http://www.aros.orgAROS-ExecAROS-Exec ArchivesPower2People
kitty mascottop logo menu

AROS System Development Manual -- Introduction


This document is not finished! It is highly likely that many parts are out-of-date, contain incorrect information, or are simply missing altogether. If you want to help rectify this, please contact us.

Licensing policy

Almost all code written by the AROS Development Team is licensed under the AROS Public License (APL), and this is the recommended choice for all new code written for the project.

However, we understand that this is not always possible; for example we often want to use good third party libraries and applications instead of reinventing the wheel ourselves. Therefore, it is allowed to import foreign code into the SVN repository that is not licensed under the APL, as long as the license of that code satisfies the requirements below.

Requirements on source code in the contrib tree

To include source code in the contrib tree, the following requirements must be satisfied:

  1. The license must allow us to:

    a. Redistribute the sources. c. Redistribute the binaries.

    If changes to the source code are necessary to make it compile and work for AROS, then additionally the license must allow us to make modifications and redistribute sources and binaries containing those modifications.

  2. The license must be explicitly stated in a file named LEGAL located in the root directory of the sources it applies to.

Requirements on source code in the main AROS tree

To include foreign source code, which is not licensed under the APL, in the main AROS tree the following requirements must be satisfied:

  1. The source code is required to build some component or is depended upon by some other component (which might or might not be APL) that we want to be included in the binary base distribution of AROS.
  2. The license must be open source as defined by the Open Source Initiative (OSI), meaning that it must be allowed for us to:
    1. Make modifications.
    2. Redistribute the source (possibly with modifications).
    3. Redistribute the binaries (possibly based on modified sources).
  3. The license does not conflict with the APL:
    1. If it is a stand-alone program, almost any license that complies with (2) is allowed.
    2. If it is a library, the license must allow linking with programs and libraries that use a different license without any problems. This means that libraries that are under GPL (as opposed to LGPL) are disallowed.
  4. The license must be explicitly stated in a file named LEGAL located in the root directory of the sources it applies to.

Coding conventions

General style

This code is used by many people and therefore you should keep some things in mind when you submit source code:

  • Keep things simple
  • Keep the source clean
  • Always know what you are doing
  • Tell what you are doing
  • Remember that you write code once but that it is read many times by many people


AROS uses some of the comments in the source to generate the documentation. Therefore it's necessary to keep a certain format so the tools can find their information. Other comments are ignored but they should explain what you thought when you wrote the code. If you really can't think of an explanation, then don't write the code a second time like this:

/* This adds 1 to t */

What we think of is this:

/* Go on with next element */

Function prototypes and headers

Every function in AROS must have a full ANSI C prototype. Prototypes should be collected in one header per file if it's needed by only a few files (no need to recompile the whole project if you change a function which is used only once), in one header per directory if it's a commonly used function in that directory or in one header per logical group (i.e. one header for all functions in a library).

The function header (i.e. the comment before the function) must be of a special format because the AutoDocs are generated from it. Here is an example for it (from <filename>AROS/exec/addhead.c</filename>):


    NAME */
#include <exec/lists.h>
#include <clib/exec_protos.h>

    AROS_LH2I(void, AddHead,

        AROS_LHA(struct List *, list, A0),
        AROS_LHA(struct Node *, node, A1),

        struct ExecBase *, SysBase, 40, Exec)

        Insert Node node as the first node of the list.

        list - The list to insert the node into
        node - This node is to be inserted



        struct List * list;
        struct Node * pred;

        // Insert Node at top
        AddHead (list, node);


        NewList(), AddTail(), Insert(), Remove(), RemHead(), RemTail(),



As you can see, comments are used to merge the function prototype and the header into one.


This field contains all necessary prototypes to use the function from the user point of view and the name of the function in an AROS_LH#?() macro. (LH for "Library Header") These macros are used to make the same code work on different kind of hardware. The name of the macro depends on the amount of parameters and whether the function needs the library base. AddHead() does not need the library base and therefore an "I" is appended to the macros name. If it had needed the library base (like AddTask()), then the "I" is omitted.

If the function is not part of a shared library, and it's arguments must be passed in specific registers (e.g. callback hooks), instead of AROS_LH#?() macros, you should use AROS_UFH#?() macros (UFH for "User Function Header"). Append the number of arguments to this macro. Since it has never a base, the field LOCATION must be omitted and it's not necessary to append the "I" to the macros name. An example for a callback hook foo() would be:

    AROS_UFHA(struct Hook, hook,  A0),
    AROS_UFHA(APTR,        obj,   A2),
    AROS_UFHA(APTR,        param, A1)

(Note that the registers need not have a particular order).

If the function is not part of a shared library and it's arguments don't have to be in specific registers, you don't need AROS_#?H#?() macros:


    NAME */
#include <header.h>

    int foo (

    int a,
    int b)



This field contains all arguments of the function one by one in AROS_LHA() macros (LHA for "Library Header Argument"). This macro makes sure each argument will be put in the right CPU register when the function is called (if possible and necessary). The first argument for the macro is the type of the parameter, followed by the name of the parameter, and the register the parameter is expected in. Valid names for registers are D0, D1, D2 up to D7 and A0 up to A6.

If the function is not part of a library but the arguments must be passed to it in registers, then use AROS_UFHA() macros (UFHA for "User Function Header Argument"), which take the same parameters as the AROS_LHA() macros. Don't forget the closing parenthesis for the AROS_UFC.

If the function is not part of a library and the arguments need not be passed in registers, no macros are necessary.

This field is necessary for shared libraries only. It contains the last four parameters for the AROS_LH#?() macro which are the type of the library, the name of the variable, in which the function expects the library base, the offset of the function in the jump table (the first vector has 1 and the first vector which may be used by a function is 5) and the name of the library.
This field contains a description of the function.
This field contains a list of all parameters of the form "name - description" or "name, name, name - description". The description should tell what the parameter is and what values can be passed to it. There is no point in explaining the parameter twice in FUNCTION and here. If the function has no parameters, say "None." here.

Optional, for functions with taglists.


name (type) - description


name (type)

Don't forget the default value.

What the function passes back. This includes return values and values passed in arguments of the function. If the function may fail, you should explain what it returns on failure and why it might fail.
Important things the user must know or take into account.

This field should contain a small or fully featured example. A good way to present an example, is to write some code which tests the function, put it into #ifdef TEST somewhere in the file and put a "See below." here. If you need comments in the code, you have two ways for this. If you need only short one-line comments, use C++ style (//) comments. Everything from the // to the end of the line is the comment. If you need more comment, then you can end the comment after the EXAMPLE and use #ifdef EXAMPLE to mask the example out:

    EXAMPLE */
#ifdef EXAMPLE
    struct List * list;
    struct Node * pred;

    /* Insert Node at top of the list */
    AddHead (list, node);

Don't use #ifdef EXAMPLE if you have a fully featured example (i.e. one which can be compiled without errors).

This field contains a list of known bugs.

This field contains a list of other functions and documents which might be of interest. This includes function which you need to initialize, create or destroy an object necessary for this function, functions which do similar and opposite things on the main object.

For example, SetAttrs() should list functions here which can create, destroy and manipulate BOOPSI objects but not taglists.

This field should contain information for other developers which are irrelevant to the user, for example an explanation of the algorithm of the function or dependencies.


Here is an example of how to format AROS code:

    /* a */
    struct RastPort * rp;
    int               a;

    /* b */
    rp = NULL;
    a  = 1;

    /* c */
    if (a == 1)
        printf ("Init worked\n");

    /* d */
        !(rp = Get_a_pointer_to_the_RastPort
                , long
                , arguments
        a <= 0
        printf ("Something failed\n");
        return FAIL;

    /* e */
    a = printf ("My RastPort is %p, a=%d\n"
        , rp
        , a

    return OK;

And here are the rules that make it look like it does:

  • If several lines contain similar code, put similar things below each other (see a and b);

  • Put spaces between operands and operators

  • Put matching braces {}, brackets [] and parentheses () below each other (d) if there is much code between. Brackets and parentheses may be in one line if the code between is small (c)

  • Indent by 4 Spaces.

    The reasons for this are:

    1. While some editors can use an arbitrary sizes for tabs, it's a bit complicated to tell another editor which tab size was used by the one used to write the code.
    2. Most code in AROS was written this way and your code should look like the rest.
    3. You can print this code on any printer without special tools to "fix" the tabs.
    4. Most editors have smart tabs which do exactly this. If your editor doesn't, write a bug report.
  • If you have a function with many arguments (d, e) you should put the parentheses in lines of their own and each argument in one line (d) or put the first argument behind the opening parenthesis (e) and each following argument in a line of its own with the comma in front. The closing parenthesis is in a line of its own and aligned with the beginning of the expression (i.e. the a and not the opening parenthesis or the printf()).

  • Use a single blank line to separate logical blocks. Large comments should have a blank line before and after them, small comments should be put before the code they explain with only one blank line before them.

Writing ROMable code

Code in AROS modules should be written in a way that makes it suitable for embedding into a ROM, FlashRAM or other kinds read-only memory. The following coding style rules are meant to make that possible. Of course they apply to all Kickstart modules and to code that may be made resident, shared or linked to other modules.

  • ROM modules must have no .data and .bss sections. Basically, we need to get rid of all non-const global data. The Amiga Kickstart proves that it's both possible and easy to achieve this.

    If you encounter an external variable (static or not) that is modified by the code, try to get rid of it or move it into the base of the library/device (or in the device node of your handler or in the userdata of your class).

  • The above applies to library bases as well. If you are writing a library, put the bases of other libraries into your own library base structure. BOOPSI classes can store library bases in their class private data.

  • Try to set the static and const attributes to all your global data. You can also use the CONST_STRPTR and CONST_APTR types defined in <exec/types.h>. Using static const allows the compiler to move data into the ".text" (AKA code) segment. If you need to pass these globals to another function, try to change its prototype to use const too. Note that, as of OS 3.5, Olaf Barthel has finally switched to using const in <clib/#?_protos.h> headers.

  • NEVER EVER touch buffers passed in by the user as an "input" parameter. The concept of input parameters is often implicit in the function description. For instance, the filename passed to Open() is clearly an input variable and Open() must not mess with it, even if it is going to fix it back later. Keep in mind that the buffer might be in read-only memory or shared among several instances of a resident or multi-threaded program.

  • Try to avoid host-OS calls such as malloc() and free() if you can do with AllocMem() and FreeMem(). This is because the pointer checking debug macros rely on finding the pointer within the Exec memory blocks with TypeOfMem().


This file describes how to port AROS to a new kind of hardware.

  1. Select an identifying name for your CPU (e.g. i386, m68k, hppa, sparc) and add "-emul" (e.g. i386-emul) if your port is to be running as a "sub-OS" or "-native" (e.g. m68k-native) if the port will be a stand-alone OS.

  2. Select an identifying name for your system (e.g. sgi, linux, amiga, etc.).

  3. Edit "configure" and make it recognize your kind of hardware and adjust the numerous variables as your system requires.


    The kind of CPU you use (see 1.)


    Name of your system (see 2.)


    The name of your C compiler


    options which should be handed to every call to the C compiler (e.g. -g -Wall -O0 etc.)

    The flags you must give to the compiler when linking to

    prevent it to use any standard libraries or start-up modules (for GCC the options are -nostartfiles -nostdlib -Xlinker -i). This is used to create AROS executables. These executables must not have any unresolved symbols and all references must be filled.

    contains the name of your ranlib program. If you don't have

    one, specify "true" here (or the name of any other shell command which simply ignores all parameters and doesn't return an error code).

  4. Type "make". It will abort because there is no $(KERNEL) yet, but it'll set up some important files and directory trees.

  5. Make a copy of i386-emul to $(KERNEL) and convert all assembler sources from x86 to your CPU.

  6. Populate $(KERNEL)/. It is recommended that you make a copy of i386-emul, because that is the most up to date version of the kernel.

  7. Type "make machine". It will compile a program and run it. The output can be used to modify $(KERNEL)/machine.h.

  8. Run "make machine.i" in $(KERNEL). It will generate a file "machine.i" which you need to compile the assembler files. "machine.i" contains the values of numerous system constants (function vector offsets, structure field offsets and system flags).

  9. Edit all #?.s files in $(KERNEL) and generate the appropriate machine code for your CPU. To compile the files, type "make".

  10. Go to the main directory and type make. If there any errors, write them down, then fix them and continue with step 10.

  11. Go to bin/$(ARCH)/AROS and start "arosshell". Now you can type some commands (e.g. "dir all", "list" or "demowin"). If all works well, you get a list of directories and files with "dir all" and "demowin" opens a window with some gadgets and renderings with which you can play. Typing "Esc" or clicking on "Exit" quits the demo. To stop the arosshell, you must press ^C (Ctrl-C) since as a real OS it has no way to stop it nicely.

Copyright © 1995-2018, Het AROS Development Team. Alle rechten voorbehouden.
Amiga® is een merknaam van Amiga Inc. Alle andere merknamen behoren tot hun respectievelijke eigenaren.