exxos's Atari, Amiga & retro forum

Hearing people are having issues with TF cards (and others too btw) with OS 3.2

Please can someone from the development team reach out to me because without knowing what you've changed i'm going to end up disassembling 3.2 just to figure it out. I have some observations already though.

Caches are turned on waaaaay earlier than in 3.1. This is causing issues with TF1260 and causes chipset reads/write probes to be cached. Ideally caches shouldnt get turned on until SetPatch is run so the 060 card library can setup the MMU etc. I have some firmware fixes that can work around this that are going out in new cards (as of 7/6/2020).

ReadGayle() maybe a victim of this. TF cards (and most "fake" Gayles will return a GayleID of $DF. GAYLE id is at $de1000 (write anything, read back 8 times, high bit has the data - pattern is $xx for current rev). Previous revisions of scsi.device checks that result from ReadGayle() matched the pattern...
AmigaOS/scsi.device should be ignoring the bottom 4 bits. I suspect it probably is but if its not get that code out now. Almost *all* fake Gayle cards return $DF (TF, Furia, Kipper2k, M608, Akkiko).

However if AmigaOS is turning on caching too early this may result in the machine caching reads to Gayle. You can diagnose this if you have this issue by disabling cpu caches on the CDIS jumper.

For the moment 3.2 seems to have issues and I cannot support this without some idea of what the guys have changed. 3.1, 3.1.4 and 3.9 all work and it was years of work to get all of these working.

Please post your OS 3.2 + TF related issues here and i'll try to get to the bottom of this. If you want perfect then avoid 3.2 for now.

Just my 2¢:
Alinea has released CD32 ROM and it seems to work nice in Akiko32 (can't test it though)

Also new version of Remus was released, it's able to work with OS3.2 now
http://www.doobreynet.co.uk/beta/index.html

Hoppefully they can fix it! But tbh, I just stick with 3.1 or on my 4000 3.9. Something is always going to break if these OS upgrades aren't tested against everything - which is a huge undertaking. I tried 3.1.4 and didnt like the "changes" it brought either.

8 Bit Dreams wrote: ↑Thu Jun 10, 2021 12:16 pm Just my 2¢:
Alinea has released CD32 ROM and it seems to work nice in Akiko32 (can't test it though)

Also new version of Remus was released, it's able to work with OS3.2 now
http://www.doobreynet.co.uk/beta/index.html

The reports i'm getting are blank screen hangs on all 030 TFs + Furias. I'm only interested in mainline Amiga OS 3.2 roms because thats what people will buy, fail to get working and then come and moan at me... possibly quoting laws to try and bully me into fixing things. As has happened before.

TBH Stephen, I wouldnt get stressed with it! Based on the volume of people using these accelerators, I would EXPECT the people behind 3.2 to have tested it on at least 1 x TF 030 board. The fact the TF cards work on every other combination of OS and hardware etc, its not your problem - its theirs.

GadgetUK164 wrote: ↑Thu Jun 10, 2021 12:42 pm TBH Stephen, I wouldnt get stressed with it! Based on the volume of people using these accelerators, I would EXPECT the people behind 3.2 to have tested it on at least 1 x TF 030 board. The fact the TF cards work on every other combination of OS and hardware etc, its not your problem - its theirs.

Well it becomes mine when people ask why it works with X accelerator and not Y.

For the people having blank screen problems and if you have installed the `Generic` 030-040-060 CPU that comes with OS 3.2, then you must remove the 040.Lib ( 040 causing probs again lol ) otherwise it will crash, remove it and it should work like on my setup on my CD32 with 030 + MMU still on 3.1 Rom below

For the record; I've tested OS3.2 with TF534, TF536 on both A500 and A2000. These have been working great.
I didn't test my TF330 and TF328 since CD32 wasn't initially a supported user-case in the beta, not until the end, and then I didn't have the time to test it.
And since I don't have access to TF1260, TF360 or TF4060 these didn't get much love.
Did test other variants, though, which all worked great ('030 to '060 CPUs).

A1200 with TF1260 and 3.2, I can no longer get the PCMCIA wifi card to initialise.

When I run "C:wirelessmanager prism2.device, I get an error message something similar to "device failed to initialize".

(Yes I have run C:Cardpatch and C:Cardreset before running wirelessmanager)

The wifi card was running perfectly with TF1260 and 3.1, had nice up and down speeds, etc etc.

matt020 wrote: ↑Thu Jun 10, 2021 2:52 pm A1200 with TF1260 and 3.2, I can no longer get the PCMCIA wifi card to initialise.

When I run "C:wirelessmanager prism2.device, I get an error message something similar to "device failed to initialize".

(Yes I have run C:Cardpatch and C:Cardreset before running wirelessmanager)

The wifi card was running perfectly with TF1260 and 3.1, had nice up and down speeds, etc etc.

Try disable caches and see what gives.