[PATCH] Remove spurious use of MI_SET_CONTEXT.

[Kristian Høgsberg]

On Dec 3, 2007 3:16 PM, Kristian Høgsberg <krh at bitplanet.net> wrote:
> On Dec 3, 2007 2:22 PM, Keith Whitwell <keith at tungstengraphics.com> wrote:
...
> > It is also necessary to do an MI_SET_CONTEXT in case there is a 3d
> > client out there *really using* MI_SET_CONTEXT.
>
> Yes, but for that to work there has to be a MI_SET_CONTEXT somewhere
> in the system that actually restores the saved state.  The patch
> removes the only MI_SET_CONTEXT use in the entires stack and all it
> ever did was to save the state.

Oh, let me just retract that.  You're right of course, if a 3d client
relied on MI_SET_CONTEXT to restore its state, the 2d driver will have
to do this step, even though the 2d driver does use the
state-restoring feature on MI_SET_CONTEXT.  Having this kind of
assumptions spread across the stack is too fragile though, and I think
enforcing this in the DRM as I described above is a much better way to
go.  That's what I'm currently trying to get running here.

Kristian