Open Lara engine on the ATARI Jaguar

[swapd0]

Ok, recompiled, tested, packed, tested the packed version and it works...

 

Download the uploaded version and yellow screen... WTF!?!?

 

 

Edited February 7, 2022 by swapd0

[swapd0]

With this link it works.

 

https://mega.nz/folder/FUcEWRTK#BqLluo_3JArKAarrxlLSVQ

 

 

[swapd0]

With VJ 2.1.3(Final) doesn't work. It hangs after the intro.

[+CyranoJ]

7 minutes ago, swapd0 said:

With this link it works.

 

https://mega.nz/folder/FUcEWRTK#BqLluo_3JArKAarrxlLSVQ

 

 

Its a green screen this time.

[swapd0]

Just now, CyranoJ said:

Its a green screen this time.

Just like in the VJ.

 

What on the hell is happening?

 

[+CyranoJ]

1 minute ago, swapd0 said:

Just like in the VJ.

 

What on the hell is happening?

 

Dunno, but red screen is simple

In all seriousness, its probably something really simple.  Hope its a quick fix.

[swapd0]

7 minutes ago, CyranoJ said:

Its a green screen this time.

Skunkboard or JagGD?

[Zerosquare]

Tested here (Jagtopus board).

 

With the BJL ROM, the title screen is corrupted and it hangs on a black screen when you press a button:

 

With the normal ROM, it works fine:

:

 

You probably forgot to initialize some RAM or registers.

[swapd0]

10 minutes ago, Zerosquare said:

You probably forgot to initialize some RAM or registers.

Probably... maybe I need to fill the RAM with a random value and see what happens in my skunk board.

[+CyranoJ]

2 minutes ago, swapd0 said:

Probably... maybe I need to fill the RAM with a random value and see what happens in my skunk board.

If its headerless... have you cleared the BSS?

[swapd0]

34 minutes ago, CyranoJ said:

If its headerless... have you cleared the BSS?

No...

 

Edited:

I've added a clear bss routine at the start, and now it works on the emulator. I've uploaded the mega file.

 

Wrong timing...

 

Edited February 7, 2022 by swapd0

[42bs]

Maybe you should remove the SKUNK debug code:

$5d40: Writing 0x37ea to 0xc00000

This is a SKUNK command register IIRC.

[42bs]

Wow, the C compiler produces horrible code:

 

| |    005DF2: 5380                     SUB.L #$1, D0
| |    005DF4: 4A80                     TST.L D0
| | -> 005DF6: 66E8                     BNE.B $5DE0 (true)

This hurts a Assembly-guy's eyes. :.(

[swapd0]

17 minutes ago, 42bs said:

Maybe you should remove the SKUNK debug code:

$5d40: Writing 0x37ea to 0xc00000

This is a SKUNK command register IIRC.

Yes, it's in my TODO list, I want to have two build versions, one with skunk debug code, the other one "naked"

Edited February 7, 2022 by swapd0

[agradeneu]

14 hours ago, XProger said:

 

It's truly shocking how wellinformed they are!  

[Chilly Willy]

Latest works here. Very nice for a slide show. ?  I can actually run around as long as I'm patient. Seems stable, and the rendering looks good.

[+selgus]

3 hours ago, 42bs said:

Wow, the C compiler produces horrible code:

 

| |    005DF2: 5380                     SUB.L #$1, D0
| |    005DF4: 4A80                     TST.L D0
| | -> 005DF6: 66E8                     BNE.B $5DE0 (true)

This hurts a Assembly-guy's eyes. :.(

That looks like debug-level code.. are you sure the optimizer was enabled when this code was compiled?

[swapd0]

6 minutes ago, selgus said:

That looks like debug-level code.. are you sure the optimizer was enabled when this code was compiled?

I'm using -Ofast flag.

 

I've seen that it's the getBuffer function from the skunk board code... so don't worry.

Edited February 7, 2022 by swapd0

[42bs]

2 minutes ago, swapd0 said:

I'm using -Ofast flag.

Is it a new GCC or the old 4.x.y one ?

[swapd0]

It's some "ataribrownart-prefix" version, IIRC 8.2 or something like that.

[42bs]

2 minutes ago, swapd0 said:

It's some "ataribrownart-prefix" version, IIRC 8.2 or something like that.

Really? And such a horrible code generation?

Even without optimization, the compiler should know that "sub.l #1,dn" will set the Z flag.

I checked an old GCC (Sourcery G++ Lite 4.4-52) and it does know about.

[42bs]

A quick check

extern int func(int);

int a()
{
  int s = 0;
  for(int i = 100; i > 0; --i){
    s += func(i);
  }
  return s;
}

=>

#NO_APP
        .file   "x.c"
        .text
        .align  2
        .globl  a
        .type   a, @function
a:
        movem.l #12320,-(%sp)
        moveq #100,%d2
        clr.w %d3
        lea func,%a2
.L2:
        move.w %d2,-(%sp)
        jsr (%a2)
        add.w %d0,%d3
        subq.w #1,%d2
        addq.l #2,%sp
        jne .L2
        move.w %d3,%d0
        movem.l (%sp)+,#1036
        rts
        .size   a, .-a
        .ident  "GCC: (Sourcery G++ Lite 4.4-52) 4.4.1"

With:-std=c99 -mc68000 -mshort -O -fomit-frame-pointer

[swapd0]

9 minutes ago, 42bs said:

Really? And such a horrible code generation?

Even without optimization, the compiler should know that "sub.l #1,dn" will set the Z flag.

I checked an old GCC (Sourcery G++ Lite 4.4-52) and it does know about.

That code is from the skunk code library and it's in asm, anyway I've removed the tst.l opcode.

[42bs]

13 minutes ago, swapd0 said:

That code is from the skunk code library and it's in asm, anyway I've removed the tst.l opcode.

Oh.Ok. So I need to send a big "sorry" to the gods of GCC.

[alucardX]

I was able to run the latest uploaded build from my new GameDrive. It is the best slideshow I have ever played on the Jag. I am seriously impressed that you got it running on the 68K and that it actually responds to control input at all, nice work!