MATLAB: What are the real_T and int_T datatypes referenced in the Simulink template CMEX S-function

cmexdatatypesint_treal_ts-functionsimulink

I was looking at sfuntmpl.c and saw that these data types are used exclusively in this file and in your examples. I would like to know if I have to define my variables in this matter, or if I can use double, int, and the other C types.

Best Answer

You do not have to use int_T or real_T anywhere in your CMEX file. We recommend that you use these type definitions, which are defined in the header file tmwtypes.h. This allows greater platform independence in your MEX-file. Using real_T allows you to switch between different data types for 16,32,64 bit systems, allowing more flexibility.

However, you do not have to use these definitions at all in your model. For example, you can edit the Simulink example S-function csfunc.c located in $MATLABROOT/simulink/src. On a PC, you can change all of the real_T to double and all of the int_T to int, compile the S-function, and it will behave the same way.

Related Solutions

MATLAB: What is the different between int_T in matlab S-function or mexfunction and int_t in C

int_T, real_T (upper case T) are data types used by MATLAB and Simulink and are defined in shipped file "tmwtypes.h". They prove more flexibility in terms of platform dependency. For example real_T can switch types for 32 or 64 bit systems. To define a custom S-Function in C, you need not use any of these data types and can use standard ones like int, double etc.

I think you can also use int_t (small case t) in your code as long as those definitions are provided to MATLAB.

MATLAB: Simulink: How to make built-in boolean a 32-bit type

Ok, I dind't find out how to change the size of the Simulink underlying type... but I found a work-around.

I have a file DataAccess.c/.h, containing the problem function DataAccess_GetBoolean(...)

The legacy code tool then generated a separate wrapper file for each function, in that case DataAccess_GetBoolean.c/.h, which containts the s-function API functions, and in mdlOutputs(..) calls the hand-code function DataAccess_GetBoolean(...)

In our case, that is where I had to adjust the code by hand and then rebuild:

This is the generated code:

   [...]
    boolean_T *y1 = (boolean_T *) ssGetOutputPortSignal(S, 0);  // this is where the pointer size mismatch caused the problem, because boolean_T is 8-bit and my boolean type is uint32
/* Call the legacy code function */
DataAccess_GetBoolean( u1, *u2, *u3, *u4, y1);

And this is the code with manual modification:

[...]
boolean_T *y1= (boolean_T *) ssGetOutputPortSignal(S, 0);
uint32_T TmpBoolean;
/* Call the legacy code function, using the uint32 pointer */
DataAccess_GetBoolean( u1, *u2, *u3, *u4, &TmpBoolean);
// map boolean value over from uint32 pointer to boolean_T(8 bit) pointer
if (TmpBoolean)
   *y1 = 1;
else 
   *y1 = 0;

After the modification I had to rebuild the legacy code and the wrapper code. The easiest way to do that was to copy the output from the previous legacy code tool run and run them again on the command line... Just looked for ### Start Compiling DataAccess_GetBoolean and grabbed the next two mex calls

Best Answer

Related Solutions

MATLAB: What is the different between int_T in matlab S-function or mexfunction and int_t in C

MATLAB: Simulink: How to make built-in boolean a 32-bit type

Related Question