XP elsewhere needs files on 1st partitoin primary master?

[dmorgan1]

Can XP be installed on-- and confined to-- the first partition of a
primary slave hard disk? That's where I want to put and use it.

I find that at installation it insists on access to the first
partition of the primary *master* -- even though I have no intention
of locating it there-- as apparent working space for use during the
installation itself. But beyond that, once installed, it doesn't
relinquish the use of first partition on primary master-- there are
files it maintains on the primary master needed for it to boot from
the primary slave. Other than those boot files, the whole filesystem
hierarchy is on the primary slave and boots and runs happily from
there. But I had other plans for first partition primary master than
to house a dozen files that an OS elsewhere can't boot without.

Is there any way for a copy of XP to have everything it needs-- not
only for running but for booting-- within primary slave first
partition and not leak its disk usage requirements to other
partitions?

 

[Jupiter Jones [MVP]]

Many have been caught short when it came time to install SP-1 and then again
with SP-2.
Both of those have space requirements that can not be deviated:
http://support.microsoft.com/?kbid=837783

Plan for the future.

 

[dmorgan1]

The reference tells "how much" but not "where" the service packs need
disk space. The needed amount, and more, is available on the first
partition of my second (ie, primary slave) disk. Can the OS, and the
service packs, tap it and be happy? or is there a hard requirement
that they put their material on the first partition of the first disk
and nowhere else?

 

[Jupiter Jones [MVP]]

Did you miuss this:
"You must also have 30 MB of free hard disk space on the first primary
system partition. The first primary system partition is the disk volume that
contains the hardware-specific files that are required to start Windows. For
example, the primary system partition contains the Ntldr file, the Boot.ini
file, and the Ntdetect.com file."
About the middle of the page.

 

[dmorgan1]

So, can Windows' "first primary system partition" (the one holding the
special files) be the first partition of the primary slave?

My special files landed on the first partion of the primary master,
making that partition by definition be the "first primary system
partition." But I don't want them there. I'd rather they be on the
hardware's first partition of primary slave along with the rest of
Windows' files. Or equivalently in other words, I'd rather Windows'
first primary system partition be the hardware's
first-partition-of-primary-slave instead of
first-partition-of-primary-master. Possible? Impossible?

 

[Ron Sommer]

Yes
The first active partition will be the first primary system partition.
This will be the C: drive.
I am guessing that the Windows files are on D:?
To get what you want, the first-partition-of-primary-slave must be the first
active partition on the slave drive and you must reinstall XP.

 

[Bruce Chambers]

Can XP be installed on-- and confined to-- the first partition of a
primary slave hard disk? That's where I want to put and use it.


Is there any way for a copy of XP to have everything it needs-- not
only for running but for booting-- within primary slave first
partition and not leak its disk usage requirements to other
partitions?

Not really, unless you're going to use a 3rd party boot manager.
WinXP, like all Microsoft operating systems, needs to put it's system
files on the bootable hard drive (the active primary partition of the
master hard drive, iow), at the very least.

--

Bruce Chambers

Help us help you:



You can have peace. Or you can have freedom. Don't ever count on having
both at once. - RAH

 

[Art]

Not really, unless you're going to use a 3rd party boot manager. WinXP,
like all Microsoft operating systems, needs to put it's system files on
the bootable hard drive (the active primary partition of the master hard
drive, iow), at the very least.

--

Bruce Chambers

Help us help you:



You can have peace. Or you can have freedom. Don't ever count on having
both at once. - RAH

dmorgan & Bruce:
Assuming I correctly understand the OP's query, there is a strong
possibility that his hard drive containing the XP OS will boot even if it is
connected as Primary Slave. It is true that some motherboards will *not*
boot to a hard drive that is configured as Slave, but many will. There's
certainly no harm in connecting the drive as PS to see if it will boot.

Why the OP wants to install the OS as a Primary Slave rather than Primary
Master is another question. I take it that it has something to do with his
question about the OS "not leak(ing) its disk usage requirements to other
partitions", but I don't understand what he's driving at.
Art

 

[dmorgan1]

dmorgan & Bruce:
Assuming I correctly understand the OP's query, there is a strong
possibility that his hard drive containing the XP OS will boot even if it is
connected as Primary Slave. It is true that some motherboards will *not*
boot to a hard drive that is configured as Slave, but many will. There's
certainly no harm in connecting the drive as PS to see if it will boot.

Why the OP wants to install the OS as a Primary Slave rather than Primary
Master is another question. I take it that it has something to do with his
question about the OS "not leak(ing) its disk usage requirements to other
partitions", but I don't understand what he's driving at.
Art

Partition map is shown below. Windows is installed on the first (and
only) partition of the primary slave. There, yes, it does indeed boot,
but not unless certain files (ntldr etc) are sitting over on the first
partition of the other drive and available to it (the thing I seek to
avoid).

Drive letter assignments by XP -
hda1, first primary partition on primary master disk, gets letter C:
hda4, extended partion on primary master disk, gets letter E:
hdb1, first primary partition on primary slave disk, gets letter I:
The other partitions get no letters and the other letters get no
partitions. But it looks like the presence of the unlettered
partitions may well influence the lettering choice for those that do.
This is too bad because if true I'm not free to add or remove
partitions on the primary master drive without Windows then
re-lettering the one on the primary slave (like, if I added a
partition, I: would probably become J:; if I removed one, I: would
become H. The registry sitting on that partition is probably full of
references to I: so Windows would thus generate a disparity between
the name it uses, and the one recorded in the registry, for the
partition.

Why I want this? I had XP installed in the first partition of the
primary master but it would not "hold." About 10 times since I bought
the machine last summer XP has failed, irrecoverably, to boot and I
would reinstall it fresh. I give up now and want to experiment with
putting XP onto a different drive altogether. So I added the primary
slave drive to the machine last week and am now experimenting.

I'm going to follow Ron Sommer's suggestion and turn off the "active"
marking on the primary master first partition, while turning it on on
the primary slave first partition. Another friend suggested installing
with the primary master drive disconnected altogether during the
installation itself, reconnecting it afterwards. We'll see what
happens. Thanks for everybody's comments.

---------------------------------------------
Partition map:

hda is primary master hard disk
hdb is primary slave hard disk

On the hda hard disk, hda1, hda2, and hda3 are primary partitions and
hda4 is an extended partition. Inside hda4, hda5, hda6, and hda7 are
contained logical partitions.

On the hdb hard disk, hdb1 is a primary partition.


Disk /dev/hda: 163.9 GB, 163928604672 bytes
255 heads, 63 sectors/track, 19929 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 637 5116671 7 HPFS/NTFS
/dev/hda2 2551 2805 2048287+ 83 Linux
/dev/hda3 2806 3060 2048287+ 82 Linux swap
/dev/hda4 3061 19929 135500242+ f W95 Ext'd
(LBA)
/dev/hda5 3061 10709 61440561 8e Linux LVM
/dev/hda6 10710 11984 10241406 83 Linux
/dev/hda7 11985 12366 3068383+ 83 Linux

Disk /dev/hdb: 8455 MB, 8455200768 bytes
255 heads, 63 sectors/track, 1027 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 1 1026 8241313+ 7 HPFS/NTFS

 

[dmorgan1]

Yes
The first active partition will be the first primary system partition.
This will be the C: drive.
I am guessing that the Windows files are on D:?
To get what you want, the first-partition-of-primary-slave must be the first
active partition on the slave drive and you must reinstall XP.

Hmmm.. the only partition marked active was the
first-partition-of-primary-master. I unmarked it, then deleted it
altogether. Then I marked active my desired XP installation target,
first-partition-of-primary-slave.

However, when booting from the XP installation CD, it insisted that I
must let it create a partition on the primary master. Having done so,
in proposing to format the first-partition-of-primary-slave it labeled
it letter I:, as before.

Apparently XP *MUST* utilize that first-partition-of-primary-master,
no matter we want it elsewhere. I also tried to have it install with
primary master disconnected physically. No go. Says "missing ntldr,
press ctl-alt-del to reboot." A hard/inflexible design requirement,
I'm concluding.

 

[cquirke (MVP Win9x)]

On Fri, 21 Jan 2005 16:15:44 -0800, dmorgan1

Hmmm.. the only partition marked active was the
first-partition-of-primary-master. I unmarked it, then deleted it
altogether. Then I marked active my desired XP installation target,
first-partition-of-primary-slave.

Apparently XP *MUST* utilize that first-partition-of-primary-master,
no matter we want it elsewhere.

This is not an OS issue, but a system-level issue that the OS must
conform to. Traditional BIOS design will always boot the active
primary partition on the first HD it sees. Modern BIOS may change
this via CMOS settings, but should an OS rely on this?

When you first turn on the PC, there is no OS, not should there be.
There is only the system code (BIOS).

Just as an OS is there to run any number of possible applications, so
it is that a system is there to run any number of possivle operating
systems. A PC that can only run Windows is as unreasonable as an OS
that can only run, say, MS Office.

So when the system boots, it finds and boots an OS by following a
standard procedure. Any OS that wants to be PC-compatible has to fit
that procedure, and that means it has to start from the active
partition on the first hard drive.

That doesn't mean it has to be installed in that partition; it can
start a small stub of code that can cross over to a different
partition to load the rest of the OS.

That's what is happening here, and all MS OSs work in this way.

MS-DOS can start from C:\IO.SYS, C:\MSDOS.SYS and C:\Command.com, and
locate the rest of DOS via an entry in the Path statement. You can
relocate or replace Command.com via a statement in Config.sys

Win9x can do the same as MS-DOS, by setting variables in the now-text
C:\MSDOS.SYS (or overriding C:\Winboot.ini); once again, only a
handful of files (under 5M) need to be on C:

NT can do the same sort of thing, though the filenames involved are
different. The first file started in NT (and Win2000 and XP) is
C:\NTLDR, and this processes C:\BOOT.INI; the second file can tell the
boot code to look elsewhere for the rest of the OS. So once again,
all but the first 5M or so files can be off C:

You may be alble to tell BIOS, via CMOS settings boot order, to treat
something else as the "first hard drive in the system". This
generally works when it's choosing SCSI before IDE, or S-ATA before
UIDE. It may or may not work when choosing a particular drive within
these interfaces, e.g. IDE primary slave before primary master.

That's because the OS has to understand what is going on. It likely
understands one interface before another, but may still assume an
order of devices within that interface.

I've covered this sort of stull in the section on multi-booting at
http://cquirke.mvps.org

--------------- ----- ---- --- -- - - -

Error Messages Are Your Friends

 

[dmorgan1]

Thanks. I got it to work by forcing the issue with trickery.
Temporarily, I disconnected the main primary master drive (which I
don't want XP to touch), and rejumpered the slave drive to master. I
installed XP on the drive under those circumstances, preventing XP
from fulfilling its instinct to write to the other drive by
eliminating it. XP installs normally, boots, runs.

Then I put things back-- jumpered the XP drive to slave again and
reconnected the original master. Only worry I had was that XP would
trip on the fact that its drive is no longer master. That problem was
at first overcome using your BIOS suggestion. My BIOS lets me direct
booting to the slave drive, making it, for booting if not wiring
purposes, "first." When I do that, Windows comes up fine, sees its
partition as letter "C" and away we go. Problem then is, I can't
conveniently boot my OS on the master. Could do with a floppy but
that's ugly. I found a bootloader that can do, I guess, the same thing
as that BIOS setting in software. Bootloader is GRUB, which has a
"map" command that "performs a virtual swap between your first and
second hard drive." More if you're interested at
http://www.gnu.org/software/grub/manual/html_node/DOS-Windows.html#DOS/Windows

One uncertainty. I understand that the PC architecture, not any OS,
is responsible for the decision to choose where to boot from. But I
think it chooses "the MBR on the first hard drive." What happens
beyond that is entirely up to whatever arbitrary code is in that MBR.
"What the first sector of the boot disk does is entirely up to whoever
programmed that code." The OS-independent phase of booting stops
short of saying anything about active (or other) partitions. Maybe the
MBR code has something to say about an active partition on the first
hard drive or maybe not. The code that DOS and Windows put there when
you install them does. That code takes special and central interest in
the active partition on the first drive. But that is beyond the
OS-independent part of the booting sequence. It's on the OS side of
the line, and is a Microsoft thing. Counterexamples: 1) Linux
installed elsewhere than the active partition of the first drive makes
no use of that partition. 2) Faked out (as describe above), neither
does XP on my machine now. ntldr and friends are currently NOT
situated in the active partition of the first drive. They're in the
first partiton of the second drive. And XP boots whole and fine from
there.