AmigaOS 3.2 and beyond

[hexaae]

Hopes to finally see C:Eval improvements supporting for example floating point ops + correct order operators like 0.01*9+3 = 3.09?

[bubbob42]

Originally Posted by tomead

tried with and without the CF adaptor, and left the amiga for over 5 mins to see if the kickstart screen comes up. May just reflash the eprom with 3.2 if I cannot perform a clean install from disk.

Re-burn or try another EPROM. Clean the legs of the EPROM by gently scrubbing it over a sheet of paper (do not use sandpaper).

The only known problems so far are unreliable Eproms or missing pullup-resistors in Rev. 3 or 5 A500s or - albeit rare - weak power supplies. That doesn’t mean there can’t be any other problems with your A600 setup (not every board revision may have been tested), but please try to rule out the most common issues, first. I successfully tried 47.102 in an unexpanded A600 myself.

[jbenam]

I think I have found a bug with the 3.2.1 (and potentially 3.2) ROM.

I have got the original Seagate ST3144A that came with my Amiga 4000. It's still in pristine condition (no bad sectors, etc. - tested on one of my DOS PCs with SpinRite) and it works well with the 3.1 ROM I had in my 4000 originally.

With 3.2.1... it's not detected at all. Has the CHS support been removed from ROM somehow? I know this one drive doesn't support LBA address but only CHS. Another drive I have got (that supports LBA) is detected with no issues.

Might also be a question of IDE timeout?

[kamelito]

Another bug found by Emu68 author.
https://twitter.com/michal_schulz/st...292078083?s=21

[bubbob42]

Originally Posted by kamelito

Another bug found by Emu68 author.
https://twitter.com/michal_schulz/st...292078083?s=21

boards.library is a binary contribution to OS3 and not part of the OS. I’m very happy about the fix, because the bug must have been there for years and I never got around to fix it.

However, Michal has now access to the source code and kindly fixed it for the next version of the library, soon to appear on Aminet with new board entries.

[tomead]

I forgot to say I have an aca620ec in my A600. It does boot from my old CF card. The same eprom that I had flashed 3.1.4 to. Timeout issue maybe if no ide drive attached. A600 motherboard is a pal rev 1.5

[tomead]

update: Flashed eprom with 3.2, same issue. Flashed 3.1.4. kickstart screen shown with CF card removed. So it is a change from 3.1.4 to 3.2 that is causing an issue.

[Lisko]

Hi, I noted that Identify.library apparently does a better job than boards.library at identifying accelerators that share the same vendor and product ID. For example it is correctly reporting that I have a blizzard 1230 mk.IV instead of the generic blizzard 12x0 reported by boards.library. Could an option be added for using the identify library instead of the boards one? Or maybe the boards.library author could take a look at the Identify.library's accelerators detection routine.

[hexaae]

Just FYI, identify.library author is back just recently, released a new version, and put the project on GitHub: https://github.com/shred/identify

[bubbob42]

Originally Posted by Lisko

Hi, I noted that Identify.library apparently does a better job than boards.library at identifying accelerators that share the same vendor and product ID. For example it is correctly reporting that I have a blizzard 1230 mk.IV instead of the generic blizzard 12x0 reported by boards.library. Could an option be added for using the identify library instead of the boards one? Or maybe the boards.library author could take a look at the Identify.library's accelerators detection routine.

identify.library merges CPU information with expansion board information, which is something I always wanted to add to boards.library, but alas….

Since we are since recently sharing sources between the libraries one solution or another might appear. However, if you use Showconfig in 3.2+ you’ll get CPU information via mmu.library in the same window displaying boards.library‘s findings. Well…you can guess what I’m after.

[tomead]

I have confirmed that the aca620ec has a compatibility issue with rom 3.2 and by extension 3.2.1. Took the card out and the rom boots correctly. checked the individual computers forum to report the issues to Jens and noticed that the aca630 had a similar issue of booting to black screen and not going further.This was due to a bug in Z3 autoconfog which has been since fixed in a later version of kickstart. The aca620ec maps 4Mb of ram using Z2 autoconfig as has a 24bit address bus. Not sure if it is a similar issue to the aca630 or a new bug. Hope it's retified in kickstart 3.2.2

[coldacid]

If the problem is because the accelerator ROM expects a bug in Kickstart and fails if the bug has been fixed, then it's on the accelerator manufacturer to rectify the issue with their ROMs, not with the OS developers to re-introduce a bug.

[bubbob42]

I have an ACA620 and will make sure it gets tested for 3.2.2. No time available right now, unfortunately.

[golem]

I think ProNET does not play nicely with OS3.2.1
I can view files but when I try to write to a ProNET server I get error 65535.
When mounting a drive it sometimes hangs as well.
Perhaps ProNET is not so system friendly but I've set up Envoy now instead which works well between my 3.9 and 3.2.1 in parallel with Roadshow.

[Ulysses13]

Originally Posted by bubbob42

A reboot is normal. Are you using a turbo board and if yes, which one?

I have a TF1260 in my A1200 which is perhaps confusing things...?

Have actually started a thread here already though, so don't want to hijack this one...

[Bruce Abbott]

Originally Posted by kamelito

Another bug found by Emu68 author.
https://twitter.com/michal_schulz/st...292078083?s=21

"Join Twitter to get the full story". No, I won't.

[paraj]

Originally Posted by Bruce Abbott

"Join Twitter to get the full story". No, I won't.

Relevant parts (I don't like twitter threads either).

Originally Posted by @michal_schulz

Good feeling when you spend hours hunting for a bug to find out it is not yours Thanks to #Emu68 I'Ve fixed boards.library from AmigaOS 3.2.1

Originally Posted by @michal_schulz

The faulty code was using 16-bit portion of D4 register to keep number of entries in the list, but decremented entire 32 bits **and** tested entire 32 bits for Zero. Too bad that several lines above the very same register was used as temporary holder of 32 bit address

[bubbob42]

Yeah, spread my shame!

Or was it there since before I inherited the poor thing…anyway, doesn’t matter. Good debugging.

[coldacid]

Originally Posted by Bruce Abbott

"Join Twitter to get the full story". No, I won't.

Twitter without needing to join.


In fact there are browser extensions that will replace twitter.com links with their Nitter equivalents using a preset or random Nitter instance. Definitely a better experience!

[hUMUNGUs]

To the wonderful AmigaOS team: Thank you for the updates, they are really good. I have a question. Could you elaborate on what you currently are working on ? Is 3.2.2 the next update ?

Thanks again