Safe-browsing PrefixSet cleanups.

chrome/browser/safe_browsing/safe_browsing_database.cc

-// Copyright (c) 2010 The Chromium Authors. All rights reserved.
+// Copyright (c) 2011 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 "chrome/browser/safe_browsing/safe_browsing_database.h"
 
 #include <algorithm>
 
 #include "base/file_util.h"
 #include "base/metrics/histogram.h"
 #include "base/metrics/stats_counters.h"
@@ skipping 195 matching lines @@
 // - Any bloom filter miss should be a prefix set miss.
 // - Any prefix set hit should be a bloom filter hit.
 // - Bloom filter false positives are prefix set misses.
 // The following is to log actual performance to verify this.
 enum PrefixSetEvent {
   PREFIX_SET_EVENT_HIT,
   PREFIX_SET_EVENT_BLOOM_HIT,
   PREFIX_SET_EVENT_BLOOM_MISS_PREFIX_HIT,
   PREFIX_SET_EVENT_BLOOM_MISS_PREFIX_HIT_INVALID,
   PREFIX_SET_GETPREFIXES_BROKEN,
+  PREFIX_SET_GETPREFIXES_BROKEN_SIZE,
+  PREFIX_SET_GETPREFIXES_FIRST_BROKEN,
+  PREFIX_SET_SBPREFIX_WAS_BROKEN,
 
   // Memory space for histograms is determined by the max.  ALWAYS ADD
   // NEW VALUES BEFORE THIS ONE.
   PREFIX_SET_EVENT_MAX
 };
 
 void RecordPrefixSetInfo(PrefixSetEvent event_type) {
   UMA_HISTOGRAM_ENUMERATION("SB2.PrefixSetEvent", event_type,
                             PREFIX_SET_EVENT_MAX);
 }
 
+// Verify that |GetPrefixes()| returns the same set of prefixes as
+// |prefixes|.  This is necessary so that the
+// PREFIX_SET_EVENT_BLOOM_MISS_PREFIX_HIT_INVALID histogram in
+// ContainsBrowseUrl() can be trustworthy.
+void CheckPrefixSet(const safe_browsing::PrefixSet& prefix_set,
+                    const std::vector<SBPrefix>& prefixes) {
+  std::vector<SBPrefix> restored;
+  prefix_set.GetPrefixes(&restored);
+
+  const std::set<SBPrefix> unique(prefixes.begin(), prefixes.end());
+
+  // Expect them to be equal.
+  if (restored.size() == unique.size() &&
+      std::equal(unique.begin(), unique.end(), restored.begin()))
+    return;
+
+  // Log BROKEN for continuity with previous release, and SIZE to
+  // distinguish which test failed.
+  NOTREACHED();
+  RecordPrefixSetInfo(PREFIX_SET_GETPREFIXES_BROKEN);
+  if (restored.size() != unique.size())
+    RecordPrefixSetInfo(PREFIX_SET_GETPREFIXES_BROKEN_SIZE);
+
+  // Try to distinguish between updates from one broken user and a
+  // distributed problem.
+  static bool logged_broken = false;
+  if (!logged_broken) {
+    RecordPrefixSetInfo(PREFIX_SET_GETPREFIXES_FIRST_BROKEN);
+    logged_broken = true;
+  }
+
+  // This seems so very very unlikely.  But if it ever were true, then
+  // it could explain why GetPrefixes() seemed broken.
+  if (sizeof(int) != sizeof(int32))
+    RecordPrefixSetInfo(PREFIX_SET_SBPREFIX_WAS_BROKEN);
+}
+
 }  // namespace
 
 // The default SafeBrowsingDatabaseFactory.
 class SafeBrowsingDatabaseFactoryImpl : public SafeBrowsingDatabaseFactory {
  public:
   virtual SafeBrowsingDatabase* CreateSafeBrowsingDatabase(
       bool enable_download_protection,
       bool enable_client_side_whitelist) {
     return new SafeBrowsingDatabaseNew(
         new SafeBrowsingStoreFile,
@@ skipping 728 matching lines @@
   }
 
   std::vector<SBPrefix> prefixes;
   for (size_t i = 0; i < add_prefixes.size(); ++i) {
     prefixes.push_back(add_prefixes[i].prefix);
   }
   std::sort(prefixes.begin(), prefixes.end());
   scoped_ptr<safe_browsing::PrefixSet>
       prefix_set(new safe_browsing::PrefixSet(prefixes));
 
-  // Verify that |GetPrefixes()| returns the same set of prefixes as
-  // was passed to the constructor.
-  std::vector<SBPrefix> restored;
-  prefix_set->GetPrefixes(&restored);
-  prefixes.erase(std::unique(prefixes.begin(), prefixes.end()), prefixes.end());
-  if (restored.size() != prefixes.size() ||
-      !std::equal(prefixes.begin(), prefixes.end(), restored.begin())) {
-    NOTREACHED();
-    RecordPrefixSetInfo(PREFIX_SET_GETPREFIXES_BROKEN);
-  }
+  CheckPrefixSet(*(prefix_set.get()), prefixes);
 
   // This needs to be in sorted order by prefix for efficient access.
   std::sort(add_full_hashes.begin(), add_full_hashes.end(),
             SBAddFullHashPrefixLess);
 
   // Swap in the newly built filter and cache.
   {
     base::AutoLock locked(lookup_lock_);
     full_browse_hashes_.swap(add_full_hashes);
 
@@ skipping 93 matching lines @@
   // Manually re-generate the prefix set from the main database.
   // TODO(shess): Write/read for prefix set.
   std::vector<SBAddPrefix> add_prefixes;
   browse_store_->GetAddPrefixes(&add_prefixes);
   std::vector<SBPrefix> prefixes;
   for (size_t i = 0; i < add_prefixes.size(); ++i) {
     prefixes.push_back(add_prefixes[i].prefix);
   }
   std::sort(prefixes.begin(), prefixes.end());
   prefix_set_.reset(new safe_browsing::PrefixSet(prefixes));
-
-  // Double-check the prefixes so that the
-  // PREFIX_SET_EVENT_BLOOM_MISS_PREFIX_HIT_INVALID histogram in
-  // ContainsBrowseUrl() can be trustworthy.
-  std::vector<SBPrefix> restored;
-  prefix_set_->GetPrefixes(&restored);
-  std::set<SBPrefix> unique(prefixes.begin(), prefixes.end());
-  if (restored.size() != unique.size() ||
-      !std::equal(unique.begin(), unique.end(), restored.begin())) {
-    NOTREACHED();
-    RecordPrefixSetInfo(PREFIX_SET_GETPREFIXES_BROKEN);
-  }
+  CheckPrefixSet(*(prefix_set_.get()), prefixes);
 }
 
 bool SafeBrowsingDatabaseNew::Delete() {
   DCHECK_EQ(creation_loop_, MessageLoop::current());
 
   const bool r1 = browse_store_->Delete();
   if (!r1)
     RecordFailure(FAILURE_DATABASE_STORE_DELETE);
 
   const bool r2 = download_store_.get() ? download_store_->Delete() : true;
@@ skipping 52 matching lines @@
   if (std::binary_search(new_csd_whitelist.begin(), new_csd_whitelist.end(),
                          kill_switch)) {
     // The kill switch is whitelisted hence we whitelist all URLs.
     CsdWhitelistAllUrls();
   } else {
     base::AutoLock locked(lookup_lock_);
     csd_whitelist_all_urls_ = false;
     csd_whitelist_.swap(new_csd_whitelist);
   }
 }