Boot Win2K -> BlackSOD TRAP 0000000D (NTLDR?)

[Frnak McKenney]

Note: The specific problem described is with Win2K Pro, but I'm
including comp.os.linux.misc because the problem appears to
have been triggered by a Linux install.

I'm seeing a _black_ screen'o'death (TRAP 0000000D) when I attempt
to boot a previously-working Win2KPro/SP4 system. Based on my
testing, the trap (at cs:eip=0008:003175D0) is occurring at some
point after NTLDR is found, but before the boot.ini list of systems
to boot is displayed.

I've run almost every Win2K recovery procedure I can find, short of
a complete reinstall, but I still see the same TRAP screen. Any
clues would be appreciated.

Background
----------

The hardware for 'mongoose' is an ASUS K8N-E-Deluxe Socket-754
motherboard with an AMD Sempron 3000 CPU, 512 Mb of ECC RAM, a
Maxtor 40Gb IDE HD, and an ASUS V9400-X AGP video adapter. I
installed Win2KPro/SP4 on it in a 10Gb extended partition last May,
with E:NTFS as the boot partition and C:FAT as the system partition.
Barring the odd Win2K hang it ran fine through January of this year.

In early February I needed to use the machine for some Linux work
(Blender and The GIMP), so I installed SuSE Linux 9.1. The SuSE
install set up its own boot loader, 'grub', in the MBR, adding an
entry on the Grub menu for booting the C: partition.

Problem
-------

I didn't need Win2K again until a few days ago, so I wasn't _too_
surprised when the 'Windows' boot menu option failed, but I was a
bit surprised to see a 'black screen of death' TRAP. No preceding
error message, no display of boot options ("safe" would have been
_really_ nice!), no nothing except the trap screen.

Trap Screen
-----------

TRAP 0000000D ==..=== GENERAL PROTECTION FAULT ==..==

tr=0028 cr0=00000011 cr2=00000000 cr3=00000000
gdt limit=03FF base=00017000 idt limit=07FF base=00017400

cs:eip=0008:003175D0 ss:esp=0010:0006202C errcode=0000
flags=00010086 NoCy NoZr IntDis Down TrapDis
eax=00000000 ebx=00000000 ecx=00000000 edx=00000000 ds=0010 es=0010
edi=00000000 esi=00000000 ebp=00000000 cr0=00000011 fs=0030 gs=0000


Already Tried
-------------

I booted from the Win2K CD and used the Recovery Console and FIXMBR
to reinstall a "proper" Win2K MBR. The TRAP screen still appeared.

I re-booted and ran FIXBOOT, which claimed C: had been fixed...
but it hadn't. I tried "automatic" recovery and "manual" recovery;
the TRAP screen appeared each time I tried to boot from the hard
drive.

I ran CHKDSK (with /P, then with /R) on both C: and E: -- no
errors or fixups were reported.

I finally gave up and re-installed the Linux boot loader, which let
me get access to the partitions.

(Query: Am I the only one who has noticed that you can't redirect
or capture the output from the Recovery Console commands? That's
why my MAP results below, for example, are hand-typed <loud
gnashing of teeth>).

Renaming my C:\NTLDR to NTLDR.000 at least reported 'NTLDR not
found' (a welcome change from the boring TRAP text). This reassured
me that my Win2K boot attempts were at least loading the C:
partition boot record and that the failure was likely somewhere in
NTLDR. Renaming NTDETECT, ARCLDR, and ARCSETUP had no effect on the
TRAP screen.

Using the Linux 'cmp' command I verified that the files NTLDR,
NTDETECT.COM, ARCLDR.EXE, and ARCSETUP.EXE on my C: partition were
byte-identical to those on the Win2K CD. (I couldn't find a Recovery
Console equivalent of 'cmp' -- did I just miss it?)

Request
-------

So what's _left_ to "fix" in my C: partition?

Or... is it possible that the portion of the NTLDR code prior to
the part that displays the boot.ini choices could be sensitive to
something in the E: partition?

Or -- since I did make some changes to the disk layout (but only to
partitions _after_ E -- could NTLDR be sensitive to non-Win2K
partitions, or maybe to partitions not created by Win2K DISKPART?
If so, why no objections from DISKPART or CHKDSK?

I've spent a number of hours with Google over the past few days
looking for similar TRAP screens or people reporting similar
problems with booting Win2K, WinXP, or even WinNT. No luck so far.

Does this remind anyone of anything? Clues, hints, or suggestions
will be muchly appreciated (no, I already thought of _that_ one,
thank you, but I'm holding off... um, "recycling" the hardware
until I get $300 worth of mad <grin?>)

My final fallback is to use Linux to FTP the most critical files
over to another system, or burn them to a CD, and then do a complete
reinstall. My concern is that, if this is a partitioning-related
problem, I might expend a lot of effort and not make much progress.

My thaks to all who reply.


Frank McKenney


Further Information
-------------------

DISKPART
--------------------------------------------------
39206 MB Disk 0 at Id 0 on bus 0 on atapi
C: FAT 8 MB ( 7 MB free)
D: FAT (LIB-BOOT) 196 MB ( 195 MB free)
E: NTFS 10001 MB ( 2689 MB free)
F: FAT (WORK-VOL) 286 MB ( 285 MB free)
Unpartitioned space 742 MB
G: FAT (LINUX-SWAP) 1028 MB ( 1027 MB free)
-- 2 MB
-- 20480 MB
-- 6454 MB
Unpartitioned space 8 MB
--------------------------------------------------

Win2K MAP command (very odd) (typed by hand)
--------------------------------------------------
? 0 MB \Device\Harddisk0\Partition0
C: FAT16 8 MB \Device\Harddisk0\Partition1
? 39190 MB \Device\Harddisk0\Partition0
? 8 MB \Device\Harddisk0\Partition0
D: FAT16 196 MB \Device\Harddisk0\Partition2
? 0 MB \Device\Harddisk0\Partition0
E: NTFS 10001 MB \Device\Harddisk0\Partition3
? 0 MB \Device\Harddisk0\Partition0
F: FAT16 286 MB \Device\Harddisk0\Partition4
? 742 MB \Device\Harddisk0\Partition0
G: FAT16 1028 MB \Device\Harddisk0\Partition5
? 0 MB \Device\Harddisk0\Partition0
? 2 MB \Device\Harddisk0\Partition6
? 0 MB \Device\Harddisk0\Partition0
? 20480 MB \Device\Harddisk0\Partition7
? 0 MB \Device\Harddisk0\Partition0
? 6454 MB \Device\Harddisk0\Partition8
? 0 MB \Device\Harddisk0\Partition0
A: \Device\Floppy0
H: \Device\CdRom0
--------------------------------------------------


Linux fdisk -l
--------------------------------------------------
Disk /dev/hda: 41.1 GB, 41110142976 bytes
16 heads, 63 sectors/track, 79656 cylinders
Units = cylinders of 1008 * 512 = 516096 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 16 8001 1 FAT12
Partition 1 does not end on cylinder boundary.
/dev/hda2 16 79640 40130370 f W95 Ext'd (LBA)
/dev/hda5 17 415 200781 6 FAT16
/dev/hda6 415 20735 10241406 7 HPFS/NTFS
/dev/hda7 20736 21316 292792+ e W95 FAT16 (LBA)
/dev/hda8 22823 24911 1052226 6 FAT16
/dev/hda9 24912 24915 1984+ 82 Linux swap
/dev/hda10 24916 66526 20971912+ 83 Linux
/dev/hda11 66527 79639 6608920+ 83 Linux
--------------------------------------------------


C: directory listing (via Linux)
--------------------------------------------------
/windows/C:
total 602
drwxrwxr-x 3 root users 16384 Dec 31 1969 .
drwxr-xr-x 7 root root 4096 Feb 19 16:38 ..
drwxrwxr-x 3 root users 2048 May 6 2005 Recycled
-r-xr-xr-x 1 root users 150528 Jul 14 2003 arcldr.exe
-r-xr-xr-x 1 root users 163840 Jul 14 2003 arcsetup.exe
-rwxrwxr-x 1 root users 0 May 6 2005 autoexec.bat
-rwxrwxr-x 1 root users 614 Mar 29 12:05 boot.ini
-rwxrwxr-x 1 root users 192 Mar 28 22:30 boot.ini0
-rwxrwxr-x 1 root users 261 Mar 28 22:43 boot.ini1
-rwxrwxr-x 1 root users 0 May 6 2005 config.sys
-rwxrwxr-x 1 root users 4109 Jan 26 23:20 ffastun.ffa
-rwxrwxr-x 1 root users 8192 Jan 26 23:20 ffastun.ffl
-rwxrwxr-x 1 root users 4096 Jan 26 23:20 ffastun.ffo
-rwxrwxr-x 1 root users 4096 Jan 26 23:20 ffastun0.ffx
-r-xr-xr-x 1 root users 0 May 6 2005 io.sys
-r-xr-xr-x 1 root users 0 May 6 2005 msdos.sys
-r-xr-xr-x 1 root users 34724 Jul 14 2003 ntdetect.com
-r-xr-xr-x 1 root users 214432 Jul 14 2003 ntldr

 

[philo]

Note: The specific problem described is with Win2K Pro, but I'm
including comp.os.linux.misc because the problem appears to
have been triggered by a Linux install.

I'm seeing a _black_ screen'o'death (TRAP 0000000D) when I attempt
to boot a previously-working Win2KPro/SP4 system. Based on my
testing, the trap (at cs:eip=0008:003175D0) is occurring at some
point after NTLDR is found, but before the boot.ini list of systems
to boot is displayed.

I've run almost every Win2K recovery procedure I can find, short of
a complete reinstall, but I still see the same TRAP screen. Any
clues would be appreciated.

<snip>

even though you compared NTLDR on your HD to the copy on your cd...
I'd still try replacing it

 

[Robert Hull]

In comp.os.linux.misc, on Fri 31 March 2006 22:48, Frnak McKenney

Note: The specific problem described is with Win2K Pro, but I'm
including comp.os.linux.misc because the problem appears to
have been triggered by a Linux install.
[snip]

In early February I needed to use the machine for some Linux work
(Blender and The GIMP), so I installed SuSE Linux 9.1. [snip]

I didn't need Win2K again until a few days ago, so I wasn't _too_
surprised when the 'Windows' boot menu option failed, [snip]


Already Tried
-------------

All good things normally, but no good in this specific case.

This is a known bug with the version of parted that came with SUSE Linux
9.1, the explanation, cause and remedy is documented at
http://en.opensuse.org/SDB:Windows_No_Longer_Boots_Following_the_Installation_of_SUSE_LINUX_9.1

 

[Douglas Mayne]

Note: The specific problem described is with Win2K Pro, but I'm
including comp.os.linux.misc because the problem appears to
have been triggered by a Linux install.

I'm seeing a _black_ screen'o'death (TRAP 0000000D) when I attempt
to boot a previously-working Win2KPro/SP4 system. Based on my
testing, the trap (at cs:eip=0008:003175D0) is occurring at some
point after NTLDR is found, but before the boot.ini list of systems
to boot is displayed.

I've run almost every Win2K recovery procedure I can find, short of
a complete reinstall, but I still see the same TRAP screen. Any
clues would be appreciated.

DISKPART
--------------------------------------------------
39206 MB Disk 0 at Id 0 on bus 0 on atapi
C: FAT 8 MB ( 7 MB free)
D: FAT (LIB-BOOT) 196 MB ( 195 MB free)
E: NTFS 10001 MB ( 2689 MB free)
F: FAT (WORK-VOL) 286 MB ( 285 MB free)
Unpartitioned space 742 MB
G: FAT (LINUX-SWAP) 1028 MB ( 1027 MB free)
-- 2 MB
-- 20480 MB
-- 6454 MB
Unpartitioned space 8 MB
--------------------------------------------------

Win2K MAP command (very odd) (typed by hand)
--------------------------------------------------
? 0 MB \Device\Harddisk0\Partition0
C: FAT16 8 MB \Device\Harddisk0\Partition1
? 39190 MB \Device\Harddisk0\Partition0
? 8 MB \Device\Harddisk0\Partition0
D: FAT16 196 MB \Device\Harddisk0\Partition2
? 0 MB \Device\Harddisk0\Partition0
E: NTFS 10001 MB \Device\Harddisk0\Partition3
? 0 MB \Device\Harddisk0\Partition0
F: FAT16 286 MB \Device\Harddisk0\Partition4
? 742 MB \Device\Harddisk0\Partition0
G: FAT16 1028 MB \Device\Harddisk0\Partition5
? 0 MB \Device\Harddisk0\Partition0
? 2 MB \Device\Harddisk0\Partition6
? 0 MB \Device\Harddisk0\Partition0
? 20480 MB \Device\Harddisk0\Partition7
? 0 MB \Device\Harddisk0\Partition0
? 6454 MB \Device\Harddisk0\Partition8
? 0 MB \Device\Harddisk0\Partition0
A: \Device\Floppy0
H: \Device\CdRom0
--------------------------------------------------


Linux fdisk -l
--------------------------------------------------
Disk /dev/hda: 41.1 GB, 41110142976 bytes
16 heads, 63 sectors/track, 79656 cylinders
Units = cylinders of 1008 * 512 = 516096 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 16 8001 1 FAT12
Partition 1 does not end on cylinder boundary.
/dev/hda2 16 79640 40130370 f W95 Ext'd (LBA)
/dev/hda5 17 415 200781 6 FAT16
/dev/hda6 415 20735 10241406 7 HPFS/NTFS
/dev/hda7 20736 21316 292792+ e W95 FAT16 (LBA)
/dev/hda8 22823 24911 1052226 6 FAT16
/dev/hda9 24912 24915 1984+ 82 Linux swap
/dev/hda10 24916 66526 20971912+ 83 Linux
/dev/hda11 66527 79639 6608920+ 83 Linux

Wow, you are more patient than me to retype all of that information.

I see the response by Robert Hull, and I agree with it. LBA mode is
important and maybe you weren't using it for some reason. Get your data
first, then start over with that disc. You might look for another disc-
since discs are cheap. That may be an easier way to start over.

Unsolicited advice...
If you are going to keep windows, put it on partition 1 using NTFS. You
probably had a reason for the FAT12 partition, but my advice is to
give Windows as little reason to bitch as possible. Maybe it will be
happier on partition 1 on a disc using LBA addressing. Then maybe not,
the little bitch that it is.

 

[Frnak McKenney]

Philo,

Thanks for responding. One is never _quite_ certain one's post hasn't
wound up at usenet.black_hole.internet. <grin!>

even though you compared NTLDR on your HD to the copy on your cd...
I'd still try replacing it

Ah! Someone who shares my sense of paranoia. Just because you're
paranoid doesn't mean Murphy isn't out to get you... <grin>

I've added this to my list. Will let you know how it turns out.


Frank McKenney, McKenney Associates
Richmond, Virginia / (804) 320-4887
Munged E-mail: frank uscore mckenney ayut minds pring dawt cahm (y'all)

 

[Frnak McKenney]

Robert,

Thanks for the reply. And thank you (and everyone else) for snipping
my rather wordy posting.

In comp.os.linux.misc, on Fri 31 March 2006 22:48, Frnak McKenney

Note: The specific problem described is with Win2K Pro, but I'm
including comp.os.linux.misc because the problem appears to
have been triggered by a Linux install.
[snip]

In early February I needed to use the machine for some Linux work
(Blender and The GIMP), so I installed SuSE Linux 9.1. [snip]

I didn't need Win2K again until a few days ago, so I wasn't _too_
surprised when the 'Windows' boot menu option failed, [snip]


Already Tried
-------------

All good things normally, but no good in this specific case.

This is a known bug with the version of parted that came with SUSE Linux
9.1, the explanation, cause and remedy is documented at
http://en.opensuse.org/SDB:Windows_No_Longer_Boots_Following_the_Installation_of_SUSE_LINUX_9.1

Thank you very much. This sounds very much like my problem, and I
read and printed off this entry in the SuSE KB (hard copy is very
handy when one is working on a boot-level problem <grin>).

For anyone reading over our shoulders, the link points to a description
of a problem with the version of 'parted', the partition editor that
was supplied with SuSE 9.1 and used during installation. The procedure
describes floppy- and CD-based approaches to applying a specific fix
to the... "improper" CHS values stored in the partition descriptions.

I say 'descriptions' because, once one creates an extended partition,
a hard drive's partition information is no longer contained solely
within the MBR; additional parts are scattered through the logical
volumes within that extended parttiion. Fixing them is no longer a
simple one-sector rewrite.

Robert,

I checked my BIOS to see if I could use the quick-fix of enabling LBA
mode. Unfortunately, my only LBA-related options are [LBA] and [Auto].
(Excuse me a sec' -- pile 35b just slid to the floor)...ah. Here's the
BIOS info:

Primary IDE Master Type: [Auto]
Device: Maxtor 6E040L0 LBA/Large Mode: [Auto] (vs. Disabled)
Size: 41.1GB PIO Mode: [Auto]
LBA Mode: Supported DMA Mode: [Auto]
Block Mode: 16sectors SMART Monitoring: [Auto]
PIO Mode: 4 32bit Data Transfer [Disabled]
Async DMA: Multihead DMA-2
Ultra DMA: Ultra DMA-6
SMART Monitoring: Supported

I applied the fixup via a CD, and saw the following dialog (the SuSE
graphics are a lot prettier <grin>):

-----------------------
Select a disk
hda 41.1Gb, CHS 79656/16/63 [broken]
<OK> <cancel>
-----------------------

"Broken" sounded like A Clue, so I chose <OK>, and this dialog
appeared:

------hda: 41.1G GB CHS 79656/16/63------
Warning: fdisk has some complaints.
FAT/NTFS partitions suggest these CHS values:
/dev/hda1: fat12, CHS 4998/255/63
/dev/hda5: fat16, CHS 4998/255/63
/dev/hda6: ntfs, CHS 4998/255/63
/dev/hda7: fat16, CHS 79656/16/63
/dev/hda8: fat16, CHS 4998/255/63
Warning: FAT/NTFS partitions with differing CHS values.
Warning: Accessing these partitions from Windows might
fail due to a disk geometry mismatch:
/dev/hda1
/dev/hda5
/dev/hda6
/dev/hda8
Would you like me to try and fix the partition table?
<OK> <cancel>
-----------------------------------------------------

I selected <OK>, and the previous dialog appeared. After two passes
with no obvious chanes int he text, I chose <cancel> and rebooted.

Same TRAP 0000000D screen.

It may be that the SuSE fixup was designed to only handle one bad
partition and, knowing its limitations, it refuses to make partial
changes to the partition information. I'm going to take a closer
look at the fixup CD and see if I can find a script I can hack or
something similar.

I'm also going back to Google, now that you've provided me with a
much better definition of the probable cause of my difficulties...
and a new set of keyoards (win2k trap 0000000d wasn't very
productive).

Query, if you happen by: assuming that the CHS information is the
_only_ bad information in the entries in the partition table(s),
which set should I use: 79656/16/63 or 4998/255/63?

Anyway, I think you've started me down the right trail. All I need
to do now is figure out what the _right_ data should be, _where_
it should go, and how to apply it without doing multiple 'dd'
writes to assorted critical sectors of a live HD. <um, grin?>


Frank McKenney, McKenney Associates
Richmond, Virginia / (804) 320-4887
Munged E-mail: frank uscore mckenney ayut minds pring dawt cahm (y'all)

 

[Frnak McKenney]

Douglas,

Three replies by people who know how to trim a reply... and _do_
it? Can this really be Usenet??? <grin!>

Thanks for answering.

--much disk info snipped--
<snip>

Wow, you are more patient than me to retype all of that information.

I confess that I _really_ hate doing it. Unfortunately, when one is
working on boot-type problems and trap screens it's really hard to
capture the output/screen information through any other mechanism.
OS/2 Standalone Dump would capture a TRAP screen and write it to the
dump floppies, but that was the last time I saw that particular
feature. (Ack! Phlbbbbt!)

I see the response by Robert Hull, and I agree with it. LBA mode is
important and maybe you weren't using it for some reason. Get your data
first, then start over with that disc. You might look for another disc-
since discs are cheap. That may be an easier way to start over.

I have a spare, but if this is an LBA-related problem (as it appears
to be) I could waste six or eight hours reinstalling Win2K, .NET,
assorted compilers and printer drivers and applications, and then
another three hours installing and tweaking SuSE only to be right back
in the same position. I'll add your suggestion to my list, but please

Unsolicited advice...
If you are going to keep windows, put it on partition 1 using NTFS. You
probably had a reason for the FAT12 partition, but my advice is to
give Windows as little reason to bitch as possible. Maybe it will be
happier on partition 1 on a disc using LBA addressing. Then maybe not,
the little bitch that it is.

I don't _believe_ I created that one; I _think_ it's the result of my
asking that Win2K be installed on an extended partition. Or maybe I
created it based on the Win2K install suggestions.

What's odd is that this drive is the one-and-only I've had in this box,
so LBA _should_ have been enabled from the very beginning, and the Win2K
installation software _should_ have used it.

Ah, well. SuSE 10.1 should be out shortly, giving me a whole _new_ set
of Things That Can Go Wrong. Maybe I'll just do this year's taxes on
something that's simpler and more reliable, like my trusty 4.77MHz
dual-floppy Gen-you-wine Eye-Bee-Emm Pee-Cee... no, I sent that down
to a niece in Marietta a few years back. Oh, well.

Thanks for the suggestions.


Frank McKenney, McKenney Associates
Richmond, Virginia / (804) 320-4887
Munged E-mail: frank uscore mckenney ayut minds pring dawt cahm (y'all)
--
I have come to believe, not in fate, but in forks in the road,
choices we make with no knowledge of the outcome. You step off
the curb in one direction, you get run over. You step off in the
other direction and and you meet someone who changes your life.
-- Baxter Black / Hey, Cowgirl, Need a Ride?
--

 

[Douglas Mayne]

Douglas,

Three replies by people who know how to trim a reply... and _do_
it? Can this really be Usenet??? <grin!>

Thanks for answering.

I have a spare, but if this is an LBA-related problem (as it appears
to be) I could waste six or eight hours reinstalling Win2K, .NET,
assorted compilers and printer drivers and applications, and then
another three hours installing and tweaking SuSE only to be right back
in the same position. I'll add your suggestion to my list, but please
don't be offended if I stick it 'way down at the bottom. <grin>

Can you back your Windows and then restore?

I don't _believe_ I created that one; I _think_ it's the result of my
asking that Win2K be installed on an extended partition. Or maybe I
created it based on the Win2K install suggestions.

What's odd is that this drive is the one-and-only I've had in this box,
so LBA _should_ have been enabled from the very beginning, and the Win2K
installation software _should_ have used it.

Ah, well. SuSE 10.1 should be out shortly, giving me a whole _new_ set
of Things That Can Go Wrong. Maybe I'll just do this year's taxes on
something that's simpler and more reliable, like my trusty 4.77MHz
dual-floppy Gen-you-wine Eye-Bee-Emm Pee-Cee... no, I sent that down
to a niece in Marietta a few years back. Oh, well.

OT: I have a Compaq Portable Plus, with 5-1/4 floppy and 10M HD. It won't
run GNU/Linux because it's a pre-386 (with 640k total memory).

Thanks for the suggestions.

<snip>
I think LBA mode is set in the BIOS. I am not too clear on these
translation modes between C/H/S and LBA. I do think that a lot of trouble
can be avoided by using LBA mode from the start. I set the BIOS to use
LBA, then initialize the partition table with GNU/Linux's fdisk. Then, I
can install Windows into the partition(s) that have been assigned. Also,
instead of running Windows setup, I sometimes restore a windows backup to
the target partition using ntfsclone from GNU/Linux.

Speaking of taxes, I also make sure I have a valid Windows license for
every system I setup using ntfsclone/sysprep. Now, I've paid my share of
those taxes and I'm looking to stop. It's past time to stop working on
Bill's house and work on the community property. I'd rather own than rent.

 

[Frnak McKenney]

I'm running again!

I've successfully booted MSWin2K three times and SuSE Linux 9.1
twice. I'm going to take that as a Good Sign. <grin!>

After a lot of research on the problem I finally realized that I'd
absorbed about as much technical detail as I was going to, so I made
yet-another copy of mongoose's MBR and partition table:

dd bs=512 count=1 if=/dev/hda of=/media/floppy/hda-mbr.img
sfdisk -d /dev/hda >/media/floppy/hda-part.txt
fdisk -l -u /dev/hda >/media/floppy/hda-fdlu.txt

I then rebooted via Grub from /dev/hda and added the following
kernel parameter:

hda=4998,255,63

On my second try I managed to type the right parameter at the right
point in the boot process and when Linux came up, the kernel was
using the 'correct' C/H/S values.

I'd run out of ways of avoiding it, so I did the following from a
root console:

sfdisk -d /dev/hda >pre_sfdisk_partitions.txt
cat pre_sfdisk_partitions.txt | sfdisk -H255 --no-reread /dev/dha

The second command failed, but adding a --force parameter let it
complete. I shut Linux down and rebooted it.

Now, even without the 'hdc=xxx' kernel parameter everyone agreed
that /dev/hda was using 255 "heads" -- another Good Sign.

Shutdown... but this time I chose 'Windows' from my Grub menu.
Ack! A black screen again!

No... Wait! _This_ one says "Starting Windows" across the bottom
of it, and there's a _lot_ less hex!

Win2K finished coming up, discovered New Hardware (I have an unused
RAID option on the K8N-E Deluxe), and sqlmangr.exe complained that
it couldn't find Ordinal 29 in 'odbcbcp.dll'. I fed the New
Hardware Whizzer my ASUS CD, it found a driver, and then it decided
not to use it (S'fine with me, I'm not using the RAID hardware
either <grin!>).

Two more boots convinced me that (a) Win2K is booting consistently,
(b) Pinball's flippers on my SuperDuper Hi-Speed ComPewTer are as
sluggish and nearly un-playable as ever, and (c) sqlmangr.exe would
just have to live without it's favorite Ordinal until tomorrow
(isn't there a College of Ordinals in Rome?).

So... end of posting. My thanks to everyone who contributed, and
my deepest sympathies to all those whose misery from dealing with
this problem produced such detailed coverage of its inner workings
and fix(es). This one hit SuSE users, RedHat users, and people
running a few other distributions I've forgotten the names and
versions for.

For the next person running into this problem, I'll offer a few
places to look for information over and above the original page
suggested by Robert Hull (Warning: URLs are manually wrapped):

SDB:Windows No Longer Boots Following the Installation
of SUSE LINUX 9.1
http://en.opensuse.org/SDB:Windows_No_Longer_Boots_Following_the
_Installation_of_SUSE_LINUX_9.1

http://lwn.net/Articles/86835/?format=printable
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=115980
http://www.win.tue.nl/~aeb/partitions/partition_types-2.html
http://www.linuxquestions.org/questions/showthread.php?
t=243931&page=3

SuSE 10.1 should be going production in the next few weeks. Gosh...
I could back everything up, wipe the partition table clean, start
over, and... run into a different bug? Noooooooooooo........

G'night everyone, and Good Luck. <grin-fer-real!>
Tomorrow I have to go get some RealWork(tm) done.


Frank McKenney, McKenney Associates
Richmond, Virginia / (804) 320-4887
Munged E-mail: frank uscore mckenney ayut minds pring dawt cahm (y'all)

 

[Frnak McKenney]

Can you back your Windows and then restore?

To be honest, I've made backup CDs of most of my user files, but I'm
not sure I could make a complete, restore-from-scratch copy of a
8Gb-used-out-of-10Gb Win2K partition even if I _had_ been able to
boot Win2K.

Fortunately Murphy was on vacation this afternoon, and I successfully
ZAPped my partition table into some semblance of usability. I admit
that, though the details vary, I seem to have been doing stuff like
this for a couple of _decades_ now... anyone _else_ willing to admit

OT: I have a Compaq Portable Plus, with 5-1/4 floppy and 10M HD. It won't
run GNU/Linux because it's a pre-386 (with 640k total memory).

Wow! Did you know that you could back up that entire drive with only
_thirty_ floppy disks? (IBM considered this a _selling_ point for the

<snip>
I think LBA mode is set in the BIOS. I am not too clear on these
translation modes between C/H/S and LBA. I do think that a lot of trouble
can be avoided by using LBA mode from the start. I set the BIOS to use
LBA, then initialize the partition table with GNU/Linux's fdisk. Then, I
can install Windows into the partition(s) that have been assigned. Also,
instead of running Windows setup, I sometimes restore a windows backup to
the target partition using ntfsclone from GNU/Linux.

Wish I could have tweaked my BIOS re LBA, but on this motherboard my only
choices are [Auto] and [Disabled]. The root cause, if I read all those
postings correctly, was that the version of the GNU 'parted' partition
editor that shipped with RedHat FC2, SuSE 9.1, and some other packages
had a bug and made changes that it 'shouldn't have orter'.

Speaking of taxes, I also make sure I have a valid Windows license for
every system I setup using ntfsclone/sysprep. Now, I've paid my share of
those taxes and I'm looking to stop. It's past time to stop working on
Bill's house and work on the community property. I'd rather own than rent.

I've got SQL-Ledger installed on 'manticore', and I'm working on a
better GL Account Detail report, but I don't think I'll have it ready
for this year's Ides'o'April. I don't know why it should be true, but I
find SQL-Ledger's interface much more... "natural"? than I found
GnuCash.

Thanks again for the help.


Frank McKenney, McKenney Associates
Richmond, Virginia / (804) 320-4887
Munged E-mail: frank uscore mckenney ayut minds pring dawt cahm (y'all)