Ship Phase A GC before parallel mutators

Platform spec ADR

Ship Phase A GC before parallel mutators

Back to Memory and GC runtime contract ◎ Open feature in platform map

Spec standingStandard

Owner

Piotr Mikstacki pmikstacki@cybernomad.it

Submitter

Piotr Mikstacki pmikstacki@cybernomad.it

Current document

Articles

• Contracts and edge cases MUST rules for allocation, barriers, roots, and Phase A mutator exclusivity. article Standard Reviewed Fri May 22
• Design model Heap ownership, tri-color GC, Phase A mutator rules, and compiler/runtime split. article Standard Reviewed Fri May 22
• Examples Allocation failures, GC pacing triggers, and external root pinning scenarios. article Standard Reviewed Fri May 22
• FAQ and troubleshooting GC pauses, mutator violations, barrier omissions, and heap growth debugging. article Standard Reviewed Fri May 22
• Flow and algorithm Allocation, barrier insertion, collection pacing, and safepoint algorithms. article Standard Reviewed Fri May 22
• Verification and traceability Runtime GC tests, Abfall integration, and compiler stack-map obligations. article Standard Reviewed Fri May 22

History

0 revisions (git unavailable at build; counts may be empty)

Open full history on GitHub

No commits recorded for this path.

Completeness

OKfeature · 3 SpecSection(s) · 1 H2 heading(s)

Section id	Required	Found
what-this-feature-specifies	yes	yes
implementation-anchors	yes	yes

Full tree: run pnpm verify:platform-spec-layout (writes src/generated/platform-spec-layout-report.json).

Related spec docs

• Feature
  Memory and GC runtime contract

  Parent feature hub

• Feature
  Fiber scheduler and stacks

  Phase A scheduling rules

Context

Parallel GC mutators need write barriers, stack-map coverage, and scheduler coordination. Shipping partial parallel GC risks data races on the Abfall heap.

Decision

Phase	Mutators	Ship order
A (v0.2)	One Beskid mutator; many cooperative fibers	Default ship
B (v0.3, opt-in)	Parallel mutators + real gc_write_barrier on pointer-payload channels + syscall-pool guard	Wired behind BESKID_RUNTIME_PHASE_B / set_runtime_phase(RuntimePhase::PhaseB); stays opt-in until preemption emission and concurrent-mark stress land
B (later)	Phase B becomes the default	Requires AOT/JIT to insert runtime_preempt_check prologues and full concurrent-mark stress coverage

Phase B must not become the default without updating this feature hub and fiber scheduler contracts.

Consequences

Phase A barriers stay correct because Phase B reuses the same Dijkstra insertion barrier; the difference is that Phase B exercises it on multi-mutator workloads and on pointer-payload channel ops. Fiber/feature docs cross-link Phase tables.

Verification anchors

GC integration tests under crates/beskid_runtime/tests/:

• tests/concurrency.rs — channel/scheduler core
• tests/gc.rs, tests/gc_concurrency.rs — Phase A heap and root handles
• tests/phase_b_concurrency.rs — Phase B multi-mutator stress, pointer-payload channels with write barriers, syscall-pool guard, optional preemption hook

Platform-spec Phase diagrams in hub and design model.