Victor 9000 / Sirius 1 Hard Disk Emulation

[1302L]

Hi djg,

here we go with a fresh thread,

Didn't print too many lines. Did you get a directory or did you get an error?

I got a directory.

I think you showed DIR worked previously.

Yes

Also did you previously get X98 (Bad Hard Disk Label) when trying to boot or did you get some other error?

I was getting the X98 most of the time (Sometimes X06 [Track 00 not found] as well), now it is solely the X98

Where did the emulator image you are using come from?

Downloaded the from your website, being curious, are there other sources for V9K HDD images?

This thread is mixing several things so getting unwieldy to track. We may want to start a separate thread.

Yes!

Also the files you attached only have some of the emulator output. Did you use the script command which should have had all or did you capture some other way? If some other way is this all the output related to the commands you ran?

I used screenshots for the shorter ones and copied the output into a text file for the longer ones, I copied just the parts which had been displayed after I executed the command. Do you wish to have the entire output from logging into the emulator? I thought it would be more clearly arranged this way and jut related to the command I was typing.

Bad thing is the low seek times went away so debug output doesn't show what I was hoping to see. If you can find any way of triggering the low seek times that's what I was hoping to see.

Yes, I noticed this too, I will try to trigger the low seek times.

Did the L command trying to read single sector work for any sector address? Did it work with the previous software version? What error is reported if it doesn't work?

I tried this several times, no output by the emulator on putty and no reaction o the machine; but there might be an issue as I have to execute DEBUG from Floppy, does the machine try to read the HDD sectors anyway or it just seeking on the floppy?

Thanks and best regards,

Martin.

 

[djg]

(trying to boot)

I was getting the X98 most of the time (Sometimes X06 [Track 00 not found] as well), now it is solely the X98

Think side effect of change is it improved seeking.

Question now is it not reading disk label correctly or is it invalid since something is different about your machine so it doesn't like the label from the other machine. If you haven't I would first try putting a new copy of the emulator image in case previous problems corrupted it. If you get the same problem can you try reformatting the disk. It didn't work previously but would be interesting to see if it does now.

Can you run the HDFIELD diagnostic SHOWSTAT command that should display the disk label?

I tried this several times, no output by the emulator on putty and no reaction o the machine; but there might be an issue as I have to execute DEBUG from Floppy, does the machine try to read the HDD sectors anyway or it just seeking on the floppy?

I think when you boot from the floppy you can still get a directory of the HDD. Debug with the correct command should then read the HDD if the correct disk is specified. If it was reading the floppy you should see the drive light if your machine has one.

By no reaction on the machine are you saying you just get another debug prompt or something else? Exactly what commands are you entering?

Also try with debug
e 100 55
l 100 0 1 1
d 100
Post the output

Downloaded the from your website, being curious, are there other sources for V9K HDD images?

Not that I know of. There are two images in the zip file. Which one are you trying to boot?

 

[1302L]

Hi djg,

sorry for my late reply, I was on business travel.

Think side effect of change is it improved seeking.

Question now is it not reading disk label correctly or is it invalid since something is different about your machine so it doesn't like the label from the other machine. If you haven't I would first try putting a new copy of the emulator image in case previous problems corrupted it. If you get the same problem can you try reformatting the disk. It didn't work previously but would be interesting to see if it does now.

Can you run the HDFIELD diagnostic SHOWSTAT command that should display the disk label?

First Thing I did was running SHOWSTAT (Picture 3611&3612) and HDDISK,(Picture 3613) afterwards I tried to use FACTORYF Super for formatting the drive (Picture 3614&3615), please see Screenshots attached.

I will now continue to use the commands you posted by running DEBUG and post the output of the machine and the log file from Putty. I will run HDDIAG as well.

At this trial I was running the 'victor-msdos3-clean-emulation' file

Best,

Martin.

 

[Gary C]

Very interested in the outcome of this.

Would be good at some point to have a summary of hardware, software etc for others to replicate.

Mind you, mine has developed yet another floppy drive problem so need to fix that first.

 

[djg]

My best guess for the disk label is your machine wants a slightly different format for the disk label than pdevine machine uses so the invalid label error is not an emulator problem. The data displayed is mostly make sense and matches what I see in the image for some of the obvious fields so its reading the sector. Some fields do seem suspicious such as available media list region 1 starting address 0 length 0.

pdevine, did you ever have time to check version? Also are you using a rev C board? "Just re-reading and realized I missed the request for the version of mfm emulator I'm running. I haven't been good about keeping current, so it's at least 1.5 years out of date. I'm out of town for a couple days but I can check on Monday when I get back." It would be good if you could do the showstat on the 'victor-msdos3-clean-emulation' file. Also if their is a was to determine ROM/BIOS revision that would be interesting to compare. I suspect its differences between machine and not emulator SW version but still good to check.

If we can get HDDISK format working then we can see what label your machine creates. Do you have a scope/logic analyzer to watch what some of the signals are doing when it generates those errors? If so I can tell you what to look at.

Also someone else with a different controller found running from ram helped with their machine. Symptoms are different so not the highest probability but you can try

mount -t tmpfs -o size=150M tmpfs /mnt
cp victor-msdos3-clean-emulation /mnt
mfm_emu --file /mnt/victor-msdos3-clean-emulation ...

Then see if HDDISK or anything else changes.

Also

mfm_emu --file /mnt/victor-msdos3-clean-emulation --drive 0
This forces the drive to always be selected so may help HDDISK. Don't think its going to change anything else. Let me know if any changes any symptoms.

Let me know if I already had you do these.

 

[1302L]

Hi djg,

I tried the following, but it did not work, would you please advise where's my mistake?

Also someone else with a different controller found running from ram helped with their machine. Symptoms are different so not the highest probability but you can try

mount -t tmpfs -o size=150M tmpfs /mnt
cp victor-msdos3-clean-emulation /mnt
mfm_emu --file /mnt/victor-msdos3-clean-emulation ...

Then see if HDDISK or anything else changes.

Also

mfm_emu --file /mnt/victor-msdos3-clean-emulation --drive 0

Best, Martin.

 

[1302L]

Subsequently I used the old command to start the emulation, I don't know if this makes sense. (emufile_a is victor-msdos3-clean-emulation)
Running HDDISK and SHOWSTAT - no changes at all, but there are very short seek times again.
In SHOWSTAT there is the possibility to dump the record, please see screenshots attached.
Please see emu log file attached
Thanks!
Martin.

 

[djg]

I tried the following, but it did not work, would you please advise where's my mistake?

You only needed sudo on the mount command. Don't use it on the cp. There are advantages to the older image where you ran everything as root but that's frowned on now.

For the normal operation specify what you normally use which I think is --drive 1. The ... was put the rest of your normal options here.

Looks like your dump of the disk label matches the image so its reading the sector fine.
0007000 0200 0100 736f 6c69 6400 0000 0000 0000
0007020 0000 0000 0002 3d00 0000 0000 0c13 0000
0007040 0000 0000 02f0 0800 8000 800b 0705 0000
0007060 7f17 0000 0100 0000 006f 8f01 0001 0000
0007100 0000 6f8f 0100 0202 0000 004e 8d01 004e
0007120 8d01 0000 0000 0000 0000 0000 0000 0000
0007140 0000 0000 0000 0000 0000 0000 0000 0000

Running HDDISK and SHOWSTAT - no changes at all, but there are very short seek times again.

The 0 ms is expected. The other isn't and the trace data doesn't make sense. It didn't go through any code path that would have sent the ARM an interrupt to trigger the second print. Since I didn't see a setup_emu in the log file had you already run setup_emu and mfm_emu during this boot? Its like it had an interrupt left over from the last run but I haven't been able to figure out how to cause that. Will need to look further. If this behavior repeatable?

 

[pdevine]

Ok, so I finally had some time to see if I could help figure out what's going on. In the process I think I may have eliminated two potential problems @1302L is having. So the Victor 9000 that I have the HD connected to is a dual floppy unit. I have the HD as drive A, which is standard for a Victor, and then I have the floppies assigned to drives B and C. The standard HD units for a Victor and ACT Sirius 1's have 1 floppy; the HD gets assigned to drive A and the floppy to drive B. I believe @1302L might be having trouble because he only has 1 floppy and the image is configured to expect a second drive. The other potential issue is the earlier image we've been using is absurdly large by the victor standards. I did a lot of experimentation to see how big I could get away with, playing with various cylinder and head combinations to see how big I could get the drive. I finally arrived as 256MB, which for MS-DOS 3.1 is crazy big. I'm worried that maybe an earlier boot ROM version is blowing up on how big that earlier image is. All the docs are excited to support 30MB drives.

In any event, to gather the information that @djg was looking for, I do have a revision C board. And the version I've been running is:
root@beaglebone:~/emu# ./mfm_emu -v
Board revision C detected
Version 2.16
Program requires options: file drive

The OS I'm using is
Linux beaglebone 3.8.13-bone84 #1 SMP Sun Feb 12 02:54:13 UTC 2017 armv7l GNU/Linux
I'm not sure if I need to upgrade anything at the OS level.

So to help us diagnose the issue, tonight I upgraded my mfm to:
root@beaglebone:~/git2/mfm/emu# ./mfm_emu --version
Board revision C detected
Version 4.14
Program requires options: file drive

To further help diagnose the issue I decided to recreate an image based on a Seagate ST412 parameters. The low level format program has an ST412 as a pre-defined choice so it seemed safer to troubleshoot with.

After upgrading the mfm_emu vesrion I created a new image:
./mfm_emu --drive 1 --file /mnt/st412_a --initialize --cylinders 306 --heads 4

I then rebooted the machine to a floppy-only MS-DOS. You can't use the HD version as it gets confused by the unformatted drive. I used the UK DOS 2.1 Floppy edition as I had the disk handy. I then ran FactoryF.exe, the low-level format program for the Victor, and I selected the F5, ST412 configuration option.

I then rebooted the machine with US MS-DOS 3.1, which includes floppy support. It's important for a machine with 2 drives to boot from the right-hand drive as at a certain point in the boot process the HD takes over the left-hand drive spot. I then ran hdsetup to create the equivalent of the bios parameter block. I configured the HD as drive A using the entire disk, and assigned the right floppy to drive B. I did not assign a second drive to prevent any issues on a machine that doesn't have one. The HD version of the Victor only has a right-hand drive and is traditionally assigned to drive B.

I then rebooted the machine again to take those drive parameters, again with the US MS-DOS 3.1 version in the right-hand drive. Afterwards I copied the OS over to the disk with sys b: a: and then copied the remainder of the boot disk I use that has some additional utilities like Kermit and pkzip. Then I rebooted the machine from the new hard drive.

The attached image is the result of the above. You can use it like:
./mfm_emu --drive 1 --file /mnt/st412_a
You don't need the --initialize after the image has been created. Also, I think you wanted a German version? If so, let me know and I can dig one up. I went with the American version as I had the disks handy.

 

[djg]

Ok, so I finally had some time to see if I could help figure out what's going on.

Thanks. Hopefully this will all work.

I would switch back to the normal software for initial testing.
cd /opt/mfm
git checkout master
make

Setup to capture messages with script command or whatever method you have been using
Then run mfm_emu with normal options.

See if the new image boots
See if you can create a new image following the procedure pdevine used

Look at the log file and see if we got any low seek times other than the expected 0.
If you captured it to a file on the beaglebone this should give you less messages to look through
grep 'seek time' logfile | sort | uniq

 

[1302L]

Dear Paul and djg,

many thanks for your posts and your effort!

Tonight I wanted to have a quick go on @pdevine 's new HDD image.

I was able to start the emulation, but my Floppy Drive played up and accepted no Disk anymore, except a Floppy-only DOS 1.25.

But nevertheless I gave the emulation a short try, you can see the outcome on the log file attached. Please not, no changes to the software done yet.

I still have to execute the debug commands before I can start the emulation:

debian@BeagleBone:~$ sudo systemctl --system daemon-reload
debian@BeagleBone:~$ sudo systemctl restart mfm_emu.service
debian@BeagleBone:~$ sudo systemctl disable mfm_emu.service

Otherwise I'll get the message that J2 might be reversed, but this is definitely not the case.

After I executed the commands above I started the emulation by

debian@BeagleBone:~$ mfm_emu --drive 1 --file st412_a

I had this floppy issue once more, as I created new floppies it worked again, but this leads to me to the question if something on my floppy drive is faulty.

I have the HD as drive A, which is standard for a Victor, and then I have the floppies assigned to drives B and C. The standard HD units for a Victor and ACT Sirius 1's have 1 floppy; the HD gets assigned to drive A and the floppy to drive B. I believe @1302L might be having trouble because he only has 1 floppy and the image is configured to expect a second drive.

My HD is also drive A, my (only) Floppy drive is drive C, when emulation is running and I booted the Victor with DOS 3.1. Drive B is existent, but cannot be accessed. When I bot a Floppy-ony Version, i.e, DOS 1.25, then the Floppy is drive B, drive A is existent, but cannot be accessed. So, nomenclature-wise it should be OK.

I finally arrived as 256MB, which for MS-DOS 3.1

That's incredible!

That's it for now. It's Valentine's day and I have to cook my wife her favorite meal!

Best,

Martin.

 

[djg]

Otherwise I'll get the message that J2 might be reversed, but this is definitely not the case.

When running manually you need to run setup_emu once before running mfm_emu. Then you won't need the systemctl commands.
http://www.pdp8online.com/mfm/revb/revb_board_build.shtml

Didn't understand if what you were able to do with emulation did or didn't work.

Hope you had a good meal.

 

[1302L]

Hi djg,

When running manually you need to run setup_emu once before running mfm_emu. Then you won't need the systemctl commands.
http://www.pdp8online.com/mfm/revb/revb_board_build.shtml

Sounds trivial, but, could this be the root cause?

Didn't understand if what you were able to do with emulation did or didn't work.

The emulation started as usual (looking on putty), but the machine did not boot, as posted above I wasn't able to boot from Floppy as my Disks have become faulty and I was getting boot errors. I will re-write the disks asap. On the HDD boot trial this I've got the HDD error code X12 (Address mark not found).The first ~10 columns of the log file after the initialization are showing the boot trial, down to the blank section.

Best,

Martin.

 

[pdevine]

@1302L My first guess about your floppy drives would be that they need to be cleaned. Do you have a cleaning disk? If not, some isopropyl alcohol with a cotton swab would do the trick.

 

[1302L]

@1302L My first guess about your floppy drives would be that they need to be cleaned. Do you have a cleaning disk? If not, some isopropyl alcohol with a cotton swab would do the trick.

Hi Paul,
as I disassembled the machine roundabout two months ago, I gave the drive a thorough cleanup with isopropyl and some thin cotton pads, can this be possible that I have to do this again after such a short time? I admit, the drive was working very well and much smoother and quieter after the cleanup, now it is nearly as loud as it was before.
I do not have a cleaning disk (yet). But I I bought one now as this is much more comfortable than disassembling the unit.
I read about head adjustments in the service manual, might this be necessary?
Thanks and best regards,
Martin.

 

[pdevine]

Hi Paul,
as I disassembled the machine roundabout two months ago, I gave the drive a thorough cleanup with isopropyl and some thin cotton pads, can this be possible that I have to do this again after such a short time? I admit, the drive was working very well and much smoother and quieter after the cleanup, now it is nearly as loud as it was before.

I doubt you need a head alignment. These days the disks we're using are 40 years old. The disks themselves are fragile and just reading them is depositing parts of the binding material on the heads. The drives get dirty quickly due to the age of the disks. I think cleaning again will likely help you out. I've also found the tiniest drop of cyclomethicone on a disk helps lubricate the disk and drive head in a way that reduces the squealing you're describing and helps keep the disks moving more smoothly, and from dirtying up the drive head so quickly.

 

[1302L]

Thanks,
of course you're right, I am just using new and (when I bought them) sealed disks, but nevertheless these are at least 30 years old...
- Is cyclomethicone a kind of silicone oil?
Do I apply this on the cleaning disk?
Tonight I will clean the heads and write the disks again, hope this helps.
Cheers,
Martin.

 

[djg]

- Is cyclomethicone a kind of silicone oil?"

"it's a mix of specific chain length (4 to 7) cyclic structured silicone molecules.". Its a liquid that evaporates and does feel oily. Frequently used in skin care products to give silky feel.

It helps to read deteriorating disks but haven't heard of people using it long term. Since it evaporates you would have to keep reapplying.
pdevine, have you had success using it long term and how often did you have to reapply? Or was this preventative when you were using a disk but didn't reuse it often?

In general if a disk is squealing in a clean drive it should be discarded since the binder is breaking down. It will leave deposits on the head that can then damage the next disk. If all your disks are squealing and they aren't all the same brand then something else may be going on. I assume drive doesn't squeal without a disk such as from bad bearings. The deterioration is brand and batch dependent. Some of my disks are fine. One sealed box I picked up all the disks shed oxide and couldn't format one.

 

[1302L]

Hi,

after creating some new disks, I was able to access pdevine's new image, same story, I can execute DIR, but if I want to execute a program on the HD emulation, I am getting the same error as before.
Running SHOWSTAT was possible, please see Screenshots and Log file attached.
No super - short seek times anymore.

Running several tests:
DMA: OK
Controller: NOK - Error- Address mark not found, poss. failures are: Radial cable, Drive, Controller
RAM: 384 kB OK
CPU: OK

I used HDFORMAT: Format was stated successful by the machine, but no data was erased, pdevines's image still intact and no data lost. What could that be?

Regarding the drive noises.

What I meant was the buzz when the drive reads or writes. This was much quieter after I cleaned the drive. Now with the new disks written, the drive is quiet again. Is the loud buzz coming from reading faults on the floppy disk?
I do not have a squeaking noise as I lubricated the drive very well.

Best,

Martin.

 

[pdevine]

I am actually not familiar with the hdformat program. I've always used factoryf.exe followed by hdsetup.exe or autoset. I have a Plus5 manual that talks about using hdformat to format their drives, but that's a totally different controller. Have you tried running hdsetup on the drive?