Basic File System question...

[Arkimedes]

OK, so I have been studying the I-node based file systems recently (largely the Minix file system, and the Ext2 file system); I understand that the partition is divided into block groups, and all that jazz, but what I don't understand is this: is there a limit, like some parameter, that constrains the maximum number of block groups?

I mean, suppose we had an ideal computer with 1 Yottabyte of space (or 2 to the 80th bytes). If we had the block size be 2^12 bytes (4 KB), there are 2^68 blocks. Here's the problem: an unsigned long long is capable of a maximum value of 2^64. Wouldn't there be 2^68 - 2^64 blocks that won't be accessible? Granted that this is so large that it wouldn't matter, but it wouldn't be that efficient (well, 2^64 over 2^68 = 0.000244141 = 0.024%, which is pretty lousy).

Am I approaching this the wrong way, or does the number of block groups not matter?

[SirDice]

IIRC most filesystems have an upper limit. They're size restricted. When bigger drives get available new filesystems must be created to allow for those sizes.

FreeBSD's UFS has a size limit of 2TB (2^32 * 512 bytes), UFS2 allows for 8ZB (ZetaByte; 2^64 * 512 bytes).

http://www.freebsd.org/projects/bigdisk/index.html

[Arkimedes]

Interesting, thanks! That really answered all of my questions, and the ones that I've been thinking about since this morning.

I greatly appreciate it!