[not-matthias]

Fixes COD-4155.

1. Stop excluding the node executable

The runner ran node -e 'console.log(process.execPath)' and added the result to callgrind's ignored objects. Node runtime cost was then missing from simulation profiles. Remove get_node_objects so the node executable is instrumented.

2. Pass --separate-callers=all --pop-on-jump to callgrind

With node instrumented, V8 builtins (Call*, CallFunction*, bytecode handlers) sit between most pairs of JS functions. As single nodes they join every JS caller to every JS callee and form false cycles in the flame graph. --separate-callers=all names every function by its whole call stack, so unrelated chains stay apart; --pop-on-jump makes tail jumps replace the jumping frame, which keeps those stacks shallow.

Dependencies:

• Needs a valgrind-codspeed release with --separate-callers=all and the --pop-on-jump fixes: fix(callgrind): correct V8 call graphs (Sparkplug returns, whole-stack contexts, pop-on-jump) valgrind-codspeed#63. The current pin rejects the flag, so the pin must be bumped before merging.
• The backend must strip 'caller context suffixes from function names.
• Output grows: node -e 0 writes 309 MB with these flags (4 MB before).

Companion change in codspeed-node removes the matching callgrindAddObjSkip(process.execPath) call: CodSpeedHQ/codspeed-node#121

[codspeed]

Merging this PR will not alter performance

1 warning

⚠️ 7 benchmarks are dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

✅ 50 untouched benchmarks
⏩ 9 skipped benchmarks¹

---

_{Comparing cod-4155-remove-obj-skip-for-codspeed-node (ae6049f) with main (2e709d9)}

Footnotes

1. 9 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩