Merge pull request #190 from shadps4-emu/stabilization_three

Stabilization three

src/core/libraries/kernel/time_management.cpp

@@ -9,9 +9,16 @@
 #include "core/libraries/libs.h"
 
 #ifdef _WIN64
+#include <Windows.h>
 #include <pthread_time.h>
+
+// http://stackoverflow.com/a/31411628/4725495
+static u32(__stdcall* NtDelayExecution)(BOOL Alertable, PLARGE_INTEGER DelayInterval) =
+    (u32(__stdcall*)(BOOL, PLARGE_INTEGER))GetProcAddress(GetModuleHandle("ntdll.dll"),
+                                                          "NtDelayExecution");
 #else
 #include <time.h>
+#include <unistd.h>
 #endif
 
 namespace Libraries::Kernel {
@@ -40,8 +47,18 @@ u64 PS4_SYSV_ABI sceKernelReadTsc() {
 }
 
 int PS4_SYSV_ABI sceKernelUsleep(u32 microseconds) {
-    ASSERT(microseconds >= 1000);
+    if (microseconds < 1000u) {
-    std::this_thread::sleep_for(std::chrono::microseconds(microseconds));
+#if _WIN64
+        LARGE_INTEGER interval{
+            .QuadPart = -1 * (microseconds * 10u),
+        };
+        NtDelayExecution(FALSE, &interval);
+    } else {
+        std::this_thread::sleep_for(std::chrono::microseconds(microseconds));
+    }
+#else
+        usleep(microseconds);
+#endif
     return 0;
 }
 

src/shader_recompiler/frontend/translate/translate.cpp

@@ -228,6 +228,9 @@ void Translate(IR::Block* block, std::span<const GcnInst> inst_list, Info& info)
         case Opcode::V_AND_B32:
             translator.V_AND_B32(inst);
             break;
+        case Opcode::V_OR_B32:
+            translator.V_OR_B32(inst);
+            break;
         case Opcode::V_LSHLREV_B32:
             translator.V_LSHLREV_B32(inst);
             break;
@@ -318,6 +321,9 @@ void Translate(IR::Block* block, std::span<const GcnInst> inst_list, Info& info)
         case Opcode::V_CMP_EQ_I32:
             translator.V_CMP_U32(ConditionOp::EQ, true, false, inst);
             break;
+        case Opcode::V_CMP_LE_I32:
+            translator.V_CMP_U32(ConditionOp::LE, true, false, inst);
+            break;
         case Opcode::V_CMP_NE_U32:
             translator.V_CMP_U32(ConditionOp::LG, false, false, inst);
             break;
@@ -378,6 +384,9 @@ void Translate(IR::Block* block, std::span<const GcnInst> inst_list, Info& info)
         case Opcode::S_CMP_GT_I32:
             translator.S_CMP(ConditionOp::GT, true, inst);
             break;
+        case Opcode::S_CMP_GE_I32:
+            translator.S_CMP(ConditionOp::GE, true, inst);
+            break;
         case Opcode::S_CMP_EQ_I32:
             translator.S_CMP(ConditionOp::EQ, true, inst);
             break;

src/shader_recompiler/frontend/translate/translate.h

@@ -62,6 +62,7 @@ public:
     void V_CVT_PKRTZ_F16_F32(const GcnInst& inst);
     void V_MUL_F32(const GcnInst& inst);
     void V_CNDMASK_B32(const GcnInst& inst);
+    void V_OR_B32(const GcnInst& inst);
     void V_AND_B32(const GcnInst& inst);
     void V_LSHLREV_B32(const GcnInst& inst);
     void V_ADD_I32(const GcnInst& inst);

src/shader_recompiler/frontend/translate/vector_alu.cpp

@@ -50,6 +50,13 @@ void Translator::V_CNDMASK_B32(const GcnInst& inst) {
     ir.SetVectorReg(dst_reg, IR::U32F32{result});
 }
 
+void Translator::V_OR_B32(const GcnInst& inst) {
+    const IR::U32 src0{GetSrc(inst.src[0])};
+    const IR::U32 src1{ir.GetVectorReg(IR::VectorReg(inst.src[1].code))};
+    const IR::VectorReg dst_reg{inst.dst[0].code};
+    ir.SetVectorReg(dst_reg, ir.BitwiseOr(src0, src1));
+}
+
 void Translator::V_AND_B32(const GcnInst& inst) {
     const IR::U32 src0{GetSrc(inst.src[0])};
     const IR::U32 src1{ir.GetVectorReg(IR::VectorReg(inst.src[1].code))};

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -429,8 +429,10 @@ void GraphicsPipeline::BindVertexBuffers(StreamBuffer& staging) const {
     for (u32 i = 0; i < num_buffers; ++i) {
         const auto& buffer = guest_buffers[i];
         const auto& host_buffer = std::ranges::find_if(
-            ranges_merged.cbegin(), ranges_merged.cend(),
+            ranges_merged.cbegin(), ranges_merged.cend(), [&](const BufferRange& range) {
-            [&](const BufferRange& range) { return (buffer.base_address >= range.base_address); });
+                return (buffer.base_address >= range.base_address &&
+                        buffer.base_address < range.end_address);
+            });
         assert(host_buffer != ranges_merged.cend());
 
         host_buffers[i] = staging.Handle();

src/video_core/renderer_vulkan/vk_stream_buffer.cpp

@@ -232,8 +232,7 @@ void StreamBuffer::WaitPendingOperations(u64 requested_upper_bound) {
 }
 
 u64 StreamBuffer::Copy(VAddr src, size_t size, size_t alignment /*= 0*/) {
-    static const u64 MinUniformAlignment = instance.UniformMinAlignment();
+    const auto [data, offset, _] = Map(size, alignment);
-    const auto [data, offset, _] = Map(size, MinUniformAlignment);
     std::memcpy(data, reinterpret_cast<const void*>(src), size);
     Commit(size);
     return offset;