Solved [LINUX] BMS 4.37u3 Invisible 3D models, err mesg: Skipping LOD load, lodIdx is negative: Help needed
-
Per @Seifer advice, I’m posting in hope anyone could explain me what is going wrong with my setup:
Problem: 3D models are invisible in the sim (TacRef and 3D world)
Steps to reproduce:- Start BMS: no matter if any (old or new) launcher is used, or Falcon BMS.exe is launched directly
- Get to any place where 3D models should be displayed: open TacRef or enter 3D world - tested with both Dogfight and training TEs.
Expected result: cockpit, planes, buildings and other objects are visible
Actual result: only F-16 cockpit canopy and content of the displays (HUD/MFDs/DED/HMD) are visible.
Any other objects are not. In debug logs there are lines: about skipping LOD loading:Log entry: 400 Skipping LOD load, lodIdx is negative: -1261039100
Extra notes:
Problem is exclusive to BMS 4.37u3. I’ve never had such issue with BMS 4.35 nor 4.37u2. If it occurs, is persistent during the whole session from BMS launch to close, ie if models are visible just after launching the sim, then they will be visible this run. If they are not, they won’t be displayed at all, until application is closed. Launching BMS again is next dice roll. In bout 70% of the cases models won’t be displayed.
BMS 4.37u3 is clean, vanilla install from scratch in fresh wine 64bit prefix. No mods, no updates on existing install, straight from Falcon4 to BMS 4.37u3.
Whole installation process was repeated several times to be sure.Screenshots:
Console output:
https://pastebin.pl/view/f2c58e99Debug log: good session,models are ok:
https://pastebin.pl/view/d4026028bad one, no models displayed
https://pastebin.pl/view/4a244c46System
Hardware:
CPU: Ryzen 5 5600X; M/B: MSI B550-A PRO; RAM: 16GB DDR4 Patriot Viper 3733MHz CL17; GPU: Saphire Pulse RX5700XT BE; P/S SuperFlower LEADEX III 750W, SSD Crucial M500;
OS:
Fedora 38 x86_64, kernel 6.5.9-200, libdrm-2.4.117, mesa-23.1.9, wine- 8.19, dxvk-2.3, Gnome (tested under X11 and Wayland session)What has been tried:
- rebooting system: makes no difference, first run of the sim may be good or broken.
- running under different window server (Wayland and Xorg X11)
- using both legacy and new launcher and running Falcon BMS.exe directly
- using DXVK DX11 -to-Vulkan and wine build -in DX11-to-OpenGL wrappers
- launching from desktop shortcut or from console.
Interestingly @Ferde is also running BMS 4.37u3 under Linux and he’s not experiencing this problem. He also uses Radoen GPU (RX 6600) with opensource drivers.
I’m totally puzzled what’s going on. In any other case BMS would be the last thing to blame, but could it be I’m hitting some extra rare race condition? -
What was also tried:
- disabled XMP profile. Runnin memory at 2133MHz instead of 3733MHz
- update BIOS to the latest version available.
-
The good news is problem didn’t occur even once on Wine 7.22/ Fedora 37 fresh install for last 4 days.
Now I’m testing same OS with Wine updated to 8.0. So far 4 launches no problem. I’ll give it some more time before I update it to newer Wine.
Sill some insight what are the conditions for “lodIdx is negative” error would be great. At least to provide some hints once I’ll be reporting a bug on WIne bugzilla. -
Small update, last known safe Wine release is 8.14, I’ve been trying it for two days, launching BMS about 30 times, no problems occurred. Next version that has Fedora packages is 8.17 and this one failed at first run.
-
Bug aws reported on WIne Bugzilla:
https://bugs.winehq.org/show_bug.cgi?id=55903It would be cool, if somebody could retest, confirm and add a results to the bugreport.
-
Update on Wine bug 55903, it’s a memory management change in 8.17 that broke BMS 4.37u3.
Again, it would be helpful if somebody could retest 8.17 release and add comment on Wine bugzilla. -
Ok so this bug is solved in Wine 9.0-rc2, till 9.0 release using 8.16 or earlier is advised.
-
-