Address review

Author: Fidget-Spinner

Misc/NEWS.d/next/Core_and_Builtins/2025-01-10-18-56-20.gh-issue-128563.baDvls.rst

@@ -1 +1 @@
-A new type of interpreter has been added to CPython. This interpreter uses tail calls for its instruction handlers. Preliminary benchmark results suggest 7.4-14.7% geometric mean faster on pyperformance (depending on platform), and up to 45% faster on Python-intensive workloads. Patch by Ken Jin, with ideas by Garret Gu, Haoran Xu, and Josh Haberman.
+A new type of interpreter has been added to CPython. This interpreter uses tail calls for its instruction handlers. Preliminary benchmark results suggest 7-14% geometric mean faster on pyperformance (depending on platform), and up to 45% faster on Python-intensive workloads. Patch by Ken Jin, with ideas by Garret Gu, Haoran Xu, and Josh Haberman.

Python/ceval.c

@@ -797,7 +797,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, _PyInterpreterFrame *frame, int
     _Py_EnsureTstateNotNULL(tstate);
     CALL_STAT_INC(pyeval_calls);
 
-#if USE_COMPUTED_GOTOS && !defined(Py_TAIL_CALL_INTERP)
+#if USE_COMPUTED_GOTOS
 /* Import the static jump table */
 #include "opcode_targets.h"
 #endif

Python/ceval_macros.h

@@ -47,7 +47,7 @@
 #define OR_DTRACE_LINE
 #endif
 
-#ifdef HAVE_COMPUTED_GOTOS
+#if defined(HAVE_COMPUTED_GOTOS) && !defined(Py_TAIL_CALL_INTERP)
     #ifndef USE_COMPUTED_GOTOS
     #define USE_COMPUTED_GOTOS 1
     #endif