[PATCH v2 04/17] etnaviv: Use only DRAW_INSTANCED on GC3000+

[Lucas Stach]

Am Mittwoch, den 22.11.2017, 18:16 +0100 schrieb Wladimir J. van der Laan:
> Hello Lucas,
> 
> On Wed, Nov 22, 2017 at 02:29:33PM +0100, Lucas Stach wrote:
> > Hi Wladimir,
> > 
> > Am Samstag, den 18.11.2017, 10:44 +0100 schrieb Wladimir J. van der Laan:
> > > The blob does this, as DRAW_INSTANCED can replace fully all the other
> > > draw commands. It is also required to handle integer vertex formats.
> > > The other path is only there for compatibility and might go away (or at
> > > least rot to become buggy due to dis-use) in newer hardware.
> > > 
> > > As a by-effect this changes the behavior for GC3000-, by no longer using
> > > the index offset for DRAW_INDEXED but instead adding it to INDEX_ADDR.
> > > This should make no difference.
> > > 
> > > Preparation for GC7000 support.
> > 
> > I haven't looked into it much yet, but this commit breaks QT5 GUI
> > rendering on GC3000 for me. It seems like the DRAWs get dropped on the
> > floor with nothing being rendered. I didn't spot anything obviously
> > wrong in this patch from a quick look, so would be glad if you could
> > look into this.
> 
> Are you possibly running an older kernel that doesn't allow DRAW_INSTANCED
> in the command stream filter?

No, this is a 4.14 kernel and it doesn't complain about the cmdstream.

> I did test this patch series on GC3000, fairly sure that included qt5.
> 
> But yes let's leave out this patch for now, or have the change in behavior only for
> GC7000.

I would really like to know what's wrong with this patch, as using the
new draw command should be fine on GC3000 and we certainly want to
support instanced drawing at some point.

Regards,
Lucas