Dracula X Chronicle stucking problem

[daniel229]

It stucks here when they are talking,but you can press START to skip it.

debug log(reame jpg to rar)

[unknownbrackets]

Hmm, hard to say. The log is from just letting it stay hung and quitting, right?

SAS seems over, but it could be somehow SAS related. Maybe more likely sceAtrac if it's waiting for the music to end or something...

-[Unknown]

[daniel229]

Yes.
The infolog report these when they are talking.
'''
22:10:717 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy
22:10:752 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy
22:10:852 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy
22:13:423 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy
22:16:058 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy
22:23:232 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy
22:25:501 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy
'''
comment return SCE_KERNEL_ERROR_ASYNC_BUSY; in that block,it will hang earlier.

[unknownbrackets]

Does it go away if you change __IoSchedAsync(f, id, 100); in sceIo.cpp to for example __IoSchedAsync(f, id, 1000);?

I'm supposing that async busy is a good thing, so if something is happening here to cause the operation to occur late and not hit async busy, that may be the issue. If that does help, it'd also be interesting if this helped:

https://github.com/unknownbrackets/ppsspp/compare/io-minor2?expand=1

It could also help to schedule io threads, since they currently are not scheduled (it just estimates time.)

-[Unknown]

[daniel229]

__IoSchedAsync(f, id, 1000); does not help,io-minor2 branch does not help either

[daniel229]

io-minor2 branch helps #6127 ,turn on BGM again,it would not hang.but not full fixes the issue. if you turn on BGM and not apply it, turn off it,then turn on again,then apply it,it still hangs.

[daniel229]

and fixes this #6557

[daniel229]

fixes #6189

[unknownbrackets]

Hmm, well, that implies that all of those are IO related. It still seems like this is IO related too.

What about an even higher number? Let's try something crazy like __IoSchedAsync(f, id, 100000);. Also, note that it appears like 3 times in sceIo.cpp.

-[Unknown]

[daniel229]

Still not help,just report more 45:57:071 user_main W[IO]: HLE\sceIo.cpp:1110 sceIoLseekAsync(5, 0, 0): async busy,change all 3 lines or just the line 1115 seems no difference.change to 10000000 bgm would be missing or delay and hangs early.

[unknownbrackets]

Well, could try changing other delays. If none of it helps it must not be related after all.

So change e.g.:

void __IoSchedAsync(FileNode *f, int fd, int usec) {

To:

void __IoSchedAsync(FileNode *f, int fd, int usec) {
   usec *= 10;

And:

void __IoSchedSync(FileNode *f, int fd, int usec) {

To:

void __IoSchedSync(FileNode *f, int fd, int usec) {
   usec *= 10;

It seemed like lseek was the last thing it usually did in the log before the async busy, but it could be something else.

-[Unknown]

[daniel229]

Not help either.even 100,1000

[daniel229]

Also fixes #5540

[daniel229]

Also fixes an invalid address issue and the camera issue in Danball Senki W #4786

[unknownbrackets]

Really. Wow, that seems to improve quite a few things then. Hmm... I should take another look at it, since I recall there were problems in some games, but maybe they were other bugs.

-[Unknown]

[daniel229]

Yes,I have try 20 or 30 times to reproduce the problem with the io-minor2 branch,still not happen.I think it's fixed.Usually would happen in 5 times

[daniel229]

That branch may cause invalid adresses in Tales of the World Radiant Mythology 2

[daniel229]

That branch makes World Neverland laggy.

[daniel229]

Don't Register sceAtrac3plus() can avoid the stucking.#6804

[daniel229]

JPCSP fixed it in https://code.google.com/p/jpcsp/source/detail?r=3629 ,

[daniel229]

Fixed in #8189