bpo-40412: Nullify inittab_copy during finalization (GH-19746)

Otherwise we leave a dangling pointer to free'd memory. If we
then initialize a new interpreter in the same process and call
PyImport_ExtendInittab, we will (likely) crash when calling
PyMem_RawRealloc(inittab_copy, ...) since the pointer address
is bogus.

Automerge-Triggered-By: @brettcannon
(cherry picked from commit 64224a4727321a8dd33e6f769edda401193ebef0)

Co-authored-by: Gregory Szorc <gregory.szorc@gmail.com>

diff --git a/Misc/NEWS.d/next/C API/2020-05-01-17-28-04.bpo-40412.dE0D8N.rst b/Misc/NEWS.d/next/C API/2020-05-01-17-28-04.bpo-40412.dE0D8N.rst
new file mode 100644 (file)
index 0000000..92bfcdd
--- /dev/null
+++ b/Misc/NEWS.d/next/C API/2020-05-01-17-28-04.bpo-40412.dE0D8N.rst
@@ -0,0 +1 @@
+Nullify inittab_copy during finalization, preventing future interpreter initializations in an embedded situation from crashing. Patch by Gregory Szorc.

diff --git a/Python/import.c b/Python/import.c
index 495012d1c7da6642c900d2045a65a1941e23376a..b4074d1dfc3fab586990d353c638ee4c2cb5e034 100644 (file)
--- a/Python/import.c
+++ b/Python/import.c
@@ -300,6 +300,7 @@ _PyImport_Fini2(void)
 
     /* Free memory allocated by PyImport_ExtendInittab() */
     PyMem_RawFree(inittab_copy);
+    inittab_copy = NULL;
 
     PyMem_SetAllocator(PYMEM_DOMAIN_RAW, &old_alloc);
 }