A question for any OS 4.0 Classic users on 604e's.....NOT Blizzard 603e's

[Hans_]

Oli_hd wrote:

So yeah Im thinking of uninstalling OS4 for a while, see if things get better later but mehh, compatiablilty is a must for the OS yet they seem to come back with "well the original software wasnt designed right and thats the problem" well sorry, if it works from 3.0-3.9 and 4.0 is the only one not working then its OS4's fault.

I can't comment on the hardware issues, as I don't have the classic version, but which programs aren't working? You can try adding them to the blacklist in the compatibility prefs. That switches off the 68k JIT for that app. I'd assume that apps that access the AGA chipset directly would still work, given that the actual hardware is srill there on the classic.

However, if the program does something naughty like attempt to read/write from the zero page, unallocated memory, or overwrite code, OS4 will stop it and bring up the grim-reaper. Also programs that do things like peek at memory before an AllocVec() to get the size won't work, because this was illegal in the first place, even though it happened to work. Even Commodore said that these programs would break in the next major OS update. These kind of programs would continue to run on OS3.x, and only cause a failure intermittently. Stopping these apps makes the system itself more stable.

BTW, you might want to check out glUAE (also on os4depot). It will allow you to use UAE to run programs that don't work. It's not ideal, but it works. I'm still waiting for someone to integrate OS4's 68k JIT with E-UAE. Hopefully that will be done at some stage.

Hans

[Hans_]

@Kin-Hell

I don't have a classic OS4 setup so I can't help too much, but I do have OS4 on an A-one.

Let's see if I have this right: you installed OS4 ok, but have networking troubles causing lockups, after which OS4 won't startup properly. Is this about right?

I'd suggest trying the following:
In s:startup-sequence, change this line:
C:AddNetInterface QUIET DEVS:NetInterfaces/~(#?.info)
to
Run >NIL: C:AddNetInterface QUIET DEVS:NetInterfaces/~(#?.info)

The original line will wait until your ethernet card is connected, has an IP address, etc. If you're ethernet driver is not working properly, this can cause it to stop at this point. Adding run means that the startup will continue, even if you can't get connected (unless the machine really freezes).

I'm guessing, that the driver and your ethernet card aren't working together properly. Not being able to startup after IBrowse locks up suggests that the ethernet card ends up in some unknown state and doesn't get reset properly. This would cause addnetinterfaces to just sit there.

BTW, what ethernet card are you using? Your sig only mentions 10/100 ethernet. Hopefully you're not the only one with that card.

I know that this is frustrating, but, with the number of different cards in use on classic Amigas, supporting all of them is a huge undertaking. They're planning to release an update that fixes at least the graphics issues and Mediator LT (or whatever sub-model) problems later this month.

Hans

[Hans_]

Kin-Hell wrote:
Not much point trying your Startup-sequence mod. The card cant find it's own Ass, never mind an IP address over @ my Router!

The startup-sequence modification should enable you to bootup to workbench despite the ethernet driver issue.

I've just looked at the OS4 classic compatibility list, and it lists the 8139 card as unsupported because it requires the Mediator DMA hack. So there's your problem; you're trying to use a card that is known not to work.

If you haven't done so already, you could try installing the 68k Mediator drivers, but you'll have to do that for absolutely every card plugged into your Mediator board. The OS4 native and classic Mediator drivers can't coexist.

Failing that, bug Elbox to write OS4 drivers that use the graphics-card DMA hack.

Hans

EDIT: I wish that we'd thought to look the card up in the compatibility list earlier as it would have saved so much time and frustration.

[Hans_]

Kin-Hell wrote:
I'm using the Mediator 3.10 drivers under OS3.9 with the 8139 card. I am not running a DMA hack as far as I am aware.

Do you have any info about this hack please? The usual searches returned ziltch!

You are actually using the DMA hack because it's used by Elbox's Mediator drivers. I can't think of any good resources that explain what happens. Basically, the Mediator board can't DMA to main memory (I can't remember why). So what it does is it reserves some memory on your PCI graphics-card and uses PCI bus-mastering to DMA to memory of the PCI graphics-card. From there it can be copied to main memory. This is a hack as it's working around hardware limitations that prevent it from doing direct DMA. It also requires that you have a graphics-card present.

This means that every card that requires DMA in order to work requires special Mediator board code that differs from the driver used by other systems (e.g., the AmigaOne, whose north-bridge supports DMA to main memory). Basically, this would mean that every PCI card that uses DMA would need two drivers, one for PCI northbridges that support DMA, and a special one specifically for the Mediator boards. It's not a minor change to the drivers.

Hopefully Elbox will support their products and release OS4 versions of their drivers. In the mean time, I've been told that you can use the 68K Mediator drivers you use under OS3.9 on OS4 too, but you have to change ALL your PCI drivers to the 68K Mediator ones. You cannot mix the OS4 native and Mediator drivers. Alternatively, you could look for an 8029 card, which does work (and a bugfix for the driver is coming).

Hans