Re: large file support? (fwd)
Janos Farkas (Janos.Farkas-#BQb9iXvHTAO527H3WKWgezR8Kve@shadow.banki.hu)
Mon, 18 Aug 1997 14:51:00 +0200
On 1997-08-18 at 10:11:54, Martin von Loewis wrote:
> > Who says you need 64bit inode numbers for a large file system. Nothing
> > stops a 64bit fs with multi-terabyte files having 8bit inodes.
>
> The LFS spec recommends this, too. They suggest an ino64_t in struct stat
> and struct dirent.
> 4 milliards of inodes sounds like a lot. However, this style of argument
> collapsed in the past many times. If we change the interface, we should
> get it right once and for good (this also involves dev_t).
It might be obsolete now. Just imagine a bit diversion from the traditional
unix filesystems, and you can easily find filesystems where inode numbers do
not "enumerate" the inodes, but represent an offset to the inode itself.
With a 32 bit inode number, this would leave us with a 4GB filesystem limit
for such funky filesystems, thus making the LFS support for them to about
to "single large file support" :)
--
Janos - Don't worry, my address works. I'm just bored of spam.