Debug problems with SVN 3486

[Raistlin]

Hi,

i encounter sporadic program crashes with SVN version 3846. So i wanted to generate a backtrace with gdb, but when starting the game in gdb i got the following message and the start was aborted.

=======================================================
(gdb) run +set vid_fullscreen 0 +set vid_grabmouse 0 +set developer 1
Starting program: /home/tron/Entwicklung/ufoai/ufo +set vid_fullscreen 0 +set vid_grabmouse 0 +set developer 1
[Thread debugging using libthread_db enabled]
[New Thread 47903566773424 (LWP 5065)]
Adding game dir: ./base
using /home/tron/.ufoai/base for writing
Adding game dir: /home/tron/.ufoai/base
execing default.cfg
execing config.cfg
execing keys.cfg
Console initialized.

------- sound initialization -------
Loading snd_sdl sound driver
Soundsystem: SDL.
SDL version: 1.2.9
SDL audio driver is "alsa".
Bits: 16
Frequency: 48000
Samples: 4096
Channels: 2
[New Thread 1082132800 (LWP 5070)]
sound sampling rate: 48000
------- Loading ref_glx.so -------

Program received signal SIG33, Real-time event 33.
[Switching to Thread 1082132800 (LWP 5070)]
0x00002b9169e705c7 in sem_post () from /lib64/libpthread.so.0

================================================

What is the error?
____________________
-Raistlin

[Raistlin]

Hi,

i tried it under windows and i had the same problem.
here is the corresponding backtrace.


Building to ensure sources are up-to-date
Build succeeded
Selecting target: default
Adding source dir: C:\Entwicklung\ufoai\src\
Adding source dir: C:\Entwicklung\ufoai\src\
Changing directory to: ../../ufoai
Adding file: ..\ufo.exe
Starting debugger: done
Registered new type: wxString
Registered new type: STL String
Registered new type: STL Vector
Setting breakpoints
Debugger name and version: GNU gdb 5.2.1
Program received signal SIGSEGV, Segmentation fault.
In _libwinmm_a_iname () ()
#0  0x7c911010 in _libwinmm_a_iname ()
No symbol table info available.
#1  0x77c0b90d in _libwinmm_a_iname ()
No symbol table info available.
#2  0x77c120e7 in _libwinmm_a_iname ()
No symbol table info available.
#3  0x003d659b in ?? ()
No symbol table info available.
#4  0x003d65f9 in ?? ()
No symbol table info available.
#5  0x611cecc0 in R_BeginFrame ()
No symbol table info available.
#6  0x00432efe in SCR_UpdateScreen () at client/cl_scrn.c:682
   numframes = 1
   i = 0
   separation = {0, 0}
#7  0x0041d927 in CL_Frame (msec=1) at client/cl_main.c:1604
   extratime = 0
   lasttimecalled = 0
#8  0x004594b9 in Qcommon_Frame (msec=1) at qcommon/common.c:1536
   s = 0x0
   time_before = 0
   time_between = 0
   time_after = 0
#9  0x0044e841 in WinMain (hInstance=0x400000, hPrevInstance=0x0, lpCmdLine=0x241f04 "", nCmdShow=10) at ports/win32/sys_win.c:622
   msg = {hwnd = 0x640380, message = 512, wParam = 0, lParam = 2031856, time = 4545375, pt = {x = 243, y = 60}}
   time = 1
   oldtime = 23389
   newtime = 23390
   timescale = 1
#10 0x00482efa in main ()
No symbol table info available.

________________________
-Raistlin

[Bandobras]

You are running SVN trunk, I suppose? Could you revert to somewhat earlier revisions and tell from revision the problem starts? Bisection is your friend...

[ubequitz]

It is possible the two errors you are seeing on Windows and Linux are different (I'm not sure). Can you try trunk SVN 3498 under Windows? A change I made to  src/client/cl_menu.c has a (very small) chance of fixing your issue.

Can you also try reproducing your crash under Linux in gdb and get a backtrace?

EDIT: oh I see you already tried getting a bt under Linux
very odd behavior crashing at such an early stage...  are you sure your Linux video card drivers are setup ok (generally you have to use the proprietry binary drivers)

[Kerner]

Code: [Select]


Program received signal SIG33, Real-time event 33.
[Switching to Thread 1082132800 (LWP 5070)]
0x00002b9169e705c7 in sem_post () from /lib64/libpthread.so.0
================================================
What is the error?
____________________

This is a position where gdb stops here to(SuSE 10.1),but if i enter "continue" it simply goes on.

[Mattn]

a crashed thread - maybe x or sound or something like that. any suspicious error messages before this crash that would help us to fix it?