[Bug 47007] HDMI monitor polling causing 100ms rendering stalls

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=47007

--- Comment #9 from Tvrtko Ursulin <tvrtko.ursulin at onelan.co.uk> 2012-03-08 01:56:12 PST ---
Created attachment 58160
  --> https://bugs.freedesktop.org/attachment.cgi?id=58160
Patch which makes the stalls go away

Here is the patch which makes the stalls go away. However, I am not that
confident we got to the bottom of this.

I was looking at how is detection supposed to work on HPD interrupt. What
happens is that poll helper gets scheduled, but, since the connector is now not
in polled mode, that won't do anything to update it's status.

In our use case things work because when no connectors are in connected status
we keep probing them manually. I _suspect_ this translates into
drm_helper_probe_single_connector_modes call, right?

But without this userspace behaviour, and with this patch, connector status
will not get updated on it's own. At least this is my understanding and why I
said that I don't think we got to the bottom of it.

Side question - are connector detect methods supposed to be re-entrant? I am
not sure they are looking at radeon_dvi_detect. And as far as I can see they
can be running simultaneously from drm_helper_probe_single_connector_modes and
output_poll_execute, no?

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.