Workbench 3.2 Issues

[Thomas Richter]

Originally Posted by AMIGASYSTEM

The Execute Command is very important, without this Command you would not be able to execute scripts from Shell or Icons.

Errr... XIcon does not require execute, actually, and the command is now a shell-built-in, so you do not need it as a binary anymore.

[Storm]

Originally Posted by Thomas Richter

Why? Because it's worth a try, that's why. MagicMenu is one of the programs that patches badly into the system, so if intuition freezes, that's a likely candidate.

You're right, I'll try anything to find the root issue of this problem. I did remove it, rebooted, even powered off to see if there was a change. Problem was still there Other than that, I haven't seen any issues caused by MagicMenu itself on any of my systems, and I have a couple setup at the moment.

[B2k_ad]

Originally Posted by Storm

2. I have to disable extra RAM to get PCMCIA cards to work, like pre OS 3.1.4 days on my A600. I never had to do this with OS 3.1.4. Something to do with card.resource?

Machine: A600 with Furia 020 (white version).

From what i read on A1K:

From Thomas Richter translated by me... so don`t be to hard on me

"The problem is that initially every PCMCIA Device consumes 4 Megs of Ram until Card.resources comes up and gives it back in case the the Device dosen`t need it.
If however in the upper 4Megs there was a stack or an iterrupt, the system would crash the moment you plug a device in"
(Loosey translated)

So i think they saw this as a bug, and it working was sheer luck.


Maybe someone already had written something to get around that on aminet

[Storm]

Thanks for that answer, I guess I can live with that on my A600 now I know what's going on.

[AMIGASYSTEM]

Originally Posted by Thomas Richter

Errr... XIcon does not require execute, actually, and the command is now a shell-built-in, so you do not need it as a binary anymore.

Yes sorry, my English is not very understandable, i meant to say that Execute is an important command, now internal and anyway it should always be inserted "without the need of the path" when running a script from Icon through the IconX command.

[FranciscoJr]

Originally Posted by Storm

Thanks for that answer, I guess I can live with that on my A600 now I know what's going on.

Hello,
What I found and modify in my A600 Furia setup....

The workaround for this, at the moment, is to use the card.resource V45 from ROM 3.1.4
The new card.resource V47 from 3.2 ROM is causing the problem. For this, You must keep the 3.1.4 ROM in A600 and install the OS 3.2 over it. After completed the 3.2 installation, just rename the card.resource or delete it from Libs/resources. Loadmodule will look for it on boot, and as you renamed it, it will use the version 45 from the 3.1.4 ROM. This way you can use the PCMCIA with all the 9.5MB ram from Furia card with no problems. do not use SHADOWROM in furiatune, but use the LoadModule command with the
NOMEMFKICK switch in your startup-sequence. this way all 3.2 modules will be in 32bit RAM.

[Storm]

Thanks for the info FrancisoJr. It was more of a curiosity than a deal breaker for me.

It's my first problem I have listed that annoys me, since it worked on 3.1.4, and even more annoying that no devs have given some sort of answer on what could be happening

[FranciscoJr]

Originally Posted by Storm

Thanks for the info FrancisoJr. It was more of a curiosity than a deal breaker for me.

It's my first problem I have listed that annoys me, since it worked on 3.1.4, and even more annoying that no devs have given some sort of answer on what could be happening

Hello Storm,
What I´ve done recently. Just modified the original 3.2 ROM with Remus, and changed the card.resource V47 for the V45 version, and then burned the new custom 3.2 ROM. It is working fine as an original 3.2 Amiga 600 ROM.

Francisco

[Thomas Richter]

Originally Posted by Storm

I have to disable extra RAM to get PCMCIA cards to work, like pre OS 3.1.4 days on my A600. I never had to do this with OS 3.1.4. Something to do with card.resource?

I don't know whether this has been answered before but... yes, that is a hardware limitation and a bug in 3.1.4. The problem is that the PCMCIA card bus occupies potentially the same address space as Z2 autoconf RAM. Now, you could say that "well, but my PCMCIA card is just an IO card and not a RAM card, so this address space is not needed".


Unfortunately, the situation is somewhat complicated. PCMCIA cards may speak "two languages". A "RAM-like" setup with random access to all contents, and an "I/O-like" setup, where it works like a harddisk. Unfortunately, the default setting for PCMCIA devices is to speak "I am RAM", even if the PCMCIA device is not really RAM, but a disk. The carddisk.resource then has to manually switch the device from "RAM mode" to "disk mode".


Unfortunately, at this point, it is already too late - if you insert the card, the RAM goes away before the card.resource can react, because the hardware (gary) does the switch.


Now, if anything is in the RAM, e.g. an interrupt, then inserting the card "pulls the rug" under the CPU, and the system crashes.


Thus, as silly as it may seem, you cannot use expansion RAM and PCMCIA at the same time. It is a silly limitation/interaction between how PCMCIA works, and how the PCMCIA RAM area is mapped in the system, even if the inserted card is not really RAM and only needs to be "switched away".

[Thomas Richter]

Originally Posted by FranciscoJr

Hello Storm,
What I´ve done recently. Just modified the original 3.2 ROM with Remus, and changed the card.resource V47 for the V45 version, and then burned the new custom 3.2 ROM. It is working fine as an original 3.2 Amiga 600 ROM.

Unfortunately, this does *not* work, it may only appear to work on your setup with your cards. For a full technical background, see my answer above.


But in short, if you have program loaded in the upper RAM area and insert the wrong type of card -even if it is an I/O card- you crash the system, and there is nothing the card.resource could do about it.

[Storm]

I should have made my post clearer, my bad. I can accept the fact that I have to reduce memory to use PCMCIA cards on the A600, so lets ignore this 'issue' part.

My issue that bugs me is this one:

1. The only problem I can't seem to solve is the Workbench screen locks up as well as keyboard input if I plug in an external hard drive via USB. I have a feeling it's got something to do with the massstorage.class of Poseidon 4.5. Weird thing is, if I load up Directory Opus 4.x and then plug the drive in, I can access the drive like normal. All my hard drives are using PFS3 as the file system, internal and external.

Machine 1: A4000, Cyberstorm MKII 060, X-Surf 100 with Rapid Road, Deneb and Picasso IV.
Machine 2: A2000, GVP 030, X-Surf 100 with Rapid Road.

[gdonner]

Originally Posted by Storm

I should have made my post clearer, my bad. I can accept the fact that I have to reduce memory to use PCMCIA cards on the A600, so lets ignore this 'issue' part.

My issue that bugs me is this one:

1. The only problem I can't seem to solve is the Workbench screen locks up as well as keyboard input if I plug in an external hard drive via USB. I have a feeling it's got something to do with the massstorage.class of Poseidon 4.5. Weird thing is, if I load up Directory Opus 4.x and then plug the drive in, I can access the drive like normal. All my hard drives are using PFS3 as the file system, internal and external.

Machine 1: A4000, Cyberstorm MKII 060, X-Surf 100 with Rapid Road, Deneb and Picasso IV.
Machine 2: A2000, GVP 030, X-Surf 100 with Rapid Road.

I'm just curious: do the lockups occur if you plug the external hard drive in before powering on your Amiga? Also, is your RapidRoad powered via floppy drive connector to the power supply, as I think it should be(?) seeing as an external hard drive (non-solid state) will be drawing more power than a typical flash drive would.

I suspect you're right that it might just be some kind of (until now) undiscovered bug in the Poseidon 4.5 massstorage.class. Too bad that someone hasn't been able (or motivated?) to continue Poseidon's development and bug fixes, seeing as numerous people still use it.

Though it's really odd that running Opus 4.x (assumingly 4.17pre) prevents the lockups... obviously something software-related; not power-related.

[Storm]

Originally Posted by gdonner

I'm just curious: do the lockups occur if you plug the external hard drive in before powering on your Amiga? Also, is your RapidRoad powered via floppy drive connector to the power supply, as I think it should be(?) seeing as an external hard drive (non-solid state) will be drawing more power than a typical flash drive would.

I suspect you're right that it might just be some kind of (until now) undiscovered bug in the Poseidon 4.5 massstorage.class. Too bad that someone hasn't been able (or motivated?) to continue Poseidon's development and bug fixes, seeing as numerous people still use it.

Though it's really odd that running Opus 4.x (assumingly 4.17pre) prevents the lockups... obviously something software-related; not power-related.

Yes, the lockups happen too if I start the machine with the external drive already plugged in. The external drive has it's own power pack. On the A4000, the lockups happen whether I connect it to the Deneb or RapidRoad, which has the floppy power connected. On the A2000, same deal with just a RapidRoad, but this one does not have the floppy power connected.

You guessed right, I am using DirOpus 4.17 pre. When the lockups occur, only the mouse pointer works, though nothing can be done with it. Icons on the screen don't work, but the drop down menus at the top pull down but no selection can be made. It's a weird one for sure. Yep, sure is a shame that Poseidon isn't worked on anymore.

Mods: Can this thread be relocated to the new 'support.AmigaOS' section?

[FranciscoJr]

Hello,

I own two Amigas (600) with Furia accelerator installed. One with Furia 68020 40MHZ firmware V12 and the other with the latest firmware V13, Furia 68020 33MHZ. Burned two custom 3.2 ROM and using without problems on the two machines. Using also PCMCIA internet cards attached on both and working fine with ROADSHOW and the 9.5MB RAM installed and working.

https://drive.google.com/file/d/1yPg...ew?usp=sharing

More pictures: with physical 3.2 Custom ROM and 9.5 RAM from Furia accelerator card
https://drive.google.com/file/d/1Y-o...ew?usp=sharing
https://drive.google.com/file/d/1flh...ew?usp=sharing
https://drive.google.com/file/d/1UJ8...ew?usp=sharing
https://drive.google.com/file/d/1EDv...ew?usp=sharing

Obs: With this setup, I can, for example, insert a CFAUX card, the same I use with ACA500Plus for sharing Amiga or PC files, and take it out, and insert again the internet card and the system goes on stable with no problems at all.

Best Regards,
Francisco

[FranciscoJr]

Hello,
I made a video, showing my Amiga 600 with Furia accelerator card 68020 33MHZ and full 9.5MB RAM working in a single boot. OS 3.2, physical custom 3.2 ROM with card.resource v45 from OS 3.1.4.

https://drive.google.com/file/d/1k5v...ew?usp=sharing

Francisco

[Nowee]

Hi, sorry for bringing the topic up from the grave. Let's get back to the USB topic, please.

I have a little theory. I've tried Poseidon 4.5 with many, many devices, including 2002 era ones, and last gen 3.2 ones (Andycine USB 3.2 sata adapter, for example). The attempts were done on several machines with different hardware (Spider, RR on XSurf100). The manufacturer's drivers were used (Jens, Elbox).

For scanners and printers, it works perfectly, same for pointer devices. It's a totally different issue with drives, and surprisingly some people claim that they never have any problem with Poseidon. That does yield a few questions, right?

Copying to USB devices will fail in most cases. It can fail after a few seconds, or a few hours of copy, but it will fail, always with the same read error. I've set Poseidon's inactivity timeout to 600s, but it doesn't seem to be related. In rare cases it works flawlessly, too, just like today:

Earlier I've copied 1.4GB of data from a FFS enabled 64 Gig single partition DOM module on the default IDE port of my 4000Tx to a PFS3 formatted with a single partition 32GB SSD through the Andycine adapter and could copy it all without problem.

Now, whenever I try to copy data from my PFS3 (latest AIO from Aminet) partitions, from a 512GB SSD, over a CyberstormPPC SCSI, to the same SSD/Andycine 32GB drive, using the same machine, and without MagicMenu, PFS3 will invariably fail at some point, which in turn invalidates the drive. Keep in mind that I'm using the exact same settings I did with the previous copy from the motherboard IDE. I can't even copy a single GB.

It is NOT a Cyberstorm SCSI issue, since I have similar problems with a Buddha IDE on my Amiga 3500. I've also tried with plugging the 512 GB internal SSD to my last gen FastATA, with similar results. Obviously, I've tried several other adapters, not just the Andycine. I also have the issue with large USB thumb drives, and large FAT32 drives.

A minute ago I flawlessly copied the 1.4GB of data back from the Andycine SSD to the same 80GB partition I couldn't copy data from. So it looks more like a read from Amiga -> write to USB issue.

I suspect some TD64 or "large drive" fix to be set by Poseidon, that could mess with PFS3's and or OS 3.14+. The whole "fix capacity" option in Poseidon's settings hints at possible additional geometry fixes.

The question is, could we disable the OS "large drive support" just for Poseidon? Is that even possible?

[Storm]

Thanks for bring this up again Nowee, as I'm still having the same problems too.