CD32Load

[Galahad/FLT]

With regards a prospective new menu, please go here:http://eab.abime.net/showthread.php?...79#post1060179

To bring up ideas, and maybe if you have ideas on display, mock up screens would be good.

[StingRay]

Originally Posted by jotd

@Stingray: no, no OS friendly that's the point

Excellent, then I'll create some code for you. What exactly do you need? Simple 8x8 1bpl text writer with the (error) text as parameter?

[jotd]

@stingray: exactly what I need: pass the string and write it on the screen. Can kill copperlist, everything since will never return.
This CD32LOAD project will be a collaboration with a lot of people! (Toni, me, you, Bert, Rob Northen , earok for intensive testing, & misc people in EAB)

@Akira: why kickemu installs don't work? because I did not code kickstart loading in JST (well, it works for some games which don't need DOS). I did not implement BCPL relocation for DOSlib (different from the other relocs, since it's a port from TripOS), and I did not implement Examine/ExNext either on JST, not to mention it would be difficult for CD32LOAD since I don't know if a command exists for directory listing in RN loader. Well, I have to perform some more tests, that could be interesing and we need that even in a modified WHDLoad with all the kickemu stuff implemented.

[Amiga1992]

But wouldn't kickstart loading, as I said, reduce the chances of a game to run because it would use a lot more RAM?

[jotd]

Yes, it would be it would still be possible. Amiga KS 1.3 is $40000 bytes, so CD32load could run a 1megabyte game anyway, provided that the data files are not too big (no way Cruise for a Corpse)

[earok]

Originally Posted by jotd

Yes, it works for some (ex: Arkanoid). Moonstone locks up indeed, but that's probably for some other reason. I recently fixed a bug related to file-based loaders, but did not find any improvements... Still searching.
When Bert releases his version (if he does it) compatibility will raise
(however some slaves work with JST but not with CD32load! strange. but the first step is making them work with JST)

Ah ok, cheers

Your idea regarding setting blue to $00, will that work with the current version of CD32Load?

A few more things I've tried:

* Alien Breed: The Story rainbow screens
* Apocalypse seems to load and handle keyboard just fine, but has severe graphical glitches during gameplay?
* Beach Volley Works. Gamepad mapping a bit temperamental, intro seems to run slow
* Chase HQ crashes to grey screen?
* Chase HQ II Loops at splash screen? Repeatedly fades out and then fades in again
* Coala both ECS and AGA crash to rainbow screen.
* Citadel seems to crash sometime during the intro (we get a blank screen instead of the FMV stuff, but the sound continues)
* Epic works, but seems to have severe keyboard handling issues (I didn't test past the name-entry screen)
* Flashback rainbow screened
* Road Rash crashes to black screen?
* Slam Tilt rainbow screened (thought I'd try this since it's one of the few AGA games to have a slave that fits in 2mb)
* Thunderhawk shows the Core logo, and then immediately crashes (looks like it's trying to show the second screen, but only one bit-plane?)

[jotd]

I have fixed a problem and now Pinball Dreams runs fine. But it's the only one!
I see a lot of KickEmu games in your list. I should add a test at startup: those are not supported.
And gamepad mapping IS temperamental. It is not perfect and will never be, the way it is designed. It's already a miracle that we can do it.

[earok]

Originally Posted by jotd

I have fixed a problem and now Pinball Dreams runs fine. But it's the only one!
I see a lot of KickEmu games in your list. I should add a test at startup: those are not supported.

Sorry, what is KickEmu? I assume KickEmu slaves are the ones that require kickstart images?

Originally Posted by jotd

And gamepad mapping IS temperamental. It is not perfect and will never be, the way it is designed. It's already a miracle that we can do it.

I understand, and agreed

[StingRay]

Originally Posted by jotd

@stingray: exactly what I need: pass the string and write it on the screen. Can kill copperlist, everything since will never return.

This is how the current output looks, displayed is the error code/text, registers and a hex dump for the address registers.

This CD32LOAD project will be a collaboration with a lot of people! (Toni, me, you, Bert, Rob Northen , earok for intensive testing, & misc people in EAB)

It's a nice and useful project which definitely deserves to be supported.

[earok]

Time for some more. Don't worry I'll start on that CSV soon-ish, I'm just having too much fun trying different slaves


* After Burner (Sega) Crash to black screen
* Alcatraz Works, sluggish during gameplay though (intro + menus are silky smooth). Keyboard issues. Maybe for the compatibility list?
* Aaargh! Crash to green screen
* Beast Busters Rainbow screen
* Castle Kingdoms Works perfectly with no keyboard issues. For the compatibility list.
* Death Trap Also works perfectly with no keyboard issues. For the compatibility list.
* Dogs of War Crashes at copyright screen. Blue background.
* Elvira Arcade Works. Some graphical glitches during intro maybe? Keyboard issues perhaps. For the compatibility list
* Indianapolis 500 Works perfectly, no keyboard issues. For the compatibility list.
* Ikari Warriors Green screen
* Mega Lo Mania Rainbow screen
* Mercs No luck. Tested with old slave (crashes to white screen on CD32load and whdload) and new slave (crashes to rainbow screen on CD32load, but works on whdload)
* Nitro I know you've already tested this but I wanted to try putting Player 3's controls on the spare buttons on the gamepad. And it works flawlessly. So if you're crazy enough to play three players on two pads, this is the game for you

[jotd]

@stingray: perfect stuff!!!
@earok: thanks for the tests. Lists updated. Soon with StingRay's code errors will be clearer.
yes kickemu = kickstart required. I'll check that before killing the OS.
WOTD seems to work with "NOJOYPAD" so must be a joypad interaction.
I have an idea: why not mapping the useful buttons on joypad port 0 / mouse? This way LMB/RMB could be used to do something useful (load, pause...) whereas not tampering with port 1 controls.
(port 0 is not implemented yet, even if command line args exist)

[StingRay]

Originally Posted by jotd

@stingray: perfect stuff!!!

Attached is the source which you can use for your CD32Load then. Short instructions are in the source, shouldn't be too hard to use, either call SHOW_DEBUGSCREEN with an error code in d0.w or call it with a0 pointing to a custom error text.

IF "TEST" is set to 0 only the plain debug screen code is assembled and it will stay in an endless loop after the debug output has been written. TEST=1 will enable the test mode which you can use to test/adapt the output to your needs, code will return to the OS once LMB has been pressed.

If you need any changes or have questions feel free to let me know.

[jotd]

nice. Integrating it now.
BTW would it be possible to display program counter & sr? by passing them to D1,D2? useful to display PC when an exception occurs.
And if I want to see the contents of A0/D0 how to do that?

Best way (maybe) would be to movem.l D0-A6,-(A7) and also save SR & PC (stack calling convention against register calling convention)
Edit: it works in cd32load, good job, I had to change some things like: making every move => address position-code-independent (relocatable part needs that), that involved adding a lot of (PC) stuff and indirect writes instead of move.l a0,TEXTPTR which could fail.

[jotd]

new CD32Load version in the zone. Updated compat list, and no more "rainbow screens" but error messages, thx Stingray!

[StingRay]

Originally Posted by jotd

nice. Integrating it now.
BTW would it be possible to display program counter & sr? by passing them to D1,D2? useful to display PC when an exception occurs.
And if I want to see the contents of A0/D0 how to do that?

I'll see about implementing that.

Originally Posted by jotd

Edit: it works in cd32load, good job, I had to change some things like: making every move => address position-code-independent (relocatable part needs that), that involved adding a lot of (PC) stuff and indirect writes instead of move.l a0,TEXTPTR which could fail.

"adding a lot of PC stuff"? The code has exactly 2 (!) relocation entries, one is the textptr thing which I added this evening to support custom error messages, I didn't care about position independent code when doing that as I didn't think it was needed for this project. The other one is the register storing which was 100% pc-relative as well until I changed stuff this evening. Anyway, 2 non-relocatable instructions are not a lot in my book.

[earok]

Originally Posted by jotd

new CD32Load version in the zone. Updated compat list, and no more "rainbow screens" but error messages, thx Stingray!

I've created a public Trello board as a way of managing compatibility testing, I've also been taking screenshots and noting the slave versions. Did you want to edit the board? If so just flick me your email address.

Is it OK if I upload that CSV to Google Docs? That way we can both edit it.

[jotd]

yeah Google Docs is the way.

[StingRay]

Originally Posted by jotd

Edit: it works in cd32load, good job, I had to change some things like

I have checked your changes, your register storing change is not correct as a7 is not stored so it will always be shown as "0" which is wrong. Then, lea SCREEN(pc),xx, which assembler do you use which allows that? Screen is at a fixed address in memory ($60000) so (pc) doesn't make sense, is not needed and should actually generate an error when assembling.

As for the custom messages/text ptr stuff, not sure I'll leave that in as everything can be done with error codes and extending the error table. Do you need this or are error codes enough for you? If so I'll remove the "pass text as parameter in a0" stuff completely.

[earok]

@ransom1122 Cheers

@riempie Ick, sorry about the inconvenience! It's a pity RAD drives can't be made to last permanently though, it'd be a great way of storing saves!

@jotd I've started updating the google docs spreadsheet and doing some more tests. I noticed Agony doesn't seem to work though? It gets past the intro, but then crashes to black screen?

[jotd]

@stingray: on the contrary, custom messages allow to print variables (PC, return codes...) and that's what I use most.

@earok: thx saw it but not much time right now. I'll add red button remapping (pinball dreams). I have a hard time making slaves work. You may be right about agony. I hope there's only 1 root cause and when I find it a lot more games will work.