Issue #15144: Fix possible integer overflow when handling pointers as integer values, by using Py_uintptr_t instead of size_t.

Patch by Serhiy Storchaka.

3 files changed, 10 insertions, 18 deletions

diff --git a/Objects/stringlib/codecs.h b/Objects/stringlib/codecs.h
index 7d55f49..2a01089 100644
--- a/Objects/stringlib/codecs.h
+++ b/Objects/stringlib/codecs.h
@@ -2,9 +2,6 @@
 
 #if STRINGLIB_IS_UNICODE
 
-/* Mask to check or force alignment of a pointer to C 'long' boundaries */
-#define LONG_PTR_MASK (size_t) (SIZEOF_LONG - 1)
-
 /* Mask to quickly check whether a C 'long' contains a
    non-ASCII, UTF8-encoded char. */
 #if (SIZEOF_LONG == 8)
@@ -25,7 +22,7 @@ STRINGLIB(utf8_decode)(const char **inptr, const char *end,
 {
     Py_UCS4 ch;
     const char *s = *inptr;
-    const char *aligned_end = (const char *) ((size_t) end & ~LONG_PTR_MASK);
+    const char *aligned_end = (const char *) _Py_ALIGN_DOWN(end, SIZEOF_LONG);
     STRINGLIB_CHAR *p = dest + *outpos;
 
     while (s < end) {
@@ -39,7 +36,7 @@ STRINGLIB(utf8_decode)(const char **inptr, const char *end,
                First, check if we can do an aligned read, as most CPUs have
                a penalty for unaligned reads.
             */
-            if (!((size_t) s & LONG_PTR_MASK)) {
+            if (_Py_IS_ALIGNED(s, SIZEOF_LONG)) {
                 /* Help register allocation */
                 register const char *_s = s;
                 register STRINGLIB_CHAR *_p = p;
@@ -453,7 +450,7 @@ STRINGLIB(utf16_decode)(const unsigned char **inptr, const unsigned char *e,
 {
     Py_UCS4 ch;
     const unsigned char *aligned_end =
-            (const unsigned char *) ((size_t) e & ~LONG_PTR_MASK);
+            (const unsigned char *) _Py_ALIGN_DOWN(e, SIZEOF_LONG);
     const unsigned char *q = *inptr;
     STRINGLIB_CHAR *p = dest + *outpos;
     /* Offsets from q for retrieving byte pairs in the right order. */
@@ -468,7 +465,7 @@ STRINGLIB(utf16_decode)(const unsigned char **inptr, const unsigned char *e,
         Py_UCS4 ch2;
         /* First check for possible aligned read of a C 'long'. Unaligned
            reads are more expensive, better to defer to another iteration. */
-        if (!((size_t) q & LONG_PTR_MASK)) {
+        if (_Py_IS_ALIGNED(q, SIZEOF_LONG)) {
             /* Fast path for runs of in-range non-surrogate chars. */
             register const unsigned char *_q = q;
             while (_q < aligned_end) {
@@ -565,7 +562,6 @@ IllegalSurrogate:
 #undef FAST_CHAR_MASK
 #undef STRIPPED_MASK
 #undef SWAB
-#undef LONG_PTR_MASK
 
 
 Py_LOCAL_INLINE(void)
@@ -588,7 +584,7 @@ STRINGLIB(utf16_encode)(unsigned short *out,
         _PyUnicode_CONVERT_BYTES(STRINGLIB_CHAR, unsigned short, in, end, out);
 # endif
     } else {
-        const STRINGLIB_CHAR *unrolled_end = in + (len & ~ (Py_ssize_t) 3);
+        const STRINGLIB_CHAR *unrolled_end = in + _Py_SIZE_ROUND_DOWN(len, 4);
         while (in < unrolled_end) {
             out[0] = SWAB2(in[0]);
             out[1] = SWAB2(in[1]);
diff --git a/Objects/stringlib/fastsearch.h b/Objects/stringlib/fastsearch.h
index 5b8d5db..ecf885e 100644
--- a/Objects/stringlib/fastsearch.h
+++ b/Objects/stringlib/fastsearch.h
@@ -43,8 +43,7 @@ STRINGLIB(fastsearch_memchr_1char)(const STRINGLIB_CHAR* s, Py_ssize_t n,
 
 #define DO_MEMCHR(memchr, s, needle, nchars) do { \
     candidate = memchr((const void *) (s), (needle), (nchars) * sizeof(STRINGLIB_CHAR)); \
-    found = (const STRINGLIB_CHAR *) \
-        ((Py_ssize_t) candidate & (~ ((Py_ssize_t) sizeof(STRINGLIB_CHAR) - 1))); \
+    found = (const STRINGLIB_CHAR *) _Py_ALIGN_DOWN(candidate, sizeof(STRINGLIB_CHAR)); \
     } while (0)
 
     if (mode == FAST_SEARCH) {
diff --git a/Objects/stringlib/find_max_char.h b/Objects/stringlib/find_max_char.h
index 9e344a0..06559c8 100644
--- a/Objects/stringlib/find_max_char.h
+++ b/Objects/stringlib/find_max_char.h
@@ -2,9 +2,6 @@
 
 #if STRINGLIB_IS_UNICODE
 
-/* Mask to check or force alignment of a pointer to C 'long' boundaries */
-#define LONG_PTR_MASK (size_t) (SIZEOF_LONG - 1)
-
 /* Mask to quickly check whether a C 'long' contains a
    non-ASCII, UTF8-encoded char. */
 #if (SIZEOF_LONG == 8)
@@ -21,10 +18,11 @@ Py_LOCAL_INLINE(Py_UCS4)
 STRINGLIB(find_max_char)(const STRINGLIB_CHAR *begin, const STRINGLIB_CHAR *end)
 {
     const unsigned char *p = (const unsigned char *) begin;
-    const unsigned char *aligned_end = (const unsigned char *) ((size_t) end & ~LONG_PTR_MASK);
+    const unsigned char *aligned_end =
+            (const unsigned char *) _Py_ALIGN_DOWN(end, SIZEOF_LONG);
 
     while (p < end) {
-        if (!((size_t) p & LONG_PTR_MASK)) {
+        if (_Py_IS_ALIGNED(p, SIZEOF_LONG)) {
             /* Help register allocation */
             register const unsigned char *_p = p;
             while (_p < aligned_end) {
@@ -43,7 +41,6 @@ STRINGLIB(find_max_char)(const STRINGLIB_CHAR *begin, const STRINGLIB_CHAR *end)
     return 127;
 }
 
-#undef LONG_PTR_MASK
 #undef ASCII_CHAR_MASK
 
 #else /* STRINGLIB_SIZEOF_CHAR == 1 */
@@ -72,7 +69,7 @@ STRINGLIB(find_max_char)(const STRINGLIB_CHAR *begin, const STRINGLIB_CHAR *end)
     register Py_UCS4 mask;
     Py_ssize_t n = end - begin;
     const STRINGLIB_CHAR *p = begin;
-    const STRINGLIB_CHAR *unrolled_end = begin + (n & ~ (Py_ssize_t) 3);
+    const STRINGLIB_CHAR *unrolled_end = begin + _Py_SIZE_ROUND_DOWN(n, 4);
     Py_UCS4 max_char;
 
     max_char = MAX_CHAR_ASCII;