When I enable Autogenerated Vegetation in the render settings and fly the UFO, I get a segfault within a couple of minutes. I'm on Debian 12 with manually build latest 'next' branch. I have many backtraces I can share.
#0 0x00007fe5ac0a794a in __dynamic_cast () from /lib/x86_64-linux-gnu/libstdc++.so.6
[Current thread is 1 (Thread 0x7fe540d496c0 (LWP 3082))]
(gdb) bt
#0 0x00007fe5ac0a794a in __dynamic_cast () from /lib/x86_64-linux-gnu/libstdc++.so.6
#1 0x0000562ab6079cfd in simgear::QuadTreeCleaner::apply (this=<optimized out>, lod=...)
at /home/daniel/FGFS/src/simgear.git/simgear/scene/tgdb/TreeBin.cxx:354
#2 0x00007fe5b05c7b77 in osg::LOD::accept(osg::NodeVisitor&) () from /lib/x86_64-linux-gnu/libosg.so.161
#3 0x00007fe5b059a693 in osg::Group::traverse(osg::NodeVisitor&) () from /lib/x86_64-linux-gnu/libosg.so.161
#4 0x0000562ab5a67658 in osg::Group::accept (nv=..., this=0x7fe4d6afc370) at /usr/include/osg/Group:38
#5 osg::Group::accept (this=0x7fe4d6afc370, nv=...) at /usr/include/osg/Group:38
#6 0x00007fe5b059a693 in osg::Group::traverse(osg::NodeVisitor&) () from /lib/x86_64-linux-gnu/libosg.so.161
#7 0x0000562ab5a67658 in osg::Group::accept (nv=..., this=0x7fe50233a100) at /usr/include/osg/Group:38
#8 osg::Group::accept (this=0x7fe50233a100, nv=...) at /usr/include/osg/Group:38
#9 0x00007fe5b059a693 in osg::Group::traverse(osg::NodeVisitor&) () from /lib/x86_64-linux-gnu/libosg.so.161
#10 0x0000562ab60796c7 in osg::MatrixTransform::accept (nv=..., this=0x7fe4e13c7730) at /usr/include/osg/MatrixTransform:37
#11 simgear::createForest (forestList=..., transform=..., options=options@entry=0x7fe5041cace0, depth=depth@entry=3)
at /home/daniel/FGFS/src/simgear.git/simgear/scene/tgdb/TreeBin.cxx:456
#12 0x0000562ab620484d in SGTileDetailsCallback::generateRandomTileObjects (this=this@entry=0x7fe4e40bd4e0, matTris=...,
matcache=0x7fe5025dfba0) at /usr/include/osg/Matrixd:448
#13 0x0000562ab620609c in SGTileDetailsCallback::readNode (this=0x7fe4e40bd4e0)
at /home/daniel/FGFS/src/simgear.git/simgear/scene/tgdb/SGTileDetailsCallback.hxx:135
#14 0x00007fe5b0e39421 in osgDB::Registry::readNode(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, osgDB::Options const*, bool) () from /lib/x86_64-linux-gnu/libosgDB.so.161
#15 0x00007fe5b0e3497e in osgDB::DatabasePager::DatabaseThread::run() () from /lib/x86_64-linux-gnu/libosgDB.so.161
#16 0x00007fe5b0c1a67e in OpenThreads::ThreadPrivateActions::StartThread(void*) ()
from /lib/x86_64-linux-gnu/libOpenThreads.so.21
#17 0x00007fe5ac2a8134 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#18 0x00007fe5ac3287dc in ?? () from /lib/x86_64-linux-gnu/libc.so.6
I believe this is resolved and I'm unable to reproduce the issue. Please retest to confirm.
It's still happening. Before there were two different faults, one segfault and one abort. This time I triggered the abort. I didn't test yet if the segfault still happens.
Last lines in the terminal:
Backtrace:
Problem confirmed by two more users (rooman, southside) in the FG forum:
https://forum.flightgear.org/viewtopic.php?f=45&p=422670#p422647
We are all on Debian/Ubuntu, if that matters.
Last edit: datsf 2024-06-25
Also on Fedora 39, Ryzen 16GBy Nvidia 750Ti on a build fetched less than 5 days ago. Running with a debugger shows various endpoints on backtraces but often during object deleting. Running fgfs with Auto Generated Scenery with 'top' in a separate terminal shows fgfs memory consumption steadily increasing until swapping starts. When the swapfile is approximately half full the kswap0d daemon bcomes the dominant cpu load and fgfs stops with a segfault .
Departing westward from KGRR, Grand Rapids at low altitude demonstrates memory consumption increases over rural terrain; memory use does not decrease as the scenery changes to plain water, over the great lake.
I'm running with no multithreading selected and when I've tried setting database-pager-threads to 1 : the sim runs slower but for much longer without error. Fgfs log shows many errors viz:
/fgfs/simgear/simgear/scene/tgdb/TreeBin.cxx:507: Error parsing tree entry in: Path "/fgScen/tsyn/Trees/w090n40/w086n45/w090n40_w086n45_1548792_SubUrban_trees_shader.txt" line: ""
Attached is a screen grab of the backtrace:
After fetch & pull 26Jun with changes to treebin: