Legacy ImageMagick Discussions Archive

The llvm openmp developer also confirmed that back porting the changes in Magick++/lib/Functions.cpp and Magick++/lib/Magick++/Functions.h from master eliminates the test suite regressions for him as well on darwin.

Also opened https://github.com/ImageMagick/ImageMagick/issues/100 as I assume that is what the ImageMagick developers use in place of bugzilla for tracking bugs.

Also confirmed that back porting the changes in Magick++/lib/Functions.cpp and Magick++/lib/Magick++/Functions.h from master eliminates the test suite regressions in mageMagick-6 branch when it is built with clang/libomp 3.9svn (llvm trunk) on x86_64-apple-darwin15.

Note for the current code in ImageMagick-6 branch , the OpenMP developer's analysis was... Looks like the culprit to the finalization problem. Again, as a final indication that this is some odd Darwin phenomenon: You can see that: 1) __kmp_cleanup_indirect_user_locks is called first (shutdown of ...

Also confirmed on a second machine under x86_64-apple-darwin13 that back porting these changes in Magick++/lib/Functions.cpp and Magick++/lib/Magick++/Functions.h from master eliminates the test suite regressions in mageMagick-6 branch when it is built with clang/libomp 3.8svn.

I am finding that a back port of the following change from master into ImageMagick-6 branch eliminates the test suite failures when built with clang/libomp 3.8svn on x86_64-apple-darwin15... --- ImageMagick-6.9.3-3/Magick++/lib/Functions.cpp 2016-01-23 05:17:00.000000000 -0500 +++ ImageMagick-master ...

It is interesting that these regressions occur in the Magick++ test suite and that a difference exists between master and ImageMagick-6 in that the latter has the following code in Magick++/lib/Functions.cpp... // // Create a local wrapper around MagickCoreTerminus // namespace Magick { extern "C ...

The OpenMP developer confirmed that these failures are only seen on darwin.

Yes, master branch on darwin works without any changes. I am still waiting for a response in https://llvm.org/bugs/show_bug.cgi?id=26184 as to whether the OpenMP developers reproduced the original lock issues on darwin or linux.

Okay, I am checking with Jonathan Peyton to see if his reproduction of the original failures that I reported was done on darwin or linux (in case that these problems are only triggered on that target). FYI, I also see failures in current ImageMagick-6 branch using llvm/libomp 3.7.1 (which requires ...

Also, in case it matters, the following are the configure options that I am using on x86_64-apple-darwin15... $ ./configure --prefix=/sw --prefix=/sw/lib/ImageMagick-6.9 --bindir=/sw/bin --docdir=/sw/share/doc/ImageMagick-6 --includedir=/sw/include --mandir=/sw/share/man --disable-silent-rules ...

Make sure that the ImageMagick-6 branch build is actually modified to link against libomp. On darwin, I used...

perl -pi -e 's|-lgomp|-L/sw/opt/llvm-3.8/lib -lomp|g' configure

Commenting the line... /* SemaphoreComponentTerminus(); */ in magick/magick.c/MagickCoreTerminus()/1406: doesn't suppress these regressions in ImageMagick-6 branch when built against clang/libomp 3.8 RC1. These failures do seem to be very random so you might need to try to repeatedly execute one of ...

FYI, prebuilt linux binaries for llvm 3.8' RC1s clang compiler (including the openmp libomp) are available for download at...
http://llvm.org/pre-releases/3.8.0/

These new regressions only occur in ImageMagick-6 branch. Master built the same way against clang/libomp 3.8svn shows clean test suite results.