Re: [MiNT] patch:MiNT:single-task

[Alan Hourihane <alanh@fairlite.co.uk>]

On Wed, 2010-06-02 at 19:04 +0000, Helmut Karlowski wrote:
> Alan Hourihane wrote:
> 
> > On Wed, 2010-06-02 at 17:56 +0000, Helmut Karlowski wrote:
> > > Alan Hourihane wrote:
> > >
> > > Also there could come up problems when the bits 16,17 are set in p_flags
> > > and the kernel does not support ST-mode, because I suspect these are
> > > also used internally by MiNT for memory-protection-control. So at least
> > > the part that resets these bits should go into any kernel.
> >
> > Eh? Not sure I follow. If MP is using these bits they you'd need to find
> > some other bits regardless.
> 
> I'm not using them - they just signal to the kernel and get translated
> into another variable, then reset. Anyway they could be set by mistake
> regardless of single-task etc. I'm not sure if it's easy to find a bit
> in p_flags that is not used by MiNT.
> 
> The high-word of p_flags should always get zeroed when launching a prog IMHO.

If they don't now then you are changing semantics.

> > But if you mean someone using the new XaAES with an older kernel would
> > break, then maybe you need to examine something in /kern to detect a
> > featureset available. As that could happen regardless of the #ifdef.
> 
> It would just not work, and XaAES would not notice anything.
> So the only place to tell the user would be the kernel by ALERT or similar.

Then there's no need to change anything. Just wrap up your code in
#ifdef SINGLE_TASK and we're done.

> >
> > Alan.
> >
> > PS. You seem to be removing the MiNT list from your replies, either that
> > or you are BCC'ing the MiNT list.
> 
> 
> You mean this:
> 
> Alan Hourihane <mint@lists.fishpool.fi>
> 
> ?
> 
> I thought it would be a real clever way to reply ;) It's done by my
> handmade macro-driven E-Mailer. Does it cause any problems?

That's pretty horrible. You are claiming people can reach me at
mint@lists.fishpool.fi in the archives. I don't like it.

Alan.