Summary: | =dev-cpp/eigen-3.2.2 version bump | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Janick Martinez Esturo <ph03> |
Component: | [OLD] Development | Assignee: | Gentoo Science Related Packages <sci> |
Status: | RESOLVED FIXED | ||
Severity: | enhancement | CC: | aballier, alunduil, dlan, hasufell, kde, toolchain |
Priority: | Normal | Keywords: | InOverlay |
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | kde overlay | ||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | 348843, 494796 | ||
Bug Blocks: | |||
Attachments: |
emerge --info '=dev-cpp/eigen-3.2.0::kde'
emerge -pqv '=dev-cpp/eigen-3.2.0::kde' Build Log Build Environment Build Log test example code |
Description
Janick Martinez Esturo
2013-11-20 18:59:40 UTC
Bumped in KDE overlay, please test there. Will move to tree in a week or so. Works like a charm for me, thanks! I've been getting a build error with the doc use flag on this package. Getting a SEGFAULT during the documentation build: /bin/sh: line 1: 568 Segmentation fault ./example_BVH_Example > /var/tmp/portage/dev-cpp/eigen-3.2.0/work/eigen-3.2.0_build/unsupported/doc/examples/BVH_Example.out I'm attaching the build logs, &c for more information but let me know if I can help fix this in any way. Created attachment 364664 [details]
emerge --info '=dev-cpp/eigen-3.2.0::kde'
Created attachment 364666 [details]
emerge -pqv '=dev-cpp/eigen-3.2.0::kde'
Created attachment 364668 [details]
Build Log
LZO Compressed due to size.
Created attachment 364670 [details]
Build Environment
(In reply to Alex Brandt from comment #6) > Created attachment 364668 [details] > Build Log > > LZO Compressed due to size. use bzip2, it a common utility and also you get more compress ratio besides for lzo I have to install app-arch/lzop Created attachment 364750 [details]
Build Log
Switched to bzip2 per request.
(In reply to Alex Brandt from comment #9) > Created attachment 364750 [details] > Build Log > > Switched to bzip2 per request. no need to attached once more.. but could do when you file another bug.. for the problem, I can re-produce in my recent system.. but in another old gentoo (haven't updated for a few time), build fine with USE=doc enabled. gdb backtrace doesn't make much sense here.. #0 0x0000000000405407 in void std::__introselect<__gnu_cxx::__normal_iterator<Eigen::internal::vector_int_pair<double, 2>*, std::vector<Eigen::internal::vector_int_pair<double, 2>, Eigen::aligned_allocator_indirection<Eigen::internal::vector_int_pair<double, 2> > > >, long, Eigen::KdBVH<double, 2, Eigen::Matrix<double, 2, 1, 0, 2, 1> >::VectorComparator>(__gnu_cxx::__normal_iterator<Eigen::internal::vector_int_pair<double, 2>*, std::vector<Eigen::internal::vector_int_pair<double, 2>, Eigen::aligned_allocator_indirection<Eigen::internal::vector_int_pair<double, 2> > > >, __gnu_cxx::__normal_iterator<Eigen::internal::vector_int_pair<double, 2>*, std::vector<Eigen::internal::vector_int_pair<double, 2>, Eigen::aligned_allocator_indirection<Eigen::internal::vector_int_pair<double, 2> > > >, __gnu_cxx::__normal_iterator<Eigen::internal::vector_int_pair<double, 2>*, std::vector<Eigen::internal::vector_int_pair<double, 2>, Eigen::aligned_allocator_indirection<Eigen::internal::vector_int_pair<double, 2> > > >, long, Eigen::KdBVH<double, 2, Eigen::Matrix<double, 2, 1, 0, 2, 1> >::VectorComparator) () #1 0x00000000004055fd in Eigen::KdBVH<double, 2, Eigen::Matrix<double, 2, 1, 0, 2, 1> >::build(std::vector<Eigen::internal::vector_int_pair<double, 2>, Eigen::aligned_allocator<Eigen::internal::vector_int_pair<double, 2> > >&, int, int, std::vector<Eigen::AlignedBox<double, 2>, Eigen::aligned_allocator<Eigen::AlignedBox<double, 2> > > const&, int) () possible of gcc issue? anyone can confirm this? the working machine has gcc-4.8.0 installed, while others have problem have 4.8.2 installed # ldd ./example_BVH_Example linux-vdso.so.1 (0x00007ffff0bfe000) libstdc++.so.6 => /usr/lib/gcc/x86_64-pc-linux-gnu/4.8.2/libstdc++.so.6 (0x00007f6b81539000) libm.so.6 => /lib64/libm.so.6 (0x00007f6b81236000) libgcc_s.so.1 => /usr/lib/gcc/x86_64-pc-linux-gnu/4.8.2/libgcc_s.so.1 (0x00007f6b8101f000) libc.so.6 => /lib64/libc.so.6 (0x00007f6b80c72000) /lib64/ld-linux-x86-64.so.2 (0x00007f6b81841000) Created attachment 365036 [details]
test example code
I can confirm this is the problem of gcc version 4.8.2
I've setup a chroot environment using stage3-20131205, compiling eigen-3.2.0 (USE=doc enabled) with gcc 4.7.3, 4.8.0, 4.8.1-r1, all works fine
for re-produce the problem, I have attached a simplified example code.
for testing, you can install eigen with <gcc-4.8.2, just do
USE="doc" emerge =eigen-3.2.0
then switch back to gcc-4.8.2
tar jxvf eigen-test.tar.bz2 && cd eigen-test && make && ./BVH_Example
you should see segmentfat.
See also bug #471258. These testcases fail in the same way: Start 555: BVH_1 555/633 Test #555: BVH_1 ............................***Failed 0.00 sec Start 556: BVH_2 556/633 Test #556: BVH_2 ............................***Failed 0.00 sec Start 557: BVH_3 557/633 Test #557: BVH_3 ............................***Failed 0.01 sec The upstream bug is http://eigen.tuxfamily.org/bz/show_bug.cgi?id=693 See also: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59557 Credit's to coolo for linking these two. what are we waiting for? (In reply to Julian Ospald (hasufell) from comment #17) > what are we waiting for? not sure how to handle it since it's miscompiled by stable gcc ~arch gcc is fine. you could argue that ~arch should not be blocked by that, but that's a matter of taste. (In reply to Andreas K. Hüttel from comment #18) > (In reply to Julian Ospald (hasufell) from comment #17) > > what are we waiting for? > > not sure how to handle it since it's miscompiled by stable gcc > > ~arch gcc is fine. > > you could argue that ~arch should not be blocked by that, but that's a > matter of taste. The next blender version will be broken by missing recent eigen version. When that happens, I'll bump eigen myself. (In reply to Julian Ospald (hasufell) from comment #19) > The next blender version will be broken by missing recent eigen version. > When that happens, I'll bump eigen myself. You won't hear any complaint; the package is not properly maintained. Bumped in the science overlay. Ebuild from science can not be pushed to the tree until alternatives-2.eclass is moved there. We need a newer version, so I've done a non-alternatives bump. + 06 Feb 2015; Michael Palimaka <kensington@gentoo.org> +eigen-3.2.4.ebuild: + Version bump wrt bug #491796. |