Misc tweaks.

src/xenia/gpu/gl4/circular_buffer.cc

@@ -47,7 +47,7 @@ bool CircularBuffer::Initialize() {
 
   if (FLAGS_vendor_gl_extensions && GLEW_NV_shader_buffer_load) {
     // To use this bindlessly we must make it resident.
-    glMakeNamedBufferResidentNV(buffer_, GL_WRITE_ONLY);
+    glMakeNamedBufferResidentNV(buffer_, GL_READ_ONLY);
     glGetNamedBufferParameterui64vNV(buffer_, GL_BUFFER_GPU_ADDRESS_NV,
                                      &gpu_base_);
   }
@@ -90,11 +90,16 @@ CircularBuffer::Allocation CircularBuffer::Acquire(size_t length) {
   return allocation;
 }
 
+void CircularBuffer::Discard(Allocation allocation) {
+  write_head_ -= allocation.aligned_length;
+}
+
 void CircularBuffer::Commit(Allocation allocation) {
   uintptr_t start = allocation.gpu_ptr - gpu_base_;
   uintptr_t end = start + allocation.aligned_length;
   dirty_start_ = std::min(dirty_start_, start);
   dirty_end_ = std::max(dirty_end_, end);
+  assert_true(dirty_end_ <= capacity_);
 }
 
 void CircularBuffer::Flush() {

src/xenia/gpu/gl4/circular_buffer.h

@@ -38,6 +38,7 @@ class CircularBuffer {
 
   bool CanAcquire(size_t length);
   Allocation Acquire(size_t length);
+  void Discard(Allocation allocation);
   void Commit(Allocation allocation);
   void Flush();
 

src/xenia/gpu/gl4/command_processor.cc

@@ -163,7 +163,6 @@ void CommandProcessor::WorkerMain() {
 }
 
 bool CommandProcessor::SetupGL() {
-  glViewport(0, 0, 1280, 720);
 
   // Circular buffer holding scratch vertex/index data.
   if (!scratch_buffer_.Initialize()) {
@@ -1580,6 +1579,8 @@ CommandProcessor::UpdateStatus CommandProcessor::UpdateRenderTargets(
     // TODO(benvanik): can we do this all named?
     // TODO(benvanik): do we want this on READ too?
     glBindFramebuffer(GL_DRAW_FRAMEBUFFER, cached_framebuffer->framebuffer);
+
+    glViewport(0, 0, 1280, 720);
   }
 
   return UpdateStatus::kMismatch;

src/xenia/gpu/gl4/gl4_shader_translator.cc

@@ -936,7 +936,8 @@ bool GL4ShaderTranslator::TranslateALU_RECIP_IEEE(const instr_alu_t& alu) {
   return true;
 }
 
-bool GL4ShaderTranslator::TranslateALU_RECIPSQ_IEEE(const ucode::instr_alu_t& alu) {
+bool GL4ShaderTranslator::TranslateALU_RECIPSQ_IEEE(
+    const ucode::instr_alu_t& alu) {
   AppendDestReg(alu.scalar_dest, alu.scalar_write_mask, alu.export_data);
   Append(" = ");
   if (alu.scalar_clamp) {
@@ -1298,8 +1299,7 @@ bool GL4ShaderTranslator::TranslateALU(const instr_alu_t* alu, int sync) {
     } else {
       Append("\t    \tOP(%u)\t", alu->scalar_opc);
     }
-    PrintDstReg(alu->scalar_dest, alu->scalar_write_mask,
+    PrintDstReg(alu->scalar_dest, alu->scalar_write_mask, alu->export_data);
-                alu->export_data);
     Append(" = ");
     if (is.num_srcs == 2) {
       // ADD_CONST_0 dest, [const], [reg]

src/xenia/gpu/shader.cc

@@ -215,6 +215,7 @@ void Shader::GatherVertexFetch(const instr_fetch_vtx_t* vtx) {
     desc.stride_words = vtx->stride;
     el = &desc.elements[desc.element_count++];
   }
+  ++inputs.total_elements_count;
 
   el->vtx_fetch = *vtx;
   el->format = static_cast<VertexFormat>(vtx->format);

src/xenia/gpu/shader.h

@@ -51,6 +51,7 @@ class Shader {
   };
   struct BufferInputs {
     uint32_t count;
+    uint32_t total_elements_count;
     BufferDesc descs[32];
   };
   const BufferInputs& buffer_inputs() { return buffer_inputs_; }