Open mt-from-sf-to-gh opened 11 years ago
Original comment by: wolfgang-ch
Original comment by: wolfgang-ch
there was no complaint
Original comment by: wolfgang-ch
Hello Wolfgang, can you reopen this bug? It's being a problem for me on Gentoo 64bit, at least since 14.10 - mostly random, hard to reproduce crashes, which would sometimes produce a stack trace file. Now with 15.3.2, the crashes seem to be much more frequent, but I have yet to see some debug info generated.
I attached one log file from 15.3.0.
I also have a core dump from 14.10 while I tried running under gdb, but that file has almost 1G. If you think it would help, I can upload it somewhere.
I'll be happy to help debugging, I use Linux and Eclipse everyday for C++ development. Last time I touched a Java code was probably around 2009, but I can give it a try :-)
Original comment by: sliwowitz
FYI, it also happens with cairo-1.14.2:
java: /home/portage-tmp/portage/x11-libs/cairo-1.14.2/work/cairo-1.14.2/src/cairo-surface.c:928: cairo_surface_reference: Assertion `((*&(&surface->ref_count)->ref_count) > 0)' failed.
Original comment by: sliwowitz
The crash occures in the Cairo code, when I google for "eclipse crash cairo" I saw a solution by setting these in the .ini file (can also be set in the commandline):
-Dorg.eclipse.swt.internal.gtk.useCairo=false -Dorg.eclipse.swt.internal.gtk.cairoGraphics=false
Can you check if this works for you ?
This is the code where it crashes:
Stack: [0x00007fe35963d000,0x00007fe35973e000], sp=0x00007fe35973b210, free space=1016k Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code) C [libcairo.so.2+0x37ba4] cairo_image_surface_get_stride+0x4
[error occurred during error reporting (printing native stack), id 0xb]
Java frames: (J=compiled Java code, j=interpreted, Vv=VM code) j org.eclipse.swt.internal.cairo.Cairo._cairo_image_surface_get_stride(J)I+0 j org.eclipse.swt.internal.cairo.Cairo.cairo_image_surface_get_stride(J)I+8 J 602 C2 org.eclipse.swt.graphics.Image.createMask()V (387 bytes) @ 0x00007fe34d23aefc [0x00007fe34d23aec0+0x3c] j org.eclipse.swt.graphics.Image.createSurface()V+19 j org.eclipse.swt.graphics.GC.drawImage(Lorg/eclipse/swt/graphics/Image;IIIIIIIIZ)V+168 j org.eclipse.swt.graphics.GC.drawImage(Lorg/eclipse/swt/graphics/Image;II)V+44 j de.byteholder.geoclipse.map.Map.paintTile20_Overlay(Lorg/eclipse/swt/graphics/GC;Lde/byteholder/geoclipse/map/Tile;Lorg/eclipse/swt/graphics/Rectangle;)V+121 j de.byteholder.geoclipse.map.Map.paintTile(Lorg/eclipse/swt/graphics/GC;IILorg/eclipse/swt/graphics/Rectangle;)V+68 j de.byteholder.geoclipse.map.Map.paint_30_Tiles(Lorg/eclipse/swt/graphics/GC;)V+96 j de.byteholder.geoclipse.map.Map.paint_10_PaintMapImage()V+69 j de.byteholder.geoclipse.map.Map.access$18(Lde/byteholder/geoclipse/map/Map;)V+1 j de.byteholder.geoclipse.map.Map$15.run()V+33
Original comment by: wolfgang-ch
I suspected Cairo as well, but didn't find the settings you mentioned. Your Google-fu looks strong :-)
With those settings in the ini file, I still get a cras, but with a different error (attached). To crash mytourbook, I have to randomly click many things. It usually takes a couple seconds to get there.
Also - 15.3.0 and earlier sometimes generated a stack trace file (like the one I posted earlier), but 15.3.2 never does (I tried about 50 times today). Any idea why?
Original comment by: sliwowitz
Posting this here for future reference: It would be nice to know how can I "run it with the --sync command line", and "break on the gdk_x_error() function". I tired the tips here http://www.sourceware.org/ml/gdb/2009-12/msg00095.html to no avail. The I found someone who also doesn't know http://blog2.vorburger.ch/2014/08/java-8-swing-related-jvm-crash-on-linux.html (last paragraph).
A bit more googling got me here https://bugs.eclipse.org/bugs/show_bug.cgi?id=183421 and https://fedoraproject.org/wiki/Java/StackTraces?rd=JavaStackTraces
With the modified settings file even 15.3.2 crashes much less than before, if I don't actively tru to crash it, I can use it for whatever I need, so the bug moved to the "mild annoyance" area. I'll get back to it when I have some time to read that Fedora wiki page.
(Edit: reflecting on all that, it seems I might be hitting a different bug now)
Original comment by: sliwowitz
Just for the record - this is still an issue. Working with the 2D map will trigger this error sooner or later.
It might be in fact this bug in Cairo https://bugs.freedesktop.org/show_bug.cgi?id=91967 I tried to apply the patch to Cairo mentioned there, but it still crashes, just with a different error
Original comment by: sliwowitz
Original comment by: wolfgang-ch
It is still an issue in the current MT version 21.6.1
https://github.com/wolfgang-ch/mytourbook/issues/372#issuecomment-892687448-permalink
This issue was also documented in 12.12 homepage https://mytourbook.sourceforge.io/mytourbook/index.php/history/history-2012
Original comment by: wolfgang-ch
Original comment by: wolfgang-ch
During my testing, Ubuntu 12.04 crashed with this log entry in the Eclipse console:
mytourbook: /build/buildd/cairo-1.10.2/src/cairo-surface.c:637: cairo_surface_destroy: Assertion `((*&(&surface->ref_count)->ref_count) > 0)' failed. Aborted (core dumped)
I can't currently say exactly when the crash occures, I've not much Linux experience and did not find a dump file.
If your Linux is crashing when using the new feature to sync photos and tours, report it in this bug tracker, hopefully with a Java stacktrace.
Reported by: wolfgang-ch
Original Ticket: mytourbook/bugs/71