Direct Rendering Infrastructure

Alan Hourihane wrote:
> Does anyone know why the radeon & r128 drivers call FLUSH_RING when in
> the BlockHandler ?

No -- seems unnecessary.  Perhaps a hangover from the days when 2D accel was 
disabled with dri?

Keith

On Tue, 2003-09-30 at 15:58, Alan Hourihane wrote:
> Does anyone know why the radeon & r128 drivers call FLUSH_RING when in
> the BlockHandler ?

This is the best way to ensure that the indirect buffer gets flushed
according to Mark Vojkovich. Without it, drawing operations could get
stuck in the indirect buffer for quite a while. See the dri-devel
archives.


-- 
Earthling Michel Dänzer   \  Debian (powerpc), XFree86 and DRI developer
Software libre enthusiast  \     http://svcs.affero.net/rm.php?r=daenzer

Michel Dänzer wrote:
> On Tue, 2003-09-30 at 15:58, Alan Hourihane wrote:
> 
>>Does anyone know why the radeon & r128 drivers call FLUSH_RING when in
>>the BlockHandler ?
> 
> 
> This is the best way to ensure that the indirect buffer gets flushed
> according to Mark Vojkovich. Without it, drawing operations could get
> stuck in the indirect buffer for quite a while. See the dri-devel
> archives.

Ah, yes.  What's confusing here is the naming of the macro:  FLUSH_RING() 
doesn't actually have anything to do with the ring, but rather sends the 
buffer of commands being built up in the X server off to the kernel.

Perhaps renaming the macro to FLUSH_INDIRECT_BUFFER() would help make things 
clearer?

Keith

On Tue, Sep 30, 2003 at 04:20:23PM +0100, Keith Whitwell wrote:
> Michel D=E4nzer wrote:
> >On Tue, 2003-09-30 at 15:58, Alan Hourihane wrote:
> >
> >>Does anyone know why the radeon & r128 drivers call FLUSH_RING when i=
n
> >>the BlockHandler ?
> >
> >
> >This is the best way to ensure that the indirect buffer gets flushed
> >according to Mark Vojkovich. Without it, drawing operations could get
> >stuck in the indirect buffer for quite a while. See the dri-devel
> >archives.
>=20
> Ah, yes.  What's confusing here is the naming of the macro:  FLUSH_RING=
()=20
> doesn't actually have anything to do with the ring, but rather sends th=
e=20
> buffer of commands being built up in the X server off to the kernel.
>=20
> Perhaps renaming the macro to FLUSH_INDIRECT_BUFFER() would help make=20
> things clearer?

Sounds good.=20

Alan.