Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote:
> - I don't understand how a folio gets to be partially cached. Cached
> should be tracked on a per-folio basis (like dirty or uptodate), not
> on a per-page basis. The point of the folio work is that managing
> memory in page-sized chunks is now too small for good performance.
Consider the following scenario:
(1) You read two 256K chunks from an nfs file. These are downloaded from the
server and cached.
(2) The nfs inode is dropped from the cache.
(3) You do a 1M read that includes both 256K chunks.
The VM might create a 1M folio to handle (3) that covers the entire read, but
only half of the data is in the cache; the other half has to be fetched from
the server.
David
