ID:2398483
 
BYOND Version:512.1448
Operating System:Windows 10 Pro 64-bit
Web Browser:Chrome 68.0.3440.106
Applies to:Dream Daemon
Status: Open

Issue hasn't been assigned a status value.
Descriptive Problem Summary: While hosting an SS13 codebase, we've been running into regular segmentation faults for quite a while now. They started roughly around the time when we made a huge push to incorporate some 511 icon and mutable appearance operations to get better performance out of the visual code. But we have been unable to nail them down to specifics, unfortunately, due to the sporadic nature of the crashes.

For the past weeks now I've also had a RAM monitor ticking away at DD to make sure that it's not us overrunning on memory. And can confidently say that, unless the build-up occurs in a time frame shorter than 60 seconds, it should not be a memory overrun issue (we regularly have ~1.4 gigabytes of allocated memory during the crashes).

On version 512.1448, the most common crash signature is as follows (5 of these over the past 24 hours):
Faulting application name: dreamdaemon.exe, version: 5.0.512.1448, time stamp: 0x5b84d13f
Faulting module name: byondcore.dll, version: 5.0.512.1448, time stamp: 0x5b84d0ca
Exception code: 0xc0000005
Fault offset: 0x000faecb
Faulting process id: 0xcc

There's also this (2 per last 48 hours):
Faulting application name: dreamdaemon.exe, version: 5.0.512.1448, time stamp: 0x5b84d13f
Faulting module name: byondcore.dll, version: 5.0.512.1448, time stamp: 0x5b84d0ca
Exception code: 0xc0000005
Fault offset: 0x000fae64
Faulting process id: 0x35c
Faulting application start time: 0x01d44867e62ca01c

Numbered Steps to Reproduce Problem: N/A

Code Snippet (if applicable) to Reproduce Problem: N/A

Expected Results: N/A

Actual Results: N/A

Does the problem occur:
Every time? Or how often? Sporadically, since late 2017.
In other games? N/A
In other user accounts? Yes.
On other computers? Yes.

When does the problem NOT occur? Not tested.

Did the problem NOT occur in any earlier versions? If so, what was the last version that worked? Issues started emerging since 511.

Workarounds: None found.

When you're up as high as 1.4 GB I wouldn't entirely rule out a memory issue, but I'll see if I can get any info from the fault offset you provided.
Well this is odd. I took a look and it appears it's failing to read density from a turf. That's the sort of thing I would have expected as fallout from the old bug where unique cell IDs got up past a certain point and started misbehaving, yet that bug was fixed in 512.1447.

Since you mentioned mutable appearances being involved, I wonder if the issue has something to do with appearance refcounts going awry. However without a way to narrow down the cause, this might be impossible to find.