To: vxp@imaginet.fr ("centek" )
From: dlanor@oden.se (Ronald Andersson)
Organization: Atari
Subject: Re: Full turbo speed for MagiC (solved)
X-Mailer: NEWSie Version 0.94 (Atari)

>Hello
>
Hello Sacha


>  I don't understand the goal of your patch. Magic doesn't
>need another patch that the centek ones.

MagiC did boot in TOS 4.x mode without extra patches, but only
without access to any FastRAM, which is what my patch provides.
It simply tells MagiC (booted in TOS 4.x mode) that it should
use the FastRAM as 'alternate' RAM at base adress $04000000.


MagiC did _NOT_ boot properly in TOS 7.x mode however, on any
of the many attempts I made (spread over several days).

What would happen is that the boot would start, as usual, in
TOS 7.x mode, and then when MAGXBOOT.PRG was executed it did
load MAGIC.RAM into the FastRAM and begin to execute it.
This would work for a while, getting access to the harddisk
boot sector and loading any XFS drivers (I also tried without).

After the XFS files were loaded, or after the harddisk driver
boot if no XFS were active, the entire system would freeze.
Regardless of variations in the boot setup this freezing
occurred 100% consistently in all attempts.

This is the kind of response I would expect from a program
that attempts to zero the TC register so as to reprogram the
PMMU while running in RAM whose address is already relocated
by the PMMU (as your FastRAM solution in TOS 7.x remaps it).
That would inevitably cause this kind of 'freezing'.


>  For the buserror in rom, it's due to the ct2 in normal mode
>only, because i make the right thing in tos 7. And Magic can
>start in tos 7, so...

No it couldn't...!
Are we really talking of the same things here ?

I am talking of MagiC 6, whereas the Centek docs only mention
MagiC 4 and 5.  In earlier mails I have been told that no patch
at all should be needed for MagiC 6, but the results of my tests
show that this is only true when booting in TOS 4.x mode, but
not when booting in TOS 7.x mode.

I've tried it a large number of times, and it _NEVER_ worked...


>  After, the fast ram is detect by magic during magxboot. The
>magic.ram is copying in it.

Sure, and it worked a little further than that too, but before
it had reached the \AUTO\ programs the system had frozen up.


>  So, i spend only a few minutes to see that you work on something
>allready done... sorry

Do you really believe I would waste time on making a patch for some
thing that worked well without it...?

Of course I wouldn't !

I wrote the patch because it was the only way I could boot MagiC 6
properly and get access to the FastRAM.  Later I also found it to
be useful for non-MagiC purposes, since the Maddalt call is also
supported by TOS 4.x.  Thus I can choose where in the booting of
the \AUTO\ folder the TOS should be told to start using FastRAM,
which proved useful to avoid some incompatibility problems that
made some software unsafe or ineffective to use under TOS 7.x.


At present I can't make any experiments with the CT2B, since it
is back at Centek for an update of the decoding logic.  It is
possible that this will change its behaviour so that I can then
boot MagiC in the way that you describe when I get it back.
I will mail you again on this subject once I have tested that.

