Struct PerCpuPageCache 

pub struct PerCpuPageCache { /* private fields */ }

Per-CPU page frame cache to reduce global FRAME_ALLOCATOR contention.

Single-frame allocations (page faults, mmap, fork) dominate. By caching frames per-CPU, we avoid acquiring the global lock on every allocation.

When the cache is empty, it batch-refills from the global allocator. When full, it batch-drains back to the global allocator. Cache-line aligned to prevent false sharing when per-CPU caches are stored in adjacent array slots accessed by different cores.

Implementations

impl PerCpuPageCache

pub const fn new() -> Self

pub fn alloc_one(&mut self) -> Option<FrameNumber>

Try to allocate a single frame from the per-CPU cache. Returns None if cache is empty (caller should refill from global).

pub fn free_one(&mut self, frame: FrameNumber) -> bool

Return a single frame to the per-CPU cache. Returns false if cache is full (caller should drain to global).

pub fn needs_refill(&self) -> bool

Is the cache below the low watermark?

pub fn needs_drain(&self) -> bool

Is the cache above the high watermark?

pub fn batch_refill(&mut self)

Batch-refill from the global frame allocator. Acquires the global lock once, filling up to BATCH_SIZE frames.

pub fn batch_drain(&mut self)

Batch-drain excess frames back to the global allocator. Acquires the global lock once, returning BATCH_SIZE frames.

pub fn cached_count(&self) -> usize

Number of cached frames

Trait Implementations

impl Default for PerCpuPageCache

fn default() -> Self

Returns the “default value” for a type.