Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 06 Nov 1997 11:51:11 +1030
From:      Mike Smith <mike@smith.net.au>
To:        Poul-Henning Kamp <phk@critter.freebsd.dk>
Cc:        Mike Smith <mike@smith.net.au>, Nate Williams <nate@mt.sri.com>, freebsd-mobile@freebsd.org
Subject:   Re: Libretto 50 - US Version and PAO 
Message-ID:  <199711060121.LAA00464@word.smith.net.au>
In-Reply-To: Your message of "Wed, 05 Nov 1997 15:29:03 BST." <1014.878740143@critter.freebsd.dk> 

next in thread | previous in thread | raw e-mail | index | archive | help
> >No, we're out.  If we want to be able to run Linux WABI, it wants 
> >(AFAIR) everything from 11 or 12 and up.  At any rate, I think I have a 
> >solution which deposes the APM descriptors and makes them "general 
> >purpose" ones.  I'll know Friday when I get to test it.
> 
> they What ??? 
> 
> They must be out of their mind :-(

Quite possibly.  Willows TWIN wants (by default) everything from 12 up. 
At least we have source for that.  Check with sos about WABI; he was 
closer to having it run than I was.

> >>  I do not like the
> >> idea of trying to go through the vm86 interface to get to the APM.
> >
> >I was only *ever* proposing this for dealing with APM biossen that 
> >don't support the 32-bit interface.  I don't actually know if it's 
> >worth the effort though, as we don't support such systems now...
> 
> Ahh, ok.

Oh, and doing the connects, for obvious reasons.

> >Hmm, I should get you to add my extra PnP rummaging and see which event 
> >signalling mechanism it uses.  In particular, I'm interested to know if 
> >anyone uses the 'system device' method rather than the 'polled flag' 
> >method.
> 
> polled flag in my case.

Thought so.  The 'system device' method would involve about 10x as much 
code I think.

> >I have an interface (the PnP protected-mode interface, FWIW) which 
> >expects to be called via a "far call", ie. lcall seg:ofs.  It also 
> >wants to take arguments on the (16-bit) stack.  I'm trying to work out 
> >how to write an assembly shim to make this possible, as gcc doesn't 
> >grok "far" in any way shape or form.
> >
> >Presuming I have the segments created correctly, how do I make a call 
> >into another segment with arbitrary arguments on the stack?  (I need to 
> >arrange a new data segment as well...)  I'm stuffed on this one, short 
> >of custom-rolling a shim for each argument arrangement...
> 
> I looked at that one.  My idea was to do two levels of wrappers:

Yeah, these are the ideas I had.  I wanted implementation suggestions 8)

> 1.  (Inline) Assembly
> 	call_pnp(segment, offset, int stackdepth, u_int16_t *stack)

You left the data segment out...

One thing I was wondering; does

pnp_func(u_int16_t a, u_int16_b)

construct the stack properly, or will we still get int-sized objects on 
the stack?

mike





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711060121.LAA00464>