2 * Copyright © 2012 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
24 #include "intel_batchbuffer.h"
25 #include "intel_fbo.h"
27 #include "brw_blorp.h"
28 #include "brw_defines.h"
29 #include "gen6_blorp.h"
30 #include "gen7_blorp.h"
32 brw_blorp_mip_info::brw_blorp_mip_info()
43 brw_blorp_surface_info::brw_blorp_surface_info()
44 : map_stencil_as_y_tiled(false),
50 brw_blorp_mip_info::set(struct intel_mipmap_tree
*mt
,
51 unsigned int level
, unsigned int layer
)
53 intel_miptree_check_level_layer(mt
, level
, layer
);
58 this->width
= mt
->level
[level
].width
;
59 this->height
= mt
->level
[level
].height
;
61 intel_miptree_get_image_offset(mt
, level
, layer
, &x_offset
, &y_offset
);
65 brw_blorp_surface_info::set(struct brw_context
*brw
,
66 struct intel_mipmap_tree
*mt
,
67 unsigned int level
, unsigned int layer
)
69 brw_blorp_mip_info::set(mt
, level
, layer
);
70 this->num_samples
= mt
->num_samples
;
71 this->array_spacing_lod0
= mt
->array_spacing_lod0
;
72 this->map_stencil_as_y_tiled
= false;
73 this->msaa_layout
= mt
->msaa_layout
;
77 /* The miptree is a W-tiled stencil buffer. Surface states can't be set
78 * up for W tiling, so we'll need to use Y tiling and have the WM
79 * program swizzle the coordinates.
81 this->map_stencil_as_y_tiled
= true;
82 this->brw_surfaceformat
= BRW_SURFACEFORMAT_R8_UNORM
;
84 case MESA_FORMAT_X8_Z24
:
85 case MESA_FORMAT_Z32_FLOAT
:
86 /* The miptree consists of 32 bits per pixel, arranged either as 24-bit
87 * depth values interleaved with 8 "don't care" bits, or as 32-bit
88 * floating point depth values. Since depth values don't require any
89 * blending, it doesn't matter how we interpret the bit pattern as long
90 * as we copy the right amount of data, so just map it as 8-bit BGRA.
92 this->brw_surfaceformat
= BRW_SURFACEFORMAT_B8G8R8A8_UNORM
;
95 /* The miptree consists of 16 bits per pixel of depth data. Since depth
96 * values don't require any blending, it doesn't matter how we interpret
97 * the bit pattern as long as we copy the right amount of data, so just
100 this->brw_surfaceformat
= BRW_SURFACEFORMAT_R8G8_UNORM
;
103 /* Blorp blits don't support any sort of format conversion (except
104 * between sRGB and linear), so we can safely assume that the format is
105 * supported as a render target, even if this is the source image. So
106 * we can convert to a surface format using brw->render_target_format.
108 assert(brw
->format_supported_as_render_target
[mt
->format
]);
109 this->brw_surfaceformat
= brw
->render_target_format
[mt
->format
];
116 * Split x_offset and y_offset into a base offset (in bytes) and a remaining
117 * x/y offset (in pixels). Note: we can't do this by calling
118 * intel_renderbuffer_tile_offsets(), because the offsets may have been
119 * adjusted to account for Y vs. W tiling differences. So we compute it
120 * directly from the adjusted offsets.
123 brw_blorp_surface_info::compute_tile_offsets(uint32_t *tile_x
,
124 uint32_t *tile_y
) const
126 struct intel_region
*region
= mt
->region
;
127 uint32_t mask_x
, mask_y
;
129 intel_region_get_tile_masks(region
, &mask_x
, &mask_y
,
130 map_stencil_as_y_tiled
);
132 *tile_x
= x_offset
& mask_x
;
133 *tile_y
= y_offset
& mask_y
;
135 return intel_region_get_aligned_offset(region
, x_offset
& ~mask_x
,
137 map_stencil_as_y_tiled
);
141 brw_blorp_params::brw_blorp_params()
147 hiz_op(GEN6_HIZ_OP_NONE
),
151 color_write_disable
[0] = false;
152 color_write_disable
[1] = false;
153 color_write_disable
[2] = false;
154 color_write_disable
[3] = false;
159 intel_hiz_exec(struct intel_context
*intel
, struct intel_mipmap_tree
*mt
,
160 unsigned int level
, unsigned int layer
, gen6_hiz_op op
)
162 brw_hiz_op_params
params(mt
, level
, layer
, op
);
163 brw_blorp_exec(intel
, ¶ms
);
169 brw_blorp_exec(struct intel_context
*intel
, const brw_blorp_params
*params
)
171 struct brw_context
*brw
= brw_context(&intel
->ctx
);
173 switch (intel
->gen
) {
175 gen6_blorp_exec(intel
, params
);
178 gen7_blorp_exec(intel
, params
);
181 /* BLORP is not supported before Gen6. */
186 if (unlikely(intel
->always_flush_batch
))
187 intel_batchbuffer_flush(intel
);
189 /* We've smashed all state compared to what the normal 3D pipeline
190 * rendering tracks for GL.
192 brw
->state
.dirty
.brw
= ~0;
193 brw
->state
.dirty
.cache
= ~0;
194 brw
->state_batch_count
= 0;
195 intel
->batch
.need_workaround_flush
= true;
197 /* Flush the sampler cache so any texturing from the destination is
200 intel_batchbuffer_emit_mi_flush(intel
);
203 brw_hiz_op_params::brw_hiz_op_params(struct intel_mipmap_tree
*mt
,
210 depth
.set(mt
, level
, layer
);
212 /* Align the rectangle primitive to 8x4 pixels.
214 * During fast depth clears, the emitted rectangle primitive must be
215 * aligned to 8x4 pixels. From the Ivybridge PRM, Vol 2 Part 1 Section
216 * 11.5.3.1 Depth Buffer Clear (and the matching section in the Sandybridge
218 * If Number of Multisamples is NUMSAMPLES_1, the rectangle must be
219 * aligned to an 8x4 pixel block relative to the upper left corner
220 * of the depth buffer [...]
222 * For hiz resolves, the rectangle must also be 8x4 aligned. Item
223 * WaHizAmbiguate8x4Aligned from the Haswell workarounds page and the
224 * Ivybridge simulator require the alignment.
226 * To be safe, let's just align the rect for all hiz operations and all
227 * hardware generations.
229 * However, for some miptree slices of a Z24 texture, emitting an 8x4
230 * aligned rectangle that covers the slice may clobber adjacent slices if
231 * we strictly adhered to the texture alignments specified in the PRM. The
232 * Ivybridge PRM, Section "Alignment Unit Size", states that
233 * SURFACE_STATE.Surface_Horizontal_Alignment should be 4 for Z24 surfaces,
234 * not 8. But commit 1f112cc increased the alignment from 4 to 8, which
235 * prevents the clobbering.
237 depth
.width
= ALIGN(depth
.width
, 8);
238 depth
.height
= ALIGN(depth
.height
, 4);
243 assert(intel_miptree_slice_has_hiz(mt
, level
, layer
));
245 switch (mt
->format
) {
246 case MESA_FORMAT_Z16
: depth_format
= BRW_DEPTHFORMAT_D16_UNORM
; break;
247 case MESA_FORMAT_Z32_FLOAT
: depth_format
= BRW_DEPTHFORMAT_D32_FLOAT
; break;
248 case MESA_FORMAT_X8_Z24
: depth_format
= BRW_DEPTHFORMAT_D24_UNORM_X8_UINT
; break;
249 default: assert(0); break;
254 brw_hiz_op_params::get_wm_prog(struct brw_context
*brw
,
255 brw_blorp_prog_data
**prog_data
) const