I'm using XFS a few days now, first the cvs-tree, now the release 1.0 with
kernel 2.4.3 and it's a really great job you did!
Maybe I may ask here some questions about filesystems (especially newer ones
like XFS and some generally questions).
In ext2 the data block is found by walking through all correspondig inodes
which are ordered in a linked list, right?
In xfs (like ReiserFS) the data blocks are referenced by inodes which are
ordered within a btree, right? Therefore the work with many files within a
directory isn't slowed down by walking through a list. But, why is ReiserFS
a few times faster than xfs when working in a directory which contains about
100.000 small files (about 1k in size)? Where's my gap in understanding this
Another questions (more in general): BSD-derivates are using a UFS or FFS
file system. How's the data referenced there? Like ext2 with some linked
lists or is there also a tree for finding the data? What's the feature named
"soft updates" there? Some people are saying it eliminates the need for
journalling. But what's behind this? Sync writes?
Thank you for your answers!