ADFFS future feature testing

Discuss development specific to the Pi version of ADFFS
JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.60 testing

Post by JonAbbott » Wed Oct 12, 2016 3:17 am

Vanfanel wrote:- Alone in the Dark: Hangs with error before protection, while passsing the credits book pages.
-Elite: Internal error.
These are a know issue, if your running on ARMv7 (Pi2 / Pi3)
Vanfanel wrote:Magic Pockets: "SWI returned a bad error pointer" after intro.
Updating TrackerMod will resolve this error.

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.60 testing

Post by Vanfanel » Wed Oct 12, 2016 10:29 am

@Jon: I am doing all this testing on a Pi1.
Also, I have updated TrackerMod before start testing.
Updated testing results with more games.

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.60 testing

Post by JonAbbott » Wed Oct 12, 2016 6:52 pm

Vanfanel wrote:@Jon: I am doing all this testing on a Pi1.
Also, I have updated TrackerMod before start testing.
My mistake, I assumed it was SWI &1E returning the error, but it appears to be another SWI.

Could you please retest the ones that fail on 2.59, as I wouldn't expect the change to VSync to cause some of those issues. They may simply be issues we've missed or ones I've been unable to reproduce.

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.60 testing

Post by Vanfanel » Wed Oct 12, 2016 8:02 pm

@Jon: retest done. Moved Curruption to working... I did something wrong but it works now on both 2.59 and 2.60.
Updated results.

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.60 testing

Post by JonAbbott » Thu Oct 13, 2016 9:33 am

Thanks for all the testing, at least most of them are unrelated to PLL VSync. Observations from my testing of the failures.

Rekall and Magic Pockets are both an unresolved bug in TrackerMod. It seems my bug fix has just moved the bug elsewhere, so I'll look at that.

BlastOn I had trouble repeating, had to run it quite a few times to get it to fail. The same issue is occurring on 2.59, so I suspect this is a bug in the code, which is probably branching to the wrong address.

Chocks Away, I've been unable to reproduce. I've tested at 50hz and 60hz on both Pi1 and Pi3. At what point did you get the crash?

Dune II, I get a crash on Pi1 with 2.59 anyhow. It's working on my Pi3 though, so this isn't PLL VSync either. If you look at R15 on the trap report, it's CLib that's crashing. I suspect it's trying to do a stack unwind. Suggest we ignore this until we get around to testing the localised CLib.

Command Ship, it does that under 2.59 on Pi3, so sounds like a bug in the code.

Elite, I've been unable to reproduce. At what point does it report the error?

Alone in the Dark - it does that under 2.59 on Pi3, so I suspect it's CLib doing a stack unwind.

Top Banana / Blowpipe - known issues that need looking into at some point.

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.60 testing

Post by JonAbbott » Thu Oct 13, 2016 5:14 pm

JonAbbott wrote:Rekall and Magic Pockets are both an unresolved bug in TrackerMod. It seems my bug fix has just moved the bug elsewhere, so I'll look at that.
Should be fixed with the updated TrackerMod I've just put in the CPU directory.

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.60 testing

Post by Vanfanel » Thu Oct 13, 2016 5:41 pm

@Jon:
-I can't reproduce the Chock's Away crash. Maybe I forgot to load Sparks or some other idiotic mistake on my side. Moved to working.
-Magnetoids works OK if I don't try to run Elite before it. So moved to working, too.
-Rekall and Magic Pockets are now fixed with today's TrackerMod!
-I have updated the info on Elite's error. I am running it installed to HDD, it was working before. Trying to boot the floppy image doesn't do anything.

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.60 testing

Post by JonAbbott » Thu Oct 13, 2016 11:24 pm

Elite sounds like a CLib problem. I'm going to ignore it for the time being as it will come off the working list when localized CLib is turned on.

It looks like PLL VSync has had no detrimental effect, but I need to look at Blaston to confirm its unrelated.

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.60 testing

Post by JonAbbott » Fri Oct 14, 2016 9:45 am

@Vanfanel Were you testing at 50Hz? We need to test at both 50Hz and a refresh rate other than 50Hz to confirm there's no knock on effect.

At 50Hz, I don't think the PLL VSync will have any effect other than possibly delaying the frame by 2cs. At other refresh rates, there's the potential for things like audio to glitch if bespoke Channel Handlers don't fill the buffer on time. ADFFS does buffer the audio slightly ahead of what's required, so I'm hoping it's not noticeable.

The other potential issue is where games are reliant on VSync and TickerV firing in a certain order or rely on specific timing. We already know about SWIV, but there may be others.

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.60 testing

Post by Vanfanel » Fri Oct 14, 2016 12:51 pm

@Jon: Yes, I did the test with my usual 50Hz setup. I can re-test all those with a 60Hz videomode, give me a couple of days.
I have moved Rekall and Magic Pokets to working.

Post Reply