1 /**************************************************************************
3 * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
4 * Copyright 2011 Intel Corporation
7 * Permission is hereby granted, free of charge, to any person obtaining a
8 * copy of this software and associated documentation files (the
9 * "Software"), to deal in the Software without restriction, including
10 * without limitation the rights to use, copy, modify, merge, publish,
11 * distribute, sub license, and/or sell copies of the Software, and to
12 * permit persons to whom the Software is furnished to do so, subject to
13 * the following conditions:
15 * The above copyright notice and this permission notice (including the
16 * next paragraph) shall be included in all copies or substantial portions
19 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
20 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
21 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
22 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
23 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
24 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
25 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
28 * Chad Versace <chad@chad-versace.us>
30 **************************************************************************/
34 #include "main/glheader.h"
35 #include "main/macros.h"
36 #include "main/mtypes.h"
37 #include "main/colormac.h"
38 #include "main/renderbuffer.h"
40 #include "intel_buffers.h"
41 #include "intel_fbo.h"
42 #include "intel_screen.h"
43 #include "intel_span.h"
44 #include "intel_regions.h"
45 #include "intel_tex.h"
47 #include "swrast/swrast.h"
50 intel_set_span_functions(struct intel_context
*intel
,
51 struct gl_renderbuffer
*rb
);
57 struct intel_renderbuffer *irb = intel_renderbuffer(rb); \
58 int minx = 0, miny = 0; \
59 int maxx = rb->Width; \
60 int maxy = rb->Height; \
61 int pitch = rb->RowStride * irb->region->cpp; \
62 void *buf = rb->Data; \
67 #define HW_ENDCLIPLOOP()
69 #define Y_FLIP(_y) (_y)
75 /* r5g6b5 color span and pixel functions */
76 #define SPANTMP_PIXEL_FMT GL_RGB
77 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
78 #define TAG(x) intel_##x##_RGB565
79 #define TAG2(x,y) intel_##x##y_RGB565
82 /* a4r4g4b4 color span and pixel functions */
83 #define SPANTMP_PIXEL_FMT GL_BGRA
84 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_4_4_4_4_REV
85 #define TAG(x) intel_##x##_ARGB4444
86 #define TAG2(x,y) intel_##x##y_ARGB4444
89 /* a1r5g5b5 color span and pixel functions */
90 #define SPANTMP_PIXEL_FMT GL_BGRA
91 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_1_5_5_5_REV
92 #define TAG(x) intel_##x##_ARGB1555
93 #define TAG2(x,y) intel_##x##y##_ARGB1555
96 /* a8r8g8b8 color span and pixel functions */
97 #define SPANTMP_PIXEL_FMT GL_BGRA
98 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
99 #define TAG(x) intel_##x##_ARGB8888
100 #define TAG2(x,y) intel_##x##y##_ARGB8888
101 #include "spantmp2.h"
103 /* x8r8g8b8 color span and pixel functions */
104 #define SPANTMP_PIXEL_FMT GL_BGR
105 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
106 #define TAG(x) intel_##x##_xRGB8888
107 #define TAG2(x,y) intel_##x##y##_xRGB8888
108 #include "spantmp2.h"
110 /* a8 color span and pixel functions */
111 #define SPANTMP_PIXEL_FMT GL_ALPHA
112 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_BYTE
113 #define TAG(x) intel_##x##_A8
114 #define TAG2(x,y) intel_##x##y##_A8
115 #include "spantmp2.h"
117 /* ------------------------------------------------------------------------- */
118 /* s8 stencil span and pixel functions */
119 /* ------------------------------------------------------------------------- */
122 * HAVE_HW_STENCIL_SPANS determines if stencil buffer read/writes are done with
123 * memcpy or for loops. Since the stencil buffer is interleaved, memcpy won't
126 #define HAVE_HW_STENCIL_SPANS 0
128 #define LOCAL_STENCIL_VARS \
132 int maxx = rb->Width; \
133 int maxy = rb->Height; \
136 * Here we ignore rb->Data and rb->RowStride as set by \
137 * intelSpanRenderStart. Since intel_offset_S8 decodes the W tile \
138 * manually, the region's *real* base address and stride is \
141 struct intel_renderbuffer *irb = intel_renderbuffer(rb); \
142 uint8_t *buf = irb->region->buffer->virtual; \
143 unsigned stride = irb->region->pitch; \
144 unsigned height = 2 * irb->region->height; \
145 bool flip = rb->Name == 0; \
146 int y_scale = flip ? -1 : 1; \
147 int y_bias = flip ? (height - 1) : 0; \
150 #define Y_FLIP(y) (y_scale * (y) + y_bias)
153 * \brief Get pointer offset into stencil buffer.
155 * The stencil buffer is W tiled. Since the GTT is incapable of W fencing, we
156 * must decode the tile's layout in software.
159 * - PRM, 2011 Sandy Bridge, Volume 1, Part 2, Section 4.5.2.1 W-Major Tile
161 * - PRM, 2011 Sandy Bridge, Volume 1, Part 2, Section 4.5.3 Tiling Algorithm
163 * Even though the returned offset is always positive, the return type is
165 * commit e8b1c6d6f55f5be3bef25084fdd8b6127517e137
166 * mesa: Fix return type of _mesa_get_format_bytes() (#37351)
168 static inline intptr_t
169 intel_offset_S8(uint32_t stride
, uint32_t x
, uint32_t y
)
171 uint32_t tile_size
= 4096;
172 uint32_t tile_width
= 64;
173 uint32_t tile_height
= 64;
174 uint32_t row_size
= 64 * stride
;
176 uint32_t tile_x
= x
/ tile_width
;
177 uint32_t tile_y
= y
/ tile_height
;
179 /* The byte's address relative to the tile's base addres. */
180 uint32_t byte_x
= x
% tile_width
;
181 uint32_t byte_y
= y
% tile_height
;
183 uintptr_t u
= tile_y
* row_size
187 + 32 * ((byte_y
/ 4) % 2)
188 + 16 * ((byte_x
/ 4) % 2)
189 + 8 * ((byte_y
/ 2) % 2)
190 + 4 * ((byte_x
/ 2) % 2)
197 * An additional offset is needed which is not documented in the PRM.
199 * if ((byte_x / 8) % 2 == 1) {
200 * if ((byte_y / 8) % 2) == 0) {
207 * The offset is expressed more tersely as
208 * u += ((int) x & 0x8) * (8 - (((int) y & 0x8) << 1));
214 #define WRITE_STENCIL(x, y, src) buf[intel_offset_S8(stride, x, y)] = src;
215 #define READ_STENCIL(dest, x, y) dest = buf[intel_offset_S8(stride, x, y)]
216 #define TAG(x) intel_##x##_S8
217 #include "stenciltmp.h"
219 /* ------------------------------------------------------------------------- */
222 intel_renderbuffer_map(struct intel_context
*intel
, struct gl_renderbuffer
*rb
)
224 struct intel_renderbuffer
*irb
= intel_renderbuffer(rb
);
229 if (irb
->wrapped_depth
)
230 intel_renderbuffer_map(intel
, irb
->wrapped_depth
);
231 if (irb
->wrapped_stencil
)
232 intel_renderbuffer_map(intel
, irb
->wrapped_stencil
);
237 drm_intel_gem_bo_map_gtt(irb
->region
->buffer
);
239 rb
->Data
= irb
->region
->buffer
->virtual;
240 rb
->RowStride
= irb
->region
->pitch
;
243 /* Flip orientation of the window system buffer */
244 rb
->Data
+= rb
->RowStride
* (irb
->region
->height
- 1) * irb
->region
->cpp
;
245 rb
->RowStride
= -rb
->RowStride
;
247 /* Adjust the base pointer of a texture image drawbuffer to the image
248 * within the miptree region (all else has draw_x/y = 0).
250 rb
->Data
+= irb
->draw_x
* irb
->region
->cpp
;
251 rb
->Data
+= irb
->draw_y
* rb
->RowStride
* irb
->region
->cpp
;
254 intel_set_span_functions(intel
, rb
);
258 intel_renderbuffer_unmap(struct intel_context
*intel
,
259 struct gl_renderbuffer
*rb
)
261 struct intel_renderbuffer
*irb
= intel_renderbuffer(rb
);
266 if (irb
->wrapped_depth
)
267 intel_renderbuffer_unmap(intel
, irb
->wrapped_depth
);
268 if (irb
->wrapped_stencil
)
269 intel_renderbuffer_unmap(intel
, irb
->wrapped_stencil
);
274 drm_intel_gem_bo_unmap_gtt(irb
->region
->buffer
);
283 intel_framebuffer_map(struct intel_context
*intel
, struct gl_framebuffer
*fb
)
287 for (i
= 0; i
< BUFFER_COUNT
; i
++) {
288 intel_renderbuffer_map(intel
, fb
->Attachment
[i
].Renderbuffer
);
291 intel_check_front_buffer_rendering(intel
);
295 intel_framebuffer_unmap(struct intel_context
*intel
, struct gl_framebuffer
*fb
)
299 for (i
= 0; i
< BUFFER_COUNT
; i
++) {
300 intel_renderbuffer_unmap(intel
, fb
->Attachment
[i
].Renderbuffer
);
305 * Prepare for software rendering. Map current read/draw framebuffers'
306 * renderbuffes and all currently bound texture objects.
308 * Old note: Moved locking out to get reasonable span performance.
311 intelSpanRenderStart(struct gl_context
* ctx
)
313 struct intel_context
*intel
= intel_context(ctx
);
316 intel_flush(&intel
->ctx
);
317 intel_prepare_render(intel
);
319 for (i
= 0; i
< ctx
->Const
.MaxTextureImageUnits
; i
++) {
320 if (ctx
->Texture
.Unit
[i
]._ReallyEnabled
) {
321 struct gl_texture_object
*texObj
= ctx
->Texture
.Unit
[i
]._Current
;
323 intel_finalize_mipmap_tree(intel
, i
);
324 intel_tex_map_images(intel
, intel_texture_object(texObj
));
328 intel_framebuffer_map(intel
, ctx
->DrawBuffer
);
329 if (ctx
->ReadBuffer
!= ctx
->DrawBuffer
) {
330 intel_framebuffer_map(intel
, ctx
->ReadBuffer
);
335 * Called when done software rendering. Unmap the buffers we mapped in
336 * the above function.
339 intelSpanRenderFinish(struct gl_context
* ctx
)
341 struct intel_context
*intel
= intel_context(ctx
);
346 for (i
= 0; i
< ctx
->Const
.MaxTextureImageUnits
; i
++) {
347 if (ctx
->Texture
.Unit
[i
]._ReallyEnabled
) {
348 struct gl_texture_object
*texObj
= ctx
->Texture
.Unit
[i
]._Current
;
349 intel_tex_unmap_images(intel
, intel_texture_object(texObj
));
353 intel_framebuffer_unmap(intel
, ctx
->DrawBuffer
);
354 if (ctx
->ReadBuffer
!= ctx
->DrawBuffer
) {
355 intel_framebuffer_unmap(intel
, ctx
->ReadBuffer
);
361 intelInitSpanFuncs(struct gl_context
* ctx
)
363 struct swrast_device_driver
*swdd
= _swrast_GetDeviceDriverReference(ctx
);
364 swdd
->SpanRenderStart
= intelSpanRenderStart
;
365 swdd
->SpanRenderFinish
= intelSpanRenderFinish
;
369 intel_map_vertex_shader_textures(struct gl_context
*ctx
)
371 struct intel_context
*intel
= intel_context(ctx
);
374 if (ctx
->VertexProgram
._Current
== NULL
)
377 for (i
= 0; i
< ctx
->Const
.MaxTextureImageUnits
; i
++) {
378 if (ctx
->Texture
.Unit
[i
]._ReallyEnabled
&&
379 ctx
->VertexProgram
._Current
->Base
.TexturesUsed
[i
] != 0) {
380 struct gl_texture_object
*texObj
= ctx
->Texture
.Unit
[i
]._Current
;
382 intel_tex_map_images(intel
, intel_texture_object(texObj
));
388 intel_unmap_vertex_shader_textures(struct gl_context
*ctx
)
390 struct intel_context
*intel
= intel_context(ctx
);
393 if (ctx
->VertexProgram
._Current
== NULL
)
396 for (i
= 0; i
< ctx
->Const
.MaxTextureImageUnits
; i
++) {
397 if (ctx
->Texture
.Unit
[i
]._ReallyEnabled
&&
398 ctx
->VertexProgram
._Current
->Base
.TexturesUsed
[i
] != 0) {
399 struct gl_texture_object
*texObj
= ctx
->Texture
.Unit
[i
]._Current
;
401 intel_tex_unmap_images(intel
, intel_texture_object(texObj
));
406 typedef void (*span_init_func
)(struct gl_renderbuffer
*rb
);
408 static span_init_func intel_span_init_funcs
[MESA_FORMAT_COUNT
] =
410 [MESA_FORMAT_A8
] = intel_InitPointers_A8
,
411 [MESA_FORMAT_RGB565
] = intel_InitPointers_RGB565
,
412 [MESA_FORMAT_ARGB4444
] = intel_InitPointers_ARGB4444
,
413 [MESA_FORMAT_ARGB1555
] = intel_InitPointers_ARGB1555
,
414 [MESA_FORMAT_XRGB8888
] = intel_InitPointers_xRGB8888
,
415 [MESA_FORMAT_ARGB8888
] = intel_InitPointers_ARGB8888
,
416 [MESA_FORMAT_SARGB8
] = intel_InitPointers_ARGB8888
,
417 [MESA_FORMAT_Z16
] = _mesa_set_renderbuffer_accessors
,
418 [MESA_FORMAT_X8_Z24
] = _mesa_set_renderbuffer_accessors
,
419 [MESA_FORMAT_S8_Z24
] = _mesa_set_renderbuffer_accessors
,
420 [MESA_FORMAT_S8
] = intel_InitStencilPointers_S8
,
421 [MESA_FORMAT_R8
] = _mesa_set_renderbuffer_accessors
,
422 [MESA_FORMAT_RG88
] = _mesa_set_renderbuffer_accessors
,
423 [MESA_FORMAT_R16
] = _mesa_set_renderbuffer_accessors
,
424 [MESA_FORMAT_RG1616
] = _mesa_set_renderbuffer_accessors
,
425 [MESA_FORMAT_RGBA_FLOAT32
] = _mesa_set_renderbuffer_accessors
,
426 [MESA_FORMAT_RG_FLOAT32
] = _mesa_set_renderbuffer_accessors
,
427 [MESA_FORMAT_R_FLOAT32
] = _mesa_set_renderbuffer_accessors
,
428 [MESA_FORMAT_INTENSITY_FLOAT32
] = _mesa_set_renderbuffer_accessors
,
429 [MESA_FORMAT_LUMINANCE_FLOAT32
] = _mesa_set_renderbuffer_accessors
,
433 intel_span_supports_format(gl_format format
)
435 return intel_span_init_funcs
[format
] != NULL
;
439 * Plug in appropriate span read/write functions for the given renderbuffer.
440 * These are used for the software fallbacks.
443 intel_set_span_functions(struct intel_context
*intel
,
444 struct gl_renderbuffer
*rb
)
446 struct intel_renderbuffer
*irb
= (struct intel_renderbuffer
*) rb
;
448 assert(intel_span_init_funcs
[irb
->Base
.Format
]);
449 intel_span_init_funcs
[irb
->Base
.Format
](rb
);