feat(prof) I/O profiling preview

[realFlowControl]

Description

This adds I/O profiling preview to the PHP profiler on Linux machines (it is only compiled for Linux, but technically this is a ELF64 feature and could work on other OS that are using ELF64 too):

• adds DD_PROFILING_EXPERIMENTAL_IO_ENABLED / datadog.profiling.experimental_io_enabled configuration (default false)
• adds new sample types:
  • socket_read_time
  • socket_write_time
  • file_read_time
  • file_write_time
  • socket_read_size
  • socket_write_size
  • file_read_size
  • file_write_size

Known limitations:

• tests, including correctness are missing
• benchmarks are missing
• GOT hooking code and I/O profiling code is mixed
• mmap() is not observed
• SQLite is not observed (probably due to missing mmap() support)

Those things will be added/handled in future PRs, also because this feature is still experimental. The sampling distances are best guess and will likely be adjusted later.

https://datadoghq.atlassian.net/browse/PROF-11288

Reviewer checklist

• Test coverage seems ok.
• Appropriate labels assigned.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 74.85%. Comparing base (4da93b3) to head (c406db0).

Additional details and impacted files

@@             Coverage Diff              @@
##             master    #3083      +/-   ##
============================================
- Coverage     74.86%   74.85%   -0.02%     
  Complexity     2792     2792              
============================================
  Files           112      112              
  Lines         11046    11046              
============================================
- Hits           8270     8268       -2     
- Misses         2776     2778       +2     

Flag	Coverage Δ
tracer-php	74.85% <ø> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

see 1 file with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4da93b3...c406db0. Read the comment docs.

[pr-commenter]

Benchmarks [ profiler ]

Benchmark execution time: 2025-02-20 17:04:02

Comparing candidate commit c406db0 in PR branch florian/poc-got-io-profiling with baseline commit 4da93b3 in branch master.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 29 metrics, 7 unstable metrics.

[bwoebi]

At this point, with that many sample types, I think this should look like:

let cpu_time = system_settings.profiling_experimental_cpu_time_enabled;
let alloc = system_settings.profiling_allocation_enabled;
#[cfg(feature = "timeline")]
let timeline = system_settings.profiling_timeline_enabled;
#[cfg(not(feature = "timeline"))]
let timeline = false;
#[cfg(feature = "exception_profiling")]
let exception = system_settings.profiling_allocation_enabled;
#[cfg(not(feature = "exception_profiling"))]
let exception = false;
#[cfg(feature = "io_profiling")]
let io = system_settings.profiling_io_enabled;
#[cfg(not(feature = "io_profiling"))]
let io = false;
let sample_types = vec![
    (ValueType::new("sample", "count"), true, mem::offset_of!(SampleValues, interrupt_count)),
    (ValueType::new("wall-time", "nanoseconds"), true, mem::offset_of!(SampleValues, wall_time)),
    (ValueType::new("cpu-time", "nanoseconds"), cpu_time, mem::offset_of!(SampleValues, cpu_time)),
    (ValueType::new("alloc-samples", "count"), allocation, mem::offset_of!(SampleValues, alloc_samples)),
    (ValueType::new("alloc-size", "bytes"), allocation, mem::offset_of!(SampleValues, alloc_size)),
    (ValueType::new("timeline", "nanoseconds"), timeline, mem::offset_of!(SampleValues, timeline)),
    (ValueType::new("exception-samples", "count"), exception, mem::offset_of!(SampleValues, exception)),
    (ValueType::new("socket-read-time", "nanoseconds"), io, mem::offset_of!(SampleValues, socket_read_time)),
    (ValueType::new("socket-write-time", "nanoseconds"), io, mem::offset_of!(SampleValues, socket_write_time)),
    (ValueType::new("file-io-read-time", "nanoseconds"), io, mem::offset_of!(SampleValues, file_read_time)),
    (ValueType::new("file-io-write-time", "nanoseconds"), io, mem::offset_of!(SampleValues, file_write_time)),
    (ValueType::new("socket-read-size", "bytes"), io, mem::offset_of!(SampleValues, socket_read_size)),
    (ValueType::new("socket-write-size", "bytes"), io, mem::offset_of!(SampleValues, socket_write_size)),
    (ValueType::new("file-io-read-size", "bytes"), io, mem::offset_of!(SampleValues, file_read_size)),
    (ValueType::new("file-io-write-size", "bytes"), io, mem::offset_of!(SampleValues, file_write_size)),
];
// and create sample_types(_mask) with a small loop from the tuples as well as offsets for filter() function

[bwoebi]

(Does not need to be done now, just a suggestion for next time at least)

[realFlowControl]

Doing this in a follow-up PR

[bwoebi]

Looks great to me now, thanks!