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 #include "main/transformfeedback.h"
  35
  36 static void
  37 gen6_update_sol_surfaces(struct brw_context *brw)
  38 {
  39    struct gl_context *ctx = &brw->intel.ctx;
  40    /* BRW_NEW_TRANSFORM_FEEDBACK */
  41    struct gl_transform_feedback_object *xfb_obj =
  42       ctx->TransformFeedback.CurrentObject;
  43    /* BRW_NEW_VERTEX_PROGRAM */
  44    const struct gl_shader_program *shaderprog =
  45       ctx->Shader.CurrentVertexProgram;
  46    const struct gl_transform_feedback_info *linked_xfb_info =
  47       &shaderprog->LinkedTransformFeedback;
  48    int i;
  49
  50    for (i = 0; i < BRW_MAX_SOL_BINDINGS; ++i) {
  51       const int surf_index = SURF_INDEX_SOL_BINDING(i);
  52       if (_mesa_is_xfb_active_and_unpaused(ctx) &&
  53           i < linked_xfb_info->NumOutputs) {
  54          unsigned buffer = linked_xfb_info->Outputs[i].OutputBuffer;
  55          unsigned buffer_offset =
  56             xfb_obj->Offset[buffer] / 4 +
  57             linked_xfb_info->Outputs[i].DstOffset;
  58          brw_update_sol_surface(
  59             brw, xfb_obj->Buffers[buffer], &brw->gs.surf_offset[surf_index],
  60             linked_xfb_info->Outputs[i].NumComponents,
  61             linked_xfb_info->BufferStride[buffer], buffer_offset);
  62       } else {
  63          brw->gs.surf_offset[surf_index] = 0;
  64       }
  65    }
  66
  67    brw->state.dirty.brw |= BRW_NEW_SURFACES;
  68 }
  69
  70 const struct brw_tracked_state gen6_sol_surface = {
  71    .dirty = {
  72       .mesa = 0,
  73       .brw = (BRW_NEW_BATCH |
  74               BRW_NEW_VERTEX_PROGRAM |
  75               BRW_NEW_TRANSFORM_FEEDBACK),
  76       .cache = 0
  77    },
  78    .emit = gen6_update_sol_surfaces,
  79 };
  80
  81 /**
  82  * Constructs the binding table for the WM surface state, which maps unit
  83  * numbers to surface state objects.
  84  */
  85 static void
  86 brw_gs_upload_binding_table(struct brw_context *brw)
  87 {
  88    struct gl_context *ctx = &brw->intel.ctx;
  89    /* BRW_NEW_VERTEX_PROGRAM */
  90    const struct gl_shader_program *shaderprog =
  91       ctx->Shader.CurrentVertexProgram;
  92    bool has_surfaces = false;
  93    uint32_t *bind;
  94
  95    if (shaderprog) {
  96       const struct gl_transform_feedback_info *linked_xfb_info =
  97          &shaderprog->LinkedTransformFeedback;
  98       /* Currently we only ever upload surfaces for SOL. */
  99       has_surfaces = linked_xfb_info->NumOutputs != 0;
 100    }
 101
 102    /* Skip making a binding table if we don't have anything to put in it. */
 103    if (!has_surfaces) {
 104       if (brw->gs.bind_bo_offset != 0) {
 105          brw->state.dirty.brw |= BRW_NEW_GS_BINDING_TABLE;
 106          brw->gs.bind_bo_offset = 0;
 107       }
 108       return;
 109    }
 110
 111    /* Might want to calculate nr_surfaces first, to avoid taking up so much
 112     * space for the binding table.
 113     */
 114    bind = brw_state_batch(brw, AUB_TRACE_BINDING_TABLE,
 115                           sizeof(uint32_t) * BRW_MAX_GS_SURFACES,
 116                           32, &brw->gs.bind_bo_offset);
 117
 118    /* BRW_NEW_SURFACES */
 119    memcpy(bind, brw->gs.surf_offset, BRW_MAX_GS_SURFACES * sizeof(uint32_t));
 120
 121    brw->state.dirty.brw |= BRW_NEW_GS_BINDING_TABLE;
 122 }
 123
 124 const struct brw_tracked_state gen6_gs_binding_table = {
 125    .dirty = {
 126       .mesa = 0,
 127       .brw = (BRW_NEW_BATCH |
 128               BRW_NEW_VERTEX_PROGRAM |
 129               BRW_NEW_SURFACES),
 130       .cache = 0
 131    },
 132    .emit = brw_gs_upload_binding_table,
 133 };
 134
 135 static void
 136 gen6_update_sol_indices(struct brw_context *brw)
 137 {
 138    struct intel_context *intel = &brw->intel;
 139
 140    BEGIN_BATCH(4);
 141    OUT_BATCH(_3DSTATE_GS_SVB_INDEX << 16 | (4 - 2));
 142    OUT_BATCH(0);
 143    OUT_BATCH(brw->sol.svbi_0_starting_index); /* BRW_NEW_SOL_INDICES */
 144    OUT_BATCH(brw->sol.svbi_0_max_index); /* BRW_NEW_SOL_INDICES */
 145    ADVANCE_BATCH();
 146 }
 147
 148 const struct brw_tracked_state gen6_sol_indices = {
 149    .dirty = {
 150       .mesa = 0,
 151       .brw = (BRW_NEW_CONTEXT |
 152               BRW_NEW_SOL_INDICES),
 153       .cache = 0
 154    },
 155    .emit = gen6_update_sol_indices,
 156 };
 157
 158 void
 159 brw_begin_transform_feedback(struct gl_context *ctx, GLenum mode,
 160                              struct gl_transform_feedback_object *obj)
 161 {
 162    struct brw_context *brw = brw_context(ctx);
 163    struct intel_context *intel = &brw->intel;
 164    const struct gl_shader_program *vs_prog =
 165       ctx->Shader.CurrentVertexProgram;
 166    const struct gl_transform_feedback_info *linked_xfb_info =
 167       &vs_prog->LinkedTransformFeedback;
 168    struct gl_transform_feedback_object *xfb_obj =
 169       ctx->TransformFeedback.CurrentObject;
 170
 171    /* Compute the maximum number of vertices that we can write without
 172     * overflowing any of the buffers currently being used for feedback.
 173     */
 174    unsigned max_index
 175       = _mesa_compute_max_transform_feedback_vertices(xfb_obj,
 176                                                       linked_xfb_info);
 177
 178    /* Initialize the SVBI 0 register to zero and set the maximum index.
 179     * These values will be sent to the hardware on the next draw.
 180     */
 181    brw->state.dirty.brw |= BRW_NEW_SOL_INDICES;
 182    brw->sol.svbi_0_starting_index = 0;
 183    brw->sol.svbi_0_max_index = max_index;
 184    brw->sol.offset_0_batch_start = 0;
 185
 186    if (intel->gen >= 7) {
 187       /* Reset the SOL buffer offset register. */
 188       for (int i = 0; i < 4; i++) {
 189          BEGIN_BATCH(3);
 190          OUT_BATCH(MI_LOAD_REGISTER_IMM | (3 - 2));
 191          OUT_BATCH(GEN7_SO_WRITE_OFFSET(i));
 192          OUT_BATCH(0);
 193          ADVANCE_BATCH();
 194       }
 195    }
 196 }
 197
 198 void
 199 brw_end_transform_feedback(struct gl_context *ctx,
 200                            struct gl_transform_feedback_object *obj)
 201 {
 202    /* After EndTransformFeedback, it's likely that the client program will try
 203     * to draw using the contents of the transform feedback buffer as vertex
 204     * input.  In order for this to work, we need to flush the data through at
 205     * least the GS stage of the pipeline, and flush out the render cache.  For
 206     * simplicity, just do a full flush.
 207     */
 208    struct brw_context *brw = brw_context(ctx);
 209    struct intel_context *intel = &brw->intel;
 210    intel_batchbuffer_emit_mi_flush(intel);
 211 }