-
Notifications
You must be signed in to change notification settings - Fork 157
Avo does no longer start #90
Comments
Logged In: YES
Original comment by: @milkypostman |
Logged In: YES kde4@moinmoin:~/build/avo/avogadro/src> ldd -v avogadro
Original comment by: @cniehaus |
Logged In: YES Ok, I have no idea if this helps or not, but it also crashes (or whatever it does) in Valgrind 3.2.3. kde4@moinmoin:~/build/avo> valgrind -v avogadro/src/avogadro ==10501== Original comment by: @cniehaus |
Logged In: YES As you can see in my previous post these are the important information we need to debug this: ==10501== at 0x4C07C4E: __lll_mutex_lock_wait (in /lib/libpthread-2.5.so) Please note that I don't know what this means, but at least we know where it happens... Original comment by: @cniehaus |
Logged In: YES I tried to debug this further. In latest trunk I changed one line in avogadro/src/main.cpp: qDebug() << "0"; Application app(argc, argv); qDebug() << "1"; On the CLI I see the 0 but not the 1. So this is somehow in the "Application app(argc, argv);" I just also tried the current branch of the bondcentric tool. It starts just fine. Original comment by: @cniehaus |
Logged In: YES As Qt 4.3.1 is out now I removed all of KDE4 and Qt4 from my system, removed almost all of my development-user, removed the $BUILDDIR, installed Qt 4.3.1 and started from scratch. The issue remains. Original comment by: @cniehaus |
Logged In: YES With that diff it "work". Works means: it starts and I can see a rendered molecule before Avo crashes. Marcus says the crash makes sense. I am now investigating further. carsten@moinmoin:~/svn/avogadro/trunk> svn diff Index: avogadro/src/main.cpp--- avogadro/src/main.cpp (Revision 631) #ifdef Q_WS_X11
Original comment by: @cniehaus |
Logged In: YES I have found some similar problems as of late (deadlock) and I am almost positive it's an Xlibs issue. Which makes me think that the problem we are having here is related to XLibs and the threading support. As far as i know we need this XLib threading support for the GL stuff, but I could be wrong and i'll have to check up on it more later (ie look back at the QT documenatation). Original comment by: @milkypostman |
Logged In: YES The problem I have been having is related to XCB (http://xcb.freedesktop.org/) which is a new X protocol C language binding (...) so anyways, in Gentoo you can get libX11 that uses XCB or one that doesn't. The one that uses XCB causes me all kinds of problems with the pthread mutex deadlock. So my question to Carsten is if he knows if OpenSuSE is using the XCB version of libX11? Original comment by: @milkypostman |
Logged In: YES Moin I just managed to crash Gwenview on the same machine. The backtrace looks very similar, so I guess this is definetly outside Avogadros scope? Note the pthread_mutex_lock () #9 0xb72403f0 in pthread_mutex_lock () from /lib/libpthread.so.0 Original comment by: @cniehaus |
Logged In: YES Still working with XCB guys and other stuff to track down this error. In the meantime I have disabled the threaded OpenGL widget but you can enable it with a compile-time flag. To enable do 'cmake -DENABLE_THREADED_GL ..' posted fix in svn head revision 650 Original comment by: @milkypostman |
Logged In: YES As I already stated on the mailinglist: Donalds commit (I think a later one, 651) fixed it for me. The interesting thing is that I don't even use XCB! ldd /usr/lib/libX11.so Original comment by: @cniehaus |
Logged In: YES As Carsten said below, this is now fixed. Marking the bug closed. Original comment by: @ghutchis |
Added undo features for all crystal operations
Since the threading-branch was merged into trunk Avogadro no longer starts.
I already reported this by mail:
http://sourceforge.net/mailarchive/forum.php?thread_name=c048b1c20707210441s144e9d3bg9443a3d12e93b852%40mail.gmail.com&forum_name=avogadro-devel
I am using OpenBabel 2.1.1, tested this with both CMake 2.4.6 and 2.4.7, using a patched Qt 4.3.0 (patches from the snapshorts to 4.3.1 as provided by OpenSUSE). All other applications (also those using threads in Qt) work without issues.
Here is what strace tells me:
chdir("/home/kde4/build/avo") = 0
stat64("/usr/lib/qt4/plugins/inputmethods/libqimsw-multi.so", {st_mode=S_IFREG|0755, st_size=27444, ...}) = 0
stat64("/usr/lib/qt4/plugins/inputmethods/libqimsw-multi.so", {st_mode=S_IFREG|0755, st_size=27444, ...}) = 0
stat64("/home/kde4/.config/Trolltech.conf", {st_mode=S_IFREG|0600, st_size=27856, ...}) = 0
open("/home/kde4/.config/Trolltech.conf", O_RDONLY|O_LARGEFILE) = 8
close(8) = 0
stat64("/home/kde4/.config/Trolltech.conf", {st_mode=S_IFREG|0600, st_size=27856, ...}) = 0
stat64("/home/kde4/.config/Trolltech.conf", {st_mode=S_IFREG|0600, st_size=27856, ...}) = 0
stat64("/etc/settings/Trolltech.conf", 0xbffdd350) = -1 ENOENT (No such file or directory)
stat64("/etc/settings/Trolltech.conf", 0x80e3878) = -1 ENOENT (No such file or directory)
stat64("/home/kde4/build/avo/avogadro/src/inputmethods/.", 0x80e3740) = -1 ENOENT (No such file or directory)
lstat64("/home/kde4/build/avo/avogadro/src/inputmethods/.", 0xbffdd730) = -1 ENOENT (No such file or directory)
stat64("/home/kde4/kde/lib/kde4/plugins/inputmethods/.", 0x80e3740) = -1 ENOENT (No such file or directory)
lstat64("/home/kde4/kde/lib/kde4/plugins/inputmethods/.", 0xbffdd730) = -1 ENOENT (No such file or directory)
write(7, "\24\0\6\0D\0\0\0\27\0\0\0\0\0\0\0\0\0\0\0\0 \0\0", 24) = 24
read(7, "\1\10z\0G\20\0\0\37\0\0\0\0\0\0\0\33A\0\0\0\0\0\0\0\0\0"..., 32) = 32
readv(7, [{"*AxeText.Translations:\t#override"..., 16667}, {"\0", 1}], 2) = 16668
writev(7, [{"b\0\6\0\17\0\0\0", 8}, {"XInputExtension", 15}, {"\0", 1}], 3) = 24
read(7, "\1\322{\0\0\0\0\0\1\221_\250\0\20\0\0\0\0\0\0\0\0\0\0\0"..., 32) = 32
futex(0x80cb468, FUTEX_WAIT, 2, NULL
I have no idea how to debug this any further so please guide me.
Reported by: @cniehaus
The text was updated successfully, but these errors were encountered: