Resolved: [Bug Report] Godzilla - Save the Earth [PAL]

[Typhox]

Yes, but It's enough to change only the second CPU_INT to get the game working again.

[Typhox]

So...
Now I found out how to use logging and logged the Game/Emu from imediately before the crash until the emulator crashed. I attached the result as emuLog.txt. I also attached an image of the logging-settings I used.
Maybe it will be easier now to fix the bug, maybe it's not.
I don't know.

[Typhox]

As I wrote above, this game does not work with 0.9.6 without the slight modification of the source code.
With the modification it goes very well through intro, menus and loadingscreen but crashes after that. r650 does not bring any difference.
btw it is te european version of the game.

[rama]

Ok, the reason it works is because VIF timing is hard to get right. This change would break other games for sure.
What it does for this game is mask a problem so it avoids it. The real issue is somewhere else.

We know about this and there are plans on how to fix it correctly (not just VIF btw ), but it'll take time.

[Typhox]

Sooo...
The game can be rated as ingame because you can sometimes view ingame-graphics and i must say that it runs at a very good framerate.
It crashes if you play with one or two monsters but with three and four there is no problem. Then the game crashes when there are special light effects on the screen.
I am very sure that it crashes through the light effects. When you play alone or with one computer you have another camera view at the beginning than if you play with a total amount of three or four monsters.
I tested it with newest gsdx (514), spu2-x (r527) and pcsx2-dev (r714)
I tried enough so that I can say: I am very sure it has something to do with something that happens when there are special lights.

[Typhox]

I tried the non-modified version of r781-dev again. There is now an (error-)message in the console.

Framelimiter rate updated (UpdateVSyncRate): 50.0 fps
vif1.cmd still set 6c
sceGsSyncPath: VIF1 does not terminate

<D1_CHCR=000000551_TADR=019000001_MADR=000000001_QWC=00000000>

<D2_CHCR=000000012_TADR=000000002_MADR=0065ff702_QWC=00000000>

<VIF1_STAT=00000041:GIF_STAT=00000004>

UNKNOWN VifCmd: 34
IOP Recompiler data reset
iR3000A Resetting recompiler memory and structures

Maybe I can find out more if you tell me what.

[Typhox]

One of the access violations is gone with one of the revisions today.
I updated the first post.

[Typhox]

I fixed the problem myself.

The problem is there (iVUzerorec):

Code:

#ifdef SUPERVU_VIBRANCHDELAY
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
///// NOTE! This could still be a hack for KH2/GoW, but until we know how it properly works, this will do for now.///
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

    if( hasSecondBranch && firstbranch != 0xff ) { //check the previous jump was conditional and there is a second branch
#else
    if( hasSecondBranch) {
#endif

it has to be just

Code:

if( hasSecondBranch) {

It doesn't crash then anymore and runs very good (playable).
It still doesn't run without x2 cycle-rate-speedhack. It shows "Unknown VifCmd: 3f" then.

[Bositman]

Requesting a status updated by using microVU and the latest beta, since it looks like a SuperVU issue.

[Typhox]

It does work under beta 1888 with micro VU enabled. However, it runs slower than it did with SuperVU and my correction. And it has a few fast-hiding graphic bugs.