Off topic Old P-III Box will boot bootable from floppy, but NOT MS-DOS

[IBMuser]

I am flummoxed. Don’t know if I had some kind of finger problem configuring the BIOS or what but an old P-III 800 MHz box of mine was running a mix of DOS, Windows 2000, and OS/2 worked for for years. I messed something up on the disk so I am in the process of rebuilding OS/2. As I am trying this that and the other thing all of a sudden I could not boot my 3.5 inch DOS rescue floppy.

I thought, after putting dozens of floppies through it, running SpinRite to refresh the surfaces, I wondered if that floppy had finally worn out.

Does not appear to be the case. Surprise surprise, I put BOOTABLE on a floppy and It boots just fine. I even have an x86 version of MemTest that boots fine so I let it run to confirm that the system ram is also error free. SpinRite from the BOOTABLE floppy with freeDOS also boots fine. Then I get to MS-DOS and all I get is the initial message “starting MS – DOS“ and that’s it. Dead stop. That same floppy that will not continue booting on this troublesome machine boots everywhere else just fine on all sorts of other retro hardware in my shack: whether it’s Pentium two or Pentium four the stand alone, MS-DOS floppy boots like a charm.

It’s just this P-III that all of a sudden has a problem. Anything spring to mind with this box that uses the code name “Whitney” and is driven by an American mega trend IBM BIOS 460024e version 1.03. MS- DOS has butut successfully from this machine in the past. Both from floppy and the SCSI hard drive on board. Is this

 

[coffeeprogrammer]

Dude, that does not read clearly. Sometimes I write like that, I think I write like that because I want some sort of emotional release. I would recommend you use something like google docs that has free voice to text functionality and perhaps make a list of everything you want to say and then talk rather than type. You might then consider using a text to speech service to then read it back to you to make sure it is clear communications. Beyond that, I believe, at least in my case, that I tend to sometime write like that because I have a lot of time, but most everything feels meaningless and I want to find something to feels meaningful. Just remember to not spend too much time on the hamster wheel. Good luck.

Chad

 

[AlanD]

SpinRite from the BOOTABLE floppy with freeDOS also boots fine. Then I get to MS-DOS and all I get is the initial message “starting MS – DOS“ and that’s it. Dead stop.

IIRC Steve came across some issues with the initial drive identification/validation on some machines. That is why SR comes with a customised version of FreeDOS. I think the issue was a time-out when a disk controller does not respond as expected.

That could explain why the Bootable floppy works but MS-DOS does not.

 

[AARC]

I can't add much to what AlanD said. We still have Point4 Mini computers here and the timing issue is a big deal as the older stuff is hard-coded to
only have a strict window.

As for OS/2 - what a flashback. We used Warp 3 and then 4 for years and it was so far ahead of anything M$ that it was staggering. It still lives on,
sort of, but is now meshed with Linux, another thing that M$ could never equal.

 

[IBMuser]

Dude, that does not read clearly. Sometimes I write like that, I think I write like that because I want some sort of emotional release. I would recommend you use something like google docs that has free voice to text functionality and perhaps make a list of everything you want to say and then talk rather than type. You might then consider using a text to speech service to then read it back to you to make sure it is clear communications. Beyond that, I believe, at least in my case, that I tend to sometime write like that because I have a lot of time, but most everything feels meaningless and I want to find something to feels meaningful. Just remember to not spend too much time on the hamster wheel. Good luck.

Chad

Well thanks for the helpful tip. I`ll just file this under ``doesn`t know the time but gives lecture on how to build a clock``

 

[IBMuser]

I can't add much to what AlanD said. We still have Point4 Mini computers here and the timing issue is a big deal as the older stuff is hard-coded to
only have a strict window.

As for OS/2 - what a flashback. We used Warp 3 and then 4 for years and it was so far ahead of anything M$ that it was staggering. It still lives on,
sort of, but is now meshed with Linux, another thing that M$ could never equal.

Well, this one's been a journey but, after ruling some things out one-by-one; success!

This all started because of a finger problem on my part - I took some Powerquest Drive Image backups as I have done many times in the past but stupidly let Drive image modify the partition table to show the Extended partition as type X'OF', an extended "X" partition. Not good. Yes a standard extended partition (X'05') for logical drives is improper as hell for any allocation in drive space beyond 1023 cylinders, 255 heads and 63 sectors, but it is what it is. The old OSes didn't mind. PowerQuest Drive Image did.

When I rebuilt the drive space and after restoring Boot Manager up front, the MS-DOS primary partition and all the logical drives in the Extended Partition (X'05') I went for re-boot. Nothing. Stopped dead. No errors. The red herring was that even an MS-DOS boot from floppy failed!!! Why a floppy boot fail for a hard drive problem?

A floppy boot of "Bootable" and the FreeDos beneath it was fine. I could see all the primary and logical drives. A FreeDos version of FDISK showed all the allocations as I intended including the non-dos Extended Partition of some 16Gb (just a small 18Gb LVD Ultra Wide SCSI) and the logical drives within.

I was sure of the integrity of my boot floppies. I started to suspect that MS-DOS was particular about accounting for the drives in the environment and could not get past reading something. Maybe FreeDos was either unaware of it or capable of accommodating it.

Now we start drilling down bit by bit. Even in a huge extended partition the MBR based partition table only writes the boundary to a maximum 1023-255-63 CHS as a place holder though the sector count vouches for the entire space.

Couldn't remember exactly how I built this in the past so I started with guarding the 1024 cylinder boundary; about 8.3Gb of this 18Gb hot little pistol of a 10,000 RPM SCSI drive.

With no logical drive allocated above cylinder 1024, everything booted fine; hard drive, MS-DOS from floppy, and FreeDos/Bootable which always booted under all conditions. The Extended Partiton was always defined right up to the end of the drive - on this drive that's CHS 2233-255-63. But to see the logical drives below the 8.3Gb line (cyl 1024) the extended partition had to stay as X'05'. If set as extendedX (X'0F') BootManager, MS-DOS via BootManager and floppy boot MS-DOS would work but any of the logical drives were not visible. Extended Partition HAD to stay marked as X'05'

I was fixated on protecting the 1024 cylinder boundary. No allocations touching it.

That's what got me. If logical drives stayed under cylinder 1024, no problem. As soon as anything was allocated to cylinder 1024, head 1, sector 1 or above; problem.

I remembered a little nugget about OS/2 booting out of a logical partition (which is what BootManager helps to do). The OS/2 logical partition had to stay below the 1024 cylinder boundary. Or at least the system files within the logical partition had to be below cylinder 1024. Data? Didn't matter. My OS/2 data partition lives happily at the end of the drive.

Here I am guarding cylinder 1024. No allocation breaking the line. But what if I built a logical partition that "straddled" the 1024 cylinder boundary? So I adjusted the OS/2 logical partition so the last 17Mb crossed the 1024 cylinder boundary. It could have been any logical partition. It just happened to by my HPFS-formatted OS/2 system partition at this location in my allocation scheme.


J-A-C-K-P-O-T ! ! !

Everything boots. All allocations above and below cylinder 1024 work fine. Windows 2000 system and data partitions are both above cylinder 1024 and working as they always did, and my HPFS data partition for OS/2 at the end.

That any of this worked before was probably a happy coincidence in how I did the additional mapping out of the partitions. So lots of notes in my files should I need to visit this again. Old software on large drive geometry - not for the squeamish.

 

[AARC]

J-A-C-K-P-O-T ! ! !

Everything boots. All allocations above and below cylinder 1024 work fine. Windows 2000 system and data partitions are both above cylinder 1024 and working as they always did, and my HPFS data partition for OS/2 at the end.

That any of this worked before was probably a happy coincidence in how I did the additional mapping out of the partitions. So lots of notes in my files should I need to visit this again. Old software on large drive geometry - not for the squeamish.

A very detailed report, thanks. Glad you were able to resolve it, but yes, not for the squeamish.