feat: expose process storage for trace to profile correlation

* include/datadog/tracer.h
* include/datadog/tracer_signature.h
* src/datadog/tracer.cpp

Author: elsa

include/datadog/tracer.h

@@ -23,6 +23,10 @@
 #include "tracer_config.h"
 #include "tracer_signature.h"
 
+#ifdef __linux__
+extern const void* elastic_apm_profiling_correlation_process_storage_v1;
+#endif
+
 namespace datadog {
 namespace tracing {
 

include/datadog/tracer_signature.h

@@ -19,6 +19,12 @@
 //    polling the Datadog Agent. See
 //    `RemoteConfigurationManager::process_response` in `remote_config.h`.
 
+#ifdef __linux__
+#include <cstring>
+#include <memory>
+#include <vector>
+#endif
+
 #include <string>
 
 #include "runtime_id.h"
@@ -31,6 +37,17 @@
 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 +64,32 @@ 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.
+  const std::unique_ptr<uint8_t*> 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());
+
+    uint8_t* res = new uint8_t[buffer.size()];
+    memcpy(res, buffer.data(), buffer.size());
+    return std::make_unique<uint8_t*>(res);
+  }
+#endif
 };
 
 }  // namespace tracing

src/datadog/tracer.cpp

@@ -13,6 +13,9 @@
 
 #include <algorithm>
 #include <cassert>
+#ifdef __linux__
+#include <memory>
+#endif
 
 #include "config_manager.h"
 #include "datadog_agent.h"
@@ -29,6 +32,10 @@
 #include "trace_sampler.h"
 #include "w3c_propagation.h"
 
+#ifdef __linux__
+const void* elastic_apm_profiling_correlation_process_storage_v1 = nullptr;
+#endif
+
 namespace datadog {
 namespace tracing {
 
@@ -94,6 +101,13 @@ 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();
+#endif
+
   if (config.log_on_startup) {
     logger_->log_startup([configuration = this->config()](std::ostream& log) {
       log << "DATADOG TRACER CONFIGURATION - " << configuration;