sifadil / pcsx2-playground

Automatically exported from code.google.com/p/pcsx2-playground
2 stars 0 forks source link

Known: Star Ocean 3 crash #47

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago
Maybe, just maybe, we could identify what makes PCSX2 to crash with 
StarOcean3... It crashes in all versions of PCSX2 since 0.9.2 and in 
Playground builds too...

I'm using PCSX2 on XP SP3 x86...
StarOcean3 - SLUS_204.88

An error is Access violation (c0000005) in pcsx2.exe

Here's the log. It crashes after VU0ExecMicro
...
loadElfFile: cdrom0:\SLUS_204.88;1; CRC = 23A97857
CDGetToc Param[0]=0, Param[1]=1
Framelimiter rate updated (UpdateVSyncRate): 59.94 fps NTSC
Framelimiter rate updated (UpdateVSyncRate): 59.94 fps NTSC
IOP Recompiler data reset
vu0ExecMicro 0
_vuExecMicro: 00000000

Original issue reported on code.google.com by eliotfur on 18 Nov 2008 at 11:01

GoogleCodeExporter commented 8 years ago
Is it single-core or dual-core mode? There is a big difference.

Original comment by Zeydl...@gmail.com on 18 Nov 2008 at 11:50

GoogleCodeExporter commented 8 years ago
"It crashes in all versions of PCSX2 since 0.9.2" << Are you saying it worked 
before?
'Cause thats extremely unlikely.

Original comment by ramapcsx2 on 18 Nov 2008 at 11:54

GoogleCodeExporter commented 8 years ago
@Zeydlitz:
Single-core, dual-core, same thing... I've tried lots of combinations
 of options... VU0/VU1 interpreter mode crashes too...

@ramapcsx2:
Ofcourse not... I mean "since earliest version I have at the moment"... =^_^= 
I'm 
sure it didn't worked before...

Original comment by eliotfur on 18 Nov 2008 at 12:02

GoogleCodeExporter commented 8 years ago
Ok then, that rules out the last hope I had for this game.
Since we know that the EErec introduced in 0.92 has flaws it might have been 
working
on earlier revs.

Original comment by ramapcsx2 on 18 Nov 2008 at 12:10

GoogleCodeExporter commented 8 years ago
Hmm... is it standard behavior for games to try to run VU0 micro-programs from
address 0x000000?  I don't know the VUs very well, so maybe it's supposed to be 
that
way.  But at first glance I always figure anything with zeros is suspicious. ;)

Original comment by Jake.Stine on 18 Nov 2008 at 10:08

GoogleCodeExporter commented 8 years ago
From what I gather Tri-ace games are anything but normal. Using all kinds of 
nasty 
things like data encryption, obfuscation, hardcoded memory check - getting 
these 
games to work is equvielant to finding the holy grail asfaik.

Original comment by kakemons...@gmail.com on 18 Nov 2008 at 11:58

GoogleCodeExporter commented 8 years ago
Well they pulled it off with zsnes with star ocean on that. So anything can be 
done. 
Given enough patience and effort.

Original comment by jfre1...@gmail.com on 19 Nov 2008 at 5:41

GoogleCodeExporter commented 8 years ago
@Jake.Stine
There's nothing special about it... Maybe it's just a bug in logging... Seems 
that 
most games start from vu1ExecMicro 0

Here's a small part from Okami:
...
Recompiling COP2:002056D8:  vmaddaz.xyzw ACC,vf06,vf11z
Recompiling COP2:002056D8:  vmaddw.xyzw vf15, vf07, vf11w
vu1ExecMicro 0
vu1ExecMicro 0 (count=0)
_vuExecMicro: 00000000
vu1ExecMicro 0
vu1ExecMicro 0 (count=1)
_vuExecMicro: 00000000
vu1ExecMicro 0
vu1ExecMicro 0 (count=2)
_vuExecMicro: 00000000
...

Here's from Final Fantasy X:
...
Recompiling COP2:001775AC:  vwaitq
Recompiling COP2:001775AC:  vaddq.x vf05,vf00,Q
vu1ExecMicro 0
vu1ExecMicro 0 (count=0)
_vuExecMicro: 00000000
vu1ExecMicro ffffffff
vu1ExecMicro ffffffff (count=1)
_vuExecMicro: 000002a0

Original comment by eliotfur on 19 Nov 2008 at 11:52

GoogleCodeExporter commented 8 years ago
Oups... Now I see the difference...
VU0 != VU1

Original comment by eliotfur on 19 Nov 2008 at 12:05

GoogleCodeExporter commented 8 years ago
Too hard to fix now, maybe someday :)

Original comment by ramapcsx2 on 23 Nov 2008 at 2:02