Re: [LI] Large disks

[Arun Sharma <adsharma@xxxxxxxxxxxxxxx>]

On Fri, Jan 21, 2000 at 11:00:45PM +0530, Atul Chitnis wrote:
> On Fri, 21 Jan 2000, Sathya Rangaswamy wrote:
> 
> > I would asssume it has a 4 gig limit on nodes in a 32 bit environement.
> 
> That can't be right - I have installed several machines with far larger
> partitions, including one with a straight 8GB / partition.
> 

The limit is on the size of a single file on Linux. From 2.3.x:

$ cd linux/include
$ grep kernel_off_t asm*/posix*.h
asm-alpha/posix_types.h:typedef long            __kernel_off_t;
asm-arm/posix_types.h:typedef long                      __kernel_off_t;
asm-i386/posix_types.h:typedef long             __kernel_off_t;
asm-m68k/posix_types.h:typedef long             __kernel_off_t;
asm-mips/posix_types.h:typedef long             __kernel_off_t;
asm-ppc/posix_types.h:typedef long              __kernel_off_t;
asm-sparc/posix_types.h:typedef long                   __kernel_off_t;
asm-sparc64/posix_types.h:typedef long                   __kernel_off_t;
asm-sparc64/posix_types.h:typedef int                    __kernel_off_t32;

As you can see, on i386, long = 32 bits. So files can't be bigger than
2^^32. But on alpha, sparc64 and the upcoming ia64, long will be 64 bits.

The data structures in the partition table talk in terms of cylinder units.
So overflow is not a problem yet.

I've also heard of claims about ext2fs not being dependent on the bitness
of the machine. In other words, you can create a fs on a disk on alpha and
take it to i386 and it will work just fine. But I haven't figured out how
that is achieved. I guess, the data structures get converted before they're
written out to the disk.

	-Arun
--------------------------------------------------------------------
The Linux India Mailing List Archives are now available.  Please search
the archive at http://lists.linux-india.org/ before posting your question
to avoid repetition and save bandwidth.