gh-146636: abi3t: Define Py_GIL_DISABLED but do not use it (GH-148142)

When compiling for abi3t, define Py_GIL_DISABLED, so that users who
check it to enable additional locking aren't broken.

But also avoid using Py_GIL_DISABLED in Python headers themselves
-- abi3 and abi3t ought to be the same except
the _Py_OPAQUE_PYOBJECT differences.

A check for this is coming in a later PR.
It will require rewriting some preprocessor conditions, some of these
changes are included in this PR.
For _Py_IsOwnedByCurrentThread & supporting functions
I opted to move them to a cpython/ header, as they're rather self-contained.

Author: encukou

Include/cpython/object.h

@@ -495,3 +495,82 @@ PyAPI_FUNC(void) PyUnstable_EnableTryIncRef(PyObject *);
 PyAPI_FUNC(int) PyUnstable_Object_IsUniquelyReferenced(PyObject *);
 
 PyAPI_FUNC(int) PyUnstable_SetImmortal(PyObject *op);
+
+#if defined(Py_GIL_DISABLED)
+PyAPI_FUNC(uintptr_t) _Py_GetThreadLocal_Addr(void);
+
+static inline uintptr_t
+_Py_ThreadId(void)
+{
+    uintptr_t tid;
+#if defined(_MSC_VER) && defined(_M_X64)
+    tid = __readgsqword(48);
+#elif defined(_MSC_VER) && defined(_M_IX86)
+    tid = __readfsdword(24);
+#elif defined(_MSC_VER) && defined(_M_ARM64)
+    tid = __getReg(18);
+#elif defined(__MINGW32__) && defined(_M_X64)
+    tid = __readgsqword(48);
+#elif defined(__MINGW32__) && defined(_M_IX86)
+    tid = __readfsdword(24);
+#elif defined(__MINGW32__) && defined(_M_ARM64)
+    tid = __getReg(18);
+#elif defined(__i386__)
+    __asm__("{movl %%gs:0, %0|mov %0, dword ptr gs:[0]}" : "=r" (tid));  // 32-bit always uses GS
+#elif defined(__MACH__) && defined(__x86_64__)
+    __asm__("{movq %%gs:0, %0|mov %0, qword ptr gs:[0]}" : "=r" (tid));  // x86_64 macOSX uses GS
+#elif defined(__x86_64__)
+    __asm__("{movq %%fs:0, %0|mov %0, qword ptr fs:[0]}" : "=r" (tid));  // x86_64 Linux, BSD uses FS
+#elif defined(__arm__) && __ARM_ARCH >= 7
+    __asm__ ("mrc p15, 0, %0, c13, c0, 3\nbic %0, %0, #3" : "=r" (tid));
+#elif defined(__aarch64__) && defined(__APPLE__)
+    __asm__ ("mrs %0, tpidrro_el0" : "=r" (tid));
+#elif defined(__aarch64__)
+    __asm__ ("mrs %0, tpidr_el0" : "=r" (tid));
+#elif defined(__powerpc64__)
+    #if defined(__clang__) && _Py__has_builtin(__builtin_thread_pointer)
+    tid = (uintptr_t)__builtin_thread_pointer();
+    #else
+    // r13 is reserved for use as system thread ID by the Power 64-bit ABI.
+    register uintptr_t tp __asm__ ("r13");
+    __asm__("" : "=r" (tp));
+    tid = tp;
+    #endif
+#elif defined(__powerpc__)
+    #if defined(__clang__) && _Py__has_builtin(__builtin_thread_pointer)
+    tid = (uintptr_t)__builtin_thread_pointer();
+    #else
+    // r2 is reserved for use as system thread ID by the Power 32-bit ABI.
+    register uintptr_t tp __asm__ ("r2");
+    __asm__ ("" : "=r" (tp));
+    tid = tp;
+    #endif
+#elif defined(__s390__) && defined(__GNUC__)
+    // Both GCC and Clang have supported __builtin_thread_pointer
+    // for s390 from long time ago.
+    tid = (uintptr_t)__builtin_thread_pointer();
+#elif defined(__riscv)
+    #if defined(__clang__) && _Py__has_builtin(__builtin_thread_pointer)
+    tid = (uintptr_t)__builtin_thread_pointer();
+    #else
+    // tp is Thread Pointer provided by the RISC-V ABI.
+    __asm__ ("mv %0, tp" : "=r" (tid));
+    #endif
+#else
+    // Fallback to a portable implementation if we do not have a faster
+    // platform-specific implementation.
+    tid = _Py_GetThreadLocal_Addr();
+#endif
+  return tid;
+}
+
+static inline Py_ALWAYS_INLINE int
+_Py_IsOwnedByCurrentThread(PyObject *ob)
+{
+#ifdef _Py_THREAD_SANITIZER
+    return _Py_atomic_load_uintptr_relaxed(&ob->ob_tid) == _Py_ThreadId();
+#else
+    return ob->ob_tid == _Py_ThreadId();
+#endif
+}
+#endif

Include/moduleobject.h

@@ -113,8 +113,10 @@ struct PyModuleDef_Slot {
 #  define Py_MOD_GIL_NOT_USED ((void *)1)
 #endif
 
-#if !defined(Py_LIMITED_API) && defined(Py_GIL_DISABLED)
+#if !defined(Py_LIMITED_API)
+#  if defined(Py_GIL_DISABLED)
 PyAPI_FUNC(int) PyUnstable_Module_SetGIL(PyObject *module, void *gil);
+#  endif
 #endif
 
 #if !defined(Py_LIMITED_API) || Py_LIMITED_API+0 >= _Py_PACK_VERSION(3, 15)

Include/object.h

@@ -186,85 +186,6 @@ typedef struct PyVarObject PyVarObject;
 PyAPI_FUNC(int) Py_Is(PyObject *x, PyObject *y);
 #define Py_Is(x, y) ((x) == (y))
 
-#if defined(Py_GIL_DISABLED) && !defined(Py_LIMITED_API)
-PyAPI_FUNC(uintptr_t) _Py_GetThreadLocal_Addr(void);
-
-static inline uintptr_t
-_Py_ThreadId(void)
-{
-    uintptr_t tid;
-#if defined(_MSC_VER) && defined(_M_X64)
-    tid = __readgsqword(48);
-#elif defined(_MSC_VER) && defined(_M_IX86)
-    tid = __readfsdword(24);
-#elif defined(_MSC_VER) && defined(_M_ARM64)
-    tid = __getReg(18);
-#elif defined(__MINGW32__) && defined(_M_X64)
-    tid = __readgsqword(48);
-#elif defined(__MINGW32__) && defined(_M_IX86)
-    tid = __readfsdword(24);
-#elif defined(__MINGW32__) && defined(_M_ARM64)
-    tid = __getReg(18);
-#elif defined(__i386__)
-    __asm__("{movl %%gs:0, %0|mov %0, dword ptr gs:[0]}" : "=r" (tid));  // 32-bit always uses GS
-#elif defined(__MACH__) && defined(__x86_64__)
-    __asm__("{movq %%gs:0, %0|mov %0, qword ptr gs:[0]}" : "=r" (tid));  // x86_64 macOSX uses GS
-#elif defined(__x86_64__)
-    __asm__("{movq %%fs:0, %0|mov %0, qword ptr fs:[0]}" : "=r" (tid));  // x86_64 Linux, BSD uses FS
-#elif defined(__arm__) && __ARM_ARCH >= 7
-    __asm__ ("mrc p15, 0, %0, c13, c0, 3\nbic %0, %0, #3" : "=r" (tid));
-#elif defined(__aarch64__) && defined(__APPLE__)
-    __asm__ ("mrs %0, tpidrro_el0" : "=r" (tid));
-#elif defined(__aarch64__)
-    __asm__ ("mrs %0, tpidr_el0" : "=r" (tid));
-#elif defined(__powerpc64__)
-    #if defined(__clang__) && _Py__has_builtin(__builtin_thread_pointer)
-    tid = (uintptr_t)__builtin_thread_pointer();
-    #else
-    // r13 is reserved for use as system thread ID by the Power 64-bit ABI.
-    register uintptr_t tp __asm__ ("r13");
-    __asm__("" : "=r" (tp));
-    tid = tp;
-    #endif
-#elif defined(__powerpc__)
-    #if defined(__clang__) && _Py__has_builtin(__builtin_thread_pointer)
-    tid = (uintptr_t)__builtin_thread_pointer();
-    #else
-    // r2 is reserved for use as system thread ID by the Power 32-bit ABI.
-    register uintptr_t tp __asm__ ("r2");
-    __asm__ ("" : "=r" (tp));
-    tid = tp;
-    #endif
-#elif defined(__s390__) && defined(__GNUC__)
-    // Both GCC and Clang have supported __builtin_thread_pointer
-    // for s390 from long time ago.
-    tid = (uintptr_t)__builtin_thread_pointer();
-#elif defined(__riscv)
-    #if defined(__clang__) && _Py__has_builtin(__builtin_thread_pointer)
-    tid = (uintptr_t)__builtin_thread_pointer();
-    #else
-    // tp is Thread Pointer provided by the RISC-V ABI.
-    __asm__ ("mv %0, tp" : "=r" (tid));
-    #endif
-#else
-    // Fallback to a portable implementation if we do not have a faster
-    // platform-specific implementation.
-    tid = _Py_GetThreadLocal_Addr();
-#endif
-  return tid;
-}
-
-static inline Py_ALWAYS_INLINE int
-_Py_IsOwnedByCurrentThread(PyObject *ob)
-{
-#ifdef _Py_THREAD_SANITIZER
-    return _Py_atomic_load_uintptr_relaxed(&ob->ob_tid) == _Py_ThreadId();
-#else
-    return ob->ob_tid == _Py_ThreadId();
-#endif
-}
-#endif
-
 PyAPI_DATA(PyTypeObject) PyLong_Type;
 PyAPI_DATA(PyTypeObject) PyBool_Type;
 
@@ -652,8 +573,10 @@ given type object has a specified feature.
 #define _Py_IMMORTAL_FLAGS (1 << 0)
 #define _Py_LEGACY_ABI_CHECK_FLAG (1 << 1) /* see PyModuleDef_Init() */
 #define _Py_STATICALLY_ALLOCATED_FLAG (1 << 2)
-#if defined(Py_GIL_DISABLED) && defined(Py_DEBUG)
-#define _Py_TYPE_REVEALED_FLAG (1 << 3)
+#if !defined(Py_LIMITED_API)
+#  if defined(Py_GIL_DISABLED) && defined(Py_DEBUG)
+#    define _Py_TYPE_REVEALED_FLAG (1 << 3)
+#  endif
 #endif
 
 #define Py_CONSTANT_NONE 0

Include/pyport.h

@@ -77,6 +77,14 @@
 #  define Py_BUILD_CORE
 #endif
 
+#if defined(Py_TARGET_ABI3T)
+#  if !defined(Py_GIL_DISABLED)
+// Define Py_GIL_DISABLED for users' needs. This macro is used to enable
+// locking needed in for free-threaded interpreters builds.
+#    define Py_GIL_DISABLED
+#  endif
+#endif
+
 
 /**************************************************************************
 Symbols and macros to supply platform-independent interfaces to basic

Include/refcount.h

@@ -5,6 +5,7 @@ extern "C" {
 #endif
 
 
+#if !defined(_Py_OPAQUE_PYOBJECT)
 /*
 Immortalization:
 
@@ -90,14 +91,16 @@ check by comparing the reference count field to the minimum immortality refcount
 #  define _Py_REF_SHARED(refcnt, flags) \
               (((refcnt) << _Py_REF_SHARED_SHIFT) + (flags))
 #endif  // Py_GIL_DISABLED
-
+#endif  // _Py_OPAQUE_PYOBJECT
 
 // Py_REFCNT() implementation for the stable ABI
 PyAPI_FUNC(Py_ssize_t) Py_REFCNT(PyObject *ob);
 
 #if defined(Py_LIMITED_API) && Py_LIMITED_API+0 >= 0x030e0000
     // Stable ABI implements Py_REFCNT() as a function call
-    // on limited C API version 3.14 and newer.
+    // on limited C API version 3.14 and newer, and on abi3t.
+#elif defined(_Py_OPAQUE_PYOBJECT)
+    // Py_REFCNT() is also a function call in abi3t.
 #else
     static inline Py_ssize_t _Py_REFCNT(PyObject *ob) {
     #if !defined(Py_GIL_DISABLED)
@@ -150,9 +153,10 @@ PyAPI_FUNC(void) _Py_SetRefcnt(PyObject *ob, Py_ssize_t refcnt);
 
 static inline void Py_SET_REFCNT(PyObject *ob, Py_ssize_t refcnt) {
     assert(refcnt >= 0);
-#if defined(Py_LIMITED_API) && Py_LIMITED_API+0 >= 0x030d0000
+#if (defined(Py_LIMITED_API) && Py_LIMITED_API+0 >= 0x030d0000) \
+    || defined(_Py_OPAQUE_PYOBJECT)
     // Stable ABI implements Py_SET_REFCNT() as a function call
-    // on limited C API version 3.13 and newer.
+    // on limited C API version 3.13 and newer, and abi3t.
     _Py_SetRefcnt(ob, refcnt);
 #else
     // This immortal check is for code that is unaware of immortal objects.
@@ -191,7 +195,7 @@ static inline void Py_SET_REFCNT(PyObject *ob, Py_ssize_t refcnt) {
         ob->ob_ref_shared = _Py_REF_SHARED(refcnt, _Py_REF_MERGED);
     }
 #endif  // Py_GIL_DISABLED
-#endif  // Py_LIMITED_API+0 < 0x030d0000
+#endif  // Py_LIMITED_API
 }
 #if !defined(Py_LIMITED_API) || Py_LIMITED_API+0 < 0x030b0000
 #  define Py_SET_REFCNT(ob, refcnt) Py_SET_REFCNT(_PyObject_CAST(ob), (refcnt))
@@ -250,10 +254,11 @@ PyAPI_FUNC(void) _Py_DecRef(PyObject *);
 
 static inline Py_ALWAYS_INLINE void Py_INCREF(PyObject *op)
 {
-#if defined(Py_LIMITED_API) && (Py_LIMITED_API+0 >= 0x030c0000 || defined(Py_REF_DEBUG))
+#if (defined(Py_LIMITED_API) && (Py_LIMITED_API+0 >= 0x030c0000 || defined(Py_REF_DEBUG))) \
+    || defined(_Py_OPAQUE_PYOBJECT)
     // Stable ABI implements Py_INCREF() as a function call on limited C API
-    // version 3.12 and newer, and on Python built in debug mode. _Py_IncRef()
-    // was added to Python 3.10.0a7, use Py_IncRef() on older Python versions.
+    // version 3.12 and newer, abi3t, and on Python built in debug mode.
+    // _Py_IncRef() was added to Python 3.10.0a7, use Py_IncRef() on older versions.
     // Py_IncRef() accepts NULL whereas _Py_IncRef() doesn't.
 #  if Py_LIMITED_API+0 >= 0x030a00A7
     _Py_IncRef(op);
@@ -305,8 +310,8 @@ static inline Py_ALWAYS_INLINE void Py_INCREF(PyObject *op)
 #  define Py_INCREF(op) Py_INCREF(_PyObject_CAST(op))
 #endif
 
-
-#if !defined(Py_LIMITED_API) && defined(Py_GIL_DISABLED)
+#if !defined(Py_LIMITED_API)
+#if defined(Py_GIL_DISABLED)
 // Implements Py_DECREF on objects not owned by the current thread.
 PyAPI_FUNC(void) _Py_DecRefShared(PyObject *);
 PyAPI_FUNC(void) _Py_DecRefSharedDebug(PyObject *, const char *, int);
@@ -316,12 +321,14 @@ PyAPI_FUNC(void) _Py_DecRefSharedDebug(PyObject *, const char *, int);
 // zero. Otherwise, the thread gives up ownership and merges the reference
 // count fields.
 PyAPI_FUNC(void) _Py_MergeZeroLocalRefcount(PyObject *);
-#endif
+#endif  // Py_GIL_DISABLED
+#endif  // Py_LIMITED_API
 
-#if defined(Py_LIMITED_API) && (Py_LIMITED_API+0 >= 0x030c0000 || defined(Py_REF_DEBUG))
+#if (defined(Py_LIMITED_API) && (Py_LIMITED_API+0 >= 0x030c0000 || defined(Py_REF_DEBUG))) \
+    || defined(_Py_OPAQUE_PYOBJECT)
 // Stable ABI implements Py_DECREF() as a function call on limited C API
-// version 3.12 and newer, and on Python built in debug mode. _Py_DecRef() was
-// added to Python 3.10.0a7, use Py_DecRef() on older Python versions.
+// version 3.12 and newer, abi3t, and on Python built in debug mode.
+// _Py_DecRef() was added to Python 3.10.0a7, use Py_DecRef() on older versions.
 // Py_DecRef() accepts NULL whereas _Py_DecRef() doesn't.
 static inline void Py_DECREF(PyObject *op) {
 #  if Py_LIMITED_API+0 >= 0x030a00A7