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_format.h"
  29 #include "util/u_math.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_OK;
  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    ret = SVGA3D_DefineShader(svga->swc,
  82                              svga->state.next_vs_id,
  83                              SVGA3D_SHADERTYPE_VS,
  84                              result->tokens,
  85                              result->nr_tokens * sizeof result->tokens[0]);
  86    if (ret)
  87       goto fail;
  88
  89    *out_result = result;
  90    result->id = svga->state.next_vs_id++;
  91    result->next = vs->base.results;
  92    vs->base.results = result;
  93    return PIPE_OK;
  94
  95 fail:
  96    if (result)
  97       svga_destroy_shader_result( result );
  98    return ret;
  99 }
 100
 101 /* SVGA_NEW_PRESCALE, SVGA_NEW_RAST, SVGA_NEW_ZERO_STRIDE
 102  */
 103 static int make_vs_key( struct svga_context *svga,
 104                         struct svga_vs_compile_key *key )
 105 {
 106    memset(key, 0, sizeof *key);
 107    key->need_prescale = svga->state.hw_clear.prescale.enabled;
 108    key->allow_psiz = svga->curr.rast->templ.point_size_per_vertex;
 109    key->zero_stride_vertex_elements =
 110       svga->curr.zero_stride_vertex_elements;
 111    key->num_zero_stride_vertex_elements =
 112       svga->curr.num_zero_stride_vertex_elements;
 113    return 0;
 114 }
 115
 116
 117
 118 static int emit_hw_vs( struct svga_context *svga,
 119                        unsigned dirty )
 120 {
 121    struct svga_shader_result *result = NULL;
 122    unsigned id = SVGA3D_INVALID_ID;
 123    int ret = 0;
 124
 125    /* SVGA_NEW_NEED_SWTNL */
 126    if (!svga->state.sw.need_swtnl) {
 127       struct svga_vertex_shader *vs = svga->curr.vs;
 128       struct svga_vs_compile_key key;
 129
 130       ret = make_vs_key( svga, &key );
 131       if (ret)
 132          return ret;
 133
 134       result = search_vs_key( vs, &key );
 135       if (!result) {
 136          ret = compile_vs( svga, vs, &key, &result );
 137          if (ret)
 138             return ret;
 139       }
 140
 141       assert (result);
 142       id = result->id;
 143    }
 144
 145    if (id != svga->state.hw_draw.shader_id[PIPE_SHADER_VERTEX]) {
 146       ret = SVGA3D_SetShader(svga->swc,
 147                              SVGA3D_SHADERTYPE_VS,
 148                              id );
 149       if (ret)
 150          return ret;
 151
 152       svga->dirty |= SVGA_NEW_VS_RESULT;
 153       svga->state.hw_draw.shader_id[PIPE_SHADER_VERTEX] = id;
 154       svga->state.hw_draw.vs = result;
 155    }
 156
 157    return 0;
 158 }
 159
 160 struct svga_tracked_state svga_hw_vs =
 161 {
 162    "vertex shader (hwtnl)",
 163    (SVGA_NEW_VS |
 164     SVGA_NEW_PRESCALE |
 165     SVGA_NEW_NEED_SWTNL |
 166     SVGA_NEW_ZERO_STRIDE),
 167    emit_hw_vs
 168 };
 169
 170
 171 /***********************************************************************
 172  */
 173 static int update_zero_stride( struct svga_context *svga,
 174                                unsigned dirty )
 175 {
 176    unsigned i;
 177
 178    svga->curr.zero_stride_vertex_elements = 0;
 179    svga->curr.num_zero_stride_vertex_elements = 0;
 180
 181    for (i = 0; i < svga->curr.num_vertex_elements; i++) {
 182       const struct pipe_vertex_element *vel = &svga->curr.ve[i];
 183       const struct pipe_vertex_buffer *vbuffer = &svga->curr.vb[
 184          vel->vertex_buffer_index];
 185       if (vbuffer->stride == 0) {
 186          unsigned const_idx =
 187             svga->curr.num_zero_stride_vertex_elements;
 188          struct translate *translate;
 189          struct translate_key key;
 190          void *mapped_buffer;
 191
 192          svga->curr.zero_stride_vertex_elements |= (1 << i);
 193          ++svga->curr.num_zero_stride_vertex_elements;
 194
 195          key.output_stride = 4 * sizeof(float);
 196          key.nr_elements = 1;
 197          key.element[0].input_format = vel->src_format;
 198          key.element[0].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
 199          key.element[0].input_buffer = vel->vertex_buffer_index;
 200          key.element[0].input_offset = vel->src_offset;
 201          key.element[0].output_offset = const_idx * 4 * sizeof(float);
 202
 203          translate_key_sanitize(&key);
 204          /* translate_generic_create is technically private but
 205           * we don't want to code-generate, just want generic
 206           * translation */
 207          translate = translate_generic_create(&key);
 208
 209          assert(vel->src_offset == 0);
 210
 211          mapped_buffer = pipe_buffer_map_range(svga->pipe.screen,
 212                                                vbuffer->buffer,
 213                                                vel->src_offset,
 214                                                util_format_get_blocksize(vel->src_format),
 215                                                PIPE_BUFFER_USAGE_CPU_READ);
 216          translate->set_buffer(translate, vel->vertex_buffer_index,
 217                                mapped_buffer,
 218                                vbuffer->stride);
 219          translate->run(translate, 0, 1,
 220                         svga->curr.zero_stride_constants);
 221
 222          pipe_buffer_unmap(svga->pipe.screen,
 223                            vbuffer->buffer);
 224          translate->release(translate);
 225       }
 226    }
 227
 228    if (svga->curr.num_zero_stride_vertex_elements)
 229       svga->dirty |= SVGA_NEW_ZERO_STRIDE;
 230
 231    return 0;
 232 }
 233
 234 struct svga_tracked_state svga_hw_update_zero_stride =
 235 {
 236    "update zero_stride",
 237    ( SVGA_NEW_VELEMENT |
 238      SVGA_NEW_VBUFFER ),
 239    update_zero_stride
 240 };