Toeic Test Portable white screen after menu.

[daniel229]

debug log
https://gist.githubusercontent.com/daniel229/d373351509bcab5d33e8/raw/Toeic%20Test%20Portable%20white%20screen

[daniel229]

Turn off BGM in game than it's working.

[sum2012]

How to turn off BGM in game as it is white screen ?

[sum2012]

Compare log of JPCSP
https://gist.github.com/sum2012/fbbbc33c8e3a46e18234
JPCSP have this warn log,but PPSSPP don't call sceKernelNotifyCallback

03:50:52 WARN hle.ThreadManForUser - user_main - hleKernelNotifyCallback(type=0) thread:'user_main' overwriting previous notifyArg 0x32 -> 0x32, newCount=2

[unknownbrackets]

I think JPCSP probably uses hleKernelNotifyCallback for any change to a callback, including callbacks generated by things like IO or UMD events.

But, considering that is right after opening an at3 file, and disabling bgm solves it, it sounds likely to be related.

-[Unknown]

[daniel229]

@sum2012 Use the save that turn off BGM.
(rename jpg to rar)

[sum2012]

PPSSPP v1.0.1-236-g1a384da log:
https://gist.github.com/sum2012/0c18847c7801798c0aae
JPCSP log:
https://gist.github.com/sum2012/7a640a36b1350e72d673

I see this from JPCSP log.......
12:01:44 DEBUG hle.sceAtrac3plus - thread atrac - sceAtracSetDataAndGetID buffer=0x09CC27C0, bufferSize=0x4E7E4

12:01:44 DEBUG hle.sceAtrac3plus - thread atrac - FileSize 0x4E7E4

12:01:44 DEBUG hle.sceAtrac3plus - thread atrac - WAVE format: magic=0x20746D66('fmt '), chunkSize=52, compressionCode=0xFFFE, channels=2, sampleRate=44100, bitrate=16021, bytesPerFrame=0x2E8, hiBytesPerSample=0, codingMode=0

12:01:44 DEBUG hle.sceAtrac3plus - thread atrac - Additional chunk data: 22 00 00 08 03 00 00 00 BF AA 23 E9 58 CB 71 44 A1 19 FF FA 01 E4 CE 62 01 00 28 5C 00 00 00 00 00 00 00 00

12:01:44 DEBUG hle.sceAtrac3plus - thread atrac - FACT Chunk: endSample=881703, sampleOffset=2232, size=12

12:01:44 DEBUG hle.sceAtrac3plus - thread atrac - DATA Chunk: data offset=0x64, data size=0x4E780

12:01:44 DEBUG hle.sceAtrac3plus - thread atrac - hleSetHalfwayBufferAndGetID returning atID=0x0

[unknownbrackets]

Any change here? Does #8105 help?

-[Unknown]

[daniel229]

Unfortunately not.

[sum2012]

The modify log
https://gist.github.com/sum2012/df683124947a3f8bcf4b

[unknownbrackets]

Per the debug log:

12:50:805 thread funct D[KERNEL]: HLE\sceKernelEventFlag.cpp:251 293=sceKernelCreateEventFlag("event atrac", 00000200, 00000000, 00000000)
13:01:105 thread atrac D[KERNEL]: HLE\sceKernelEventFlag.cpp:336 sceKernelSetEventFlag(293, 00000001)
13:01:106 thread wave  D[KERNEL]: HLE\sceKernelEventFlag.cpp:336 sceKernelSetEventFlag(293, 00000002)
13:01:107 user_main    D[KERNEL]: HLE\sceKernelEventFlag.cpp:532 sceKernelWaitEventFlagCB(293, 00000003, 32, 09fff510, 00000000): waiting
13:01:107 user_main    D[KERNEL]: HLE\sceKernelEventFlag.cpp:210 sceKernelWaitEventFlagCB: Suspending lock wait for callback
13:01:107 user_main    D[KERNEL]: HLE\sceKernelEventFlag.cpp:221 sceKernelWaitEventFlagCB: Resuming lock wait from callback
** everything stops **
13:55:344 idle0        I[KERNEL]: HLE\sceKernel.cpp:513 KO 293: EventFlag "event atrac": init=00000000 cur=00001000 numwait=0

Wait 32 means "OR", not that it matters since 1 and 2 are both set anyway. Anyway, this is a bug in callback handling I think. Fixing.

-[Unknown]

[sum2012]

v1.1.1-101-gc14d1f7
Now it is not hang but display this screen everytime

info log:https://gist.github.com/sum2012/ebbd24f7b1f0413c4954
debug log:https://drive.google.com/file/d/0B3OaSdeV0L8kOUR2YzJQZlM5ejg/view?usp=sharing

[sum2012]

Unfortunately @unknownbrackets 's JPCSPtrace log not work

[unknownbrackets]

Looks like this is doing a wait inside a callback, which we don't handle well still...

-[Unknown]

[daniel229]

At least,it's playable now,I don't see any errors in game,and may help other games of this series,good job.@sum2012 @unknownbrackets

[unknownbrackets]

Just for context, SCE_KERNEL_ERROR_WAIT_DELETE is that error code. It's different from SCE_KERNEL_ERROR_WAIT_CANCEL which should happen for sceUmdWaitDriveStatCB.

My theory is that this is caused by nested callback waits (CB), which don't work exactly right. Callbacks can be a bit tricky.

-[Unknown]