Struct HostcallReactorMesh 

pub struct HostcallReactorMesh { /* private fields */ }

Deterministic SPSC reactor mesh for hostcall traffic.

Routes fast-lane hostcall requests to per-shard SPSC lanes using stable hash routing (by call_id for affinity) or opcode-class routing.

Routing policy:

• Session opcodes: hash-routed by call_id (shard affinity preserves per-call ordering for streaming scenarios).
• Events opcodes: round-robin across shards for load distribution.
• Tool opcodes: hash-routed by call_id.

Drain policy:

• drain_shard(shard_id, budget) for per-shard processing.
• drain_global_order(budget) for deterministic cross-shard ordering.

Implementations

impl HostcallReactorMesh

pub fn new(config: HostcallReactorConfig) -> Self

Create a new reactor mesh with the given configuration.

pub fn shard_count(&self) -> usize

Number of shard lanes.

pub fn total_depth(&self) -> usize

Total pending requests across all shards.

pub fn has_pending(&self) -> bool

Whether any lane has pending requests.

pub fn telemetry(&self) -> HostcallReactorTelemetry

Snapshot queueing telemetry.

pub fn core_id_for_shard(&self, shard_id: usize) -> Option<usize>

Core affinity configuration (if any).

pub const fn numa_pool(&self) -> Option<&NumaSlabPool>

NUMA-aware slab pool (if core-ids were configured).

pub fn affinity_advice(&self) -> &[ThreadAffinityAdvice]

Thread affinity advice derived from reactor core mapping.

pub fn drain_shard( &mut self, shard_id: usize, budget: usize, ) -> Vec<HostcallReactorRequest>

Drain up to budget requests from a specific shard.

pub fn drain_global_order( &mut self, budget: usize, ) -> Vec<HostcallReactorRequest>

Drain across all shards in deterministic global sequence order.

pub const fn record_completions(&mut self, count: u64)

Record that a batch of completions was produced by shard processing.

Trait Implementations

impl Debug for HostcallReactorMesh

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

Formats the value using the given formatter.