bpo-46430: Fix memory leak in interned strings of deep-frozen modules (GH-31549)

Include/internal/pycore_pylifecycle.h

@@ -65,7 +65,7 @@ extern PyStatus _Py_HashRandomization_Init(const PyConfig *);
 extern PyStatus _PyImportZip_Init(PyThreadState *tstate);
 extern PyStatus _PyGC_Init(PyInterpreterState *interp);
 extern PyStatus _PyAtExit_Init(PyInterpreterState *interp);
-
+extern void _Py_Deepfreeze_Init(void);
 
 /* Various internal finalizers */
 

Misc/NEWS.d/next/Core and Builtins/2022-02-24-07-33-29.bpo-46430.c91TAg.rst

@@ -0,0 +1 @@
+Fix memory leak in interned strings of deep-frozen modules.

Programs/_bootstrap_python.c

@@ -14,9 +14,13 @@
 #include "Python/frozen_modules/importlib._bootstrap_external.h"
 /* End includes */
 
-/* Empty finalizer for deepfrozen modules*/
+/* Empty initializer for deepfrozen modules */
+void _Py_Deepfreeze_Init(void)
+{
+}
+/* Empty finalizer for deepfrozen modules */
 void
-_Py_Deepfreeze_Fini(void) 
+_Py_Deepfreeze_Fini(void)
 {
 }
 

Programs/_freeze_module.c

@@ -22,6 +22,10 @@
 #include <unistd.h>
 #endif
 
+/* Empty initializer for deepfrozen modules */
+void _Py_Deepfreeze_Init(void)
+{
+}
 /* Empty finalizer for deepfrozen modules */
 void
 _Py_Deepfreeze_Fini(void)

Python/pylifecycle.c

@@ -754,7 +754,6 @@ pycore_init_types(PyInterpreterState *interp)
     if (_PyStatus_EXCEPTION(status)) {
         return status;
     }
-
     return _PyStatus_OK();
 }
 
@@ -827,7 +826,10 @@ pycore_interp_init(PyThreadState *tstate)
     if (_PyStatus_EXCEPTION(status)) {
         return status;
     }
-
+    // Intern strings in deep-frozen modules first so that others
+    // can use it instead of creating a heap allocated string.
+    _Py_Deepfreeze_Init();
+    
     status = pycore_init_types(interp);
     if (_PyStatus_EXCEPTION(status)) {
         goto done;

Tools/scripts/deepfreeze.py

@@ -110,6 +110,7 @@ class Printer:
         self.hits, self.misses = 0, 0
         self.patchups: list[str] = []
         self.deallocs: list[str] = []
+        self.interns: list[str] = []
         self.write('#include "Python.h"')
         self.write('#include "internal/pycore_gc.h"')
         self.write('#include "internal/pycore_code.h"')
@@ -279,7 +280,7 @@ class Printer:
             self.write(f".co_cellvars = {co_cellvars},")
             self.write(f".co_freevars = {co_freevars},")
         self.deallocs.append(f"_PyStaticCode_Dealloc(&{name});")
-        self.patchups.append(f"_PyStaticCode_InternStrings(&{name});")
+        self.interns.append(f"_PyStaticCode_InternStrings(&{name});")
         return f"& {name}.ob_base"
 
     def generate_tuple(self, name: str, t: Tuple[object, ...]) -> str:
@@ -446,6 +447,9 @@ def generate(args: list[str], output: TextIO) -> None:
     with printer.block(f"void\n_Py_Deepfreeze_Fini(void)"):
             for p in printer.deallocs:
                 printer.write(p)
+    with printer.block(f"void\n_Py_Deepfreeze_Init(void)"):
+            for p in printer.interns:
+                printer.write(p)
     if verbose:
         print(f"Cache hits: {printer.hits}, misses: {printer.misses}")