Date: Wed, 01 Jun 2011 21:13:09 +0200 From: "Hartmann, O." <ohartman@zedat.fu-berlin.de> To: Andriy Gapon <avg@FreeBSD.org> Cc: Olivier Smedts <olivier@gid0.org>, "O. Hartmann" <ohartman@mail.zedat.fu-berlin.de>, freebsd-ports@FreeBSD.org Subject: Re: editors/libreoffice: can not compile, crash on, FreeBSD 9.0-CUR/amd64 - msg#00540 Message-ID: <4DE68F45.2030201@zedat.fu-berlin.de> In-Reply-To: <4DE4E887.2090806@FreeBSD.org> References: <4DE13892.5030909@mail.zedat.fu-berlin.de> <4DE152E6.1000801@FreeBSD.org> <BANLkTimX44PPzw7U-fccFDJ_j8=HSDp3Lg@mail.gmail.com> <4DE4BB34.508@FreeBSD.org> <BANLkTinmO9Ybgx=P8pFv%2BNeO72hYS-G6-A@mail.gmail.com> <4DE4E887.2090806@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 05/31/11 15:09, Andriy Gapon wrote: > on 31/05/2011 15:22 Olivier Smedts said the following: >> 2011/5/31 Andriy Gapon<avg@freebsd.org>: >>> on 31/05/2011 12:11 Olivier Smedts said the following: >>>> Segmentation fault: 11 (core dumped) >>> That core dump should be investigated / debugged. >>> At the very least it is not clear from the messages which program crashes. >> It's more clear if you read the output of the "build" command. >> >> Well I'm not very familiar with debugging, I loaded the program and >> the core dump with gdb and asked for a backtrace, but that's all I >> know. Is there anything useful I can do ? I can also provide the core >> dump. > Speaking from general point of view, it would be nice to compile the failing > program (regcomp) and its shared libraries with debug information. However I > don't have any advice on how to actually do it in this build environment. > > [snip] >> (gdb) bt >> #0 0x000000080202c640 in ?? () >> #1 0x0000000802461467 in stoc_impreg::insert_singletons () >> from /usr/ports/editors/libreoffice/work/libreoffice-build-3.3.2.2/build/libreoffice/solver/330/unxfbsdx.pro/lib/bootstrap.uno.so >> #2 0x0000000802463713 in stoc_impreg::ImplementationRegistration::doRegister () >> from /usr/ports/editors/libreoffice/work/libreoffice-build-3.3.2.2/build/libreoffice/solver/330/unxfbsdx.pro/lib/bootstrap.uno.so >> #3 0x0000000802464f8d in >> stoc_impreg::ImplementationRegistration::prepareRegister () >> from /usr/ports/editors/libreoffice/work/libreoffice-build-3.3.2.2/build/libreoffice/solver/330/unxfbsdx.pro/lib/bootstrap.uno.so >> #4 0x00000000004093ba in DoIt::operator() () >> #5 0x000000000041aeff in >> std::for_each<__gnu_cxx::__normal_iterator<rtl::OUString*, >> std::vector<rtl::OUString, std::allocator<rtl::OUString> > >, DoIt> () >> #6 0x0000000000409bef in main () > From the backtrace and some googling it seems like the problem could be - and it's > just a wild guess - because of exceptions. In particular I've seen something like > this when libgcc_s.so used at run-time came from a compiler different than a > compiler which was used to produce an executable. In my case it was libgcc_s.so > from base gcc at run-time versus libgcc_s.so from gcc45 at link-time. My program > crashed on any 'throw' statement (even when an exception should have been caught). > Again, I am not saying that this is what happens here, I am just hinting at > things to try. > > P.S. I have not debugged the actual mechanics of the crash that I have described > above. > As I reported eralier, LibreOffice 3.3.2 crashes on all of my FreeBSD 9.0-CURRENT/amd64 boxes compiled with CLANG. Today I compiled the OS on one of those boxes with the optimization option "-O3" instead of the suggested "-O2" - and voila, LibreOffice 3.3.2 as installed seems to work again! Strange, strange ...
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4DE68F45.2030201>