Config MD5 a bit more deterministic

```
$ ./build/darwin/osquery/osqueryi --config_path=/asdfasdfadfs
E0903 11:45:02.050308 1990836992 init.cpp:370] Error reading config: config file does not exist
Using a virtual database. Need help, type '.help'
osquery> .mode line
osquery> .all osquery_info
           pid = 33700
       version = 1.5.2-43-gb06fa92
    config_md5 =
  config_valid = 0
   config_path = /asdfasdfadfs
    extensions = active
build_platform = darwin
  build_distro = 10.10
osquery> .exit

$ ./build/darwin/osquery/osqueryi
osquery> .mode line
osquery> .all osquery_info
           pid = 33781
       version = 1.5.2-43-gb06fa92
    config_md5 = 8a432ac93d3de080c62d77ba99b89783
  config_valid = 1
   config_path = /var/osquery/osquery.conf
    extensions = active
build_platform = darwin
  build_distro = 10.10
osquery> .exit
```

include/osquery/config.h

@@ -91,7 +91,7 @@ class Schedule {
  */
 class Config {
  private:
-  Config() : schedule_(Schedule()){};
+  Config() : schedule_(Schedule()), valid_(false){};
 
  protected:
   typedef boost::unique_lock<boost::shared_mutex> WriteLock;
@@ -172,6 +172,17 @@ class Config {
    */
   Status getMD5(std::string& hash);
 
+  /**
+   * @brief Hash a source's config data
+   *
+   * @param source is the place where the config content came from
+   * @param content is the content of the config data for a given source
+   */
+  void hashSource(const std::string& source, const std::string& content);
+
+  /// Whether or not the last loaded config was valid
+  bool isValid();
+
   /**
    * @brief Add a pack to the osquery schedule
    */
@@ -270,7 +281,8 @@ class Config {
   Schedule schedule_;
   std::map<std::string, QueryPerformance> performance_;
   std::map<std::string, std::vector<std::string> > files_;
-  std::string hash_;
+  std::map<std::string, std::string> hash_;
+  bool valid_;
 
  private:
   friend class ConfigTests;

osquery/config/config.cpp

@@ -34,6 +34,7 @@ boost::shared_mutex config_schedule_mutex_;
 boost::shared_mutex config_performance_mutex_;
 boost::shared_mutex config_files_mutex_;
 boost::shared_mutex config_hash_mutex_;
+boost::shared_mutex config_valid_mutex_;
 
 void Config::addPack(const Pack& pack) {
   WriteLock wlock(config_schedule_mutex_);
@@ -78,7 +79,7 @@ void Config::clearSchedule() {
 
 void Config::clearHash() {
   WriteLock wlock(config_hash_mutex_);
-  std::string().swap(hash_);
+  hash_.erase(hash_.begin(), hash_.end());
 }
 
 void Config::clearFiles() {
@@ -86,7 +87,13 @@ void Config::clearFiles() {
   files_.erase(files_.begin(), files_.end());
 }
 
+bool Config::isValid() {
+  ReadLock rlock(config_valid_mutex_);
+  return valid_;
+}
+
 Status Config::load() {
+  valid_ = false;
   auto& config_plugin = Registry::getActive("config");
   if (!Registry::exists("config", config_plugin)) {
     return Status(1, "Missing config plugin " + config_plugin);
@@ -102,6 +109,7 @@ Status Config::load() {
   clearSchedule();
   clearHash();
   clearFiles();
+  valid_ = true;
 
   // if there was a response, parse it and update internal state
   if (response.size() > 0) {
@@ -130,6 +138,8 @@ Status Config::update(const std::map<std::string, std::string>& config) {
   }
 
   for (const auto& source : config) {
+    hashSource(source.first, source.second);
+
     // load the config (source.second) into a pt::ptree
     std::stringstream json;
     json << source.second;
@@ -270,53 +280,30 @@ void Config::getPerformanceStats(
   }
 }
 
+void Config::hashSource(const std::string& source, const std::string& content) {
+  WriteLock wlock(config_hash_mutex_);
+  hash_[source] =
+      hashFromBuffer(HASH_TYPE_MD5, &(content.c_str())[0], content.size());
+}
+
 Status Config::getMD5(std::string& hash) {
-  if (hash_.empty()) {
-    std::vector<char> buffer;
-    auto add = [&buffer](const std::string& text) {
-      for (const auto& c : text) {
-        buffer.push_back(c);
-      }
-    };
-    scheduledQueries(
-        [&add, &buffer](const std::string& name, const ScheduledQuery& query) {
-          add(name);
-          add(query.query);
-          add(std::to_string(query.interval));
-          for (const auto& it : query.options) {
-            add(it.first);
-            add(it.second ? "true" : "false");
-          }
-        });
-
-    auto parsers = Registry::all("config_parser");
-    for (const auto& parser : parsers) {
-      add(parser.first);
-      try {
-        if (parser.second == nullptr || parser.second.get() == nullptr) {
-          continue;
-        }
-        auto plugin =
-            std::static_pointer_cast<ConfigParserPlugin>(parser.second);
-        if (plugin == nullptr || plugin.get() == nullptr) {
-          continue;
-        }
-        std::stringstream ss;
-        pt::write_json(ss, plugin->getData());
-        add(ss.str());
-      } catch (const std::bad_cast& e) {
-        LOG(ERROR) << "Error casting config parser plugin: " << e.what();
-      } catch (const pt::ptree_error& e) {
-        LOG(ERROR)
-            << "Error writing config parser content to JSON: " << e.what();
-      }
-    }
-
-    std::sort(buffer.begin(), buffer.end());
-    hash_ = hashFromBuffer(HASH_TYPE_MD5, &buffer[0], buffer.size());
+  if (valid_) {
+    return Status(1, "Current config is not valid");
   }
 
-  hash = hash_;
+  ReadLock rlock(config_hash_mutex_);
+  std::vector<char> buffer;
+  buffer.reserve(hash_.size() * 32);
+  auto add = [&buffer](const std::string& text) {
+    for (const auto& c : text) {
+      buffer.push_back(c);
+    }
+  };
+  for (const auto it : hash_) {
+    add(it.second);
+  }
+
+  hash = hashFromBuffer(HASH_TYPE_MD5, &buffer[0], buffer.size());
   return Status(0, "OK");
 }
 

osquery/core/init.cpp

@@ -365,7 +365,10 @@ void Initializer::start() {
   }
 
   // Load the osquery config using the default/active config plugin.
-  Config::getInstance().load();
+  auto s = Config::getInstance().load();
+  if (!s.ok()) {
+    LOG(ERROR) << "Error reading config: " << s.toString();
+  }
 
   // Initialize the status and result plugin logger.
   initActivePlugin("logger", FLAGS_logger_plugin);

osquery/tables/utility/osquery.cpp

@@ -144,6 +144,8 @@ QueryData genOsqueryInfo(QueryContext& context) {
     VLOG(1) << "Could not retrieve config hash: " << s.toString();
   }
 
+  r["config_valid"] = Config::getInstance().isValid() ? INTEGER(1) : INTEGER(0);
+
   r["config_path"] = Flag::getValue("config_path");
   r["extensions"] =
       (pingExtension(FLAGS_extensions_socket).ok()) ? "active" : "inactive";

specs/utility/osquery_info.table

@@ -3,13 +3,12 @@ description("Top level information about the running version of osquery.")
 schema([
     Column("pid", INTEGER, "Process (or thread) ID"),
     Column("version", TEXT, "osquery toolkit version"),
-    Column("config_md5", TEXT, "md5 hash of the working configuration"),
-    Column("config_path", TEXT,
-      "Optional: path to filesystem config plugin content"),
+    Column("config_md5", TEXT, "md5 hash of the working configuration state"),
+    Column("config_valid", INTEGER, "1 if the config was loaded and considered valid, else 0"),
+    Column("config_path", TEXT, "Optional: path to filesystem config plugin content"),
     Column("extensions", TEXT, "osquery extensions status"),
     Column("build_platform", TEXT, "osquery toolkit build platform"),
-    Column("build_distro", TEXT,
-      "osquery toolkit platform distribution name (os version)"),
+    Column("build_distro", TEXT, "osquery toolkit platform distribution name (os version)"),
 ])
 attributes(utility=True)
 implementation("osquery@genOsqueryInfo")