Subject: Re: Are there any limits on hard-disc size?
To: <>
From: Ignatios Souvatzis <ignatios@theory.cs.uni-bonn.de>
List: port-arm32
Date: 10/12/1998 12:09:16
On Sun, Oct 11, 1998 at 12:10:12PM +0100, Neil A. Carson wrote:
> Mark Brinicombe wrote:
> 
> > the arm32 port does not impose any specific limits. With FFS disk
> > addresses are 64 bit so you can have some big file systems.
> 
> They weren't last time I looked :-)

Uhm, sort of, both.

The 4.4BSD-ish operating systems have a an offset_t (e.g., used by lseek() and
for the file size in the *stat() output) thats 64 bit, measured in bytes.

(However, note that the stdio fseek() parameters are black boxes; I think the
current implementation uses 32 bit file offsets, however.)

However, both values are just one of about three limits.

The 2nd one is the size of addresses the filesystem code can handle. At
the user interface, it is offset_t (64bit measured in bytes), however,
at the disk interfaces 32bit daddr_t's (measured in disk blocks) are
used. Thats 512*4 Gigabytes, or 2097152 Megabytes, or 2199023255552
bytes. Still a big disk even for our times, but we are approaching.

This happens to coincide with what the SCSI protocol can address,
which would otherwise be the 3rd limit. I don't know off-hand what the
IDE addressing limit is.

The fourth limit is what partition tables etc. can address, which causes
lots of nightmares on some PCs with big disks. We're not independent of
this as the boot code on disk needs to be found by the BIOS. I don't
know what the Shark firmware is using.

Regards,
	-is