feat: add option to enable correlation

* include/datadog/environment.h
* include/datadog/tracer.h
* include/datadog/tracer_config.h
* src/datadog/tracer.cpp
* src/datadog/tracer_config.cpp

Author: elsa

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

@@ -64,6 +64,7 @@ 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;

src/datadog/trace_segment.cpp

@@ -247,7 +247,8 @@ void TraceSegment::span_finished() {
 
 #ifdef __linux__
   // When all spans are finished, so is the current trace.
-  elastic_apm_profiling_correlation_tls_v1->trace_present = 0;
+  if (elastic_apm_profiling_correlation_process_storage_v1 != nullptr)
+    elastic_apm_profiling_correlation_tls_v1->trace_present = 0;
 #endif
 
   telemetry::metrics().tracer.trace_segments_closed.inc();

src/datadog/tracer.cpp

@@ -68,7 +68,8 @@ 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,
                   std::vector<std::shared_ptr<telemetry::Metric>>{},
                   config.event_scheduler, config.agent_url);
@@ -107,9 +108,9 @@ Tracer::Tracer(const FinalizedTracerConfig& config,
 
 #ifdef __linux__
   // TODO: change the way this is done to handle programs that fork.
-  // TODO: add a flag to disable this by default.
-  elastic_apm_profiling_correlation_process_storage_v1 =
-      *signature_.generate_process_correlation_storage();
+  if (correlate_full_host_profiles_)
+    elastic_apm_profiling_correlation_process_storage_v1 =
+        *signature_.generate_process_correlation_storage();
 #endif
 
   if (config.log_on_startup) {
@@ -239,7 +240,7 @@ Span Tracer::create_span(const SpanConfig& config) {
             [generator = generator_]() { return generator->span_id(); },
             clock_};
 #ifdef __linux__
-  correlate(span);
+  if (correlate_full_host_profiles_) correlate(span);
 #endif
   return span;
 }
@@ -444,7 +445,7 @@ Expected<Span> Tracer::extract_span(const DictReader& reader,
             [generator = generator_]() { return generator->span_id(); },
             clock_};
 #ifdef __linux__
-  correlate(span);
+  if (correlate_full_host_profiles_) correlate(span);
 #endif
   return span;
 }

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");