Reimplement -s option to fix unACKed data

Old code used to reduce packet size by one byte to remove space in
HTTP Host header. This introduces one unACKed byte which OS later
tried to send to the host. This byte was \n (the last byte in original
packet) which broke POST requests.

New code in this commit moves "stolen" space in the end of User-Agent
header value and do not reduce packet size anymore.
User-Agent value is used because not all web servers are compatible
with additional space in the end of Host value.

Fix #3

README.md

@@ -18,14 +18,17 @@ Usage: goodbyedpi.exe [OPTION...]
  -s          remove space between host header and its value
  -f [value]  set HTTP fragmentation to value
  -e [value]  set HTTPS fragmentation to value
+ -a          additional space between Method and Request-URI (enables -s, may break sites)
 
- -1          enables all options, -f 2 -e 2 (most compatible mode, default)
+ -1          -p -r -s -f 2 -e 2 (most compatible mode, default)
- -2          enables all options, -f 2 -e 40 (better speed yet still compatible)
+ -2          -p -r -s -f 2 -e 40 (better speed yet still compatible)
- -3          all options except HTTP fragmentation, -e 40 (even better speed)
+ -3          -p -r -s -e 40 (even better speed)
- -4          all options except fragmentation (best speed)
+ -4          -p -r -s (best speed)
 ```
 
-Try to run `goodbyedpi.exe` without any arguments first. If you can open blocked websites it means your ISP has DPI which can be circumvented. This is the slowest mode.
+Try to run `goodbyedpi.exe -1 -a` first. If you can open blocked websites it means your ISP has DPI which can be circumvented. This is the slowest and prone to break websites mode, but suitable for most DPI.
+
+Try `-1` to see if it works too.
 
 Then try `goodbyedpi.exe -2`. It should be faster for HTTPS sites. Mode `-3` speed ups HTTP websites.
 
@@ -35,17 +38,18 @@ Use `goodbyedpi.exe -4` if it works for your ISP's DPI. This is the fastest mode
 
 ### Passive DPI
 
-Most Passive DPI send HTTP 301 Redirect if you try to access blocked website over HTTP and TCP Reset in case of HTTPS, faster than destination website. Packets sent by DPI have always have IP Identification field equal to `0x0000` or `0x0001`, as seen with Russian providers. These packets are blocked by GoodbyeDPI.
+Most Passive DPI send HTTP 302 Redirect if you try to access blocked website over HTTP and TCP Reset in case of HTTPS, faster than destination website. Packets sent by DPI have always have IP Identification field equal to `0x0000` or `0x0001`, as seen with Russian providers. These packets, if they redirect you to another website (censorship page), are blocked by GoodbyeDPI.
 
 ### Active DPI
 
-Active DPI is more tricky to fool. Currently the software uses 3 methods to circumvent Active DPI:
+Active DPI is more tricky to fool. Currently the software uses 4 methods to circumvent Active DPI:
 
 * TCP-level fragmentation for first data packet
 * Replacing `Host` header with `hoSt`
 * Removing space between header name and value in `Host` header
+* Adding additional space between HTTP Method (GET, POST etc) and URI
 
-These methods do not break any website as are fully compatible with TCP and HTTP standards, yet it's sufficient to prevent DPI data classification and to circumvent censorship.
+These methods should not break any website as are fully compatible with TCP and HTTP standards, yet it's sufficient to prevent DPI data classification and to circumvent censorship. Additional space may break some websites, although it's acceptable by HTTP/1.1 specification (see 19.3 Tolerant Applications).
 
 The program loads WinDivert driver which uses Windows Filtering Platform to set filters and redirect packets to the userspace. It's running as long as console window is visible and terminates when you close the window.
 

goodbyedpi.c

@@ -27,7 +27,26 @@ static const char *http10_redirect_302 = "HTTP/1.0 302 ";
 static const char *http11_redirect_302 = "HTTP/1.1 302 ";
 static const char *http_host_find = "\r\nHost: ";
 static const char *http_host_replace = "\r\nhoSt: ";
+static const char *http_useragent_find = "\r\nUser-Agent: ";
 static const char *location_http = "\r\nLocation: http://";
+static const char *http_methods[] = {
+    "GET ",
+    "HEAD ",
+    "POST ",
+    "PUT ",
+    "DELETE ",
+    "CONNECT ",
+    "OPTIONS ",
+    "TRACE ",
+    "PATCH ",
+    "PROPFIND ",
+    "PROPPATCH ",
+    "MKCOL ",
+    "COPY ",
+    "MOVE ",
+    "LOCK ",
+    "UNLOCK ",
+};
 
 static char* dumb_memmem(const char* haystack, int hlen, const char* needle, int nlen) {
     // naive implementation
@@ -86,10 +105,27 @@ static PVOID find_host_header(const char *pktdata, int pktlen) {
                 http_host_find, strlen(http_host_find));
 }
 
-static void change_window_size(char *pkt, int size) {
+/* Finds User-Agent header with \r\n before it */
+static PVOID find_useragent_header(const char *pktdata, int pktlen) {
+    return dumb_memmem(pktdata, pktlen,
+                http_useragent_find, strlen(http_useragent_find));
+}
+
+static void change_window_size(const char *pkt, int size) {
     *(uint16_t*)(pkt + IPV4_HDR_LEN + TCP_WINDOWSIZE_OFFSET) = htons(size);
 }
 
+/* HTTP method end without trailing space */
+static PVOID find_http_method_end(const char *pkt) {
+    int i;
+    for (i = 0; i<(sizeof(http_methods) / sizeof(*http_methods)); i++) {
+        if (memcmp(pkt, http_methods[i], strlen(http_methods[i])) == 0) {
+            return (char*)pkt+strlen(http_methods[i]) - 1;
+        }
+    }
+    return NULL;
+}
+
 int main(int argc, char *argv[]) {
     static const char fragment_size_message[] =
                 "Fragment size should be in range [0 - 65535]\n";
@@ -106,11 +142,11 @@ int main(int argc, char *argv[]) {
 
     int do_passivedpi = 0, do_fragment_http = 0,
         do_fragment_https = 0, do_host = 0,
-        do_host_removespace = 0;
+        do_host_removespace = 0, do_additional_space = 0;
     int http_fragment_size = 2;
     int https_fragment_size = 2;
-    char *data_addr, *data_addr_rn, *host_addr;
+    char *data_addr, *data_addr_rn, *host_addr, *useragent_addr, *method_addr;
-    int host_len, fromhost_uptoend_len;
+    int data_len, host_len;
 
     printf("GoodbyeDPI: Passive DPI blocker and Active DPI circumvention utility\n");
 
@@ -120,7 +156,7 @@ int main(int argc, char *argv[]) {
             = do_fragment_http = do_fragment_https = 1;
     }
 
-    while ((opt = getopt(argc, argv, "1234prsf:e:")) != -1) {
+    while ((opt = getopt(argc, argv, "1234prsaf:e:")) != -1) {
         switch (opt) {
             case '1':
                 do_passivedpi = do_host = do_host_removespace \
@@ -147,6 +183,10 @@ int main(int argc, char *argv[]) {
             case 's':
                 do_host_removespace = 1;
                 break;
+            case 'a':
+                do_additional_space = 1;
+                do_host_removespace = 1;
+                break;
             case 'f':
                 do_fragment_http = 1;
                 http_fragment_size = atoi(optarg);
@@ -168,22 +208,23 @@ int main(int argc, char *argv[]) {
                 " -p          block passive DPI\n"
                 " -r          replace Host with hoSt\n"
                 " -s          remove space between host header and its value\n"
+                " -a          additional space between Method and Request-URI (enables -s, may break sites)\n"
                 " -f [value]  set HTTP fragmentation to value\n"
                 " -e [value]  set HTTPS fragmentation to value\n"
                 "\n"
-                " -1          enables all options, -f 2 -e 2 (most compatible mode, default)\n"
+                " -1          -p -r -s -f 2 -e 2 (most compatible mode, default)\n"
-                " -2          enables all options, -f 2 -e 40 (better speed yet still compatible)\n"
+                " -2          -p -r -s -f 2 -e 40 (better speed yet still compatible)\n"
-                " -3          all options except HTTP fragmentation, -e 40 (even better speed)\n"
+                " -3          -p -r -s -e 40 (even better speed)\n"
-                " -4          all options except fragmentation (best speed)\n");
+                " -4          -p -r -s (best speed)\n");
                 exit(EXIT_FAILURE);
         }
     }
 
     printf("Block passive: %d, Fragment HTTP: %d, Fragment HTTPS: %d, "
-           "hoSt: %d, Host no space: %d\n",
+           "hoSt: %d, Host no space: %d, Additional space: %d\n",
            do_passivedpi, (do_fragment_http ? http_fragment_size : 0),
            (do_fragment_https ? https_fragment_size : 0),
-           do_host, do_host_removespace);
+           do_host, do_host_removespace, do_additional_space);
 
     printf("\nOpening filter\n");
     filter_num = 0;
@@ -240,40 +281,73 @@ int main(int argc, char *argv[]) {
                 /* Handle OUTBOUND packet, search for Host header */
                 else if (addr.Direction == WINDIVERT_DIRECTION_OUTBOUND && 
                         packet_dataLen > 16 && ppTcpHdr->DstPort == htons(80) &&
+                        find_http_method_end(packet_data) &&
                         (do_host || do_host_removespace)) {
 
                     data_addr = find_host_header(packet_data, packet_dataLen);
+                    if (data_addr) {
 
-                    if (do_host && data_addr) {
+                        if (do_host) {
                             /* Replace "Host: " with "hoSt: " */
                             memcpy(data_addr, http_host_replace, strlen(http_host_replace));
                             //printf("Replaced Host header!\n");
                         }
 
-                    if (do_host_removespace && data_addr) {
+                        if (do_additional_space && do_host_removespace) {
+                            /* End of "Host:" without trailing space */
+                            host_addr = data_addr + strlen(http_host_find) - 1;
+                            method_addr = find_http_method_end(packet_data);
+
+                            if (method_addr) {
+                                memmove(method_addr + 1, method_addr, (PVOID)host_addr - (PVOID)method_addr);
+                            }
+                        }
+                        else if (do_host_removespace) {
                             host_addr = data_addr + strlen(http_host_find);
 
-                        fromhost_uptoend_len = packet_dataLen - ((PVOID)host_addr - packet_data);
                             data_addr_rn = dumb_memmem(host_addr,
-                                                    fromhost_uptoend_len,
+                                                       packet_dataLen - ((PVOID)host_addr - packet_data),
                                                        "\r\n", 2);
                             if (data_addr_rn) {
+                                /* We move Host header value by one byte to the left and then
+                                 * "insert" stolen space to the end of User-Agent value because
+                                 * some web servers are not tolerant to additional space in the
+                                 * end of Host header.
+                                 *
+                                 * Nothing is done if User-Agent header is missing.
+                                 */
                                 host_len = data_addr_rn - host_addr;
-                            if (host_len <= 64) {
+                                useragent_addr = find_useragent_header(packet_data, packet_dataLen);
-                                /* Move memory left by 1 byte and reduce packet size for 1 byte */
+                                if (host_len <= 253 && useragent_addr && useragent_addr > host_addr) {
-                                memmove(host_addr - 1, host_addr, fromhost_uptoend_len);
+                                    /* Performing action only if User-Agent header goes after Host */
-                                /* Reduce "Total Length" in IP header by 1 byte */
+
-                                *(uint16_t*)(packet + IPV4_TOTALLEN_OFFSET) = ntohs(
+                                    useragent_addr += strlen(http_useragent_find);
-                                     htons(*(uint16_t*)(packet + IPV4_TOTALLEN_OFFSET)) - 1);
+                                    /* useragent_addr is in the beginning of User-Agent value */
-                                /* Reduce packetLen by 1 byte */
+
-                                packetLen--;
+                                    data_len = packet_dataLen - ((PVOID)useragent_addr - packet_data);
+                                    data_addr_rn = dumb_memmem(useragent_addr,
+                                                               data_len, "\r\n", 2);
+                                    /* data_addr_rn is in the end of User-Agent value */
+
+                                    if (data_addr_rn) {
+                                        data_len = (PVOID)data_addr_rn - (PVOID)host_addr;
+
+                                        /* Move one byte to the left from "Host:"
+                                         * to the end of User-Agen
+                                         */
+                                        memmove(host_addr - 1, host_addr, data_len);
+                                        /* Put space in the end of User-Agent header */
+                                        *(char*)(data_addr_rn - 1) = ' ';
                                         //printf("Replaced Host header!\n");
                                     }
                                 }
                             }
+                        }
+
                         WinDivertHelperCalcChecksums(packet, packetLen, 0);
                     }
                 }
+            }
             /* Else if we got TCP packet without data */
             else if (WinDivertHelperParsePacket(packet, packetLen, &ppIpHdr,
                 NULL, NULL, NULL, &ppTcpHdr, NULL, NULL, NULL)) {