test(bpf): Add eBPF Test Suite

[dave-tucker]

This commit adds a more comprehensive eBPF test suite.
Currently it tests the operation of a number of key functions
within the eBPF code - for example the main sched_switch
tracepoint that we run. In addition, it runs a number
of micro benchmarks so we can track performance of these
key pieces of code.

[github-actions]

🤖 SeineSailor

Here is a concise summary of the pull request changes:

Summary: This pull request, titled "WIP: BPF Testing," introduces a comprehensive eBPF test suite, refactors BPF code for testing, and migrates to cilium/ebpf. The changes are grouped into three main areas:

1. Migration to cilium/ebpf and package creation: A new pkg/bpftest package is introduced for testing BPF-related functions, facilitated by the migration to cilium/ebpf. This change extends PR chore(pkg/bpf): Replace libbpfgo with cilium/ebpf #1438.

2. eBPF test suite and benchmarking: The pull request adds a comprehensive eBPF test suite, including tests for key functions, micro benchmarks, and new types (e.g., ExperimentCache, enumSupport, Experiment, SamplingConfig, Measurement, Ranking, Stat, StatsType, Stats, PrecisionBundle, and Stopwatch) to support benchmarking and measuring code.

3. eBPF code refactoring and additions: The pull request refactors BPF code for testing, adding several map structures, new functions (e.g., get_on_cpu_elapsed_time_us(), get_on_cpu_cycles(), and get_on_cpu_instr()), and modifying existing ones. New global variables (e.g., TEST, HW, and SAMPLE_RATE) are introduced to configure test modes and sampling rates.

Impact: These changes significantly enhance the testing capabilities of the eBPF codebase, providing a comprehensive test suite and benchmarking support. The migration to cilium/ebpf and the introduction of the pkg/bpftest package improve the code's maintainability and testability.

Observations and suggestions:

• The pull request is marked as "WIP," indicating that it is still a work-in-progress. It would be beneficial to break down the changes into smaller, more focused pull requests to facilitate easier review and testing.
• The addition of new types and functions may require updates to the documentation to ensure that users are aware of the changes and can effectively utilize the new features.
• It would be helpful to include more detailed comments and explanations within the code to facilitate understanding and maintenance of the new test suite and benchmarking features.

[dave-tucker]

This will fail CI (for now at least) due to needing access to hardware perf counters.
I'm going to look in to creating some software (or dummy) counters to solve this.
Suite output is as follows:

=== RUN   TestBpf
Running Suite: Bpf Suite - /home/dave/dev/kepler
================================================
Random Seed: 1720176038

Will run 6 of 6 specs
••
------------------------------
• [1.593 seconds]
BPF Exporter test should increment the page hit counter efficiently
/home/dave/dev/kepler/pkg/bpftest/bpf_suite_test.go:107

  Report Entries >>
  Increment the page hit counter - /home/dave/dev/kepler/pkg/bpftest/bpf_suite_test.go:109 @ 07/05/24 11:40:38.399
    Increment the page hit counter
    Name                                  | N       | Min     | Median  | Mean    | StdDev  | Max      
    ===================================================================================================
    page hit counter increment [duration] | 1000000 | 1.172µs | 1.223µs | 1.286µs | 1.125µs | 759.055µs
  << Report Entries
------------------------------
•
------------------------------
• [7.402 seconds]
BPF Exporter test efficiently collects hardware counter metrics for sched_switch events [perf_event]
/home/dave/dev/kepler/pkg/bpftest/bpf_suite_test.go:217

  Report Entries >>
  sched_switch tracepoint - /home/dave/dev/kepler/pkg/bpftest/bpf_suite_test.go:219 @ 07/05/24 11:40:40.035
    sched_switch tracepoint
    Name                                       | N       | Min     | Median  | Mean   | StdDev | Max      
    ======================================================================================================
    sampled sched_switch tracepoint [duration] | 1000000 | 2.124µs | 2.786µs | 2.77µs | 664ns  | 251.505µs
  << Report Entries
------------------------------
• [7.084 seconds]
BPF Exporter test uses sample rate to reduce CPU time [perf_event]
/home/dave/dev/kepler/pkg/bpftest/bpf_suite_test.go:256

  Report Entries >>
  sampled sched_switch tracepoint - /home/dave/dev/kepler/pkg/bpftest/bpf_suite_test.go:258 @ 07/05/24 11:40:47.483
    sampled sched_switch tracepoint
    Name                                       | N       | Min     | Median  | Mean    | StdDev | Max      
    =======================================================================================================
    sampled sched_switch tracepoint [duration] | 1000000 | 1.863µs | 2.535µs | 2.474µs | 664ns  | 244.011µs
  << Report Entries
------------------------------

Ran 6 of 6 Specs in 16.228 seconds
SUCCESS! -- 6 Passed | 0 Failed | 0 Pending | 0 Skipped
--- PASS: TestBpf (16.23s)
PASS

[sthaha]

looks good over!
could you please resolve the unit-test failure ?

[sthaha]

LGTM!
Please do consider if the file descriptors must be closed before merging.