Random crashes

[Blue Shadow]

Tested with: 2.9pre-1487-g6e6249f (32-bit)
Test material: Jenesis (This just happened to be the WAD I was playing)


Unfortunately, nothing to say other than ZDoom just crashes randomly, as I'm unable to identify a pattern which causes it.

Here are two crash reports: one I got on MAP01, and another on MAP02.

CrashReport.zip

CrashReport2.zip

Another one with the map file of the build:
https://www.dropbox.com/sh/5jr1mrivtyck ... iwvwa?dl=0

[Rachael]

Can you provide the map file? I cannot do anything with this crash data as this build does not appear on DRD Team...

Nevertheless, I suspect that it is related to this bug.

Also, in the future if you are using a 32-bit build, it would be good to specify.

[Blue Shadow]

Can you provide the map file?

OP updated (I hope I understood what you meant about the map file).

Also, in the future if you are using a 32-bit build, it would be good to specify.

I usually do, I just forgot this time - sorry.

[Rachael]

Yes - that is the file I needed - thank you.

As far as I can tell, the crash happened somewhere beyond the bounds of the map file. Unfortunately I am not experienced enough to know what this means. But my best guess is, that's where the assembly code is located. *If* I am correct, that means it's a drawer responsible for it, which points back to the bug I linked from QZDoom. If I am incorrect, then I really don't know what it could mean. (One thing that suggests I am wrong, is, the crash address is *way* out of bounds from the map file, from what I can tell)

[dpJudas]

Nevertheless, I suspect that it is related to this bug.

There's a very good chance it could be. I'll take a closer look and submit a patch as it may been introduced by my dancing sprites patch.

[Graf Zahl]

It's most likely in the renderer. The recursive call 013F0850 strongly hints at R_RenderBSPNode. But I cannot trace it from there, It seems that something in Windows has changed that makes it very hard to do address matching. Is there a reason you use 32 bit? It's a bit odd that nobody ever posted a 64 bit crash log.

[Blue Shadow]

Is there a reason you use 32 bit?

My compilation environment is currently set up to build 32-bit binaries only. I guess I'm going to have to look into switching to 64 bit, then.

[dpJudas]

Nevertheless, I suspect that it is related to this bug.

There's a very good chance it could be. I'll take a closer look and submit a patch as it may been introduced by my dancing sprites patch.

Sadly, this bug probably cannot explain this as all the FTexture::GetColumn implementations wrap if the U texture coordinate is out of bounds.

[Graf Zahl]

I think these have long been solved. If not, please reopen a report on the new tracker.