Hello community, here is the log from the commit of package openEMS for openSUSE:Factory checked in at 2018-03-29 11:56:35 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/openEMS (Old) and /work/SRC/openSUSE:Factory/.openEMS.new (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Package is "openEMS" Thu Mar 29 11:56:35 2018 rev:9 rq:591673 version:0.0.35 Changes: -------- --- /work/SRC/openSUSE:Factory/openEMS/openEMS.changes 2018-01-23 13:53:16.252825205 +0100 +++ /work/SRC/openSUSE:Factory/.openEMS.new/openEMS.changes 2018-03-29 11:56:38.932125677 +0200 @@ -1,0 +2,15 @@ +Tue Mar 27 00:30:23 UTC 2018 - stefan.bruens@rwth-aachen.de + +- Fix build on e.g. ARM, use x86 intrinsics only on x86 + add 0001-Guard-xmmintrin.h-include-so-it-is-only-used-when-ne.patch + +------------------------------------------------------------------- +Tue Mar 27 00:12:19 UTC 2018 - stefan.bruens@rwth-aachen.de + +- Change the Qt BuildRequires: for older distributions, Leap 42.x + has a VTK build with Qt4, use the same conditional as the VTK + package (i.e. Qt4 on Leap 42.x, Qt5 on TW and Leap 15.x) +- Fix build when using -std=gnu++11 and glibc <= 2.25 + add 0001-Fix-build-error-due-to-ambigous-overload-of-isnan-is.patch + +------------------------------------------------------------------- New: ---- 0001-Fix-build-error-due-to-ambigous-overload-of-isnan-is.patch 0001-Guard-xmmintrin.h-include-so-it-is-only-used-when-ne.patch ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ openEMS.spec ++++++ --- /var/tmp/diff_new_pack.kpTOIz/_old 2018-03-29 11:56:39.860092169 +0200 +++ /var/tmp/diff_new_pack.kpTOIz/_new 2018-03-29 11:56:39.860092169 +0200 @@ -21,7 +21,7 @@ Version: 0.0.35 Release: 0 Summary: Electromagnetic field solver using the EC-FDTD method -License: GPL-3.0 +License: GPL-3.0-only Group: Productivity/Scientific/Physics Url: http://openems.de # source - openEMS component only, not openEMS-Project @@ -38,6 +38,10 @@ Patch5: 0005-Fix-build-with-HDF5.patch # PATCH-FIX-UPSTREAM openEMS-no-return.patch Patch6: 0006-Add-missing-return-statement.patch +# PATCH-FIX-OPENSUSE 0001-Fix-build-error-due-to-ambigous-overload-of-isnan-is.patch -- Fix ambigous isnan/std::isnan +Patch7: 0001-Fix-build-error-due-to-ambigous-overload-of-isnan-is.patch +# PATCH-FIX-OPENSUSE 0001-Guard-xmmintrin.h-include-so-it-is-only-used-when-ne.patch -- Only include xmmintrin.h on x86 +Patch8: 0001-Guard-xmmintrin.h-include-so-it-is-only-used-when-ne.patch BuildRequires: CSXCAD-devel BuildRequires: cmake @@ -59,8 +63,12 @@ BuildRequires: vtk-devel BuildRequires: vtk-java BuildRequires: vtk-tcl +%if %{suse_version} >= 1500 BuildRequires: cmake(Qt5Sql) BuildRequires: cmake(Qt5Widgets) +%else +BuildRequires: libqt4-devel +%endif BuildRequires: pkgconfig(fparser) BuildRequires: pkgconfig(python3) BuildRequires: pkgconfig(sm) @@ -127,6 +135,8 @@ %patch4 -p1 %patch5 -p1 %patch6 -p1 +%patch7 -p1 +%patch8 -p1 echo "Name: %{octpkg}" >> DESCRIPTION echo "Version: %{version}" >> DESCRIPTION @@ -176,6 +186,12 @@ export CFLAGS="%{optflags} -mavx" export CXXFLAGS="%{optflags} -mavx" %endif +%ifnarch %ix86 x86_64 +# Always handle subnormals according to IEEE754 (gradual underflow), +# as the code for enabling Flush-To-Zero is x86 specific. This may have +# a performance impact, arch specific code for non-x86 should be used. +export CXXFLAGS="%{optflags} -DSSE_CORRECT_DENORMALS" +%endif %cmake make %{?_smp_mflags} ++++++ 0001-Fix-build-error-due-to-ambigous-overload-of-isnan-is.patch ++++++
From e802c612af3311f2b290c98cc175b236d8b013d4 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Stefan=20Br=C3=BCns?= <stefan.bruens@rwth-aachen.de> Date: Tue, 27 Mar 2018 02:10:18 +0200 Subject: [PATCH] Fix build error due to ambigous overload of isnan/isinf MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit
In case the code is build with -std=c++11, there may be conflicting definitions of isnan/isinf vs std::isnan/std::isinf, due to the using namespace std directive. This happens for glibc versions 2.25 and older, see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=48891 for details. Signed-off-by: Stefan Brüns <stefan.bruens@rwth-aachen.de> --- Common/processmodematch.cpp | 2 +- FDTD/extensions/operator_ext_tfsf.cpp | 2 +- FDTD/operator.cpp | 6 +++--- tools/sar_calculation.cpp | 2 +- 4 files changed, 6 insertions(+), 6 deletions(-) diff --git a/Common/processmodematch.cpp b/Common/processmodematch.cpp index 620257d..85d2d58 100644 --- a/Common/processmodematch.cpp +++ b/Common/processmodematch.cpp @@ -174,7 +174,7 @@ void ProcessModeMatch::InitProcess() for (int n=0; n<2; ++n) { m_ModeDist[n][posP][posPP] = m_ModeParser[n]->Eval(var); //calc mode template - if ((isnan(m_ModeDist[n][posP][posPP])) || (isinf(m_ModeDist[n][posP][posPP]))) + if ((std::isnan(m_ModeDist[n][posP][posPP])) || (std::isinf(m_ModeDist[n][posP][posPP]))) m_ModeDist[n][posP][posPP] = 0.0; norm += pow(m_ModeDist[n][posP][posPP],2) * area; } diff --git a/FDTD/extensions/operator_ext_tfsf.cpp b/FDTD/extensions/operator_ext_tfsf.cpp index 663dc1f..62fd580 100644 --- a/FDTD/extensions/operator_ext_tfsf.cpp +++ b/FDTD/extensions/operator_ext_tfsf.cpp @@ -159,7 +159,7 @@ bool Operator_Ext_TFSF::BuildExtension() else m_PhVel=m_Op->CalcNumericPhaseVelocity(m_Start,m_Stop,m_PropDir,m_Frequency); - if ((m_PhVel<0) || (m_PhVel>__C0__/ref_index) || isnan(m_PhVel)) + if ((m_PhVel<0) || (m_PhVel>__C0__/ref_index) || std::isnan(m_PhVel)) { cerr << "Operator_Ext_TFSF::BuildExtension: Warning, invalid phase velocity found, resetting to c0! " << endl; m_PhVel = __C0__/ref_index; diff --git a/FDTD/operator.cpp b/FDTD/operator.cpp index a7582aa..259b1fb 100644 --- a/FDTD/operator.cpp +++ b/FDTD/operator.cpp @@ -1395,7 +1395,7 @@ bool Operator::AverageMatCellCenter(int ny, const unsigned int* pos, double* Eff if (EffMat[3]) EffMat[3]=length / EffMat[3]; for (int n=0; n<4; ++n) - if (isnan(EffMat[n]) || isinf(EffMat[n])) + if (std::isnan(EffMat[n]) || std::isinf(EffMat[n])) { cerr << "Operator::" << __func__ << ": Error, an effective material parameter is not a valid result, this should NOT have happend... exit..." << endl; cerr << ny << "@" << n << " : " << pos[0] << "," << pos[1] << "," << pos[2] << endl; @@ -1508,7 +1508,7 @@ bool Operator::AverageMatQuarterCell(int ny, const unsigned int* pos, double* Ef if (EffMat[3]) EffMat[3]=length / EffMat[3]; for (int n=0; n<4; ++n) - if (isnan(EffMat[n]) || isinf(EffMat[n])) + if (std::isnan(EffMat[n]) || std::isinf(EffMat[n])) { cerr << "Operator::" << __func__ << ": Error, An effective material parameter is not a valid result, this should NOT have happend... exit..." << endl; cerr << ny << "@" << n << " : " << pos[0] << "," << pos[1] << "," << pos[2] << endl; @@ -1555,7 +1555,7 @@ bool Operator::Calc_LumpedElements() if (R<0) R = NAN; - if ((isnan(R)) && (isnan(C))) + if ((std::isnan(R)) && (std::isnan(C))) { cerr << "Operator::Calc_LumpedElements(): Warning: Lumped Element R or C not specified! skipping. " << " ID: " << prims.at(bn)->GetID() << " @ Property: " << PLE->GetName() << endl; diff --git a/tools/sar_calculation.cpp b/tools/sar_calculation.cpp index 343aa07..e135aca 100644 --- a/tools/sar_calculation.cpp +++ b/tools/sar_calculation.cpp @@ -292,7 +292,7 @@ int SAR_Calculation::FindFittingCubicalMass(unsigned int pos[3], float box_size, bool SAR_Calculation::GetCubicalMass(unsigned int pos[3], double box_size, unsigned int start[3], unsigned int stop[3], float partial_start[3], float partial_stop[3], double &mass, double &volume, double &bg_ratio, int disabledFace) { - if ((box_size<=0) || isnan(box_size) || isinf(box_size)) + if ((box_size<=0) || std::isnan(box_size) || std::isinf(box_size)) { cerr << "SAR_Calculation::GetCubicalMass: critical error: invalid averaging box size!! EXIT" << endl; exit(-1); -- 2.16.2 ++++++ 0001-Guard-xmmintrin.h-include-so-it-is-only-used-when-ne.patch ++++++
From 120691d8a7a3bd951e74a5436a2b26fa4c0478d1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Stefan=20Br=C3=BCns?= <stefan.bruens@rwth-aachen.de> Date: Tue, 27 Mar 2018 02:28:22 +0200 Subject: [PATCH] Guard xmmintrin.h include so it is only used when necessary MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit
The x86/SSE specific code for Flush-To-Zero is only used when SSE_CORRECT_DENORMALS is not defined. Guarding the include allows the code to compile on e.g. ARM. Signed-off-by: Stefan Brüns <stefan.bruens@rwth-aachen.de> --- FDTD/engine_multithread.cpp | 3 +++ FDTD/engine_sse.cpp | 3 +++ 2 files changed, 6 insertions(+) diff --git a/FDTD/engine_multithread.cpp b/FDTD/engine_multithread.cpp index 7370198..7ebe0d7 100644 --- a/FDTD/engine_multithread.cpp +++ b/FDTD/engine_multithread.cpp @@ -32,7 +32,10 @@ #include "boost/date_time/posix_time/posix_time.hpp" #include "boost/date_time/gregorian/gregorian.hpp" #include <iomanip> + +#ifndef SSE_CORRECT_DENORMALS #include <xmmintrin.h> +#endif //! \brief construct an Engine_Multithread instance //! it's the responsibility of the caller to free the returned pointer diff --git a/FDTD/engine_sse.cpp b/FDTD/engine_sse.cpp index 660e6d6..2e06685 100644 --- a/FDTD/engine_sse.cpp +++ b/FDTD/engine_sse.cpp @@ -15,7 +15,10 @@ * along with this program. If not, see <http://www.gnu.org/licenses/>. */ +#ifndef SSE_CORRECT_DENORMALS #include <xmmintrin.h> +#endif + #include "engine_sse.h" //! \brief construct an Engine_sse instance -- 2.16.2