[Mesa-dev] [PATCH 19.5/64] isl/state: Don't set RenderTargetViewExtent for texture surfaces

[Jason Ekstrand]

On Mon, Jun 20, 2016 at 2:47 PM, Chad Versace <chad.versace at intel.com>
wrote:

> On Sat 18 Jun 2016, Jason Ekstrand wrote:
> > The docs specify that this only matters for render targets and surfaces
> > used with typed dataport messages.  On some platforms (gen4-6) the Depth
> > field has more bits than RenderTargetViewExtent so we can have textures
> > with more levels than we can render to.
>
>
> I really want to see the commit message as a comment in the code,
> because it's subtle and surprising. Without such a comment, it's
> a complete mystery why the assignment is guarded by the usage flags.
> After all, the PRMs say RenderTargetViewExtent is ignored when the
> surface is neither a render target nor dataport surface; so
> (rhetorically) why doesn't set RenderTargetViewExtent unconditionally?
>

Fair enough.  I'll add a comment


>
> Insert some variant of the commit message as comment, and this is
> Reviewed-by: Chad Versace <chad.versace at intel.com>
>

Thanks!


> Also, how did you find this subtle problem? Did a pack function throw an
> assertion failure?


Ian wanted me to leave the max size of 3-D textures at 2048 on gen4-5
(since they don't do layered rendering) and it would have had problems in
that case.  How did I find it?  I knew it was a problem from when I found
the 3-D texture size issue.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20160620/ebabebc6/attachment-0001.html>