Simpsons crash on GLES fixed #16483
Simpsons crash on GLES fixed #16483
Conversation
This reverts commit cbfa4bf.
|
Since all of the If it only fixes it when it's virtual, it makes me worry there's some heap corruption happening and this only "fixes" it because it moves memory around and makes the thing crashing dodge that corruption (but some other unknown thing is getting corrupted now.) -[Unknown] |
I tried to move it into |
|
This one is quite hard to justify heh, like, are we somehow saving/loading the vtable, or how could that possibly matter? That can't possibly work with ASLR... |
| @@ -1197,7 +1197,6 @@ bool TextureCacheCommon::GetCurrentFramebufferTextureDebug(GPUDebugBuffer &buffe | |||
| } | |||
|
|
|||
| void TextureCacheCommon::NotifyConfigChanged() { | |||
| clearCacheNextFrame_ = true; | |||
There was a problem hiding this comment.
What if all you do is remove this line move this to CheckConfigChanged? This actually didn't seem to happen before the same way, though I don't see why it would be crashy.
There was a problem hiding this comment.
Tested removing and that fixes the crash. Just turned everything off, I can try to move it tomorrow.
|
Actually this is a small change in behavior. If my suggestion above doesn't help, I'm ok with reverting this, I'll do a second version of the cleanup later on top. |
Tested on Pico Neo 3 and Pico 4 VR headsets. Loading saved game renders one frame and then it crashes on texture destroying with a broken pointer (0x1FFF...FFF). No idea why is it happening, I only found the commit from which it started happening and reverted it.
This reverts commit cbfa4bf.