[Intel-gfx] [RFC] drm/i915/dp: Remove i915.enable_dp_mst module parameter

[Dhinakaran Pandiyan]

On Thu, 2018-10-04 at 08:54 -0700, Rodrigo Vivi wrote:
> On Thu, Oct 04, 2018 at 01:59:52PM +0300, Ville Syrjälä wrote:
> > On Thu, Oct 04, 2018 at 10:03:39AM +0300, Jani Nikula wrote:
> > > On Wed, 03 Oct 2018, Dhinakaran Pandiyan <
> > > dhinakaran.pandiyan at intel.com> wrote:
> > > > MST is enabled by default on all platforms that support it. I
> > > > don't think
> > > > we should be providing a switch to work around MST issues as
> > > > the feature
> > > > has been supported for a while now. Let's kill this module
> > > > parameter
> > > > that we also do not test in CI.
> > > 
> > > I agree we don't want to provide this to users to *work around*
> > > issues. But maybe we want something like this to *debug* issues?
> > 
> > Yes. I was using it for that just a few days ago when looking at a
> > bug.
> 
> so it seems useful and it means that we need to move to debugfs :)
> 
It also allows us to have a switch for each primary connector instead
of disabling MST completely.


> > 
> > Also the mst code lacks a bunch of features I think we'd want
> > (remote dpcd,
> > remote i2c write, maybe others). It's still the unloved stepchild
> > with no
> > one really focusing on improving it.
> > 
Because things work (mostly) without them :) 

But yeah, remote dpcd reads can be very useful for debugging. Why are
remote i2c writes needed though?


-DK


> > So I think it's way too early to think about removing this
> > outright.
> > Not sure we should ever remove it really. What happens if in the
> > future
> > most of our ci displays are mst capable? Do we just not test sst at
> > all?
> > Granted a modparam is a probably a bit too coarse for that, but I
> > think
> > we may want *something* to force sst.
> > 
> > -- 
> > Ville Syrjälä
> > Intel
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/intel-gfx