aperature driver

Michael Richardson

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

Unless there are some objections, I'd like to commit an aperature driver
to at least -i386 that will compile into a kernel. I don't like LKMs, but
I need the aperature driver to make my notebook work. (XFree crashes
when it tries to bank switch, which I have reported)
It would not be enabled in GENERIC.

I'd love to do this for 1.5, but I think it is too late?

:!mcr!: | Solidum Systems Corporation, http://www.solidum.com
Michael Richardson | now at 1575 Carling Avenue... still moving in
Personal: <A HREF="http://www.sandelman.ottawa.on.ca/People/Michael_Richardson/Bio.html">m...@sandelman.ottawa.on.ca</A>. PGP key available.
Corporate: <A HREF="mailto:m...@solidum.com">m...@solidum.com</A>.

John Hawkinson

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

| I'd love to do this for 1.5, but I think it is too late?

If it is self-contained it may be pulled up to 1.5.

--jhawk

Jason R Thorpe

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

On Fri, Jul 21, 2000 at 10:44:06AM -0400, Michael Richardson wrote:

> I'd love to do this for 1.5, but I think it is too late?

If it includes the XF86 MTRR code, please no!

--
-- Jason R. Thorpe <tho...@zembu.com>

Oleg Polyanski

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

Jason R Thorpe <tho...@zembu.com> writes:

> > I'd love to do this for 1.5, but I think it is too late?
>
> If it includes the XF86 MTRR code, please no!

Why?

Jason R Thorpe

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

On Fri, Jul 21, 2000 at 09:45:02PM +0400, Oleg Polyanski wrote:

> > > I'd love to do this for 1.5, but I think it is too late?
> >
> > If it includes the XF86 MTRR code, please no!
>
> Why?

Because that code needs a lot of cleaning up.

Jason R Thorpe

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

On Fri, Jul 21, 2000 at 09:45:02PM +0400, Oleg Polyanski wrote:

> > > I'd love to do this for 1.5, but I think it is too late?
> >
> > If it includes the XF86 MTRR code, please no!
>
> Why?

Also, using ioctls for it is totally wrong; it should be done w/
sysarch(2).

Michael Richardson

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

>>>>> "Jason" == Jason R Thorpe <tho...@zembu.com> writes:

Jason> On Fri, Jul 21, 2000 at 10:44:06AM -0400, Michael Richardson wrote:

>> I'd love to do this for 1.5, but I think it is too late?

Jason> If it includes the XF86 MTRR code, please no!

I'm not certain what that is. It is just the "ap.c" that was in
pub/NetBSD/arch/i386 for ever.

Oleg Polyanski

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

Jason R Thorpe <tho...@zembu.com> writes:

> > > > I'd love to do this for 1.5, but I think it is too late?
> > >

> > > If it includes the XF86 MTRR code, please no!
> >

> > Why?
>
> Also, using ioctls for it is totally wrong; it should be done w/
> sysarch(2).

Agreed. But this API was invented by FreeBSD crowd not by XFree86
team. I've cleaned up this code once and replaced ioctl's with
sysarch calls but serious hardware failure prevented me from
finalizing the work. Hope, I will finish this weekend and could be
able to submit version 2 of patches to Frank. Patches include MTRR
support for P2+ and K6+ CPUs, some syscalls routed through sysarch
and piece of code for XFree86.

Oleg Polyanski

unread,

Jul 21, 2000, 3:00:00 AM7/21/00

to

Michael Richardson <m...@sandelman.ottawa.on.ca> writes:

> >> I'd love to do this for 1.5, but I think it is too late?
>

> Jason> If it includes the XF86 MTRR code, please no!
>
> I'm not certain what that is. It is just the "ap.c" that was in
> pub/NetBSD/arch/i386 for ever.

No, Jason means another (updated) version of aperture driver that
requires MTRR support in kernel to work with XFree86 4 and later.