[Intel-gfx] [PATCH 16/37] drm/hdlcd|mali: Drop drm_vblank_cleanup

[Daniel Vetter]

On Wed, May 31, 2017 at 05:57:07PM +0100, Liviu Dudau wrote:
> On Wed, May 31, 2017 at 06:41:05PM +0200, Daniel Vetter wrote:
> > On Wed, May 31, 2017 at 1:22 PM, Liviu Dudau <liviu.dudau at arm.com> wrote:
> > > On Wed, May 31, 2017 at 01:03:34PM +0200, Daniel Vetter wrote:
> > >> On Wed, May 31, 2017 at 12:57 PM, Liviu Dudau <liviu.dudau at arm.com> wrote:
> > >> > On Wed, May 24, 2017 at 04:51:51PM +0200, Daniel Vetter wrote:
> > >> >> IRQs are properly shut down, so it almost works as race-free shutdown.
> > >> >> Except the irq is stopped after the vblank stuff, so boom anyway.
> > >> >> Proper way would be to call drm_atomic_helper_shutdown before any of
> > >> >> the kms things gets stopped. So no harm in removing the
> > >> >> drm_vblank_cleanup here really.
> > >> >
> > >> > Slightly confused on the implied message in the commit text: is "Proper way
> > >> > would be to call drm_atomic_helper_shutdown" a hint? A promise of a future
> > >> > patch? A message to the future us on how to fix things if they blow up?
> > >> >
> > >> > If calling drm_atomic_helper_shutdown() is the proper thing to do, why does the
> > >> > patch (and the series) avoids doing that? Lack of understanding of the driver's
> > >> > internal workings? Then I want to help, if I can understand the new direction.
> > >>
> > >> Yes, I wanted to not make things worse. If you look at the overall
> > >> result (especially last patch) I'm also trying to better document
> > >> stuff in the vblank area, but summarized, if you want to make sure
> > >> that vblank processing has stopped, you need to call drm_vblank_off on
> > >> each active crtc. The simplest way to get that is by using
> > >> drm_atomic_helper_shutdown(). Calling drm_vblank_cleanup doesn't
> > >> really do anything useful (see the last patch for the only valid
> > >> usecase there ever was).
> > >
> > > OK, so there should be a follow up patch adding drm_atomic_helper_shutdown(). I guess
> > > I need to call dibs on it? :)
> > 
> > I googled what "to call dibs on smth" means and still can't figure out
> > what you mean here. Can you pls explain?
> 
> Means I put my name on it. Urban dictionary on "call dibs" gives it a very gender
> specific usage, my intent was to convey the message that I will take it on my TODO list.

Ah ok, thanks for explaining. I was indeed thrown off be the ub entry ...
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch