mojira.dev
MC-159315

Chunks generate in a non-threadsafe manner after special conditions

note: this MAY duplicate MC-315, however, the conditions to replicate this bug and crash are very different.

Chunk generation is not thread-safe on linux (ubuntu 18.4) under certain conditions,  leading to crashes.  Steps to reproduce: Can occur at random, but this is very rare and requires teleportation, a slow ticks-per-second, or other strange conditions. Otherwise, create a virtual partition of small size. mount it, move minecraft inside this partion, create a world, generate enough world-space that the partition becomes full and the game crashes. Move minecraft out of the virtual partition and load up the game file again. go to the part of the world where the crash occurred. Here, new chunks will generate in the wrong location (you can clearly see where they should have generated, but they are in the wrong position) and the game will likely crash after a minute or so. May be reproducible on windows, but I haven't tested, because virtual partitions are a lot less friendly on windows.

See these screenshots I took after spending a while filling up a 500mb virtual partition, the moving the save to a different directory:

Clearly, the chunks have generated in the wrong positions; the correct position can be seen in the other screenshots.

after getting these screenshots, the game crashed. log file is attached.

 

The savefile may keep producing these errors, so, if necessary, I can provide it via dropbox; the player should appear in a village that is quite far from spawn; fly a few more chunks away from spawn and the glitch will begin to occur as new chunks are generated.

Related issues

Attachments

Comments

migrated
[media][media][media][media]
violine1101

Does MC-150202 describe your issue?

CubeTheThird

This seems related to, if not duplicating these three reports: MC-150202, MC-158988, MC-74762.
Note these have all been fixed for a future release.

migrated

 MC-74762 makes note of the full disk null-pointer-error that caused my issue. Seems I was barking up the wrong tree when I was looking at MC-315. Thanks, I think this should be marked as a duplicate of that issue

migrated

(Unassigned)

Unconfirmed

(Unassigned)

crash

1.14.4

Retrieved