intel: Use rb->Data and rb->RowStride to handle spans Y flipping.
[mesa.git] / src / mesa / drivers / dri / intel / intel_span.c
1 /**************************************************************************
2 *
3 * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
4 * Copyright 2011 Intel Corporation
5 * All Rights Reserved.
6 *
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:
14 *
15 * The above copyright notice and this permission notice (including the
16 * next paragraph) shall be included in all copies or substantial portions
17 * of the Software.
18 *
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.
26 *
27 * Authors:
28 * Chad Versace <chad@chad-versace.us>
29 *
30 **************************************************************************/
31
32 #include <stdbool.h>
33 #include <stdint.h>
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"
39
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"
46
47 #include "swrast/swrast.h"
48
49 static void
50 intel_set_span_functions(struct intel_context *intel,
51 struct gl_renderbuffer *rb);
52
53 #undef DBG
54 #define DBG 0
55
56 #define LOCAL_VARS \
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; \
63 GLuint p; \
64 (void) p;
65
66 #define HW_CLIPLOOP()
67 #define HW_ENDCLIPLOOP()
68
69 #define Y_FLIP(_y) (_y)
70
71 #define HW_LOCK()
72
73 #define HW_UNLOCK()
74
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
80 #include "spantmp2.h"
81
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
87 #include "spantmp2.h"
88
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
94 #include "spantmp2.h"
95
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"
102
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"
109
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"
116
117 /* ------------------------------------------------------------------------- */
118 /* s8 stencil span and pixel functions */
119 /* ------------------------------------------------------------------------- */
120
121 /*
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
124 * work.
125 */
126 #define HAVE_HW_STENCIL_SPANS 0
127
128 #define LOCAL_STENCIL_VARS \
129 (void) ctx; \
130 int minx = 0; \
131 int miny = 0; \
132 int maxx = rb->Width; \
133 int maxy = rb->Height; \
134 int stride = rb->RowStride; \
135 uint8_t *buf = rb->Data; \
136
137 /* Don't flip y. */
138 #undef Y_FLIP
139 #define Y_FLIP(y) y
140
141 /**
142 * \brief Get pointer offset into stencil buffer.
143 *
144 * The stencil buffer interleaves two rows into one. Yay for crazy hardware.
145 * The table below demonstrates how the pointer arithmetic behaves for a buffer
146 * with positive stride (s=stride).
147 *
148 * x | y | byte offset
149 * --------------------------
150 * 0 | 0 | 0
151 * 0 | 0 | 1
152 * 1 | 0 | 2
153 * 1 | 1 | 3
154 * ... | ... | ...
155 * 0 | 2 | s
156 * 0 | 3 | s + 1
157 * 1 | 2 | s + 2
158 * 1 | 3 | s + 3
159 *
160 *
161 */
162 static inline intptr_t
163 intel_offset_S8(int stride, GLint x, GLint y)
164 {
165 return 2 * ((y / 2) * stride + x) + y % 2;
166 }
167
168 #define WRITE_STENCIL(x, y, src) buf[intel_offset_S8(stride, x, y)] = src;
169 #define READ_STENCIL(dest, x, y) dest = buf[intel_offset_S8(stride, x, y)]
170 #define TAG(x) intel_##x##_S8
171 #include "stenciltmp.h"
172
173 /* ------------------------------------------------------------------------- */
174
175 void
176 intel_renderbuffer_map(struct intel_context *intel, struct gl_renderbuffer *rb)
177 {
178 struct intel_renderbuffer *irb = intel_renderbuffer(rb);
179
180 if (irb == NULL || irb->region == NULL)
181 return;
182
183 drm_intel_gem_bo_map_gtt(irb->region->buffer);
184
185 rb->Data = irb->region->buffer->virtual;
186 rb->RowStride = irb->region->pitch;
187
188 /* Flip orientation if it's the window system buffer */
189 if (!rb->Name) {
190 rb->Data += rb->RowStride * (irb->region->height - 1) * irb->region->cpp;
191 rb->RowStride = -rb->RowStride;
192 }
193
194 intel_set_span_functions(intel, rb);
195 }
196
197 void
198 intel_renderbuffer_unmap(struct intel_context *intel,
199 struct gl_renderbuffer *rb)
200 {
201 struct intel_renderbuffer *irb = intel_renderbuffer(rb);
202
203 if (irb == NULL || irb->region == NULL)
204 return;
205
206 drm_intel_gem_bo_unmap_gtt(irb->region->buffer);
207
208 rb->GetRow = NULL;
209 rb->PutRow = NULL;
210 rb->Data = NULL;
211 rb->RowStride = 0;
212 }
213
214 /**
215 * Map or unmap all the renderbuffers which we may need during
216 * software rendering.
217 * XXX in the future, we could probably convey extra information to
218 * reduce the number of mappings needed. I.e. if doing a glReadPixels
219 * from the depth buffer, we really only need one mapping.
220 *
221 * XXX Rewrite this function someday.
222 * We can probably just loop over all the renderbuffer attachments,
223 * map/unmap all of them, and not worry about the _ColorDrawBuffers
224 * _ColorReadBuffer, _DepthBuffer or _StencilBuffer fields.
225 */
226 static void
227 intel_map_unmap_framebuffer(struct intel_context *intel,
228 struct gl_framebuffer *fb,
229 GLboolean map)
230 {
231 GLuint i;
232
233 /* color draw buffers */
234 for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
235 if (map)
236 intel_renderbuffer_map(intel, fb->_ColorDrawBuffers[i]);
237 else
238 intel_renderbuffer_unmap(intel, fb->_ColorDrawBuffers[i]);
239 }
240
241 /* color read buffer */
242 if (map)
243 intel_renderbuffer_map(intel, fb->_ColorReadBuffer);
244 else
245 intel_renderbuffer_unmap(intel, fb->_ColorReadBuffer);
246
247 /* check for render to textures */
248 for (i = 0; i < BUFFER_COUNT; i++) {
249 struct gl_renderbuffer_attachment *att =
250 fb->Attachment + i;
251 struct gl_texture_object *tex = att->Texture;
252 if (tex) {
253 /* render to texture */
254 ASSERT(att->Renderbuffer);
255 if (map)
256 intel_tex_map_images(intel, intel_texture_object(tex));
257 else
258 intel_tex_unmap_images(intel, intel_texture_object(tex));
259 }
260 }
261
262 /* depth buffer (Note wrapper!) */
263 if (fb->_DepthBuffer) {
264 if (map)
265 intel_renderbuffer_map(intel, fb->_DepthBuffer->Wrapped);
266 else
267 intel_renderbuffer_unmap(intel, fb->_DepthBuffer->Wrapped);
268 }
269
270 /* stencil buffer (Note wrapper!) */
271 if (fb->_StencilBuffer) {
272 if (map)
273 intel_renderbuffer_map(intel, fb->_StencilBuffer->Wrapped);
274 else
275 intel_renderbuffer_unmap(intel, fb->_StencilBuffer->Wrapped);
276 }
277
278 intel_check_front_buffer_rendering(intel);
279 }
280
281 /**
282 * Prepare for software rendering. Map current read/draw framebuffers'
283 * renderbuffes and all currently bound texture objects.
284 *
285 * Old note: Moved locking out to get reasonable span performance.
286 */
287 void
288 intelSpanRenderStart(struct gl_context * ctx)
289 {
290 struct intel_context *intel = intel_context(ctx);
291 GLuint i;
292
293 intel_flush(&intel->ctx);
294 intel_prepare_render(intel);
295
296 for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
297 if (ctx->Texture.Unit[i]._ReallyEnabled) {
298 struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
299
300 intel_finalize_mipmap_tree(intel, i);
301 intel_tex_map_images(intel, intel_texture_object(texObj));
302 }
303 }
304
305 intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_TRUE);
306 if (ctx->ReadBuffer != ctx->DrawBuffer)
307 intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE);
308 }
309
310 /**
311 * Called when done software rendering. Unmap the buffers we mapped in
312 * the above function.
313 */
314 void
315 intelSpanRenderFinish(struct gl_context * ctx)
316 {
317 struct intel_context *intel = intel_context(ctx);
318 GLuint i;
319
320 _swrast_flush(ctx);
321
322 for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
323 if (ctx->Texture.Unit[i]._ReallyEnabled) {
324 struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
325 intel_tex_unmap_images(intel, intel_texture_object(texObj));
326 }
327 }
328
329 intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_FALSE);
330 if (ctx->ReadBuffer != ctx->DrawBuffer)
331 intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_FALSE);
332 }
333
334
335 void
336 intelInitSpanFuncs(struct gl_context * ctx)
337 {
338 struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
339 swdd->SpanRenderStart = intelSpanRenderStart;
340 swdd->SpanRenderFinish = intelSpanRenderFinish;
341 }
342
343 void
344 intel_map_vertex_shader_textures(struct gl_context *ctx)
345 {
346 struct intel_context *intel = intel_context(ctx);
347 int i;
348
349 if (ctx->VertexProgram._Current == NULL)
350 return;
351
352 for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
353 if (ctx->Texture.Unit[i]._ReallyEnabled &&
354 ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) {
355 struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
356
357 intel_tex_map_images(intel, intel_texture_object(texObj));
358 }
359 }
360 }
361
362 void
363 intel_unmap_vertex_shader_textures(struct gl_context *ctx)
364 {
365 struct intel_context *intel = intel_context(ctx);
366 int i;
367
368 if (ctx->VertexProgram._Current == NULL)
369 return;
370
371 for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
372 if (ctx->Texture.Unit[i]._ReallyEnabled &&
373 ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) {
374 struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
375
376 intel_tex_unmap_images(intel, intel_texture_object(texObj));
377 }
378 }
379 }
380
381 typedef void (*span_init_func)(struct gl_renderbuffer *rb);
382
383 static span_init_func intel_span_init_funcs[MESA_FORMAT_COUNT] =
384 {
385 [MESA_FORMAT_A8] = intel_InitPointers_A8,
386 [MESA_FORMAT_RGB565] = intel_InitPointers_RGB565,
387 [MESA_FORMAT_ARGB4444] = intel_InitPointers_ARGB4444,
388 [MESA_FORMAT_ARGB1555] = intel_InitPointers_ARGB1555,
389 [MESA_FORMAT_XRGB8888] = intel_InitPointers_xRGB8888,
390 [MESA_FORMAT_ARGB8888] = intel_InitPointers_ARGB8888,
391 [MESA_FORMAT_SARGB8] = intel_InitPointers_ARGB8888,
392 [MESA_FORMAT_Z16] = _mesa_set_renderbuffer_accessors,
393 [MESA_FORMAT_X8_Z24] = _mesa_set_renderbuffer_accessors,
394 [MESA_FORMAT_S8_Z24] = _mesa_set_renderbuffer_accessors,
395 [MESA_FORMAT_S8] = intel_InitStencilPointers_S8,
396 [MESA_FORMAT_R8] = _mesa_set_renderbuffer_accessors,
397 [MESA_FORMAT_RG88] = _mesa_set_renderbuffer_accessors,
398 [MESA_FORMAT_R16] = _mesa_set_renderbuffer_accessors,
399 [MESA_FORMAT_RG1616] = _mesa_set_renderbuffer_accessors,
400 [MESA_FORMAT_RGBA_FLOAT32] = _mesa_set_renderbuffer_accessors,
401 [MESA_FORMAT_RG_FLOAT32] = _mesa_set_renderbuffer_accessors,
402 [MESA_FORMAT_R_FLOAT32] = _mesa_set_renderbuffer_accessors,
403 [MESA_FORMAT_INTENSITY_FLOAT32] = _mesa_set_renderbuffer_accessors,
404 [MESA_FORMAT_LUMINANCE_FLOAT32] = _mesa_set_renderbuffer_accessors,
405 };
406
407 bool
408 intel_span_supports_format(gl_format format)
409 {
410 return intel_span_init_funcs[format] != NULL;
411 }
412
413 /**
414 * Plug in appropriate span read/write functions for the given renderbuffer.
415 * These are used for the software fallbacks.
416 */
417 static void
418 intel_set_span_functions(struct intel_context *intel,
419 struct gl_renderbuffer *rb)
420 {
421 struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb;
422
423 assert(intel_span_init_funcs[irb->Base.Format]);
424 intel_span_init_funcs[irb->Base.Format](rb);
425 }