introduced suhosin.upload.allow_utf8

7 files changed, 88 insertions, 11 deletions

diff --git a/php_suhosin.h b/php_suhosin.h
index e5604a7..6d2ea80 100644
--- a/php_suhosin.h
+++ b/php_suhosin.h
@@ -150,6 +150,7 @@ ZEND_BEGIN_MODULE_GLOBALS(suhosin)
         zend_bool  upload_disallow_elf;
         zend_bool  upload_disallow_binary;
         zend_bool  upload_remove_binary;
+        zend_bool  upload_allow_utf8;
         char *upload_verification_script;
         
         zend_bool  no_more_variables;
diff --git a/suhosin.c b/suhosin.c
index 0d1eba0..f2533cb 100644
--- a/suhosin.c
+++ b/suhosin.c
@@ -986,6 +986,7 @@ 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)
         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/tests/filter/suhosin_upload_disallow_binary_utf8.phpt b/tests/filter/suhosin_upload_disallow_binary_utf8.phpt
index 4661dc9..557a8d5 100644
--- a/tests/filter/suhosin_upload_disallow_binary_utf8.phpt
+++ b/tests/filter/suhosin_upload_disallow_binary_utf8.phpt
@@ -7,6 +7,7 @@ suhosin.log.stdout=255
 suhosin.log.script=0
 file_uploads=1
 suhosin.upload.disallow_binary=On
+suhosin.upload.allow_utf8=On
 max_file_uploads=40
 suhosin.upload.max_uploads=40
 --SKIPIF--
diff --git a/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt b/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt
new file mode 100644
index 0000000..413d25a
--- /dev/null
+++ b/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt
@@ -0,0 +1,45 @@
+--TEST--
+Testing: suhosin.upload.disallow_binary=On with UTF-8 and allow_utf8=Off
+--INI--
+suhosin.log.syslog=0
+suhosin.log.sapi=0
+suhosin.log.stdout=255
+suhosin.log.script=0
+file_uploads=1
+suhosin.upload.disallow_binary=On
+suhosin.upload.allow_utf8=Off
+max_file_uploads=40
+suhosin.upload.max_uploads=40
+--SKIPIF--
+<?php include('skipif.inc'); ?>
+--COOKIE--
+--GET--
+--POST_RAW--
+Content-Type: multipart/form-data; boundary=bound
+--bound
+Content-Disposition: form-data; name="test"; filename="test"
+
+Spaß am Gerät!
+
+--bound--
+--FILE--
+<?php
+var_dump($_FILES);
+?>
+--EXPECTF--
+array(1) {
+  ["test"]=>
+  array(5) {
+    ["name"]=>
+    string(4) "test"
+    ["type"]=>
+    string(0) ""
+    ["tmp_name"]=>
+    string(0) ""
+    ["error"]=>
+    int(8)
+    ["size"]=>
+    int(0)
+  }
+}
+ALERT - uploaded file contains binary data - file dropped (attacker 'REMOTE_ADDR not set', file '%s')
diff --git a/tests/filter/suhosin_upload_remove_binary_utf8.phpt b/tests/filter/suhosin_upload_remove_binary_utf8.phpt
index 2d10eaa..6fbd240 100644
--- a/tests/filter/suhosin_upload_remove_binary_utf8.phpt
+++ b/tests/filter/suhosin_upload_remove_binary_utf8.phpt
@@ -8,6 +8,7 @@ suhosin.log.script=0
 file_uploads=1
 suhosin.upload.disallow_binary=Off
 suhosin.upload.remove_binary=On
+suhosin.upload.allow_utf8=On
 max_file_uploads=40
 suhosin.upload.max_uploads=40
 --SKIPIF--
diff --git a/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt b/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt
new file mode 100644
index 0000000..5c31115
--- /dev/null
+++ b/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt
@@ -0,0 +1,32 @@
+--TEST--
+Testing: suhosin.upload.remove_binary=On with UTF-8 and allow_utf8=Off
+--INI--
+suhosin.log.syslog=0
+suhosin.log.sapi=0
+suhosin.log.stdout=255
+suhosin.log.script=0
+file_uploads=1
+suhosin.upload.disallow_binary=Off
+suhosin.upload.remove_binary=On
+suhosin.upload.allow_utf8=Off
+max_file_uploads=40
+suhosin.upload.max_uploads=40
+--SKIPIF--
+<?php include('skipif.inc'); ?>
+--COOKIE--
+--GET--
+--POST_RAW--
+Content-Type: multipart/form-data; boundary=bound
+--bound
+Content-Disposition: form-data; name="test"; filename="test"
+
+Spaß am Gerät!
+
+--bound--
+--FILE--
+<?php
+var_dump(file_get_contents($_FILES['test']['tmp_name']));
+?>
+--EXPECTF--
+string(13) "Spa am Gert!
+"
\ No newline at end of file
diff --git a/ufilter.c b/ufilter.c
index 2eeed56..6d9669f 100644
--- a/ufilter.c
+++ b/ufilter.c
@@ -281,10 +281,10 @@ int suhosin_rfc1867_filter(unsigned int event, void *event_data, void **extra TS
                                 int n;
                                 cpend = mefd->data + mefd->length;
                                 for (char *cp = mefd->data; cp < cpend; cp++) {
-                                        if (*cp >= 32) {
+                                        if (*cp >= 32 || isspace(*cp)) {
                                                 continue;
                                         }
-                                        if (*cp & 0x80) {
+                                        if ((*cp & 0x80) && SUHOSIN_G(upload_allow_utf8)) {
                                                 SDEBUG("checking char %x", *cp);
                                                 if ((n = suhosin_validate_utf8_multibyte(cp))) { // valid UTF8 multibyte character
                                                         cp += n - 1;
@@ -292,16 +292,12 @@ int suhosin_rfc1867_filter(unsigned int event, void *event_data, void **extra TS
                                                 }
                                         }
                                         
-                                        if (!isspace(*cp)) {
-                                                suhosin_log(S_FILES, "uploaded file contains binary data - file dropped");
-                                                if (!SUHOSIN_G(simulation)) {
-                                                        goto continue_with_failure;
-                                                }
-                                                break;
+                                        suhosin_log(S_FILES, "uploaded file contains binary data - file dropped");
+                                        if (!SUHOSIN_G(simulation)) {
+                                                goto continue_with_failure;
                                         }
-
+                                        break;
                                 }
-                                
                         }
 
                         if (SUHOSIN_G(upload_remove_binary)) {
@@ -313,7 +309,7 @@ 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 (mefd->data[i] & 0x80) {
+                                        } else if (SUHOSIN_G(upload_allow_utf8) && mefd->data[i] & 0x80) {
                                                 n = suhosin_validate_utf8_multibyte(mefd->data + i);
                                                 if (!n) { continue; }
                                                 while (n) {