pOS - portable Operating System
Still Questions
05/06/03


0. IMPORTANT: What for do I use kernel.h include ???
- because we use the pos.c library that contains a
   list of function pointers set to the kernel symbols
   preceded with "__"

1. How to manage to use more than one fs at a time ?

	- see /linux/include/linux/fs.h !
	
	- we must implement a "mount" and "umount" commands
	  to mount/unmount a device using a specified fs,
	  the kernel would then store a pointer to a fs
	  access struct containing pointers to all fs
	  functions

	- smallest chunks in linux fs's are called inodes,
	  this struct contains an union (yes !) with space
	  for all supported filesystem-structs


2. How integrate the bios-code into kernel (boot.c instead of boot.S->inline asm ?) ?
   A. BIOS compiled into kernel: kernel build is not portable
   B. BIOS external, kernel accesses via system.map

   A would be better: more compact, BIOS is guaranteed to be latest:

   compile bios.S with luna, link with lld to *.obj from kernel

   OR

   integrate all code for hardware i/o into kernel
   (linux-like)

  - there's no bios code anymore, we now use drivers and some
    few arch-specific code (entry.s, boot.s), so this question
    is obsolete and thus irrelevant

3. There is a concept missing how to manage disk (pfs).
   - yes, as of 05/06/03 there's still no native filesystem,
     but first kernel versions will presumably use
     an existing filesystem like minix etc

4. How to enable convenient kernel configuration ?
   - at first config.h in pos/include will be edited
     manually, later there will be configuration
     tool allowing convinient configuration
     (see config.txt)

5. How to manage the problem with differnt compilers on different CPU's
   having different sizes for all types (int, long etc) ?
   - we do not use the types directly, but rather define the
     required types (BYTE, WORD, DWORD etc) from the original
     types, e.g.: #define BYTE unsigned char
     (not anymore, I don't like these #define-declarations,
      they remind me too much of Micro$oft code)

6. How at all does Linux I/O Subsystem work, what drivers does Linux
   bring for HDD/FDD I/O, does it replace BIOS completely ?

   - linux does replace almost all i/o drivers for serial,
     ide/scsi, except for vesa bios and apm

   linux is layered as follows:

   kernel
	  |
	  --> driver
	         |
                 --> system related assembler code as macros (include/asm/*.*) (e.g."out", "in" on x86)

7. How does Linux process the "system.map" file, which sub-routines are listed
   there, is this list created at link time ?

   - the file is created at link time (by the linker), it is obsolete and a holdover from older
     kernel versions, it is still created for old linux applications compatibility

8. Does Linux take advantage of the protected mode architecture of i386 ?
   If yes, does 68k support protected mode, if no, how is the system
   protected from illegal memory accesses ?

  Of course it does !! Indeed the 68k series also have protected
  modes.

9. There is still the global problem unanswered how "char *" is
   treated by any compiler (MSVC: "It is a very special, const type")

   - "char *" is a constant pointer to char, right, it can be
     initialized with a constant (!) string, but not with
     a volatile string (i.e. user input), we have to use
     malloc () to allocate memory for dynamic strings
     (who the hell asked such stupid newbie-question ???)

10. Have another look at the 1541 file system, max # of sectors, tracks, heads

11. In what language is the linux mem-management written (speed issue, 
    portability, c vs asm) ?

	- in C

12. How does the printf function work (VAR ARG !) ?
   - there is a variable va_alist of type va_list
     automatically, internally defined by the compiler,
     it stores the address to the function parameters
     and these are accessed through a set of macros
     (va_start, va_arg, va_end ...)
     see /usr/include/vararg.h for more information

13. Is UART 8250 register compatible with 16550 ?

14. How to call sysinit code as entry code with the code
    being the first part in the binary kernel image like
    depicted here:

	-----------------
	| Sysinit	| $0002 (on C64)
	|		|
	-----------------
	| other Kernel	|
	| code		|
	-----------------

    - just specify the entry object code as
      first object in the object list

15. What is the difference between an idle status and an sleeping
    status of a process ???
    - interesting question, I don't know actually :), is there
      really an "idle"-status ???? (I only know right now from
      an "sleeping"-status)

16. why does Lunix interpret ESCAPE codes after "sh" has been
    started but "init" does interpret these codes, i thought
    that the escape codes are interpreted by "console.s" that is
    internal to the kernal (same goes for Linux)

To answer these questions...

...check out the Linux kernel source
...check out the LUnix kernel source
...check out some DO$/Window$ resources