PPSSPP 1.1 release - what to fix before?

[hrydgard]

I think it's soon time to push out another official release. The emulator seems to be in a fairly stable state, there are a couple of important features I want to get in first though.

• Ability to delete save state
• Enable the homebrew "store", even if it's still a bit ugly
• Non-"fast memory" on 64-bit ARM is broken
• Tenchu San Portable major issues! #7234
• Homebrew states naming : Homebrew states naming #7795
• UI text regression : UI text regression #7884
• Regression: Breath of Fire III now jumps up and down rapidly since a3078a6ae #7974 (Breath of Fire regression)
• Texture glitch in CGI video #7939 (not a bug)
• XP: game specific configs ignored in recent build ? #7967 (fixed, awaiting tests)
• Version string corrupted #7991
• OpenGL Nvidia GPU Driver crashes while changing to fullscreen mode #7957 (fullscreen crash)
• Broken translations #8001
• Auto rendering resolution is broken #8002
• GTA LCS,VCS and CHW Radio loading delays still there #7863 (GTA loading delays)
• PSPKVM?

As usual, of course, add any bad regressions to this thread. That is, things that used to work, but don't anymore.

[vsub]

For regression...at least here,the game will ALWAYS crash when you start a new game and go talk to the principal and when he say
"Ah,Killian.Just the person I wanted to see."
if you press X to continue,ppsspp will either freeze or crash.

v1.0-83-g1ecfd6c is the last working version

[Bigpet]

@vsub you should probably mention which game you are talking about and if you have already opened an issue mention the issue number

[vsub]

Sorry about that...I'm talking about Crimson Gem Saga and as I said v1.0-83-g1ecfd6c is the last working version and I tried a lot of different newer revs including the latest 1.0.1-551 x86(default settings)
And no,I didn't open an issue...I just report it in the forum

[unknownbrackets]

For context: http://forums.ppsspp.org/showthread.php?tid=1243&pid=106178#pid106178

I reproduced this issue with replacements off, but it seems mostly triggered by the Jak & Daxter changes. Also, I was able to get it to work even with the latest changes. So it's a timing issue and not 100% of the time, which makes being sure it was caused by anything annoyingly hard.

-[Unknown]

[vnctdj]

Ah, the homebrew store :)
It makes me think of an issue...
When we save a state in a game, PPSSPP creates 2 files (a .jpg file and a .ppst file) which are named as the following : "UMD serial" + "_#".
Given that each game has an unique UMD serial, there is no problem :)
But homebrews don't have an UMD serial, therefore when you save a state with homebrew example1 PPSSPP names the 2 files "_#". And when you save a state with homebrew example2, PPSSPP also names the 2 files "_#".
It means that all homebrews will share the same states, which is incoherent...
We have to solve this issue : how should we name the 2 files created when we save a state in a homebrew ?

Edit : created a proper issue >> #7795

[asbel123]

and new version of ppsspp is 64 support for new andriod phone update too please i have to use fast mermoy on samsung galaxy s6 edge without fastmermory made graphcis problem u cant see the game. please fix it for new version. i will be happy ur emulator is a best now. it kills pcsx2 emulator now and try opengl on pcsx2 it same is pppsspp problem is cloud issues on the game that mean ps2 emulator coming to andriod soon.

[hrydgard]

Yes, ARM64 support is working and will be in 1.1.

What game are you talking about with graphics problems?

[iOSSOi]

Please, please fix this issue #7302 before the 1.1 release

I think i,ve tested every build since early January, it still crashes.

[asbel123]

look final fantasy 7 crisis core and kingdom hearts birth by sleep without fast mermory on samsung galaxy s6 edge u get green screen and when i turn it on fast mermory fix the graphics problem on 3d games no 2d game problem just 3d game issues. i think 64 bit issues on new phones

[hrydgard]

Oh, right, indeed non-fast memory is broken on 64-bit. Thanks, I'll add it to the list.

[FibYar]

It would be great if you could fix "Rewrite". It was broken in 0.9.8:
#6981

[ofry]

Maybe fix "sceAtrac" issue in Silverfall?

#7708
#7709

[ofry]

And maybe add xBRZ 1.3 ?

#7651

[jbeich]

Nominating #7717 as more distros start to use GCC 5.x by default. It should probably be done after update in #7651 to more easily replace with upstream fix.

[inactive123]

PPSSPP has been broken for a few months now on Linux (compiler: GCC).

http://pastie.org/10218708

I think this was pointed out in a separate issue but it was never addressed.

[ernestbone]

how about implementing "tunneling all communications through a single port"

[unknownbrackets]

how about implementing "tunneling all communications through a single port"

That's not a regression. This isn't about "hey what features do you want to make PPSSPP v1.1 take 3 more months to come out for?" but instead, "what had been broken since v1.0.1?"

PPSSPP has been broken for a few months now on Linux (compiler: GCC).

This doesn't seem to be the case. Every commit is built on Linux via travis. I would be getting an email for every merge at least for the past few months if it had not been building on Linux for the past few months.

Rather, something specific to your gcc version or build configuration has not been building for the past few months.

I think PPSSPP already contains most of xBRZ 1.3 (including the alpha performance stuff, which I think was the code I wrote?)

-[Unknown]

[hrydgard]

@Fadi5555 Already using some ES 3.0 features on iOS. I'm planning a larger rewrite of the GPU code which might allow further speedups on both ES 3 and newer APIs like Vulkan/DX12/Metal but that's further into the future than 1.1...

pspkvm is a curious case that needs more investigation.

I'll enable the builtin little app store for installing some homebrew apps, and I'll have a look at what PSPinstaller does later..

[inactive123]

@unknownbrackets -

#7717

I get the exact same errors when trying to compile xbrz as detailed here. This issue hasn't been closed yet hence it should still be an ongoing issue.

EDIT: For now I have just followed the suggested quick 'fix' in order to get it compiling for now in the absence of a better fix.

https://github.com/libretro/libretro-ppsspp/commit/86812754f8a443699c1c96c7d11136e34833e1a8

[enomat]

You should fix this bug before the next PPSSPP release:
#7774

It shouldn't be much work to fix it, because the multithreaded feature works well. Except that one bug when the multithreaded feature doesn't have any effect if it's enabled when you start the game.

[LunaMoo]

Any new thoughts about the somewhat ancient regression discussed in 4561440 ?
Known issues caused by it/fixed by re-adding 2 mentioned lines:
#1782
#5999

[unknownbrackets]

Those changes still fixed games also. Unpausing the list on stall address change doesn't reproduce. Those games have lived with being broken and imho they should be able to live longer until a correct solution has been found, rather than breaking other games by adding back the incorrect behavior.

AFAICT, updating the stall address just.. updates the stall address. It doesn't change the list state at all. The assumption those lines of code had was that updating the stall address maybe implied that the list ought to automatically be resumed also. This isn't correct and it's not how games use it either.

What's happening here is that the list isn't being unpaused by something, and this random place where the list was being unpaused happened to "work around that". We could also put "chaos list unpauses" in other places in the code (maybe in audio mixing, sounds like a good place?) and that would also "fix" this bug. It would be no more correct, however, and no less likely to break other games.

-[Unknown]

[LunaMoo]

Ha! I had a feeling that if I missuse the word "fix" I might get a lecture:X.(and yet I still did it;c)
So this was not a regression, might as well be another timing issue which japanese games are already famous from.

[hrydgard]

@unknownbrackets There is also the uncomfortable possibility that the display list unpausing behaviour has changed between firmware versions, too, but I don't know if we have seen any correlation between what behaviour the game "needs" and with the firmware version they request.

[unknownbrackets]

Yeah, I agree, that could definitely be involved here. But I do know that someone from uofw told me that the stall addr update function appeared to be handwritten in assembly and was quite simple. The tests I've run against it have never shown it specifically doing anything with list state.

-[Unknown]

[vsub]

The Savedata management feature looks cool but I have some suggestions:
1.In the Save State tab,show the save state preview image(I think it's either a but or the feature is not added)...you see the preview when you click on the file but there is no image like when you are just looking at all of the save games

2.If I understand the code correctly,I see that you are reading the game images to get data.
That's not really needed because all the info is in the PARAM.SFO file in the save folder

Save Type at 0x110
Save Description at 0x1230
Game ID at 0x510
Game Name at 0x12B0

3.Again about the Save State tab.
Display the game name and add the save state slot after it rather than using the file name(you can add the file name in the right side or in the preview window)
Add info about which ppspp version was used to create the save state...the info about it is in the save state at 0x10

4.You can also place the feature(but limited to the current game)in the "Delete Savedata" when you right click\click and hold on a game)...otherwise,I don't really see a point in that menu there with this feature in Tools

And btw,just curious...is it possible in some way to start a game directly from choosing a save state in the Savedata management screen(since you are getting the info from the game images now,maybe it is possible)

[sum2012]

@hrydgard
Need fix graphic issue of #7234

[hrydgard]

@sum2012 thanks for the reminder, I have some ideas about that one.. added it to the list.

[ppmeis]

It's only an idea, but we could add load game to RAM for Android devices with >=4GB RAM? Also I notice there's a lot of graphics bugs with Intel Z3580/Z3560 CPU's, but I don't have time to report them.

[LionMagnus]

@hrydgard #5999 I found a old version that someone changed/hack that works with Armored Core games http://forums.ppsspp.org/showthread.php?tid=2965&pid=108005#pid108005 , only problem is the blackbox on boosters which was a common issue before, that means potentially you can fix the issue of freezing on all Armored Core series on psp? If so PLEASE do! As a die-hard AC fan, that would be a dream come true

[sum2012]

@LionMagnus it is me,
It is the old code change
sum2012@212230e
I don't know that how to hack in new version

[hrydgard]

pspkvm is broken in current versions for tricky reasons, and adding it to the homebrew store won't help, only fixing the bug. I obviously won't add it until it runs..

[xperia64]

Is LamecraftMod working for anyone? On the latest Windows (x64) and Android versions it crashes PPSSPP itself whenever it tries to generate terrain. Also happens with the interpreter. There are a lot of log messages like these:

07-14 13:42:05.747: E/NativeApp(26696): 42:05:747 user_main    E[FileSys]: FileSystems/MetaFileSystem.cpp:116 RealPath: inPath "Assets/Sounds/ambient/track1.bgm" is relative, but current directory "umd0:" is all prefix and no path. Using "/" as path for current directory.

[vnctdj]

@hrydgard
A reminder about 2 relevant issues, as they are not on the list of the first post :

• Homebrew states naming : Homebrew states naming #7795
• UI text regression : UI text regression #7884

[hrydgard]

@vnctdj , thanks, copied to the first post.

BTW, I got inspired and am busy with a total rethink of how PPSSPP draws stuff, which might hopefully improve performance considerably, so the new release is on hold for a little bit, but I'll have to get to it soon.. will take too long to get my new GPU code ready to make it to this release.

[vnctdj]

Oh, I guess it explains why no commit has been pushed to master since one week :)
It sounds really interesting !

[sum2012]

@hrydgard Please don't forget update README.md

[GamerzHell9137]

#7886

Still an issue.

[asbel123]

hygard can u do pppsspp gold version add cw cheats db for andriod version beacuse drastic ds did that for andriod its sell crazy for google play please only 1.1 version ppssspp gold.

[LunaMoo]

@asbel123 I'm not against cheats, heck I make a lot of them by myself, but that would be really bad idea for a simple reason - we don't have a single database created specially for ppsspp with only tested cheats and maintained to reflect code changes that affect addresses. You could start a thread in the forums to collect working cheats from people who safely played with them and build such database, but for now there's really nothing that could be safely shipped with ppsspp.

Adding any of those big databases flying around the net and even shared on ppsspp forums would just cause an outbreak of people submitting issues caused by cheats, since they include lots of cheats which would be broken even on psp by bad conversions from other formats, incorrectly assigned versions etc. there's 0 quality control there, just mass copy&paste:|. That get's worse with ppsspp since it's not perfect yet and in many popular games memory differs from actual psp ~ which pretty much for affected titles breaks all simple cheats that use static addresses.

[hrydgard]

Alright, getting back to this. Been doing some UI bug fixing today ( @GamerzHell9137 will be happy), will see if I can find a workaround for #7863 (GTA music issue) later this week, then will start to wrap this up.

[vnctdj]

@hrydgard, just a reminder.
A regression to add to the list of the first post :

• Texture glitch in CGI video : Texture glitch in CGI video #7939

[daniel229]

How about the Depth range hack?

[sergiobenrocha2]

SDL and Qt build broken on ARM (again?) #7969

[GamerzHell9137]

This should be fixed too #7957

[sum2012]

@hrydgard
Fix Windows XP compatibility #7967

[LunaMoo]

Not much of an blocking issue not so new either althrough not long ago it was asked a few times in the forums ~ so are there any plans for postprocessing shaders in D3D backend?
I guess shaders themselves could be translated by defining data types and common functions from one to another, for example:

#if HLSL
    (...)
    #define vec2 float2
    #define mix(x,y,s) lerp(x,y,s)
    (...)
#endif

etc., but there's probably much more stuff left to do.

~ Which might not be worth doing, since for the most part it's used by people with very weak/outdated hardware that couldn't run too many shaders without affecting performance anyway. Maybe we should just disable the option for postprocessing shaders in that backend before 1.1 release.

[sergiobenrocha2]

version string corrupted, #7991

[sergiobenrocha2]

#7730 still happens. If I load only one game, it forgets in the next time. I had to open 2 games to appears at least one in the recent files.

[sum2012]

@hrydgard
Can merge #7965 ?

[hrydgard]

@sum2012 I will take a look soon. It may need some changes.

I've updated the status, major regressions left are Breath of Fire and GTA.

[vnctdj]

@hrydgard
You forgot to add this, as I wrote earlier :

A regression to add to the list of the first post :

• Texture glitch in CGI video : Texture glitch in CGI video #7939

[sum2012]

@hrydgard OK

[vnctdj]

@hrydgard
Unfortunately, recent changes have caused 2 new issues :

• Broken translations : Broken translations #8001
• Auto rendering resolution is broken : Auto rendering resolution is broken #8002

[sum2012]

@hrydgard add #8003

[sum2012]

Can this issue can close ?
Or used for for v1.1.1 ?

[unknownbrackets]

I'm going to close it. If we want one for 1.1.1 we can open one, but I think we've largely fixed the things we want fixed in that release (barring new regressions.)

-[Unknown]