corrupted size vs prev_size

[przem360]

Hi!
Today I was trying to run Tic-80 on my Linux Mint 19.1 Cinnamon, but it failed wit a message:

corrupted size vs. prev_size
aborted (memory dump)

It also may by problem with my pc because tic-80 was working just fine few weeks ago. Perhaps this is due to some dependency updates?

[mt12345]

The same error on Ubuntu MATE 18.04

[przem360]

I do not know why, I do not know how... but it works again.
In history.log of my apt I cant see nothing related to TIC-80.
But it works :)

[nesbox]

Hmm, can't reproduce...

[dreamwebspace]

I had this problem on Ubuntu 18.04, after upgrading to 19.10 it works fine.

[farvardin]

I'd suggest to reopen this one.

This problem occurs on one of my computers, and I can't understand why. I can investigate further if you can point me to some tests.

I've made all the updates on my Linux mint 19.3, and it's the same. Some weeks ago it was working ok on the same computer. I have also other computers with the same OS, without any problem on them.

This computer is running Intel graphical Card ("Intel Corporation 4 Series Chipset Integrated Graphics Controller")

I've run it with gdb, and got:

Reading symbols from tic80...done.
(gdb) run
Starting program: /temp/github/TIC-80/build/bin/tic80
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[New Thread 0x7fffe8fd1700 (LWP 16267)]
[New Thread 0x7fffe87d0700 (LWP 16268)]
[New Thread 0x7fffe7fcf700 (LWP 16269)]
[New Thread 0x7fffe77ce700 (LWP 16270)]
[New Thread 0x7ffff7fa6700 (LWP 16271)]
[New Thread 0x7ffff7fe0700 (LWP 16272)]

TIC-80 tiny computer 0.90.1392-dev Pro
http://tic80.com (C) 2020
corrupted size vs. prev_size

Thread 1 "tic80" received signal SIGABRT, Aborted.
__GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
51 ../sysdeps/unix/sysv/linux/raise.c: Aucun fichier ou dossier de ce type.
(gdb)

but not other information.

[nesbox]

@farvardin investigate please, I can't reproduce it on my side.
Thank you.

[farvardin]

I can investigate a bit yes, but I'm not a great developper :D

Here is what I've tried so far:

Computer side

I've run a testdisk with no reported error. I suppose my memory is not faulty.

TIC-80 versions (from the release tab in github) :

• The 0.60.3 (32 and 64 bit) version is working, I can launch it without problem on this computer
• 0.70.1, 0.70.6 and 0.80 : same "corrupted size vs. prev_size" error
• 0.80 windows version: I can run it on the same OS with wine.

but I've tried again the 0.70.1 version after running wine, and I ran once, without problem. I've closed it, ran it again, and it crashed with the corrupted size vs. prev_size error

So I've started this 0.70.1 version several times in a row (8 times), I crashed all the time. I've started again the wine version, which used some memory, closed it, started the 0.70.1 version, it crashed again the first time, but after I've started the wine version again, then the 0.70.1 version (on linux) started correctly!

I remember some months ago, when TIC-80 was working (I suppose I was running a pre-80 version I've compiled myself), it crashed once during start-up, then I closed firefox which freed some memory and I could start TIC-80 again. I thought it was a temporary memory error. It really seldom crashed. Now it's most of the time.

I finally managed to run the 0.80 linux version (a post-0.80 dev version I've compiled myself) after running wine several time + the 0.70.1 version.

It also worked once after I started it 12 times!

The 60.3 version works all the time on this computer.

Libraries

I've compared the library used.

basically, what is in addition in the 0.7 version are:
libGLdispatch.so.0 => /usr/lib/x86_64-linux-gnu/libGLdispatch.so.0
libGL.so.1 => /usr/lib/x86_64-linux-gnu/libGL.so.1
libGLX.so.0 => /usr/lib/x86_64-linux-gnu/libGLX.so.0

and in 0.8 :
libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1
libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6

what is strange is

libSDL2-2.0.so.0
libvorbis.so.0
libpulse.so.0

from the 0.6 version are no longer present in further releases. I suppose it is handled differently.

Code

I don't know C very well (it's rather almost nothing). I've just searched for "corrupted size vs prev_size".
I've found this: pjreddie/darknet#105

Someone said: "This happens when you are using memory which was not allocated to you or when you are using free() twice in a row on a pointer."

But I don't know more.

[nesbox]

I added Ubuntu build without SDLGPU lib support, could you please check it?
Download here https://github.com/nesbox/TIC-80/actions/runs/361286385
tic80-ubuntu-sdl and tic80-ubuntu-sdlgpu

Thank you.