Date: Thu, 1 Apr 1999 19:34:45 +0000 (GMT) From: Terry Lambert <tlambert@primenet.com> To: ksmm@threespace.com (The Classiest Man Alive) Cc: doconnor@gsoft.com.au, freebsd-chat@FreeBSD.ORG Subject: Re: FreeBSD is running out of time Message-ID: <199904011934.MAA08863@usr08.primenet.com> In-Reply-To: <199904011659.LAA24524@geek.grf.ov.com> from "The Classiest Man Alive" at Apr 1, 99 11:40:13 am
next in thread | previous in thread | raw e-mail | index | archive | help
> I just assumed that once we made a change to something as fundamental as > the architecture's word size that everything else built upon that > (including the filesystem structures) would follow suit. In other words, I > wasn't expecting that backward compatibility would be paramount after going > 64-bit. Nope. Try talking to a 21041 chip on the other side of a PCI bridge in anything except device register size units. Device drivers don't change because of memory bus size changes. Also note that on-disk structure sizes are chosen for their ability to be represented in integral multiples or fractions of a physical disk block. It is drastically easier to deal in terms of an atomic write to a single disk block (check out the FFS directory entry code to see what it would take to allow it to span two 512b disk blocks -- I've made the change for Unicode support... it's not pretty). Inodes are currently 128 bytes; the next integral fraction of a disk block is 256 bytes. Note that this would effectively double the amount of disk space that must be dedicated to inode storage, even though it meets the <= one disk block pseudo-requirement. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199904011934.MAA08863>