Jim Sweet wrote: > Could someone explain why caching is even an issue? > > I'm familiar with the Solaris kernel but not the Linux kernel, but > by analogy I would imagine that the VM's page cache is completely > transparent to any layer above it. Thus a read following a given combination > of writes will always return the same data regardless of whether the device > is backed by a page-layer or not. The same must > surely be true of a write-caching disk drive. It is true that at a > given point in time, there will be no one single location (e.g. > disk platters, disk cache or VM page layer) containing a > contiguous representation of the data. However, I would have thought > that this would be an issue only if there was a sudden power failure. Journaling file systems are supposed to do the right thing even in cases where mains power is temporarily lost. Journaling file systems need to make sure that writes reach disk platters in known order. Example: if some "transaction completed" journal write reached disk platters *before* the actual transaction writes hit disk platters, and mains power was lost, that transaction would not be re-played because file system code found "transaction completed" on the journal. -- Jari Ruusu 1024R/3A220F51 5B 4B F9 BB D3 3F 52 E9 DB 1D EB E3 24 0E A9 DD - Linux-crypto: cryptography in and on the Linux system Archive: http://mail.nl.linux.org/linux-crypto/
