Do not use random numbers as IDs for the frames

[javierhonduco]

Do not use random numbers as IDs for the frames

[manuelfelipe]

Nice, thanks @javierhonduco. Just gave it another try after #74. I'm no longer seeing the previous occasional mismatch frame count errors

RUST_LOG=debug RUST_BACKTRACE=1 rbperf record -d 10 --pid 100743 syscall enter_getpid
[2023-02-28T13:05:58Z DEBUG rbperf::rbperf] verbose_bpf_logging set to false
[2023-02-28T13:05:58Z DEBUG rbperf::rbperf] use_ringbuf set to false
[2023-02-28T13:05:58Z DEBUG rbperf::rbperf] open prog: walk_ruby_stack has 10873 intructions
[2023-02-28T13:05:58Z DEBUG rbperf::rbperf] open prog: on_event has 217 intructions
[2023-02-28T13:05:58Z DEBUG rbperf::binary] Checking symbol ruby_version in binary /proc/100743/exe
[2023-02-28T13:05:58Z DEBUG rbperf::process] Binary "/proc/100743/exe"
[2023-02-28T13:05:58Z DEBUG rbperf::binary] Checking symbol ruby_current_vm in binary /proc/100743/exe
pid: 100743
statically linked
ruby main thread address: 0x5572c8a56bd0
process base address: 0x5572c8425000
ruby version: "3.2.1"

[2023-02-28T13:05:58Z INFO  rbperf::rbperf] Adding config for version starting with RubyVersion { major_version: 3, minor_version: 2, patch_version: 1 } at index 10
[2023-02-28T13:05:58Z DEBUG rbperf::rbperf] profiling started
[2023-02-28T13:05:58Z DEBUG rbperf::events] syscall with id 206 found in /sys/kernel/debug/tracing/events/syscalls/sys_enter_getpid/id
[2023-02-28T13:05:58Z DEBUG rbperf::rbperf] program type Tracepoint
[2023-02-28T13:05:58Z DEBUG rbperf::rbperf] program type Tracepoint
Got 6453 samples and 0 errors
Flamegraph written to: rbperf_flame_02282023_13h06m09s.svg

[javierhonduco]

Great to hear! Thanks so much for trying 😄 .

Just for context, frame count mismatches can still happen if the application that's being profiled or traced produces events very rapidly or if the duration is very long. I have some ideas on how to solve this problem more generally, will update you on this once the PR is ready.