{RSX [0x02a6238]} VM: Access violation reading location 0xcf900000 (unmapped memory) (Dynasty Warriors 8 XL)

The game crashes when entering a stage, always within the first few seconds. I can get through a few minutes of the tutorial before the same happens.
Debug console mode changes the error from: F {RSX [0x0399668]} VM: Access violation reading location 0xcf900000 (unmapped memory)
To: F {PPU[0x1000000] Thread (main_thread) [0x00441ca4]} VM: Access violation writing location 0x0 (unmapped memory)
@Archim119 Did you use a RPCS3 patch ? I got a similar message on another game https://github.com/RPCS3/rpcs3/issues/8985
@jy95 With or without patches enabled, the game still freezes at the same point. Did you manage to get yours to work?
I've tried enabling certain libraries to solve the issue, while it does get rid of the access violation it is then replaced with this: E {PPU[0x1000000] Thread (main_thread) [0x014b8a30]} SYS: 'sys_mutex_create' failed with 0x80010005 : CELL_ESRCH [1]
I've also tried disabling accurate LLVM DFMA, but that didn't work either.
Last comment is approximately 2 years old. Is this still an issue?
@Darkhost1999 yes, sadly still an issue
I've recently come back to emulating PS3 games and after playing most of my favorite games i played Dynasty Warriors 8 and i discovered this crash too. I tested the game for a while and now i discovered the issue and why this happens. It is a problem with the rendering of the level of detail with the soldiers. After a few seconds the game crashes because the game has an issue with loading the low level of details of the soldiers. The RAM has an error that might just be one byte that is unmapped. This error happens on every Dynasty Warriors 8 version, including the japanese and chinese versions, except Dynasty Warriors 8 Empires. I made a screenshot of one of my last tests and this is my conclusion. You can finish the tutorial and you can finish every stage but you need to point the camera at the floor to not load the low level of details and if there are not enough soldiers around you could look up but only in the tutorial. This screenshot is the frame that is crashed on. This is literally 60 frames after the low level of detail of this one soldier was loaded

3 years later and still not fixed
although i got a different access violation code F {RSX [0x004cfbc]} VM: Access violation reading location 0xcf90f7c0 (unmapped memory) which might or might not help the dev figure something out
Game ID is: NPJB00511 (DW8XL JPN Digital version)
Upload updated log and savedata.
I think I figured it out: We assume that the highest vertex index is the frequency divider when modulo op is used because there are vetex indices higher than it. But in reality we should do max_index = max(index % frequency, max_index) for each index. This is of course slower, but it can be optimized to not do it if we have a lot of free memory range left.
RPCS3.log.gz NPJB00510-SAVEDATA.zip
Here is my log and savedata (mind you im not the original poster) and also save data is maybe useless, due to the crash literally happening on any stage that you starts
WIP
Test #14505
sorry for not replying i couldnt test DW8XL since i deleted the game install both from PS3 and RPCS3 and would not rather redownload a 23GB game on my current network situation
BUT i have another DW game that suffer from the same issue and surprise surprise using #14505 build it worked
this was the error code before i used the build 0xcf906420 (nearly identical with DW8XL error code) and here is my log for this other DW game (BLJM61347) RPCS3.log.gz
as pointed out on the comments in #14505 the fallback handler "This sucks" appears also 1x every like 1-2 second
Somewhere in the update broke this again RPCS3.log.gz
Last known good build: Honestly idk probably #14505
Old error code: 0xcf906420 New error code: 0xcf905960
seems like its in the same ballpark of issue
Just got the same error while playing MGS Rising :Revengeance:
E RSX: [attrib buffer] Pool limit was reached. Will attempt to swap out the current heap.
F {RSX [0x010f094]} VM: Access violation reading location 0xcf900000 (unmapped memory)
And this shows when trying to stop the game and RPCS3 crashes:
Segfault reading location 0000000000000028 at 000076767776f722.
Thread: Main Thread.
Title: "METAL GEAR RISING: REVENGEANCE [BLES01750]" (emulation is stopping)
Build: "0.0.34-17339-1b87e186 Alpha | HEAD"
Date: "2025-01-13T21:29:02"
Thanks!
Is this still an issue?
Sometimes the game still crashes because of that reason or a similar reason but it has become better. For me it mostly crashes in the Ambition Mode. I was able to port my 100% save file to the emulator and i can say that if it runs well without crashes it's actually running like it is supposed to run but it uses a lot of CPU (which isn't a problem but it's important to say that). Give it another year or two and this will be completely resolved :)
is that still an issue?
Yes, it also appeared in BLUS30690 (Dynasty Warriors 7). F {RSX Offloader} VM: Access violation reading location 0xcf900000 (unmapped memory)
Got the error on NPUB31633 (Metal Gear Solid 4). F {RSX [0x0a2aacc]} VM: Access violation reading location 0x4239b7c0 (unmapped memory)
Can't even try to avoid it with camera or menu since it happens in the opening cutscene.
@guss0ar this is only a thread for Dynasty Warrior games
also im pretty sure you just havent done MGS4 correct settings if it happens on opening cutscene
I would also tag @kd-11 for a possible issue on RSX
I changed some codes in RSXThread.cpp, it seems can fix this problem. idk whether it'll cause other issue. @elad335 @digant73
This is wrong, continue is the right action there. The real problem appears to be the index range, but I'm 99% sure it would also have crashed a real PS3 when going out of bounds. How do you reproduce the issue quickly?
I changed some codes in RSXThread.cpp, it seems can fix this problem. idk whether it'll cause other issue. @elad335 @digant73
This is wrong, continue is the right action there. The real problem appears to be the index range, but I'm 99% sure it would also have crashed a real PS3 when going out of bounds. How do you reproduce the issue quickly?
Because it'll happen after enter some certain battles immediately. After change this code, the issue is disappeared.
I have recorded and uploaded a video to compare them.
https://www.bilibili.com/video/BV16aUkB3ETC/
I changed some codes in RSXThread.cpp, it seems can fix this problem. idk whether it'll cause other issue. @elad335 @digant73
This is wrong, continue is the right action there. The real problem appears to be the index range, but I'm 99% sure it would also have crashed a real PS3 when going out of bounds. How do you reproduce the issue quickly?
Just by playing, itll crash in like 3 minute of gameplay usually, especially using Lu Bu
It's still wrong. You can also just disable the function and it wont crash. Restart indices are a normal part of graphics rendering. first + count is generating a value too large, that's the real issue.
I'll give the game a shot later and try to recreate it. I have no idea what the series is about, so hopefully just standing still or watching some cutscenes can cause it to fail.
Thanks. Hope this issue can be fixed correctly!

I changed some codes in RSXThread.cpp, it seems can fix this problem. idk whether it'll cause other issue.