Compile MatrixX Autocode under Linux

[Charley_007]

Hi, I try to compile my autocoded code under Linux and I got problem at the compilation and at the link.

Everything is working fine under windows....

First Problem. When I switch to linux I got a error at the compilation :
     error: ‘Matrix_Inverse’ was not declared in this scope
Matrix_Inverse is part of sa_matrix.c that was compile before without any problem. Any Idea ? You can find the makefile in attachement.

Second error is during the link. If I put the Matrix_Inverse callout, I able to compile. When it's time to link everything, I got those error:
       TEST_CONTROLPC.o: In function `SCHEDULER':TEST_CONTROLPC.c:(.text+0x1fb41): undefined reference to `External_Input'
        :TEST_CONTROLPC.c:(.text+0x1fb91): undefined reference to `External_Output'
        :TEST_CONTROLPC.c:(.text+0x1fceb): undefined reference to `Error'
        :TEST_CONTROLPC.c:(.text+0x1fe0f): undefined reference to `Error'
        TEST_CONTROLPC.o: In function `main':TEST_CONTROLPC.c:(.text+0x218eb): undefined reference to `Implementation_Initialize'
        :TEST_CONTROLPC.c:(.text+0x218f0): undefined reference to `Background'
        :TEST_CONTROLPC.c:(.text+0x218fa): undefined reference to `Implementation_Terminate'
        NGC-sa_utils_bin.o: In function `Background()':NGC-sa_utils_bin.c:(.text+0x313): undefined reference to `SCHEDULER

It's look that it's not able to find function declare in the NGC-sa_utils_bin.c (little adaptation of sa_utils.c). Any idea on that ?

Thanks a lot

[CMal]

Charley_007,

It sounds like your compiler may be interpreting certain symbols incorrectly.  One thing to try would be to add another define to your makefile: -DANSI_PROTOTYPES, then do a rebuild of all files includeing the SA lib source files.  Also, what type of Linux are you running?  Please let me know if this has any effect.

Chris M.
Applications Engineer
National Instruments

[Charley007]

All right, let me resume :

We have successfully AutoCode a MATRIXx model into C code (using default AutoCode template) and compile it on a windows based computer with our own makefile. On the windows based computer, we have define –DMSWIN32 in the makefile in order to compile the model.

The result from the AutoCoded version and the result from the MATRIXx model are the same (diff of x10^-6). Now we want to move the AutocCoded model on a RT-Linux based computer.

Is there any other define we should include in the makefile? Target computer: RTAI_Linux on a FEDORA Linux

 And Now to solve the first problem :

        NGC-sa_utils_bin.o: In function `Background()':NGC-sa_utils_bin.c .text+0x313): undefined reference to `SCHEDULER

To bypass the problem, we have made a small modification in order to be able to compile the code. We have remove the declaration of SCHEDULER() in the procedure Background in the sa_utils.h and sa_utils.c file.

sa_utils.c

//extern void SCHEDULER ();

sa_utils.h

//EXTERN_FUNC (void SCHEDULER, ( ));

We have then created the header file (same name as the model) to declare this SCHEDULER procedure: EXTERN_FUNC (void SCHEDULER, ( )); We then run the code and write outputs in binary to a file. We read them back on a windows based computer to conclude that the results are not the same of the MATRIXx model.What do you think ?

Charley

[NathanT]

Hello Charley,
        Do you think you could attach the code that is causing these errors?

Thanks,

Nathan

[Charley_007]

I can't sorry, but I will try to create a simple model that I can attach.

[Charley007]

Hello!

I continued investigating the error and here's what I found out:
    In the sa_sys.h file where WS_TARGET and AC100_TARGET are defined, I discovered that no matter which OS I define, the #define of those variable is not done properly.
    From other files, I discovered that I want WS_TARGET=1 and AC100_TARGET=0 ( so to get RT_FLOAT defined as a double).
    If I force:  #define WS_TARGET 1 and #define AC100_TARGET 0 and comment out the part of the code in as_sys where the definition is supposed to be done, then everything compiles and seems to run properly. Through incremental steps in the testing, we concluded that the #define inside an if seems to be the problem but why?

In other words, this part of the code seems to cause the problem:

#if (APOLLO||DEC||HP||HP700||IBM||SGI||SOLARIS||SUN||VAX||MSWIN32||OSF1)
#define WS_TARGET 1
#else
#define WS_TARGET 0
#endif

#if (AC1000 || RS_PSOS || RS_VXWORKS || RS_UNIX || C_PC || _TMS320C30 || _TMS320C40 || _NT)
#define AC100_TARGET 1
/*
** Cannot be both a WS_TARGET and an AC100_TARGET
*/
#undef  WS_TARGET
#define WS_TARGET 0
#else
#define AC100_TARGET 0
#endif


#if (!AC100_TARGET && !WS_TARGET)
 --> Error invalid system type
#endif

If that is the case, this could also explain the previous problem I had with the external function (home made scheduler) which was defined in the .c file and had to be moved in a header file in order to work properly. (see previous post)

I hope I gave you enough information so you can help me...
Sorry I cannot send the code, but the company is ruled by the Canadian Controled Goods Program...
Thanks,

Charley
NGC Aerospace