Struct TierSetupGate 

pub struct TierSetupGate { /* private fields */ }

Issue 1733 — the ONE gate a tier runner holds across each tier’s open-and-setup so no two tiers declare entities on the shared session at once.

Every tier executor borrows the boot executor’s ONE RMW session, and a tier’s setup declares its nodes and entities on it. The backend’s entity-creation path is not reentrant across threads: zenoh’s ZenohSession::create_* takes &mut self and mutates the per-node liveliness table, and issue 0447 measured the topic-binding scramble that concurrent declares cause. The RTOS runners serialize by CHAINING the spawns (issue #144: tier N+1 starts only after tier N’s setup returned); a runner that spawns every tier at once holds this instead. Setup runs once per boot and off the hot path, so a sleeping spin costs nothing that matters, and the spin loops that follow stay fully concurrent.

no_std and allocation-free (one atomic flag), so the same gate serves the native C++ runner (nros-cpp), the Linux Rust runner and the NuttX Rust runner. The backoff sleeps through the platform ABI, never spins hot: a waiter that outranks the holder under SCHED_FIFO would otherwise starve it on one CPU.

Implementations

impl TierSetupGate

pub const fn new() -> Self

An open gate.

pub fn enter(&self) -> TierSetupGuard<'_>

Block until this caller holds the gate; it is released when the returned guard drops (on every return path, including a failed setup).

Trait Implementations

impl Debug for TierSetupGate

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for TierSetupGate

fn default() -> TierSetupGate

Returns the “default value” for a type.