Re: [PATCH i-g-t 03/15] tests/kms_plane_scaling: Use for_each_crtc*()

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Fri, Jan 30, 2026 at 02:03:44PM +0200, Jani Nikula wrote:
> On Fri, 30 Jan 2026, Ville Syrjala <ville.syrjala@linux.intel.com> wrote:
> > -				for_each_pipe(&data.display, pipe) {
> > -					igt_dynamic_f("pipe-%s", kmstest_pipe_name(pipe)) {
> > -						for_each_valid_output_on_pipe(&data.display, pipe, output) {
> > +				for_each_crtc(&data.display, crtc) {
> > +					igt_dynamic_f("pipe-%s", igt_crtc_name(crtc)) {
> 
> First, a higher level meta comment here. I'm having mild second thoughts
> about igt_crtc_name() returning kmstest_pipe_name().
> 
> What if, in the future, we want to use CRTC name? And make that
> different from the pipe name? If we percolate igt_crtc_name() all over
> the place, the switch is going to be difficult. (Mostly because it's
> being used in test names.)
> 
> Should we have igt_crtc_name() return something based on the CRTC index
> instead? And have, say, igt_crtc_pipe_name() return kmstest_pipe_name(),
> and use that in these conversions?
> 
> Or, more generally, maybe igt_crtc_name() should be the vendor neutral
> DRM KMS name, and, say, igt_crtc_vendor_name() should return something
> that's specific to the vendor. For Intel, it would be pipe, since that's
> mostly Intel specific thing. For others, it might be something else.
> 
> Or am I overcomplicating, and we should just cross that bridge when we
> come to it?

Shrug. I suppose for now igt_crtc_name() is always the pipe name,
so if at some point want to start using some other naming scheme
we could start that off with a global
s/igt_crtc_name()/igt_crtc_pipe_name()/ or whatever. So doesn't
really seem to matter what we call it for now. But I can rename
it now as well, if you think it's a good idea.

-- 
Ville Syrjälä
Intel