Re: [Emutos-devel] Undocumented Pexec features

[Martin <mdo...@us...>]

On Tue, 13 May 2003 18:58:00 +0200
Thomas Huth <tho...@in...> wrote:

>=20
>  Hi all!
=20
Hi, Thomas!

This weekend I did come back from my holidays in Hamburg (well, it was
really nice this time). I'm very happy to see, that you did go a bit
deeper into this problem. I will do some work on EmuTOS too today.
Though, I will go on with my work at the VDI.



> I've found another problem that was causing trouble with EmuTOS at the
> moment: In proc_go() (in bdos/proc.c) that is used by Pexec() to
> launch programs, there was following piece of code:
>=20
> #if 0
>     /* the following settings are not documented, and hence
>     theoretically=20
>      * the assignments below are not necessary.
>      */
>     {   /* d1-d7/a0-a2 and dummy return address set to zero */
>         int i;
>         for(i =3D 0; i < 11 ; i++)=20
>             sp->fill[i] =3D 0;
>     }
>     /* ... */
> #endif

Here I thought, how I ever should have know this thing about ACCs... :-|
 Why does A0 show, how a program is started? Is this the "official" way
to determine it, or is it just a hack? Hmm, you tell it undocumented...
Can I read about this somewhere?



Well, I have a question about some comment. I read:

struct gouser_stack {
  LONG other_sp;   /* a4, the other stack pointer */
  WORD sr;         /* d0, the status register */
  LONG retaddr;    /* a3, the return address */
  LONG fill[11];   /* 10 registers d1-d7/a0-a2 and one dummy so that ...
*/
  PD * basepage;   /* ... upon startup the basepage is in 4(sp) */
};


What is meant with "upon startup"? Does it mean at the end of the
initialisation, which is documented in the first comment ?

What do you think... Should we wait with a release, till we solved the
pending problem?

--
Gru=DF, Martin