[Mesa-dev] [PATCH 1/7] gallium: fold u_trim_pipe_prim call from st/mesa to drivers

[Ilia Mirkin]

On Fri, Apr 14, 2017 at 12:42 PM, Marek Olšák <maraeo at gmail.com> wrote:
> On Fri, Apr 14, 2017 at 5:45 PM, Ilia Mirkin <imirkin at alum.mit.edu> wrote:
>> On Fri, Apr 14, 2017 at 11:07 AM, Marek Olšák <maraeo at gmail.com> wrote:
>>> diff --git a/src/gallium/drivers/nouveau/nv30/nv30_vbo.c b/src/gallium/drivers/nouveau/nv30/nv30_vbo.c
>>> index bc9b9a1..295c394 100644
>>> --- a/src/gallium/drivers/nouveau/nv30/nv30_vbo.c
>>> +++ b/src/gallium/drivers/nouveau/nv30/nv30_vbo.c
>>> @@ -543,20 +543,23 @@ nv30_draw_elements(struct nv30_context *nv30, bool shorten,
>>>     }
>>>  }
>>>
>>>  static void
>>>  nv30_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
>>>  {
>>>     struct nv30_context *nv30 = nv30_context(pipe);
>>>     struct nouveau_pushbuf *push = nv30->base.pushbuf;
>>>     int i;
>>>
>>> +   if (!u_trim_pipe_prim(info->mode, (unsigned*)&info->count))
>>> +      return;
>>> +
>>
>> Should this also have a !info->primitive_restart? It's supported on
>> nv4x (covered by this driver).
>
> In that case, I wonder if u_trim_pipe_prim is required with this
> driver. It might be better to just remove that call.

Based on a quick look, this seems to exist to prevent short draws and
trim the count to the nearest prim size, i.e. if you try to draw a tri
with %3 != 0 vertices, or a line with %2 != 0? I'm not 100% sure that
the NV30 HW handles those correctly, but it probably does. I can
double-check tonight, as I have one plugged in these days.

Cheers,

  -ilia