Mahou Shoujo Madoka Magica Portable in-game problem

[sum2012]

Load the save
https://drive.google.com/file/d/0B3OaSdeV0L8kb0ZRRTN2amltdDA/edit?usp=sharing

after die,W[MM]: MemmapFunctions.cpp:93 ReadFromHardware: Invalid address 00000000

v0.9.8-1614-g459e244
info log:
https://gist.github.com/sum2012/654b7f4370045f5fa286
debug log:
https://drive.google.com/file/d/0B3OaSdeV0L8kcGRBOER1NXBhcEU/edit?usp=sharing

[unknownbrackets]

Has it ever worked? Does it work in JPCSP?

-[Unknown]

[daniel229]

PPSSPP always hangs there,0.7 to the latest,and for JPCSP,the game go to intro and playable since this https://code.google.com/p/jpcsp/source/detail?r=2268 ,and does not has problem there.

[unknownbrackets]

Hmm, interesting. Well, I asked because I found out Gods Eater Burst also does not load real PSP savedata correctly, so I was wondering if it was related. But now I think I'm understanding that the savedata is just to get to the problem, and the problem is there either way.

Unknown GetPointer 00000000 PC 00000000 LR 0889f2ac

Seems bad. What's at 0889f2ac?

Might possibly be sas or dispatch thread related...

-[Unknown]

[daniel229]

[unknownbrackets]

Hmm, so the pointer at t9 is becoming 0. Most likely we're either corrupting memory there, not scheduling properly, and not writing something there we ought to...

-[Unknown]

[daniel229]

By the way,this is not a savedata issue.

[unknownbrackets]

So it seems this is io delay related, or possibly async io related.

-[Unknown]

[unknownbrackets]

Reported not fixed by the async IO changes in #7636.

-[Unknown]

[Panderner]

Simulated UMD delays helps it to prevent crashes

[sum2012]

Update savedata to easier to reproduce.
https://drive.google.com/file/d/1wvDHLaYCReY0wmH6DZscW1hoWN3kcN4n/view?usp=sharing
v1.10.3-1639-g4eee5df2a
bad log: https://drive.google.com/file/d/1vOw7m5_kpIYehpA5dc9PnoXB1GI-83L-/view?usp=sharing
Simulate UMD delay good log: https://drive.google.com/file/d/1c03HMxkfaqopsUxKrT8QbnFz_H1knTnu/view?usp=sharing

The game maybe related to I\O add sceKernelUnlockMutex,sceKernelLockMutex staff

[sum2012]

@unknownbrackets There is two case that we need reschedule (from jpcsp)
https://github.com/jpcsp/jpcsp/blob/master/src/jpcsp/HLE/kernel/managers/MutexManager.java#L121
https://github.com/jpcsp/jpcsp/blob/master/src/jpcsp/HLE/kernel/managers/MutexManager.java#L155

edit:We have already reschedule in sceKernelUnlockMutex ,hmm

[sum2012]

@unknownbrackets Can you try translate it ? seem ppsspp don't have it. https://github.com/jpcsp/jpcsp/blob/master/src/jpcsp/HLE/kernel/managers/MutexManager.java#L154

[sum2012]

v1.13.1-603-g59cb7a075
Still need Simulated UMD delays

[unknownbrackets]

@unknownbrackets Can you try translate it ? seem ppsspp don't have it. https://github.com/jpcsp/jpcsp/blob/master/src/jpcsp/HLE/kernel/managers/MutexManager.java#L154

Sorry, I think I never saw this before. PPSSPP does do that, here:

ppsspp/Core/HLE/sceKernelThread.cpp

Line 3027 in 80e481b

__KernelChangeReadyState(this, GetUID(), true);

-[Unknown]

[sum2012]

I was too deep to jpcsp's code ,don't think the game's code itself

[sum2012]

Reopen as it is wrong change