[Mesa-dev] [PATCH 1/3] st/xorg: reorder exa context creation and use screen param queries

[Lucas Stach]

Hello Michel,

I've just sent out a v2 of this series, which also fixes a resource leak
in the EDID patch.

Am Mittwoch, den 18.07.2012, 11:59 +0200 schrieb Michel Dänzer:
> On Mit, 2012-07-18 at 01:26 +0200, Lucas Stach wrote: 
> > Gives the x-server a more accurate description of the exa hardware
> > capabilities.
> 
> [...]
> 
> > @@ -1012,8 +1017,11 @@ xorg_exa_init(ScrnInfoPtr pScrn, Bool accel)
> >  #ifdef EXA_MIXED_PIXMAPS
> >     pExa->flags            |= EXA_MIXED_PIXMAPS;
> >  #endif
> > -   pExa->maxX              = 8191; /* FIXME */
> > -   pExa->maxY              = 8191; /* FIXME */
> > +   if (!exa->scrn->get_param(exa->scrn, PIPE_CAP_NPOT_TEXTURES))
> > +      pExa->flags         |= EXA_OFFSCREEN_ALIGN_POT;
> 
> I don't think EXA_OFFSCREEN_ALIGN_POT really makes sense for st/xorg. It
> only affects the pitch, the pixmap dimensions can still be NPOT. I'm not
> sure it makes sense trying to make st/xorg work with Gallium drivers
> that don't support PIPE_CAP_NPOT_TEXTURES.
> 
Ok, I've dropped this hunk.

> Ideally, st/xorg should provide an EXA CreatePixmap2 hook instead of the
> current CreatePixmap and ModifyPixmapHeader hooks.
> 
> 
> > +   pExa->maxX = pExa->maxY =
> > +   1 << (exa->scrn->get_param(exa->scrn, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1);
> 
> This is okay for now, though in the long run we'll probably need
> separate caps for hardware such as R300 with different rendering /
> sampling size limits.
> 
Yep, will revisit later.

> 
> Your other patches look good, do you have access to push them?

No, I don't have commit access. Could you please push them for me?

Thanks,
Lucas