[Therion] loch segmentation fault at launch

Thu Nov 21 14:14:18 CET 2024

Hmmm...   I made a coredump file available.  Is that not useful?

How is a debug version of loch created?

The crash does not show up in abrt or the ProblemReporting application.

I used gdb to run loch.  Here are the last few lines of gdb output. 
After loch crashed I ran the "bt" command in gdb.

=======================
Downloading 182.06 K separate debug info for 
/usr/lib64/gio/modules/libdconfsettings.so
Downloading 28.11 K separate debug info for 
/home/bgee/.cache/debuginfod_client/55594807e24e127c88a7d823b293bb334e0cea3c/debuginfo
[New Thread 0x7fffe26006c0 (LWP 2587396)] 

[New Thread 0x7fffe1c006c0 (LWP 2587397)]
Downloading 716.23 K separate debug info for 
/usr/lib64/gio/modules/libgvfsdbus.so
Downloading 320.29 K separate debug info for 
/home/bgee/.cache/debuginfod_client/1c817f19fbcb65d0d279cc2d0867d3e3a018e048/debuginfo
Downloading 757.83 K separate debug info for 
/usr/lib64/gvfs/libgvfscommon.so

Thread 1 "loch" received signal SIGSEGV, Segmentation fault.
Downloading 24.37 K source file 
/usr/src/debug/libX11-1.8.10-2.fc41.x86_64/src/xcb_io.c
0x00007ffff6f5f344 in require_socket (dpy=0x777860) at 
/usr/src/debug/libX11-1.8.10-2.fc41.x86_64/src/xcb_io.c:70
70                      if(dpy->xcb->event_owner != XlibOwnsEventQueue)

(gdb) bt
#0  0x00007ffff6f5f344 in require_socket (dpy=0x777860) at 
/usr/src/debug/libX11-1.8.10-2.fc41.x86_64/src/xcb_io.c:70
#1  _XFlush (dpy=0x777860) at 
/usr/src/debug/libX11-1.8.10-2.fc41.x86_64/src/xcb_io.c:606
#2  0x00007ffff6f5f64d in _XGetRequest (dpy=0x777860, type=98 'b', len=8)
     at /usr/src/debug/libX11-1.8.10-2.fc41.x86_64/src/XlibInt.c:1778
#3  0x00007ffff6f50217 in XQueryExtension
     (dpy=dpy at entry=0x777860, name=name at entry=0x7ffff748b002 "GLX", 
major_opcode=major_opcode at entry=0x85efc4, 
first_event=first_event at entry=0x7fffffffd0f4, 
first_error=first_error at entry=0x85efc8)
     at /usr/src/debug/libX11-1.8.10-2.fc41.x86_64/src/QuExt.c:49
#4  0x00007ffff74872b0 in InitDisplayInfoEntry (dpy=0x777860)
     at 
/usr/src/debug/libglvnd-1.7.0-5.fc41.x86_64/src/GLX/libglxmapping.c:643
#5  __glXLookupDisplay (dpy=dpy at entry=0x777860)
     at 
/usr/src/debug/libglvnd-1.7.0-5.fc41.x86_64/src/GLX/libglxmapping.c:759
#6  0x00007ffff748842b in glXQueryVersion (dpy=0x777860, 
major=0x7fffffffd1b4, minor=0x7fffffffd1b0)
     at /usr/src/debug/libglvnd-1.7.0-5.fc41.x86_64/src/GLX/libglx.c:1166
#7  0x00007ffff7f7ac35 in wxGLCanvasX11::GetGLXVersion () at 
../src/unix/glx11.cpp:769
#8  0x00007ffff7f7acb3 in wxGLAttributes::RGBA 
(this=this at entry=0x7fffffffd270) at ../src/unix/glx11.cpp:268
#9  0x00007ffff7f7b53c in wxGLCanvasBase::ParseAttribList
     (attribList=attribList at entry=0x7fffffffd300, dispAttrs=..., 
ctxAttrs=ctxAttrs at entry=0x0)
     at ../src/common/glcmn.cpp:212
#10 0x00007ffff7f7bb6c in wxGLCanvasX11::InitDefaultVisualInfo 
(attribList=0x7fffffffd300) at ../src/unix/glx11.cpp:752
#11 0x000000000042ba58 in ??? ()
#12 0x00007ffff729b9d2 in wxEntry (argc=<optimized out>, argv=<optimized 
out>) at ../src/common/init.cpp:481
#13 0x0000000000413ce8 in ??? ()
#14 0x00007ffff4c11248 in __libc_start_call_main
     (main=main at entry=0x413cc0, argc=argc at entry=1, 
argv=argv at entry=0x7fffffffd598)
     at ../sysdeps/nptl/libc_start_call_main.h:58
#15 0x00007ffff4c1130b in __libc_start_main_impl
     (main=0x413cc0, argc=1, argv=0x7fffffffd598, init=<optimized out>, 
fini=<optimized out>, rtld_fini=<optimized out>, 
stack_end=0x7fffffffd588) at ../csu/libc-start.c:360
#16 0x0000000000418945 in ??? ()
(gdb)

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

I see one interesting thing in there ...  The segmentation fault is 
apparently happening in the X11 library.  My system is running Fedora 
41, Wayland and KDE.  That means X programs must run through XWayland. 
Could there be a problem between loch and XWayland?

Confirmed - This problem has something to do with Wayland.  I logged 
into a test computer using X instead of Wayland.  Loch starts without 
error.  Logging out and back in with a Wayland session causes loch to fail.

===============
Bill Gee

On 11/21/24 00:45, Martin Budaj wrote:
> Hi,
> 
> no problem on Debian testing. It would help if you could debug it on 
> your system (backtrace output from gdb running loch compiled in the 
> debug mode).
> 
> Martin
> 
> 
> On Wed, Nov 20, 2024, 23:10 Bill Gee <bgee at campercaver.net 
> <mailto:bgee at campercaver.net>> wrote:
> 
>     I have heard nothing on this.  Therion 6.3.0 was released today.  I
>     downloaded from github and compiled.  It compiled with no errors.  :-)
> 
>     Loch still does not run.  :-(  Same problem - it throws a segmentation
>     fault during launch.
> 
>     Does anyone else see this problem?
> 
>     ===============
>     Bill Gee
> 
>     On 11/9/24 07:23, Bill Gee wrote:
>      > Testing out Therion 6.2.1+dev, I find that loch does not work.  When
>      > launched from a command line, it runs for one or two seconds and
>     then gives
>      >
>      > bgee at main2:~$ loch
>      > Segmentation fault (core dumped)
>      > bgee at main2:~$
>      >
>      > This is on Fedora 41, and it happens on several machines.
>      >
>      > If it helps any, I captured the core dump file and put it on a
>     server.
>      >
>      > https://campercaver.net/MiscFiles/lochcore.dump.zip <https://
>     campercaver.net/MiscFiles/lochcore.dump.zip>
>      >
> 
>     _______________________________________________
>     Therion mailing list
>     Therion at speleo.sk <mailto:Therion at speleo.sk>
>     https://mailman.speleo.sk/listinfo/therion <https://
>     mailman.speleo.sk/listinfo/therion>
> 
> 
> _______________________________________________
> Therion mailing list
> Therion at speleo.sk
> https://mailman.speleo.sk/listinfo/therion