From: Martin Stromberg <Martin DOT Stromberg AT epl DOT ericsson DOT se>
Message-Id: <200108271601.SAA15957@merope.lu.erisoft.se>
Subject: Re: UMB access with %gs selector
To: djgpp-workers AT delorie DOT com
Date: Mon, 27 Aug 2001 18:01:39 +0200 (MET DST)
In-Reply-To: <4.2.0.58.20010827163125.00bf6e90@ics.u-strasbg.fr> from "Pierre Muller" at Aug 27, 2001 04:41:28 PM
X-Mailer: ELM [version 2.5 PL3]
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
X-MIME-Autoconverted: from 8bit to quoted-printable by penguin.wise.edt.ericsson.se id f7RG1ev12911
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by delorie.com id MAA10452
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> 
> At 15:56 27/08/01 , vous avez écrit:
> > >     Doesn't this mean that the higher part of a index is not used?
> > > (i.e. if  $esi contain 0x10008,
> > > will
> > >     movl  %gs:(%esi),%eax
> > > load into %eax
> > > the content of linear address 0x10008 or 0x00008 ??)
> > >
> > > Can we really access to UMB area with a 16 bit selector ?

> I think that you misunderstood my question.

Oh, sorry!

>    I know the segment register are 16 bit registers but I was talking about
> something completely different:
>    The descriptor of each selector allocated as a field that specifies if 
> the OS should
> consider it as a 16 bit or a 32 bit data area.
> 
>    But as the instruction explicitly have a 16 bit or 32 bit nature
> (depending on some CR0 field IIRC) I don't know really if this

No, it depends on the presence/absense of the address prefix and what
the entry in GDT/LDT is set to. If I understood you correctly that
selector points to a 16-bit segment, so the presence of the address
prefix would imply 32-bit.

> does influence the reading past the first megabyte or not.

Hmm. Above you're talking about UMB and now past the first megabyte.

>    My question is more to know if the OS shouldn't forbid to
> access anything above the 0xffff memory position relative to selector base 
> if the
> selector is marked as a 16 bit selector.

This ought to depend on limit of the selector, not on the "bitness" of
the selector, but I'm not sure.


Right,

						MartinS