KernelTrap - David Howells

Local Caching For Network Filesystems

Jeremy — Thu, 21 Feb 2008 15:35:41 +0000

"These patches add local caching for network filesystems such as NFS," began David Howells describing an updated set of thirty-seven patches to introduce FS-Cache. When asked how the patches affect performance, he noted that this was dependent on the use case, highlighting issues when dealing with lots of metadata, "getting metadata from the local disk fs is slower than pulling it across an unshared gigabit ethernet from a server that already has it in memory."

David continued "these points don't mean that fscache is no use, just that you have to consider carefully whether it's of use to *you* given your particular situation, and that depends on various factors," adding, "note that currently FS-Caching is disabled for individual NFS files opened for writing as there's no way to handle the coherency problems thereby introduced." He concluded with a number of simple performance benchmarks.

Removing iget() and read_inode()

Jeremy — Mon, 15 Oct 2007 00:19:53 +0000

"Here's a set of patches that remove all calls to iget() and all read_inode() functions," began David Howells describing a collection of 32 patches posted to the lkml. He went on to explain the reason for removing these functions, "they should be removed for two reasons: firstly they don't lend themselves to good error handling, and secondly their presence is a temptation for code outside a filesystem to call iget() to access inodes within that filesystem." He then suggested three benefits:

"(1) Error handling gets simpler as you can return an error code rather than having to call is_bad_inode(). (2) You can now tell the difference between ENOMEM and EIO occurring in the read_inode() path. (3) The code should get smaller. iget() is an inline function that is typically called 2-3 times per filesystem that uses it. By folding the iget code into the read_inode code for each filesystem, it eliminates some duplication."