C99 Math library incomplete in GCC 4.9.* ARM

I try to build the Eigen library on a ARM with cxx11. The library builds with ARM GCC 4.8.* but not with ARM 4.9.*
It also builds on a native environment. See: https://travis-ci.org/diydrones/ardupilot/builds/88184030
Other people reported the same problems: http://lists.freebsd.org/pipermail/freebsd-current/2014-January/048130.html
As parts of GCC's standard library are protected by guard DEFs, compilation on GCC 4.8 was only possible with such a construct, but with GCC 4.9.* I get new problems (see below):

    #pragma once
    #pragma GCC diagnostic push
        #define RAND_MAX __RAND_MAX
        #pragma GCC diagnostic ignored "-Wshadow"
        #pragma GCC diagnostic ignored "-Wfloat-equal"
        #define _GLIBCXX_USE_C99_FP_MACROS_DYNAMIC 1
    ..

Here is the short version of my compiler error:

/home/travis/build/diydrones/ardupilot/libraries/AP_Eigen/AP_Eigen.h:17:81: note: #pragma message: std::lgamma*() does not exist in this environment.
             #pragma message("std::lgamma*() does not exist in this environment.")
                                                                                 ^
In file included from /home/travis/opt/gcc-arm-none-eabi-4_9-2015q3/arm-none-eabi/include/c++/4.9.3/random:51:0,
                 from /home/travis/opt/gcc-arm-none-eabi-4_9-2015q3/arm-none-eabi/include/c++/4.9.3/bits/stl_algo.h:66,
                 from /home/travis/opt/gcc-arm-none-eabi-4_9-2015q3/arm-none-eabi/include/c++/4.9.3/algorithm:62,
                 from /home/travis/build/diydrones/ardupilot/libraries/AP_Eigen/eigen/Eigen/Core:160,
                 from /home/travis/build/diydrones/ardupilot/libraries/AP_Eigen/AP_Eigen.h:32,
                 from /home/travis/build/diydrones/ardupilot/libraries/Filter/LowPassFilter2p.cpp:100:
/home/travis/opt/gcc-arm-none-eabi-4_9-2015q3/arm-none-eabi/include/c++/4.9.3/bits/random.tcc: In member function 'void std::poisson_distribution<_IntType>::param_type::_M_initialize()':
/home/travis/opt/gcc-arm-none-eabi-4_9-2015q3/arm-none-eabi/include/c++/4.9.3/bits/random.tcc:1408:55: error: no matching function for call to 'max(float, const double&)'
    _M_d = std::round(std::max(6.0, std::min(__m, __dx)));
                                                       ^
compilation terminated due to -Wfatal-errors.

Please fix this fast, it cannot be that half of the standard library cannot be used with ARM.