This is one of two remaining "definitely lost" leaks in py3k. It first
appeared in r70152. How to reproduce:

make distclean && ./configure OPT="-O0 -g" --without-pymalloc && make

valgrind --leak-check=full --suppressions=Misc/valgrind-python.supp ./python > VGOUT 2>&1

Then search for 'definitely'. This leak is not present in release-2.7.


==25233== 106 (56 direct, 50 indirect) bytes in 1 blocks are definitely lost in loss record 1,432 of 2,121
==25233==    at 0x4C2412C: malloc (vg_replace_malloc.c:195)
==25233==    by 0x4167AE: _PyObject_New (object.c:243)
==25233==    by 0x42C1C4: _PyUnicode_New (unicodeobject.c:341)
==25233==    by 0x430562: PyUnicodeUCS2_DecodeUTF8Stateful (unicodeobject.c:2036)
==25233==    by 0x430516: PyUnicodeUCS2_DecodeUTF8 (unicodeobject.c:2001)
==25233==    by 0x479F81: r_object (marshal.c:726)
==25233==    by 0x47A03E: r_object (marshal.c:745)
==25233==    by 0x47A720: r_object (marshal.c:873)
==25233==    by 0x47AF4B: PyMarshal_ReadObjectFromString (marshal.c:1053)
==25233==    by 0x47AE2A: PyMarshal_ReadLastObjectFromFile (marshal.c:1012)
==25233==    by 0x471C5B: read_compiled_module (import.c:823)
==25233==    by 0x47230C: load_source_module (import.c:1043)

I tracked it down to r68683, which is still a large commit.

Your traceback suggests it's a string allocated when reading a module file...

I also thought that it might be a pointer that's lost in marshal.c.
However, perhaps a pointer is lost in Py_Finalize. With this ...

Index: Modules/main.c
===================================================================
--- Modules/main.c      (revision 85766)
+++ Modules/main.c      (working copy)
@@ -697,6 +697,7 @@
         sts = PyRun_AnyFileFlags(stdin, "<stdin>", &cf) != 0;
     }
 
+    exit(0);
     Py_Finalize();
 
 #ifdef __INSURE__


... Valgrind does not report the leak.

Hello. Does this patch fix the problem?

Hirokazu, the patch looks good to me. Unfortunately Valgrind still
reports the leak.

After taking the scenic route through half of the tree[1], I finally
found another leak in pythonrun.c. I'm closing #10153, merging those
two leaks into the new patch.


Does it look ok?


[1] Valgrind stack traces should be approached with caution.

I think your patch looks good.

Thank you, I committed your patch in r85980(py3k) and
r85981(release31-maint).