Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 11 Mar 2004 06:16:43 +1100 (EST)
From:      Bruce Evans <bde@zeta.org.au>
To:        John Baldwin <jhb@FreeBSD.org>
Cc:        freebsd-current@FreeBSD.org
Subject:   Re: Expand libdisk to use 64bit size/offset
Message-ID:  <20040311054522.T2459@gamplex.bde.org>
In-Reply-To: <200403101309.32292.jhb@FreeBSD.org>
References:  <7m7jxth70h.wl@black.imgsrc.co.jp> <200403101309.32292.jhb@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 10 Mar 2004, John Baldwin wrote:

> On Tuesday 09 March 2004 07:37 pm, Jun Kuriyama wrote:
> > I found sysinstall(8) cannot handle TB class storage.  It looks
> > both of libdisk and sysinstall should be fixed.
> >
> > I'm trying to improve this situation, but I'm not sure I'm going to
> > the right direction.  Please look this patch (including my debugging
> > stuff and XXX comments which should be resolved/removed).

This will probably need EFI, since DOS partitions are limited to 2^31
or 2^32 sectors and the sector size may be restricted to 512.

> > http://www.imgsrc.co.jp/~kuriyama/BSD/libdisk-20040310.diff

I didn't look closely.

> > I choose daddr_t as larger storage for offset, size and end member of
> > "struct chunk".  I think offset and end is okay, but I don't know
> > daddr_t for offset is correct or not.
>
> dadd_t is a uint32_t, so it won't help.  Just use off_t for chunk
> offset/size/end.

Erm, daddr_t is (signed) int64_t.  It is the correct type for storing
counts of DEV_BSIZE'd blocks (DEV_BSIZE = 512), so it is correct for
sector counts in libdisk to the extent that libdisk is limited to
DEV_BSIZE'd sectors.  Counts of blocks of other sizes should use a
different typedef.  E.g., ufs uses types ufs1_daddr_t (32-bit ufs1
block numbers), ufs2_daddr_t (64-bit ufs2 block numbers) and ufs_lbn_t
(64-bit ufs logical block numbers).

libdisk never uses DEV_BSIZE or even its own macro for the units.  It
hard-codes 512-all over :-(.  So using daddr_t would not be logically
correct but would be less worse than what libdisk already does.

Using off_t for block counts is very bogus.  off_t is for byte offsets.

libdisk's "offset" is actually a block number, so it should have the
same [foo_]daddr_t type as "size" and "end".

Bruce



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040311054522.T2459>