feat: write process storage

* CMakeLists.txt
* include/datadog/environment.h
* include/datadog/tracer.h
* include/datadog/tracer_config.h
* include/datadog/tracer_signature.h
* src/datadog/tracer.cpp
* src/datadog/tracer_config.cpp

Author: elsa

CMakeLists.txt

@@ -203,6 +203,7 @@ if (BUILD_SHARED_LIBS)
     PUBLIC
       dd_trace::obj
       CURL::libcurl_shared
+      customlabels
     PRIVATE
       dd_trace::specs
   )

include/datadog/environment.h

@@ -50,6 +50,7 @@ namespace environment {
   MACRO(DD_TRACE_TAGS_PROPAGATION_MAX_LENGTH)        \
   MACRO(DD_VERSION)                                  \
   MACRO(DD_TRACE_128_BIT_TRACEID_GENERATION_ENABLED) \
+  MACRO(DD_TRACE_CORRELATE_FULL_HOST_PROFILES)       \
   MACRO(DD_TELEMETRY_HEARTBEAT_INTERVAL)             \
   MACRO(DD_TELEMETRY_METRICS_ENABLED)                \
   MACRO(DD_TELEMETRY_METRICS_INTERVAL_SECONDS)       \

include/datadog/tracer.h

@@ -54,6 +54,7 @@ class Tracer {
   Baggage::Options baggage_opts_;
   bool baggage_injection_enabled_;
   bool baggage_extraction_enabled_;
+  bool correlate_full_host_profiles_;
 
  public:
   // Create a tracer configured using the specified `config`, and optionally:

include/datadog/tracer_config.h

@@ -133,6 +133,14 @@ struct TracerConfig {
   // the `DD_TRACE_128_BIT_TRACEID_GENERATION_ENABLED` environment variable.
   Optional<bool> generate_128bit_trace_ids;
 
+  // `correlate_full_host_profiles` indicates whether we want to correlate
+  // traces and spans with profiles generated by the eBPF full host profiler.
+  // This correlation only works on linux, due to the eBPF-based nature of
+  // the profiler. It implies writing some process-level and thread-level
+  // data in variables which the profiler will then read from the process's
+  // memory.
+  Optional<bool> correlate_full_host_profiles;
+
   // `runtime_id` denotes the current run of the application in which the tracer
   // is embedded. If `runtime_id` is not specified, then it defaults to a
   // pseudo-randomly generated value. A server that contains multiple tracers,
@@ -197,6 +205,7 @@ class FinalizedTracerConfig final {
   std::shared_ptr<Logger> logger;
   bool log_on_startup;
   bool generate_128bit_trace_ids;
+  bool correlate_full_host_profiles;
   Optional<RuntimeID> runtime_id;
   Clock clock;
   std::string integration_name;

include/datadog/tracer_signature.h

@@ -19,6 +19,14 @@
 //    polling the Datadog Agent. See
 //    `RemoteConfigurationManager::process_response` in `remote_config.h`.
 
+#ifdef __linux__
+#include <cstring>
+#include <vector>
+extern "C" {
+#include <customlabels/customlabels.h>
+}
+#endif
+
 #include <string>
 
 #include "runtime_id.h"
@@ -31,6 +39,17 @@
 namespace datadog {
 namespace tracing {
 
+#ifdef __linux__
+namespace {
+void write_utf8_string(std::vector<uint8_t>& buffer, const std::string& str) {
+  uint32_t length = str.length();
+  buffer.insert(buffer.end(), reinterpret_cast<uint8_t*>(&length),
+                reinterpret_cast<uint8_t*>(&length) + sizeof(length));
+  buffer.insert(buffer.end(), str.begin(), str.end());
+}
+}  // namespace
+#endif
+
 struct TracerSignature {
   RuntimeID runtime_id;
   std::string default_service;
@@ -47,6 +66,30 @@ struct TracerSignature {
         library_version(tracer_version),
         library_language("cpp"),
         library_language_version(DD_TRACE_STRINGIFY(__cplusplus), 6) {}
+
+#ifdef __linux__
+  // The process correlation storage contains information needed to
+  // correlate traces to profiles generated by dd-otel-host-profiler.
+  void generate_process_correlation_storage() {
+    std::vector<uint8_t> buffer;
+
+    // Currently, layout minor version is 2 to differ from Elastic's
+    // version which includes a socket path.
+    // Layout:
+    // https://github.com/elastic/apm/blob/149cd3e39a77a58002344270ed2ad35357bdd02d/specs/agents/universal-profiling-integration.md#process-storage-layout
+    uint16_t layout_minor_version = 2;
+    buffer.insert(buffer.end(),
+                  reinterpret_cast<uint8_t*>(&layout_minor_version),
+                  reinterpret_cast<uint8_t*>(&layout_minor_version) +
+                      sizeof(layout_minor_version));
+
+    write_utf8_string(buffer, default_service);
+    write_utf8_string(buffer, default_environment);
+    write_utf8_string(buffer, runtime_id.string());
+
+    proc_storage_set(buffer.data(), buffer.size());
+  }
+#endif
 };
 
 }  // namespace tracing

src/datadog/tracer.cpp

@@ -58,7 +58,8 @@ Tracer::Tracer(const FinalizedTracerConfig& config,
       tags_header_max_size_(config.tags_header_size),
       baggage_opts_(config.baggage_opts),
       baggage_injection_enabled_(false),
-      baggage_extraction_enabled_(false) {
+      baggage_extraction_enabled_(false),
+      correlate_full_host_profiles_(config.correlate_full_host_profiles) {
   telemetry::init(config.telemetry, logger_, config.http_client,
                   config.event_scheduler, config.agent_url);
   if (config.report_hostname) {
@@ -92,6 +93,12 @@ Tracer::Tracer(const FinalizedTracerConfig& config,
     }
   }
 
+#ifdef __linux__
+  // TODO: change the way this is done to handle programs that fork.
+  if (correlate_full_host_profiles_)
+    signature_.generate_process_correlation_storage();
+#endif
+
   if (config.log_on_startup) {
     logger_->log_startup([configuration = this->config()](std::ostream& log) {
       log << "DATADOG TRACER CONFIGURATION - " << configuration;

src/datadog/tracer_config.cpp

@@ -127,6 +127,10 @@ Expected<TracerConfig> load_tracer_env_config(Logger &logger) {
           lookup(environment::DD_TRACE_128_BIT_TRACEID_GENERATION_ENABLED)) {
     env_cfg.generate_128bit_trace_ids = !falsy(*enabled_env);
   }
+  if (auto enabled_env =
+          lookup(environment::DD_TRACE_CORRELATE_FULL_HOST_PROFILES)) {
+    env_cfg.correlate_full_host_profiles = !falsy(*enabled_env);
+  }
 
   // Baggage
   if (auto baggage_items_env =
@@ -361,6 +365,11 @@ Expected<FinalizedTracerConfig> finalize_config(const TracerConfig &user_config,
       ConfigMetadata(ConfigName::GENEREATE_128BIT_TRACE_IDS,
                      to_string(final_config.generate_128bit_trace_ids), origin);
 
+  // Correlate with Full Host Profiles
+  final_config.correlate_full_host_profiles =
+      value_or(env_config->correlate_full_host_profiles,
+               user_config.correlate_full_host_profiles, false);
+
   // Integration name & version
   final_config.integration_name = value_or(
       env_config->integration_name, user_config.integration_name, "datadog");