Internal Error. MathLib::toLongNumber: input was not completely consumed: 0.0f16

[vix]

This comes from CMSIS-DSP which provides optimized compute kernels for Cortex-M and for Cortex-A.
16-bits-wide floating points are used in some functions and they're represente din this way:
float16_t res=0.0f16;
https://github.com/ARM-software/CMSIS-DSP/blob/99e15d3d95084b6868c5579b6791aff5cbeefcd6/Source/FastMathFunctions/arm_atan2_f16.c#L70

With cppcheck 2.12.0 this gives the following error:
Internal Error. MathLib::toLongNumber: input was not completely consumed: 0.0f16

[CHR]

This is fixed in head.

[Melroy van den Berg]

I got this exact error message when just compiling on x86-64 platform with basic C++ code. It's in head but not yet a release, I think I need to go back to version 2.11 for now.

src/main.cpp:0:0: error: Internal Error. MathLib::toLongNumber: input was not completely consumed: 1ms [cppcheckError]