ddtrace 1.14.0 → 1.15.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ae7238650b45da9d8bf8f5834af2a29153081e798c2c8733c9bffb8768b6c430
-  data.tar.gz: b9bc8f36d37b91d18754892bd8dbc92cd515b9e38caf8561e23b665f14ee5e9e
+  metadata.gz: 18896f24a83a74a5ea6f4adfb3baad9fec922f0d3c3e8d087d28805744584d40
+  data.tar.gz: 173d93233e49670ed94a9d3cd7501f50c596ceb7624293e4c1992513e6121fa6
 SHA512:
-  metadata.gz: 4b4cdffa9d4ee39e6763be4a87dbdf3e351fd29915d4eaea45dc9c800543c65c8eb21375138422280cfed5715d8f23e95e58190752af6631c14453d39b29b263
-  data.tar.gz: fbc970e48e5e5038775368618e672ab7b2405cd6ea9b899136b34e9ae776f26a92436087fb0810ef7d66574c2f694cdea03197cf980b08dc35a9defcf4c3dc74
+  metadata.gz: 174e727aad6f68b2873e88ac71b5a563075c501f4fbd75cad563735c765962d5753f6025a0a75c257f12a7a89fc83713e3ff249f2e2815803786ed709e35daff
+  data.tar.gz: bc2bbcc9d601473e871dee51e53745045b37204bb902dd2f29deac2ea4c94bdc7aa6c016eecfd2648ccac13393b612400c9167bdbd115271893e561fcb12a8ce

data/CHANGELOG.md

@@ -2,6 +2,135 @@
 
 ## [Unreleased]
 
+## [1.15.0] - 2023-10-09
+
+### Highlights
+
+#### Timeline view for Profiler beta
+
+As of ddtrace 1.15.0, the Profiler now supports gathering data for the new
+[Timeline view](https://docs.datadoghq.com/profiler/profile_visualizations/#timeline-view).
+
+The Timeline view allows you to look at time-based patterns and work distribution over the period of a single profile: you can look at what individual threads were doing, and when 🎉
+
+You can use the timeline view both when looking at individual profiles, as well as when scoped to a given trace.
+
+You can enable it:
+
+* Using an environment variable by setting `DD_PROFILING_EXPERIMENTAL_TIMELINE_ENABLED=true`
+* Or via code by adding to your `Datadog.configure` block:
+
+```ruby
+Datadog.configure do |c|
+  # … existing configuration …
+  c.profiling.advanced.experimental_timeline_enabled = true
+end
+```
+
+Give it a try, let us know what you think!
+
+(Note: We do not recommend enabling this feature prior to 1.15.0!)
+
+#### google-protobuf dependency is no longer needed by the Profiler
+
+As of ddtrace version 1.15.0, the `google-protobuf` gem is no longer needed to enable the Profiler.
+
+If you've added this gem to your `Gemfile`/`gems.rb` file as part of enabling the Profiler, you can
+remove it now. (If you're curious, we've internally replaced this dependency with the `libdatadog` gem.)
+
+#### Configure blocking responses for AppSec via configuration or Remote Configuration
+
+As of dd-trace-rb 1.15.0, AppSec supports configuring the blocking response.
+
+You can configure the blocking response via:
+- Using the ENV variables: `DD_APPSEC_HTTP_BLOCKED_TEMPLATE_HTML=#{file_name}`, and `DD_APPSEC_HTTP_BLOCKED_TEMPLATE_JSON=#{file_name}`
+- Via code by adding to your `Datadog.configure` block:
+
+```ruby
+Datadog.configure do |c|
+  # … existing configuration …
+  c.appsec.block.templates.html = "#{file_name}"
+  c.appsec.block.templates.json = "#{file_name}"
+end
+```
+
+- Using the Remote configuration UI. This option allow you to configure the status code and the blocking behaviour. You can redirect malicious attacker to custome pages.
+  You can find more information on the [official documentation](https://docs.datadoghq.com/security/application_security/threats/protection/#customize-protection-behavior)
+
+#### Configure agentless mode for CI visibility
+
+If you are using CI visibility with a cloud CI provider without access to the underlying worker nodes, such as GitHub Actions or CircleCI, configure the library to use the Agentless mode.
+
+For this, set the following environment variables:
+- DD_CIVISIBILITY_AGENTLESS_ENABLED=true
+- DD_API_KEY=<your_api_key>
+
+Additionally, configure which Datadog site you want to send your data to:
+- DD_SITE (default: datadoghq.com)
+
+You can also enable agentless mode with `Datadog.configure` block:
+
+```ruby
+Datadog.configure do |c|
+  # … existing configuration …
+  c.ci.agentless_mode_enabled = true
+  # don't forget to set DD_API_KEY env variable!
+end
+```
+
+### Added
+* Profiling: Import java-profiler PID controller and port it to C ([#3190][])
+* Tracing: Support Opensearch 3 ([#3189][])
+* bump datadog-ci dependency to 0.2 ([#3186][])
+* Enable allocation counting feature by default for some Ruby 3 versions ([#3176][])
+* Tracing: Introduce async configuration for test mode to use standard writer when needed ([#3158][])
+* Appsec: Update AppSec rules to 1.8.0 ([#3140][])
+* Appsec: Update AppSec rules to 1.7.2 ([#3139][])
+* Appsec: ASM API security. Schema extraction ([#3131][], [#3166][], [#3177][])
+* Tracing: peer.service adjustment for sql propagation with DBM ([#3127][])
+* Ci-app: CI visibility: validate git tags ([#3100][])
+* Appsec: Enable configuring blocking response via Remote Configuration  ([#3099][])
+* Profiling: Record allocation type when sampling objects ([#3096][])
+* Appsec: Uppgrade libddwaf-rb version to 1.11.0 ([#3087][])
+* Profiling: Mergequeue-status: removed: Include 'ruby vm type' in profiler allocation samples ([#3074][])
+* Detect WebMock to disable telemetry and remote configuration  ([#3065][])
+* Ensure Rails testing is considered a development environment ([#3062][])
+* Tracing: Implements `_dd.base_service` tag ([#3018][])
+* Appsec: Allow blocking response template configuration via ENV variables  ([#2975][])
+* Ci-app: agentless mode ([#3186][])
+
+### Changed
+* Appsec: skip passing waf addresses when the value is empty ([#3188][])
+* Tracing: Disable memcached command tag by default ([#3171][])
+* Profiling: Upgrade to libdatadog 5 ([#3169][])
+* Profiling: Restore support for Ruby 3.3 ([#3167][])
+* Bump debase-ruby_core_source dependency to 3.2.2 ([#3163][])
+* Profiling: Add approximate thread state categorization for timeline ([#3162][])
+* Tracing: remove variable helpers module from our configuration DSL ([#3152][])
+* Profiling: Tracing: ekump/decouple core transport ([#3150][])
+* Test:Remove explicit dependency on `addressable` ([#3148][])
+* Detect Cucumber as a development environment ([#3145][])
+* Tracing: rename core configuration option on_set to after_set ([#3107][])
+* Profiling: Upgrade to libdatadog 4 ([#3104][])
+* Profiling: Wire up allocation sampling into `CpuAndWallTimeWorker` ([#3103][])
+* Tracing: rename experimental_default_proc to default_proc ([#3091][])
+* remove `delegate_to` configuration option from our DSL ([#3086][])
+* Ci-app: Fix Datadog::CI::Environment to support the new CI specs ([#3080][])
+* Tracing: Use first valid extracted style for distributed tracing ([#2879][])
+
+### Fixed
+* Profiling: Add workaround for incorrect invoke location when logging gem is in use ([#3183][])
+* Profiling: Fix missing endpoint profiling when request_queuing is enabled in rack instrumentation ([#3109][])
+* Appsec: Fix a bug with ASM span tags reporting the number of WAF failed loaded rules ([#3106][])
+* Tracing: Fix tagging with empty data ([#3102][])
+* Tracing: fix(grpc): allow custom error_handler for client interceptor ([#3095][])
+* Tracing: Correctly set `rails.cache.backend` span tag for multiple stores ([#3060][])
+
+### Removed
+* Profiling: Remove legacy profiler codepath ([#3172][])
+* Ci-app: remove CI module and add a dependency on datadog-ci gem ([#3128][])
+* Tracing: Remove `depends_on` from Core Configuration option ([#3085][])
+
 ## [1.14.0] - 2023-08-24
 
 ### Added
@@ -2534,7 +2663,8 @@ 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/v1.14.0...master
+[Unreleased]: https://github.com/DataDog/dd-trace-rb/compare/v1.15.0...master
+[1.15.0]: https://github.com/DataDog/dd-trace-rb/compare/v1.14.0...v1.15.0
 [1.14.0]: https://github.com/DataDog/dd-trace-rb/compare/v1.13.1...1.14.0
 [1.13.1]: https://github.com/DataDog/dd-trace-rb/compare/v1.13.0...1.13.1
 [1.13.0]: https://github.com/DataDog/dd-trace-rb/compare/v1.12.1...v1.13.0
@@ -3624,6 +3754,7 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [#2874]: https://github.com/DataDog/dd-trace-rb/issues/2874
 [#2875]: https://github.com/DataDog/dd-trace-rb/issues/2875
 [#2877]: https://github.com/DataDog/dd-trace-rb/issues/2877
+[#2879]: https://github.com/DataDog/dd-trace-rb/issues/2879
 [#2882]: https://github.com/DataDog/dd-trace-rb/issues/2882
 [#2883]: https://github.com/DataDog/dd-trace-rb/issues/2883
 [#2890]: https://github.com/DataDog/dd-trace-rb/issues/2890
@@ -3659,6 +3790,7 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [#2972]: https://github.com/DataDog/dd-trace-rb/issues/2972
 [#2973]: https://github.com/DataDog/dd-trace-rb/issues/2973
 [#2974]: https://github.com/DataDog/dd-trace-rb/issues/2974
+[#2975]: https://github.com/DataDog/dd-trace-rb/issues/2975
 [#2977]: https://github.com/DataDog/dd-trace-rb/issues/2977
 [#2978]: https://github.com/DataDog/dd-trace-rb/issues/2978
 [#2982]: https://github.com/DataDog/dd-trace-rb/issues/2982
@@ -3671,6 +3803,7 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [#3005]: https://github.com/DataDog/dd-trace-rb/issues/3005
 [#3007]: https://github.com/DataDog/dd-trace-rb/issues/3007
 [#3011]: https://github.com/DataDog/dd-trace-rb/issues/3011
+[#3018]: https://github.com/DataDog/dd-trace-rb/issues/3018
 [#3019]: https://github.com/DataDog/dd-trace-rb/issues/3019
 [#3020]: https://github.com/DataDog/dd-trace-rb/issues/3020
 [#3022]: https://github.com/DataDog/dd-trace-rb/issues/3022
@@ -3685,8 +3818,51 @@ Git diff: https://github.com/DataDog/dd-trace-rb/compare/v0.3.0...v0.3.1
 [#3054]: https://github.com/DataDog/dd-trace-rb/issues/3054
 [#3056]: https://github.com/DataDog/dd-trace-rb/issues/3056
 [#3057]: https://github.com/DataDog/dd-trace-rb/issues/3057
+[#3060]: https://github.com/DataDog/dd-trace-rb/issues/3060
 [#3061]: https://github.com/DataDog/dd-trace-rb/issues/3061
+[#3062]: https://github.com/DataDog/dd-trace-rb/issues/3062
+[#3065]: https://github.com/DataDog/dd-trace-rb/issues/3065
 [#3070]: https://github.com/DataDog/dd-trace-rb/issues/3070
+[#3074]: https://github.com/DataDog/dd-trace-rb/issues/3074
+[#3080]: https://github.com/DataDog/dd-trace-rb/issues/3080
+[#3085]: https://github.com/DataDog/dd-trace-rb/issues/3085
+[#3086]: https://github.com/DataDog/dd-trace-rb/issues/3086
+[#3087]: https://github.com/DataDog/dd-trace-rb/issues/3087
+[#3091]: https://github.com/DataDog/dd-trace-rb/issues/3091
+[#3095]: https://github.com/DataDog/dd-trace-rb/issues/3095
+[#3096]: https://github.com/DataDog/dd-trace-rb/issues/3096
+[#3099]: https://github.com/DataDog/dd-trace-rb/issues/3099
+[#3100]: https://github.com/DataDog/dd-trace-rb/issues/3100
+[#3102]: https://github.com/DataDog/dd-trace-rb/issues/3102
+[#3103]: https://github.com/DataDog/dd-trace-rb/issues/3103
+[#3104]: https://github.com/DataDog/dd-trace-rb/issues/3104
+[#3106]: https://github.com/DataDog/dd-trace-rb/issues/3106
+[#3107]: https://github.com/DataDog/dd-trace-rb/issues/3107
+[#3109]: https://github.com/DataDog/dd-trace-rb/issues/3109
+[#3127]: https://github.com/DataDog/dd-trace-rb/issues/3127
+[#3128]: https://github.com/DataDog/dd-trace-rb/issues/3128
+[#3131]: https://github.com/DataDog/dd-trace-rb/issues/3131
+[#3139]: https://github.com/DataDog/dd-trace-rb/issues/3139
+[#3140]: https://github.com/DataDog/dd-trace-rb/issues/3140
+[#3145]: https://github.com/DataDog/dd-trace-rb/issues/3145
+[#3148]: https://github.com/DataDog/dd-trace-rb/issues/3148
+[#3150]: https://github.com/DataDog/dd-trace-rb/issues/3150
+[#3152]: https://github.com/DataDog/dd-trace-rb/issues/3152
+[#3158]: https://github.com/DataDog/dd-trace-rb/issues/3158
+[#3162]: https://github.com/DataDog/dd-trace-rb/issues/3162
+[#3163]: https://github.com/DataDog/dd-trace-rb/issues/3163
+[#3166]: https://github.com/DataDog/dd-trace-rb/issues/3166
+[#3167]: https://github.com/DataDog/dd-trace-rb/issues/3167
+[#3169]: https://github.com/DataDog/dd-trace-rb/issues/3169
+[#3171]: https://github.com/DataDog/dd-trace-rb/issues/3171
+[#3172]: https://github.com/DataDog/dd-trace-rb/issues/3172
+[#3176]: https://github.com/DataDog/dd-trace-rb/issues/3176
+[#3177]: https://github.com/DataDog/dd-trace-rb/issues/3177
+[#3183]: https://github.com/DataDog/dd-trace-rb/issues/3183
+[#3186]: https://github.com/DataDog/dd-trace-rb/issues/3186
+[#3188]: https://github.com/DataDog/dd-trace-rb/issues/3188
+[#3189]: https://github.com/DataDog/dd-trace-rb/issues/3189
+[#3190]: https://github.com/DataDog/dd-trace-rb/issues/3190
 [@AdrianLC]: https://github.com/AdrianLC
 [@Azure7111]: https://github.com/Azure7111
 [@BabyGroot]: https://github.com/BabyGroot
@@ -3838,4 +4014,4 @@ 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/ext/ddtrace_profiling_native_extension/NativeExtensionDesign.md

@@ -16,8 +16,7 @@ the gem. Setting `DD_PROFILING_NO_EXTENSION` at installation time skips compilat
 
 (If you're a customer and needed to use this, please tell us why on <https://github.com/DataDog/dd-trace-rb/issues/new>.)
 
-In past releases, it was possible for the profiler to run without the native extension, but that's no longer the case,
-and disabling the extension will disable profiling.
+Disabling the profiler extension will disable profiling.
 
 ## Who is this page for?
 
@@ -83,7 +82,7 @@ contribute upstream changes so that they become official public VM APIs.
 Ruby versions 2.6 to 3.2 shipped a JIT compiler called MJIT. This compiler does not directly generate machine code;
 instead it generates C code and uses the system C compiler to turn it into machine code.
 
-The generated C code `#include`s a private header -- which we reference as "the MJIT header" everywhere.
+The generated C code `#include`s a private header -- which we call "the MJIT header".
 The MJIT header gets shipped with all MJIT-enabled Rubies and includes the layout of many internal VM structures;
 and of course the intention is that it is only used by the Ruby MJIT compiler.
 
@@ -104,8 +103,7 @@ Thus, even though a regular Ruby installation does not include these files, we c
 * **OS support**: Linux
 
 To enable CPU-time profiling, we use the `pthread_getcpuclockid(pthread_t thread, clockid_t *clockid)` C function to
-obtain a `clockid_t` that can then be used with the `Process.clock_gettime` method (or directly with the
-`clock_gettime()` C function).
+obtain a `clockid_t` that can then be used with the `clock_gettime` function.
 
 The challenge with using `pthread_getcpuclockid()` is that we need to get the `pthread_t` for a given Ruby `Thread`
 object. We previously did this with a weird combination of monkey patching and `pthread_self()` (effectively patching

data/ext/ddtrace_profiling_native_extension/clock_id.h

@@ -17,9 +17,6 @@ typedef struct thread_cpu_time {
 
 void self_test_clock_id(void);
 
-// TODO: Remove this after the OldStack profiler gets removed
-VALUE clock_id_for(VALUE self, VALUE thread);
-
 // Safety: This function is assumed never to raise exceptions by callers
 thread_cpu_time_id thread_cpu_time_id_for(VALUE thread);
 thread_cpu_time thread_cpu_time_for(thread_cpu_time_id time_id);

data/ext/ddtrace_profiling_native_extension/clock_id_from_pthread.c

@@ -22,28 +22,6 @@ void self_test_clock_id(void) {
   if (expected_pthread_id != actual_pthread_id) rb_raise(rb_eRuntimeError, "pthread_id_for() self-test failed");
 }
 
-// TODO: Remove this after the OldStack profiler gets removed
-VALUE clock_id_for(DDTRACE_UNUSED VALUE _self, VALUE thread) {
-  rb_nativethread_id_t thread_id = pthread_id_for(thread);
-
-  clockid_t clock_id;
-  int error = pthread_getcpuclockid(thread_id, &clock_id);
-
-  if (error == 0) {
-    return CLOCKID2NUM(clock_id);
-  } else {
-    switch(error) {
-      // The more specific error messages are based on the pthread_getcpuclockid(3) man page
-      case ENOENT:
-        rb_exc_raise(rb_syserr_new(error, "Failed to get clock_id for given thread: Per-thread CPU time clocks are not supported by the system."));
-      case ESRCH:
-        rb_exc_raise(rb_syserr_new(error, "Failed to get clock_id for given thread: No thread could be found."));
-      default:
-        rb_exc_raise(rb_syserr_new(error, "Failed to get clock_id for given thread"));
-    }
-  }
-}
-
 // Safety: This function is assumed never to raise exceptions by callers
 thread_cpu_time_id thread_cpu_time_id_for(VALUE thread) {
   rb_nativethread_id_t thread_id = pthread_id_for(thread);

data/ext/ddtrace_profiling_native_extension/clock_id_noop.c

@@ -10,7 +10,6 @@
 #include "helpers.h"
 
 void self_test_clock_id(void) { } // Nothing to check
-VALUE clock_id_for(DDTRACE_UNUSED VALUE _self, DDTRACE_UNUSED VALUE _thread) { return Qnil; } // Nothing to return
 
 thread_cpu_time_id thread_cpu_time_id_for(DDTRACE_UNUSED VALUE _thread) {
   return (thread_cpu_time_id) {.valid = false};

data/ext/ddtrace_profiling_native_extension/collectors_cpu_and_wall_time_worker.c

@@ -83,6 +83,7 @@ struct cpu_and_wall_time_worker_state {
   bool allocation_counting_enabled;
   bool no_signals_workaround_enabled;
   bool dynamic_sampling_rate_enabled;
+  int allocation_sample_every; // Temporarily used for development/testing of allocation profiling
   VALUE self_instance;
   VALUE thread_context_collector_instance;
   VALUE idle_sampling_helper_instance;
@@ -150,7 +151,8 @@ static VALUE _native_initialize(
   VALUE idle_sampling_helper_instance,
   VALUE allocation_counting_enabled,
   VALUE no_signals_workaround_enabled,
-  VALUE dynamic_sampling_rate_enabled
+  VALUE dynamic_sampling_rate_enabled,
+  VALUE allocation_sample_every
 );
 static void cpu_and_wall_time_worker_typed_data_mark(void *state_ptr);
 static VALUE _native_sampling_loop(VALUE self, VALUE instance);
@@ -183,9 +185,10 @@ static void grab_gvl_and_sample(void);
 static void reset_stats(struct cpu_and_wall_time_worker_state *state);
 static void sleep_for(uint64_t time_ns);
 static VALUE _native_allocation_count(DDTRACE_UNUSED VALUE self);
-static void on_newobj_event(DDTRACE_UNUSED VALUE tracepoint_data, DDTRACE_UNUSED void *unused);
+static void on_newobj_event(VALUE tracepoint_data, DDTRACE_UNUSED void *unused);
 static void disable_tracepoints(struct cpu_and_wall_time_worker_state *state);
 static VALUE _native_with_blocked_sigprof(DDTRACE_UNUSED VALUE self);
+static VALUE rescued_sample_allocation(VALUE tracepoint_data);
 
 // Note on sampler global state safety:
 //
@@ -223,7 +226,7 @@ void collectors_cpu_and_wall_time_worker_init(VALUE profiling_module) {
   // https://bugs.ruby-lang.org/issues/18007 for a discussion around this.
   rb_define_alloc_func(collectors_cpu_and_wall_time_worker_class, _native_new);
 
-  rb_define_singleton_method(collectors_cpu_and_wall_time_worker_class, "_native_initialize", _native_initialize, 7);
+  rb_define_singleton_method(collectors_cpu_and_wall_time_worker_class, "_native_initialize", _native_initialize, 8);
   rb_define_singleton_method(collectors_cpu_and_wall_time_worker_class, "_native_sampling_loop", _native_sampling_loop, 1);
   rb_define_singleton_method(collectors_cpu_and_wall_time_worker_class, "_native_stop", _native_stop, 2);
   rb_define_singleton_method(collectors_cpu_and_wall_time_worker_class, "_native_reset_after_fork", _native_reset_after_fork, 1);
@@ -257,10 +260,14 @@ static const rb_data_type_t cpu_and_wall_time_worker_typed_data = {
 static VALUE _native_new(VALUE klass) {
   struct cpu_and_wall_time_worker_state *state = ruby_xcalloc(1, sizeof(struct cpu_and_wall_time_worker_state));
 
+  // Note: Any exceptions raised from this note until the TypedData_Wrap_Struct call will lead to the state memory
+  // being leaked.
+
   state->gc_profiling_enabled = false;
   state->allocation_counting_enabled = false;
   state->no_signals_workaround_enabled = false;
   state->dynamic_sampling_rate_enabled = true;
+  state->allocation_sample_every = 0;
   state->thread_context_collector_instance = Qnil;
   state->idle_sampling_helper_instance = Qnil;
   state->owner_thread = Qnil;
@@ -287,12 +294,14 @@ static VALUE _native_initialize(
   VALUE idle_sampling_helper_instance,
   VALUE allocation_counting_enabled,
   VALUE no_signals_workaround_enabled,
-  VALUE dynamic_sampling_rate_enabled
+  VALUE dynamic_sampling_rate_enabled,
+  VALUE allocation_sample_every
 ) {
   ENFORCE_BOOLEAN(gc_profiling_enabled);
   ENFORCE_BOOLEAN(allocation_counting_enabled);
   ENFORCE_BOOLEAN(no_signals_workaround_enabled);
   ENFORCE_BOOLEAN(dynamic_sampling_rate_enabled);
+  ENFORCE_TYPE(allocation_sample_every, T_FIXNUM);
 
   struct cpu_and_wall_time_worker_state *state;
   TypedData_Get_Struct(self_instance, struct cpu_and_wall_time_worker_state, &cpu_and_wall_time_worker_typed_data, state);
@@ -301,6 +310,12 @@ static VALUE _native_initialize(
   state->allocation_counting_enabled = (allocation_counting_enabled == Qtrue);
   state->no_signals_workaround_enabled = (no_signals_workaround_enabled == Qtrue);
   state->dynamic_sampling_rate_enabled = (dynamic_sampling_rate_enabled == Qtrue);
+  state->allocation_sample_every = NUM2INT(allocation_sample_every);
+
+  if (state->allocation_sample_every < 0) {
+    rb_raise(rb_eArgError, "Unexpected value for allocation_sample_every: %d. This value must be >= 0.", state->allocation_sample_every);
+  }
+
   state->thread_context_collector_instance = enforce_thread_context_collector_instance(thread_context_collector_instance);
   state->idle_sampling_helper_instance = idle_sampling_helper_instance;
   state->gc_tracepoint = rb_tracepoint_new(Qnil, RUBY_INTERNAL_EVENT_GC_ENTER | RUBY_INTERNAL_EVENT_GC_EXIT, on_gc_event, NULL /* unused */);
@@ -880,7 +895,7 @@ static VALUE _native_allocation_count(DDTRACE_UNUSED VALUE self) {
 
 // Implements memory-related profiling events. This function is called by Ruby via the `object_allocation_tracepoint`
 // when the RUBY_INTERNAL_EVENT_NEWOBJ event is triggered.
-static void on_newobj_event(DDTRACE_UNUSED VALUE tracepoint_data, DDTRACE_UNUSED void *unused) {
+static void on_newobj_event(VALUE tracepoint_data, DDTRACE_UNUSED void *unused) {
   // Update thread-local allocation count
   if (RB_UNLIKELY(allocation_count == UINT64_MAX)) {
     allocation_count = 0;
@@ -907,7 +922,12 @@ static void on_newobj_event(DDTRACE_UNUSED VALUE tracepoint_data, DDTRACE_UNUSED
   // defined as not being able to allocate) sets this.
   state->during_sample = true;
 
-  // TODO: Sampling goes here (calling into `thread_context_collector_sample_allocation`)
+  // TODO: This is a placeholder sampling decision strategy. We plan to replace it with a better one soon (e.g. before
+  // beta), and having something here allows us to test the rest of feature, sampling decision aside.
+  if (state->allocation_sample_every > 0 && ((allocation_count % state->allocation_sample_every) == 0)) {
+    // Rescue against any exceptions that happen during sampling
+    safely_call(rescued_sample_allocation, tracepoint_data, state->self_instance);
+  }
 
   state->during_sample = false;
 }
@@ -929,3 +949,18 @@ static VALUE _native_with_blocked_sigprof(DDTRACE_UNUSED VALUE self) {
     return result;
   }
 }
+
+static VALUE rescued_sample_allocation(VALUE tracepoint_data) {
+  struct cpu_and_wall_time_worker_state *state = active_sampler_instance_state; // Read from global variable, see "sampler global state safety" note above
+
+  // This should not happen in a normal situation because on_newobj_event already checked for this, but just in case...
+  if (state == NULL) return Qnil;
+
+  rb_trace_arg_t *data = rb_tracearg_from_tracepoint(tracepoint_data);
+  VALUE new_object = rb_tracearg_object(data);
+
+  thread_context_collector_sample_allocation(state->thread_context_collector_instance, state->allocation_sample_every, new_object);
+
+  // Return a dummy VALUE because we're called from rb_rescue2 which requires it
+  return Qnil;
+}

data/ext/ddtrace_profiling_native_extension/collectors_idle_sampling_helper.c

@@ -78,6 +78,9 @@ static const rb_data_type_t idle_sampling_helper_typed_data = {
 static VALUE _native_new(VALUE klass) {
   struct idle_sampling_loop_state *state = ruby_xcalloc(1, sizeof(struct idle_sampling_loop_state));
 
+  // Note: Any exceptions raised from this note until the TypedData_Wrap_Struct call will lead to the state memory
+  // being leaked.
+
   reset_state(state);
 
   return TypedData_Wrap_Struct(klass, &idle_sampling_helper_typed_data, state);