Date:	Mon, 29 Nov 1999 08:45:54 +0200 (IST)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender:	eliz AT is
To:	djgpp AT delorie DOT com
Subject:	Re: nearptr access to memory mapped devices
In-Reply-To:	<81s582$bmt$1@nntp8.atl.mindspring.net>
Message-ID:	<Pine.SUN.3.91.991129084537.14681I-100000@is>
MIME-Version:	1.0
Reply-To:	djgpp AT delorie DOT com
X-Mailing-List:	djgpp AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On Sun, 28 Nov 1999, Marp wrote:

> I have a question about accessing a memory mapped device using a near
> pointer. I was reading through the vbe 3.0 spec and it suggested the
> following to access the linear frame buffer:

I'm not sure VBE spec is relevant for your case, because VBE talks
about a device that is already mapped into the address space (SVGA),
while an arbitrary memory-mapped device is not.  See below.

> void *lfbptr;
> __dpmi_meminfo meminf;
> unsigned long baseaddr;
> 
> /* step 1 */
> meminf.address = address_of_lfb;
> meminf.size = size_of_lfb;
> __dpmi_physical_address_mapping(&meminf);
> 
> /*step 2 */
> __dpmi_get_segment_base_address(_my_ds(), &baseaddr);
> 
> /* step 3 */
> lfbptr = meminf.address - baseaddr;
> /* code ends */
> 
> The problem is that when I go to use lfbptr, I get SIGSEGV.

Please re-read section 18.7 of the FAQ.  It shows you that a few
additional DPMI calls are needed before you can access the
memory-mapped device.  Specifically, your call to
__dpmi_physical_address_mapping mapped the device into memory, but you
don't have any selector set up to access that memory.  That's why it
SIGSEGVs.

If you want nearptr access to that device, you need to extend the
limit of the DS selector to cover the addresses up to the top of the
memory on board the device.

Why do you need to mess with nearptr, anyway?  Isn't the method
suggested in the FAQ good enough?  It *does* work.

- Raw text -