[ros-kernel] install problem

[Mike Nordell]

Just some misc comments, which may or may not apply.

Hartmut Birr wrote:

> I've a problem to install ros on my test computer. The installation
(usetup)
> works, but creates wrong entries in freeldr.ini for ros on the second
disk.
> The configuration is:
>
> primary ide master:      Seagate ST36531A
> primary ide slave:       Iomega ZIP 250
>
> secondary ide master:    Yamaha CDRW4416E
> secondary ide slave:     -
>
> SCSI controller SYM875: IBM DDYS-T09170N
>
> I'm using a modified scsiport driver and setupldr to load the driver for
the
> scsi controller. Usetup creates for the second disk entries like this:
>
> SystemPath=multi(0)disk(0)rdisk(2)partition(1)\ReactOS

This seems to me to be a case where code reuse would have been good - to use
the *exact* same code when detecting disks for creating entries in boot.ini
as when actually using them.

> The rdisk number is one too much.

Did you have a media in the Zip-drive during boot, and did the BIOS (of the
PC) deem it a "bootable disk" (it seems from your dump of usetup this wasn't
the case, but just for completeness)?

> I think that isn't a problem of the scsi
> controller. It seems that usetup interprets the ZIP drive as disk.

If media was present, and the BIOS of the PC gave it "bootable disk" status,
I think this would be correct. Else it's incorrect. Please note that this
can, if one knows about it, be used as a feature on the PC (even that it's
just a workaround for a flaw) - have a removable drive preceed a
non-removable drive, to assume its role if machine is booted with media in
drive.

> Usetup shows also the extended partition on the second disk.

Your second disk is the Zip-drive, if media was present. Else second disk
would be the SCSI drive (but then I'm not sure an rdisk() ARC entry would be
correct - I'd almost expect a scsi() entry, though I might be wrong and
scsi() entries are only present on SCSI-only systems).

> Output from usetup:

So long as partition() isn't a problem, disk geometry isn't of interest.
If/once partition() becomes an issue, partition information such as "active"
and "in extended partition" becomes vital.

[snip]
>   --  Type 15                           8738 MB

What is "15" supposed to mean?

/Mike