Fix bug #77418 - Heap overflow in utf32be_mbc_to_code

(cherry picked from commit aeec40cb50eca6a97975765e2bacc28a5950cfa9)

6 files changed, 22 insertions, 2 deletions

diff --git a/NEWS b/NEWS
index f5bb842826..e4d4bd51e4 100644
--- a/NEWS
+++ b/NEWS
@@ -33,6 +33,7 @@ PHP                                                                        NEWS
 
 - MBString:
   . Fixed bug #77367 (Negative size parameter in mb_split). (Stas)
+  . Fixed bug #77418 (Heap overflow in utf32be_mbc_to_code). (Stas)
 
 - OCI8:
   . Fixed bug #76804 (oci_pconnect with OCI_CRED_EXT not working). (KoenigsKind)
diff --git a/ext/mbstring/oniguruma/src/utf16_be.c b/ext/mbstring/oniguruma/src/utf16_be.c
index 8f5b8bf64b..0fd298d2b4 100644
--- a/ext/mbstring/oniguruma/src/utf16_be.c
+++ b/ext/mbstring/oniguruma/src/utf16_be.c
@@ -128,16 +128,18 @@ utf16be_is_mbc_newline(const UChar* p, const UChar* end)
 }
 
 static OnigCodePoint
-utf16be_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
+utf16be_mbc_to_code(const UChar* p, const UChar* end)
 {
   OnigCodePoint code;
 
   if (UTF16_IS_SURROGATE_FIRST(*p)) {
+    if (end - p < 4) return 0;
     code = ((((p[0] - 0xd8) << 2) + ((p[1] & 0xc0) >> 6) + 1) << 16)
          + ((((p[1] & 0x3f) << 2) + (p[2] - 0xdc)) << 8)
          + p[3];
   }
   else {
+    if (end - p < 2) return 0;
     code = p[0] * 256 + p[1];
   }
   return code;
diff --git a/ext/mbstring/oniguruma/src/utf16_le.c b/ext/mbstring/oniguruma/src/utf16_le.c
index 92bf3186f5..47cacffdaf 100644
--- a/ext/mbstring/oniguruma/src/utf16_le.c
+++ b/ext/mbstring/oniguruma/src/utf16_le.c
@@ -141,13 +141,14 @@ utf16le_is_mbc_newline(const UChar* p, const UChar* end)
 }
 
 static OnigCodePoint
-utf16le_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
+utf16le_mbc_to_code(const UChar* p, const UChar* end)
 {
   OnigCodePoint code;
   UChar c0 = *p;
   UChar c1 = *(p+1);
 
   if (UTF16_IS_SURROGATE_FIRST(c1)) {
+    if (end - p < 4) return 0;
     code = ((((c1 - 0xd8) << 2) + ((c0  & 0xc0) >> 6) + 1) << 16)
          + ((((c0 & 0x3f) << 2) + (p[3] - 0xdc)) << 8)
          + p[2];
diff --git a/ext/mbstring/oniguruma/src/utf32_be.c b/ext/mbstring/oniguruma/src/utf32_be.c
index 92476ec033..db353afc50 100644
--- a/ext/mbstring/oniguruma/src/utf32_be.c
+++ b/ext/mbstring/oniguruma/src/utf32_be.c
@@ -67,6 +67,7 @@ utf32be_is_mbc_newline(const UChar* p, const UChar* end)
 static OnigCodePoint
 utf32be_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
 {
+  if (end - p < 4) return 0;
   return (OnigCodePoint )(((p[0] * 256 + p[1]) * 256 + p[2]) * 256 + p[3]);
 }
 
diff --git a/ext/mbstring/oniguruma/src/utf32_le.c b/ext/mbstring/oniguruma/src/utf32_le.c
index dc3fd92806..f32ce9a61c 100644
--- a/ext/mbstring/oniguruma/src/utf32_le.c
+++ b/ext/mbstring/oniguruma/src/utf32_le.c
@@ -67,6 +67,7 @@ utf32le_is_mbc_newline(const UChar* p, const UChar* end)
 static OnigCodePoint
 utf32le_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
 {
+  if (end - p < 4) return 0;
   return (OnigCodePoint )(((p[3] * 256 + p[2]) * 256 + p[1]) * 256 + p[0]);
 }
 
diff --git a/ext/mbstring/tests/bug77418.phpt b/ext/mbstring/tests/bug77418.phpt
new file mode 100644
index 0000000000..b4acc45c21
--- /dev/null
+++ b/ext/mbstring/tests/bug77418.phpt
@@ -0,0 +1,14 @@
+--TEST--
+Bug #77371 (Heap overflow in utf32be_mbc_to_code)
+--SKIPIF--
+<?php extension_loaded('mbstring') or die('skip mbstring not available'); ?>
+--FILE--
+<?php
+mb_regex_encoding("UTF-32");
+var_dump(mb_split("\x00\x00\x00\x5c\x00\x00\x00B","000000000000000000000000000000"));
+?>
+--EXPECT--
+array(1) {
+  [0]=>
+  string(30) "000000000000000000000000000000"
+}