[audit] executor: FailFast cancel leaves unstarted goroutine slot as zero-value success

[lugassawan]

Audit findings — executor

Scope: reliability · Severity range: Low–Low
Source: codebase audit

reliability

Severity	File:Line	Symptom	Suggested fix
Low	internal/executor/executor.go:76	On FailFast cancel(), a goroutine scheduled-but-not-started leaves results[idx] at zero-value Result{} (ExitCode 0, no error, Cancelled: false) — indistinguishable from clean success.	Pre-initialize each slot to Result{Target: target, Cancelled: true} before the goroutine is dispatched.

Root cause: Cancellation is sampled at two discrete checkpoints; there is no atomic "never executed" marker for goroutines that lose the race between cancel() and their first checkpoint.

Reasoning chain: 1. FailFast calls cancel() (lines 92-93). 2. Goroutines past lines 76-81 run to completion; one not yet scheduled never writes its slot. 3. Zero-value Result{} presents as ExitCode 0 / nil error / Cancelled: false. 4. allOK callers count the slot as a successful run.

Counter-evidence considered: wg.Wait() guarantees every goroutine writes something, but only the pre-semaphore branch sets Cancelled: true; the narrow post-start/pre-write window still leaves a zero-value slot. Low probability → Low severity, but genuine.

---

Provenance

Filed by workflow-audit-emit-issues from an in-session codebase audit.
Re-running re-files — check for duplicates before confirming.