ID: 19493 User updated by: [EMAIL PROTECTED] Reported By: [EMAIL PROTECTED] -Status: Feedback +Status: Open Bug Type: Session related Operating System: Linux 2.4.18 PHP Version: 4.2.3 New Comment:
Hello, there is a new patch (without C++-comments :-): --- ext/standard/var_unserializer.re.old Thu Sep 19 16:34:32 2002 +++ ext/standard/var_unserializer.re Fri Sep 20 09:07:25 2002 @@ -144,31 +144,29 @@ if (!php_var_unserialize(&key, p, max, NULL TSRMLS_CC)) { zval_dtor(key); FREE_ZVAL(key); - return 0; - } - - ALLOC_INIT_ZVAL(data); - - if (!php_var_unserialize(&data, p, max, var_hash TSRMLS_CC)) { - zval_dtor(key); - FREE_ZVAL(key); - zval_dtor(data); - FREE_ZVAL(data); - return 0; - } + } else { + ALLOC_INIT_ZVAL(data); - switch (Z_TYPE_P(key)) { - case IS_LONG: - zend_hash_index_update(ht, Z_LVAL_P(key), &data, sizeof(data), NULL); - break; - case IS_STRING: - zend_hash_update(ht, Z_STRVAL_P(key), Z_STRLEN_P(key) + 1, &data, sizeof(data), NULL); - break; + if (!php_var_unserialize(&data, p, max, var_hash TSRMLS_CC)) { + zval_dtor(key); + FREE_ZVAL(key); + zval_dtor(data); + FREE_ZVAL(data); + } else { + switch (Z_TYPE_P(key)) { + case IS_LONG: + zend_hash_index_update(ht, +Z_LVAL_P(key), &data, sizeof(data), NULL); + break; + case IS_STRING: + zend_hash_update(ht, Z_STRVAL_P(key), +Z_STRLEN_P(key) + 1, &data, sizeof(data), NULL); + break; - } + } - zval_dtor(key); - FREE_ZVAL(key); + zval_dtor(key); + FREE_ZVAL(key); + } + } } return 1; Previous Comments: ------------------------------------------------------------------------ [2002-09-19 13:17:02] [EMAIL PROTECTED] No C++ comment styles allowed! Can you fix that and resubmit the patch? ------------------------------------------------------------------------ [2002-09-19 09:44:58] [EMAIL PROTECTED] There is the diff: --- ext/standard/var_unserializer.re.old Thu Sep 19 16:34:32 2002 +++ ext/standard/var_unserializer.re Thu Sep 19 11:20:08 2002 @@ -144,31 +144,29 @@ if (!php_var_unserialize(&key, p, max, NULL TSRMLS_CC)) { zval_dtor(key); FREE_ZVAL(key); - return 0; - } - - ALLOC_INIT_ZVAL(data); - - if (!php_var_unserialize(&data, p, max, var_hash TSRMLS_CC)) { - zval_dtor(key); - FREE_ZVAL(key); - zval_dtor(data); - FREE_ZVAL(data); - return 0; - } + } else { + ALLOC_INIT_ZVAL(data); - switch (Z_TYPE_P(key)) { - case IS_LONG: - zend_hash_index_update(ht, Z_LVAL_P(key), &data, sizeof(data), NULL); - break; - case IS_STRING: - zend_hash_update(ht, Z_STRVAL_P(key), Z_STRLEN_P(key) + 1, &data, sizeof(data), NULL); - break; + if (!php_var_unserialize(&data, p, max, var_hash TSRMLS_CC)) { + zval_dtor(key); + FREE_ZVAL(key); + zval_dtor(data); + FREE_ZVAL(data); + } else { + switch (Z_TYPE_P(key)) { + case IS_LONG: + zend_hash_index_update(ht, +Z_LVAL_P(key), &data, sizeof(data), NULL); + break; + case IS_STRING: + zend_hash_update(ht, Z_STRVAL_P(key), +Z_STRLEN_P(key) + 1, &data, sizeof(data), NULL); + break; - } + } - zval_dtor(key); - FREE_ZVAL(key); + zval_dtor(key); + FREE_ZVAL(key); + } + } } return 1; @@ -396,9 +394,9 @@ } "}" { - /* this is the case where we have less data than planned */ + // this is the case where we have less data than planned zend_error(E_NOTICE, "Unexpected end of serialized data"); - return 0; /* not sure if it should be 0 or 1 here? */ + return 0; // not sure if it should be 0 or 1 here? } any { return 0; } ------------------------------------------------------------------------ [2002-09-19 09:05:16] [EMAIL PROTECTED] Please supply a unified diff here. ------------------------------------------------------------------------ [2002-09-19 04:53:25] [EMAIL PROTECTED] I've changed thd category ... ------------------------------------------------------------------------ [2002-09-19 04:38:19] [EMAIL PROTECTED] There was an error in older versions (4.1.0) of the serializer code (it's possible the error is in the current code, too): some times, the element-count for objects was to hight. The old deserializer-code seams to ignore this problem. The new code has problems with that issue. We have many huge databases with serialized structures at some points. Thats why, we must use the old data :-( I've fixed the problem with a little change in ext/standard/var_unserializer.re, function process_nested_data : static inline int process_nested_data(UNSERIALIZE_PARAMETER, HashTable *ht, int elements) { while (elements-- > 0) { zval *key, *data; ALLOC_INIT_ZVAL(key); if (!php_var_unserialize(&key, p, max, NULL TSRMLS_CC)) { zval_dtor(key); FREE_ZVAL(key); } else { ALLOC_INIT_ZVAL(data); if (!php_var_unserialize(&data, p, max, var_hash TSRMLS_CC)) { zval_dtor(key); FREE_ZVAL(key); zval_dtor(data); FREE_ZVAL(data); } else { switch (Z_TYPE_P(key)) { case IS_LONG: zend_hash_index_update(ht, Z_LVAL_P(key), &data, sizeof(data), NULL); break; case IS_STRING: zend_hash_update(ht, Z_STRVAL_P(key), Z_STRLEN_P(key) + 1, &data, sizeof(data), NULL); break; } zval_dtor(key); FREE_ZVAL(key); } } } return 1; } Now, it doesn't have a problem with to hight element counters. It would be great, if the developer of the new deserializer-code could look at this change and eventually integrate it into the CVS-tree. Greetings from Berlin, Matthias ------------------------------------------------------------------------ -- Edit this bug report at http://bugs.php.net/?id=19493&edit=1