Changeset 0600976 in mainline for common/stdc/uchar.c

Timestamp:

2025-04-14T11:23:38Z (3 months ago)

Author:

Jiří Zárevúcky <zarevucky.jiri@…>

Branches:

master

Children:

5d2bdaa

Parents:

11782da

git-author:

Jiří Zárevúcky <zarevucky.jiri@…> (2025-04-14 10:57:38)

git-committer:

Jiří Zárevúcky <zarevucky.jiri@…> (2025-04-14 11:23:38)

Message:

Reject invalid non-shortest UTF-8 forms and fix some other issues in str

File:

• common/stdc/uchar.c (modified) (4 diffs)

common/stdc/uchar.c

@@ -84 +84 @@
 }
 
+static bool _is_non_shortest(unsigned short cont, uint8_t b)
+{
+        return (cont == 0b1111110000000000 && !(b & 0b00100000)) ||
+            (cont == 0b1111111111110000 && !(b & 0b00110000));
+}
+
 size_t mbrtoc32(char32_t *c, const char *s, size_t n, mbstate_t *mb)
 {
@@ -139 +145 @@
 
                 if (_is_2_byte(b)) {
+                        /* Reject non-shortest form. */
+                        if (!(b & 0b00011110)) {
+                                _set_ilseq();
+                                return UCHAR_ILSEQ;
+                        }
+
                         /* 2 byte encoding               110xxxxx */
                         mb->continuation = b ^ 0b0000000011000000;
@@ -152 +164 @@
         }
 
-        while (i < n) {
+        for (; i < n; i++) {
                 /* Read continuation bytes. */
-
-                if (!_is_continuation(s[i])) {
+                uint8_t b = s[i];
+
+                if (!_is_continuation(b) || _is_non_shortest(mb->continuation, b)) {
                         _set_ilseq();
                         return UCHAR_ILSEQ;
@@ -162 +175 @@
                 /* Top bit becomes zero just before the last byte is shifted in. */
                 if (!(mb->continuation & 0x8000)) {
-                        *c = ((char32_t) mb->continuation) << 6 | (s[i++] & 0x3f);
+                        *c = ((char32_t) mb->continuation) << 6 | (b & 0x3f);
                         mb->continuation = 0;
-                        return i;
-                }
-
-                mb->continuation = mb->continuation << 6 | (s[i++] & 0x3f);
+                        return ++i;
+                }
+
+                mb->continuation = mb->continuation << 6 | (b & 0x3f);
         }