9 files changed, 233 insertions, 32 deletions

diff --git a/ifilter.c b/ifilter.c
index cd02869..8b2e8a3 100644
--- a/ifilter.c
+++ b/ifilter.c
@@ -366,7 +366,7 @@ unsigned int suhosin_input_filter(int arg, char *var, char **val, unsigned int v
         }
         
         /* Drop this variable if it begins with whitespace which is disallowed */
-        if (*var == ' ') {
+        if (isspace(*var)) {
                 if (SUHOSIN_G(disallow_ws)) {
                         suhosin_log(S_VARS, "request variable name begins with disallowed whitespace - dropped variable '%s'", var);
                         if (!SUHOSIN_G(simulation)) {
diff --git a/php_suhosin.h b/php_suhosin.h
index e85fc38..4b460e4 100644
--- a/php_suhosin.h
+++ b/php_suhosin.h
@@ -151,6 +151,7 @@ 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 469e88f..f57e85a 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
new file mode 100644
index 0000000..557a8d5
--- /dev/null
+++ b/tests/filter/suhosin_upload_disallow_binary_utf8.phpt
@@ -0,0 +1,44 @@
+--TEST--
+Testing: suhosin.upload.disallow_binary=On with UTF-8
+--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=On
+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($_FILES);
+?>
+--EXPECTF--
+array(1) {
+  ["test"]=>
+  array(5) {
+    ["name"]=>
+    string(4) "test"
+    ["type"]=>
+    string(0) ""
+    ["tmp_name"]=>
+    string(%d) "%s"
+    ["error"]=>
+    int(0)
+    ["size"]=>
+    int(17)
+  }
+}
diff --git a/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt b/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt
new file mode 100644
index 0000000..413d25a
--- /dev/null
+++ b/tests/filter/suhosin_upload_disallow_binary_utf8fail.phpt
@@ -0,0 +1,45 @@
+--TEST--
+Testing: suhosin.upload.disallow_binary=On with UTF-8 and allow_utf8=Off
+--INI--
+suhosin.log.syslog=0
+suhosin.log.sapi=0
+suhosin.log.stdout=255
+suhosin.log.script=0
+file_uploads=1
+suhosin.upload.disallow_binary=On
+suhosin.upload.allow_utf8=Off
+max_file_uploads=40
+suhosin.upload.max_uploads=40
+--SKIPIF--
+<?php include('skipif.inc'); ?>
+--COOKIE--
+--GET--
+--POST_RAW--
+Content-Type: multipart/form-data; boundary=bound
+--bound
+Content-Disposition: form-data; name="test"; filename="test"
+
+Spaß am Gerät!
+
+--bound--
+--FILE--
+<?php
+var_dump($_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.phpt b/tests/filter/suhosin_upload_remove_binary.phpt
new file mode 100644
index 0000000..f4337d9
--- /dev/null
+++ b/tests/filter/suhosin_upload_remove_binary.phpt
diff --git a/tests/filter/suhosin_upload_remove_binary_utf8.phpt b/tests/filter/suhosin_upload_remove_binary_utf8.phpt
new file mode 100644
index 0000000..6fbd240
--- /dev/null
+++ b/tests/filter/suhosin_upload_remove_binary_utf8.phpt
@@ -0,0 +1,32 @@
+--TEST--
+Testing: suhosin.upload.remove_binary=On with UTF-8
+--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=On
+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($_FILES['test']['tmp_name']));
+?>
+--EXPECTF--
+string(17) "Spaß am Gerät!
+"
\ No newline at end of file
diff --git a/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt b/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt
new file mode 100644
index 0000000..5c31115
--- /dev/null
+++ b/tests/filter/suhosin_upload_remove_binary_utf8fail.phpt
@@ -0,0 +1,32 @@
+--TEST--
+Testing: suhosin.upload.remove_binary=On with UTF-8 and allow_utf8=Off
+--INI--
+suhosin.log.syslog=0
+suhosin.log.sapi=0
+suhosin.log.stdout=255
+suhosin.log.script=0
+file_uploads=1
+suhosin.upload.disallow_binary=Off
+suhosin.upload.remove_binary=On
+suhosin.upload.allow_utf8=Off
+max_file_uploads=40
+suhosin.upload.max_uploads=40
+--SKIPIF--
+<?php include('skipif.inc'); ?>
+--COOKIE--
+--GET--
+--POST_RAW--
+Content-Type: multipart/form-data; boundary=bound
+--bound
+Content-Disposition: form-data; name="test"; filename="test"
+
+Spaß am Gerät!
+
+--bound--
+--FILE--
+<?php
+var_dump(file_get_contents($_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 5a85b54..6d9669f 100644
--- a/ufilter.c
+++ b/ufilter.c
@@ -197,6 +197,29 @@ return_failure:
 }
 /* }}} */
 
+static inline int suhosin_validate_utf8_multibyte(const char* cp)
+{
+        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 ((*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 ((*cp & 0xf8) == 0xf0 &&                             // 1st byte is 11110xxx
+                (*(cp+1) & 0xc0) == 0x80 &&                             // 2nd byte is 10xxxxxx
+                (*(cp+2) & 0xc0) == 0x80 &&                             // 3rd byte is 10xxxxxx
+                (*(cp+3) & 0xc0) == 0x80 &&                             // 4th byte is 10xxxxxx
+                ((*cp & 0x07) | (*(cp+1) & 0x30))) {    // 11110[xxx] 10[xx]xxxx 10xxxxxx 10xxxxxx
+                        return 4;
+        }
+        return 0;
+}
+
 int suhosin_rfc1867_filter(unsigned int event, void *event_data, void **extra TSRMLS_DC)
 {
         int retval = SUCCESS;
@@ -250,38 +273,61 @@ int suhosin_rfc1867_filter(unsigned int event, void *event_data, void **extra TS
                             }
                     }
                     
-                    if (SUHOSIN_G(upload_disallow_binary)) {
-                    
-                            multipart_event_file_data *mefd = (multipart_event_file_data *) event_data;
-                            size_t i;
-                            
-                            for (i=0; i<mefd->length; i++) {
-                                    if (mefd->data[i] < 32 && !isspace(mefd->data[i])) {
-                                            suhosin_log(S_FILES, "uploaded file contains binary data - file dropped");
-                                            if (!SUHOSIN_G(simulation)) {
-                                                    goto continue_with_failure;
-                                            }
-                                    }
-                            }
-                    }
+                        if (SUHOSIN_G(upload_disallow_binary)) {
+                        
+                                multipart_event_file_data *mefd = (multipart_event_file_data *) event_data;
 
-                    if (SUHOSIN_G(upload_remove_binary)) {
-                    
-                            multipart_event_file_data *mefd = (multipart_event_file_data *) event_data;
-                            size_t i, j;
-                            
-                            for (i=0, j=0; i<mefd->length; i++) {
-                                    if (mefd->data[i] >= 32 || isspace(mefd->data[i])) {
-                                            mefd->data[j++] = mefd->data[i];
-                                    }
-                            }
-                            SDEBUG("removing binary %u %u",i,j);
-                            /* IMPORTANT FOR DAISY CHAINING */
-                            mefd->length = j;
-                            if (mefd->newlength) {
-                                    *mefd->newlength = j;
-                            }
-                    }
+                                char *cp, *cpend;
+                                int n;
+                                cpend = mefd->data + mefd->length;
+                                for (char *cp = mefd->data; cp < cpend; cp++) {
+                                        if (*cp >= 32 || isspace(*cp)) {
+                                                continue;
+                                        }
+                                        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;
+                                                        continue;
+                                                }
+                                        }
+                                        
+                                        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)) {
+                        
+                                multipart_event_file_data *mefd = (multipart_event_file_data *) event_data;
+                                size_t i, j;
+                                int n;
+                                
+                                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);
+                                                if (!n) { continue; }
+                                                while (n) {
+                                                        mefd->data[j++] = mefd->data[i++];
+                                                        n--;
+                                                }
+                                                i--;
+                                        }
+                                }
+                                mefd->data[j] = '\0';
+                                
+                                SDEBUG("removing binary %zu %zu",i,j);
+                                /* IMPORTANT FOR DAISY CHAINING */
+                                mefd->length = j;
+                                if (mefd->newlength) {
+                                        *mefd->newlength = j;
+                                }
+                        }
                     
                     break;