Home | History | Annotate | Download | only in i965
      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    bool has_surfaces = false;
     91    uint32_t *bind;
     92 
     93    if (shaderprog) {
     94       const struct gl_transform_feedback_info *linked_xfb_info =
     95 	 &shaderprog->LinkedTransformFeedback;
     96       /* Currently we only ever upload surfaces for SOL. */
     97       has_surfaces = linked_xfb_info->NumOutputs != 0;
     98    }
     99 
    100    /* Skip making a binding table if we don't have anything to put in it. */
    101    if (!has_surfaces) {
    102       if (brw->gs.bind_bo_offset != 0) {
    103 	 brw->state.dirty.brw |= BRW_NEW_GS_BINDING_TABLE;
    104 	 brw->gs.bind_bo_offset = 0;
    105       }
    106       return;
    107    }
    108 
    109    /* Might want to calculate nr_surfaces first, to avoid taking up so much
    110     * space for the binding table.
    111     */
    112    bind = brw_state_batch(brw, AUB_TRACE_BINDING_TABLE,
    113 			  sizeof(uint32_t) * BRW_MAX_GS_SURFACES,
    114 			  32, &brw->gs.bind_bo_offset);
    115 
    116    /* BRW_NEW_SURFACES */
    117    memcpy(bind, brw->gs.surf_offset, BRW_MAX_GS_SURFACES * sizeof(uint32_t));
    118 
    119    brw->state.dirty.brw |= BRW_NEW_GS_BINDING_TABLE;
    120 }
    121 
    122 const struct brw_tracked_state gen6_gs_binding_table = {
    123    .dirty = {
    124       .mesa = 0,
    125       .brw = (BRW_NEW_BATCH |
    126 	      BRW_NEW_VERTEX_PROGRAM |
    127 	      BRW_NEW_SURFACES),
    128       .cache = 0
    129    },
    130    .emit = brw_gs_upload_binding_table,
    131 };
    132 
    133 static void
    134 gen6_update_sol_indices(struct brw_context *brw)
    135 {
    136    struct intel_context *intel = &brw->intel;
    137 
    138    BEGIN_BATCH(4);
    139    OUT_BATCH(_3DSTATE_GS_SVB_INDEX << 16 | (4 - 2));
    140    OUT_BATCH(0);
    141    OUT_BATCH(brw->sol.svbi_0_starting_index); /* BRW_NEW_SOL_INDICES */
    142    OUT_BATCH(brw->sol.svbi_0_max_index); /* BRW_NEW_SOL_INDICES */
    143    ADVANCE_BATCH();
    144 }
    145 
    146 const struct brw_tracked_state gen6_sol_indices = {
    147    .dirty = {
    148       .mesa = 0,
    149       .brw = (BRW_NEW_BATCH |
    150               BRW_NEW_SOL_INDICES),
    151       .cache = 0
    152    },
    153    .emit = gen6_update_sol_indices,
    154 };
    155 
    156 void
    157 brw_begin_transform_feedback(struct gl_context *ctx, GLenum mode,
    158 			     struct gl_transform_feedback_object *obj)
    159 {
    160    struct brw_context *brw = brw_context(ctx);
    161    const struct gl_shader_program *vs_prog =
    162       ctx->Shader.CurrentVertexProgram;
    163    const struct gl_transform_feedback_info *linked_xfb_info =
    164       &vs_prog->LinkedTransformFeedback;
    165    struct gl_transform_feedback_object *xfb_obj =
    166       ctx->TransformFeedback.CurrentObject;
    167 
    168    unsigned max_index = 0xffffffff;
    169 
    170    /* Compute the maximum number of vertices that we can write without
    171     * overflowing any of the buffers currently being used for feedback.
    172     */
    173    for (int i = 0; i < BRW_MAX_SOL_BUFFERS; ++i) {
    174       unsigned stride = linked_xfb_info->BufferStride[i];
    175 
    176       /* Skip any inactive buffers, which have a stride of 0. */
    177       if (stride == 0)
    178 	 continue;
    179 
    180       unsigned max_for_this_buffer = xfb_obj->Size[i] / (4 * stride);
    181       max_index = MIN2(max_index, max_for_this_buffer);
    182    }
    183 
    184    /* Initialize the SVBI 0 register to zero and set the maximum index.
    185     * These values will be sent to the hardware on the next draw.
    186     */
    187    brw->state.dirty.brw |= BRW_NEW_SOL_INDICES;
    188    brw->sol.svbi_0_starting_index = 0;
    189    brw->sol.svbi_0_max_index = max_index;
    190    brw->sol.offset_0_batch_start = 0;
    191 }
    192 
    193 void
    194 brw_end_transform_feedback(struct gl_context *ctx,
    195                            struct gl_transform_feedback_object *obj)
    196 {
    197    /* After EndTransformFeedback, it's likely that the client program will try
    198     * to draw using the contents of the transform feedback buffer as vertex
    199     * input.  In order for this to work, we need to flush the data through at
    200     * least the GS stage of the pipeline, and flush out the render cache.  For
    201     * simplicity, just do a full flush.
    202     */
    203    struct brw_context *brw = brw_context(ctx);
    204    struct intel_context *intel = &brw->intel;
    205    intel_batchbuffer_emit_mi_flush(intel);
    206 }
    207