Spring 0.79.0.2 frozen on "Finalizing..." stage

[billyboy747]

Spring 0.78 was working fine on my platform:

Ubuntu 8.10 Intrepid Ibex 64 bits
ATI Radeon HD 3870 512 MB
Intel Dual Core 3.16 GHz
8 GB RAM

Since I upgraded to 0.79, I often get stuck when loading a game on the "Finalizing..." stage, after loading Lua scripts and before watching the map. The hard disk keeps working and almost all my CPU power is sucked by Spring. I need to immediately switch to a terminal after launching the game so I can issue the "kill -KILL xxxxx" command for killing the spring process, then I relaunch the game and so on until it works. When I manage to load the game, the "Finalizing..." stage lasts fractions of a second. Very few times Spring crashed on version 0.78: the spring process got terminated by itself but leaved and orphan "pulseaudio" process I had to kill so that when relaunching spring I still had sound. Now, the "spring" process seems to enter into an infinite loop and I must kill it, and no "pulseaudio" process is left afterwards.

I tried to compile Spring and Springlobby myself, but I'm getting the same result. I compiled the last version of both (0.79.0.2 and 0.0.1.10467, respectively) with gcc/g++ version 4.3.2.

I also tried to completely remove Spring and Springlobby and reinstall everything, and deleting the $HOME/spring folder so it gets initialized from the scratch, then adding the maps and mods and executing spring with no Lua widgets at all, but there is no difference at all. When it works, it works with all my precedent widgets.

[Tobi]

If you start spring while having something like system guard open (something which can show graph of memory usage), check how memory usage develops.

Also post the last 3 lines of your infolog.txt after you killed it.

This sounds like the det/Bashar bug. It is being investigated

EDIT: added to mantis, http://springrts.com/mantis/view.php?id=1415

[billyboy747]

The memory usage increases up to consuming all my memory, then the whole system becomes barely responsive. The last 3 lines of the infolog.txt after killing spring are:

[ 0] Finalizing...
[ 0] Spring 0.79.0.2
[ 0] Build date/time: May 25 2009 12:39:44

As I said, sometimes it works, sometimes it gets frozen; I just retry under the same conditions until it works.

[billyboy747]

If the issue I'm having is due to some problem with the use of both cores of my CPU (or multithreading), I could try to compile spring for a single core use (or no multithreading), provided that somebody gives me the corresponding cmake parameter . . .

[lurker]

If you're in a position to compile spring, then can you compile it as 32-bit for now?

[billyboy747]

I'm on a 64 bit platform . . . will 32 bit code work? If so, which would be the cmake parameter for forcing 32 bit code generation? Well, it would be nice to know the complete list of cmake parameters.

[lurker]

If you get the libs, sure. It's a matter of how much you want to play spring right now.

[bashar]

This sounds like the det/Bashar bug. It is being investigated

I'm proud having a bug called Bashar. Begin of celebrity