diff --git a/Include/internal/pycore_cell.h b/Include/internal/pycore_cell.h
index 27f67d57b2fb79..5a12a2f14571af 100644
--- a/Include/internal/pycore_cell.h
+++ b/Include/internal/pycore_cell.h
@@ -2,6 +2,7 @@
 #define Py_INTERNAL_CELL_H
 
 #include "pycore_critical_section.h"
+#include "pycore_object.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -19,7 +20,7 @@ PyCell_SwapTakeRef(PyCellObject *cell, PyObject *value)
     PyObject *old_value;
     Py_BEGIN_CRITICAL_SECTION(cell);
     old_value = cell->ob_ref;
-    cell->ob_ref = value;
+    FT_ATOMIC_STORE_PTR_RELEASE(cell->ob_ref, value);
     Py_END_CRITICAL_SECTION();
     return old_value;
 }
@@ -42,6 +43,28 @@ PyCell_GetRef(PyCellObject *cell)
     return res;
 }
 
+static inline
+_PyStackRef _PyCell_GetStackRef(PyCellObject *cell)
+{
+    PyObject *value;
+#ifdef Py_GIL_DISABLED
+    value = _Py_atomic_load_ptr(&cell->ob_ref);
+    if (value != NULL) {
+        if (_Py_IsImmortal(value) || _PyObject_HasDeferredRefcount(value)) {
+             return (_PyStackRef){ .bits = (uintptr_t)value | Py_TAG_DEFERRED };
+         }
+        if (_Py_TryIncrefFast(value)) {
+            return _PyStackRef_FromPyObjectSteal(value);
+        }
+    }
+#endif
+    value = PyCell_GetRef(cell);
+    if (value == NULL) {
+        return PyStackRef_NULL;
+    }
+    return PyStackRef_FromPyObjectSteal(value);
+}
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/Include/internal/pycore_object.h b/Include/internal/pycore_object.h
index 0af13b1bcda20b..cc6389dc6a4f2a 100644
--- a/Include/internal/pycore_object.h
+++ b/Include/internal/pycore_object.h
@@ -16,7 +16,6 @@ extern "C" {
 #include "pycore_pystate.h"       // _PyInterpreterState_GET()
 #include "pycore_uniqueid.h"      // _PyType_IncrefSlow
 
-
 #define _Py_IMMORTAL_REFCNT_LOOSE ((_Py_IMMORTAL_REFCNT >> 1) + 1)
 
 // This value is added to `ob_ref_shared` for objects that use deferred
diff --git a/Include/internal/pycore_opcode_metadata.h b/Include/internal/pycore_opcode_metadata.h
index 3344ede5e92c07..b5fede10758f2e 100644
--- a/Include/internal/pycore_opcode_metadata.h
+++ b/Include/internal/pycore_opcode_metadata.h
@@ -1154,7 +1154,7 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[266] = {
     [LOAD_BUILD_CLASS] = { true, INSTR_FMT_IX, HAS_ERROR_FLAG | HAS_ESCAPES_FLAG },
     [LOAD_COMMON_CONSTANT] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
     [LOAD_CONST] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_CONST_FLAG | HAS_PURE_FLAG },
-    [LOAD_DEREF] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_FREE_FLAG | HAS_ERROR_FLAG | HAS_ESCAPES_FLAG },
+    [LOAD_DEREF] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_LOCAL_FLAG | HAS_ERROR_FLAG | HAS_ESCAPES_FLAG },
     [LOAD_FAST] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_LOCAL_FLAG | HAS_PURE_FLAG },
     [LOAD_FAST_AND_CLEAR] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_LOCAL_FLAG },
     [LOAD_FAST_CHECK] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_LOCAL_FLAG | HAS_ERROR_FLAG | HAS_ESCAPES_FLAG },
diff --git a/Include/internal/pycore_uop_metadata.h b/Include/internal/pycore_uop_metadata.h
index 07606135d7a356..6171af241e845d 100644
--- a/Include/internal/pycore_uop_metadata.h
+++ b/Include/internal/pycore_uop_metadata.h
@@ -123,7 +123,7 @@ const uint16_t _PyUop_Flags[MAX_UOP_ID+1] = {
     [_MAKE_CELL] = HAS_ARG_FLAG | HAS_FREE_FLAG | HAS_ERROR_FLAG | HAS_ERROR_NO_POP_FLAG,
     [_DELETE_DEREF] = HAS_ARG_FLAG | HAS_FREE_FLAG | HAS_ERROR_FLAG | HAS_ERROR_NO_POP_FLAG | HAS_ESCAPES_FLAG,
     [_LOAD_FROM_DICT_OR_DEREF] = HAS_ARG_FLAG | HAS_FREE_FLAG | HAS_ERROR_FLAG | HAS_ERROR_NO_POP_FLAG | HAS_ESCAPES_FLAG,
-    [_LOAD_DEREF] = HAS_ARG_FLAG | HAS_FREE_FLAG | HAS_ERROR_FLAG | HAS_ESCAPES_FLAG,
+    [_LOAD_DEREF] = HAS_ARG_FLAG | HAS_LOCAL_FLAG | HAS_ERROR_FLAG | HAS_ESCAPES_FLAG,
     [_STORE_DEREF] = HAS_ARG_FLAG | HAS_FREE_FLAG | HAS_ESCAPES_FLAG,
     [_COPY_FREE_VARS] = HAS_ARG_FLAG,
     [_BUILD_STRING] = HAS_ARG_FLAG | HAS_ERROR_FLAG,
diff --git a/Python/bytecodes.c b/Python/bytecodes.c
index c712c772201e10..82ed7b2c099347 100644
--- a/Python/bytecodes.c
+++ b/Python/bytecodes.c
@@ -1649,12 +1649,11 @@ dummy_func(
 
         inst(LOAD_DEREF, ( -- value)) {
             PyCellObject *cell = (PyCellObject *)PyStackRef_AsPyObjectBorrow(GETLOCAL(oparg));
-            PyObject *value_o = PyCell_GetRef(cell);
-            if (value_o == NULL) {
+            value = _PyCell_GetStackRef(cell);
+            if (PyStackRef_IsNull(value)) {
                 _PyEval_FormatExcUnbound(tstate, _PyFrame_GetCode(frame), oparg);
                 ERROR_IF(true, error);
             }
-            value = PyStackRef_FromPyObjectSteal(value_o);
         }
 
         inst(STORE_DEREF, (v --)) {
diff --git a/Python/executor_cases.c.h b/Python/executor_cases.c.h
index fdfec66b73c730..9f36420133b648 100644
--- a/Python/executor_cases.c.h
+++ b/Python/executor_cases.c.h
@@ -1824,12 +1824,11 @@
             _PyStackRef value;
             oparg = CURRENT_OPARG();
             PyCellObject *cell = (PyCellObject *)PyStackRef_AsPyObjectBorrow(GETLOCAL(oparg));
-            PyObject *value_o = PyCell_GetRef(cell);
-            if (value_o == NULL) {
+            value = _PyCell_GetStackRef(cell);
+            if (PyStackRef_IsNull(value)) {
                 _PyEval_FormatExcUnbound(tstate, _PyFrame_GetCode(frame), oparg);
                 if (true) JUMP_TO_ERROR();
             }
-            value = PyStackRef_FromPyObjectSteal(value_o);
             stack_pointer[0] = value;
             stack_pointer += 1;
             assert(WITHIN_STACK_BOUNDS());
diff --git a/Python/generated_cases.c.h b/Python/generated_cases.c.h
index 9de7554d4dfd55..753c3c94c88fad 100644
--- a/Python/generated_cases.c.h
+++ b/Python/generated_cases.c.h
@@ -5502,12 +5502,11 @@
             INSTRUCTION_STATS(LOAD_DEREF);
             _PyStackRef value;
             PyCellObject *cell = (PyCellObject *)PyStackRef_AsPyObjectBorrow(GETLOCAL(oparg));
-            PyObject *value_o = PyCell_GetRef(cell);
-            if (value_o == NULL) {
+            value = _PyCell_GetStackRef(cell);
+            if (PyStackRef_IsNull(value)) {
                 _PyEval_FormatExcUnbound(tstate, _PyFrame_GetCode(frame), oparg);
                 if (true) goto error;
             }
-            value = PyStackRef_FromPyObjectSteal(value_o);
             stack_pointer[0] = value;
             stack_pointer += 1;
             assert(WITHIN_STACK_BOUNDS());
diff --git a/Tools/tsan/suppressions_free_threading.txt b/Tools/tsan/suppressions_free_threading.txt
index e5eb665ae212de..127d710bc58bdc 100644
--- a/Tools/tsan/suppressions_free_threading.txt
+++ b/Tools/tsan/suppressions_free_threading.txt
@@ -41,6 +41,8 @@ race_top:tstate_is_freed
 race_top:type_modified_unlocked
 race_top:write_thread_id
 race_top:PyThreadState_Clear
+# see: https://github.com/python/cpython/issues/117721
+race_top:lock_PyThread_release_lock
 # Only seen on macOS, sample: https://gist.github.com/aisk/dda53f5d494a4556c35dde1fce03259c
 race_top:set_default_allocator_unlocked