MotorStorm Arctic Edge : game hangs after the sony logo

[hdd60311]

Test 0.9.1-15, after logo page it got black screen and no response.

log : http://www.mediafire.com/?affmmid58a23vfm

[unknownbrackets]

This game uses a not-yet-supported relocation type. I have an idea how to implement it, but I wanted to actually test it in some game.

http://report.ppsspp.org/logs/game/UCES01250_1.01
http://report.ppsspp.org/logs/game/UCKS45124_1.00

Can you try this version? It may help or do nothing:

http://temp-share.com/show/gFHcdyG3Y

It adds this for R_MIPS_16: op = (op & 0xFFFF0000) | (((int)(op & 0xFFFF) + (int)relocateTo) & 0xFFFF);

-[Unknown]

[hdd60311]

@unknownbrackets
I tried it and still don't work.

[hdd60311]

Tested with 0.9.1-619, it can run after manufacturer logo but title video is blackscreen, then getting into main menu page the background image is losed and bgm doesn't play, finally it hangs at loading page.
http://i765.photobucket.com/albums/xx296/hdd60311/screen00000_zpsf23c40df.jpg
http://i765.photobucket.com/albums/xx296/hdd60311/screen00001_zps20595a3b.jpg
http://i765.photobucket.com/albums/xx296/hdd60311/2_zps138ce6f1.jpg

log : http://www.mediafire.com/?41rsc77c8vuc08g

[scraggles80]

it is an atrac3+ problem when it hangs at the loading screen
this started to work in Jpcsp svn 3339
Improved sceAtracGetRemainFrame for small atrac3+ files.
Improved compatibility of sceAtracDecode: the PSP is returning a lower number of samples at the first call to sceAtracDecode.
and the no videos is a scempeg problem

[dbz400]

I just tried to disable atrac3+/audio in ppsspp , it still hangs at the loading screen.

[hdd60311]

I tried JPCSP 3375 to test MotorStorm but it also hangs at the loading screen. ( I use ATI card, don't know nvidia card can run it or not.)　

[scraggles80]

goes ingame for me using svn 3375 windows 64 bit and i have an ATI card
must have use media engine and decode audio files with sonic stage ticked

[hdd60311]

Alright I tried it with sonicstage again and can work now, wonder the improvement of JPCSP 3339 can be transplanted to PPSSPP or not?
@shenweip
@CPkmn

[unknownbrackets]

Definitely the first frame returns less samples, but it varies. I saw, all in ATRAC3+, 0x2CA, 0x690, 0x341, and 0x280. For ATRAC3, I saw 0x3BB.

-[Unknown]

[hdd60311]

http://www.emunewz.net/forum/showthread.php?tid=1826&pid=14916#pid14916
seems JPCSP fixed the video playback in r2135

[hdd60311]

Have progress with the stuck issue at loading screen?

[asorcinelli]

I've tested the last version on Android and the stuck issue is present. After the loading screen the emulator hang.

[hdd60311]

Tested latest build seems new ffmepg atrac3+ doesn't help this.

[hdd60311]

The graphic gets worse since 0.9.5-94 with this pullrq #4399

[dbz400]

Should be fixed by latest build now ( I mean the graphic gets worse issue)

[hdd60311]

Yep it has been fixed.

[hrydgard]

I reverted the fix though as it broke other things.

[hdd60311]

debug log by @sum2012
https://drive.google.com/file/d/0B3OaSdeV0L8kU0lfVUltQjA5Nk0/edit?usp=sharing

[sum2012]

@hdd60311 you can also produce debug log by run DebugLog.bat
edit:Just read tiebie forum know that you cannot run DebugLog.bat

[hdd60311]

#3338 (comment)
According to this the video starts working in latest build but still cannot in-game.

[sum2012]

PPSSPP v0.9.6-406 debug log
https://gist.github.com/sum2012/6820cb48f5b322b5a3a8

Not sure sceIoWrite have problem
31:25:981 idle0 D[IO]: hle\sceio.cpp:899 35=sceIoWrite(1, 08aba080, 35)
31:25:983 idle0 E[CPU]: mips\mipsint.cpp:184 BREAK!

[unknownbrackets]

No, sceIoWrite wrote this:

stdout: Error: pthread API call made from non-pthread thread

It's almost definitely because it's within an interrupt (Entering interrupt handler 08976f70). Probably some syscall needs to return an error when in an interrupt and isn't currently.

For example, maybe sceKernelReferThreadStatus or sceKernelSetEventFlag need the HLE_NOT_IN_INTERRUPT flag (I think I tested the latter though already, not sure...) Possibly sceKernelGetThreadId(), which I have not tested. I need to test more of those...

-[Unknown]

[sum2012]

In sceKernal.cpp and sceKernalnterrupt.cpp there are two
{0x293b45b8,WrapI_V, "sceKernelGetThreadId"},
Do they are the same ?

edit:I find it one is ThreadManForUser another is Kernel_Library

[unknownbrackets]

Right. You can add it to both. In practice, the kernel one is never called except by Chinese-translated games and cfw plugins. Mostly they don't work anyway (which is why I haven't worked on my cfw plugins branch more.)

-[Unknown]

[sum2012]

In sceKernal.cpp ,add sceKernelGetThreadId HLE_NOT_IN_INTERRUPT flag can move a bit
video:https://drive.google.com/file/d/0B3OaSdeV0L8kakRsTWNLQlhIaDQ/edit?usp=sharing
Info log: https://gist.github.com/sum2012/090f8a55dbe6492c07cc

[dbz400]

Humm i did try to use our latest 412 here and also can go upto the following screen and no need to set the above flag.

[sum2012]

@raven02
Yes,i forget Press start to enter.
But the log full of stdout: Error: pthread API call made from non-pthread thread

[hdd60311]

#5150 (comment) graphic glitch is fixed now.
http://www.youtube.com/watch?v=wJrDTPAYqAs
According to the psp gameplay there should have the bgm at main menu page but emu doesn't play it, maybe is also the atrac3+ problem.

[unknownbrackets]

Okay, actually, I think SCE_ERROR_AUDIO_CHANNEL_ALREADY_RESERVED might mean that the channel is busy (e.g. it's being output/read from), so it can't be released when returned from Release.

-[Unknown]

[sum2012]

I go out to work now.
Will come back home in 12 hours

[sum2012]

@unknownbrackets The game do not call sceAudioSRCChRelease / sceAudioSRCChReserve
JPCSPTrace log
https://gist.github.com/sum2012/d1bda3f48142fe167c10
PPSSPP v0.9.8-441-g669dc74 sceaudio log:
https://gist.github.com/sum2012/9578235fe5bbcad51a2f

@hrydgard Yes, after load the save, real psp have a click sounds while ppsspp freeze

[dbz400]

I think the description below is wrong , should be "channel not reserved" ?

u32 sceAudioSRCChRelease() {
if (!chans[PSP_AUDIO_CHANNEL_SRC].reserved) {
ERROR_LOG(SCEAUDIO, "sceAudioSRCChRelease() - channel already reserved ");
return SCE_ERROR_AUDIO_CHANNEL_NOT_RESERVED;
}

[unknownbrackets]

Yes, that looks like a typo.

-[Unknown]

[dbz400]

Yep , fixed here #5931

[hdd60311]

@sum2012 any new progress about this?

[sum2012]

@hdd60311 No finding.

[hdd60311]

The image gets wrong size since 0.9.8-1012 #6211

[unknownbrackets]

Debug log?

-[Unknown]

[ppmeis]

Confirmed Issue:

Debuglog: http://goo.gl/U94I0g

[unknownbrackets]

512x296, region : 512x296 , scissor: 512x296, stride: 512

Hmm, well, it would seem like the framebuffer is an odd, large size. Even so, I thought it would display the top left of it... theoretically it should not matter.

-[Unknown]

[hdd60311]

in latest build the wrong size trouble is fixed, still hangs at loading page.

[unknownbrackets]

So it was probably like that Airplane game, I guess. But sounds like the hang is likely related to sceSas, or audio somewhere.

-[Unknown]

[sum2012]

v0.9.9.1-370-g31ce0e7 debug log:
https://drive.google.com/file/d/0B3OaSdeV0L8kNlNLNElRaUFYX00/edit?usp=sharing

[daniel229]

The hanging and the missing BGM I think they are an atrac3p issue,and disable atrac3p also works.
JPCSP fixed missing bgm and the hanging in https://code.google.com/p/jpcsp/source/detail?r=3432

[unknownbrackets]

So, if I understand correctly, this no longer hangs, but the music doesn't play correctly, right?

-[Unknown]

[daniel229]

Yeah,no loger hangs,and the title music is back too,but during the match all the sounds are gone.
debug log
(rename jpg to rar)

[hdd60311]

It has some graphic issue, like abnormal shadows under cars and seems over bright.

[hdd60311]

And seems that the hanging issue is not really fixed, first time it can go into match, but if quit match
and go into match again it still hangs at loading page

[hdd60311]

Alright I got the point that seems strange, I tried to change cpu clock like this image
http://forums.ppsspp.org/attachment.php?aid=12863
and the hanging issue is gone, but if use default cpu clock setting the hanging issue is still there...

[hdd60311]

http://forums.ppsspp.org/showthread.php?tid=430&pid=95518#pid95518
According to this report switching to read framebuffers to memory and then back to buffered rendering fixes the bloom temporarily.

[ppmeis]

Tested with latest build. Bloomy still present. But now music in menus are fine. When you enter to the race, all sound stops, just like before.

[daniel229]

The shadow in Gedebugger

[hrydgard]

Closing in favor of #7445 as much of this report isn't relevant anymore, but the sound part, also mentioned there, is (and it still seems possible to trigger some hangs, plus overbloom).