Regressions from #4611

[Asinin3]

After #4611 merged I have found the following issues:

Zuma doesn't display graphics properly anymore (gpu texture scaling disabled)


Fixed if i use Aspect ratio auto instead of 16:9, still somewhat a regression but other games working fine still.

F1 Race Stars freezes at boot with the following error when launching with Write Color Buffers
F {rsx::thread} class std::runtime_error thrown: Unreachable
RPCS3.log.gz
Fixed in new PR

[micaelis72]

Something to add: After #4611, Soul Calibur IV now hangs when generating new shaders, even after they were cleaned previously.

This kind of error happens with other games, but it's the first time with this game, now easily to point out:

E {PPU[0x1000001] Thread (mainThread) [0x005dd27c]} 'sys_lwmutex_trylock' failed with 0x80010008 : CELL_EDEADLK [1]
E {PPU[0x1000001] Thread (mainThread) [0x0061fc64]} 'sys_lwmutex_lock' failed with 0x80010008 : CELL_EDEADLK [2]
E {PPU[0x1000001] Thread (mainThread) [0x0061fc64]} 'sys_lwmutex_lock' failed with 0x80010008 : CELL_EDEADLK [3]
E {PPU[0x1000001] Thread (mainThread) [0x0061fc30]} 'sys_lwmutex_unlock' failed with 0x80010009 : CELL_EPERM [2]
E {PPU[0x1000001] Thread (mainThread) [0x0061fc30]} 'sys_lwmutex_unlock' failed with 0x80010009 : CELL_EPERM [3]
E {PPU[0x1000001] Thread (mainThread) [0x005dd27c]} 'sys_lwmutex_trylock' failed with 0x80010008 : CELL_EDEADLK [2]
E {PPU[0x1000001] Thread (mainThread) [0x005dd27c]} 'sys_lwmutex_trylock' failed with 0x80010008 : CELL_EDEADLK [3]
E {PPU[0x1000016] Thread (DLC Check) [0x01522000]} 'sys_fs_opendir' failed with 0x80010006 : CELL_ENOENT, “/dev_hdd0/game/BLUS30160/USRDIR” [1]
S {rsx::thread} RSX: New program compiled successfully x3
E {PPU[0x1000001] Thread (mainThread) [0x01375e5c]} Stat: 'sys_lwmutex_trylock' failed with 0x80010008 : CELL_EDEADLK [x7145]
E {PPU[0x1000001] Thread (mainThread) [0x01375e5c]} Stat: 'sys_lwmutex_unlock' failed with 0x80010009 : CELL_EPERM [x109393]
E {PPU[0x1000001] Thread (mainThread) [0x01375e5c]} Stat: 'sys_lwmutex_lock' failed with 0x80010008 : CELL_EDEADLK [x109393]

Follows the logs to help further investigation:
RPCS3.log.gz

Cheers!

[Asinin3]

F1 Race Stars issue when using Write Color Buffers has been fixed with #4649

[kd-11]

What settings break zuma? In builds before #4611 I could only get half screen. Works fine with all settings default + no gpu texture scaling. Tested on both AMD and NVIDIA.

[Asinin3]

Strange, before 4611 i even managed to go ingame once and thought it was playable apart from text rendering issues. Upon secon launch with builds before 4611 I only got stuck on the screen above but it did take up the full screen. Not using anything fancy here.

[kd-11]

Here is my before

[Asinin3]

Well, I figured it out. If you use Aspect ratio 16:9 this game has the blackscreen and if you use automatic it's fine. But on old builds 16:9 didn't cause these issues. Feel free to close if you don't think this is a regression.

[kd-11]

It should stay open for a little while longer in case other regressions are found

[raveskirza]

Can confirm that Soul Calibur IV hangs reliably after #4611 was merged. So this issue should remain open.

[micaelis72]

The Soul Calibur IV issue remains after #4649

[kd-11]

Sounds like a timing issue. Does the emulator close if you click the stop button?
If rsx hangs, clicking stop or closing the window will not actually close rpcs3, a zombie will remain in task manager. I suspect the changes in the PR may have unhidden something that already existed. Keep checking, hopefully this can be reproduced on other titles.

[raveskirza]

When clicking on Stop, the game stops gracefully. So far this is the only game that is affected badly by 4611.

This game is very sensitive to memory allocation
Before jarveson fixed this game probably in commit dc0f2ca (Pull jarves SPU fixes) , it used to hang with error St13runtime_error thrown: Verification failed (e=11) (0):
(in file /home/mnunes/apps/rpcs3/rpcs3/Emu/Cell/lv2/sys_memory.cpp:52), while loading the game texture files (#3663).

After 4611, the sys_memory.cpp:52 is not seen but the hang is very similar(no error message though) and happens when it is loading the game texture files.
The last working build which does not hang this game is rpcs3-v0.0.5-6831-f8a0be8c

[Emulator-Team-2]

@kd-11
After pr #4611 this game crashes before menu on the intro and loading screen.

Dragon Ball Z: Battle of Z-Demo / Full Version
F {PPU[0x1000000] Thread (main_thread) [0x006b540c]} MEM: Access violation reading location 0xbfffff38
Log:
RPCS3.log.gz

Forum:
-Dragon Ball Z: Battle of Z [NPEB01823]
-Dragon Ball Z - Battle of Z [BLUS31314]
-Dragon Ball Z: Battle of Z [BLES01941]

[jenci8888]

After PR #4611 this game freezes before ingame on loading screen.

Need For Speed: Most Wanted (Any Disc and PSN)

E {rsx::thread} RSX: Invalid FIFO queue get/put registers found, get=0x11337000, put=0x106C
E {rsx::thread} HLE: Getting the real address failed. (addr=0x11337000)

Log: RPCS3.log.gz

[PeterTh]

Starting with #4611 Tales of Xillia (BLES01815) shadows sometimes flicker.

As in, the shadow will be cast correctly one frame but be gone completely on another, alternating every other frame. I've re-checked using older builds and am certain that it was caused by that PR.

It doesn't apply to all shadows equally. Note that Tales games on this engine are somewhat special w.r.t. shadow mapping, as they use a separate shadow map for each dynamically cast character shadow.

[theoldsport]

Since #4611, Skate 3's replay editor no longer functions.

After reaching ingame and pressing Select (entering the replay editor), the game hangs, occasionally shows some graphical errors, and spams the log with FIFO errors or semaphore locks. No combination of settings, including debug settings, seem to affect the issue. Present in Vulkan and OpenGL

RPCS3 (1).log.gz

[ghost]

Since #4611, Whenever I boot up Skate 3 in the emulator, I can use a preset skater a free skate around but when ever I want to do a mission, place an object, look at a replay, or edit my skater it gets stuck in this infinite loading screen or it will freeze. In free skate I am getting consistent 30+ FPS and its really smooth and few audio issues but whenever I try to do stuff like this it fails, any help would be appreciated.

My Specs : Windows 10
Intel Core i7-6700K
GeForce GTX 980 Ti

[spyropt]

@kd-11
Sorry to post a regression after more than two months
Soulcalibur V BLES01250 regressed with this pr it just hangs on the fmv after a second sometimes there is a fainted image of the fmv, but it crashs the emulator
pr 4635 (build 0.0.5-6831 ) it works fine
RPCS3-#4635.log.gz

pr 4611 build ( 0.0.5-6841 ) does not
RPCS3-#4611log.gz

i dont know if this is related to my cpu but i tested on a cpu with just AVX (3770k) and without AVX (x5675) and i got the same result

3770k log result build 7117
RPCS3.zip

I tryed every ppu an spu combination and i got the same result

[raveskirza]

Soul calibur IV hang seems to be fixed in #4973

[kd-11]

Retest with build from #5099. There is a good possibility this is fixed now.

[PeterTh]

@kd-11
I just checked, and sadly the shadow regression in Tales of Xillia I pointed out above seems unaffected by #5099 .

[daaaaaavis]

Skate 3 replay editor, and any mission isn't working properly.
It goes into the kind of "camera" view, sticks to 0 - 0.5fps and never comes back.
Using latest compiled version.
Game works fine in free skate though, just can't do anything regarding the story, replays etc., like mentioned above.
Specs : R7 1700@3.5ghz, 980TI, 16GB DDR4 3200mhz.
No change of settings help.

[kd-11]

Remnants of this bug should be fixed by #5427

[Emulator-Team-2]

Regression in Dragon Ball Z: Battle of Z is fixed by #5427

[kd-11]

Closing as fixed. If any bugs remain, open a dedicated ticket for it.