panda3d/panda/src/display/graphicsStateGuardian.cxx

/**
 * PANDA 3D SOFTWARE
 * Copyright (c) Carnegie Mellon University.  All rights reserved.
 *
 * All use of this software is subject to the terms of the revised BSD
 * license.  You should have received a copy of this license along
 * with this source code in a file named "LICENSE."
 *
 * @file graphicsStateGuardian.cxx
 * @author drose
 * @date 1999-02-02
 * @author fperazzi, PandaSE
 * @date 2010-05-05
 *  _max_2d_texture_array_layers, _supports_2d_texture_array,
 *  get_supports_cg_profile)
 */

#include "graphicsStateGuardian.h"
#include "graphicsEngine.h"
#include "config_display.h"
#include "textureContext.h"
#include "vertexBufferContext.h"
#include "indexBufferContext.h"
#include "renderBuffer.h"
#include "light.h"
#include "planeNode.h"
#include "ambientLight.h"
#include "throw_event.h"
#include "clockObject.h"
#include "pStatTimer.h"
#include "pStatGPUTimer.h"
#include "geomTristrips.h"
#include "geomTrifans.h"
#include "geomLinestrips.h"
#include "colorWriteAttrib.h"
#include "shader.h"
#include "pnotify.h"
#include "drawableRegion.h"
#include "displayRegion.h"
#include "graphicsOutput.h"
#include "texturePool.h"
#include "geomMunger.h"
#include "stateMunger.h"
#include "ambientLight.h"
#include "directionalLight.h"
#include "pointLight.h"
#include "spotlight.h"
#include "textureReloadRequest.h"
#include "shaderAttrib.h"
#include "materialAttrib.h"
#include "depthWriteAttrib.h"
#include "lightAttrib.h"
#include "texGenAttrib.h"
#include "shaderGenerator.h"
#include "lightLensNode.h"
#include "colorAttrib.h"
#include "colorScaleAttrib.h"
#include "clipPlaneAttrib.h"
#include "fogAttrib.h"
#include "config_pstats.h"

#include <algorithm>
#include <limits.h>

PStatCollector GraphicsStateGuardian::_vertex_buffer_switch_pcollector("Buffer switch:Vertex");
PStatCollector GraphicsStateGuardian::_index_buffer_switch_pcollector("Buffer switch:Index");
PStatCollector GraphicsStateGuardian::_shader_buffer_switch_pcollector("Buffer switch:Shader");
PStatCollector GraphicsStateGuardian::_load_vertex_buffer_pcollector("Draw:Transfer data:Vertex buffer");
PStatCollector GraphicsStateGuardian::_load_index_buffer_pcollector("Draw:Transfer data:Index buffer");
PStatCollector GraphicsStateGuardian::_load_shader_buffer_pcollector("Draw:Transfer data:Shader buffer");
PStatCollector GraphicsStateGuardian::_create_vertex_buffer_pcollector("Draw:Transfer data:Create Vertex buffer");
PStatCollector GraphicsStateGuardian::_create_index_buffer_pcollector("Draw:Transfer data:Create Index buffer");
PStatCollector GraphicsStateGuardian::_create_shader_buffer_pcollector("Draw:Transfer data:Create Shader buffer");
PStatCollector GraphicsStateGuardian::_load_texture_pcollector("Draw:Transfer data:Texture");
PStatCollector GraphicsStateGuardian::_data_transferred_pcollector("Data transferred");
PStatCollector GraphicsStateGuardian::_texmgrmem_total_pcollector("Texture manager");
PStatCollector GraphicsStateGuardian::_texmgrmem_resident_pcollector("Texture manager:Resident");
PStatCollector GraphicsStateGuardian::_primitive_batches_pcollector("Primitive batches");
PStatCollector GraphicsStateGuardian::_primitive_batches_tristrip_pcollector("Primitive batches:Triangle strips");
PStatCollector GraphicsStateGuardian::_primitive_batches_trifan_pcollector("Primitive batches:Triangle fans");
PStatCollector GraphicsStateGuardian::_primitive_batches_tri_pcollector("Primitive batches:Triangles");
PStatCollector GraphicsStateGuardian::_primitive_batches_patch_pcollector("Primitive batches:Patches");
PStatCollector GraphicsStateGuardian::_primitive_batches_other_pcollector("Primitive batches:Other");
PStatCollector GraphicsStateGuardian::_vertices_tristrip_pcollector("Vertices:Triangle strips");
PStatCollector GraphicsStateGuardian::_vertices_trifan_pcollector("Vertices:Triangle fans");
PStatCollector GraphicsStateGuardian::_vertices_tri_pcollector("Vertices:Triangles");
PStatCollector GraphicsStateGuardian::_vertices_patch_pcollector("Vertices:Patches");
PStatCollector GraphicsStateGuardian::_vertices_other_pcollector("Vertices:Other");
PStatCollector GraphicsStateGuardian::_state_pcollector("State changes");
PStatCollector GraphicsStateGuardian::_transform_state_pcollector("State changes:Transforms");
PStatCollector GraphicsStateGuardian::_texture_state_pcollector("State changes:Textures");
PStatCollector GraphicsStateGuardian::_draw_primitive_pcollector("Draw:Primitive:Draw");
PStatCollector GraphicsStateGuardian::_draw_set_state_pcollector("Draw:Set State");
PStatCollector GraphicsStateGuardian::_clear_pcollector("Draw:Clear");
PStatCollector GraphicsStateGuardian::_flush_pcollector("Draw:Flush");
PStatCollector GraphicsStateGuardian::_compute_dispatch_pcollector("Draw:Compute dispatch");

PStatCollector GraphicsStateGuardian::_wait_occlusion_pcollector("Wait:Occlusion");
PStatCollector GraphicsStateGuardian::_wait_timer_pcollector("Wait:Timer Queries");
PStatCollector GraphicsStateGuardian::_timer_queries_pcollector("Timer queries");
PStatCollector GraphicsStateGuardian::_command_latency_pcollector("Command latency");

PStatCollector GraphicsStateGuardian::_prepare_pcollector("Draw:Prepare");
PStatCollector GraphicsStateGuardian::_prepare_texture_pcollector("Draw:Prepare:Texture");
PStatCollector GraphicsStateGuardian::_prepare_sampler_pcollector("Draw:Prepare:Sampler");
PStatCollector GraphicsStateGuardian::_prepare_geom_pcollector("Draw:Prepare:Geom");
PStatCollector GraphicsStateGuardian::_prepare_shader_pcollector("Draw:Prepare:Shader");
PStatCollector GraphicsStateGuardian::_prepare_vertex_buffer_pcollector("Draw:Prepare:Vertex buffer");
PStatCollector GraphicsStateGuardian::_prepare_index_buffer_pcollector("Draw:Prepare:Index buffer");
PStatCollector GraphicsStateGuardian::_prepare_shader_buffer_pcollector("Draw:Prepare:Shader buffer");

PStatCollector GraphicsStateGuardian::_draw_set_state_transform_pcollector("Draw:Set State:Transform");
PStatCollector GraphicsStateGuardian::_draw_set_state_alpha_test_pcollector("Draw:Set State:Alpha test");
PStatCollector GraphicsStateGuardian::_draw_set_state_antialias_pcollector("Draw:Set State:Antialias");
PStatCollector GraphicsStateGuardian::_draw_set_state_clip_plane_pcollector("Draw:Set State:Clip plane");
PStatCollector GraphicsStateGuardian::_draw_set_state_color_pcollector("Draw:Set State:Color");
PStatCollector GraphicsStateGuardian::_draw_set_state_cull_face_pcollector("Draw:Set State:Cull face");
PStatCollector GraphicsStateGuardian::_draw_set_state_depth_offset_pcollector("Draw:Set State:Depth offset");
PStatCollector GraphicsStateGuardian::_draw_set_state_depth_test_pcollector("Draw:Set State:Depth test");
PStatCollector GraphicsStateGuardian::_draw_set_state_depth_write_pcollector("Draw:Set State:Depth write");
PStatCollector GraphicsStateGuardian::_draw_set_state_render_mode_pcollector("Draw:Set State:Render mode");
PStatCollector GraphicsStateGuardian::_draw_set_state_rescale_normal_pcollector("Draw:Set State:Rescale normal");
PStatCollector GraphicsStateGuardian::_draw_set_state_shade_model_pcollector("Draw:Set State:Shade model");
PStatCollector GraphicsStateGuardian::_draw_set_state_blending_pcollector("Draw:Set State:Blending");
PStatCollector GraphicsStateGuardian::_draw_set_state_shader_pcollector("Draw:Set State:Shader");
PStatCollector GraphicsStateGuardian::_draw_set_state_shader_parameters_pcollector("Draw:Set State:Shader Parameters");
PStatCollector GraphicsStateGuardian::_draw_set_state_texture_pcollector("Draw:Set State:Texture");
PStatCollector GraphicsStateGuardian::_draw_set_state_tex_matrix_pcollector("Draw:Set State:Tex matrix");
PStatCollector GraphicsStateGuardian::_draw_set_state_tex_gen_pcollector("Draw:Set State:Tex gen");
PStatCollector GraphicsStateGuardian::_draw_set_state_material_pcollector("Draw:Set State:Material");
PStatCollector GraphicsStateGuardian::_draw_set_state_light_pcollector("Draw:Set State:Light");
PStatCollector GraphicsStateGuardian::_draw_set_state_stencil_pcollector("Draw:Set State:Stencil");
PStatCollector GraphicsStateGuardian::_draw_set_state_fog_pcollector("Draw:Set State:Fog");
PStatCollector GraphicsStateGuardian::_draw_set_state_scissor_pcollector("Draw:Set State:Scissor");

PT(TextureStage) GraphicsStateGuardian::_alpha_scale_texture_stage = NULL;

TypeHandle GraphicsStateGuardian::_type_handle;

/**
 *
 */

GraphicsStateGuardian::
GraphicsStateGuardian(CoordinateSystem internal_coordinate_system,
                      GraphicsEngine *engine, GraphicsPipe *pipe) :
  _internal_coordinate_system(internal_coordinate_system),
  _pipe(pipe),
  _engine(engine)
{
  _coordinate_system = CS_invalid;
  _internal_transform = TransformState::make_identity();

  if (_internal_coordinate_system == CS_default) {
    _internal_coordinate_system = get_default_coordinate_system();
  }

  set_coordinate_system(get_default_coordinate_system());

  _data_reader = (GeomVertexDataPipelineReader *)NULL;
  _current_display_region = (DisplayRegion*)NULL;
  _current_stereo_channel = Lens::SC_mono;
  _current_tex_view_offset = 0;
  _current_lens = (Lens *)NULL;
  _projection_mat = TransformState::make_identity();
  _projection_mat_inv = TransformState::make_identity();

  _needs_reset = true;
  _is_valid = false;
  _current_properties = NULL;
  _closing_gsg = false;
  _active = true;
  _prepared_objects = new PreparedGraphicsObjects;
  _stereo_buffer_mask = ~0;
  _incomplete_render = allow_incomplete_render;
  _effective_incomplete_render = false;
  _loader = Loader::get_global_ptr();

  _is_hardware = false;
  _prefers_triangle_strips = false;
  _max_vertices_per_array = INT_MAX;
  _max_vertices_per_primitive = INT_MAX;

  // Initially, we set this to 1 (the default--no multitexturing supported).
  // A derived GSG may set this differently if it supports multitexturing.
  _max_texture_stages = 1;

  // Also initially, we assume there are no limits on texture sizes, and that
  // 3-d and cube-map textures are not supported.
  _max_texture_dimension = -1;
  _max_3d_texture_dimension = 0;
  _max_2d_texture_array_layers = 0;
  _max_cube_map_dimension = 0;
  _max_buffer_texture_size = 0;

  // Assume we don't support these fairly advanced texture combiner modes.
  _supports_texture_combine = false;
  _supports_texture_saved_result = false;
  _supports_texture_dot3 = false;

  _supports_3d_texture = false;
  _supports_2d_texture_array = false;
  _supports_cube_map = false;
  _supports_buffer_texture = false;
  _supports_cube_map_array = false;
  _supports_tex_non_pow2 = false;
  _supports_texture_srgb = false;
  _supports_compressed_texture = false;
  _compressed_texture_formats.clear();
  _compressed_texture_formats.set_bit(Texture::CM_off);

  // Assume no limits on number of lights or clip planes.
  _max_lights = -1;
  _max_clip_planes = -1;

  // Assume no vertex blending capability.
  _max_vertex_transforms = 0;
  _max_vertex_transform_indices = 0;

  _supports_occlusion_query = false;
  _supports_timer_query = false;

#ifdef DO_PSTATS
  _timer_queries_active = false;
  _last_query_frame = 0;
  _last_num_queried = 0;
  // _timer_delta = 0.0;

  _pstats_gpu_thread = -1;
#endif

  // Initially, we set this to false; a GSG that knows it has this property
  // should set it to true.
  _copy_texture_inverted = false;

  // Similarly with these capabilities flags.
  _supports_multisample = false;
  _supports_generate_mipmap = false;
  _supports_depth_texture = false;
  _supports_depth_stencil = false;
  _supports_shadow_filter = false;
  _supports_sampler_objects = false;
  _supports_basic_shaders = false;
  _supports_geometry_shaders = false;
  _supports_tessellation_shaders = false;
  _supports_compute_shaders = false;
  _supports_glsl = false;
  _supports_hlsl = false;

  _supports_stencil = false;
  _supports_stencil_wrap = false;
  _supports_two_sided_stencil = false;
  _supports_geometry_instancing = false;
  _supports_indirect_draw = false;

  // Assume a maximum of 1 render target in absence of MRT.
  _max_color_targets = 1;
  _supports_dual_source_blending = false;

  _supported_geom_rendering = 0;

  // If this is true, then we can apply a color andor color scale by twiddling
  // the material andor ambient light (which could mean enabling lighting even
  // without a LightAttrib).
  _color_scale_via_lighting = color_scale_via_lighting;

  // Similarly for applying a texture to achieve uniform alpha scaling.
  _alpha_scale_via_texture = alpha_scale_via_texture;

  // Few GSG's can do this, since it requires touching each vertex as it is
  // rendered.
  _runtime_color_scale = false;

  // The default is no shader support.
  _auto_detect_shader_model = SM_00;
  _shader_model = SM_00;

  _gamma = 1.0f;
  _texture_quality_override = Texture::QL_default;
}

/**
 *
 */
GraphicsStateGuardian::
~GraphicsStateGuardian() {
  remove_gsg(this);
  GeomMunger::unregister_mungers_for_gsg(this);
}

/**
 * Returns the graphics engine that created this GSG. Since there is normally
 * only one GraphicsEngine object in an application, this is usually the same
 * as the global GraphicsEngine.
 */
GraphicsEngine *GraphicsStateGuardian::
get_engine() const {
  nassertr(_engine != (GraphicsEngine *)NULL, GraphicsEngine::get_global_ptr());
  return _engine;
}

/**
 * Returns true if this particular GSG supports using the multisample bits to
 * provide antialiasing, and also supports M_multisample and
 * M_multisample_mask transparency modes.  If this is not true for a
 * particular GSG, Panda will map the M_multisample modes to M_binary.
 *
 * This method is declared virtual solely so that it can be queried from
 * cullResult.cxx.
 */
bool GraphicsStateGuardian::
get_supports_multisample() const {
  return _supports_multisample;
}

/**
 * Returns the union of Geom::GeomRendering values that this particular GSG
 * can support directly.  If a Geom needs to be rendered that requires some
 * additional properties, the StandardMunger and/or the CullableObject will
 * convert it as needed.
 *
 * This method is declared virtual solely so that it can be queried from
 * cullableObject.cxx.
 */
int GraphicsStateGuardian::
get_supported_geom_rendering() const {
  return _supported_geom_rendering;
}

/**
 * Returns true if this particular GSG supports the specified Cg Shader
 * Profile.
 */
bool GraphicsStateGuardian::
get_supports_cg_profile(const string &name) const {
  return false;
}

/**
 * Changes the coordinate system in effect on this particular gsg.  This is
 * also called the "external" coordinate system, since it is the coordinate
 * system used by the scene graph, external to to GSG.
 *
 * Normally, this will be the default coordinate system, but it might be set
 * differently at runtime.  It will automatically be copied from the current
 * lens's coordinate system as each DisplayRegion is rendered.
 */
void GraphicsStateGuardian::
set_coordinate_system(CoordinateSystem cs) {
  if (cs == CS_default) {
    cs = get_default_coordinate_system();
  }
  if (_coordinate_system == cs) {
    return;
  }
  _coordinate_system = cs;

  // Changing the external coordinate system changes the cs_transform.
  if (_internal_coordinate_system == CS_default ||
      _internal_coordinate_system == _coordinate_system) {
    _cs_transform = TransformState::make_identity();
    _inv_cs_transform = TransformState::make_identity();

  } else {
    _cs_transform =
      TransformState::make_mat
      (LMatrix4::convert_mat(_coordinate_system,
                              _internal_coordinate_system));
    _inv_cs_transform =
      TransformState::make_mat
      (LMatrix4::convert_mat(_internal_coordinate_system,
                              _coordinate_system));
  }
}

/**
 * Returns the coordinate system used internally by the GSG.  This may be the
 * same as the external coordinate system reported by get_coordinate_system(),
 * or it may be something different.
 *
 * In any case, vertices that have been transformed before being handed to the
 * GSG (that is, vertices with a contents value of C_clip_point) will be
 * expected to be in this coordinate system.
 */
CoordinateSystem GraphicsStateGuardian::
get_internal_coordinate_system() const {
  return _internal_coordinate_system;
}

/**
 * Returns the set of texture and geom objects that have been prepared with
 * this GSG (and possibly other GSG's that share objects).
 */
PreparedGraphicsObjects *GraphicsStateGuardian::
get_prepared_objects() {
  return _prepared_objects;
}

/**
 * Set gamma.  Returns true on success.
 */
bool GraphicsStateGuardian::
set_gamma(PN_stdfloat gamma) {
  _gamma = gamma;

  return false;
}

/**
 * Get the current gamma setting.
 */
PN_stdfloat GraphicsStateGuardian::
get_gamma() const {
  return _gamma;
}

/**
 * Restore original gamma setting.
 */
void GraphicsStateGuardian::
restore_gamma() {
}

/**
 * Calls the indicated function on all currently-prepared textures, or until
 * the callback function returns false.
 */
void GraphicsStateGuardian::
traverse_prepared_textures(GraphicsStateGuardian::TextureCallback *func,
                           void *callback_arg) {
  ReMutexHolder holder(_prepared_objects->_lock);
  PreparedGraphicsObjects::Textures::const_iterator ti;
  for (ti = _prepared_objects->_prepared_textures.begin();
       ti != _prepared_objects->_prepared_textures.end();
       ++ti) {
    bool result = (*func)(*ti, callback_arg);
    if (!result) {
      return;
    }
  }
}

#ifndef NDEBUG
/**
 * Sets the "flash texture".  This is a debug feature; when enabled, the
 * specified texture will begin flashing in the scene, helping you to find it
 * visually.
 *
 * The texture also flashes with a color code: blue for mipmap level 0, yellow
 * for mipmap level 1, and red for mipmap level 2 or higher (even for textures
 * that don't have mipmaps).  This gives you an idea of the choice of the
 * texture size.  If it is blue, the texture is being drawn the proper size or
 * magnified; if it is yellow, it is being minified a little bit; and if it
 * red, it is being minified considerably.  If you see a red texture when you
 * are right in front of it, you should consider reducing the size of the
 * texture to avoid wasting texture memory.
 *
 * Not all rendering backends support the flash_texture feature.  Presently,
 * it is only supported by OpenGL.
 */
void GraphicsStateGuardian::
set_flash_texture(Texture *tex) {
  _flash_texture = tex;
}
#endif  // NDEBUG

#ifndef NDEBUG
/**
 * Resets the "flash texture", so that no textures will flash.  See
 * set_flash_texture().
 */
void GraphicsStateGuardian::
clear_flash_texture() {
  _flash_texture = NULL;
}
#endif  // NDEBUG

#ifndef NDEBUG
/**
 * Returns the current "flash texture", if any, or NULL if none.  See
 * set_flash_texture().
 */
Texture *GraphicsStateGuardian::
get_flash_texture() const {
  return _flash_texture;
}
#endif  // NDEBUG

/**
 * Sets the SceneSetup object that indicates the initial camera position, etc.
 * This must be called before traversal begins.  Returns true if the scene is
 * acceptable, false if something's wrong.  This should be called in the draw
 * thread only.
 */
bool GraphicsStateGuardian::
set_scene(SceneSetup *scene_setup) {
  _scene_setup = scene_setup;
  _current_lens = scene_setup->get_lens();
  if (_current_lens == (Lens *)NULL) {
    return false;
  }

  set_coordinate_system(_current_lens->get_coordinate_system());

  _projection_mat = calc_projection_mat(_current_lens);
  if (_projection_mat == 0) {
    return false;
  }
  _projection_mat_inv = _projection_mat->get_inverse();
  return prepare_lens();
}

/**
 * Returns the current SceneSetup object.
 */
SceneSetup *GraphicsStateGuardian::
get_scene() const {
  return _scene_setup;
}

/**
 * Creates whatever structures the GSG requires to represent the texture
 * internally, and returns a newly-allocated TextureContext object with this
 * data.  It is the responsibility of the calling function to later call
 * release_texture() with this same pointer (which will also delete the
 * pointer).
 *
 * This function should not be called directly to prepare a texture.  Instead,
 * call Texture::prepare().
 */
TextureContext *GraphicsStateGuardian::
prepare_texture(Texture *) {
  return (TextureContext *)NULL;
}

/**
 * Ensures that the current Texture data is refreshed onto the GSG.  This
 * means updating the texture properties and/or re-uploading the texture
 * image, if necessary.  This should only be called within the draw thread.
 *
 * If force is true, this function will not return until the texture has been
 * fully uploaded.  If force is false, the function may choose to upload a
 * simple version of the texture instead, if the texture is not fully resident
 * (and if get_incomplete_render() is true).
 */
bool GraphicsStateGuardian::
update_texture(TextureContext *, bool) {
  return true;
}

/**
 * Frees the resources previously allocated via a call to prepare_texture(),
 * including deleting the TextureContext itself, if it is non-NULL.
 */
void GraphicsStateGuardian::
release_texture(TextureContext *) {
}

/**
 * This method should only be called by the GraphicsEngine.  Do not call it
 * directly; call GraphicsEngine::extract_texture_data() instead.
 *
 * This method will be called in the draw thread to download the texture
 * memory's image into its ram_image value.  It returns true on success, false
 * otherwise.
 */
bool GraphicsStateGuardian::
extract_texture_data(Texture *) {
  return false;
}

/**
 * Creates whatever structures the GSG requires to represent the sampler
 * internally, and returns a newly-allocated SamplerContext object with this
 * data.  It is the responsibility of the calling function to later call
 * release_sampler() with this same pointer (which will also delete the
 * pointer).
 *
 * This function should not be called directly to prepare a sampler.  Instead,
 * call Texture::prepare().
 */
SamplerContext *GraphicsStateGuardian::
prepare_sampler(const SamplerState &sampler) {
  return (SamplerContext *)NULL;
}

/**
 * Frees the resources previously allocated via a call to prepare_sampler(),
 * including deleting the SamplerContext itself, if it is non-NULL.
 */
void GraphicsStateGuardian::
release_sampler(SamplerContext *) {
}

/**
 * Prepares the indicated Geom for retained-mode rendering, by creating
 * whatever structures are necessary in the GSG (for instance, vertex
 * buffers). Returns the newly-allocated GeomContext that can be used to
 * render the geom.
 */
GeomContext *GraphicsStateGuardian::
prepare_geom(Geom *) {
  return (GeomContext *)NULL;
}

/**
 * Frees the resources previously allocated via a call to prepare_geom(),
 * including deleting the GeomContext itself, if it is non-NULL.
 *
 * This function should not be called directly to prepare a Geom.  Instead,
 * call Geom::prepare().
 */
void GraphicsStateGuardian::
release_geom(GeomContext *) {
}

/**
 * Compile a vertex/fragment shader body.
 */
ShaderContext *GraphicsStateGuardian::
prepare_shader(Shader *shader) {
  return (ShaderContext *)NULL;
}

/**
 * Releases the resources allocated by prepare_shader
 */
void GraphicsStateGuardian::
release_shader(ShaderContext *sc) {
}

/**
 * Prepares the indicated buffer for retained-mode rendering.
 */
VertexBufferContext *GraphicsStateGuardian::
prepare_vertex_buffer(GeomVertexArrayData *) {
  return (VertexBufferContext *)NULL;
}

/**
 * Frees the resources previously allocated via a call to prepare_data(),
 * including deleting the VertexBufferContext itself, if necessary.
 */
void GraphicsStateGuardian::
release_vertex_buffer(VertexBufferContext *) {
}

/**
 * Prepares the indicated buffer for retained-mode rendering.
 */
IndexBufferContext *GraphicsStateGuardian::
prepare_index_buffer(GeomPrimitive *) {
  return (IndexBufferContext *)NULL;
}

/**
 * Frees the resources previously allocated via a call to prepare_data(),
 * including deleting the IndexBufferContext itself, if necessary.
 */
void GraphicsStateGuardian::
release_index_buffer(IndexBufferContext *) {
}

/**
 * Prepares the indicated buffer for retained-mode rendering.
 */
BufferContext *GraphicsStateGuardian::
prepare_shader_buffer(ShaderBuffer *) {
  return (BufferContext *)NULL;
}

/**
 * Frees the resources previously allocated via a call to prepare_data(),
 * including deleting the BufferContext itself, if necessary.
 */
void GraphicsStateGuardian::
release_shader_buffer(BufferContext *) {
}

/**
 * Begins a new occlusion query.  After this call, you may call
 * begin_draw_primitives() and draw_triangles()/draw_whatever() repeatedly.
 * Eventually, you should call end_occlusion_query() before the end of the
 * frame; that will return a new OcclusionQueryContext object that will tell
 * you how many pixels represented by the bracketed geometry passed the depth
 * test.
 *
 * It is not valid to call begin_occlusion_query() between another
 * begin_occlusion_query() .. end_occlusion_query() sequence.
 */
void GraphicsStateGuardian::
begin_occlusion_query() {
  nassertv(_current_occlusion_query == (OcclusionQueryContext *)NULL);
}

/**
 * Ends a previous call to begin_occlusion_query(). This call returns the
 * OcclusionQueryContext object that will (eventually) report the number of
 * pixels that passed the depth test between the call to
 * begin_occlusion_query() and end_occlusion_query().
 */
PT(OcclusionQueryContext) GraphicsStateGuardian::
end_occlusion_query() {
  nassertr(_current_occlusion_query != (OcclusionQueryContext *)NULL, NULL);
  PT(OcclusionQueryContext) result = _current_occlusion_query;
  _current_occlusion_query = NULL;
  return result;
}

/**
 * Adds a timer query to the command stream, associated with the given PStats
 * collector index.
 */
PT(TimerQueryContext) GraphicsStateGuardian::
issue_timer_query(int pstats_index) {
  return NULL;
}

/**
 * Dispatches a currently bound compute shader using the given work group
 * counts.
 */
void GraphicsStateGuardian::
dispatch_compute(int num_groups_x, int num_groups_y, int num_groups_z) {
  nassertv(false /* Compute shaders not supported by GSG */);
}

/**
 * Looks up or creates a GeomMunger object to munge vertices appropriate to
 * this GSG for the indicated state.
 */
PT(GeomMunger) GraphicsStateGuardian::
get_geom_munger(const RenderState *state, Thread *current_thread) {
  RenderState::Mungers &mungers = state->_mungers;

  if (!mungers.is_empty()) {
    // Before we even look up the map, see if the _last_mi value points to
    // this GSG.  This is likely because we tend to visit the same state
    // multiple times during a frame.  Also, this might well be the only GSG
    // in the world anyway.
    int mi = state->_last_mi;
    if (mi >= 0 && mungers.has_element(mi) && mungers.get_key(mi) == this) {
      PT(GeomMunger) munger = mungers.get_data(mi);
      if (munger->is_registered()) {
        return munger;
      }
    }

    // Nope, we have to look it up in the map.
    mi = mungers.find(this);
    if (mi >= 0) {
      PT(GeomMunger) munger = mungers.get_data(mi);
      if (munger->is_registered()) {
        state->_last_mi = mi;
        return munger;
      } else {
        // This GeomMunger is no longer registered.  Remove it from the map.
        mungers.remove_element(mi);
      }
    }
  }

  // Nothing in the map; create a new entry.
  PT(GeomMunger) munger = make_geom_munger(state, current_thread);
  nassertr(munger != (GeomMunger *)NULL && munger->is_registered(), munger);
  nassertr(munger->is_of_type(StateMunger::get_class_type()), munger);

  state->_last_mi = mungers.store(this, munger);
  return munger;
}

/**
 * Creates a new GeomMunger object to munge vertices appropriate to this GSG
 * for the indicated state.
 */
PT(GeomMunger) GraphicsStateGuardian::
make_geom_munger(const RenderState *state, Thread *current_thread) {
  // The default implementation returns no munger at all, but presumably,
  // every kind of GSG needs some special munging action, so real GSG's will
  // override this to return something more useful.
  return NULL;
}

/**
 * This function will compute the distance to the indicated point, assumed to
 * be in eye coordinates, from the camera plane.  The point is assumed to be
 * in the GSG's internal coordinate system.
 */
PN_stdfloat GraphicsStateGuardian::
compute_distance_to(const LPoint3 &point) const {
  switch (_internal_coordinate_system) {
  case CS_zup_right:
    return point[1];

  case CS_yup_right:
    return -point[2];

  case CS_zup_left:
    return -point[1];

  case CS_yup_left:
    return point[2];

  default:
    gsg_cat.error()
      << "Invalid coordinate system in compute_distance_to: "
      << (int)_internal_coordinate_system << "\n";
    return 0.0f;
  }
}

/**
 * The gsg contains a large number of useful matrices:
 *
 * * the world transform, * the modelview matrix, * the cs_transform, * etc,
 * etc.
 *
 * A shader can request any of these values, and furthermore, it can request
 * that various compositions, inverses, and transposes be performed.  The
 * ShaderMatSpec is a data structure indicating what datum is desired and what
 * conversions to perform.  This routine, fetch_specified_value, is
 * responsible for doing the actual retrieval and conversions.
 *
 * Some values, like the following, aren't matrices:
 *
 * * window size * texture coordinates of card center
 *
 * This routine can fetch these values as well, by shoehorning them into a
 * matrix.  In this way, we avoid the need for a separate routine to fetch
 * these values.
 *
 * The "altered" bits indicate what parts of the state_and_transform have
 * changed since the last time this particular ShaderMatSpec was evaluated.
 * This may allow data to be cached and not reevaluated.
 *
 */
const LMatrix4 *GraphicsStateGuardian::
fetch_specified_value(Shader::ShaderMatSpec &spec, int altered) {
  LVecBase3 v;

  if (altered & spec._dep[0]) {
    const LMatrix4 *t = fetch_specified_part(spec._part[0], spec._arg[0], spec._cache[0], spec._index);
    if (t != &spec._cache[0]) {
      spec._cache[0] = *t;
    }
  }
  if (altered & spec._dep[1]) {
    const LMatrix4 *t = fetch_specified_part(spec._part[1], spec._arg[1], spec._cache[1], spec._index);
    if (t != &spec._cache[1]) {
      spec._cache[1] = *t;
    }
  }

  switch(spec._func) {
  case Shader::SMF_compose:
    spec._value.multiply(spec._cache[0], spec._cache[1]);
    return &spec._value;
  case Shader::SMF_transform_dlight:
    spec._value = spec._cache[0];
    v = spec._cache[1].xform_vec(spec._cache[0].get_row3(2));
    v.normalize();
    spec._value.set_row(2, v);
    v = spec._cache[1].xform_vec(spec._cache[0].get_row3(3));
    v.normalize();
    spec._value.set_row(3, v);
    return &spec._value;
  case Shader::SMF_transform_plight:
    {
      // Careful not to touch the w component, which contains the near value.
      spec._value = spec._cache[0];
      LPoint3 point = spec._cache[1].xform_point(spec._cache[0].get_row3(2));
      spec._value(2, 0) = point[0];
      spec._value(2, 1) = point[1];
      spec._value(2, 2) = point[2];
      return &spec._value;
    }
  case Shader::SMF_transform_slight:
    spec._value = spec._cache[0];
    spec._value.set_row(2, spec._cache[1].xform_point(spec._cache[0].get_row3(2)));
    v = spec._cache[1].xform_vec(spec._cache[0].get_row3(3));
    v.normalize();
    spec._value.set_row(3, v);
    return &spec._value;
  case Shader::SMF_first:
    return &spec._cache[0];
  default:
    // should never get here
    spec._value = LMatrix4::ident_mat();
    return &spec._value;
  }
}

/**
 * See fetch_specified_value
 */
const LMatrix4 *GraphicsStateGuardian::
fetch_specified_part(Shader::ShaderMatInput part, InternalName *name,
                     LMatrix4 &t, int index) {
  switch (part) {
  case Shader::SMO_identity: {
    return &LMatrix4::ident_mat();
  }
  case Shader::SMO_window_size: {
    t = LMatrix4::translate_mat(_current_display_region->get_pixel_width(),
                                 _current_display_region->get_pixel_height(),
                                 0.0);
    return &t;
  }
  case Shader::SMO_pixel_size: {
    t = LMatrix4::translate_mat(_current_display_region->get_pixel_width(),
                                 _current_display_region->get_pixel_height(),
                                 0.0);
    return &t;
  }
  case Shader::SMO_frame_time: {
    PN_stdfloat time = ClockObject::get_global_clock()->get_frame_time();
    t = LMatrix4(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, time, time, time, time);
    return &t;
  }
  case Shader::SMO_frame_delta: {
    PN_stdfloat dt = ClockObject::get_global_clock()->get_dt();
    t = LMatrix4(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, dt, dt, dt, dt);
    return &t;
  }
  case Shader::SMO_texpad_x: {
    Texture *tex = _target_shader->get_shader_input_texture(name);
    nassertr(tex != 0, &LMatrix4::zeros_mat());
    int sx = tex->get_x_size() - tex->get_pad_x_size();
    int sy = tex->get_y_size() - tex->get_pad_y_size();
    int sz = tex->get_z_size() - tex->get_pad_z_size();
    double cx = (sx * 0.5) / tex->get_x_size();
    double cy = (sy * 0.5) / tex->get_y_size();
    double cz = (sz * 0.5) / tex->get_z_size();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,cx,cy,cz,0);
    return &t;
  }
  case Shader::SMO_texpix_x: {
    Texture *tex = _target_shader->get_shader_input_texture(name);
    nassertr(tex != 0, &LMatrix4::zeros_mat());
    double px = 1.0 / tex->get_x_size();
    double py = 1.0 / tex->get_y_size();
    double pz = 1.0 / tex->get_z_size();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,px,py,pz,0);
    return &t;
  }
  case Shader::SMO_attr_material: {
    const MaterialAttrib *target_material = (const MaterialAttrib *)
      _target_rs->get_attrib_def(MaterialAttrib::get_class_slot());
    // Material matrix contains AMBIENT, DIFFUSE, EMISSION, SPECULAR+SHININESS
    if (target_material->is_off()) {
      t = LMatrix4(1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0);
      return &t;
    }
    Material *m = target_material->get_material();
    LVecBase4 const &amb = m->get_ambient();
    LVecBase4 const &dif = m->get_diffuse();
    LVecBase4 const &emm = m->get_emission();
    LVecBase4 spc = m->get_specular();
    spc[3] = m->get_shininess();
    t = LMatrix4(amb[0],amb[1],amb[2],amb[3],
                  dif[0],dif[1],dif[2],dif[3],
                  emm[0],emm[1],emm[2],emm[3],
                  spc[0],spc[1],spc[2],spc[3]);
    return &t;
  }
  case Shader::SMO_attr_material2: {
    const MaterialAttrib *target_material = (const MaterialAttrib *)
      _target_rs->get_attrib_def(MaterialAttrib::get_class_slot());
    if (target_material->is_off()) {
      t = LMatrix4(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1);
      return &t;
    }
    Material *m = target_material->get_material();
    t.set_row(0, m->get_base_color());
    t.set_row(3, LVecBase4(m->get_metallic(), m->get_refractive_index(), 0, m->get_roughness()));
    return &t;
  }
  case Shader::SMO_attr_color: {
    const ColorAttrib *target_color = (const ColorAttrib *)
      _target_rs->get_attrib_def(ColorAttrib::get_class_slot());
    if (target_color->get_color_type() != ColorAttrib::T_flat) {
      return &LMatrix4::ones_mat();
    }
    LVecBase4 c = target_color->get_color();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,c[0],c[1],c[2],c[3]);
    return &t;
  }
  case Shader::SMO_attr_colorscale: {
    const ColorScaleAttrib *target_color = (const ColorScaleAttrib *)
      _target_rs->get_attrib_def(ColorScaleAttrib::get_class_slot());
    if (target_color->is_identity()) {
      return &LMatrix4::ones_mat();
    }
    LVecBase4 cs = target_color->get_scale();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,cs[0],cs[1],cs[2],cs[3]);
    return &t;
  }
  case Shader::SMO_attr_fog: {
    const FogAttrib *target_fog = (const FogAttrib *)
      _target_rs->get_attrib_def(FogAttrib::get_class_slot());
    Fog *fog = target_fog->get_fog();
    if (fog == (Fog*) NULL) {
      return &LMatrix4::ones_mat();
    }
    PN_stdfloat start, end;
    fog->get_linear_range(start, end);
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,fog->get_exp_density(),start,end,1.0f/(end-start));
    return &t;
  }
  case Shader::SMO_attr_fogcolor: {
    const FogAttrib *target_fog = (const FogAttrib *)
      _target_rs->get_attrib_def(FogAttrib::get_class_slot());
    Fog *fog = target_fog->get_fog();
    if (fog == (Fog*) NULL) {
      return &LMatrix4::ones_mat();
    }
    LVecBase4 c = fog->get_color();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,c[0],c[1],c[2],c[3]);
    return &t;
  }
  case Shader::SMO_alight_x: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::zeros_mat());
    AmbientLight *lt;
    DCAST_INTO_R(lt, np.node(), &LMatrix4::zeros_mat());
    LColor const &c = lt->get_color();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,c[0],c[1],c[2],c[3]);
    return &t;
  }
  case Shader::SMO_satten_x: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ones_mat());
    Spotlight *lt;
    DCAST_INTO_R(lt, np.node(), &LMatrix4::ones_mat());
    LVecBase3 const &a = lt->get_attenuation();
    PN_stdfloat x = lt->get_exponent();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,a[0],a[1],a[2],x);
    return &t;
  }
  case Shader::SMO_dlight_x: {
    // The dlight matrix contains COLOR, SPECULAR, DIRECTION, PSEUDOHALFANGLE
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::zeros_mat());
    DirectionalLight *lt;
    DCAST_INTO_R(lt, np.node(), &LMatrix4::zeros_mat());
    LColor const &c = lt->get_color();
    LColor const &s = lt->get_specular_color();
    t = np.get_net_transform()->get_mat() *
      get_scene()->get_world_transform()->get_mat();
    LVecBase3 d = -(t.xform_vec(lt->get_direction()));
    d.normalize();
    LVecBase3 h = d + LVecBase3(0,-1,0);
    h.normalize();
    t = LMatrix4(c[0],c[1],c[2],c[3],s[0],s[1],s[2],c[3],d[0],d[1],d[2],0,h[0],h[1],h[2],0);
    return &t;
  }
  case Shader::SMO_plight_x: {
    // The plight matrix contains COLOR, SPECULAR, POINT, ATTENUATION
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ones_mat());
    PointLight *lt;
    DCAST_INTO_R(lt, np.node(), &LMatrix4::zeros_mat());
    LColor const &c = lt->get_color();
    LColor const &s = lt->get_specular_color();
    t = np.get_net_transform()->get_mat() *
      get_scene()->get_world_transform()->get_mat();
    LVecBase3 p = (t.xform_point(lt->get_point()));
    LVecBase3 a = lt->get_attenuation();
    PN_stdfloat lnear = lt->get_lens(0)->get_near();
    PN_stdfloat lfar = lt->get_lens(0)->get_far();
    t = LMatrix4(c[0],c[1],c[2],c[3],s[0],s[1],s[2],s[3],p[0],p[1],p[2],lnear,a[0],a[1],a[2],lfar);
    return &t;
  }
  case Shader::SMO_slight_x: {
    // The slight matrix contains COLOR, SPECULAR, POINT, DIRECTION
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::zeros_mat());
    Spotlight *lt;
    DCAST_INTO_R(lt, np.node(), &LMatrix4::zeros_mat());
    Lens *lens = lt->get_lens();
    nassertr(lens != (Lens *)NULL, &LMatrix4::zeros_mat());
    LColor const &c = lt->get_color();
    LColor const &s = lt->get_specular_color();
    PN_stdfloat cutoff = ccos(deg_2_rad(lens->get_hfov() * 0.5f));
    t = np.get_net_transform()->get_mat() *
      get_scene()->get_world_transform()->get_mat();
    LVecBase3 p = t.xform_point(lens->get_nodal_point());
    LVecBase3 d = -(t.xform_vec(lens->get_view_vector()));
    t = LMatrix4(c[0],c[1],c[2],c[3],s[0],s[1],s[2],s[3],p[0],p[1],p[2],0,d[0],d[1],d[2],cutoff);
    return &t;
  }
  case Shader::SMO_light_ambient: {
    LColor cur_ambient_light(0.0f, 0.0f, 0.0f, 0.0f);
    const LightAttrib *target_light = (const LightAttrib *)
      _target_rs->get_attrib_def(LightAttrib::get_class_slot());

    int num_on_lights = target_light->get_num_on_lights();
    if (num_on_lights == 0) {
      // There are no lights at all.  This means, to follow the fixed-
      // function model, we pretend there is an all-white ambient light.
      t.set_row(3, LVecBase4(1, 1, 1, 1));
    } else {
      for (int li = 0; li < num_on_lights; li++) {
        NodePath light = target_light->get_on_light(li);
        nassertr(!light.is_empty(), &LMatrix4::zeros_mat());
        Light *light_obj = light.node()->as_light();
        nassertr(light_obj != (Light *)NULL, &LMatrix4::zeros_mat());

        if (light_obj->get_type() == AmbientLight::get_class_type()) {
          cur_ambient_light += light_obj->get_color();
        }
      }
      t.set_row(3, cur_ambient_light);
    }
    return &t;
  }
  case Shader::SMO_texmat_i: {
    const TexMatrixAttrib *tma;
    const TextureAttrib *ta;
    if (_target_rs->get_attrib(ta) && _target_rs->get_attrib(tma) &&
        index < ta->get_num_on_stages()) {
      return &tma->get_mat(ta->get_on_stage(index));
    } else {
      return &LMatrix4::ident_mat();
    }
  }
  case Shader::SMO_inv_texmat_i: {
    const TexMatrixAttrib *tma;
    const TextureAttrib *ta;
    if (_target_rs->get_attrib(ta) && _target_rs->get_attrib(tma) &&
        index < ta->get_num_on_stages()) {
      t = tma->get_transform(ta->get_on_stage(index))->get_inverse()->get_mat();
      return &t;
    } else {
      return &LMatrix4::ident_mat();
    }
  }
  case Shader::SMO_tex_is_alpha_i: {
    // This is a hack so we can support both F_alpha and other formats in the
    // default shader, to fix font rendering in GLES2
    const TextureAttrib *ta;
    if (_target_rs->get_attrib(ta) &&
        index < ta->get_num_on_stages()) {
      TextureStage *ts = ta->get_on_stage(index);
      PN_stdfloat v = (ta->get_on_texture(ts)->get_format() == Texture::F_alpha);
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,v,v,v,0);
      return &t;
    } else {
      return &LMatrix4::zeros_mat();
    }
  }
  case Shader::SMO_plane_x: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::zeros_mat());
    nassertr(np.node()->is_of_type(PlaneNode::get_class_type()), &LMatrix4::zeros_mat());
    LPlane p = DCAST(PlaneNode, np.node())->get_plane();
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,p[0],p[1],p[2],p[3]);
    return &t;
  }
  case Shader::SMO_clipplane_x: {
    const ClipPlaneAttrib *cpa = DCAST(ClipPlaneAttrib, _target_rs->get_attrib_def(ClipPlaneAttrib::get_class_slot()));
    int planenr = atoi(name->get_name().c_str());
    if (planenr >= cpa->get_num_on_planes()) {
      return &LMatrix4::zeros_mat();
    }
    const NodePath &np = cpa->get_on_plane(planenr);
    nassertr(!np.is_empty(), &LMatrix4::zeros_mat());
    nassertr(np.node()->is_of_type(PlaneNode::get_class_type()), &LMatrix4::zeros_mat());
    LPlane p (DCAST(PlaneNode, np.node())->get_plane());
    p.xform(np.get_net_transform()->get_mat()); // World-space
    t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,p[0],p[1],p[2],p[3]);
    return &t;
  }
  case Shader::SMO_apiview_clipplane_i: {
    const ClipPlaneAttrib *cpa = DCAST(ClipPlaneAttrib, _target_rs->get_attrib_def(ClipPlaneAttrib::get_class_slot()));
    if (index >= cpa->get_num_on_planes()) {
      return &LMatrix4::zeros_mat();
    }

    const NodePath &plane = cpa->get_on_plane(index);
    nassertr(!plane.is_empty(), &LMatrix4::zeros_mat());
    const PlaneNode *plane_node;
    DCAST_INTO_R(plane_node, plane.node(), &LMatrix4::zeros_mat());

    CPT(TransformState) transform =
      get_scene()->get_cs_world_transform()->compose(
        plane.get_transform(_scene_setup->get_scene_root().get_parent()));

    LPlane xformed_plane = plane_node->get_plane() * transform->get_mat();
    t.set_row(3, xformed_plane);
    return &t;
  }
  case Shader::SMO_mat_constant_x: {
    return &_target_shader->get_shader_input_matrix(name, t);
  }
  case Shader::SMO_vec_constant_x: {
    const LVecBase4 &input = _target_shader->get_shader_input_vector(name);
    const PN_stdfloat *data = input.get_data();
    t = LMatrix4(data[0],data[1],data[2],data[3],
                 data[0],data[1],data[2],data[3],
                 data[0],data[1],data[2],data[3],
                 data[0],data[1],data[2],data[3]);
    return &t;
  }
  case Shader::SMO_world_to_view: {
    return &(get_scene()->get_world_transform()->get_mat());
    break;
  }
  case Shader::SMO_view_to_world: {
    return &(get_scene()->get_camera_transform()->get_mat());
  }
  case Shader::SMO_model_to_view: {
    return &(get_external_transform()->get_mat());
  }
  case Shader::SMO_model_to_apiview: {
    return &(get_internal_transform()->get_mat());
  }
  case Shader::SMO_view_to_model: {
    t = get_external_transform()->get_inverse()->get_mat();
    return &t;
  }
  case Shader::SMO_apiview_to_model: {
    t = get_internal_transform()->get_inverse()->get_mat();
    return &t;
  }
  case Shader::SMO_apiview_to_view: {
    return &(_inv_cs_transform->get_mat());
  }
  case Shader::SMO_view_to_apiview: {
    return &(_cs_transform->get_mat());
  }
  case Shader::SMO_clip_to_view: {
    if (_current_lens->get_coordinate_system() == _coordinate_system) {
      return &(_current_lens->get_projection_mat_inv(_current_stereo_channel));
    } else {
      t = _current_lens->get_projection_mat_inv(_current_stereo_channel) *
        LMatrix4::convert_mat(_current_lens->get_coordinate_system(), _coordinate_system);
      return &t;
    }
  }
  case Shader::SMO_view_to_clip: {
    if (_current_lens->get_coordinate_system() == _coordinate_system) {
      return &(_current_lens->get_projection_mat(_current_stereo_channel));
    } else {
      t = LMatrix4::convert_mat(_coordinate_system, _current_lens->get_coordinate_system()) *
        _current_lens->get_projection_mat(_current_stereo_channel);
      return &t;
    }
  }
  case Shader::SMO_apiclip_to_view: {
    t = _projection_mat_inv->get_mat() * _inv_cs_transform->get_mat();
    return &t;
  }
  case Shader::SMO_view_to_apiclip: {
    t = _cs_transform->get_mat() * _projection_mat->get_mat();
    return &t;
  }
  case Shader::SMO_apiclip_to_apiview: {
    return &(_projection_mat_inv->get_mat());
  }
  case Shader::SMO_apiview_to_apiclip: {
    return &(_projection_mat->get_mat());
  }
  case Shader::SMO_view_x_to_view: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    t = np.get_net_transform()->get_mat() *
      get_scene()->get_world_transform()->get_mat();
    return &t;
  }
  case Shader::SMO_view_to_view_x: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    t = get_scene()->get_camera_transform()->get_mat() *
      np.get_net_transform()->get_inverse()->get_mat();
    return &t;
  }
  case Shader::SMO_apiview_x_to_view: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    t = LMatrix4::convert_mat(_internal_coordinate_system, _coordinate_system) *
      np.get_net_transform()->get_mat() *
      get_scene()->get_world_transform()->get_mat();
    return &t;
  }
  case Shader::SMO_view_to_apiview_x: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    t = (get_scene()->get_camera_transform()->get_mat() *
         np.get_net_transform()->get_inverse()->get_mat() *
         LMatrix4::convert_mat(_coordinate_system, _internal_coordinate_system));
    return &t;
  }
  case Shader::SMO_clip_x_to_view: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    nassertr(np.node()->is_of_type(LensNode::get_class_type()), &LMatrix4::ident_mat());
    Lens *lens = DCAST(LensNode, np.node())->get_lens();
    t = lens->get_projection_mat_inv(_current_stereo_channel) *
      LMatrix4::convert_mat(lens->get_coordinate_system(), _coordinate_system) *
      np.get_net_transform()->get_mat() *
      get_scene()->get_world_transform()->get_mat();
    return &t;
  }
  case Shader::SMO_view_to_clip_x: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    nassertr(np.node()->is_of_type(LensNode::get_class_type()), &LMatrix4::ident_mat());
    Lens *lens = DCAST(LensNode, np.node())->get_lens();
    t = get_scene()->get_camera_transform()->get_mat() *
      np.get_net_transform()->get_inverse()->get_mat() *
      LMatrix4::convert_mat(_coordinate_system, lens->get_coordinate_system()) *
      lens->get_projection_mat(_current_stereo_channel);
    return &t;
  }
  case Shader::SMO_apiclip_x_to_view: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    nassertr(np.node()->is_of_type(LensNode::get_class_type()), &LMatrix4::ident_mat());
    Lens *lens = DCAST(LensNode, np.node())->get_lens();
    t = calc_projection_mat(lens)->get_inverse()->get_mat() *
      get_cs_transform_for(lens->get_coordinate_system())->get_inverse()->get_mat() *
      np.get_net_transform()->get_mat() *
      get_scene()->get_world_transform()->get_mat();
    return &t;
  }
  case Shader::SMO_view_to_apiclip_x: {
    const NodePath &np = _target_shader->get_shader_input_nodepath(name);
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());
    nassertr(np.node()->is_of_type(LensNode::get_class_type()), &LMatrix4::ident_mat());
    Lens *lens = DCAST(LensNode, np.node())->get_lens();
    t = get_scene()->get_camera_transform()->get_mat() *
      np.get_net_transform()->get_inverse()->get_mat() *
      get_cs_transform_for(lens->get_coordinate_system())->get_mat() *
      calc_projection_mat(lens)->get_mat();
    return &t;
  }
  case Shader::SMO_mat_constant_x_attrib: {
    if (_target_shader->has_shader_input(name)) {
      // There is an input specifying precisely this whole thing, with dot and
      // all.  Support this, even if only for backward compatibility.
      return &_target_shader->get_shader_input_matrix(name, t);
    }

    const NodePath &np = _target_shader->get_shader_input_nodepath(name->get_parent());
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());

    return fetch_specified_member(np, name->get_basename(), t);
  }
  case Shader::SMO_vec_constant_x_attrib: {
    if (_target_shader->has_shader_input(name)) {
      // There is an input specifying precisely this whole thing, with dot and
      // all.  Support this, even if only for backward compatibility.
      const LVecBase4 &data = _target_shader->get_shader_input_vector(name);
      t = LMatrix4(data[0],data[1],data[2],data[3],
                   data[0],data[1],data[2],data[3],
                   data[0],data[1],data[2],data[3],
                   data[0],data[1],data[2],data[3]);
      return &t;
    }

    const NodePath &np = _target_shader->get_shader_input_nodepath(name->get_parent());
    nassertr(!np.is_empty(), &LMatrix4::ident_mat());

    return fetch_specified_member(np, name->get_basename(), t);
  }
  case Shader::SMO_light_source_i_attrib: {
    const LightAttrib *target_light;
    _target_rs->get_attrib_def(target_light);

    // We want to ignore ambient lights.  To that effect, iterate through the
    // list of lights.  In the future, we will improve this system, by also
    // filtering down to the number of lights specified by the shader.
    int i = 0;

    int num_on_lights = target_light->get_num_on_lights();
    for (int li = 0; li < num_on_lights; li++) {
      NodePath light = target_light->get_on_light(li);
      nassertr(!light.is_empty(), &LMatrix4::ident_mat());
      Light *light_obj = light.node()->as_light();
      nassertr(light_obj != (Light *)NULL, &LMatrix4::ident_mat());

      if (light_obj->get_type() != AmbientLight::get_class_type()) {
        if (i++ == index) {
          return fetch_specified_member(light, name, t);
        }
      }
    }

    // Apply the default OpenGL lights otherwise.
    // Special exception for light 0, which defaults to white.
    if (index == 0) {
      string basename = name->get_basename();
      if (basename == "color" || basename == "diffuse") {
        t.set_row(3, _light_color_scale);
        return &t;
      } else if (basename == "specular") {
        return &LMatrix4::ones_mat();
      }
    }
    return fetch_specified_member(NodePath(), name, t);
  }
  default:
    nassertr(false /*should never get here*/, &LMatrix4::ident_mat());
    return &LMatrix4::ident_mat();
  }
}

/**
 * Given a NodePath passed into a shader input that is a structure, fetches
 * the value for the given member.
 */
const LMatrix4 *GraphicsStateGuardian::
fetch_specified_member(const NodePath &np, CPT_InternalName attrib, LMatrix4 &t) {
  // This system is not ideal.  It will be improved in the future.
  static const CPT_InternalName IN_color("color");
  static const CPT_InternalName IN_ambient("ambient");
  static const CPT_InternalName IN_diffuse("diffuse");
  static const CPT_InternalName IN_specular("specular");
  static const CPT_InternalName IN_position("position");
  static const CPT_InternalName IN_halfVector("halfVector");
  static const CPT_InternalName IN_spotDirection("spotDirection");
  static const CPT_InternalName IN_spotCutoff("spotCutoff");
  static const CPT_InternalName IN_spotCosCutoff("spotCosCutoff");
  static const CPT_InternalName IN_spotExponent("spotExponent");
  static const CPT_InternalName IN_attenuation("attenuation");
  static const CPT_InternalName IN_constantAttenuation("constantAttenuation");
  static const CPT_InternalName IN_linearAttenuation("linearAttenuation");
  static const CPT_InternalName IN_quadraticAttenuation("quadraticAttenuation");
  static const CPT_InternalName IN_shadowMatrix("shadowMatrix");

  PandaNode *node = NULL;
  if (!np.is_empty()) {
    node = np.node();
  }

  if (attrib == IN_color) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::ident_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ident_mat());
    LColor c = light->get_color();
    c.componentwise_mult(_light_color_scale);
    t.set_row(3, c);
    return &t;

  } else if (attrib == IN_ambient) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::ident_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ident_mat());
    if (node->is_of_type(AmbientLight::get_class_type())) {
      LColor c = light->get_color();
      c.componentwise_mult(_light_color_scale);
      t.set_row(3, c);
    } else {
      // Non-ambient lights don't currently have an ambient color in Panda3D.
      t.set_row(3, LColor(0.0f, 0.0f, 0.0f, 1.0f));
    }
    return &t;

  } else if (attrib == IN_diffuse) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::ident_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ones_mat());
    if (node->is_of_type(AmbientLight::get_class_type())) {
      // Ambient light has no diffuse color.
      t.set_row(3, LColor(0.0f, 0.0f, 0.0f, 1.0f));
    } else {
      LColor c = light->get_color();
      c.componentwise_mult(_light_color_scale);
      t.set_row(3, c);
    }
    return &t;

  } else if (attrib == IN_specular) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::ident_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ones_mat());
    t.set_row(3, light->get_specular_color());
    return &t;

  } else if (attrib == IN_position) {
    if (np.is_empty()) {
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0);
      return &t;
    } else if (node->is_of_type(AmbientLight::get_class_type())) {
      // Ambient light has no position.
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0);
      return &t;
    } else if (node->is_of_type(DirectionalLight::get_class_type())) {
      DirectionalLight *light;
      DCAST_INTO_R(light, node, &LMatrix4::ident_mat());

      CPT(TransformState) transform = np.get_transform(_scene_setup->get_scene_root().get_parent());
      LVector3 dir = -(light->get_direction() * transform->get_mat());
      dir *= get_scene()->get_cs_world_transform()->get_mat();
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,dir[0],dir[1],dir[2],0);
      return &t;
    } else {
      LightLensNode *light;
      DCAST_INTO_R(light, node, &LMatrix4::ident_mat());
      Lens *lens = light->get_lens();
      nassertr(lens != (Lens *)NULL, &LMatrix4::ident_mat());

      CPT(TransformState) transform =
        get_scene()->get_cs_world_transform()->compose(
          np.get_transform(_scene_setup->get_scene_root().get_parent()));

      const LMatrix4 &light_mat = transform->get_mat();
      LPoint3 pos = lens->get_nodal_point() * light_mat;
      t = LMatrix4::translate_mat(pos);
      return &t;
    }

  } else if (attrib == IN_halfVector) {
    if (np.is_empty()) {
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0);
      return &t;
    } else if (node->is_of_type(AmbientLight::get_class_type())) {
      // Ambient light has no half-vector.
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0);
      return &t;
    } else if (node->is_of_type(DirectionalLight::get_class_type())) {
      DirectionalLight *light;
      DCAST_INTO_R(light, node, &LMatrix4::ident_mat());

      CPT(TransformState) transform = np.get_transform(_scene_setup->get_scene_root().get_parent());
      LVector3 dir = -(light->get_direction() * transform->get_mat());
      dir *= get_scene()->get_cs_world_transform()->get_mat();
      dir.normalize();
      dir += LVector3(0, 0, 1);
      dir.normalize();
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,dir[0],dir[1],dir[2],1);
      return &t;
    } else {
      LightLensNode *light;
      DCAST_INTO_R(light, node, &LMatrix4::ident_mat());
      Lens *lens = light->get_lens();
      nassertr(lens != (Lens *)NULL, &LMatrix4::ident_mat());

      CPT(TransformState) transform =
        get_scene()->get_cs_world_transform()->compose(
          np.get_transform(_scene_setup->get_scene_root().get_parent()));

      const LMatrix4 &light_mat = transform->get_mat();
      LPoint3 pos = lens->get_nodal_point() * light_mat;
      pos.normalize();
      pos += LVector3(0, 0, 1);
      pos.normalize();
      t = LMatrix4(0,0,0,0,0,0,0,0,0,0,0,0,pos[0],pos[1],pos[2],1);
      return &t;
    }

  } else if (attrib == IN_spotDirection) {
    if (node == (PandaNode *)NULL) {
      t.set_row(3, LVector3(0.0f, 0.0f, -1.0f));
      return &t;
    } else if (node->is_of_type(AmbientLight::get_class_type())) {
      // Ambient light has no spot direction.
      t.set_row(3, LVector3(0.0f, 0.0f, 0.0f));
      return &t;
    } else {
      LightLensNode *light;
      DCAST_INTO_R(light, node, &LMatrix4::ident_mat());
      Lens *lens = light->get_lens();
      nassertr(lens != (Lens *)NULL, &LMatrix4::ident_mat());

      CPT(TransformState) transform =
        get_scene()->get_cs_world_transform()->compose(
          np.get_transform(_scene_setup->get_scene_root().get_parent()));

      const LMatrix4 &light_mat = transform->get_mat();
      LVector3 dir = lens->get_view_vector() * light_mat;
      t.set_row(3, dir);
      return &t;
    }

  } else if (attrib == IN_spotCutoff) {
    if (node != (PandaNode *)NULL &&
        node->is_of_type(Spotlight::get_class_type())) {
      LightLensNode *light;
      DCAST_INTO_R(light, node, &LMatrix4::ident_mat());
      Lens *lens = light->get_lens();
      nassertr(lens != (Lens *)NULL, &LMatrix4::ident_mat());

      float cutoff = lens->get_hfov() * 0.5f;
      t.set_row(3, LVecBase4(cutoff));
      return &t;
    } else {
      // Other lights have no cut-off.
      t.set_row(3, LVecBase4(180));
      return &t;
    }

  } else if (attrib == IN_spotCosCutoff) {
    if (node != (PandaNode *)NULL &&
        node->is_of_type(Spotlight::get_class_type())) {
      LightLensNode *light;
      DCAST_INTO_R(light, node, &LMatrix4::ident_mat());
      Lens *lens = light->get_lens();
      nassertr(lens != (Lens *)NULL, &LMatrix4::ident_mat());

      float cutoff = lens->get_hfov() * 0.5f;
      t.set_row(3, LVecBase4(ccos(deg_2_rad(cutoff))));
      return &t;
    } else {
      // Other lights have no cut-off.
      t.set_row(3, LVecBase4(-1));
      return &t;
    }

  } else if (attrib == IN_spotExponent) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::zeros_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ident_mat());

    t.set_row(3, LVecBase4(light->get_exponent()));
    return &t;

  } else if (attrib == IN_attenuation) {
    if (node != (PandaNode *)NULL) {
      Light *light = node->as_light();
      nassertr(light != (Light *)NULL, &LMatrix4::ones_mat());

      t.set_row(3, LVecBase4(light->get_attenuation(), 0));
    } else {
      t.set_row(3, LVecBase4(1, 0, 0, 0));
    }
    return &t;

  } else if (attrib == IN_constantAttenuation) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::ones_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ones_mat());

    t.set_row(3, LVecBase4(light->get_attenuation()[0]));
    return &t;

  } else if (attrib == IN_linearAttenuation) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::zeros_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ident_mat());

    t.set_row(3, LVecBase4(light->get_attenuation()[1]));
    return &t;

  } else if (attrib == IN_quadraticAttenuation) {
    if (node == (PandaNode *)NULL) {
      return &LMatrix4::zeros_mat();
    }
    Light *light = node->as_light();
    nassertr(light != (Light *)NULL, &LMatrix4::ident_mat());

    t.set_row(3, LVecBase4(light->get_attenuation()[2]));
    return &t;

  } else if (attrib == IN_shadowMatrix) {
    static const LMatrix4 biasmat(0.5f, 0.0f, 0.0f, 0.0f,
                                  0.0f, 0.5f, 0.0f, 0.0f,
                                  0.0f, 0.0f, 0.5f, 0.0f,
                                  0.5f, 0.5f, 0.5f, 1.0f);

    if (node == (PandaNode *)NULL) {
      return &biasmat;
    }

    LensNode *lnode;
    DCAST_INTO_R(lnode, node, &LMatrix4::ident_mat());
    Lens *lens = lnode->get_lens();

    t = get_external_transform()->get_mat() *
      get_scene()->get_camera_transform()->get_mat() *
      np.get_net_transform()->get_inverse()->get_mat() *
      LMatrix4::convert_mat(_coordinate_system, lens->get_coordinate_system());

    if (!node->is_of_type(PointLight::get_class_type())) {
      t *= lens->get_projection_mat() * biasmat;
    }
    return &t;

  } else {
    display_cat.error()
      << "Shader input requests invalid attribute " << *attrib
      << " from node " << np << "\n";
    return &LMatrix4::ident_mat();
  }
}

/**
 * Like fetch_specified_value, but for texture inputs.
 */
PT(Texture) GraphicsStateGuardian::
fetch_specified_texture(Shader::ShaderTexSpec &spec, SamplerState &sampler,
                        int &view) {
  switch (spec._part) {
  case Shader::STO_named_input:
    // Named texture input.
    if (!_target_shader->has_shader_input(spec._name)) {
      // Is this a member of something, like a node?
      const InternalName *parent = spec._name->get_parent();
      if (parent != InternalName::get_root() &&
          _target_shader->has_shader_input(parent)) {

        // Yes, grab the node.
        const string &basename = spec._name->get_basename();
        NodePath np = _target_shader->get_shader_input_nodepath(parent);

        if (basename == "shadowMap") {
          PT(Texture) tex = get_shadow_map(np);
          if (tex != (Texture *)NULL) {
            sampler = tex->get_default_sampler();
          }
          return tex;

        } else {
          if (spec._stage == 0) {
            display_cat.error()
              << "Shader input " << *parent
              << " has no member named " << basename << ".\n";
            spec._stage = -1;
          }
        }
      } else {
        // This used to be legal for some reason, so don't trigger the assert.
        // Prevent flood, though, so abuse the _stage flag to indicate whether
        // we've already errored about this.
        if (spec._stage == 0) {
          display_cat.error()
            << "Shader input " << *spec._name << " is not present.\n";
          spec._stage = -1;
        }
      }
    } else {
      // Just a regular texture input.
      return _target_shader->get_shader_input_texture(spec._name, &sampler);
    }
    break;

  case Shader::STO_stage_i:
    {
      // We get the TextureAttrib directly from the _target_rs, not the
      // filtered TextureAttrib in _target_texture.
      const TextureAttrib *texattrib;
      _target_rs->get_attrib_def(texattrib);

      if (spec._stage < texattrib->get_num_on_stages()) {
        TextureStage *stage = texattrib->get_on_stage(spec._stage);
        sampler = texattrib->get_on_sampler(stage);
        view += stage->get_tex_view_offset();
        return texattrib->get_on_texture(stage);
      }
    }
    break;

  case Shader::STO_light_i_shadow_map:
    {
      const LightAttrib *target_light;
      _target_rs->get_attrib_def(target_light);

      // We want to ignore ambient lights.  To that effect, iterate through
      // the list of lights.  In the future, we will improve this system, by
      // also filtering down to the number of lights specified by the shader.
      int i = 0;

      int num_on_lights = target_light->get_num_on_lights();
      for (int li = 0; li < num_on_lights; li++) {
        NodePath light = target_light->get_on_light(li);
        nassertr(!light.is_empty(), NULL);
        Light *light_obj = light.node()->as_light();
        nassertr(light_obj != (Light *)NULL, NULL);

        if (light_obj->get_type() != AmbientLight::get_class_type()) {
          if (i++ == spec._stage) {
            PT(Texture) tex = get_shadow_map(light);
            if (tex != (Texture *)NULL) {
              sampler = tex->get_default_sampler();
            }
            return tex;
          }
        }
      }
    }
    break;

  default:
    nassertr(false, NULL);
    break;
  }

  return NULL;
}

/**
 * Return a pointer to struct ShaderPtrData
 */
const Shader::ShaderPtrData *GraphicsStateGuardian::
fetch_ptr_parameter(const Shader::ShaderPtrSpec& spec) {
  return (_target_shader->get_shader_input_ptr(spec._arg));
}

/**
 * Makes the specified DisplayRegion current.  All future drawing and clear
 * operations will be constrained within the given DisplayRegion.
 */
void GraphicsStateGuardian::
prepare_display_region(DisplayRegionPipelineReader *dr) {
  _current_display_region = dr->get_object();
  _current_stereo_channel = dr->get_stereo_channel();
  _current_tex_view_offset = dr->get_tex_view_offset();
  _effective_incomplete_render = _incomplete_render && _current_display_region->get_incomplete_render();

  _stereo_buffer_mask = ~0;

  Lens::StereoChannel output_channel = dr->get_stereo_channel();
  if (dr->get_window()->get_swap_eyes()) {
    // Reverse the output channel.
    switch (output_channel) {
    case Lens::SC_left:
      output_channel = Lens::SC_right;
      break;

    case Lens::SC_right:
      output_channel = Lens::SC_left;
      break;

    default:
      break;
    }
  }

  switch (output_channel) {
  case Lens::SC_left:
    _color_write_mask = dr->get_window()->get_left_eye_color_mask();
    if (_current_properties->is_stereo()) {
      _stereo_buffer_mask = ~RenderBuffer::T_right;
    }
    break;

  case Lens::SC_right:
    _color_write_mask = dr->get_window()->get_right_eye_color_mask();
    if (_current_properties->is_stereo()) {
      _stereo_buffer_mask = ~RenderBuffer::T_left;
    }
    break;

  case Lens::SC_mono:
  case Lens::SC_stereo:
    _color_write_mask = ColorWriteAttrib::C_all;
  }
}

/**
 * Resets any non-standard graphics state that might give a callback apoplexy.
 * Some drivers require that the graphics state be restored to neutral before
 * performing certain operations.  In OpenGL, for instance, this closes any
 * open vertex buffers.
 */
void GraphicsStateGuardian::
clear_before_callback() {
}

/**
 * Forgets the current graphics state and current transform, so that the next
 * call to set_state_and_transform() will have to reload everything.  This is
 * a good thing to call when you are no longer sure what the graphics state
 * is.  This should only be called from the draw thread.
 */
void GraphicsStateGuardian::
clear_state_and_transform() {
  // Re-issue the modelview and projection transforms.
  reissue_transforms();

  // Now clear the state flags to unknown.
  _state_rs = RenderState::make_empty();
  _state_mask.clear();
}

/**
 * This is simply a transparent call to GraphicsEngine::remove_window().  It
 * exists primary to support removing a window from that compiles before the
 * display module, and therefore has no knowledge of a GraphicsEngine object.
 */
void GraphicsStateGuardian::
remove_window(GraphicsOutputBase *window) {
  nassertv(_engine != (GraphicsEngine *)NULL);
  GraphicsOutput *win;
  DCAST_INTO_V(win, window);
  _engine->remove_window(win);
}

/**
 * Makes the current lens (whichever lens was most recently specified with
 * set_scene()) active, so that it will transform future rendered geometry.
 * Normally this is only called from the draw process, and usually it is
 * called by set_scene().
 *
 * The return value is true if the lens is acceptable, false if it is not.
 */
bool GraphicsStateGuardian::
prepare_lens() {
  return false;
}

/**
 * Given a lens, this function calculates the appropriate projection matrix
 * for this gsg.  The result depends on the peculiarities of the rendering
 * API.
 */
CPT(TransformState) GraphicsStateGuardian::
calc_projection_mat(const Lens *lens) {
  if (lens == (Lens *)NULL) {
    return NULL;
  }

  if (!lens->is_linear()) {
    return NULL;
  }

  return TransformState::make_identity();
}

/**
 * Called before each frame is rendered, to allow the GSG a chance to do any
 * internal cleanup before beginning the frame.
 *
 * The return value is true if successful (in which case the frame will be
 * drawn and end_frame() will be called later), or false if unsuccessful (in
 * which case nothing will be drawn and end_frame() will not be called).
 */
bool GraphicsStateGuardian::
begin_frame(Thread *current_thread) {
  _prepared_objects->begin_frame(this, current_thread);

  // We should reset the state to the default at the beginning of every frame.
  // Although this will incur additional overhead, particularly in a simple
  // scene, it helps ensure that states that have changed properties since
  // last time without changing attribute pointers--like textures, lighting,
  // or fog--will still be accurately updated.
  _state_rs = RenderState::make_empty();
  _state_mask.clear();

#ifdef DO_PSTATS
  // We have to do this here instead of in GraphicsEngine because we need a
  // current context to issue timer queries.
  int frame = ClockObject::get_global_clock()->get_frame_count();
  if (_last_query_frame < frame) {
    _last_query_frame = frame;
    _timer_queries_pcollector.clear_level();

    // Now is a good time to flush previous frame's queries.  We may not
    // actually have all of the previous frame's results in yet, but that's
    // okay; the GPU data is allowed to lag a few frames behind.
    flush_timer_queries();

    if (_timer_queries_active) {
      // Issue a stop and start event for collector 0, marking the beginning
      // of the new frame.
      issue_timer_query(0x8000);
      issue_timer_query(0x0000);
    }
  }
#endif

  return !_needs_reset;
}

/**
 * Called between begin_frame() and end_frame() to mark the beginning of
 * drawing commands for a "scene" (usually a particular DisplayRegion) within
 * a frame.  All 3-D drawing commands, except the clear operation, must be
 * enclosed within begin_scene() .. end_scene(). This must be called in the
 * draw thread.
 *
 * The return value is true if successful (in which case the scene will be
 * drawn and end_scene() will be called later), or false if unsuccessful (in
 * which case nothing will be drawn and end_scene() will not be called).
 */
bool GraphicsStateGuardian::
begin_scene() {
  return true;
}

/**
 * Called between begin_frame() and end_frame() to mark the end of drawing
 * commands for a "scene" (usually a particular DisplayRegion) within a frame.
 * All 3-D drawing commands, except the clear operation, must be enclosed
 * within begin_scene() .. end_scene().
 */
void GraphicsStateGuardian::
end_scene() {
  // We should clear this pointer now, so that we don't keep unneeded
  // reference counts dangling.  We keep around a "null" scene setup object
  // instead of using a null pointer to avoid special-case code in
  // set_state_and_transform.
  _scene_setup = _scene_null;

  // Undo any lighting we had enabled last scene, to force the lights to be
  // reissued, in case their parameters or positions have changed between
  // scenes.
  int i;
  for (i = 0; i < _num_lights_enabled; ++i) {
    enable_light(i, false);
  }
  _num_lights_enabled = 0;

  // Ditto for the clipping planes.
  for (i = 0; i < _num_clip_planes_enabled; ++i) {
    enable_clip_plane(i, false);
  }
  _num_clip_planes_enabled = 0;

  // Put the state into the 'unknown' state, forcing a reload.
  _state_rs = RenderState::make_empty();
  _state_mask.clear();
}

/**
 * Called after each frame is rendered, to allow the GSG a chance to do any
 * internal cleanup after rendering the frame, and before the window flips.
 */
void GraphicsStateGuardian::
end_frame(Thread *current_thread) {
  _prepared_objects->end_frame(current_thread);

  // Flush any PStatCollectors.
  _data_transferred_pcollector.flush_level();

  _primitive_batches_pcollector.flush_level();
  _primitive_batches_tristrip_pcollector.flush_level();
  _primitive_batches_trifan_pcollector.flush_level();
  _primitive_batches_tri_pcollector.flush_level();
  _primitive_batches_patch_pcollector.flush_level();
  _primitive_batches_other_pcollector.flush_level();
  _vertices_tristrip_pcollector.flush_level();
  _vertices_trifan_pcollector.flush_level();
  _vertices_tri_pcollector.flush_level();
  _vertices_patch_pcollector.flush_level();
  _vertices_other_pcollector.flush_level();

  _state_pcollector.flush_level();
  _texture_state_pcollector.flush_level();
  _transform_state_pcollector.flush_level();
  _draw_primitive_pcollector.flush_level();

  // Evict any textures andor vbuffers that exceed our texture memory.
  _prepared_objects->_graphics_memory_lru.begin_epoch();
}

/**
 * Called by the graphics engine on the draw thread to check the status of the
 * running timer queries and submit their results to the PStats server.
 */
void GraphicsStateGuardian::
flush_timer_queries() {
#ifdef DO_PSTATS
  // This uses the lower-level PStats interfaces for now because of all the
  // unnecessary overhead that would otherwise be incurred when adding such a
  // large amount of data at once.

  PStatClient *client = PStatClient::get_global_pstats();

  if (!client->client_is_connected()) {
    _timer_queries_active = false;
    return;
  }

  if (!_timer_queries_active) {
    if (pstats_gpu_timing && _supports_timer_query) {
      // Check if timer queries should be enabled.
      _timer_queries_active = true;
    } else {
      return;
    }
  }

  // Currently, we use one thread per GSG, for convenience.  In the future, we
  // may want to try and use one thread per graphics card.
  if (_pstats_gpu_thread == -1) {
    _pstats_gpu_thread = client->make_gpu_thread(get_driver_renderer()).get_index();
  }
  PStatThread gpu_thread(client, _pstats_gpu_thread);

  // Get the results of all the timer queries.
  int first = 0;
  if (!_pending_timer_queries.empty()) {
    int count = _pending_timer_queries.size();
    if (count == 0) {
      return;
    }

    PStatGPUTimer timer(this, _wait_timer_pcollector);

    if (_last_num_queried > 0) {
      // We know how many queries were available last frame, and this usually
      // stays fairly constant, so use this as a starting point.
      int i = min(_last_num_queried, count) - 1;

      if (_pending_timer_queries[i]->is_answer_ready()) {
        first = count;
        while (i < count - 1) {
          if (!_pending_timer_queries[++i]->is_answer_ready()) {
            first = i;
            break;
          }
        }
      } else {
        first = 0;
        while (i > 0) {
          if (_pending_timer_queries[--i]->is_answer_ready()) {
            first = i + 1;
            break;
          }
        }
      }
    } else {
      // We figure out which tasks the GPU has already finished by doing a
      // binary search for the first query that does not have an answer ready.
      // We know then that everything before that must be ready.
      while (count > 0) {
        int step = count / 2;
        int i = first + step;
        if (_pending_timer_queries[i]->is_answer_ready()) {
          first += step + 1;
          count -= step + 1;
        } else {
          count = step;
        }
      }
    }

    if (first <= 0) {
      return;
    }

    _last_num_queried = first;

    for (int i = 0; i < first; ++i) {
      CPT(TimerQueryContext) query = _pending_timer_queries[i];

      double time_data = query->get_timestamp(); //  + _timer_delta;

      if (query->_pstats_index == _command_latency_pcollector.get_index()) {
        // Special case for the latency pcollector.
        PStatCollectorDef *cdef;
        cdef = client->get_collector_ptr(query->_pstats_index)->get_def(client, query->_pstats_index);
        _pstats_gpu_data.add_level(query->_pstats_index, time_data * cdef->_factor);

      } else if (query->_pstats_index & 0x8000) {
        _pstats_gpu_data.add_stop(query->_pstats_index & 0x7fff, time_data);

      } else {
        _pstats_gpu_data.add_start(query->_pstats_index & 0x7fff, time_data);
      }

      // We found an end-frame marker (a stop event for collector 0). This
      // means that the GPU actually caught up with that frame, and we can
      // flush the GPU thread's frame data to the pstats server.
      if (query->_pstats_index == 0x8000) {
        gpu_thread.add_frame(_pstats_gpu_data);
        _pstats_gpu_data.clear();
      }
    }
  }

  if (first > 0) {
    // Do this out of the scope of _wait_timer_pcollector.
    _pending_timer_queries.erase(
      _pending_timer_queries.begin(),
      _pending_timer_queries.begin() + first
    );
    _timer_queries_pcollector.add_level_now(first);
  }
#endif
}

/**
 * Returns true if this GSG can implement decals using a DepthOffsetAttrib, or
 * false if that is unreliable and the three-step rendering process should be
 * used instead.
 */
bool GraphicsStateGuardian::
depth_offset_decals() {
  return true;
}

/**
 * Called during draw to begin a three-step rendering phase to draw decals.
 * The first step, begin_decal_base_first(), is called prior to drawing the
 * base geometry.  It should set up whatever internal state is appropriate, as
 * well as returning a RenderState object that should be applied to the base
 * geometry for rendering.
 */
CPT(RenderState) GraphicsStateGuardian::
begin_decal_base_first() {
  // Turn off writing the depth buffer to render the base geometry.
  static CPT(RenderState) decal_base_first;
  if (decal_base_first == (const RenderState *)NULL) {
    decal_base_first = RenderState::make
      (DepthWriteAttrib::make(DepthWriteAttrib::M_off),
       RenderState::get_max_priority());
  }
  return decal_base_first;
}

/**
 * Called during draw to begin a three-step rendering phase to draw decals.
 * The second step, begin_decal_nested(), is called after drawing the base
 * geometry and prior to drawing any of the nested decal geometry that is to
 * be applied to the base geometry.
 */
CPT(RenderState) GraphicsStateGuardian::
begin_decal_nested() {
  // We should keep the depth buffer off during this operation, so that decals
  // on decals will render properly.
  static CPT(RenderState) decal_nested;
  if (decal_nested == (const RenderState *)NULL) {
    decal_nested = RenderState::make
      (DepthWriteAttrib::make(DepthWriteAttrib::M_off),
       RenderState::get_max_priority());
  }
  return decal_nested;
}

/**
 * Called during draw to begin a three-step rendering phase to draw decals.
 * The third step, begin_decal_base_second(), is called after drawing the base
 * geometry and the nested decal geometry, and prior to drawing the base
 * geometry one more time (if needed).
 *
 * It should return a RenderState object appropriate for rendering the base
 * geometry the second time, or NULL if it is not necessary to re-render the
 * base geometry.
 */
CPT(RenderState) GraphicsStateGuardian::
begin_decal_base_second() {
  // Now let the depth buffer go back on, but turn off writing the color
  // buffer to render the base geometry after the second pass.  Also, turn off
  // texturing since there's no need for it now.
  static CPT(RenderState) decal_base_second;
  if (decal_base_second == (const RenderState *)NULL) {
    decal_base_second = RenderState::make
      (ColorWriteAttrib::make(ColorWriteAttrib::C_off),
       // On reflection, we need to leave texturing on so the alpha test
       // mechanism can work (if it is enabled, e.g.  we are rendering an
       // object with M_dual transparency). TextureAttrib::make_off(),
       RenderState::get_max_priority());
  }
  return decal_base_second;
}

/**
 * Called during draw to clean up after decals are finished.
 */
void GraphicsStateGuardian::
finish_decal() {
  // No need to do anything special here.
}

/**
 * Called before a sequence of draw_primitive() functions are called, this
 * should prepare the vertex data for rendering.  It returns true if the
 * vertices are ok, false to abort this group of primitives.
 */
bool GraphicsStateGuardian::
begin_draw_primitives(const GeomPipelineReader *geom_reader,
                      const GeomMunger *munger,
                      const GeomVertexDataPipelineReader *data_reader,
                      bool force) {
  _munger = munger;
  _data_reader = data_reader;

  // Always draw if we have a shader, since the shader might use a different
  // mechanism for fetching vertex data.
  return _data_reader->has_vertex() || (_target_shader && _target_shader->has_shader());
}

/**
 * Draws a series of disconnected triangles.
 */
bool GraphicsStateGuardian::
draw_triangles(const GeomPrimitivePipelineReader *, bool) {
  return false;
}

/**
 * Draws a series of triangle strips.
 */
bool GraphicsStateGuardian::
draw_tristrips(const GeomPrimitivePipelineReader *, bool) {
  return false;
}

/**
 * Draws a series of triangle fans.
 */
bool GraphicsStateGuardian::
draw_trifans(const GeomPrimitivePipelineReader *, bool) {
  return false;
}

/**
 * Draws a series of "patches", which can only be processed by a tessellation
 * shader.
 */
bool GraphicsStateGuardian::
draw_patches(const GeomPrimitivePipelineReader *, bool) {
  return false;
}

/**
 * Draws a series of disconnected line segments.
 */
bool GraphicsStateGuardian::
draw_lines(const GeomPrimitivePipelineReader *, bool) {
  return false;
}

/**
 * Draws a series of line strips.
 */
bool GraphicsStateGuardian::
draw_linestrips(const GeomPrimitivePipelineReader *, bool) {
  return false;
}

/**
 * Draws a series of disconnected points.
 */
bool GraphicsStateGuardian::
draw_points(const GeomPrimitivePipelineReader *, bool) {
  return false;
}

/**
 * Called after a sequence of draw_primitive() functions are called, this
 * should do whatever cleanup is appropriate.
 */
void GraphicsStateGuardian::
end_draw_primitives() {
  _munger = NULL;
  _data_reader = NULL;
}

/**
 * Resets all internal state as if the gsg were newly created.
 */
void GraphicsStateGuardian::
reset() {
  _needs_reset = false;
  _is_valid = false;

  _state_rs = RenderState::make_empty();
  _target_rs = NULL;
  _state_mask.clear();
  _internal_transform = _cs_transform;
  _scene_null = new SceneSetup;
  _scene_setup = _scene_null;

  _color_write_mask = ColorWriteAttrib::C_all;

  _has_scene_graph_color = false;
  _scene_graph_color.set(1.0f, 1.0f, 1.0f, 1.0f);
  _transform_stale = true;
  _color_blend_involves_color_scale = false;
  _texture_involves_color_scale = false;
  _vertex_colors_enabled = true;
  _lighting_enabled = false;
  _num_lights_enabled = 0;
  _num_clip_planes_enabled = 0;
  _clip_planes_enabled = false;

  _color_scale_enabled = false;
  _current_color_scale.set(1.0f, 1.0f, 1.0f, 1.0f);
  _has_texture_alpha_scale = false;

  _has_material_force_color = false;
  _material_force_color.set(1.0f, 1.0f, 1.0f, 1.0f);
  _light_color_scale.set(1.0f, 1.0f, 1.0f, 1.0f);

  _tex_gen_modifies_mat = false;
  _last_max_stage_index = 0;

  _is_valid = true;
}

/**
 * Simultaneously resets the render state and the transform state.
 *
 * This transform specified is the "internal" net transform, already converted
 * into the GSG's internal coordinate space by composing it to
 * get_cs_transform().  (Previously, this used to be the "external" net
 * transform, with the assumption that that GSG would convert it internally,
 * but that is no longer the case.)
 *
 * Special case: if (state==NULL), then the target state is already stored in
 * _target.
 */
void GraphicsStateGuardian::
set_state_and_transform(const RenderState *state,
                        const TransformState *trans) {
}

/**
 * Clears the framebuffer within the current DisplayRegion, according to the
 * flags indicated by the given DrawableRegion object.
 *
 * This does not set the DisplayRegion first.  You should call
 * prepare_display_region() to specify the region you wish the clear operation
 * to apply to.
 */
void GraphicsStateGuardian::
clear(DrawableRegion *clearable) {
}

/**
 * Returns a RenderBuffer object suitable for operating on the requested set
 * of buffers.  buffer_type is the union of all the desired RenderBuffer::Type
 * values.
 */
RenderBuffer GraphicsStateGuardian::
get_render_buffer(int buffer_type, const FrameBufferProperties &prop) {
  return RenderBuffer(this, buffer_type & prop.get_buffer_mask() & _stereo_buffer_mask);
}

/**
 * Returns what the cs_transform would be set to after a call to
 * set_coordinate_system(cs).  This is another way of saying the cs_transform
 * when rendering the scene for a camera with the indicated coordinate system.
 */
CPT(TransformState) GraphicsStateGuardian::
get_cs_transform_for(CoordinateSystem cs) const {
  if (_coordinate_system == cs) {
    // We've already calculated this.
    return _cs_transform;

  } else if (_internal_coordinate_system == CS_default ||
             _internal_coordinate_system == cs) {
    return TransformState::make_identity();

  } else {
    return TransformState::make_mat
      (LMatrix4::convert_mat(cs, _internal_coordinate_system));
  }
}

/**
 * Returns a transform that converts from the GSG's external coordinate system
 * (as returned by get_coordinate_system()) to its internal coordinate system
 * (as returned by get_internal_coordinate_system()).  This is used for
 * rendering.
 */
CPT(TransformState) GraphicsStateGuardian::
get_cs_transform() const {
  return _cs_transform;
}

/**
 * This is fundametically similar to do_issue_light(), with calls to
 * apply_clip_plane() and enable_clip_planes(), as appropriate.
 */
void GraphicsStateGuardian::
do_issue_clip_plane() {
  int num_enabled = 0;
  int num_on_planes = 0;

  const ClipPlaneAttrib *target_clip_plane = (const ClipPlaneAttrib *)
    _target_rs->get_attrib_def(ClipPlaneAttrib::get_class_slot());

  if (target_clip_plane != (ClipPlaneAttrib *)NULL) {
    CPT(ClipPlaneAttrib) new_plane = target_clip_plane->filter_to_max(_max_clip_planes);

    num_on_planes = new_plane->get_num_on_planes();
    for (int li = 0; li < num_on_planes; li++) {
      NodePath plane = new_plane->get_on_plane(li);
      nassertv(!plane.is_empty());
      PlaneNode *plane_node;
      DCAST_INTO_V(plane_node, plane.node());
      if ((plane_node->get_clip_effect() & PlaneNode::CE_visible) != 0) {
        // Clipping should be enabled before we apply any planes.
        if (!_clip_planes_enabled) {
          enable_clip_planes(true);
          _clip_planes_enabled = true;
        }

        enable_clip_plane(num_enabled, true);
        if (num_enabled == 0) {
          begin_bind_clip_planes();
        }

        bind_clip_plane(plane, num_enabled);
        num_enabled++;
      }
    }
  }

  int i;
  for (i = num_enabled; i < _num_clip_planes_enabled; ++i) {
    enable_clip_plane(i, false);
  }
  _num_clip_planes_enabled = num_enabled;

  // If no planes were set, disable clipping
  if (num_enabled == 0) {
    if (_clip_planes_enabled) {
      enable_clip_planes(false);
      _clip_planes_enabled = false;
    }
  } else {
    end_bind_clip_planes();
  }
}

/**
 * This method is defined in the base class because it is likely that this
 * functionality will be used for all (or at least most) kinds of
 * GraphicsStateGuardians--it's not specific to any one rendering backend.
 *
 * The ColorAttribute just changes the interpretation of the color on the
 * vertices, and fiddles with _vertex_colors_enabled, etc.
 */
void GraphicsStateGuardian::
do_issue_color() {
  const ColorAttrib *target_color = (const ColorAttrib *)
    _target_rs->get_attrib_def(ColorAttrib::get_class_slot());

  switch (target_color->get_color_type()) {
  case ColorAttrib::T_flat:
    // Color attribute flat: it specifies a scene graph color that overrides
    // the vertex color.
    _scene_graph_color = target_color->get_color();
    _has_scene_graph_color = true;
    _vertex_colors_enabled = false;
    break;

  case ColorAttrib::T_off:
    // Color attribute off: it specifies that no scene graph color is in
    // effect, and vertex color is not important either.
    _scene_graph_color.set(1.0f, 1.0f, 1.0f, 1.0f);
    _has_scene_graph_color = false;
    _vertex_colors_enabled = false;
    break;

  case ColorAttrib::T_vertex:
    // Color attribute vertex: it specifies that vertex color should be
    // revealed.
    _scene_graph_color.set(1.0f, 1.0f, 1.0f, 1.0f);
    _has_scene_graph_color = false;
    _vertex_colors_enabled = true;
    break;
  }

  if (_color_scale_via_lighting) {
    _state_mask.clear_bit(LightAttrib::get_class_slot());
    _state_mask.clear_bit(MaterialAttrib::get_class_slot());

    determine_light_color_scale();
  }
}

/**
 *
 */
void GraphicsStateGuardian::
do_issue_color_scale() {
  // If the previous color scale had set a special texture, clear the texture
  // now.
  if (_has_texture_alpha_scale) {
    _state_mask.clear_bit(TextureAttrib::get_class_slot());
  }

  const ColorScaleAttrib *target_color_scale = (const ColorScaleAttrib *)
    _target_rs->get_attrib_def(ColorScaleAttrib::get_class_slot());

  _color_scale_enabled = target_color_scale->has_scale();
  _current_color_scale = target_color_scale->get_scale();
  _has_texture_alpha_scale = false;

  if (_color_blend_involves_color_scale) {
    _state_mask.clear_bit(TransparencyAttrib::get_class_slot());
  }
  if (_texture_involves_color_scale) {
    _state_mask.clear_bit(TextureAttrib::get_class_slot());
  }
  if (_color_scale_via_lighting) {
    _state_mask.clear_bit(LightAttrib::get_class_slot());
    _state_mask.clear_bit(MaterialAttrib::get_class_slot());

    determine_light_color_scale();
  }

  if (_alpha_scale_via_texture && !_has_scene_graph_color &&
      target_color_scale->has_alpha_scale()) {
    // This color scale will set a special texture--so again, clear the
    // texture.
    _state_mask.clear_bit(TextureAttrib::get_class_slot());
    _state_mask.clear_bit(TexMatrixAttrib::get_class_slot());

    _has_texture_alpha_scale = true;
  }
}

/**
 * This implementation of do_issue_light() assumes we have a limited number of
 * hardware lights available.  This function assigns each light to a different
 * hardware light id, trying to keep each light associated with the same id
 * where possible, but reusing id's when necessary.  When it is no longer
 * possible to reuse existing id's (e.g.  all id's are in use), the next
 * sequential id is assigned (if available).
 *
 * It will call apply_light() each time a light is assigned to a particular id
 * for the first time in a given frame, and it will subsequently call
 * enable_light() to enable or disable each light as the frame is rendered, as
 * well as enable_lighting() to enable or disable overall lighting.
 */
void GraphicsStateGuardian::
do_issue_light() {
  // Initialize the current ambient light total and newly enabled light list
  LColor cur_ambient_light(0.0f, 0.0f, 0.0f, 0.0f);
  int i;

  int num_enabled = 0;
  int num_on_lights = 0;

  const LightAttrib *target_light;
  _target_rs->get_attrib_def(target_light);

  if (display_cat.is_spam()) {
    display_cat.spam()
      << "do_issue_light: " << target_light << "\n";
  }
  if (target_light != (LightAttrib *)NULL) {
    CPT(LightAttrib) new_light = target_light->filter_to_max(_max_lights);
    if (display_cat.is_spam()) {
      new_light->write(display_cat.spam(false), 2);
    }

    num_on_lights = new_light->get_num_on_lights();
    for (int li = 0; li < num_on_lights; li++) {
      NodePath light = new_light->get_on_light(li);
      nassertv(!light.is_empty());
      Light *light_obj = light.node()->as_light();
      nassertv(light_obj != (Light *)NULL);

      // Lighting should be enabled before we apply any lights.
      if (!_lighting_enabled) {
        enable_lighting(true);
        _lighting_enabled = true;
      }

      if (light_obj->get_type() == AmbientLight::get_class_type()) {
        // Ambient lights don't require specific light ids; simply add in the
        // ambient contribution to the current total
        cur_ambient_light += light_obj->get_color();

      } else {
        const LColor &color = light_obj->get_color();
        // Don't bother binding the light if it has no color to contribute.
        if (color[0] != 0.0 || color[1] != 0.0 || color[2] != 0.0) {
          enable_light(num_enabled, true);
          if (num_enabled == 0) {
            begin_bind_lights();
          }

          light_obj->bind(this, light, num_enabled);
          num_enabled++;
        }
      }
    }
  }

  for (i = num_enabled; i < _num_lights_enabled; ++i) {
    enable_light(i, false);
  }
  _num_lights_enabled = num_enabled;

  // If no lights were set, disable lighting
  if (num_on_lights == 0) {
    if (_color_scale_via_lighting && (_has_material_force_color || _light_color_scale != LVecBase4(1.0f, 1.0f, 1.0f, 1.0f))) {
      // Unless we need lighting anyway to apply a color or color scale.
      if (!_lighting_enabled) {
        enable_lighting(true);
        _lighting_enabled = true;
      }
      set_ambient_light(LColor(1.0f, 1.0f, 1.0f, 1.0f));

    } else {
      if (_lighting_enabled) {
        enable_lighting(false);
        _lighting_enabled = false;
      }
    }

  } else {
    set_ambient_light(cur_ambient_light);
  }

  if (num_enabled != 0) {
    end_bind_lights();
  }
}

/**
 * Copy the pixels within the indicated display region from the framebuffer
 * into texture memory.
 *
 * If z > -1, it is the cube map index into which to copy.
 */
bool GraphicsStateGuardian::
framebuffer_copy_to_texture(Texture *, int, int, const DisplayRegion *,
                            const RenderBuffer &) {
  return false;
}


/**
 * Copy the pixels within the indicated display region from the framebuffer
 * into system memory, not texture memory.  Returns true on success, false on
 * failure.
 *
 * This completely redefines the ram image of the indicated texture.
 */
bool GraphicsStateGuardian::
framebuffer_copy_to_ram(Texture *, int, int, const DisplayRegion *,
                        const RenderBuffer &) {
  return false;
}

/**
 * Called the first time a particular light has been bound to a given id
 * within a frame, this should set up the associated hardware light with the
 * light's properties.
 */
void GraphicsStateGuardian::
bind_light(PointLight *light_obj, const NodePath &light, int light_id) {
}

/**
 * Called the first time a particular light has been bound to a given id
 * within a frame, this should set up the associated hardware light with the
 * light's properties.
 */
void GraphicsStateGuardian::
bind_light(DirectionalLight *light_obj, const NodePath &light, int light_id) {
}

/**
 * Called the first time a particular light has been bound to a given id
 * within a frame, this should set up the associated hardware light with the
 * light's properties.
 */
void GraphicsStateGuardian::
bind_light(Spotlight *light_obj, const NodePath &light, int light_id) {
}

#ifdef DO_PSTATS
/**
 * Initializes the relevant PStats data at the beginning of the frame.
 */
void GraphicsStateGuardian::
init_frame_pstats() {
  if (PStatClient::is_connected()) {
    _data_transferred_pcollector.clear_level();
    _vertex_buffer_switch_pcollector.clear_level();
    _index_buffer_switch_pcollector.clear_level();

    _primitive_batches_pcollector.clear_level();
    _primitive_batches_tristrip_pcollector.clear_level();
    _primitive_batches_trifan_pcollector.clear_level();
    _primitive_batches_tri_pcollector.clear_level();
    _primitive_batches_patch_pcollector.clear_level();
    _primitive_batches_other_pcollector.clear_level();
    _vertices_tristrip_pcollector.clear_level();
    _vertices_trifan_pcollector.clear_level();
    _vertices_tri_pcollector.clear_level();
    _vertices_patch_pcollector.clear_level();
    _vertices_other_pcollector.clear_level();

    _state_pcollector.clear_level();
    _transform_state_pcollector.clear_level();
    _texture_state_pcollector.clear_level();
  }
}
#endif  // DO_PSTATS


/**
 * Create a gamma table.
 */
void GraphicsStateGuardian::
create_gamma_table (PN_stdfloat gamma, unsigned short *red_table, unsigned short *green_table, unsigned short *blue_table) {
  int i;

  if (gamma <= 0.0) {
    // avoid divide by zero and negative exponents
    gamma = 1.0;
  }

  for (i = 0; i < 256; i++) {
    double g;
    double x;
    PN_stdfloat gamma_correction;

    x = ((double) i / 255.0);
    gamma_correction = 1.0 / gamma;
    x = pow (x, (double) gamma_correction);
    if (x > 1.00) {
      x = 1.0;
    }

    g = x * 65535.0;
    red_table [i] = (int)g;
    green_table [i] = (int)g;
    blue_table [i] = (int)g;
  }
}

/**
 * Called by clear_state_and_transform() to ensure that the current modelview
 * and projection matrices are properly loaded in the graphics state, after a
 * callback might have mucked them up.
 */
void GraphicsStateGuardian::
reissue_transforms() {
}

/**
 * Intended to be overridden by a derived class to enable or disable the use
 * of lighting overall.  This is called by do_issue_light() according to
 * whether any lights are in use or not.
 */
void GraphicsStateGuardian::
enable_lighting(bool enable) {
}

/**
 * Intended to be overridden by a derived class to indicate the color of the
 * ambient light that should be in effect.  This is called by do_issue_light()
 * after all other lights have been enabled or disabled.
 */
void GraphicsStateGuardian::
set_ambient_light(const LColor &color) {
}

/**
 * Intended to be overridden by a derived class to enable the indicated light
 * id.  A specific Light will already have been bound to this id via
 * bind_light().
 */
void GraphicsStateGuardian::
enable_light(int light_id, bool enable) {
}

/**
 * Called immediately before bind_light() is called, this is intended to
 * provide the derived class a hook in which to set up some state (like
 * transform) that might apply to several lights.
 *
 * The sequence is: begin_bind_lights() will be called, then one or more
 * bind_light() calls, then end_bind_lights().
 */
void GraphicsStateGuardian::
begin_bind_lights() {
}

/**
 * Called after before bind_light() has been called one or more times (but
 * before any geometry is issued or additional state is changed), this is
 * intended to clean up any temporary changes to the state that may have been
 * made by begin_bind_lights().
 */
void GraphicsStateGuardian::
end_bind_lights() {
}

/**
 * Intended to be overridden by a derived class to enable or disable the use
 * of clipping planes overall.  This is called by do_issue_clip_plane()
 * according to whether any planes are in use or not.
 */
void GraphicsStateGuardian::
enable_clip_planes(bool enable) {
}

/**
 * Intended to be overridden by a derived class to enable the indicated plane
 * id.  A specific PlaneNode will already have been bound to this id via
 * bind_clip_plane().
 */
void GraphicsStateGuardian::
enable_clip_plane(int plane_id, bool enable) {
}

/**
 * Called immediately before bind_clip_plane() is called, this is intended to
 * provide the derived class a hook in which to set up some state (like
 * transform) that might apply to several planes.
 *
 * The sequence is: begin_bind_clip_planes() will be called, then one or more
 * bind_clip_plane() calls, then end_bind_clip_planes().
 */
void GraphicsStateGuardian::
begin_bind_clip_planes() {
}

/**
 * Called the first time a particular clipping plane has been bound to a given
 * id within a frame, this should set up the associated hardware (or API)
 * clipping plane with the plane's properties.
 */
void GraphicsStateGuardian::
bind_clip_plane(const NodePath &plane, int plane_id) {
}

/**
 * Called after before bind_clip_plane() has been called one or more times
 * (but before any geometry is issued or additional state is changed), this is
 * intended to clean up any temporary changes to the state that may have been
 * made by begin_bind_clip_planes().
 */
void GraphicsStateGuardian::
end_bind_clip_planes() {
}

/**
 * Assigns _target_texture and _target_tex_gen based on the _target_rs.
 */
void GraphicsStateGuardian::
determine_target_texture() {
  const TextureAttrib *target_texture = (const TextureAttrib *)
    _target_rs->get_attrib_def(TextureAttrib::get_class_slot());
  const TexGenAttrib *target_tex_gen = (const TexGenAttrib *)
    _target_rs->get_attrib_def(TexGenAttrib::get_class_slot());

  nassertv(target_texture != (TextureAttrib *)NULL &&
           target_tex_gen != (TexGenAttrib *)NULL);
  _target_texture = target_texture;
  _target_tex_gen = target_tex_gen;

  if (_has_texture_alpha_scale) {
    PT(TextureStage) stage = get_alpha_scale_texture_stage();
    PT(Texture) texture = TexturePool::get_alpha_scale_map();

    _target_texture = DCAST(TextureAttrib, _target_texture->add_on_stage(stage, texture));
    _target_tex_gen = DCAST(TexGenAttrib, _target_tex_gen->add_stage
                               (stage, TexGenAttrib::M_constant, LTexCoord3(_current_color_scale[3], 0.0f, 0.0f)));
  }

  int max_texture_stages = get_max_texture_stages();
  _target_texture = _target_texture->filter_to_max(max_texture_stages);
  nassertv(_target_texture->get_num_on_stages() <= max_texture_stages);
}

/**
 * Frees some memory that was explicitly allocated within the glgsg.
 */
void GraphicsStateGuardian::
free_pointers() {
}

/**
 * This is called by the associated GraphicsWindow when close_window() is
 * called.  It should null out the _win pointer and possibly free any open
 * resources associated with the GSG.
 */
void GraphicsStateGuardian::
close_gsg() {
  // Protect from multiple calls, and also inform any other functions not to
  // try to create new stuff while we're going down.
  if (_closing_gsg) {
    return;
  }
  _closing_gsg = true;

  if (display_cat.is_debug()) {
    display_cat.debug()
      << this << " close_gsg " << get_type() << "\n";
  }

  // As tempting as it may be to try to release all the textures and geoms
  // now, we can't, because we might not be the currently-active GSG (this is
  // particularly important in OpenGL, which maintains one currently-active GL
  // state in each thread).  If we start deleting textures, we'll be
  // inadvertently deleting textures from some other OpenGL state.

  // Fortunately, it doesn't really matter, since the graphics API will be
  // responsible for cleaning up anything we don't clean up explicitly.  We'll
  // just let them drop.

  // Make sure that all the contexts belonging to the GSG are deleted.
  _prepared_objects.clear();
#ifdef DO_PSTATS
  _pending_timer_queries.clear();
#endif

  free_pointers();
}

/**
 * This is called internally when it is determined that things are just fubar.
 * It temporarily deactivates the GSG just so things don't get out of hand,
 * and throws an event so the application can deal with this if it needs to.
 */
void GraphicsStateGuardian::
panic_deactivate() {
  if (_active) {
    display_cat.error()
      << "Deactivating " << get_type() << ".\n";
    set_active(false);
    throw_event("panic-deactivate-gsg", this);
  }
}

/**
 * Called whenever the color or color scale is changed, if
 * _color_scale_via_lighting is true.  This will rederive
 * _material_force_color and _light_color_scale appropriately.
 */
void GraphicsStateGuardian::
determine_light_color_scale() {
  if (_has_scene_graph_color) {
    // If we have a scene graph color, it, plus the color scale, goes directly
    // into the material; we don't color scale the lights--this allows an
    // alpha color scale to work properly.
    _has_material_force_color = true;
    _material_force_color = _scene_graph_color;
    _light_color_scale.set(1.0f, 1.0f, 1.0f, 1.0f);
    if (!_color_blend_involves_color_scale && _color_scale_enabled) {
      _material_force_color.set(_scene_graph_color[0] * _current_color_scale[0],
                                _scene_graph_color[1] * _current_color_scale[1],
                                _scene_graph_color[2] * _current_color_scale[2],
                                _scene_graph_color[3] * _current_color_scale[3]);
    }

  } else {
    // Otherise, leave the materials alone, but we might still scale the
    // lights.
    _has_material_force_color = false;
    _light_color_scale.set(1.0f, 1.0f, 1.0f, 1.0f);
    if (!_color_blend_involves_color_scale && _color_scale_enabled) {
      _light_color_scale = _current_color_scale;
    }
  }
}

/**
 *
 */
CPT(RenderState) GraphicsStateGuardian::
get_unlit_state() {
  static CPT(RenderState) state = NULL;
  if (state == (const RenderState *)NULL) {
    state = RenderState::make(LightAttrib::make_all_off());
  }
  return state;
}

/**
 *
 */
CPT(RenderState) GraphicsStateGuardian::
get_unclipped_state() {
  static CPT(RenderState) state = NULL;
  if (state == (const RenderState *)NULL) {
    state = RenderState::make(ClipPlaneAttrib::make_all_off());
  }
  return state;
}

/**
 *
 */
CPT(RenderState) GraphicsStateGuardian::
get_untextured_state() {
  static CPT(RenderState) state = NULL;
  if (state == (const RenderState *)NULL) {
    state = RenderState::make(TextureAttrib::make_off());
  }
  return state;
}

/**
 * Should be called when a texture is encountered that needs to have its RAM
 * image reloaded, and get_incomplete_render() is true.  This will fire off a
 * thread on the current Loader object that will request the texture to load
 * its image.  The image will be available at some point in the future (no
 * event will be generated).
 */
void GraphicsStateGuardian::
async_reload_texture(TextureContext *tc) {
  nassertv(_loader != (Loader *)NULL);

  int priority = 0;
  if (_current_display_region != (DisplayRegion *)NULL) {
    priority = _current_display_region->get_texture_reload_priority();
  }

  string task_name = string("reload:") + tc->get_texture()->get_name();
  PT(AsyncTaskManager) task_mgr = _loader->get_task_manager();

  // See if we are already loading this task.
  AsyncTaskCollection orig_tasks = task_mgr->find_tasks(task_name);
  int num_tasks = orig_tasks.get_num_tasks();
  for (int ti = 0; ti < num_tasks; ++ti) {
    AsyncTask *task = orig_tasks.get_task(ti);
    if (task->is_exact_type(TextureReloadRequest::get_class_type()) &&
        DCAST(TextureReloadRequest, task)->get_texture() == tc->get_texture()) {
      // This texture is already queued to be reloaded.  Don't queue it again,
      // just make sure the priority is updated, and return.
      task->set_priority(max(task->get_priority(), priority));
      return;
    }
  }

  // This texture has not yet been queued to be reloaded.  Queue it up now.
  PT(AsyncTask) request =
    new TextureReloadRequest(task_name,
                             _prepared_objects, tc->get_texture(),
                             _supports_compressed_texture);
  request->set_priority(priority);
  _loader->load_async(request);
}

/**
 * Returns a shadow map for the given light source.  If none exists, it is
 * created, using the given host window to create the buffer, or the current
 * window if that is set to NULL.
 */
PT(Texture) GraphicsStateGuardian::
get_shadow_map(const NodePath &light_np, GraphicsOutputBase *host) {
  nassertr(light_np.node()->is_of_type(DirectionalLight::get_class_type()) ||
           light_np.node()->is_of_type(PointLight::get_class_type()) ||
           light_np.node()->is_of_type(Spotlight::get_class_type()), NULL);

  PT(LightLensNode) light = DCAST(LightLensNode, light_np.node());
  if (light == NULL || !light->_shadow_caster) {
    // TODO: return dummy shadow map (all white).
    return NULL;
  }

  // See if we already have a buffer.  If not, create one.
  if (light->_sbuffers.count(this) == 0) {
    if (host == (GraphicsOutputBase *)NULL) {
      host = _current_display_region->get_window();
    }
    nassertr(host != NULL, NULL);

    // Nope, the light doesn't have a buffer for our GSG. Make one.
    return make_shadow_buffer(light_np, host);

  } else {
    // There's already a buffer - use that.
    return light->_sbuffers[this]->get_texture();
  }
}

/**
 * Creates a depth buffer for shadow mapping.  This is a convenience function
 * for the ShaderGenerator; putting this directly in the ShaderGenerator would
 * cause circular dependency issues.  Returns the depth texture.
 */
PT(Texture) GraphicsStateGuardian::
make_shadow_buffer(const NodePath &light_np, GraphicsOutputBase *host) {
  // Make sure everything is valid.
  nassertr(light_np.node()->is_of_type(DirectionalLight::get_class_type()) ||
           light_np.node()->is_of_type(PointLight::get_class_type()) ||
           light_np.node()->is_of_type(Spotlight::get_class_type()), NULL);

  PT(LightLensNode) light = DCAST(LightLensNode, light_np.node());
  if (light == NULL || !light->_shadow_caster) {
    return NULL;
  }

  bool is_point = light->is_of_type(PointLight::get_class_type());

  nassertr(light->_sbuffers.count(this) == 0, NULL);

  if (display_cat.is_debug()) {
    display_cat.debug()
      << "Constructing shadow buffer for light '" << light->get_name()
      << "', size=" << light->_sb_size[0] << "x" << light->_sb_size[1]
      << ", sort=" << light->_sb_sort << "\n";
  }

  // Determine the properties for creating the depth buffer.
  FrameBufferProperties fbp;
  fbp.set_depth_bits(shadow_depth_bits);

  WindowProperties props = WindowProperties::size(light->_sb_size[0], light->_sb_size[1]);
  int flags = GraphicsPipe::BF_refuse_window;
  if (is_point) {
    flags |= GraphicsPipe::BF_size_square;
  }

  // Create the buffer
  PT(GraphicsOutput) sbuffer = get_engine()->make_output(get_pipe(), light->get_name(),
      light->_sb_sort, fbp, props, flags, this, DCAST(GraphicsOutput, host));
  nassertr(sbuffer != NULL, NULL);

  // Create a texture and fill it in with some data to workaround an OpenGL
  // error
  PT(Texture) tex = new Texture(light->get_name());
  if (is_point) {
    if (light->_sb_size[0] != light->_sb_size[1]) {
      display_cat.error()
        << "PointLight shadow buffers must have an equal width and height!\n";
    }
    tex->setup_cube_map(light->_sb_size[0], Texture::T_unsigned_byte, Texture::F_depth_component);
  } else {
    tex->setup_2d_texture(light->_sb_size[0], light->_sb_size[1], Texture::T_unsigned_byte, Texture::F_depth_component);
  }
  tex->make_ram_image();
  sbuffer->add_render_texture(tex, GraphicsOutput::RTM_bind_or_copy, GraphicsOutput::RTP_depth);

  // Set the wrap mode
  if (is_point) {
    tex->set_wrap_u(SamplerState::WM_clamp);
    tex->set_wrap_v(SamplerState::WM_clamp);
  } else {
    tex->set_wrap_u(SamplerState::WM_border_color);
    tex->set_wrap_v(SamplerState::WM_border_color);
    tex->set_border_color(LVecBase4(1, 1, 1, 1));
  }

  // Note: cube map shadow filtering doesn't seem to work in Cg.
  if (get_supports_shadow_filter() && !is_point) {
    // If we have the ARB_shadow extension, enable shadow filtering.
    tex->set_minfilter(SamplerState::FT_shadow);
    tex->set_magfilter(SamplerState::FT_shadow);
  } else {
    tex->set_minfilter(SamplerState::FT_linear);
    tex->set_magfilter(SamplerState::FT_linear);
  }

  // Assign display region(s) to the buffer and camera
  if (is_point) {
    for (int i = 0; i < 6; ++i) {
      PT(DisplayRegion) dr = sbuffer->make_mono_display_region(0, 1, 0, 1);
      dr->set_lens_index(i);
      dr->set_target_tex_page(i);
      dr->set_camera(light_np);
      dr->set_clear_depth_active(true);
    }
  } else {
    PT(DisplayRegion) dr = sbuffer->make_mono_display_region(0, 1, 0, 1);
    dr->set_camera(light_np);
    dr->set_clear_depth_active(true);
  }
  light->_sbuffers[this] = sbuffer;

  return tex;
}

/**
 * Returns true if the GSG implements the extension identified by the given
 * string.  This currently is only implemented by the OpenGL back-end.
 */
bool GraphicsStateGuardian::
has_extension(const string &extension) const {
  return false;
}

/**
 * Returns the vendor of the video card driver
 */
string GraphicsStateGuardian::
get_driver_vendor() {
  return string();
}

/**
 * Returns GL_Renderer
 */
string GraphicsStateGuardian::get_driver_renderer() {
  return string();
}

/**
 * Returns driver version This has an implementation-defined meaning, and may
 * be "" if the particular graphics implementation does not provide a way to
 * query this information.
 */
string GraphicsStateGuardian::
get_driver_version() {
  return string();
}

/**
 * Returns major version of the video driver.  This has an implementation-
 * defined meaning, and may be -1 if the particular graphics implementation
 * does not provide a way to query this information.
 */
int GraphicsStateGuardian::
get_driver_version_major() {
  return -1;
}

/**
 * Returns the minor version of the video driver.  This has an implementation-
 * defined meaning, and may be -1 if the particular graphics implementation
 * does not provide a way to query this information.
 */
int GraphicsStateGuardian::
get_driver_version_minor() {
  return -1;
}

/**
 * Returns the major version of the shader model.
 */
int GraphicsStateGuardian::
get_driver_shader_version_major() {
  return -1;
}

/**
 * Returns the minor version of the shader model.
 */
int GraphicsStateGuardian::
get_driver_shader_version_minor() {
  return -1;
}

ostream &
operator << (ostream &out, GraphicsStateGuardian::ShaderModel sm) {
  static const char *sm_strings[] = {"none", "1.1", "2.0", "2.x", "3.0", "4.0", "5.0", "5.1"};
  nassertr(sm >= 0 && sm <= GraphicsStateGuardian::SM_51, out);
  out << sm_strings[sm];
  return out;
}