Final Fantasy X/X-2 HD [BLES01880]: Mem access violation during bossfight

[maxgerhardt]

• RPCS3 version: latest master (ca5158a)
• System Specs
  • Intel i5 3450
  • AMD R9 270X
  • 8GB DDR3 RAM

Problem:

Final Fantasy X HD remaster can't be played through because of a fatal error during a boss fight. This error always occurs reproducably.

During the second-last boss fight of the game with Yunalesca, once you defeat her first-stage (24,000HP), the game crashes at the place where the transformation sequence into her second form should play.

·F 0:27:13.311372 {PPU[0x1000000] Thread (main_thread) [0x026551a4]} MEM: Access violation reading location 0x300c6d8
·E 0:27:14.315552 {rsx::thread} RSX: nv406e::semaphore_acquire has timed out. semaphore_address=0x40300400
·! 0:27:48.064769 Stopping emulator...

Related messages for that address

·! 0:15:15.387976 {PPU[0x1000011] Thread (fios mediathread 8) [0x026609ac]} LDR: **** MagicScriptLib export: [0x131A6569] at 0x300c6d8
·E 0:15:15.387978 {PPU[0x1000011] Thread (fios mediathread 8) [0x026609ac]} LDR: Already linked function '0x131A6569' in module 'MagicScriptLib'

Settings:

• PPU: LLVM
• SPU: ASMJIT
• Renderer: Vulkan
• Preffered SPU Threads: 2, lower SPU thread priority
• resolution to 1920x1080
• everything else default
• all caches were previously cleared

This bug must have existed for a long time, because I've started that playthrough at the beginning of the year and that bug already stopped me, and it's still there.

Attached are the RPCS3 logs and the savefile for easy reproduction.

RPCS3.log.gz
BLES01880SAVEDIR--------------1.zip

[maxgerhardt]

1080p is natively supported by the game, but I'll try with 720p tomorrow, if that fixes a memory access failure at all..

[AniLeo]

Upload TTY as well, the log indicates stuff is being written to TTY.log too

[maxgerhardt]

Here are the logfiles for a 720p play, crashes at the exact same place.

Address is slightly different 0x302afe0 this time, maybe ASLR?

RPCS3.log.gz
TTY.log

[kd-11]

Likely fixed in the past few days. Retest.

[maxgerhardt]

@kd-11: Tested with latest master ed8aa77, this was not fixed. Crashes at the exact same place. I have previously cleared all possible caches.

[...]
·! 0:25:05.571516 {PPU[0x1000011] Thread (fios mediathread 8) [0x026609ac]} LDR: **** MagicScriptLib export: [0x131A6569] at 0x3000390
·E 0:25:05.571518 {PPU[0x1000011] Thread (fios mediathread 8) [0x026609ac]} LDR: Already linked function '0x131A6569' in module 'MagicScriptLib'
[...]
·F 0:27:42.610970 {PPU[0x1000000] Thread (main_thread) [0x026551a4]} MEM: Access violation reading location 0x3000390

Log files are attached.

RPCS3.log.gz
TTY.log

[maxgerhardt]

Still occuring on latest master.
logs.zip

[Whacker007]

I signed up to make this report.

Bug and crash still exist in the 7470-43b75ccf build, can be reproduced consistently and without fail. Always occurs at Yunalesca's first transition after losing 24000 HP. Note that I'm using the NA release, BLUS-31211.

System specs are a 2600K Intel CPU, 980ti card, 16gb of ram, Windows 7 64bit. My system is patched and drivers are up to date.

Crash occurs regardless of settings. I have tried a large variety of settings including up what what is arguably the most "safe mode" with precise decoders, 1 SPU and compile thread(s), Vulkan renderer, default res, and no scaling. The error is same as others reported, MEM: Access violation at xxxxxxx, however the exact address is different. I have tried this with a fresh install of RPCS3 so cache should not be a factor, and always restart the program after making any changes to settings or after any crashes.

[Whacker007]

One quick addition. I went back to an older build, 7022-3d8223fcc, and the same problem exists. The error is the same MEM access violation at the same address.

Edit - Did some more research. Tried about twenty different builds from the latest 7480 today all the way back to a 0.0.3.6 from Oct of 2017 I found on EmuCR. All produce the same crash, but the exact memory address is varies depending on build date. Also realized I should have posted the log files, apologies for that. The log files below are from the 7480-11681f02 build.
rpcs3.zip

[maxgerhardt]

Still occuring in 45942c4.

TTY.log
RPCS3.log

[akien-mga]

Finally reached this stage in my playthrough of FFX HD [BLUS31211] and I can confirm that the bug still occurs in ad10eb3. The game is otherwise emulated flawlessly, so it would be really nice to get a fix for this issue.

System specs (Mageia 7 x86_64):

CPU: Quad Core Intel Core i7-8705G (-MT MCP-) speed/min/max: 1032/800/4100 MHz Kernel: 5.1.14-desktop-1.mga7 x86_64 
Up: 2d 6h 27m Mem: 2956.8/15792.2 MiB (18.7%) Storage: 953.87 GiB (44.5% used) Procs: 256 Shell: bash 4.4.23 inxi: 3.0.33 

I'd be happy to provide logs and save files, though I believe the ones in OP might still be relevant.

If anyone can advise on how to get more debugging information around the crash, I can do some testing.

[akien-mga]

I'd be happy to provide logs and save files, though I believe the ones in OP might still be relevant.

Reproduced the issue anyway on b9130dd to be thorough. As for others, the crash is due to a memory access violation:

F {PPU[0x1000000] Thread (main_thread) [0x00549f74]} MEM: Access violation reading location 0x328d378

• BLUS31211 - RPCS3.log
• BLUS31211 - TTY.log
• BLUS31211 - Main thread.txt
• BLUS31211 - SAVEDIR.zip

Steps to reproduce:

• Use the provided game save
• Proceed forward to the North, wait through the cutscene until the battle starts
• To crash quickly, switch character to Yuna (L2 by default to change formation), use Summon > Bahamut. Then cast Bahamut's already preloaded Overdrive (Dpad to the left to see Overdrive action), it should be strong enough to kill Yunalesca's first form and trigger the crash when she starts her transformation.

When it crashes, the game is paused in the debugger, and pressing Play will make the same memory access violation happen again. When I tried, pressing Play several times in a row allowed the game's animation to proceed further by a few frames, until it was fully stuck (at least no visual progress anymore after ~300 presses, FPS at 0.06).

[maxgerhardt]

Huh, still occurring. @Nekotekina @kd-11 can you make a patreon donation with a specific request, e.g. to investigate this specific bug? At the one year anniversary of this bug I'd be happy to give you money for this to be fixed.

[kd-11]

Not directly, you can't post bounty through patreon.

[elad335]

Upload log with latest master.

[jorgensond]

0.0.6-8368-c7277605
NPUB31247
F {PPU[0x1000000] Thread (main_thread) [0x00549a64]} MEM: Access violation reading location 0x31e8308
RPCS3.log
TTY.log

[akien-mga]

Sorry for the delay, my laptop was in repair. I think my logs are similar to @jorgensond's, but uploading nevertheless:

0.0.6-8395 1b140c8
BLUS31211
RPCS3.log
TTY.log

Would it be useful to take a capture at the moment of the access violation issue?

[elad335]

Can you retest with this build? It's a hack but the answer may lead me to a proper fix.

[akien-mga]

I tried that Windows build through WINE but the initial PPU compilation was taking ages, so in the end I compiled it from source (current master fcc75c8) + the two patches from https://github.com/elad335/rpcs3/commits/sys_prx

It stalls at the same event. There seems to be one instruction more in the call stack.

RPCS3.log
TTY.log

[AniLeo]

English only, as other people need to understand what you write

[laurilucio]

sorry

hello good morning how are you
I'm Laurilucio, Brazil
I have the same problem. the game crashes when I finish Yunalesca in the first stage.
Has anyone figured out how to solve?
Is there anything the user can do to solve this problem?

[Asinin3]

No, no one has figured out a way around it yet and the problem still exists in the latest builds of RPCS3, If you're already at the final boss, I'd suggest just watching the ending on YouTube. No point in waiting for a fix if you've already got this far in the game, it could take awhile before a dev looks into it.

[digitaldude555]

@akien-mga still issue?

[elad335]

Reupload log if it is.

[akien-mga]

I tried but rpcs3-v0.0.8-9666-6bb083a7_linux64.AppImage doesn't run on my system (while the last version I tried, rpcs3-v0.0.7-9125-0ff5938b_linux64.AppImage, runs fine):

$ ./rpcs3-v0.0.8-9666-6bb083a7_linux64.AppImage 
DEBUG: 2 log messages accumulated. 178 channels registered.

(no window appears)

I tested rpcs3-v0.0.8-9395-ca5bc512_linux64.AppImage which starts fine. I'll try to bisect further to see what introduces the issue, but I'll move this to a dedicated issue as it's unrelated to FFX.

[Asinin3]

This issue is been ongoing for what 2 years and the devs I am guessing have not shown any interest in resolving the issue guessing Final Fantasy X/X-2 HD is not worth the fix

If you read the thread, you would see that Eladash looked into it back in March this year, however sadly he wasn't successful.

RPCS3 is open-source, anyone who wants to is free to investigate the issue and fix it. RPCS3 is in Alpha, looking into game specific issues when there is a slew of major general things to work on is going to be a low priority. There is also 554 issues open right now, and a lot more not reported on GitHub. Not to mention when changes are made sometimes there is a regression.

Be patient, and update the issue every 4-6 months to ensure devs know it is actually still broken other wise they might be wary of even looking at it.

[dmsun]

just want to highlight this for visibility, still an issue on version 0.0.15-12048-f64a7bb8. I'm having issues getting the logs, as it appears they're not being saved somewhere. I'll try and get a copy from the Console next time I try this out.

[chanfriendly]

Hi all! Bumping that as of 0.0.17-12573-8c3a4c26, also running into the freeze after the first transformation. Logs also disappear, RPCS3 fully crashes after I try to close the game

[dmsun]

Hi all! Bumping that as of 0.0.17-12660-f44e50b9, persisting issue relating to a freeze. I've notived that the memory access address changes between version releases. Not a strong software developer, so unsure if that is a useful observation. Mem address is 0x2e0cb80

playing on Linux build. screenshot below, the log appears to have not been updated in a long time - could be something wrong with it found in ~/.config/rpcs3/RPCS3.log.gz sudo gunzip results in unexpected end of file

[elad335]

Maybe we should ban sceeenshots of main window, anyways if you want to update an issue upload the log file.

[tavarchynn]

A lot of persons have uploaded the log file myself included

…

On Tue, Aug 24, 2021, 8:59 PM Eladash ***@***.***> wrote: Maybe we should ban sceeenshots of main window, anyways if you want to update an issue upload the log file. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5080 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADVI4V3GDWAL7L24WBIF3L3T6RE6RANCNFSM4FTGNF3A> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

[AniLeo]

The last log file is from a year ago

[tavarchynn]

On Wed, Aug 25, 2021 at 11:18 AM Ani ***@***.***> wrote: The last log file is from a year ago — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5080 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADVI4V6O4PZPDLMKOSQDD3TT6UJWFANCNFSM4FTGNF3A> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

Yes and I believe there where other uploads before all with the same memory access violation error

[AniLeo]

Not the point. why are you bumping an issue without an updated log file

[tavarchynn]

The last time I tried playing was a few months ago and it was the same so ignored it will run it later and upload a updated log file

[maxgerhardt]

Original reporter from 3 years ago here, I've gathered new logs against 4e139ee, see:

new_logs_25_aug_2021.zip

RPCS3 also crashes after the access violation has occurred, the stop-button is pressed and the OK button of this dialog is clicked.

But the logs were retrievable in their .txt format.

[elad335]

Are any of you available on discord? Github is really annoying to debug something with anyone. Not today though but for later.

[maxgerhardt]

Yes, I've joined the rpcs3 discord server and shot you a message there.

[aurelaureel]

Hey guys i have the same issue, does anyone figured out a way to bypass it ? Maybe a save post fight on the thread might just do it for now, if any can be found.

And of course if someone have the knowledge and time to fix it i would love to end my story :)

Good luck

[Pyroxenz]

Add me to the list of folks who has run up against this wall of a bug.

Sad that the problem has stayed unresolved 4 years after it was first reported.

[kksanders]

Has anyone had any luck resolving this issue? Playing on steamdeck, just ran into this problem. Literally the only issue I have had running it, but obviously I am stuck. Please help

[kksanders]

RPCS3.log
TTY.log
Just tried again. Still crashes.

"F {PPU[0x1000000] Thread (main_thread) [0x00549f74]} VM: Access violation reading location 0x2dff240 (unmapped memory)"

Edit: attached log files

[Asinin3]

Why would you retest it on master build when ELAD opened a pull request a few hours ago when a fix that needs testing? See #12921

[kksanders]

I thought it was already pushed to the master build. I'm sorry, I don't fully understand how to test if it is still in pull request status. I would love to test if someone can give me a hand

[Darkhost1999]

Go to the PR
Then checks
Select your OS
Click view more details
Then download the artifact and extract the zip ontop of your current rpcs3 folder
When you boot the game you'll get a popup "this is a pr and can damage your data. Do you consent to using this PR?" Yes or no

[kksanders]

Thanks for the help! Was able to load up the PR 12921 Build. Successfully made it passed the phase 1 part of the battle! Thanks so much! I will also post directly on the PR #12921 to confirm working.

[kksanders]

Was able to get past the boss fight. However, trying to save and "overwrite" the old save results in another crash. I will attach logs.
"E cellSaveData: cellSaveDataUserListSave(version=1, userId=0, setList=*0xd026dcc4, setBuf=*0xd026dcd4, funcList=*0x874890, funcStat=*0x874870, funcFile=*0x874878, container=0xffffffff, userdata=*0x300fbf20)
F {PPU[0x1000037] Thread (save_data_thread) [HLE:0x026302c4, LR:0x001592d4]} SIG: Thread terminated due to fatal error: Failed to move directory C:/Users/kevin/Downloads/rpcs3-v0.0.25-testing/dev_hdd0/home/00000001/savedata/BLUS31211SAVEDIR---------------/ (Already exists)
(in file C:\Users\ContainerAdministrator\AppData\Local\Temp\cirrus-ci-build\rpcs3\Emu\Cell\Modules\cellSaveData.cpp:2001[:24], in function savedata_op) (error=0xb7)"

Edit: I will try to create a new save, instead of overwriting, and see if it fails or not.

Edit 2: Seems to have work to create a new save, instead of overwriting. Not sure what was going on there, but seems fine this way.

RPCS3.log
TTY.log

[Pyroxenz]

"Then download the artifact and extract the zip ontop of your current rpcs3 folder"

I downloaded the artifact and extracted the zip file inside, to my rpcs3 folder. When booting up the game, no pop-up message occurred and the game still hung upon defeating the first form of the boss. What am I doing wrong?

[kksanders]

There is another .7z file titled rpcs3-v0.0.25-.......u get the rest. unzip that into the rpcs3 folder. Then you should get the message

[Pyroxenz]

Nice - the fix works!

Thanks Elad! You're the man! Finally we can get past that damn Yunalesca now!

[Dpera85]

Hi guys! amazing you fixed the issue!!! i've been following this issue since years waiting for a solution :D
All i have to do is to put the save file into rpcs3 folder? where can i download the save file?
thanks a lot!!!

[dmsun]

I've been watching this issue for too long...I maxed the Sphere Grid, made some Break Damage Limit Weapons, I did everything I could think of waiting for this moment. 140 hours of Gameplay, most of it stuck at this boss lol.

I can confirm in the related PR this issue appears resolved!! We may have FFX HD finally be fully playable after so long!

thank you Elad, this really means a lot to me. It's my favourite game.

[Pyroxenz]

On another note, the FMVs are now emulated perfectly. Previously they had been choppy and glitchy.

Good job!