src/gallium/drivers/svga/svga_state_vs.c

   1 /**********************************************************
   2  * Copyright 2008-2009 VMware, Inc.  All rights reserved.
   3  *
   4  * Permission is hereby granted, free of charge, to any person
   5  * obtaining a copy of this software and associated documentation
   6  * files (the "Software"), to deal in the Software without
   7  * restriction, including without limitation the rights to use, copy,
   8  * modify, merge, publish, distribute, sublicense, and/or sell copies
   9  * of the Software, and to permit persons to whom the Software is
  10  * furnished to do so, subject to the following conditions:
  11  *
  12  * The above copyright notice and this permission notice shall be
  13  * included in all copies or substantial portions of the Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  16  * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  17  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  18  * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
  19  * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
  20  * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
  21  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  22  * SOFTWARE.
  23  *
  24  **********************************************************/
  25
  26 #include "pipe/p_inlines.h"
  27 #include "pipe/p_defines.h"
  28 #include "util/u_math.h"
  29 #include "util/u_bitmask.h"
  30 #include "translate/translate.h"
  31
  32 #include "svga_context.h"
  33 #include "svga_state.h"
  34 #include "svga_cmd.h"
  35 #include "svga_tgsi.h"
  36
  37 #include "svga_hw_reg.h"
  38
  39 /***********************************************************************
  40  */
  41
  42
  43 static INLINE int compare_vs_keys( const struct svga_vs_compile_key *a,
  44                                    const struct svga_vs_compile_key *b )
  45 {
  46    unsigned keysize = svga_vs_key_size( a );
  47    return memcmp( a, b, keysize );
  48 }
  49
  50
  51 static struct svga_shader_result *search_vs_key( struct svga_vertex_shader *vs,
  52                                                  const struct svga_vs_compile_key *key )
  53 {
  54    struct svga_shader_result *result = vs->base.results;
  55
  56    assert(key);
  57
  58    for ( ; result; result = result->next) {
  59       if (compare_vs_keys( key, &result->key.vkey ) == 0)
  60          return result;
  61    }
  62
  63    return NULL;
  64 }
  65
  66
  67 static enum pipe_error compile_vs( struct svga_context *svga,
  68                                    struct svga_vertex_shader *vs,
  69                                    const struct svga_vs_compile_key *key,
  70                                    struct svga_shader_result **out_result )
  71 {
  72    struct svga_shader_result *result;
  73    enum pipe_error ret = PIPE_ERROR;
  74
  75    result = svga_translate_vertex_program( vs, key );
  76    if (result == NULL) {
  77       ret = PIPE_ERROR_OUT_OF_MEMORY;
  78       goto fail;
  79    }
  80
  81    result->id = util_bitmask_add(svga->vs_bm);
  82    if(result->id == UTIL_BITMASK_INVALID_INDEX) {
  83       ret = PIPE_ERROR_OUT_OF_MEMORY;
  84       goto fail;
  85    }
  86
  87    ret = SVGA3D_DefineShader(svga->swc,
  88                              result->id,
  89                              SVGA3D_SHADERTYPE_VS,
  90                              result->tokens,
  91                              result->nr_tokens * sizeof result->tokens[0]);
  92    if (ret)
  93       goto fail;
  94
  95    *out_result = result;
  96    result->next = vs->base.results;
  97    vs->base.results = result;
  98    return PIPE_OK;
  99
 100 fail:
 101    if (result) {
 102       if (result->id != UTIL_BITMASK_INVALID_INDEX)
 103          util_bitmask_clear( svga->vs_bm, result->id );
 104       svga_destroy_shader_result( result );
 105    }
 106    return ret;
 107 }
 108
 109 /* SVGA_NEW_PRESCALE, SVGA_NEW_RAST, SVGA_NEW_ZERO_STRIDE
 110  */
 111 static int make_vs_key( struct svga_context *svga,
 112                         struct svga_vs_compile_key *key )
 113 {
 114    memset(key, 0, sizeof *key);
 115    key->need_prescale = svga->state.hw_clear.prescale.enabled;
 116    key->allow_psiz = svga->curr.rast->templ.point_size_per_vertex;
 117    key->zero_stride_vertex_elements =
 118       svga->curr.zero_stride_vertex_elements;
 119    key->num_zero_stride_vertex_elements =
 120       svga->curr.num_zero_stride_vertex_elements;
 121    return 0;
 122 }
 123
 124
 125
 126 static int emit_hw_vs( struct svga_context *svga,
 127                        unsigned dirty )
 128 {
 129    struct svga_shader_result *result = NULL;
 130    unsigned id = SVGA3D_INVALID_ID;
 131    int ret = 0;
 132
 133    /* SVGA_NEW_NEED_SWTNL */
 134    if (!svga->state.sw.need_swtnl) {
 135       struct svga_vertex_shader *vs = svga->curr.vs;
 136       struct svga_vs_compile_key key;
 137
 138       ret = make_vs_key( svga, &key );
 139       if (ret)
 140          return ret;
 141
 142       result = search_vs_key( vs, &key );
 143       if (!result) {
 144          ret = compile_vs( svga, vs, &key, &result );
 145          if (ret)
 146             return ret;
 147       }
 148
 149       assert (result);
 150       id = result->id;
 151    }
 152
 153    if (result != svga->state.hw_draw.vs) {
 154       ret = SVGA3D_SetShader(svga->swc,
 155                              SVGA3D_SHADERTYPE_VS,
 156                              id );
 157       if (ret)
 158          return ret;
 159
 160       svga->dirty |= SVGA_NEW_VS_RESULT;
 161       svga->state.hw_draw.vs = result;
 162    }
 163
 164    return 0;
 165 }
 166
 167 struct svga_tracked_state svga_hw_vs =
 168 {
 169    "vertex shader (hwtnl)",
 170    (SVGA_NEW_VS |
 171     SVGA_NEW_PRESCALE |
 172     SVGA_NEW_NEED_SWTNL |
 173     SVGA_NEW_ZERO_STRIDE),
 174    emit_hw_vs
 175 };
 176
 177
 178 /***********************************************************************
 179  */
 180 static int update_zero_stride( struct svga_context *svga,
 181                                unsigned dirty )
 182 {
 183    unsigned i;
 184
 185    svga->curr.zero_stride_vertex_elements = 0;
 186    svga->curr.num_zero_stride_vertex_elements = 0;
 187
 188    for (i = 0; i < svga->curr.num_vertex_elements; i++) {
 189       const struct pipe_vertex_element *vel = &svga->curr.ve[i];
 190       const struct pipe_vertex_buffer *vbuffer = &svga->curr.vb[
 191          vel->vertex_buffer_index];
 192       if (vbuffer->stride == 0) {
 193          unsigned const_idx =
 194             svga->curr.num_zero_stride_vertex_elements;
 195          struct translate *translate;
 196          struct translate_key key;
 197          void *mapped_buffer;
 198
 199          svga->curr.zero_stride_vertex_elements |= (1 << i);
 200          ++svga->curr.num_zero_stride_vertex_elements;
 201
 202          key.output_stride = 4 * sizeof(float);
 203          key.nr_elements = 1;
 204          key.element[0].input_format = vel->src_format;
 205          key.element[0].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
 206          key.element[0].input_buffer = vel->vertex_buffer_index;
 207          key.element[0].input_offset = vel->src_offset;
 208          key.element[0].output_offset = const_idx * 4 * sizeof(float);
 209
 210          translate_key_sanitize(&key);
 211          /* translate_generic_create is technically private but
 212           * we don't want to code-generate, just want generic
 213           * translation */
 214          translate = translate_generic_create(&key);
 215
 216          assert(vel->src_offset == 0);
 217
 218          mapped_buffer = pipe_buffer_map_range(svga->pipe.screen,
 219                                                vbuffer->buffer,
 220                                                vel->src_offset,
 221                                                pf_get_size(vel->src_format),
 222                                                PIPE_BUFFER_USAGE_CPU_READ);
 223          translate->set_buffer(translate, vel->vertex_buffer_index,
 224                                mapped_buffer,
 225                                vbuffer->stride);
 226          translate->run(translate, 0, 1,
 227                         svga->curr.zero_stride_constants);
 228
 229          pipe_buffer_unmap(svga->pipe.screen,
 230                            vbuffer->buffer);
 231          translate->release(translate);
 232       }
 233    }
 234
 235    if (svga->curr.num_zero_stride_vertex_elements)
 236       svga->dirty |= SVGA_NEW_ZERO_STRIDE;
 237
 238    return 0;
 239 }
 240
 241 struct svga_tracked_state svga_hw_update_zero_stride =
 242 {
 243    "update zero_stride",
 244    ( SVGA_NEW_VELEMENT |
 245      SVGA_NEW_VBUFFER ),
 246    update_zero_stride
 247 };