[Intel-gfx] [PATCH 2/2] Fix up some pI830->memory_manager versus use_drm_mode confusion.

[Jesse Barnes]

On Wed, 22 Apr 2009 12:48:24 -0700
Eric Anholt <eric at anholt.net> wrote:

> On Wed, 2009-04-22 at 08:46 -0700, Jesse Barnes wrote:
> > On Tue, 21 Apr 2009 20:27:35 -0700
> > Eric Anholt <eric at anholt.net> wrote:
> > 
> > > We'd end up doing direct drm_intel_bo_maps of our tiled buffers
> > > for sw fallbacks, resulting in bad rendering with KMS.  Also
> > > cleans up a spurious error message.x
> > > ---
> > >  src/i830_driver.c |    2 +-
> > >  src/i830_exa.c    |    4 ++--
> > >  src/i830_memory.c |    7 ++++---
> > >  3 files changed, 7 insertions(+), 6 deletions(-)
> > 
> > Wow this really makes me wonder how things ever worked, nice catch.
> 
> Nope, this was what you mentioned on IRC, the broken warning fix.  We
> do currently allocate "aperture space" (not really) for the memory
> manager, it just has no effect except for filling in the pointer.

Yeah I always meant to clean that up.  I optimistically thought we'd be
doing our big cleanup way back when KMS went in so it wouldn't be in
there for long.  Now that we're cleaning things up though we may as
well rip that stuff out and apply something like your patch (or just
remove all the ->memory_manager checks altogether and assume we have
one).

-- 
Jesse Barnes, Intel Open Source Technology Center