spu/arm64: clean up assembly code generation

Clean up asmjit usage so we don't unnecessarily allocate memory
anymore for SPURecompiler functions.

rpcs3/Emu/Cell/SPURecompiler.cpp

@@ -1636,18 +1636,26 @@ void spu_recompiler_base::dispatch(spu_thread& spu, void*, u8* rip)
 
 		atomic_storage<u64>::release(*reinterpret_cast<u64*>(rip - 8), result);
 #elif defined(ARCH_ARM64)
-		auto jump_instrs = build_function_asm<spu_function_t>("", [](native_asm& c, auto& args)
-			{
-				using namespace asmjit;
+		union
+		{
+			u8 bytes[16];
+			u128 result;
+		};
 
-				Label branch_target = c.newLabel();
-				c.ldr(a64::x9, arm::Mem(branch_target)); // PC rel load
-				c.br(a64::x9);
+		// ldr x9, #8
+		bytes[0] = 0x49;
+		bytes[1] = 0x00;
+		bytes[2] = 0x00;
+		bytes[3] = 0x58;
 
-				c.bind(branch_target);
-				c.embedUInt64(reinterpret_cast<u64>(spu_runtime::tr_all));
-			});
-		u128 result = *reinterpret_cast<u128*>(jump_instrs);
+		// br x9
+		bytes[4] = 0x20;
+		bytes[5] = 0x01;
+		bytes[6] = 0x1F;
+		bytes[7] = 0xD6;
+
+		const u64 target = reinterpret_cast<u64>(spu_runtime::tr_all);
+		std::memcpy(bytes + 8, &target, 8);
 #if defined(__APPLE__)
 		pthread_jit_write_protect_np(false);
 #endif
@@ -1768,18 +1776,26 @@ void spu_recompiler_base::branch(spu_thread& spu, void*, u8* rip)
 
 	atomic_storage<u64>::release(*reinterpret_cast<u64*>(rip), result);
 #elif defined(ARCH_ARM64)
-	auto jmp_instrs = build_function_asm<spu_function_t>("", [&](native_asm& c, auto& args)
-		{
-			using namespace asmjit;
+	union
+	{
+		u8 bytes[16];
+		u128 result;
+	};
 
-			Label branch_target = c.newLabel();
-			c.ldr(a64::x9, arm::Mem(branch_target)); // PC rel load
-			c.br(a64::x9);
+	// ldr x9, #8
+	bytes[0] = 0x49;
+	bytes[1] = 0x00;
+	bytes[2] = 0x00;
+	bytes[3] = 0x58;
 
-			c.bind(branch_target);
-			c.embedUInt64(reinterpret_cast<u64>(func));
-		});
-	u128 result = *reinterpret_cast<u128*>(jmp_instrs);
+	// br x9
+	bytes[4] = 0x20;
+	bytes[5] = 0x01;
+	bytes[6] = 0x1F;
+	bytes[7] = 0xD6;
+
+	const u64 target = reinterpret_cast<u64>(func);
+	std::memcpy(bytes + 8, &target, 8);
 #if defined(__APPLE__)
 	pthread_jit_write_protect_np(false);
 #endif