From 36c6178d372b075e9c74b786cfb5e47702976b1c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?B=C3=A9n=C3=A9dikt=20Tran?=
 <10796600+picnixz@users.noreply.github.com>
Date: Fri, 6 Dec 2024 15:31:30 +0100
Subject: gh-126024: fix UBSan failure in `unicodeobject.c:find_first_nonascii`
 (GH-127566)

---
 Objects/unicodeobject.c | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 463da06..33c4747 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -5083,12 +5083,9 @@ find_first_nonascii(const unsigned char *start, const unsigned char *end)
         const unsigned char *p2 = _Py_ALIGN_UP(p, SIZEOF_SIZE_T);
 #if PY_LITTLE_ENDIAN && HAVE_CTZ
         if (p < p2) {
-#if defined(_M_AMD64) || defined(_M_IX86) || defined(__x86_64__) || defined(__i386__)
-            // x86 and amd64 are little endian and can load unaligned memory.
-            size_t u = *(const size_t*)p & ASCII_CHAR_MASK;
-#else
-            size_t u = load_unaligned(p, p2 - p) & ASCII_CHAR_MASK;
-#endif
+            size_t u;
+            memcpy(&u, p, sizeof(size_t));
+            u &= ASCII_CHAR_MASK;
             if (u) {
                 return (ctz(u) - 7) / 8;
             }
-- 
cgit v0.12