chrome/browser/ssl/ssl_error_classification.cc

   1 // Copyright 2014 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include <vector>
   6
   7 #include "chrome/browser/ssl/ssl_error_classification.h"
   8
   9 #include "base/build_time.h"
  10 #include "base/metrics/field_trial.h"
  11 #include "base/metrics/histogram.h"
  12 #include "base/strings/string_split.h"
  13 #include "base/strings/utf_string_conversions.h"
  14 #include "base/time/time.h"
  15 #include "chrome/browser/browser_process.h"
  16 #include "chrome/browser/chrome_notification_types.h"
  17 #include "chrome/browser/profiles/profile.h"
  18 #include "chrome/browser/ssl/ssl_error_info.h"
  19 #include "content/public/browser/notification_service.h"
  20 #include "content/public/browser/web_contents.h"
  21 #include "net/base/net_util.h"
  22 #include "net/base/registry_controlled_domains/registry_controlled_domain.h"
  23 #include "net/cert/x509_cert_types.h"
  24 #include "net/cert/x509_certificate.h"
  25 #include "url/gurl.h"
  26
  27 #if defined(ENABLE_CAPTIVE_PORTAL_DETECTION)
  28 #include "chrome/browser/captive_portal/captive_portal_service.h"
  29 #include "chrome/browser/captive_portal/captive_portal_service_factory.h"
  30 #endif
  31
  32 #if defined(OS_WIN)
  33 #include "base/win/win_util.h"
  34 #include "base/win/windows_version.h"
  35 #endif
  36
  37 using base::Time;
  38 using base::TimeTicks;
  39 using base::TimeDelta;
  40
  41 namespace {
  42
  43 // Events for UMA. Do not reorder or change!
  44 enum SSLInterstitialCause {
  45   CLOCK_PAST,
  46   CLOCK_FUTURE,
  47   WWW_SUBDOMAIN_MATCH,
  48   SUBDOMAIN_MATCH,
  49   SUBDOMAIN_INVERSE_MATCH,
  50   SUBDOMAIN_OUTSIDE_WILDCARD,
  51   HOST_NAME_NOT_KNOWN_TLD,
  52   LIKELY_MULTI_TENANT_HOSTING,
  53   LOCALHOST,
  54   PRIVATE_URL,
  55   AUTHORITY_ERROR_CAPTIVE_PORTAL,
  56   SELF_SIGNED,
  57   EXPIRED_RECENTLY,
  58   LIKELY_SAME_DOMAIN,
  59   UNUSED_INTERSTITIAL_CAUSE_ENTRY,
  60 };
  61
  62 // Events for UMA. Do not reorder or change!
  63 enum SSLInterstitialCauseCaptivePortal {
  64   CAPTIVE_PORTAL_ALL,
  65   CAPTIVE_PORTAL_DETECTION_ENABLED,
  66   CAPTIVE_PORTAL_DETECTION_ENABLED_OVERRIDABLE,
  67   CAPTIVE_PORTAL_PROBE_COMPLETED,
  68   CAPTIVE_PORTAL_PROBE_COMPLETED_OVERRIDABLE,
  69   CAPTIVE_PORTAL_NO_RESPONSE,
  70   CAPTIVE_PORTAL_NO_RESPONSE_OVERRIDABLE,
  71   CAPTIVE_PORTAL_DETECTED,
  72   CAPTIVE_PORTAL_DETECTED_OVERRIDABLE,
  73   UNUSED_CAPTIVE_PORTAL_EVENT,
  74 };
  75
  76 void RecordSSLInterstitialCause(bool overridable, SSLInterstitialCause event) {
  77   if (overridable) {
  78     UMA_HISTOGRAM_ENUMERATION("interstitial.ssl.cause.overridable", event,
  79                               UNUSED_INTERSTITIAL_CAUSE_ENTRY);
  80   } else {
  81     UMA_HISTOGRAM_ENUMERATION("interstitial.ssl.cause.nonoverridable", event,
  82                               UNUSED_INTERSTITIAL_CAUSE_ENTRY);
  83   }
  84 }
  85
  86 #if defined(ENABLE_CAPTIVE_PORTAL_DETECTION)
  87 void RecordCaptivePortalEventStats(SSLInterstitialCauseCaptivePortal event) {
  88   UMA_HISTOGRAM_ENUMERATION("interstitial.ssl.captive_portal",
  89                             event,
  90                             UNUSED_CAPTIVE_PORTAL_EVENT);
  91 }
  92 #endif
  93
  94 int GetLevensteinDistance(const std::string& str1,
  95                           const std::string& str2) {
  96   if (str1 == str2)
  97     return 0;
  98   if (str1.size() == 0)
  99     return str2.size();
 100   if (str2.size() == 0)
 101     return str1.size();
 102   std::vector<int> kFirstRow(str2.size() + 1, 0);
 103   std::vector<int> kSecondRow(str2.size() + 1, 0);
 104
 105   for (size_t i = 0; i < kFirstRow.size(); ++i)
 106     kFirstRow[i] = i;
 107   for (size_t i = 0; i < str1.size(); ++i) {
 108     kSecondRow[0] = i + 1;
 109     for (size_t j = 0; j < str2.size(); ++j) {
 110       int cost = str1[i] == str2[j] ? 0 : 1;
 111       kSecondRow[j+1] = std::min(std::min(
 112           kSecondRow[j] + 1, kFirstRow[j + 1] + 1), kFirstRow[j] + cost);
 113     }
 114     for (size_t j = 0; j < kFirstRow.size(); j++)
 115       kFirstRow[j] = kSecondRow[j];
 116   }
 117   return kSecondRow[str2.size()];
 118 }
 119
 120 } // namespace
 121
 122 SSLErrorClassification::SSLErrorClassification(
 123     content::WebContents* web_contents,
 124     const base::Time& current_time,
 125     const GURL& url,
 126     int cert_error,
 127     const net::X509Certificate& cert)
 128   : web_contents_(web_contents),
 129     current_time_(current_time),
 130     request_url_(url),
 131     cert_error_(cert_error),
 132     cert_(cert),
 133     captive_portal_detection_enabled_(false),
 134     captive_portal_probe_completed_(false),
 135     captive_portal_no_response_(false),
 136     captive_portal_detected_(false) {
 137 #if defined(ENABLE_CAPTIVE_PORTAL_DETECTION)
 138   Profile* profile = Profile::FromBrowserContext(
 139       web_contents_->GetBrowserContext());
 140   captive_portal_detection_enabled_ =
 141       CaptivePortalServiceFactory::GetForProfile(profile)->enabled();
 142   registrar_.Add(this,
 143                  chrome::NOTIFICATION_CAPTIVE_PORTAL_CHECK_RESULT,
 144                  content::Source<Profile>(profile));
 145 #endif
 146 }
 147
 148 SSLErrorClassification::~SSLErrorClassification() { }
 149
 150 void SSLErrorClassification::RecordCaptivePortalUMAStatistics(
 151     bool overridable) const {
 152 #if defined(ENABLE_CAPTIVE_PORTAL_DETECTION)
 153   RecordCaptivePortalEventStats(CAPTIVE_PORTAL_ALL);
 154   if (captive_portal_detection_enabled_)
 155     RecordCaptivePortalEventStats(
 156         overridable ?
 157         CAPTIVE_PORTAL_DETECTION_ENABLED_OVERRIDABLE :
 158         CAPTIVE_PORTAL_DETECTION_ENABLED);
 159   if (captive_portal_probe_completed_)
 160     RecordCaptivePortalEventStats(
 161         overridable ?
 162         CAPTIVE_PORTAL_PROBE_COMPLETED_OVERRIDABLE :
 163         CAPTIVE_PORTAL_PROBE_COMPLETED);
 164   // Log only one of portal detected and no response results.
 165   if (captive_portal_detected_)
 166     RecordCaptivePortalEventStats(
 167         overridable ?
 168         CAPTIVE_PORTAL_DETECTED_OVERRIDABLE :
 169         CAPTIVE_PORTAL_DETECTED);
 170   else if (captive_portal_no_response_)
 171     RecordCaptivePortalEventStats(
 172         overridable ?
 173         CAPTIVE_PORTAL_NO_RESPONSE_OVERRIDABLE :
 174         CAPTIVE_PORTAL_NO_RESPONSE);
 175 #endif
 176 }
 177
 178 void SSLErrorClassification::RecordUMAStatistics(
 179     bool overridable) const {
 180   SSLErrorInfo::ErrorType type =
 181       SSLErrorInfo::NetErrorToErrorType(cert_error_);
 182   UMA_HISTOGRAM_ENUMERATION(
 183       "interstitial.ssl_error_type", type, SSLErrorInfo::END_OF_ENUM);
 184   switch (type) {
 185     case SSLErrorInfo::CERT_DATE_INVALID: {
 186       if (IsUserClockInThePast(base::Time::NowFromSystemTime())) {
 187         RecordSSLInterstitialCause(overridable, CLOCK_PAST);
 188       } else if (IsUserClockInTheFuture(base::Time::NowFromSystemTime())) {
 189         RecordSSLInterstitialCause(overridable, CLOCK_FUTURE);
 190       } else if (cert_.HasExpired() && TimePassedSinceExpiry().InDays() < 28) {
 191         RecordSSLInterstitialCause(overridable, EXPIRED_RECENTLY);
 192       }
 193       break;
 194     }
 195     case SSLErrorInfo::CERT_COMMON_NAME_INVALID: {
 196       std::string host_name = request_url_.host();
 197       if (IsHostNameKnownTLD(host_name)) {
 198         Tokens host_name_tokens = Tokenize(host_name);
 199         if (IsWWWSubDomainMatch())
 200           RecordSSLInterstitialCause(overridable, WWW_SUBDOMAIN_MATCH);
 201         if (IsSubDomainOutsideWildcard(host_name_tokens))
 202           RecordSSLInterstitialCause(overridable, SUBDOMAIN_OUTSIDE_WILDCARD);
 203         std::vector<std::string> dns_names;
 204         cert_.GetDNSNames(&dns_names);
 205         std::vector<Tokens> dns_name_tokens = GetTokenizedDNSNames(dns_names);
 206         if (NameUnderAnyNames(host_name_tokens, dns_name_tokens))
 207           RecordSSLInterstitialCause(overridable, SUBDOMAIN_MATCH);
 208         if (AnyNamesUnderName(dns_name_tokens, host_name_tokens))
 209           RecordSSLInterstitialCause(overridable, SUBDOMAIN_INVERSE_MATCH);
 210         if (IsCertLikelyFromMultiTenantHosting())
 211           RecordSSLInterstitialCause(overridable, LIKELY_MULTI_TENANT_HOSTING);
 212         if (IsCertLikelyFromSameDomain())
 213           RecordSSLInterstitialCause(overridable, LIKELY_SAME_DOMAIN);
 214       } else {
 215          RecordSSLInterstitialCause(overridable, HOST_NAME_NOT_KNOWN_TLD);
 216       }
 217       break;
 218     }
 219     case SSLErrorInfo::CERT_AUTHORITY_INVALID: {
 220       const std::string& hostname = request_url_.HostNoBrackets();
 221       if (net::IsLocalhost(hostname))
 222         RecordSSLInterstitialCause(overridable, LOCALHOST);
 223       if (IsHostnameNonUniqueOrDotless(hostname))
 224         RecordSSLInterstitialCause(overridable, PRIVATE_URL);
 225       if (captive_portal_probe_completed_ && captive_portal_detected_)
 226         RecordSSLInterstitialCause(overridable, AUTHORITY_ERROR_CAPTIVE_PORTAL);
 227       if (net::X509Certificate::IsSelfSigned(cert_.os_cert_handle()))
 228         RecordSSLInterstitialCause(overridable, SELF_SIGNED);
 229       break;
 230     }
 231     default:
 232       break;
 233   }
 234   UMA_HISTOGRAM_ENUMERATION("interstitial.ssl.connection_type",
 235                             net::NetworkChangeNotifier::GetConnectionType(),
 236                             net::NetworkChangeNotifier::CONNECTION_LAST);
 237 }
 238
 239 base::TimeDelta SSLErrorClassification::TimePassedSinceExpiry() const {
 240   base::TimeDelta delta = current_time_ - cert_.valid_expiry();
 241   return delta;
 242 }
 243
 244 bool SSLErrorClassification::IsUserClockInThePast(const base::Time& time_now) {
 245 #if defined(DONT_EMBED_BUILD_METADATA) && !defined(OFFICIAL_BUILD)
 246   return false;
 247 #else
 248   base::Time build_time = base::GetBuildTime();
 249   if (time_now < build_time - base::TimeDelta::FromDays(2))
 250     return true;
 251   return false;
 252 #endif
 253 }
 254
 255 bool SSLErrorClassification::IsUserClockInTheFuture(
 256     const base::Time& time_now) {
 257 #if defined(DONT_EMBED_BUILD_METADATA) && !defined(OFFICIAL_BUILD)
 258   return false;
 259 #else
 260   base::Time build_time = base::GetBuildTime();
 261   if (time_now > build_time + base::TimeDelta::FromDays(365))
 262     return true;
 263   return false;
 264 #endif
 265 }
 266
 267 bool SSLErrorClassification::MaybeWindowsLacksSHA256Support() {
 268 #if defined(OS_WIN)
 269   return !base::win::MaybeHasSHA256Support();
 270 #else
 271   return false;
 272 #endif
 273 }
 274
 275 bool SSLErrorClassification::IsHostNameKnownTLD(const std::string& host_name) {
 276   size_t tld_length =
 277       net::registry_controlled_domains::GetRegistryLength(
 278           host_name,
 279           net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES,
 280           net::registry_controlled_domains::INCLUDE_PRIVATE_REGISTRIES);
 281   if (tld_length == 0 || tld_length == std::string::npos)
 282     return false;
 283   return true;
 284 }
 285
 286 std::vector<SSLErrorClassification::Tokens> SSLErrorClassification::
 287 GetTokenizedDNSNames(const std::vector<std::string>& dns_names) {
 288   std::vector<std::vector<std::string>> dns_name_tokens;
 289   for (size_t i = 0; i < dns_names.size(); ++i) {
 290     std::vector<std::string> dns_name_token_single;
 291     if (dns_names[i].empty() || dns_names[i].find('\0') != std::string::npos
 292         || !(IsHostNameKnownTLD(dns_names[i]))) {
 293       dns_name_token_single.push_back(std::string());
 294     } else {
 295       dns_name_token_single = Tokenize(dns_names[i]);
 296     }
 297     dns_name_tokens.push_back(dns_name_token_single);
 298   }
 299   return dns_name_tokens;
 300 }
 301
 302 size_t SSLErrorClassification::FindSubDomainDifference(
 303     const Tokens& potential_subdomain, const Tokens& parent) const {
 304   // A check to ensure that the number of tokens in the tokenized_parent is
 305   // less than the tokenized_potential_subdomain.
 306   if (parent.size() >= potential_subdomain.size())
 307     return 0;
 308
 309   size_t tokens_match = 0;
 310   size_t diff_size = potential_subdomain.size() - parent.size();
 311   for (size_t i = 0; i < parent.size(); ++i) {
 312     if (parent[i] == potential_subdomain[i + diff_size])
 313       tokens_match++;
 314   }
 315   if (tokens_match == parent.size())
 316     return diff_size;
 317   return 0;
 318 }
 319
 320 SSLErrorClassification::Tokens SSLErrorClassification::
 321 Tokenize(const std::string& name) {
 322   return base::SplitString(
 323       name, ".", base::KEEP_WHITESPACE, base::SPLIT_WANT_ALL);
 324 }
 325
 326 // We accept the inverse case for www for historical reasons.
 327 bool SSLErrorClassification::GetWWWSubDomainMatch(
 328     const std::string& host_name,
 329     const std::vector<std::string>& dns_names,
 330     std::string* www_match_host_name) {
 331   if (IsHostNameKnownTLD(host_name)) {
 332     // Need to account for all possible domains given in the SSL certificate.
 333     for (size_t i = 0; i < dns_names.size(); ++i) {
 334       if (dns_names[i].empty() ||
 335           dns_names[i].find('\0') != std::string::npos ||
 336           dns_names[i].length() == host_name.length() ||
 337           !IsHostNameKnownTLD(dns_names[i])) {
 338         continue;
 339       } else if (dns_names[i].length() > host_name.length()) {
 340         if (net::StripWWW(base::ASCIIToUTF16(dns_names[i])) ==
 341             base::ASCIIToUTF16(host_name)) {
 342           *www_match_host_name = dns_names[i];
 343           return true;
 344         }
 345       } else {
 346         if (net::StripWWW(base::ASCIIToUTF16(host_name)) ==
 347             base::ASCIIToUTF16(dns_names[i])) {
 348           *www_match_host_name = dns_names[i];
 349           return true;
 350         }
 351       }
 352     }
 353   }
 354   return false;
 355 }
 356
 357 bool SSLErrorClassification::IsWWWSubDomainMatch() const {
 358   const std::string& host_name = request_url_.host();
 359   std::vector<std::string> dns_names;
 360   cert_.GetDNSNames(&dns_names);
 361   std::string www_host;
 362   return GetWWWSubDomainMatch(host_name, dns_names, &www_host);
 363 }
 364
 365 bool SSLErrorClassification::NameUnderAnyNames(
 366     const Tokens& child,
 367     const std::vector<Tokens>& potential_parents) const {
 368   bool result = false;
 369   // Need to account for all the possible domains given in the SSL certificate.
 370   for (size_t i = 0; i < potential_parents.size(); ++i) {
 371     if (potential_parents[i].empty() ||
 372         potential_parents[i].size() >= child.size()) {
 373       result = result || false;
 374     } else {
 375       size_t domain_diff = FindSubDomainDifference(child,
 376                                                    potential_parents[i]);
 377       if (domain_diff == 1 &&  child[0] != "www")
 378         result = result || true;
 379     }
 380   }
 381   return result;
 382 }
 383
 384 bool SSLErrorClassification::AnyNamesUnderName(
 385     const std::vector<Tokens>& potential_children,
 386     const Tokens& parent) const {
 387   bool result = false;
 388   // Need to account for all the possible domains given in the SSL certificate.
 389   for (size_t i = 0; i < potential_children.size(); ++i) {
 390     if (potential_children[i].empty() ||
 391         potential_children[i].size() <= parent.size()) {
 392       result = result || false;
 393     } else {
 394       size_t domain_diff = FindSubDomainDifference(potential_children[i],
 395                                                    parent);
 396       if (domain_diff == 1 &&  potential_children[i][0] != "www")
 397         result = result || true;
 398     }
 399   }
 400   return result;
 401 }
 402
 403 bool SSLErrorClassification::IsSubDomainOutsideWildcard(
 404     const Tokens& host_name_tokens) const {
 405   std::string host_name = request_url_.host();
 406   std::vector<std::string> dns_names;
 407   cert_.GetDNSNames(&dns_names);
 408   bool result = false;
 409
 410   // This method requires that the host name be longer than the dns name on
 411   // the certificate.
 412   for (size_t i = 0; i < dns_names.size(); ++i) {
 413     const std::string& name = dns_names[i];
 414     if (name.length() < 2 || name.length() >= host_name.length() ||
 415         name.find('\0') != std::string::npos ||
 416         !IsHostNameKnownTLD(name)
 417         || name[0] != '*' || name[1] != '.') {
 418       continue;
 419     }
 420
 421     // Move past the "*.".
 422     std::string extracted_dns_name = name.substr(2);
 423     if (FindSubDomainDifference(
 424         host_name_tokens, Tokenize(extracted_dns_name)) == 2) {
 425       return true;
 426     }
 427   }
 428   return result;
 429 }
 430
 431 bool SSLErrorClassification::IsCertLikelyFromMultiTenantHosting() const {
 432   std::string host_name = request_url_.host();
 433   std::vector<std::string> dns_names;
 434   std::vector<std::string> dns_names_domain;
 435   cert_.GetDNSNames(&dns_names);
 436   size_t dns_names_size = dns_names.size();
 437
 438   // If there is only 1 DNS name then it is definitely not a shared certificate.
 439   if (dns_names_size == 0 || dns_names_size == 1)
 440     return false;
 441
 442   // Check to see if all the domains in the SAN field in the SSL certificate are
 443   // the same or not.
 444   for (size_t i = 0; i < dns_names_size; ++i) {
 445     dns_names_domain.push_back(
 446         net::registry_controlled_domains::
 447         GetDomainAndRegistry(
 448             dns_names[i],
 449             net::registry_controlled_domains::INCLUDE_PRIVATE_REGISTRIES));
 450   }
 451   for (size_t i = 1; i < dns_names_domain.size(); ++i) {
 452     if (dns_names_domain[i] != dns_names_domain[0])
 453       return false;
 454   }
 455
 456   // If the number of DNS names is more than 5 then assume that it is a shared
 457   // certificate.
 458   static const int kDistinctNameThreshold = 5;
 459   if (dns_names_size > kDistinctNameThreshold)
 460     return true;
 461
 462   // Heuristic - The edit distance between all the strings should be at least 5
 463   // for it to be counted as a shared SSLCertificate. If even one pair of
 464   // strings edit distance is below 5 then the certificate is no longer
 465   // considered as a shared certificate. Include the host name in the URL also
 466   // while comparing.
 467   dns_names.push_back(host_name);
 468   static const int kMinimumEditDsitance = 5;
 469   for (size_t i = 0; i < dns_names_size; ++i) {
 470     for (size_t j = i + 1; j < dns_names_size; ++j) {
 471       int edit_distance = GetLevensteinDistance(dns_names[i], dns_names[j]);
 472       if (edit_distance < kMinimumEditDsitance)
 473         return false;
 474     }
 475   }
 476   return true;
 477 }
 478
 479 bool SSLErrorClassification::IsCertLikelyFromSameDomain() const {
 480   std::string host_name = request_url_.host();
 481   std::vector<std::string> dns_names;
 482   cert_.GetDNSNames(&dns_names);
 483
 484   dns_names.push_back(host_name);
 485   std::vector<std::string> dns_names_domain;
 486
 487   for (const std::string& dns_name : dns_names) {
 488     dns_names_domain.push_back(
 489         net::registry_controlled_domains::GetDomainAndRegistry(
 490             dns_name,
 491             net::registry_controlled_domains::INCLUDE_PRIVATE_REGISTRIES));
 492   }
 493
 494   DCHECK(!dns_names_domain.empty());
 495   const std::string& host_name_domain = dns_names_domain.back();
 496
 497   // Last element is the original domain. So, excluding it.
 498   return std::find(dns_names_domain.begin(), dns_names_domain.end() - 1,
 499                    host_name_domain) != dns_names_domain.end() - 1;
 500 }
 501
 502 // static
 503 bool SSLErrorClassification::IsHostnameNonUniqueOrDotless(
 504     const std::string& hostname) {
 505   return net::IsHostnameNonUnique(hostname) ||
 506          hostname.find('.') == std::string::npos;
 507 }
 508
 509 void SSLErrorClassification::Observe(
 510     int type,
 511     const content::NotificationSource& source,
 512     const content::NotificationDetails& details) {
 513 #if defined(ENABLE_CAPTIVE_PORTAL_DETECTION)
 514   // When detection is disabled, captive portal service always sends
 515   // RESULT_INTERNET_CONNECTED. Ignore any probe results in that case.
 516   if (!captive_portal_detection_enabled_)
 517     return;
 518   if (type == chrome::NOTIFICATION_CAPTIVE_PORTAL_CHECK_RESULT) {
 519     captive_portal_probe_completed_ = true;
 520     CaptivePortalService::Results* results =
 521         content::Details<CaptivePortalService::Results>(details).ptr();
 522     // If a captive portal was detected at any point when the interstitial was
 523     // displayed, assume that the interstitial was caused by a captive portal.
 524     // Example scenario:
 525     // 1- Interstitial displayed and captive portal detected, setting the flag.
 526     // 2- Captive portal detection automatically opens portal login page.
 527     // 3- User logs in on the portal login page.
 528     // A notification will be received here for RESULT_INTERNET_CONNECTED. Make
 529     // sure we don't clear the captive protal flag, since the interstitial was
 530     // potentially caused by the captive portal.
 531     captive_portal_detected_ = captive_portal_detected_ ||
 532         (results->result == captive_portal::RESULT_BEHIND_CAPTIVE_PORTAL);
 533     // Also keep track of non-HTTP portals and error cases.
 534     captive_portal_no_response_ = captive_portal_no_response_ ||
 535         (results->result == captive_portal::RESULT_NO_RESPONSE);
 536   }
 537 #endif
 538 }