Merge pull request #989 from berkayurun/fix-m_eventPtr

Stop updating event pointer on exit for reentry (fixes #980)

expat/Changes

@@ -38,6 +38,16 @@
 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
 
 Release 2.7.1 ??? ????? ?? ????
+        Bug fixes:
+       #980 #989  Restore event pointer behavior from Expat 2.6.4
+                    (that the fix to CVE-2024-8176 changed in 2.7.0);
+                    affected API functions are:
+                    - XML_GetCurrentByteCount
+                    - XML_GetCurrentByteIndex
+                    - XML_GetCurrentColumnNumber
+                    - XML_GetCurrentLineNumber
+                    - XML_GetInputContext
+
         Other changes:
        #976 #977  Autotools: Integrate files "fuzz/xml_lpm_fuzzer.{cpp,proto}"
                     with Automake that were missing from 2.7.0 release tarballs
@@ -54,6 +64,11 @@ Release 2.7.1 ??? ????? ?? ????
        #976 #977  CI: Protect against fuzzer files missing from future
                     release archives
 
+        Special thanks to:
+            Berkay Eren Ürün
+                 and
+            Perl XML::Parser
+
 Release 2.7.0 Thu March 13 2025
         Security fixes:
        #893 #973  CVE-2024-8176 -- Fix crash from chaining a large number

expat/lib/xmlparse.c

@@ -3402,12 +3402,13 @@ doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc,
       break;
       /* LCOV_EXCL_STOP */
     }
-    *eventPP = s = next;
     switch (parser->m_parsingStatus.parsing) {
     case XML_SUSPENDED:
+      *eventPP = next;
       *nextPtr = next;
       return XML_ERROR_NONE;
     case XML_FINISHED:
+      *eventPP = next;
       return XML_ERROR_ABORTED;
     case XML_PARSING:
       if (parser->m_reenter) {
@@ -3416,6 +3417,7 @@ doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc,
       }
       /* Fall through */
     default:;
+      *eventPP = s = next;
     }
   }
   /* not reached */
@@ -4332,12 +4334,13 @@ doCdataSection(XML_Parser parser, const ENCODING *enc, const char **startPtr,
       /* LCOV_EXCL_STOP */
     }
 
-    *eventPP = s = next;
     switch (parser->m_parsingStatus.parsing) {
     case XML_SUSPENDED:
+      *eventPP = next;
       *nextPtr = next;
       return XML_ERROR_NONE;
     case XML_FINISHED:
+      *eventPP = next;
       return XML_ERROR_ABORTED;
     case XML_PARSING:
       if (parser->m_reenter) {
@@ -4345,6 +4348,7 @@ doCdataSection(XML_Parser parser, const ENCODING *enc, const char **startPtr,
       }
       /* Fall through */
     default:;
+      *eventPP = s = next;
     }
   }
   /* not reached */
@@ -5951,12 +5955,13 @@ epilogProcessor(XML_Parser parser, const char *s, const char *end,
     default:
       return XML_ERROR_JUNK_AFTER_DOC_ELEMENT;
     }
-    parser->m_eventPtr = s = next;
     switch (parser->m_parsingStatus.parsing) {
     case XML_SUSPENDED:
+      parser->m_eventPtr = next;
       *nextPtr = next;
       return XML_ERROR_NONE;
     case XML_FINISHED:
+      parser->m_eventPtr = next;
       return XML_ERROR_ABORTED;
     case XML_PARSING:
       if (parser->m_reenter) {
@@ -5964,6 +5969,7 @@ epilogProcessor(XML_Parser parser, const char *s, const char *end,
       }
     /* Fall through */
     default:;
+      parser->m_eventPtr = s = next;
     }
   }
 }

expat/tests/common.c

@@ -42,6 +42,8 @@
 */
 
 #include <assert.h>
+#include <errno.h>
+#include <stdint.h> // for SIZE_MAX
 #include <stdio.h>
 #include <string.h>
 
@@ -300,3 +302,26 @@ duff_reallocator(void *ptr, size_t size) {
     g_reallocation_count--;
   return realloc(ptr, size);
 }
+
+// Portable remake of strndup(3) for C99; does not care about space efficiency
+char *
+portable_strndup(const char *s, size_t n) {
+  if ((s == NULL) || (n == SIZE_MAX)) {
+    errno = EINVAL;
+    return NULL;
+  }
+
+  char *const buffer = (char *)malloc(n + 1);
+  if (buffer == NULL) {
+    errno = ENOMEM;
+    return NULL;
+  }
+
+  errno = 0;
+
+  memcpy(buffer, s, n);
+
+  buffer[n] = '\0';
+
+  return buffer;
+}

expat/tests/common.h

@@ -146,6 +146,8 @@ extern void *duff_allocator(size_t size);
 
 extern void *duff_reallocator(void *ptr, size_t size);
 
+extern char *portable_strndup(const char *s, size_t n);
+
 #endif /* XML_COMMON_H */
 
 #ifdef __cplusplus

expat/tests/misc_tests.c

@@ -618,6 +618,66 @@ START_TEST(test_renter_loop_finite_content) {
 }
 END_TEST
 
+// Inspired by function XML_OriginalString of Perl's XML::Parser
+static char *
+dup_original_string(XML_Parser parser) {
+  const int byte_count = XML_GetCurrentByteCount(parser);
+
+  assert_true(byte_count >= 0);
+
+  int offset = -1;
+  int size = -1;
+
+  const char *const context = XML_GetInputContext(parser, &offset, &size);
+
+#if XML_CONTEXT_BYTES > 0
+  assert_true(context != NULL);
+  assert_true(offset >= 0);
+  assert_true(size >= 0);
+  return portable_strndup(context + offset, byte_count);
+#else
+  assert_true(context == NULL);
+  return NULL;
+#endif
+}
+
+static void
+on_characters_issue_980(void *userData, const XML_Char *s, int len) {
+  (void)s;
+  (void)len;
+  XML_Parser parser = (XML_Parser)userData;
+
+  char *const original_string = dup_original_string(parser);
+
+#if XML_CONTEXT_BYTES > 0
+  assert_true(original_string != NULL);
+  assert_true(strcmp(original_string, "&draft.day;") == 0);
+  free(original_string);
+#else
+  assert_true(original_string == NULL);
+#endif
+}
+
+START_TEST(test_misc_expected_event_ptr_issue_980) {
+  // NOTE: This is a tiny subset of sample "REC-xml-19980210.xml"
+  //       from Perl's XML::Parser
+  const char *const doc = "<!DOCTYPE day [\n"
+                          "  <!ENTITY draft.day '10'>\n"
+                          "]>\n"
+                          "<day>&draft.day;</day>\n";
+
+  XML_Parser parser = XML_ParserCreate(NULL);
+  XML_SetUserData(parser, parser);
+  XML_SetCharacterDataHandler(parser, on_characters_issue_980);
+
+  assert_true(_XML_Parse_SINGLE_BYTES(parser, doc, (int)strlen(doc),
+                                      /*isFinal=*/XML_TRUE)
+              == XML_STATUS_OK);
+
+  XML_ParserFree(parser);
+}
+END_TEST
+
 void
 make_miscellaneous_test_case(Suite *s) {
   TCase *tc_misc = tcase_create("miscellaneous tests");
@@ -645,4 +705,5 @@ make_miscellaneous_test_case(Suite *s) {
   tcase_add_test(tc_misc, test_misc_resumeparser_not_crashing);
   tcase_add_test(tc_misc, test_misc_stopparser_rejects_unstarted_parser);
   tcase_add_test__if_xml_ge(tc_misc, test_renter_loop_finite_content);
+  tcase_add_test(tc_misc, test_misc_expected_event_ptr_issue_980);
 }