[Solved] Crash on the 1st cell change

[monsta88]

Latest AE 1170, w10 Pro.

 

The LO is unusually stable except for the 1st cell change after loading a game. It is especially true, when I try to load from a small cell (house) into a big one (world). If the game survives the first 2-3 loads, it remains stable for the rest of the play session (several hours).

 

Here is the latest crash log, this time I was loading from the world into a dungeon.

 

crash-2025-06-29-17-27-07.log

 

Another one, same place.

 

crash-2025-06-29-17-51-38.log

 

 

The common thing in these crash logs is the FSMP. 

 

Please note that I could not install FSMP properly with MO2. When I run the FOMOD installer, I can't choose my game version (1170), or any other version. I used an earlier install from the SE times (1.5.97) and then manually copied the 1170 .dll and other files from the installer zip. The FSMP version is v2.5.1 (latest).

Edited October 12, 2025 by monsta88

[traison]

These crashes were both caused by EngineFixes.dll. Its potentially having some issues with the usvfs and the NtCreateFile function exported from ntdll.dll, but this seems quite unlikely. It's doing something related to the ScrapHeap.

 

Your SSE Engine Fixes version is old. Update it and test again. If its still crashing, post a new crash log and I'll have a look at the dll in a disassembler. You can also fiddle with its config file. Try changing everything related to memory management and the scrap pool/heap.

Edited June 29, 2025 by traison

[monsta88]

4 hours ago, traison said:

Your SSE Engine Fixes version is old.

 

It was up to date, all my installed files had the same creation date/size as the ones in the newest download (v6.2).

 

I downloaded this now, it looks it is just a config file though.

 

New Crashlog

 

crash-2025-06-29-22-12-53.log

 

 

 

Edited June 29, 2025 by monsta88

[traison]

9 hours ago, monsta88 said:

It was up to date...

 

Oh, its reporting its version incorrectly then. 6.2 is available for 1170, and yours is reporting 6.1.1. The code offsets line up with 6.2 though so regardless of what it says, that's the version you have most likely.

 

skse64_1_6_1170.dll+00108D8 has something to do with saving/loading. That matches with the NtCreateFile hint. It's most likely using this to either write or read the autosave when changing cells.

 

There is no pdb file for SSE Engine Fixes unfortunately, so no easy answers for us.

 

Both EngineFixes.dll+004F910 and 004E0B8 are also called on every frame when the game is running (not in the main menu). It's using tbbmalloc.scalable_realloc to resize some memory space.

 

So unfortunately for the time being I'm going to go back to my previous statement:

13 hours ago, traison said:

...fiddle with its config file. Try changing everything related to memory management and the scrap pool/heap.

Edited June 30, 2025 by traison

[monsta88]

The latest Engine Fixes (7.0) solved my crashing problem.

[traison]

Not really seeing where specifically this bug might have been fixed. Seems there's been like 10 versions since 7.0 that messed with memory allocation settings and such, probably one of them did it.

 

7.0 included a PDB file finally.

 

bOverrideHavokMemorySystem might be interesting for Pandora. Just guessing.

[PippinTom]

6 hours ago, traison said:

bOverrideHavokMemorySystem might be interesting for Pandora. Just guessing.

 

That would be weird, because in my game, this is the only one of "new" features that I had to change from (default) ON to OFF on the first day because out of the blue it caused a streak of CTDs in ridiculous moments.

 

BTW: I don't even have Pandora.

[traison]

10 hours ago, PippinTom said:

...it caused a streak of CTDs...

 

I was thinking this might change the behaviour of the Helgen Havok crashes. I haven't really looked into them tbh, never bothered me enough since I use LAL. I just put it on my todo list as a thing to try in the future should I change my mind.