Sol Trigger : MIPSInt BREAK!

[dbz400]

Just for record . Break with _sceAtracGetContextAddress() . It is a known issue previously .May be oioitff has some idea since he already implemented _sceAtracGetContextAddress() recently .

[unknownbrackets]

I recently found I have a game hitting a problem with the same func, possibly related - Mad Blocker Alpha.

Still need to write tests for this func...

-[Unknown]

[dbz400]

Indeed , not many games use this function _sceAtracGetContextAddress() .

[oioitff]

I guess those games need more detail infos for atracContext or sasVoice. I'm not familiar to write PSP tests, but I had written some psuedo codes in #2313 to dump detail infos for atracContext and sasVoice. Hopes it may help something.

[kension]

There's a "new" bug in the 0.76-1422 build (that merged #2313) during a certain enemy encounter (that caused freezing some time during the fight). This bug is still present in 0.8-54. This bug seems to be not present in build 0.76-1414.

[solarmystic]

@kension

Confirmed, getting that no free ID bug in Sol Trigger on the latest git revision (0.8-54)

EDIT:- It seems to happen even more if you skip the cutscenes manually. If you dont skip them, the game will function as intended.

EDIT2:- Moving this comment to an issue page since it's a different one and unrelated to the MIPS Break! issue

[unknownbrackets]

I don't think this is unrelated. They're both related to how atrac IDs are being managed by the game.

-[Unknown]

[solarmystic]

@unknownbrackets

Fair enough, thought it was worth a go for the added exposure.

[unknownbrackets]

@oioitff sorry, haven't done much with this yet, but initial results seem positive. It's not printing much yet but here's some of the output:

[x] Init
[r] Load avcodec: 00000000
[r] Load sascore: 00000000
[r] Load atrac3plus: 00000000
[x] __sceSasInit: 00000000
[r] Loaded sample.at3 data.
[r] Got atracID: 00000000
[x] Valid AtracID? 00000000
[x] Context: 09d3dbc0
[x] decodePos=84936193, endSample=0, loopStart=0
[x] loopEnd=1619584, samplesPerChan=102400, numFrame=1
[x] state=0
[x] __sceSasSetVoiceATRAC3: 00000000
[x] Valid AtracID? 00000000
[x] __sceSasSetKeyOn: 00000000
[x] __sceSasGetEndFlag: ffffffff
[r] __sceSasGetEndFlag (after __sasCores): ffffffff
[x] Valid AtracID? 00000000
[x] __sceSasUnsetATRAC3: 00000000
[x] Valid AtracID? 00000000
[x] sceAtracReleaseAtracID: 00000000

It's possible the sample I'm using isn't valid...

Also, my example makes PPSSPP crash in __AtracUpdateOutputMode(). pCodecCtx is NULL. I must be doing something wrong...

-[Unknown]

[oioitff]

Hmm, I just find out that it require state = 0x10 when atrac3 voice is finished. Maybe I could give out a solution later if everything goes fine :)

[unknownbrackets]

I tried calling the lowlevel func:

    SceAtracLowLevelParams params;
    memset(&params, 0, sizeof(params));
    params.channels = 2;
    params.outputChannels = 2;
    params.bufSize = 0x2000;
    checkpoint("sceAtracLowLevelInitDecoder: %08x", sceAtracLowLevelInitDecoder(atracID, &params));

But it just failed:

[x] sceAtracLowLevelInitDecoder: 80630001

Anyway, cool.

-[Unknown]

[oioitff]

Thanks.
By the way, the params.bufSize should be the atracbytesPerFrame,such as 0xC0, 0x180, 0x230 and so on...

[unknownbrackets]

Oh, I think I got the struct wrong, oops...

-[Unknown]

[solarmystic]

#2466 fixed it. Thanks @oioitff

Only one small "issue" remains that does not affect the game's playability, if you're a cutscene skipping fiend in this game, characters voices in ingame cutscenes will still get muted if you skip them often enough.

They do not return until you reset the game in your next playthrough.

Do not skip the cutscenes if you wish to get full functional character voices for now.

Could be related to Danganronpa's character audio problem if you skip the dialogue boxes too fast. #1686 (comment)

That's the only one left.

[unknownbrackets]

Hmm, sounds like it's still running out of atrac IDs, but at least it's running. Reopening.

-[Unknown]

[CPkmn]

@solarmystic Danganronpa voices work fine when I skip dialog boxes (both in the demo and full version).

#2466 seems to fix God Eater Burst's sceKernelPrintf's after _sceAtracGetContextAddress().

(Don't mind the "The system cannot find the path specified." - I used my own modified sceAtrac and I added support to dump and change atrac audio; it's just reporting that it can't find a replacement audio file since I'm not using any for that audio).

[unknownbrackets]

@oioitff okay, now that I didn't make a stupid mistake (I had used SceAtracIdInfo), it went a lot better:

[x] Init
[r] Load avcodec: 00000000
[r] Load sascore: 00000000
[r] Load atrac3plus: 00000000
[x] __sceSasInit: 00000000
[r] Loaded sample.at3 data.
[r] Got atracID: 00000002
[x] Valid AtracID? 00000000
[x] Context: 09d3ddc0
[x] decodePos=1093, endSample=1691647, loopStart=0
[x] loopEnd=0, samplesPerChan=1093, numFrame=0
[x] state=4
[x] __sceSasSetVoiceATRAC3: 00000000
[x] Valid AtracID? 00000000
[x] __sceSasSetKeyOn: 00000000
[x] __sceSasGetEndFlag: ffffffff
[r] __sceSasConcatenateATRAC3 (curOff=444): 80420042

It doesn't like __sceSasConcatenateATRAC3() though... hmm.

-[Unknown]

[oioitff]

@unknownbrackets Hmm, what does error 80420042 mean? By the way, "__sceSasGetEndFlag: ffffffff" means there is no voice playing?

[unknownbrackets]

No idea what that error code means yet. Could be related to yes - seems like the voice is not considered playing. I'm not sure what it's missing other than keyon...

-[Unknown]

[oioitff]

Maybe it needs to call sceSasSetPitch after __sceSasSetVoiceATRAC3?

[unknownbrackets]

Does this still hit the break?

-[Unknown]

[dbz400]

No more break hit now .I'm close it now then.