src/mesa/drivers/dri/i965/gen6_sol.c

   1 /*
   2  * Copyright © 2011 Intel Corporation
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a
   5  * copy of this software and associated documentation files (the "Software"),
   6  * to deal in the Software without restriction, including without limitation
   7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   8  * and/or sell copies of the Software, and to permit persons to whom the
   9  * Software is furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice (including the next
  12  * paragraph) shall be included in all copies or substantial portions of the
  13  * Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  21  * IN THE SOFTWARE.
  22  */
  23
  24 /** \file gen6_sol.c
  25  *
  26  * Code to initialize the binding table entries used by transform feedback.
  27  */
  28
  29 #include "main/macros.h"
  30 #include "brw_context.h"
  31 #include "intel_batchbuffer.h"
  32 #include "brw_defines.h"
  33 #include "brw_state.h"
  34
  35 static void
  36 gen6_update_sol_surfaces(struct brw_context *brw)
  37 {
  38    struct gl_context *ctx = &brw->intel.ctx;
  39    /* _NEW_TRANSFORM_FEEDBACK */
  40    struct gl_transform_feedback_object *xfb_obj =
  41       ctx->TransformFeedback.CurrentObject;
  42    /* BRW_NEW_VERTEX_PROGRAM */
  43    const struct gl_shader_program *shaderprog =
  44       ctx->Shader.CurrentVertexProgram;
  45    const struct gl_transform_feedback_info *linked_xfb_info =
  46       &shaderprog->LinkedTransformFeedback;
  47    int i;
  48
  49    for (i = 0; i < BRW_MAX_SOL_BINDINGS; ++i) {
  50       const int surf_index = SURF_INDEX_SOL_BINDING(i);
  51       if (xfb_obj->Active && !xfb_obj->Paused &&
  52           i < linked_xfb_info->NumOutputs) {
  53          unsigned buffer = linked_xfb_info->Outputs[i].OutputBuffer;
  54          unsigned buffer_offset =
  55             xfb_obj->Offset[buffer] / 4 +
  56             linked_xfb_info->Outputs[i].DstOffset;
  57          brw_update_sol_surface(
  58             brw, xfb_obj->Buffers[buffer], &brw->gs.surf_offset[surf_index],
  59             linked_xfb_info->Outputs[i].NumComponents,
  60             linked_xfb_info->BufferStride[buffer], buffer_offset);
  61       } else {
  62          brw->gs.surf_offset[surf_index] = 0;
  63       }
  64    }
  65
  66    brw->state.dirty.brw |= BRW_NEW_SURFACES;
  67 }
  68
  69 const struct brw_tracked_state gen6_sol_surface = {
  70    .dirty = {
  71       .mesa = _NEW_TRANSFORM_FEEDBACK,
  72       .brw = (BRW_NEW_BATCH |
  73               BRW_NEW_VERTEX_PROGRAM),
  74       .cache = 0
  75    },
  76    .emit = gen6_update_sol_surfaces,
  77 };
  78
  79 /**
  80  * Constructs the binding table for the WM surface state, which maps unit
  81  * numbers to surface state objects.
  82  */
  83 static void
  84 brw_gs_upload_binding_table(struct brw_context *brw)
  85 {
  86    struct gl_context *ctx = &brw->intel.ctx;
  87    /* BRW_NEW_VERTEX_PROGRAM */
  88    const struct gl_shader_program *shaderprog =
  89       ctx->Shader.CurrentVertexProgram;
  90    const struct gl_transform_feedback_info *linked_xfb_info =
  91       &shaderprog->LinkedTransformFeedback;
  92    /* Currently we only ever upload surfaces for SOL. */
  93    bool has_surfaces = linked_xfb_info->NumOutputs != 0;
  94
  95    uint32_t *bind;
  96
  97    /* Skip making a binding table if we don't have anything to put in it. */
  98    if (!has_surfaces) {
  99       if (brw->gs.bind_bo_offset != 0) {
 100          brw->state.dirty.brw |= BRW_NEW_GS_BINDING_TABLE;
 101          brw->gs.bind_bo_offset = 0;
 102       }
 103       return;
 104    }
 105
 106    /* Might want to calculate nr_surfaces first, to avoid taking up so much
 107     * space for the binding table.
 108     */
 109    bind = brw_state_batch(brw, AUB_TRACE_BINDING_TABLE,
 110                           sizeof(uint32_t) * BRW_MAX_SURFACES,
 111                           32, &brw->gs.bind_bo_offset);
 112
 113    /* BRW_NEW_SURFACES */
 114    memcpy(bind, brw->gs.surf_offset, BRW_MAX_GS_SURFACES * sizeof(uint32_t));
 115
 116    brw->state.dirty.brw |= BRW_NEW_GS_BINDING_TABLE;
 117 }
 118
 119 const struct brw_tracked_state gen6_gs_binding_table = {
 120    .dirty = {
 121       .mesa = 0,
 122       .brw = (BRW_NEW_BATCH |
 123               BRW_NEW_VERTEX_PROGRAM |
 124               BRW_NEW_SURFACES),
 125       .cache = 0
 126    },
 127    .emit = brw_gs_upload_binding_table,
 128 };
 129
 130 static void
 131 gen6_update_sol_indices(struct brw_context *brw)
 132 {
 133    struct intel_context *intel = &brw->intel;
 134
 135    BEGIN_BATCH(4);
 136    OUT_BATCH(_3DSTATE_GS_SVB_INDEX << 16 | (4 - 2));
 137    OUT_BATCH(0);
 138    OUT_BATCH(brw->sol.svbi_0_starting_index); /* BRW_NEW_SOL_INDICES */
 139    OUT_BATCH(brw->sol.svbi_0_max_index); /* BRW_NEW_SOL_INDICES */
 140    ADVANCE_BATCH();
 141 }
 142
 143 const struct brw_tracked_state gen6_sol_indices = {
 144    .dirty = {
 145       .mesa = 0,
 146       .brw = (BRW_NEW_BATCH |
 147               BRW_NEW_SOL_INDICES),
 148       .cache = 0
 149    },
 150    .emit = gen6_update_sol_indices,
 151 };
 152
 153 void
 154 brw_begin_transform_feedback(struct gl_context *ctx, GLenum mode,
 155                              struct gl_transform_feedback_object *obj)
 156 {
 157    struct brw_context *brw = brw_context(ctx);
 158    const struct gl_shader_program *vs_prog =
 159       ctx->Shader.CurrentVertexProgram;
 160    const struct gl_transform_feedback_info *linked_xfb_info =
 161       &vs_prog->LinkedTransformFeedback;
 162    struct gl_transform_feedback_object *xfb_obj =
 163       ctx->TransformFeedback.CurrentObject;
 164
 165    unsigned max_index = 0xffffffff;
 166
 167    /* Compute the maximum number of vertices that we can write without
 168     * overflowing any of the buffers currently being used for feedback.
 169     */
 170    for (int i = 0; i < BRW_MAX_SOL_BUFFERS; ++i) {
 171       unsigned stride = linked_xfb_info->BufferStride[i];
 172
 173       /* Skip any inactive buffers, which have a stride of 0. */
 174       if (stride == 0)
 175          continue;
 176
 177       unsigned max_for_this_buffer = xfb_obj->Size[i] / (4 * stride);
 178       max_index = MIN2(max_index, max_for_this_buffer);
 179    }
 180
 181    /* Initialize the SVBI 0 register to zero and set the maximum index.
 182     * These values will be sent to the hardware on the next draw.
 183     */
 184    brw->state.dirty.brw |= BRW_NEW_SOL_INDICES;
 185    brw->sol.svbi_0_starting_index = 0;
 186    brw->sol.svbi_0_max_index = max_index;
 187    brw->sol.offset_0_batch_start = 0;
 188 }
 189
 190 void
 191 brw_end_transform_feedback(struct gl_context *ctx,
 192                            struct gl_transform_feedback_object *obj)
 193 {
 194    /* After EndTransformFeedback, it's likely that the client program will try
 195     * to draw using the contents of the transform feedback buffer as vertex
 196     * input.  In order for this to work, we need to flush the data through at
 197     * least the GS stage of the pipeline, and flush out the render cache.  For
 198     * simplicity, just do a full flush.
 199     */
 200    struct brw_context *brw = brw_context(ctx);
 201    struct intel_context *intel = &brw->intel;
 202    intel_batchbuffer_emit_mi_flush(intel);
 203 }