> now THIS is the kind of info i like from an Open Source based community. ;o)

thanks :)

> by the way, my old Operating Systems book says something about Unix filesystems and 
>inodes,

An inode contains information about where to find a file, like what
blocks on the disk it occupies, and also things like its name, and 
permission bits and access times. The inode (AFAIK) in and of itself
is not the limiter on a file's size -- at least in a 32-bit environment,
our limiter is the size of 'offset', and if the system can't seek
past a certain point because the value can't exceed 4G, then that
is the limiter of a file's size. Of course, a filesystem design would 
specify the format of the inodes, and how many blocks there could be,
which would in turn limit a file's size, since you can't have more than
one inode describing a file -- at least you can't have a file that 
partly belongs to one inode and finishes with another.

In ext2 at least, an inode has space for 14 (1k) blocks, and any
file bigger than that has allocated to it a secondary block of
pointers to blocks (indirect blocks). That basically lets you 
specify a megabyte's worth (give or take a few K) of blocks, assuming 
a 4K indirect block (i.e., 1024 32-bit unsigned block numbers), which
I think is right. Anything bigger than that requires a double indirect
scheme, which provides another block which specifies a bunch of 
pointers to (indirect) blocks. I imagine this all scales up to the
maximum file size, as I would think that it would be a poor design if
it shortchanged in this area. 


> Damian



Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Reply via email to