path: root/src/mesa/drivers/dri/i965/brw_draw.c

/*
 * Copyright 2003 VMware, Inc.
 * All Rights Reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the
 * "Software"), to deal in the Software without restriction, including
 * without limitation the rights to use, copy, modify, merge, publish,
 * distribute, sublicense, and/or sell copies of the Software, and to
 * permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 *
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
 * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 */

#include <sys/errno.h>

#include "main/arrayobj.h"
#include "main/blend.h"
#include "main/context.h"
#include "main/condrender.h"
#include "main/samplerobj.h"
#include "main/state.h"
#include "main/enums.h"
#include "main/macros.h"
#include "main/transformfeedback.h"
#include "main/framebuffer.h"
#include "main/varray.h"
#include "tnl/tnl.h"
#include "vbo/vbo.h"
#include "swrast/swrast.h"
#include "swrast_setup/swrast_setup.h"
#include "drivers/common/meta.h"
#include "util/bitscan.h"
#include "util/bitset.h"

#include "brw_blorp.h"
#include "brw_draw.h"
#include "brw_defines.h"
#include "compiler/brw_eu_defines.h"
#include "brw_context.h"
#include "brw_state.h"

#include "intel_batchbuffer.h"
#include "intel_buffers.h"
#include "intel_fbo.h"
#include "intel_mipmap_tree.h"
#include "intel_buffer_objects.h"

#define FILE_DEBUG_FLAG DEBUG_PRIMS


static const GLenum reduced_prim[GL_POLYGON+1] = {
   [GL_POINTS] = GL_POINTS,
   [GL_LINES] = GL_LINES,
   [GL_LINE_LOOP] = GL_LINES,
   [GL_LINE_STRIP] = GL_LINES,
   [GL_TRIANGLES] = GL_TRIANGLES,
   [GL_TRIANGLE_STRIP] = GL_TRIANGLES,
   [GL_TRIANGLE_FAN] = GL_TRIANGLES,
   [GL_QUADS] = GL_TRIANGLES,
   [GL_QUAD_STRIP] = GL_TRIANGLES,
   [GL_POLYGON] = GL_TRIANGLES
};

/* When the primitive changes, set a state bit and re-validate.  Not
 * the nicest and would rather deal with this by having all the
 * programs be immune to the active primitive (ie. cope with all
 * possibilities).  That may not be realistic however.
 */
static void
brw_set_prim(struct brw_context *brw, const struct _mesa_prim *prim)
{
   struct gl_context *ctx = &brw->ctx;
   uint32_t hw_prim = get_hw_prim_for_gl_prim(prim->mode);

   DBG("PRIM: %s\n", _mesa_enum_to_string(prim->mode));

   /* Slight optimization to avoid the GS program when not needed:
    */
   if (prim->mode == GL_QUAD_STRIP &&
       ctx->Light.ShadeModel != GL_FLAT &&
       ctx->Polygon.FrontMode == GL_FILL &&
       ctx->Polygon.BackMode == GL_FILL)
      hw_prim = _3DPRIM_TRISTRIP;

   if (prim->mode == GL_QUADS && prim->count == 4 &&
       ctx->Light.ShadeModel != GL_FLAT &&
       ctx->Polygon.FrontMode == GL_FILL &&
       ctx->Polygon.BackMode == GL_FILL) {
      hw_prim = _3DPRIM_TRIFAN;
   }

   if (hw_prim != brw->primitive) {
      brw->primitive = hw_prim;
      brw->ctx.NewDriverState |= BRW_NEW_PRIMITIVE;

      if (reduced_prim[prim->mode] != brw->reduced_primitive) {
         brw->reduced_primitive = reduced_prim[prim->mode];
         brw->ctx.NewDriverState |= BRW_NEW_REDUCED_PRIMITIVE;
      }
   }
}

static void
gen6_set_prim(struct brw_context *brw, const struct _mesa_prim *prim)
{
   const struct gl_context *ctx = &brw->ctx;
   uint32_t hw_prim;

   DBG("PRIM: %s\n", _mesa_enum_to_string(prim->mode));

   if (prim->mode == GL_PATCHES) {
      hw_prim = _3DPRIM_PATCHLIST(ctx->TessCtrlProgram.patch_vertices);
   } else {
      hw_prim = get_hw_prim_for_gl_prim(prim->mode);
   }

   if (hw_prim != brw->primitive) {
      brw->primitive = hw_prim;
      brw->ctx.NewDriverState |= BRW_NEW_PRIMITIVE;
      if (prim->mode == GL_PATCHES)
         brw->ctx.NewDriverState |= BRW_NEW_PATCH_PRIMITIVE;
   }
}


/**
 * The hardware is capable of removing dangling vertices on its own; however,
 * prior to Gen6, we sometimes convert quads into trifans (and quad strips
 * into tristrips), since pre-Gen6 hardware requires a GS to render quads.
 * This function manually trims dangling vertices from a draw call involving
 * quads so that those dangling vertices won't get drawn when we convert to
 * trifans/tristrips.
 */
static GLuint
trim(GLenum prim, GLuint length)
{
   if (prim == GL_QUAD_STRIP)
      return length > 3 ? (length - length % 2) : 0;
   else if (prim == GL_QUADS)
      return length - length % 4;
   else
      return length;
}


static void
brw_emit_prim(struct brw_context *brw,
              const struct _mesa_prim *prim,
              uint32_t hw_prim,
              struct brw_transform_feedback_object *xfb_obj,
              unsigned stream)
{
   const struct gen_device_info *devinfo = &brw->screen->devinfo;
   int verts_per_instance;
   int vertex_access_type;
   int indirect_flag;

   DBG("PRIM: %s %d %d\n", _mesa_enum_to_string(prim->mode),
       prim->start, prim->count);

   int start_vertex_location = prim->start;
   int base_vertex_location = prim->basevertex;

   if (prim->indexed) {
      vertex_access_type = devinfo->gen >= 7 ?
         GEN7_3DPRIM_VERTEXBUFFER_ACCESS_RANDOM :
         GEN4_3DPRIM_VERTEXBUFFER_ACCESS_RANDOM;
      start_vertex_location += brw->ib.start_vertex_offset;
      base_vertex_location += brw->vb.start_vertex_bias;
   } else {
      vertex_access_type = devinfo->gen >= 7 ?
         GEN7_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL :
         GEN4_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL;
      start_vertex_location += brw->vb.start_vertex_bias;
   }

   /* We only need to trim the primitive count on pre-Gen6. */
   if (devinfo->gen < 6)
      verts_per_instance = trim(prim->mode, prim->count);
   else
      verts_per_instance = prim->count;

   /* If nothing to emit, just return. */
   if (verts_per_instance == 0 && !prim->is_indirect && !xfb_obj)
      return;

   /* If we're set to always flush, do it before and after the primitive emit.
    * We want to catch both missed flushes that hurt instruction/state cache
    * and missed flushes of the render cache as it heads to other parts of
    * the besides the draw code.
    */
   if (brw->always_flush_cache)
      brw_emit_mi_flush(brw);

   /* If indirect, emit a bunch of loads from the indirect BO. */
   if (xfb_obj) {
      indirect_flag = GEN7_3DPRIM_INDIRECT_PARAMETER_ENABLE;

      brw_load_register_mem(brw, GEN7_3DPRIM_VERTEX_COUNT,
                            xfb_obj->prim_count_bo,
                            stream * sizeof(uint32_t));
      BEGIN_BATCH(9);
      OUT_BATCH(MI_LOAD_REGISTER_IMM | (9 - 2));
      OUT_BATCH(GEN7_3DPRIM_INSTANCE_COUNT);
      OUT_BATCH(prim->num_instances);
      OUT_BATCH(GEN7_3DPRIM_START_VERTEX);
      OUT_BATCH(0);
      OUT_BATCH(GEN7_3DPRIM_BASE_VERTEX);
      OUT_BATCH(0);
      OUT_BATCH(GEN7_3DPRIM_START_INSTANCE);
      OUT_BATCH(0);
      ADVANCE_BATCH();
   } else if (prim->is_indirect) {
      struct gl_buffer_object *indirect_buffer = brw->ctx.DrawIndirectBuffer;
      struct brw_bo *bo = intel_bufferobj_buffer(brw,
            intel_buffer_object(indirect_buffer),
            prim->indirect_offset, 5 * sizeof(GLuint), false);

      indirect_flag = GEN7_3DPRIM_INDIRECT_PARAMETER_ENABLE;

      brw_load_register_mem(brw, GEN7_3DPRIM_VERTEX_COUNT, bo,
                            prim->indirect_offset + 0);
      brw_load_register_mem(brw, GEN7_3DPRIM_INSTANCE_COUNT, bo,
                            prim->indirect_offset + 4);

      brw_load_register_mem(brw, GEN7_3DPRIM_START_VERTEX, bo,
                            prim->indirect_offset + 8);
      if (prim->indexed) {
         brw_load_register_mem(brw, GEN7_3DPRIM_BASE_VERTEX, bo,
                               prim->indirect_offset + 12);
         brw_load_register_mem(brw, GEN7_3DPRIM_START_INSTANCE, bo,
                               prim->indirect_offset + 16);
      } else {
         brw_load_register_mem(brw, GEN7_3DPRIM_START_INSTANCE, bo,
                               prim->indirect_offset + 12);
         brw_load_register_imm32(brw, GEN7_3DPRIM_BASE_VERTEX, 0);
      }
   } else {
      indirect_flag = 0;
   }

   BEGIN_BATCH(devinfo->gen >= 7 ? 7 : 6);

   if (devinfo->gen >= 7) {
      const int predicate_enable =
         (brw->predicate.state == BRW_PREDICATE_STATE_USE_BIT)
         ? GEN7_3DPRIM_PREDICATE_ENABLE : 0;

      OUT_BATCH(CMD_3D_PRIM << 16 | (7 - 2) | indirect_flag | predicate_enable);
      OUT_BATCH(hw_prim | vertex_access_type);
   } else {
      OUT_BATCH(CMD_3D_PRIM << 16 | (6 - 2) |
                hw_prim << GEN4_3DPRIM_TOPOLOGY_TYPE_SHIFT |
                vertex_access_type);
   }
   OUT_BATCH(verts_per_instance);
   OUT_BATCH(start_vertex_location);
   OUT_BATCH(prim->num_instances);
   OUT_BATCH(prim->base_instance);
   OUT_BATCH(base_vertex_location);
   ADVANCE_BATCH();

   if (brw->always_flush_cache)
      brw_emit_mi_flush(brw);
}


static void
brw_merge_inputs(struct brw_context *brw)
{
   const struct gen_device_info *devinfo = &brw->screen->devinfo;
   const struct gl_context *ctx = &brw->ctx;
   GLuint i;

   for (i = 0; i < brw->vb.nr_buffers; i++) {
      brw_bo_unreference(brw->vb.buffers[i].bo);
      brw->vb.buffers[i].bo = NULL;
   }
   brw->vb.nr_buffers = 0;

   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
      struct brw_vertex_element *input = &brw->vb.inputs[i];
      input->buffer = -1;
      _mesa_draw_attrib_and_binding(ctx, i,
                                    &input->glattrib, &input->glbinding);
   }

   if (devinfo->gen < 8 && !devinfo->is_haswell) {
      uint64_t mask = ctx->VertexProgram._Current->info.inputs_read;
      /* Prior to Haswell, the hardware can't natively support GL_FIXED or
       * 2_10_10_10_REV vertex formats.  Set appropriate workaround flags.
       */
      while (mask) {
         const struct gl_vertex_format *glformat;
         uint8_t wa_flags = 0;

         i = u_bit_scan64(&mask);
         glformat = &brw->vb.inputs[i].glattrib->Format;

         switch (glformat->Type) {

         case GL_FIXED:
            wa_flags = glformat->Size;
            break;

         case GL_INT_2_10_10_10_REV:
            wa_flags |= BRW_ATTRIB_WA_SIGN;
            /* fallthough */

         case GL_UNSIGNED_INT_2_10_10_10_REV:
            if (glformat->Format == GL_BGRA)
               wa_flags |= BRW_ATTRIB_WA_BGRA;

            if (glformat->Normalized)
               wa_flags |= BRW_ATTRIB_WA_NORMALIZE;
            else if (!glformat->Integer)
               wa_flags |= BRW_ATTRIB_WA_SCALE;

            break;
         }

         if (brw->vb.attrib_wa_flags[i] != wa_flags) {
            brw->vb.attrib_wa_flags[i] = wa_flags;
            brw->ctx.NewDriverState |= BRW_NEW_VS_ATTRIB_WORKAROUNDS;
         }
      }
   }
}

/* Disable auxiliary buffers if a renderbuffer is also bound as a texture
 * or shader image.  This causes a self-dependency, where both rendering
 * and sampling may concurrently read or write the CCS buffer, causing
 * incorrect pixels.
 */
static bool
intel_disable_rb_aux_buffer(struct brw_context *brw,
                            bool *draw_aux_buffer_disabled,
                            struct intel_mipmap_tree *tex_mt,
                            unsigned min_level, unsigned num_levels,
                            const char *usage)
{
   const struct gl_framebuffer *fb = brw->ctx.DrawBuffer;
   bool found = false;

   /* We only need to worry about color compression and fast clears. */
   if (tex_mt->aux_usage != ISL_AUX_USAGE_CCS_D &&
       tex_mt->aux_usage != ISL_AUX_USAGE_CCS_E)
      return false;

   for (unsigned i = 0; i < fb->_NumColorDrawBuffers; i++) {
      const struct intel_renderbuffer *irb =
         intel_renderbuffer(fb->_ColorDrawBuffers[i]);

      if (irb && irb->mt->bo == tex_mt->bo &&
          irb->mt_level >= min_level &&
          irb->mt_level < min_level + num_levels) {
         found = draw_aux_buffer_disabled[i] = true;
      }
   }

   if (found) {
      perf_debug("Disabling CCS because a renderbuffer is also bound %s.\n",
                 usage);
   }

   return found;
}

/** Implement the ASTC 5x5 sampler workaround
 *
 * Gen9 sampling hardware has a bug where an ASTC 5x5 compressed surface
 * cannot live in the sampler cache at the same time as an aux compressed
 * surface.  In order to work around the bug we have to stall rendering with a
 * CS and pixel scoreboard stall (implicit in the CS stall) and invalidate the
 * texture cache whenever one of ASTC 5x5 or aux compressed may be in the
 * sampler cache and we're about to render with something which samples from
 * the other.
 *
 * In the case of a single shader which textures from both ASTC 5x5 and
 * a texture which is CCS or HiZ compressed, we have to resolve the aux
 * compressed texture prior to rendering.  This second part is handled in
 * brw_predraw_resolve_inputs() below.
 *
 * We have observed this issue to affect CCS and HiZ sampling but whether or
 * not it also affects MCS is unknown.  Because MCS has no concept of a
 * resolve (and doing one would be stupid expensive), we choose to simply
 * ignore the possibility and hope for the best.
 */
static void
gen9_apply_astc5x5_wa_flush(struct brw_context *brw,
                            enum gen9_astc5x5_wa_tex_type curr_mask)
{
   assert(brw->screen->devinfo.gen == 9);

   if (((brw->gen9_astc5x5_wa_tex_mask & GEN9_ASTC5X5_WA_TEX_TYPE_ASTC5x5) &&
        (curr_mask & GEN9_ASTC5X5_WA_TEX_TYPE_AUX)) ||
       ((brw->gen9_astc5x5_wa_tex_mask & GEN9_ASTC5X5_WA_TEX_TYPE_AUX) &&
        (curr_mask & GEN9_ASTC5X5_WA_TEX_TYPE_ASTC5x5))) {
      brw_emit_pipe_control_flush(brw, PIPE_CONTROL_CS_STALL);
      brw_emit_pipe_control_flush(brw, PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE);
   }

   brw->gen9_astc5x5_wa_tex_mask = curr_mask;
}

static enum gen9_astc5x5_wa_tex_type
gen9_astc5x5_wa_bits(mesa_format format, enum isl_aux_usage aux_usage)
{
   if (aux_usage != ISL_AUX_USAGE_NONE &&
       aux_usage != ISL_AUX_USAGE_MCS)
      return GEN9_ASTC5X5_WA_TEX_TYPE_AUX;

   if (format == MESA_FORMAT_RGBA_ASTC_5x5 ||
       format == MESA_FORMAT_SRGB8_ALPHA8_ASTC_5x5)
      return GEN9_ASTC5X5_WA_TEX_TYPE_ASTC5x5;

   return 0;
}

/* Helper for the gen9 ASTC 5x5 workaround.  This version exists for BLORP's
 * use-cases where only a single texture is bound.
 */
void
gen9_apply_single_tex_astc5x5_wa(struct brw_context *brw,
                                 mesa_format format,
                                 enum isl_aux_usage aux_usage)
{
   gen9_apply_astc5x5_wa_flush(brw, gen9_astc5x5_wa_bits(format, aux_usage));
}

static void
mark_textures_used_for_txf(BITSET_WORD *used_for_txf,
                           const struct gl_program *prog)
{
   if (!prog)
      return;

   uint32_t mask = prog->info.textures_used_by_txf;
   while (mask) {
      int s = u_bit_scan(&mask);
      BITSET_SET(used_for_txf, prog->SamplerUnits[s]);
   }
}

/**
 * \brief Resolve buffers before drawing.
 *
 * Resolve the depth buffer's HiZ buffer, resolve the depth buffer of each
 * enabled depth texture, and flush the render cache for any dirty textures.
 */
void
brw_predraw_resolve_inputs(struct brw_context *brw, bool rendering,
                           bool *draw_aux_buffer_disabled)
{
   struct gl_context *ctx = &brw->ctx;
   struct intel_texture_object *tex_obj;

   BITSET_DECLARE(used_for_txf, MAX_COMBINED_TEXTURE_IMAGE_UNITS);
   memset(used_for_txf, 0, sizeof(used_for_txf));
   if (rendering) {
      mark_textures_used_for_txf(used_for_txf, ctx->VertexProgram._Current);
      mark_textures_used_for_txf(used_for_txf, ctx->TessCtrlProgram._Current);
      mark_textures_used_for_txf(used_for_txf, ctx->TessEvalProgram._Current);
      mark_textures_used_for_txf(used_for_txf, ctx->GeometryProgram._Current);
      mark_textures_used_for_txf(used_for_txf, ctx->FragmentProgram._Current);
   } else {
      mark_textures_used_for_txf(used_for_txf, ctx->ComputeProgram._Current);
   }

   int maxEnabledUnit = ctx->Texture._MaxEnabledTexImageUnit;

   enum gen9_astc5x5_wa_tex_type astc5x5_wa_bits = 0;
   if (brw->screen->devinfo.gen == 9) {
      /* In order to properly implement the ASTC 5x5 workaround for an
       * arbitrary draw or dispatch call, we have to walk the entire list of
       * textures looking for ASTC 5x5.  If there is any ASTC 5x5 in this draw
       * call, all aux compressed textures must be resolved and have aux
       * compression disabled while sampling.
       */
      for (int i = 0; i <= maxEnabledUnit; i++) {
         if (!ctx->Texture.Unit[i]._Current)
            continue;
         tex_obj = intel_texture_object(ctx->Texture.Unit[i]._Current);
         if (!tex_obj || !tex_obj->mt)
            continue;

         astc5x5_wa_bits |= gen9_astc5x5_wa_bits(tex_obj->_Format,
                                                 tex_obj->mt->aux_usage);
      }
      gen9_apply_astc5x5_wa_flush(brw, astc5x5_wa_bits);
   }

   /* Resolve depth buffer and render cache of each enabled texture. */
   for (int i = 0; i <= maxEnabledUnit; i++) {
      if (!ctx->Texture.Unit[i]._Current)
	 continue;
      tex_obj = intel_texture_object(ctx->Texture.Unit[i]._Current);
      if (!tex_obj || !tex_obj->mt)
	 continue;

      struct gl_sampler_object *sampler = _mesa_get_samplerobj(ctx, i);
      enum isl_format view_format =
         translate_tex_format(brw, tex_obj->_Format, sampler->sRGBDecode);

      unsigned min_level, min_layer, num_levels, num_layers;
      if (tex_obj->base.Immutable) {
         min_level  = tex_obj->base.MinLevel;
         num_levels = MIN2(tex_obj->base.NumLevels, tex_obj->_MaxLevel + 1);
         min_layer  = tex_obj->base.MinLayer;
         num_layers = tex_obj->base.Target != GL_TEXTURE_3D ?
                      tex_obj->base.NumLayers : INTEL_REMAINING_LAYERS;
      } else {
         min_level  = tex_obj->base.BaseLevel;
         num_levels = tex_obj->_MaxLevel - tex_obj->base.BaseLevel + 1;
         min_layer  = 0;
         num_layers = INTEL_REMAINING_LAYERS;
      }

      if (rendering) {
         intel_disable_rb_aux_buffer(brw, draw_aux_buffer_disabled,
                                     tex_obj->mt, min_level, num_levels,
                                     "for sampling");
      }

      intel_miptree_prepare_texture(brw, tex_obj->mt, view_format,
                                    min_level, num_levels,
                                    min_layer, num_layers,
                                    astc5x5_wa_bits);

      /* If any programs are using it with texelFetch, we may need to also do
       * a prepare with an sRGB format to ensure texelFetch works "properly".
       */
      if (BITSET_TEST(used_for_txf, i)) {
         enum isl_format txf_format =
            translate_tex_format(brw, tex_obj->_Format, GL_DECODE_EXT);
         if (txf_format != view_format) {
            intel_miptree_prepare_texture(brw, tex_obj->mt, txf_format,
                                          min_level, num_levels,
                                          min_layer, num_layers,
                                          astc5x5_wa_bits);
         }
      }

      brw_cache_flush_for_read(brw, tex_obj->mt->bo);

      if (tex_obj->base.StencilSampling ||
          tex_obj->mt->format == MESA_FORMAT_S_UINT8) {
         intel_update_r8stencil(brw, tex_obj->mt);
      }
   }

   /* Resolve color for each active shader image. */
   for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
      const struct gl_program *prog = ctx->_Shader->CurrentProgram[i];

      if (unlikely(prog && prog->info.num_images)) {
         for (unsigned j = 0; j < prog->info.num_images; j++) {
            struct gl_image_unit *u =
               &ctx->ImageUnits[prog->sh.ImageUnits[j]];
            tex_obj = intel_texture_object(u->TexObj);

            if (tex_obj && tex_obj->mt) {
               if (rendering) {
                  intel_disable_rb_aux_buffer(brw, draw_aux_buffer_disabled,
                                              tex_obj->mt, 0, ~0,
                                              "as a shader image");
               }

               intel_miptree_prepare_image(brw, tex_obj->mt);

               brw_cache_flush_for_read(brw, tex_obj->mt->bo);
            }
         }
      }
   }
}

static void
brw_predraw_resolve_framebuffer(struct brw_context *brw,
                                bool *draw_aux_buffer_disabled)
{
   struct gl_context *ctx = &brw->ctx;
   struct intel_renderbuffer *depth_irb;

   /* Resolve the depth buffer's HiZ buffer. */
   depth_irb = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_DEPTH);
   if (depth_irb && depth_irb->mt) {
      intel_miptree_prepare_depth(brw, depth_irb->mt,
                                  depth_irb->mt_level,
                                  depth_irb->mt_layer,
                                  depth_irb->layer_count);
   }

   /* Resolve color buffers for non-coherent framebuffer fetch. */
   if (!ctx->Extensions.EXT_shader_framebuffer_fetch &&
       ctx->FragmentProgram._Current &&
       ctx->FragmentProgram._Current->info.outputs_read) {
      const struct gl_framebuffer *fb = ctx->DrawBuffer;

      /* This is only used for non-coherent framebuffer fetch, so we don't
       * need to worry about CCS_E and can simply pass 'false' below.
       */
      assert(brw->screen->devinfo.gen < 9);

      for (unsigned i = 0; i < fb->_NumColorDrawBuffers; i++) {
         const struct intel_renderbuffer *irb =
            intel_renderbuffer(fb->_ColorDrawBuffers[i]);

         if (irb) {
            intel_miptree_prepare_texture(brw, irb->mt, irb->mt->surf.format,
                                          irb->mt_level, 1,
                                          irb->mt_layer, irb->layer_count,
                                          brw->gen9_astc5x5_wa_tex_mask);
         }
      }
   }

   struct gl_framebuffer *fb = ctx->DrawBuffer;
   for (int i = 0; i < fb->_NumColorDrawBuffers; i++) {
      struct intel_renderbuffer *irb =
         intel_renderbuffer(fb->_ColorDrawBuffers[i]);

      if (irb == NULL || irb->mt == NULL)
         continue;

      mesa_format mesa_format =
         _mesa_get_render_format(ctx, intel_rb_format(irb));
      enum isl_format isl_format = brw_isl_format_for_mesa_format(mesa_format);
      bool blend_enabled = ctx->Color.BlendEnabled & (1 << i);
      enum isl_aux_usage aux_usage =
         intel_miptree_render_aux_usage(brw, irb->mt, isl_format,
                                        blend_enabled,
                                        draw_aux_buffer_disabled[i]);
      if (brw->draw_aux_usage[i] != aux_usage) {
         brw->ctx.NewDriverState |= BRW_NEW_AUX_STATE;
         brw->draw_aux_usage[i] = aux_usage;
      }

      intel_miptree_prepare_render(brw, irb->mt, irb->mt_level,
                                   irb->mt_layer, irb->layer_count,
                                   aux_usage);

      brw_cache_flush_for_render(brw, irb->mt->bo,
                                 isl_format, aux_usage);
   }
}

/**
 * \brief Call this after drawing to mark which buffers need resolving
 *
 * If the depth buffer was written to and if it has an accompanying HiZ
 * buffer, then mark that it needs a depth resolve.
 *
 * If the stencil buffer was written to then mark that it may need to be
 * copied to an R8 texture.
 *
 * If the color buffer is a multisample window system buffer, then
 * mark that it needs a downsample.
 *
 * Also mark any render targets which will be textured as needing a render
 * cache flush.
 */
static void
brw_postdraw_set_buffers_need_resolve(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   struct gl_framebuffer *fb = ctx->DrawBuffer;

   struct intel_renderbuffer *front_irb = NULL;
   struct intel_renderbuffer *back_irb = intel_get_renderbuffer(fb, BUFFER_BACK_LEFT);
   struct intel_renderbuffer *depth_irb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
   struct intel_renderbuffer *stencil_irb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
   struct gl_renderbuffer_attachment *depth_att = &fb->Attachment[BUFFER_DEPTH];

   if (_mesa_is_front_buffer_drawing(fb))
      front_irb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);

   if (front_irb)
      front_irb->need_downsample = true;
   if (back_irb)
      back_irb->need_downsample = true;
   if (depth_irb) {
      bool depth_written = brw_depth_writes_enabled(brw);
      if (depth_att->Layered) {
         intel_miptree_finish_depth(brw, depth_irb->mt,
                                    depth_irb->mt_level,
                                    depth_irb->mt_layer,
                                    depth_irb->layer_count,
                                    depth_written);
      } else {
         intel_miptree_finish_depth(brw, depth_irb->mt,
                                    depth_irb->mt_level,
                                    depth_irb->mt_layer, 1,
                                    depth_written);
      }
      if (depth_written)
         brw_depth_cache_add_bo(brw, depth_irb->mt->bo);
   }

   if (stencil_irb && brw->stencil_write_enabled) {
      struct intel_mipmap_tree *stencil_mt =
         stencil_irb->mt->stencil_mt != NULL ?
         stencil_irb->mt->stencil_mt : stencil_irb->mt;
      brw_depth_cache_add_bo(brw, stencil_mt->bo);
      intel_miptree_finish_write(brw, stencil_mt, stencil_irb->mt_level,
                                 stencil_irb->mt_layer,
                                 stencil_irb->layer_count, ISL_AUX_USAGE_NONE);
   }

   for (unsigned i = 0; i < fb->_NumColorDrawBuffers; i++) {
      struct intel_renderbuffer *irb =
         intel_renderbuffer(fb->_ColorDrawBuffers[i]);

      if (!irb)
         continue;

      mesa_format mesa_format =
         _mesa_get_render_format(ctx, intel_rb_format(irb));
      enum isl_format isl_format = brw_isl_format_for_mesa_format(mesa_format);
      enum isl_aux_usage aux_usage = brw->draw_aux_usage[i];

      brw_render_cache_add_bo(brw, irb->mt->bo, isl_format, aux_usage);

      intel_miptree_finish_render(brw, irb->mt, irb->mt_level,
                                  irb->mt_layer, irb->layer_count,
                                  aux_usage);
   }
}

static void
intel_renderbuffer_move_temp_back(struct brw_context *brw,
                                  struct intel_renderbuffer *irb)
{
   if (irb->align_wa_mt == NULL)
      return;

   brw_cache_flush_for_read(brw, irb->align_wa_mt->bo);

   intel_miptree_copy_slice(brw, irb->align_wa_mt, 0, 0,
                            irb->mt,
                            irb->Base.Base.TexImage->Level, irb->mt_layer);

   intel_miptree_reference(&irb->align_wa_mt, NULL);

   /* Finally restore the x,y to correspond to full miptree. */
   intel_renderbuffer_set_draw_offset(irb);

   /* Make sure render surface state gets re-emitted with updated miptree. */
   brw->NewGLState |= _NEW_BUFFERS;
}

static void
brw_postdraw_reconcile_align_wa_slices(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   struct gl_framebuffer *fb = ctx->DrawBuffer;

   struct intel_renderbuffer *depth_irb =
      intel_get_renderbuffer(fb, BUFFER_DEPTH);
   struct intel_renderbuffer *stencil_irb =
      intel_get_renderbuffer(fb, BUFFER_STENCIL);

   if (depth_irb && depth_irb->align_wa_mt)
      intel_renderbuffer_move_temp_back(brw, depth_irb);

   if (stencil_irb && stencil_irb->align_wa_mt)
      intel_renderbuffer_move_temp_back(brw, stencil_irb);

   for (unsigned i = 0; i < fb->_NumColorDrawBuffers; i++) {
      struct intel_renderbuffer *irb =
         intel_renderbuffer(fb->_ColorDrawBuffers[i]);

      if (!irb || irb->align_wa_mt == NULL)
         continue;

      intel_renderbuffer_move_temp_back(brw, irb);
   }
}

static void
brw_prepare_drawing(struct gl_context *ctx,
                    const struct _mesa_index_buffer *ib,
                    bool index_bounds_valid,
                    GLuint min_index,
                    GLuint max_index)
{
   struct brw_context *brw = brw_context(ctx);

   if (ctx->NewState)
      _mesa_update_state(ctx);

   /* We have to validate the textures *before* checking for fallbacks;
    * otherwise, the software fallback won't be able to rely on the
    * texture state, the firstLevel and lastLevel fields won't be
    * set in the intel texture object (they'll both be 0), and the
    * software fallback will segfault if it attempts to access any
    * texture level other than level 0.
    */
   brw_validate_textures(brw);

   /* Find the highest sampler unit used by each shader program.  A bit-count
    * won't work since ARB programs use the texture unit number as the sampler
    * index.
    */
   brw->wm.base.sampler_count =
      util_last_bit(ctx->FragmentProgram._Current->info.textures_used);
   brw->gs.base.sampler_count = ctx->GeometryProgram._Current ?
      util_last_bit(ctx->GeometryProgram._Current->info.textures_used) : 0;
   brw->tes.base.sampler_count = ctx->TessEvalProgram._Current ?
      util_last_bit(ctx->TessEvalProgram._Current->info.textures_used) : 0;
   brw->tcs.base.sampler_count = ctx->TessCtrlProgram._Current ?
      util_last_bit(ctx->TessCtrlProgram._Current->info.textures_used) : 0;
   brw->vs.base.sampler_count =
      util_last_bit(ctx->VertexProgram._Current->info.textures_used);

   intel_prepare_render(brw);

   /* This workaround has to happen outside of brw_upload_render_state()
    * because it may flush the batchbuffer for a blit, affecting the state
    * flags.
    */
   brw_workaround_depthstencil_alignment(brw, 0);

   /* Resolves must occur after updating renderbuffers, updating context state,
    * and finalizing textures but before setting up any hardware state for
    * this draw call.
    */
   bool draw_aux_buffer_disabled[MAX_DRAW_BUFFERS] = { };
   brw_predraw_resolve_inputs(brw, true, draw_aux_buffer_disabled);
   brw_predraw_resolve_framebuffer(brw, draw_aux_buffer_disabled);

   /* Bind all inputs, derive varying and size information:
    */
   brw_merge_inputs(brw);

   brw->ib.ib = ib;
   brw->ctx.NewDriverState |= BRW_NEW_INDICES;

   brw->vb.index_bounds_valid = index_bounds_valid;
   brw->vb.min_index = min_index;
   brw->vb.max_index = max_index;
   brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
}

static void
brw_finish_drawing(struct gl_context *ctx)
{
   struct brw_context *brw = brw_context(ctx);

   if (brw->always_flush_batch)
      intel_batchbuffer_flush(brw);

   brw_program_cache_check_size(brw);
   brw_postdraw_reconcile_align_wa_slices(brw);
   brw_postdraw_set_buffers_need_resolve(brw);

   if (brw->draw.draw_params_count_bo) {
      brw_bo_unreference(brw->draw.draw_params_count_bo);
      brw->draw.draw_params_count_bo = NULL;
   }
}

/**
 * Implement workarounds for preemption:
 *    - WaDisableMidObjectPreemptionForGSLineStripAdj
 *    - WaDisableMidObjectPreemptionForTrifanOrPolygon
 *    - WaDisableMidObjectPreemptionForLineLoop
 *    - WA#0798
 */
static void
gen9_emit_preempt_wa(struct brw_context *brw,
                     const struct _mesa_prim *prim)
{
   bool object_preemption = true;
   const struct gen_device_info *devinfo = &brw->screen->devinfo;

   /* Only apply these workarounds for gen9 */
   assert(devinfo->gen == 9);

   /* WaDisableMidObjectPreemptionForGSLineStripAdj
    *
    *    WA: Disable mid-draw preemption when draw-call is a linestrip_adj and
    *    GS is enabled.
    */
   if (brw->primitive == _3DPRIM_LINESTRIP_ADJ && brw->gs.enabled)
      object_preemption = false;

   /* WaDisableMidObjectPreemptionForTrifanOrPolygon
    *
    *    TriFan miscompare in Execlist Preemption test. Cut index that is on a
    *    previous context. End the previous, the resume another context with a
    *    tri-fan or polygon, and the vertex count is corrupted. If we prempt
    *    again we will cause corruption.
    *
    *    WA: Disable mid-draw preemption when draw-call has a tri-fan.
    */
   if (brw->primitive == _3DPRIM_TRIFAN)
      object_preemption = false;

   /* WaDisableMidObjectPreemptionForLineLoop
    *
    *    VF Stats Counters Missing a vertex when preemption enabled.
    *
    *    WA: Disable mid-draw preemption when the draw uses a lineloop
    *    topology.
    */
   if (brw->primitive == _3DPRIM_LINELOOP)
      object_preemption = false;

   /* WA#0798
    *
    *    VF is corrupting GAFS data when preempted on an instance boundary and
    *    replayed with instancing enabled.
    *
    *    WA: Disable preemption when using instanceing.
    */
   if (prim->num_instances > 1)
      object_preemption = false;

   brw_enable_obj_preemption(brw, object_preemption);
}

/* May fail if out of video memory for texture or vbo upload, or on
 * fallback conditions.
 */
static void
brw_draw_single_prim(struct gl_context *ctx,
                     const struct _mesa_prim *prim,
                     unsigned prim_id,
                     struct brw_transform_feedback_object *xfb_obj,
                     unsigned stream,
                     struct gl_buffer_object *indirect)
{
   struct brw_context *brw = brw_context(ctx);
   const struct gen_device_info *devinfo = &brw->screen->devinfo;
   bool fail_next;

   /* Flag BRW_NEW_DRAW_CALL on every draw.  This allows us to have
    * atoms that happen on every draw call.
    */
   brw->ctx.NewDriverState |= BRW_NEW_DRAW_CALL;

   /* Flush the batch if the batch/state buffers are nearly full.  We can
    * grow them if needed, but this is not free, so we'd like to avoid it.
    */
   intel_batchbuffer_require_space(brw, 1500);
   brw_require_statebuffer_space(brw, 2400);
   intel_batchbuffer_save_state(brw);
   fail_next = intel_batchbuffer_saved_state_is_empty(brw);

   if (brw->num_instances != prim->num_instances ||
       brw->basevertex != prim->basevertex ||
       brw->baseinstance != prim->base_instance) {
      brw->num_instances = prim->num_instances;
      brw->basevertex = prim->basevertex;
      brw->baseinstance = prim->base_instance;
      if (prim_id > 0) { /* For i == 0 we just did this before the loop */
         brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
         brw_merge_inputs(brw);
      }
   }

   /* Determine if we need to flag BRW_NEW_VERTICES for updating the
    * gl_BaseVertexARB or gl_BaseInstanceARB values. For indirect draw, we
    * always flag if the shader uses one of the values. For direct draws,
    * we only flag if the values change.
    */
   const int new_firstvertex =
      prim->indexed ? prim->basevertex : prim->start;
   const int new_baseinstance = prim->base_instance;
   const struct brw_vs_prog_data *vs_prog_data =
      brw_vs_prog_data(brw->vs.base.prog_data);
   if (prim_id > 0) {
      const bool uses_draw_parameters =
         vs_prog_data->uses_firstvertex ||
         vs_prog_data->uses_baseinstance;

      if ((uses_draw_parameters && prim->is_indirect) ||
          (vs_prog_data->uses_firstvertex &&
           brw->draw.params.firstvertex != new_firstvertex) ||
          (vs_prog_data->uses_baseinstance &&
           brw->draw.params.gl_baseinstance != new_baseinstance))
         brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
   }

   brw->draw.params.firstvertex = new_firstvertex;
   brw->draw.params.gl_baseinstance = new_baseinstance;
   brw_bo_unreference(brw->draw.draw_params_bo);

   if (prim->is_indirect) {
      /* Point draw_params_bo at the indirect buffer. */
      brw->draw.draw_params_bo =
         intel_buffer_object(ctx->DrawIndirectBuffer)->buffer;
      brw_bo_reference(brw->draw.draw_params_bo);
      brw->draw.draw_params_offset =
         prim->indirect_offset + (prim->indexed ? 12 : 8);
   } else {
      /* Set draw_params_bo to NULL so brw_prepare_vertices knows it
       * has to upload gl_BaseVertex and such if they're needed.
       */
      brw->draw.draw_params_bo = NULL;
      brw->draw.draw_params_offset = 0;
   }

   /* gl_DrawID always needs its own vertex buffer since it's not part of
    * the indirect parameter buffer. Same for is_indexed_draw, which shares
    * the buffer with gl_DrawID. If the program uses gl_DrawID, we need to
    * flag BRW_NEW_VERTICES. For the first iteration, we don't have valid
    * vs_prog_data, but we always flag BRW_NEW_VERTICES before the loop.
    */
   if (prim_id > 0 && vs_prog_data->uses_drawid)
      brw->ctx.NewDriverState |= BRW_NEW_VERTICES;

   brw->draw.derived_params.gl_drawid = prim->draw_id;
   brw->draw.derived_params.is_indexed_draw = prim->indexed ? ~0 : 0;

   brw_bo_unreference(brw->draw.derived_draw_params_bo);
   brw->draw.derived_draw_params_bo = NULL;
   brw->draw.derived_draw_params_offset = 0;

   if (devinfo->gen < 6)
      brw_set_prim(brw, prim);
   else
      gen6_set_prim(brw, prim);

retry:

   /* Note that before the loop, brw->ctx.NewDriverState was set to != 0, and
    * that the state updated in the loop outside of this block is that in
    * *_set_prim or intel_batchbuffer_flush(), which only impacts
    * brw->ctx.NewDriverState.
    */
   if (brw->ctx.NewDriverState) {
      brw->batch.no_wrap = true;
      brw_upload_render_state(brw);
   }

   if (devinfo->gen == 9)
      gen9_emit_preempt_wa(brw, prim);

   brw_emit_prim(brw, prim, brw->primitive, xfb_obj, stream);

   brw->batch.no_wrap = false;

   if (!brw_batch_has_aperture_space(brw, 0)) {
      if (!fail_next) {
         intel_batchbuffer_reset_to_saved(brw);
         intel_batchbuffer_flush(brw);
         fail_next = true;
         goto retry;
      } else {
         int ret = intel_batchbuffer_flush(brw);
         WARN_ONCE(ret == -ENOSPC,
                   "i965: Single primitive emit exceeded "
                   "available aperture space\n");
      }
   }

   /* Now that we know we haven't run out of aperture space, we can safely
    * reset the dirty bits.
    */
   if (brw->ctx.NewDriverState)
      brw_render_state_finished(brw);

   return;
}



void
brw_draw_prims(struct gl_context *ctx,
               const struct _mesa_prim *prims,
               GLuint nr_prims,
               const struct _mesa_index_buffer *ib,
               GLboolean index_bounds_valid,
               GLuint min_index,
               GLuint max_index,
               struct gl_transform_feedback_object *gl_xfb_obj,
               unsigned stream,
               struct gl_buffer_object *indirect)
{
   unsigned i;
   struct brw_context *brw = brw_context(ctx);
   int predicate_state = brw->predicate.state;
   struct brw_transform_feedback_object *xfb_obj =
      (struct brw_transform_feedback_object *) gl_xfb_obj;

   if (!brw_check_conditional_render(brw))
      return;

   /* Handle primitive restart if needed */
   if (brw_handle_primitive_restart(ctx, prims, nr_prims, ib, indirect)) {
      /* The draw was handled, so we can exit now */
      return;
   }

   /* Do GL_SELECT and GL_FEEDBACK rendering using swrast, even though it
    * won't support all the extensions we support.
    */
   if (ctx->RenderMode != GL_RENDER) {
      perf_debug("%s render mode not supported in hardware\n",
                 _mesa_enum_to_string(ctx->RenderMode));
      _swsetup_Wakeup(ctx);
      _tnl_wakeup(ctx);
      _tnl_draw(ctx, prims, nr_prims, ib,
                index_bounds_valid, min_index, max_index, NULL, 0, NULL);
      return;
   }

   /* If we're going to have to upload any of the user's vertex arrays, then
    * get the minimum and maximum of their index buffer so we know what range
    * to upload.
    */
   if (!index_bounds_valid && _mesa_draw_user_array_bits(ctx) != 0) {
      perf_debug("Scanning index buffer to compute index buffer bounds.  "
                 "Use glDrawRangeElements() to avoid this.\n");
      vbo_get_minmax_indices(ctx, prims, ib, &min_index, &max_index, nr_prims);
      index_bounds_valid = true;
   }

   brw_prepare_drawing(ctx, ib, index_bounds_valid, min_index, max_index);
   /* Try drawing with the hardware, but don't do anything else if we can't
    * manage it.  swrast doesn't support our featureset, so we can't fall back
    * to it.
    */

   for (i = 0; i < nr_prims; i++) {
      /* Implementation of ARB_indirect_parameters via predicates */
      if (brw->draw.draw_params_count_bo) {
         brw_emit_pipe_control_flush(brw, PIPE_CONTROL_FLUSH_ENABLE);

         /* Upload the current draw count from the draw parameters buffer to
          * MI_PREDICATE_SRC0.
          */
         brw_load_register_mem(brw, MI_PREDICATE_SRC0,
                               brw->draw.draw_params_count_bo,
                               brw->draw.draw_params_count_offset);
         /* Zero the top 32-bits of MI_PREDICATE_SRC0 */
         brw_load_register_imm32(brw, MI_PREDICATE_SRC0 + 4, 0);
         /* Upload the id of the current primitive to MI_PREDICATE_SRC1. */
         brw_load_register_imm64(brw, MI_PREDICATE_SRC1, prims[i].draw_id);

         BEGIN_BATCH(1);
         if (i == 0 && brw->predicate.state != BRW_PREDICATE_STATE_USE_BIT) {
            OUT_BATCH(GEN7_MI_PREDICATE | MI_PREDICATE_LOADOP_LOADINV |
                      MI_PREDICATE_COMBINEOP_SET |
                      MI_PREDICATE_COMPAREOP_SRCS_EQUAL);
         } else {
            OUT_BATCH(GEN7_MI_PREDICATE |
                      MI_PREDICATE_LOADOP_LOAD | MI_PREDICATE_COMBINEOP_XOR |
                      MI_PREDICATE_COMPAREOP_SRCS_EQUAL);
         }
         ADVANCE_BATCH();

         brw->predicate.state = BRW_PREDICATE_STATE_USE_BIT;
      }

      brw_draw_single_prim(ctx, &prims[i], i, xfb_obj, stream, indirect);
   }

   brw_finish_drawing(ctx);
   brw->predicate.state = predicate_state;
}

void
brw_draw_indirect_prims(struct gl_context *ctx,
                        GLuint mode,
                        struct gl_buffer_object *indirect_data,
                        GLsizeiptr indirect_offset,
                        unsigned draw_count,
                        unsigned stride,
                        struct gl_buffer_object *indirect_params,
                        GLsizeiptr indirect_params_offset,
                        const struct _mesa_index_buffer *ib)
{
   struct brw_context *brw = brw_context(ctx);
   struct _mesa_prim *prim;
   GLsizei i;

   prim = calloc(draw_count, sizeof(*prim));
   if (prim == NULL) {
      _mesa_error(ctx, GL_OUT_OF_MEMORY, "gl%sDraw%sIndirect%s",
                  (draw_count > 1) ? "Multi" : "",
                  ib ? "Elements" : "Arrays",
                  indirect_params ? "CountARB" : "");
      return;
   }

   prim[0].begin = 1;
   prim[draw_count - 1].end = 1;
   for (i = 0; i < draw_count; ++i, indirect_offset += stride) {
      prim[i].mode = mode;
      prim[i].indexed = ib != NULL;
      prim[i].indirect_offset = indirect_offset;
      prim[i].is_indirect = 1;
      prim[i].draw_id = i;
   }

   if (indirect_params) {
      brw->draw.draw_params_count_bo =
         intel_buffer_object(indirect_params)->buffer;
      brw_bo_reference(brw->draw.draw_params_count_bo);
      brw->draw.draw_params_count_offset = indirect_params_offset;
   }

   brw_draw_prims(ctx, prim, draw_count,
                  ib, false, 0, ~0,
                  NULL, 0,
                  indirect_data);

   free(prim);
}

void
brw_init_draw_functions(struct dd_function_table *functions)
{
   /* Register our drawing function:
    */
   functions->Draw = brw_draw_prims;
   functions->DrawIndirect = brw_draw_indirect_prims;
}

void
brw_draw_init(struct brw_context *brw)
{
   for (int i = 0; i < VERT_ATTRIB_MAX; i++)
      brw->vb.inputs[i].buffer = -1;
   brw->vb.nr_buffers = 0;
   brw->vb.nr_enabled = 0;
}

void
brw_draw_destroy(struct brw_context *brw)
{
   unsigned i;

   for (i = 0; i < brw->vb.nr_buffers; i++) {
      brw_bo_unreference(brw->vb.buffers[i].bo);
      brw->vb.buffers[i].bo = NULL;
   }
   brw->vb.nr_buffers = 0;

   for (i = 0; i < brw->vb.nr_enabled; i++) {
      brw->vb.enabled[i]->buffer = -1;
   }
   brw->vb.nr_enabled = 0;

   brw_bo_unreference(brw->ib.bo);
   brw->ib.bo = NULL;
}