[Mesa-dev] RFC: 0001-r600g-R700-can-do-more-than-8-tex-and-vtx-clauses.patch

[Christian König]

Am Mittwoch, den 09.03.2011, 12:03 +0100 schrieb Henri Verbeet:
> 2011/3/9 Christian König <deathsimple at vodafone.de>:
> > before I create another regression with my patches, could you guys
> > please test the attached patch and see if it causes another regression
> > on your hardware? Especially evergreen and any of the R600 chipsets?
> >
> I didn't test the patch, do you have a specific application that needs
> this? 
Piglits glsl-routing and texrect-many tests are failing if this isn't
working right for example.

> I have some general comments anyway. First, the patch introduces
> several whitespace errors. I've noticed that in some of your other
> commits as well, so you may want to spend some time on fixing your
> editor there. I'm not really a fan of using tabs for indentation
> either, but at least try to keep things consistent. 
I'm using Code::Blocks most of the time, but since mesa doesn't seems to
have a consistent coding convention I sometimes have trouble switching
back and forth between the different editor settings. I even reverted to
using a good old vim for editing some parts of the code. Any ideas how
to improve this?

> Second,
> introducing a function to build the CF vtx/tex clause initiation
> instructions and raising the vtx/tex clause instruction limits are
> really two different changes. Note that
> r600_bc_num_tex_and_vtx_clauses() may be a bit misnamed, since the
> limit is on the number of instructions in a clause, not the number of
> clauses.
> 
> > +static int r600_bc_num_tex_and_vtx_clauses(struct r600_bc *bc)
> "bc" should be const, and I think unsigned makes more sense as a return type.
> 
> > +static void r600_bc_cf_vtx_build(struct r600_bc *bc, struct r600_bc_cf *cf, unsigned id)
> "cf" should be const, and I don't think it makes much sense to pass
> "bc" and "id" separately.
> 
> How about the attached two patches instead?
Yes they indeed look better now, thanks. If you don't find any more
regressions I will push them.

By the way: What should be the correct line to add to the commit
message, "Reviewed-by: Henri Verbeet <hverbeet at gmail.com>"?

Thanks,
Christian.