Added-checksum filesystem

[Sander Vesik]

from more or less the department of semi-idle thoughts:

	lets assume that there is a disk / filesystem where 
	its important to know that the data coming back from
	the disk - via disk, interconnect & pci is the same as
	was originaly sent with common errors outruled, and that
	the result would not overly suck performance-wise. so
	the checksum should be fast tocompute, relatively short
	(4-8 bytes per fragment?), and be placed close to data
	- probably in the same cylinder group and distributed to
	be next to the data it covers.

	

-- 
	Sander

+++ Out of cheese error +++