datadog 2.21.0 → 2.38.0

data/ext/libdatadog_api/crashtracker.c

@@ -2,10 +2,13 @@
 #include <datadog/crashtracker.h>
 
 #include "datadog_ruby_common.h"
+#include "helpers.h"
 
 static VALUE _native_start_or_update_on_fork(int argc, VALUE *argv, DDTRACE_UNUSED VALUE _self);
 static VALUE _native_stop(DDTRACE_UNUSED VALUE _self);
 
+void crashtracker_report_exception_init(VALUE crashtracker_class);
+
 static bool first_init = true;
 
 // Used to report Ruby VM crashes.
@@ -17,6 +20,9 @@ void crashtracker_init(VALUE core_module) {
 
   rb_define_singleton_method(crashtracker_class, "_native_start_or_update_on_fork", _native_start_or_update_on_fork, -1);
   rb_define_singleton_method(crashtracker_class, "_native_stop", _native_stop, 0);
+
+  // Initialize Ruby non-signal-crash reporting
+  crashtracker_report_exception_init(crashtracker_class);
 }
 
 static VALUE _native_start_or_update_on_fork(int argc, VALUE *argv, DDTRACE_UNUSED VALUE _self) {
@@ -41,16 +47,12 @@ static VALUE _native_start_or_update_on_fork(int argc, VALUE *argv, DDTRACE_UNUS
   ENFORCE_TYPE(action, T_SYMBOL);
   ENFORCE_TYPE(upload_timeout_seconds, T_FIXNUM);
 
-  if (action != start_action && action != update_on_fork_action) rb_raise(rb_eArgError, "Unexpected action: %+"PRIsVALUE, action);
+  if (action != start_action && action != update_on_fork_action) raise_error(rb_eArgError, "Unexpected action: %+"PRIsVALUE, action);
 
   VALUE version = datadog_gem_version();
 
-  // Tags and endpoint are heap-allocated, so after here we can't raise exceptions otherwise we'll leak this memory
+  // Tags are heap-allocated, so after here we can't raise exceptions otherwise we'll leak this memory
   // Start of exception-free zone to prevent leaks {{
-  ddog_Endpoint *endpoint = ddog_endpoint_from_url(char_slice_from_ruby_string(agent_base_url));
-  if (endpoint == NULL) {
-    rb_raise(rb_eRuntimeError, "Failed to create endpoint from agent_base_url: %"PRIsVALUE, agent_base_url);
-  }
   ddog_Vec_Tag tags = convert_tags(tags_as_array);
 
   ddog_crasht_Config config = {
@@ -67,9 +69,11 @@ static VALUE _native_start_or_update_on_fork(int argc, VALUE *argv, DDTRACE_UNUS
     // overriding what Ruby set up seems a saner default to keep anyway.
     .create_alt_stack = false,
     .use_alt_stack = true,
-    .endpoint = endpoint,
+    .endpoint = {.url = char_slice_from_ruby_string(agent_base_url)},
     .resolve_frames = DDOG_CRASHT_STACKTRACE_COLLECTION_ENABLED_WITH_SYMBOLS_IN_RECEIVER,
     .timeout_ms = FIX2INT(upload_timeout_seconds) * 1000,
+    .collect_all_threads = true,
+    .max_threads = 128,
   };
 
   ddog_crasht_Metadata metadata = {
@@ -100,16 +104,15 @@ static VALUE _native_start_or_update_on_fork(int argc, VALUE *argv, DDTRACE_UNUS
       ) :
       ddog_crasht_update_on_fork(config, receiver_config, metadata);
 
-  first_init = false;
+  // We use first_init to know which of [init, reconfigure] needs to be called. BUT if init failed we actually need
+  // to call init next time again, not reconfigure.
+  if (result.tag == DDOG_VOID_RESULT_OK) first_init = false;
 
   // Clean up before potentially raising any exceptions
   ddog_Vec_Tag_drop(tags);
-  ddog_endpoint_drop(endpoint);
   // }} End of exception-free zone to prevent leaks
 
-  if (result.tag == DDOG_VOID_RESULT_ERR) {
-    rb_raise(rb_eRuntimeError, "Failed to start/update the crash tracker: %"PRIsVALUE, get_error_details_and_drop(&result.err));
-  }
+  CHECK_VOID_RESULT("Failed to start/update the crash tracker", result);
 
   return Qtrue;
 }
@@ -117,9 +120,7 @@ static VALUE _native_start_or_update_on_fork(int argc, VALUE *argv, DDTRACE_UNUS
 static VALUE _native_stop(DDTRACE_UNUSED VALUE _self) {
   ddog_VoidResult result = ddog_crasht_disable();
 
-  if (result.tag == DDOG_VOID_RESULT_ERR) {
-    rb_raise(rb_eRuntimeError, "Failed to stop the crash tracker: %"PRIsVALUE, get_error_details_and_drop(&result.err));
-  }
+  CHECK_VOID_RESULT("Failed to stop the crash tracker", result);
 
   return Qtrue;
 }

data/ext/libdatadog_api/crashtracker_report_exception.c

@@ -0,0 +1,126 @@
+#include <datadog/common.h>
+#include <datadog/crashtracker.h>
+#include <ruby.h>
+
+#include "datadog_ruby_common.h"
+
+static VALUE _native_report_ruby_exception(VALUE _self, VALUE exception_type,
+                                          VALUE message, VALUE frames_data);
+
+static bool process_crash_frames(VALUE frames_data, ddog_crasht_Handle_StackTrace *stack_trace);
+
+void crashtracker_report_exception_init(VALUE crashtracker_class) {
+  rb_define_singleton_method(crashtracker_class, "_native_report_ruby_exception",
+                            _native_report_ruby_exception, 3);
+}
+
+static VALUE _native_report_ruby_exception(DDTRACE_UNUSED VALUE _self, VALUE exception_type,
+                                          VALUE message, VALUE frames_data) {
+  ENFORCE_TYPE(exception_type, T_STRING);
+  ENFORCE_TYPE(message, T_STRING);
+  ENFORCE_TYPE(frames_data, T_ARRAY);
+
+  // Build stack trace
+  ddog_crasht_StackTrace_NewResult stack_result = ddog_crasht_StackTrace_new();
+  if (stack_result.tag != DDOG_CRASHT_STACK_TRACE_NEW_RESULT_OK) {
+    ddog_Error_drop(&stack_result.err);
+    return Qfalse;
+  }
+
+  ddog_crasht_Handle_StackTrace *stack_trace = &stack_result.ok;
+
+  bool frames_ok = process_crash_frames(frames_data, stack_trace);
+  if (frames_ok) {
+    ddog_VoidResult complete_result = ddog_crasht_StackTrace_set_complete(stack_trace);
+    if (complete_result.tag != DDOG_VOID_RESULT_OK) {
+      ddog_crasht_StackTrace_drop(stack_trace);
+      ddog_Error_drop(&complete_result.err);
+      return Qfalse;
+    }
+  }
+
+  // ddog_crasht_report_unhandled_exception takes ownership of stack_trace
+  ddog_VoidResult result = ddog_crasht_report_unhandled_exception(
+    char_slice_from_ruby_string(exception_type),
+    char_slice_from_ruby_string(message),
+    stack_trace
+  );
+
+  if (result.tag != DDOG_VOID_RESULT_OK) {
+    ddog_Error_drop(&result.err);
+    return Qfalse;
+  }
+
+  return Qtrue;
+}
+
+static bool process_crash_frames(VALUE frames_data, ddog_crasht_Handle_StackTrace *stack_trace) {
+  size_t frame_count = RARRAY_LEN(frames_data);
+
+  // Return false and early so we can mark the stack as incomplete
+  // libdatadog's definition of an incomplete stack is that it has no frames
+  // or that report generation died in the middle of unwinding frames
+  if (frame_count == 0) {
+    return false;
+  }
+
+  for (size_t i = 0; i < frame_count; i++) {
+    VALUE frame_array = RARRAY_AREF(frames_data, i);
+
+    // ruby should guarantee [String, String, Integer]
+    if (!RB_TYPE_P(frame_array, T_ARRAY) || RARRAY_LEN(frame_array) != 3) {
+      // Malformed data from Ruby; this is a bug, bail out
+      return false;
+    }
+
+    VALUE file_val = RARRAY_AREF(frame_array, 0);
+    VALUE func_val = RARRAY_AREF(frame_array, 1);
+    VALUE line_val = RARRAY_AREF(frame_array, 2);
+
+    // validate types; Ruby should guarantee these
+    if (!RB_TYPE_P(file_val, T_STRING) || !RB_TYPE_P(func_val, T_STRING) || !RB_TYPE_P(line_val, T_FIXNUM)) {
+      // Type mismatch from Ruby; this is a bug, bail out
+      return false;
+    }
+
+    ddog_crasht_StackFrame_NewResult frame_result = ddog_crasht_StackFrame_new();
+    if (frame_result.tag != DDOG_CRASHT_STACK_FRAME_NEW_RESULT_OK) {
+      ddog_Error_drop(&frame_result.err);
+      return false;
+    }
+
+    ddog_crasht_Handle_StackFrame *frame = &frame_result.ok;
+
+    ddog_VoidResult file_result = ddog_crasht_StackFrame_with_file(frame, char_slice_from_ruby_string(file_val));
+    if (file_result.tag != DDOG_VOID_RESULT_OK) {
+      ddog_crasht_StackFrame_drop(frame);
+      ddog_Error_drop(&file_result.err);
+      return false;
+    }
+    ddog_VoidResult func_result = ddog_crasht_StackFrame_with_function(frame, char_slice_from_ruby_string(func_val));
+    if (func_result.tag != DDOG_VOID_RESULT_OK) {
+      ddog_crasht_StackFrame_drop(frame);
+      ddog_Error_drop(&func_result.err);
+      return false;
+    }
+
+    uint32_t line = (uint32_t)FIX2INT(line_val);
+    if (line > 0) {
+      ddog_VoidResult line_result = ddog_crasht_StackFrame_with_line(frame, line);
+      if (line_result.tag != DDOG_VOID_RESULT_OK) {
+        ddog_crasht_StackFrame_drop(frame);
+        ddog_Error_drop(&line_result.err);
+        return false;
+      }
+    }
+
+    ddog_VoidResult push_result = ddog_crasht_StackTrace_push_frame(stack_trace, frame, true);
+    if (push_result.tag != DDOG_VOID_RESULT_OK) {
+      ddog_crasht_StackFrame_drop(frame);
+      ddog_Error_drop(&push_result.err);
+      return false;
+    }
+  }
+
+  return true;
+}

data/ext/libdatadog_api/datadog_ruby_common.c

@@ -1,8 +1,11 @@
 #include "datadog_ruby_common.h"
+#include <stdarg.h>
 
 // IMPORTANT: Currently this file is copy-pasted between extensions. Make sure to update all versions when doing any change!
 
-void raise_unexpected_type(VALUE value, const char *value_name, const char *type_name, const char *file, int line, const char* function_name) {
+static ID telemetry_message_id;
+
+void raise_unexpected_type(VALUE value, const char *value_name, const char *type_name, const char *file, int line, const char *function_name) {
   rb_exc_raise(
     rb_exc_new_str(
       rb_eTypeError,
@@ -18,6 +21,25 @@ void raise_unexpected_type(VALUE value, const char *value_name, const char *type
   );
 }
 
+// Raises an exception with separate telemetry-safe and detailed messages.
+// NOTE: Raising an exception always invokes Ruby code so it requires the GVL and is not compatible with "debug_enter_unsafe_context".
+// @see debug_enter_unsafe_context
+void private_raise_exception(VALUE exception, const char *static_message) {
+  rb_ivar_set(exception, telemetry_message_id, rb_str_new_cstr(static_message));
+  rb_exc_raise(exception);
+}
+
+// Use `raise_error` the macro instead, as it provides additional argument checks.
+void private_raise_error(VALUE exception_class, const char *fmt, ...) {
+  va_list args;
+  va_start(args, fmt);
+  VALUE detailed_message = rb_vsprintf(fmt, args);
+  va_end(args);
+
+  VALUE exception = rb_exc_new_str(exception_class, detailed_message);
+  private_raise_exception(exception, fmt);
+}
+
 VALUE datadog_gem_version(void) {
   VALUE ddtrace_module = rb_const_get(rb_cObject, rb_intern("Datadog"));
   ENFORCE_TYPE(ddtrace_module, T_MODULE);
@@ -78,3 +100,45 @@ ddog_Vec_Tag convert_tags(VALUE tags_as_array) {
 
   return tags;
 }
+
+size_t read_ddogerr_string_and_drop(ddog_Error *error, char *string, size_t capacity) {
+  if (capacity == 0 || string == NULL) {
+    // short-circuit, we can't write anything
+    ddog_Error_drop(error);
+    return 0;
+  }
+
+  ddog_CharSlice error_msg_slice = ddog_Error_message(error);
+  size_t error_msg_size = error_msg_slice.len;
+  // Account for extra null char for proper cstring
+  if (error_msg_size >= capacity) {
+    // Error message too big, lets truncate it to capacity - 1 to allow for extra null at end
+    error_msg_size = capacity - 1;
+  }
+  strncpy(string, error_msg_slice.ptr, error_msg_size);
+  string[error_msg_size] = '\0';
+  ddog_Error_drop(error);
+  return error_msg_size;
+}
+
+static void verify_libdatadog_version(void) {
+  rb_eval_string(
+    "require 'libdatadog';"
+    "expected = '" EXPECTED_LIBDATADOG_VERSION "';"
+    "if expected != Libdatadog::VERSION;"
+      "raise(LoadError, <<MSG\n"
+        "The `datadog` gem needs to be reinstalled whenever the `libdatadog` gem version is changed. "
+        "The currently-installed version of `datadog` was built to work with `libdatadog` gem version #{expected} "
+        "but the currently-loaded version of `libdatadog` is #{Libdatadog::VERSION}. "
+        "To fix this, reinstall the `datadog` gem (e.g. `bundle exec gem pristine datadog`) "
+        "or contact Datadog support for help at <https://docs.datadoghq.com/help/>.\n"
+        "MSG\n"
+      ");"
+    "end"
+  );
+}
+
+void datadog_ruby_common_init(void) {
+  telemetry_message_id = rb_intern("@telemetry_message");
+  verify_libdatadog_version();
+}

data/ext/libdatadog_api/datadog_ruby_common.h

@@ -3,7 +3,10 @@
 // IMPORTANT: Currently this file is copy-pasted between extensions. Make sure to update all versions when doing any change!
 
 #include <ruby.h>
-#include <datadog/profiling.h>
+#include <datadog/common.h>
+
+// Must be called once during initialization
+void datadog_ruby_common_init(void);
 
 // Used to mark symbols to be exported to the outside of the extension.
 // Consider very carefully before tagging a function with this.
@@ -32,6 +35,27 @@
 
 NORETURN(void raise_unexpected_type(VALUE value, const char *value_name, const char *type_name, const char *file, int line, const char* function_name));
 
+// Raises an exception of the specified class with the formatted string as its message.
+// This macro ensures that the literal string is sent for telemetry, while the formatted
+// message is the default `Exception#message`.
+// *Ruby exceptions not raised through this function will not be reported via telemetry.*
+#define raise_error(exception_class, fmt, ...) \
+  private_raise_error(exception_class, "" fmt, ##__VA_ARGS__)
+
+NORETURN(
+  void private_raise_error(VALUE exception_class, const char *fmt, ...)
+  __attribute__ ((format (printf, 2, 3)));
+);
+
+// Raises an exception with separate telemetry-safe and detailed messages.
+// NOTE: Raising an exception always invokes Ruby code so it requires the GVL and is not compatible with "debug_enter_unsafe_context".
+// @see debug_enter_unsafe_context
+NORETURN(
+  void private_raise_exception(VALUE exception, const char *static_message)
+);
+
+#define MAX_RAISE_MESSAGE_SIZE 256
+
 // Helper to retrieve Datadog::VERSION::STRING
 VALUE datadog_gem_version(void);
 
@@ -61,3 +85,18 @@ static inline VALUE get_error_details_and_drop(ddog_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).
+size_t read_ddogerr_string_and_drop(ddog_Error *error, char *string, size_t capacity);
+
+#define IMEMO_MASK 0x0f
+
+// Returns the imemo type of an imemo object.
+// This mask is the same between Ruby 2.5 and 3.3-preview3. Furthermore, the intention of this method is to be used
+// to call `rb_imemo_name` which correctly handles invalid numbers so even if the mask changes in the future, at most
+// we'll get incorrect results (and never a VM crash)
+static inline int ddtrace_imemo_type(VALUE imemo) {
+  return (RBASIC(imemo)->flags >> FL_USHIFT) & IMEMO_MASK;
+}

data/ext/libdatadog_api/ddsketch.c

@@ -0,0 +1,102 @@
+#include <ruby.h>
+#include <datadog/ddsketch.h>
+
+#include "datadog_ruby_common.h"
+#include "helpers.h"
+
+static VALUE _native_new(VALUE klass);
+static void ddsketch_free(void *ptr);
+static VALUE native_add(VALUE self, VALUE point);
+static VALUE native_add_with_count(VALUE self, VALUE point, VALUE count);
+static VALUE native_count(VALUE self);
+static VALUE native_encode(VALUE self);
+
+void ddsketch_init(VALUE core_module) {
+  VALUE ddsketch_class = rb_define_class_under(core_module, "DDSketch", rb_cObject);
+
+  rb_define_alloc_func(ddsketch_class, _native_new);
+  rb_define_method(ddsketch_class, "add", native_add, 1);
+  rb_define_method(ddsketch_class, "add_with_count", native_add_with_count, 2);
+  rb_define_method(ddsketch_class, "count", native_count, 0);
+  rb_define_method(ddsketch_class, "encode", native_encode, 0);
+}
+
+// This structure is used to define a Ruby object that stores a pointer to a ddsketch_Handle_DDSketch
+// See also https://github.com/ruby/ruby/blob/master/doc/extension.rdoc for how this works
+static const rb_data_type_t ddsketch_typed_data = {
+  .wrap_struct_name = "Datadog::DDSketch",
+  .function = {
+    .dmark = NULL, // We don't store references to Ruby objects so we don't need to mark any of them
+    .dfree = ddsketch_free,
+    .dsize = NULL, // We don't track memory usage (although it'd be cool if we did!)
+    //.dcompact = NULL, // Not needed -- we don't store references to Ruby objects
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY
+};
+
+static VALUE _native_new(VALUE klass) {
+  ddsketch_Handle_DDSketch *state = ruby_xcalloc(1, sizeof(ddsketch_Handle_DDSketch));
+
+  *state = ddog_ddsketch_new();
+
+  return TypedData_Wrap_Struct(klass, &ddsketch_typed_data, state);
+}
+
+static void ddsketch_free(void *ptr) {
+  ddsketch_Handle_DDSketch *state = (ddsketch_Handle_DDSketch *) ptr;
+  ddog_ddsketch_drop(state);
+  ruby_xfree(ptr);
+}
+
+static VALUE native_add(VALUE self, VALUE point) {
+  ddsketch_Handle_DDSketch *state;
+  TypedData_Get_Struct(self, ddsketch_Handle_DDSketch, &ddsketch_typed_data, state);
+
+  ddog_VoidResult result = ddog_ddsketch_add(state, NUM2DBL(point));
+
+  CHECK_VOID_RESULT("DDSketch add failed", result);
+
+  return self;
+}
+
+static VALUE native_add_with_count(VALUE self, VALUE point, VALUE count) {
+  ddsketch_Handle_DDSketch *state;
+  TypedData_Get_Struct(self, ddsketch_Handle_DDSketch, &ddsketch_typed_data, state);
+
+  ddog_VoidResult result = ddog_ddsketch_add_with_count(state, NUM2DBL(point), NUM2DBL(count));
+
+  CHECK_VOID_RESULT("DDSketch add_with_count failed", result);
+
+  return self;
+}
+
+static VALUE native_count(VALUE self) {
+  ddsketch_Handle_DDSketch *state;
+  TypedData_Get_Struct(self, ddsketch_Handle_DDSketch, &ddsketch_typed_data, state);
+
+  double count_out;
+  ddog_VoidResult result = ddog_ddsketch_count(state, &count_out);
+
+  CHECK_VOID_RESULT("DDSketch count failed", result);
+
+  return DBL2NUM(count_out);
+}
+
+static VALUE native_encode(VALUE self) {
+  ddsketch_Handle_DDSketch *state;
+  TypedData_Get_Struct(self, ddsketch_Handle_DDSketch, &ddsketch_typed_data, state);
+
+  ddog_Vec_U8 encoded = ddog_ddsketch_encode(state);
+
+  // Copy into a Ruby string
+  VALUE bytes = rb_str_new((const char *) encoded.ptr, encoded.len);
+
+  ddog_Vec_U8_drop(encoded);
+
+  // The sketch is consumed by encode; to make this a bit more user-friendly for
+  // a Ruby API (since we can't "kill" the Ruby object), let's re-initialize it so
+  // it can be used again.
+  *state = ddog_ddsketch_new();
+
+  return bytes;
+}

data/ext/libdatadog_api/di.c

@@ -0,0 +1,203 @@
+#include <stdbool.h>
+
+#include "datadog_ruby_common.h"
+
+// Prototypes for Ruby functions declared in internal Ruby headers.
+// rb_iseqw_new wraps an internal iseq pointer into a Ruby-visible
+// RubyVM::InstructionSequence object.
+VALUE rb_iseqw_new(const void *iseq);
+// rb_iseqw_to_iseq unwraps a RubyVM::InstructionSequence object back
+// to its internal iseq pointer.
+const void *rb_iseqw_to_iseq(VALUE iseqw);
+int rb_objspace_internal_object_p(VALUE obj);
+void rb_objspace_each_objects(
+    int (*callback)(void *start, void *end, size_t stride, void *data),
+    void *data);
+
+#define IMEMO_TYPE_ISEQ 7
+
+// The ID value of the string "mesg" which is used in Ruby source as
+// id_mesg or idMesg, and is used to set and retrieve the exception message
+// from standard library exception classes like NameError.
+static ID id_mesg;
+
+// ID for the fiber-local key that backs the method-probe re-entrancy guard.
+// Storage is the same hashtable that backs Thread#[] / Thread#[]=, but accessed
+// directly via rb_thread_local_aref / rb_thread_local_aset so that user-installed
+// method probes on Thread#[] / Thread#[]= cannot intercept guard reads/writes.
+static ID id_datadog_di_in_probe;
+
+// Returns whether the argument is an IMEMO of type ISEQ.
+static bool ddtrace_imemo_iseq_p(VALUE v) {
+  return rb_objspace_internal_object_p(v) && RB_TYPE_P(v, T_IMEMO) && ddtrace_imemo_type(v) == IMEMO_TYPE_ISEQ;
+}
+
+static int ddtrace_di_os_obj_of_i(void *vstart, void *vend, size_t stride, void *data)
+{
+  VALUE *array = (VALUE *)data;
+
+  VALUE v = (VALUE)vstart;
+  for (; v != (VALUE)vend; v += stride) {
+    if (ddtrace_imemo_iseq_p(v)) {
+      VALUE iseq = rb_iseqw_new((void *) v);
+      rb_ary_push(*array, iseq);
+    }
+  }
+
+  return 0;
+}
+
+/*
+Returns all RubyVM::InstructionSequence objects existing in the current process.
+
+This uses the same approach as ruby/debug's iseq_collector.c:
+https://github.com/ruby/debug/blob/master/ext/debug/iseq_collector.c
+*/
+static VALUE all_iseqs(DDTRACE_UNUSED VALUE _self) {
+  VALUE array = rb_ary_new();
+  rb_objspace_each_objects(ddtrace_di_os_obj_of_i, &array);
+  return array;
+}
+
+/*
+ * call-seq:
+ *   DI.exception_message(exception) -> String | Object
+ *
+ * Returns the exception message associated with the exception via the
+ * exception's constructor.
+ *
+ * This method does not invoke Ruby code and as such will not call
+ * the +message+ method, if one is defined on the exception object.
+ *
+ * Normally, the exception message is a string, however there is no
+ * type enforcement done by Ruby for the messages and objects of arbitrary
+ * classes can be passed to exception constructors and will, subsequently,
+ * be returned by this method.
+ *
+ * @param exception [Exception] The exception object
+ * @return [String | Object] The exception message
+ */
+static VALUE exception_message(DDTRACE_UNUSED VALUE _self, VALUE exception) {
+  return rb_ivar_get(exception, id_mesg);
+}
+
+/*
+ * call-seq:
+ *   DI.in_probe? -> true | false
+ *
+ * Returns whether the current fiber is currently inside DI probe processing.
+ * Reads the same fiber-local storage as Thread.current[:datadog_di_in_probe]
+ * but bypasses Thread#[] method dispatch — a user method probe on Thread#[]
+ * cannot observe or intercept this call.
+ *
+ * @api private
+ */
+static VALUE in_probe_p(DDTRACE_UNUSED VALUE _self) {
+  VALUE v = rb_thread_local_aref(rb_thread_current(), id_datadog_di_in_probe);
+  return RTEST(v) ? Qtrue : Qfalse;
+}
+
+/*
+ * call-seq:
+ *   DI.enter_probe -> nil
+ *
+ * Marks the current fiber as inside DI probe processing. Writes to the same
+ * fiber-local storage as Thread.current[:datadog_di_in_probe] = true, but
+ * bypasses Thread#[]= method dispatch — a user method probe on Thread#[]=
+ * cannot observe or intercept this call.
+ *
+ * @api private
+ */
+static VALUE enter_probe(DDTRACE_UNUSED VALUE _self) {
+  rb_thread_local_aset(rb_thread_current(), id_datadog_di_in_probe, Qtrue);
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *   DI.leave_probe -> nil
+ *
+ * Marks the current fiber as no longer inside DI probe processing. Writes to
+ * the same fiber-local storage as Thread.current[:datadog_di_in_probe] = nil,
+ * but bypasses Thread#[]= method dispatch — a user method probe on Thread#[]=
+ * cannot observe or intercept this call.
+ *
+ * @api private
+ */
+static VALUE leave_probe(DDTRACE_UNUSED VALUE _self) {
+  rb_thread_local_aset(rb_thread_current(), id_datadog_di_in_probe, Qnil);
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *   DI.hash?(obj) -> true | false
+ *
+ * Returns whether the given object is a Hash via a direct type check
+ * (RB_TYPE_P), bypassing Kernel#is_a? method dispatch. Used in the method
+ * probe wrapper (the Ruby < 3 argument-serialization split) to test whether
+ * the trailing argument is a hash without giving a user-installed method
+ * probe on Kernel#is_a? a chance to recurse. Matches Hash and its
+ * subclasses, like is_a?(Hash), since their instances are still T_HASH.
+ *
+ * @api private
+ */
+static VALUE is_hash(DDTRACE_UNUSED VALUE _self, VALUE obj) {
+  return RB_TYPE_P(obj, T_HASH) ? Qtrue : Qfalse;
+}
+
+// rb_iseq_type was added in Ruby 3.1 (commit 89a02d89 by Koichi Sasada,
+// 2021-12-19). It returns the iseq type as a Symbol. On Ruby < 3.1 this
+// function does not exist, so have_func('rb_iseq_type') in extconf.rb
+// gates compilation. When unavailable, backfill_registry falls back to
+// the first_lineno == 0 heuristic.
+#ifdef HAVE_RB_ISEQ_TYPE
+VALUE rb_iseq_type(const void *iseq);
+
+/*
+ * call-seq:
+ *   DI.iseq_type(iseq) -> Symbol
+ *
+ * Returns the type of an InstructionSequence as a symbol by calling
+ * the internal rb_iseq_type() function (available since Ruby 3.1).
+ *
+ * This method is only defined when rb_iseq_type is detected at compile
+ * time via have_func in extconf.rb. On Ruby < 3.1 it is not available
+ * and callers must use an alternative (e.g. first_lineno heuristic).
+ *
+ * Possible return values: :top, :method, :block, :class, :rescue,
+ * :ensure, :eval, :main, :plain.
+ *
+ * :top and :main represent whole-file iseqs (from require/load and the
+ * entry point script respectively). Other types represent sub-file
+ * constructs (method definitions, class bodies, blocks, etc.).
+ *
+ * Used by CodeTracker#backfill_registry to distinguish whole-file iseqs
+ * from per-method/block/class iseqs when populating the registry from
+ * the object space.
+ *
+ * @param iseq [RubyVM::InstructionSequence] The instruction sequence
+ * @return [Symbol] The iseq type
+ */
+static VALUE iseq_type(DDTRACE_UNUSED VALUE _self, VALUE iseq_val) {
+  const void *iseq = rb_iseqw_to_iseq(iseq_val);
+  if (!iseq) return Qnil;
+  return rb_iseq_type(iseq);
+}
+#endif
+
+void di_init(VALUE datadog_module) {
+  id_mesg = rb_intern("mesg");
+  id_datadog_di_in_probe = rb_intern("datadog_di_in_probe");
+
+  VALUE di_module = rb_define_module_under(datadog_module, "DI");
+  rb_define_singleton_method(di_module, "all_iseqs", all_iseqs, 0);
+  rb_define_singleton_method(di_module, "exception_message", exception_message, 1);
+  rb_define_singleton_method(di_module, "in_probe?", in_probe_p, 0);
+  rb_define_singleton_method(di_module, "enter_probe", enter_probe, 0);
+  rb_define_singleton_method(di_module, "leave_probe", leave_probe, 0);
+  rb_define_singleton_method(di_module, "hash?", is_hash, 1);
+#ifdef HAVE_RB_ISEQ_TYPE
+  rb_define_singleton_method(di_module, "iseq_type", iseq_type, 1);
+#endif
+}