the current oldest item in the container in seconds.

Returns the alignment size for device io operations

Usable memory for this allocation. The caller is free to do whatever he wants with it and needs to ensure concurrency for access into this piece of me

returns the reaper stats

returns the slab in which the memory resides, irrespective of the validity of the memory. The caller can use isValidSlab to check if the returned slab

Return the byte size of all pieces

Whether NvmCache is currently enabled

Helper to create a FindByKeyFn for tests.

number of unique filters

return the hashtable size in bytes

Return the number of elements in the array

Adjust the data start offset forwards to include less valid data This moves the data pointer forwards so that the first amount bytes are no longer con

adding extra config after generating the config: tailSize

alloc size corresponding to the class id

Get an iterator to iterate over all items in object-cache. The returned iterator provides getMemory() for safe aligned access to the ObjectCacheItem

helper function.

@param dst Destination buffer into which to copy data from this view

furcHash -- a consistent hash function using a binary decision tree. Based on an algorithm by Mark Rabkin with two changes: 1) Uses murmurHash64A to h

@return pointer to slab marked for release

FOLLY_EXPORT ensures the dynamic linker dedups the function so there's not a copy of thread local variables per inline instance of this function

marks the shared memory resource to be detached once all the mappings are unmapped.

return the number of pools the cache was configured with.

Moves ownership into a regular item handle.

Return an unordered map.

Current bit mapping from the hash (64bits) to the bucket id and others bit 0 ~ 31 : Reserved for bucket offset within chunk bit 32 ~ : Used for chunk

Clones Item handle. returns an empty handle if it is null. @return HandleImpl return a handle to this item @throw std::overflow_error is the maximum

Returns true if key found and can be actually evicted in the real world

Convenient function to divide integers as floats. Much better than static_cast<double>(a) / (a + b), just fdiv(a, a + b).

Helper function to generate allocation sizes as powers of 2. This provides a simple, uniform distribution of allocation sizes where each size is doubl

overall byte footprint of the array of filters.