Chrome Participated Tamper Detect

components/data_reduction_proxy/browser/data_reduction_proxy_tamper_detect.cc

Index: components/data_reduction_proxy/browser/data_reduction_proxy_tamper_detect.cc
diff --git a/components/data_reduction_proxy/browser/data_reduction_proxy_tamper_detect.cc b/components/data_reduction_proxy/browser/data_reduction_proxy_tamper_detect.cc
new file mode 100644
index 0000000000000000000000000000000000000000..26f25ab8608f4af8f126e7bdad3734a18238f3f7
--- /dev/null
+++ b/components/data_reduction_proxy/browser/data_reduction_proxy_tamper_detect.cc
@@ -0,0 +1,281 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include <string.h>
+#include <algorithm>
+#include <vector>
+
+#include "base/base64.h"
+#include "base/md5.h"
+#include "base/metrics/sparse_histogram.h"
+#include "base/strings/string_number_conversions.h"
+#include "components/data_reduction_proxy/browser/data_reduction_proxy_tamper_detect.h"
+
+#include "net/android/network_library.h"
+#include "net/http/http_request_headers.h"
+#include "net/http/http_util.h"
+
+std::vector<std::string> GetHeaderValues(
+    const net::HttpResponseHeaders* headers, const std::string header_name) {
+  std::vector<std::string> values;
+  std::string value;
+  void* iter = NULL;
+  while (headers->EnumerateHeader(&iter, header_name, &value)) {
+    values.push_back(value);
+  }
+  return values;
+}
+
+std::string ValuesToSortedString(std::vector<std::string> &values) {
+  std::string aggregated_values = "";
+
+  std::sort(values.begin(), values.end());
+  for (size_t i = 0; i < values.size(); ++i)
+    aggregated_values += values[i] + ",";
+  return aggregated_values;
+}
+
+std::string GetMD5(const std::string input) {
+  base::MD5Context context;
+  base::MD5Init(&context);
+  base::MD5Update(&context, input);
+  base::MD5Digest new_digest;
+  base::MD5Final(&new_digest, &context);
+  return std::string((char*)new_digest.a, 16);

[bolian, 2014/06/27 00:49:02]

Is this constant defined somewhere?

+}
+
+namespace data_reduction_proxy {
+
+bool CheckHeaderChromeProxy(const std::string fingerprint,
+                            const net::HttpResponseHeaders* headers) {
+  // I call fingerprint from FW, received_fingerprint; the fingerprint

[bolian, 2014/06/27 00:49:02]

Don't mention FW. I think readers want to see more comments about the mechanism
used here. You can have a header doc in the cc file as high level description,
and then func doc about each one here. Be concise in both cases.

+  // generated from contents is called actual_fingerprint
+  std::string received_fingerprint;
+  if (!base::Base64Decode(fingerprint, &received_fingerprint)) {
+    LOG(WARNING) << "Xing f1 base64 decode fails"; // remove later

[bolian, 2014/06/27 00:49:02]

It is time to remove now. :)

+    return false;
+  }
+
+  // need to get all the values of Chrome-Proxy, remove value fp=xxx,

[bolian, 2014/06/27 00:49:02]

Capitalize the first letter and add period at the end of the sentence. Here and
everywhere else.

+  // and calculate hash value
+  std::vector<std::string> values = GetHeaderValues(headers, "Chrome-Proxy");
+
+  for (size_t i = 0; i < values.size(); ++i)
+    if (values[i].find("fp=") == 0) {
+      values.erase(values.begin() + i);
+      break;
+    }
+
+  // from vector of values of "Chrome-Proxy", forma a string and
+  // calculate the MD5 value on the string
+  std::string actual_fingerprint = GetMD5(ValuesToSortedString(values));
+
+  return received_fingerprint.compare(actual_fingerprint) != 0;

[bolian, 2014/06/27 00:49:02]

return received_fingerprint != actual_fingerprint;

[xingx, 2014/06/27 16:34:37]

Done.

+}
+
+bool CheckHeaderVia(const std::string fingerprint,
+                    const net::HttpResponseHeaders* headers) {
+
+  // right now we get f2 value from FW, we may remove this later
+  // since for FW's f2 value, it should always be 0
+
+  std::vector<std::string> vias = GetHeaderValues(headers, "via");
+
+  // exist_chrome represents whether there is Chrome proxy
+  // in Via header;
+  // exist_hidden represents whether there is middlebox between
+  // FW and phone.
+  //bool exist_chrome = false;
+  bool exist_hidden = false;
+  for (int i = vias.size() - 1; i >= 0; --i)
+    if (vias[i].find("Chrome") != std::string::npos) {

[bolian, 2014/06/27 02:16:58]

Use the real header value. Expose that from
components/data_reduction_proxy/common/data_reduction_proxy_headers.cc

+      // exist_chrome = true;

[bolian, 2014/06/27 00:49:02]

Use HasDataReductionProxyViaHeader in
components/data_reduction_proxy/common/data_reduction_proxy_headers.h.

+      exist_hidden = (i < (int)(vias.size() - 1));
+      break;
+    }
+
+  return exist_hidden;
+}
+
+bool CheckHeaderOtherHeaders(const std::string fingerprint,
+                             const net::HttpResponseHeaders* headers) {
+  std::string received_fingerprint;
+
+  // f3 format:
+  // f3 = [base64fingerprint]:header_name1:header_namer2:...
+  net::HttpUtil::ValuesIterator it(fingerprint.begin(),
+                                   fingerprint.end(), ':');
+  if (!(it.GetNext() && base::Base64Decode(
+        std::string(it.value_begin(), it.value_end()),

[bolian, 2014/06/27 00:49:02]

it.value()?

[xingx, 2014/06/27 16:34:37]

Done.

+        &received_fingerprint))) {
+    LOG(WARNING) << "Xing f3 base64 decode fails";
+    return false;
+  }
+
+  // get header value for each header specified in f3
+  std::string header_values = "";

[bolian, 2014/06/27 00:49:02]

Remove the empty string literal. That is the default value.

[xingx, 2014/06/27 16:34:37]

Done.

+  while (it.GetNext()) {
+    std::vector<std::string> values = GetHeaderValues(headers,
+                                        std::string(it.value_begin(),
+                                        it.value_end()));
+    header_values += ValuesToSortedString(values)  + ";";
+  }
+
+  // calculate actual_f3
+  std::string actual_fingerprint = GetMD5(header_values);
+
+  return received_fingerprint.compare(actual_fingerprint) != 0;
+}
+
+bool CheckHeaderContentLength(const std::string fingerprint,
+                              const net::HttpResponseHeaders* headers) {
+  bool equal = true;
+  // if content_length from FW is not empty, check;
+  // otherwise, pass.
+  if (fingerprint.size()) {
+    int received_content_length, actual_content_length;
+    if (!base::StringToInt(fingerprint, &received_content_length))
+      return false;
+
+    std::string actual_content_length_;
+    if (headers->GetNormalizedHeader("Content-Length",
+                                     &actual_content_length_)) {
+      if (!base::StringToInt(actual_content_length_, &actual_content_length))

[bolian, 2014/06/27 00:49:02]

You can just compare the strings.

+        return true;
+      // equal marks whether received length == length sent by FW
+      equal = (received_content_length == actual_content_length);
+    }
+  }
+
+  return !equal;
+}
+
+void CheckResponseFingerprint(const net::HttpResponseHeaders* headers,
+                              const bool is_secure_scheme)
+{
+  // schemeIsSecure, we may need to change name, it means it's default
+  // FW on HTTPS or fallback FW on HTTP
+
+  // put values of Chrome-Proxy into a vector, check if it has a "fp=" value
+  std::vector<std::string> values = GetHeaderValues(headers, "Chrome-Proxy");

[bolian, 2014/06/27 00:49:02]

I think we should have a class that wraps these intermediate result so that you
don't parse it again, and also wraps the helper funcs as private. Use
FRIEND_TEST_ALL_PREFIXES to expose them to test.

+
+  // enumerate and check if we see "fp=" value
+  int fingerprint_index = -1;
+  for (size_t i=0; i<values.size(); ++i) {
+    if (values[i].find("fp=") == 0) {

[bolian, 2014/06/27 00:49:02]

define fp= and other keywords as constants.

[xingx, 2014/06/27 16:34:37]

Done.

+      fingerprint_index = i;
+      break;
+    }
+  }
+
+  if (fingerprint_index == -1)
+    return;
+
+  // delimiter "|", separate fp= string: fp=f1|f2|f3|f4

[bolian, 2014/06/27 00:49:02]

I think it might be better to encode the whole value of fp and add some basic
integrity check. If fx value is tampered somehow, we can know it. Otherwise, we
will report false positive for x header. WDYT?

[bolian, 2014/06/27 00:49:02]

What is the assumption here? All four values must be appear and in this order?
Do you always have three delimiters even if some values are missing?
I think that is hard to maintain, why not make each field a name-value pair?

+  net::HttpUtil::ValuesIterator it(values[fingerprint_index].begin() + 3,

[bolian, 2014/06/27 00:49:02]

compute 3 from the keyword "fp=" you defined.

[xingx, 2014/06/27 16:34:37]

Done.

+                                   values[fingerprint_index].end(), '|');
+
+  // we found "fp=" value, need to check fingerprint, first get carrier ID
+  unsigned mcc_mnc = 0;
+  base::StringToUint(net::android::GetTelephonyNetworkOperator(), &mcc_mnc);
+
+  // log total number for tamper detect
+  UMA_HISTOGRAM_SPARSE_SLOWLY(
+      is_secure_scheme ?
+    "DataReductionProxy.HTTPSHeaderTampereDetected" :

[bolian, 2014/06/27 00:49:02]

The name sounds like tamper happened. How about ...HeaderTamperDetection

[xingx, 2014/06/27 16:34:37]

Done.

+    "DataReductionProxy.HTTPHeaderTampereDetected",
+    mcc_mnc);
+
+
+  // check fingerprint one by one
+  if (!it.GetNext()) return;
+  if (CheckHeaderChromeProxy(std::string(it.value_begin(), it.value_end()),
+          headers)) {
+    UMA_HISTOGRAM_SPARSE_SLOWLY(
+        is_secure_scheme ?
+      "DataReductionProxy.HTTPSHeaderTampered_ChromeProxy" :
+      "DataReductionProxy.HTTPHeaderTampered_ChromeProxy",
+      mcc_mnc);
+
+    LOG(WARNING)<<"Xing f1 not equal";

[bolian, 2014/06/27 00:49:02]

Time to remove all these.

+  }
+
+  if (!it.GetNext()) return;
+  if (CheckHeaderVia(std::string(it.value_begin(), it.value_end()),
+          headers)) {
+    UMA_HISTOGRAM_SPARSE_SLOWLY(
+          is_secure_scheme ?
+          "DataReductionProxy.HTTPSHeaderTampered_Via" :
+          "DataReductionProxy.HTTPHeaderTampered_Via",
+          mcc_mnc);
+
+    LOG(WARNING)<<"Xing f2 not equal";
+  }
+
+  if (!it.GetNext()) return;
+  if (CheckHeaderOtherHeaders(std::string(it.value_begin(), it.value_end()),
+          headers)) {
+    UMA_HISTOGRAM_SPARSE_SLOWLY(
+        is_secure_scheme ?
+          "DataReductionProxy.HTTPSHeaderTampered_OtherHeaders" :
+          "DataReductionProxy.HTTPHeaderTampered_OtherHeaders",
+          mcc_mnc);
+    LOG(WARNING)<<"Xing f3 not equal";
+  }
+
+  std::string mime_type;
+  if (!it.GetNext()) return;
+  if (CheckHeaderContentLength(std::string(it.value_begin(), it.value_end()),
+          headers)) {
+    headers->GetMimeType(&mime_type);
+    LOG(WARNING) << "xing type "<<mime_type;
+    UMA_HISTOGRAM_SPARSE_SLOWLY(
+        is_secure_scheme ?
+          "DataReductionProxy.HTTPSHeaderTampered_ContentLength" :
+          "DataReductionProxy.HTTPHeaderTampered_ContentLength",
+          mcc_mnc);
+
+    if (mime_type.compare("text/javascript") == 0 ||
+        mime_type.compare("application/x-javascript") == 0 ||
+        mime_type.compare("application/javascript") == 0) {
+      UMA_HISTOGRAM_SPARSE_SLOWLY(
+          is_secure_scheme ?
+            "DataReductionProxy.HTTPSHeaderTampered_ContentLength_JS" :
+            "DataReductionProxy.HTTPHeaderTampered_ContentLength_JS",
+            mcc_mnc);
+
+      LOG(WARNING) << "Xing mimetype JS";
+    }
+    else if (mime_type.compare("text/css") == 0) {
+      UMA_HISTOGRAM_SPARSE_SLOWLY(
+          is_secure_scheme ?
+            "DataReductionProxy.HTTPSHeaderTampered_ContentLength_CSS" :
+            "DataReductionProxy.HTTPHeaderTampered_ContentLength_CSS",
+            mcc_mnc);
+
+      LOG(WARNING) << "Xing mimetype CSS";
+    }
+    else if (mime_type.find("image") == 0) {
+      UMA_HISTOGRAM_SPARSE_SLOWLY(
+          is_secure_scheme ?
+            "DataReductionProxy.HTTPSHeaderTampered_ContentLength_Image" :
+            "DataReductionProxy.HTTPHeaderTampered_ContentLength_Image",
+            mcc_mnc);
+
+      LOG(WARNING) << "Xing mimetype Image";
+    }
+    else {
+      UMA_HISTOGRAM_SPARSE_SLOWLY(
+          is_secure_scheme ?
+            "DataReductionProxy.HTTPSHeaderTampered_ContentLength_Other" :
+            "DataReductionProxy.HTTPHeaderTampered_ContentLength_Other",
+            mcc_mnc);
+
+      LOG(WARNING) << "Xing mimetype Other";
+    }
+    LOG(WARNING)<<"Xing f4 not equal";
+  }
+}
+
+}  // namespace data_reduction_proxy