Vertex fetching.

src/xenia/gpu/d3d11/d3d11_graphics_driver.cc

@@ -10,6 +10,7 @@
 #include <xenia/gpu/d3d11/d3d11_graphics_driver.h>
 
 #include <xenia/gpu/gpu-private.h>
+#include <xenia/gpu/d3d11/d3d11_shader.h>
 #include <xenia/gpu/d3d11/d3d11_shader_cache.h>
 
 
@@ -24,13 +25,35 @@ D3D11GraphicsDriver::D3D11GraphicsDriver(
     GraphicsDriver(memory) {
   device_ = device;
   device_->AddRef();
-
+  device_->GetImmediateContext(&context_);
   shader_cache_ = new D3D11ShaderCache(device_);
+
+  xe_zero_struct(&state_, sizeof(state_));
+
+  HRESULT hr;
+  D3D11_BUFFER_DESC buffer_desc;
+  xe_zero_struct(&buffer_desc, sizeof(buffer_desc));
+  buffer_desc.Usage           = D3D11_USAGE_DYNAMIC;
+  buffer_desc.BindFlags       = D3D11_BIND_CONSTANT_BUFFER;
+  buffer_desc.CPUAccessFlags  = D3D11_CPU_ACCESS_WRITE;
+  buffer_desc.ByteWidth       = (512 * 4) * sizeof(float);
+  hr = device_->CreateBuffer(
+      &buffer_desc, NULL, &state_.constant_buffers.float_constants);
+  buffer_desc.ByteWidth       = (8) * sizeof(int);
+  hr = device_->CreateBuffer(
+      &buffer_desc, NULL, &state_.constant_buffers.bool_constants);
+  buffer_desc.ByteWidth       = (32) * sizeof(int);
+  hr = device_->CreateBuffer(
+      &buffer_desc, NULL, &state_.constant_buffers.loop_constants);
 }
 
 D3D11GraphicsDriver::~D3D11GraphicsDriver() {
+  XESAFERELEASE(state_.constant_buffers.float_constants);
+  XESAFERELEASE(state_.constant_buffers.bool_constants);
+  XESAFERELEASE(state_.constant_buffers.loop_constants);
   delete shader_cache_;
-  device_->Release();
+  XESAFERELEASE(context_);
+  XESAFERELEASE(device_);
 }
 
 void D3D11GraphicsDriver::Initialize() {
@@ -70,22 +93,264 @@ void D3D11GraphicsDriver::SetShader(
     xe_free(source);
   }
 
-  // Prepare for use.
+  // Stash for later.
-  if (shader->Prepare()) {
+  switch (type) {
-    XELOGGPU("D3D11: failed to prepare shader");
+  case XE_GPU_SHADER_TYPE_VERTEX:
+    state_.vertex_shader = (D3D11VertexShader*)shader;
+    break;
+  case XE_GPU_SHADER_TYPE_PIXEL:
+    state_.pixel_shader = (D3D11PixelShader*)shader;
+    break;
   }
 }
 
-void D3D11GraphicsDriver::DrawIndexed(
+void D3D11GraphicsDriver::DrawAutoIndexed(
     XE_GPU_PRIMITIVE_TYPE prim_type,
     uint32_t index_count) {
+  RegisterFile& rf = register_file_;
+
   XELOGGPU("D3D11: draw indexed %d (%d indicies)",
            prim_type, index_count);
 
-  // TODO(benvanik):
+  // Misc state.
-  // program control
+  UpdateState();
-  // context misc
+
-  // interpolator control
+  // Build constant buffers.
-  // shader constants / bools / integers
+  UpdateConstantBuffers();
-  // fetch constants
+
+  // Bind shaders.
+  BindShaders();
+
+  // Switch primitive topology.
+  // Some are unsupported on D3D11 and must be emulated.
+  D3D11_PRIMITIVE_TOPOLOGY primitive_topology;
+  switch (prim_type) {
+  default:
+  case XE_GPU_PRIMITIVE_TYPE_POINT_LIST:
+    primitive_topology = D3D_PRIMITIVE_TOPOLOGY_POINTLIST;
+    break;
+  case XE_GPU_PRIMITIVE_TYPE_LINE_LIST:
+    primitive_topology = D3D_PRIMITIVE_TOPOLOGY_LINELIST;
+    break;
+  case XE_GPU_PRIMITIVE_TYPE_LINE_STRIP:
+    primitive_topology = D3D_PRIMITIVE_TOPOLOGY_LINESTRIP;
+    break;
+  case XE_GPU_PRIMITIVE_TYPE_TRIANGLE_LIST:
+    primitive_topology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST;
+    break;
+  case XE_GPU_PRIMITIVE_TYPE_TRIANGLE_STRIP:
+    primitive_topology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP;
+    break;
+  case XE_GPU_PRIMITIVE_TYPE_TRIANGLE_FAN:
+  case XE_GPU_PRIMITIVE_TYPE_UNKNOWN_07:
+  case XE_GPU_PRIMITIVE_TYPE_RECTANGLE_LIST:
+  case XE_GPU_PRIMITIVE_TYPE_LINE_LOOP:
+    XELOGE("D3D11: unsupported primitive type %d", prim_type);
+    return;
+  }
+  context_->IASetPrimitiveTopology(primitive_topology);
+
+  // Setup all fetchers (vertices/textures).
+  PrepareFetchers();
+
+  // Setup index buffer.
+  PrepareIndexBuffer();
+
+  // Issue draw.
+  uint32_t start_index = rf.values[XE_GPU_REG_VGT_INDX_OFFSET].u32;
+  uint32_t base_vertex = 0;
+  //context_->DrawIndexed(index_count, start_index, base_vertex);
+}
+
+void D3D11GraphicsDriver::UpdateState() {
+  //context_->OMSetBlendState(blend_state, blend_factor, sample_mask);
+  //context_->OMSetDepthStencilState
+  //context_->RSSetScissorRects
+  //context_->RSSetState
+  //context_->RSSetViewports
+}
+
+void D3D11GraphicsDriver::UpdateConstantBuffers() {
+  RegisterFile& rf = register_file_;
+
+  D3D11_MAPPED_SUBRESOURCE res;
+  context_->Map(
+      state_.constant_buffers.float_constants, 0,
+      D3D11_MAP_WRITE_DISCARD, 0, &res);
+  memcpy(res.pData,
+         &rf.values[XE_GPU_REG_SHADER_CONSTANT_000_X],
+         (512 * 4) * sizeof(float));
+  context_->Unmap(state_.constant_buffers.float_constants, 0);
+
+  context_->Map(
+      state_.constant_buffers.loop_constants, 0,
+      D3D11_MAP_WRITE_DISCARD, 0, &res);
+  memcpy(res.pData,
+      &rf.values[XE_GPU_REG_SHADER_CONSTANT_LOOP_00],
+      (32) * sizeof(int));
+  context_->Unmap(state_.constant_buffers.loop_constants, 0);
+
+  context_->Map(
+      state_.constant_buffers.bool_constants, 0,
+      D3D11_MAP_WRITE_DISCARD, 0, &res);
+  memcpy(res.pData,
+      &rf.values[XE_GPU_REG_SHADER_CONSTANT_BOOL_000_031],
+      (8) * sizeof(int));
+  context_->Unmap(state_.constant_buffers.bool_constants, 0);
+}
+
+void D3D11GraphicsDriver::BindShaders() {
+  RegisterFile& rf = register_file_;
+  xe_gpu_program_cntl_t program_cntl;
+  program_cntl.dword_0 = rf.values[XE_GPU_REG_SQ_PROGRAM_CNTL].u32;
+
+  // Vertex shader setup.
+  D3D11VertexShader* vs = state_.vertex_shader;
+  if (vs) {
+    if (!vs->is_prepared()) {
+      // Prepare for use.
+      if (vs->Prepare(&program_cntl)) {
+        XELOGGPU("D3D11: failed to prepare vertex shader");
+        state_.vertex_shader = NULL;
+        return;
+      }
+    }
+
+    // Bind.
+    context_->VSSetShader(vs->handle(), NULL, 0);
+
+    // Set constant buffers.
+    context_->VSSetConstantBuffers(
+        0,
+        sizeof(state_.constant_buffers) / sizeof(ID3D11Buffer*),
+        (ID3D11Buffer**)&state_.constant_buffers);
+
+    // Setup input layout (as encoded in vertex shader).
+    context_->IASetInputLayout(vs->input_layout());
+
+    //context_->VSSetSamplers
+    //context_->VSSetShaderResources
+  }
+
+  // Pixel shader setup.
+  D3D11PixelShader* ps = state_.pixel_shader;
+  if (ps) {
+    if (!ps->is_prepared()) {
+      // Prepare for use.
+      if (ps->Prepare(&program_cntl)) {
+        XELOGGPU("D3D11: failed to prepare pixel shader");
+        state_.pixel_shader = NULL;
+        return;
+      }
+    }
+
+    // Bind.
+    context_->PSSetShader(ps->handle(), NULL, 0);
+
+    // Set constant buffers.
+    context_->PSSetConstantBuffers(
+        0,
+        sizeof(state_.constant_buffers) / sizeof(ID3D11Buffer*),
+        (ID3D11Buffer**)&state_.constant_buffers);
+
+    //context_->PSSetSamplers
+    //context_->PSSetShaderResources
+  }
+}
+
+void D3D11GraphicsDriver::PrepareFetchers() {
+  RegisterFile& rf = register_file_;
+  for (int n = 0; n < 32; n++) {
+    int r = XE_GPU_REG_SHADER_CONSTANT_FETCH_00_0 + n * 6;
+    xe_gpu_fetch_group_t* group = (xe_gpu_fetch_group_t*)&rf.values[r];
+    if (group->type_0 == 0x2) {
+      PrepareTextureFetcher(n, &group->texture_fetch);
+    } else {
+      // TODO(benvanik): verify register numbering.
+      if (group->type_0 == 0x3) {
+        PrepareVertexFetcher(n * 3 + 0, &group->vertex_fetch_0);
+      }
+      if (group->type_1 == 0x3) {
+        PrepareVertexFetcher(n * 3 + 1, &group->vertex_fetch_1);
+      }
+      if (group->type_2 == 0x3) {
+        PrepareVertexFetcher(n * 3 + 2, &group->vertex_fetch_2);
+      }
+    }
+  }
+}
+
+void D3D11GraphicsDriver::PrepareVertexFetcher(
+    int slot, xe_gpu_vertex_fetch_t* fetch) {
+  uint32_t address = (fetch->address << 2) + address_translation_;
+  uint32_t size_dwords = fetch->size;
+
+  ID3D11Buffer* buffer = 0;
+  D3D11_BUFFER_DESC buffer_desc;
+  xe_zero_struct(&buffer_desc, sizeof(buffer_desc));
+  buffer_desc.ByteWidth       = size_dwords * 4;
+  buffer_desc.Usage           = D3D11_USAGE_DYNAMIC;
+  buffer_desc.BindFlags       = D3D11_BIND_VERTEX_BUFFER;
+  buffer_desc.CPUAccessFlags  = D3D11_CPU_ACCESS_WRITE;
+  device_->CreateBuffer(&buffer_desc, NULL, &buffer);
+  D3D11_MAPPED_SUBRESOURCE res;
+  context_->Map(buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &res);
+  uint32_t* src = (uint32_t*)xe_memory_addr(memory_, address);
+  uint32_t* dest = (uint32_t*)res.pData;
+  for (uint32_t n = 0; n < size_dwords; n++) {
+    union {
+      uint32_t i;
+      float f;
+    } d = {XESWAP32(src[n])};
+    XELOGGPU("v%.3d %0.8X %g", n, d.i, d.f);
+
+    dest[n] = XESWAP32(src[n]);
+  }
+  context_->Unmap(buffer, 0);
+
+  // TODO(benvanik): fetch from VS.
+  /*uint32_t stride = 0;
+  uint32_t offset = 0;
+  context_->IASetVertexBuffers(slot, 1, &buffer, &stride, &offset);*/
+
+  buffer->Release();
+}
+
+void D3D11GraphicsDriver::PrepareTextureFetcher(
+    int slot, xe_gpu_texture_fetch_t* fetch) {
+}
+
+void D3D11GraphicsDriver::PrepareIndexBuffer() {
+  RegisterFile& rf = register_file_;
+
+/*
+  ID3D11Buffer* buffer = 0;
+  D3D11_BUFFER_DESC buffer_desc;
+  xe_zero_struct(&buffer_desc, sizeof(buffer_desc));
+  buffer_desc.ByteWidth       = size_dwords * 4;
+  buffer_desc.Usage           = D3D11_USAGE_DYNAMIC;
+  buffer_desc.BindFlags       = D3D11_BIND_INDEX_BUFFER;
+  buffer_desc.CPUAccessFlags  = D3D11_CPU_ACCESS_WRITE;
+  device_->CreateBuffer(&buffer_desc, NULL, &buffer);
+  D3D11_MAPPED_SUBRESOURCE res;
+  context_->Map(buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &res);
+  uint32_t* src = (uint32_t*)xe_memory_addr(memory_, address);
+  uint32_t* dest = (uint32_t*)res.pData;
+  for (uint32_t n = 0; n < size_dwords; n++) {
+    union {
+      uint32_t i;
+      float f;
+    } d = {XESWAP32(src[n])};
+    XELOGGPU("v%.3d %0.8X %g", n, d.i, d.f);
+
+    dest[n] = XESWAP32(src[n]);
+  }
+  context_->Unmap(buffer, 0);
+
+  DXGI_FORMAT format;
+  format = DXGI_FORMAT_R16_UINT;
+  format = DXGI_FORMAT_R32_UINT;
+  context_->IASetIndexBuffer(buffer, format, 0);
+
+  buffer->Release();*/
 }

src/xenia/gpu/d3d11/d3d11_graphics_driver.h

@@ -23,7 +23,9 @@ namespace xe {
 namespace gpu {
 namespace d3d11 {
 
+class D3D11PixelShader;
 class D3D11ShaderCache;
+class D3D11VertexShader;
 
 
 class D3D11GraphicsDriver : public GraphicsDriver {
@@ -40,14 +42,36 @@ public:
       uint32_t address,
       uint32_t start,
       uint32_t length);
-  virtual void DrawIndexed(
+  virtual void DrawAutoIndexed(
       xenos::XE_GPU_PRIMITIVE_TYPE prim_type,
       uint32_t index_count);
 
 private:
-  ID3D11Device*     device_;
+  void UpdateState();
+  void UpdateConstantBuffers();
+  void BindShaders();
+  void PrepareFetchers();
+  void PrepareVertexFetcher(
+      int slot, xenos::xe_gpu_vertex_fetch_t* fetch);
+  void PrepareTextureFetcher(
+      int slot, xenos::xe_gpu_texture_fetch_t* fetch);
+  void PrepareIndexBuffer();
 
+private:
+  ID3D11Device*         device_;
+  ID3D11DeviceContext*  context_;
   D3D11ShaderCache*     shader_cache_;
+
+  struct {
+    D3D11VertexShader*  vertex_shader;
+    D3D11PixelShader*   pixel_shader;
+
+    struct {
+      ID3D11Buffer*     float_constants;
+      ID3D11Buffer*     bool_constants;
+      ID3D11Buffer*     loop_constants;
+    } constant_buffers;
+  } state_;
 };
 
 

src/xenia/gpu/d3d11/d3d11_graphics_system.cc

@@ -30,8 +30,8 @@ D3D11GraphicsSystem::D3D11GraphicsSystem(const CreationParams* params) :
 }
 
 D3D11GraphicsSystem::~D3D11GraphicsSystem() {
-  if (device_)        device_->Release();
+  XESAFERELEASE(device_);
-  if (dxgi_factory_)  dxgi_factory_->Release();
+  XESAFERELEASE(dxgi_factory_);
   delete window_;
 }
 

src/xenia/gpu/d3d11/d3d11_shader.cc

@@ -27,7 +27,7 @@ D3D11Shader::D3D11Shader(
 }
 
 D3D11Shader::~D3D11Shader() {
-  device_->Release();
+  XESAFERELEASE(device_);
 }
 
 
@@ -35,21 +35,48 @@ D3D11VertexShader::D3D11VertexShader(
     ID3D11Device* device,
     const uint8_t* src_ptr, size_t length,
     uint64_t hash) :
-    handle_(0),
+    handle_(0), input_layout_(0),
     D3D11Shader(device, XE_GPU_SHADER_TYPE_VERTEX,
                 src_ptr, length, hash) {
 }
 
 D3D11VertexShader::~D3D11VertexShader() {
-  if (handle_) handle_->Release();
+  XESAFERELEASE(input_layout_);
+  XESAFERELEASE(handle_);
 }
 
-int D3D11VertexShader::Prepare() {
+int D3D11VertexShader::Prepare(xe_gpu_program_cntl_t* program_cntl) {
   if (handle_) {
     return 0;
   }
 
-  // TODO(benvanik): translate/etc.
+  const void* byte_code = NULL;
+  size_t byte_code_length = 0;
+
+  // Create shader.
+  HRESULT hr = device_->CreateVertexShader(
+      byte_code, byte_code_length,
+      NULL,
+      &handle_);
+  if (FAILED(hr)) {
+    XELOGE("D3D11: failed to create vertex shader");
+    return 1;
+  }
+
+  // Create input layout.
+  uint32_t element_count = 0;
+  D3D11_INPUT_ELEMENT_DESC* element_descs = 0;
+  hr = device_->CreateInputLayout(
+      element_descs,
+      element_count,
+      byte_code, byte_code_length,
+      &input_layout_);
+  if (FAILED(hr)) {
+    XELOGE("D3D11: failed to create vertex shader input layout");
+    return 1;
+  }
+
+  is_prepared_ = true;
   return 0;
 }
 
@@ -64,14 +91,26 @@ D3D11PixelShader::D3D11PixelShader(
 }
 
 D3D11PixelShader::~D3D11PixelShader() {
-  if (handle_) handle_->Release();
+  XESAFERELEASE(handle_);
 }
 
-int D3D11PixelShader::Prepare() {
+int D3D11PixelShader::Prepare(xe_gpu_program_cntl_t* program_cntl) {
   if (handle_) {
     return 0;
   }
 
-  // TODO(benvanik): translate/etc.
+  const void* byte_code = NULL;
+  size_t byte_code_length = 0;
+
+  // Create shader.
+  HRESULT hr = device_->CreatePixelShader(
+      byte_code, byte_code_length,
+      NULL,
+      &handle_);
+  if (FAILED(hr)) {
+    XELOGE("D3D11: failed to create vertex shader");
+    return 1;
+  }
+  is_prepared_ = true;
   return 0;
 }

src/xenia/gpu/d3d11/d3d11_shader.h

@@ -13,6 +13,7 @@
 #include <xenia/core.h>
 
 #include <xenia/gpu/shader.h>
+#include <xenia/gpu/xenos/xenos.h>
 
 #include <d3d11.h>
 
@@ -46,10 +47,14 @@ public:
       uint64_t hash);
   virtual ~D3D11VertexShader();
 
-  virtual int Prepare();
+  ID3D11VertexShader* handle() const { return handle_; }
+  ID3D11InputLayout* input_layout() const { return input_layout_; }
+
+  int Prepare(xenos::xe_gpu_program_cntl_t* program_cntl);
 
 private:
   ID3D11VertexShader* handle_;
+  ID3D11InputLayout*  input_layout_;
 };
 
 
@@ -61,7 +66,9 @@ public:
       uint64_t hash);
   virtual ~D3D11PixelShader();
 
-  virtual int Prepare();
+  ID3D11PixelShader* handle() const { return handle_; }
+
+  int Prepare(xenos::xe_gpu_program_cntl_t* program_cntl);
 
 private:
   ID3D11PixelShader*  handle_;

src/xenia/gpu/d3d11/d3d11_shader_cache.cc

@@ -30,7 +30,7 @@ D3D11ShaderCache::~D3D11ShaderCache() {
 Shader* D3D11ShaderCache::CreateCore(
     xenos::XE_GPU_SHADER_TYPE type,
     const uint8_t* src_ptr, size_t length,
-    uint32_t hash) {
+    uint64_t hash) {
   switch (type) {
   case XE_GPU_SHADER_TYPE_VERTEX:
     return new D3D11VertexShader(

src/xenia/gpu/d3d11/d3d11_shader_cache.h

@@ -31,7 +31,7 @@ protected:
   virtual Shader* CreateCore(
       xenos::XE_GPU_SHADER_TYPE type,
       const uint8_t* src_ptr, size_t length,
-      uint32_t hash);
+      uint64_t hash);
 
 protected:
   ID3D11Device* device_;

src/xenia/gpu/d3d11/d3d11_window.cc

@@ -82,14 +82,14 @@ D3D11Window::D3D11Window(
 }
 
 D3D11Window::~D3D11Window() {
-  if (render_target_view_)  render_target_view_->Release();
   if (context_) {
     context_->ClearState();
-    context_->Release();
   }
-  if (swap_chain_)    swap_chain_->Release();
+  XESAFERELEASE(render_target_view_);
-  if (device_)        device_->Release();
+  XESAFERELEASE(context_);
-  if (dxgi_factory_)  dxgi_factory_->Release();
+  XESAFERELEASE(swap_chain_);
+  XESAFERELEASE(device_);
+  XESAFERELEASE(dxgi_factory_);
 }
 
 void D3D11Window::Swap() {

src/xenia/gpu/graphics_driver.cc

@@ -14,7 +14,8 @@ using namespace xe;
 using namespace xe::gpu;
 
 
-GraphicsDriver::GraphicsDriver(xe_memory_ref memory) {
+GraphicsDriver::GraphicsDriver(xe_memory_ref memory) :
+    address_translation_(0) {
   memory_ = xe_memory_retain(memory);
 
   memset(&register_file_, 0, sizeof(register_file_));

src/xenia/gpu/graphics_driver.h

@@ -25,6 +25,9 @@ public:
 
   xe_memory_ref memory();
   xenos::RegisterFile* register_file() { return &register_file_; };
+  void set_address_translation(uint32_t value) {
+    address_translation_ = value;
+  }
 
   virtual void Initialize() = 0;
 
@@ -35,7 +38,7 @@ public:
       uint32_t address,
       uint32_t start,
       uint32_t length) = 0;
-  virtual void DrawIndexed(
+  virtual void DrawAutoIndexed(
       xenos::XE_GPU_PRIMITIVE_TYPE prim_type,
       uint32_t index_count) = 0;
 
@@ -45,6 +48,7 @@ protected:
   xe_memory_ref memory_;
 
   xenos::RegisterFile register_file_;
+  uint32_t address_translation_;
 };
 
 

src/xenia/gpu/nop/nop_graphics_driver.cc

@@ -64,14 +64,9 @@ void NopGraphicsDriver::SetShader(
   if (source) {
     xe_free(source);
   }
-
-  // Prepare for use.
-  if (shader->Prepare()) {
-    XELOGGPU("NOP: failed to prepare shader");
-  }
 }
 
-void NopGraphicsDriver::DrawIndexed(
+void NopGraphicsDriver::DrawAutoIndexed(
     XE_GPU_PRIMITIVE_TYPE prim_type,
     uint32_t index_count) {
   XELOGGPU("NOP: draw indexed %d (%d indicies)",

src/xenia/gpu/nop/nop_graphics_driver.h

@@ -39,7 +39,7 @@ public:
       uint32_t address,
       uint32_t start,
       uint32_t length);
-  virtual void DrawIndexed(
+  virtual void DrawAutoIndexed(
       xenos::XE_GPU_PRIMITIVE_TYPE prim_type,
       uint32_t index_count);
 

src/xenia/gpu/ring_buffer_worker.cc

@@ -98,6 +98,9 @@ void RingBufferWorker::ExecuteSegment(uint32_t ptr, uint32_t length) {
   // Adjust pointer base.
   ptr = (primary_buffer_ptr_ & ~0x1FFFFFFF) | (ptr & 0x1FFFFFFF);
 
+  // Tell the driver what to use for translation.
+  driver_->set_address_translation(primary_buffer_ptr_ & ~0x1FFFFFFF);
+
 #define LOG_DATA(count) \
   for (uint32_t __m = 0; __m < count; __m++) { \
     XELOGGPU("  %.8X", XEGETUINT32BE(packet_base + 1 * 4 + __m * 4)); \
@@ -200,6 +203,7 @@ void RingBufferWorker::ExecuteSegment(uint32_t ptr, uint32_t length) {
             XELOGGPU("Packet(%.8X): PM4_INDIRECT_BUFFER %.8X (%dw)",
                      packet, list_ptr, list_length);
             ExecuteSegment(list_ptr, list_length);
+            driver_->set_address_translation(primary_buffer_ptr_ & ~0x1FFFFFFF);
           }
           break;
 
@@ -252,8 +256,9 @@ void RingBufferWorker::ExecuteSegment(uint32_t ptr, uint32_t length) {
             uint32_t d1 = XEGETUINT32BE(packet_base + 2 * 4);
             uint32_t index_count = d1 >> 16;
             uint32_t prim_type = d1 & 0x3F;
-            // Not sure what the other bits mean - 'SrcSel=AutoIndex'?
+            uint32_t src_sel = (d0 >> 6) & 0x3;
-            driver_->DrawIndexed(
+            XEASSERT(src_sel == 0x2); // 'SrcSel=AutoIndex'
+            driver_->DrawAutoIndexed(
                 (XE_GPU_PRIMITIVE_TYPE)prim_type,
                 index_count);
           }
@@ -266,9 +271,9 @@ void RingBufferWorker::ExecuteSegment(uint32_t ptr, uint32_t length) {
             uint32_t d0 = XEGETUINT32BE(packet_base + 1 * 4);
             uint32_t index_count = d0 >> 16;
             uint32_t prim_type = d0 & 0x3F;
-            // Not sure what the other bits mean - 'SrcSel=AutoIndex'?
+            uint32_t src_sel = (d0 >> 6) & 0x3;
-            // TODO(benvanik): verify this matches DRAW_INDX
+            XEASSERT(src_sel == 0x2); // 'SrcSel=AutoIndex'
-            driver_->DrawIndexed(
+            driver_->DrawAutoIndexed(
                 (XE_GPU_PRIMITIVE_TYPE)prim_type,
                 index_count);
           }

src/xenia/gpu/shader.cc

@@ -41,7 +41,3 @@ Shader::~Shader() {
 char* Shader::Disassemble() {
   return DisassembleShader(type_, dwords_, dword_count_);
 }
-
-int Shader::Prepare() {
-  return 0;
-}

src/xenia/gpu/shader.h

@@ -29,6 +29,7 @@ public:
   const uint32_t* dwords() const { return dwords_; }
   size_t dword_count() const { return dword_count_; }
   uint64_t hash() const { return hash_; }
+  bool is_prepared() const { return is_prepared_; }
 
   // vfetch formats
   // sampler formats
@@ -37,13 +38,12 @@ public:
   // NOTE: xe_free() the returned string!
   char* Disassemble();
 
-  virtual int Prepare();
-
 protected:
   xenos::XE_GPU_SHADER_TYPE type_;
   uint32_t*   dwords_;
   size_t      dword_count_;
   uint64_t    hash_;
+  bool        is_prepared_;
 };
 
 

src/xenia/gpu/xenos/xenos.h

@@ -42,6 +42,90 @@ typedef enum {
   XE_GPU_PRIMITIVE_TYPE_LINE_LOOP         = 0x0C,
 } XE_GPU_PRIMITIVE_TYPE;
 
+// XE_GPU_REG_SQ_PROGRAM_CNTL
+typedef union {
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t vs_regs            : 8;
+    uint32_t ps_regs            : 8;
+    uint32_t vs_resource        : 1;
+    uint32_t ps_resource        : 1;
+    uint32_t param_gen          : 1;
+    uint32_t unknown0           : 1;
+    uint32_t vs_export_count    : 4;
+    uint32_t vs_export_mode     : 3;
+    uint32_t ps_export_depth    : 1;
+    uint32_t ps_export_count    : 3;
+    uint32_t gen_index_vtx      : 1;
+  });
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t dword_0;
+  });
+} xe_gpu_program_cntl_t;
+
+// XE_GPU_REG_SHADER_CONSTANT_FETCH_*
+XEPACKEDUNION(xe_gpu_vertex_fetch_t, {
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t type               : 2;
+    uint32_t address            : 30;
+    uint32_t unk0               : 2;
+    uint32_t size               : 24;
+    uint32_t unk1               : 6;
+  });
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t dword_0;
+    uint32_t dword_1;
+  });
+});
+
+// XE_GPU_REG_SHADER_CONSTANT_FETCH_*
+XEPACKEDUNION(xe_gpu_texture_fetch_t, {
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t unk0;
+    uint32_t unk1;
+    uint32_t unk2;
+    uint32_t unk3;
+    uint32_t unk4;
+    uint32_t unk5;
+  });
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t dword_0;
+    uint32_t dword_1;
+    uint32_t dword_2;
+    uint32_t dword_3;
+    uint32_t dword_4;
+    uint32_t dword_5;
+  });
+});
+
+// XE_GPU_REG_SHADER_CONSTANT_FETCH_*
+XEPACKEDUNION(xe_gpu_fetch_group_t, {
+  xe_gpu_texture_fetch_t texture_fetch;
+  XEPACKEDSTRUCTANONYMOUS({
+    xe_gpu_vertex_fetch_t vertex_fetch_0;
+    xe_gpu_vertex_fetch_t vertex_fetch_1;
+    xe_gpu_vertex_fetch_t vertex_fetch_2;
+  });
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t dword_0;
+    uint32_t dword_1;
+    uint32_t dword_2;
+    uint32_t dword_3;
+    uint32_t dword_4;
+    uint32_t dword_5;
+  });
+  XEPACKEDSTRUCTANONYMOUS({
+    uint32_t type_0 : 2;
+    uint32_t        : 30;
+    uint32_t        : 32;
+    uint32_t type_1 : 2;
+    uint32_t        : 30;
+    uint32_t        : 32;
+    uint32_t type_2 : 2;
+    uint32_t        : 30;
+    uint32_t        : 32;
+  });
+});
+
 
 }  // namespace xenos
 }  // namespace gpu

src/xenia/types.h

@@ -119,6 +119,10 @@ typedef XECACHEALIGN volatile void xe_aligned_void_t;
 #endif  // GNUC
 #endif  // !MIN
 
+#if XE_PLATFORM(WIN32)
+#define XESAFERELEASE(p)        if (p) { p->Release(); }
+#endif  // WIN32
+
 #define XEBITMASK(a, b) (((unsigned) -1 >> (31 - (b))) & ~((1U << (a)) - 1))
 #define XESELECTBITS(value, a, b) ((value & XEBITMASK(a, b)) >> a)