diff --git a/3rdparty/libsdl-org/SDL b/3rdparty/libsdl-org/SDL
index 877399b2b2..5ac37a8ffc 160000
--- a/3rdparty/libsdl-org/SDL
+++ b/3rdparty/libsdl-org/SDL
@@ -1 +1 @@
-Subproject commit 877399b2b2cf21e67554ed9046410f268ce1d1b2
+Subproject commit 5ac37a8ffcf89da390404c1016833d56e2d67ae4
diff --git a/3rdparty/libsdl-org/SDL.vcxproj b/3rdparty/libsdl-org/SDL.vcxproj
index e062060fab..81b7d853a5 100644
--- a/3rdparty/libsdl-org/SDL.vcxproj
+++ b/3rdparty/libsdl-org/SDL.vcxproj
@@ -532,8 +532,6 @@
       <AdditionalIncludeDirectories>SDL\include;SDL\include\build_config;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
       <DebugInformationFormat Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">ProgramDatabase</DebugInformationFormat>
       <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|x64'">MaxSpeed</Optimization>
-      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">SDL_HIDAPI_DISABLED;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|x64'">SDL_HIDAPI_DISABLED;%(PreprocessorDefinitions)</PreprocessorDefinitions>
     </ClCompile>
   </ItemDefinitionGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
diff --git a/Utilities/lockless.h b/Utilities/lockless.h
index 63c2774bc8..5028b398c3 100644
--- a/Utilities/lockless.h
+++ b/Utilities/lockless.h
@@ -384,7 +384,7 @@ public:
 template <typename T>
 class lf_queue final
 {
-public:
+private:
 	struct fat_ptr
 	{
 		u64 ptr{};
@@ -392,7 +392,6 @@ public:
 		u32 reserved{};
 	};
 
-private:
 	atomic_t<fat_ptr> m_head{fat_ptr{}};
 
 	lf_queue_item<T>* load(fat_ptr value) const noexcept
@@ -439,8 +438,7 @@ public:
 			return *this;
 		}
 
-		delete load(m_head);
-		m_head.release(other.m_head.exchange(fat_ptr{}));
+		delete load(m_head.exchange(other.m_head.exchange(fat_ptr{})));
 		return *this;
 	}
 
@@ -453,10 +451,15 @@ public:
 	{
 		if (!operator bool())
 		{
-			utils::bless<atomic_t<u32>>(&m_head.raw().is_non_null)->wait(0);
+			get_wait_atomic().wait(0);
 		}
 	}
 
+	atomic_t<u32> &get_wait_atomic()
+	{
+		return *utils::bless<atomic_t<u32>>(&m_head.raw().is_non_null);
+	}
+
 	const volatile void* observe() const noexcept
 	{
 		return load(m_head);
@@ -491,7 +494,7 @@ public:
 	{
 		if (force || operator bool())
 		{
-			utils::bless<atomic_t<u32>>(&m_head.raw().is_non_null)->notify_one();
+			get_wait_atomic().notify_one();
 		}
 	}
 
diff --git a/rpcs3/Emu/Cell/SPUCommonRecompiler.cpp b/rpcs3/Emu/Cell/SPUCommonRecompiler.cpp
index ae1d5bd889..d73efc8e84 100644
--- a/rpcs3/Emu/Cell/SPUCommonRecompiler.cpp
+++ b/rpcs3/Emu/Cell/SPUCommonRecompiler.cpp
@@ -7376,7 +7376,7 @@ struct spu_llvm_worker
 				set_relax_flag = false;
 			}
 
-			thread_ctrl::wait_on(utils::bless<atomic_t<u32>>(&registered)[1], 0);
+			thread_ctrl::wait_on(registered.get_wait_atomic(), 0);
 			slice = registered.pop_all();
 		}())
 		{
@@ -7491,7 +7491,7 @@ struct spu_llvm
 		while (!registered && thread_ctrl::state() != thread_state::aborting)
 		{
 			// Wait for the first SPU block before launching any thread
-			thread_ctrl::wait_on(utils::bless<atomic_t<u32>>(&registered)[1], 0);
+			thread_ctrl::wait_on(registered.get_wait_atomic(), 0);
 		}
 
 		if (thread_ctrl::state() == thread_state::aborting)
@@ -7594,7 +7594,7 @@ struct spu_llvm
 
 				// Interrupt profiler thread and put it to sleep
 				static_cast<void>(prof_mutex.reset());
-				thread_ctrl::wait_on(utils::bless<atomic_t<u32>>(&registered)[1], 0);
+				thread_ctrl::wait_on(registered.get_wait_atomic(), 0);
 				std::fill(notify_compile.begin(), notify_compile.end(), 0); // Reset notification flags
 				notify_compile_count = 0;
 				compile_pending = 0;
diff --git a/rpcs3/util/atomic.hpp b/rpcs3/util/atomic.hpp
index 85c8b10482..80d9e34f9d 100644
--- a/rpcs3/util/atomic.hpp
+++ b/rpcs3/util/atomic.hpp
@@ -207,7 +207,7 @@ namespace atomic_wait
 			static_assert(Index < Max);
 			static_assert(sizeof(var) == sizeof(uptr) * 2);
 
-			m_info[Index].data = reinterpret_cast<char*>(&var) + offsetof(typename lf_queue<T2>::fat_ptr, is_non_null);
+			m_info[Index].data = std::bit_cast<char*>(&var.get_wait_atomic().raw());
 			m_info[Index].old = 0;
 		}
 
@@ -217,7 +217,7 @@ namespace atomic_wait
 			static_assert(Index < Max);
 			static_assert(sizeof(var) == sizeof(uptr) * 2);
 
-			m_info[Index].data = reinterpret_cast<char*>(&var) + offsetof(typename stx::atomic_ptr<T2>::fat_ptr, is_non_null);
+			m_info[Index].data = std::bit_cast<char*>(&var.get_wait_atomic().raw());
 			m_info[Index].old = 0;
 		}
 
diff --git a/rpcs3/util/shared_ptr.hpp b/rpcs3/util/shared_ptr.hpp
index 375784e7dd..093d90ef7e 100644
--- a/rpcs3/util/shared_ptr.hpp
+++ b/rpcs3/util/shared_ptr.hpp
@@ -576,7 +576,7 @@ namespace stx
 	template <typename T>
 	class atomic_ptr
 	{
-	public:
+	private:
 		struct fat_ptr
 		{
 			uptr ptr{};
@@ -584,8 +584,6 @@ namespace stx
 			u32 ref_ctr{};
 		};
 
-	private:
-
 		mutable atomic_t<fat_ptr> m_val{fat_ptr{}};
 
 		static shared_counter* d(fat_ptr val) noexcept
@@ -1117,19 +1115,24 @@ namespace stx
 			return static_cast<volatile const void*>(observe()) == r.get();
 		}
 
+		atomic_t<u32> &get_wait_atomic()
+		{
+			return *utils::bless<atomic_t<u32>>(&m_val.raw().is_non_null);
+		}
+
 		void wait(std::nullptr_t, atomic_wait_timeout timeout = atomic_wait_timeout::inf)
 		{
-			utils::bless<atomic_t<u32>>(&m_val.raw().is_non_null)->wait(0, timeout);
+			get_wait_atomic().wait(0, timeout);
 		}
 
 		void notify_one()
 		{
-			utils::bless<atomic_t<u32>>(&m_val.raw().is_non_null)->notify_one();
+			get_wait_atomic().notify_one();
 		}
 
 		void notify_all()
 		{
-			utils::bless<atomic_t<u32>>(&m_val.raw().is_non_null)->notify_all();
+			get_wait_atomic().notify_all();
 		}
 	};