Created attachment 411196 [details] qt-creator-3.2.2 backtrace when trying to open (CMake) projects Hi, I'm no more able to open my projects with qt-creator-3.2.2: it instantly segfaults. This is a regression as I can do it with qt-creator-2.8.1 without a problem. It's noteworthy that, while I can't open my projects, I can open individual .h/.cpp files of the same projects without a problem. I don't know whether being CMake projects or not is important here. Although I've rebuilt qt-creator-3.2.2 with debug symbols, gdb still doesn't seem to find all the required resources: Program received signal SIGSEGV, Segmentation fault. 0x20000000073a1b01 in QmlJS::QmlBundle::readFrom (this=0x600ffffffffecbf8, path= ..., errors=0x600ffffffffeca50) at qmljsbundle.cpp:263 263 qmljsbundle.cpp: No such file or directory. I'm nevertheless attaching the complete backtrace in the hopes something will provide useful information. A similar issue was reported upstream [1], but since I'm building qt-creator using current ia64 stable gcc-4.7.4, the cause (and thus solution) is probably different. Émeric [1] https://bugreports.qt.io/browse/QTBUG-38808
I'm not able to test whether this issue has been fixed in =dev-qt/qt-creator-3.4.2 as I can't emerge qt5 on ia64 at the moment (see bug #559802) Émeric
Can you try building with -fno-devirtualize?
(In reply to Davide Pesavento from comment #2) > Can you try building with -fno-devirtualize? I didn't have time to try with -fno-devirtualize yet, but this issue also affects qt-creator-2.8.1 when I try to open Qt's own qt.pro project! So, it's not limited to >=dev-qt/qt-creator-3.2.2, it's just that I didn't noticed this issue before, as qt-creator-2.8.1 never crashed with my CMake projects. Bah, probably some random combination :-S Will check with older Qt Creator releases if I still own a backup of my Debian ia64 install. Émeric
Created attachment 411390 [details] qt-creator-2.8.1 backtrace when trying to open Qt's own qt.pro project
(In reply to Émeric Maschino from comment #3) > I didn't have time to try with -fno-devirtualize yet, but this issue also > affects qt-creator-2.8.1 when I try to open Qt's own qt.pro project! > > So, it's not limited to >=dev-qt/qt-creator-3.2.2 I don't doubt it, since the crash seems to happen in one of the qt libraries. So the bug is probably not in qt-creator itself. However I suggest you focus your efforts on making qt5 work on ia64 rather than spending time on old versions. Upstream (sadly) doesn't care about qt4 anymore, and even if you submit a patch they won't care... :-/ So if the -fno-devirtualize workaround works for you, good. Otherwise I'm afraid you'll have to take this upstream (if reproducible with current qt-creator and qt5 versions), because we don't have the resources to handle this kind of bugs at the distribution level. Thanks.
(In reply to Davide Pesavento from comment #5) > I don't doubt it, since the crash seems to happen in one of the qt > libraries. So the bug is probably not in qt-creator itself. The crash is in some Qt library _bundled_ with Qt Creator. It's not in Qt per se (hence the QmlBundle name, probably). > However I suggest you focus your efforts on making qt5 work on ia64 rather > than spending time on old versions. Upstream (sadly) doesn't care about qt4 > anymore, and even if you submit a patch they won't care... :-/ Good news is that I've started fixing ia64 compilation errors in upstream Qt stable 5.5 :-) Well, I'm actually stuck with an alignment issue (that was previously only warned, but compiler severity seems to have increased) in qtbase/src/openglextensions. > So if the -fno-devirtualize workaround works for you, good. Otherwise I'm > afraid you'll have to take this upstream (if reproducible with current > qt-creator and qt5 versions), because we don't have the resources to handle > this kind of bugs at the distribution level. Thanks. Will try the -fno-devirtualize workaround once I (hopefully) completed Qt 5 ia64 fixing. Émeric
(In reply to Émeric Maschino from comment #6) > (In reply to Davide Pesavento from comment #5) > > I don't doubt it, since the crash seems to happen in one of the qt > > libraries. So the bug is probably not in qt-creator itself. > > The crash is in some Qt library _bundled_ with Qt Creator. It's not in Qt > per se (hence the QmlBundle name, probably). > oh ok, the rest my comment still applies though, as upstream doesn't maintain older versions of qt-creator. > > However I suggest you focus your efforts on making qt5 work on ia64 rather > > than spending time on old versions. Upstream (sadly) doesn't care about qt4 > > anymore, and even if you submit a patch they won't care... :-/ > > Good news is that I've started fixing ia64 compilation errors in upstream Qt > stable 5.5 :-) > Awesome, that's greatly appreciated!
