Re: A89: Re: FYI: added info to J89hw.txt
[Prev][Next][Index][Thread]
Re: A89: Re: FYI: added info to J89hw.txt
thanks!
so why do we put these nop? to be sure that's it's at least three reads, but
what could do that it's not without these nops?
Jeremie Knusel
----- Original Message -----
From: "Johan" <johei804@student.liu.se>
To: <assembly-89@lists.ticalc.org>
Sent: Tuesday, August 08, 2000 6:37 PM
Subject: Re: A89: Re: FYI: added info to J89hw.txt
>
> On Tue, Aug 08, 2000 at 05:30:51PM +0200, Olle Hedman wrote:
> > Jeremie Knusel wrote:
> > > nop ; Execute three NOPs to make SURE it's
actually
> > > nop ; three read accesses from within $21xxxx
> > > nop
> > > move.w #$2700,sr ; "make sure nobody's trying to do stupid things"
> > > move.w d0,(a0) ; Do the $1Cxxxx write (any value in D0 will do)
> > >
> > > where are the three reads from 20xxxx or 21xxxx that should be done
just
> > > before
> > > move.w d0,(a0) ?
> >
> > The three nops. The code is run from somewhere in 20xxxx to 21xxxx and
the
> > reads are the instruction fetches.
>
>
> The bus activity when executing the code fragment above is:
>
> fetch NOP instruction
> fetch NOP instruction
> fetch NOP instruction
> fetch MOVE #xxx,SR instruction
> -> fetch immediate data to MOVE instruction ("$2700")
> -> fetch MOVE D0,(A0) instruction
> -> fetch instruction following the MOVE
> -> write to $1Cxxxx (execute the MOVE instruction)
> fetch next instruction etc...
>
> The four lines with "->" are the important ones. As you can see, the NOPs
> are not necessary! Note that the write is "delayed" -- the 68000 has a
> one-word prefetch queue. (Who said you never learn anything useful at
> school?!? :-) )
>
>
> /Johan "100 bytes is enough."
>
Follow-Ups:
References: