Qt 5.5 doesn't build with -qreal float on ARM

painting/qpdf.cpp:1982:70: error: no matching function for call to 'qBound(double, const float&, double)'
                 b.start = step + 1 - qBound(0., stops.at(i).first, 1.);
                                                                      ^
painting/qpdf.cpp:1982:70: note: candidate is:
In file included from ../../include/QtCore/qglobal.h:1:0,
                 from kernel/qt_gui_pch.h:44:
../../include/QtCore/../../src/corelib/global/qglobal.h:543:34: note: template<class T> constexpr const T& qBound(const T&, const T&, const T&)
 Q_DECL_CONSTEXPR inline const T &qBound(const T &min, const T &val, const T &max)
                                  ^
../../include/QtCore/../../src/corelib/global/qglobal.h:543:34: note:   template argument deduction/substitution failed:
painting/qpdf.cpp:1982:70: note:   deduced conflicting types for parameter 'const T' ('double' and 'float')
                 b.start = step + 1 - qBound(0., stops.at(i).first, 1.);
                                                                      ^
painting/qpdf.cpp:1983:73: error: no matching function for call to 'qBound(double, const float&, double)'
                 b.stop = step + 1 - qBound(0., stops.at(i - 1).first, 1.);
                                                                         ^