Memory Leak from a mesh existing in your presence
Describe the bug?
When working with convex hulls in a new thing for a game jam i started crashing, thinking it was bepu crash i then went down that avanue. i then had my friend techno test it out and, they ran out of ram!, so turns out, this mesh ive created in game causes a runaway memory leak that can crash games in a matter of a mere minutia of minutes!
To Reproduce
spawn out this item: resdb:///fc91ce3528a9eddb42d7544e1540882c4cfd9635e4b68870c517ccdd8bc3b14b.brson
wait till the game eventually dies or your system runs out of ram
Reproduction Item/World
resdb:///fc91ce3528a9eddb42d7544e1540882c4cfd9635e4b68870c517ccdd8bc3b14b.brson
Expected behavior
the game to not infinitely gain ram from an item simply existing
Screenshots
No response
Resonite Version Number
2025.11.24.1340
What Platforms does this occur on?
Linux, Windows
What headset if any do you use?
No response
Log Files
reddspc - 2025.11.24.1340 - 2025-11-28 16_02_58.log due to it being a system level crash its not extremely usefull but theres the fe log. heres the other more important stuff tho:
Additional Context
im not sure exactly which mesh out of the 20ish causes the leak but ill try to figure it out
Reporters
redd56 navy3001 technocatron
Hello! Here are the results of the automated log parsing:
| Version | OS | CPU | GPU | VRAM | RAM | Headset | Plug-ins/Mods | Renderer | Clean Exit |
|---|---|---|---|---|---|---|---|---|---|
| Beta 2025.11.24.1340 | Steam Runtime | AMD Ryzen 7 5800X 8-Core Processor | Navi 32 [Radeon RX 7700 XT / 7800 XT] (rev c8) | 31.36 GB | 62.72 GB | SteamVR | no | Renderite.Renderer.Unity 2025.11.18.1147 (2019.4.19f1) (WindowPtr: 0x30080 | ❌ |
This message has been auto-generated using logscanner.
forgot the bootstrapper and coredump, give me like 2 minutes to nab em (if i have coredumps enabled)
okay the coredump is gonna take a few minutes as its too big for github. 375 megs to be exact, ill link it via a google drive link
This is a known bug with the current version of Bepu. It will be fixed once it's updated as mentioned here https://github.com/Yellow-Dog-Man/Resonite-Issues/issues/4748#issuecomment-3043632524
coredump is uploaded, if your on the team and want to take a crack at it shoot me a dm and ill send ya the link since it might have my reso password in it
this kind of problem hasnt been documented yet for the "bepu bug". ive ran into pretty much every single other bepu bug possible (and made like 4-5 issues on them) yet this one is new. plus it might not even be a bepu bug, i still have a bit more i can do to narrow it down
It may not be tracked in the resonite issues currently, but was reported here with more details (and the PR that fixed this issue) https://github.com/bepu/bepuphysics2/issues/325
quite possibly it could be that but 325 talks about explosive levels of growth, this bug takes about 2-3 minutes to go from 2 gigs to 5 (then crash) a few orders of magnitude slower than the described bug
I was also able to narrow it down to an exact mesh and exact cause, it ineed seems to be a convex hull collider, but wierdly enough way slower than the bepu repo describes. about 100 megs a second at most. specificly this mesh: resdb:///66074bf5ec0c28a49acc684e148fb50884989520d034efc8d1a5b2295ddf8a9e.meshx so feel free to paste that into a static mesh component to test or just use resdb:///833653d89143e0e2d021d7e04c5b41e47f24389203c825dc467ea958657590bb.brson since its already set up. none the less its a much slower burn than in https://github.com/bepu/bepuphysics2/issues/325 and thusly probably not the same issue