Re: what is the status of >2gig filesizes

Theodore Y. Ts'o (tytso@mit.edu)
Tue, 11 Aug 1998 16:39:13 -0400


From: "Michael H. Warfield" <mhw@alcove.wittsend.com>
Date: Mon, 10 Aug 1998 23:37:59 -0400 (EDT)

I think the problem is NOT in the filesystems per se. I have
several 9 Gig systems that are working just fine. If I remember correctly
the discussion a few months back involving the maintainer of some of
the file system utilities, much of the problem involved some critical
maintenance utilites such as e2fsck which don't properly handle filesystems
that large under certain circumstances. It may well crop up when the
files on the file systems start growing larger than 2 Gig that e2fsck
and such start having problems dealing with them. All of this is conjecture
and, as I said, I have numerous file systems in the 9 Gig range, but I
watch those file systems nervously any time they have to be cleaned!

Yet another kernel FAQ....

There is no problem supporting filesystems greater than 2 gigabytes.
That works just fine. The hard part is supporting *files* greater than
2 gigabytes. There, the problem is that there are limitations in the
kernel which make it impossible to fully support > 2gig files on 32 bit
platforms. The work on ext2 filesystems to support > 2gig files was
done in the 2.1 development tree for 64-bit platforms. In order to
support that, there were changes that had to be made in e2fsck, which
are in the latest released version of e2fsprogs.

Fully supporting >2gig files on 32-bit platforms is hard, because the
VFS and memory management layer simply don't have support for it. It
would be possible to bypass these lyers, but it would require writing
quite a bit of new code in ext2fs, and in any case mmap of files greater
than 2 gigabytes would still not be supported.

At this point, support for >2gig files is a 2.3 issue, as I very much
doubt a new feature like this would make it past Linus's code freeze!

- Ted

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html