From: "Albert D. Cahalan" <acahalan@cs.uml.edu>
To: linux-kernel@vger.rutgers.edu
Subject: Re: VFS/ext2fs - large files on the Alpha (fails for 17GB+)
Date: Wed, 19 Aug 1998 03:17:07 -0400 (EDT)
Message-Id: <199808190717.DAA09901@jupiter.cs.uml.edu>
http://www.cs.uml.edu/~acahalan/linux/ext2.gif
Summary:
1 kB -> 16 GB
2 kB -> 256 GB
4 kB -> 4 TB
8 kB -> 32 TB (limit is block counter)
(8kB is _not_ portabel, it requires 8k pages, so you can't use it on ix86!!)
That assumes you won't hit a 32-bit sector number limit in some
other part of the kernel, cutting you off at 1 TB or 2 TB.
From: Matti Aarnio <matti.aarnio@sonera.fi>
To: linux-kernel@vger.rutgers.edu
Subject: Re: VFS/ext2fs - large files on the Alpha (fails for 17GB+)
Date: Thu, 20 Aug 1998 09:17:11 +0300 (EEST)
Message-Id: <19980820061717Z92213-453+18@mea.tmt.tele.fi>
> Does that means we *can* have a disk partition (or a RAID system) with a
> max. size of 70TB with ext2fs on Intel platforms with a block size of 8192
> bytes? If so, will that be enough to satisfy one's lust for ever expanding
> disk space....
Umm... Propably not. While the theorethical block addressability
goes up like (blksize/4)**3 function, the maximum offset encodable
with blocks is different (non-exponential) function:
32-bits per block offset within FS, giving us 4G blocks maximum.
blksize offset (blksize/4)^3 max size
1k 4 TB 16 GB 16 GB
2k 8 TB 256 GB 256 GB
4k 16 TB 4 TB 4 TB
8k 32 TB 64 TB 32 TB
So, you can't have 70TB filesystem with EXT2 even if you had the
hardware... (counting with 1024 byte kilos that is 64 TB)
Also I think that at Intel you are better off to limit yourself
to 4k blocks as those match paging system page size, and using
larger blocks than pagesize is rather difficult...
Harald
--
All SCSI disks will from now on ___ _____
be required to send an email notice 0--,| /OOOOOOO\
24 hours prior to complete hardware failure! <_/ / /OOOOOOOOOOO\
\ \/OOOOOOOOOOOOOOO\
\ OOOOOOOOOOOOOOOOO|//
Harald Koenig, \/\/\/\/\/\/\/\/\/
Inst.f.Theoret.Astrophysik // / \\ \
koenig@tat.physik.uni-tuebingen.de ^^^^^ ^^^^^
-
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.tux.org/lkml/