TF4060 Beta Program

[terriblefire]

For anyone having DMA issues with Z3 SCSI Cards try running this tool.

It will check the masks on all drives and if there is upper ram present and the DMA mask would cause issues it will show a recoverable alert and then patch your drive mask. Will eventually get this into the FPGA rom.

dmamaskpatch.zip

[chucky]

It should be loaded with the loadmodule command

[Chritoph]

The huge problem is:
It would trigger messages in PFS, but the crash on re4091 could happen, still.
Only a driver patch would help, but even then a tiny utility to mark the mem non-DMA capable will be needed.
I will start working on it as soon as the AmiGUS drivers are done, but that will be next year earliest.

[terriblefire]

The huge problem is:
It would trigger messages in PFS, but the crash on re4091 could happen, still.

Please elaborate on the scenario that has the issue. Does the re4091 still have the issue?

Only a driver patch would help, but even then a tiny utility to mark the mem non-DMA capable will be needed.
I will start working on it as soon as the AmiGUS drivers are done, but that will be next year earliest.

How do you mark memory as non-dma capable? There is nothing in the OS for that that i can see.

[Chritoph]

Please elaborate on the scenario that has the issue. Does the re4091 still have the issue?

I will retest ASAP, but from my current point of knowledge, I can crash the box as described even if mask is set as suggested.

How do you mark memory as non-dma capable? There is nothing in the OS for that that i can see.

Worst case: some additional settings tool or whatever.
OS boots fine, until first 128MB are used up. ;)
A better solution could be a patch in PFS actually. But that is kind of not cool either, somehow.

[Chritoph]

Well ...

20241021_225220.jpg

The module is loaded, did not complain about anything and did not show any alert or alike.

Does the pic answer the question?

Same old PFS error message re:mem and mask not fitting together. Pushing OK causes a crash...

[alexh]

That can happen if you have a typo in your memory mask

https://eab.abime.net/showthread.php?t=87567

[terriblefire]

I'm sort of the opinion that PFS should not do this. I know why it does but i feel this is not a filesystem level responsibility.

[donix]

Hi,

I apologise being a bit late to the party but when I received my TF4060 back in late august my A4000TX wasn't quite ready yet. Now I have had the pleasure of trying it out for a few weeks and with my initial, rather basic, setup it has been running just fine. The setup being:

A4000TX 1.1 with 112 MB of fastRAM
Kickstart/os 3.2.2 with ehide booting of TF4060 IDE
Spinning disk and CDROM attached to A4000 internal IDE
X-surf 100 with RoadShow TCP stack

What doesn't seem to want to play nicely is my attempt to run a Prometheus resurrector PCI brigeboard with a Radeon graphics card. Specifically the graphics card seem to work just fine running a 1920x1080 workbench in 24-bit color the A4000 will do a 8000 0004 guru when I also try to use the only other Zorro card that I own, and that is an X-surf 100. I can get a various amount of pings over the network but eventually and typically sooner rather than later i get the aforementioned guru.

I am aware that this is a very hacky setup indeed but it does seem to be super stable when using it with a BFG9060. Any kind of debugging information I can provide regarding this?

I am also not sure which firmware Chucky put on my card - any chance the updated firmware versions in this thread could fix this issue?

[terriblefire]

I would try removing the onboard fastram as a first step. If that doesnt help we can try other things.