[Hellzed]

Hi,
I was playing BP:WiH with the 3.6.12 stable build of fs2_open, or the 3.6.13 unstable either... I compile it myself for Ubuntu maverick 64 (tested : no CFLAGS, or custom ones). I also tested antipodes 6, changes nothing, the same crach happens.
The game keeps segfaulting after checkpoint 1, while i'm killing some bombers if i started the mission at the checkpoint, or before the Indus disengages if i starded from the beginning.

I want to know the end of WiH !
Thanks for your help.

I attached the log (made with fs2_open 3.6.13 debug build), and i can use some more debug tool if you can explain me how to use them !

[attachment deleted by admin]

[General Battuta]

Blue Planet requires the 3.6.12 MediaVPs; it looks like you're still using the 3.6.10s.

Download the 3.6.12s, put them in a folder called 'mediavps_3612' (it MUST have this name), and edit your mod.ini back to the default so it points to mediavps_3612.

Hopefully it'll fix it. Also everything will get much prettier. 

[Hellzed]

I am already using the mediavps_3612 (I use my own launcher script, and in the mod flag i have "-mod blueplanet2,blueplanet,mediavps", it already lokks great !). That was just a name problem !

Here's the complete command line :  "-glow -spec -env -ambient_factor 35 -ogl_spec 20 -spec_exp 15 -spec_point 1.2 -spec_static 1.5 -spec_tube 1.5 -missile_lighting -normal -3dshockwave -3dwarp -warp_flash -snd_preload -orbradar -ship_choice_3d -weapon_choice_3d -ballistic_gauge -dualscanlines -mod blueplanet2,blueplanet,mediavps_3612" (I renamed my mediavps).

[Herra Tohtori]

Download the 3.6.12s, put them in a folder called 'mediavps_3612' (it MUST have this name), and edit your mod.ini back to the default so it points to mediavps_3612.

You're assuming he's using a launcher of some sort that actually uses the mod.ini file...

EDIT: There you go, like I suspected from a linux user he's using a script.

Hellzed: You can use a debug filter: Make a new file in ~/.fs2_open/data/debug_filter.cfg

If you leave it empty, it should automatically generate a setup that shows most debug info that is normally filtered; my debug_filter.cfg is

Code: [Select]

+General
+Parse
+Error
+BmpMan
+Paging
+Network
+JOYSTICK
+BmpFastLoad
+Sound
+SEXP
+Model
+Maps
+Warning
+wash
+AI
+AB TRAIL
+Weapons
+BmpInfo
+BMP DEBUG
+Alan
+lethality
+Messaging
+Particles
+Sandeep
+EVENTMUSIC
+Physics
+Jim
+missionlog
+allender
+BEAM
+lightning
+OpenGL
+ModelAnim
+Fred routing
+OpenAL
+SHADER-DEBUG

Though I do not know if this will actually help at all in solving the actual crash. Also be warned that the debug log will be huge with this setup for obvious reasons...

[General Battuta]

Okay. Both blueplanet2 and blueplanet look for the mediavps in a folder called mediavps_3612, so just make sure it's there.

Past that, I'm...

...well, I'm stumped for the moment.

Make sure you've updated your AoA Director's Cut install to the latest. I'll direct higher-level tech support ninjas this way.

[Hellzed]

I remember already having this problem with fsport 3.1, something was going wrong and the same mission always crashed...
Then fsport 3.2 fixed it, and i never knew what happened.
I just noticed a group of things appearing like Erynies on the radar with the icon pack, i can see them warping in, but i can't see the models or target anything. But they are targeted and destroyed quickly by my wingmen and by the cruisers... Has this something to do with the bug ?
I already have AoA director's cut. I checked all checksums (mediavps and bp2 too), they are all OK.

When I use low details, the orion class is transparent blue when i'm far (not invisible like the Erynies) and turns good when i'm get nearer, but it seems to be a minor LoD bug.

New log here : http://ubuntuone.com/p/IT0/

I just tested antipodes 6, performance is better, but the game still crashes (definitely not a perf issue)...

[The E]

Is there any sort of callstack in the terminal window after the crash?

[Hellzed]

Code: [Select]

Segmentation fault
** Fatal error : Error while running "/home/loris/Jeux/FreeSpace2/fs2_open_r" with arguments   "-glow -spec -env -ambient_factor 35 -ogl_spec 20 -spec_exp 15 -spec_point 1.2 -spec_static 1.5 -spec_tube 1.5 -missile_lighting -normal -3dshockwave -3dwarp -warp_flash -snd_preload -orbradar -ship_choice_3d -weapon_choice_3d -ballistic_gauge -dualscanlines -mod blueplanet2,blueplanet,mediavps_3612" (139)


That's all...

[Hellzed]

Can I use some BP developpement data ? Perhaps it's a small problem in the data (like it seemed to be with fsport 3.1, fsport 3.2 solved the bug)...

[General Battuta]

Can I use some BP developpement data ? Perhaps it's a small problem in the data (like it seemed to be with fsport 3.1, fsport 3.2 solved the bug)...

How so? You can feel free to redownload the War in Heaven VPs, if that's what you're asking, but this error is...really puzzling me. It seems like it should be present for everybody.

[Hellzed]

I reinstalled blueplanet2, built fs2_open again... I cleared the cache in .fs2_open too...
I checked all the md5 for mediavps, they are still ok.
I still can't see some GTVA ships after they warped in, they vanish !
The bug still happens, and still during the withdrawal of the Indus if i begin the mission from the start, and when fighting bombers after checkpoint 1 if i begin from this point !

I checked the CFLAGS, the game behaves the same.

I'll need help from fs2_open gurus, because i'm running out of options.

Could anyone send me some pilot profile to check if mine is corrupted ?

[General Battuta]

 

I'm sorry, mate, I don't know enough to help you. Sounds awful.

You can go to the next mission by hitting ctrl-shift-S in the techroom if you want to see the ending, but you may spoil yourself.

[Hellzed]

Wow ! the end is great ! I'm looking forward to play WiH part 2 !

It seems I didn't miss anyting important, because the bug happened probably just before the Indus crash jumped (triggered by a bug in the SEXP ?).

I noticed Na'ima is the aramean for Noemi... Sounds shivan !

[Hellzed]

I think i found something : i played with the compile options among other things, and i found this :

Code: [Select]

Loading SunGlow.dds for the first time.
Bitmap is to small at 0x0.
Int3(): From graphics/gropengltexture.cpp at line 1069
Is it somehow important ?

[The E]

What compile options did you play with?

[Hellzed]

The CFLAGS ( CFLAGS="-march=core2 -O3 -pipe" CXXFLAGS="${CFLAGS}" ), and i always use inferno builds. I tried without it, the result is the same. The flags never caused me any trouble (on my computer, they speed up the game if used). But i don't remember exactly how i did get to this. Now i have some doubts about the importance of the information, since my problem seems architecture related (is it even possible ?)
Debug builds (3.6.12, 3.6.13, antipodes 6) fail on loading this mission, but no other mission. Release builds fail always at the same point during the mission.
It could be a model problem combined with a fs2_open bug, or something with the save and restore script (using checkpoints change the crash point). The other mission with a checkpoint crashed once, i had to clear the cache in .fs3_open, load the mission then load the checkpoint to get it passed.
There's definitely something odd here, while the game installation is not corrupted, and the same bug happens with all recent builds.
I could try an older graphics driver, but it will cause me some trouble for now (performance loss in other apps).

I also tried disabling all video enhancement settings (-normal -glow ..., keeping just the mod and some other flags needed by the mod). Changes nothing.

I'm running the game on a Dell XPS m1330 (Core2Duo T9300, 4Gb RAM, nvidia 8400m GS). It's Ubuntu 10.10, amd64, and nvidia proprietary driver 260.19.06. FS2 main campaign is ok, fsport 3.2 too (while i remember having trouble with fsport 3.1, don't know why, but likely the same problem, segfault caused by a mod). Derelict seems ok.

I also tested creating a new pilot then skip missions to bp2-15.fs2. The bug still happens.

[Iss Mneur]

FS2_open should never be built with -O3.
Never!
Never! 

Have I made myself clear yet?

-O1 and -Os are fine, but FS2_open breaks some assumptions of -O3 (and -O2 for that matter) and that will cause random, unexplainable errors.

Never build FS2_Open with -O3!

[Hellzed]

Compiled with O1, still crashes.

[Iss Mneur]

Compiled with O1, still crashes.

Have you tried without any -O at all?  I know said -O1 should be fine, but we may broke that since last time.  As far as I know we don't build FSO with any optimizations at all, so...

[Hellzed]

Yeah, I tried, it's still crashing.
Though, -O3 never caused me any trouble one Core2Duo architecture (some trouble on i3), and this speeds up the game a lot.
thanks for helping ! We'll have to search harder !