2026-09-08 05:16:42

Signed-off-by: lizzie <lizzie@eden-emu.dev>

src/dynarmic/src/dynarmic/backend/arm64/a32_address_space.cpp

@@ -24,29 +24,65 @@
 namespace Dynarmic::Backend::Arm64 {
 
 template<auto mfp, typename T>
-static void* EmitCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+static void* EmitCallReadTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
     using namespace oaknut::util;
-
     const auto info = Devirtualize<mfp>(this_);
-
     oaknut::Label l_addr, l_this;
-
     void* target = code.xptr<void*>();
+    // params = { this, vaddr, bitsize }
     code.LDR(X0, l_this);
+    // X1 = vaddr
+    code.MOV(X2, bitsize);
     code.LDR(Xscratch0, l_addr);
     code.BR(Xscratch0);
-
     code.align(8);
     code.l(l_this);
     code.dx(info.this_ptr);
     code.l(l_addr);
     code.dx(info.fn_ptr);
-
     return target;
 }
 
 template<auto mfp, typename T>
-static void* EmitWrappedReadCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+static void* EmitCallWriteTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
+    using namespace oaknut::util;
+    const auto info = Devirtualize<mfp>(this_);
+    oaknut::Label l_addr, l_this;
+    void* target = code.xptr<void*>();
+    // params = { this, vaddr, value, bitsize }
+    code.LDR(X0, l_this);
+    // X1 = vaddr
+    // X2 = value
+    code.MOV(X3, bitsize);
+    code.LDR(Xscratch0, l_addr);
+    code.BR(Xscratch0);
+    code.align(8);
+    code.l(l_this);
+    code.dx(info.this_ptr);
+    code.l(l_addr);
+    code.dx(info.fn_ptr);
+    return target;
+}
+
+template<auto mfp, typename T>
+static void* EmitCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+    using namespace oaknut::util;
+    const auto info = Devirtualize<mfp>(this_);
+    oaknut::Label l_addr, l_this;
+    void* target = code.xptr<void*>();
+    code.LDR(X0, l_this);
+    code.LDR(Xscratch0, l_addr);
+    code.BR(Xscratch0);
+    code.align(8);
+    code.l(l_this);
+    code.dx(info.this_ptr);
+    code.l(l_addr);
+    code.dx(info.fn_ptr);
+    return target;
+}
+
+template<auto mfp, typename T>
+static void* EmitWrappedReadCallTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
     using namespace oaknut::util;
 
     const auto info = Devirtualize<mfp>(this_);
@@ -59,6 +95,7 @@ static void* EmitWrappedReadCallTrampoline(oaknut::CodeGenerator& code, T* this_
     ABI_PushRegisters(code, save_regs, 0);
     code.LDR(X0, l_this);
     code.MOV(X1, Xscratch0);
+    code.MOV(X2, bitsize);
     code.LDR(Xscratch0, l_addr);
     code.BLR(Xscratch0);
     code.MOV(Xscratch0, X0);
@@ -101,7 +138,7 @@ static void* EmitExclusiveReadCallTrampoline(oaknut::CodeGenerator& code, const
 }
 
 template<auto mfp, typename T>
-static void* EmitWrappedWriteCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+static void* EmitWrappedWriteCallTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
     using namespace oaknut::util;
 
     const auto info = Devirtualize<mfp>(this_);
@@ -115,6 +152,7 @@ static void* EmitWrappedWriteCallTrampoline(oaknut::CodeGenerator& code, T* this
     code.LDR(X0, l_this);
     code.MOV(X1, Xscratch0);
     code.MOV(X2, Xscratch1);
+    code.MOV(X3, bitsize);
     code.LDR(Xscratch0, l_addr);
     code.BLR(Xscratch0);
     ABI_PopRegisters(code, save_regs, 0);
@@ -176,14 +214,27 @@ void A32AddressSpace::EmitPrelude() {
 
     UnprotectCodeMemory();
 
-    prelude_info.read_memory = EmitCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks);
+
+    prelude_info.read_memory_8 = EmitCallReadTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u8));
+    prelude_info.read_memory_16 = EmitCallReadTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u16));
+    prelude_info.read_memory_32 = EmitCallReadTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u32));
+    prelude_info.read_memory_64 = EmitCallReadTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u64));
+    prelude_info.wrapped_read_memory_8 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u8));
+    prelude_info.wrapped_read_memory_16 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u16));
+    prelude_info.wrapped_read_memory_32 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u32));
+    prelude_info.wrapped_read_memory_64 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u64));
     prelude_info.exclusive_read_memory_8 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u8>(code, conf);
     prelude_info.exclusive_read_memory_16 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u16>(code, conf);
     prelude_info.exclusive_read_memory_32 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u32>(code, conf);
     prelude_info.exclusive_read_memory_64 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u64>(code, conf);
-    prelude_info.write_memory = EmitCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks);
+    prelude_info.write_memory_8 = EmitCallWriteTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u8));
+    prelude_info.write_memory_16 = EmitCallWriteTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u16));
+    prelude_info.write_memory_32 = EmitCallWriteTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u32));
+    prelude_info.write_memory_64 = EmitCallWriteTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u64));
+    prelude_info.wrapped_write_memory_8 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u8));
+    prelude_info.wrapped_write_memory_16 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u16));
+    prelude_info.wrapped_write_memory_32 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u32));
+    prelude_info.wrapped_write_memory_64 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u64));
     prelude_info.exclusive_write_memory_8 = EmitExclusiveWriteCallTrampoline<&A32::UserCallbacks::MemoryWriteExclusive8, u8>(code, conf);
     prelude_info.exclusive_write_memory_16 = EmitExclusiveWriteCallTrampoline<&A32::UserCallbacks::MemoryWriteExclusive16, u16>(code, conf);
     prelude_info.exclusive_write_memory_32 = EmitExclusiveWriteCallTrampoline<&A32::UserCallbacks::MemoryWriteExclusive32, u32>(code, conf);

src/dynarmic/src/dynarmic/backend/arm64/a64_address_space.cpp

@@ -23,29 +23,65 @@
 namespace Dynarmic::Backend::Arm64 {
 
 template<auto mfp, typename T>
-static void* EmitCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+static void* EmitCallReadTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
     using namespace oaknut::util;
-
     const auto info = Devirtualize<mfp>(this_);
-
     oaknut::Label l_addr, l_this;
-
     void* target = code.xptr<void*>();
+    // params = { this, vaddr, bitsize }
     code.LDR(X0, l_this);
+    // X1 = vaddr
+    code.MOV(X2, bitsize);
     code.LDR(Xscratch0, l_addr);
     code.BR(Xscratch0);
-
     code.align(8);
     code.l(l_this);
     code.dx(info.this_ptr);
     code.l(l_addr);
     code.dx(info.fn_ptr);
-
     return target;
 }
 
 template<auto mfp, typename T>
-static void* EmitWrappedReadCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+static void* EmitCallWriteTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
+    using namespace oaknut::util;
+    const auto info = Devirtualize<mfp>(this_);
+    oaknut::Label l_addr, l_this;
+    void* target = code.xptr<void*>();
+    // params = { this, vaddr, value, bitsize }
+    code.LDR(X0, l_this);
+    // X1 = vaddr
+    // X2 = value
+    code.MOV(X3, bitsize);
+    code.LDR(Xscratch0, l_addr);
+    code.BR(Xscratch0);
+    code.align(8);
+    code.l(l_this);
+    code.dx(info.this_ptr);
+    code.l(l_addr);
+    code.dx(info.fn_ptr);
+    return target;
+}
+
+template<auto mfp, typename T>
+static void* EmitCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+    using namespace oaknut::util;
+    const auto info = Devirtualize<mfp>(this_);
+    oaknut::Label l_addr, l_this;
+    void* target = code.xptr<void*>();
+    code.LDR(X0, l_this);
+    code.LDR(Xscratch0, l_addr);
+    code.BR(Xscratch0);
+    code.align(8);
+    code.l(l_this);
+    code.dx(info.this_ptr);
+    code.l(l_addr);
+    code.dx(info.fn_ptr);
+    return target;
+}
+
+template<auto mfp, typename T>
+static void* EmitWrappedReadCallTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
     using namespace oaknut::util;
 
     const auto info = Devirtualize<mfp>(this_);
@@ -58,6 +94,7 @@ static void* EmitWrappedReadCallTrampoline(oaknut::CodeGenerator& code, T* this_
     ABI_PushRegisters(code, save_regs, 0);
     code.LDR(X0, l_this);
     code.MOV(X1, Xscratch0);
+    code.MOV(X2, bitsize);
     code.LDR(Xscratch0, l_addr);
     code.BLR(Xscratch0);
     code.MOV(Xscratch0, X0);
@@ -100,7 +137,7 @@ static void* EmitExclusiveReadCallTrampoline(oaknut::CodeGenerator& code, const
 }
 
 template<auto mfp, typename T>
-static void* EmitWrappedWriteCallTrampoline(oaknut::CodeGenerator& code, T* this_) {
+static void* EmitWrappedWriteCallTrampoline(oaknut::CodeGenerator& code, T* this_, size_t bitsize) {
     using namespace oaknut::util;
 
     const auto info = Devirtualize<mfp>(this_);
@@ -114,6 +151,7 @@ static void* EmitWrappedWriteCallTrampoline(oaknut::CodeGenerator& code, T* this
     code.LDR(X0, l_this);
     code.MOV(X1, Xscratch0);
     code.MOV(X2, Xscratch1);
+    code.MOV(X3, bitsize);
     code.LDR(Xscratch0, l_addr);
     code.BLR(Xscratch0);
     ABI_PopRegisters(code, save_regs, 0);
@@ -342,18 +380,30 @@ void A64AddressSpace::EmitPrelude() {
 
     UnprotectCodeMemory();
 
-    prelude_info.read_memory = EmitCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks);
+    prelude_info.read_memory_8 = EmitCallReadTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u8));
+    prelude_info.read_memory_16 = EmitCallReadTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u16));
+    prelude_info.read_memory_32 = EmitCallReadTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u32));
+    prelude_info.read_memory_64 = EmitCallReadTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u64));
     prelude_info.read_memory_128 = EmitRead128CallTrampoline(code, conf.callbacks);
-    prelude_info.wrapped_read_memory = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks);
+    prelude_info.wrapped_read_memory_8 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u8));
+    prelude_info.wrapped_read_memory_16 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u16));
+    prelude_info.wrapped_read_memory_32 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u32));
+    prelude_info.wrapped_read_memory_64 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks, sizeof(u64));
     prelude_info.wrapped_read_memory_128 = EmitWrappedRead128CallTrampoline(code, conf.callbacks);
     prelude_info.exclusive_read_memory_8 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u8>(code, conf);
     prelude_info.exclusive_read_memory_16 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u16>(code, conf);
     prelude_info.exclusive_read_memory_32 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u32>(code, conf);
     prelude_info.exclusive_read_memory_64 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u64>(code, conf);
     prelude_info.exclusive_read_memory_128 = EmitExclusiveRead128CallTrampoline(code, conf);
-    prelude_info.write_memory = EmitCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks);
+    prelude_info.write_memory_8 = EmitCallWriteTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u8));
+    prelude_info.write_memory_16 = EmitCallWriteTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u16));
+    prelude_info.write_memory_32 = EmitCallWriteTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u32));
+    prelude_info.write_memory_64 = EmitCallWriteTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u64));
     prelude_info.write_memory_128 = EmitWrite128CallTrampoline(code, conf.callbacks);
-    prelude_info.wrapped_write_memory = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks);
+    prelude_info.wrapped_write_memory_8 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u8));
+    prelude_info.wrapped_write_memory_16 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u16));
+    prelude_info.wrapped_write_memory_32 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u32));
+    prelude_info.wrapped_write_memory_64 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks, sizeof(u64));
     prelude_info.wrapped_write_memory_128 = EmitWrappedWrite128CallTrampoline(code, conf.callbacks);
     prelude_info.exclusive_write_memory_8 = EmitExclusiveWriteCallTrampoline<&A64::UserCallbacks::MemoryWriteExclusive8, u8>(code, conf);
     prelude_info.exclusive_write_memory_16 = EmitExclusiveWriteCallTrampoline<&A64::UserCallbacks::MemoryWriteExclusive16, u16>(code, conf);

src/dynarmic/src/dynarmic/backend/arm64/address_space.cpp

@@ -147,44 +147,37 @@ void AddressSpace::Link(EmittedBlockInfo& block_info) {
             break;
         // { this, vaddr, size }
         case LinkTarget::ReadMemory8:
-            c.LDR(X2, 1);
-            c.BL(prelude_info.read_memory);
+            c.BL(prelude_info.read_memory_8);
             break;
         case LinkTarget::ReadMemory16:
-            c.LDR(X2, 2);
-            c.BL(prelude_info.read_memory);
+            c.BL(prelude_info.read_memory_16);
             break;
         case LinkTarget::ReadMemory32:
-            c.LDR(X2, 4);
-            c.BL(prelude_info.read_memory);
+            c.BL(prelude_info.read_memory_32);
             break;
         case LinkTarget::ReadMemory64:
-            c.LDR(X2, 8);
-            c.BL(prelude_info.read_memory);
+            c.BL(prelude_info.read_memory_64);
             break;
         case LinkTarget::ReadMemory128:
             c.BL(prelude_info.read_memory_128);
             break;
         // { this, vaddr, size }
         case LinkTarget::WrappedReadMemory8:
-            c.LDR(X2, 1);
-            c.BL(prelude_info.wrapped_read_memory);
+            c.BL(prelude_info.wrapped_read_memory_8);
             break;
         case LinkTarget::WrappedReadMemory16:
-            c.LDR(X2, 2);
-            c.BL(prelude_info.wrapped_read_memory);
+            c.BL(prelude_info.wrapped_read_memory_16);
             break;
         case LinkTarget::WrappedReadMemory32:
-            c.LDR(X2, 4);
-            c.BL(prelude_info.wrapped_read_memory);
+            c.BL(prelude_info.wrapped_read_memory_32);
             break;
         case LinkTarget::WrappedReadMemory64:
-            c.LDR(X2, 8);
-            c.BL(prelude_info.wrapped_read_memory);
+            c.BL(prelude_info.wrapped_read_memory_64);
             break;
         case LinkTarget::WrappedReadMemory128:
             c.BL(prelude_info.wrapped_read_memory_128);
             break;
+        //
         case LinkTarget::ExclusiveReadMemory8:
             c.BL(prelude_info.exclusive_read_memory_8);
             break;
@@ -202,43 +195,37 @@ void AddressSpace::Link(EmittedBlockInfo& block_info) {
             break;
         // { this, vaddr, value, size }
         case LinkTarget::WriteMemory8:
-            c.LDR(X3, 1);
-            c.BL(prelude_info.write_memory);
+            c.BL(prelude_info.write_memory_8);
             break;
         case LinkTarget::WriteMemory16:
-            c.LDR(X3, 2);
-            c.BL(prelude_info.write_memory);
+            c.BL(prelude_info.write_memory_16);
             break;
         case LinkTarget::WriteMemory32:
-            c.LDR(X3, 4);
-            c.BL(prelude_info.write_memory);
+            c.BL(prelude_info.write_memory_32);
             break;
         case LinkTarget::WriteMemory64:
-            c.LDR(X3, 8);
-            c.BL(prelude_info.write_memory);
+            c.BL(prelude_info.write_memory_64);
             break;
         case LinkTarget::WriteMemory128:
             c.BL(prelude_info.write_memory_128);
             break;
+        //
         case LinkTarget::WrappedWriteMemory8:
-            c.LDR(X3, 1);
-            c.BL(prelude_info.wrapped_write_memory);
+            c.BL(prelude_info.wrapped_write_memory_8);
             break;
         case LinkTarget::WrappedWriteMemory16:
-            c.LDR(X3, 2);
-            c.BL(prelude_info.wrapped_write_memory);
+            c.BL(prelude_info.wrapped_write_memory_16);
             break;
         case LinkTarget::WrappedWriteMemory32:
-            c.LDR(X3, 4);
-            c.BL(prelude_info.wrapped_write_memory);
+            c.BL(prelude_info.wrapped_write_memory_32);
             break;
         case LinkTarget::WrappedWriteMemory64:
-            c.LDR(X3, 8);
-            c.BL(prelude_info.wrapped_write_memory);
+            c.BL(prelude_info.wrapped_write_memory_64);
             break;
         case LinkTarget::WrappedWriteMemory128:
             c.BL(prelude_info.wrapped_write_memory_128);
             break;
+        //
         case LinkTarget::ExclusiveWriteMemory8:
             c.BL(prelude_info.exclusive_write_memory_8);
             break;

src/dynarmic/src/dynarmic/backend/arm64/address_space.h

@@ -93,18 +93,30 @@ protected:
         void* return_to_dispatcher;
         void* return_from_run_code;
 
-        void* read_memory;
+        void* read_memory_8;
+        void* read_memory_16;
+        void* read_memory_32;
+        void* read_memory_64;
         void* read_memory_128;
-        void* wrapped_read_memory;
+        void* wrapped_read_memory_8;
+        void* wrapped_read_memory_16;
+        void* wrapped_read_memory_32;
+        void* wrapped_read_memory_64;
         void* wrapped_read_memory_128;
         void* exclusive_read_memory_8;
         void* exclusive_read_memory_16;
         void* exclusive_read_memory_32;
         void* exclusive_read_memory_64;
         void* exclusive_read_memory_128;
-        void* write_memory;
+        void* write_memory_8;
+        void* write_memory_16;
+        void* write_memory_32;
+        void* write_memory_64;
         void* write_memory_128;
-        void* wrapped_write_memory;
+        void* wrapped_write_memory_8;
+        void* wrapped_write_memory_16;
+        void* wrapped_write_memory_32;
+        void* wrapped_write_memory_64;
         void* wrapped_write_memory_128;
         void* exclusive_write_memory_8;
         void* exclusive_write_memory_16;

src/dynarmic/src/dynarmic/ir/basic_block.cpp

@@ -39,16 +39,8 @@ Block::iterator Block::PrependNewInst(iterator insertion_point, Opcode opcode, s
     // just pool it!!! - reason why there is an inline buffer is because many small blocks are created
     // with few instructions due to subpar optimisations on other passes... plus branch-heavy code will
     // hugely benefit from the coherency of faster allocations...
-    IR::Inst* inst;
-    if (inlined_inst.size() < inlined_inst.max_size()) {
-        inlined_inst.emplace_back(opcode);
-        inst = &inlined_inst[inlined_inst.size() - 1];
-    } else {
-        if (pooled_inst.empty() || pooled_inst.back().size() == pooled_inst.back().max_size())
-            pooled_inst.emplace_back();
-        pooled_inst.back().emplace_back(opcode);
-        inst = &pooled_inst.back()[pooled_inst.back().size() - 1];
-    }
+    inlined_inst.emplace_back(opcode);
+    auto* inst = &inlined_inst[inlined_inst.size() - 1];
     DEBUG_ASSERT(args.size() == inst->NumArgs());
     std::for_each(args.begin(), args.end(), [&inst, index = size_t(0)](const auto& arg) mutable {
         inst->SetArg(index, arg);
@@ -61,7 +53,6 @@ void Block::Reset(LocationDescriptor location_) noexcept {
     mcl::intrusive_list<IR::Inst> tmp = {};
     instructions.swap(tmp);
     inlined_inst.clear();
-    pooled_inst.clear();
     cond_failed.reset();
     location = location_;
     end_location = location_;

src/dynarmic/src/dynarmic/ir/basic_block.h

@@ -142,11 +142,9 @@ public:
     }
 
     /// "Hot cache" for small blocks so we don't call global allocator
-    boost::container::static_vector<Inst, 30> inlined_inst;
+    boost::container::static_vector<Inst, 65536> inlined_inst = {};
     /// List of instructions in this block.
     instruction_list_type instructions;
-    /// "Long/far" memory pool
-    boost::container::stable_vector<boost::container::static_vector<Inst, 32>> pooled_inst;
     /// Block to execute next if `cond` did not pass.
     std::optional<LocationDescriptor> cond_failed = {};
     /// Description of the starting location of this block
@@ -162,7 +160,7 @@ public:
     /// Number of cycles this block takes to execute.
     size_t cycle_count = 0;
 };
-static_assert(sizeof(Block) == 4096);
+//static_assert(sizeof(Block) == 4096);
 
 /// Returns a string representation of the contents of block. Intended for debugging.
 std::string DumpBlock(const IR::Block& block) noexcept;

src/hid_core/frontend/emulated_controller.cpp

@@ -761,6 +761,9 @@ void EmulatedController::StartMotionCalibration() {
 }
 
 void EmulatedController::SetButton(const Common::Input::CallbackStatus& callback, std::size_t index, Common::UUID uuid) {
+    const auto player_index = Service::HID::NpadIdTypeToIndex(npad_id_type);
+    const auto& player = Settings::values.players.GetValue()[player_index];
+
     if (index >= controller.button_values.size()) {
         return;
     }
@@ -913,10 +916,21 @@ void EmulatedController::SetButton(const Common::Input::CallbackStatus& callback
         break;
     }
 
-    const auto player_index = Service::HID::NpadIdTypeToIndex(npad_id_type);
-    const auto& player = Settings::values.players.GetValue()[player_index];
-    if (player.connected) {
-        Connect();
+    if (!is_connected) {
+        if (npad_type == NpadStyleIndex::Handheld) {
+            if (npad_id_type == NpadIdType::Handheld) {
+                Connect();
+                controller_connected[player_index] = true;
+            }
+        } else if (npad_type != NpadStyleIndex::Handheld) {
+            if (npad_id_type == NpadIdType::Player1) {
+                Connect();
+                controller_connected[player_index] = true;
+            } else if (player.connected && !controller_connected[player_index]) {
+                Connect();
+                controller_connected[player_index] = true;
+            }
+        }
     }
 
     TriggerOnChange(ControllerTriggerType::Button, true);

src/hid_core/frontend/emulated_controller.h

@@ -22,6 +22,7 @@
 #include "common/settings.h"
 #include "common/vector_math.h"
 #include "hid_core/frontend/motion_input.h"
+#include "hid_core/hid_core.h"
 #include "hid_core/hid_types.h"
 #include "hid_core/irsensor/irs_types.h"
 
@@ -584,6 +585,7 @@ private:
     std::array<VibrationValue, 2> last_vibration_value{DEFAULT_VIBRATION_VALUE,
                                                        DEFAULT_VIBRATION_VALUE};
     std::array<std::chrono::steady_clock::time_point, 2> last_vibration_timepoint{};
+    std::array<bool, HIDCore::available_controllers> controller_connected{};
 
     // Atomically synched values
     std::atomic<HID::NpadStyleIndex> npad_type{HID::NpadStyleIndex::None};