Re: Install problem on Supermicro X6DHT-G system
On Tue, Jan 03, 2006 at 05:44:42PM +0000, A J Stiles wrote:
> CD-ROM drives usually used to be factory-set to "slave" so that a clueless
> n00b would be unlikely to do any harm by attaching it straight to the
> existing HDD cable {the HDD is invariably set to master}. However, DVD+RW
> writers {which need a high throughput if you don't want many expensive
> laser-decorated beer mats} are commonly factory-set to "master" and supplied
> with a new cable.
>
> Most operating systems, and most motherboards, really don't care about this;
> the drives are not really "master and slave" in the sense that communications
> meant for the "slave" are not relayed through the "master". These were just
> convenient designations that seem to have stuck.
>
> MS-DOS *does* care about this, though Windows 95/98/ME could access CD-ROM
> devices natively without going through DOS. So it was posible for a really
> misconfigured Win9X machine to have some applications {true 32-bit ones}
> seeing the CD drive and other applications {16-bit ones} not seeing it.
>
> 80386 Linux will install onto, and boot from, hdb quite happily with no hda
> present {I've done it by mistake myself once}. And it doesn't care if the
> CD-ROM device is hdd with no hdc.
>
> However, I *have* encountered *one* motherboard which *did* complain about
> slave IDE devices with no master present. That was an ASRock board with some
> kind of AMD 32-bit CPU. Whether this is still done on 64-bit motherboards, I
> do not know.
As far as I understood it (and I am not an expert on ATAPI), the master
does have something to do with the ide bus timings that the slave
doesn't, although many systems may just let the slave pick them if there
is no master, while others don't. I always make sure to have a master
only, or a master and a slave, since there really isn't any good reason
to not do it properly.
Len Sorensen
Reply to: