On Wed, Dec 17, 2014 at 09:12:31AM -0500, Rob Clark wrote:
> But I am still not a huge fan of hiding props for atomic drivers via
> legacy getprop/setprop interfaces.. if nothing else it will prevent
> them from showing up in modetest.  And it seems kind of inconsistent
> and unnecessary.

By default at least the intel X driver reads/writes _all_ props it finds.
Maybe just on connectors, but even there we have the crtc prop. Which
means with the current legacy set_prop helpers we'll end up doing tons of
modesets, and that will piss of users. So I think we don't have a choice
but hide them.

In i915 all the legacy props have careful checks to short-circuit such
noop changes out, exactly for this reasons. Adding all these short-circuit
checks would pretty badly complicate helpers and hence is something I've
explicitly left out (for now at least).

And the PROP_ATOMIC flag is a useful hint to userspace to maybe not do
stupid things with that property just for fun.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

Reply via email to