OS-9 Assembly Code Development: Difference between revisions

From F256 Foenix
Jump to navigationJump to search
 
(16 intermediate revisions by 2 users not shown)
Line 1: Line 1:
== Overview ==
== Overview ==
Each process in NitrOS-9 is allocated its own 64K address space.  When run, the program object code is loaded into the end of the memory space, and the variables and data structures are loaded into the beginning.  In addition, NitrOS-9 keeps an separate per-process MMU table (DAT Image).  So it is important to use the OS-9 calls <code>F$MapBlk</code> and <code>F$ClrBlk</code> calls to map in memory blocks with the MMU.   
Each process in NitrOS-9 is allocated its own 64K address space.  When run, the program object code is loaded into the end of the memory space, and the variables and data structures are loaded into the beginning.  In addition, NitrOS-9 keeps an separate per-process MMU table (DAT Image), so you must use <code>F$MapBlk</code> to map and <code>F$ClrBlk</code> to unmap 8K blocks in your process' address space. The operating system has its own 64K address space as well.   


When a new process is created, registers U, Y, DP and SP are set with U = start of data area, Y = end of data area, DP = page # of beginning page, SP = end of data area + 1.  Each process includes its own execution and data directories, and <code>std in</code>, <code>std out</code> and <code>std err</code>.
When a new process is created, registers U, Y, DP and SP are set with U = start of data area, Y = end of data area, DP = page # of beginning page, SP = end of data area + 1.  Each process includes its own execution and data directories, along with <code>std in</code>, <code>std out</code> and <code>std err</code>.


=== Memory and DAT Image ===
=== Memory and DAT Image ===
<code>mmap</code> and <code>pmap</code> will output the current status of the memory in the system. <code>mmap</code> shows which memory blocks are currently in use.  <code>pmap</code> shows the blocks that are mapped in the DAT Image for a particular process.
<code>mmap</code> and <code>pmap</code> output the current status of the memory in the system. <code>mmap</code> shows which memory blocks are currently in use.  <code>pmap</code> shows the blocks that are mapped in the DAT Image for a particular process.
[[File:Mmappmap.png|none|thumb]]
[[File:Mmappmap.png|none|thumb]]


Line 90: Line 90:


=== Program Modules ===
=== Program Modules ===
The format of a program module varies depending on whether you have a single source file for your program or multiple source files. 


=== Data Modules ===
Note: multiple source files format cannot use direct page addressing due to the way they are compiled.
Here is a portion of the font module, which is a data module in NitrOS-9.


<code>               nam       font</code>
==== Single Source File Format ====
The following is a skeleton for a single source file memory module (program).
<code>********************************************************************</code>
<code>* [Program Name]</code>
<code>* [Description]</code>
<code>*</code>
<code>* by [Author Name]</code>
<code>*</code>
<code>* Edt/Rev  YYYY/MM/DD  Modified by</code>
<code>* Comment</code>
<code>* ------------------------------------------------------------------</code>
<code>              nam      [executablename]</code>
<code>              ttl      [Full Program Name]</code>
<code>              ifp1</code>
<code>              use      defsfile</code>
<code>              endc</code>
<code>tylg          set      Prgrm+Objct</code>
<code>atrv          set      ReEnt+rev</code>
<code>rev            set      $00</code>
<code>edition        set      1</code>
<code>              '''mod      eom,name,tylg,atrv,start,size'''</code>
<code>        [Data Section - Variables and Data Structures Here]</code>
<code>size          equ      .</code>
<code>name          fcs      /[executablename]/</code>
<code>              fcb      edition</code>
<code>start</code>
<code>        [Program Section - Program Code Here]</code>
<code>        </code>
<code>              os9      F$Exit</code>
<code>              '''emod'''</code>
<code>eom            equ *</code>
<code>              end</code>
OS9 can load programs (modules) only if they are in module header format.  A module (program) is defined between the '''mod''' and '''emod''' pseudo instructions.


<code>               ttl       F256 font</code>
* '''mod:''' Creates the module header and starts CRC check. '''mod''' has 6 attributes:


<code>               use       defsfile</code>
# Module size (computed from the ''<u>eom</u>'' label at the end of the code - 2 bytes)
# Module name offset (offset where the string of the name is contained, here that is defined at the ''<u>name</u>'' label with the ''fcs'' pseudo instruction - 2 bytes
# Module type & language (in this case a Program in 6809 object code - 1 byte)
# Module attribute-revision (1 byte).  Attributes are reentrant (ReEnt) or not reentrant  (0).  Revision is a number 0-15.  If two or more modules with the same name are in the module directory, then OS-9 only keeps the highest revision number module.
# Module execution offset (here that is defined with the ''<u>start</u>'' label - 2 bytes)
# Module variable storage size (defined by the ''<u>size</u>'' label immediately after the variable area, includes stack space - 2 bytes)


<code>tylg           set       Data</code>
* '''emod''': Outputs the correct 3 byte CRC.  This is generated by the assembler over the entire module.


<code>atrv           set       ReEnt+rev</code>


<code>rev            set       $01</code>
For more information about memory modules, see OS-9 Fundamentals:


<code>               mod       eom,name,tylg,atrv,start,0</code>
[https://www.youtube.com/watch?v=YpoqTyrty8k OS-9 Fundamentals on YouTube by Boisy Pitre]


<code>name           fcs       /font/</code>
==== Multiple Source File Format ====


<code>start</code>


<code>L0000          fcb   $00,$00,$00,$00,$00,$00,$00,$00 </code>
=== Data Modules ===
 
Here is a portion of the font module, which is a data module in NitrOS-9.
<code>L0008          fcb   $7C,$82,$AA,$82,$BA,$92,$82,$7C </code>
 
<nowiki>*</nowiki>More data here
 
<code>               emod</code>
 
<code>eom            equ       *</code>


<code>               end</code>
<code>               nam       font</code>
<code>               ttl       F256 font</code>
</code>
<code>               use       defsfile</code>
</code>
<code>tylg           set       Data</code>
<code>atrv           set       ReEnt+rev</code>
<code>rev            set       $01</code>
</code>
<code>               mod       eom,name,tylg,atrv,start,0</code>
<code>name           fcs       /font/</code>
</code>
<code>start</code>
<code>L0000          fcb   $00,$00,$00,$00,$00,$00,$00,$00 </code>
<code>L0008          fcb   $7C,$82,$AA,$82,$BA,$92,$82,$7C </code>
</code>
<code>*More data here</code>
</code>
<code>               emod</code>
<code>eom            equ       *</code>
<code>               end</code>


== Writing Position Independent Code ==
== Writing Position Independent Code ==

Latest revision as of 03:43, 8 October 2024

Overview[edit | edit source]

Each process in NitrOS-9 is allocated its own 64K address space. When run, the program object code is loaded into the end of the memory space, and the variables and data structures are loaded into the beginning. In addition, NitrOS-9 keeps an separate per-process MMU table (DAT Image), so you must use F$MapBlk to map and F$ClrBlk to unmap 8K blocks in your process' address space. The operating system has its own 64K address space as well.

When a new process is created, registers U, Y, DP and SP are set with U = start of data area, Y = end of data area, DP = page # of beginning page, SP = end of data area + 1. Each process includes its own execution and data directories, along with std in, std out and std err.

Memory and DAT Image[edit | edit source]

mmap and pmap output the current status of the memory in the system. mmap shows which memory blocks are currently in use. pmap shows the blocks that are mapped in the DAT Image for a particular process.

Program and Data Module Structure[edit | edit source]

There are 9 different types of modules in 4 possible languages:

Module Types Language
Code Module Type Name Code Language
$1x Program Module Prgm $x0 Data (non-executable)
$2x Subroutine Module Sbrtn $x1 6809 Object Code
$3x Multi-module Multi $x2 BASIC09 I-Code
$4x Data Module Data $x3 PASCAL I-Code
$5x-$Bx User-definable module $x4-$xF Reserved for Future Use
$Cx OS-9 System Module Systm
$Dx OS-9 File Manager Module FlMgr
$Ex OS-9 Device Driver Module Drivr
$Fx OS-9 Device Descriptor Module Devic

The two most used will be Program Modules and Data Modules.

Program Modules[edit | edit source]

The format of a program module varies depending on whether you have a single source file for your program or multiple source files.

Note: multiple source files format cannot use direct page addressing due to the way they are compiled.

Single Source File Format[edit | edit source]

The following is a skeleton for a single source file memory module (program).

********************************************************************
* [Program Name]
* [Description]
*
* by [Author Name]
*
* Edt/Rev  YYYY/MM/DD  Modified by
* Comment
* ------------------------------------------------------------------
               nam       [executablename]
               ttl       [Full Program Name]

               ifp1
               use       defsfile
               endc

tylg           set       Prgrm+Objct
atrv           set       ReEnt+rev
rev            set       $00
edition        set       1

               mod       eom,name,tylg,atrv,start,size

        [Data Section - Variables and Data Structures Here]

size           equ       .
name           fcs       /[executablename]/
               fcb       edition

start

        [Program Section - Program Code Here]
        
               os9       F$Exit

               emod
eom            equ *
               end

OS9 can load programs (modules) only if they are in module header format. A module (program) is defined between the mod and emod pseudo instructions.

  • mod: Creates the module header and starts CRC check. mod has 6 attributes:
  1. Module size (computed from the eom label at the end of the code - 2 bytes)
  2. Module name offset (offset where the string of the name is contained, here that is defined at the name label with the fcs pseudo instruction - 2 bytes
  3. Module type & language (in this case a Program in 6809 object code - 1 byte)
  4. Module attribute-revision (1 byte). Attributes are reentrant (ReEnt) or not reentrant (0). Revision is a number 0-15. If two or more modules with the same name are in the module directory, then OS-9 only keeps the highest revision number module.
  5. Module execution offset (here that is defined with the start label - 2 bytes)
  6. Module variable storage size (defined by the size label immediately after the variable area, includes stack space - 2 bytes)
  • emod: Outputs the correct 3 byte CRC. This is generated by the assembler over the entire module.


For more information about memory modules, see OS-9 Fundamentals:

OS-9 Fundamentals on YouTube by Boisy Pitre

Multiple Source File Format[edit | edit source]

Data Modules[edit | edit source]

Here is a portion of the font module, which is a data module in NitrOS-9.

               nam       font
               ttl       F256 font

               use       defsfile

tylg           set       Data
atrv           set       ReEnt+rev
rev            set       $01

               mod       eom,name,tylg,atrv,start,0
name           fcs       /font/

start
L0000          fcb   $00,$00,$00,$00,$00,$00,$00,$00  
L0008          fcb   $7C,$82,$AA,$82,$BA,$92,$82,$7C 

*More data here

               emod
eom            equ       *
               end

Writing Position Independent Code[edit | edit source]

All code in NitrOS-9 must be position independent. Addresses are not known until a program module is loaded and executed. All addressing must be "program counter relative addressing". All branch and long branch instructions are program counter relative.

Rules:

Use BRA and LBRA instead of JMP

Use BSR and LBSR instead of JSR

Use program counter relative indexed addressing for all load, store, arithmetic and logical instructions.

Accessing Variables and Data Structures[edit | edit source]

Accessing Table Data with PCR[edit | edit source]