fixed utf8 detection + made feature experimental

8 files changed, 34 insertions, 20 deletions

diff --git a/php_suhosin.h b/php_suhosin.h
index d567877..df7c877 100644
--- a/php_suhosin.h
+++ b/php_suhosin.h
@@ -250,7 +250,9 @@ ZEND_BEGIN_MODULE_GLOBALS(suhosin)
         zend_bool  upload_disallow_elf;
         zend_bool  upload_disallow_binary;
         zend_bool  upload_remove_binary;
+#ifdef SUHOSIN_EXPERIMENTAL
         zend_bool  upload_allow_utf8;
+#endif
         char *upload_verification_script;
         
         zend_bool  no_more_variables;
diff --git a/suhosin.c b/suhosin.c
index fc84a94..8d25acc 100644
--- a/suhosin.c
+++ b/suhosin.c
@@ -824,7 +824,9 @@ PHP_INI_BEGIN()
         STD_PHP_INI_ENTRY("suhosin.upload.disallow_elf", "1", PHP_INI_SYSTEM|PHP_INI_PERDIR, OnUpdateUploadBool, upload_disallow_elf, zend_suhosin_globals, suhosin_globals)
         STD_PHP_INI_ENTRY("suhosin.upload.disallow_binary", "0", PHP_INI_SYSTEM|PHP_INI_PERDIR, OnUpdateUploadBool, upload_disallow_binary, zend_suhosin_globals, suhosin_globals)
         STD_PHP_INI_ENTRY("suhosin.upload.remove_binary", "0", PHP_INI_SYSTEM|PHP_INI_PERDIR, OnUpdateUploadBool, upload_remove_binary, zend_suhosin_globals, suhosin_globals)
-        STD_PHP_INI_ENTRY("suhosin.upload.allow_utf8", "0", PHP_INI_SYSTEM|PHP_INI_PERDIR, OnUpdateUploadBool, upload_allow_utf8, zend_suhosin_globals, suhosin_globals)
+#ifdef SUHOSIN_EXPERIMENTAL
+        STD_PHP_INI_BOOLEAN("suhosin.upload.allow_utf8", "0", PHP_INI_SYSTEM|PHP_INI_PERDIR, OnUpdateUploadBool, upload_allow_utf8, zend_suhosin_globals, suhosin_globals)
+#endif
         STD_PHP_INI_ENTRY("suhosin.upload.verification_script", NULL, PHP_INI_SYSTEM|PHP_INI_PERDIR, OnUpdateUploadString, upload_verification_script, zend_suhosin_globals, suhosin_globals)
 
 
diff --git a/suhosin.ini b/suhosin.ini
index 3ae8ff5..311bca7 100644
--- a/suhosin.ini
+++ b/suhosin.ini
@@ -105,9 +105,6 @@
 ; | LOG_LOCAL7   | 31    |
 ; +--------------+-------+
 ; 
-; Using constant names is only supported with the Suhosin-Patch. If in doubt, use
-; the numeric value.
-; 
 ;suhosin.log.syslog.facility = LOG_USER
 ;
 
@@ -141,9 +138,6 @@
 ; |LOG_ERR     | 7     |
 ; +------------+-------+
 ; 
-; Using constant names is only supported with the Suhosin-Patch. If in doubt, use
-; the numeric value.
-; 
 ;suhosin.log.syslog.priority = LOG_ALERT
 ;
 
@@ -1325,8 +1319,8 @@
 ; * Type: Boolean
 ; * Default: Off
 ; 
-; This option allows UTF-8 along with ASCII when using
-; `suhosin.upload.disallow_binary` or `suhosin.upload.remove_binary`.
+; This is an experimental feature. This option allows UTF-8 along with ASCII when
+; using `suhosin.upload.disallow_binary` or `suhosin.upload.remove_binary`.
 ; 
 ;suhosin.upload.allow_utf8 = Off
 ;
diff --git a/tests/filter/suhosin_upload_disallow_binary_utf8.phpt b/tests/filter/suhosin_upload_disallow_binary_utf8.phpt
index 557a8d5..b4627c7 100644
--- a/tests/filter/suhosin_upload_disallow_binary_utf8.phpt
+++ b/tests/filter/suhosin_upload_disallow_binary_utf8.phpt
@@ -11,7 +11,9 @@ suhosin.upload.allow_utf8=On
 max_file_uploads=40
 suhosin.upload.max_uploads=40
 --SKIPIF--
-<?php include('skipif.inc'); ?>
+<?php include('skipif.inc');
+if (ini_get('suhosin.upload.allow_utf8') === FALSE) { die("skip feature not compiled in"); }
+?>
 --COOKIE--
 --GET--
 --POST_RAW--
diff --git a/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt b/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt
index 413d25a..0bf122b 100644
--- a/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt
+++ b/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt
@@ -11,7 +11,9 @@ suhosin.upload.allow_utf8=Off
 max_file_uploads=40
 suhosin.upload.max_uploads=40
 --SKIPIF--
-<?php include('skipif.inc'); ?>
+<?php include('skipif.inc');
+if (ini_get('suhosin.upload.allow_utf8') === FALSE) { die("skip feature not compiled in"); }
+?>
 --COOKIE--
 --GET--
 --POST_RAW--
diff --git a/tests/filter/suhosin_upload_remove_binary_utf8.phpt b/tests/filter/suhosin_upload_remove_binary_utf8.phpt
index 6fbd240..1920dc5 100644
--- a/tests/filter/suhosin_upload_remove_binary_utf8.phpt
+++ b/tests/filter/suhosin_upload_remove_binary_utf8.phpt
@@ -12,7 +12,9 @@ suhosin.upload.allow_utf8=On
 max_file_uploads=40
 suhosin.upload.max_uploads=40
 --SKIPIF--
-<?php include('skipif.inc'); ?>
+<?php include('skipif.inc');
+if (ini_get('suhosin.upload.allow_utf8') === FALSE) { die("skip feature not compiled in"); }
+?>
 --COOKIE--
 --GET--
 --POST_RAW--
diff --git a/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt b/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt
index 5c31115..56de404 100644
--- a/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt
+++ b/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt
@@ -12,7 +12,9 @@ suhosin.upload.allow_utf8=Off
 max_file_uploads=40
 suhosin.upload.max_uploads=40
 --SKIPIF--
-<?php include('skipif.inc'); ?>
+<?php include('skipif.inc');
+if (ini_get('suhosin.upload.allow_utf8') === FALSE) { die("skip feature not compiled in"); }
+?>
 --COOKIE--
 --GET--
 --POST_RAW--
diff --git a/ufilter.c b/ufilter.c
index 1669e88..f3c3054 100644
--- a/ufilter.c
+++ b/ufilter.c
@@ -149,19 +149,23 @@ return_failure:
 }
 /* }}} */
 
-static inline int suhosin_validate_utf8_multibyte(const char* cp)
+#ifdef SUHOSIN_EXPERIMENTAL
+static inline int suhosin_validate_utf8_multibyte(const char* cp, size_t maxlen)
 {
+        if (maxlen < 2 || !(*cp & 0x80)) { return 0; }
         if ((*cp & 0xe0) == 0xc0 &&                                     // 1st byte is 110xxxxx
                 (*(cp+1) & 0xc0) == 0x80 &&                             // 2nd byte is 10xxxxxx
                 (*cp & 0x1e)) {                                                 // overlong check 110[xxxx]x 10xxxxxx
                          return 2;
         }
+        if (maxlen < 3) { return 0; }
         if ((*cp & 0xf0) == 0xe0 &&                                     // 1st byte is 1110xxxx
                 (*(cp+1) & 0xc0) == 0x80 &&                             // 2nd byte is 10xxxxxx
                 (*(cp+2) & 0xc0) == 0x80 &&                             // 3rd byte is 10xxxxxx
                 ((*cp & 0x0f) | (*(cp+1) & 0x20))) {    // 1110[xxxx] 10[x]xxxxx 10xxxxxx
                         return 3;
         }
+        if (maxlen < 4) { return 0; }
         if ((*cp & 0xf8) == 0xf0 &&                             // 1st byte is 11110xxx
                 (*(cp+1) & 0xc0) == 0x80 &&                             // 2nd byte is 10xxxxxx
                 (*(cp+2) & 0xc0) == 0x80 &&                             // 3rd byte is 10xxxxxx
@@ -171,6 +175,7 @@ static inline int suhosin_validate_utf8_multibyte(const char* cp)
         }
         return 0;
 }
+#endif
 
 int suhosin_rfc1867_filter(unsigned int event, void *event_data, void **extra TSRMLS_DC)
 {
@@ -236,14 +241,15 @@ int suhosin_rfc1867_filter(unsigned int event, void *event_data, void **extra TS
                                         if (*cp >= 32 || isspace(*cp)) {
                                                 continue;
                                         }
+#ifdef SUHOSIN_EXPERIMENTAL
                                         if ((*cp & 0x80) && SUHOSIN_G(upload_allow_utf8)) {
                                                 SDEBUG("checking char %x", *cp);
-                                                if ((n = suhosin_validate_utf8_multibyte(cp))) { // valid UTF8 multibyte character
+                                                if ((n = suhosin_validate_utf8_multibyte(cp, cpend-cp))) { // valid UTF8 multibyte character
                                                         cp += n - 1;
                                                         continue;
                                                 }
                                         }
-                                        
+#endif
                                         suhosin_log(S_FILES, "uploaded file contains binary data - file dropped");
                                         if (!SUHOSIN_G(simulation)) {
                                                 goto continue_with_failure;
@@ -261,15 +267,17 @@ int suhosin_rfc1867_filter(unsigned int event, void *event_data, void **extra TS
                                 for (i=0, j=0; i<mefd->length; i++) {
                                         if (mefd->data[i] >= 32 || isspace(mefd->data[i])) {
                                                 mefd->data[j++] = mefd->data[i];
-                                        } else if (SUHOSIN_G(upload_allow_utf8) && mefd->data[i] & 0x80) {
-                                                n = suhosin_validate_utf8_multibyte(mefd->data + i);
+                                        }
+#ifdef SUHOSIN_EXPERIMENTAL
+                                        else if (SUHOSIN_G(upload_allow_utf8) && mefd->data[i] & 0x80) {
+                                                n = suhosin_validate_utf8_multibyte(mefd->data + i, mefd->length - i);
                                                 if (!n) { continue; }
-                                                while (n) {
+                                                while (n--) {
                                                         mefd->data[j++] = mefd->data[i++];
-                                                        n--;
                                                 }
                                                 i--;
                                         }
+#endif
                                 }
                                 mefd->data[j] = '\0';