[PATCH] drm: Avoid forcing a detection cycle following a hotplug event

[Chris Wilson]

On Sat, Jun 08, 2013 at 09:28:17AM +0200, Laurent Pinchart wrote:
> Could you please also update Documentation/DocBook/drm.tmpl ?

It looks out of context there, as nothing explains the hotplug ->
fill_modes -> probe -> detect loop...

How about:

  <title>Modes</title>
  <synopsis>int (*fill_modes)(struct drm_connector *connector, uint32_t max_width,
	      uint32_t max_height, bool force);</synopsis>
  <para>
    Fill the mode list with all supported modes for the connector. If the
    <parameter>max_width</parameter> and <parameter>max_height</parameter>
    arguments are non-zero, the implementation must ignore all modes wider
    than <parameter>max_width</parameter> or higher than
    <parameter>max_height</parameter>. The driver may use the existing
    connector status, unless <parameter>force</parameter> is passed. During a
    hotplug event, the driver may already have updated its knowledge of the
    output and so may simply refresh the modes list from the information it
    acquired whilst handling the event. However, the caller may explicitly
    request that any cached information be dropped, and for the output to be
    queried for its current status and modes - under such circumstances
    <parameter>force</parameter> is true.
  </para>
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre