pagecache is a lock-free pagecache and log for building high-performance databases.
Top-level configuration for the system.
A sequential store which allows users to create reservations placed at known log offsets, used for writing persistent data structures that need to know where to find persisted bits in the future.
A metric collector for all pagecache users running in this process.
A materializer for things that have nothing to
materialize or recover, like a standalone
A lock-free pagecache which supports fragmented pages for dramatically improving write throughput.
A simple lock-free radix tree.
A pending log reservation which can be aborted or completed.
NB the holder should quickly call
A simple lock-free stack, with the ability to atomically append or entirely swap-out entries.
Points to either a memory location or a disk location to page-in data from.
An Error type encapsulating various issues that may come up in both the expected and unexpected operation of a PageCache.
The result of a
The log may be configured to write data in several different ways, depending on the constraints of the system using it.
A user of a
This function is useful for inducing random jitter into our atomic operations, shaking out more possible interleavings quickly. It gets fully elliminated by the compiler in non-test code.
The top-level result type for dealing with the PageCache.
A log file offset.
A logical sequence number.
Allows arbitrary logic to be injected into mere operations of the
A page identifier.
A pointer to shared lock-free state bound by a pinned epoch's lifetime.
An offset for a storage file segment.