSTATE_WM, dw1.29. Despite
530 * that, thread dispatch info must still be specified.
531 * - Maximum Number of Threads (dw4.24:31) must be nonzero, as the
532 * valid range for this field is [0x3, 0x2f].
533 * - A dispatch mode must be given; that is, at least one of the
534 * "N Pixel Dispatch Enable" (N=8,16,32) fields must be set. This was
535 * discovered through simulator error messages.
538 gen7_blorp_emit_ps_config(struct brw_context
*brw
,
539 const brw_blorp_params
*params
,
540 uint32_t prog_offset
,
541 brw_blorp_prog_data
*prog_data
)
543 uint32_t dw2
, dw4
, dw5
;
544 const int max_threads_shift
= brw
->is_haswell
?
545 HSW_PS_MAX_THREADS_SHIFT
: IVB_PS_MAX_THREADS_SHIFT
;
548 dw4
|= (brw
->max_wm_threads
- 1) << max_threads_shift
;
550 /* If there's a WM program, we need to do 16-pixel dispatch since that's
551 * what the program is compiled for. If there isn't, then it shouldn't
552 * matter because no program is actually being run. However, the hardware
553 * gets angry if we don't enable at least one dispatch mode, so just enable
554 * 16-pixel dispatch unconditionally.
556 dw4
|= GEN7_PS_16_DISPATCH_ENABLE
;
559 dw4
|= SET_FIELD(1, HSW_PS_SAMPLE_MASK
); /* 1 sample for now */
560 if (params
->use_wm_prog
) {
561 dw2
|= 1 << GEN7_PS_SAMPLER_COUNT_SHIFT
; /* Up to 4 samplers */
562 dw4
|= GEN7_PS_PUSH_CONSTANT_ENABLE
;
563 dw5
|= prog_data
->first_curbe_grf
<< GEN7_PS_DISPATCH_START_GRF_SHIFT_0
;
566 switch (params
->fast_clear_op
) {
567 case GEN7_FAST_CLEAR_OP_FAST_CLEAR
:
568 dw4
|= GEN7_PS_RENDER_TARGET_FAST_CLEAR_ENABLE
;
570 case GEN7_FAST_CLEAR_OP_RESOLVE
:
571 dw4
|= GEN7_PS_RENDER_TARGET_RESOLVE_ENABLE
;
578 OUT_BATCH(_3DSTATE_PS
<< 16 | (8 - 2));
579 OUT_BATCH(params
->use_wm_prog
? prog_offset
: 0);
591 gen7_blorp_emit_binding_table_pointers_ps(struct brw_context
*brw
,
592 const brw_blorp_params
*params
,
593 uint32_t wm_bind_bo_offset
)
596 OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_PS
<< 16 | (2 - 2));
597 OUT_BATCH(wm_bind_bo_offset
);
603 gen7_blorp_emit_sampler_state_pointers_ps(struct brw_context
*brw
,
604 const brw_blorp_params
*params
,
605 uint32_t sampler_offset
)
608 OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_PS
<< 16 | (2 - 2));
609 OUT_BATCH(sampler_offset
);
615 gen7_blorp_emit_constant_ps(struct brw_context
*brw
,
616 const brw_blorp_params
*params
,
617 uint32_t wm_push_const_offset
)
619 uint8_t mocs
= brw
->is_haswell
? GEN7_MOCS_L3
: 0;
621 /* Make sure the push constants fill an exact integer number of
624 assert(sizeof(brw_blorp_wm_push_constants
) % 32 == 0);
626 /* There must be at least one register worth of push constant data. */
627 assert(BRW_BLORP_NUM_PUSH_CONST_REGS
> 0);
629 /* Enable push constant buffer 0. */
631 OUT_BATCH(_3DSTATE_CONSTANT_PS
<< 16 |
633 OUT_BATCH(BRW_BLORP_NUM_PUSH_CONST_REGS
);
635 OUT_BATCH(wm_push_const_offset
| mocs
);
643 gen7_blorp_emit_constant_ps_disable(struct brw_context
*brw
,
644 const brw_blorp_params
*params
)
647 OUT_BATCH(_3DSTATE_CONSTANT_PS
<< 16 | (7 - 2));
658 gen7_blorp_emit_depth_stencil_config(struct brw_context
*brw
,
659 const brw_blorp_params
*params
)
661 struct gl_context
*ctx
= &brw
->ctx
;
662 uint32_t draw_x
= params
->depth
.x_offset
;
663 uint32_t draw_y
= params
->depth
.y_offset
;
664 uint32_t tile_mask_x
, tile_mask_y
;
665 uint8_t mocs
= brw
->is_haswell
? GEN7_MOCS_L3
: 0;
666 uint32_t surfwidth
, surfheight
;
668 unsigned int depth
= MAX2(params
->depth
.mt
->logical_depth0
, 1);
669 unsigned int min_array_element
;
670 GLenum gl_target
= params
->depth
.mt
->target
;
673 brw_get_depthstencil_tile_masks(params
->depth
.mt
,
677 &tile_mask_x
, &tile_mask_y
);
679 case GL_TEXTURE_CUBE_MAP_ARRAY
:
680 case GL_TEXTURE_CUBE_MAP
:
681 /* The PRM claims that we should use BRW_SURFACE_CUBE for this
682 * situation, but experiments show that gl_Layer doesn't work when we do
683 * this. So we use BRW_SURFACE_2D, since for rendering purposes this is
686 surftype
= BRW_SURFACE_2D
;
690 surftype
= translate_tex_target(gl_target
);
694 min_array_element
= params
->depth
.layer
;
695 if (params
->depth
.mt
->num_samples
> 1) {
696 /* Convert physical layer to logical layer. */
697 min_array_element
/= params
->depth
.mt
->num_samples
;
700 lod
= params
->depth
.level
- params
->depth
.mt
->first_level
;
702 if (params
->hiz_op
!= GEN6_HIZ_OP_NONE
&& lod
== 0) {
703 /* HIZ ops for lod 0 may set the width & height a little
704 * larger to allow the fast depth clear to fit the hardware
705 * alignment requirements. (8x4)
707 surfwidth
= params
->depth
.width
;
708 surfheight
= params
->depth
.height
;
710 surfwidth
= params
->depth
.mt
->physical_width0
;
711 surfheight
= params
->depth
.mt
->physical_height0
;
714 /* 3DSTATE_DEPTH_BUFFER */
716 uint32_t tile_x
= draw_x
& tile_mask_x
;
717 uint32_t tile_y
= draw_y
& tile_mask_y
;
719 intel_region_get_aligned_offset(params
->depth
.mt
->region
,
720 draw_x
& ~tile_mask_x
,
721 draw_y
& ~tile_mask_y
, false);
723 /* According to the Sandy Bridge PRM, volume 2 part 1, pp326-327
724 * (3DSTATE_DEPTH_BUFFER dw5), in the documentation for "Depth
725 * Coordinate Offset X/Y":
727 * "The 3 LSBs of both offsets must be zero to ensure correct
730 * We have no guarantee that tile_x and tile_y are correctly aligned,
731 * since they are determined by the mipmap layout, which is only aligned
734 * So, to avoid hanging the GPU, just smash the low order 3 bits of
735 * tile_x and tile_y to 0. This is a temporary workaround until we come
736 * up with a better solution.
738 WARN_ONCE((tile_x
& 7) || (tile_y
& 7),
739 "Depth/stencil buffer needs alignment to 8-pixel boundaries.\n"
740 "Truncating offset, bad rendering may occur.\n");
744 intel_emit_depth_stall_flushes(brw
);
747 OUT_BATCH(GEN7_3DSTATE_DEPTH_BUFFER
<< 16 | (7 - 2));
748 OUT_BATCH((params
->depth
.mt
->region
->pitch
- 1) |
749 params
->depth_format
<< 18 |
750 1 << 22 | /* hiz enable */
751 1 << 28 | /* depth write */
752 BRW_SURFACE_2D
<< 29);
753 OUT_RELOC(params
->depth
.mt
->region
->bo
,
754 I915_GEM_DOMAIN_RENDER
, I915_GEM_DOMAIN_RENDER
,
756 OUT_BATCH((params
->depth
.width
+ tile_x
- 1) << 4 |
757 (params
->depth
.height
+ tile_y
- 1) << 18);
765 /* 3DSTATE_HIER_DEPTH_BUFFER */
767 struct intel_region
*hiz_region
= params
->depth
.mt
->hiz_mt
->region
;
768 uint32_t hiz_offset
=
769 intel_region_get_aligned_offset(hiz_region
,
770 draw_x
& ~tile_mask_x
,
771 (draw_y
& ~tile_mask_y
) / 2, false);
774 OUT_BATCH((GEN7_3DSTATE_HIER_DEPTH_BUFFER
<< 16) | (3 - 2));
775 OUT_BATCH((mocs
<< 25) |
776 (hiz_region
->pitch
- 1));
777 OUT_RELOC(hiz_region
->bo
,
778 I915_GEM_DOMAIN_RENDER
, I915_GEM_DOMAIN_RENDER
,
783 /* 3DSTATE_STENCIL_BUFFER */
786 OUT_BATCH((GEN7_3DSTATE_STENCIL_BUFFER
<< 16) | (3 - 2));
795 gen7_blorp_emit_depth_disable(struct brw_context
*brw
,
796 const brw_blorp_params
*params
)
799 OUT_BATCH(GEN7_3DSTATE_DEPTH_BUFFER
<< 16 | (7 - 2));
800 OUT_BATCH(BRW_DEPTHFORMAT_D32_FLOAT
<< 18 | (BRW_SURFACE_NULL
<< 29));
810 /* 3DSTATE_CLEAR_PARAMS
812 * From the Ivybridge PRM, Volume 2 Part 1, Section 11.5.5.4
813 * 3DSTATE_CLEAR_PARAMS:
814 * 3DSTATE_CLEAR_PARAMS must always be programmed in the along
815 * with the other Depth/Stencil state commands(i.e. 3DSTATE_DEPTH_BUFFER,
816 * 3DSTATE_STENCIL_BUFFER, or 3DSTATE_HIER_DEPTH_BUFFER).
819 gen7_blorp_emit_clear_params(struct brw_context
*brw
,
820 const brw_blorp_params
*params
)
823 OUT_BATCH(GEN7_3DSTATE_CLEAR_PARAMS
<< 16 | (3 - 2));
824 OUT_BATCH(params
->depth
.mt
? params
->depth
.mt
->depth_clear_value
: 0);
825 OUT_BATCH(GEN7_DEPTH_CLEAR_VALID
);
832 gen7_blorp_emit_primitive(struct brw_context
*brw
,
833 const brw_blorp_params
*params
)
836 OUT_BATCH(CMD_3D_PRIM
<< 16 | (7 - 2));
837 OUT_BATCH(GEN7_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL
|
839 OUT_BATCH(3); /* vertex count per instance */
841 OUT_BATCH(1); /* instance count */
849 * \copydoc gen6_blorp_exec()
852 gen7_blorp_exec(struct brw_context
*brw
,
853 const brw_blorp_params
*params
)
855 brw_blorp_prog_data
*prog_data
= NULL
;
856 uint32_t cc_blend_state_offset
= 0;
857 uint32_t cc_state_offset
= 0;
858 uint32_t depthstencil_offset
;
859 uint32_t wm_push_const_offset
= 0;
860 uint32_t wm_bind_bo_offset
= 0;
861 uint32_t sampler_offset
= 0;
863 uint32_t prog_offset
= params
->get_wm_prog(brw
, &prog_data
);
864 gen6_blorp_emit_batch_head(brw
, params
);
865 gen6_emit_3dstate_multisample(brw
, params
->num_samples
);
866 gen6_emit_3dstate_sample_mask(brw
, params
->num_samples
, 1.0, false, ~0u);
867 gen6_blorp_emit_state_base_address(brw
, params
);
868 gen6_blorp_emit_vertices(brw
, params
);
869 gen7_blorp_emit_urb_config(brw
, params
);
870 if (params
->use_wm_prog
) {
871 cc_blend_state_offset
= gen6_blorp_emit_blend_state(brw
, params
);
872 cc_state_offset
= gen6_blorp_emit_cc_state(brw
, params
);
873 gen7_blorp_emit_blend_state_pointer(brw
, params
, cc_blend_state_offset
);
874 gen7_blorp_emit_cc_state_pointer(brw
, params
, cc_state_offset
);
876 depthstencil_offset
= gen6_blorp_emit_depth_stencil_state(brw
, params
);
877 gen7_blorp_emit_depth_stencil_state_pointers(brw
, params
,
878 depthstencil_offset
);
879 if (params
->use_wm_prog
) {
880 uint32_t wm_surf_offset_renderbuffer
;
881 uint32_t wm_surf_offset_texture
= 0;
882 wm_push_const_offset
= gen6_blorp_emit_wm_constants(brw
, params
);
883 intel_miptree_used_for_rendering(params
->dst
.mt
);
884 wm_surf_offset_renderbuffer
=
885 gen7_blorp_emit_surface_state(brw
, params
, ¶ms
->dst
,
886 I915_GEM_DOMAIN_RENDER
,
887 I915_GEM_DOMAIN_RENDER
,
888 true /* is_render_target */);
889 if (params
->src
.mt
) {
890 wm_surf_offset_texture
=
891 gen7_blorp_emit_surface_state(brw
, params
, ¶ms
->src
,
892 I915_GEM_DOMAIN_SAMPLER
, 0,
893 false /* is_render_target */);
896 gen6_blorp_emit_binding_table(brw
, params
,
897 wm_surf_offset_renderbuffer
,
898 wm_surf_offset_texture
);
899 sampler_offset
= gen7_blorp_emit_sampler_state(brw
, params
);
901 gen7_blorp_emit_vs_disable(brw
, params
);
902 gen7_blorp_emit_hs_disable(brw
, params
);
903 gen7_blorp_emit_te_disable(brw
, params
);
904 gen7_blorp_emit_ds_disable(brw
, params
);
905 gen7_blorp_emit_gs_disable(brw
, params
);
906 gen7_blorp_emit_streamout_disable(brw
, params
);
907 gen6_blorp_emit_clip_disable(brw
, params
);
908 gen7_blorp_emit_sf_config(brw
, params
);
909 gen7_blorp_emit_wm_config(brw
, params
, prog_data
);
910 if (params
->use_wm_prog
) {
911 gen7_blorp_emit_binding_table_pointers_ps(brw
, params
,
913 gen7_blorp_emit_sampler_state_pointers_ps(brw
, params
, sampler_offset
);
914 gen7_blorp_emit_constant_ps(brw
, params
, wm_push_const_offset
);
916 gen7_blorp_emit_constant_ps_disable(brw
, params
);
918 gen7_blorp_emit_ps_config(brw
, params
, prog_offset
, prog_data
);
919 gen7_blorp_emit_cc_viewport(brw
, params
);
921 if (params
->depth
.mt
)
922 gen7_blorp_emit_depth_stencil_config(brw
, params
);
924 gen7_blorp_emit_depth_disable(brw
, params
);
925 gen7_blorp_emit_clear_params(brw
, params
);
926 gen6_blorp_emit_drawing_rectangle(brw
, params
);
927 gen7_blorp_emit_primitive(brw
, params
);