Skip to content

[release/6.0] Always write sample events in EventPipe based on sample frequency. #58761

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Anipik merged 2 commits into from
Sep 8, 2021
Merged

[release/6.0] Always write sample events in EventPipe based on sample frequency. #58761

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7a45fe2
Always write sample events in EventPipe based on sample frequency.
lateralusX Sep 6, 2021
c3380dd
Adjust included profiler samples.
lateralusX Sep 7, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 11 additions & 1 deletion src/mono/mono/eventpipe/ep-rt-mono.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2553,6 +2553,16 @@ ep_rt_mono_sample_profiler_write_sampling_event_for_threads (
// as second, re-classify current callstack to be executing managed code.
data->payload_data = EP_SAMPLE_PROFILER_SAMPLE_TYPE_MANAGED;
}
if (data->stack_walk_data.top_frame && ep_stack_contents_get_length (&data->stack_contents) == 0) {
// If no managed frames (including helper frames) are located on stack, mark sample as beginning in external code.
// This can happen on attached embedding threads returning to native code between runtime invokes.
// Make sure sample is still written into EventPipe for all attached threads even if they are currently not having
// any managed frames on stack. Prevents some tools applying thread time heuristics to prolong duration of last sample
// when embedding thread returns to native code. It also opens ability to visualize number of samples in unmanaged code
// on attached threads when executing outside of runtime. If tooling is not interested in these sample events, they are easy
// to identify and filter out.
data->payload_data = EP_SAMPLE_PROFILER_SAMPLE_TYPE_EXTERNAL;
}

sampled_thread_count++;
}
Expand All @@ -2570,7 +2580,7 @@ ep_rt_mono_sample_profiler_write_sampling_event_for_threads (
THREAD_INFO_TYPE adapter = { { 0 } };
for (uint32_t i = 0; i < sampled_thread_count; ++i) {
EventPipeSampleProfileStackWalkData *data = &g_array_index (_ep_rt_mono_sampled_thread_callstacks, EventPipeSampleProfileStackWalkData, i);
if (data->payload_data != EP_SAMPLE_PROFILER_SAMPLE_TYPE_ERROR && ep_stack_contents_get_length(&data->stack_contents) > 0) {
if ((data->stack_walk_data.top_frame && data->payload_data == EP_SAMPLE_PROFILER_SAMPLE_TYPE_EXTERNAL) || (data->payload_data != EP_SAMPLE_PROFILER_SAMPLE_TYPE_ERROR && ep_stack_contents_get_length (&data->stack_contents) > 0)) {
// Check if we have an async frame, if so we will need to make sure all frames are registered in regular jit info table.
// TODO: An async frame can contain wrapper methods (no way to check during stackwalk), we could skip writing profile event
// for this specific stackwalk or we could cleanup stack_frames before writing profile event.
Expand Down