Build with -O0, that is, with no optimizations. Compiler optimization levels can re-arrange your code. If it runs with -O0 builds, try -O1, etc.
If it doesn't work with -O0, start that version with a debugger. You will need to step thru the code to find any misaligned pointers, and it helps to have no optimizations to do that. Perhaps the vendor has run into similar problems and can help you out. Dale Smith, Ph.D. Senior Financial Quantitative Analyst Financial & Risk Management Solutions Fiserv Office: 678-375-5315 www.fiserv.com -----Original Message----- From: Tobias Schmidt [mailto:[email protected]] Sent: Friday, January 10, 2014 8:24 AM To: Smith, Dale (Norcross) Cc: [email protected] Subject: Re: [NLopt-discuss] Using NLopt on BeagleBone Black with QNX Hi, thanks for your reply. ldqnx.so links to libc.so which is used by a simple Hello World program, too. I successfully run such a test program on the board. I don't think that Cygwin causes the fault. Within Cygwin I did only build the Makefiles. The "make all" command is executed from QNX Momentics IDE. If there would be any dependency to Cygwin, the program won't start on the board because there is no Cygwin library. I tried to use NLopt both as dynamic and static library. Both versions fails with indicated SIGBUS. Do anybody have any other suggestions? Regards, Tobias Schmidt Am 09.01.2014 15:42, schrieb Smith, Dale (Norcross): > The only other thing I can think of is to build nlopt using --enable-shared > and compile with -mno-cygwin to avoid a reference to the Cygwin dll. This is > the way to get code that doesn't reference Cygwin when running outside of > Cygwin. > > http://www.delorie.com/howto/cygwin/mno-cygwin-howto.html > > Dale Smith, Ph.D. > Senior Financial Quantitative Analyst > Financial & Risk Management Solutions > Fiserv > Office: 678-375-5315 > www.fiserv.com > > > -----Original Message----- > From: NLopt-discuss [mailto:[email protected]] > On Behalf Of Smith, Dale (Norcross) > Sent: Thursday, January 09, 2014 9:17 AM > To: Tobias Schmidt; [email protected] > Subject: Re: [NLopt-discuss] Using NLopt on BeagleBone Black with QNX > > Can you successfully build and run a simple test program that loads ldqnx.so > but does not link with nlopt? If not, then it's not an nlopt problem. Does > the board have a test suite you can run to verify programs built in Cygwin > work properly? > > Have you tried building nlopt as a static lib and linking with that? > > Dale Smith, Ph.D. > Senior Financial Quantitative Analyst > Financial & Risk Management Solutions > Fiserv > Office: 678-375-5315 > www.fiserv.com > > > -----Original Message----- > From: NLopt-discuss [mailto:[email protected]] > On Behalf Of Tobias Schmidt > Sent: Thursday, January 09, 2014 5:06 AM > To: [email protected] > Subject: [NLopt-discuss] Using NLopt on BeagleBone Black with QNX > > Hi nlopt-community, > > I'm trying to use NLopt to solve optimization problems on my BeagleBone Black > running QNX 6.5.0. The board has an ARM Cortex-A8 processor. > > Did anybody cross-compile the NLopt library for QNX on BBB or a similar board > with ARM processor? > > Up to now I did following steps to configure and build the library: > > 1. Created a new QNX C Project in Momentics 2. Configured the Makefile > by running <code>./configure --without-octave --without-python > --without-guile --enable-shared > --host=arm-unknown-nto-qnx6.5.0 --build=i486-pc-nto-qnx6.5.0 CC="qcc > -V4.4.2,gcc_ntoarmv7le" CPP="qcc -V4.4.2,gcc_ntoarmv7le -E" CXX="QCC > -V4.4.2,gcc_ntoarmv7le" CXXCPP="QCC -V4.4.2,gcc_ntoarmv7le -E" > LD="arm-unknown-nto-qnx6.5.0-ld" NM="arm-unknown-nto-qnx6.5.0-nm -B"</code> > within Cygwin environment. > 3. Run "Build Project" in Momentics > > The build process succeed, but if I want to use the library within another > Project the execution fails with the message: > > Process 36886 (nlopt_test_ctschmidt138926111541656) terminated SIGBUS > code=1 fltno=5 > ip=01024c34(/usr/lib/ldqnx.so.2@_band_get_aligned+0x168) > mapaddr=00024c34. ref=bee2e36a > > Does anybody know what I'm doing wrong? Does anybody has an idea where the > fault could be located? > > Regards, > Tobias Schmidt > > _______________________________________________ > NLopt-discuss mailing list > [email protected] > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss > > _______________________________________________ > NLopt-discuss mailing list > [email protected] > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss > _______________________________________________ NLopt-discuss mailing list [email protected] http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
