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_mipmap_tree.h"
  43 #include "intel_screen.h"
  44 #include "intel_span.h"
  45 #include "intel_regions.h"
  46 #include "intel_tex.h"
  47
  48 #include "swrast/swrast.h"
  49 #include "swrast/s_renderbuffer.h"
  50
  51 /**
  52  * \brief Get pointer offset into stencil buffer.
  53  *
  54  * The stencil buffer is W tiled. Since the GTT is incapable of W fencing, we
  55  * must decode the tile's layout in software.
  56  *
  57  * See
  58  *   - PRM, 2011 Sandy Bridge, Volume 1, Part 2, Section 4.5.2.1 W-Major Tile
  59  *     Format.
  60  *   - PRM, 2011 Sandy Bridge, Volume 1, Part 2, Section 4.5.3 Tiling Algorithm
  61  *
  62  * Even though the returned offset is always positive, the return type is
  63  * signed due to
  64  *    commit e8b1c6d6f55f5be3bef25084fdd8b6127517e137
  65  *    mesa: Fix return type of  _mesa_get_format_bytes() (#37351)
  66  */
  67 intptr_t
  68 intel_offset_S8(uint32_t stride, uint32_t x, uint32_t y)
  69 {
  70    uint32_t tile_size = 4096;
  71    uint32_t tile_width = 64;
  72    uint32_t tile_height = 64;
  73    uint32_t row_size = 64 * stride;
  74
  75    uint32_t tile_x = x / tile_width;
  76    uint32_t tile_y = y / tile_height;
  77
  78    /* The byte's address relative to the tile's base addres. */
  79    uint32_t byte_x = x % tile_width;
  80    uint32_t byte_y = y % tile_height;
  81
  82    uintptr_t u = tile_y * row_size
  83                + tile_x * tile_size
  84                + 512 * (byte_x / 8)
  85                +  64 * (byte_y / 8)
  86                +  32 * ((byte_y / 4) % 2)
  87                +  16 * ((byte_x / 4) % 2)
  88                +   8 * ((byte_y / 2) % 2)
  89                +   4 * ((byte_x / 2) % 2)
  90                +   2 * (byte_y % 2)
  91                +   1 * (byte_x % 2);
  92
  93    /*
  94     * Errata for Gen5:
  95     *
  96     * An additional offset is needed which is not documented in the PRM.
  97     *
  98     * if ((byte_x / 8) % 2 == 1) {
  99     *    if ((byte_y / 8) % 2) == 0) {
 100     *       u += 64;
 101     *    } else {
 102     *       u -= 64;
 103     *    }
 104     * }
 105     *
 106     * The offset is expressed more tersely as
 107     * u += ((int) x & 0x8) * (8 - (((int) y & 0x8) << 1));
 108     */
 109
 110    return u;
 111 }
 112
 113
 114 /**
 115  * Resolve all buffers that will be mapped by intelSpanRenderStart().
 116  *
 117  * Resolve the depth buffer of each enabled texture and of the read and draw
 118  * buffers.
 119  *
 120  * (Note: In the future this will also perform MSAA resolves.)
 121  */
 122 static void
 123 intel_span_resolve_buffers(struct intel_context *intel)
 124 {
 125    struct gl_context *ctx = &intel->ctx;
 126    struct intel_renderbuffer *draw_irb;
 127    struct intel_renderbuffer *read_irb;
 128    struct intel_texture_object *tex_obj;
 129
 130    /* Resolve depth buffer of each enabled texture. */
 131    for (int i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
 132       if (!ctx->Texture.Unit[i]._ReallyEnabled)
 133          continue;
 134       tex_obj = intel_texture_object(ctx->Texture.Unit[i]._Current);
 135       intel_finalize_mipmap_tree(intel, i);
 136       if (!tex_obj || !tex_obj->mt)
 137          continue;
 138       intel_miptree_all_slices_resolve_depth(intel, tex_obj->mt);
 139    }
 140
 141    /* Resolve each attached depth buffer. */
 142    draw_irb = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_DEPTH);
 143    read_irb = intel_get_renderbuffer(ctx->ReadBuffer, BUFFER_DEPTH);
 144    if (draw_irb)
 145       intel_renderbuffer_resolve_depth(intel, draw_irb);
 146    if (read_irb != draw_irb && read_irb)
 147       intel_renderbuffer_resolve_depth(intel, read_irb);
 148 }
 149
 150 /**
 151  * Map the regions needed by intelSpanRenderStart().
 152  */
 153 static void
 154 intel_span_map_buffers(struct intel_context *intel)
 155 {
 156    struct gl_context *ctx = &intel->ctx;
 157    struct intel_texture_object *tex_obj;
 158
 159    for (int i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
 160       if (!ctx->Texture.Unit[i]._ReallyEnabled)
 161          continue;
 162       tex_obj = intel_texture_object(ctx->Texture.Unit[i]._Current);
 163       intel_finalize_mipmap_tree(intel, i);
 164       intel_tex_map_images(intel, tex_obj,
 165                            GL_MAP_READ_BIT | GL_MAP_WRITE_BIT);
 166    }
 167
 168    _swrast_map_renderbuffers(ctx);
 169 }
 170
 171 /**
 172  * Prepare for software rendering.  Map current read/draw framebuffers'
 173  * renderbuffes and all currently bound texture objects.
 174  *
 175  * Old note: Moved locking out to get reasonable span performance.
 176  */
 177 void
 178 intelSpanRenderStart(struct gl_context * ctx)
 179 {
 180    struct intel_context *intel = intel_context(ctx);
 181
 182    intel_flush(ctx);
 183    intel_prepare_render(intel);
 184    intel_span_resolve_buffers(intel);
 185    intel_flush(ctx);
 186    intel_span_map_buffers(intel);
 187 }
 188
 189 /**
 190  * Called when done software rendering.  Unmap the buffers we mapped in
 191  * the above function.
 192  */
 193 void
 194 intelSpanRenderFinish(struct gl_context * ctx)
 195 {
 196    struct intel_context *intel = intel_context(ctx);
 197    GLuint i;
 198
 199    _swrast_flush(ctx);
 200
 201    for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
 202       if (ctx->Texture.Unit[i]._ReallyEnabled) {
 203          struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
 204          intel_tex_unmap_images(intel, intel_texture_object(texObj));
 205       }
 206    }
 207
 208    _swrast_unmap_renderbuffers(ctx);
 209 }
 210
 211
 212 void
 213 intelInitSpanFuncs(struct gl_context * ctx)
 214 {
 215    struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
 216    swdd->SpanRenderStart = intelSpanRenderStart;
 217    swdd->SpanRenderFinish = intelSpanRenderFinish;
 218 }
 219
 220 void
 221 intel_map_vertex_shader_textures(struct gl_context *ctx)
 222 {
 223    struct intel_context *intel = intel_context(ctx);
 224    int i;
 225
 226    if (ctx->VertexProgram._Current == NULL)
 227       return;
 228
 229    for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
 230       if (ctx->Texture.Unit[i]._ReallyEnabled &&
 231           ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) {
 232          struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
 233
 234          intel_tex_map_images(intel, intel_texture_object(texObj),
 235                               GL_MAP_READ_BIT | GL_MAP_WRITE_BIT);
 236       }
 237    }
 238 }
 239
 240 void
 241 intel_unmap_vertex_shader_textures(struct gl_context *ctx)
 242 {
 243    struct intel_context *intel = intel_context(ctx);
 244    int i;
 245
 246    if (ctx->VertexProgram._Current == NULL)
 247       return;
 248
 249    for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
 250       if (ctx->Texture.Unit[i]._ReallyEnabled &&
 251           ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) {
 252          struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
 253
 254          intel_tex_unmap_images(intel, intel_texture_object(texObj));
 255       }
 256    }
 257 }