rtcw linux - signal 11 in wolfmp

[pinkname]

Hello,

I couldn't resolve the issue of signal 11 in wolfmp. I used to play on linux, now I tried it again (same machine, same install, but years later) and now wolfmp segfaults (wolfsp works nicely - but who cares :d). Checked display modes, refesh rates (even 640x480@60), cfg removed, tested as root too, googled for an hour, and couldn't find a solution. The system is 32 bit x86. I repeat, sp works with sound perfectly. Video driver is radeon (not fglrx).

Any ideas, any clues by anyone? Any thoughts or links to be read would be welcome.

thx:

J.

(solving the issue would mean +1 player to join the scene)

['L0,]

Signal 11 is a generic error and one of the dumbest ones since it can be just about anything.

Would help if you would give some more info - what distro/version, how are you running it (wine/native)?

It is solvable. I've encountered it many times in past, but I'm not a linux (desktop) user so don't really recall the solution atm. Giving more detailed description may help me remember it, or at least someone else that's a native user may figure it out easier.

[Jun1oR]

I can't get rtcw to run on my linux machine either. I have tried multiple distros both x86 and x86_64. I am able to launch the game fine but all the menus are missing, I am also using the Radeon drivers from RPM Fusion. Both ET and ioRTCW run fine.

Since the game is old you need to make sure you have the required packages and files to run it, on some newer distros they are hard to find and you must manually download and install.

[pinkname]

Thanks for your words!

However let me repeat, _sp works perfectly_, I'm using linux as a profession since kernel 2.0, and rtcw should work just perfectly as it did a couple years ago on this same sytem (2.4, 2.6 kernels those times, but didn't touch rtcw setup since). Signal 11 is not a generic error, it shows segmentation fault (that can be caused by many things, true), rtcw has this problem usually when it can't touch legacy oss devices. In my current case it's not a problem, it can use /dev/dsp fine (both sp and mp), the log is nice and clean, segfault occurs after activating XFree86-VidModeExtension. Sometimes I have signal 4, depending on the resolution chosen, so I guess it's somehow related to X settings, maybe Gnome. I've tried 640x480@60Hz (r_mode 3, displayrefresh 60), I think it should work safely whatever I do. It switches to the given mode and leaves it there when quitting back to the shell.

I hope some of you already knows when this stuff happens and can spare me a night dealing with X settings whatever. Testing it on a clean X client without any window manager is yet to be done, atm the gui is in use, so I couldn't yet try, also I yet have to try it with a valid serial (even if I think it's unrelated).


Anyhow:

distro: debian testing
kernel: 3.9-1-686-pae
glx: gallium 0.4 (X.Org R300)
videohw: ati x1600 agp
compozite managers: none
gnome: gnome2 (not 3!)
sound: alsa (no pulseaudio or other softmixer) using an emu10k1 and jackd. Issue is the same without jackd.
rtcw: the latest native (can't check version atm, but I know it's the latest whatever it is), no wine involved, just whine :D


thx again:

J.

['L0,]

I'm not a native linux user so never really bothered with a roll signal plays in overall scheme of it. As far as rtcw goes, it usually segfaults with 11 and rarely with 7, at least that are my experiences. I got signal 11 for banal things like sending udp packets in past (modified version) thus it's pretty much generic use of it as far as rtcw goes imho - But in either case, if you say it's related to oss devices then great and I wont argue with you, I guess something new every time. :)

As far as your problem. After a nap I recalled how I solved my problem, it was with ia32-libs but since you're already on 32bit that's clearly not the reason. By checking google a little most of the problems are tied to sp rather mp.

In case if it helps, here's few solutions that were purposed (note they're for sp but may help):
http://ubuntuforums.org/archive/index.p ... 91932.html

Otherwise you can try to troubleshoot it with running +set developer 1 and see if there's any more intel about it. Personally I think it's just as usual, graphic drivers and packages..at least that's what people mostly complain about when facing problems with rtcw on linux.

In anyway, hope it helped in any way and GL. :)

[MAN-AT-ARMS]

Do you get the crash when using fglrx driver? You probably have to use the legacy driver for that card. I think its in the 'backports' repo (at least it is for stable)

How are you using Gnome2 on debian testing? Do you mean MATE?

Do you get the crash when using r_allowextensions 0?

[pinkname]

Thanks for the replay mr MAA!

- gnome: you are right it's gnome 3.16, I was misled by seeing the "gnome classic" desktop.

Didn't have much time experimenting since the first troubles, but I tried r_allowexetensions, tried running without a wm, without sound (s_initsound 0) but nothing changed. Sometimes I get signal 4, more often signal 11.

Getting lagacy fglrx (needed for X1600) work with newer kernels is a pain, since those don't support KMS, I spent some time with it bout a year ago before switching to gallium.

My next steps will be:

Check more xorg setups (ignore edid, different display - even if the modechange seems working), also I'll check the checksums of the paks, try a working fglrx setup.

I can hardly understand why sp does work and why mp does not, but I have no idea what might the engine differences be.

Any more ideas are more than welcome.

thx again:

J.

[pinkname]

Re dear helpers! :)

I had some time again to read and experiment, and I'm out of ideas for now.

fglrx: I won't be able to use it, since the latest version working with an x1600 (rv530) is 8.593 (catalyst 9.3), and this version doesn't work with kernels>2.6.28, nor x.org>1.12 (I have x.org 1.2.4-6 atm), so I'm stuck (with the otherwise nicely working) gallium driver.

gallium: I still have to resolve an issue with it, I can't set all modes that the display plugged on it supports, these _might_ (I'd give only a very slight chance to it) interfere somehow. I've even tried a different display, but no wonders did happen.

checksums: all fine

log: working sp log attached, interestingly mp log overwrites/creates logfile but it is empty, I copied last words from the shell to the also attached shellout.txt.

thx for reading!

J.


UPDATE: I don't see the attachments, so here come those last words inline:

----- Client Initialization Complete -----
----- R_Init -----
...loading libGL.so.1: Initializing OpenGL display
...setting mode 3: 640 480
Using XFree86-VidModeExtension Version 2.2
XF86DGA Mouse (Version 2.0) initialized
XFree86-VidModeExtension Activated at 640x480
Received signal 11, exiting...

[pinkname]

UPDATE:

I've checked everything I could, refresh issues solved, q3 works, et works, wolfsp (still) works, reinstalled rtcw (to be sure of the integrity) and everything remained the same.

I had an idea, maybe wolfmp uses some gl extensions that sp does not and that are not supported by the gallium driver. "Stippled" and "smoothed" primitives are such (http://www.x.org/wiki/RadeonFeature/), but I have no clue if they're even used in rtcw and if yes, how they are called in a q3/rtcw cfg.

Any further ideas are quite welcome. :X

J.

['L0,]

r_primitives (2) is usually used to solve the FPS issue. Default is 0 but 0 = auto and always results in 1 (single pass) which causes major fps drop for most modern ati users so try with 2 (multi pass).

[pinkname]

Thanks for the idea!

Don't know (didn't google after it) if it's related to galliums shortcomings or not, but I've tried. Didn't do a thing.

I've also tried copying the autogenerated sp cfg (even if it hasn't got all variables) to autoexec, didn't work either. I wouldn't think however there's any difference in sp or mp engine, I guess the fault is some very simple thingie.

I feel so stupid. I did some opengl programming on SGI some 20 yrs ago, but don't remember a thing (mkay, maybe I could rotate a textured cube :D), nor I'm sure these stuffs were implemented those times :D Anyhow, I'll look into the sources to know if these suspicions might even be valid or not. Quite an effort, would be cheaper to buy a new machine and not care about, but it's an office machine I can't really touch physically. :X

GRRR!

J.

[pinkname]

UPDATE:

The mp code definitely uses stippled primitives (linux_qgl.c). But the sp code also (same linux_qgl.c). As I see it's also linked the same way (except for linebreaks) in headers either sp or mp. Hmmh. I feel even more stupid than before.


J.

['L0,]

Engines are basically the same but since at one point they got separated into different branches so there are some differences between them, as well as different bugs etc. I never really bothered much with SP engine so can't state much but I do know that there are some differences.

As far as ati goes, it's general bs in any os. I can't even play 1.4 as whole computer freezes after a round or two and only hard boot solves it. On a side note, after 10.4 catalyst windows drivers went to hell as far as rtcw goes..so same crap i guess.

[pinkname]

Well, atm I think it's quite a waste of time to look at and compare mp and sp code, but it's interesting at least a little bit :>

off:

I'm not a fanguy of nvidia or ati (mkay, ati cards used to have better image quality about 10 yrs ago than geforces), I always hated drivers of both (linux and windows too), I guess it's the usual sw industry stuff: creating good software is expensive and takes time, why invest into it if marketing works better, faster and cheaper. Anyhow, this same machine was capable of running rtcw with fglrx drivers, and now after some yrs it doesn't work with gallium (and a completely different kernel and libs ofc). Linux sucks too, it just sucks differently, even if we can mostly blame hardware vendors for that :DD


J.

[MAN-AT-ARMS]

You just have an old card and Gallium driver isn't feature complete.