Inlined RtLinux Drivers with S-Functions and TLC: Full Guide

Inlined RtLinux Drivers with S-Functions and TLC: Full Guide

This technical series presents a complete methodology for developing fully inlined hardware drivers under RtLinux using MATLAB Real-Time Workshop (RTW) S-Functions and Target Language Compiler (TLC). The approach reduces runtime overhead, improves determinism, and seamlessly integrates with real-time applications.

⚡ Chapter 1: Introduction – The Need for Inlined Drivers

#

In real-time environments like RtLinux, precise timing is essential. Standard non-inlined S-Functions in RTW often fail to meet stringent performance requirements when interacting with hardware I/O boards.

Solution Overview:

• Develop hardware drivers (e.g., PCL-726) using S-Functions with TLC.
• Achieve fully inlined drivers inside RTW-generated code.

Benefits:

• Minimized memory footprint
• Elimination of extra function calls and SimStruct overhead
• Enhanced real-time performance
• Single driver usable in both simulation and real-time execution

🛠 Chapter 2: Fundamentals of S-Functions and TLC

#

S-Functions are custom dynamic blocks in Simulink. This guide focuses on Fully Inlined S-Functions:

• Simulation: Implemented as C MEX S-Functions
• Code Generation: Re-implemented using TLC for inlined code

TLC (Target Language Compiler) enables inlining, optimization, and embedding of hardware-specific code directly into the generated C output.

🔧 Chapter 3: TLC Implementation Mechanism

#

RTW Code Generation Flow (mymodel example):

1. Simulink generates mymodel.rtw intermediate representation.
2. TLC processes .rtw with system, block, and inlined S-Function target files.
3. Optimized C code (mymodel.c) and support files are produced.
4. Output resides in mymodel_rtw_rtw/.

Note: TLC is only required for inlined S-Functions, replacing calls with direct inline code.

💻 Chapter 4: Implementing Inlined S-Functions

#

Steps:

1. Create C MEX S-Function for simulation.
2. Implement mdlRTW() to export parameters to .rtw.
3. Write corresponding .tlc file (same base name).
4. Place .tlc in the TLC search path.

Example: mdlRTW() in pcl726fordi.c:

#define MDL_RTW
#if defined(MDL_RTW) && ( defined(MATLAB_MEX_FILE) || defined(NRT))
static void mdlRTW(SimStruct *S)
{
    char _T addrStr[15];
    mxGetString(_PARAM_1(S), addrStr, sizeof(addrStr));
    
    if (!ssWriteRTWParamSettings(S, 1,
            SSWRITE_VALUE_QSTR, "P1", addrStr, 10))
    {
        return; /* Error reported by Simulink */
    }
}
#endif

This routine saves parameters like board addresses for TLC to use in generated code.

⚙️ Chapter 5: Detailed Analysis of pcl726fordi.tlc

#

Hardware-specific code (e.g., #include <asm/io.h>, register access) resides solely in the TLC file.

TLC Structure:

%function BlockTypeSetup(block, system) void
  /* Parameter configuration */
%endfunction

%function Start(block, system) Output
  /* Initialization code */
%endfunction

%function Outputs(block, system) Output
  int _T flag = 0;
  %assign u = LibBlockInputSignal(0, "", "", 0)
  %assign y = LibBlockOutputSignal(0, "", "", 0)
  
  outb(%<u> >> 8, %<base> + 0);
  outb(%<u> & 0xff, %<base> + 1);
%endfunction

%function Terminate(block, system) Output
  /* Cleanup */
%endfunction

Workflow:

• BlockTypeSetup: Configure parameters
• Start: Hardware initialization
• Outputs: Main I/O logic
• Terminate: Cleanup at termination

🏗 Chapter 6: Complete Development Workflow

#

1. Write C MEX S-Function (pcl726fordi.c)
2. Compile MEX for Simulink simulation
3. Create pcl726fordi.tlc
4. Add S-Function block to Simulink model
5. Run RTW build → generates fully inlined code
6. Compile and execute on RtLinux

Advantages:

• No extra system calls during real-time execution
• Lower latency, improved determinism
• Reduced code footprint

✅ Chapter 7: Conclusion and Recommendations

#

By combining S-Functions with TLC, developers can produce fully inlined, high-performance drivers compatible with RTW real-time applications on RtLinux.

Modern Considerations (2026):

• Embedded Coder offers similar functionality
• Consider upgrading to modern real-time targets (e.g., Speedgoat, Linux RT PREEMPT)
• TLC remains vital for advanced code customization

References:

• System Simulation Technology
• Linux Device Drivers (Alessandro Rubini)
• MATLAB & C Mixed Programming resources