[Intel-gfx] [PATCH] drm/i915: Handle changing enable_fbc parameter at runtime better.

[Rodrigo Vivi]

On Tue, Mar 06, 2018 at 09:12:02PM +0100, Maarten Lankhorst wrote:
> Op 06-03-18 om 21:02 schreef Rodrigo Vivi:
> > On Tue, Mar 06, 2018 at 11:22:16AM +0100, Maarten Lankhorst wrote:
> >> Op 05-03-18 om 19:50 schreef Rodrigo Vivi:
> >>> On Mon, Mar 05, 2018 at 01:36:08PM +0100, Maarten Lankhorst wrote:
> >>>> If i915.enable_fbc is cleared at runtime, but FBC was previously enabled
> >>>> then we don't disable FBC until the next time the crtc is disabled.
> >>>>
> >>>> Make sure that if the module param is changed, we disable FBC in
> >>>> intel_fbc_post_update so we never have to worry about disabling.
> >>> What about switching this from a parameter to debugfs toggle like drrs?
> >> There are still places where people recommend booting with i915.enable_fbc=0, this would break if we moved it to debugfs.
> > I think this was exactly what Chris and Joonas were trying to avoid when removing enable_rc6.
> > To avoid random recommendations of unvalidated combinations.
> >
> >> Having a small fix is enough.
> >>
> > Well... it would be fine for me.
> >
> > But I went to look the patch and I don't know what exactly we are trying to fix here.
> >
> > I don't know if it is safe to disable it like this when crtc is already enabled
> > and mostly: why?! why just not wait until next modeset?
> >
> > Cc: Paulo
> I'm trying to get rid of all modesets in kms_frontbuffer_tracking, they add up on glk where it's 50% of the total runtime for kms tests. :)

oh! this is a very good reason indeed. ;)

Patch looks correct to me. my only concern is with the reliability of toggling this on/off without
the modeset. But if there is problem with this I believe the test cases on CI will get this.

Besides I just remembered that I did some toggle on/off on sysfs at somepoint when playing with
that fbc false color and I never saw a problem... so:

Reviewed-by: Rodrigo Vivi <rodrigo.vivi at intel.com>

> 
> ~Maarten