Date: Mon, 8 Dec 1997 16:13:13 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Fabrice ILPONSE <fabrice AT trash DOT lip6 DOT fr>
cc: djgpp AT delorie DOT com, satoh AT asd DOT tnes DOT nec DOT co DOT jp
Subject: Re: Physical address of convenstional memory
In-Reply-To: <34869DF7.55D0@trash.lip6.fr>
Message-ID: <Pine.SUN.3.91.971208161220.4200Z-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Thu, 4 Dec 1997, Fabrice ILPONSE wrote:

> To answer the question, it's very dangerous. As the memory is remap, it
> can be swapped to the disk so the information is no more at a physical
> adress. First i think the DOS memory have to be locked (i think it's
> possible) so the data won't be swapped. This is one point. Now where to
> get this adress. Perhaps in the TSS or a V86 structure(if it's not the
> same) but i don't know enough to help you about those structures.

You don't need to go to these lengths, AFAIK.  If the physical address
is required for the DMA transfers, then just use whatever
`__dpmi_allocate_dos_memory' returned as if it were mapped 1:1.
AFAIK, Windows transparently traps all DMA requests and substitutes
the actual physical address for the logical one you use.  (Otherwise,
real-mode DOS programs that use DMA would never work.)