datadog 2.2.0 → 2.4.0

data/ext/datadog_profiling_native_extension/heap_recorder.c

@@ -166,6 +166,12 @@ struct heap_recorder {
     size_t objects_frozen;
   } stats_last_update;
 };
+
+struct end_heap_allocation_args {
+  struct heap_recorder *heap_recorder;
+  ddog_prof_Slice_Location locations;
+};
+
 static heap_record* get_or_create_heap_record(heap_recorder*, ddog_prof_Slice_Location);
 static void cleanup_heap_record_if_unused(heap_recorder*, heap_record*);
 static void on_committed_object_record_cleanup(heap_recorder *heap_recorder, object_record *record);
@@ -176,6 +182,7 @@ static int st_object_records_iterate(st_data_t, st_data_t, st_data_t);
 static int st_object_records_debug(st_data_t key, st_data_t value, st_data_t extra);
 static int update_object_record_entry(st_data_t*, st_data_t*, st_data_t, int);
 static void commit_recording(heap_recorder*, heap_record*, recording);
+static VALUE end_heap_allocation_recording(VALUE end_heap_allocation_args);
 
 // ==========================
 // Heap Recorder External API
@@ -219,7 +226,7 @@ void heap_recorder_free(heap_recorder *heap_recorder) {
   st_foreach(heap_recorder->heap_records, st_heap_record_entry_free, 0);
   st_free_table(heap_recorder->heap_records);
 
-  if (heap_recorder->active_recording.object_record != NULL) {
+  if (heap_recorder->active_recording.object_record != NULL && heap_recorder->active_recording.object_record != &SKIPPED_RECORD) {
     // If there's a partial object record, clean it up as well
     object_record_free(heap_recorder->active_recording.object_record);
   }
@@ -340,9 +347,28 @@ void start_heap_allocation_recording(heap_recorder *heap_recorder, VALUE new_obj
   };
 }
 
-void end_heap_allocation_recording(struct heap_recorder *heap_recorder, ddog_prof_Slice_Location locations) {
+// end_heap_allocation_recording_with_rb_protect gets called while the stack_recorder is holding one of the profile
+// locks. To enable us to correctly unlock the profile on exception, we wrap the call to end_heap_allocation_recording
+// with an rb_protect.
+__attribute__((warn_unused_result))
+int end_heap_allocation_recording_with_rb_protect(struct heap_recorder *heap_recorder, ddog_prof_Slice_Location locations) {
+  int exception_state;
+  struct end_heap_allocation_args end_heap_allocation_args = {
+    .heap_recorder = heap_recorder,
+    .locations = locations,
+  };
+  rb_protect(end_heap_allocation_recording, (VALUE) &end_heap_allocation_args, &exception_state);
+  return exception_state;
+}
+
+static VALUE end_heap_allocation_recording(VALUE end_heap_allocation_args) {
+  struct end_heap_allocation_args *args = (struct end_heap_allocation_args *) end_heap_allocation_args;
+
+  struct heap_recorder *heap_recorder = args->heap_recorder;
+  ddog_prof_Slice_Location locations = args->locations;
+
   if (heap_recorder == NULL) {
-    return;
+    return Qnil;
   }
 
   recording active_recording = heap_recorder->active_recording;
@@ -356,15 +382,16 @@ void end_heap_allocation_recording(struct heap_recorder *heap_recorder, ddog_pro
   // data required for committing though.
   heap_recorder->active_recording = (recording) {0};
 
-  if (active_recording.object_record == &SKIPPED_RECORD) {
-    // special marker when we decided to skip due to sampling
-    return;
+  if (active_recording.object_record == &SKIPPED_RECORD) { // special marker when we decided to skip due to sampling
+    return Qnil;
   }
 
   heap_record *heap_record = get_or_create_heap_record(heap_recorder, locations);
 
   // And then commit the new allocation.
   commit_recording(heap_recorder, heap_record, active_recording);
+
+  return Qnil;
 }
 
 void heap_recorder_prepare_iteration(heap_recorder *heap_recorder) {
@@ -605,12 +632,14 @@ static int st_object_records_iterate(DDTRACE_UNUSED st_data_t key, st_data_t val
   ddog_prof_Location *locations = recorder->reusable_locations;
   for (uint16_t i = 0; i < stack->frames_len; i++) {
     const heap_frame *frame = &stack->frames[i];
-    ddog_prof_Location *location = &locations[i];
-    location->function.name.ptr = frame->name;
-    location->function.name.len = strlen(frame->name);
-    location->function.filename.ptr = frame->filename;
-    location->function.filename.len = strlen(frame->filename);
-    location->line = frame->line;
+    locations[i] = (ddog_prof_Location) {
+      .mapping = {.filename = DDOG_CHARSLICE_C(""), .build_id = DDOG_CHARSLICE_C("")},
+      .function = {
+        .name = {.ptr = frame->name, .len = strlen(frame->name)},
+        .filename = {.ptr = frame->filename, .len = strlen(frame->filename)},
+      },
+      .line = frame->line,
+    };
   }
 
   heap_recorder_iteration_data iteration_data;
@@ -755,8 +784,20 @@ static void cleanup_heap_record_if_unused(heap_recorder *heap_recorder, heap_rec
 }
 
 static void on_committed_object_record_cleanup(heap_recorder *heap_recorder, object_record *record) {
+  // @ivoanjo: We've seen a segfault crash in the field in this function (October 2024) which we're still trying to investigate.
+  // (See PROF-10656 Datadog-internal for details). Just in case, I've sprinkled a bunch of NULL tests in this function for now.
+  // Once we figure out the issue we can get rid of them again.
+
+  if (heap_recorder == NULL) rb_raise(rb_eRuntimeError, "heap_recorder was NULL in on_committed_object_record_cleanup");
+  if (heap_recorder->heap_records == NULL) rb_raise(rb_eRuntimeError, "heap_recorder->heap_records was NULL in on_committed_object_record_cleanup");
+  if (record == NULL) rb_raise(rb_eRuntimeError, "record was NULL in on_committed_object_record_cleanup");
+
   // Starting with the associated heap record. There will now be one less tracked object pointing to it
   heap_record *heap_record = record->heap_record;
+
+  if (heap_record == NULL) rb_raise(rb_eRuntimeError, "heap_record was NULL in on_committed_object_record_cleanup");
+  if (heap_record->stack == NULL) rb_raise(rb_eRuntimeError, "heap_record->stack was NULL in on_committed_object_record_cleanup");
+
   heap_record->num_tracked_objects--;
 
   // One less object using this heap record, it may have become unused...
@@ -815,6 +856,7 @@ VALUE object_record_inspect(object_record *record) {
   if (!ruby_ref_from_id(LONG2NUM(record->obj_id), &ref)) {
     rb_str_catf(inspect, "object=<invalid>");
   } else {
+    rb_str_catf(inspect, "value=%p ", (void *) ref);
     VALUE ruby_inspect = ruby_safe_inspect(ref);
     if (ruby_inspect != Qnil) {
       rb_str_catf(inspect, "object=%"PRIsVALUE, ruby_inspect);

data/ext/datadog_profiling_native_extension/heap_recorder.h

@@ -114,7 +114,9 @@ void start_heap_allocation_recording(heap_recorder *heap_recorder, VALUE new_obj
 // @param locations The stacktrace representing the location of the allocation.
 //
 // WARN: It is illegal to call this without previously having called ::start_heap_allocation_recording.
-void end_heap_allocation_recording(heap_recorder *heap_recorder, ddog_prof_Slice_Location locations);
+// WARN: This method rescues exceptions with `rb_protect`, returning the exception state integer for the caller to handle.
+__attribute__((warn_unused_result))
+int end_heap_allocation_recording_with_rb_protect(heap_recorder *heap_recorder, ddog_prof_Slice_Location locations);
 
 // Update the heap recorder to reflect the latest state of the VM and prepare internal structures
 // for efficient iteration.

data/ext/datadog_profiling_native_extension/helpers.h

@@ -2,22 +2,11 @@
 
 #include <stdint.h>
 
-// Used to mark symbols to be exported to the outside of the extension.
-// Consider very carefully before tagging a function with this.
-#define DDTRACE_EXPORT __attribute__ ((visibility ("default")))
-
-// Used to mark function arguments that are deliberately left unused
-#ifdef __GNUC__
-  #define DDTRACE_UNUSED  __attribute__((unused))
-#else
-  #define DDTRACE_UNUSED
-#endif
-
 // @ivoanjo: After trying to read through https://stackoverflow.com/questions/3437404/min-and-max-in-c I decided I
 // don't like C and I just implemented this as a function.
-inline static uint64_t uint64_max_of(uint64_t a, uint64_t b) { return a > b ? a : b; }
-inline static uint64_t uint64_min_of(uint64_t a, uint64_t b) { return a > b ? b : a; }
-inline static long long_max_of(long a, long b) { return a > b ? a : b; }
-inline static long long_min_of(long a, long b) { return a > b ? b : a; }
-inline static double double_max_of(double a, double b) { return a > b ? a : b; }
-inline static double double_min_of(double a, double b) { return a > b ? b : a; }
+static inline uint64_t uint64_max_of(uint64_t a, uint64_t b) { return a > b ? a : b; }
+static inline uint64_t uint64_min_of(uint64_t a, uint64_t b) { return a > b ? b : a; }
+static inline long long_max_of(long a, long b) { return a > b ? a : b; }
+static inline long long_min_of(long a, long b) { return a > b ? b : a; }
+static inline double double_max_of(double a, double b) { return a > b ? a : b; }
+static inline double double_min_of(double a, double b) { return a > b ? b : a; }

data/ext/datadog_profiling_native_extension/http_transport.c

@@ -21,7 +21,7 @@ struct call_exporter_without_gvl_arguments {
   bool send_ran;
 };
 
-inline static ddog_ByteSlice byte_slice_from_ruby_string(VALUE string);
+static inline ddog_ByteSlice byte_slice_from_ruby_string(VALUE string);
 static VALUE _native_validate_exporter(VALUE self, VALUE exporter_configuration);
 static ddog_prof_Exporter_NewResult create_exporter(VALUE exporter_configuration, VALUE tags_as_array);
 static VALUE handle_exporter_failure(ddog_prof_Exporter_NewResult exporter_result);
@@ -53,11 +53,11 @@ void http_transport_init(VALUE profiling_module) {
   ok_symbol = ID2SYM(rb_intern_const("ok"));
   error_symbol = ID2SYM(rb_intern_const("error"));
 
-  library_version_string = ddtrace_version();
+  library_version_string = datadog_gem_version();
   rb_global_variable(&library_version_string);
 }
 
-inline static ddog_ByteSlice byte_slice_from_ruby_string(VALUE string) {
+static inline ddog_ByteSlice byte_slice_from_ruby_string(VALUE string) {
   ENFORCE_TYPE(string, T_STRING);
   ddog_ByteSlice byte_slice = {.ptr = (uint8_t *) StringValuePtr(string), .len = RSTRING_LEN(string)};
   return byte_slice;
@@ -77,6 +77,32 @@ static VALUE _native_validate_exporter(DDTRACE_UNUSED VALUE _self, VALUE exporte
   return rb_ary_new_from_args(2, ok_symbol, Qnil);
 }
 
+static ddog_prof_Endpoint endpoint_from(VALUE exporter_configuration) {
+  ENFORCE_TYPE(exporter_configuration, T_ARRAY);
+
+  VALUE exporter_working_mode = rb_ary_entry(exporter_configuration, 0);
+  ENFORCE_TYPE(exporter_working_mode, T_SYMBOL);
+  ID working_mode = SYM2ID(exporter_working_mode);
+
+  ID agentless_id = rb_intern("agentless");
+  ID agent_id = rb_intern("agent");
+
+  if (working_mode != agentless_id && working_mode != agent_id) {
+    rb_raise(rb_eArgError, "Failed to initialize transport: Unexpected working mode, expected :agentless or :agent");
+  }
+
+  if (working_mode == agentless_id) {
+    VALUE site = rb_ary_entry(exporter_configuration, 1);
+    VALUE api_key = rb_ary_entry(exporter_configuration, 2);
+
+    return ddog_prof_Endpoint_agentless(char_slice_from_ruby_string(site), char_slice_from_ruby_string(api_key));
+  } else { // agent_id
+    VALUE base_url = rb_ary_entry(exporter_configuration, 1);
+
+    return ddog_prof_Endpoint_agent(char_slice_from_ruby_string(base_url));
+  }
+}
+
 static ddog_prof_Exporter_NewResult create_exporter(VALUE exporter_configuration, VALUE tags_as_array) {
   ENFORCE_TYPE(exporter_configuration, T_ARRAY);
   ENFORCE_TYPE(tags_as_array, T_ARRAY);
@@ -115,8 +141,7 @@ static VALUE perform_export(
   ddog_prof_Exporter_Slice_File files_to_export_unmodified,
   ddog_Vec_Tag *additional_tags,
   ddog_CharSlice internal_metadata,
-  ddog_CharSlice info,
-  uint64_t timeout_milliseconds
+  ddog_CharSlice info
 ) {
   ddog_prof_ProfiledEndpointsStats *endpoints_stats = NULL; // Not in use yet
   ddog_prof_Exporter_Request_BuildResult build_result = ddog_prof_Exporter_Request_build(
@@ -128,8 +153,7 @@ static VALUE perform_export(
     additional_tags,
     endpoints_stats,
     &internal_metadata,
-    &info,
-    timeout_milliseconds
+    &info
   );
 
   if (build_result.tag == DDOG_PROF_EXPORTER_REQUEST_BUILD_RESULT_ERR) {
@@ -254,6 +278,15 @@ static VALUE _native_do_export(
   VALUE failure_tuple = handle_exporter_failure(exporter_result);
   if (!NIL_P(failure_tuple)) return failure_tuple;
 
+  ddog_prof_MaybeError timeout_result = ddog_prof_Exporter_set_timeout(exporter_result.ok, timeout_milliseconds);
+  if (timeout_result.tag == DDOG_PROF_OPTION_ERROR_SOME_ERROR) {
+    // NOTE: Seems a bit harsh to fail the upload if we can't set a timeout. OTOH, this is only expected to fail
+    // if the exporter is not well built. Because such a situation should already be caught above I think it's
+    // preferable to leave this here as a virtually unreachable exception rather than ignoring it.
+    ddog_prof_Exporter_drop(exporter_result.ok);
+    return rb_ary_new_from_args(2, error_symbol, get_error_details_and_drop(&timeout_result.some));
+  }
+
   return perform_export(
     exporter_result.ok,
     start,
@@ -262,8 +295,7 @@ static VALUE _native_do_export(
     files_to_export_unmodified,
     null_additional_tags,
     internal_metadata,
-    info,
-    timeout_milliseconds
+    info
   );
 }
 

data/ext/datadog_profiling_native_extension/libdatadog_helpers.c

@@ -2,8 +2,6 @@
 
 #include <ruby.h>
 
-static VALUE log_failure_to_process_tag(VALUE err_details);
-
 const char *ruby_value_type_to_string(enum ruby_value_type type) {
   return ruby_value_type_to_char_slice(type).ptr;
 }
@@ -62,87 +60,3 @@ size_t read_ddogerr_string_and_drop(ddog_Error *error, char *string, size_t capa
   ddog_Error_drop(error);
   return error_msg_size;
 }
-
-__attribute__((warn_unused_result))
-ddog_prof_Endpoint endpoint_from(VALUE exporter_configuration) {
-  ENFORCE_TYPE(exporter_configuration, T_ARRAY);
-
-  VALUE exporter_working_mode = rb_ary_entry(exporter_configuration, 0);
-  ENFORCE_TYPE(exporter_working_mode, T_SYMBOL);
-  ID working_mode = SYM2ID(exporter_working_mode);
-
-  ID agentless_id = rb_intern("agentless");
-  ID agent_id = rb_intern("agent");
-
-  if (working_mode != agentless_id && working_mode != agent_id) {
-    rb_raise(rb_eArgError, "Failed to initialize transport: Unexpected working mode, expected :agentless or :agent");
-  }
-
-  if (working_mode == agentless_id) {
-    VALUE site = rb_ary_entry(exporter_configuration, 1);
-    VALUE api_key = rb_ary_entry(exporter_configuration, 2);
-    ENFORCE_TYPE(site, T_STRING);
-    ENFORCE_TYPE(api_key, T_STRING);
-
-    return ddog_prof_Endpoint_agentless(char_slice_from_ruby_string(site), char_slice_from_ruby_string(api_key));
-  } else { // agent_id
-    VALUE base_url = rb_ary_entry(exporter_configuration, 1);
-    ENFORCE_TYPE(base_url, T_STRING);
-
-    return ddog_prof_Endpoint_agent(char_slice_from_ruby_string(base_url));
-  }
-}
-
-__attribute__((warn_unused_result))
-ddog_Vec_Tag convert_tags(VALUE tags_as_array) {
-  ENFORCE_TYPE(tags_as_array, T_ARRAY);
-
-  long tags_count = RARRAY_LEN(tags_as_array);
-  ddog_Vec_Tag tags = ddog_Vec_Tag_new();
-
-  for (long i = 0; i < tags_count; i++) {
-    VALUE name_value_pair = rb_ary_entry(tags_as_array, i);
-
-    if (!RB_TYPE_P(name_value_pair, T_ARRAY)) {
-      ddog_Vec_Tag_drop(tags);
-      ENFORCE_TYPE(name_value_pair, T_ARRAY);
-    }
-
-    // Note: We can index the array without checking its size first because rb_ary_entry returns Qnil if out of bounds
-    VALUE tag_name = rb_ary_entry(name_value_pair, 0);
-    VALUE tag_value = rb_ary_entry(name_value_pair, 1);
-
-    if (!(RB_TYPE_P(tag_name, T_STRING) && RB_TYPE_P(tag_value, T_STRING))) {
-      ddog_Vec_Tag_drop(tags);
-      ENFORCE_TYPE(tag_name, T_STRING);
-      ENFORCE_TYPE(tag_value, T_STRING);
-    }
-
-    ddog_Vec_Tag_PushResult push_result =
-      ddog_Vec_Tag_push(&tags, char_slice_from_ruby_string(tag_name), char_slice_from_ruby_string(tag_value));
-
-    if (push_result.tag == DDOG_VEC_TAG_PUSH_RESULT_ERR) {
-      // libdatadog validates tags and may catch invalid tags that ddtrace didn't actually catch.
-      // We warn users about such tags, and then just ignore them.
-
-      int exception_state;
-      rb_protect(log_failure_to_process_tag, get_error_details_and_drop(&push_result.err), &exception_state);
-
-      // Since we are calling into Ruby code, it may raise an exception. Ensure that dynamically-allocated tags
-      // get cleaned before propagating the exception.
-      if (exception_state) {
-        ddog_Vec_Tag_drop(tags);
-        rb_jump_tag(exception_state);  // "Re-raise" exception
-      }
-    }
-  }
-
-  return tags;
-}
-
-static VALUE log_failure_to_process_tag(VALUE err_details) {
-  VALUE datadog_module = rb_const_get(rb_cObject, rb_intern("Datadog"));
-  VALUE logger = rb_funcall(datadog_module, rb_intern("logger"), 0);
-
-  return rb_funcall(logger, rb_intern("warn"), 1, rb_sprintf("Failed to add tag to profiling request: %"PRIsVALUE, err_details));
-}

data/ext/datadog_profiling_native_extension/libdatadog_helpers.h

@@ -3,27 +3,10 @@
 #include <datadog/profiling.h>
 #include "ruby_helpers.h"
 
-inline static ddog_CharSlice char_slice_from_ruby_string(VALUE string) {
-  ENFORCE_TYPE(string, T_STRING);
-  ddog_CharSlice char_slice = {.ptr = StringValuePtr(string), .len = RSTRING_LEN(string)};
-  return char_slice;
-}
-
-inline static VALUE ruby_string_from_vec_u8(ddog_Vec_U8 string) {
+static inline VALUE ruby_string_from_vec_u8(ddog_Vec_U8 string) {
   return rb_str_new((char *) string.ptr, string.len);
 }
 
-inline static VALUE ruby_string_from_error(const ddog_Error *error) {
-  ddog_CharSlice char_slice = ddog_Error_message(error);
-  return rb_str_new(char_slice.ptr, char_slice.len);
-}
-
-inline static VALUE get_error_details_and_drop(ddog_Error *error) {
-  VALUE result = ruby_string_from_error(error);
-  ddog_Error_drop(error);
-  return result;
-}
-
 // Utility function to be able to extract an error cstring from a ddog_Error.
 // Returns the amount of characters written to string (which are necessarily
 // bounded by capacity - 1 since the string will be null-terminated).
@@ -37,10 +20,6 @@ ddog_CharSlice ruby_value_type_to_char_slice(enum ruby_value_type type);
 
 // Returns a dynamically allocated string from the provided char slice.
 // WARN: The returned string must be explicitly freed with ruby_xfree.
-inline static char* string_from_char_slice(ddog_CharSlice slice) {
+static inline char* string_from_char_slice(ddog_CharSlice slice) {
   return ruby_strndup(slice.ptr, slice.len);
 }
-
-ddog_prof_Endpoint endpoint_from(VALUE exporter_configuration);
-
-ddog_Vec_Tag convert_tags(VALUE tags_as_array);