gh-161242: Cache strerror() messages per errno

Include/internal/pycore_exceptions.h

@@ -22,6 +22,8 @@ extern void _PyExc_Fini(PyInterpreterState *);
 struct _Py_exc_state {
     // The dict mapping from errno codes to OSError subclasses
     PyObject *errnomap;
+    // errno code -> _Py_strerror() message
+    PyObject *strerror_cache;
     PyBaseExceptionObject *memerrors_freelist;
     int memerrors_numfree;
 #ifdef Py_GIL_DISABLED

Lib/test/test_exception_hierarchy.py

@@ -5,6 +5,7 @@
 import unittest
 import errno
 from errno import EEXIST
+from test.support import cpython_only
 
 
 class SubOSError(OSError):
@@ -136,6 +137,16 @@ def test_posix_error(self):
         if os.name == "nt":
             self.assertEqual(e.winerror, None)
 
+    @cpython_only
+    @unittest.skipIf(os.name == "nt", "POSIX-specific test")
+    def test_strerror_cached(self):
+        def strerror():
+            with self.assertRaises(FileNotFoundError) as cm:
+                open("some_hopefully_non_existing_file")
+            return cm.exception.strerror
+        self.assertIs(strerror(), strerror())
+        self.assertIs(strerror(), os.strerror(errno.ENOENT))
+
     @unittest.skipUnless(os.name == "nt", "Windows-specific test")
     def test_errno_translation(self):
         # ERROR_ALREADY_EXISTS (183) -> EEXIST

Modules/_localemodule.c

@@ -12,6 +12,7 @@ This software comes with no warranty. Use at your own risk.
 #include "Python.h"
 #include "pycore_fileutils.h"     // _Py_GetLocaleconvNumeric()
 #include "pycore_pymem.h"         // _PyMem_Strdup()
+#include "pycore_pystate.h"       // _PyInterpreterState_GET()
 
 #include <locale.h>               // setlocale()
 #include <string.h>               // strlen()
@@ -169,6 +170,8 @@ _locale_setlocale_impl(PyObject *module, int category, const char *locale)
                             "unsupported locale setting");
             return NULL;
         }
+        // strerror() messages depend on LC_MESSAGES and LC_CTYPE
+        PyDict_Clear(_PyInterpreterState_GET()->exc_state.strerror_cache);
         result_object = PyUnicode_DecodeLocale(result, NULL);
         if (!result_object)
             return NULL;

Objects/exceptions.c

@@ -4608,7 +4608,8 @@ _PyExc_InitState(PyInterpreterState *interp)
     /* Add exceptions to errnomap */
     assert(state->errnomap == NULL);
     state->errnomap = PyDict_New();
-    if (!state->errnomap) {
+    state->strerror_cache = PyDict_New();
+    if (!state->errnomap || !state->strerror_cache) {
         return _PyStatus_NO_MEMORY();
     }
 
@@ -4699,6 +4700,7 @@ _PyExc_Fini(PyInterpreterState *interp)
     struct _Py_exc_state *state = &interp->exc_state;
     free_preallocated_memerrors(state);
     Py_CLEAR(state->errnomap);
+    Py_CLEAR(state->strerror_cache);
 
     _PyExc_FiniTypes(interp);
 }

Python/fileutils.c

@@ -1,5 +1,6 @@
 #include "Python.h"
 #include "pycore_fileutils.h"     // fileutils definitions
+#include "pycore_long.h"          // _PyLong_FromUnsignedChar()
 #include "pycore_runtime.h"       // _PyRuntime
 #include "pycore_pystate.h"       // _Py_AssertHoldsTstate()
 #include "pycore_unicodeobject.h" // _Py_MAX_UNICODE
@@ -3304,13 +3305,8 @@ _Py_IsValidFD(int fd)
 }
 
 
-// Call strerror_r(code) if available, or use strerror() otherwise. Decode the
-// result from the locale encoding using surrogateescape error handler.
-//
-// On success, return a Unicode string. On error, set an exception and return
-// NULL.
-PyObject*
-_Py_strerror(int code)
+static PyObject *
+decode_strerror(int code)
 /*[clinic end generated code: output=baebf09fa02a78f2 input=75a8673d97915a91]*/
 {
     const char *errors = "surrogateescape";
@@ -3396,3 +3392,29 @@ _Py_strerror(int code)
     return PyUnicode_DecodeLocale(message, errors);
 #endif
 }
+
+
+// Call strerror_r(code) if available, or use strerror() otherwise. Decode the
+// result from the locale encoding using surrogateescape error handler.
+// Messages of codes 0-255 are cached until the next locale.setlocale():
+// decoding is a large part of the cost of raising OSError.
+//
+// On success, return a Unicode string. On error, set an exception and return
+// NULL.
+PyObject*
+_Py_strerror(int code)
+{
+    PyObject *cache = _PyInterpreterState_GET()->exc_state.strerror_cache;
+    if (cache == NULL || code < 0 || code >= 256) {
+        return decode_strerror(code);
+    }
+    PyObject *key = _PyLong_FromUnsignedChar((unsigned char)code);
+    PyObject *message;
+    if (PyDict_GetItemRef(cache, key, &message) == 0) {
+        message = decode_strerror(code);
+        if (message != NULL && PyDict_SetItem(cache, key, message) < 0) {
+            Py_CLEAR(message);
+        }
+    }
+    return message;
+}