Re: Induced crashes

[tb@becket.net (Thomas Bushnell, BSG)]

neal@cs.uml.edu (Neal H. Walfield) writes:

> > Disk hardware guarantees that a sector write can always be completed
> > even if the power goes out partway through.  That means that writing a
> > single sector *is* always atomic.
> 
> The size of a single sector does not necessarily equal the size of of
> a disk block.

If the metadata is properly designed, this doesn't matter.

The ext2fs format actually has a bug in which it permits directory
entries to cross sector boundaries (ufs does not), which can result in
unrepairable inconsistencies if there is a crash in the middle of
writing such a pair of blocks.

But there is no theoretical problem with getting it right.