Vamiga: Sword of Sodan graphics errors

[Alessandro1970]

vAmiga hangs as soon as this disc is inserted, but the disk works with other emulators.
I tried with or without sloram etc. but the incompatibility starts as soon as this disc is inserted.
It looks like nothing is loaded, either it's a format problem or something in the bootblock.
Insert in DF0: do not load from WB.

Report:
report.txt

Adf:

Simple text_test.adf.zip

[dirkwhoffmann]

You did run into an intentional crash in function pokeDSKSYNC:

void
DiskController::pokeDSKSYNC(uint16_t value)
{
    assert(false);
    debug(DSK_DEBUG, "pokeDSKSYNC(%X)\n", value);
    dsksync = value;
}

The DSKSYNC register is not implemented properly yet. Because of this, I decided to let the emulator crash (doing so let's me collect test-cases using this register). I'll get back to this when DSKSYNC has been implemented.

[dirkwhoffmann]

Fixed in http://www.dirkwhoffmann.de/vAmiga/vAmiga_0.23.zip

[Alessandro1970]

If I insert Sword of Sodan disk 1 it hangs:
report.txt

Sword of Sodan (1989)(Discovery Software)[cr MFC][t +2 MFC](Disk 1 of 4).adf.zip

[dirkwhoffmann]

  for (int i = currentPixel; i < 2 * agnus->diwHstop; i++) {
        assert(i < sizeof(rasterline));
        rasterline[i] = openR;
    }

😯 This game writes an unusually high value into DIWSTOP (right side of display window). I'll look into it ...

[dirkwhoffmann]

No more write accesses outside the visible screen area 😃:

Yeeeeeeessssss, I remember that one. Horizontal scrolling at it's best 😎:

Oh no... 🙈

[Alessandro1970]

With release 0.29 the intro menu is not more visible:

[Alessandro1970]

...the game starts pressing some Fx keys but there are some graphic errors in the upper left corner:

[dirkwhoffmann]

Intro screen got broken by commit b18a277 (July 1st)

[dirkwhoffmann]

Intro is good again 😎.

When the intro screen comes up, DIWSTRT and DIWSTOP are set to unusual coordinates ($00 and $FF, respectively) and trigger an "out of bounds" condition in vAmiga. Removing the check solves the issue. However, I think that I added the bound check because some vAmigaTS test was failing without the check. I need to doublecheck all DIW related test cases and probably write more of them.

The game itself seems to work fine now:

A little bit bloody though 😬

Anyway, let's see what comes next 🤔

Oooups 🙄

Hmmm, all games I tested so far seem to have gotten more difficult over time. Probably some kind of mysterious ageing issue with ADF files? 🤔

[Alessandro1970]

Could you let us to download a the new beta release ?

[dirkwhoffmann]

Could you let us to download a the new beta telease ?

Coming soon. I want to walk through all existing vAmigaTS tests first. To implement display hierarchy (sprites before or behind playfields), I had to overhaul the complete graphics backend. Unfortunately, this has broken some existing functionality.

BTW, did you get back the recapped A2000 mainboard yet?

[Alessandro1970]

I have sent my A2000 MoBo to a repair center...
At the moment they have not told me anything.
I have to wait...

[dirkwhoffmann]

Findings:

DIW settings in the intro:

[1070] ( 12, 24)  0601A8 BCB-D- 606C 0000 [074014] Agnus: pokeDIWSTRT<Copper>(2801)
[1070] ( 12, 26)  0601A8 BCB-D- 606C 0000 Agnus: setDIWSTRT(2801)
[1070] ( 12, 26)  0601A8 BCB-D- 606C 0000 Agnus: newDiwVstrt = 40 newDiwHstrt = 1
[1070] ( 12, 26)  0601A8 BCB-D- 606C 0000 Agnus: newDiwHstrt is too small
[1070] ( 12, 28)  0601A8 BCB-D- 606C 0000 [074018] Agnus: pokeDIWSTOP<Copper>(30FF)
[1070] ( 12, 30)  0601A2 BCB-D- 606C 0000 Agnus: setDIWSTOP(30FF)
[1070] ( 12, 30)  0601A2 BCB-D- 606C 0000 Agnus: newDiwVstop = 304 newDiwHstop = 511
[1070] ( 12, 30)  0601A2 BCB-D- 606C 0000 Agnus: newDiwHstop is too large

According to my model, only the bg color can be drawn. Indeed, when the particular values for DIWSTART and DIWSTOP are replicated in a test case, the screen remains black in SAE (as predicted by the model).

TODO: Check other parameters such as the DDF settings and replicate them in the test case.

[dirkwhoffmann]

Added tests diw9, diw10, diw11 to vAmigaTS repo.

The three tests provide strong evidence for my hypothesis that the DIW H flip-flop is automatically set at the beginning of each frame. I am pretty sure this fixes the Sword of Sodan intro bug. We'll see once it's implemented...

[dirkwhoffmann]

Fixed in V0.32