ddtrace 1.6.1 → 1.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7a13232b0a3e34d8c05f4e9cd14083d3da7899e4cd4a8a52a95495193dce81b4
-  data.tar.gz: '0975d24547f318e657422c35ef54f8f68cff797436594d30514d163b55170220'
+  metadata.gz: '02285d37473e8fe6c731e55ae7864860a311e2b6d509f56a41f31094cd198f09'
+  data.tar.gz: 8857a0f43f96fef23309cd7fced88b846d5ad48834b6c87d0f40181cc3bd5dd9
 SHA512:
-  metadata.gz: da8917c06511999a28152037efa177fceba67d0329534bc8a462d17ffb151bfad0f360318262be095fc73ce42990051e6856b2cc3663317091b552d76730116c
-  data.tar.gz: 44a8062936ae2f20fdca4d524d6390825b87092d58ebb934d2e4b715f442b595be020752b3a2c0fd432d4dc3901ad611725adf4d70139c095d78e7b2b127fdd1
+  metadata.gz: afcde0e01258f5ab997fa6bcfa51c29a82f0e730258056a9efae86130445dbd63471269c047878b1ee554cc4ec956fe6e95a597ce65930e73549b0e32cbec50c
+  data.tar.gz: 292867c7c22146d0bd3147bb8863dd51fc89200a08efe640f05ae053fb93bdaeb068eefb6c89466b40ebbe749174844aa6b874c5a2b6aae30cb5055c9f2dc3cd

data/CHANGELOG.md

@@ -2,6 +2,58 @@
 
 ## [Unreleased]
 
+## [1.8.0] - 2022-12-14
+
+Release notes: https://github.com/DataDog/dd-trace-rb/releases/tag/v1.8.0
+
+As of ddtrace 1.8.0, CPU Profiling 2.0 is now in opt-in (that is, disabled by default) public beta. For more details,
+check the release notes.
+
+### Added
+
+* Core: Profiling: [PROF-6559] Mark Ruby CPU Profiling 2.0 as being in beta ([#2489][])
+* Tracing: Attempt to parse future version of TraceContext ([#2473][])
+* Tracing: Add DD_TRACE_PROPAGATION_STYLE option ([#2466][])
+* Integrations: Tracing: SQL comment propagation full mode with traceparent ([#2464][])
+* Integrations: Tracing: Wire W3C propagator to HTTP & gRPC propagation ([#2458][])
+* Integrations: Tracing: Auto-instrumentation with service_name from environmental variable ([#2455][])
+* Core: Integrations: Tracing: Deprecation notice for B3 propagation configuration ([#2454][])
+* Tracing: Add W3C Trace Context propagator ([#2451][])
+* Integrations: Tracing: Redis 5 Instrumentation ([#2428][])
+
+### Changed
+
+* Tracing: Changes `error.msg` to `error.message` for UNC ([#2469][])
+* Tracing: Semicolons not allowed in 'origin' ([#2461][])
+* Core: Dev/refactor: Tracing: Dev/internal: Move Utils#next_id and constants to Tracing::Utils ([#2463][])
+* Core: Dev/refactor: Tracing: Dev/internal: Move Tracing config settings from Core to Tracing ([#2459][])
+* Core: Dev/refactor: Tracing: Dev/internal: Move Tracing diagnostic code from Core to Tracing ([#2453][])
+
+### Fixed
+
+* Integrations: Tracing: Improve redis integration patching ([#2470][])
+* Tracing: Extra testing from W3C spec ([#2460][])
+
+## [1.7.0] - 2022-11-29
+
+### Added
+* Integrations: Support que 2 ([#2382][]) ([@danhodge][])
+* Tracing: Unified tagging `span.kind` as `server` and `client` ([#2365][])
+* Tracing: Adds `span.kind` tag for `kafka`, `sidekiq`, `racecar`,  `que`, `shoryuken`, `sneakers`, and `resque` ([#2420][], [#2419][], [#2413][], [#2394][])
+* Tracing: Adds `span.kind` with values `producer` and `consumer` for `delayed_job` ([#2393][])
+* Tracing: Adds `span.kind` as `client` for `redis` ([#2392][])
+* Appsec: Pass HTTP client IP to WAF ([#2316][])
+* Unified tagging `process_id` ([#2276][])
+
+### Changed
+* Allow `debase-ruby_core_source` 0.10.18 to be used ([#2435][])
+* Update AppSec ruleset to v1.4.2 ([#2390][])
+* Refactored clearing of profile data after Ruby app forks ([#2362][], [#2367][])
+* Tracing: Move distributed propagation to Contrib ([#2352][])
+
+### Fixed
+* Fix ddtrace installation issue when users have CI=true ([#2378][])
+
 ## [1.6.1] - 2022-11-16
 
 ### Changed
@@ -2198,7 +2250,9 @@ Release notes: https://github.com/DataDog/dd-trace-rb/releases/tag/v0.3.1
 
 Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 
-[Unreleased]: https://github.com/DataDog/dd-trace-rb/compare/v1.6.1...master
+[Unreleased]: https://github.com/DataDog/dd-trace-rb/compare/v1.8.0...master
+[1.8.0]: https://github.com/DataDog/dd-trace-rb/compare/v1.7.0...v1.8.0
+[1.7.0]: https://github.com/DataDog/dd-trace-rb/compare/v1.6.1...v1.7.0
 [1.6.1]: https://github.com/DataDog/dd-trace-rb/compare/v1.6.0...v1.6.1
 [1.6.0]: https://github.com/DataDog/dd-trace-rb/compare/v1.5.2...v1.6.0
 [1.5.2]: https://github.com/DataDog/dd-trace-rb/compare/v1.5.1...v1.5.2
@@ -3102,6 +3156,7 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [#2260]: https://github.com/DataDog/dd-trace-rb/issues/2260
 [#2265]: https://github.com/DataDog/dd-trace-rb/issues/2265
 [#2267]: https://github.com/DataDog/dd-trace-rb/issues/2267
+[#2276]: https://github.com/DataDog/dd-trace-rb/issues/2276
 [#2279]: https://github.com/DataDog/dd-trace-rb/issues/2279
 [#2283]: https://github.com/DataDog/dd-trace-rb/issues/2283
 [#2289]: https://github.com/DataDog/dd-trace-rb/issues/2289
@@ -3114,6 +3169,7 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [#2310]: https://github.com/DataDog/dd-trace-rb/issues/2310
 [#2311]: https://github.com/DataDog/dd-trace-rb/issues/2311
 [#2313]: https://github.com/DataDog/dd-trace-rb/issues/2313
+[#2316]: https://github.com/DataDog/dd-trace-rb/issues/2316
 [#2317]: https://github.com/DataDog/dd-trace-rb/issues/2317
 [#2318]: https://github.com/DataDog/dd-trace-rb/issues/2318
 [#2319]: https://github.com/DataDog/dd-trace-rb/issues/2319
@@ -3124,7 +3180,37 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [#2331]: https://github.com/DataDog/dd-trace-rb/issues/2331
 [#2335]: https://github.com/DataDog/dd-trace-rb/issues/2335
 [#2339]: https://github.com/DataDog/dd-trace-rb/issues/2339
+[#2352]: https://github.com/DataDog/dd-trace-rb/issues/2352
+[#2362]: https://github.com/DataDog/dd-trace-rb/issues/2362
 [#2363]: https://github.com/DataDog/dd-trace-rb/issues/2363
+[#2365]: https://github.com/DataDog/dd-trace-rb/issues/2365
+[#2367]: https://github.com/DataDog/dd-trace-rb/issues/2367
+[#2378]: https://github.com/DataDog/dd-trace-rb/issues/2378
+[#2382]: https://github.com/DataDog/dd-trace-rb/issues/2382
+[#2390]: https://github.com/DataDog/dd-trace-rb/issues/2390
+[#2392]: https://github.com/DataDog/dd-trace-rb/issues/2392
+[#2393]: https://github.com/DataDog/dd-trace-rb/issues/2393
+[#2394]: https://github.com/DataDog/dd-trace-rb/issues/2394
+[#2413]: https://github.com/DataDog/dd-trace-rb/issues/2413
+[#2419]: https://github.com/DataDog/dd-trace-rb/issues/2419
+[#2420]: https://github.com/DataDog/dd-trace-rb/issues/2420
+[#2428]: https://github.com/DataDog/dd-trace-rb/issues/2428
+[#2435]: https://github.com/DataDog/dd-trace-rb/issues/2435
+[#2451]: https://github.com/DataDog/dd-trace-rb/issues/2451
+[#2453]: https://github.com/DataDog/dd-trace-rb/issues/2453
+[#2454]: https://github.com/DataDog/dd-trace-rb/issues/2454
+[#2455]: https://github.com/DataDog/dd-trace-rb/issues/2455
+[#2458]: https://github.com/DataDog/dd-trace-rb/issues/2458
+[#2459]: https://github.com/DataDog/dd-trace-rb/issues/2459
+[#2460]: https://github.com/DataDog/dd-trace-rb/issues/2460
+[#2461]: https://github.com/DataDog/dd-trace-rb/issues/2461
+[#2463]: https://github.com/DataDog/dd-trace-rb/issues/2463
+[#2464]: https://github.com/DataDog/dd-trace-rb/issues/2464
+[#2466]: https://github.com/DataDog/dd-trace-rb/issues/2466
+[#2469]: https://github.com/DataDog/dd-trace-rb/issues/2469
+[#2470]: https://github.com/DataDog/dd-trace-rb/issues/2470
+[#2473]: https://github.com/DataDog/dd-trace-rb/issues/2473
+[#2489]: https://github.com/DataDog/dd-trace-rb/issues/2489
 [@AdrianLC]: https://github.com/AdrianLC
 [@Azure7111]: https://github.com/Azure7111
 [@BabyGroot]: https://github.com/BabyGroot
@@ -3170,6 +3256,7 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [@components]: https://github.com/components
 [@cswatt]: https://github.com/cswatt
 [@cwoodcox]: https://github.com/cwoodcox
+[@danhodge]: https://github.com/danhodge
 [@dasch]: https://github.com/dasch
 [@dim]: https://github.com/dim
 [@dirk]: https://github.com/dirk
@@ -3268,4 +3355,4 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [@y-yagi]: https://github.com/y-yagi
 [@yujideveloper]: https://github.com/yujideveloper
 [@yukimurasawa]: https://github.com/yukimurasawa
-[@zachmccormick]: https://github.com/zachmccormick
+[@zachmccormick]: https://github.com/zachmccormick

data/README.md

@@ -1,5 +1,3 @@
-**We've recently released the 1.x version series. If you're upgrading from a 0.x version, check out our [upgrade guide](https://github.com/DataDog/dd-trace-rb/blob/master/docs/UpgradeGuide.md#from-0x-to-10).**
-
 # Datadog Trace Client
 
 [![Gem](https://img.shields.io/gem/v/ddtrace)](https://rubygems.org/gems/ddtrace/)
@@ -12,6 +10,8 @@ databases and microservices so that developers have great visiblity into bottlen
 
 ## Getting started
 
+**If you're upgrading from a 0.x version, check out our [upgrade guide](https://github.com/DataDog/dd-trace-rb/blob/master/docs/UpgradeGuide.md#from-0x-to-10).**
+
 For a basic product overview, check out our [setup documentation][setup docs].
 
 For installation, configuration, and details about using the API, check out our [API documentation][api docs] and [gem documentation][gem docs].

data/ext/ddtrace_profiling_loader/extconf.rb

@@ -26,7 +26,7 @@ end
 
 # Because we can't control what compiler versions our customers use, shipping with -Werror by default is a no-go.
 # But we can enable it in CI, so that we quickly spot any new warnings that just got introduced.
-add_compiler_flag '-Werror' if ENV['CI'] == 'true'
+add_compiler_flag '-Werror' if ENV['DDTRACE_CI'] == 'true'
 
 # Older gcc releases may not default to C99 and we need to ask for this. This is also used:
 # * by upstream Ruby -- search for gnu99 in the codebase
@@ -34,7 +34,7 @@ add_compiler_flag '-Werror' if ENV['CI'] == 'true'
 #   (https://github.com/msgpack/msgpack-ruby/blob/18ce08f6d612fe973843c366ac9a0b74c4e50599/ext/msgpack/extconf.rb#L8)
 add_compiler_flag '-std=gnu99'
 
-# Gets really noisy when we include the MJIT header, let's omit it
+# Gets really noisy when we include the MJIT header, let's omit it (TODO: Use #pragma GCC diagnostic instead?)
 add_compiler_flag '-Wno-unused-function'
 
 # Allow defining variables at any point in a function
@@ -55,6 +55,9 @@ add_compiler_flag '-Wunused-parameter'
 # For more details see https://gcc.gnu.org/wiki/Visibility
 add_compiler_flag '-fvisibility=hidden'
 
+# Avoid legacy C definitions
+add_compiler_flag '-Wold-style-definition'
+
 # Enable all other compiler warnings
 add_compiler_flag '-Wall'
 add_compiler_flag '-Wextra'

data/ext/ddtrace_profiling_native_extension/NativeExtensionDesign.md

@@ -57,7 +57,7 @@ Non-exhaustive list of APIs that cause exceptions to be raised:
 
 * `Check_TypedStruct`, `Check_Type`, `ENFORCE_TYPE`
 * `rb_funcall`
-* `rb_thread_call_without_gvl`, `rb_thread_call_without_gvl2`
+* `rb_thread_call_without_gvl`
 * [Numeric conversion APIs, e.g. `NUM2LONG`, `NUM2INT`, etc.](https://silverhammermba.github.io/emberb/c/?#translation)
 * Our `char_slice_from_ruby_string` helper
 

data/ext/ddtrace_profiling_native_extension/clock_id_from_pthread.c

@@ -7,11 +7,12 @@
 #include <pthread.h>
 #include <time.h>
 #include <errno.h>
-
 #include <ruby.h>
+
 #include "helpers.h"
 #include "private_vm_api_access.h"
 #include "clock_id.h"
+#include "time_helpers.h"
 
 // Validate that our home-cooked pthread_id_for() matches pthread_self() for the current thread
 void self_test_clock_id(void) {
@@ -68,7 +69,7 @@ thread_cpu_time thread_cpu_time_for(thread_cpu_time_id time_id) {
   // TODO: Include the error code in some way in the output?
   if (clock_gettime(time_id.clock_id, &current_cpu) != 0) return error;
 
-  return (thread_cpu_time) {.valid = true, .result_ns = current_cpu.tv_nsec + (current_cpu.tv_sec * 1000 * 1000 * 1000)};
+  return (thread_cpu_time) {.valid = true, .result_ns = current_cpu.tv_nsec + SECONDS_AS_NS(current_cpu.tv_sec)};
 }
 
 #endif

data/ext/ddtrace_profiling_native_extension/collectors_cpu_and_wall_time.c

@@ -1,11 +1,13 @@
 #include <ruby.h>
-#include "helpers.h"
+
 #include "clock_id.h"
+#include "collectors_cpu_and_wall_time.h"
 #include "collectors_stack.h"
+#include "helpers.h"
 #include "libdatadog_helpers.h"
 #include "private_vm_api_access.h"
 #include "stack_recorder.h"
-#include "collectors_cpu_and_wall_time.h"
+#include "time_helpers.h"
 
 // Used to periodically sample threads, recording elapsed CPU-time and Wall-time between samples.
 //
@@ -60,16 +62,16 @@
 
 #define INVALID_TIME -1
 #define THREAD_ID_LIMIT_CHARS 44 // Why 44? "#{2**64} (#{2**64})".size + 1 for \0
-#define RAISE_ON_FAILURE true
-#define DO_NOT_RAISE_ON_FAILURE false
 #define IS_WALL_TIME true
 #define IS_NOT_WALL_TIME false
 #define MISSING_TRACER_CONTEXT_KEY 0
 
-static ID at_active_trace_id; // id of :@active_trace in Ruby
-static ID at_root_span_id;    // id of :@root_span in Ruby
 static ID at_active_span_id;  // id of :@active_span in Ruby
+static ID at_active_trace_id; // id of :@active_trace in Ruby
 static ID at_id_id;           // id of :@id in Ruby
+static ID at_resource_id;     // id of :@resource in Ruby
+static ID at_root_span_id;    // id of :@root_span in Ruby
+static ID at_type_id;         // id of :@type in Ruby
 
 // Contains state for a single CpuAndWallTime instance
 struct cpu_and_wall_time_collector_state {
@@ -91,7 +93,7 @@ struct cpu_and_wall_time_collector_state {
   // is not (just) a stat.
   unsigned int sample_count;
 
-  struct {
+  struct stats {
     // Track how many garbage collection samples we've taken.
     unsigned int gc_samples;
     // See cpu_and_wall_time_collector_on_gc_start for details
@@ -126,9 +128,9 @@ struct trace_identifiers {
 
   bool valid;
   ddog_CharSlice local_root_span_id;
-  ddog_CharSlice span_id;
+  uint64_t span_id;
   char local_root_span_id_buffer[MAXIMUM_LENGTH_64_BIT_IDENTIFIER];
-  char span_id_buffer[MAXIMUM_LENGTH_64_BIT_IDENTIFIER];
+  VALUE trace_endpoint;
 };
 
 static void cpu_and_wall_time_collector_typed_data_mark(void *state_ptr);
@@ -161,10 +163,11 @@ static int remove_if_dead_thread(st_data_t key_thread, st_data_t value_context,
 static VALUE _native_per_thread_context(VALUE self, VALUE collector_instance);
 static long update_time_since_previous_sample(long *time_at_previous_sample_ns, long current_time_ns, long gc_start_time_ns, bool is_wall_time);
 static long cpu_time_now_ns(struct per_thread_context *thread_context);
-static long wall_time_now_ns(bool raise_on_failure);
 static long thread_id_for(VALUE thread);
 static VALUE _native_stats(VALUE self, VALUE collector_instance);
 static void trace_identifiers_for(struct cpu_and_wall_time_collector_state *state, VALUE thread, struct trace_identifiers *trace_identifiers_result);
+static bool is_type_web(VALUE root_span_type);
+static VALUE _native_reset_after_fork(DDTRACE_UNUSED VALUE self, VALUE collector_instance);
 
 void collectors_cpu_and_wall_time_init(VALUE profiling_module) {
   VALUE collectors_module = rb_define_module_under(profiling_module, "Collectors");
@@ -184,6 +187,7 @@ void collectors_cpu_and_wall_time_init(VALUE profiling_module) {
 
   rb_define_singleton_method(collectors_cpu_and_wall_time_class, "_native_initialize", _native_initialize, 4);
   rb_define_singleton_method(collectors_cpu_and_wall_time_class, "_native_inspect", _native_inspect, 1);
+  rb_define_singleton_method(collectors_cpu_and_wall_time_class, "_native_reset_after_fork", _native_reset_after_fork, 1);
   rb_define_singleton_method(testing_module, "_native_sample", _native_sample, 1);
   rb_define_singleton_method(testing_module, "_native_on_gc_start", _native_on_gc_start, 1);
   rb_define_singleton_method(testing_module, "_native_on_gc_finish", _native_on_gc_finish, 1);
@@ -192,10 +196,12 @@ void collectors_cpu_and_wall_time_init(VALUE profiling_module) {
   rb_define_singleton_method(testing_module, "_native_per_thread_context", _native_per_thread_context, 1);
   rb_define_singleton_method(testing_module, "_native_stats", _native_stats, 1);
 
-  at_active_trace_id = rb_intern_const("@active_trace");
-  at_root_span_id = rb_intern_const("@root_span");
   at_active_span_id = rb_intern_const("@active_span");
+  at_active_trace_id = rb_intern_const("@active_trace");
   at_id_id = rb_intern_const("@id");
+  at_resource_id = rb_intern_const("@resource");
+  at_root_span_id = rb_intern_const("@root_span");
+  at_type_id = rb_intern_const("@type");
 }
 
 // This structure is used to define a Ruby object that stores a pointer to a struct cpu_and_wall_time_collector_state
@@ -292,7 +298,7 @@ static VALUE _native_initialize(DDTRACE_UNUSED VALUE _self, VALUE collector_inst
 // This method exists only to enable testing Datadog::Profiling::Collectors::CpuAndWallTime behavior using RSpec.
 // It SHOULD NOT be used for other purposes.
 static VALUE _native_sample(DDTRACE_UNUSED VALUE _self, VALUE collector_instance) {
-  cpu_and_wall_time_collector_sample(collector_instance);
+  cpu_and_wall_time_collector_sample(collector_instance, monotonic_wall_time_now_ns(RAISE_ON_FAILURE));
   return Qtrue;
 }
 
@@ -324,12 +330,11 @@ static VALUE _native_sample_after_gc(DDTRACE_UNUSED VALUE self, VALUE collector_
 // Assumption 3: This function IS NOT called from a signal handler. This function is not async-signal-safe.
 // Assumption 4: This function IS NOT called in a reentrant way.
 // Assumption 5: This function is called from the main Ractor (if Ruby has support for Ractors).
-VALUE cpu_and_wall_time_collector_sample(VALUE self_instance) {
+void cpu_and_wall_time_collector_sample(VALUE self_instance, long current_monotonic_wall_time_ns) {
   struct cpu_and_wall_time_collector_state *state;
   TypedData_Get_Struct(self_instance, struct cpu_and_wall_time_collector_state, &cpu_and_wall_time_collector_typed_data, state);
 
   VALUE threads = ddtrace_thread_list();
-  long current_wall_time_ns = wall_time_now_ns(RAISE_ON_FAILURE);
 
   const long thread_count = RARRAY_LEN(threads);
   for (long i = 0; i < thread_count; i++) {
@@ -346,7 +351,7 @@ VALUE cpu_and_wall_time_collector_sample(VALUE self_instance) {
     );
     long wall_time_elapsed_ns = update_time_since_previous_sample(
       &thread_context->wall_time_at_previous_sample_ns,
-      current_wall_time_ns,
+      current_monotonic_wall_time_ns,
       thread_context->gc_tracking.wall_time_at_start_ns,
       IS_WALL_TIME
     );
@@ -368,12 +373,9 @@ VALUE cpu_and_wall_time_collector_sample(VALUE self_instance) {
 
   state->sample_count++;
 
-  // TODO: This seems somewhat overkill and inefficient to do often; right now we just doing every few samples
+  // TODO: This seems somewhat overkill and inefficient to do often; right now we just do it every few samples
   // but there's probably a better way to do this if we actually track when threads finish
   if (state->sample_count % 100 == 0) remove_context_for_dead_threads(state);
-
-  // Return a VALUE to make it easier to call this function from Ruby APIs that expect a return value (such as rb_rescue2)
-  return Qnil;
 }
 
 // This function gets called when Ruby is about to start running the Garbage Collector on the current thread.
@@ -417,7 +419,7 @@ void cpu_and_wall_time_collector_on_gc_start(VALUE self_instance) {
     thread_context->gc_tracking.wall_time_at_finish_ns != INVALID_TIME) return;
 
   // Here we record the wall-time first and in on_gc_finish we record it second to avoid having wall-time be slightly < cpu-time
-  thread_context->gc_tracking.wall_time_at_start_ns = wall_time_now_ns(DO_NOT_RAISE_ON_FAILURE);
+  thread_context->gc_tracking.wall_time_at_start_ns = monotonic_wall_time_now_ns(DO_NOT_RAISE_ON_FAILURE);
   thread_context->gc_tracking.cpu_time_at_start_ns = cpu_time_now_ns(thread_context);
 }
 
@@ -453,7 +455,7 @@ void cpu_and_wall_time_collector_on_gc_finish(VALUE self_instance) {
 
   // Here we record the wall-time second and in on_gc_start we record it first to avoid having wall-time be slightly < cpu-time
   thread_context->gc_tracking.cpu_time_at_finish_ns = cpu_time_now_ns(thread_context);
-  thread_context->gc_tracking.wall_time_at_finish_ns = wall_time_now_ns(DO_NOT_RAISE_ON_FAILURE);
+  thread_context->gc_tracking.wall_time_at_finish_ns = monotonic_wall_time_now_ns(DO_NOT_RAISE_ON_FAILURE);
 }
 
 // This function gets called shortly after Ruby has finished running the Garbage Collector.
@@ -570,12 +572,29 @@ static void trigger_sample_for_thread(
     };
   }
 
-  struct trace_identifiers trace_identifiers_result = {.valid = false};
+  struct trace_identifiers trace_identifiers_result = {.valid = false, .trace_endpoint = Qnil};
   trace_identifiers_for(state, thread, &trace_identifiers_result);
 
   if (trace_identifiers_result.valid) {
     labels[label_pos++] = (ddog_Label) {.key = DDOG_CHARSLICE_C("local root span id"), .str = trace_identifiers_result.local_root_span_id};
-    labels[label_pos++] = (ddog_Label) {.key = DDOG_CHARSLICE_C("span id"), .str = trace_identifiers_result.span_id};
+    labels[label_pos++] = (ddog_Label) {.key = DDOG_CHARSLICE_C("span id"), .num = trace_identifiers_result.span_id};
+
+    if (trace_identifiers_result.trace_endpoint != Qnil) {
+      // The endpoint gets recorded in a different way because it is mutable in the tracer and can change during a
+      // trace.
+      //
+      // Instead of each sample for the same local_root_span_id getting a potentially-different endpoint,
+      // `record_endpoint` (via libdatadog) keeps a list of local_root_span_id values and their most-recently-seen
+      // endpoint values, and at serialization time the most-recently-seen endpoint is applied to all relevant samples.
+      //
+      // This is why the endpoint is not directly added in this function to the labels array, although it will later
+      // show up in the array in the output pprof.
+      record_endpoint(
+        state->recorder_instance,
+        trace_identifiers_result.local_root_span_id,
+        char_slice_from_ruby_string(trace_identifiers_result.trace_endpoint)
+      );
+    }
   }
 
   // The number of times `label_pos++` shows up in this function needs to match `max_label_count`. To avoid "oops I
@@ -668,8 +687,6 @@ static VALUE per_thread_context_st_table_as_ruby_hash(struct cpu_and_wall_time_c
   return result;
 }
 
-#define VALUE_COUNT(array) (sizeof(array) / sizeof(VALUE))
-
 static int per_thread_context_as_ruby_hash(st_data_t key_thread, st_data_t value_context, st_data_t result_hash) {
   VALUE thread = (VALUE) key_thread;
   struct per_thread_context *thread_context = (struct per_thread_context*) value_context;
@@ -698,8 +715,8 @@ static VALUE stats_as_ruby_hash(struct cpu_and_wall_time_collector_state *state)
   // Update this when modifying state struct (stats inner struct)
   VALUE stats_as_hash = rb_hash_new();
   VALUE arguments[] = {
-    ID2SYM(rb_intern("gc_samples")),                               /* => */ INT2NUM(state->stats.gc_samples),
-    ID2SYM(rb_intern("gc_samples_missed_due_to_missing_context")), /* => */ INT2NUM(state->stats.gc_samples_missed_due_to_missing_context),
+    ID2SYM(rb_intern("gc_samples")),                               /* => */ UINT2NUM(state->stats.gc_samples),
+    ID2SYM(rb_intern("gc_samples_missed_due_to_missing_context")), /* => */ UINT2NUM(state->stats.gc_samples_missed_due_to_missing_context),
   };
   for (long unsigned int i = 0; i < VALUE_COUNT(arguments); i += 2) rb_hash_aset(stats_as_hash, arguments[i], arguments[i+1]);
   return stats_as_hash;
@@ -769,18 +786,6 @@ static long update_time_since_previous_sample(long *time_at_previous_sample_ns,
   return elapsed_time_ns;
 }
 
-// Safety: This function is assumed never to raise exceptions by callers when raise_on_failure == false
-static long wall_time_now_ns(bool raise_on_failure) {
-  struct timespec current_monotonic;
-
-  if (clock_gettime(CLOCK_MONOTONIC, &current_monotonic) != 0) {
-    if (raise_on_failure) rb_sys_fail("Failed to read CLOCK_MONOTONIC");
-    else return 0;
-  }
-
-  return current_monotonic.tv_nsec + (current_monotonic.tv_sec * 1000 * 1000 * 1000);
-}
-
 // Safety: This function is assumed never to raise exceptions by callers
 static long cpu_time_now_ns(struct per_thread_context *thread_context) {
   thread_cpu_time cpu_time = thread_cpu_time_for(thread_context->thread_cpu_time_id);
@@ -847,19 +852,48 @@ static void trace_identifiers_for(struct cpu_and_wall_time_collector_state *stat
   if (numeric_local_root_span_id == Qnil || numeric_span_id == Qnil) return;
 
   unsigned long long local_root_span_id = NUM2ULL(numeric_local_root_span_id);
-  unsigned long long span_id = NUM2ULL(numeric_span_id);
-
   snprintf(trace_identifiers_result->local_root_span_id_buffer, MAXIMUM_LENGTH_64_BIT_IDENTIFIER, "%llu", local_root_span_id);
-  snprintf(trace_identifiers_result->span_id_buffer, MAXIMUM_LENGTH_64_BIT_IDENTIFIER, "%llu", span_id);
 
   trace_identifiers_result->local_root_span_id = (ddog_CharSlice) {
     .ptr = trace_identifiers_result->local_root_span_id_buffer,
     .len = strlen(trace_identifiers_result->local_root_span_id_buffer)
   };
-  trace_identifiers_result->span_id = (ddog_CharSlice) {
-    .ptr = trace_identifiers_result->span_id_buffer,
-    .len = strlen(trace_identifiers_result->span_id_buffer)
-  };
+  trace_identifiers_result->span_id = NUM2ULL(numeric_span_id);
 
   trace_identifiers_result->valid = true;
+
+  VALUE root_span_type = rb_ivar_get(root_span, at_type_id /* @type */);
+  if (root_span_type == Qnil || !is_type_web(root_span_type)) return;
+
+  VALUE trace_resource = rb_ivar_get(active_trace, at_resource_id /* @resource */);
+  if (RB_TYPE_P(trace_resource, T_STRING)) {
+    trace_identifiers_result->trace_endpoint = trace_resource;
+  } else if (trace_resource == Qnil) {
+    // Fall back to resource from span, if any
+    trace_identifiers_result->trace_endpoint = rb_ivar_get(root_span, at_resource_id /* @resource */);
+  }
+}
+
+static bool is_type_web(VALUE root_span_type) {
+  ENFORCE_TYPE(root_span_type, T_STRING);
+
+  return RSTRING_LEN(root_span_type) == strlen("web") &&
+    (memcmp("web", StringValuePtr(root_span_type), strlen("web")) == 0);
+}
+
+// After the Ruby VM forks, this method gets called in the child process to clean up any leftover state from the parent.
+//
+// Assumption: This method gets called BEFORE restarting profiling -- e.g. there are no components attempting to
+// trigger samples at the same time.
+static VALUE _native_reset_after_fork(DDTRACE_UNUSED VALUE self, VALUE collector_instance) {
+  struct cpu_and_wall_time_collector_state *state;
+  TypedData_Get_Struct(collector_instance, struct cpu_and_wall_time_collector_state, &cpu_and_wall_time_collector_typed_data, state);
+
+  st_clear(state->hash_map_per_thread_context);
+
+  state->stats = (struct stats) {}; // Resets all stats back to zero
+
+  rb_funcall(state->recorder_instance, rb_intern("reset_after_fork"), 0);
+
+  return Qtrue;
 }

data/ext/ddtrace_profiling_native_extension/collectors_cpu_and_wall_time.h

@@ -2,7 +2,7 @@
 
 #include <ruby.h>
 
-VALUE cpu_and_wall_time_collector_sample(VALUE self_instance);
+void cpu_and_wall_time_collector_sample(VALUE self_instance, long current_monotonic_wall_time_ns);
 VALUE cpu_and_wall_time_collector_sample_after_gc(VALUE self_instance);
 void cpu_and_wall_time_collector_on_gc_start(VALUE self_instance);
 void cpu_and_wall_time_collector_on_gc_finish(VALUE self_instance);