#248 Inline Math funcs gives SegFault

[Carlos]

I'm getting a SegFault when I try to use a math function, but only when this function is inline and only when I compile in release mode. If I put it inside a CPP all works fine.

I get the error using versions 4.8.1 and 4.9.2.

These are the release compiling flags:

-O2 -march=pentium4 -mtune=pentium4 -msse -msse2 -mfpmath=sse -Wextra -Wall -pipe -fno-strict-aliasing -std=gnu++11

It fails with or without -g

These are the debug compiling flags:

-g -O0 -march=pentium4 -mtune=pentium4 -msse -msse2 -mfpmath=sse -Wextra -Wall -pipe -fno-strict-aliasing -std=gnu++11

It never fails in debug mode.

This is the function:

MATH_INLINE

CMatrix4_GL

CMatrix4_GL::Perspective(float fovy, float aspect, float zNear, float zFar)

{

   ymax = zNear * MindShake::Tan(fovy * Float32::PI / 360.0f);

   ymin = -ymax;

   xmin = ymin * aspect;

   xmax = ymax * aspect;

return Frustum(xmin, xmax, ymin, ymax, zNear, zFar);

}

• fovy = 45.0f
• aspect = 1.7f
• zNear = 512.0f;

• zFar = 3000.0f

• MindShake::Tan is an inline tenplate function (float / double)

• Frustum is an inline function too (similar to Perspective).

GDB / CodeBlocks get the SigFault when I try to execute this line:

ymin = -ymax;

All works fine using Visual Studio 2010, 2012 and 2013.

I think the problem is because the various inline functions.

Thanks in advance!

[ralph engels]

You should allways use a .cpp extension with C++ code anything else is just asking for trouble.

hint

CMatrix4_GL::Perspective is not C so i would be very surprised if gcc would not croak :)