Struct PageCache 

pub struct PageCache { /* private fields */ }

Sharded page cache.

Routes each page_id to one of at most NUM_SHARDS independent PageCacheShards by the low bits of the id. Readers and writers for disjoint pages hit different shards and do not contend on the shard-internal RwLocks/Mutexes. Each shard runs its own SIEVE eviction loop over its share of the requested slots; hot/cold asymmetry across shards is accepted in exchange for the contention win. Tiny budgets use fewer shards so one requested page remains one physical page slot.

Implementations

impl PageCache

pub fn new(capacity: usize) -> PageCache

pub fn with_default_capacity() -> PageCache

pub fn len(&self) -> usize

pub fn is_empty(&self) -> bool

pub fn capacity(&self) -> usize

pub fn stats(&self) -> CacheStats

pub fn reset_stats(&self)

pub fn get(&self, page_id: u32) -> Option<Page>

pub fn insert(&self, page_id: u32, page: Page) -> Option<Page>

pub fn insert_dirty(&self, page_id: u32, page: Page) -> Option<Page>

pub fn insert_if_absent(&self, page_id: u32, page: Page) -> Option<Page>

pub fn mark_dirty(&self, page_id: u32)

pub fn mark_clean(&self, page_id: u32)

pub fn pin(&self, page_id: u32) -> bool

pub fn unpin(&self, page_id: u32) -> bool

pub fn remove(&self, page_id: u32) -> Option<Page>

pub fn contains(&self, page_id: u32) -> bool

pub fn flush_dirty(&self) -> Vec<(u32, Page)>

pub fn flush_dirty_pages(&self, page_ids: &[u32]) -> Vec<(u32, Page)>

Snapshot only the requested dirty images, bounding payload copies by the caller’s batch size rather than the full cache capacity.

pub fn flush_some_dirty(&self, max: usize) -> Vec<(u32, Page)>

pub fn dirty_count(&self) -> usize

pub fn clear(&self)

pub fn page_ids(&self) -> Vec<u32>

pub fn dirty_page_ids(&self, max: usize) -> Vec<u32>

Clean entries must not split a small dirty set into many I/O batches.

Trait Implementations

impl Default for PageCache

fn default() -> PageCache

Returns the “default value” for a type.