[1756] in linux-scsi channel archive
Re: Is SCSI partitioning universally readable/writable?
daemon@ATHENA.MIT.EDU (Art S. Kagel)
Fri Apr 25 11:02:06 1997
Date: Fri, 25 Apr 1997 10:58:06 -0400 (EDT)
From: "Art S. Kagel" <kagel@ns1.bloomberg.com>
To: Bradley M Keryan <keryan@andrew.cmu.edu>
Cc: Marc SCHAEFER <schaefer@alphanet.ch>, linux-scsi@vger.rutgers.edu
In-Reply-To: <Pine.LNX.3.96L.970425084703.24630A-100000@fatale.res.cmu.edu>
On Fri, 25 Apr 1997, Bradley M Keryan wrote:
> On 22 Apr 1997, Marc SCHAEFER wrote:
>
> > lbliao (lbliao@alumnae.caltech.edu) wrote:
> > > For example partitions were made using aha 1515, and not the card is 2940,
> > > will it read the disk? Is there any portability to the SCSI disk or format?
> > > Is there upward compatibility?
> >
> > Yes, as long as you use PC based partitionning. There is no
> > portability between architectures on the partition nor
> > filesystem level (the latter mainly for endianess issues).
>
> Not quite. This might be an extreme case, but my father has used a Syquest
> EZ135 drive to transfer files between an Amiga (running AmigaDOS, not
> Linux 68k) and a PC running Linux x86 2.0.25 before, with the amiga ffs
> filesystem driver. Perhaps Linux has rigid disk block support (Amiga
> partition tables) from the work done with Linux 68k?
>
> However, the x86 machine *was* able to read the disk, which had been
> partitioned and formatted on the Amiga, and there were no problems with
> endianness.
>
> Ironically, the EZ135 was about 30% bigger than the Amiga's hard drive ...
>
> On the other hand, don't PC SCSI BIOSes that use "extended translation"
> sometimes cause problems, when different controllers support different
> translation schemes? (Read: it may be possible to mount drives from a
> 12-year old Amiga 1000, but not from another PC SCSI controller? Or am I
> missing something here?)
You are correct. I had just such a problem switching controllers. I had
partitioned with a Future Domain controller and switched to a BusLogic
controller which used a different mapping scheme. In DOS everything
seemed fine but Linux had problems (could not install LILO, sometimes
hung trying to format a file system). But this can only happen on a
drive with >1024 cylinders which is the DOS/BIOS cylinder limit, and only if
you have cylinder translation turned on (if you do not have DOS or your
DOS partitions and your Linux boot partition are entirely contained in the
first 1024 cylinders so that the BIOS boot function can find it, you do
not need the extended cylinder translation feature).
>
> [ rest of message deleted ]
>
Art S. Kagel, kagel@bloomberg.com
