Trouble firing up DRI with a PCI ATi Rage Mobility M

Leif Delgass ldelgass at gmail.com
Wed Nov 2 11:30:51 PST 2005


On 11/2/05, Eric Anholt <eta at lclark.edu> wrote:
> On Wed, 2005-11-02 at 10:15 -0500, Joshua Coombs wrote:
> > "Joshua Coombs" <jcoombs at gwi.net> wrote in message
> > news:dk5da9$i6p$1 at sea.gmane.org...
> > > I'm running FreeBSD 6.0RC1, xorg 6.8.2 built from ports, dri driver
> > > package built from ports, based on a cvs snap from 7/19/2005.
> > >
> > > FreeBSD is seeing the card as dri capable:
> > > drm0: <Rage Mobility P/M> port 0x1800-0x18ff mem
> > > 0xfd000000-0xfdffffff,0xfc103000-0xfc103fff irq 9 at device 20.0 on
> > > pci0
> > > info: [drm] Initialized mach64 1.0.0 20020904 on minor 0
> > >
> > > xorg however never seems to load dri even though it is conf'd to do
> > > so.  I have my dmesg/conf/logs at http://test.x386.net/x11 and can
> > > post more info as needed.  The laptop in question is a Fujitsu
> > > Lifebook P2040 if that helps.
> > >
> > > Joshua Coombs
> >
> > No suggestions eh?  Drat.
> >
> > Anyone care to offer a description on mmio_cache and test_mmio_cache?
> > I can't find them in r128 or Radeon man page, nor can I find an ATi
> > driver man page.  Just curious if there is any potential gain or if I
> > should avoid those like the plague with my mach_64 derived video?
>
> Since the mach64 DRI is insecure, it's disabled by default in
> xorg-server.  It's turned on in xorg-server-snap, though.
>
> I have no idea what you're talking about with mmio_cache.
>
> --
> Eric Anholt                                     eta at lclark.edu
> http://people.freebsd.org/~anholt/              anholt at FreeBSD.org

mmio_cache is an option to do lazy reads/writes of registers.  The
driver keeps a cache of register values and only reads/writes over the
bus when necessary to update the hardware values or to refresh the
driver's view of registers after e.g. a VT switch.  It is enabled by
default.  The test_mmio_cache option is just for debugging, it
verifies that the driver's cache matches the actual register values in
the hardware.

--
Leif Delgass



More information about the xorg mailing list