PFS3-AIO problems

[-Acid-]

I turned my A4000 and it loaded WB, double clicked on a disk icon to load something and it froze up. Now on reboot I get these errors as can be seen on the attached pictures.

If I immediately go to the ESM I can see all partitions listed but cannot boot whatsoever. I decided to whip the drive out and put it in my PC and try it in WinUAE. When I try that WinUAE sits on a black screen before it stops responding and then it took a good 5 minutes to end the process using task manager. Then when I reloaded WinUAE and go to add the drive again is it gone from the list and it doesn't even show up in CP>Admin Tools>Disk Management anymore (will reboot in a minute to see if it helps).

Any ideas how to repair this? the drive is a huge 160gb with about 60gb of data on it so I do not want to lose it. Partitions are set up 2gb x 2 and 51gb x 3.

[dlfrsilver]

what was the application you launched before this happened ?

[Toni Wilen]

Unfortunately it sounds like dying drive.

Drive access becoming really slow in Windows points to drive or driver problem. It is normal that program that accesses the disk becomes (temporarily) unkillable if disk access call is stuck in kernel mode driver (which is waiting for disk to respond).

[-Acid-]

Originally Posted by dlfrsilver

what was the application you launched before this happened ?

On the Amiga nothing, I was double clicking on the Work disk icon to open it on the WB screen.

When I restart the PC WinUAE sees the drive again but has went into meltdown again when I try to start up the emulation. It's not making any odd noises or anything along those lines either.

Can anyone suggest any recovery action I might attempt?

[emufan]

on the linux platform you could use the dd command - this will (try) to backup the whole hdd or just a partition
and writes it into file. eg: dd if=/dev/sda1 of=amihdd.img
this would save the first partition of the first hardrive into the file amihdd.img
if (in-file) is the source harddrive/partition. of (out-file) is the destination, in this case a file.

it is a bit confusing if you will do it for the first time.
dd ignores the filesystem of the partition and just writes raw data.
other partition tools will offten only works with filesystems known to them.

i'm not sure how winuae will handle the "amihdd.img" file, but it is worth a try,
if everything else fails. can anyone say if the rescued partion can be used with winuae?

there are different small linux distros out there, I think dd command comes with all of them.
i was using DamnSmallLinux in the past but i think there are more modern out there.

here a link which could work as a start:
Drive and partition backups with dd

[thomas]

I agree with Toni, the drive probably is beyond repair.

If you are lucky you can manage to create an image backup. But you need a program which does not stop on read errors. I am not sure if dd can do this.

TSGUI on AmigaOS can. But I never tested it on such a huge drive. Not to mention that Amiga file systems have a file size limit of 2GB. You'll have to write the image to a network drive or use WinUAE with a mounted NTFS directory.

http://thomas-rapp.homepage.t-online...oads/tsgui.lha

[emufan]

dd command has a "noerror" option, which let it continue on read errors.
just to collect some usefull websites/info - two more links:
dd - man page
How to use dd command.
if there is an amiga based solution, of course try that first.

[idrougge]

For reading damaged disks, "ddrescue" is a better alternative than pure "dd".

[Jope]

Also try dd_rescue (it is a different program) and dcfldd..

[utri007]

Is pfsdoctor availlable somewhere and does it still work with current version?

[emufan]

Originally Posted by utri007

Is pfsdoctor availlable somewhere and does it still work with current version?

it is in the PFS3_v5.3 archive on aminet. no idea if there is another version around.

btw: on Thomas website there is PFSsalv v2.2 which one can use to rescue files from a pfs partition.

[-Acid-]

Thanks for the suggestions but it seems like as soon as drive access takes place the system has a problem communicating with the drive and only a reset makes the drive visible to my PC again. Cannot really see a way around this, has anyone else ever seen those error requester I posted and recovered from it?

[thomas]

In your first post you wrote that you can boot into AmigaOS if you disable the partition in the early startup menu. Is this still true? And does it work in WinUAE?

If you can boot into AmigaOS you should be able to create an image with TSGUI. If you can do this in WinUAE, it shouldn't be much slower than with dd.

In TSGUI enable "extended restart" from the menu. Be sure to save the restart file on a disk which does not lose the latest update on a system freeze (i.e. not PFS and not SFS). As already mentioned you have to save the image file on a disk which allows laaaaaarge files, i.e. NTFS or maybe SFS\2. In order to select the harddrive if no partitions are mounted, you can use the "select raw device" option from the project menu. In any case the "full hdd image" checkbox should be enabled to get an image with partition table and all partitions.

Now when the system freezes during image creation, you can reset and when you run TSGUI again it should recognise that part of the image already exists and enter restart mode. It will first ask you it you want to skip some blocks (which you should do because otherwise it would freeze again immediately) and then continue reading.

If read errors occur regularly (e.g. every 1000 blocks or so) you can tell TSGUI to automatically skip some blocks after a certain amount of read blocks, e.g. you can tell it to skip 2 blocks every 998 blocks.

I think, for this to work you also have to enable "read block by block" from the menu. This will make the operation much slower than reading a big number of blocks at once, but only so you can approach the bad block as close as possible.

[-Acid-]

Originally Posted by thomas

In your first post you wrote that you can boot into AmigaOS if you disable the partition in the early startup menu. Is this still true? And does it work in WinUAE?

Yes i have been trying to fix it today.

If I boot it up and let it boot once and reset it does load in the scsi.device to enable the rest of the drive (after wrestling the error requesters away). Then I go straight to the ESM and disable DH0 and DH3 since those are the ones giving errors. I can then boot from a Workbench floppy and DH1, DH2 and DH4 are accessible so I think the drive itself is working. I cannot do this in WinUAE though it just causes a long hang of the emulation and then takes ages until you can end the process.

The drive has 2 partitions in the first 4gb and with luck I had an install of OS3.9 in the second one, so I made it bootable and then disabled DH0 and chose to boot from DH4. OS3.9 did boot to WB but then I get the read error saying make sure disk is inserted.

On the plus side I was looking around my backups on the PC in WinUAE and I seem to have all of the files on the drive archived (must be a first for me!) so i'm not going to lose anything. I just want to make the drive usable again, i'm going to try deleting the partitions and starting over with it again.

[thomas]

I just want to make the drive usable again

You can't. The drive is dead. If the drive hangs on access of certain blocks, then it has encountered so many bad blocks that the firmware cannot handle them any more. You should trash it and only use it for experiments with unimportant data.

One experiment could be to try a low-level format with OS 3.1's HDToolbox. The IDE drive certainly won't format anything, but at least once I got a drive with read errors to do some kind of firmware-reset so that it starts from scratch to replace bad blocks. This way I could make a drive with a read error in the RDSK block to run again. One could clearly hear that the head moves to the end of the drive and back when reading the first block.

But as said above, actiually the drive is dead. A reset does not remove the bad blocks from the drive. It just reassigns replacement blocks in a different order. Once as many bad blocks have been found as before, it will start to hang again.